1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001 Free Software Foundation, Inc.
2 @c This is part of the GCC manual.
3 @c For copying conditions, see the file gcc.texi.
7 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
8 1998, 1999, 2000, 2001 Free Software Foundation, Inc.
10 Permission is granted to make and distribute verbatim copies of this
11 manual provided the copyright notice and this permission notice are
12 preserved on all copies.
14 Permission is granted to copy and distribute modified versions of this
15 manual under the conditions for verbatim copying, provided also that the
16 entire resulting derived work is distributed under the terms of a
17 permission notice identical to this one.
19 Permission is granted to copy and distribute translations of this manual
20 into another language, under the above conditions for modified versions,
21 except that this permission notice may be included in translations
22 approved by the Free Software Foundation instead of in the original
25 @c Set file name and title for the man page.
27 @settitle GNU project C and C++ compiler
29 gcc [@samp{-c}|@samp{-S}|@samp{-E}] [@samp{-std=}@var{standard}]
30 [@samp{-g}] [@samp{-pg}] [@samp{-O}@var{level}]
31 [@samp{-W}@var{warn}...] [@samp{-pedantic}]
32 [@samp{-I}@var{dir}...] [@samp{-L}@var{dir}...]
33 [@samp{-D}@var{macro}[=@var{defn}]...] [@samp{-U}@var{macro}]
34 [@samp{-f}@var{option}...] [@samp{-m}@var{machine-option}...]
35 [@samp{-o} @var{outfile}] @var{infile}...
37 Only the most useful options are listed here; see below for the
38 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
41 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
42 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
43 @file{ld}, @file{binutils} and @file{gdb}.
46 For instructions on reporting bugs, see
47 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
48 script to report bugs is recommended.
51 See the Info entry for @file{gcc}, or
52 @w{@uref{http://gcc.gnu.org/thanks.html}}, for contributors to GCC.
57 @chapter GCC Command Options
58 @cindex GCC command options
59 @cindex command options
60 @cindex options, GCC command
62 @c man begin DESCRIPTION
64 When you invoke GCC, it normally does preprocessing, compilation,
65 assembly and linking. The ``overall options'' allow you to stop this
66 process at an intermediate stage. For example, the @samp{-c} option
67 says not to run the linker. Then the output consists of object files
68 output by the assembler.
70 Other options are passed on to one stage of processing. Some options
71 control the preprocessor and others the compiler itself. Yet other
72 options control the assembler and linker; most of these are not
73 documented here, since you rarely need to use any of them.
75 @cindex C compilation options
76 Most of the command line options that you can use with GCC are useful
77 for C programs; when an option is only useful with another language
78 (usually C++), the explanation says so explicitly. If the description
79 for a particular option does not mention a source language, you can use
80 that option with all supported languages.
82 @cindex C++ compilation options
83 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
84 options for compiling C++ programs.
86 @cindex grouping options
87 @cindex options, grouping
88 The @command{gcc} program accepts options and file names as operands. Many
89 options have multi-letter names; therefore multiple single-letter options
90 may @emph{not} be grouped: @samp{-dr} is very different from @w{@samp{-d
93 @cindex order of options
94 @cindex options, order
95 You can mix options and other arguments. For the most part, the order
96 you use doesn't matter. Order does matter when you use several options
97 of the same kind; for example, if you specify @samp{-L} more than once,
98 the directories are searched in the order specified.
100 Many options have long names starting with @samp{-f} or with
101 @samp{-W}---for example, @samp{-fforce-mem},
102 @samp{-fstrength-reduce}, @samp{-Wformat} and so on. Most of
103 these have both positive and negative forms; the negative form of
104 @samp{-ffoo} would be @samp{-fno-foo}. This manual documents
105 only one of these two forms, whichever one is not the default.
110 * Option Summary:: Brief list of all options, without explanations.
111 * Overall Options:: Controlling the kind of output:
112 an executable, object files, assembler files,
113 or preprocessed source.
114 * Invoking G++:: Compiling C++ programs.
115 * C Dialect Options:: Controlling the variant of C language compiled.
116 * C++ Dialect Options:: Variations on C++.
117 * Language Independent Options:: Controlling how diagnostics should be
119 * Warning Options:: How picky should the compiler be?
120 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
121 * Optimize Options:: How much optimization?
122 * Preprocessor Options:: Controlling header files and macro definitions.
123 Also, getting dependency information for Make.
124 * Assembler Options:: Passing options to the assembler.
125 * Link Options:: Specifying libraries and so on.
126 * Directory Options:: Where to find header files and libraries.
127 Where to find the compiler executable files.
128 * Spec Files:: How to pass switches to sub-processes.
129 * Target Options:: Running a cross-compiler, or an old version of GCC.
130 * Submodel Options:: Specifying minor hardware or convention variations,
131 such as 68010 vs 68020.
132 * Code Gen Options:: Specifying conventions for function calls, data layout
134 * Environment Variables:: Env vars that affect GCC.
135 * Running Protoize:: Automatically adding or removing function prototypes.
139 @section Option Summary
141 Here is a summary of all the options, grouped by type. Explanations are
142 in the following sections.
145 @item Overall Options
146 @xref{Overall Options,,Options Controlling the Kind of Output}.
148 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language}
149 -v --target-help --help
152 @item C Language Options
153 @xref{C Dialect Options,,Options Controlling C Dialect}.
155 -ansi -std=@var{standard} -fno-asm -fno-builtin
156 -fhosted -ffreestanding
157 -trigraphs -traditional -traditional-cpp
158 -fallow-single-precision -fcond-mismatch
159 -fsigned-bitfields -fsigned-char
160 -funsigned-bitfields -funsigned-char
161 -fwritable-strings -fshort-wchar
164 @item C++ Language Options
165 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
167 -fno-access-control -fcheck-new -fconserve-space
168 -fdollars-in-identifiers -fno-elide-constructors
169 -fno-enforce-eh-specs -fexternal-templates
170 -falt-external-templates
171 -ffor-scope -fno-for-scope -fno-gnu-keywords -fhonor-std
172 -fhuge-objects -fno-implicit-templates
173 -fno-implicit-inline-templates
174 -fno-implement-inlines -fms-extensions
175 -fname-mangling-version-@var{n} -fno-operator-names
176 -fno-optional-diags -fpermissive
177 -frepo -fno-rtti -fsquangle -ftemplate-depth-@var{n}
178 -fuse-cxa-atexit -fvtable-thunks -nostdinc++
179 -fno-default-inline -Wctor-dtor-privacy
180 -Wnon-virtual-dtor -Wreorder
181 -Weffc++ -Wno-deprecated
182 -Wno-non-template-friend -Wold-style-cast
183 -Woverloaded-virtual -Wno-pmf-conversions
187 @item Language Independent Options
188 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
190 -fmessage-length=@var{n}
191 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}
194 @item Warning Options
195 @xref{Warning Options,,Options to Request or Suppress Warnings}.
197 -fsyntax-only -pedantic -pedantic-errors
198 -w -W -Wall -Waggregate-return
199 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment
200 -Wconversion -Wdisabled-optimization -Werror
201 -Wfloat-equal -Wformat -Wformat=2
202 -Wformat-nonliteral -Wformat-security
203 -Wid-clash-@var{len} -Wimplicit -Wimplicit-int
204 -Wimplicit-function-declaration
205 -Werror-implicit-function-declaration
207 -Wlarger-than-@var{len} -Wlong-long
208 -Wmain -Wmissing-declarations
209 -Wmissing-format-attribute -Wmissing-noreturn
210 -Wmultichar -Wno-format-extra-args -Wno-format-y2k
211 -Wno-import -Wpacked -Wpadded
212 -Wparentheses -Wpointer-arith -Wredundant-decls
213 -Wreturn-type -Wsequence-point -Wshadow
214 -Wsign-compare -Wswitch -Wsystem-headers
215 -Wtrigraphs -Wundef -Wuninitialized
216 -Wunknown-pragmas -Wunreachable-code
217 -Wunused -Wunused-function -Wunused-label -Wunused-parameter
218 -Wunused-value -Wunused-variable -Wwrite-strings
221 @item C-only Warning Options
223 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs
224 -Wstrict-prototypes -Wtraditional
227 @item Debugging Options
228 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
230 -a -ax -d@var{letters} -fdump-unnumbered -fdump-translation-unit-@var{file}
231 -fpretend-float -fprofile-arcs -ftest-coverage
232 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2
233 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+
234 -p -pg -print-file-name=@var{library} -print-libgcc-file-name
235 -print-prog-name=@var{program} -print-search-dirs -Q
239 @item Optimization Options
240 @xref{Optimize Options,,Options that Control Optimization}.
242 -falign-functions=@var{n} -falign-jumps=@var{n}
243 -falign-labels=@var{n} -falign-loops=@var{n}
244 -fbranch-probabilities -fcaller-saves
245 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections -fdce
246 -fdelayed-branch -fdelete-null-pointer-checks
247 -fexpensive-optimizations -ffast-math -ffloat-store
248 -fforce-addr -fforce-mem -ffunction-sections -fgcse
249 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions
250 -fkeep-static-consts -fmove-all-movables
251 -fno-default-inline -fno-defer-pop
252 -fno-function-cse -fno-inline -fno-math-errno -fno-peephole
253 -fomit-frame-pointer -foptimize-register-move
254 -foptimize-sibling-calls -freduce-all-givs
255 -fregmove -frename-registers
256 -frerun-cse-after-loop -frerun-loop-opt
257 -fschedule-insns -fschedule-insns2
258 -fsingle-precision-constant -fssa
259 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv
260 -funroll-all-loops -funroll-loops
261 -O -O0 -O1 -O2 -O3 -Os
264 @item Preprocessor Options
265 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
267 -$ -A@var{question}=@var{answer} -A-@var{question}[=@var{answer}]
269 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H
271 -include @var{file} -imacros @var{file}
272 -iprefix @var{file} -iwithprefix @var{dir}
273 -iwithprefixbefore @var{dir} -isystem @var{dir} -isystem-c++ @var{dir}
274 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap
275 -trigraphs -undef -U@var{macro} -Wp,@var{option}
278 @item Assembler Option
279 @xref{Assembler Options,,Passing Options to the Assembler}.
285 @xref{Link Options,,Options for Linking}.
287 @var{object-file-name} -l@var{library}
288 -nostartfiles -nodefaultlibs -nostdlib
289 -s -static -static-libgcc -shared -shared-libgcc -symbolic
290 -Wl,@var{option} -Xlinker @var{option}
294 @item Directory Options
295 @xref{Directory Options,,Options for Directory Search}.
297 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}
301 @c I wrote this xref this way to avoid overfull hbox. -- rms
302 @xref{Target Options}.
304 -b @var{machine} -V @var{version}
307 @item Machine Dependent Options
308 @xref{Submodel Options,,Hardware Models and Configurations}.
310 @emph{M680x0 Options}
311 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040
312 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020
313 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel
314 -malign-int -mstrict-align
316 @emph{M68hc1x Options}
317 -m6811 -m6812 -m68hc11 -m68hc12
318 -mauto-incdec -mshort -msoft-reg-count=@var{count}
325 -mtune=@var{cpu type}
326 -mcmodel=@var{code model}
328 -mapp-regs -mbroken-saverestore -mcypress
329 -mepilogue -mfaster-structs -mflat
330 -mfpu -mhard-float -mhard-quad-float
331 -mimpure-text -mlive-g0 -mno-app-regs
332 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu
333 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles
334 -msoft-float -msoft-quad-float -msparclite -mstack-bias
335 -msupersparc -munaligned-doubles -mv8
337 @emph{Convex Options}
338 -mc1 -mc2 -mc32 -mc34 -mc38
339 -margcount -mnoargcount
341 -mvolatile-cache -mvolatile-nocache
343 @emph{AMD29K Options}
344 -m29000 -m29050 -mbw -mnbw -mdw -mndw
345 -mlarge -mnormal -msmall
346 -mkernel-registers -mno-reuse-arg-regs
347 -mno-stack-check -mno-storem-bug
348 -mreuse-arg-regs -msoft-float -mstack-check
349 -mstorem-bug -muser-registers
352 -mapcs-frame -mno-apcs-frame
354 -mapcs-stack-check -mno-apcs-stack-check
355 -mapcs-float -mno-apcs-float
356 -mapcs-reentrant -mno-apcs-reentrant
357 -msched-prolog -mno-sched-prolog
358 -mlittle-endian -mbig-endian -mwords-little-endian
359 -malignment-traps -mno-alignment-traps
360 -msoft-float -mhard-float -mfpe
361 -mthumb-interwork -mno-thumb-interwork
362 -mcpu= -march= -mfpe=
363 -mstructure-size-boundary=
364 -mbsd -mxopen -mno-symrename
366 -mlong-calls -mno-long-calls
367 -mnop-fun-dllimport -mno-nop-fun-dllimport
368 -msingle-pic-base -mno-single-pic-base
372 -mtpcs-frame -mno-tpcs-frame
373 -mtpcs-leaf-frame -mno-tpcs-leaf-frame
374 -mlittle-endian -mbig-endian
375 -mthumb-interwork -mno-thumb-interwork
376 -mstructure-size-boundary=
377 -mnop-fun-dllimport -mno-nop-fun-dllimport
378 -mcallee-super-interworking -mno-callee-super-interworking
379 -mcaller-super-interworking -mno-caller-super-interworking
380 -msingle-pic-base -mno-single-pic-base
383 @emph{MN10200 Options}
386 @emph{MN10300 Options}
393 @emph{M32R/D Options}
394 -mcode-model=@var{model type} -msdata=@var{sdata type}
398 -m88000 -m88100 -m88110 -mbig-pic
399 -mcheck-zero-division -mhandle-large-shift
400 -midentify-revision -mno-check-zero-division
401 -mno-ocs-debug-info -mno-ocs-frame-position
402 -mno-optimize-arg-area -mno-serialize-volatile
403 -mno-underscores -mocs-debug-info
404 -mocs-frame-position -moptimize-arg-area
405 -mserialize-volatile -mshort-data-@var{num} -msvr3
406 -msvr4 -mtrap-large-shift -muse-div-instruction
407 -mversion-03.00 -mwarn-passed-structs
409 @emph{RS/6000 and PowerPC Options}
411 -mtune=@var{cpu type}
412 -mpower -mno-power -mpower2 -mno-power2
413 -mpowerpc -mpowerpc64 -mno-powerpc
414 -mpowerpc-gpopt -mno-powerpc-gpopt
415 -mpowerpc-gfxopt -mno-powerpc-gfxopt
416 -mnew-mnemonics -mold-mnemonics
417 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc
418 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe
419 -msoft-float -mhard-float -mmultiple -mno-multiple
420 -mstring -mno-string -mupdate -mno-update
421 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align
422 -mstrict-align -mno-strict-align -mrelocatable
423 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib
424 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian
425 -mcall-aix -mcall-sysv -mprototype -mno-prototype
426 -msim -mmvme -mads -myellowknife -memb -msdata
427 -msdata=@var{opt} -mvxworks -G @var{num}
430 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs
431 -mfull-fp-blocks -mhc-struct-return -min-line-mul
432 -mminimum-fp-blocks -mnohc-struct-return
435 -mabicalls -mcpu=@var{cpu type}
436 -membedded-data -muninit-const-in-rodata
437 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64
438 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1
439 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy
440 -mmips-as -mmips-tfile -mno-abicalls
441 -mno-embedded-data -mno-uninit-const-in-rodata -mno-embedded-pic
442 -mno-gpopt -mno-long-calls
443 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats
444 -mrnames -msoft-float
445 -m4650 -msingle-float -mmad
446 -mstats -EL -EB -G @var{num} -nocpp
447 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi
451 -mcpu=@var{cpu type} -march=@var{cpu type}
452 -mintel-syntax -mieee-fp -mno-fancy-math-387
453 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib
454 -mno-wide-multiply -mrtd -malign-double
455 -mreg-alloc=@var{list} -mregparm=@var{num}
456 -malign-jumps=@var{num} -malign-loops=@var{num}
457 -malign-functions=@var{num} -mpreferred-stack-boundary=@var{num}
458 -mthreads -mno-align-stringops -minline-all-stringops
459 -mpush-args -maccumulate-outgoing-args -m128bit-long-double
463 -march=@var{architecture type}
464 -mbig-switch -mdisable-fpregs -mdisable-indexing
465 -mfast-indirect-calls -mgas -mjump-in-delay
466 -mlong-load-store -mno-big-switch -mno-disable-fpregs
467 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas
468 -mno-jump-in-delay -mno-long-load-store
469 -mno-portable-runtime -mno-soft-float
470 -mno-space-regs -msoft-float -mpa-risc-1-0
471 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime
472 -mschedule=@var{cpu type} -mspace-regs
474 @emph{Intel 960 Options}
475 -m@var{cpu type} -masm-compat -mclean-linkage
476 -mcode-align -mcomplex-addr -mleaf-procedures
477 -mic-compat -mic2.0-compat -mic3.0-compat
478 -mintel-asm -mno-clean-linkage -mno-code-align
479 -mno-complex-addr -mno-leaf-procedures
480 -mno-old-align -mno-strict-align -mno-tail-call
481 -mnumerics -mold-align -msoft-float -mstrict-align
484 @emph{DEC Alpha Options}
485 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float
487 -mieee -mieee-with-inexact -mieee-conformant
488 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode}
489 -mtrap-precision=@var{mode} -mbuild-constants
491 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max
492 -mmemory-latency=@var{time}
494 @emph{Clipper Options}
497 @emph{H8/300 Options}
498 -mrelax -mh -ms -mint32 -malign-300
502 -m4-nofpu -m4-single-only -m4-single -m4
503 -mb -ml -mdalign -mrelax
504 -mbigtable -mfmovd -mhitachi -mnomacsave
505 -misize -mpadstruct -mspace
509 @emph{System V Options}
510 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}
514 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section}
515 -mdata=@var{data section} -mrodata=@var{readonly data section}
517 @emph{TMS320C3x/C4x Options}
518 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm
519 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload
520 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned
521 -mparallel-insns -mparallel-mpy -mpreserve-float
524 -mlong-calls -mno-long-calls -mep -mno-ep
525 -mprolog-function -mno-prolog-function -mspace
526 -mtda=@var{n} -msda=@var{n} -mzda=@var{n}
530 -m32032 -m32332 -m32532 -m32081 -m32381 -mmult-add -mnomult-add
531 -msoft-float -mrtd -mnortd -mregparam -mnoregparam -msb -mnosb
532 -mbitfield -mnobitfield -mhimem -mnohimem
535 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts
536 -mcall-prologues -mno-tablejump -mtiny-stack
539 -mhardlit, -mno-hardlit -mdiv -mno-div -mrelax-immediates
540 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields
541 -m4byte-functions -mno-4byte-functions -mcallgraph-data
542 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim
543 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment
546 @item Code Generation Options
547 @xref{Code Gen Options,,Options for Code Generation Conventions}.
549 -fcall-saved-@var{reg} -fcall-used-@var{reg}
550 -fexceptions -funwind-tables -ffixed-@var{reg}
551 -finhibit-size-directive -finstrument-functions
552 -fcheck-memory-usage -fprefix-function-name
553 -fno-common -fno-ident -fno-gnu-linker
554 -fpcc-struct-return -fpic -fPIC
555 -freg-struct-return -fshared-data -fshort-enums
556 -fshort-double -fvolatile -fvolatile-global -fvolatile-static
557 -fverbose-asm -fpack-struct -fstack-check
558 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym}
559 -fargument-alias -fargument-noalias
560 -fargument-noalias-global
566 * Overall Options:: Controlling the kind of output:
567 an executable, object files, assembler files,
568 or preprocessed source.
569 * C Dialect Options:: Controlling the variant of C language compiled.
570 * C++ Dialect Options:: Variations on C++.
571 * Language Independent Options:: Controlling how diagnostics should be
573 * Warning Options:: How picky should the compiler be?
574 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
575 * Optimize Options:: How much optimization?
576 * Preprocessor Options:: Controlling header files and macro definitions.
577 Also, getting dependency information for Make.
578 * Assembler Options:: Passing options to the assembler.
579 * Link Options:: Specifying libraries and so on.
580 * Directory Options:: Where to find header files and libraries.
581 Where to find the compiler executable files.
582 * Spec Files:: How to pass switches to sub-processes.
583 * Target Options:: Running a cross-compiler, or an old version of GCC.
588 @node Overall Options
589 @section Options Controlling the Kind of Output
591 Compilation can involve up to four stages: preprocessing, compilation
592 proper, assembly and linking, always in that order. The first three
593 stages apply to an individual source file, and end by producing an
594 object file; linking combines all the object files (those newly
595 compiled, and those specified as input) into an executable file.
597 @cindex file name suffix
598 For any given input file, the file name suffix determines what kind of
603 C source code which must be preprocessed.
606 C source code which should not be preprocessed.
609 C++ source code which should not be preprocessed.
612 Objective-C source code. Note that you must link with the library
613 @file{libobjc.a} to make an Objective-C program work.
616 Objective-C source code which should not be preprocessed.
619 C header file (not to be compiled or linked).
623 @itemx @var{file}.cxx
624 @itemx @var{file}.cpp
625 @itemx @var{file}.c++
627 C++ source code which must be preprocessed. Note that in @samp{.cxx},
628 the last two letters must both be literally @samp{x}. Likewise,
629 @samp{.C} refers to a literal capital C.
632 @itemx @var{file}.for
633 @itemx @var{file}.FOR
634 Fortran source code which should not be preprocessed.
637 @itemx @var{file}.fpp
638 @itemx @var{file}.FPP
639 Fortran source code which must be preprocessed (with the traditional
643 Fortran source code which must be preprocessed with a RATFOR
644 preprocessor (not included with GCC).
646 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
647 Using and Porting GNU Fortran}, for more details of the handling of
650 @c FIXME: Descriptions of Java file types.
656 @c GCC also knows about some suffixes for languages not yet included:
666 @itemx @var{file}.chi
667 CHILL source code (preprocessed with the traditional preprocessor).
673 Assembler code which must be preprocessed.
676 An object file to be fed straight into linking.
677 Any file name with no recognized suffix is treated this way.
680 You can specify the input language explicitly with the @samp{-x} option:
683 @item -x @var{language}
684 Specify explicitly the @var{language} for the following input files
685 (rather than letting the compiler choose a default based on the file
686 name suffix). This option applies to all following input files until
687 the next @samp{-x} option. Possible values for @var{language} are:
689 c c-header cpp-output
691 objective-c objc-cpp-output
692 assembler assembler-with-cpp
693 f77 f77-cpp-input ratfor
696 @c Also f77-version, for internal use only.
699 Turn off any specification of a language, so that subsequent files are
700 handled according to their file name suffixes (as they are if @samp{-x}
701 has not been used at all).
703 @item -pass-exit-codes
704 Normally the @command{gcc} program will exit with the code of 1 if any
705 phase of the compiler returns a non-success return code. If you specify
706 @samp{-pass-exit-codes}, the @command{gcc} program will instead return with
707 numerically highest error produced by any phase that returned an error
711 If you only want some of the stages of compilation, you can use
712 @samp{-x} (or filename suffixes) to tell @command{gcc} where to start, and
713 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
714 @command{gcc} is to stop. Note that some combinations (for example,
715 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
719 Compile or assemble the source files, but do not link. The linking
720 stage simply is not done. The ultimate output is in the form of an
721 object file for each source file.
723 By default, the object file name for a source file is made by replacing
724 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
726 Unrecognized input files, not requiring compilation or assembly, are
730 Stop after the stage of compilation proper; do not assemble. The output
731 is in the form of an assembler code file for each non-assembler input
734 By default, the assembler file name for a source file is made by
735 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
737 Input files that don't require compilation are ignored.
740 Stop after the preprocessing stage; do not run the compiler proper. The
741 output is in the form of preprocessed source code, which is sent to the
744 Input files which don't require preprocessing are ignored.
746 @cindex output file option
748 Place output in file @var{file}. This applies regardless to whatever
749 sort of output is being produced, whether it be an executable file,
750 an object file, an assembler file or preprocessed C code.
752 Since only one output file can be specified, it does not make sense to
753 use @samp{-o} when compiling more than one input file, unless you are
754 producing an executable file as output.
756 If @samp{-o} is not specified, the default is to put an executable file
757 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
758 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
759 all preprocessed C source on standard output.@refill
762 Print (on standard error output) the commands executed to run the stages
763 of compilation. Also print the version number of the compiler driver
764 program and of the preprocessor and the compiler proper.
767 Use pipes rather than temporary files for communication between the
768 various stages of compilation. This fails to work on some systems where
769 the assembler is unable to read from a pipe; but the GNU assembler has
773 Print (on the standard output) a description of the command line options
774 understood by @command{gcc}. If the @option{-v} option is also specified
775 then @option{--help} will also be passed on to the various processes
776 invoked by @command{gcc}, so that they can display the command line options
777 they accept. If the @option{-W} option is also specified then command
778 line options which have no documentation associated with them will also
782 Print (on the standard output) a description of target specific command
783 line options for each tool.
787 @section Compiling C++ Programs
789 @cindex suffixes for C++ source
790 @cindex C++ source file suffixes
791 C++ source files conventionally use one of the suffixes @samp{.C},
792 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
793 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
794 files with these names and compiles them as C++ programs even if you
795 call the compiler the same way as for compiling C programs (usually with
796 the name @command{gcc}).
800 However, C++ programs often require class libraries as well as a
801 compiler that understands the C++ language---and under some
802 circumstances, you might want to compile programs from standard input,
803 or otherwise without a suffix that flags them as C++ programs.
804 @command{g++} is a program that calls GCC with the default language
805 set to C++, and automatically specifies linking against the C++
806 library. On many systems, @command{g++} is also
807 installed with the name @command{c++}.
809 @cindex invoking @command{g++}
810 When you compile C++ programs, you may specify many of the same
811 command-line options that you use for compiling programs in any
812 language; or command-line options meaningful for C and related
813 languages; or options that are meaningful only for C++ programs.
814 @xref{C Dialect Options,,Options Controlling C Dialect}, for
815 explanations of options for languages related to C.
816 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
817 explanations of options that are meaningful only for C++ programs.
819 @node C Dialect Options
820 @section Options Controlling C Dialect
821 @cindex dialect options
822 @cindex language dialect options
823 @cindex options, dialect
825 The following options control the dialect of C (or languages derived
826 from C, such as C++ and Objective C) that the compiler accepts:
832 In C mode, support all ISO C89 programs. In C++ mode,
833 remove GNU extensions that conflict with ISO C++.
835 This turns off certain features of GCC that are incompatible with ISO
836 C (when compiling C code), or of standard C++ (when compiling C++ code),
837 such as the @code{asm} and @code{typeof} keywords, and
838 predefined macros such as @code{unix} and @code{vax} that identify the
839 type of system you are using. It also enables the undesirable and
840 rarely used ISO trigraph feature. For the C compiler,
841 it disables recognition of C++ style @samp{//} comments as well as
842 the @code{inline} keyword.
844 The alternate keywords @code{__asm__}, @code{__extension__},
845 @code{__inline__} and @code{__typeof__} continue to work despite
846 @samp{-ansi}. You would not want to use them in an ISO C program, of
847 course, but it is useful to put them in header files that might be included
848 in compilations done with @samp{-ansi}. Alternate predefined macros
849 such as @code{__unix__} and @code{__vax__} are also available, with or
850 without @samp{-ansi}.
852 The @samp{-ansi} option does not cause non-ISO programs to be
853 rejected gratuitously. For that, @samp{-pedantic} is required in
854 addition to @samp{-ansi}. @xref{Warning Options}.
856 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
857 option is used. Some header files may notice this macro and refrain
858 from declaring certain functions or defining certain macros that the
859 ISO standard doesn't call for; this is to avoid interfering with any
860 programs that might use these names for other things.
862 Functions which would normally be builtin but do not have semantics
863 defined by ISO C (such as @code{alloca} and @code{ffs}) are not builtin
864 functions with @samp{-ansi} is used. @xref{Other Builtins,,Other
865 built-in functions provided by GNU CC}, for details of the functions
869 Determine the language standard. A value for this option must be provided;
874 Same as @option{-ansi}
877 ISO C as modified in amend. 1
880 ISO C99. Note that this standard is not yet fully supported; see
881 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information.
884 same as @option{-std=iso9899:1990}
887 same as @option{-std=iso9899:1999}
890 default, iso9899:1990 + gnu extensions
893 iso9899:1999 + gnu extensions
896 same as @option{-std=iso9899:1999}, deprecated
899 same as @option{-std=iso9899:1999}, deprecated
902 same as @option{-std=gnu99}, deprecated
906 Even when this option is not specified, you can still use some of the
907 features of newer standards in so far as they do not conflict with
908 previous C standards. For example, you may use @code{__restrict__} even
909 when @option{-std=c99} is not specified.
911 @xref{Standards,,Language Standards Supported by GCC}, for details of
912 these standard versions.
915 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
916 keyword, so that code can use these words as identifiers. You can use
917 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
918 instead. @samp{-ansi} implies @samp{-fno-asm}.
920 In C++, this switch only affects the @code{typeof} keyword, since
921 @code{asm} and @code{inline} are standard keywords. You may want to
922 use the @samp{-fno-gnu-keywords} flag instead, which has the same effect.
925 @cindex builtin functions
926 Don't recognize builtin functions that do not begin with
927 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
928 functions provided by GNU CC}, for details of the functions affected,
929 including those which are not builtin functions when @samp{-ansi} is
930 used because they do not have an ISO standard meaning.
932 GCC normally generates special code to handle certain builtin functions
933 more efficiently; for instance, calls to @code{alloca} may become single
934 instructions that adjust the stack directly, and calls to @code{memcpy}
935 may become inline copy loops. The resulting code is often both smaller
936 and faster, but since the function calls no longer appear as such, you
937 cannot set a breakpoint on those calls, nor can you change the behavior
938 of the functions by linking with a different library.
941 @cindex hosted environment
943 Assert that compilation takes place in a hosted environment. This implies
944 @samp{-fbuiltin}. A hosted environment is one in which the
945 entire standard library is available, and in which @code{main} has a return
946 type of @code{int}. Examples are nearly everything except a kernel.
947 This is equivalent to @samp{-fno-freestanding}.
950 @cindex hosted environment
952 Assert that compilation takes place in a freestanding environment. This
953 implies @samp{-fno-builtin}. A freestanding environment
954 is one in which the standard library may not exist, and program startup may
955 not necessarily be at @code{main}. The most obvious example is an OS kernel.
956 This is equivalent to @samp{-fno-hosted}.
958 @xref{Standards,,Language Standards Supported by GCC}, for details of
959 freestanding and hosted environments.
962 Support ISO C trigraphs. You don't want to know about this
963 brain-damage. The @samp{-ansi} option implies @samp{-trigraphs}.
965 @cindex traditional C language
966 @cindex C language, traditional
968 Attempt to support some aspects of traditional C compilers.
973 All @code{extern} declarations take effect globally even if they
974 are written inside of a function definition. This includes implicit
975 declarations of functions.
978 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
979 and @code{volatile} are not recognized. (You can still use the
980 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
984 Comparisons between pointers and integers are always allowed.
987 Integer types @code{unsigned short} and @code{unsigned char} promote
988 to @code{unsigned int}.
991 Out-of-range floating point literals are not an error.
994 Certain constructs which ISO regards as a single invalid preprocessing
995 number, such as @samp{0xe-0xd}, are treated as expressions instead.
998 String ``constants'' are not necessarily constant; they are stored in
999 writable space, and identical looking constants are allocated
1000 separately. (This is the same as the effect of
1001 @samp{-fwritable-strings}.)
1003 @cindex @code{longjmp} and automatic variables
1005 All automatic variables not declared @code{register} are preserved by
1006 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1007 not declared @code{volatile} may be clobbered.
1012 @cindex escape sequences, traditional
1013 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1014 literal characters @samp{x} and @samp{a} respectively. Without
1015 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1016 representation of a character, and @samp{\a} produces a bell.
1019 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1020 if your program uses names that are normally GNU C builtin functions for
1021 other purposes of its own.
1023 You cannot use @samp{-traditional} if you include any header files that
1024 rely on ISO C features. Some vendors are starting to ship systems with
1025 ISO C header files and you cannot use @samp{-traditional} on such
1026 systems to compile files that include any system headers.
1028 The @samp{-traditional} option also enables @samp{-traditional-cpp},
1029 which is described next.
1031 @item -traditional-cpp
1032 Attempt to support some aspects of traditional C preprocessors.
1037 Comments convert to nothing at all, rather than to a space. This allows
1038 traditional token concatenation.
1041 In a preprocessing directive, the @samp{#} symbol must appear as the first
1042 character of a line.
1045 Macro arguments are recognized within string constants in a macro
1046 definition (and their values are stringified, though without additional
1047 quote marks, when they appear in such a context). The preprocessor
1048 always considers a string constant to end at a newline.
1051 @cindex detecting @w{@samp{-traditional}}
1052 The predefined macro @code{__STDC__} is not defined when you use
1053 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
1054 which @code{__GNUC__} indicates are not affected by
1055 @samp{-traditional}). If you need to write header files that work
1056 differently depending on whether @samp{-traditional} is in use, by
1057 testing both of these predefined macros you can distinguish four
1058 situations: GNU C, traditional GNU C, other ISO C compilers, and other
1059 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
1060 not defined when you use @samp{-traditional}. @xref{Standard
1061 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
1062 for more discussion of these and other predefined macros.
