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}.
48 @chapter GCC Command Options
49 @cindex GCC command options
50 @cindex command options
51 @cindex options, GCC command
53 @c man begin DESCRIPTION
55 When you invoke GCC, it normally does preprocessing, compilation,
56 assembly and linking. The ``overall options'' allow you to stop this
57 process at an intermediate stage. For example, the @samp{-c} option
58 says not to run the linker. Then the output consists of object files
59 output by the assembler.
61 Other options are passed on to one stage of processing. Some options
62 control the preprocessor and others the compiler itself. Yet other
63 options control the assembler and linker; most of these are not
64 documented here, since you rarely need to use any of them.
66 @cindex C compilation options
67 Most of the command line options that you can use with GCC are useful
68 for C programs; when an option is only useful with another language
69 (usually C++), the explanation says so explicitly. If the description
70 for a particular option does not mention a source language, you can use
71 that option with all supported languages.
73 @cindex C++ compilation options
74 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
75 options for compiling C++ programs.
77 @cindex grouping options
78 @cindex options, grouping
79 The @code{gcc} program accepts options and file names as operands. Many
80 options have multi-letter names; therefore multiple single-letter options
81 may @emph{not} be grouped: @samp{-dr} is very different from @w{@samp{-d
84 @cindex order of options
85 @cindex options, order
86 You can mix options and other arguments. For the most part, the order
87 you use doesn't matter. Order does matter when you use several options
88 of the same kind; for example, if you specify @samp{-L} more than once,
89 the directories are searched in the order specified.
91 Many options have long names starting with @samp{-f} or with
92 @samp{-W}---for example, @samp{-fforce-mem},
93 @samp{-fstrength-reduce}, @samp{-Wformat} and so on. Most of
94 these have both positive and negative forms; the negative form of
95 @samp{-ffoo} would be @samp{-fno-foo}. This manual documents
96 only one of these two forms, whichever one is not the default.
101 * Option Summary:: Brief list of all options, without explanations.
102 * Overall Options:: Controlling the kind of output:
103 an executable, object files, assembler files,
104 or preprocessed source.
105 * Invoking G++:: Compiling C++ programs.
106 * C Dialect Options:: Controlling the variant of C language compiled.
107 * C++ Dialect Options:: Variations on C++.
108 * Language Independent Options:: Controlling how diagnostics should be
110 * Warning Options:: How picky should the compiler be?
111 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
112 * Optimize Options:: How much optimization?
113 * Preprocessor Options:: Controlling header files and macro definitions.
114 Also, getting dependency information for Make.
115 * Assembler Options:: Passing options to the assembler.
116 * Link Options:: Specifying libraries and so on.
117 * Directory Options:: Where to find header files and libraries.
118 Where to find the compiler executable files.
119 * Spec Files:: How to pass switches to sub-processes.
120 * Target Options:: Running a cross-compiler, or an old version of GCC.
121 * Submodel Options:: Specifying minor hardware or convention variations,
122 such as 68010 vs 68020.
123 * Code Gen Options:: Specifying conventions for function calls, data layout
125 * Environment Variables:: Env vars that affect GCC.
126 * Running Protoize:: Automatically adding or removing function prototypes.
130 @section Option Summary
132 Here is a summary of all the options, grouped by type. Explanations are
133 in the following sections.
136 @item Overall Options
137 @xref{Overall Options,,Options Controlling the Kind of Output}.
139 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language}
140 -v --target-help --help
143 @item C Language Options
144 @xref{C Dialect Options,,Options Controlling C Dialect}.
146 -ansi -std=@var{standard} -fno-asm -fno-builtin
147 -fhosted -ffreestanding
148 -trigraphs -traditional -traditional-cpp
149 -fallow-single-precision -fcond-mismatch
150 -fsigned-bitfields -fsigned-char
151 -funsigned-bitfields -funsigned-char
152 -fwritable-strings -fshort-wchar
155 @item C++ Language Options
156 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
158 -fno-access-control -fcheck-new -fconserve-space
159 -fdollars-in-identifiers -fno-elide-constructors
160 -fno-enforce-eh-specs -fexternal-templates
161 -falt-external-templates
162 -ffor-scope -fno-for-scope -fno-gnu-keywords -fhonor-std
163 -fhuge-objects -fno-implicit-templates
164 -fno-implicit-inline-templates
165 -fno-implement-inlines -fms-extensions
166 -fname-mangling-version-@var{n} -fno-operator-names
167 -fno-optional-diags -fpermissive
168 -frepo -fno-rtti -fsquangle -ftemplate-depth-@var{n}
169 -fuse-cxa-atexit -fvtable-thunks -nostdinc++
170 -fno-default-inline -Wctor-dtor-privacy
171 -Wnon-virtual-dtor -Wreorder
172 -Weffc++ -Wno-deprecated
173 -Wno-non-template-friend -Wold-style-cast
174 -Woverloaded-virtual -Wno-pmf-conversions
178 @item Language Independent Options
179 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
181 -fmessage-length=@var{n}
182 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}
185 @item Warning Options
186 @xref{Warning Options,,Options to Request or Suppress Warnings}.
188 -fsyntax-only -pedantic -pedantic-errors
189 -w -W -Wall -Waggregate-return
190 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment
191 -Wconversion -Wdisabled-optimization -Werror
192 -Wfloat-equal -Wformat -Wformat=2
193 -Wformat-nonliteral -Wformat-security
194 -Wid-clash-@var{len} -Wimplicit -Wimplicit-int
195 -Wimplicit-function-declaration
196 -Werror-implicit-function-declaration
198 -Wlarger-than-@var{len} -Wlong-long
199 -Wmain -Wmissing-declarations
200 -Wmissing-format-attribute -Wmissing-noreturn
201 -Wmultichar -Wno-format-extra-args -Wno-format-y2k
202 -Wno-import -Wpacked -Wpadded
203 -Wparentheses -Wpointer-arith -Wredundant-decls
204 -Wreturn-type -Wsequence-point -Wshadow
205 -Wsign-compare -Wswitch -Wsystem-headers
206 -Wtrigraphs -Wundef -Wuninitialized
207 -Wunknown-pragmas -Wunreachable-code
208 -Wunused -Wunused-function -Wunused-label -Wunused-parameter
209 -Wunused-value -Wunused-variable -Wwrite-strings
212 @item C-only Warning Options
214 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs
215 -Wstrict-prototypes -Wtraditional
218 @item Debugging Options
219 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
221 -a -ax -d@var{letters} -fdump-unnumbered -fdump-translation-unit-@var{file}
222 -fpretend-float -fprofile-arcs -ftest-coverage
223 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2
224 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+
225 -p -pg -print-file-name=@var{library} -print-libgcc-file-name
226 -print-prog-name=@var{program} -print-search-dirs -Q
230 @item Optimization Options
231 @xref{Optimize Options,,Options that Control Optimization}.
233 -falign-functions=@var{n} -falign-jumps=@var{n}
234 -falign-labels=@var{n} -falign-loops=@var{n}
235 -fbranch-probabilities -fcaller-saves
236 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections -fdce
237 -fdelayed-branch -fdelete-null-pointer-checks
238 -fexpensive-optimizations -ffast-math -ffloat-store
239 -fforce-addr -fforce-mem -ffunction-sections -fgcse
240 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions
241 -fkeep-static-consts -fmove-all-movables
242 -fno-default-inline -fno-defer-pop
243 -fno-function-cse -fno-inline -fno-math-errno -fno-peephole
244 -fomit-frame-pointer -foptimize-register-move
245 -foptimize-sibling-calls -freduce-all-givs
246 -fregmove -frename-registers
247 -frerun-cse-after-loop -frerun-loop-opt
248 -fschedule-insns -fschedule-insns2
249 -fsingle-precision-constant -fssa
250 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv
251 -funroll-all-loops -funroll-loops
252 -O -O0 -O1 -O2 -O3 -Os
255 @item Preprocessor Options
256 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
258 -A@var{question}(@var{answer}) -C -dD -dM -dN
259 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H
261 -include @var{file} -imacros @var{file}
262 -iprefix @var{file} -iwithprefix @var{dir}
263 -iwithprefixbefore @var{dir} -isystem @var{dir} -isystem-c++ @var{dir}
264 -M -MD -MM -MMD -MG -nostdinc -P -trigraphs
265 -undef -U@var{macro} -Wp,@var{option}
268 @item Assembler Option
269 @xref{Assembler Options,,Passing Options to the Assembler}.
275 @xref{Link Options,,Options for Linking}.
277 @var{object-file-name} -l@var{library}
278 -nostartfiles -nodefaultlibs -nostdlib
279 -s -static -shared -symbolic
280 -Wl,@var{option} -Xlinker @var{option}
284 @item Directory Options
285 @xref{Directory Options,,Options for Directory Search}.
287 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}
291 @c I wrote this xref this way to avoid overfull hbox. -- rms
292 @xref{Target Options}.
294 -b @var{machine} -V @var{version}
297 @item Machine Dependent Options
298 @xref{Submodel Options,,Hardware Models and Configurations}.
300 @emph{M680x0 Options}
301 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040
302 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020
303 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel
304 -malign-int -mstrict-align
306 @emph{M68hc1x Options}
307 -m6811 -m6812 -m68hc11 -m68hc12
308 -mauto-incdec -mshort -msoft-reg-count=@var{count}
315 -mtune=@var{cpu type}
316 -mcmodel=@var{code model}
318 -mapp-regs -mbroken-saverestore -mcypress
319 -mepilogue -mfaster-structs -mflat
320 -mfpu -mhard-float -mhard-quad-float
321 -mimpure-text -mlive-g0 -mno-app-regs
322 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu
323 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles
324 -msoft-float -msoft-quad-float -msparclite -mstack-bias
325 -msupersparc -munaligned-doubles -mv8
327 @emph{Convex Options}
328 -mc1 -mc2 -mc32 -mc34 -mc38
329 -margcount -mnoargcount
331 -mvolatile-cache -mvolatile-nocache
333 @emph{AMD29K Options}
334 -m29000 -m29050 -mbw -mnbw -mdw -mndw
335 -mlarge -mnormal -msmall
336 -mkernel-registers -mno-reuse-arg-regs
337 -mno-stack-check -mno-storem-bug
338 -mreuse-arg-regs -msoft-float -mstack-check
339 -mstorem-bug -muser-registers
342 -mapcs-frame -mno-apcs-frame
344 -mapcs-stack-check -mno-apcs-stack-check
345 -mapcs-float -mno-apcs-float
346 -mapcs-reentrant -mno-apcs-reentrant
347 -msched-prolog -mno-sched-prolog
348 -mlittle-endian -mbig-endian -mwords-little-endian
349 -malignment-traps -mno-alignment-traps
350 -msoft-float -mhard-float -mfpe
351 -mthumb-interwork -mno-thumb-interwork
352 -mcpu= -march= -mfpe=
353 -mstructure-size-boundary=
354 -mbsd -mxopen -mno-symrename
356 -mlong-calls -mno-long-calls
357 -mnop-fun-dllimport -mno-nop-fun-dllimport
358 -msingle-pic-base -mno-single-pic-base
362 -mtpcs-frame -mno-tpcs-frame
363 -mtpcs-leaf-frame -mno-tpcs-leaf-frame
364 -mlittle-endian -mbig-endian
365 -mthumb-interwork -mno-thumb-interwork
366 -mstructure-size-boundary=
367 -mnop-fun-dllimport -mno-nop-fun-dllimport
368 -mcallee-super-interworking -mno-callee-super-interworking
369 -mcaller-super-interworking -mno-caller-super-interworking
370 -msingle-pic-base -mno-single-pic-base
373 @emph{MN10200 Options}
376 @emph{MN10300 Options}
383 @emph{M32R/D Options}
384 -mcode-model=@var{model type} -msdata=@var{sdata type}
388 -m88000 -m88100 -m88110 -mbig-pic
389 -mcheck-zero-division -mhandle-large-shift
390 -midentify-revision -mno-check-zero-division
391 -mno-ocs-debug-info -mno-ocs-frame-position
392 -mno-optimize-arg-area -mno-serialize-volatile
393 -mno-underscores -mocs-debug-info
394 -mocs-frame-position -moptimize-arg-area
395 -mserialize-volatile -mshort-data-@var{num} -msvr3
396 -msvr4 -mtrap-large-shift -muse-div-instruction
397 -mversion-03.00 -mwarn-passed-structs
399 @emph{RS/6000 and PowerPC Options}
401 -mtune=@var{cpu type}
402 -mpower -mno-power -mpower2 -mno-power2
403 -mpowerpc -mpowerpc64 -mno-powerpc
404 -mpowerpc-gpopt -mno-powerpc-gpopt
405 -mpowerpc-gfxopt -mno-powerpc-gfxopt
406 -mnew-mnemonics -mold-mnemonics
407 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc
408 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe
409 -msoft-float -mhard-float -mmultiple -mno-multiple
410 -mstring -mno-string -mupdate -mno-update
411 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align
412 -mstrict-align -mno-strict-align -mrelocatable
413 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib
414 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian
415 -mcall-aix -mcall-sysv -mprototype -mno-prototype
416 -msim -mmvme -mads -myellowknife -memb -msdata
417 -msdata=@var{opt} -mvxworks -G @var{num}
420 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs
421 -mfull-fp-blocks -mhc-struct-return -min-line-mul
422 -mminimum-fp-blocks -mnohc-struct-return
425 -mabicalls -mcpu=@var{cpu type}
426 -membedded-data -muninit-const-in-rodata
427 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64
428 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1
429 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy
430 -mmips-as -mmips-tfile -mno-abicalls
431 -mno-embedded-data -mno-uninit-const-in-rodata -mno-embedded-pic
432 -mno-gpopt -mno-long-calls
433 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats
434 -mrnames -msoft-float
435 -m4650 -msingle-float -mmad
436 -mstats -EL -EB -G @var{num} -nocpp
437 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi
441 -mcpu=@var{cpu type} -march=@var{cpu type}
442 -mintel-syntax -mieee-fp -mno-fancy-math-387
443 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib
444 -mno-wide-multiply -mrtd -malign-double
445 -mreg-alloc=@var{list} -mregparm=@var{num}
446 -malign-jumps=@var{num} -malign-loops=@var{num}
447 -malign-functions=@var{num} -mpreferred-stack-boundary=@var{num}
448 -mthreads -mno-align-stringops -minline-all-stringops
449 -mpush-args -maccumulate-outgoing-args -m128bit-long-double
453 -march=@var{architecture type}
454 -mbig-switch -mdisable-fpregs -mdisable-indexing
455 -mfast-indirect-calls -mgas -mjump-in-delay
456 -mlong-load-store -mno-big-switch -mno-disable-fpregs
457 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas
458 -mno-jump-in-delay -mno-long-load-store
459 -mno-portable-runtime -mno-soft-float
460 -mno-space-regs -msoft-float -mpa-risc-1-0
461 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime
462 -mschedule=@var{cpu type} -mspace-regs
464 @emph{Intel 960 Options}
465 -m@var{cpu type} -masm-compat -mclean-linkage
466 -mcode-align -mcomplex-addr -mleaf-procedures
467 -mic-compat -mic2.0-compat -mic3.0-compat
468 -mintel-asm -mno-clean-linkage -mno-code-align
469 -mno-complex-addr -mno-leaf-procedures
470 -mno-old-align -mno-strict-align -mno-tail-call
471 -mnumerics -mold-align -msoft-float -mstrict-align
474 @emph{DEC Alpha Options}
475 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float
477 -mieee -mieee-with-inexact -mieee-conformant
478 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode}
479 -mtrap-precision=@var{mode} -mbuild-constants
481 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max
482 -mmemory-latency=@var{time}
484 @emph{Clipper Options}
487 @emph{H8/300 Options}
488 -mrelax -mh -ms -mint32 -malign-300
492 -m4-nofpu -m4-single-only -m4-single -m4
493 -mb -ml -mdalign -mrelax
494 -mbigtable -mfmovd -mhitachi -mnomacsave
495 -misize -mpadstruct -mspace
499 @emph{System V Options}
500 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}
504 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section}
505 -mdata=@var{data section} -mrodata=@var{readonly data section}
507 @emph{TMS320C3x/C4x Options}
508 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm
509 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload
510 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned
511 -mparallel-insns -mparallel-mpy -mpreserve-float
514 -mlong-calls -mno-long-calls -mep -mno-ep
515 -mprolog-function -mno-prolog-function -mspace
516 -mtda=@var{n} -msda=@var{n} -mzda=@var{n}
520 -m32032 -m32332 -m32532 -m32081 -m32381 -mmult-add -mnomult-add
521 -msoft-float -mrtd -mnortd -mregparam -mnoregparam -msb -mnosb
522 -mbitfield -mnobitfield -mhimem -mnohimem
525 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts
526 -mcall-prologues -mno-tablejump -mtiny-stack
529 -mhardlit, -mno-hardlit -mdiv -mno-div -mrelax-immediates
530 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields
531 -m4byte-functions -mno-4byte-functions -mcallgraph-data
532 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim
533 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment
536 @item Code Generation Options
537 @xref{Code Gen Options,,Options for Code Generation Conventions}.
539 -fcall-saved-@var{reg} -fcall-used-@var{reg}
540 -fexceptions -funwind-tables -ffixed-@var{reg}
541 -finhibit-size-directive -finstrument-functions
542 -fcheck-memory-usage -fprefix-function-name
543 -fno-common -fno-ident -fno-gnu-linker
544 -fpcc-struct-return -fpic -fPIC
545 -freg-struct-return -fshared-data -fshort-enums
546 -fshort-double -fvolatile -fvolatile-global -fvolatile-static
547 -fverbose-asm -fpack-struct -fstack-check
548 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym}
549 -fargument-alias -fargument-noalias
550 -fargument-noalias-global
556 * Overall Options:: Controlling the kind of output:
557 an executable, object files, assembler files,
558 or preprocessed source.