1065 @cindex string constants vs newline
1066 @cindex newline vs string constants
1067 The preprocessor considers a string constant to end at a newline (unless
1068 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
1069 string constants can contain the newline character as typed.)
1072 @item -fcond-mismatch
1073 Allow conditional expressions with mismatched types in the second and
1074 third arguments. The value of such an expression is void. This option
1075 is not supported for C++.
1077 @item -funsigned-char
1078 Let the type @code{char} be unsigned, like @code{unsigned char}.
1080 Each kind of machine has a default for what @code{char} should
1081 be. It is either like @code{unsigned char} by default or like
1082 @code{signed char} by default.
1084 Ideally, a portable program should always use @code{signed char} or
1085 @code{unsigned char} when it depends on the signedness of an object.
1086 But many programs have been written to use plain @code{char} and
1087 expect it to be signed, or expect it to be unsigned, depending on the
1088 machines they were written for. This option, and its inverse, let you
1089 make such a program work with the opposite default.
1091 The type @code{char} is always a distinct type from each of
1092 @code{signed char} or @code{unsigned char}, even though its behavior
1093 is always just like one of those two.
1096 Let the type @code{char} be signed, like @code{signed char}.
1098 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
1099 the negative form of @samp{-funsigned-char}. Likewise, the option
1100 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
1102 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1103 if your program uses names that are normally GNU C builtin functions for
1104 other purposes of its own.
1106 You cannot use @samp{-traditional} if you include any header files that
1107 rely on ISO C features. Some vendors are starting to ship systems with
1108 ISO C header files and you cannot use @samp{-traditional} on such
1109 systems to compile files that include any system headers.
1111 @item -fsigned-bitfields
1112 @itemx -funsigned-bitfields
1113 @itemx -fno-signed-bitfields
1114 @itemx -fno-unsigned-bitfields
1115 These options control whether a bitfield is signed or unsigned, when the
1116 declaration does not use either @code{signed} or @code{unsigned}. By
1117 default, such a bitfield is signed, because this is consistent: the
1118 basic integer types such as @code{int} are signed types.
1120 However, when @samp{-traditional} is used, bitfields are all unsigned
1123 @item -fwritable-strings
1124 Store string constants in the writable data segment and don't uniquize
1125 them. This is for compatibility with old programs which assume they can
1126 write into string constants. The option @samp{-traditional} also has
1129 Writing into string constants is a very bad idea; ``constants'' should
1132 @item -fallow-single-precision
1133 Do not promote single precision math operations to double precision,
1134 even when compiling with @samp{-traditional}.
1136 Traditional K&R C promotes all floating point operations to double
1137 precision, regardless of the sizes of the operands. On the
1138 architecture for which you are compiling, single precision may be faster
1139 than double precision. If you must use @samp{-traditional}, but want
1140 to use single precision operations when the operands are single
1141 precision, use this option. This option has no effect when compiling
1142 with ISO or GNU C conventions (the default).
1145 Override the underlying type for @samp{wchar_t} to be @samp{short
1146 unsigned int} instead of the default for the target. This option is
1147 useful for building programs to run under WINE.
1150 @node C++ Dialect Options
1151 @section Options Controlling C++ Dialect
1153 @cindex compiler options, C++
1154 @cindex C++ options, command line
1155 @cindex options, C++
1156 This section describes the command-line options that are only meaningful
1157 for C++ programs; but you can also use most of the GNU compiler options
1158 regardless of what language your program is in. For example, you
1159 might compile a file @code{firstClass.C} like this:
1162 g++ -g -frepo -O -c firstClass.C
1166 In this example, only @samp{-frepo} is an option meant
1167 only for C++ programs; you can use the other options with any
1168 language supported by GCC.
1170 Here is a list of options that are @emph{only} for compiling C++ programs:
1173 @item -fno-access-control
1174 Turn off all access checking. This switch is mainly useful for working
1175 around bugs in the access control code.
1178 Check that the pointer returned by @code{operator new} is non-null
1179 before attempting to modify the storage allocated. The current Working
1180 Paper requires that @code{operator new} never return a null pointer, so
1181 this check is normally unnecessary.
1183 An alternative to using this option is to specify that your
1184 @code{operator new} does not throw any exceptions; if you declare it
1185 @samp{throw()}, g++ will check the return value. See also @samp{new
1188 @item -fconserve-space
1189 Put uninitialized or runtime-initialized global variables into the
1190 common segment, as C does. This saves space in the executable at the
1191 cost of not diagnosing duplicate definitions. If you compile with this
1192 flag and your program mysteriously crashes after @code{main()} has
1193 completed, you may have an object that is being destroyed twice because
1194 two definitions were merged.
1196 This option is no longer useful on most targets, now that support has
1197 been added for putting variables into BSS without making them common.
1199 @item -fdollars-in-identifiers
1200 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1201 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
1202 @samp{$} by default on most target systems, but there are a few exceptions.)
1203 Traditional C allowed the character @samp{$} to form part of
1204 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1206 @item -fno-elide-constructors
1207 The C++ standard allows an implementation to omit creating a temporary
1208 which is only used to initialize another object of the same type.
1209 Specifying this option disables that optimization, and forces g++ to
1210 call the copy constructor in all cases.
1212 @item -fno-enforce-eh-specs
1213 Don't check for violation of exception specifications at runtime. This
1214 option violates the C++ standard, but may be useful for reducing code
1215 size in production builds, much like defining @samp{NDEBUG}. The compiler
1216 will still optimize based on the exception specifications.
1218 @item -fexternal-templates
1219 Cause template instantiations to obey @samp{#pragma interface} and
1220 @samp{implementation}; template instances are emitted or not according
1221 to the location of the template definition. @xref{Template
1222 Instantiation}, for more information.
1224 This option is deprecated.
1226 @item -falt-external-templates
1227 Similar to -fexternal-templates, but template instances are emitted or
1228 not according to the place where they are first instantiated.
1229 @xref{Template Instantiation}, for more information.
1231 This option is deprecated.
1234 @itemx -fno-for-scope
1235 If -ffor-scope is specified, the scope of variables declared in
1236 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1237 as specified by the C++ standard.
1238 If -fno-for-scope is specified, the scope of variables declared in
1239 a @i{for-init-statement} extends to the end of the enclosing scope,
1240 as was the case in old versions of gcc, and other (traditional)
1241 implementations of C++.
1243 The default if neither flag is given to follow the standard,
1244 but to allow and give a warning for old-style code that would
1245 otherwise be invalid, or have different behavior.
1247 @item -fno-gnu-keywords
1248 Do not recognize @code{typeof} as a keyword, so that code can use this
1249 word as an identifier. You can use the keyword @code{__typeof__} instead.
1250 @samp{-ansi} implies @samp{-fno-gnu-keywords}.
1253 Treat the @code{namespace std} as a namespace, instead of ignoring
1254 it. For compatibility with earlier versions of g++, the compiler will,
1255 by default, ignore @code{namespace-declarations},
1256 @code{using-declarations}, @code{using-directives}, and
1257 @code{namespace-names}, if they involve @code{std}.
1259 @item -fhuge-objects
1260 Support virtual function calls for objects that exceed the size
1261 representable by a @samp{short int}. Users should not use this flag by
1262 default; if you need to use it, the compiler will tell you so.
1264 This flag is not useful when compiling with -fvtable-thunks.
1266 Like all options that change the ABI, all C++ code, @emph{including
1267 libgcc} must be built with the same setting of this option.
1269 @item -fno-implicit-templates
1270 Never emit code for non-inline templates which are instantiated
1271 implicitly (i.e. by use); only emit code for explicit instantiations.
1272 @xref{Template Instantiation}, for more information.
1274 @item -fno-implicit-inline-templates
1275 Don't emit code for implicit instantiations of inline templates, either.
1276 The default is to handle inlines differently so that compiles with and
1277 without optimization will need the same set of explicit instantiations.
1279 @item -fno-implement-inlines
1280 To save space, do not emit out-of-line copies of inline functions
1281 controlled by @samp{#pragma implementation}. This will cause linker
1282 errors if these functions are not inlined everywhere they are called.
1284 @item -fms-extensions
1285 Disable pedantic warnings about constructs used in MFC, such as implicit
1286 int and getting a pointer to member function via non-standard syntax.
1288 @item -fname-mangling-version-@var{n}
1289 Control the way in which names are mangled. Version 0 is compatible
1290 with versions of g++ before 2.8. Version 1 is the default. Version 1
1291 will allow correct mangling of function templates. For example,
1292 version 0 mangling does not mangle foo<int, double> and foo<int, char>
1293 given this declaration:
1296 template <class T, class U> void foo(T t);
1299 Like all options that change the ABI, all C++ code, @emph{including
1300 libgcc} must be built with the same setting of this option.
1302 @item -fno-operator-names
1303 Do not treat the operator name keywords @code{and}, @code{bitand},
1304 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1305 synonyms as keywords.
1307 @item -fno-optional-diags
1308 Disable diagnostics that the standard says a compiler does not need to
1309 issue. Currently, the only such diagnostic issued by g++ is the one for
1310 a name having multiple meanings within a class.
1313 Downgrade messages about nonconformant code from errors to warnings. By
1314 default, g++ effectively sets @samp{-pedantic-errors} without
1315 @samp{-pedantic}; this option reverses that. This behavior and this
1316 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1319 Enable automatic template instantiation. This option also implies
1320 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1324 Disable generation of information about every class with virtual
1325 functions for use by the C++ runtime type identification features
1326 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1327 of the language, you can save some space by using this flag. Note that
1328 exception handling uses the same information, but it will generate it as
1332 @itemx -fno-squangle
1333 @samp{-fsquangle} will enable a compressed form of name mangling for
1334 identifiers. In particular, it helps to shorten very long names by recognizing
1335 types and class names which occur more than once, replacing them with special
1336 short ID codes. This option also requires any C++ libraries being used to
1337 be compiled with this option as well. The compiler has this disabled (the
1338 equivalent of @samp{-fno-squangle}) by default.
1340 Like all options that change the ABI, all C++ code, @emph{including
1341 libgcc.a} must be built with the same setting of this option.
1343 @item -ftemplate-depth-@var{n}
1344 Set the maximum instantiation depth for template classes to @var{n}.
1345 A limit on the template instantiation depth is needed to detect
1346 endless recursions during template class instantiation. ANSI/ISO C++
1347 conforming programs must not rely on a maximum depth greater than 17.
1349 @item -fuse-cxa-atexit
1350 Register destructors for objects with static storage duration with the
1351 @code{__cxa_atexit} function rather than the @code{atexit} function.
1352 This option is required for fully standards-compliant handling of static
1353 destructors, but will only work if your C library supports
1354 @code{__cxa_atexit}.
1356 @item -fvtable-thunks
1357 Use @samp{thunks} to implement the virtual function dispatch table
1358 (@samp{vtable}). The traditional (cfront-style) approach to
1359 implementing vtables was to store a pointer to the function and two
1360 offsets for adjusting the @samp{this} pointer at the call site. Newer
1361 implementations store a single pointer to a @samp{thunk} function which
1362 does any necessary adjustment and then calls the target function.
1364 This option also enables a heuristic for controlling emission of
1365 vtables; if a class has any non-inline virtual functions, the vtable
1366 will be emitted in the translation unit containing the first one of
1369 Like all options that change the ABI, all C++ code, @emph{including
1370 libgcc.a} must be built with the same setting of this option.
1373 Do not search for header files in the standard directories specific to
1374 C++, but do still search the other standard directories. (This option
1375 is used when building the C++ library.)
1378 In addition, these optimization, warning, and code generation options
1379 have meanings only for C++ programs:
1382 @item -fno-default-inline
1383 Do not assume @samp{inline} for functions defined inside a class scope.
1384 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1385 functions will have linkage like inline functions; they just won't be
1388 @item -Wctor-dtor-privacy (C++ only)
1389 Warn when a class seems unusable, because all the constructors or
1390 destructors in a class are private and the class has no friends or
1391 public static member functions.
1393 @item -Wnon-virtual-dtor (C++ only)
1394 Warn when a class declares a non-virtual destructor that should probably
1395 be virtual, because it looks like the class will be used polymorphically.
1397 @item -Wreorder (C++ only)
1398 @cindex reordering, warning
1399 @cindex warning for reordering of member initializers
1400 Warn when the order of member initializers given in the code does not
1401 match the order in which they must be executed. For instance:
1407 A(): j (0), i (1) @{ @}
1411 Here the compiler will warn that the member initializers for @samp{i}
1412 and @samp{j} will be rearranged to match the declaration order of the
1416 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1419 @item -Weffc++ (C++ only)
1420 Warn about violations of various style guidelines from Scott Meyers'
1421 @cite{Effective C++} books. If you use this option, you should be aware
1422 that the standard library headers do not obey all of these guidelines;
1423 you can use @samp{grep -v} to filter out those warnings.
1425 @item -Wno-deprecated (C++ only)
1426 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1428 @item -Wno-non-template-friend (C++ only)
1429 Disable warnings when non-templatized friend functions are declared
1430 within a template. With the advent of explicit template specification
1431 support in g++, if the name of the friend is an unqualified-id (ie,
1432 @samp{friend foo(int)}), the C++ language specification demands that the
1433 friend declare or define an ordinary, nontemplate function. (Section
1434 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1435 could be interpreted as a particular specialization of a templatized
1436 function. Because this non-conforming behavior is no longer the default
1437 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1438 check existing code for potential trouble spots, and is on by default.
1439 This new compiler behavior can be turned off with
1440 @samp{-Wno-non-template-friend} which keeps the conformant compiler code
1441 but disables the helpful warning.
1443 @item -Wold-style-cast (C++ only)
1444 Warn if an old-style (C-style) cast is used within a C++ program. The
1445 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1446 @samp{const_cast}) are less vulnerable to unintended effects.
1448 @item -Woverloaded-virtual (C++ only)
1449 @cindex overloaded virtual fn, warning
1450 @cindex warning for overloaded virtual fn
1451 Warn when a derived class function declaration may be an error in
1452 defining a virtual function. In a derived class, the
1453 definitions of virtual functions must match the type signature of a
1454 virtual function declared in the base class. With this option, the
1455 compiler warns when you define a function with the same name as a
1456 virtual function, but with a type signature that does not match any
1457 declarations from the base class.
1459 @item -Wno-pmf-conversions (C++ only)
1460 Disable the diagnostic for converting a bound pointer to member function
1463 @item -Wsign-promo (C++ only)
1464 Warn when overload resolution chooses a promotion from unsigned or
1465 enumeral type to a signed type over a conversion to an unsigned type of
1466 the same size. Previous versions of g++ would try to preserve
1467 unsignedness, but the standard mandates the current behavior.
1469 @item -Wsynth (C++ only)
1470 @cindex warning for synthesized methods
1471 @cindex synthesized methods, warning
1472 Warn when g++'s synthesis behavior does not match that of cfront. For
1478 A& operator = (int);
1488 In this example, g++ will synthesize a default @samp{A& operator =
1489 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1492 @node Language Independent Options
1493 @section Options to Control Diagnostic Messages Formatting
1494 @cindex options to control diagnostics formatting
1495 @cindex diagnostic messages
1496 @cindex message formatting
1498 Traditionally, diagnostic messages have been formatted irrespective of
1499 the output device's aspect (e.g. its width, ...). The options described
1500 below can be used to control the diagnostic messages formatting
1501 algorithm, e.g. how many characters per line, how often source location
1502 information should be reported. Right now, only the C++ front-end can
1503 honor these options. However it is expected, in the near future, that
1504 the remaining front-ends would be able to digest them correctly.
1507 @item -fmessage-length=@var{n}
1508 Try to format error messages so that they fit on lines of about @var{n}
1509 characters. The default is 72 characters for g++ and 0 for the rest of
1510 the front-ends supported by GCC. If @var{n} is zero, then no
1511 line-wrapping will be done; each error message will appear on a single
1514 @item -fdiagnostics-show-location=once
1515 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1516 reporter to emit @emph{once} source location information; that is, in
1517 case the message is too long to fit on a single physical line and has to
1518 be wrapped, the source location won't be emitted (as prefix) again,
1519 over and over, in subsequent continuation lines. This is the default
1522 @item -fdiagnostics-show-location=every-line
1523 Only meaningful in line-wrapping mode. Instructs the diagnostic
1524 messages reporter to emit the same source location information (as
1525 prefix) for physical lines that result from the process of breaking a
1526 a message which is too long to fit on a single line.
1530 @node Warning Options
1531 @section Options to Request or Suppress Warnings
1532 @cindex options to control warnings
1533 @cindex warning messages
1534 @cindex messages, warning
1535 @cindex suppressing warnings
1537 Warnings are diagnostic messages that report constructions which
1538 are not inherently erroneous but which are risky or suggest there
1539 may have been an error.
1541 You can request many specific warnings with options beginning @samp{-W},
1542 for example @samp{-Wimplicit} to request warnings on implicit
1543 declarations. Each of these specific warning options also has a
1544 negative form beginning @samp{-Wno-} to turn off warnings;
1545 for example, @samp{-Wno-implicit}. This manual lists only one of the
1546 two forms, whichever is not the default.
1548 These options control the amount and kinds of warnings produced by GCC:
1551 @cindex syntax checking
1553 Check the code for syntax errors, but don't do anything beyond that.
1556 Issue all the warnings demanded by strict ISO C and ISO C++;
1557 reject all programs that use forbidden extensions, and some other
1558 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1559 version of the ISO C standard specified by any @samp{-std} option used.
1561 Valid ISO C and ISO C++ programs should compile properly with or without
1562 this option (though a rare few will require @samp{-ansi}). However,
1563 without this option, certain GNU extensions and traditional C and C++
1564 features are supported as well. With this option, they are rejected.
1566 @samp{-pedantic} does not cause warning messages for use of the
1567 alternate keywords whose names begin and end with @samp{__}. Pedantic
1568 warnings are also disabled in the expression that follows
1569 @code{__extension__}. However, only system header files should use
1570 these escape routes; application programs should avoid them.
1571 @xref{Alternate Keywords}.
1573 Some users try to use @samp{-pedantic} to check programs for strict ISO
1574 C conformance. They soon find that it does not do quite what they want:
1575 it finds some non-ISO practices, but not all---only those for which
1576 ISO C @emph{requires} a diagnostic, and some others for which
1577 diagnostics have been added.
1579 A feature to report any failure to conform to ISO C might be useful in
1580 some instances, but would require considerable additional work and would
1581 be quite different from @samp{-pedantic}. We don't have plans to
1582 support such a feature in the near future.
1584 @item -pedantic-errors
1585 Like @samp{-pedantic}, except that errors are produced rather than
1589 Inhibit all warning messages.
1592 Inhibit warning messages about the use of @samp{#import}.
1594 @item -Wchar-subscripts
1595 Warn if an array subscript has type @code{char}. This is a common cause
1596 of error, as programmers often forget that this type is signed on some
1600 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1601 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1604 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1605 the arguments supplied have types appropriate to the format string
1606 specified, and that the conversions specified in the format string make
1607 sense. This includes standard functions, and others specified by format
1608 attributes (@pxref{Function Attributes}), in the @code{printf},
1609 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1610 not in the C standard) families.
1612 The formats are checked against the format features supported by GNU
1613 libc version 2.2. These include all ISO C89 and C99 features, as well
1614 as features from the Single Unix Specification and some BSD and GNU
1615 extensions. Other library implementations may not support all these
1616 features; GCC does not support warning about features that go beyond a
1617 particular library's limitations. However, if @samp{-pedantic} is used
1618 with @samp{-Wformat}, warnings will be given about format features not
1619 in the selected standard version (but not for @code{strfmon} formats,
1620 since those are not in any version of the C standard). @xref{C Dialect
1621 Options,,Options Controlling C Dialect}.
1623 @samp{-Wformat} is included in @samp{-Wall}. For more control over some
1624 aspects of format checking, the options @samp{-Wno-format-y2k},
1625 @samp{-Wno-format-extra-args}, @samp{-Wformat-nonliteral},
1626 @samp{-Wformat-security} and @samp{-Wformat=2} are available, but are
1627 not included in @samp{-Wall}.
1629 @item -Wno-format-y2k
1630 If @samp{-Wformat} is specified, do not warn about @code{strftime}
1631 formats which may yield only a two-digit year.
1633 @item -Wno-format-extra-args
1634 If @samp{-Wformat} is specified, do not warn about excess arguments to a
1635 @code{printf} or @code{scanf} format function. The C standard specifies
1636 that such arguments are ignored.
1638 @item -Wformat-nonliteral
1639 If @samp{-Wformat} is specified, also warn if the format string is not a
1640 string literal and so cannot be checked, unless the format function
1641 takes its format arguments as a @code{va_list}.
1643 @item -Wformat-security
1644 If @samp{-Wformat} is specified, also warn about uses of format
1645 functions that represent possible security problems. At present, this
1646 warns about calls to @code{printf} and @code{scanf} functions where the
1647 format string is not a string literal and there are no format arguments,
1648 as in @code{printf (foo);}. This may be a security hole if the format
1649 string came from untrusted input and contains @samp{%n}. (This is
1650 currently a subset of what @samp{-Wformat-nonliteral} warns about, but
1651 in future warnings may be added to @samp{-Wformat-security} that are not
1652 included in @samp{-Wformat-nonliteral}.)
1655 Enable @samp{-Wformat} plus format checks not included in
1656 @samp{-Wformat}. Currently equivalent to @samp{-Wformat
1657 -Wformat-nonliteral -Wformat-security}.
1659 @item -Wimplicit-int
1660 Warn when a declaration does not specify a type.
1662 @item -Wimplicit-function-declaration
1663 @itemx -Werror-implicit-function-declaration
1664 Give a warning (or error) whenever a function is used before being
1668 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1672 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1673 function with external linkage, returning int, taking either zero
1674 arguments, two, or three arguments of appropriate types.
1677 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1678 indicate a typo in the user's code, as they have implementation-defined
1679 values, and should not be used in portable code.
1682 Warn if parentheses are omitted in certain contexts, such
1683 as when there is an assignment in a context where a truth value
1684 is expected, or when operators are nested whose precedence people
1685 often get confused about.
1687 Also warn about constructions where there may be confusion to which
1688 @code{if} statement an @code{else} branch belongs. Here is an example of
1701 In C, every @code{else} branch belongs to the innermost possible @code{if}
1702 statement, which in this example is @code{if (b)}. This is often not
1703 what the programmer expected, as illustrated in the above example by
1704 indentation the programmer chose. When there is the potential for this
1705 confusion, GNU C will issue a warning when this flag is specified.
1706 To eliminate the warning, add explicit braces around the innermost
1707 @code{if} statement so there is no way the @code{else} could belong to
1708 the enclosing @code{if}. The resulting code would look like this:
1722 @item -Wsequence-point
1723 Warn about code that may have undefined semantics because of violations
1724 of sequence point rules in the C standard.
1726 The C standard defines the order in which expressions in a C program are
1727 evaluated in terms of @dfn{sequence points}, which represent a partial
1728 ordering between the execution of parts of the program: those executed
1729 before the sequence point, and those executed after it. These occur
1730 after the evaluation of a full expression (one which is not part of a
1731 larger expression), after the evaluation of the first operand of a
1732 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1733 function is called (but after the evaluation of its arguments and the
1734 expression denoting the called function), and in certain other places.
1735 Other than as expressed by the sequence point rules, the order of
1736 evaluation of subexpressions of an expression is not specified. All
1737 these rules describe only a partial order rather than a total order,
1738 since, for example, if two functions are called within one expression
1739 with no sequence point between them, the order in which the functions
1740 are called is not specified. However, the standards committee have
1741 ruled that function calls do not overlap.
1743 It is not specified when between sequence points modifications to the
1744 values of objects take effect. Programs whose behavior depends on this
1745 have undefined behavior; the C standard specifies that ``Between the
1746 previous and next sequence point an object shall have its stored value
1747 modified at most once by the evaluation of an expression. Furthermore,
1748 the prior value shall be read only to determine the value to be
1749 stored.''. If a program breaks these rules, the results on any
1750 particular implementation are entirely unpredictable.
1752 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1753 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1754 diagnosed by this option, and it may give an occasional false positive
1755 result, but in general it has been found fairly effective at detecting
1756 this sort of problem in programs.
1758 The present implementation of this option only works for C programs. A
1759 future implementation may also work for C++ programs.
1761 There is some controversy over the precise meaning of the sequence point
1762 rules in subtle cases. Alternative formal definitions may be found in
1763 Clive Feather's ``Annex S''
1764 @w{@uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n925.htm}} and in
1765 Michael Norrish's thesis
1766 @w{@uref{http://www.cl.cam.ac.uk/users/mn200/PhD/thesis-report.ps.gz}}.
1767 Other discussions are by Raymond Mak
1768 @w{@uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n926.htm}} and
1770 @w{@uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n927.htm}}.
1773 Warn whenever a function is defined with a return-type that defaults to
1774 @code{int}. Also warn about any @code{return} statement with no
1775 return-value in a function whose return-type is not @code{void}.
1777 For C++, a function without return type always produces a diagnostic
1778 message, even when @samp{-Wno-return-type} is specified. The only
1779 exceptions are @samp{main} and functions defined in system headers.
1782 Warn whenever a @code{switch} statement has an index of enumeral type
1783 and lacks a @code{case} for one or more of the named codes of that
1784 enumeration. (The presence of a @code{default} label prevents this
1785 warning.) @code{case} labels outside the enumeration range also
1786 provoke warnings when this option is used.
1789 Warn if any trigraphs are encountered that might change the meaning of
1790 the program (trigraphs within comments are not warned about).
1792 @item -Wunused-function
1793 Warn whenever a static function is declared but not defined or a
1794 non\-inline static function is unused.
1796 @item -Wunused-label
1797 Warn whenever a label is declared but not used.
1799 To suppress this warning use the @samp{unused} attribute
1800 (@pxref{Variable Attributes}).
1802 @item -Wunused-parameter
1803 Warn whenever a function parameter is unused aside from its declaration.
1805 To suppress this warning use the @samp{unused} attribute
1806 (@pxref{Variable Attributes}).
1808 @item -Wunused-variable
1809 Warn whenever a local variable or non-constant static variable is unused
1810 aside from its declaration
1812 To suppress this warning use the @samp{unused} attribute
1813 (@pxref{Variable Attributes}).
1815 @item -Wunused-value
1816 Warn whenever a statement computes a result that is explicitly not used.
1818 To suppress this warning cast the expression to @samp{void}.
1821 All all the above @samp{-Wunused} options combined.
1823 In order to get a warning about an unused function parameter, you must
1824 either specify @samp{-W -Wunused} or separately specify
1825 @samp{-Wunused-parameter}.
1827 @item -Wuninitialized
1828 Warn if an automatic variable is used without first being initialized or
1829 if a variable may be clobbered by a @code{setjmp} call.
1831 These warnings are possible only in optimizing compilation,
1832 because they require data flow information that is computed only
1833 when optimizing. If you don't specify @samp{-O}, you simply won't
1836 These warnings occur only for variables that are candidates for
1837 register allocation. Therefore, they do not occur for a variable that
1838 is declared @code{volatile}, or whose address is taken, or whose size
1839 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1840 structures, unions or arrays, even when they are in registers.
1842 Note that there may be no warning about a variable that is used only
1843 to compute a value that itself is never used, because such
1844 computations may be deleted by data flow analysis before the warnings
1847 These warnings are made optional because GCC is not smart
1848 enough to see all the reasons why the code might be correct
1849 despite appearing to have an error. Here is one example of how
1868 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1869 always initialized, but GCC doesn't know this. Here is
1870 another common case:
1875 if (change_y) save_y = y, y = new_y;
1877 if (change_y) y = save_y;
1882 This has no bug because @code{save_y} is used only if it is set.
1884 @cindex @code{longjmp} warnings
1885 This option also warns when a non-volatile automatic variable might be
1886 changed by a call to @code{longjmp}. These warnings as well are possible
1887 only in optimizing compilation.
1889 The compiler sees only the calls to @code{setjmp}. It cannot know
1890 where @code{longjmp} will be called; in fact, a signal handler could
1891 call it at any point in the code. As a result, you may get a warning
1892 even when there is in fact no problem because @code{longjmp} cannot
1893 in fact be called at the place which would cause a problem.
1895 Some spurious warnings can be avoided if you declare all the functions
1896 you use that never return as @code{noreturn}. @xref{Function
1899 @item -Wreorder (C++ only)
1900 @cindex reordering, warning
1901 @cindex warning for reordering of member initializers
1902 Warn when the order of member initializers given in the code does not
1903 match the order in which they must be executed. For instance:
1905 @item -Wunknown-pragmas
1906 @cindex warning for unknown pragmas
1907 @cindex unknown pragmas, warning
1908 @cindex pragmas, warning of unknown
1909 Warn when a #pragma directive is encountered which is not understood by
1910 GCC. If this command line option is used, warnings will even be issued
1911 for unknown pragmas in system header files. This is not the case if
1912 the warnings were only enabled by the @samp{-Wall} command line option.
1915 All of the above @samp{-W} options combined. This enables all the
1916 warnings about constructions that some users consider questionable, and
1917 that are easy to avoid (or modify to prevent the warning), even in
1918 conjunction with macros.
1920 @item -Wsystem-headers
1921 @cindex warnings from system headers
1922 @cindex system headers, warnings from
1923 Print warning messages for constructs found in system header files.
1924 Warnings from system headers are normally suppressed, on the assumption
1925 that they usually do not indicate real problems and would only make the
1926 compiler output harder to read. Using this command line option tells
1927 GCC to emit warnings from system headers as if they occurred in user
1928 code. However, note that using @samp{-Wall} in conjunction with this
1929 option will @emph{not} warn about unknown pragmas in system
1930 headers---for that, @samp{-Wunknown-pragmas} must also be used.
1933 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1934 Some of them warn about constructions that users generally do not
1935 consider questionable, but which occasionally you might wish to check
1936 for; others warn about constructions that are necessary or hard to avoid
1937 in some cases, and there is no simple way to modify the code to suppress
1942 Print extra warning messages for these events:
1946 A function can return either with or without a value. (Falling
1947 off the end of the function body is considered returning without
1948 a value.) For example, this function would evoke such a
1962 An expression-statement or the left-hand side of a comma expression
1963 contains no side effects.
1964 To suppress the warning, cast the unused expression to void.
1965 For example, an expression such as @samp{x[i,j]} will cause a warning,
1966 but @samp{x[(void)i,j]} will not.
1969 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1972 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1973 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1974 that of ordinary mathematical notation.
1977 Storage-class specifiers like @code{static} are not the first things in
1978 a declaration. According to the C Standard, this usage is obsolescent.
1981 The return type of a function has a type qualifier such as @code{const}.
1982 Such a type qualifier has no effect, since the value returned by a
1983 function is not an lvalue. (But don't warn about the GNU extension of
1984 @code{volatile void} return types. That extension will be warned about
1985 if @samp{-pedantic} is specified.)
1988 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
1992 A comparison between signed and unsigned values could produce an
1993 incorrect result when the signed value is converted to unsigned.
1994 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
1997 An aggregate has a partly bracketed initializer.
1998 For example, the following code would evoke such a warning,
1999 because braces are missing around the initializer for @code{x.h}:
2002 struct s @{ int f, g; @};
2003 struct t @{ struct s h; int i; @};
2004 struct t x = @{ 1, 2, 3 @};
2008 An aggregate has an initializer which does not initialize all members.
2009 For example, the following code would cause such a warning, because
2010 @code{x.h} would be implicitly initialized to zero:
2013 struct s @{ int f, g, h; @};
2014 struct s x = @{ 3, 4 @};
2019 Warn if floating point values are used in equality comparisons.
2021 The idea behind this is that sometimes it is convenient (for the
2022 programmer) to consider floating-point values as approximations to
2023 infinitely precise real numbers. If you are doing this, then you need
2024 to compute (by analysing the code, or in some other way) the maximum or
2025 likely maximum error that the computation introduces, and allow for it
2026 when performing comparisons (and when producing output, but that's a
2027 different problem). In particular, instead of testing for equality, you
2028 would check to see whether the two values have ranges that overlap; and
2029 this is done with the relational operators, so equality comparisons are
2032 @item -Wtraditional (C only)
2033 Warn about certain constructs that behave differently in traditional and
2038 Macro arguments occurring within string constants in the macro body.
2039 These would substitute the argument in traditional C, but are part of
2040 the constant in ISO C.
2043 A function declared external in one block and then used after the end of
2047 A @code{switch} statement has an operand of type @code{long}.
2050 A non-@code{static} function declaration follows a @code{static} one.
2051 This construct is not accepted by some traditional C compilers.
2054 The ISO type of an integer constant has a different width or
2055 signedness from its traditional type. This warning is only issued if
2056 the base of the constant is ten. I.e. hexadecimal or octal values, which
2057 typically represent bit patterns, are not warned about.
2060 Usage of ISO string concatenation is detected.
2063 A function macro appears without arguments.
2066 The unary plus operator.
2069 Initialization of automatic aggregates.
2072 Identifier conflicts with labels. Traditional C lacks a separate
2073 namespace for labels.
2076 Initialization of unions. If the initializer is zero, the warning is
2077 omitted. This is done under the assumption that the zero initializer in
2078 user code appears conditioned on e.g. @code{__STDC__} to avoid missing
2079 initializer warnings and relies on default initialization to zero in the
2083 The `U' integer constant suffix, or the `F' or `L' floating point
2084 constant suffixes. (Traditonal C does support the `L' suffix on integer
2085 constants.) Note, these suffixes appear in macros defined in the system
2086 headers of most modern systems, e.g. the _MIN/_MAX macros in limits.h.
2087 Use of these macros can lead to spurious warnings as they do not
2088 necessarily reflect whether the code in question is any less portable to
2089 traditional C given that suitable backup definitions are provided.
2093 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2096 Warn whenever a local variable shadows another local variable.
2098 @item -Wid-clash-@var{len}
2099 Warn whenever two distinct identifiers match in the first @var{len}
2100 characters. This may help you prepare a program that will compile
2101 with certain obsolete, brain-damaged compilers.
2103 @item -Wlarger-than-@var{len}
2104 Warn whenever an object of larger than @var{len} bytes is defined.
2106 @item -Wpointer-arith
2107 Warn about anything that depends on the ``size of'' a function type or
2108 of @code{void}. GNU C assigns these types a size of 1, for
2109 convenience in calculations with @code{void *} pointers and pointers
2112 @item -Wbad-function-cast (C only)
2113 Warn whenever a function call is cast to a non-matching type.
2114 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2117 Warn whenever a pointer is cast so as to remove a type qualifier from
2118 the target type. For example, warn if a @code{const char *} is cast
2119 to an ordinary @code{char *}.
2122 Warn whenever a pointer is cast such that the required alignment of the
2123 target is increased. For example, warn if a @code{char *} is cast to
2124 an @code{int *} on machines where integers can only be accessed at
2125 two- or four-byte boundaries.
2127 @item -Wwrite-strings
2128 Give string constants the type @code{const char[@var{length}]} so that
2129 copying the address of one into a non-@code{const} @code{char *}
2130 pointer will get a warning. These warnings will help you find at
2131 compile time code that can try to write into a string constant, but
2132 only if you have been very careful about using @code{const} in
2133 declarations and prototypes. Otherwise, it will just be a nuisance;
2134 this is why we did not make @samp{-Wall} request these warnings.
2137 Warn if a prototype causes a type conversion that is different from what
2138 would happen to the same argument in the absence of a prototype. This
2139 includes conversions of fixed point to floating and vice versa, and
2140 conversions changing the width or signedness of a fixed point argument
2141 except when the same as the default promotion.
2143 Also, warn if a negative integer constant expression is implicitly
2144 converted to an unsigned type. For example, warn about the assignment
2145 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2146 casts like @code{(unsigned) -1}.
2148 @item -Wsign-compare
2149 @cindex warning for comparison of signed and unsigned values
2150 @cindex comparison of signed and unsigned values, warning
2151 @cindex signed and unsigned values, comparison warning
2152 Warn when a comparison between signed and unsigned values could produce
2153 an incorrect result when the signed value is converted to unsigned.
2154 This warning is also enabled by @samp{-W}; to get the other warnings
2155 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2157 @item -Waggregate-return
2158 Warn if any functions that return structures or unions are defined or
2159 called. (In languages where you can return an array, this also elicits
2162 @item -Wstrict-prototypes (C only)
2163 Warn if a function is declared or defined without specifying the
2164 argument types. (An old-style function definition is permitted without
2165 a warning if preceded by a declaration which specifies the argument
2168 @item -Wmissing-prototypes (C only)
2169 Warn if a global function is defined without a previous prototype
2170 declaration. This warning is issued even if the definition itself
2171 provides a prototype. The aim is to detect global functions that fail
2172 to be declared in header files.
2174 @item -Wmissing-declarations
2175 Warn if a global function is defined without a previous declaration.
2176 Do so even if the definition itself provides a prototype.
2177 Use this option to detect global functions that are not declared in
2180 @item -Wmissing-noreturn
2181 Warn about functions which might be candidates for attribute @code{noreturn}.
2182 Note these are only possible candidates, not absolute ones. Care should
2183 be taken to manually verify functions actually do not ever return before
2184 adding the @code{noreturn} attribute, otherwise subtle code generation
2185 bugs could be introduced. You will not get a warning for @code{main} in
2186 hosted C environments.
2188 @item -Wmissing-format-attribute
2189 If @samp{-Wformat} is enabled, also warn about functions which might be
2190 candidates for @code{format} attributes. Note these are only possible
2191 candidates, not absolute ones. GCC will guess that @code{format}
2192 attributes might be appropriate for any function that calls a function
2193 like @code{vprintf} or @code{vscanf}, but this might not always be the
2194 case, and some functions for which @code{format} attributes are
2195 appropriate may not be detected. This option has no effect unless
2196 @samp{-Wformat} is enabled (possibly by @samp{-Wall}).
2199 Warn if a structure is given the packed attribute, but the packed
2200 attribute has no effect on the layout or size of the structure.
2201 Such structures may be mis-aligned for little benefit. For
2202 instance, in this code, the variable @code{f.x} in @code{struct bar}
2203 will be misaligned even though @code{struct bar} does not itself
2204 have the packed attribute:
2211 @} __attribute__((packed));
2220 Warn if padding is included in a structure, either to align an element
2221 of the structure or to align the whole structure. Sometimes when this
2222 happens it is possible to rearrange the fields of the structure to
2223 reduce the padding and so make the structure smaller.
2225 @item -Wredundant-decls
2226 Warn if anything is declared more than once in the same scope, even in
2227 cases where multiple declaration is valid and changes nothing.
2229 @item -Wnested-externs (C only)
2230 Warn if an @code{extern} declaration is encountered within a function.
2232 @item -Wunreachable-code
2233 Warn if the compiler detects that code will never be executed.
2235 This option is intended to warn when the compiler detects that at
2236 least a whole line of source code will never be executed, because
2237 some condition is never satisfied or because it is after a
2238 procedure that never returns.
2240 It is possible for this option to produce a warning even though there
2241 are circumstances under which part of the affected line can be executed,
2242 so care should be taken when removing apparently-unreachable code.
2244 For instance, when a function is inlined, a warning may mean that the
2245 line is unreachable in only one inlined copy of the function.
2247 This option is not made part of @samp{-Wall} because in a debugging
2248 version of a program there is often substantial code which checks
2249 correct functioning of the program and is, hopefully, unreachable
2250 because the program does work. Another common use of unreachable
2251 code is to provide behaviour which is selectable at compile-time.
2254 Warn if a function can not be inlined and it was declared as inline.
2257 Warn if @samp{long long} type is used. This is default. To inhibit
2258 the warning messages, use @samp{-Wno-long-long}. Flags
2259 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
2260 only when @samp{-pedantic} flag is used.
2262 @item -Wdisabled-optimization
2263 Warn if a requested optimization pass is disabled. This warning does
2264 not generally indicate that there is anything wrong with your code; it
2265 merely indicates that GCC's optimizers were unable to handle the code
2266 effectively. Often, the problem is that your code is too big or too
2267 complex; GCC will refuse to optimize programs when the optimization
2268 itself is likely to take inordinate amounts of time.
2271 Make all warnings into errors.
2274 @node Debugging Options
2275 @section Options for Debugging Your Program or GCC
2276 @cindex options, debugging
2277 @cindex debugging information options
2279 GCC has various special options that are used for debugging
2280 either your program or GCC:
2284 Produce debugging information in the operating system's native format
2285 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
2288 On most systems that use stabs format, @samp{-g} enables use of extra
2289 debugging information that only GDB can use; this extra information
2290 makes debugging work better in GDB but will probably make other debuggers
2292 refuse to read the program. If you want to control for certain whether
2293 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
2294 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
2297 Unlike most other C compilers, GCC allows you to use @samp{-g} with
2298 @samp{-O}. The shortcuts taken by optimized code may occasionally
2299 produce surprising results: some variables you declared may not exist
2300 at all; flow of control may briefly move where you did not expect it;
2301 some statements may not be executed because they compute constant
2302 results or their values were already at hand; some statements may
2303 execute in different places because they were moved out of loops.
2305 Nevertheless it proves possible to debug optimized output. This makes
2306 it reasonable to use the optimizer for programs that might have bugs.
2308 The following options are useful when GCC is generated with the
2309 capability for more than one debugging format.
2312 Produce debugging information for use by GDB. This means to use the
2313 most expressive format available (DWARF 2, stabs, or the native format
2314 if neither of those are supported), including GDB extensions if at all
2318 Produce debugging information in stabs format (if that is supported),
2319 without GDB extensions. This is the format used by DBX on most BSD
2320 systems. On MIPS, Alpha and System V Release 4 systems this option
2321 produces stabs debugging output which is not understood by DBX or SDB.
2322 On System V Release 4 systems this option requires the GNU assembler.
2325 Produce debugging information in stabs format (if that is supported),
2326 using GNU extensions understood only by the GNU debugger (GDB). The
2327 use of these extensions is likely to make other debuggers crash or
2328 refuse to read the program.
2331 Produce debugging information in COFF format (if that is supported).
2332 This is the format used by SDB on most System V systems prior to
2336 Produce debugging information in XCOFF format (if that is supported).
2337 This is the format used by the DBX debugger on IBM RS/6000 systems.
2340 Produce debugging information in XCOFF format (if that is supported),
2341 using GNU extensions understood only by the GNU debugger (GDB). The
2342 use of these extensions is likely to make other debuggers crash or
2343 refuse to read the program, and may cause assemblers other than the GNU
2344 assembler (GAS) to fail with an error.
2347 Produce debugging information in DWARF version 1 format (if that is
2348 supported). This is the format used by SDB on most System V Release 4
2352 Produce debugging information in DWARF version 1 format (if that is
2353 supported), using GNU extensions understood only by the GNU debugger
2354 (GDB). The use of these extensions is likely to make other debuggers
2355 crash or refuse to read the program.
2358 Produce debugging information in DWARF version 2 format (if that is
2359 supported). This is the format used by DBX on IRIX 6.
2362 @itemx -ggdb@var{level}
2363 @itemx -gstabs@var{level}
2364 @itemx -gcoff@var{level}
2365 @itemx -gxcoff@var{level}
2366 @itemx -gdwarf@var{level}
2367 @itemx -gdwarf-2@var{level}
2368 Request debugging information and also use @var{level} to specify how
2369 much information. The default level is 2.
2371 Level 1 produces minimal information, enough for making backtraces in
2372 parts of the program that you don't plan to debug. This includes
2373 descriptions of functions and external variables, but no information
2374 about local variables and no line numbers.
2376 Level 3 includes extra information, such as all the macro definitions
2377 present in the program. Some debuggers support macro expansion when
2382 Generate extra code to write profile information suitable for the
2383 analysis program @code{prof}. You must use this option when compiling
2384 the source files you want data about, and you must also use it when
2387 @cindex @code{gprof}
2389 Generate extra code to write profile information suitable for the
2390 analysis program @code{gprof}. You must use this option when compiling
2391 the source files you want data about, and you must also use it when
2396 Generate extra code to write profile information for basic blocks, which will
2397 record the number of times each basic block is executed, the basic block start
2398 address, and the function name containing the basic block. If @samp{-g} is
2399 used, the line number and filename of the start of the basic block will also be
2400 recorded. If not overridden by the machine description, the default action is
2401 to append to the text file @file{bb.out}.
2403 This data could be analyzed by a program like @code{tcov}. Note,
2404 however, that the format of the data is not what @code{tcov} expects.
2405 Eventually GNU @code{gprof} should be extended to process this data.
2408 Makes the compiler print out each function name as it is compiled, and
2409 print some statistics about each pass when it finishes.
2412 Generate extra code to profile basic blocks. Your executable will
2413 produce output that is a superset of that produced when @samp{-a} is
2414 used. Additional output is the source and target address of the basic
2415 blocks where a jump takes place, the number of times a jump is executed,
2416 and (optionally) the complete sequence of basic blocks being executed.
2417 The output is appended to file @file{bb.out}.
2419 You can examine different profiling aspects without recompilation. Your
2420 executable will read a list of function names from file @file{bb.in}.
2421 Profiling starts when a function on the list is entered and stops when
2422 that invocation is exited. To exclude a function from profiling, prefix
2423 its name with `-'. If a function name is not unique, you can
2424 disambiguate it by writing it in the form
2425 @samp{/path/filename.d:functionname}. Your executable will write the
2426 available paths and filenames in file @file{bb.out}.
2428 Several function names have a special meaning:
2431 Write source, target and frequency of jumps to file @file{bb.out}.
2432 @item __bb_hidecall__
2433 Exclude function calls from frequency count.
2434 @item __bb_showret__
2435 Include function returns in frequency count.
2437 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2438 The file will be compressed using the program @samp{gzip}, which must
2439 exist in your @env{PATH}. On systems without the @samp{popen}
2440 function, the file will be named @file{bbtrace} and will not be
2441 compressed. @strong{Profiling for even a few seconds on these systems
2442 will produce a very large file.} Note: @code{__bb_hidecall__} and
2443 @code{__bb_showret__} will not affect the sequence written to
2447 Here's a short example using different profiling parameters
2448 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2449 1 and 2 and is called twice from block 3 of function @code{main}. After
2450 the calls, block 3 transfers control to block 4 of @code{main}.
2452 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2453 the following sequence of blocks is written to file @file{bbtrace.gz}:
2454 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2455 the return is to a point inside the block and not to the top. The
2456 block address 0 always indicates, that control is transferred
2457 to the trace from somewhere outside the observed functions. With
2458 @samp{-foo} added to @file{bb.in}, the blocks of function
2459 @code{foo} are removed from the trace, so only 0 3 4 remains.
2461 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2462 jump frequencies will be written to file @file{bb.out}. The
2463 frequencies are obtained by constructing a trace of blocks
2464 and incrementing a counter for every neighbouring pair of blocks
2465 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2469 Jump from block 0x0 to block 0x3 executed 1 time(s)
2470 Jump from block 0x3 to block 0x1 executed 1 time(s)
2471 Jump from block 0x1 to block 0x2 executed 2 time(s)
2472 Jump from block 0x2 to block 0x1 executed 1 time(s)
2473 Jump from block 0x2 to block 0x4 executed 1 time(s)
2476 With @code{__bb_hidecall__}, control transfer due to call instructions
2477 is removed from the trace, that is the trace is cut into three parts: 0
2478 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2479 to return instructions is added to the trace. The trace becomes: 0 3 1
2480 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2481 written to @file{bbtrace.gz}. It is solely used for counting jump
2484 @item -fprofile-arcs
2485 Instrument @dfn{arcs} during compilation. For each function of your
2486 program, GCC creates a program flow graph, then finds a spanning tree
2487 for the graph. Only arcs that are not on the spanning tree have to be
2488 instrumented: the compiler adds code to count the number of times that these
2489 arcs are executed. When an arc is the only exit or only entrance to a
2490 block, the instrumentation code can be added to the block; otherwise, a
2491 new basic block must be created to hold the instrumentation code.
2493 Since not every arc in the program must be instrumented, programs
2494 compiled with this option run faster than programs compiled with
2495 @samp{-a}, which adds instrumentation code to every basic block in the
2496 program. The tradeoff: since @code{gcov} does not have
2497 execution counts for all branches, it must start with the execution
2498 counts for the instrumented branches, and then iterate over the program
2499 flow graph until the entire graph has been solved. Hence, @code{gcov}
2500 runs a little more slowly than a program which uses information from
2503 @samp{-fprofile-arcs} also makes it possible to estimate branch
2504 probabilities, and to calculate basic block execution counts. In
2505 general, basic block execution counts do not give enough information to
2506 estimate all branch probabilities. When the compiled program exits, it
2507 saves the arc execution counts to a file called
2508 @file{@var{sourcename}.da}. Use the compiler option
2509 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2510 Control Optimization}) when recompiling, to optimize using estimated
2511 branch probabilities.
2514 @item -ftest-coverage
2515 Create data files for the @code{gcov} code-coverage utility
2516 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2517 The data file names begin with the name of your source file:
2520 @item @var{sourcename}.bb
2521 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2522 associate basic block execution counts with line numbers.
2524 @item @var{sourcename}.bbg
2525 A list of all arcs in the program flow graph. This allows @code{gcov}
2526 to reconstruct the program flow graph, so that it can compute all basic
2527 block and arc execution counts from the information in the
2528 @code{@var{sourcename}.da} file (this last file is the output from
2529 @samp{-fprofile-arcs}).
2532 @item -d@var{letters}
2533 Says to make debugging dumps during compilation at times specified by
2534 @var{letters}. This is used for debugging the compiler. The file names
2535 for most of the dumps are made by appending a pass number and a word to
2536 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2537 Here are the possible letters for use in @var{letters}, and their meanings:
2541 Annotate the assembler output with miscellaneous debugging information.
2543 Dump after computing branch probabilities, to @file{@var{file}.11.bp}.
2545 Dump after block reordering, to @file{@var{file}.26.bbro}.
2547 Dump after instruction combination, to the file @file{@var{file}.14.combine}.
2549 Dump after the first if conversion, to the file @file{@var{file}.15.ce}.
2551 Dump after delayed branch scheduling, to @file{@var{file}.29.dbr}.
2553 Dump all macro definitions, at the end of preprocessing, in addition to
2556 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
2557 @file{@var{file}.06.ussa}.
2559 Dump after the second if conversion, to @file{@var{file}.24.ce2}.
2561 Dump after life analysis, to @file{@var{file}.13.life}.
2563 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.04.addressof}.
2565 Dump after global register allocation, to @file{@var{file}.19.greg}.
2567 Dump after post-reload CSE and other optimizations, to @file{@var{file}.20.postreload}.
2569 Dump after GCSE, to @file{@var{file}.08.gcse}.
2571 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2573 Dump after the first jump optimization, to @file{@var{file}.02.jump}.
2575 Dump after the last jump optimization, to @file{@var{file}.27.jump2}.
2577 Dump after conversion from registers to stack, to @file{@var{file}.29.stack}.
2579 Dump after local register allocation, to @file{@var{file}.18.lreg}.
2581 Dump after loop optimization, to @file{@var{file}.09.loop}.
2583 Dump after performing the machine dependent reorganisation pass, to
2584 @file{@var{file}.28.mach}.
2586 Dump after register renumbering, to @file{@var{file}.23.rnreg}.
2588 Dump after the register move pass, to @file{@var{file}.16.regmove}.
2590 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2592 Dump after the second instruction scheduling pass, to
2593 @file{@var{file}.25.sched2}.
2595 Dump after CSE (including the jump optimization that sometimes follows
2596 CSE), to @file{@var{file}.03.cse}.
2598 Dump after the first instruction scheduling pass, to
2599 @file{@var{file}.17.sched}.
2601 Dump after the second CSE pass (including the jump optimization that
2602 sometimes follows CSE), to @file{@var{file}.10.cse2}.
2604 Dump after the second flow pass, to @file{@var{file}.21.flow2}.
2606 Dump after dead code elimination, to @file{@var{file}.06.dce}.
2608 Dump after the peephole pass, to @file{@var{file}.22.peephole2}.
2610 Produce all the dumps listed above.
2612 Print statistics on memory usage, at the end of the run, to
2615 Annotate the assembler output with a comment indicating which
2616 pattern and alternative was used. The length of each instruction is
2619 Dump the RTL in the assembler output as a comment before each instruction.
2620 Also turns on @samp{-dp} annotation.
2622 For each of the other indicated dump files (except for
2623 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2624 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2626 Just generate RTL for a function instead of compiling it. Usually used
2629 Dump debugging information during parsing, to standard error.
2632 @item -fdump-unnumbered
2633 When doing debugging dumps (see -d option above), suppress instruction
2634 numbers and line number note output. This makes it more feasible to
2635 use diff on debugging dumps for compiler invocations with different
2636 options, in particular with and without -g.
2638 @item -fdump-translation-unit-@var{file} (C and C++ only)
2639 Dump a representation of the tree structure for the entire translation
2642 @item -fpretend-float
2643 When running a cross-compiler, pretend that the target machine uses the
2644 same floating point format as the host machine. This causes incorrect
2645 output of the actual floating constants, but the actual instruction
2646 sequence will probably be the same as GCC would make when running on
2650 Store the usual ``temporary'' intermediate files permanently; place them
2651 in the current directory and name them based on the source file. Thus,
2652 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2653 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
2654 preprocessed @file{foo.i} output file even though the compiler now
2655 normally uses an integrated preprocessor.
2658 Report the CPU time taken by each subprocess in the compilation
2659 sequence. For C source files, this is the compiler proper and assembler
2660 (plus the linker if linking is done). The output looks like this:
2667 The first number on each line is the ``user time,'' that is time spent
2668 executing the program itself. The second number is ``system time,''
2669 time spent executing operating system routines on behalf of the program.
2670 Both numbers are in seconds.
2672 @item -print-file-name=@var{library}
2673 Print the full absolute name of the library file @var{library} that
2674 would be used when linking---and don't do anything else. With this
2675 option, GCC does not compile or link anything; it just prints the
2678 @item -print-prog-name=@var{program}
2679 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2681 @item -print-libgcc-file-name
2682 Same as @samp{-print-file-name=libgcc.a}.
2684 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2685 but you do want to link with @file{libgcc.a}. You can do
2688 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2691 @item -print-search-dirs
2692 Print the name of the configured installation directory and a list of
2693 program and library directories gcc will search---and don't do anything else.
2695 This is useful when gcc prints the error message
2696 @samp{installation problem, cannot exec cpp0: No such file or directory}.
2697 To resolve this you either need to put @file{cpp0} and the other compiler
2698 components where gcc expects to find them, or you can set the environment
2699 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
2700 Don't forget the trailing '/'.
2701 @xref{Environment Variables}.
2704 @node Optimize Options
2705 @section Options That Control Optimization
2706 @cindex optimize options
2707 @cindex options, optimization
2709 These options control various sorts of optimizations:
2714 Optimize. Optimizing compilation takes somewhat more time, and a lot
2715 more memory for a large function.
2717 Without @samp{-O}, the compiler's goal is to reduce the cost of
2718 compilation and to make debugging produce the expected results.
2719 Statements are independent: if you stop the program with a breakpoint
2720 between statements, you can then assign a new value to any variable or
2721 change the program counter to any other statement in the function and
2722 get exactly the results you would expect from the source code.
2724 Without @samp{-O}, the compiler only allocates variables declared
2725 @code{register} in registers. The resulting compiled code is a little
2726 worse than produced by PCC without @samp{-O}.
2728 With @samp{-O}, the compiler tries to reduce code size and execution
2731 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2732 and @samp{-fdefer-pop} on all machines. The compiler turns on
2733 @samp{-fdelayed-branch} on machines that have delay slots, and
2734 @samp{-fomit-frame-pointer} on machines that can support debugging even
2735 without a frame pointer. On some machines the compiler also turns
2736 on other flags.@refill
2739 Optimize even more. GCC performs nearly all supported optimizations
2740 that do not involve a space-speed tradeoff. The compiler does not
2741 perform loop unrolling or function inlining when you specify @samp{-O2}.
2742 As compared to @samp{-O}, this option increases both compilation time
2743 and the performance of the generated code.
2745 @samp{-O2} turns on all optional optimizations except for loop unrolling,
2746 function inlining, and register renaming. It also turns on the
2747 @samp{-fforce-mem} option on all machines and frame pointer elimination
2748 on machines where doing so does not interfere with debugging.
2751 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2752 @samp{-O2} and also turns on the @samp{-finline-functions} and
2753 @samp{-frename-registers} options.
2759 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2760 do not typically increase code size. It also performs further
2761 optimizations designed to reduce code size.
2763 If you use multiple @samp{-O} options, with or without level numbers,
2764 the last such option is the one that is effective.
2767 Options of the form @samp{-f@var{flag}} specify machine-independent
2768 flags. Most flags have both positive and negative forms; the negative
2769 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2770 only one of the forms is listed---the one which is not the default.
2771 You can figure out the other form by either removing @samp{no-} or
2776 Do not store floating point variables in registers, and inhibit other
2777 options that might change whether a floating point value is taken from a
2780 @cindex floating point precision
2781 This option prevents undesirable excess precision on machines such as
2782 the 68000 where the floating registers (of the 68881) keep more
2783 precision than a @code{double} is supposed to have. Similarly for the
2784 x86 architecture. For most programs, the excess precision does only
2785 good, but a few programs rely on the precise definition of IEEE floating
2786 point. Use @samp{-ffloat-store} for such programs, after modifying
2787 them to store all pertinent intermediate computations into variables.
2789 @item -fno-default-inline
2790 Do not make member functions inline by default merely because they are
2791 defined inside the class scope (C++ only). Otherwise, when you specify
2792 @w{@samp{-O}}, member functions defined inside class scope are compiled
2793 inline by default; i.e., you don't need to add @samp{inline} in front of
2794 the member function name.
2796 @item -fno-defer-pop
2797 Always pop the arguments to each function call as soon as that function
2798 returns. For machines which must pop arguments after a function call,
2799 the compiler normally lets arguments accumulate on the stack for several
2800 function calls and pops them all at once.
2803 Force memory operands to be copied into registers before doing
2804 arithmetic on them. This produces better code by making all memory
2805 references potential common subexpressions. When they are not common
2806 subexpressions, instruction combination should eliminate the separate
2807 register-load. The @samp{-O2} option turns on this option.
2810 Force memory address constants to be copied into registers before
2811 doing arithmetic on them. This may produce better code just as
2812 @samp{-fforce-mem} may.
2814 @item -fomit-frame-pointer
2815 Don't keep the frame pointer in a register for functions that
2816 don't need one. This avoids the instructions to save, set up and
2817 restore frame pointers; it also makes an extra register available
2818 in many functions. @strong{It also makes debugging impossible on
2822 On some machines, such as the Vax, this flag has no effect, because
2823 the standard calling sequence automatically handles the frame pointer
2824 and nothing is saved by pretending it doesn't exist. The
2825 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2826 whether a target machine supports this flag. @xref{Registers}.@refill
2829 On some machines, such as the Vax, this flag has no effect, because
2830 the standard calling sequence automatically handles the frame pointer
2831 and nothing is saved by pretending it doesn't exist. The
2832 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2833 whether a target machine supports this flag. @xref{Registers,,Register
2834 Usage, gcc.info, Using and Porting GCC}.@refill
2837 @item -foptimize-sibling-calls
2838 Optimize sibling and tail recursive calls.
2841 This option generates traps for signed overflow on addition, subtraction,
2842 multiplication operations.
2845 Don't pay attention to the @code{inline} keyword. Normally this option
2846 is used to keep the compiler from expanding any functions inline.
2847 Note that if you are not optimizing, no functions can be expanded inline.
2849 @item -finline-functions
2850 Integrate all simple functions into their callers. The compiler
2851 heuristically decides which functions are simple enough to be worth
2852 integrating in this way.
2854 If all calls to a given function are integrated, and the function is
2855 declared @code{static}, then the function is normally not output as
2856 assembler code in its own right.
2858 @item -finline-limit=@var{n}
2859 By default, gcc limits the size of functions that can be inlined. This flag
2860 allows the control of this limit for functions that are explicitly marked as
2861 inline (ie marked with the inline keyword or defined within the class
2862 definition in c++). @var{n} is the size of functions that can be inlined in
2863 number of pseudo instructions (not counting parameter handling). The default
2864 value of n is 10000. Increasing this value can result in more inlined code at
2865 the cost of compilation time and memory consumption. Decreasing usually makes
2866 the compilation faster and less code will be inlined (which presumably
2867 means slower programs). This option is particularly useful for programs that
2868 use inlining heavily such as those based on recursive templates with c++.
2870 @emph{Note:} pseudo instruction represents, in this particular context, an
2871 abstract measurement of function's size. In no way, it represents a count
2872 of assembly instructions and as such its exact meaning might change from one
2873 release to an another.
2875 @item -fkeep-inline-functions
2876 Even if all calls to a given function are integrated, and the function
2877 is declared @code{static}, nevertheless output a separate run-time
2878 callable version of the function. This switch does not affect
2879 @code{extern inline} functions.
2881 @item -fkeep-static-consts
2882 Emit variables declared @code{static const} when optimization isn't turned
2883 on, even if the variables aren't referenced.
2885 GCC enables this option by default. If you want to force the compiler to
2886 check if the variable was referenced, regardless of whether or not
2887 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2889 @item -fno-function-cse
2890 Do not put function addresses in registers; make each instruction that
2891 calls a constant function contain the function's address explicitly.
2893 This option results in less efficient code, but some strange hacks
2894 that alter the assembler output may be confused by the optimizations
2895 performed when this option is not used.
2898 This option allows GCC to violate some ISO or IEEE rules and/or
2899 specifications in the interest of optimizing code for speed. For
2900 example, it allows the compiler to assume arguments to the @code{sqrt}
2901 function are non-negative numbers and that no floating-point values
2904 This option should never be turned on by any @samp{-O} option since
2905 it can result in incorrect output for programs which depend on
2906 an exact implementation of IEEE or ISO rules/specifications for
2909 @item -fno-math-errno
2910 Do not set ERRNO after calling math functions that are executed
2911 with a single instruction, e.g., sqrt. A program that relies on
2912 IEEE exceptions for math error handling may want to use this flag
2913 for speed while maintaining IEEE arithmetic compatibility.
2915 The default is @samp{-fmath-errno}. The @samp{-ffast-math} option
2916 sets @samp{-fno-math-errno}.
2919 @c following causes underfulls.. they don't look great, but we deal.
2921 The following options control specific optimizations. The @samp{-O2}
2922 option turns on all of these optimizations except @samp{-funroll-loops}
2923 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
2924 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
2925 but specific machines may handle it differently.
2927 You can use the following flags in the rare cases when ``fine-tuning''
2928 of optimizations to be performed is desired.
2931 @item -fstrength-reduce
2932 Perform the optimizations of loop strength reduction and
2933 elimination of iteration variables.
2935 @item -fthread-jumps
2936 Perform optimizations where we check to see if a jump branches to a
2937 location where another comparison subsumed by the first is found. If
2938 so, the first branch is redirected to either the destination of the
2939 second branch or a point immediately following it, depending on whether
2940 the condition is known to be true or false.
2942 @item -fcse-follow-jumps
2943 In common subexpression elimination, scan through jump instructions
2944 when the target of the jump is not reached by any other path. For
2945 example, when CSE encounters an @code{if} statement with an
2946 @code{else} clause, CSE will follow the jump when the condition
2949 @item -fcse-skip-blocks
2950 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
2951 follow jumps which conditionally skip over blocks. When CSE
2952 encounters a simple @code{if} statement with no else clause,
2953 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
2954 body of the @code{if}.
2956 @item -frerun-cse-after-loop
2957 Re-run common subexpression elimination after loop optimizations has been
2960 @item -frerun-loop-opt
2961 Run the loop optimizer twice.
2964 Perform a global common subexpression elimination pass.
2965 This pass also performs global constant and copy propagation.
2967 @item -fdelete-null-pointer-checks
2968 Use global dataflow analysis to identify and eliminate useless null
2969 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
2970 halting the program may not work properly with this option. Use
2971 -fno-delete-null-pointer-checks to disable this optimizing for programs
2972 which depend on that behavior.
2974 @item -fexpensive-optimizations
2975 Perform a number of minor optimizations that are relatively expensive.
2977 @item -foptimize-register-move
2979 Attempt to reassign register numbers in move instructions and as
2980 operands of other simple instructions in order to maximize the amount of
2981 register tying. This is especially helpful on machines with two-operand
2982 instructions. GCC enables this optimization by default with @samp{-O2}
2985 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
2988 @item -fdelayed-branch
2989 If supported for the target machine, attempt to reorder instructions
2990 to exploit instruction slots available after delayed branch
2993 @item -fschedule-insns
2994 If supported for the target machine, attempt to reorder instructions to
2995 eliminate execution stalls due to required data being unavailable. This
2996 helps machines that have slow floating point or memory load instructions
2997 by allowing other instructions to be issued until the result of the load
2998 or floating point instruction is required.
3000 @item -fschedule-insns2
3001 Similar to @samp{-fschedule-insns}, but requests an additional pass of
3002 instruction scheduling after register allocation has been done. This is
3003 especially useful on machines with a relatively small number of
3004 registers and where memory load instructions take more than one cycle.
3006 @item -ffunction-sections
3007 @itemx -fdata-sections
3008 Place each function or data item into its own section in the output
3009 file if the target supports arbitrary sections. The name of the
3010 function or the name of the data item determines the section's name
3013 Use these options on systems where the linker can perform optimizations
3014 to improve locality of reference in the instruction space. HPPA
3015 processors running HP-UX and Sparc processors running Solaris 2 have
3016 linkers with such optimizations. Other systems using the ELF object format
3017 as well as AIX may have these optimizations in the future.
3019 Only use these options when there are significant benefits from doing
3020 so. When you specify these options, the assembler and linker will
3021 create larger object and executable files and will also be slower.
3022 You will not be able to use @code{gprof} on all systems if you
3023 specify this option and you may have problems with debugging if
3024 you specify both this option and @samp{-g}.
3026 @item -fcaller-saves
3027 Enable values to be allocated in registers that will be clobbered by
3028 function calls, by emitting extra instructions to save and restore the
3029 registers around such calls. Such allocation is done only when it
3030 seems to result in better code than would otherwise be produced.
3032 This option is always enabled by default on certain machines, usually
3033 those which have no call-preserved registers to use instead.