559 * C Dialect Options:: Controlling the variant of C language compiled.
560 * C++ Dialect Options:: Variations on C++.
561 * Language Independent Options:: Controlling how diagnostics should be
563 * Warning Options:: How picky should the compiler be?
564 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
565 * Optimize Options:: How much optimization?
566 * Preprocessor Options:: Controlling header files and macro definitions.
567 Also, getting dependency information for Make.
568 * Assembler Options:: Passing options to the assembler.
569 * Link Options:: Specifying libraries and so on.
570 * Directory Options:: Where to find header files and libraries.
571 Where to find the compiler executable files.
572 * Spec Files:: How to pass switches to sub-processes.
573 * Target Options:: Running a cross-compiler, or an old version of GCC.
578 @node Overall Options
579 @section Options Controlling the Kind of Output
581 Compilation can involve up to four stages: preprocessing, compilation
582 proper, assembly and linking, always in that order. The first three
583 stages apply to an individual source file, and end by producing an
584 object file; linking combines all the object files (those newly
585 compiled, and those specified as input) into an executable file.
587 @cindex file name suffix
588 For any given input file, the file name suffix determines what kind of
593 C source code which must be preprocessed.
596 C source code which should not be preprocessed.
599 C++ source code which should not be preprocessed.
602 Objective-C source code. Note that you must link with the library
603 @file{libobjc.a} to make an Objective-C program work.
606 Objective-C source code which should not be preprocessed.
609 C header file (not to be compiled or linked).
613 @itemx @var{file}.cxx
614 @itemx @var{file}.cpp
615 @itemx @var{file}.c++
617 C++ source code which must be preprocessed. Note that in @samp{.cxx},
618 the last two letters must both be literally @samp{x}. Likewise,
619 @samp{.C} refers to a literal capital C.
622 @itemx @var{file}.for
623 @itemx @var{file}.FOR
624 Fortran source code which should not be preprocessed.
627 @itemx @var{file}.fpp
628 @itemx @var{file}.FPP
629 Fortran source code which must be preprocessed (with the traditional
633 Fortran source code which must be preprocessed with a RATFOR
634 preprocessor (not included with GCC).
636 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
637 Using and Porting GNU Fortran}, for more details of the handling of
640 @c FIXME: Descriptions of Java file types.
646 @c GCC also knows about some suffixes for languages not yet included:
656 @itemx @var{file}.chi
657 CHILL source code (preprocessed with the traditional preprocessor).
663 Assembler code which must be preprocessed.
666 An object file to be fed straight into linking.
667 Any file name with no recognized suffix is treated this way.
670 You can specify the input language explicitly with the @samp{-x} option:
673 @item -x @var{language}
674 Specify explicitly the @var{language} for the following input files
675 (rather than letting the compiler choose a default based on the file
676 name suffix). This option applies to all following input files until
677 the next @samp{-x} option. Possible values for @var{language} are:
679 c c-header cpp-output
681 objective-c objc-cpp-output
682 assembler assembler-with-cpp
683 f77 f77-cpp-input ratfor
686 @c Also f77-version, for internal use only.
689 Turn off any specification of a language, so that subsequent files are
690 handled according to their file name suffixes (as they are if @samp{-x}
691 has not been used at all).
693 @item -pass-exit-codes
694 Normally the @code{gcc} program will exit with the code of 1 if any
695 phase of the compiler returns a non-success return code. If you specify
696 @samp{-pass-exit-codes}, the @code{gcc} program will instead return with
697 numerically highest error produced by any phase that returned an error
701 If you only want some of the stages of compilation, you can use
702 @samp{-x} (or filename suffixes) to tell @code{gcc} where to start, and
703 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
704 @code{gcc} is to stop. Note that some combinations (for example,
705 @samp{-x cpp-output -E} instruct @code{gcc} to do nothing at all.
709 Compile or assemble the source files, but do not link. The linking
710 stage simply is not done. The ultimate output is in the form of an
711 object file for each source file.
713 By default, the object file name for a source file is made by replacing
714 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
716 Unrecognized input files, not requiring compilation or assembly, are
720 Stop after the stage of compilation proper; do not assemble. The output
721 is in the form of an assembler code file for each non-assembler input
724 By default, the assembler file name for a source file is made by
725 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
727 Input files that don't require compilation are ignored.
730 Stop after the preprocessing stage; do not run the compiler proper. The
731 output is in the form of preprocessed source code, which is sent to the
734 Input files which don't require preprocessing are ignored.
736 @cindex output file option
738 Place output in file @var{file}. This applies regardless to whatever
739 sort of output is being produced, whether it be an executable file,
740 an object file, an assembler file or preprocessed C code.
742 Since only one output file can be specified, it does not make sense to
743 use @samp{-o} when compiling more than one input file, unless you are
744 producing an executable file as output.
746 If @samp{-o} is not specified, the default is to put an executable file
747 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
748 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
749 all preprocessed C source on standard output.@refill
752 Print (on standard error output) the commands executed to run the stages
753 of compilation. Also print the version number of the compiler driver
754 program and of the preprocessor and the compiler proper.
757 Use pipes rather than temporary files for communication between the
758 various stages of compilation. This fails to work on some systems where
759 the assembler is unable to read from a pipe; but the GNU assembler has
763 Print (on the standard output) a description of the command line options
764 understood by @code{gcc}. If the @code{-v} option is also specified
765 then @code{--help} will also be passed on to the various processes
766 invoked by @code{gcc}, so that they can display the command line options
767 they accept. If the @code{-W} option is also specified then command
768 line options which have no documentation associated with them will also
772 Print (on the standard output) a description of target specific command
773 line options for each tool.
777 @section Compiling C++ Programs
779 @cindex suffixes for C++ source
780 @cindex C++ source file suffixes
781 C++ source files conventionally use one of the suffixes @samp{.C},
782 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
783 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
784 files with these names and compiles them as C++ programs even if you
785 call the compiler the same way as for compiling C programs (usually with
786 the name @code{gcc}).
790 However, C++ programs often require class libraries as well as a
791 compiler that understands the C++ language---and under some
792 circumstances, you might want to compile programs from standard input,
793 or otherwise without a suffix that flags them as C++ programs.
794 @code{g++} is a program that calls GCC with the default language
795 set to C++, and automatically specifies linking against the C++
796 library. On many systems, @code{g++} is also
797 installed with the name @code{c++}.
799 @cindex invoking @code{g++}
800 When you compile C++ programs, you may specify many of the same
801 command-line options that you use for compiling programs in any
802 language; or command-line options meaningful for C and related
803 languages; or options that are meaningful only for C++ programs.
804 @xref{C Dialect Options,,Options Controlling C Dialect}, for
805 explanations of options for languages related to C.
806 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
807 explanations of options that are meaningful only for C++ programs.
809 @node C Dialect Options
810 @section Options Controlling C Dialect
811 @cindex dialect options
812 @cindex language dialect options
813 @cindex options, dialect
815 The following options control the dialect of C (or languages derived
816 from C, such as C++ and Objective C) that the compiler accepts:
822 In C mode, support all ISO C89 programs. In C++ mode,
823 remove GNU extensions that conflict with ISO C++.
825 This turns off certain features of GCC that are incompatible with ISO
826 C (when compiling C code), or of standard C++ (when compiling C++ code),
827 such as the @code{asm} and @code{typeof} keywords, and
828 predefined macros such as @code{unix} and @code{vax} that identify the
829 type of system you are using. It also enables the undesirable and
830 rarely used ISO trigraph feature. For the C compiler,
831 it disables recognition of C++ style @samp{//} comments as well as
832 the @code{inline} keyword.
834 The alternate keywords @code{__asm__}, @code{__extension__},
835 @code{__inline__} and @code{__typeof__} continue to work despite
836 @samp{-ansi}. You would not want to use them in an ISO C program, of
837 course, but it is useful to put them in header files that might be included
838 in compilations done with @samp{-ansi}. Alternate predefined macros
839 such as @code{__unix__} and @code{__vax__} are also available, with or
840 without @samp{-ansi}.
842 The @samp{-ansi} option does not cause non-ISO programs to be
843 rejected gratuitously. For that, @samp{-pedantic} is required in
844 addition to @samp{-ansi}. @xref{Warning Options}.
846 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
847 option is used. Some header files may notice this macro and refrain
848 from declaring certain functions or defining certain macros that the
849 ISO standard doesn't call for; this is to avoid interfering with any
850 programs that might use these names for other things.
852 Functions which would normally be builtin but do not have semantics
853 defined by ISO C (such as @code{alloca} and @code{ffs}) are not builtin
854 functions with @samp{-ansi} is used. @xref{Other Builtins,,Other
855 built-in functions provided by GNU CC}, for details of the functions
859 Determine the language standard. A value for this option must be provided;
867 ISO C as modified in amend. 1
870 ISO C99. Note that this standard is not yet fully supported; see
871 @uref{http://gcc.gnu.org/c99status.html} for more information.
874 same as -std=iso9899:1990
877 same as -std=iso9899:1999
880 default, iso9899:1990 + gnu extensions
883 iso9899:1999 + gnu extensions
886 same as -std=iso9899:1999, deprecated
889 same as -std=iso9899:1999, deprecated
892 same as -std=gnu99, deprecated
896 Even when this option is not specified, you can still use some of the
897 features of newer standards in so far as they do not conflict with
898 previous C standards. For example, you may use @code{__restrict__} even
899 when -std=c99 is not specified.
901 @xref{Standards,,Language Standards Supported by GCC}, for details of
902 these standard versions.
905 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
906 keyword, so that code can use these words as identifiers. You can use
907 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
908 instead. @samp{-ansi} implies @samp{-fno-asm}.
910 In C++, this switch only affects the @code{typeof} keyword, since
911 @code{asm} and @code{inline} are standard keywords. You may want to
912 use the @samp{-fno-gnu-keywords} flag instead, which has the same effect.
915 @cindex builtin functions
916 Don't recognize builtin functions that do not begin with
917 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
918 functions provided by GNU CC}, for details of the functions affected,
919 including those which are not builtin functions when @samp{-ansi} is
920 used because they do not have an ISO standard meaning.
922 GCC normally generates special code to handle certain builtin functions
923 more efficiently; for instance, calls to @code{alloca} may become single
924 instructions that adjust the stack directly, and calls to @code{memcpy}
925 may become inline copy loops. The resulting code is often both smaller
926 and faster, but since the function calls no longer appear as such, you
927 cannot set a breakpoint on those calls, nor can you change the behavior
928 of the functions by linking with a different library.
931 @cindex hosted environment
933 Assert that compilation takes place in a hosted environment. This implies
934 @samp{-fbuiltin}. A hosted environment is one in which the
935 entire standard library is available, and in which @code{main} has a return
936 type of @code{int}. Examples are nearly everything except a kernel.
937 This is equivalent to @samp{-fno-freestanding}.
940 @cindex hosted environment
942 Assert that compilation takes place in a freestanding environment. This
943 implies @samp{-fno-builtin}. A freestanding environment
944 is one in which the standard library may not exist, and program startup may
945 not necessarily be at @code{main}. The most obvious example is an OS kernel.
946 This is equivalent to @samp{-fno-hosted}.
948 @xref{Standards,,Language Standards Supported by GCC}, for details of
949 freestanding and hosted environments.
952 Support ISO C trigraphs. You don't want to know about this
953 brain-damage. The @samp{-ansi} option implies @samp{-trigraphs}.
955 @cindex traditional C language
956 @cindex C language, traditional
958 Attempt to support some aspects of traditional C compilers.
963 All @code{extern} declarations take effect globally even if they
964 are written inside of a function definition. This includes implicit
965 declarations of functions.
968 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
969 and @code{volatile} are not recognized. (You can still use the
970 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
974 Comparisons between pointers and integers are always allowed.
977 Integer types @code{unsigned short} and @code{unsigned char} promote
978 to @code{unsigned int}.
981 Out-of-range floating point literals are not an error.
984 Certain constructs which ISO regards as a single invalid preprocessing
985 number, such as @samp{0xe-0xd}, are treated as expressions instead.
988 String ``constants'' are not necessarily constant; they are stored in
989 writable space, and identical looking constants are allocated
990 separately. (This is the same as the effect of
991 @samp{-fwritable-strings}.)
993 @cindex @code{longjmp} and automatic variables
995 All automatic variables not declared @code{register} are preserved by
996 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
997 not declared @code{volatile} may be clobbered.
1002 @cindex escape sequences, traditional
1003 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1004 literal characters @samp{x} and @samp{a} respectively. Without
1005 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1006 representation of a character, and @samp{\a} produces a bell.
1009 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1010 if your program uses names that are normally GNU C builtin functions for
1011 other purposes of its own.
1013 You cannot use @samp{-traditional} if you include any header files that
1014 rely on ISO C features. Some vendors are starting to ship systems with
1015 ISO C header files and you cannot use @samp{-traditional} on such
1016 systems to compile files that include any system headers.
1018 The @samp{-traditional} option also enables @samp{-traditional-cpp},
1019 which is described next.
1021 @item -traditional-cpp
1022 Attempt to support some aspects of traditional C preprocessors.
1027 Comments convert to nothing at all, rather than to a space. This allows
1028 traditional token concatenation.
1031 In a preprocessing directive, the @samp{#} symbol must appear as the first
1032 character of a line.
1035 Macro arguments are recognized within string constants in a macro
1036 definition (and their values are stringified, though without additional
1037 quote marks, when they appear in such a context). The preprocessor
1038 always considers a string constant to end at a newline.
1041 @cindex detecting @w{@samp{-traditional}}
1042 The predefined macro @code{__STDC__} is not defined when you use
1043 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
1044 which @code{__GNUC__} indicates are not affected by
1045 @samp{-traditional}). If you need to write header files that work
1046 differently depending on whether @samp{-traditional} is in use, by
1047 testing both of these predefined macros you can distinguish four
1048 situations: GNU C, traditional GNU C, other ISO C compilers, and other
1049 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
1050 not defined when you use @samp{-traditional}. @xref{Standard
1051 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
1052 for more discussion of these and other predefined macros.
1055 @cindex string constants vs newline
1056 @cindex newline vs string constants
1057 The preprocessor considers a string constant to end at a newline (unless
1058 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
1059 string constants can contain the newline character as typed.)
1062 @item -fcond-mismatch
1063 Allow conditional expressions with mismatched types in the second and
1064 third arguments. The value of such an expression is void.
1066 @item -funsigned-char
1067 Let the type @code{char} be unsigned, like @code{unsigned char}.
1069 Each kind of machine has a default for what @code{char} should
1070 be. It is either like @code{unsigned char} by default or like
1071 @code{signed char} by default.
1073 Ideally, a portable program should always use @code{signed char} or
1074 @code{unsigned char} when it depends on the signedness of an object.
1075 But many programs have been written to use plain @code{char} and
1076 expect it to be signed, or expect it to be unsigned, depending on the
1077 machines they were written for. This option, and its inverse, let you
1078 make such a program work with the opposite default.
1080 The type @code{char} is always a distinct type from each of
1081 @code{signed char} or @code{unsigned char}, even though its behavior
1082 is always just like one of those two.
1085 Let the type @code{char} be signed, like @code{signed char}.
1087 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
1088 the negative form of @samp{-funsigned-char}. Likewise, the option
1089 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
1091 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1092 if your program uses names that are normally GNU C builtin functions for
1093 other purposes of its own.
1095 You cannot use @samp{-traditional} if you include any header files that
1096 rely on ISO C features. Some vendors are starting to ship systems with
1097 ISO C header files and you cannot use @samp{-traditional} on such
1098 systems to compile files that include any system headers.
1100 @item -fsigned-bitfields
1101 @itemx -funsigned-bitfields
1102 @itemx -fno-signed-bitfields
1103 @itemx -fno-unsigned-bitfields
1104 These options control whether a bitfield is signed or unsigned, when the
1105 declaration does not use either @code{signed} or @code{unsigned}. By
1106 default, such a bitfield is signed, because this is consistent: the
1107 basic integer types such as @code{int} are signed types.
1109 However, when @samp{-traditional} is used, bitfields are all unsigned
1112 @item -fwritable-strings
1113 Store string constants in the writable data segment and don't uniquize
1114 them. This is for compatibility with old programs which assume they can
1115 write into string constants. The option @samp{-traditional} also has
1118 Writing into string constants is a very bad idea; ``constants'' should
1121 @item -fallow-single-precision
1122 Do not promote single precision math operations to double precision,
1123 even when compiling with @samp{-traditional}.
1125 Traditional K&R C promotes all floating point operations to double
1126 precision, regardless of the sizes of the operands. On the
1127 architecture for which you are compiling, single precision may be faster
1128 than double precision. If you must use @samp{-traditional}, but want
1129 to use single precision operations when the operands are single
1130 precision, use this option. This option has no effect when compiling
1131 with ISO or GNU C conventions (the default).
1134 Override the underlying type for @samp{wchar_t} to be @samp{short
1135 unsigned int} instead of the default for the target. This option is
1136 useful for building programs to run under WINE.
1139 @node C++ Dialect Options
1140 @section Options Controlling C++ Dialect
1142 @cindex compiler options, C++
1143 @cindex C++ options, command line
1144 @cindex options, C++
1145 This section describes the command-line options that are only meaningful
1146 for C++ programs; but you can also use most of the GNU compiler options
1147 regardless of what language your program is in. For example, you
1148 might compile a file @code{firstClass.C} like this:
1151 g++ -g -frepo -O -c firstClass.C
1155 In this example, only @samp{-frepo} is an option meant
1156 only for C++ programs; you can use the other options with any
1157 language supported by GCC.
1159 Here is a list of options that are @emph{only} for compiling C++ programs:
1162 @item -fno-access-control
1163 Turn off all access checking. This switch is mainly useful for working
1164 around bugs in the access control code.