3035 For all machines, optimization level 2 and higher enables this flag by
3038 @item -funroll-loops
3039 Perform the optimization of loop unrolling. This is only done for loops
3040 whose number of iterations can be determined at compile time or run time.
3041 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
3042 @samp{-frerun-cse-after-loop}.
3044 @item -funroll-all-loops
3045 Perform the optimization of loop unrolling. This is done for all loops
3046 and usually makes programs run more slowly. @samp{-funroll-all-loops}
3047 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
3049 @item -fmove-all-movables
3050 Forces all invariant computations in loops to be moved
3053 @item -freduce-all-givs
3054 Forces all general-induction variables in loops to be
3057 @emph{Note:} When compiling programs written in Fortran,
3058 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
3059 by default when you use the optimizer.
3061 These options may generate better or worse code; results are highly
3062 dependent on the structure of loops within the source code.
3064 These two options are intended to be removed someday, once
3065 they have helped determine the efficacy of various
3066 approaches to improving loop optimizations.
3068 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3069 know how use of these options affects
3070 the performance of your production code.
3071 We're very interested in code that runs @emph{slower}
3072 when these options are @emph{enabled}.
3075 Disable any machine-specific peephole optimizations.
3077 @item -fbranch-probabilities
3078 After running a program compiled with @samp{-fprofile-arcs}
3079 (@pxref{Debugging Options,, Options for Debugging Your Program or
3080 @command{gcc}}), you can compile it a second time using
3081 @samp{-fbranch-probabilities}, to improve optimizations based on
3082 guessing the path a branch might take.
3085 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3086 note on the first instruction of each basic block, and a
3087 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3088 These can be used to improve optimization. Currently, they are only
3089 used in one place: in @file{reorg.c}, instead of guessing which path a
3090 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3091 exactly determine which path is taken more often.
3094 @item -fstrict-aliasing
3095 Allows the compiler to assume the strictest aliasing rules applicable to
3096 the language being compiled. For C (and C++), this activates
3097 optimizations based on the type of expressions. In particular, an
3098 object of one type is assumed never to reside at the same address as an
3099 object of a different type, unless the types are almost the same. For
3100 example, an @code{unsigned int} can alias an @code{int}, but not a
3101 @code{void*} or a @code{double}. A character type may alias any other
3104 Pay special attention to code like this:
3117 The practice of reading from a different union member than the one most
3118 recently written to (called ``type-punning'') is common. Even with
3119 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
3120 is accessed through the union type. So, the code above will work as
3121 expected. However, this code might not:
3133 Every language that wishes to perform language-specific alias analysis
3134 should define a function that computes, given an @code{tree}
3135 node, an alias set for the node. Nodes in different alias sets are not
3136 allowed to alias. For an example, see the C front-end function
3137 @code{c_get_alias_set}.
3140 @item -falign-functions
3141 @itemx -falign-functions=@var{n}
3142 Align the start of functions to the next power-of-two greater than
3143 @var{n}, skipping up to @var{n} bytes. For instance,
3144 @samp{-falign-functions=32} aligns functions to the next 32-byte
3145 boundary, but @samp{-falign-functions=24} would align to the next
3146 32-byte boundary only if this can be done by skipping 23 bytes or less.
3148 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
3149 equivalent and mean that functions will not be aligned.
3151 Some assemblers only support this flag when @var{n} is a power of two;
3152 in that case, it is rounded up.
3154 If @var{n} is not specified, use a machine-dependent default.
3156 @item -falign-labels
3157 @itemx -falign-labels=@var{n}
3158 Align all branch targets to a power-of-two boundary, skipping up to
3159 @var{n} bytes like @samp{-falign-functions}. This option can easily
3160 make code slower, because it must insert dummy operations for when the
3161 branch target is reached in the usual flow of the code.
3163 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
3164 are greater than this value, then their values are used instead.
3166 If @var{n} is not specified, use a machine-dependent default which is
3167 very likely to be @samp{1}, meaning no alignment.
3170 @itemx -falign-loops=@var{n}
3171 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3172 like @samp{-falign-functions}. The hope is that the loop will be
3173 executed many times, which will make up for any execution of the dummy
3176 If @var{n} is not specified, use a machine-dependent default.
3179 @itemx -falign-jumps=@var{n}
3180 Align branch targets to a power-of-two boundary, for branch targets
3181 where the targets can only be reached by jumping, skipping up to @var{n}
3182 bytes like @samp{-falign-functions}. In this case, no dummy operations
3185 If @var{n} is not specified, use a machine-dependent default.
3188 Perform optimizations in static single assignment form. Each function's
3189 flow graph is translated into SSA form, optimizations are performed, and
3190 the flow graph is translated back from SSA form. User's should not
3191 specify this option, since it is not yet ready for production use.
3194 Perform dead-code elimination in SSA form. Requires @samp{-fssa}. Like
3195 @samp{-fssa}, this is an experimental feature.
3197 @item -fsingle-precision-constant
3198 Treat floating point constant as single precision constant instead of
3199 implicitly converting it to double precision constant.
3201 @item -frename-registers
3202 Attempt to avoid false dependancies in scheduled code by making use
3203 of registers left over after register allocation. This optimization
3204 will most benefit processors with lots of registers. It can, however,
3205 make debugging impossible, since variables will no longer stay in
3206 a ``home register''.
3209 @node Preprocessor Options
3210 @section Options Controlling the Preprocessor
3211 @cindex preprocessor options
3212 @cindex options, preprocessor
3214 These options control the C preprocessor, which is run on each C source
3215 file before actual compilation.
3217 If you use the @samp{-E} option, nothing is done except preprocessing.
3218 Some of these options make sense only together with @samp{-E} because
3219 they cause the preprocessor output to be unsuitable for actual
3223 @item -include @var{file}
3224 Process @var{file} as input before processing the regular input file.
3225 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
3226 and @samp{-U} options on the command line are always processed before
3227 @samp{-include @var{file}}, regardless of the order in which they are
3228 written. All the @samp{-include} and @samp{-imacros} options are
3229 processed in the order in which they are written.
3231 @item -imacros @var{file}
3232 Process @var{file} as input, discarding the resulting output, before
3233 processing the regular input file. Because the output generated from
3234 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
3235 is to make the macros defined in @var{file} available for use in the
3236 main input. All the @samp{-include} and @samp{-imacros} options are
3237 processed in the order in which they are written.
3239 @item -idirafter @var{dir}
3240 @cindex second include path
3241 Add the directory @var{dir} to the second include path. The directories
3242 on the second include path are searched when a header file is not found
3243 in any of the directories in the main include path (the one that
3246 @item -iprefix @var{prefix}
3247 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
3250 @item -iwithprefix @var{dir}
3251 Add a directory to the second include path. The directory's name is
3252 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3253 specified previously with @samp{-iprefix}. If you have not specified a
3254 prefix yet, the directory containing the installed passes of the
3255 compiler is used as the default.
3257 @item -iwithprefixbefore @var{dir}
3258 Add a directory to the main include path. The directory's name is made
3259 by concatenating @var{prefix} and @var{dir}, as in the case of
3260 @samp{-iwithprefix}.
3262 @item -isystem @var{dir}
3263 Add a directory to the beginning of the second include path, marking it
3264 as a system directory, so that it gets the same special treatment as
3265 is applied to the standard system directories.
3268 Do not search the standard system directories for header files. Only
3269 the directories you have specified with @samp{-I} options (and the
3270 current directory, if appropriate) are searched. @xref{Directory
3271 Options}, for information on @samp{-I}.
3273 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
3274 search path to only those directories you specify explicitly.
3278 When searching for a header file in a directory, remap file names if a
3279 file named @file{header.gcc} exists in that directory. This can be used
3280 to work around limitations of file systems with file name restrictions.
3281 The @file{header.gcc} file should contain a series of lines with two
3282 tokens on each line: the first token is the name to map, and the second
3283 token is the actual name to use.
3286 Do not predefine any nonstandard macros. (Including architecture flags).
3289 Run only the C preprocessor. Preprocess all the C source files
3290 specified and output the results to standard output or to the
3291 specified output file.
3294 Tell the preprocessor not to discard comments. Used with the
3298 Tell the preprocessor not to generate @samp{#line} directives.
3299 Used with the @samp{-E} option.
3302 @cindex dependencies, make
3305 Instead of outputting the result of preprocessing, output a rule
3306 suitable for @code{make} describing the dependencies of the main source
3307 file. The preprocessor outputs one @code{make} rule containing the
3308 object file name for that source file, a colon, and the names of all the
3309 included files. If there are many included files then the rule is split
3310 into several lines using @samp{\}-newline.
3312 @samp{-M} implies @samp{-E}.
3316 Like @samp{-M}, but mention only the files included with @samp{#include
3317 "@var{file}"}. System header files included with @samp{#include
3318 <@var{file}>} are omitted.
3320 @item -MF @var{file}
3322 When used with @samp{-M} or @samp{-MM}, specifies a file to write the
3323 dependencies to. This allows the preprocessor to write the preprocessed
3324 file to stdout normally. If no @samp{-MF} switch is given, CPP sends
3325 the rules to stdout and suppresses normal preprocessed output.
3327 Another way to specify output of a @code{make} rule is by setting
3328 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
3333 When used with @samp{-M} or @samp{-MM}, @samp{-MG} says to treat missing
3334 header files as generated files and assume they live in the same
3335 directory as the source file. It suppresses preprocessed output, as a
3336 missing header file is ordinarily an error.
3338 This feature is used in automatic updating of makefiles.
3342 This option instructs CPP to add a phony target for each dependency
3343 other than the main file, causing each to depend on nothing. These
3344 dummy rules work around errors @code{make} gives if you remove header
3345 files without updating the @code{Makefile} to match.
3347 This is typical output:-
3350 /tmp/test.o: /tmp/test.c /tmp/test.h
3355 @item -MQ @var{target}
3356 @item -MT @var{target}
3359 By default CPP uses the main file name, including any path, and appends
3360 the object suffix, normally ``.o'', to it to obtain the name of the
3361 target for dependency generation. With @samp{-MT} you can specify a
3362 target yourself, overriding the default one.
3364 If you want multiple targets, you can specify them as a single argument
3365 to @samp{-MT}, or use multiple @samp{-MT} options.
3367 The targets you specify are output in the order they appear on the
3368 command line. @samp{-MQ} is identical to @samp{-MT}, except that the
3369 target name is quoted for Make, but with @samp{-MT} it isn't. For
3370 example, -MT '$(objpfx)foo.o' gives
3373 $(objpfx)foo.o: /tmp/foo.c
3376 but -MQ '$(objpfx)foo.o' gives
3379 $$(objpfx)foo.o: /tmp/foo.c
3382 The default target is automatically quoted, as if it were given with
3386 Print the name of each header file used, in addition to other normal
3389 @item -A@var{question}(@var{answer})
3390 Assert the answer @var{answer} for @var{question}, in case it is tested
3391 with a preprocessing conditional such as @samp{#if
3392 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
3393 assertions that normally describe the target machine.
3396 Define macro @var{macro} with the string @samp{1} as its definition.
3398 @item -D@var{macro}=@var{defn}
3399 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
3400 the command line are processed before any @samp{-U} options.
3402 Any @samp{-D} and @samp{-U} options on the command line are processed in
3403 order, and always before @samp{-imacros @var{file}}, regardless of the
3404 order in which they are written.
3407 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
3408 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
3411 Any @samp{-D} and @samp{-U} options on the command line are processed in
3412 order, and always before @samp{-imacros @var{file}}, regardless of the
3413 order in which they are written.
3416 Tell the preprocessor to output only a list of the macro definitions
3417 that are in effect at the end of preprocessing. Used with the @samp{-E}
3421 Tell the preprocessing to pass all macro definitions into the output, in
3422 their proper sequence in the rest of the output.
3425 Like @samp{-dD} except that the macro arguments and contents are omitted.
3426 Only @samp{#define @var{name}} is included in the output.
3430 Output @samp{#include} directives in addition to the result of
3435 Process ISO standard trigraph sequences. These are three-character
3436 sequences, all starting with @samp{??}, that are defined by ISO C to
3437 stand for single characters. For example, @samp{??/} stands for
3438 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
3439 default, GCC ignores trigraphs, but in standard-conforming modes it
3440 converts them. See the @samp{-std} and @samp{-ansi} options.
3442 The nine trigraph sequences are
3473 Trigraph support is not popular, so many compilers do not implement it
3474 properly. Portable code should not rely on trigraphs being either
3475 converted or ignored.
3477 @item -Wp,@var{option}
3478 Pass @var{option} as an option to the preprocessor. If @var{option}
3479 contains commas, it is split into multiple options at the commas.
3482 @node Assembler Options
3483 @section Passing Options to the Assembler
3485 @c prevent bad page break with this line
3486 You can pass options to the assembler.
3489 @item -Wa,@var{option}
3490 Pass @var{option} as an option to the assembler. If @var{option}
3491 contains commas, it is split into multiple options at the commas.
3495 @section Options for Linking
3496 @cindex link options
3497 @cindex options, linking
3499 These options come into play when the compiler links object files into
3500 an executable output file. They are meaningless if the compiler is
3501 not doing a link step.
3505 @item @var{object-file-name}
3506 A file name that does not end in a special recognized suffix is
3507 considered to name an object file or library. (Object files are
3508 distinguished from libraries by the linker according to the file
3509 contents.) If linking is done, these object files are used as input
3515 If any of these options is used, then the linker is not run, and
3516 object file names should not be used as arguments. @xref{Overall
3520 @item -l@var{library}
3521 Search the library named @var{library} when linking.
3523 It makes a difference where in the command you write this option; the
3524 linker searches processes libraries and object files in the order they
3525 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
3526 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
3527 to functions in @samp{z}, those functions may not be loaded.
3529 The linker searches a standard list of directories for the library,
3530 which is actually a file named @file{lib@var{library}.a}. The linker
3531 then uses this file as if it had been specified precisely by name.
3533 The directories searched include several standard system directories
3534 plus any that you specify with @samp{-L}.
3536 Normally the files found this way are library files---archive files
3537 whose members are object files. The linker handles an archive file by
3538 scanning through it for members which define symbols that have so far
3539 been referenced but not defined. But if the file that is found is an
3540 ordinary object file, it is linked in the usual fashion. The only
3541 difference between using an @samp{-l} option and specifying a file name
3542 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
3543 and searches several directories.
3546 You need this special case of the @samp{-l} option in order to
3547 link an Objective C program.
3550 Do not use the standard system startup files when linking.
3551 The standard system libraries are used normally, unless @option{-nostdlib}
3552 or @option{-nodefaultlibs} is used.
3554 @item -nodefaultlibs
3555 Do not use the standard system libraries when linking.
3556 Only the libraries you specify will be passed to the linker.
3557 The standard startup files are used normally, unless @option{-nostartfiles}
3558 is used. The compiler may generate calls to memcmp, memset, and memcpy
3559 for System V (and ISO C) environments or to bcopy and bzero for
3560 BSD environments. These entries are usually resolved by entries in
3561 libc. These entry points should be supplied through some other
3562 mechanism when this option is specified.
3565 Do not use the standard system startup files or libraries when linking.
3566 No startup files and only the libraries you specify will be passed to
3567 the linker. The compiler may generate calls to memcmp, memset, and memcpy
3568 for System V (and ISO C) environments or to bcopy and bzero for
3569 BSD environments. These entries are usually resolved by entries in
3570 libc. These entry points should be supplied through some other
3571 mechanism when this option is specified.
3573 @cindex @code{-lgcc}, use with @code{-nostdlib}
3574 @cindex @code{-nostdlib} and unresolved references
3575 @cindex unresolved references and @code{-nostdlib}
3576 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
3577 @cindex @code{-nodefaultlibs} and unresolved references
3578 @cindex unresolved references and @code{-nodefaultlibs}
3579 One of the standard libraries bypassed by @samp{-nostdlib} and
3580 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
3581 that GCC uses to overcome shortcomings of particular machines, or special
3582 needs for some languages.
3584 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
3588 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
3589 for more discussion of @file{libgcc.a}.)
3591 In most cases, you need @file{libgcc.a} even when you want to avoid
3592 other standard libraries. In other words, when you specify @samp{-nostdlib}
3593 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
3594 This ensures that you have no unresolved references to internal GCC
3595 library subroutines. (For example, @samp{__main}, used to ensure C++
3596 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
3599 Remove all symbol table and relocation information from the executable.
3602 On systems that support dynamic linking, this prevents linking with the shared
3603 libraries. On other systems, this option has no effect.
3606 Produce a shared object which can then be linked with other objects to
3607 form an executable. Not all systems support this option. For predictable
3608 results, you must also specify the same set of options that were used to
3609 generate code (@samp{-fpic}, @samp{-fPIC}, or model suboptions)
3610 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
3611 needs to build supplementary stub code for constructors to work. On
3612 multi-libbed systems, @samp{gcc -shared} must select the correct support
3613 libraries to link against. Failing to supply the correct flags may lead
3614 to subtle defects. Supplying them in cases where they are not necessary
3617 @item -shared-libgcc
3618 @itemx -static-libgcc
3619 On systems that provide @file{libgcc} as a shared library, these options
3620 force the use of either the shared or static version respectively.
3621 If no shared version of @file{libgcc} was built when the compiler was
3622 configured, these options have no effect.
3624 There are several situations in which an application should use the
3625 shared @file{libgcc} instead of the static version. The most common
3626 of these is when the application wishes to throw and catch exceptions
3627 across different shared libraries. In that case, each of the libraries
3628 as well as the application itself should use the shared @file{libgcc}.
3630 At present the GCC driver makes no attempt to recognize the situations
3631 in which the shared @file{libgcc} should be used, and defaults to using
3632 the static @file{libgcc} always. This will likely change in the future,
3633 at which time @samp{-static-libgcc} becomes useful as a means for
3634 overriding GCC's choice.
3637 Bind references to global symbols when building a shared object. Warn
3638 about any unresolved references (unless overridden by the link editor
3639 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
3642 @item -Xlinker @var{option}
3643 Pass @var{option} as an option to the linker. You can use this to
3644 supply system-specific linker options which GCC does not know how to
3647 If you want to pass an option that takes an argument, you must use
3648 @samp{-Xlinker} twice, once for the option and once for the argument.
3649 For example, to pass @samp{-assert definitions}, you must write
3650 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
3651 @samp{-Xlinker "-assert definitions"}, because this passes the entire
3652 string as a single argument, which is not what the linker expects.
3654 @item -Wl,@var{option}
3655 Pass @var{option} as an option to the linker. If @var{option} contains
3656 commas, it is split into multiple options at the commas.
3658 @item -u @var{symbol}
3659 Pretend the symbol @var{symbol} is undefined, to force linking of
3660 library modules to define it. You can use @samp{-u} multiple times with
3661 different symbols to force loading of additional library modules.
3664 @node Directory Options
3665 @section Options for Directory Search
3666 @cindex directory options
3667 @cindex options, directory search
3670 These options specify directories to search for header files, for
3671 libraries and for parts of the compiler:
3675 Add the directory @var{dir} to the head of the list of directories to be
3676 searched for header files. This can be used to override a system header
3677 file, substituting your own version, since these directories are
3678 searched before the system header file directories. If you use more
3679 than one @samp{-I} option, the directories are scanned in left-to-right
3680 order; the standard system directories come after.
3683 Any directories you specify with @samp{-I} options before the @samp{-I-}
3684 option are searched only for the case of @samp{#include "@var{file}"};
3685 they are not searched for @samp{#include <@var{file}>}.
3687 If additional directories are specified with @samp{-I} options after
3688 the @samp{-I-}, these directories are searched for all @samp{#include}
3689 directives. (Ordinarily @emph{all} @samp{-I} directories are used
3692 In addition, the @samp{-I-} option inhibits the use of the current
3693 directory (where the current input file came from) as the first search
3694 directory for @samp{#include "@var{file}"}. There is no way to
3695 override this effect of @samp{-I-}. With @samp{-I.} you can specify
3696 searching the directory which was current when the compiler was
3697 invoked. That is not exactly the same as what the preprocessor does
3698 by default, but it is often satisfactory.
3700 @samp{-I-} does not inhibit the use of the standard system directories
3701 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
3705 Add directory @var{dir} to the list of directories to be searched
3708 @item -B@var{prefix}
3709 This option specifies where to find the executables, libraries,
3710 include files, and data files of the compiler itself.
3712 The compiler driver program runs one or more of the subprograms
3713 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
3714 @var{prefix} as a prefix for each program it tries to run, both with and
3715 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
3717 For each subprogram to be run, the compiler driver first tries the
3718 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
3719 was not specified, the driver tries two standard prefixes, which are
3720 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
3721 those results in a file name that is found, the unmodified program
3722 name is searched for using the directories specified in your
3723 @env{PATH} environment variable.
3725 @samp{-B} prefixes that effectively specify directory names also apply
3726 to libraries in the linker, because the compiler translates these
3727 options into @samp{-L} options for the linker. They also apply to
3728 includes files in the preprocessor, because the compiler translates these
3729 options into @samp{-isystem} options for the preprocessor. In this case,
3730 the compiler appends @samp{include} to the prefix.
3732 The run-time support file @file{libgcc.a} can also be searched for using
3733 the @samp{-B} prefix, if needed. If it is not found there, the two
3734 standard prefixes above are tried, and that is all. The file is left
3735 out of the link if it is not found by those means.
3737 Another way to specify a prefix much like the @samp{-B} prefix is to use
3738 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
3741 @item -specs=@var{file}
3742 Process @var{file} after the compiler reads in the standard @file{specs}
3743 file, in order to override the defaults that the @file{gcc} driver
3744 program uses when determining what switches to pass to @file{cc1},
3745 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
3746 @samp{-specs=}@var{file} can be specified on the command line, and they
3747 are processed in order, from left to right.
3753 @section Specifying subprocesses and the switches to pass to them
3755 @command{gcc} is a driver program. It performs its job by invoking a
3756 sequence of other programs to do the work of compiling, assembling and
3757 linking. GCC interprets its command-line parameters and uses these to
3758 deduce which programs it should invoke, and which command-line options
3759 it ought to place on their command lines. This behaviour is controlled
3760 by @dfn{spec strings}. In most cases there is one spec string for each
3761 program that GCC can invoke, but a few programs have multiple spec
3762 strings to control their behaviour. The spec strings built into GCC can
3763 be overridden by using the @samp{-specs=} command-line switch to specify
3766 @dfn{Spec files} are plaintext files that are used to construct spec
3767 strings. They consist of a sequence of directives separated by blank
3768 lines. The type of directive is determined by the first non-whitespace
3769 character on the line and it can be one of the following:
3772 @item %@var{command}
3773 Issues a @var{command} to the spec file processor. The commands that can
3777 @item %include <@var{file}>
3779 Search for @var{file} and insert its text at the current point in the
3782 @item %include_noerr <@var{file}>
3783 @cindex %include_noerr
3784 Just like @samp{%include}, but do not generate an error message if the include
3785 file cannot be found.
3787 @item %rename @var{old_name} @var{new_name}
3789 Rename the spec string @var{old_name} to @var{new_name}.
3793 @item *[@var{spec_name}]:
3794 This tells the compiler to create, override or delete the named spec
3795 string. All lines after this directive up to the next directive or
3796 blank line are considered to be the text for the spec string. If this
3797 results in an empty string then the spec will be deleted. (Or, if the
3798 spec did not exist, then nothing will happened.) Otherwise, if the spec
3799 does not currently exist a new spec will be created. If the spec does
3800 exist then its contents will be overridden by the text of this
3801 directive, unless the first character of that text is the @samp{+}
3802 character, in which case the text will be appended to the spec.
3804 @item [@var{suffix}]:
3805 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
3806 and up to the next directive or blank line are considered to make up the
3807 spec string for the indicated suffix. When the compiler encounters an
3808 input file with the named suffix, it will processes the spec string in
3809 order to work out how to compile that file. For example:
3816 This says that any input file whose name ends in @samp{.ZZ} should be
3817 passed to the program @samp{z-compile}, which should be invoked with the
3818 command-line switch @samp{-input} and with the result of performing the
3819 @samp{%i} substitution. (See below.)
3821 As an alternative to providing a spec string, the text that follows a
3822 suffix directive can be one of the following:
3825 @item @@@var{language}
3826 This says that the suffix is an alias for a known @var{language}. This is
3827 similar to using the @option{-x} command-line switch to GCC to specify a
3828 language explicitly. For example:
3835 Says that .ZZ files are, in fact, C++ source files.
3838 This causes an error messages saying:
3841 @var{name} compiler not installed on this system.
3845 GCC already has an extensive list of suffixes built into it.
3846 This directive will add an entry to the end of the list of suffixes, but
3847 since the list is searched from the end backwards, it is effectively
3848 possible to override earlier entries using this technique.
3852 GCC has the following spec strings built into it. Spec files can
3853 override these strings or create their own. Note that individual
3854 targets can also add their own spec strings to this list.
3857 asm Options to pass to the assembler
3858 asm_final Options to pass to the assembler post-processor
3859 cpp Options to pass to the C preprocessor
3860 cc1 Options to pass to the C compiler
3861 cc1plus Options to pass to the C++ compiler
3862 endfile Object files to include at the end of the link
3863 link Options to pass to the linker
3864 lib Libraries to include on the command line to the linker
3865 libgcc Decides which GCC support library to pass to the linker
3866 linker Sets the name of the linker
3867 predefines Defines to be passed to the C preprocessor
3868 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
3869 startfile Object files to include at the start of the link
3872 Here is a small example of a spec file:
3878 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
3881 This example renames the spec called @samp{lib} to @samp{old_lib} and
3882 then overrides the previous definition of @samp{lib} with a new one.
3883 The new definition adds in some extra command-line options before
3884 including the text of the old definition.
3886 @dfn{Spec strings} are a list of command-line options to be passed to their
3887 corresponding program. In addition, the spec strings can contain
3888 @samp{%}-prefixed sequences to substitute variable text or to
3889 conditionally insert text into the command line. Using these constructs
3890 it is possible to generate quite complex command lines.
3892 Here is a table of all defined @samp{%}-sequences for spec
3893 strings. Note that spaces are not generated automatically around the
3894 results of expanding these sequences. Therefore you can concatenate them
3895 together or combine them with constant text in a single argument.
3899 Substitute one @samp{%} into the program name or argument.
3902 Substitute the name of the input file being processed.
3905 Substitute the basename of the input file being processed.
3906 This is the substring up to (and not including) the last period
3907 and not including the directory.
3910 Marks the argument containing or following the @samp{%d} as a
3911 temporary file name, so that that file will be deleted if GCC exits
3912 successfully. Unlike @samp{%g}, this contributes no text to the
3915 @item %g@var{suffix}
3916 Substitute a file name that has suffix @var{suffix} and is chosen
3917 once per compilation, and mark the argument in the same way as
3918 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
3919 name is now chosen in a way that is hard to predict even when previously
3920 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
3921 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
3922 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
3923 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
3924 was simply substituted with a file name chosen once per compilation,
3925 without regard to any appended suffix (which was therefore treated
3926 just like ordinary text), making such attacks more likely to succeed.
3928 @item %u@var{suffix}
3929 Like @samp{%g}, but generates a new temporary file name even if
3930 @samp{%u@var{suffix}} was already seen.
3932 @item %U@var{suffix}
3933 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
3934 new one if there is no such last file name. In the absence of any
3935 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
3936 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
3937 would involve the generation of two distinct file names, one
3938 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
3939 simply substituted with a file name chosen for the previous @samp{%u},
3940 without regard to any appended suffix.
3943 Marks the argument containing or following the @samp{%w} as the
3944 designated output file of this compilation. This puts the argument
3945 into the sequence of arguments that @samp{%o} will substitute later.
3948 Substitutes the names of all the output files, with spaces
3949 automatically placed around them. You should write spaces
3950 around the @samp{%o} as well or the results are undefined.
3951 @samp{%o} is for use in the specs for running the linker.
3952 Input files whose names have no recognized suffix are not compiled
3953 at all, but they are included among the output files, so they will
3957 Substitutes the suffix for object files. Note that this is
3958 handled specially when it immediately follows @samp{%g, %u, or %U},
3959 because of the need for those to form complete file names. The
3960 handling is such that @samp{%O} is treated exactly as if it had already
3961 been substituted, except that @samp{%g, %u, and %U} do not currently
3962 support additional @var{suffix} characters following @samp{%O} as they would
3963 following, for example, @samp{.o}.
3966 Substitutes the standard macro predefinitions for the
3967 current target machine. Use this when running @code{cpp}.
3970 Like @samp{%p}, but puts @samp{__} before and after the name of each
3971 predefined macro, except for macros that start with @samp{__} or with
3972 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
3976 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
3979 Current argument is the name of a library or startup file of some sort.
3980 Search for that file in a standard list of directories and substitute
3981 the full name found.
3984 Print @var{str} as an error message. @var{str} is terminated by a newline.
3985 Use this when inconsistent options are detected.
3988 Output @samp{-} if the input for the current command is coming from a pipe.
3991 Substitute the contents of spec string @var{name} at this point.
3994 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
3996 @item %x@{@var{option}@}
3997 Accumulate an option for @samp{%X}.
4000 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
4004 Output the accumulated assembler options specified by @samp{-Wa}.
4007 Output the accumulated preprocessor options specified by @samp{-Wp}.
4010 Substitute the major version number of GCC.
4011 (For version 2.9.5, this is 2.)
4014 Substitute the minor version number of GCC.
4015 (For version 2.9.5, this is 9.)
4018 Process the @code{asm} spec. This is used to compute the
4019 switches to be passed to the assembler.
4022 Process the @code{asm_final} spec. This is a spec string for
4023 passing switches to an assembler post-processor, if such a program is
4027 Process the @code{link} spec. This is the spec for computing the
4028 command line passed to the linker. Typically it will make use of the
4029 @samp{%L %G %S %D and %E} sequences.
4032 Dump out a @samp{-L} option for each directory that GCC believes might
4033 contain startup files. If the target supports multilibs then the
4034 current multilib directory will be prepended to each of these paths.
4037 Process the @code{lib} spec. This is a spec string for deciding which
4038 libraries should be included on the command line to the linker.
4041 Process the @code{libgcc} spec. This is a spec string for deciding
4042 which GCC support library should be included on the command line to the linker.
4045 Process the @code{startfile} spec. This is a spec for deciding which
4046 object files should be the first ones passed to the linker. Typically
4047 this might be a file named @file{crt0.o}.
4050 Process the @code{endfile} spec. This is a spec string that specifies
4051 the last object files that will be passed to the linker.
4054 Process the @code{cpp} spec. This is used to construct the arguments
4055 to be passed to the C preprocessor.
4058 Process the @code{signed_char} spec. This is intended to be used
4059 to tell cpp whether a char is signed. It typically has the definition:
4061 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4065 Process the @code{cc1} spec. This is used to construct the options to be
4066 passed to the actual C compiler (@samp{cc1}).
4069 Process the @code{cc1plus} spec. This is used to construct the options to be
4070 passed to the actual C++ compiler (@samp{cc1plus}).
4073 Substitute the variable part of a matched option. See below.
4074 Note that each comma in the substituted string is replaced by
4078 Substitutes the @code{-S} switch, if that switch was given to GCC.
4079 If that switch was not specified, this substitutes nothing. Note that
4080 the leading dash is omitted when specifying this option, and it is
4081 automatically inserted if the substitution is performed. Thus the spec
4082 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
4083 and would output the command line option @samp{-foo}.
4085 @item %W@{@code{S}@}
4086 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4089 @item %@{@code{S}*@}
4090 Substitutes all the switches specified to GCC whose names start
4091 with @code{-S}, but which also take an argument. This is used for
4092 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
4093 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4094 text, including the space. Thus two arguments would be generated.
4096 @item %@{^@code{S}*@}
4097 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4098 argument. Thus %@{^o*@} would only generate one argument, not two.
4100 @item %@{<@code{S}@}
4101 Remove all occurrences of @code{-S} from the command line. Note - this
4102 command is position dependent. @samp{%} commands in the spec string
4103 before this option will see @code{-S}, @samp{%} commands in the spec
4104 string after this option will not.
4106 @item %@{@code{S}*:@code{X}@}
4107 Substitutes @code{X} if one or more switches whose names start with
4108 @code{-S} are specified to GCC. Note that the tail part of the
4109 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
4110 for each occurrence of @samp{%*} within @code{X}.
4112 @item %@{@code{S}:@code{X}@}
4113 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
4115 @item %@{!@code{S}:@code{X}@}
4116 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
4118 @item %@{|@code{S}:@code{X}@}
4119 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4121 @item %@{|!@code{S}:@code{X}@}
4122 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4124 @item %@{.@code{S}:@code{X}@}
4125 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4127 @item %@{!.@code{S}:@code{X}@}
4128 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4130 @item %@{@code{S}|@code{P}:@code{X}@}
4131 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
4132 combined with @samp{!} and @samp{.} sequences as well, although they
4133 have a stronger binding than the @samp{|}. For example a spec string
4137 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4140 will output the following command-line options from the following input
4141 command-line options:
4146 -d fred.c -foo -baz -boggle
4147 -d jim.d -bar -baz -boggle
4152 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4153 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4154 or spaces, or even newlines. They are processed as usual, as described
4157 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
4158 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
4159 -W} switch is found later in the command line, the earlier switch
4160 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4161 letter, which passes all matching options.
4163 The character @samp{|} at the beginning of the predicate text is used to indicate
4164 that a command should be piped to the following command, but only if @samp{-pipe}
4167 It is built into GCC which switches take arguments and which do not.
4168 (You might think it would be useful to generalize this to allow each
4169 compiler's spec to say which switches take arguments. But this cannot
4170 be done in a consistent fashion. GCC cannot even decide which input
4171 files have been specified without knowing which switches take arguments,
4172 and it must know which input files to compile in order to tell which
4175 GCC also knows implicitly that arguments starting in @samp{-l} are to be
4176 treated as compiler output files, and passed to the linker in their
4177 proper position among the other output files.
4179 @c man begin OPTIONS
4181 @node Target Options
4182 @section Specifying Target Machine and Compiler Version
4183 @cindex target options
4184 @cindex cross compiling
4185 @cindex specifying machine version
4186 @cindex specifying compiler version and target machine
4187 @cindex compiler version, specifying
4188 @cindex target machine, specifying
4190 By default, GCC compiles code for the same type of machine that you
4191 are using. However, it can also be installed as a cross-compiler, to
4192 compile for some other type of machine. In fact, several different
4193 configurations of GCC, for different target machines, can be
4194 installed side by side. Then you specify which one to use with the
4197 In addition, older and newer versions of GCC can be installed side
4198 by side. One of them (probably the newest) will be the default, but
4199 you may sometimes wish to use another.
4202 @item -b @var{machine}
4203 The argument @var{machine} specifies the target machine for compilation.
4204 This is useful when you have installed GCC as a cross-compiler.
4206 The value to use for @var{machine} is the same as was specified as the
4207 machine type when configuring GCC as a cross-compiler. For
4208 example, if a cross-compiler was configured with @samp{configure
4209 i386v}, meaning to compile for an 80386 running System V, then you
4210 would specify @samp{-b i386v} to run that cross compiler.
4212 When you do not specify @samp{-b}, it normally means to compile for
4213 the same type of machine that you are using.
4215 @item -V @var{version}
4216 The argument @var{version} specifies which version of GCC to run.
4217 This is useful when multiple versions are installed. For example,
4218 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
4220 The default version, when you do not specify @samp{-V}, is the last
4221 version of GCC that you installed.
4224 The @samp{-b} and @samp{-V} options actually work by controlling part of
4225 the file name used for the executable files and libraries used for
4226 compilation. A given version of GCC, for a given target machine, is
4227 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
4229 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
4230 changing the names of these directories or adding alternate names (or
4231 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
4232 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
4233 80386} becomes an alias for @samp{-b i386v}.
4235 In one respect, the @samp{-b} or @samp{-V} do not completely change
4236 to a different compiler: the top-level driver program @command{gcc}
4237 that you originally invoked continues to run and invoke the other
4238 executables (preprocessor, compiler per se, assembler and linker)
4239 that do the real work. However, since no real work is done in the
4240 driver program, it usually does not matter that the driver program
4241 in use is not the one for the specified target. It is common for the
4242 interface to the other executables to change incompatibly between
4243 compiler versions, so unless the version specified is very close to that
4244 of the driver (for example, @samp{-V 3.0} with a driver program from GCC
4245 version 3.0.1), use of @samp{-V} may not work; for example, using
4246 @samp{-V 2.95.2} will not work with a driver program from GCC 3.0.
4248 The only way that the driver program depends on the target machine is
4249 in the parsing and handling of special machine-specific options.
4250 However, this is controlled by a file which is found, along with the
4251 other executables, in the directory for the specified version and
4252 target machine. As a result, a single installed driver program adapts
4253 to any specified target machine, and sufficiently similar compiler
4256 The driver program executable does control one significant thing,
4257 however: the default version and target machine. Therefore, you can
4258 install different instances of the driver program, compiled for
4259 different targets or versions, under different names.
4261 For example, if the driver for version 2.0 is installed as @command{ogcc}
4262 and that for version 2.1 is installed as @command{gcc}, then the command
4263 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
4264 2.0 by default. However, you can choose either version with either
4265 command with the @samp{-V} option.
4267 @node Submodel Options
4268 @section Hardware Models and Configurations
4269 @cindex submodel options
4270 @cindex specifying hardware config
4271 @cindex hardware models and configurations, specifying
4272 @cindex machine dependent options
4274 Earlier we discussed the standard option @samp{-b} which chooses among
4275 different installed compilers for completely different target
4276 machines, such as Vax vs. 68000 vs. 80386.
4278 In addition, each of these target machine types can have its own
4279 special options, starting with @samp{-m}, to choose among various
4280 hardware models or configurations---for example, 68010 vs 68020,
4281 floating coprocessor or none. A single installed version of the
4282 compiler can compile for any model or configuration, according to the
4285 Some configurations of the compiler also support additional special
4286 options, usually for compatibility with other compilers on the same
4290 These options are defined by the macro @code{TARGET_SWITCHES} in the
4291 machine description. The default for the options is also defined by
4292 that macro, which enables you to change the defaults.
4308 * RS/6000 and PowerPC Options::
4313 * Intel 960 Options::
4314 * DEC Alpha Options::
4318 * System V Options::
4319 * TMS320C3x/C4x Options::
4328 @node M680x0 Options
4329 @subsection M680x0 Options
4330 @cindex M680x0 options
4332 These are the @samp{-m} options defined for the 68000 series. The default
4333 values for these options depends on which style of 68000 was selected when
4334 the compiler was configured; the defaults for the most common choices are
4340 Generate output for a 68000. This is the default
4341 when the compiler is configured for 68000-based systems.
4343 Use this option for microcontrollers with a 68000 or EC000 core,
4344 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
4348 Generate output for a 68020. This is the default
4349 when the compiler is configured for 68020-based systems.
4352 Generate output containing 68881 instructions for floating point.
4353 This is the default for most 68020 systems unless @samp{-nfp} was
4354 specified when the compiler was configured.
4357 Generate output for a 68030. This is the default when the compiler is
4358 configured for 68030-based systems.
4361 Generate output for a 68040. This is the default when the compiler is
4362 configured for 68040-based systems.
4364 This option inhibits the use of 68881/68882 instructions that have to be
4365 emulated by software on the 68040. Use this option if your 68040 does not
4366 have code to emulate those instructions.
4369 Generate output for a 68060. This is the default when the compiler is
4370 configured for 68060-based systems.
4372 This option inhibits the use of 68020 and 68881/68882 instructions that
4373 have to be emulated by software on the 68060. Use this option if your 68060
4374 does not have code to emulate those instructions.
4377 Generate output for a CPU32. This is the default
4378 when the compiler is configured for CPU32-based systems.
4380 Use this option for microcontrollers with a
4381 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
4382 68336, 68340, 68341, 68349 and 68360.
4385 Generate output for a 520X "coldfire" family cpu. This is the default
4386 when the compiler is configured for 520X-based systems.
4388 Use this option for microcontroller with a 5200 core, including
4389 the MCF5202, MCF5203, MCF5204 and MCF5202.
4393 Generate output for a 68040, without using any of the new instructions.
4394 This results in code which can run relatively efficiently on either a
4395 68020/68881 or a 68030 or a 68040. The generated code does use the
4396 68881 instructions that are emulated on the 68040.
4399 Generate output for a 68060, without using any of the new instructions.
4400 This results in code which can run relatively efficiently on either a
4401 68020/68881 or a 68030 or a 68040. The generated code does use the
4402 68881 instructions that are emulated on the 68060.
4405 Generate output containing Sun FPA instructions for floating point.
4408 Generate output containing library calls for floating point.
4409 @strong{Warning:} the requisite libraries are not available for all m68k
4410 targets. Normally the facilities of the machine's usual C compiler are
4411 used, but this can't be done directly in cross-compilation. You must
4412 make your own arrangements to provide suitable library functions for
4413 cross-compilation. The embedded targets @samp{m68k-*-aout} and
4414 @samp{m68k-*-coff} do provide software floating point support.
4417 Consider type @code{int} to be 16 bits wide, like @code{short int}.
4420 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
4421 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
4424 Do use the bit-field instructions. The @samp{-m68020} option implies
4425 @samp{-mbitfield}. This is the default if you use a configuration
4426 designed for a 68020.
4429 Use a different function-calling convention, in which functions
4430 that take a fixed number of arguments return with the @code{rtd}
4431 instruction, which pops their arguments while returning. This
4432 saves one instruction in the caller since there is no need to pop
4433 the arguments there.
4435 This calling convention is incompatible with the one normally
4436 used on Unix, so you cannot use it if you need to call libraries
4437 compiled with the Unix compiler.
4439 Also, you must provide function prototypes for all functions that
4440 take variable numbers of arguments (including @code{printf});
4441 otherwise incorrect code will be generated for calls to those
4444 In addition, seriously incorrect code will result if you call a
4445 function with too many arguments. (Normally, extra arguments are
4446 harmlessly ignored.)
4448 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
4449 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
4452 @itemx -mno-align-int
4453 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
4454 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
4455 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
4456 Aligning variables on 32-bit boundaries produces code that runs somewhat
4457 faster on processors with 32-bit busses at the expense of more memory.
4459 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
4460 align structures containing the above types differently than
4461 most published application binary interface specifications for the m68k.
4464 Use the pc-relative addressing mode of the 68000 directly, instead of
4465 using a global offset table. At present, this option implies -fpic,
4466 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
4467 not presently supported with -mpcrel, though this could be supported for
4468 68020 and higher processors.
4470 @item -mno-strict-align
4471 @itemx -mstrict-align
4472 @kindex -mstrict-align
4473 Do not (do) assume that unaligned memory references will be handled by
4478 @node M68hc1x Options
4479 @subsection M68hc1x Options
4480 @cindex M68hc1x options
4482 These are the @samp{-m} options defined for the 68hc11 and 68hc12
4483 microcontrollers. The default values for these options depends on
4484 which style of microcontroller was selected when the compiler was configured;
4485 the defaults for the most common choices are given below.
4490 Generate output for a 68HC11. This is the default
4491 when the compiler is configured for 68HC11-based systems.
4495 Generate output for a 68HC12. This is the default
4496 when the compiler is configured for 68HC12-based systems.
4499 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
4503 Consider type @code{int} to be 16 bits wide, like @code{short int}.
4505 @item -msoft-reg-count=@var{count}
4506 Specify the number of pseudo-soft registers which are used for the
4507 code generation. The maximum number is 32. Using more pseudo-soft
4508 register may or may not result in better code depending on the program.
4509 The default is 4 for 68HC11 and 2 for 68HC12.
4514 @subsection VAX Options
4517 These @samp{-m} options are defined for the Vax:
4521 Do not output certain jump instructions (@code{aobleq} and so on)
4522 that the Unix assembler for the Vax cannot handle across long
4526 Do output those jump instructions, on the assumption that you
4527 will assemble with the GNU assembler.
4530 Output code for g-format floating point numbers instead of d-format.
4534 @subsection SPARC Options
4535 @cindex SPARC options
4537 These @samp{-m} switches are supported on the SPARC:
4542 Specify @samp{-mapp-regs} to generate output using the global registers
4543 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
4546 To be fully SVR4 ABI compliant at the cost of some performance loss,
4547 specify @samp{-mno-app-regs}. You should compile libraries and system
4548 software with this option.
4552 Generate output containing floating point instructions. This is the
4557 Generate output containing library calls for floating point.
4558 @strong{Warning:} the requisite libraries are not available for all SPARC
4559 targets. Normally the facilities of the machine's usual C compiler are
4560 used, but this cannot be done directly in cross-compilation. You must make
4561 your own arrangements to provide suitable library functions for
4562 cross-compilation. The embedded targets @samp{sparc-*-aout} and
4563 @samp{sparclite-*-*} do provide software floating point support.
4565 @samp{-msoft-float} changes the calling convention in the output file;
4566 therefore, it is only useful if you compile @emph{all} of a program with
4567 this option. In particular, you need to compile @file{libgcc.a}, the
4568 library that comes with GCC, with @samp{-msoft-float} in order for
4571 @item -mhard-quad-float
4572 Generate output containing quad-word (long double) floating point
4575 @item -msoft-quad-float
4576 Generate output containing library calls for quad-word (long double)
4577 floating point instructions. The functions called are those specified
4578 in the SPARC ABI. This is the default.
4580 As of this writing, there are no sparc implementations that have hardware
4581 support for the quad-word floating point instructions. They all invoke
4582 a trap handler for one of these instructions, and then the trap handler
4583 emulates the effect of the instruction. Because of the trap handler overhead,
4584 this is much slower than calling the ABI library routines. Thus the
4585 @samp{-msoft-quad-float} option is the default.
4589 With @samp{-mepilogue} (the default), the compiler always emits code for
4590 function exit at the end of each function. Any function exit in
4591 the middle of the function (such as a return statement in C) will
4592 generate a jump to the exit code at the end of the function.
4594 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
4595 at every function exit.
4599 With @samp{-mflat}, the compiler does not generate save/restore instructions
4600 and will use a "flat" or single register window calling convention.
4601 This model uses %i7 as the frame pointer and is compatible with the normal
4602 register window model. Code from either may be intermixed.
4603 The local registers and the input registers (0-5) are still treated as
4604 "call saved" registers and will be saved on the stack as necessary.
4606 With @samp{-mno-flat} (the default), the compiler emits save/restore
4607 instructions (except for leaf functions) and is the normal mode of operation.
4609 @item -mno-unaligned-doubles
4610 @itemx -munaligned-doubles
4611 Assume that doubles have 8 byte alignment. This is the default.
4613 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
4614 alignment only if they are contained in another type, or if they have an
4615 absolute address. Otherwise, it assumes they have 4 byte alignment.
4616 Specifying this option avoids some rare compatibility problems with code
4617 generated by other compilers. It is not the default because it results
4618 in a performance loss, especially for floating point code.
4620 @item -mno-faster-structs
4621 @itemx -mfaster-structs
4622 With @samp{-mfaster-structs}, the compiler assumes that structures
4623 should have 8 byte alignment. This enables the use of pairs of
4624 @code{ldd} and @code{std} instructions for copies in structure
4625 assignment, in place of twice as many @code{ld} and @code{st} pairs.
4626 However, the use of this changed alignment directly violates the Sparc
4627 ABI. Thus, it's intended only for use on targets where the developer
4628 acknowledges that their resulting code will not be directly in line with
4629 the rules of the ABI.
4633 These two options select variations on the SPARC architecture.
4635 By default (unless specifically configured for the Fujitsu SPARClite),
4636 GCC generates code for the v7 variant of the SPARC architecture.
4638 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
4639 code is that the compiler emits the integer multiply and integer
4640 divide instructions which exist in SPARC v8 but not in SPARC v7.
4642 @samp{-msparclite} will give you SPARClite code. This adds the integer
4643 multiply, integer divide step and scan (@code{ffs}) instructions which
4644 exist in SPARClite but not in SPARC v7.
4646 These options are deprecated and will be deleted in a future GCC release.
4647 They have been replaced with @samp{-mcpu=xxx}.
4651 These two options select the processor for which the code is optimised.
4653 With @samp{-mcypress} (the default), the compiler optimizes code for the
4654 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
4655 This is also appropriate for the older SparcStation 1, 2, IPX etc.
4657 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
4658 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
4659 of the full SPARC v8 instruction set.
4661 These options are deprecated and will be deleted in a future GCC release.
4662 They have been replaced with @samp{-mcpu=xxx}.
4664 @item -mcpu=@var{cpu_type}
4665 Set the instruction set, register set, and instruction scheduling parameters
4666 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
4667 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
4668 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
4669 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
4671 Default instruction scheduling parameters are used for values that select
4672 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
4673 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
4675 Here is a list of each supported architecture and their supported
4680 v8: supersparc, hypersparc
4681 sparclite: f930, f934, sparclite86x
4686 @item -mtune=@var{cpu_type}
4687 Set the instruction scheduling parameters for machine type
4688 @var{cpu_type}, but do not set the instruction set or register set that the
4689 option @samp{-mcpu=}@var{cpu_type} would.
4691 The same values for @samp{-mcpu=}@var{cpu_type} are used for
4692 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
4693 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
4694 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
4695 @samp{tsc701}, @samp{ultrasparc}.
4699 These @samp{-m} switches are supported in addition to the above
4700 on the SPARCLET processor.
4703 @item -mlittle-endian
4704 Generate code for a processor running in little-endian mode.
4707 Treat register @code{%g0} as a normal register.
4708 GCC will continue to clobber it as necessary but will not assume
4709 it always reads as 0.
4711 @item -mbroken-saverestore
4712 Generate code that does not use non-trivial forms of the @code{save} and
4713 @code{restore} instructions. Early versions of the SPARCLET processor do
4714 not correctly handle @code{save} and @code{restore} instructions used with
4715 arguments. They correctly handle them used without arguments. A @code{save}
4716 instruction used without arguments increments the current window pointer
4717 but does not allocate a new stack frame. It is assumed that the window
4718 overflow trap handler will properly handle this case as will interrupt
4722 These @samp{-m} switches are supported in addition to the above
4723 on SPARC V9 processors in 64 bit environments.
4726 @item -mlittle-endian
4727 Generate code for a processor running in little-endian mode.
4731 Generate code for a 32 bit or 64 bit environment.
4732 The 32 bit environment sets int, long and pointer to 32 bits.
4733 The 64 bit environment sets int to 32 bits and long and pointer
4736 @item -mcmodel=medlow
4737 Generate code for the Medium/Low code model: the program must be linked
4738 in the low 32 bits of the address space. Pointers are 64 bits.
4739 Programs can be statically or dynamically linked.
4741 @item -mcmodel=medmid
4742 Generate code for the Medium/Middle code model: the program must be linked
4743 in the low 44 bits of the address space, the text segment must be less than
4744 2G bytes, and data segment must be within 2G of the text segment.
4745 Pointers are 64 bits.
4747 @item -mcmodel=medany
4748 Generate code for the Medium/Anywhere code model: the program may be linked
4749 anywhere in the address space, the text segment must be less than
4750 2G bytes, and data segment must be within 2G of the text segment.
4751 Pointers are 64 bits.
4753 @item -mcmodel=embmedany
4754 Generate code for the Medium/Anywhere code model for embedded systems:
4755 assume a 32 bit text and a 32 bit data segment, both starting anywhere
4756 (determined at link time). Register %g4 points to the base of the
4757 data segment. Pointers still 64 bits.
4758 Programs are statically linked, PIC is not supported.
4761 @itemx -mno-stack-bias
4762 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
4763 frame pointer if present, are offset by -2047 which must be added back
4764 when making stack frame references.
4765 Otherwise, assume no such offset is present.
4768 @node Convex Options
4769 @subsection Convex Options
4770 @cindex Convex options
4772 These @samp{-m} options are defined for Convex:
4776 Generate output for C1. The code will run on any Convex machine.
4777 The preprocessor symbol @code{__convex__c1__} is defined.
4780 Generate output for C2. Uses instructions not available on C1.
4781 Scheduling and other optimizations are chosen for max performance on C2.
4782 The preprocessor symbol @code{__convex_c2__} is defined.
4785 Generate output for C32xx. Uses instructions not available on C1.
4786 Scheduling and other optimizations are chosen for max performance on C32.
4787 The preprocessor symbol @code{__convex_c32__} is defined.
4790 Generate output for C34xx. Uses instructions not available on C1.
4791 Scheduling and other optimizations are chosen for max performance on C34.
4792 The preprocessor symbol @code{__convex_c34__} is defined.
4795 Generate output for C38xx. Uses instructions not available on C1.
4796 Scheduling and other optimizations are chosen for max performance on C38.
4797 The preprocessor symbol @code{__convex_c38__} is defined.
4800 Generate code which puts an argument count in the word preceding each
4801 argument list. This is compatible with regular CC, and a few programs
4802 may need the argument count word. GDB and other source-level debuggers
4803 do not need it; this info is in the symbol table.
4806 Omit the argument count word. This is the default.
4808 @item -mvolatile-cache
4809 Allow volatile references to be cached. This is the default.
4811 @item -mvolatile-nocache
4812 Volatile references bypass the data cache, going all the way to memory.
4813 This is only needed for multi-processor code that does not use standard
4814 synchronization instructions. Making non-volatile references to volatile
4815 locations will not necessarily work.
4818 Type long is 32 bits, the same as type int. This is the default.
4821 Type long is 64 bits, the same as type long long. This option is useless,
4822 because no library support exists for it.
4825 @node AMD29K Options
4826 @subsection AMD29K Options
4827 @cindex AMD29K options
4829 These @samp{-m} options are defined for the AMD Am29000:
4834 @cindex DW bit (29k)
4835 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
4836 halfword operations are directly supported by the hardware. This is the
4841 Generate code that assumes the @code{DW} bit is not set.
4845 @cindex byte writes (29k)
4846 Generate code that assumes the system supports byte and halfword write
4847 operations. This is the default.
4851 Generate code that assumes the systems does not support byte and
4852 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
4856 @cindex memory model (29k)
4857 Use a small memory model that assumes that all function addresses are
4858 either within a single 256 KB segment or at an absolute address of less
4859 than 256k. This allows the @code{call} instruction to be used instead
4860 of a @code{const}, @code{consth}, @code{calli} sequence.
4864 Use the normal memory model: Generate @code{call} instructions only when
4865 calling functions in the same file and @code{calli} instructions
4866 otherwise. This works if each file occupies less than 256 KB but allows
4867 the entire executable to be larger than 256 KB. This is the default.
4870 Always use @code{calli} instructions. Specify this option if you expect
4871 a single file to compile into more than 256 KB of code.
4875 @cindex processor selection (29k)
4876 Generate code for the Am29050.
4880 Generate code for the Am29000. This is the default.
4882 @item -mkernel-registers
4883 @kindex -mkernel-registers
4884 @cindex kernel and user registers (29k)
4885 Generate references to registers @code{gr64-gr95} instead of to
4886 registers @code{gr96-gr127}. This option can be used when compiling
4887 kernel code that wants a set of global registers disjoint from that used
4890 Note that when this option is used, register names in @samp{-f} flags
4891 must use the normal, user-mode, names.
4893 @item -muser-registers
4894 @kindex -muser-registers
4895 Use the normal set of global registers, @code{gr96-gr127}. This is the
4899 @itemx -mno-stack-check
4900 @kindex -mstack-check
4901 @cindex stack checks (29k)
4902 Insert (or do not insert) a call to @code{__msp_check} after each stack
4903 adjustment. This is often used for kernel code.
4906 @itemx -mno-storem-bug
4907 @kindex -mstorem-bug
4908 @cindex storem bug (29k)
4909 @samp{-mstorem-bug} handles 29k processors which cannot handle the
4910 separation of a mtsrim insn and a storem instruction (most 29000 chips
4911 to date, but not the 29050).
4913 @item -mno-reuse-arg-regs
4914 @itemx -mreuse-arg-regs
4915 @kindex -mreuse-arg-regs
4916 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
4917 registers for copying out arguments. This helps detect calling a function
4918 with fewer arguments than it was declared with.
4920 @item -mno-impure-text
4921 @itemx -mimpure-text
4922 @kindex -mimpure-text
4923 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
4924 not pass @samp{-assert pure-text} to the linker when linking a shared object.
4927 @kindex -msoft-float
4928 Generate output containing library calls for floating point.
4929 @strong{Warning:} the requisite libraries are not part of GCC.
4930 Normally the facilities of the machine's usual C compiler are used, but
4931 this can't be done directly in cross-compilation. You must make your
4932 own arrangements to provide suitable library functions for
4937 Do not generate multm or multmu instructions. This is useful for some embedded
4938 systems which do not have trap handlers for these instructions.
4942 @subsection ARM Options
4945 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
4950 @kindex -mapcs-frame
4951 Generate a stack frame that is compliant with the ARM Procedure Call
4952 Standard for all functions, even if this is not strictly necessary for
4953 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
4954 with this option will cause the stack frames not to be generated for
4955 leaf functions. The default is @samp{-mno-apcs-frame}.
4959 This is a synonym for @samp{-mapcs-frame}.
4963 Generate code for a processor running with a 26-bit program counter,
4964 and conforming to the function calling standards for the APCS 26-bit
4965 option. This option replaces the @samp{-m2} and @samp{-m3} options
4966 of previous releases of the compiler.
4970 Generate code for a processor running with a 32-bit program counter,
4971 and conforming to the function calling standards for the APCS 32-bit
4972 option. This option replaces the @samp{-m6} option of previous releases
4975 @item -mapcs-stack-check
4976 @kindex -mapcs-stack-check
4977 @kindex -mno-apcs-stack-check
4978 Generate code to check the amount of stack space available upon entry to
4979 every function (that actually uses some stack space). If there is
4980 insufficient space available then either the function
4981 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
4982 called, depending upon the amount of stack space required. The run time
4983 system is required to provide these functions. The default is
4984 @samp{-mno-apcs-stack-check}, since this produces smaller code.
4987 @kindex -mapcs-float
4988 @kindex -mno-apcs-float
4989 Pass floating point arguments using the float point registers. This is
4990 one of the variants of the APCS. This option is recommended if the
4991 target hardware has a floating point unit or if a lot of floating point
4992 arithmetic is going to be performed by the code. The default is
4993 @samp{-mno-apcs-float}, since integer only code is slightly increased in
4994 size if @samp{-mapcs-float} is used.
4996 @item -mapcs-reentrant
4997 @kindex -mapcs-reentrant
4998 @kindex -mno-apcs-reentrant
4999 Generate reentrant, position independent code. This is the equivalent
5000 to specifying the @samp{-fpic} option. The default is
5001 @samp{-mno-apcs-reentrant}.
5003 @item -mthumb-interwork
5004 @kindex -mthumb-interwork
5005 @kindex -mno-thumb-interwork
5006 Generate code which supports calling between the ARM and THUMB
5007 instruction sets. Without this option the two instruction sets cannot
5008 be reliably used inside one program. The default is
5009 @samp{-mno-thumb-interwork}, since slightly larger code is generated
5010 when @samp{-mthumb-interwork} is specified.
5012 @item -mno-sched-prolog
5013 @kindex -mno-sched-prolog
5014 @kindex -msched-prolog
5015 Prevent the reordering of instructions in the function prolog, or the
5016 merging of those instruction with the instructions in the function's
5017 body. This means that all functions will start with a recognizable set
5018 of instructions (or in fact one of a choice from a small set of
5019 different function prologues), and this information can be used to
5020 locate the start if functions inside an executable piece of code. The
5021 default is @samp{-msched-prolog}.
5024 Generate output containing floating point instructions. This is the
5028 Generate output containing library calls for floating point.
5029 @strong{Warning:} the requisite libraries are not available for all ARM
5030 targets. Normally the facilities of the machine's usual C compiler are
5031 used, but this cannot be done directly in cross-compilation. You must make
5032 your own arrangements to provide suitable library functions for
5035 @samp{-msoft-float} changes the calling convention in the output file;
5036 therefore, it is only useful if you compile @emph{all} of a program with
5037 this option. In particular, you need to compile @file{libgcc.a}, the
5038 library that comes with GCC, with @samp{-msoft-float} in order for
5041 @item -mlittle-endian
5042 Generate code for a processor running in little-endian mode. This is
5043 the default for all standard configurations.
5046 Generate code for a processor running in big-endian mode; the default is
5047 to compile code for a little-endian processor.
5049 @item -mwords-little-endian
5050 This option only applies when generating code for big-endian processors.
5051 Generate code for a little-endian word order but a big-endian byte
5052 order. That is, a byte order of the form @samp{32107654}. Note: this
5053 option should only be used if you require compatibility with code for
5054 big-endian ARM processors generated by versions of the compiler prior to
5057 @item -malignment-traps
5058 @kindex -malignment-traps
5059 Generate code that will not trap if the MMU has alignment traps enabled.
5060 On ARM architectures prior to ARMv4, there were no instructions to
5061 access half-word objects stored in memory. However, when reading from
5062 memory a feature of the ARM architecture allows a word load to be used,
5063 even if the address is unaligned, and the processor core will rotate the
5064 data as it is being loaded. This option tells the compiler that such
5065 misaligned accesses will cause a MMU trap and that it should instead
5066 synthesise the access as a series of byte accesses. The compiler can
5067 still use word accesses to load half-word data if it knows that the
5068 address is aligned to a word boundary.
5070 This option is ignored when compiling for ARM architecture 4 or later,
5071 since these processors have instructions to directly access half-word
5074 @item -mno-alignment-traps
5075 @kindex -mno-alignment-traps
5076 Generate code that assumes that the MMU will not trap unaligned
5077 accesses. This produces better code when the target instruction set
5078 does not have half-word memory operations (implementations prior to
5081 Note that you cannot use this option to access unaligned word objects,
5082 since the processor will only fetch one 32-bit aligned object from
5085 The default setting for most targets is -mno-alignment-traps, since
5086 this produces better code when there are no half-word memory
5087 instructions available.
5089 @item -mshort-load-bytes
5090 @kindex -mshort-load-bytes
5091 This is a deprecated alias for @samp{-malignment-traps}.
5093 @item -mno-short-load-bytes
5094 @kindex -mno-short-load-bytes
5095 This is a deprecated alias for @samp{-mno-alignment-traps}.
5097 @item -mshort-load-words
5098 @kindex -mshort-load-words
5099 This is a deprecated alias for @samp{-mno-alignment-traps}.
5101 @item -mno-short-load-words
5102 @kindex -mno-short-load-words
5103 This is a deprecated alias for @samp{-malignment-traps}.
5107 This option only applies to RISC iX. Emulate the native BSD-mode
5108 compiler. This is the default if @samp{-ansi} is not specified.
5112 This option only applies to RISC iX. Emulate the native X/Open-mode
5115 @item -mno-symrename
5116 @kindex -mno-symrename
5117 This option only applies to RISC iX. Do not run the assembler
5118 post-processor, @samp{symrename}, after code has been assembled.
5119 Normally it is necessary to modify some of the standard symbols in
5120 preparation for linking with the RISC iX C library; this option
5121 suppresses this pass. The post-processor is never run when the
5122 compiler is built for cross-compilation.
5126 This specifies the name of the target ARM processor. GCC uses this name
5127 to determine what kind of instructions it can use when generating
5128 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
5129 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
5130 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
5131 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
5132 arm9, arm920, arm920t, arm9tdmi.
5134 @itemx -mtune=<name>
5136 This option is very similar to the @samp{-mcpu=} option, except that
5137 instead of specifying the actual target processor type, and hence
5138 restricting which instructions can be used, it specifies that GCC should
5139 tune the performance of the code as if the target were of the type
5140 specified in this option, but still choosing the instructions that it
5141 will generate based on the cpu specified by a @samp{-mcpu=} option.
5142 For some arm implementations better performance can be obtained by using
5147 This specifies the name of the target ARM architecture. GCC uses this
5148 name to determine what kind of instructions it can use when generating
5149 assembly code. This option can be used in conjunction with or instead
5150 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
5151 armv3, armv3m, armv4, armv4t, armv5.
5153 @item -mfpe=<number>
5154 @itemx -mfp=<number>
5157 This specifies the version of the floating point emulation available on
5158 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
5159 for @samp{-mfpe=} to support older versions of GCC.
5161 @item -mstructure-size-boundary=<n>
5162 @kindex -mstructure-size-boundary
5163 The size of all structures and unions will be rounded up to a multiple
5164 of the number of bits set by this option. Permissible values are 8 and
5165 32. The default value varies for different toolchains. For the COFF
5166 targeted toolchain the default value is 8. Specifying the larger number
5167 can produce faster, more efficient code, but can also increase the size
5168 of the program. The two values are potentially incompatible. Code
5169 compiled with one value cannot necessarily expect to work with code or
5170 libraries compiled with the other value, if they exchange information
5171 using structures or unions. Programmers are encouraged to use the 32
5172 value as future versions of the toolchain may default to this value.
5174 @item -mabort-on-noreturn
5175 @kindex -mabort-on-noreturn
5176 @kindex -mnoabort-on-noreturn
5177 Generate a call to the function abort at the end of a noreturn function.
5178 It will be executed if the function tries to return.
5181 @itemx -mno-long-calls
5182 Tells the compiler to perform function calls by first loading the
5183 address of the function into a register and then performing a subroutine
5184 call on this register. This switch is needed if the target function
5185 will lie outside of the 64 megabyte addressing range of the offset based
5186 version of subroutine call instruction.
5188 Even if this switch is enabled, not all function calls will be turned
5189 into long calls. The heuristic is that static functions, functions
5190 which have the @samp{short-call} attribute, functions that are inside
5191 the scope of a @samp{#pragma no_long_calls} directive and functions whose
5192 definitions have already been compiled within the current compilation
5193 unit, will not be turned into long calls. The exception to this rule is
5194 that weak function definitions, functions with the @samp{long-call}
5195 attribute or the @samp{section} attribute, and functions that are within
5196 the scope of a @samp{#pragma long_calls} directive, will always be
5197 turned into long calls.
5199 This feature is not enabled by default. Specifying
5200 @samp{--no-long-calls} will restore the default behaviour, as will
5201 placing the function calls within the scope of a @samp{#pragma
5202 long_calls_off} directive. Note these switches have no effect on how
5203 the compiler generates code to handle function calls via function
5206 @item -mnop-fun-dllimport
5207 @kindex -mnop-fun-dllimport
5208 Disable the support for the @emph{dllimport} attribute.
5210 @item -msingle-pic-base
5211 @kindex -msingle-pic-base
5212 Treat the register used for PIC addressing as read-only, rather than
5213 loading it in the prologue for each function. The run-time system is
5214 responsible for initialising this register with an appropriate value
5215 before execution begins.