1167 Check that the pointer returned by @code{operator new} is non-null
1168 before attempting to modify the storage allocated. The current Working
1169 Paper requires that @code{operator new} never return a null pointer, so
1170 this check is normally unnecessary.
1172 An alternative to using this option is to specify that your
1173 @code{operator new} does not throw any exceptions; if you declare it
1174 @samp{throw()}, g++ will check the return value. See also @samp{new
1177 @item -fconserve-space
1178 Put uninitialized or runtime-initialized global variables into the
1179 common segment, as C does. This saves space in the executable at the
1180 cost of not diagnosing duplicate definitions. If you compile with this
1181 flag and your program mysteriously crashes after @code{main()} has
1182 completed, you may have an object that is being destroyed twice because
1183 two definitions were merged.
1185 This option is no longer useful on most targets, now that support has
1186 been added for putting variables into BSS without making them common.
1188 @item -fdollars-in-identifiers
1189 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1190 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
1191 @samp{$} by default on most target systems, but there are a few exceptions.)
1192 Traditional C allowed the character @samp{$} to form part of
1193 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1195 @item -fno-elide-constructors
1196 The C++ standard allows an implementation to omit creating a temporary
1197 which is only used to initialize another object of the same type.
1198 Specifying this option disables that optimization, and forces g++ to
1199 call the copy constructor in all cases.
1201 @item -fno-enforce-eh-specs
1202 Don't check for violation of exception specifications at runtime. This
1203 option violates the C++ standard, but may be useful for reducing code
1204 size in production builds, much like defining @samp{NDEBUG}. The compiler
1205 will still optimize based on the exception specifications.
1207 @item -fexternal-templates
1208 Cause template instantiations to obey @samp{#pragma interface} and
1209 @samp{implementation}; template instances are emitted or not according
1210 to the location of the template definition. @xref{Template
1211 Instantiation}, for more information.
1213 This option is deprecated.
1215 @item -falt-external-templates
1216 Similar to -fexternal-templates, but template instances are emitted or
1217 not according to the place where they are first instantiated.
1218 @xref{Template Instantiation}, for more information.
1220 This option is deprecated.
1223 @itemx -fno-for-scope
1224 If -ffor-scope is specified, the scope of variables declared in
1225 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1226 as specified by the C++ standard.
1227 If -fno-for-scope is specified, the scope of variables declared in
1228 a @i{for-init-statement} extends to the end of the enclosing scope,
1229 as was the case in old versions of gcc, and other (traditional)
1230 implementations of C++.
1232 The default if neither flag is given to follow the standard,
1233 but to allow and give a warning for old-style code that would
1234 otherwise be invalid, or have different behavior.
1236 @item -fno-gnu-keywords
1237 Do not recognize @code{typeof} as a keyword, so that code can use this
1238 word as an identifier. You can use the keyword @code{__typeof__} instead.
1239 @samp{-ansi} implies @samp{-fno-gnu-keywords}.
1242 Treat the @code{namespace std} as a namespace, instead of ignoring
1243 it. For compatibility with earlier versions of g++, the compiler will,
1244 by default, ignore @code{namespace-declarations},
1245 @code{using-declarations}, @code{using-directives}, and
1246 @code{namespace-names}, if they involve @code{std}.
1248 @item -fhuge-objects
1249 Support virtual function calls for objects that exceed the size
1250 representable by a @samp{short int}. Users should not use this flag by
1251 default; if you need to use it, the compiler will tell you so.
1253 This flag is not useful when compiling with -fvtable-thunks.
1255 Like all options that change the ABI, all C++ code, @emph{including
1256 libgcc} must be built with the same setting of this option.
1258 @item -fno-implicit-templates
1259 Never emit code for non-inline templates which are instantiated
1260 implicitly (i.e. by use); only emit code for explicit instantiations.
1261 @xref{Template Instantiation}, for more information.
1263 @item -fno-implicit-inline-templates
1264 Don't emit code for implicit instantiations of inline templates, either.
1265 The default is to handle inlines differently so that compiles with and
1266 without optimization will need the same set of explicit instantiations.
1268 @item -fno-implement-inlines
1269 To save space, do not emit out-of-line copies of inline functions
1270 controlled by @samp{#pragma implementation}. This will cause linker
1271 errors if these functions are not inlined everywhere they are called.
1273 @item -fms-extensions
1274 Disable pedantic warnings about constructs used in MFC, such as implicit
1275 int and getting a pointer to member function via non-standard syntax.
1277 @item -fname-mangling-version-@var{n}
1278 Control the way in which names are mangled. Version 0 is compatible
1279 with versions of g++ before 2.8. Version 1 is the default. Version 1
1280 will allow correct mangling of function templates. For example,
1281 version 0 mangling does not mangle foo<int, double> and foo<int, char>
1282 given this declaration:
1285 template <class T, class U> void foo(T t);
1288 Like all options that change the ABI, all C++ code, @emph{including
1289 libgcc} must be built with the same setting of this option.
1291 @item -fno-operator-names
1292 Do not treat the operator name keywords @code{and}, @code{bitand},
1293 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1294 synonyms as keywords.
1296 @item -fno-optional-diags
1297 Disable diagnostics that the standard says a compiler does not need to
1298 issue. Currently, the only such diagnostic issued by g++ is the one for
1299 a name having multiple meanings within a class.
1302 Downgrade messages about nonconformant code from errors to warnings. By
1303 default, g++ effectively sets @samp{-pedantic-errors} without
1304 @samp{-pedantic}; this option reverses that. This behavior and this
1305 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1308 Enable automatic template instantiation. This option also implies
1309 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1313 Disable generation of information about every class with virtual
1314 functions for use by the C++ runtime type identification features
1315 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1316 of the language, you can save some space by using this flag. Note that
1317 exception handling uses the same information, but it will generate it as
1321 @itemx -fno-squangle
1322 @samp{-fsquangle} will enable a compressed form of name mangling for
1323 identifiers. In particular, it helps to shorten very long names by recognizing
1324 types and class names which occur more than once, replacing them with special
1325 short ID codes. This option also requires any C++ libraries being used to
1326 be compiled with this option as well. The compiler has this disabled (the
1327 equivalent of @samp{-fno-squangle}) by default.
1329 Like all options that change the ABI, all C++ code, @emph{including
1330 libgcc.a} must be built with the same setting of this option.
1332 @item -ftemplate-depth-@var{n}
1333 Set the maximum instantiation depth for template classes to @var{n}.
1334 A limit on the template instantiation depth is needed to detect
1335 endless recursions during template class instantiation. ANSI/ISO C++
1336 conforming programs must not rely on a maximum depth greater than 17.
1338 @item -fuse-cxa-atexit
1339 Register destructors for objects with static storage duration with the
1340 @code{__cxa_atexit} function rather than the @code{atexit} function.
1341 This option is required for fully standards-compliant handling of static
1342 destructors, but will only work if your C library supports
1343 @code{__cxa_atexit}.
1345 @item -fvtable-thunks
1346 Use @samp{thunks} to implement the virtual function dispatch table
1347 (@samp{vtable}). The traditional (cfront-style) approach to
1348 implementing vtables was to store a pointer to the function and two
1349 offsets for adjusting the @samp{this} pointer at the call site. Newer
1350 implementations store a single pointer to a @samp{thunk} function which
1351 does any necessary adjustment and then calls the target function.
1353 This option also enables a heuristic for controlling emission of
1354 vtables; if a class has any non-inline virtual functions, the vtable
1355 will be emitted in the translation unit containing the first one of
1358 Like all options that change the ABI, all C++ code, @emph{including
1359 libgcc.a} must be built with the same setting of this option.
1362 Do not search for header files in the standard directories specific to
1363 C++, but do still search the other standard directories. (This option
1364 is used when building the C++ library.)
1367 In addition, these optimization, warning, and code generation options
1368 have meanings only for C++ programs:
1371 @item -fno-default-inline
1372 Do not assume @samp{inline} for functions defined inside a class scope.
1373 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1374 functions will have linkage like inline functions; they just won't be
1377 @item -Wctor-dtor-privacy (C++ only)
1378 Warn when a class seems unusable, because all the constructors or
1379 destructors in a class are private and the class has no friends or
1380 public static member functions.
1382 @item -Wnon-virtual-dtor (C++ only)
1383 Warn when a class declares a non-virtual destructor that should probably
1384 be virtual, because it looks like the class will be used polymorphically.
1386 @item -Wreorder (C++ only)
1387 @cindex reordering, warning
1388 @cindex warning for reordering of member initializers
1389 Warn when the order of member initializers given in the code does not
1390 match the order in which they must be executed. For instance:
1396 A(): j (0), i (1) @{ @}
1400 Here the compiler will warn that the member initializers for @samp{i}
1401 and @samp{j} will be rearranged to match the declaration order of the
1405 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1408 @item -Weffc++ (C++ only)
1409 Warn about violations of various style guidelines from Scott Meyers'
1410 @cite{Effective C++} books. If you use this option, you should be aware
1411 that the standard library headers do not obey all of these guidelines;
1412 you can use @samp{grep -v} to filter out those warnings.
1414 @item -Wno-deprecated (C++ only)
1415 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1417 @item -Wno-non-template-friend (C++ only)
1418 Disable warnings when non-templatized friend functions are declared
1419 within a template. With the advent of explicit template specification
1420 support in g++, if the name of the friend is an unqualified-id (ie,
1421 @samp{friend foo(int)}), the C++ language specification demands that the
1422 friend declare or define an ordinary, nontemplate function. (Section
1423 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1424 could be interpreted as a particular specialization of a templatized
1425 function. Because this non-conforming behavior is no longer the default
1426 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1427 check existing code for potential trouble spots, and is on by default.
1428 This new compiler behavior can be turned off with
1429 @samp{-Wno-non-template-friend} which keeps the conformant compiler code
1430 but disables the helpful warning.
1432 @item -Wold-style-cast (C++ only)
1433 Warn if an old-style (C-style) cast is used within a C++ program. The
1434 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1435 @samp{const_cast}) are less vulnerable to unintended effects.
1437 @item -Woverloaded-virtual (C++ only)
1438 @cindex overloaded virtual fn, warning
1439 @cindex warning for overloaded virtual fn
1440 Warn when a derived class function declaration may be an error in
1441 defining a virtual function. In a derived class, the
1442 definitions of virtual functions must match the type signature of a
1443 virtual function declared in the base class. With this option, the
1444 compiler warns when you define a function with the same name as a
1445 virtual function, but with a type signature that does not match any
1446 declarations from the base class.
1448 @item -Wno-pmf-conversions (C++ only)
1449 Disable the diagnostic for converting a bound pointer to member function
1452 @item -Wsign-promo (C++ only)
1453 Warn when overload resolution chooses a promotion from unsigned or
1454 enumeral type to a signed type over a conversion to an unsigned type of
1455 the same size. Previous versions of g++ would try to preserve
1456 unsignedness, but the standard mandates the current behavior.
1458 @item -Wsynth (C++ only)
1459 @cindex warning for synthesized methods
1460 @cindex synthesized methods, warning
1461 Warn when g++'s synthesis behavior does not match that of cfront. For
1467 A& operator = (int);
1477 In this example, g++ will synthesize a default @samp{A& operator =
1478 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1481 @node Language Independent Options
1482 @section Options to Control Diagnostic Messages Formatting
1483 @cindex options to control diagnostics formatting
1484 @cindex diagnostic messages
1485 @cindex message formatting
1487 Traditionally, diagnostic messages have been formatted irrespective of
1488 the output device's aspect (e.g. its width, ...). The options described
1489 below can be used to control the diagnostic messages formatting
1490 algorithm, e.g. how many characters per line, how often source location
1491 information should be reported. Right now, only the C++ front-end can
1492 honor these options. However it is expected, in the near future, that
1493 the remaining front-ends would be able to digest them correctly.
1496 @item -fmessage-length=@var{n}
1497 Try to format error messages so that they fit on lines of about @var{n}
1498 characters. The default is 72 characters for g++ and 0 for the rest of
1499 the front-ends supported by GCC. If @var{n} is zero, then no
1500 line-wrapping will be done; each error message will appear on a single
1503 @item -fdiagnostics-show-location=once
1504 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1505 reporter to emit @emph{once} source location information; that is, in
1506 case the message is too long to fit on a single physical line and has to
1507 be wrapped, the source location won't be emitted (as prefix) again,
1508 over and over, in subsequent continuation lines. This is the default
1511 @item -fdiagnostics-show-location=every-line
1512 Only meaningful in line-wrapping mode. Instructs the diagnostic
1513 messages reporter to emit the same source location information (as
1514 prefix) for physical lines that result from the process of breaking a
1515 a message which is too long to fit on a single line.
1519 @node Warning Options
1520 @section Options to Request or Suppress Warnings
1521 @cindex options to control warnings
1522 @cindex warning messages
1523 @cindex messages, warning
1524 @cindex suppressing warnings
1526 Warnings are diagnostic messages that report constructions which
1527 are not inherently erroneous but which are risky or suggest there
1528 may have been an error.
1530 You can request many specific warnings with options beginning @samp{-W},
1531 for example @samp{-Wimplicit} to request warnings on implicit
1532 declarations. Each of these specific warning options also has a
1533 negative form beginning @samp{-Wno-} to turn off warnings;
1534 for example, @samp{-Wno-implicit}. This manual lists only one of the
1535 two forms, whichever is not the default.
1537 These options control the amount and kinds of warnings produced by GCC:
1540 @cindex syntax checking
1542 Check the code for syntax errors, but don't do anything beyond that.
1545 Issue all the warnings demanded by strict ISO C and ISO C++;
1546 reject all programs that use forbidden extensions, and some other
1547 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1548 version of the ISO C standard specified by any @samp{-std} option used.
1550 Valid ISO C and ISO C++ programs should compile properly with or without
1551 this option (though a rare few will require @samp{-ansi}). However,
1552 without this option, certain GNU extensions and traditional C and C++
1553 features are supported as well. With this option, they are rejected.
1555 @samp{-pedantic} does not cause warning messages for use of the
1556 alternate keywords whose names begin and end with @samp{__}. Pedantic
1557 warnings are also disabled in the expression that follows
1558 @code{__extension__}. However, only system header files should use
1559 these escape routes; application programs should avoid them.
1560 @xref{Alternate Keywords}.
1562 Some users try to use @samp{-pedantic} to check programs for strict ISO
1563 C conformance. They soon find that it does not do quite what they want:
1564 it finds some non-ISO practices, but not all---only those for which
1565 ISO C @emph{requires} a diagnostic, and some others for which
1566 diagnostics have been added.
1568 A feature to report any failure to conform to ISO C might be useful in
1569 some instances, but would require considerable additional work and would
1570 be quite different from @samp{-pedantic}. We don't have plans to
1571 support such a feature in the near future.
1573 @item -pedantic-errors
1574 Like @samp{-pedantic}, except that errors are produced rather than
1578 Inhibit all warning messages.
1581 Inhibit warning messages about the use of @samp{#import}.
1583 @item -Wchar-subscripts
1584 Warn if an array subscript has type @code{char}. This is a common cause
1585 of error, as programmers often forget that this type is signed on some
1589 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1590 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1593 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1594 the arguments supplied have types appropriate to the format string
1595 specified, and that the conversions specified in the format string make
1596 sense. This includes standard functions, and others specified by format
1597 attributes (@pxref{Function Attributes}), in the @code{printf},
1598 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1599 not in the C standard) families.
1601 The formats are checked against the format features supported by GNU
1602 libc version 2.2. These include all ISO C89 and C99 features, as well
1603 as features from the Single Unix Specification and some BSD and GNU
1604 extensions. Other library implementations may not support all these
1605 features; GCC does not support warning about features that go beyond a
1606 particular library's limitations. However, if @samp{-pedantic} is used
1607 with @samp{-Wformat}, warnings will be given about format features not
1608 in the selected standard version (but not for @code{strfmon} formats,
1609 since those are not in any version of the C standard). @xref{C Dialect
1610 Options,,Options Controlling C Dialect}.
1612 @samp{-Wformat} is included in @samp{-Wall}. For more control over some
1613 aspects of format checking, the options @samp{-Wno-format-y2k},
1614 @samp{-Wno-format-extra-args}, @samp{-Wformat-nonliteral},
1615 @samp{-Wformat-security} and @samp{-Wformat=2} are available, but are
1616 not included in @samp{-Wall}.
1618 @item -Wno-format-y2k
1619 If @samp{-Wformat} is specified, do not warn about @code{strftime}
1620 formats which may yield only a two-digit year.
1622 @item -Wno-format-extra-args
1623 If @samp{-Wformat} is specified, do not warn about excess arguments to a
1624 @code{printf} or @code{scanf} format function. The C standard specifies
1625 that such arguments are ignored.
1627 @item -Wformat-nonliteral
1628 If @samp{-Wformat} is specified, also warn if the format string is not a
1629 string literal and so cannot be checked, unless the format function
1630 takes its format arguments as a @code{va_list}.
1632 @item -Wformat-security
1633 If @samp{-Wformat} is specified, also warn about uses of format
1634 functions that represent possible security problems. At present, this
1635 warns about calls to @code{printf} and @code{scanf} functions where the
1636 format string is not a string literal and there are no format arguments,
1637 as in @code{printf (foo);}. This may be a security hole if the format
1638 string came from untrusted input and contains @samp{%n}. (This is
1639 currently a subset of what @samp{-Wformat-nonliteral} warns about, but
1640 in future warnings may be added to @samp{-Wformat-security} that are not
1641 included in @samp{-Wformat-nonliteral}.)
1644 Enable @samp{-Wformat} plus format checks not included in
1645 @samp{-Wformat}. Currently equivalent to @samp{-Wformat
1646 -Wformat-nonliteral -Wformat-security}.
1648 @item -Wimplicit-int
1649 Warn when a declaration does not specify a type.
1651 @item -Wimplicit-function-declaration
1652 @itemx -Werror-implicit-function-declaration
1653 Give a warning (or error) whenever a function is used before being
1657 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1661 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1662 function with external linkage, returning int, taking either zero
1663 arguments, two, or three arguments of appropriate types.
1666 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1667 indicate a typo in the user's code, as they have implementation-defined
1668 values, and should not be used in portable code.
1671 Warn if parentheses are omitted in certain contexts, such
1672 as when there is an assignment in a context where a truth value
1673 is expected, or when operators are nested whose precedence people
1674 often get confused about.
1676 Also warn about constructions where there may be confusion to which
1677 @code{if} statement an @code{else} branch belongs. Here is an example of
1690 In C, every @code{else} branch belongs to the innermost possible @code{if}
1691 statement, which in this example is @code{if (b)}. This is often not
1692 what the programmer expected, as illustrated in the above example by
1693 indentation the programmer chose. When there is the potential for this
1694 confusion, GNU C will issue a warning when this flag is specified.
1695 To eliminate the warning, add explicit braces around the innermost
1696 @code{if} statement so there is no way the @code{else} could belong to
1697 the enclosing @code{if}. The resulting code would look like this:
1711 @item -Wsequence-point
1712 Warn about code that may have undefined semantics because of violations
1713 of sequence point rules in the C standard.
1715 The C standard defines the order in which expressions in a C program are
1716 evaluated in terms of @dfn{sequence points}, which represent a partial
1717 ordering between the execution of parts of the program: those executed
1718 before the sequence point, and those executed after it. These occur
1719 after the evaluation of a full expression (one which is not part of a
1720 larger expression), after the evaluation of the first operand of a
1721 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1722 function is called (but after the evaluation of its arguments and the
1723 expression denoting the called function), and in certain other places.
1724 Other than as expressed by the sequence point rules, the order of
1725 evaluation of subexpressions of an expression is not specified. All
1726 these rules describe only a partial order rather than a total order,
1727 since, for example, if two functions are called within one expression
1728 with no sequence point between them, the order in which the functions
1729 are called is not specified. However, the standards committee have
1730 ruled that function calls do not overlap.
1732 It is not specified when between sequence points modifications to the
1733 values of objects take effect. Programs whose behavior depends on this
1734 have undefined behavior; the C standard specifies that ``Between the
1735 previous and next sequence point an object shall have its stored value
1736 modified at most once by the evaluation of an expression. Furthermore,
1737 the prior value shall be read only to determine the value to be
1738 stored.''. If a program breaks these rules, the results on any
1739 particular implementation are entirely unpredictable.
1741 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1742 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1743 diagnosed by this option, and it may give an occasional false positive
1744 result, but in general it has been found fairly effective at detecting
1745 this sort of problem in programs.
1747 The present implementation of this option only works for C programs. A
1748 future implementation may also work for C++ programs.
1750 There is some controversy over the precise meaning of the sequence point
1751 rules in subtle cases. Alternative formal definitions may be found in
1752 Clive Feather's ``Annex S''
1753 @uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n925.htm} and in
1754 Michael Norrish's thesis
1755 @uref{http://www.cl.cam.ac.uk/users/mn200/PhD/thesis-report.ps.gz}.
1756 Other discussions are by Raymond Mak
1757 @uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n926.htm} and
1759 @uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n927.htm}.
1762 Warn whenever a function is defined with a return-type that defaults to
1763 @code{int}. Also warn about any @code{return} statement with no
1764 return-value in a function whose return-type is not @code{void}.
1766 For C++, a function without return type always produces a diagnostic
1767 message, even when @samp{-Wno-return-type} is specified. The only
1768 exceptions are @samp{main} and functions defined in system headers.
1771 Warn whenever a @code{switch} statement has an index of enumeral type
1772 and lacks a @code{case} for one or more of the named codes of that
1773 enumeration. (The presence of a @code{default} label prevents this
1774 warning.) @code{case} labels outside the enumeration range also
1775 provoke warnings when this option is used.
1778 Warn if any trigraphs are encountered that might change the meaning of
1779 the program (trigraphs within comments are not warned about).
1781 @item -Wunused-function
1782 Warn whenever a static function is declared but not defined or a
1783 non\-inline static function is unused.
1785 @item -Wunused-label
1786 Warn whenever a label is declared but not used.
1788 To suppress this warning use the @samp{unused} attribute
1789 (@pxref{Variable Attributes}).
1791 @item -Wunused-parameter
1792 Warn whenever a function parameter is unused aside from its declaration.
1794 To suppress this warning use the @samp{unused} attribute
1795 (@pxref{Variable Attributes}).
1797 @item -Wunused-variable
1798 Warn whenever a local variable or non-constant static variable is unused
1799 aside from its declaration
1801 To suppress this warning use the @samp{unused} attribute
1802 (@pxref{Variable Attributes}).
1804 @item -Wunused-value
1805 Warn whenever a statement computes a result that is explicitly not used.
1807 To suppress this warning cast the expression to @samp{void}.
1810 All all the above @samp{-Wunused} options combined.
1812 In order to get a warning about an unused function parameter, you must
1813 either specify @samp{-W -Wunused} or separately specify
1814 @samp{-Wunused-parameter}.
1816 @item -Wuninitialized
1817 Warn if an automatic variable is used without first being initialized or
1818 if a variable may be clobbered by a @code{setjmp} call.
1820 These warnings are possible only in optimizing compilation,
1821 because they require data flow information that is computed only
1822 when optimizing. If you don't specify @samp{-O}, you simply won't
1825 These warnings occur only for variables that are candidates for
1826 register allocation. Therefore, they do not occur for a variable that
1827 is declared @code{volatile}, or whose address is taken, or whose size
1828 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1829 structures, unions or arrays, even when they are in registers.
1831 Note that there may be no warning about a variable that is used only
1832 to compute a value that itself is never used, because such
1833 computations may be deleted by data flow analysis before the warnings
1836 These warnings are made optional because GCC is not smart
1837 enough to see all the reasons why the code might be correct
1838 despite appearing to have an error. Here is one example of how
1857 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1858 always initialized, but GCC doesn't know this. Here is
1859 another common case:
1864 if (change_y) save_y = y, y = new_y;
1866 if (change_y) y = save_y;
1871 This has no bug because @code{save_y} is used only if it is set.
1873 @cindex @code{longjmp} warnings
1874 This option also warns when a non-volatile automatic variable might be
1875 changed by a call to @code{longjmp}. These warnings as well are possible
1876 only in optimizing compilation.
1878 The compiler sees only the calls to @code{setjmp}. It cannot know
1879 where @code{longjmp} will be called; in fact, a signal handler could
1880 call it at any point in the code. As a result, you may get a warning
1881 even when there is in fact no problem because @code{longjmp} cannot
1882 in fact be called at the place which would cause a problem.
1884 Some spurious warnings can be avoided if you declare all the functions
1885 you use that never return as @code{noreturn}. @xref{Function
1888 @item -Wreorder (C++ only)
1889 @cindex reordering, warning
1890 @cindex warning for reordering of member initializers
1891 Warn when the order of member initializers given in the code does not
1892 match the order in which they must be executed. For instance:
1894 @item -Wunknown-pragmas
1895 @cindex warning for unknown pragmas
1896 @cindex unknown pragmas, warning
1897 @cindex pragmas, warning of unknown
1898 Warn when a #pragma directive is encountered which is not understood by
1899 GCC. If this command line option is used, warnings will even be issued
1900 for unknown pragmas in system header files. This is not the case if
1901 the warnings were only enabled by the @samp{-Wall} command line option.
1904 All of the above @samp{-W} options combined. This enables all the
1905 warnings about constructions that some users consider questionable, and
1906 that are easy to avoid (or modify to prevent the warning), even in
1907 conjunction with macros.
1909 @item -Wsystem-headers
1910 @cindex warnings from system headers
1911 @cindex system headers, warnings from
1912 Print warning messages for constructs found in system header files.
1913 Warnings from system headers are normally suppressed, on the assumption
1914 that they usually do not indicate real problems and would only make the
1915 compiler output harder to read. Using this command line option tells
1916 GCC to emit warnings from system headers as if they occurred in user
1917 code. However, note that using @samp{-Wall} in conjunction with this
1918 option will @emph{not} warn about unknown pragmas in system
1919 headers---for that, @samp{-Wunknown-pragmas} must also be used.
1922 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1923 Some of them warn about constructions that users generally do not
1924 consider questionable, but which occasionally you might wish to check
1925 for; others warn about constructions that are necessary or hard to avoid
1926 in some cases, and there is no simple way to modify the code to suppress
1931 Print extra warning messages for these events:
1935 A function can return either with or without a value. (Falling
1936 off the end of the function body is considered returning without
1937 a value.) For example, this function would evoke such a
1951 An expression-statement or the left-hand side of a comma expression
1952 contains no side effects.
1953 To suppress the warning, cast the unused expression to void.
1954 For example, an expression such as @samp{x[i,j]} will cause a warning,
1955 but @samp{x[(void)i,j]} will not.
1958 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1961 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1962 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1963 that of ordinary mathematical notation.
1966 Storage-class specifiers like @code{static} are not the first things in
1967 a declaration. According to the C Standard, this usage is obsolescent.
1970 The return type of a function has a type qualifier such as @code{const}.
1971 Such a type qualifier has no effect, since the value returned by a
1972 function is not an lvalue. (But don't warn about the GNU extension of
1973 @code{volatile void} return types. That extension will be warned about
1974 if @samp{-pedantic} is specified.)
1977 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
1981 A comparison between signed and unsigned values could produce an
1982 incorrect result when the signed value is converted to unsigned.
1983 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
1986 An aggregate has a partly bracketed initializer.
1987 For example, the following code would evoke such a warning,
1988 because braces are missing around the initializer for @code{x.h}:
1991 struct s @{ int f, g; @};
1992 struct t @{ struct s h; int i; @};
1993 struct t x = @{ 1, 2, 3 @};
1997 An aggregate has an initializer which does not initialize all members.
1998 For example, the following code would cause such a warning, because
1999 @code{x.h} would be implicitly initialized to zero:
2002 struct s @{ int f, g, h; @};
2003 struct s x = @{ 3, 4 @};
2008 Warn if floating point values are used in equality comparisons.
2010 The idea behind this is that sometimes it is convenient (for the
2011 programmer) to consider floating-point values as approximations to
2012 infinitely precise real numbers. If you are doing this, then you need
2013 to compute (by analysing the code, or in some other way) the maximum or
2014 likely maximum error that the computation introduces, and allow for it
2015 when performing comparisons (and when producing output, but that's a
2016 different problem). In particular, instead of testing for equality, you
2017 would check to see whether the two values have ranges that overlap; and
2018 this is done with the relational operators, so equality comparisons are
2021 @item -Wtraditional (C only)
2022 Warn about certain constructs that behave differently in traditional and
2027 Macro arguments occurring within string constants in the macro body.
2028 These would substitute the argument in traditional C, but are part of
2029 the constant in ISO C.
2032 A function declared external in one block and then used after the end of
2036 A @code{switch} statement has an operand of type @code{long}.
2039 A non-@code{static} function declaration follows a @code{static} one.
2040 This construct is not accepted by some traditional C compilers.
2043 The ISO type of an integer constant has a different width or
2044 signedness from its traditional type. This warning is only issued if
2045 the base of the constant is ten. I.e. hexadecimal or octal values, which
2046 typically represent bit patterns, are not warned about.
2049 Usage of ISO string concatenation is detected.
2052 A function macro appears without arguments.
2055 The unary plus operator.
2058 Initialization of automatic aggregates.
2061 Identifier conflicts with labels. Traditional C lacks a separate
2062 namespace for labels.
2065 Initialization of unions. If the initializer is zero, the warning is
2066 omitted. This is done under the assumption that the zero initializer in
2067 user code appears conditioned on e.g. @code{__STDC__} to avoid missing
2068 initializer warnings and relies on default initialization to zero in the
2072 The `U' integer constant suffix, or the `F' or `L' floating point
2073 constant suffixes. (Traditonal C does support the `L' suffix on integer
2074 constants.) Note, these suffixes appear in macros defined in the system
2075 headers of most modern systems, e.g. the _MIN/_MAX macros in limits.h.
2076 Use of these macros can lead to spurious warnings as they do not
2077 necessarily reflect whether the code in question is any less portable to
2078 traditional C given that suitable backup definitions are provided.
2082 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2085 Warn whenever a local variable shadows another local variable.
2087 @item -Wid-clash-@var{len}
2088 Warn whenever two distinct identifiers match in the first @var{len}
2089 characters. This may help you prepare a program that will compile
2090 with certain obsolete, brain-damaged compilers.
2092 @item -Wlarger-than-@var{len}
2093 Warn whenever an object of larger than @var{len} bytes is defined.
2095 @item -Wpointer-arith
2096 Warn about anything that depends on the ``size of'' a function type or
2097 of @code{void}. GNU C assigns these types a size of 1, for
2098 convenience in calculations with @code{void *} pointers and pointers
2101 @item -Wbad-function-cast (C only)
2102 Warn whenever a function call is cast to a non-matching type.
2103 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2106 Warn whenever a pointer is cast so as to remove a type qualifier from
2107 the target type. For example, warn if a @code{const char *} is cast
2108 to an ordinary @code{char *}.
2111 Warn whenever a pointer is cast such that the required alignment of the
2112 target is increased. For example, warn if a @code{char *} is cast to
2113 an @code{int *} on machines where integers can only be accessed at
2114 two- or four-byte boundaries.
2116 @item -Wwrite-strings
2117 Give string constants the type @code{const char[@var{length}]} so that
2118 copying the address of one into a non-@code{const} @code{char *}
2119 pointer will get a warning. These warnings will help you find at
2120 compile time code that can try to write into a string constant, but
2121 only if you have been very careful about using @code{const} in
2122 declarations and prototypes. Otherwise, it will just be a nuisance;
2123 this is why we did not make @samp{-Wall} request these warnings.
2126 Warn if a prototype causes a type conversion that is different from what
2127 would happen to the same argument in the absence of a prototype. This
2128 includes conversions of fixed point to floating and vice versa, and
2129 conversions changing the width or signedness of a fixed point argument
2130 except when the same as the default promotion.
2132 Also, warn if a negative integer constant expression is implicitly
2133 converted to an unsigned type. For example, warn about the assignment
2134 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2135 casts like @code{(unsigned) -1}.
2137 @item -Wsign-compare
2138 @cindex warning for comparison of signed and unsigned values
2139 @cindex comparison of signed and unsigned values, warning
2140 @cindex signed and unsigned values, comparison warning
2141 Warn when a comparison between signed and unsigned values could produce
2142 an incorrect result when the signed value is converted to unsigned.
2143 This warning is also enabled by @samp{-W}; to get the other warnings
2144 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2146 @item -Waggregate-return
2147 Warn if any functions that return structures or unions are defined or
2148 called. (In languages where you can return an array, this also elicits
2151 @item -Wstrict-prototypes (C only)
2152 Warn if a function is declared or defined without specifying the
2153 argument types. (An old-style function definition is permitted without
2154 a warning if preceded by a declaration which specifies the argument
2157 @item -Wmissing-prototypes (C only)
2158 Warn if a global function is defined without a previous prototype
2159 declaration. This warning is issued even if the definition itself
2160 provides a prototype. The aim is to detect global functions that fail
2161 to be declared in header files.
2163 @item -Wmissing-declarations
2164 Warn if a global function is defined without a previous declaration.
2165 Do so even if the definition itself provides a prototype.
2166 Use this option to detect global functions that are not declared in
2169 @item -Wmissing-noreturn
2170 Warn about functions which might be candidates for attribute @code{noreturn}.
2171 Note these are only possible candidates, not absolute ones. Care should
2172 be taken to manually verify functions actually do not ever return before
2173 adding the @code{noreturn} attribute, otherwise subtle code generation
2174 bugs could be introduced. You will not get a warning for @code{main} in
2175 hosted C environments.
2177 @item -Wmissing-format-attribute
2178 If @samp{-Wformat} is enabled, also warn about functions which might be
2179 candidates for @code{format} attributes. Note these are only possible
2180 candidates, not absolute ones. GCC will guess that @code{format}
2181 attributes might be appropriate for any function that calls a function
2182 like @code{vprintf} or @code{vscanf}, but this might not always be the
2183 case, and some functions for which @code{format} attributes are
2184 appropriate may not be detected. This option has no effect unless
2185 @samp{-Wformat} is enabled (possibly by @samp{-Wall}).
2188 Warn if a structure is given the packed attribute, but the packed
2189 attribute has no effect on the layout or size of the structure.
2190 Such structures may be mis-aligned for little benefit. For
2191 instance, in this code, the variable @code{f.x} in @code{struct bar}
2192 will be misaligned even though @code{struct bar} does not itself
2193 have the packed attribute:
2200 @} __attribute__((packed));
2209 Warn if padding is included in a structure, either to align an element
2210 of the structure or to align the whole structure. Sometimes when this
2211 happens it is possible to rearrange the fields of the structure to
2212 reduce the padding and so make the structure smaller.
2214 @item -Wredundant-decls
2215 Warn if anything is declared more than once in the same scope, even in
2216 cases where multiple declaration is valid and changes nothing.
2218 @item -Wnested-externs (C only)
2219 Warn if an @code{extern} declaration is encountered within a function.
2221 @item -Wunreachable-code
2222 Warn if the compiler detects that code will never be executed.
2224 This option is intended to warn when the compiler detects that at
2225 least a whole line of source code will never be executed, because
2226 some condition is never satisfied or because it is after a
2227 procedure that never returns.
2229 It is possible for this option to produce a warning even though there
2230 are circumstances under which part of the affected line can be executed,
2231 so care should be taken when removing apparently-unreachable code.
2233 For instance, when a function is inlined, a warning may mean that the
2234 line is unreachable in only one inlined copy of the function.
2236 This option is not made part of @samp{-Wall} because in a debugging
2237 version of a program there is often substantial code which checks
2238 correct functioning of the program and is, hopefully, unreachable
2239 because the program does work. Another common use of unreachable
2240 code is to provide behaviour which is selectable at compile-time.