5217 @item -mpic-register=<reg>
5218 @kindex -mpic-register=
5219 Specify the register to be used for PIC addressing. The default is R10
5220 unless stack-checking is enabled, when R9 is used.
5225 @subsection Thumb Options
5226 @cindex Thumb Options
5230 @item -mthumb-interwork
5231 @kindex -mthumb-interwork
5232 @kindex -mno-thumb-interwork
5233 Generate code which supports calling between the THUMB and ARM
5234 instruction sets. Without this option the two instruction sets cannot
5235 be reliably used inside one program. The default is
5236 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
5240 @kindex -mtpcs-frame
5241 @kindex -mno-tpcs-frame
5242 Generate a stack frame that is compliant with the Thumb Procedure Call
5243 Standard for all non-leaf functions. (A leaf function is one that does
5244 not call any other functions). The default is @samp{-mno-apcs-frame}.
5246 @item -mtpcs-leaf-frame
5247 @kindex -mtpcs-leaf-frame
5248 @kindex -mno-tpcs-leaf-frame
5249 Generate a stack frame that is compliant with the Thumb Procedure Call
5250 Standard for all leaf functions. (A leaf function is one that does
5251 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
5253 @item -mlittle-endian
5254 @kindex -mlittle-endian
5255 Generate code for a processor running in little-endian mode. This is
5256 the default for all standard configurations.
5259 @kindex -mbig-endian
5260 Generate code for a processor running in big-endian mode.
5262 @item -mstructure-size-boundary=<n>
5263 @kindex -mstructure-size-boundary
5264 The size of all structures and unions will be rounded up to a multiple
5265 of the number of bits set by this option. Permissible values are 8 and
5266 32. The default value varies for different toolchains. For the COFF
5267 targeted toolchain the default value is 8. Specifying the larger number
5268 can produced faster, more efficient code, but can also increase the size
5269 of the program. The two values are potentially incompatible. Code
5270 compiled with one value cannot necessarily expect to work with code or
5271 libraries compiled with the other value, if they exchange information
5272 using structures or unions. Programmers are encouraged to use the 32
5273 value as future versions of the toolchain may default to this value.
5275 @item -mnop-fun-dllimport
5276 @kindex -mnop-fun-dllimport
5277 Disable the support for the @emph{dllimport} attribute.
5279 @item -mcallee-super-interworking
5280 @kindex -mcallee-super-interworking
5281 Gives all externally visible functions in the file being compiled an ARM
5282 instruction set header which switches to Thumb mode before executing the
5283 rest of the function. This allows these functions to be called from
5284 non-interworking code.
5286 @item -mcaller-super-interworking
5287 @kindex -mcaller-super-interworking
5288 Allows calls via function pointers (including virtual functions) to
5289 execute correctly regardless of whether the target code has been
5290 compiled for interworking or not. There is a small overhead in the cost
5291 of executing a function pointer if this option is enabled.
5293 @item -msingle-pic-base
5294 @kindex -msingle-pic-base
5295 Treat the register used for PIC addressing as read-only, rather than
5296 loading it in the prologue for each function. The run-time system is
5297 responsible for initialising this register with an appropriate value
5298 before execution begins.
5300 @item -mpic-register=<reg>
5301 @kindex -mpic-register=
5302 Specify the register to be used for PIC addressing. The default is R10.
5306 @node MN10200 Options
5307 @subsection MN10200 Options
5308 @cindex MN10200 options
5309 These @samp{-m} options are defined for Matsushita MN10200 architectures:
5313 Indicate to the linker that it should perform a relaxation optimization pass
5314 to shorten branches, calls and absolute memory addresses. This option only
5315 has an effect when used on the command line for the final link step.
5317 This option makes symbolic debugging impossible.
5320 @node MN10300 Options
5321 @subsection MN10300 Options
5322 @cindex MN10300 options
5323 These @samp{-m} options are defined for Matsushita MN10300 architectures:
5327 Generate code to avoid bugs in the multiply instructions for the MN10300
5328 processors. This is the default.
5331 Do not generate code to avoid bugs in the multiply instructions for the
5335 Generate code which uses features specific to the AM33 processor.
5338 Do not generate code which uses features specific to the AM33 processor. This
5342 Indicate to the linker that it should perform a relaxation optimization pass
5343 to shorten branches, calls and absolute memory addresses. This option only
5344 has an effect when used on the command line for the final link step.
5346 This option makes symbolic debugging impossible.
5350 @node M32R/D Options
5351 @subsection M32R/D Options
5352 @cindex M32R/D options
5354 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
5357 @item -mcode-model=small
5358 Assume all objects live in the lower 16MB of memory (so that their addresses
5359 can be loaded with the @code{ld24} instruction), and assume all subroutines
5360 are reachable with the @code{bl} instruction.
5361 This is the default.
5363 The addressability of a particular object can be set with the
5364 @code{model} attribute.
5366 @item -mcode-model=medium
5367 Assume objects may be anywhere in the 32 bit address space (the compiler
5368 will generate @code{seth/add3} instructions to load their addresses), and
5369 assume all subroutines are reachable with the @code{bl} instruction.
5371 @item -mcode-model=large
5372 Assume objects may be anywhere in the 32 bit address space (the compiler
5373 will generate @code{seth/add3} instructions to load their addresses), and
5374 assume subroutines may not be reachable with the @code{bl} instruction
5375 (the compiler will generate the much slower @code{seth/add3/jl}
5376 instruction sequence).
5379 Disable use of the small data area. Variables will be put into
5380 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
5381 @code{section} attribute has been specified).
5382 This is the default.
5384 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
5385 Objects may be explicitly put in the small data area with the
5386 @code{section} attribute using one of these sections.
5389 Put small global and static data in the small data area, but do not
5390 generate special code to reference them.
5393 Put small global and static data in the small data area, and generate
5394 special instructions to reference them.
5397 @cindex smaller data references
5398 Put global and static objects less than or equal to @var{num} bytes
5399 into the small data or bss sections instead of the normal data or bss
5400 sections. The default value of @var{num} is 8.
5401 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
5402 for this option to have any effect.
5404 All modules should be compiled with the same @samp{-G @var{num}} value.
5405 Compiling with different values of @var{num} may or may not work; if it
5406 doesn't the linker will give an error message - incorrect code will not be
5412 @subsection M88K Options
5413 @cindex M88k options
5415 These @samp{-m} options are defined for Motorola 88k architectures:
5420 Generate code that works well on both the m88100 and the
5425 Generate code that works best for the m88100, but that also
5430 Generate code that works best for the m88110, and may not run
5435 Obsolete option to be removed from the next revision.
5438 @item -midentify-revision
5439 @kindex -midentify-revision
5441 @cindex identifying source, compiler (88k)
5442 Include an @code{ident} directive in the assembler output recording the
5443 source file name, compiler name and version, timestamp, and compilation
5446 @item -mno-underscores
5447 @kindex -mno-underscores
5448 @cindex underscores, avoiding (88k)
5449 In assembler output, emit symbol names without adding an underscore
5450 character at the beginning of each name. The default is to use an
5451 underscore as prefix on each name.
5453 @item -mocs-debug-info
5454 @itemx -mno-ocs-debug-info
5455 @kindex -mocs-debug-info
5456 @kindex -mno-ocs-debug-info
5458 @cindex debugging, 88k OCS
5459 Include (or omit) additional debugging information (about registers used
5460 in each stack frame) as specified in the 88open Object Compatibility
5461 Standard, ``OCS''. This extra information allows debugging of code that
5462 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
5463 Delta 88 SVr3.2 is to include this information; other 88k configurations
5464 omit this information by default.
5466 @item -mocs-frame-position
5467 @kindex -mocs-frame-position
5468 @cindex register positions in frame (88k)
5469 When emitting COFF debugging information for automatic variables and
5470 parameters stored on the stack, use the offset from the canonical frame
5471 address, which is the stack pointer (register 31) on entry to the
5472 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
5473 @samp{-mocs-frame-position}; other 88k configurations have the default
5474 @samp{-mno-ocs-frame-position}.
5476 @item -mno-ocs-frame-position
5477 @kindex -mno-ocs-frame-position
5478 @cindex register positions in frame (88k)
5479 When emitting COFF debugging information for automatic variables and
5480 parameters stored on the stack, use the offset from the frame pointer
5481 register (register 30). When this option is in effect, the frame
5482 pointer is not eliminated when debugging information is selected by the
5485 @item -moptimize-arg-area
5486 @itemx -mno-optimize-arg-area
5487 @kindex -moptimize-arg-area
5488 @kindex -mno-optimize-arg-area
5489 @cindex arguments in frame (88k)
5490 Control how function arguments are stored in stack frames.
5491 @samp{-moptimize-arg-area} saves space by optimizing them, but this
5492 conflicts with the 88open specifications. The opposite alternative,
5493 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
5494 GCC does not optimize the argument area.
5496 @item -mshort-data-@var{num}
5497 @kindex -mshort-data-@var{num}
5498 @cindex smaller data references (88k)
5499 @cindex r0-relative references (88k)
5500 Generate smaller data references by making them relative to @code{r0},
5501 which allows loading a value using a single instruction (rather than the
5502 usual two). You control which data references are affected by
5503 specifying @var{num} with this option. For example, if you specify
5504 @samp{-mshort-data-512}, then the data references affected are those
5505 involving displacements of less than 512 bytes.
5506 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
5509 @item -mserialize-volatile
5510 @kindex -mserialize-volatile
5511 @itemx -mno-serialize-volatile
5512 @kindex -mno-serialize-volatile
5513 @cindex sequential consistency on 88k
5514 Do, or don't, generate code to guarantee sequential consistency
5515 of volatile memory references. By default, consistency is
5518 The order of memory references made by the MC88110 processor does
5519 not always match the order of the instructions requesting those
5520 references. In particular, a load instruction may execute before
5521 a preceding store instruction. Such reordering violates
5522 sequential consistency of volatile memory references, when there
5523 are multiple processors. When consistency must be guaranteed,
5524 GNU C generates special instructions, as needed, to force
5525 execution in the proper order.
5527 The MC88100 processor does not reorder memory references and so
5528 always provides sequential consistency. However, by default, GNU
5529 C generates the special instructions to guarantee consistency
5530 even when you use @samp{-m88100}, so that the code may be run on an
5531 MC88110 processor. If you intend to run your code only on the
5532 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
5534 The extra code generated to guarantee consistency may affect the
5535 performance of your application. If you know that you can safely
5536 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
5542 @cindex assembler syntax, 88k
5544 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
5545 related to System V release 4 (SVr4). This controls the following:
5549 Which variant of the assembler syntax to emit.
5551 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
5552 that is used on System V release 4.
5554 @samp{-msvr4} makes GCC issue additional declaration directives used in
5558 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
5559 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
5560 other m88k configurations.
5562 @item -mversion-03.00
5563 @kindex -mversion-03.00
5564 This option is obsolete, and is ignored.
5565 @c ??? which asm syntax better for GAS? option there too?
5567 @item -mno-check-zero-division
5568 @itemx -mcheck-zero-division
5569 @kindex -mno-check-zero-division
5570 @kindex -mcheck-zero-division
5571 @cindex zero division on 88k
5572 Do, or don't, generate code to guarantee that integer division by
5573 zero will be detected. By default, detection is guaranteed.
5575 Some models of the MC88100 processor fail to trap upon integer
5576 division by zero under certain conditions. By default, when
5577 compiling code that might be run on such a processor, GNU C
5578 generates code that explicitly checks for zero-valued divisors
5579 and traps with exception number 503 when one is detected. Use of
5580 mno-check-zero-division suppresses such checking for code
5581 generated to run on an MC88100 processor.
5583 GNU C assumes that the MC88110 processor correctly detects all
5584 instances of integer division by zero. When @samp{-m88110} is
5585 specified, both @samp{-mcheck-zero-division} and
5586 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
5587 zero-valued divisors are generated.
5589 @item -muse-div-instruction
5590 @kindex -muse-div-instruction
5591 @cindex divide instruction, 88k
5592 Use the div instruction for signed integer division on the
5593 MC88100 processor. By default, the div instruction is not used.
5595 On the MC88100 processor the signed integer division instruction
5596 div) traps to the operating system on a negative operand. The
5597 operating system transparently completes the operation, but at a
5598 large cost in execution time. By default, when compiling code
5599 that might be run on an MC88100 processor, GNU C emulates signed
5600 integer division using the unsigned integer division instruction
5601 divu), thereby avoiding the large penalty of a trap to the
5602 operating system. Such emulation has its own, smaller, execution
5603 cost in both time and space. To the extent that your code's
5604 important signed integer division operations are performed on two
5605 nonnegative operands, it may be desirable to use the div
5606 instruction directly.
5608 On the MC88110 processor the div instruction (also known as the
5609 divs instruction) processes negative operands without trapping to
5610 the operating system. When @samp{-m88110} is specified,
5611 @samp{-muse-div-instruction} is ignored, and the div instruction is used
5612 for signed integer division.
5614 Note that the result of dividing INT_MIN by -1 is undefined. In
5615 particular, the behavior of such a division with and without
5616 @samp{-muse-div-instruction} may differ.
5618 @item -mtrap-large-shift
5619 @itemx -mhandle-large-shift
5620 @kindex -mtrap-large-shift
5621 @kindex -mhandle-large-shift
5622 @cindex bit shift overflow (88k)
5623 @cindex large bit shifts (88k)
5624 Include code to detect bit-shifts of more than 31 bits; respectively,
5625 trap such shifts or emit code to handle them properly. By default GCC
5626 makes no special provision for large bit shifts.
5628 @item -mwarn-passed-structs
5629 @kindex -mwarn-passed-structs
5630 @cindex structure passing (88k)
5631 Warn when a function passes a struct as an argument or result.
5632 Structure-passing conventions have changed during the evolution of the C
5633 language, and are often the source of portability problems. By default,
5634 GCC issues no such warning.
5637 @node RS/6000 and PowerPC Options
5638 @subsection IBM RS/6000 and PowerPC Options
5639 @cindex RS/6000 and PowerPC Options
5640 @cindex IBM RS/6000 and PowerPC Options
5642 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
5650 @itemx -mpowerpc-gpopt
5651 @itemx -mno-powerpc-gpopt
5652 @itemx -mpowerpc-gfxopt
5653 @itemx -mno-powerpc-gfxopt
5655 @itemx -mno-powerpc64
5659 @kindex -mpowerpc-gpopt
5660 @kindex -mpowerpc-gfxopt
5662 GCC supports two related instruction set architectures for the
5663 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
5664 instructions supported by the @samp{rios} chip set used in the original
5665 RS/6000 systems and the @dfn{PowerPC} instruction set is the
5666 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
5667 the IBM 4xx microprocessors.
5669 Neither architecture is a subset of the other. However there is a
5670 large common subset of instructions supported by both. An MQ
5671 register is included in processors supporting the POWER architecture.
5673 You use these options to specify which instructions are available on the
5674 processor you are using. The default value of these options is
5675 determined when configuring GCC. Specifying the
5676 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
5677 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
5678 rather than the options listed above.
5680 The @samp{-mpower} option allows GCC to generate instructions that
5681 are found only in the POWER architecture and to use the MQ register.
5682 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
5683 to generate instructions that are present in the POWER2 architecture but
5684 not the original POWER architecture.
5686 The @samp{-mpowerpc} option allows GCC to generate instructions that
5687 are found only in the 32-bit subset of the PowerPC architecture.
5688 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
5689 GCC to use the optional PowerPC architecture instructions in the
5690 General Purpose group, including floating-point square root. Specifying
5691 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
5692 use the optional PowerPC architecture instructions in the Graphics
5693 group, including floating-point select.
5695 The @samp{-mpowerpc64} option allows GCC to generate the additional
5696 64-bit instructions that are found in the full PowerPC64 architecture
5697 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
5698 @samp{-mno-powerpc64}.
5700 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
5701 will use only the instructions in the common subset of both
5702 architectures plus some special AIX common-mode calls, and will not use
5703 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
5704 permits GCC to use any instruction from either architecture and to
5705 allow use of the MQ register; specify this for the Motorola MPC601.
5707 @item -mnew-mnemonics
5708 @itemx -mold-mnemonics
5709 @kindex -mnew-mnemonics
5710 @kindex -mold-mnemonics
5711 Select which mnemonics to use in the generated assembler code.
5712 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
5713 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
5714 requests the assembler mnemonics defined for the POWER architecture.
5715 Instructions defined in only one architecture have only one mnemonic;
5716 GCC uses that mnemonic irrespective of which of these options is
5719 GCC defaults to the mnemonics appropriate for the architecture in
5720 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
5721 value of these option. Unless you are building a cross-compiler, you
5722 should normally not specify either @samp{-mnew-mnemonics} or
5723 @samp{-mold-mnemonics}, but should instead accept the default.
5725 @item -mcpu=@var{cpu_type}
5727 Set architecture type, register usage, choice of mnemonics, and
5728 instruction scheduling parameters for machine type @var{cpu_type}.
5729 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
5730 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
5731 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
5732 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
5733 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
5734 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
5735 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
5736 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
5737 and 64-bit PowerPC architecture machine types, with an appropriate,
5738 generic processor model assumed for scheduling purposes.@refill
5740 Specifying any of the following options:
5741 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
5742 @samp{-mcpu=power}, or @samp{-mcpu=power2}
5743 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
5744 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
5745 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
5746 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
5747 @samp{-mcpu=740}, and @samp{-mcpu=750}
5748 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5749 Exactly similarly, all of @samp{-mcpu=403},
5750 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
5751 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5752 @samp{-mcpu=common} disables both the
5753 @samp{-mpower} and @samp{-mpowerpc} options.@refill
5755 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
5756 that code will operate on all members of the RS/6000 POWER and PowerPC
5757 families. In that case, GCC will use only the instructions in the
5758 common subset of both architectures plus some special AIX common-mode
5759 calls, and will not use the MQ register. GCC assumes a generic
5760 processor model for scheduling purposes.
5762 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
5763 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
5764 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
5765 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
5766 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
5767 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
5768 the @samp{new-mnemonics} option.@refill
5770 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
5771 enables the @samp{-msoft-float} option.
5773 @item -mtune=@var{cpu_type}
5774 Set the instruction scheduling parameters for machine type
5775 @var{cpu_type}, but do not set the architecture type, register usage,
5776 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
5777 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
5778 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
5779 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
5780 instruction scheduling parameters.
5783 @itemx -mno-fp-in-toc
5784 @itemx -mno-sum-in-toc
5785 @itemx -mminimal-toc
5786 @kindex -mminimal-toc
5787 Modify generation of the TOC (Table Of Contents), which is created for
5788 every executable file. The @samp{-mfull-toc} option is selected by
5789 default. In that case, GCC will allocate at least one TOC entry for
5790 each unique non-automatic variable reference in your program. GCC
5791 will also place floating-point constants in the TOC. However, only
5792 16,384 entries are available in the TOC.
5794 If you receive a linker error message that saying you have overflowed
5795 the available TOC space, you can reduce the amount of TOC space used
5796 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
5797 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
5798 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
5799 generate code to calculate the sum of an address and a constant at
5800 run-time instead of putting that sum into the TOC. You may specify one
5801 or both of these options. Each causes GCC to produce very slightly
5802 slower and larger code at the expense of conserving TOC space.
5804 If you still run out of space in the TOC even when you specify both of
5805 these options, specify @samp{-mminimal-toc} instead. This option causes
5806 GCC to make only one TOC entry for every file. When you specify this
5807 option, GCC will produce code that is slower and larger but which
5808 uses extremely little TOC space. You may wish to use this option
5809 only on files that contain less frequently executed code. @refill
5815 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
5816 @code{long} type, and the infrastructure needed to support them.
5817 Specifying @samp{-maix64} implies @samp{-mpowerpc64} and
5818 @samp{-mpowerpc}, while @samp{-maix32} disables the 64-bit ABI and
5819 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-maix32}.
5824 On AIX, pass floating-point arguments to prototyped functions beyond the
5825 register save area (RSA) on the stack in addition to argument FPRs. The
5826 AIX calling convention was extended but not initially documented to
5827 handle an obscure K&R C case of calling a function that takes the
5828 address of its arguments with fewer arguments than declared. AIX XL
5829 compilers access floating point arguments which do not fit in the
5830 RSA from the stack when a subroutine is compiled without
5831 optimization. Because always storing floating-point arguments on the
5832 stack is inefficient and rarely needed, this option is not enabled by
5833 default and only is necessary when calling subroutines compiled by AIX
5834 XL compilers without optimization.
5838 Support @dfn{AIX Threads}. Link an application written to use
5839 @dfn{pthreads} with special libraries and startup code to enable the
5844 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
5845 application written to use message passing with special startup code to
5846 enable the application to run. The system must have PE installed in the
5847 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
5848 must be overridden with the @samp{-specs=} option to specify the
5849 appropriate directory location. The Parallel Environment does not
5850 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
5851 option are incompatible.
5855 @kindex -msoft-float
5856 Generate code that does not use (uses) the floating-point register set.
5857 Software floating point emulation is provided if you use the
5858 @samp{-msoft-float} option, and pass the option to GCC when linking.
5861 @itemx -mno-multiple
5862 Generate code that uses (does not use) the load multiple word
5863 instructions and the store multiple word instructions. These
5864 instructions are generated by default on POWER systems, and not
5865 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
5866 endian PowerPC systems, since those instructions do not work when the
5867 processor is in little endian mode. The exceptions are PPC740 and
5868 PPC750 which permit the instructions usage in little endian mode.
5873 Generate code that uses (does not use) the load string instructions
5874 and the store string word instructions to save multiple registers and
5875 do small block moves. These instructions are generated by default on
5876 POWER systems, and not generated on PowerPC systems. Do not use
5877 @samp{-mstring} on little endian PowerPC systems, since those
5878 instructions do not work when the processor is in little endian mode.
5879 The exceptions are PPC740 and PPC750 which permit the instructions
5880 usage in little endian mode.
5885 Generate code that uses (does not use) the load or store instructions
5886 that update the base register to the address of the calculated memory
5887 location. These instructions are generated by default. If you use
5888 @samp{-mno-update}, there is a small window between the time that the
5889 stack pointer is updated and the address of the previous frame is
5890 stored, which means code that walks the stack frame across interrupts or
5891 signals may get corrupted data.
5894 @itemx -mno-fused-madd
5895 @kindex -mfused-madd
5896 Generate code that uses (does not use) the floating point multiply and
5897 accumulate instructions. These instructions are generated by default if
5898 hardware floating is used.
5900 @item -mno-bit-align
5903 On System V.4 and embedded PowerPC systems do not (do) force structures
5904 and unions that contain bit fields to be aligned to the base type of the
5907 For example, by default a structure containing nothing but 8
5908 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
5909 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
5910 the structure would be aligned to a 1 byte boundary and be one byte in
5913 @item -mno-strict-align
5914 @itemx -mstrict-align
5915 @kindex -mstrict-align
5916 On System V.4 and embedded PowerPC systems do not (do) assume that
5917 unaligned memory references will be handled by the system.
5920 @itemx -mno-relocatable
5921 @kindex -mrelocatable
5922 On embedded PowerPC systems generate code that allows (does not allow)
5923 the program to be relocated to a different address at runtime. If you
5924 use @samp{-mrelocatable} on any module, all objects linked together must
5925 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
5927 @item -mrelocatable-lib
5928 @itemx -mno-relocatable-lib
5929 On embedded PowerPC systems generate code that allows (does not allow)
5930 the program to be relocated to a different address at runtime. Modules
5931 compiled with @samp{-mrelocatable-lib} can be linked with either modules
5932 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
5933 with modules compiled with the @samp{-mrelocatable} options.
5937 On System V.4 and embedded PowerPC systems do not (do) assume that
5938 register 2 contains a pointer to a global area pointing to the addresses
5939 used in the program.
5942 @itemx -mlittle-endian
5943 On System V.4 and embedded PowerPC systems compile code for the
5944 processor in little endian mode. The @samp{-mlittle-endian} option is
5945 the same as @samp{-mlittle}.
5949 On System V.4 and embedded PowerPC systems compile code for the
5950 processor in big endian mode. The @samp{-mbig-endian} option is
5951 the same as @samp{-mbig}.
5954 On System V.4 and embedded PowerPC systems compile code using calling
5955 conventions that adheres to the March 1995 draft of the System V
5956 Application Binary Interface, PowerPC processor supplement. This is the
5957 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
5959 @item -mcall-sysv-eabi
5960 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
5962 @item -mcall-sysv-noeabi
5963 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
5966 On System V.4 and embedded PowerPC systems compile code using calling
5967 conventions that are similar to those used on AIX. This is the
5968 default if you configured GCC using @samp{powerpc-*-eabiaix}.
5970 @item -mcall-solaris
5971 On System V.4 and embedded PowerPC systems compile code for the Solaris
5975 On System V.4 and embedded PowerPC systems compile code for the
5976 Linux-based GNU system.
5979 @itemx -mno-prototype
5980 On System V.4 and embedded PowerPC systems assume that all calls to
5981 variable argument functions are properly prototyped. Otherwise, the
5982 compiler must insert an instruction before every non prototyped call to
5983 set or clear bit 6 of the condition code register (@var{CR}) to
5984 indicate whether floating point values were passed in the floating point
5985 registers in case the function takes a variable arguments. With
5986 @samp{-mprototype}, only calls to prototyped variable argument functions
5987 will set or clear the bit.
5990 On embedded PowerPC systems, assume that the startup module is called
5991 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
5992 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
5996 On embedded PowerPC systems, assume that the startup module is called
5997 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6001 On embedded PowerPC systems, assume that the startup module is called
6002 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6006 On embedded PowerPC systems, assume that the startup module is called
6007 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6011 On System V.4 and embedded PowerPC systems, specify that you are
6012 compiling for a VxWorks system.
6015 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6016 header to indicate that @samp{eabi} extended relocations are used.
6020 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6021 Embedded Applications Binary Interface (eabi) which is a set of
6022 modifications to the System V.4 specifications. Selecting @option{-meabi}
6023 means that the stack is aligned to an 8 byte boundary, a function
6024 @code{__eabi} is called to from @code{main} to set up the eabi
6025 environment, and the @samp{-msdata} option can use both @code{r2} and
6026 @code{r13} to point to two separate small data areas. Selecting
6027 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6028 do not call an initialization function from @code{main}, and the
6029 @samp{-msdata} option will only use @code{r13} to point to a single
6030 small data area. The @samp{-meabi} option is on by default if you
6031 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
6034 On System V.4 and embedded PowerPC systems, put small initialized
6035 @code{const} global and static data in the @samp{.sdata2} section, which
6036 is pointed to by register @code{r2}. Put small initialized
6037 non-@code{const} global and static data in the @samp{.sdata} section,
6038 which is pointed to by register @code{r13}. Put small uninitialized
6039 global and static data in the @samp{.sbss} section, which is adjacent to
6040 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
6041 incompatible with the @samp{-mrelocatable} option. The
6042 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
6045 On System V.4 and embedded PowerPC systems, put small global and static
6046 data in the @samp{.sdata} section, which is pointed to by register
6047 @code{r13}. Put small uninitialized global and static data in the
6048 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
6049 The @samp{-msdata=sysv} option is incompatible with the
6050 @samp{-mrelocatable} option.
6052 @item -msdata=default
6054 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
6055 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
6056 same as @samp{-msdata=sysv}.
6059 On System V.4 and embedded PowerPC systems, put small global and static
6060 data in the @samp{.sdata} section. Put small uninitialized global and
6061 static data in the @samp{.sbss} section. Do not use register @code{r13}
6062 to address small data however. This is the default behavior unless
6063 other @samp{-msdata} options are used.
6067 On embedded PowerPC systems, put all initialized global and static data
6068 in the @samp{.data} section, and all uninitialized data in the
6069 @samp{.bss} section.
6072 @cindex smaller data references (PowerPC)
6073 @cindex .sdata/.sdata2 references (PowerPC)
6074 On embedded PowerPC systems, put global and static items less than or
6075 equal to @var{num} bytes into the small data or bss sections instead of
6076 the normal data or bss section. By default, @var{num} is 8. The
6077 @samp{-G @var{num}} switch is also passed to the linker.
6078 All modules should be compiled with the same @samp{-G @var{num}} value.
6081 @itemx -mno-regnames
6082 On System V.4 and embedded PowerPC systems do (do not) emit register
6083 names in the assembly language output using symbolic forms.
6088 @subsection IBM RT Options
6090 @cindex IBM RT options
6092 These @samp{-m} options are defined for the IBM RT PC:
6096 Use an in-line code sequence for integer multiplies. This is the
6099 @item -mcall-lib-mul
6100 Call @code{lmul$$} for integer multiples.
6102 @item -mfull-fp-blocks
6103 Generate full-size floating point data blocks, including the minimum
6104 amount of scratch space recommended by IBM. This is the default.
6106 @item -mminimum-fp-blocks
6107 Do not include extra scratch space in floating point data blocks. This
6108 results in smaller code, but slower execution, since scratch space must
6109 be allocated dynamically.
6111 @cindex @file{varargs.h} and RT PC
6112 @cindex @file{stdarg.h} and RT PC
6113 @item -mfp-arg-in-fpregs
6114 Use a calling sequence incompatible with the IBM calling convention in
6115 which floating point arguments are passed in floating point registers.
6116 Note that @code{varargs.h} and @code{stdargs.h} will not work with
6117 floating point operands if this option is specified.
6119 @item -mfp-arg-in-gregs
6120 Use the normal calling convention for floating point arguments. This is
6123 @item -mhc-struct-return
6124 Return structures of more than one word in memory, rather than in a
6125 register. This provides compatibility with the MetaWare HighC (hc)
6126 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
6127 with the Portable C Compiler (pcc).
6129 @item -mnohc-struct-return
6130 Return some structures of more than one word in registers, when
6131 convenient. This is the default. For compatibility with the
6132 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
6133 option @samp{-mhc-struct-return}.
6137 @subsection MIPS Options
6138 @cindex MIPS options
6140 These @samp{-m} options are defined for the MIPS family of computers:
6143 @item -mcpu=@var{cpu type}
6144 Assume the defaults for the machine type @var{cpu type} when scheduling
6145 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
6146 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
6147 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
6148 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
6149 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
6150 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
6151 @var{cpu type} will schedule things appropriately for that particular
6152 chip, the compiler will not generate any code that does not meet level 1
6153 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
6154 or @samp{-mabi} switch being used.
6157 Issue instructions from level 1 of the MIPS ISA. This is the default.
6158 @samp{r3000} is the default @var{cpu type} at this ISA level.
6161 Issue instructions from level 2 of the MIPS ISA (branch likely, square
6162 root instructions). @samp{r6000} is the default @var{cpu type} at this
6166 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
6167 @samp{r4000} is the default @var{cpu type} at this ISA level.
6170 Issue instructions from level 4 of the MIPS ISA (conditional move,
6171 prefetch, enhanced FPU instructions). @samp{r8000} is the default
6172 @var{cpu type} at this ISA level.
6175 Assume that 32 32-bit floating point registers are available. This is
6179 Assume that 32 64-bit floating point registers are available. This is
6180 the default when the @samp{-mips3} option is used.
6183 Assume that 32 32-bit general purpose registers are available. This is
6187 Assume that 32 64-bit general purpose registers are available. This is
6188 the default when the @samp{-mips3} option is used.
6191 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
6192 explanation of the default, and the width of pointers.
6195 Force long types to be 64 bits wide. See @samp{-mlong32} for an
6196 explanation of the default, and the width of pointers.
6199 Force long, int, and pointer types to be 32 bits wide.
6201 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
6202 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
6203 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
6204 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
6205 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
6206 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
6207 are 32 bits, and longs are 64 bits wide. The width of pointer types is
6208 the smaller of the width of longs or the width of general purpose
6209 registers (which in turn depends on the ISA).
6216 Generate code for the indicated ABI. The default instruction level is
6217 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
6218 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
6219 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
6223 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
6224 add normal debug information. This is the default for all
6225 platforms except for the OSF/1 reference platform, using the OSF/rose
6226 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
6227 switches are used, the @file{mips-tfile} program will encapsulate the
6228 stabs within MIPS ECOFF.
6231 Generate code for the GNU assembler. This is the default on the OSF/1
6232 reference platform, using the OSF/rose object format. Also, this is
6233 the default if the configure option @samp{--with-gnu-as} is used.
6235 @item -msplit-addresses
6236 @itemx -mno-split-addresses
6237 Generate code to load the high and low parts of address constants separately.
6238 This allows @code{gcc} to optimize away redundant loads of the high order
6239 bits of addresses. This optimization requires GNU as and GNU ld.
6240 This optimization is enabled by default for some embedded targets where
6241 GNU as and GNU ld are standard.
6245 The @samp{-mrnames} switch says to output code using the MIPS software
6246 names for the registers, instead of the hardware names (ie, @var{a0}
6247 instead of @var{$4}). The only known assembler that supports this option
6248 is the Algorithmics assembler.
6252 The @samp{-mgpopt} switch says to write all of the data declarations
6253 before the instructions in the text section, this allows the MIPS
6254 assembler to generate one word memory references instead of using two
6255 words for short global or static data items. This is on by default if
6256 optimization is selected.