2243 Warn if a function can not be inlined and it was declared as inline.
2246 Warn if @samp{long long} type is used. This is default. To inhibit
2247 the warning messages, use @samp{-Wno-long-long}. Flags
2248 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
2249 only when @samp{-pedantic} flag is used.
2251 @item -Wdisabled-optimization
2252 Warn if a requested optimization pass is disabled. This warning does
2253 not generally indicate that there is anything wrong with your code; it
2254 merely indicates that GCC's optimizers were unable to handle the code
2255 effectively. Often, the problem is that your code is too big or too
2256 complex; GCC will refuse to optimize programs when the optimization
2257 itself is likely to take inordinate amounts of time.
2260 Make all warnings into errors.
2263 @node Debugging Options
2264 @section Options for Debugging Your Program or GCC
2265 @cindex options, debugging
2266 @cindex debugging information options
2268 GCC has various special options that are used for debugging
2269 either your program or GCC:
2273 Produce debugging information in the operating system's native format
2274 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
2277 On most systems that use stabs format, @samp{-g} enables use of extra
2278 debugging information that only GDB can use; this extra information
2279 makes debugging work better in GDB but will probably make other debuggers
2281 refuse to read the program. If you want to control for certain whether
2282 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
2283 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
2286 Unlike most other C compilers, GCC allows you to use @samp{-g} with
2287 @samp{-O}. The shortcuts taken by optimized code may occasionally
2288 produce surprising results: some variables you declared may not exist
2289 at all; flow of control may briefly move where you did not expect it;
2290 some statements may not be executed because they compute constant
2291 results or their values were already at hand; some statements may
2292 execute in different places because they were moved out of loops.
2294 Nevertheless it proves possible to debug optimized output. This makes
2295 it reasonable to use the optimizer for programs that might have bugs.
2297 The following options are useful when GCC is generated with the
2298 capability for more than one debugging format.
2301 Produce debugging information for use by GDB. This means to use the
2302 most expressive format available (DWARF 2, stabs, or the native format
2303 if neither of those are supported), including GDB extensions if at all
2307 Produce debugging information in stabs format (if that is supported),
2308 without GDB extensions. This is the format used by DBX on most BSD
2309 systems. On MIPS, Alpha and System V Release 4 systems this option
2310 produces stabs debugging output which is not understood by DBX or SDB.
2311 On System V Release 4 systems this option requires the GNU assembler.
2314 Produce debugging information in stabs format (if that is supported),
2315 using GNU extensions understood only by the GNU debugger (GDB). The
2316 use of these extensions is likely to make other debuggers crash or
2317 refuse to read the program.
2320 Produce debugging information in COFF format (if that is supported).
2321 This is the format used by SDB on most System V systems prior to
2325 Produce debugging information in XCOFF format (if that is supported).
2326 This is the format used by the DBX debugger on IBM RS/6000 systems.
2329 Produce debugging information in XCOFF format (if that is supported),
2330 using GNU extensions understood only by the GNU debugger (GDB). The
2331 use of these extensions is likely to make other debuggers crash or
2332 refuse to read the program, and may cause assemblers other than the GNU
2333 assembler (GAS) to fail with an error.
2336 Produce debugging information in DWARF version 1 format (if that is
2337 supported). This is the format used by SDB on most System V Release 4
2341 Produce debugging information in DWARF version 1 format (if that is
2342 supported), using GNU extensions understood only by the GNU debugger
2343 (GDB). The use of these extensions is likely to make other debuggers
2344 crash or refuse to read the program.
2347 Produce debugging information in DWARF version 2 format (if that is
2348 supported). This is the format used by DBX on IRIX 6.
2351 @itemx -ggdb@var{level}
2352 @itemx -gstabs@var{level}
2353 @itemx -gcoff@var{level}
2354 @itemx -gxcoff@var{level}
2355 @itemx -gdwarf@var{level}
2356 @itemx -gdwarf-2@var{level}
2357 Request debugging information and also use @var{level} to specify how
2358 much information. The default level is 2.
2360 Level 1 produces minimal information, enough for making backtraces in
2361 parts of the program that you don't plan to debug. This includes
2362 descriptions of functions and external variables, but no information
2363 about local variables and no line numbers.
2365 Level 3 includes extra information, such as all the macro definitions
2366 present in the program. Some debuggers support macro expansion when
2371 Generate extra code to write profile information suitable for the
2372 analysis program @code{prof}. You must use this option when compiling
2373 the source files you want data about, and you must also use it when
2376 @cindex @code{gprof}
2378 Generate extra code to write profile information suitable for the
2379 analysis program @code{gprof}. You must use this option when compiling
2380 the source files you want data about, and you must also use it when
2385 Generate extra code to write profile information for basic blocks, which will
2386 record the number of times each basic block is executed, the basic block start
2387 address, and the function name containing the basic block. If @samp{-g} is
2388 used, the line number and filename of the start of the basic block will also be
2389 recorded. If not overridden by the machine description, the default action is
2390 to append to the text file @file{bb.out}.
2392 This data could be analyzed by a program like @code{tcov}. Note,
2393 however, that the format of the data is not what @code{tcov} expects.
2394 Eventually GNU @code{gprof} should be extended to process this data.
2397 Makes the compiler print out each function name as it is compiled, and
2398 print some statistics about each pass when it finishes.
2401 Generate extra code to profile basic blocks. Your executable will
2402 produce output that is a superset of that produced when @samp{-a} is
2403 used. Additional output is the source and target address of the basic
2404 blocks where a jump takes place, the number of times a jump is executed,
2405 and (optionally) the complete sequence of basic blocks being executed.
2406 The output is appended to file @file{bb.out}.
2408 You can examine different profiling aspects without recompilation. Your
2409 executable will read a list of function names from file @file{bb.in}.
2410 Profiling starts when a function on the list is entered and stops when
2411 that invocation is exited. To exclude a function from profiling, prefix
2412 its name with `-'. If a function name is not unique, you can
2413 disambiguate it by writing it in the form
2414 @samp{/path/filename.d:functionname}. Your executable will write the
2415 available paths and filenames in file @file{bb.out}.
2417 Several function names have a special meaning:
2420 Write source, target and frequency of jumps to file @file{bb.out}.
2421 @item __bb_hidecall__
2422 Exclude function calls from frequency count.
2423 @item __bb_showret__
2424 Include function returns in frequency count.
2426 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2427 The file will be compressed using the program @samp{gzip}, which must
2428 exist in your @code{PATH}. On systems without the @samp{popen}
2429 function, the file will be named @file{bbtrace} and will not be
2430 compressed. @strong{Profiling for even a few seconds on these systems
2431 will produce a very large file.} Note: @code{__bb_hidecall__} and
2432 @code{__bb_showret__} will not affect the sequence written to
2436 Here's a short example using different profiling parameters
2437 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2438 1 and 2 and is called twice from block 3 of function @code{main}. After
2439 the calls, block 3 transfers control to block 4 of @code{main}.
2441 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2442 the following sequence of blocks is written to file @file{bbtrace.gz}:
2443 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2444 the return is to a point inside the block and not to the top. The
2445 block address 0 always indicates, that control is transferred
2446 to the trace from somewhere outside the observed functions. With
2447 @samp{-foo} added to @file{bb.in}, the blocks of function
2448 @code{foo} are removed from the trace, so only 0 3 4 remains.
2450 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2451 jump frequencies will be written to file @file{bb.out}. The
2452 frequencies are obtained by constructing a trace of blocks
2453 and incrementing a counter for every neighbouring pair of blocks
2454 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2458 Jump from block 0x0 to block 0x3 executed 1 time(s)
2459 Jump from block 0x3 to block 0x1 executed 1 time(s)
2460 Jump from block 0x1 to block 0x2 executed 2 time(s)
2461 Jump from block 0x2 to block 0x1 executed 1 time(s)
2462 Jump from block 0x2 to block 0x4 executed 1 time(s)
2465 With @code{__bb_hidecall__}, control transfer due to call instructions
2466 is removed from the trace, that is the trace is cut into three parts: 0
2467 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2468 to return instructions is added to the trace. The trace becomes: 0 3 1
2469 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2470 written to @file{bbtrace.gz}. It is solely used for counting jump
2473 @item -fprofile-arcs
2474 Instrument @dfn{arcs} during compilation. For each function of your
2475 program, GCC creates a program flow graph, then finds a spanning tree
2476 for the graph. Only arcs that are not on the spanning tree have to be
2477 instrumented: the compiler adds code to count the number of times that these
2478 arcs are executed. When an arc is the only exit or only entrance to a
2479 block, the instrumentation code can be added to the block; otherwise, a
2480 new basic block must be created to hold the instrumentation code.
2482 Since not every arc in the program must be instrumented, programs
2483 compiled with this option run faster than programs compiled with
2484 @samp{-a}, which adds instrumentation code to every basic block in the
2485 program. The tradeoff: since @code{gcov} does not have
2486 execution counts for all branches, it must start with the execution
2487 counts for the instrumented branches, and then iterate over the program
2488 flow graph until the entire graph has been solved. Hence, @code{gcov}
2489 runs a little more slowly than a program which uses information from
2492 @samp{-fprofile-arcs} also makes it possible to estimate branch
2493 probabilities, and to calculate basic block execution counts. In
2494 general, basic block execution counts do not give enough information to
2495 estimate all branch probabilities. When the compiled program exits, it
2496 saves the arc execution counts to a file called
2497 @file{@var{sourcename}.da}. Use the compiler option
2498 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2499 Control Optimization}) when recompiling, to optimize using estimated
2500 branch probabilities.
2503 @item -ftest-coverage
2504 Create data files for the @code{gcov} code-coverage utility
2505 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2506 The data file names begin with the name of your source file:
2509 @item @var{sourcename}.bb
2510 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2511 associate basic block execution counts with line numbers.
2513 @item @var{sourcename}.bbg
2514 A list of all arcs in the program flow graph. This allows @code{gcov}
2515 to reconstruct the program flow graph, so that it can compute all basic
2516 block and arc execution counts from the information in the
2517 @code{@var{sourcename}.da} file (this last file is the output from
2518 @samp{-fprofile-arcs}).
2521 @item -d@var{letters}
2522 Says to make debugging dumps during compilation at times specified by
2523 @var{letters}. This is used for debugging the compiler. The file names
2524 for most of the dumps are made by appending a pass number and a word to
2525 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2526 Here are the possible letters for use in @var{letters}, and their meanings:
2530 Annotate the assembler output with miscellaneous debugging information.
2532 Dump after computing branch probabilities, to @file{@var{file}.11.bp}.
2534 Dump after block reordering, to @file{@var{file}.26.bbro}.
2536 Dump after instruction combination, to the file @file{@var{file}.14.combine}.
2538 Dump after the first if conversion, to the file @file{@var{file}.15.ce}.
2540 Dump after delayed branch scheduling, to @file{@var{file}.29.dbr}.
2542 Dump all macro definitions, at the end of preprocessing, in addition to
2545 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
2546 @file{@var{file}.06.ussa}.
2548 Dump after the second if conversion, to @file{@var{file}.24.ce2}.
2550 Dump after life analysis, to @file{@var{file}.13.life}.
2552 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.04.addressof}.
2554 Dump after global register allocation, to @file{@var{file}.19.greg}.
2556 Dump after post-reload CSE and other optimizations, to @file{@var{file}.20.postreload}.
2558 Dump after GCSE, to @file{@var{file}.08.gcse}.
2560 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2562 Dump after the first jump optimization, to @file{@var{file}.02.jump}.
2564 Dump after the last jump optimization, to @file{@var{file}.27.jump2}.
2566 Dump after conversion from registers to stack, to @file{@var{file}.29.stack}.
2568 Dump after local register allocation, to @file{@var{file}.18.lreg}.
2570 Dump after loop optimization, to @file{@var{file}.09.loop}.
2572 Dump after performing the machine dependent reorganisation pass, to
2573 @file{@var{file}.28.mach}.
2575 Dump after register renumbering, to @file{@var{file}.23.rnreg}.
2577 Dump after the register move pass, to @file{@var{file}.16.regmove}.
2579 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2581 Dump after the second instruction scheduling pass, to
2582 @file{@var{file}.25.sched2}.
2584 Dump after CSE (including the jump optimization that sometimes follows
2585 CSE), to @file{@var{file}.03.cse}.
2587 Dump after the first instruction scheduling pass, to
2588 @file{@var{file}.17.sched}.
2590 Dump after the second CSE pass (including the jump optimization that
2591 sometimes follows CSE), to @file{@var{file}.10.cse2}.
2593 Dump after the second flow pass, to @file{@var{file}.21.flow2}.
2595 Dump after dead code elimination, to @file{@var{file}.06.dce}.
2597 Dump after the peephole pass, to @file{@var{file}.22.peephole2}.
2599 Produce all the dumps listed above.
2601 Print statistics on memory usage, at the end of the run, to
2604 Annotate the assembler output with a comment indicating which
2605 pattern and alternative was used. The length of each instruction is
2608 Dump the RTL in the assembler output as a comment before each instruction.
2609 Also turns on @samp{-dp} annotation.
2611 For each of the other indicated dump files (except for
2612 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2613 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2615 Just generate RTL for a function instead of compiling it. Usually used
2618 Dump debugging information during parsing, to standard error.
2621 @item -fdump-unnumbered
2622 When doing debugging dumps (see -d option above), suppress instruction
2623 numbers and line number note output. This makes it more feasible to
2624 use diff on debugging dumps for compiler invocations with different
2625 options, in particular with and without -g.
2627 @item -fdump-translation-unit-@var{file} (C and C++ only)
2628 Dump a representation of the tree structure for the entire translation
2631 @item -fpretend-float
2632 When running a cross-compiler, pretend that the target machine uses the
2633 same floating point format as the host machine. This causes incorrect
2634 output of the actual floating constants, but the actual instruction
2635 sequence will probably be the same as GCC would make when running on
2639 Store the usual ``temporary'' intermediate files permanently; place them
2640 in the current directory and name them based on the source file. Thus,
2641 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2642 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
2643 preprocessed @file{foo.i} output file even though the compiler now
2644 normally uses an integrated preprocessor.
2647 Report the CPU time taken by each subprocess in the compilation
2648 sequence. For C source files, this is the compiler proper and assembler
2649 (plus the linker if linking is done). The output looks like this:
2656 The first number on each line is the ``user time,'' that is time spent
2657 executing the program itself. The second number is ``system time,''
2658 time spent executing operating system routines on behalf of the program.
2659 Both numbers are in seconds.
2661 @item -print-file-name=@var{library}
2662 Print the full absolute name of the library file @var{library} that
2663 would be used when linking---and don't do anything else. With this
2664 option, GCC does not compile or link anything; it just prints the
2667 @item -print-prog-name=@var{program}
2668 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2670 @item -print-libgcc-file-name
2671 Same as @samp{-print-file-name=libgcc.a}.
2673 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2674 but you do want to link with @file{libgcc.a}. You can do
2677 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2680 @item -print-search-dirs
2681 Print the name of the configured installation directory and a list of
2682 program and library directories gcc will search---and don't do anything else.
2684 This is useful when gcc prints the error message
2685 @samp{installation problem, cannot exec cpp0: No such file or directory}.
2686 To resolve this you either need to put @file{cpp0} and the other compiler
2687 components where gcc expects to find them, or you can set the environment
2688 variable @code{GCC_EXEC_PREFIX} to the directory where you installed them.
2689 Don't forget the trailing '/'.
2690 @xref{Environment Variables}.
2693 @node Optimize Options
2694 @section Options That Control Optimization
2695 @cindex optimize options
2696 @cindex options, optimization
2698 These options control various sorts of optimizations:
2703 Optimize. Optimizing compilation takes somewhat more time, and a lot
2704 more memory for a large function.
2706 Without @samp{-O}, the compiler's goal is to reduce the cost of
2707 compilation and to make debugging produce the expected results.
2708 Statements are independent: if you stop the program with a breakpoint
2709 between statements, you can then assign a new value to any variable or
2710 change the program counter to any other statement in the function and
2711 get exactly the results you would expect from the source code.
2713 Without @samp{-O}, the compiler only allocates variables declared
2714 @code{register} in registers. The resulting compiled code is a little
2715 worse than produced by PCC without @samp{-O}.
2717 With @samp{-O}, the compiler tries to reduce code size and execution
2720 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2721 and @samp{-fdefer-pop} on all machines. The compiler turns on
2722 @samp{-fdelayed-branch} on machines that have delay slots, and
2723 @samp{-fomit-frame-pointer} on machines that can support debugging even
2724 without a frame pointer. On some machines the compiler also turns
2725 on other flags.@refill
2728 Optimize even more. GCC performs nearly all supported optimizations
2729 that do not involve a space-speed tradeoff. The compiler does not
2730 perform loop unrolling or function inlining when you specify @samp{-O2}.
2731 As compared to @samp{-O}, this option increases both compilation time
2732 and the performance of the generated code.
2734 @samp{-O2} turns on all optional optimizations except for loop unrolling,
2735 function inlining, and register renaming. It also turns on the
2736 @samp{-fforce-mem} option on all machines and frame pointer elimination
2737 on machines where doing so does not interfere with debugging.
2740 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2741 @samp{-O2} and also turns on the @samp{-finline-functions} and
2742 @samp{-frename-registers} options.
2748 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2749 do not typically increase code size. It also performs further
2750 optimizations designed to reduce code size.
2752 If you use multiple @samp{-O} options, with or without level numbers,
2753 the last such option is the one that is effective.
2756 Options of the form @samp{-f@var{flag}} specify machine-independent
2757 flags. Most flags have both positive and negative forms; the negative
2758 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2759 only one of the forms is listed---the one which is not the default.
2760 You can figure out the other form by either removing @samp{no-} or
2765 Do not store floating point variables in registers, and inhibit other
2766 options that might change whether a floating point value is taken from a
2769 @cindex floating point precision
2770 This option prevents undesirable excess precision on machines such as
2771 the 68000 where the floating registers (of the 68881) keep more
2772 precision than a @code{double} is supposed to have. Similarly for the
2773 x86 architecture. For most programs, the excess precision does only
2774 good, but a few programs rely on the precise definition of IEEE floating
2775 point. Use @samp{-ffloat-store} for such programs, after modifying
2776 them to store all pertinent intermediate computations into variables.
2778 @item -fno-default-inline
2779 Do not make member functions inline by default merely because they are
2780 defined inside the class scope (C++ only). Otherwise, when you specify
2781 @w{@samp{-O}}, member functions defined inside class scope are compiled
2782 inline by default; i.e., you don't need to add @samp{inline} in front of
2783 the member function name.
2785 @item -fno-defer-pop
2786 Always pop the arguments to each function call as soon as that function
2787 returns. For machines which must pop arguments after a function call,
2788 the compiler normally lets arguments accumulate on the stack for several
2789 function calls and pops them all at once.
2792 Force memory operands to be copied into registers before doing
2793 arithmetic on them. This produces better code by making all memory
2794 references potential common subexpressions. When they are not common
2795 subexpressions, instruction combination should eliminate the separate
2796 register-load. The @samp{-O2} option turns on this option.
2799 Force memory address constants to be copied into registers before
2800 doing arithmetic on them. This may produce better code just as
2801 @samp{-fforce-mem} may.
2803 @item -fomit-frame-pointer
2804 Don't keep the frame pointer in a register for functions that
2805 don't need one. This avoids the instructions to save, set up and
2806 restore frame pointers; it also makes an extra register available
2807 in many functions. @strong{It also makes debugging impossible on
2811 On some machines, such as the Vax, this flag has no effect, because
2812 the standard calling sequence automatically handles the frame pointer
2813 and nothing is saved by pretending it doesn't exist. The
2814 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2815 whether a target machine supports this flag. @xref{Registers}.@refill
2818 On some machines, such as the Vax, this flag has no effect, because
2819 the standard calling sequence automatically handles the frame pointer
2820 and nothing is saved by pretending it doesn't exist. The
2821 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2822 whether a target machine supports this flag. @xref{Registers,,Register
2823 Usage, gcc.info, Using and Porting GCC}.@refill
2826 @item -foptimize-sibling-calls
2827 Optimize sibling and tail recursive calls.
2830 This option generates traps for signed overflow on addition, subtraction,
2831 multiplication operations.
2834 Don't pay attention to the @code{inline} keyword. Normally this option
2835 is used to keep the compiler from expanding any functions inline.
2836 Note that if you are not optimizing, no functions can be expanded inline.
2838 @item -finline-functions
2839 Integrate all simple functions into their callers. The compiler
2840 heuristically decides which functions are simple enough to be worth
2841 integrating in this way.
2843 If all calls to a given function are integrated, and the function is
2844 declared @code{static}, then the function is normally not output as
2845 assembler code in its own right.
2847 @item -finline-limit=@var{n}
2848 By default, gcc limits the size of functions that can be inlined. This flag
2849 allows the control of this limit for functions that are explicitly marked as
2850 inline (ie marked with the inline keyword or defined within the class
2851 definition in c++). @var{n} is the size of functions that can be inlined in
2852 number of pseudo instructions (not counting parameter handling). The default
2853 value of n is 10000. Increasing this value can result in more inlined code at
2854 the cost of compilation time and memory consumption. Decreasing usually makes
2855 the compilation faster and less code will be inlined (which presumably
2856 means slower programs). This option is particularly useful for programs that
2857 use inlining heavily such as those based on recursive templates with c++.
2859 @emph{Note:} pseudo instruction represents, in this particular context, an
2860 abstract measurement of function's size. In no way, it represents a count
2861 of assembly instructions and as such its exact meaning might change from one
2862 release to an another.
2864 @item -fkeep-inline-functions
2865 Even if all calls to a given function are integrated, and the function
2866 is declared @code{static}, nevertheless output a separate run-time
2867 callable version of the function. This switch does not affect
2868 @code{extern inline} functions.
2870 @item -fkeep-static-consts
2871 Emit variables declared @code{static const} when optimization isn't turned
2872 on, even if the variables aren't referenced.
2874 GCC enables this option by default. If you want to force the compiler to
2875 check if the variable was referenced, regardless of whether or not
2876 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2878 @item -fno-function-cse
2879 Do not put function addresses in registers; make each instruction that
2880 calls a constant function contain the function's address explicitly.
2882 This option results in less efficient code, but some strange hacks
2883 that alter the assembler output may be confused by the optimizations
2884 performed when this option is not used.
2887 This option allows GCC to violate some ISO or IEEE rules and/or
2888 specifications in the interest of optimizing code for speed. For
2889 example, it allows the compiler to assume arguments to the @code{sqrt}
2890 function are non-negative numbers and that no floating-point values
2893 This option should never be turned on by any @samp{-O} option since
2894 it can result in incorrect output for programs which depend on
2895 an exact implementation of IEEE or ISO rules/specifications for
2898 @item -fno-math-errno
2899 Do not set ERRNO after calling math functions that are executed
2900 with a single instruction, e.g., sqrt. A program that relies on
2901 IEEE exceptions for math error handling may want to use this flag
2902 for speed while maintaining IEEE arithmetic compatibility.
2904 The default is @samp{-fmath-errno}. The @samp{-ffast-math} option
2905 sets @samp{-fno-math-errno}.
2908 @c following causes underfulls.. they don't look great, but we deal.
2910 The following options control specific optimizations. The @samp{-O2}
2911 option turns on all of these optimizations except @samp{-funroll-loops}
2912 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
2913 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
2914 but specific machines may handle it differently.
2916 You can use the following flags in the rare cases when ``fine-tuning''
2917 of optimizations to be performed is desired.
2920 @item -fstrength-reduce
2921 Perform the optimizations of loop strength reduction and
2922 elimination of iteration variables.
2924 @item -fthread-jumps
2925 Perform optimizations where we check to see if a jump branches to a
2926 location where another comparison subsumed by the first is found. If
2927 so, the first branch is redirected to either the destination of the
2928 second branch or a point immediately following it, depending on whether
2929 the condition is known to be true or false.
2931 @item -fcse-follow-jumps
2932 In common subexpression elimination, scan through jump instructions
2933 when the target of the jump is not reached by any other path. For
2934 example, when CSE encounters an @code{if} statement with an
2935 @code{else} clause, CSE will follow the jump when the condition
2938 @item -fcse-skip-blocks
2939 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
2940 follow jumps which conditionally skip over blocks. When CSE
2941 encounters a simple @code{if} statement with no else clause,
2942 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
2943 body of the @code{if}.
2945 @item -frerun-cse-after-loop
2946 Re-run common subexpression elimination after loop optimizations has been
2949 @item -frerun-loop-opt
2950 Run the loop optimizer twice.
2953 Perform a global common subexpression elimination pass.
2954 This pass also performs global constant and copy propagation.
2956 @item -fdelete-null-pointer-checks
2957 Use global dataflow analysis to identify and eliminate useless null
2958 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
2959 halting the program may not work properly with this option. Use
2960 -fno-delete-null-pointer-checks to disable this optimizing for programs
2961 which depend on that behavior.
2963 @item -fexpensive-optimizations
2964 Perform a number of minor optimizations that are relatively expensive.
2966 @item -foptimize-register-move
2968 Attempt to reassign register numbers in move instructions and as
2969 operands of other simple instructions in order to maximize the amount of
2970 register tying. This is especially helpful on machines with two-operand
2971 instructions. GCC enables this optimization by default with @samp{-O2}
2974 Note @code{-fregmove} and @code{-foptimize-register-move} are the same
2977 @item -fdelayed-branch
2978 If supported for the target machine, attempt to reorder instructions
2979 to exploit instruction slots available after delayed branch
2982 @item -fschedule-insns
2983 If supported for the target machine, attempt to reorder instructions to
2984 eliminate execution stalls due to required data being unavailable. This
2985 helps machines that have slow floating point or memory load instructions
2986 by allowing other instructions to be issued until the result of the load
2987 or floating point instruction is required.
2989 @item -fschedule-insns2
2990 Similar to @samp{-fschedule-insns}, but requests an additional pass of
2991 instruction scheduling after register allocation has been done. This is
2992 especially useful on machines with a relatively small number of
2993 registers and where memory load instructions take more than one cycle.
2995 @item -ffunction-sections
2996 @itemx -fdata-sections
2997 Place each function or data item into its own section in the output
2998 file if the target supports arbitrary sections. The name of the
2999 function or the name of the data item determines the section's name
3002 Use these options on systems where the linker can perform optimizations
3003 to improve locality of reference in the instruction space. HPPA
3004 processors running HP-UX and Sparc processors running Solaris 2 have
3005 linkers with such optimizations. Other systems using the ELF object format
3006 as well as AIX may have these optimizations in the future.
3008 Only use these options when there are significant benefits from doing
3009 so. When you specify these options, the assembler and linker will
3010 create larger object and executable files and will also be slower.
3011 You will not be able to use @code{gprof} on all systems if you
3012 specify this option and you may have problems with debugging if
3013 you specify both this option and @samp{-g}.
3015 @item -fcaller-saves
3016 Enable values to be allocated in registers that will be clobbered by
3017 function calls, by emitting extra instructions to save and restore the
3018 registers around such calls. Such allocation is done only when it
3019 seems to result in better code than would otherwise be produced.
3021 This option is always enabled by default on certain machines, usually
3022 those which have no call-preserved registers to use instead.
3024 For all machines, optimization level 2 and higher enables this flag by
3027 @item -funroll-loops
3028 Perform the optimization of loop unrolling. This is only done for loops
3029 whose number of iterations can be determined at compile time or run time.
3030 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
3031 @samp{-frerun-cse-after-loop}.
3033 @item -funroll-all-loops
3034 Perform the optimization of loop unrolling. This is done for all loops
3035 and usually makes programs run more slowly. @samp{-funroll-all-loops}
3036 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
3038 @item -fmove-all-movables
3039 Forces all invariant computations in loops to be moved
3042 @item -freduce-all-givs
3043 Forces all general-induction variables in loops to be
3046 @emph{Note:} When compiling programs written in Fortran,
3047 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
3048 by default when you use the optimizer.
3050 These options may generate better or worse code; results are highly
3051 dependent on the structure of loops within the source code.
3053 These two options are intended to be removed someday, once
3054 they have helped determine the efficacy of various
3055 approaches to improving loop optimizations.
3057 Please let us (@email{gcc@@gcc.gnu.org} and @email{fortran@@gnu.org})
3058 know how use of these options affects
3059 the performance of your production code.
3060 We're very interested in code that runs @emph{slower}
3061 when these options are @emph{enabled}.
3064 Disable any machine-specific peephole optimizations.
3066 @item -fbranch-probabilities
3067 After running a program compiled with @samp{-fprofile-arcs}
3068 (@pxref{Debugging Options,, Options for Debugging Your Program or
3069 @code{gcc}}), you can compile it a second time using
3070 @samp{-fbranch-probabilities}, to improve optimizations based on
3071 guessing the path a branch might take.
3074 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3075 note on the first instruction of each basic block, and a
3076 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3077 These can be used to improve optimization. Currently, they are only
3078 used in one place: in @file{reorg.c}, instead of guessing which path a
3079 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3080 exactly determine which path is taken more often.
3083 @item -fstrict-aliasing
3084 Allows the compiler to assume the strictest aliasing rules applicable to
3085 the language being compiled. For C (and C++), this activates
3086 optimizations based on the type of expressions. In particular, an
3087 object of one type is assumed never to reside at the same address as an
3088 object of a different type, unless the types are almost the same. For
3089 example, an @code{unsigned int} can alias an @code{int}, but not a
3090 @code{void*} or a @code{double}. A character type may alias any other
3093 Pay special attention to code like this:
3106 The practice of reading from a different union member than the one most
3107 recently written to (called ``type-punning'') is common. Even with
3108 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
3109 is accessed through the union type. So, the code above will work as
3110 expected. However, this code might not:
3122 Every language that wishes to perform language-specific alias analysis
3123 should define a function that computes, given an @code{tree}
3124 node, an alias set for the node. Nodes in different alias sets are not
3125 allowed to alias. For an example, see the C front-end function
3126 @code{c_get_alias_set}.
3129 @item -falign-functions
3130 @itemx -falign-functions=@var{n}
3131 Align the start of functions to the next power-of-two greater than
3132 @var{n}, skipping up to @var{n} bytes. For instance,
3133 @samp{-falign-functions=32} aligns functions to the next 32-byte
3134 boundary, but @samp{-falign-functions=24} would align to the next
3135 32-byte boundary only if this can be done by skipping 23 bytes or less.
3137 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
3138 equivalent and mean that functions will not be aligned.
3140 Some assemblers only support this flag when @var{n} is a power of two;
3141 in that case, it is rounded up.
3143 If @var{n} is not specified, use a machine-dependent default.
3145 @item -falign-labels
3146 @itemx -falign-labels=@var{n}
3147 Align all branch targets to a power-of-two boundary, skipping up to
3148 @var{n} bytes like @samp{-falign-functions}. This option can easily
3149 make code slower, because it must insert dummy operations for when the
3150 branch target is reached in the usual flow of the code.
3152 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
3153 are greater than this value, then their values are used instead.
3155 If @var{n} is not specified, use a machine-dependent default which is
3156 very likely to be @samp{1}, meaning no alignment.
3159 @itemx -falign-loops=@var{n}
3160 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3161 like @samp{-falign-functions}. The hope is that the loop will be
3162 executed many times, which will make up for any execution of the dummy
3165 If @var{n} is not specified, use a machine-dependent default.
3168 @itemx -falign-jumps=@var{n}
3169 Align branch targets to a power-of-two boundary, for branch targets
3170 where the targets can only be reached by jumping, skipping up to @var{n}
3171 bytes like @samp{-falign-functions}. In this case, no dummy operations
3174 If @var{n} is not specified, use a machine-dependent default.
3177 Perform optimizations in static single assignment form. Each function's
3178 flow graph is translated into SSA form, optimizations are performed, and
3179 the flow graph is translated back from SSA form. User's should not
3180 specify this option, since it is not yet ready for production use.
3183 Perform dead-code elimination in SSA form. Requires @samp{-fssa}. Like
3184 @samp{-fssa}, this is an experimental feature.
3186 @item -fsingle-precision-constant
3187 Treat floating point constant as single precision constant instead of
3188 implicitly converting it to double precision constant.
3190 @item -frename-registers
3191 Attempt to avoid false dependancies in scheduled code by making use
3192 of registers left over after register allocation. This optimization
3193 will most benefit processors with lots of registers. It can, however,
3194 make debugging impossible, since variables will no longer stay in
3195 a ``home register''.
3198 @node Preprocessor Options
3199 @section Options Controlling the Preprocessor
3200 @cindex preprocessor options
3201 @cindex options, preprocessor
3203 These options control the C preprocessor, which is run on each C source
3204 file before actual compilation.
3206 If you use the @samp{-E} option, nothing is done except preprocessing.
3207 Some of these options make sense only together with @samp{-E} because
3208 they cause the preprocessor output to be unsuitable for actual
3212 @item -include @var{file}
3213 Process @var{file} as input before processing the regular input file.
3214 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
3215 and @samp{-U} options on the command line are always processed before
3216 @samp{-include @var{file}}, regardless of the order in which they are
3217 written. All the @samp{-include} and @samp{-imacros} options are
3218 processed in the order in which they are written.
3220 @item -imacros @var{file}
3221 Process @var{file} as input, discarding the resulting output, before
3222 processing the regular input file. Because the output generated from
3223 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
3224 is to make the macros defined in @var{file} available for use in the
3227 Any @samp{-D} and @samp{-U} options on the command line are always
3228 processed before @samp{-imacros @var{file}}, regardless of the order in
3229 which they are written. All the @samp{-include} and @samp{-imacros}
3230 options are processed in the order in which they are written.
3232 @item -idirafter @var{dir}
3233 @cindex second include path
3234 Add the directory @var{dir} to the second include path. The directories
3235 on the second include path are searched when a header file is not found
3236 in any of the directories in the main include path (the one that
3239 @item -iprefix @var{prefix}
3240 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
3243 @item -iwithprefix @var{dir}
3244 Add a directory to the second include path. The directory's name is
3245 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3246 specified previously with @samp{-iprefix}. If you have not specified a
3247 prefix yet, the directory containing the installed passes of the
3248 compiler is used as the default.
3250 @item -iwithprefixbefore @var{dir}
3251 Add a directory to the main include path. The directory's name is made
3252 by concatenating @var{prefix} and @var{dir}, as in the case of
3253 @samp{-iwithprefix}.
3255 @item -isystem @var{dir}
3256 Add a directory to the beginning of the second include path, marking it
3257 as a system directory, so that it gets the same special treatment as
3258 is applied to the standard system directories.
3261 Do not search the standard system directories for header files. Only
3262 the directories you have specified with @samp{-I} options (and the
3263 current directory, if appropriate) are searched. @xref{Directory
3264 Options}, for information on @samp{-I}.
3266 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
3267 search path to only those directories you specify explicitly.
3270 Do not predefine any nonstandard macros. (Including architecture flags).
3273 Run only the C preprocessor. Preprocess all the C source files
3274 specified and output the results to standard output or to the
3275 specified output file.
3278 Tell the preprocessor not to discard comments. Used with the
3282 Tell the preprocessor not to generate @samp{#line} directives.
3283 Used with the @samp{-E} option.
3286 @cindex dependencies, make
3288 Tell the preprocessor to output a rule suitable for @code{make}
3289 describing the dependencies of each object file. For each source file,
3290 the preprocessor outputs one @code{make}-rule whose target is the object
3291 file name for that source file and whose dependencies are all the
3292 @code{#include} header files it uses. This rule may be a single line or
3293 may be continued with @samp{\}-newline if it is long. The list of rules
3294 is printed on standard output instead of the preprocessed C program.