6260 For each non-inline function processed, the @samp{-mstats} switch
6261 causes the compiler to emit one line to the standard error file to
6262 print statistics about the program (number of registers saved, stack
6267 The @samp{-mmemcpy} switch makes all block moves call the appropriate
6268 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
6269 generating inline code.
6272 @itemx -mno-mips-tfile
6273 The @samp{-mno-mips-tfile} switch causes the compiler not
6274 postprocess the object file with the @file{mips-tfile} program,
6275 after the MIPS assembler has generated it to add debug support. If
6276 @file{mips-tfile} is not run, then no local variables will be
6277 available to the debugger. In addition, @file{stage2} and
6278 @file{stage3} objects will have the temporary file names passed to the
6279 assembler embedded in the object file, which means the objects will
6280 not compare the same. The @samp{-mno-mips-tfile} switch should only
6281 be used when there are bugs in the @file{mips-tfile} program that
6282 prevents compilation.
6285 Generate output containing library calls for floating point.
6286 @strong{Warning:} the requisite libraries are not part of GCC.
6287 Normally the facilities of the machine's usual C compiler are used, but
6288 this can't be done directly in cross-compilation. You must make your
6289 own arrangements to provide suitable library functions for
6293 Generate output containing floating point instructions. This is the
6294 default if you use the unmodified sources.
6297 @itemx -mno-abicalls
6298 Emit (or do not emit) the pseudo operations @samp{.abicalls},
6299 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
6300 position independent code.
6303 @itemx -mno-long-calls
6304 Do all calls with the @samp{JALR} instruction, which requires
6305 loading up a function's address into a register before the call.
6306 You need to use this switch, if you call outside of the current
6307 512 megabyte segment to functions that are not through pointers.
6310 @itemx -mno-half-pic
6311 Put pointers to extern references into the data section and load them
6312 up, rather than put the references in the text section.
6314 @item -membedded-pic
6315 @itemx -mno-embedded-pic
6316 Generate PIC code suitable for some embedded systems. All calls are
6317 made using PC relative address, and all data is addressed using the $gp
6318 register. No more than 65536 bytes of global data may be used. This
6319 requires GNU as and GNU ld which do most of the work. This currently
6320 only works on targets which use ECOFF; it does not work with ELF.
6322 @item -membedded-data
6323 @itemx -mno-embedded-data
6324 Allocate variables to the read-only data section first if possible, then
6325 next in the small data section if possible, otherwise in data. This gives
6326 slightly slower code than the default, but reduces the amount of RAM required
6327 when executing, and thus may be preferred for some embedded systems.
6329 @item -muninit-const-in-rodata
6330 @itemx -mno-uninit-const-in-rodata
6331 When used together with -membedded-data, it will always store uninitialized
6332 const variables in the read-only data section.
6334 @item -msingle-float
6335 @itemx -mdouble-float
6336 The @samp{-msingle-float} switch tells gcc to assume that the floating
6337 point coprocessor only supports single precision operations, as on the
6338 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
6339 double precision operations. This is the default.
6343 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
6344 as on the @samp{r4650} chip.
6347 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
6352 Enable 16-bit instructions.
6355 Use the entry and exit pseudo ops. This option can only be used with
6359 Compile code for the processor in little endian mode.
6360 The requisite libraries are assumed to exist.
6363 Compile code for the processor in big endian mode.
6364 The requisite libraries are assumed to exist.
6367 @cindex smaller data references (MIPS)
6368 @cindex gp-relative references (MIPS)
6369 Put global and static items less than or equal to @var{num} bytes into
6370 the small data or bss sections instead of the normal data or bss
6371 section. This allows the assembler to emit one word memory reference
6372 instructions based on the global pointer (@var{gp} or @var{$28}),
6373 instead of the normal two words used. By default, @var{num} is 8 when
6374 the MIPS assembler is used, and 0 when the GNU assembler is used. The
6375 @samp{-G @var{num}} switch is also passed to the assembler and linker.
6376 All modules should be compiled with the same @samp{-G @var{num}}
6380 Tell the MIPS assembler to not run its preprocessor over user
6381 assembler files (with a @samp{.s} suffix) when assembling them.
6384 Pass an option to gas which will cause nops to be inserted if
6385 the read of the destination register of an mfhi or mflo instruction
6386 occurs in the following two instructions.
6389 Do not include the default crt0.
6393 These options are defined by the macro
6394 @code{TARGET_SWITCHES} in the machine description. The default for the
6395 options is also defined by that macro, which enables you to change the
6400 @subsection Intel 386 Options
6401 @cindex i386 Options
6402 @cindex Intel 386 Options
6404 These @samp{-m} options are defined for the i386 family of computers:
6407 @item -mcpu=@var{cpu type}
6408 Assume the defaults for the machine type @var{cpu type} when scheduling
6409 instructions. The choices for @var{cpu type} are @samp{i386},
6410 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
6411 @samp{pentiumpro}, @samp{k6}, and @samp{athlon}
6413 While picking a specific @var{cpu type} will schedule things appropriately
6414 for that particular chip, the compiler will not generate any code that
6415 does not run on the i386 without the @samp{-march=@var{cpu type}} option
6416 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
6417 is equivalent to @samp{pentiumpro}. @samp{k6} is the AMD chip as
6418 opposed to the Intel ones.
6420 @item -march=@var{cpu type}
6421 Generate instructions for the machine type @var{cpu type}. The choices
6422 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
6423 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
6429 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
6430 respectively. These synonyms are deprecated.
6432 @item -mintel-syntax
6433 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
6437 Control whether or not the compiler uses IEEE floating point
6438 comparisons. These handle correctly the case where the result of a
6439 comparison is unordered.
6442 Generate output containing library calls for floating point.
6443 @strong{Warning:} the requisite libraries are not part of GCC.
6444 Normally the facilities of the machine's usual C compiler are used, but
6445 this can't be done directly in cross-compilation. You must make your
6446 own arrangements to provide suitable library functions for
6449 On machines where a function returns floating point results in the 80387
6450 register stack, some floating point opcodes may be emitted even if
6451 @samp{-msoft-float} is used.
6453 @item -mno-fp-ret-in-387
6454 Do not use the FPU registers for return values of functions.
6456 The usual calling convention has functions return values of types
6457 @code{float} and @code{double} in an FPU register, even if there
6458 is no FPU. The idea is that the operating system should emulate
6461 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
6462 in ordinary CPU registers instead.
6464 @item -mno-fancy-math-387
6465 Some 387 emulators do not support the @code{sin}, @code{cos} and
6466 @code{sqrt} instructions for the 387. Specify this option to avoid
6467 generating those instructions. This option is the default on FreeBSD.
6468 As of revision 2.6.1, these instructions are not generated unless you
6469 also use the @samp{-ffast-math} switch.
6471 @item -malign-double
6472 @itemx -mno-align-double
6473 Control whether GCC aligns @code{double}, @code{long double}, and
6474 @code{long long} variables on a two word boundary or a one word
6475 boundary. Aligning @code{double} variables on a two word boundary will
6476 produce code that runs somewhat faster on a @samp{Pentium} at the
6477 expense of more memory.
6479 @item -m128bit-long-double
6480 @itemx -m128bit-long-double
6481 Control the size of @code{long double} type. i386 application binary interface
6482 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
6483 preffer @code{long double} aligned to 8 or 16 byte boundary. This is
6484 impossible to reach with 12 byte long doubles in the array accesses.
6486 @strong{Warning:} if you use the @samp{-m128bit-long-double} switch, the
6487 structures and arrays containing @code{long double} will change their size as
6488 well as function calling convention for function taking @code{long double}
6491 @item -m96bit-long-double
6492 @itemx -m96bit-long-double
6493 Set the size of @code{long double} to 96 bits as required by the i386
6494 application binary interface. This is the default.
6497 @itemx -mno-svr3-shlib
6498 Control whether GCC places uninitialized locals into @code{bss} or
6499 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
6500 These options are meaningful only on System V Release 3.
6502 @item -mno-wide-multiply
6503 @itemx -mwide-multiply
6504 Control whether GCC uses the @code{mul} and @code{imul} that produce
6505 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
6506 long} multiplies and 32-bit division by constants.
6509 Use a different function-calling convention, in which functions that
6510 take a fixed number of arguments return with the @code{ret} @var{num}
6511 instruction, which pops their arguments while returning. This saves one
6512 instruction in the caller since there is no need to pop the arguments
6515 You can specify that an individual function is called with this calling
6516 sequence with the function attribute @samp{stdcall}. You can also
6517 override the @samp{-mrtd} option by using the function attribute
6518 @samp{cdecl}. @xref{Function Attributes}.
6520 @strong{Warning:} this calling convention is incompatible with the one
6521 normally used on Unix, so you cannot use it if you need to call
6522 libraries compiled with the Unix compiler.
6524 Also, you must provide function prototypes for all functions that
6525 take variable numbers of arguments (including @code{printf});
6526 otherwise incorrect code will be generated for calls to those
6529 In addition, seriously incorrect code will result if you call a
6530 function with too many arguments. (Normally, extra arguments are
6531 harmlessly ignored.)
6533 @item -mreg-alloc=@var{regs}
6534 Control the default allocation order of integer registers. The
6535 string @var{regs} is a series of letters specifying a register. The
6536 supported letters are: @code{a} allocate EAX; @code{b} allocate EBX;
6537 @code{c} allocate ECX; @code{d} allocate EDX; @code{S} allocate ESI;
6538 @code{D} allocate EDI; @code{B} allocate EBP.
6540 @item -mregparm=@var{num}
6541 Control how many registers are used to pass integer arguments. By
6542 default, no registers are used to pass arguments, and at most 3
6543 registers can be used. You can control this behavior for a specific
6544 function by using the function attribute @samp{regparm}.
6545 @xref{Function Attributes}.
6547 @strong{Warning:} if you use this switch, and
6548 @var{num} is nonzero, then you must build all modules with the same
6549 value, including any libraries. This includes the system libraries and
6552 @item -malign-loops=@var{num}
6553 Align loops to a 2 raised to a @var{num} byte boundary. If
6554 @samp{-malign-loops} is not specified, the default is 2 unless
6555 gas 2.8 (or later) is being used in which case the default is
6556 to align the loop on a 16 byte boundary if it is less than 8
6559 @item -malign-jumps=@var{num}
6560 Align instructions that are only jumped to to a 2 raised to a @var{num}
6561 byte boundary. If @samp{-malign-jumps} is not specified, the default is
6562 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
6563 gas 2.8 (or later) is being used in which case the default is
6564 to align the instruction on a 16 byte boundary if it is less
6567 @item -malign-functions=@var{num}
6568 Align the start of functions to a 2 raised to @var{num} byte boundary.
6569 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
6570 for a 386, and 4 if optimizing for a 486.
6572 @item -mpreferred-stack-boundary=@var{num}
6573 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
6574 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
6575 the default is 4 (16 bytes or 128 bits).
6577 The stack is required to be aligned on a 4 byte boundary. On Pentium
6578 and PentiumPro, @code{double} and @code{long double} values should be
6579 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
6580 significant run time performance penalties. On Pentium III, the
6581 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
6582 penalties if it is not 16 byte aligned.
6584 To ensure proper alignment of this values on the stack, the stack boundary
6585 must be as aligned as that required by any value stored on the stack.
6586 Further, every function must be generated such that it keeps the stack
6587 aligned. Thus calling a function compiled with a higher preferred
6588 stack boundary from a function compiled with a lower preferred stack
6589 boundary will most likely misalign the stack. It is recommended that
6590 libraries that use callbacks always use the default setting.
6592 This extra alignment does consume extra stack space. Code that is sensitive
6593 to stack space usage, such as embedded systems and operating system kernels,
6594 may want to reduce the preferred alignment to
6595 @samp{-mpreferred-stack-boundary=2}.
6599 Use PUSH operations to store outgoing parameters. This method is shorter
6600 and usually equally fast as method using SUB/MOV operations and is enabled
6601 by default. In some cases disabling it may improve performance because of
6602 improved scheduling and reduced dependencies.
6604 @item -maccumulate-outgoing-args
6605 @kindex -maccumulate-outgoing-args
6606 If enabled, the maximum amount of space required for outgoing arguments will be
6607 computed in the function prologue. This in faster on most modern CPUs
6608 because of reduced dependencies, improved scheduling and reduced stack usage
6609 when preferred stack boundary is not equal to 2. The drawback is a notable
6610 increase in code size. This switch implies -mno-push-args.
6614 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
6615 on thread-safe exception handling must compile and link all code with the
6616 @samp{-mthreads} option. When compiling, @samp{-mthreads} defines
6617 @samp{-D_MT}; when linking, it links in a special thread helper library
6618 @samp{-lmingwthrd} which cleans up per thread exception handling data.
6620 @item -mno-align-stringops
6621 @kindex -mno-align-stringops
6622 Do not align destination of inlined string operations. This switch reduces
6623 code size and improves performance in case the destination is already aligned,
6624 but gcc don't know about it.
6626 @item -minline-all-stringops
6627 @kindex -minline-all-stringops
6628 By default GCC inlines string operations only when destination is known to be
6629 aligned at least to 4 byte boundary. This enables more inlining, increase code
6630 size, but may improve performance of code that depends on fast memcpy, strlen
6631 and memset for short lengths.
6635 @subsection HPPA Options
6636 @cindex HPPA Options
6638 These @samp{-m} options are defined for the HPPA family of computers:
6641 @item -march=@var{architecture type}
6642 Generate code for the specified architecture. The choices for
6643 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
6644 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
6645 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
6646 architecture option for your machine. Code compiled for lower numbered
6647 architectures will run on higher numbered architectures, but not the
6650 PA 2.0 support currently requires gas snapshot 19990413 or later. The
6651 next release of binutils (current is 2.9.1) will probably contain PA 2.0
6655 @itemx -mpa-risc-1-1
6656 @itemx -mpa-risc-2-0
6657 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
6660 Generate code suitable for big switch tables. Use this option only if
6661 the assembler/linker complain about out of range branches within a switch
6664 @item -mjump-in-delay
6665 Fill delay slots of function calls with unconditional jump instructions
6666 by modifying the return pointer for the function call to be the target
6667 of the conditional jump.
6669 @item -mdisable-fpregs
6670 Prevent floating point registers from being used in any manner. This is
6671 necessary for compiling kernels which perform lazy context switching of
6672 floating point registers. If you use this option and attempt to perform
6673 floating point operations, the compiler will abort.
6675 @item -mdisable-indexing
6676 Prevent the compiler from using indexing address modes. This avoids some
6677 rather obscure problems when compiling MIG generated code under MACH.
6679 @item -mno-space-regs
6680 Generate code that assumes the target has no space registers. This allows
6681 GCC to generate faster indirect calls and use unscaled index address modes.
6683 Such code is suitable for level 0 PA systems and kernels.
6685 @item -mfast-indirect-calls
6686 Generate code that assumes calls never cross space boundaries. This
6687 allows GCC to emit code which performs faster indirect calls.
6689 This option will not work in the presence of shared libraries or nested
6692 @item -mlong-load-store
6693 Generate 3-instruction load and store sequences as sometimes required by
6694 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
6697 @item -mportable-runtime
6698 Use the portable calling conventions proposed by HP for ELF systems.
6701 Enable the use of assembler directives only GAS understands.
6703 @item -mschedule=@var{cpu type}
6704 Schedule code according to the constraints for the machine type
6705 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
6706 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
6707 @file{/usr/lib/sched.models} on an HP-UX system to determine the
6708 proper scheduling option for your machine.
6711 Enable the optimization pass in the HPUX linker. Note this makes symbolic
6712 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
6713 in which they give bogus error messages when linking some programs.
6716 Generate output containing library calls for floating point.
6717 @strong{Warning:} the requisite libraries are not available for all HPPA
6718 targets. Normally the facilities of the machine's usual C compiler are
6719 used, but this cannot be done directly in cross-compilation. You must make
6720 your own arrangements to provide suitable library functions for
6721 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
6722 does provide software floating point support.
6724 @samp{-msoft-float} changes the calling convention in the output file;
6725 therefore, it is only useful if you compile @emph{all} of a program with
6726 this option. In particular, you need to compile @file{libgcc.a}, the
6727 library that comes with GCC, with @samp{-msoft-float} in order for
6731 @node Intel 960 Options
6732 @subsection Intel 960 Options
6734 These @samp{-m} options are defined for the Intel 960 implementations:
6737 @item -m@var{cpu type}
6738 Assume the defaults for the machine type @var{cpu type} for some of
6739 the other options, including instruction scheduling, floating point
6740 support, and addressing modes. The choices for @var{cpu type} are
6741 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
6742 @samp{sa}, and @samp{sb}.
6748 The @samp{-mnumerics} option indicates that the processor does support
6749 floating-point instructions. The @samp{-msoft-float} option indicates
6750 that floating-point support should not be assumed.
6752 @item -mleaf-procedures
6753 @itemx -mno-leaf-procedures
6754 Do (or do not) attempt to alter leaf procedures to be callable with the
6755 @code{bal} instruction as well as @code{call}. This will result in more
6756 efficient code for explicit calls when the @code{bal} instruction can be
6757 substituted by the assembler or linker, but less efficient code in other
6758 cases, such as calls via function pointers, or using a linker that doesn't
6759 support this optimization.
6762 @itemx -mno-tail-call
6763 Do (or do not) make additional attempts (beyond those of the
6764 machine-independent portions of the compiler) to optimize tail-recursive
6765 calls into branches. You may not want to do this because the detection of
6766 cases where this is not valid is not totally complete. The default is
6767 @samp{-mno-tail-call}.
6769 @item -mcomplex-addr
6770 @itemx -mno-complex-addr
6771 Assume (or do not assume) that the use of a complex addressing mode is a
6772 win on this implementation of the i960. Complex addressing modes may not
6773 be worthwhile on the K-series, but they definitely are on the C-series.
6774 The default is currently @samp{-mcomplex-addr} for all processors except
6778 @itemx -mno-code-align
6779 Align code to 8-byte boundaries for faster fetching (or don't bother).
6780 Currently turned on by default for C-series implementations only.
6783 @item -mclean-linkage
6784 @itemx -mno-clean-linkage
6785 These options are not fully implemented.
6789 @itemx -mic2.0-compat
6790 @itemx -mic3.0-compat
6791 Enable compatibility with iC960 v2.0 or v3.0.
6795 Enable compatibility with the iC960 assembler.
6797 @item -mstrict-align
6798 @itemx -mno-strict-align
6799 Do not permit (do permit) unaligned accesses.
6802 Enable structure-alignment compatibility with Intel's gcc release version
6803 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
6805 @item -mlong-double-64
6806 Implement type @samp{long double} as 64-bit floating point numbers.
6807 Without the option @samp{long double} is implemented by 80-bit
6808 floating point numbers. The only reason we have it because there is
6809 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
6810 is only useful for people using soft-float targets. Otherwise, we
6811 should recommend against use of it.
6815 @node DEC Alpha Options
6816 @subsection DEC Alpha Options
6818 These @samp{-m} options are defined for the DEC Alpha implementations:
6821 @item -mno-soft-float
6823 Use (do not use) the hardware floating-point instructions for
6824 floating-point operations. When @option{-msoft-float} is specified,
6825 functions in @file{libgcc1.c} will be used to perform floating-point
6826 operations. Unless they are replaced by routines that emulate the
6827 floating-point operations, or compiled in such a way as to call such
6828 emulations routines, these routines will issue floating-point
6829 operations. If you are compiling for an Alpha without floating-point
6830 operations, you must ensure that the library is built so as not to call
6833 Note that Alpha implementations without floating-point operations are
6834 required to have floating-point registers.
6838 Generate code that uses (does not use) the floating-point register set.
6839 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
6840 register set is not used, floating point operands are passed in integer
6841 registers as if they were integers and floating-point results are passed
6842 in $0 instead of $f0. This is a non-standard calling sequence, so any
6843 function with a floating-point argument or return value called by code
6844 compiled with @option{-mno-fp-regs} must also be compiled with that
6847 A typical use of this option is building a kernel that does not use,
6848 and hence need not save and restore, any floating-point registers.
6851 The Alpha architecture implements floating-point hardware optimized for
6852 maximum performance. It is mostly compliant with the IEEE floating
6853 point standard. However, for full compliance, software assistance is
6854 required. This option generates code fully IEEE compliant code
6855 @emph{except} that the @var{inexact flag} is not maintained (see below).
6856 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
6857 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
6858 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
6859 code is less efficient but is able to correctly support denormalized
6860 numbers and exceptional IEEE values such as not-a-number and plus/minus
6861 infinity. Other Alpha compilers call this option
6862 @option{-ieee_with_no_inexact}.
6864 @item -mieee-with-inexact
6865 @c overfull hbox here --bob 22 jul96
6866 @c original text between ignore ... end ignore
6868 This is like @samp{-mieee} except the generated code also maintains the
6869 IEEE @var{inexact flag}. Turning on this option causes the generated
6870 code to implement fully-compliant IEEE math. The option is a shorthand
6871 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
6872 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
6873 implementations the resulting code may execute significantly slower than
6874 the code generated by default. Since there is very little code that
6875 depends on the @var{inexact flag}, you should normally not specify this
6876 option. Other Alpha compilers call this option
6877 @samp{-ieee_with_inexact}.
6879 @c changed paragraph
6880 This is like @samp{-mieee} except the generated code also maintains the
6881 IEEE @var{inexact flag}. Turning on this option causes the generated
6882 code to implement fully-compliant IEEE math. The option is a shorthand
6883 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
6884 @samp{-mieee-conformant},
6885 @samp{-mfp-trap-mode=sui},
6886 and @samp{-mtrap-precision=i}.
6887 On some Alpha implementations the resulting code may execute
6888 significantly slower than the code generated by default. Since there
6889 is very little code that depends on the @var{inexact flag}, you should
6890 normally not specify this option. Other Alpha compilers call this
6891 option @samp{-ieee_with_inexact}.
6892 @c end changes to prevent overfull hboxes
6894 @item -mfp-trap-mode=@var{trap mode}
6895 This option controls what floating-point related traps are enabled.
6896 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
6897 The trap mode can be set to one of four values:
6901 This is the default (normal) setting. The only traps that are enabled
6902 are the ones that cannot be disabled in software (e.g., division by zero
6906 In addition to the traps enabled by @samp{n}, underflow traps are enabled
6910 Like @samp{su}, but the instructions are marked to be safe for software
6911 completion (see Alpha architecture manual for details).
6914 Like @samp{su}, but inexact traps are enabled as well.
6917 @item -mfp-rounding-mode=@var{rounding mode}
6918 Selects the IEEE rounding mode. Other Alpha compilers call this option
6919 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
6924 Normal IEEE rounding mode. Floating point numbers are rounded towards
6925 the nearest machine number or towards the even machine number in case
6929 Round towards minus infinity.
6932 Chopped rounding mode. Floating point numbers are rounded towards zero.
6935 Dynamic rounding mode. A field in the floating point control register
6936 (@var{fpcr}, see Alpha architecture reference manual) controls the
6937 rounding mode in effect. The C library initializes this register for
6938 rounding towards plus infinity. Thus, unless your program modifies the
6939 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
6942 @item -mtrap-precision=@var{trap precision}
6943 In the Alpha architecture, floating point traps are imprecise. This
6944 means without software assistance it is impossible to recover from a
6945 floating trap and program execution normally needs to be terminated.
6946 GCC can generate code that can assist operating system trap handlers
6947 in determining the exact location that caused a floating point trap.
6948 Depending on the requirements of an application, different levels of
6949 precisions can be selected:
6953 Program precision. This option is the default and means a trap handler
6954 can only identify which program caused a floating point exception.
6957 Function precision. The trap handler can determine the function that
6958 caused a floating point exception.
6961 Instruction precision. The trap handler can determine the exact
6962 instruction that caused a floating point exception.
6965 Other Alpha compilers provide the equivalent options called
6966 @samp{-scope_safe} and @samp{-resumption_safe}.
6968 @item -mieee-conformant
6969 This option marks the generated code as IEEE conformant. You must not
6970 use this option unless you also specify @samp{-mtrap-precision=i} and either
6971 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
6972 is to emit the line @samp{.eflag 48} in the function prologue of the
6973 generated assembly file. Under DEC Unix, this has the effect that
6974 IEEE-conformant math library routines will be linked in.
6976 @item -mbuild-constants
6977 Normally GCC examines a 32- or 64-bit integer constant to
6978 see if it can construct it from smaller constants in two or three
6979 instructions. If it cannot, it will output the constant as a literal and
6980 generate code to load it from the data segment at runtime.
6982 Use this option to require GCC to construct @emph{all} integer constants
6983 using code, even if it takes more instructions (the maximum is six).
6985 You would typically use this option to build a shared library dynamic
6986 loader. Itself a shared library, it must relocate itself in memory
6987 before it can find the variables and constants in its own data segment.
6991 Select whether to generate code to be assembled by the vendor-supplied
6992 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
7000 Indicate whether GCC should generate code to use the optional BWX,
7001 CIX, and MAX instruction sets. The default is to use the instruction sets
7002 supported by the CPU type specified via @samp{-mcpu=} option or that
7003 of the CPU on which GCC was built if none was specified.
7005 @item -mcpu=@var{cpu_type}
7006 Set the instruction set, register set, and instruction scheduling
7007 parameters for machine type @var{cpu_type}. You can specify either the
7008 @samp{EV} style name or the corresponding chip number. GCC
7009 supports scheduling parameters for the EV4 and EV5 family of processors
7010 and will choose the default values for the instruction set from
7011 the processor you specify. If you do not specify a processor type,
7012 GCC will default to the processor on which the compiler was built.
7014 Supported values for @var{cpu_type} are
7019 Schedules as an EV4 and has no instruction set extensions.
7023 Schedules as an EV5 and has no instruction set extensions.
7027 Schedules as an EV5 and supports the BWX extension.
7032 Schedules as an EV5 and supports the BWX and MAX extensions.
7036 Schedules as an EV5 (until Digital releases the scheduling parameters
7037 for the EV6) and supports the BWX, CIX, and MAX extensions.
7040 @item -mmemory-latency=@var{time}
7041 Sets the latency the scheduler should assume for typical memory
7042 references as seen by the application. This number is highly
7043 dependent on the memory access patterns used by the application
7044 and the size of the external cache on the machine.
7046 Valid options for @var{time} are
7050 A decimal number representing clock cycles.
7056 The compiler contains estimates of the number of clock cycles for
7057 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7058 (also called Dcache, Scache, and Bcache), as well as to main memory.
7059 Note that L3 is only valid for EV5.
7064 @node Clipper Options
7065 @subsection Clipper Options
7067 These @samp{-m} options are defined for the Clipper implementations:
7071 Produce code for a C300 Clipper processor. This is the default.
7074 Produce code for a C400 Clipper processor i.e. use floating point
7078 @node H8/300 Options
7079 @subsection H8/300 Options
7081 These @samp{-m} options are defined for the H8/300 implementations:
7085 Shorten some address references at link time, when possible; uses the
7086 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
7087 ld.info, Using ld}, for a fuller description.
7090 Generate code for the H8/300H.
7093 Generate code for the H8/S.
7096 Generate code for the H8/S2600. This switch must be used with -ms.
7099 Make @code{int} data 32 bits by default.
7102 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
7103 The default for the H8/300H and H8/S is to align longs and floats on 4
7105 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
7106 This option has no effect on the H8/300.
7110 @subsection SH Options
7112 These @samp{-m} options are defined for the SH implementations:
7116 Generate code for the SH1.
7119 Generate code for the SH2.
7122 Generate code for the SH3.
7125 Generate code for the SH3e.
7128 Generate code for the SH4 without a floating-point unit.
7130 @item -m4-single-only
7131 Generate code for the SH4 with a floating-point unit that only
7132 supports single-precision arithmentic.
7135 Generate code for the SH4 assuming the floating-point unit is in
7136 single-precision mode by default.
7139 Generate code for the SH4.
7142 Compile code for the processor in big endian mode.
7145 Compile code for the processor in little endian mode.
7148 Align doubles at 64 bit boundaries. Note that this changes the calling
7149 conventions, and thus some functions from the standard C library will
7150 not work unless you recompile it first with -mdalign.
7153 Shorten some address references at link time, when possible; uses the
7154 linker option @samp{-relax}.
7157 Use 32-bit offsets in @code{switch} tables. The default is to use
7161 Enable the use of the instruction @code{fmovd}.
7164 Comply with the calling conventions defined by Hitachi.
7167 Mark the @code{MAC} register as call-clobbered, even if
7168 @option{-mhitachi} is given.
7171 Dump instruction size and location in the assembly code.
7174 This option is deprecated. It pads structures to multiple of 4 bytes,
7175 which is incompatible with the SH ABI.
7178 Optimize for space instead of speed. Implied by @option{-Os}.
7181 When generating position-independent code, emit function calls using
7182 the Global Offset Table instead of the Procedure Linkage Table.
7185 Generate a library function call to invalidate instruction cache
7186 entries, after fixing up a trampoline. This library function call
7187 doesn't assume it can write to the whole memory address space. This
7188 is the default when the target is @code{sh-*-linux*}.
7191 @node System V Options
7192 @subsection Options for System V
7194 These additional options are available on System V Release 4 for
7195 compatibility with other compilers on those systems:
7199 Create a shared object.
7200 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
7203 Identify the versions of each tool used by the compiler, in a
7204 @code{.ident} assembler directive in the output.
7207 Refrain from adding @code{.ident} directives to the output file (this is
7210 @item -YP,@var{dirs}
7211 Search the directories @var{dirs}, and no others, for libraries
7212 specified with @samp{-l}.
7215 Look in the directory @var{dir} to find the M4 preprocessor.
7216 The assembler uses this option.
7217 @c This is supposed to go with a -Yd for predefined M4 macro files, but
7218 @c the generic assembler that comes with Solaris takes just -Ym.
7221 @node TMS320C3x/C4x Options
7222 @subsection TMS320C3x/C4x Options
7223 @cindex TMS320C3x/C4x Options
7225 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
7229 @item -mcpu=@var{cpu_type}
7230 Set the instruction set, register set, and instruction scheduling
7231 parameters for machine type @var{cpu_type}. Supported values for
7232 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
7233 @samp{c44}. The default is @samp{c40} to generate code for the
7238 @itemx -msmall-memory
7240 Generates code for the big or small memory model. The small memory
7241 model assumed that all data fits into one 64K word page. At run-time
7242 the data page (DP) register must be set to point to the 64K page
7243 containing the .bss and .data program sections. The big memory model is
7244 the default and requires reloading of the DP register for every direct
7249 Allow (disallow) allocation of general integer operands into the block
7254 Enable (disable) generation of code using decrement and branch,
7255 DBcond(D), instructions. This is enabled by default for the C4x. To be
7256 on the safe side, this is disabled for the C3x, since the maximum
7257 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
7258 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
7259 that it can utilise the decrement and branch instruction, but will give
7260 up if there is more than one memory reference in the loop. Thus a loop
7261 where the loop counter is decremented can generate slightly more
7262 efficient code, in cases where the RPTB instruction cannot be utilised.
7264 @item -mdp-isr-reload
7266 Force the DP register to be saved on entry to an interrupt service
7267 routine (ISR), reloaded to point to the data section, and restored on
7268 exit from the ISR. This should not be required unless someone has
7269 violated the small memory model by modifying the DP register, say within
7274 For the C3x use the 24-bit MPYI instruction for integer multiplies
7275 instead of a library call to guarantee 32-bit results. Note that if one
7276 of the operands is a constant, then the multiplication will be performed
7277 using shifts and adds. If the -mmpyi option is not specified for the C3x,
7278 then squaring operations are performed inline instead of a library call.
7281 @itemx -mno-fast-fix
7282 The C3x/C4x FIX instruction to convert a floating point value to an
7283 integer value chooses the nearest integer less than or equal to the
7284 floating point value rather than to the nearest integer. Thus if the
7285 floating point number is negative, the result will be incorrectly
7286 truncated an additional code is necessary to detect and correct this
7287 case. This option can be used to disable generation of the additional
7288 code required to correct the result.
7292 Enable (disable) generation of repeat block sequences using the RPTB
7293 instruction for zero overhead looping. The RPTB construct is only used
7294 for innermost loops that do not call functions or jump across the loop
7295 boundaries. There is no advantage having nested RPTB loops due to the
7296 overhead required to save and restore the RC, RS, and RE registers.
7297 This is enabled by default with -O2.
7299 @item -mrpts=@var{count}
7301 Enable (disable) the use of the single instruction repeat instruction
7302 RPTS. If a repeat block contains a single instruction, and the loop
7303 count can be guaranteed to be less than the value @var{count}, GCC will
7304 emit a RPTS instruction instead of a RPTB. If no value is specified,
7305 then a RPTS will be emitted even if the loop count cannot be determined
7306 at compile time. Note that the repeated instruction following RPTS does
7307 not have to be reloaded from memory each iteration, thus freeing up the
7308 CPU buses for operands. However, since interrupts are blocked by this
7309 instruction, it is disabled by default.
7311 @item -mloop-unsigned
7312 @itemx -mno-loop-unsigned
7313 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
7314 is 2^31 + 1 since these instructions test if the iteration count is
7315 negative to terminate the loop. If the iteration count is unsigned
7316 there is a possibility than the 2^31 + 1 maximum iteration count may be
7317 exceeded. This switch allows an unsigned iteration count.