3296 @samp{-M} implies @samp{-E}.
3298 Another way to specify output of a @code{make} rule is by setting
3299 the environment variable @code{DEPENDENCIES_OUTPUT} (@pxref{Environment
3303 Like @samp{-M} but the output mentions only the user header files
3304 included with @samp{#include "@var{file}"}. System header files
3305 included with @samp{#include <@var{file}>} are omitted.
3308 Like @samp{-M} but the dependency information is written to a file made by
3309 replacing ".c" with ".d" at the end of the input file names.
3310 This is in addition to compiling the file as specified---@samp{-MD} does
3311 not inhibit ordinary compilation the way @samp{-M} does.
3313 In Mach, you can use the utility @code{md} to merge multiple dependency
3314 files into a single dependency file suitable for using with the @samp{make}
3318 Like @samp{-MD} except mention only user header files, not system
3322 Treat missing header files as generated files and assume they live in the
3323 same directory as the source file. If you specify @samp{-MG}, you
3324 must also specify either @samp{-M} or @samp{-MM}. @samp{-MG} is not
3325 supported with @samp{-MD} or @samp{-MMD}.
3328 Print the name of each header file used, in addition to other normal
3331 @item -A@var{question}(@var{answer})
3332 Assert the answer @var{answer} for @var{question}, in case it is tested
3333 with a preprocessing conditional such as @samp{#if
3334 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
3335 assertions that normally describe the target machine.
3338 Define macro @var{macro} with the string @samp{1} as its definition.
3340 @item -D@var{macro}=@var{defn}
3341 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
3342 the command line are processed before any @samp{-U} options.
3345 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
3346 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
3350 Tell the preprocessor to output only a list of the macro definitions
3351 that are in effect at the end of preprocessing. Used with the @samp{-E}
3355 Tell the preprocessing to pass all macro definitions into the output, in
3356 their proper sequence in the rest of the output.
3359 Like @samp{-dD} except that the macro arguments and contents are omitted.
3360 Only @samp{#define @var{name}} is included in the output.
3363 Support ISO C trigraphs. The @samp{-ansi} option also has this effect.
3365 @item -Wp,@var{option}
3366 Pass @var{option} as an option to the preprocessor. If @var{option}
3367 contains commas, it is split into multiple options at the commas.
3370 @node Assembler Options
3371 @section Passing Options to the Assembler
3373 @c prevent bad page break with this line
3374 You can pass options to the assembler.
3377 @item -Wa,@var{option}
3378 Pass @var{option} as an option to the assembler. If @var{option}
3379 contains commas, it is split into multiple options at the commas.
3383 @section Options for Linking
3384 @cindex link options
3385 @cindex options, linking
3387 These options come into play when the compiler links object files into
3388 an executable output file. They are meaningless if the compiler is
3389 not doing a link step.
3393 @item @var{object-file-name}
3394 A file name that does not end in a special recognized suffix is
3395 considered to name an object file or library. (Object files are
3396 distinguished from libraries by the linker according to the file
3397 contents.) If linking is done, these object files are used as input
3403 If any of these options is used, then the linker is not run, and
3404 object file names should not be used as arguments. @xref{Overall
3408 @item -l@var{library}
3409 Search the library named @var{library} when linking.
3411 It makes a difference where in the command you write this option; the
3412 linker searches processes libraries and object files in the order they
3413 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
3414 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
3415 to functions in @samp{z}, those functions may not be loaded.
3417 The linker searches a standard list of directories for the library,
3418 which is actually a file named @file{lib@var{library}.a}. The linker
3419 then uses this file as if it had been specified precisely by name.
3421 The directories searched include several standard system directories
3422 plus any that you specify with @samp{-L}.
3424 Normally the files found this way are library files---archive files
3425 whose members are object files. The linker handles an archive file by
3426 scanning through it for members which define symbols that have so far
3427 been referenced but not defined. But if the file that is found is an
3428 ordinary object file, it is linked in the usual fashion. The only
3429 difference between using an @samp{-l} option and specifying a file name
3430 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
3431 and searches several directories.
3434 You need this special case of the @samp{-l} option in order to
3435 link an Objective C program.
3438 Do not use the standard system startup files when linking.
3439 The standard system libraries are used normally, unless @code{-nostdlib}
3440 or @code{-nodefaultlibs} is used.
3442 @item -nodefaultlibs
3443 Do not use the standard system libraries when linking.
3444 Only the libraries you specify will be passed to the linker.
3445 The standard startup files are used normally, unless @code{-nostartfiles}
3446 is used. The compiler may generate calls to memcmp, memset, and memcpy
3447 for System V (and ISO C) environments or to bcopy and bzero for
3448 BSD environments. These entries are usually resolved by entries in
3449 libc. These entry points should be supplied through some other
3450 mechanism when this option is specified.
3453 Do not use the standard system startup files or libraries when linking.
3454 No startup files and only the libraries you specify will be passed to
3455 the linker. The compiler may generate calls to memcmp, memset, and memcpy
3456 for System V (and ISO C) environments or to bcopy and bzero for
3457 BSD environments. These entries are usually resolved by entries in
3458 libc. These entry points should be supplied through some other
3459 mechanism when this option is specified.
3461 @cindex @code{-lgcc}, use with @code{-nostdlib}
3462 @cindex @code{-nostdlib} and unresolved references
3463 @cindex unresolved references and @code{-nostdlib}
3464 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
3465 @cindex @code{-nodefaultlibs} and unresolved references
3466 @cindex unresolved references and @code{-nodefaultlibs}
3467 One of the standard libraries bypassed by @samp{-nostdlib} and
3468 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
3469 that GCC uses to overcome shortcomings of particular machines, or special
3470 needs for some languages.
3472 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
3476 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
3477 for more discussion of @file{libgcc.a}.)
3479 In most cases, you need @file{libgcc.a} even when you want to avoid
3480 other standard libraries. In other words, when you specify @samp{-nostdlib}
3481 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
3482 This ensures that you have no unresolved references to internal GCC
3483 library subroutines. (For example, @samp{__main}, used to ensure C++
3484 constructors will be called; @pxref{Collect2,,@code{collect2}}.)
3487 Remove all symbol table and relocation information from the executable.
3490 On systems that support dynamic linking, this prevents linking with the shared
3491 libraries. On other systems, this option has no effect.
3494 Produce a shared object which can then be linked with other objects to
3495 form an executable. Not all systems support this option. For predictable
3496 results, you must also specify the same set of options that were used to
3497 generate code (@samp{-fpic}, @samp{-fPIC}, or model suboptions)
3498 when you specify this option.@footnote{On some systems, @code{gcc -shared}
3499 needs to build supplementary stub code for constructors to work. On
3500 multi-libbed systems, @code{gcc -shared} must select the correct support
3501 libraries to link against. Failing to supply the correct flags may lead
3502 to subtle defects. Supplying them in cases where they are not necessary
3506 Bind references to global symbols when building a shared object. Warn
3507 about any unresolved references (unless overridden by the link editor
3508 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
3511 @item -Xlinker @var{option}
3512 Pass @var{option} as an option to the linker. You can use this to
3513 supply system-specific linker options which GCC does not know how to
3516 If you want to pass an option that takes an argument, you must use
3517 @samp{-Xlinker} twice, once for the option and once for the argument.
3518 For example, to pass @samp{-assert definitions}, you must write
3519 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
3520 @samp{-Xlinker "-assert definitions"}, because this passes the entire
3521 string as a single argument, which is not what the linker expects.
3523 @item -Wl,@var{option}
3524 Pass @var{option} as an option to the linker. If @var{option} contains
3525 commas, it is split into multiple options at the commas.
3527 @item -u @var{symbol}
3528 Pretend the symbol @var{symbol} is undefined, to force linking of
3529 library modules to define it. You can use @samp{-u} multiple times with
3530 different symbols to force loading of additional library modules.
3533 @node Directory Options
3534 @section Options for Directory Search
3535 @cindex directory options
3536 @cindex options, directory search
3539 These options specify directories to search for header files, for
3540 libraries and for parts of the compiler:
3544 Add the directory @var{dir} to the head of the list of directories to be
3545 searched for header files. This can be used to override a system header
3546 file, substituting your own version, since these directories are
3547 searched before the system header file directories. If you use more
3548 than one @samp{-I} option, the directories are scanned in left-to-right
3549 order; the standard system directories come after.
3552 Any directories you specify with @samp{-I} options before the @samp{-I-}
3553 option are searched only for the case of @samp{#include "@var{file}"};
3554 they are not searched for @samp{#include <@var{file}>}.
3556 If additional directories are specified with @samp{-I} options after
3557 the @samp{-I-}, these directories are searched for all @samp{#include}
3558 directives. (Ordinarily @emph{all} @samp{-I} directories are used
3561 In addition, the @samp{-I-} option inhibits the use of the current
3562 directory (where the current input file came from) as the first search
3563 directory for @samp{#include "@var{file}"}. There is no way to
3564 override this effect of @samp{-I-}. With @samp{-I.} you can specify
3565 searching the directory which was current when the compiler was
3566 invoked. That is not exactly the same as what the preprocessor does
3567 by default, but it is often satisfactory.
3569 @samp{-I-} does not inhibit the use of the standard system directories
3570 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
3574 Add directory @var{dir} to the list of directories to be searched
3577 @item -B@var{prefix}
3578 This option specifies where to find the executables, libraries,
3579 include files, and data files of the compiler itself.
3581 The compiler driver program runs one or more of the subprograms
3582 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
3583 @var{prefix} as a prefix for each program it tries to run, both with and
3584 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
3586 For each subprogram to be run, the compiler driver first tries the
3587 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
3588 was not specified, the driver tries two standard prefixes, which are
3589 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
3590 those results in a file name that is found, the unmodified program
3591 name is searched for using the directories specified in your
3592 @samp{PATH} environment variable.
3594 @samp{-B} prefixes that effectively specify directory names also apply
3595 to libraries in the linker, because the compiler translates these
3596 options into @samp{-L} options for the linker. They also apply to
3597 includes files in the preprocessor, because the compiler translates these
3598 options into @samp{-isystem} options for the preprocessor. In this case,
3599 the compiler appends @samp{include} to the prefix.
3601 The run-time support file @file{libgcc.a} can also be searched for using
3602 the @samp{-B} prefix, if needed. If it is not found there, the two
3603 standard prefixes above are tried, and that is all. The file is left
3604 out of the link if it is not found by those means.
3606 Another way to specify a prefix much like the @samp{-B} prefix is to use
3607 the environment variable @code{GCC_EXEC_PREFIX}. @xref{Environment
3610 @item -specs=@var{file}
3611 Process @var{file} after the compiler reads in the standard @file{specs}
3612 file, in order to override the defaults that the @file{gcc} driver
3613 program uses when determining what switches to pass to @file{cc1},
3614 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
3615 @samp{-specs=}@var{file} can be specified on the command line, and they
3616 are processed in order, from left to right.
3622 @section Specifying subprocesses and the switches to pass to them
3624 @code{GCC} is a driver program. It performs its job by invoking a
3625 sequence of other programs to do the work of compiling, assembling and
3626 linking. GCC interprets its command-line parameters and uses these to
3627 deduce which programs it should invoke, and which command-line options
3628 it ought to place on their command lines. This behaviour is controlled
3629 by @dfn{spec strings}. In most cases there is one spec string for each
3630 program that GCC can invoke, but a few programs have multiple spec
3631 strings to control their behaviour. The spec strings built into GCC can
3632 be overridden by using the @samp{-specs=} command-line switch to specify
3635 @dfn{Spec files} are plaintext files that are used to construct spec
3636 strings. They consist of a sequence of directives separated by blank
3637 lines. The type of directive is determined by the first non-whitespace
3638 character on the line and it can be one of the following:
3641 @item %@var{command}
3642 Issues a @var{command} to the spec file processor. The commands that can
3646 @item %include <@var{file}>
3648 Search for @var{file} and insert its text at the current point in the
3651 @item %include_noerr <@var{file}>
3652 @cindex %include_noerr
3653 Just like @samp{%include}, but do not generate an error message if the include
3654 file cannot be found.
3656 @item %rename @var{old_name} @var{new_name}
3658 Rename the spec string @var{old_name} to @var{new_name}.
3662 @item *[@var{spec_name}]:
3663 This tells the compiler to create, override or delete the named spec
3664 string. All lines after this directive up to the next directive or
3665 blank line are considered to be the text for the spec string. If this
3666 results in an empty string then the spec will be deleted. (Or, if the
3667 spec did not exist, then nothing will happened.) Otherwise, if the spec
3668 does not currently exist a new spec will be created. If the spec does
3669 exist then its contents will be overridden by the text of this
3670 directive, unless the first character of that text is the @samp{+}
3671 character, in which case the text will be appended to the spec.
3673 @item [@var{suffix}]:
3674 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
3675 and up to the next directive or blank line are considered to make up the
3676 spec string for the indicated suffix. When the compiler encounters an
3677 input file with the named suffix, it will processes the spec string in
3678 order to work out how to compile that file. For example:
3685 This says that any input file whose name ends in @samp{.ZZ} should be
3686 passed to the program @samp{z-compile}, which should be invoked with the
3687 command-line switch @samp{-input} and with the result of performing the
3688 @samp{%i} substitution. (See below.)
3690 As an alternative to providing a spec string, the text that follows a
3691 suffix directive can be one of the following:
3694 @item @@@var{language}
3695 This says that the suffix is an alias for a known @var{language}. This is
3696 similar to using the @code{-x} command-line switch to GCC to specify a
3697 language explicitly. For example:
3704 Says that .ZZ files are, in fact, C++ source files.
3707 This causes an error messages saying:
3710 @var{name} compiler not installed on this system.
3714 GCC already has an extensive list of suffixes built into it.
3715 This directive will add an entry to the end of the list of suffixes, but
3716 since the list is searched from the end backwards, it is effectively
3717 possible to override earlier entries using this technique.
3721 GCC has the following spec strings built into it. Spec files can
3722 override these strings or create their own. Note that individual
3723 targets can also add their own spec strings to this list.
3726 asm Options to pass to the assembler
3727 asm_final Options to pass to the assembler post-processor
3728 cpp Options to pass to the C preprocessor
3729 cc1 Options to pass to the C compiler
3730 cc1plus Options to pass to the C++ compiler
3731 endfile Object files to include at the end of the link
3732 link Options to pass to the linker
3733 lib Libraries to include on the command line to the linker
3734 libgcc Decides which GCC support library to pass to the linker
3735 linker Sets the name of the linker
3736 predefines Defines to be passed to the C preprocessor
3737 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
3738 startfile Object files to include at the start of the link
3741 Here is a small example of a spec file:
3747 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
3750 This example renames the spec called @samp{lib} to @samp{old_lib} and
3751 then overrides the previous definition of @samp{lib} with a new one.
3752 The new definition adds in some extra command-line options before
3753 including the text of the old definition.
3755 @dfn{Spec strings} are a list of command-line options to be passed to their
3756 corresponding program. In addition, the spec strings can contain
3757 @samp{%}-prefixed sequences to substitute variable text or to
3758 conditionally insert text into the command line. Using these constructs
3759 it is possible to generate quite complex command lines.
3761 Here is a table of all defined @samp{%}-sequences for spec
3762 strings. Note that spaces are not generated automatically around the
3763 results of expanding these sequences. Therefore you can concatenate them
3764 together or combine them with constant text in a single argument.
3768 Substitute one @samp{%} into the program name or argument.
3771 Substitute the name of the input file being processed.
3774 Substitute the basename of the input file being processed.
3775 This is the substring up to (and not including) the last period
3776 and not including the directory.
3779 Marks the argument containing or following the @samp{%d} as a
3780 temporary file name, so that that file will be deleted if GCC exits
3781 successfully. Unlike @samp{%g}, this contributes no text to the
3784 @item %g@var{suffix}
3785 Substitute a file name that has suffix @var{suffix} and is chosen
3786 once per compilation, and mark the argument in the same way as
3787 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
3788 name is now chosen in a way that is hard to predict even when previously
3789 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
3790 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
3791 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
3792 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
3793 was simply substituted with a file name chosen once per compilation,
3794 without regard to any appended suffix (which was therefore treated
3795 just like ordinary text), making such attacks more likely to succeed.
3797 @item %u@var{suffix}
3798 Like @samp{%g}, but generates a new temporary file name even if
3799 @samp{%u@var{suffix}} was already seen.
3801 @item %U@var{suffix}
3802 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
3803 new one if there is no such last file name. In the absence of any
3804 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
3805 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
3806 would involve the generation of two distinct file names, one
3807 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
3808 simply substituted with a file name chosen for the previous @samp{%u},
3809 without regard to any appended suffix.
3812 Marks the argument containing or following the @samp{%w} as the
3813 designated output file of this compilation. This puts the argument
3814 into the sequence of arguments that @samp{%o} will substitute later.
3817 Substitutes the names of all the output files, with spaces
3818 automatically placed around them. You should write spaces
3819 around the @samp{%o} as well or the results are undefined.
3820 @samp{%o} is for use in the specs for running the linker.
3821 Input files whose names have no recognized suffix are not compiled
3822 at all, but they are included among the output files, so they will
3826 Substitutes the suffix for object files. Note that this is
3827 handled specially when it immediately follows @samp{%g, %u, or %U},
3828 because of the need for those to form complete file names. The
3829 handling is such that @samp{%O} is treated exactly as if it had already
3830 been substituted, except that @samp{%g, %u, and %U} do not currently
3831 support additional @var{suffix} characters following @samp{%O} as they would
3832 following, for example, @samp{.o}.
3835 Substitutes the standard macro predefinitions for the
3836 current target machine. Use this when running @code{cpp}.