7320 Try to emit an assembler syntax that the TI assembler (asm30) is happy
7321 with. This also enforces compatibility with the API employed by the TI
7322 C3x C compiler. For example, long doubles are passed as structures
7323 rather than in floating point registers.
7327 Generate code that uses registers (stack) for passing arguments to functions.
7328 By default, arguments are passed in registers where possible rather
7329 than by pushing arguments on to the stack.
7331 @item -mparallel-insns
7332 @itemx -mno-parallel-insns
7333 Allow the generation of parallel instructions. This is enabled by
7336 @item -mparallel-mpy
7337 @itemx -mno-parallel-mpy
7338 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
7339 provided -mparallel-insns is also specified. These instructions have
7340 tight register constraints which can pessimize the code generation
7346 @subsection V850 Options
7347 @cindex V850 Options
7349 These @samp{-m} options are defined for V850 implementations:
7353 @itemx -mno-long-calls
7354 Treat all calls as being far away (near). If calls are assumed to be
7355 far away, the compiler will always load the functions address up into a
7356 register, and call indirect through the pointer.
7360 Do not optimize (do optimize) basic blocks that use the same index
7361 pointer 4 or more times to copy pointer into the @code{ep} register, and
7362 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
7363 option is on by default if you optimize.
7365 @item -mno-prolog-function
7366 @itemx -mprolog-function
7367 Do not use (do use) external functions to save and restore registers at
7368 the prolog and epilog of a function. The external functions are slower,
7369 but use less code space if more than one function saves the same number
7370 of registers. The @samp{-mprolog-function} option is on by default if
7374 Try to make the code as small as possible. At present, this just turns
7375 on the @samp{-mep} and @samp{-mprolog-function} options.
7378 Put static or global variables whose size is @var{n} bytes or less into
7379 the tiny data area that register @code{ep} points to. The tiny data
7380 area can hold up to 256 bytes in total (128 bytes for byte references).
7383 Put static or global variables whose size is @var{n} bytes or less into
7384 the small data area that register @code{gp} points to. The small data
7385 area can hold up to 64 kilobytes.
7388 Put static or global variables whose size is @var{n} bytes or less into
7389 the first 32 kilobytes of memory.
7392 Specify that the target processor is the V850.
7395 Generate code suitable for big switch tables. Use this option only if
7396 the assembler/linker complain about out of range branches within a switch
7401 @subsection ARC Options
7404 These options are defined for ARC implementations:
7408 Compile code for little endian mode. This is the default.
7411 Compile code for big endian mode.
7414 Prepend the name of the cpu to all public symbol names.
7415 In multiple-processor systems, there are many ARC variants with different
7416 instruction and register set characteristics. This flag prevents code
7417 compiled for one cpu to be linked with code compiled for another.
7418 No facility exists for handling variants that are "almost identical".
7419 This is an all or nothing option.
7421 @item -mcpu=@var{cpu}
7422 Compile code for ARC variant @var{cpu}.
7423 Which variants are supported depend on the configuration.
7424 All variants support @samp{-mcpu=base}, this is the default.
7426 @item -mtext=@var{text section}
7427 @itemx -mdata=@var{data section}
7428 @itemx -mrodata=@var{readonly data section}
7429 Put functions, data, and readonly data in @var{text section},
7430 @var{data section}, and @var{readonly data section} respectively
7431 by default. This can be overridden with the @code{section} attribute.
7432 @xref{Variable Attributes}.
7437 @subsection NS32K Options
7438 @cindex NS32K options
7440 These are the @samp{-m} options defined for the 32000 series. The default
7441 values for these options depends on which style of 32000 was selected when
7442 the compiler was configured; the defaults for the most common choices are
7448 Generate output for a 32032. This is the default
7449 when the compiler is configured for 32032 and 32016 based systems.
7453 Generate output for a 32332. This is the default
7454 when the compiler is configured for 32332-based systems.
7458 Generate output for a 32532. This is the default
7459 when the compiler is configured for 32532-based systems.
7462 Generate output containing 32081 instructions for floating point.
7463 This is the default for all systems.
7466 Generate output containing 32381 instructions for floating point. This
7467 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
7468 and 32532 cpus. This is the default for the pc532-netbsd configuration.
7471 Try and generate multiply-add floating point instructions @code{polyF}
7472 and @code{dotF}. This option is only available if the @samp{-m32381}
7473 option is in effect. Using these instructions requires changes to to
7474 register allocation which generally has a negative impact on
7475 performance. This option should only be enabled when compiling code
7476 particularly likely to make heavy use of multiply-add instructions.
7479 Do not try and generate multiply-add floating point instructions
7480 @code{polyF} and @code{dotF}. This is the default on all platforms.
7483 Generate output containing library calls for floating point.
7484 @strong{Warning:} the requisite libraries may not be available.
7487 Do not use the bit-field instructions. On some machines it is faster to
7488 use shifting and masking operations. This is the default for the pc532.
7491 Do use the bit-field instructions. This is the default for all platforms
7495 Use a different function-calling convention, in which functions
7496 that take a fixed number of arguments return pop their
7497 arguments on return with the @code{ret} instruction.
7499 This calling convention is incompatible with the one normally
7500 used on Unix, so you cannot use it if you need to call libraries
7501 compiled with the Unix compiler.
7503 Also, you must provide function prototypes for all functions that
7504 take variable numbers of arguments (including @code{printf});
7505 otherwise incorrect code will be generated for calls to those
7508 In addition, seriously incorrect code will result if you call a
7509 function with too many arguments. (Normally, extra arguments are
7510 harmlessly ignored.)
7512 This option takes its name from the 680x0 @code{rtd} instruction.
7516 Use a different function-calling convention where the first two arguments
7517 are passed in registers.
7519 This calling convention is incompatible with the one normally
7520 used on Unix, so you cannot use it if you need to call libraries
7521 compiled with the Unix compiler.
7524 Do not pass any arguments in registers. This is the default for all
7528 It is OK to use the sb as an index register which is always loaded with
7529 zero. This is the default for the pc532-netbsd target.
7532 The sb register is not available for use or has not been initialized to
7533 zero by the run time system. This is the default for all targets except
7534 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
7535 @samp{-fpic} is set.
7538 Many ns32000 series addressing modes use displacements of up to 512MB.
7539 If an address is above 512MB then displacements from zero can not be used.
7540 This option causes code to be generated which can be loaded above 512MB.
7541 This may be useful for operating systems or ROM code.
7544 Assume code will be loaded in the first 512MB of virtual address space.
7545 This is the default for all platforms.
7551 @subsection AVR Options
7554 These options are defined for AVR implementations:
7557 @item -mmcu=@var{mcu}
7558 Specify ATMEL AVR instruction set or MCU type.
7560 Instruction set avr1 is for the minimal AVR core, not supported by the C
7561 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7562 attiny11, attiny12, attiny15, attiny28).
7564 Instruction set avr2 (default) is for the classic AVR core with up to
7565 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7566 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7567 at90c8534, at90s8535).
7569 Instruction set avr3 is for the classic AVR core with up to 128K program
7570 memory space (MCU types: atmega103, atmega603).
7572 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7573 memory space (MCU types: atmega83, atmega85).
7575 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7576 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
7579 Output instruction sizes to the asm file.
7581 @item -minit-stack=@var{N}
7582 Specify the initial stack address, which may be a symbol or numeric value,
7583 __stack is the default.
7585 @item -mno-interrupts
7586 Generated code is not compatible with hardware interrupts.
7587 Code size will be smaller.
7589 @item -mcall-prologues
7590 Functions prologues/epilogues expanded as call to appropriate
7591 subroutines. Code size will be smaller.
7593 @item -mno-tablejump
7594 Do not generate tablejump insns which sometimes increase code size.
7597 Change only the low 8 bits of the stack pointer.
7601 @subsection MCore Options
7602 @cindex MCore options
7604 These are the @samp{-m} options defined for the Motorola M*Core
7612 Inline constants into the code stream if it can be done in two
7613 instructions or less.
7618 Use the divide instruction. (Enabled by default).
7620 @item -mrelax-immediate
7621 @itemx -mrelax-immediate
7622 @itemx -mno-relax-immediate
7623 Allow arbitrary sized immediates in bit operations.
7625 @item -mwide-bitfields
7626 @itemx -mwide-bitfields
7627 @itemx -mno-wide-bitfields
7628 Always treat bitfields as int-sized.
7630 @item -m4byte-functions
7631 @itemx -m4byte-functions
7632 @itemx -mno-4byte-functions
7633 Force all functions to be aligned to a four byte boundary.
7635 @item -mcallgraph-data
7636 @itemx -mcallgraph-data
7637 @itemx -mno-callgraph-data
7638 Emit callgraph information.
7642 @itemx -mno-slow-bytes
7643 Prefer word access when reading byte quantities.
7645 @item -mlittle-endian
7646 @itemx -mlittle-endian
7648 Generate code for a little endian target.
7653 Generate code for the 210 processor.
7657 @subsection D30V Options
7658 @cindex D30V Options
7660 These @samp{-m} options are defined for D30V implementations:
7664 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
7665 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
7666 memory, which starts at location @code{0x80000000}.
7669 Same as the @samp{-mextmem} switch.
7672 Link the @samp{.text} section into onchip text memory, which starts at
7673 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
7674 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
7675 into onchip data memory, which starts at location @code{0x20000000}.
7677 @item -mno-asm-optimize
7678 @itemx -masm-optimize
7679 Disable (enable) passing @samp{-O} to the assembler when optimizing.
7680 The assembler uses the @samp{-O} option to automatically parallelize
7681 adjacent short instructions where possible.
7683 @item -mbranch-cost=@var{n}
7684 Increase the internal costs of branches to @var{n}. Higher costs means
7685 that the compiler will issue more instructions to avoid doing a branch.
7688 @item -mcond-exec=@var{n}
7689 Specify the maximum number of conditionally executed instructions that
7690 replace a branch. The default is 4.
7693 @node Code Gen Options
7694 @section Options for Code Generation Conventions
7695 @cindex code generation conventions
7696 @cindex options, code generation
7697 @cindex run-time options
7699 These machine-independent options control the interface conventions
7700 used in code generation.
7702 Most of them have both positive and negative forms; the negative form
7703 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
7704 one of the forms is listed---the one which is not the default. You
7705 can figure out the other form by either removing @samp{no-} or adding
7710 Enable exception handling. Generates extra code needed to propagate
7711 exceptions. For some targets, this implies GNU CC will generate frame
7712 unwind information for all functions, which can produce significant data
7713 size overhead, although it does not affect execution. If you do not
7714 specify this option, GNU CC will enable it by default for languages like
7715 C++ which normally require exception handling, and disable itfor
7716 languages like C that do not normally require it. However, you may need
7717 to enable this option when compiling C code that needs to interoperate
7718 properly with exception handlers written in C++. You may also wish to
7719 disable this option if you are compiling older C++ programs that don't
7720 use exception handling.
7722 @item -funwind-tables
7723 Similar to @option{-fexceptions}, except that it will just generate any needed
7724 static data, but will not affect the generated code in any other way.
7725 You will normally not enable this option; instead, a language processor
7726 that needs this handling would enable it on your behalf.
7728 @item -fpcc-struct-return
7729 Return ``short'' @code{struct} and @code{union} values in memory like
7730 longer ones, rather than in registers. This convention is less
7731 efficient, but it has the advantage of allowing intercallability between
7732 GCC-compiled files and files compiled with other compilers.
7734 The precise convention for returning structures in memory depends
7735 on the target configuration macros.
7737 Short structures and unions are those whose size and alignment match
7738 that of some integer type.
7740 @item -freg-struct-return
7741 Use the convention that @code{struct} and @code{union} values are
7742 returned in registers when possible. This is more efficient for small
7743 structures than @samp{-fpcc-struct-return}.
7745 If you specify neither @samp{-fpcc-struct-return} nor its contrary
7746 @samp{-freg-struct-return}, GCC defaults to whichever convention is
7747 standard for the target. If there is no standard convention, GCC
7748 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
7749 is the principal compiler. In those cases, we can choose the standard,
7750 and we chose the more efficient register return alternative.
7753 Allocate to an @code{enum} type only as many bytes as it needs for the
7754 declared range of possible values. Specifically, the @code{enum} type
7755 will be equivalent to the smallest integer type which has enough room.
7757 @item -fshort-double
7758 Use the same size for @code{double} as for @code{float}.
7761 Requests that the data and non-@code{const} variables of this
7762 compilation be shared data rather than private data. The distinction
7763 makes sense only on certain operating systems, where shared data is
7764 shared between processes running the same program, while private data
7765 exists in one copy per process.
7768 Allocate even uninitialized global variables in the data section of the
7769 object file, rather than generating them as common blocks. This has the
7770 effect that if the same variable is declared (without @code{extern}) in
7771 two different compilations, you will get an error when you link them.
7772 The only reason this might be useful is if you wish to verify that the
7773 program will work on other systems which always work this way.
7776 Ignore the @samp{#ident} directive.
7778 @item -fno-gnu-linker
7779 Do not output global initializations (such as C++ constructors and
7780 destructors) in the form used by the GNU linker (on systems where the GNU
7781 linker is the standard method of handling them). Use this option when
7782 you want to use a non-GNU linker, which also requires using the
7783 @command{collect2} program to make sure the system linker includes
7784 constructors and destructors. (@command{collect2} is included in the GCC
7785 distribution.) For systems which @emph{must} use @command{collect2}, the
7786 compiler driver @command{gcc} is configured to do this automatically.
7788 @item -finhibit-size-directive
7789 Don't output a @code{.size} assembler directive, or anything else that
7790 would cause trouble if the function is split in the middle, and the
7791 two halves are placed at locations far apart in memory. This option is
7792 used when compiling @file{crtstuff.c}; you should not need to use it
7796 Put extra commentary information in the generated assembly code to
7797 make it more readable. This option is generally only of use to those
7798 who actually need to read the generated assembly code (perhaps while
7799 debugging the compiler itself).
7801 @samp{-fno-verbose-asm}, the default, causes the
7802 extra information to be omitted and is useful when comparing two assembler
7806 Consider all memory references through pointers to be volatile.
7808 @item -fvolatile-global
7809 Consider all memory references to extern and global data items to
7810 be volatile. GCC does not consider static data items to be volatile
7811 because of this switch.
7813 @item -fvolatile-static
7814 Consider all memory references to static data to be volatile.
7817 @cindex global offset table
7819 Generate position-independent code (PIC) suitable for use in a shared
7820 library, if supported for the target machine. Such code accesses all
7821 constant addresses through a global offset table (GOT). The dynamic
7822 loader resolves the GOT entries when the program starts (the dynamic
7823 loader is not part of GCC; it is part of the operating system). If
7824 the GOT size for the linked executable exceeds a machine-specific
7825 maximum size, you get an error message from the linker indicating that
7826 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
7827 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
7828 on the m68k and RS/6000. The 386 has no such limit.)
7830 Position-independent code requires special support, and therefore works
7831 only on certain machines. For the 386, GCC supports PIC for System V
7832 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
7833 position-independent.
7836 If supported for the target machine, emit position-independent code,
7837 suitable for dynamic linking and avoiding any limit on the size of the
7838 global offset table. This option makes a difference on the m68k, m88k,
7841 Position-independent code requires special support, and therefore works
7842 only on certain machines.
7844 @item -ffixed-@var{reg}
7845 Treat the register named @var{reg} as a fixed register; generated code
7846 should never refer to it (except perhaps as a stack pointer, frame
7847 pointer or in some other fixed role).
7849 @var{reg} must be the name of a register. The register names accepted
7850 are machine-specific and are defined in the @code{REGISTER_NAMES}
7851 macro in the machine description macro file.
7853 This flag does not have a negative form, because it specifies a
7856 @item -fcall-used-@var{reg}
7857 Treat the register named @var{reg} as an allocable register that is
7858 clobbered by function calls. It may be allocated for temporaries or
7859 variables that do not live across a call. Functions compiled this way
7860 will not save and restore the register @var{reg}.
7862 It is an error to used this flag with the frame pointer or stack pointer.
7863 Use of this flag for other registers that have fixed pervasive roles in
7864 the machine's execution model will produce disastrous results.
7866 This flag does not have a negative form, because it specifies a
7869 @item -fcall-saved-@var{reg}
7870 Treat the register named @var{reg} as an allocable register saved by
7871 functions. It may be allocated even for temporaries or variables that
7872 live across a call. Functions compiled this way will save and restore
7873 the register @var{reg} if they use it.
7875 It is an error to used this flag with the frame pointer or stack pointer.
7876 Use of this flag for other registers that have fixed pervasive roles in
7877 the machine's execution model will produce disastrous results.
7879 A different sort of disaster will result from the use of this flag for
7880 a register in which function values may be returned.
7882 This flag does not have a negative form, because it specifies a
7886 Pack all structure members together without holes. Usually you would
7887 not want to use this option, since it makes the code suboptimal, and
7888 the offsets of structure members won't agree with system libraries.
7890 @item -fcheck-memory-usage
7891 Generate extra code to check each memory access. GCC will generate
7892 code that is suitable for a detector of bad memory accesses such as
7895 Normally, you should compile all, or none, of your code with this option.
7897 If you do mix code compiled with and without this option,
7898 you must ensure that all code that has side effects
7899 and that is called by code compiled with this option
7900 is, itself, compiled with this option.
7901 If you do not, you might get erroneous messages from the detector.
7903 If you use functions from a library that have side-effects (such as
7904 @code{read}), you might not be able to recompile the library and
7905 specify this option. In that case, you can enable the
7906 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
7907 your code and make other functions look as if they were compiled with
7908 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
7909 which are provided by the detector. If you cannot find or build
7910 stubs for every function you call, you might have to specify
7911 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
7913 If you specify this option, you can not use the @code{asm} or
7914 @code{__asm__} keywords in functions with memory checking enabled. GNU
7915 CC cannot understand what the @code{asm} statement may do, and therefore
7916 cannot generate the appropriate code, so it will reject it. However, if
7917 you specify the function attribute @code{no_check_memory_usage}
7918 (@pxref{Function Attributes}), GNU CC will disable memory checking within a
7919 function; you may use @code{asm} statements inside such functions. You
7920 may have an inline expansion of a non-checked function within a checked
7921 function; in that case GNU CC will not generate checks for the inlined
7922 function's memory accesses.
7924 If you move your @code{asm} statements to non-checked inline functions
7925 and they do access memory, you can add calls to the support code in your
7926 inline function, to indicate any reads, writes, or copies being done.
7927 These calls would be similar to those done in the stubs described above.
7929 @item -fprefix-function-name
7930 Request GCC to add a prefix to the symbols generated for function names.
7931 GCC adds a prefix to the names of functions defined as well as
7932 functions called. Code compiled with this option and code compiled
7933 without the option can't be linked together, unless stubs are used.
7935 If you compile the following code with @samp{-fprefix-function-name}
7937 extern void bar (int);
7946 GCC will compile the code as if it was written:
7948 extern void prefix_bar (int);
7952 return prefix_bar (a + 5);
7955 This option is designed to be used with @samp{-fcheck-memory-usage}.
7957 @item -finstrument-functions
7958 Generate instrumentation calls for entry and exit to functions. Just
7959 after function entry and just before function exit, the following
7960 profiling functions will be called with the address of the current
7961 function and its call site. (On some platforms,
7962 @code{__builtin_return_address} does not work beyond the current
7963 function, so the call site information may not be available to the
7964 profiling functions otherwise.)
7967 void __cyg_profile_func_enter (void *this_fn, void *call_site);
7968 void __cyg_profile_func_exit (void *this_fn, void *call_site);
7971 The first argument is the address of the start of the current function,
7972 which may be looked up exactly in the symbol table.
7974 This instrumentation is also done for functions expanded inline in other
7975 functions. The profiling calls will indicate where, conceptually, the
7976 inline function is entered and exited. This means that addressable
7977 versions of such functions must be available. If all your uses of a
7978 function are expanded inline, this may mean an additional expansion of
7979 code size. If you use @samp{extern inline} in your C code, an
7980 addressable version of such functions must be provided. (This is
7981 normally the case anyways, but if you get lucky and the optimizer always
7982 expands the functions inline, you might have gotten away without
7983 providing static copies.)
7985 A function may be given the attribute @code{no_instrument_function}, in
7986 which case this instrumentation will not be done. This can be used, for
7987 example, for the profiling functions listed above, high-priority
7988 interrupt routines, and any functions from which the profiling functions
7989 cannot safely be called (perhaps signal handlers, if the profiling
7990 routines generate output or allocate memory).
7993 Generate code to verify that you do not go beyond the boundary of the
7994 stack. You should specify this flag if you are running in an
7995 environment with multiple threads, but only rarely need to specify it in
7996 a single-threaded environment since stack overflow is automatically
7997 detected on nearly all systems if there is only one stack.
7999 Note that this switch does not actually cause checking to be done; the
8000 operating system must do that. The switch causes generation of code
8001 to ensure that the operating system sees the stack being extended.
8003 @item -fstack-limit-register=@var{reg}
8004 @itemx -fstack-limit-symbol=@var{sym}
8005 @itemx -fno-stack-limit
8006 Generate code to ensure that the stack does not grow beyond a certain value,
8007 either the value of a register or the address of a symbol. If the stack
8008 would grow beyond the value, a signal is raised. For most targets,
8009 the signal is raised before the stack overruns the boundary, so
8010 it is possible to catch the signal without taking special precautions.
8012 For instance, if the stack starts at address @samp{0x80000000} and grows
8013 downwards you can use the flags
8014 @samp{-fstack-limit-symbol=__stack_limit}
8015 @samp{-Wl,--defsym,__stack_limit=0x7ffe0000} which will enforce a stack
8018 @cindex aliasing of parameters
8019 @cindex parameters, aliased
8020 @item -fargument-alias
8021 @itemx -fargument-noalias
8022 @itemx -fargument-noalias-global
8023 Specify the possible relationships among parameters and between
8024 parameters and global data.
8026 @samp{-fargument-alias} specifies that arguments (parameters) may
8027 alias each other and may alias global storage.
8028 @samp{-fargument-noalias} specifies that arguments do not alias
8029 each other, but may alias global storage.
8030 @samp{-fargument-noalias-global} specifies that arguments do not
8031 alias each other and do not alias global storage.
8033 Each language will automatically use whatever option is required by
8034 the language standard. You should not need to use these options yourself.
8036 @item -fleading-underscore
8037 This option and its counterpart, -fno-leading-underscore, forcibly
8038 change the way C symbols are represented in the object file. One use
8039 is to help link with legacy assembly code.
8041 Be warned that you should know what you are doing when invoking this
8042 option, and that not all targets provide complete support for it.
8047 @node Environment Variables
8048 @section Environment Variables Affecting GCC
8049 @cindex environment variables
8051 @c man begin ENVIRONMENT
8053 This section describes several environment variables that affect how GCC
8054 operates. Some of them work by specifying directories or prefixes to use
8055 when searching for various kinds of files. Some are used to specify other
8056 aspects of the compilation environment.
8059 Note that you can also specify places to search using options such as
8060 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
8061 take precedence over places specified using environment variables, which
8062 in turn take precedence over those specified by the configuration of GCC.
8066 Note that you can also specify places to search using options such as
8067 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
8068 take precedence over places specified using environment variables, which
8069 in turn take precedence over those specified by the configuration of GCC.
8076 @c @itemx LC_COLLATE
8078 @c @itemx LC_MONETARY
8079 @c @itemx LC_NUMERIC
8084 @c @findex LC_COLLATE
8086 @c @findex LC_MONETARY
8087 @c @findex LC_NUMERIC
8091 These environment variables control the way that GCC uses
8092 localization information that allow GCC to work with different
8093 national conventions. GCC inspects the locale categories
8094 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
8095 so. These locale categories can be set to any value supported by your
8096 installation. A typical value is @samp{en_UK} for English in the United
8099 The @env{LC_CTYPE} environment variable specifies character
8100 classification. GCC uses it to determine the character boundaries in
8101 a string; this is needed for some multibyte encodings that contain quote
8102 and escape characters that would otherwise be interpreted as a string
8105 The @env{LC_MESSAGES} environment variable specifies the language to
8106 use in diagnostic messages.
8108 If the @env{LC_ALL} environment variable is set, it overrides the value
8109 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
8110 and @env{LC_MESSAGES} default to the value of the @env{LANG}
8111 environment variable. If none of these variables are set, GCC
8112 defaults to traditional C English behavior.
8116 If @env{TMPDIR} is set, it specifies the directory to use for temporary
8117 files. GCC uses temporary files to hold the output of one stage of
8118 compilation which is to be used as input to the next stage: for example,
8119 the output of the preprocessor, which is the input to the compiler
8122 @item GCC_EXEC_PREFIX
8123 @findex GCC_EXEC_PREFIX
8124 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
8125 names of the subprograms executed by the compiler. No slash is added
8126 when this prefix is combined with the name of a subprogram, but you can
8127 specify a prefix that ends with a slash if you wish.
8129 If @env{GCC_EXEC_PREFIX} is not set, GNU CC will attempt to figure out
8130 an appropriate prefix to use based on the pathname it was invoked with.
8132 If GCC cannot find the subprogram using the specified prefix, it
8133 tries looking in the usual places for the subprogram.
8135 The default value of @env{GCC_EXEC_PREFIX} is
8136 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
8137 of @code{prefix} when you ran the @file{configure} script.
8139 Other prefixes specified with @samp{-B} take precedence over this prefix.
8141 This prefix is also used for finding files such as @file{crt0.o} that are
8144 In addition, the prefix is used in an unusual way in finding the
8145 directories to search for header files. For each of the standard
8146 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
8147 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
8148 replacing that beginning with the specified prefix to produce an
8149 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
8150 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
8151 These alternate directories are searched first; the standard directories
8155 @findex COMPILER_PATH
8156 The value of @env{COMPILER_PATH} is a colon-separated list of
8157 directories, much like @env{PATH}. GCC tries the directories thus
8158 specified when searching for subprograms, if it can't find the
8159 subprograms using @env{GCC_EXEC_PREFIX}.
8162 @findex LIBRARY_PATH
8163 The value of @env{LIBRARY_PATH} is a colon-separated list of
8164 directories, much like @env{PATH}. When configured as a native compiler,
8165 GCC tries the directories thus specified when searching for special
8166 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
8167 using GCC also uses these directories when searching for ordinary
8168 libraries for the @samp{-l} option (but directories specified with
8169 @samp{-L} come first).
8171 @item C_INCLUDE_PATH
8172 @itemx CPLUS_INCLUDE_PATH
8173 @itemx OBJC_INCLUDE_PATH
8174 @findex C_INCLUDE_PATH
8175 @findex CPLUS_INCLUDE_PATH
8176 @findex OBJC_INCLUDE_PATH
8177 @c @itemx OBJCPLUS_INCLUDE_PATH
8178 These environment variables pertain to particular languages. Each
8179 variable's value is a colon-separated list of directories, much like
8180 @env{PATH}. When GCC searches for header files, it tries the
8181 directories listed in the variable for the language you are using, after
8182 the directories specified with @samp{-I} but before the standard header
8185 @item DEPENDENCIES_OUTPUT
8186 @findex DEPENDENCIES_OUTPUT
8187 @cindex dependencies for make as output
8188 If this variable is set, its value specifies how to output dependencies
8189 for Make based on the header files processed by the compiler. This
8190 output looks much like the output from the @samp{-M} option
8191 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
8192 in addition to the usual results of compilation.
8194 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
8195 which case the Make rules are written to that file, guessing the target
8196 name from the source file name. Or the value can have the form
8197 @samp{@var{file} @var{target}}, in which case the rules are written to
8198 file @var{file} using @var{target} as the target name.
8202 @cindex locale definition
8203 This variable is used to pass locale information to the compiler. One way in
8204 which this information is used is to determine the character set to be used
8205 when character literals, string literals and comments are parsed in C and C++.
8206 When the compiler is configured to allow multibyte characters,
8207 the following values for @env{LANG} are recognized:
8211 Recognize JIS characters.
8213 Recognize SJIS characters.
8215 Recognize EUCJP characters.
8218 If @env{LANG} is not defined, or if it has some other value, then the
8219 compiler will use mblen and mbtowc as defined by the default locale to
8220 recognize and translate multibyte characters.
8225 @node Running Protoize
8226 @section Running Protoize
8228 The program @code{protoize} is an optional part of GNU C. You can use
8229 it to add prototypes to a program, thus converting the program to ISO
8230 C in one respect. The companion program @code{unprotoize} does the
8231 reverse: it removes argument types from any prototypes that are found.
8233 When you run these programs, you must specify a set of source files as
8234 command line arguments. The conversion programs start out by compiling
8235 these files to see what functions they define. The information gathered
8236 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
8238 After scanning comes actual conversion. The specified files are all
8239 eligible to be converted; any files they include (whether sources or
8240 just headers) are eligible as well.
8242 But not all the eligible files are converted. By default,
8243 @code{protoize} and @code{unprotoize} convert only source and header
8244 files in the current directory. You can specify additional directories
8245 whose files should be converted with the @samp{-d @var{directory}}
8246 option. You can also specify particular files to exclude with the
8247 @samp{-x @var{file}} option. A file is converted if it is eligible, its
8248 directory name matches one of the specified directory names, and its
8249 name within the directory has not been excluded.
8251 Basic conversion with @code{protoize} consists of rewriting most
8252 function definitions and function declarations to specify the types of
8253 the arguments. The only ones not rewritten are those for varargs
8256 @code{protoize} optionally inserts prototype declarations at the
8257 beginning of the source file, to make them available for any calls that
8258 precede the function's definition. Or it can insert prototype
8259 declarations with block scope in the blocks where undeclared functions
8262 Basic conversion with @code{unprotoize} consists of rewriting most
8263 function declarations to remove any argument types, and rewriting
8264 function definitions to the old-style pre-ISO form.
8266 Both conversion programs print a warning for any function declaration or
8267 definition that they can't convert. You can suppress these warnings
8270 The output from @code{protoize} or @code{unprotoize} replaces the
8271 original source file. The original file is renamed to a name ending
8272 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
8273 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
8274 for DOS) file already exists, then the source file is simply discarded.
8276 @code{protoize} and @code{unprotoize} both depend on GCC itself to
8277 scan the program and collect information about the functions it uses.
8278 So neither of these programs will work until GCC is installed.
8280 Here is a table of the options you can use with @code{protoize} and
8281 @code{unprotoize}. Each option works with both programs unless
8285 @item -B @var{directory}
8286 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
8287 usual directory (normally @file{/usr/local/lib}). This file contains
8288 prototype information about standard system functions. This option
8289 applies only to @code{protoize}.
8291 @item -c @var{compilation-options}
8292 Use @var{compilation-options} as the options when running @code{gcc} to
8293 produce the @samp{.X} files. The special option @samp{-aux-info} is
8294 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
8296 Note that the compilation options must be given as a single argument to
8297 @code{protoize} or @code{unprotoize}. If you want to specify several
8298 @code{gcc} options, you must quote the entire set of compilation options
8299 to make them a single word in the shell.
8301 There are certain @code{gcc} arguments that you cannot use, because they
8302 would produce the wrong kind of output. These include @samp{-g},
8303 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
8304 the @var{compilation-options}, they are ignored.
8307 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
8308 systems) instead of @samp{.c}. This is convenient if you are converting
8309 a C program to C++. This option applies only to @code{protoize}.
8312 Add explicit global declarations. This means inserting explicit
8313 declarations at the beginning of each source file for each function
8314 that is called in the file and was not declared. These declarations
8315 precede the first function definition that contains a call to an
8316 undeclared function. This option applies only to @code{protoize}.
8318 @item -i @var{string}
8319 Indent old-style parameter declarations with the string @var{string}.
8320 This option applies only to @code{protoize}.
8322 @code{unprotoize} converts prototyped function definitions to old-style
8323 function definitions, where the arguments are declared between the
8324 argument list and the initial @samp{@{}. By default, @code{unprotoize}
8325 uses five spaces as the indentation. If you want to indent with just
8326 one space instead, use @samp{-i " "}.
8329 Keep the @samp{.X} files. Normally, they are deleted after conversion
8333 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
8334 a prototype declaration for each function in each block which calls the
8335 function without any declaration. This option applies only to
8339 Make no real changes. This mode just prints information about the conversions
8340 that would have been done without @samp{-n}.
8343 Make no @samp{.save} files. The original files are simply deleted.
8344 Use this option with caution.
8346 @item -p @var{program}
8347 Use the program @var{program} as the compiler. Normally, the name
8351 Work quietly. Most warnings are suppressed.
8354 Print the version number, just like @samp{-v} for @code{gcc}.
8357 If you need special compiler options to compile one of your program's
8358 source files, then you should generate that file's @samp{.X} file
8359 specially, by running @code{gcc} on that source file with the
8360 appropriate options and the option @samp{-aux-info}. Then run
8361 @code{protoize} on the entire set of files. @code{protoize} will use
8362 the existing @samp{.X} file because it is newer than the source file.
8366 gcc -Dfoo=bar file1.c -aux-info
8371 You need to include the special files along with the rest in the
8372 @code{protoize} command, even though their @samp{.X} files already
8373 exist, because otherwise they won't get converted.
8375 @xref{Protoize Caveats}, for more information on how to use
8376 @code{protoize} successfully.