3839 Like @samp{%p}, but puts @samp{__} before and after the name of each
3840 predefined macro, except for macros that start with @samp{__} or with
3841 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
3845 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
3848 Current argument is the name of a library or startup file of some sort.
3849 Search for that file in a standard list of directories and substitute
3850 the full name found.
3853 Print @var{str} as an error message. @var{str} is terminated by a newline.
3854 Use this when inconsistent options are detected.
3857 Output @samp{-} if the input for the current command is coming from a pipe.
3860 Substitute the contents of spec string @var{name} at this point.
3863 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
3865 @item %x@{@var{option}@}
3866 Accumulate an option for @samp{%X}.
3869 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
3873 Output the accumulated assembler options specified by @samp{-Wa}.
3876 Output the accumulated preprocessor options specified by @samp{-Wp}.
3879 Substitute the major version number of GCC.
3880 (For version 2.9.5, this is 2.)
3883 Substitute the minor version number of GCC.
3884 (For version 2.9.5, this is 9.)
3887 Process the @code{asm} spec. This is used to compute the
3888 switches to be passed to the assembler.
3891 Process the @code{asm_final} spec. This is a spec string for
3892 passing switches to an assembler post-processor, if such a program is
3896 Process the @code{link} spec. This is the spec for computing the
3897 command line passed to the linker. Typically it will make use of the
3898 @samp{%L %G %S %D and %E} sequences.
3901 Dump out a @samp{-L} option for each directory that GCC believes might
3902 contain startup files. If the target supports multilibs then the
3903 current multilib directory will be prepended to each of these paths.
3906 Process the @code{lib} spec. This is a spec string for deciding which
3907 libraries should be included on the command line to the linker.
3910 Process the @code{libgcc} spec. This is a spec string for deciding
3911 which GCC support library should be included on the command line to the linker.
3914 Process the @code{startfile} spec. This is a spec for deciding which
3915 object files should be the first ones passed to the linker. Typically
3916 this might be a file named @file{crt0.o}.
3919 Process the @code{endfile} spec. This is a spec string that specifies
3920 the last object files that will be passed to the linker.
3923 Process the @code{cpp} spec. This is used to construct the arguments
3924 to be passed to the C preprocessor.
3927 Process the @code{signed_char} spec. This is intended to be used
3928 to tell cpp whether a char is signed. It typically has the definition:
3930 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
3934 Process the @code{cc1} spec. This is used to construct the options to be
3935 passed to the actual C compiler (@samp{cc1}).
3938 Process the @code{cc1plus} spec. This is used to construct the options to be
3939 passed to the actual C++ compiler (@samp{cc1plus}).
3942 Substitute the variable part of a matched option. See below.
3943 Note that each comma in the substituted string is replaced by
3947 Substitutes the @code{-S} switch, if that switch was given to GCC.
3948 If that switch was not specified, this substitutes nothing. Note that
3949 the leading dash is omitted when specifying this option, and it is
3950 automatically inserted if the substitution is performed. Thus the spec
3951 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
3952 and would output the command line option @samp{-foo}.
3954 @item %W@{@code{S}@}
3955 Like %@{@code{S}@} but mark last argument supplied within as a file to be
3958 @item %@{@code{S}*@}
3959 Substitutes all the switches specified to GCC whose names start
3960 with @code{-S}, but which also take an argument. This is used for
3961 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
3962 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
3963 text, including the space. Thus two arguments would be generated.
3965 @item %@{^@code{S}*@}
3966 Like %@{@code{S}*@}, but don't put a blank between a switch and its
3967 argument. Thus %@{^o*@} would only generate one argument, not two.
3969 @item %@{<@code{S}@}
3970 Remove all occurrences of @code{-S} from the command line. Note - this
3971 command is position dependent. @samp{%} commands in the spec string
3972 before this option will see @code{-S}, @samp{%} commands in the spec
3973 string after this option will not.
3975 @item %@{@code{S}*:@code{X}@}
3976 Substitutes @code{X} if one or more switches whose names start with
3977 @code{-S} are specified to GCC. Note that the tail part of the
3978 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
3979 for each occurrence of @samp{%*} within @code{X}.
3981 @item %@{@code{S}:@code{X}@}
3982 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
3984 @item %@{!@code{S}:@code{X}@}
3985 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
3987 @item %@{|@code{S}:@code{X}@}
3988 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
3990 @item %@{|!@code{S}:@code{X}@}
3991 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
3993 @item %@{.@code{S}:@code{X}@}
3994 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
3996 @item %@{!.@code{S}:@code{X}@}
3997 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
3999 @item %@{@code{S}|@code{P}:@code{X}@}
4000 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
4001 combined with @samp{!} and @samp{.} sequences as well, although they
4002 have a stronger binding than the @samp{|}. For example a spec string
4006 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4009 will output the following command-line options from the following input
4010 command-line options:
4015 -d fred.c -foo -baz -boggle
4016 -d jim.d -bar -baz -boggle
4021 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4022 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4023 or spaces, or even newlines. They are processed as usual, as described
4026 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
4027 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
4028 -W} switch is found later in the command line, the earlier switch
4029 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4030 letter, which passes all matching options.
4032 The character @samp{|} at the beginning of the predicate text is used to indicate
4033 that a command should be piped to the following command, but only if @samp{-pipe}
4036 It is built into GCC which switches take arguments and which do not.
4037 (You might think it would be useful to generalize this to allow each
4038 compiler's spec to say which switches take arguments. But this cannot
4039 be done in a consistent fashion. GCC cannot even decide which input
4040 files have been specified without knowing which switches take arguments,
4041 and it must know which input files to compile in order to tell which
4044 GCC also knows implicitly that arguments starting in @samp{-l} are to be
4045 treated as compiler output files, and passed to the linker in their
4046 proper position among the other output files.
4048 @c man begin OPTIONS
4050 @node Target Options
4051 @section Specifying Target Machine and Compiler Version
4052 @cindex target options
4053 @cindex cross compiling
4054 @cindex specifying machine version
4055 @cindex specifying compiler version and target machine
4056 @cindex compiler version, specifying
4057 @cindex target machine, specifying
4059 By default, GCC compiles code for the same type of machine that you
4060 are using. However, it can also be installed as a cross-compiler, to
4061 compile for some other type of machine. In fact, several different
4062 configurations of GCC, for different target machines, can be
4063 installed side by side. Then you specify which one to use with the
4066 In addition, older and newer versions of GCC can be installed side
4067 by side. One of them (probably the newest) will be the default, but
4068 you may sometimes wish to use another.
4071 @item -b @var{machine}
4072 The argument @var{machine} specifies the target machine for compilation.
4073 This is useful when you have installed GCC as a cross-compiler.
4075 The value to use for @var{machine} is the same as was specified as the
4076 machine type when configuring GCC as a cross-compiler. For
4077 example, if a cross-compiler was configured with @samp{configure
4078 i386v}, meaning to compile for an 80386 running System V, then you
4079 would specify @samp{-b i386v} to run that cross compiler.
4081 When you do not specify @samp{-b}, it normally means to compile for
4082 the same type of machine that you are using.
4084 @item -V @var{version}
4085 The argument @var{version} specifies which version of GCC to run.
4086 This is useful when multiple versions are installed. For example,
4087 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
4089 The default version, when you do not specify @samp{-V}, is the last
4090 version of GCC that you installed.
4093 The @samp{-b} and @samp{-V} options actually work by controlling part of
4094 the file name used for the executable files and libraries used for
4095 compilation. A given version of GCC, for a given target machine, is
4096 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
4098 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
4099 changing the names of these directories or adding alternate names (or
4100 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
4101 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
4102 80386} becomes an alias for @samp{-b i386v}.
4104 In one respect, the @samp{-b} or @samp{-V} do not completely change
4105 to a different compiler: the top-level driver program @code{gcc}
4106 that you originally invoked continues to run and invoke the other
4107 executables (preprocessor, compiler per se, assembler and linker)
4108 that do the real work. However, since no real work is done in the
4109 driver program, it usually does not matter that the driver program
4110 in use is not the one for the specified target. It is common for the
4111 interface to the other executables to change incompatibly between
4112 compiler versions, so unless the version specified is very close to that
4113 of the driver (for example, @samp{-V 3.0} with a driver program from GCC
4114 version 3.0.1), use of @samp{-V} may not work; for example, using
4115 @samp{-V 2.95.2} will not work with a driver program from GCC 3.0.
4117 The only way that the driver program depends on the target machine is
4118 in the parsing and handling of special machine-specific options.
4119 However, this is controlled by a file which is found, along with the
4120 other executables, in the directory for the specified version and
4121 target machine. As a result, a single installed driver program adapts
4122 to any specified target machine, and sufficiently similar compiler
4125 The driver program executable does control one significant thing,
4126 however: the default version and target machine. Therefore, you can
4127 install different instances of the driver program, compiled for
4128 different targets or versions, under different names.
4130 For example, if the driver for version 2.0 is installed as @code{ogcc}
4131 and that for version 2.1 is installed as @code{gcc}, then the command
4132 @code{gcc} will use version 2.1 by default, while @code{ogcc} will use
4133 2.0 by default. However, you can choose either version with either
4134 command with the @samp{-V} option.
4136 @node Submodel Options
4137 @section Hardware Models and Configurations
4138 @cindex submodel options
4139 @cindex specifying hardware config
4140 @cindex hardware models and configurations, specifying
4141 @cindex machine dependent options
4143 Earlier we discussed the standard option @samp{-b} which chooses among
4144 different installed compilers for completely different target
4145 machines, such as Vax vs. 68000 vs. 80386.
4147 In addition, each of these target machine types can have its own
4148 special options, starting with @samp{-m}, to choose among various
4149 hardware models or configurations---for example, 68010 vs 68020,
4150 floating coprocessor or none. A single installed version of the
4151 compiler can compile for any model or configuration, according to the
4154 Some configurations of the compiler also support additional special
4155 options, usually for compatibility with other compilers on the same
4159 These options are defined by the macro @code{TARGET_SWITCHES} in the
4160 machine description. The default for the options is also defined by
4161 that macro, which enables you to change the defaults.
4177 * RS/6000 and PowerPC Options::
4182 * Intel 960 Options::
4183 * DEC Alpha Options::
4187 * System V Options::
4188 * TMS320C3x/C4x Options::
4197 @node M680x0 Options
4198 @subsection M680x0 Options
4199 @cindex M680x0 options
4201 These are the @samp{-m} options defined for the 68000 series. The default
4202 values for these options depends on which style of 68000 was selected when
4203 the compiler was configured; the defaults for the most common choices are
4209 Generate output for a 68000. This is the default
4210 when the compiler is configured for 68000-based systems.
4212 Use this option for microcontrollers with a 68000 or EC000 core,
4213 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
4217 Generate output for a 68020. This is the default
4218 when the compiler is configured for 68020-based systems.
4221 Generate output containing 68881 instructions for floating point.
4222 This is the default for most 68020 systems unless @samp{-nfp} was
4223 specified when the compiler was configured.
4226 Generate output for a 68030. This is the default when the compiler is
4227 configured for 68030-based systems.
4230 Generate output for a 68040. This is the default when the compiler is
4231 configured for 68040-based systems.
4233 This option inhibits the use of 68881/68882 instructions that have to be
4234 emulated by software on the 68040. Use this option if your 68040 does not
4235 have code to emulate those instructions.
4238 Generate output for a 68060. This is the default when the compiler is
4239 configured for 68060-based systems.
4241 This option inhibits the use of 68020 and 68881/68882 instructions that
4242 have to be emulated by software on the 68060. Use this option if your 68060
4243 does not have code to emulate those instructions.
4246 Generate output for a CPU32. This is the default
4247 when the compiler is configured for CPU32-based systems.
4249 Use this option for microcontrollers with a
4250 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
4251 68336, 68340, 68341, 68349 and 68360.
4254 Generate output for a 520X "coldfire" family cpu. This is the default
4255 when the compiler is configured for 520X-based systems.
4257 Use this option for microcontroller with a 5200 core, including
4258 the MCF5202, MCF5203, MCF5204 and MCF5202.
4262 Generate output for a 68040, without using any of the new instructions.
4263 This results in code which can run relatively efficiently on either a
4264 68020/68881 or a 68030 or a 68040. The generated code does use the
4265 68881 instructions that are emulated on the 68040.
4268 Generate output for a 68060, without using any of the new instructions.
4269 This results in code which can run relatively efficiently on either a
4270 68020/68881 or a 68030 or a 68040. The generated code does use the
4271 68881 instructions that are emulated on the 68060.
4274 Generate output containing Sun FPA instructions for floating point.
4277 Generate output containing library calls for floating point.
4278 @strong{Warning:} the requisite libraries are not available for all m68k
4279 targets. Normally the facilities of the machine's usual C compiler are
4280 used, but this can't be done directly in cross-compilation. You must
4281 make your own arrangements to provide suitable library functions for
4282 cross-compilation. The embedded targets @samp{m68k-*-aout} and
4283 @samp{m68k-*-coff} do provide software floating point support.
4286 Consider type @code{int} to be 16 bits wide, like @code{short int}.
4289 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
4290 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
4293 Do use the bit-field instructions. The @samp{-m68020} option implies
4294 @samp{-mbitfield}. This is the default if you use a configuration
4295 designed for a 68020.
4298 Use a different function-calling convention, in which functions
4299 that take a fixed number of arguments return with the @code{rtd}
4300 instruction, which pops their arguments while returning. This
4301 saves one instruction in the caller since there is no need to pop
4302 the arguments there.
4304 This calling convention is incompatible with the one normally
4305 used on Unix, so you cannot use it if you need to call libraries
4306 compiled with the Unix compiler.
4308 Also, you must provide function prototypes for all functions that
4309 take variable numbers of arguments (including @code{printf});
4310 otherwise incorrect code will be generated for calls to those
4313 In addition, seriously incorrect code will result if you call a
4314 function with too many arguments. (Normally, extra arguments are
4315 harmlessly ignored.)
4317 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
4318 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
4321 @itemx -mno-align-int
4322 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
4323 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
4324 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
4325 Aligning variables on 32-bit boundaries produces code that runs somewhat
4326 faster on processors with 32-bit busses at the expense of more memory.
4328 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
4329 align structures containing the above types differently than
4330 most published application binary interface specifications for the m68k.
4333 Use the pc-relative addressing mode of the 68000 directly, instead of
4334 using a global offset table. At present, this option implies -fpic,
4335 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
4336 not presently supported with -mpcrel, though this could be supported for
4337 68020 and higher processors.
4339 @item -mno-strict-align
4340 @itemx -mstrict-align
4341 @kindex -mstrict-align
4342 Do not (do) assume that unaligned memory references will be handled by
4347 @node M68hc1x Options
4348 @subsection M68hc1x Options
4349 @cindex M68hc1x options
4351 These are the @samp{-m} options defined for the 68hc11 and 68hc12
4352 microcontrollers. The default values for these options depends on
4353 which style of microcontroller was selected when the compiler was configured;
4354 the defaults for the most common choices are given below.
4359 Generate output for a 68HC11. This is the default
4360 when the compiler is configured for 68HC11-based systems.
4364 Generate output for a 68HC12. This is the default
4365 when the compiler is configured for 68HC12-based systems.
4368 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
4372 Consider type @code{int} to be 16 bits wide, like @code{short int}.
4374 @item -msoft-reg-count=@var{count}
4375 Specify the number of pseudo-soft registers which are used for the
4376 code generation. The maximum number is 32. Using more pseudo-soft
4377 register may or may not result in better code depending on the program.
4378 The default is 4 for 68HC11 and 2 for 68HC12.
4383 @subsection VAX Options
4386 These @samp{-m} options are defined for the Vax:
4390 Do not output certain jump instructions (@code{aobleq} and so on)
4391 that the Unix assembler for the Vax cannot handle across long
4395 Do output those jump instructions, on the assumption that you
4396 will assemble with the GNU assembler.
4399 Output code for g-format floating point numbers instead of d-format.
4403 @subsection SPARC Options
4404 @cindex SPARC options
4406 These @samp{-m} switches are supported on the SPARC:
4411 Specify @samp{-mapp-regs} to generate output using the global registers
4412 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
4415 To be fully SVR4 ABI compliant at the cost of some performance loss,
4416 specify @samp{-mno-app-regs}. You should compile libraries and system
4417 software with this option.
4421 Generate output containing floating point instructions. This is the
4426 Generate output containing library calls for floating point.
4427 @strong{Warning:} the requisite libraries are not available for all SPARC
4428 targets. Normally the facilities of the machine's usual C compiler are
4429 used, but this cannot be done directly in cross-compilation. You must make
4430 your own arrangements to provide suitable library functions for
4431 cross-compilation. The embedded targets @samp{sparc-*-aout} and
4432 @samp{sparclite-*-*} do provide software floating point support.
4434 @samp{-msoft-float} changes the calling convention in the output file;
4435 therefore, it is only useful if you compile @emph{all} of a program with
4436 this option. In particular, you need to compile @file{libgcc.a}, the
4437 library that comes with GCC, with @samp{-msoft-float} in order for
4440 @item -mhard-quad-float
4441 Generate output containing quad-word (long double) floating point
4444 @item -msoft-quad-float
4445 Generate output containing library calls for quad-word (long double)
4446 floating point instructions. The functions called are those specified
4447 in the SPARC ABI. This is the default.
4449 As of this writing, there are no sparc implementations that have hardware
4450 support for the quad-word floating point instructions. They all invoke
4451 a trap handler for one of these instructions, and then the trap handler
4452 emulates the effect of the instruction. Because of the trap handler overhead,
4453 this is much slower than calling the ABI library routines. Thus the
4454 @samp{-msoft-quad-float} option is the default.
4458 With @samp{-mepilogue} (the default), the compiler always emits code for
4459 function exit at the end of each function. Any function exit in
4460 the middle of the function (such as a return statement in C) will
4461 generate a jump to the exit code at the end of the function.
4463 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
4464 at every function exit.
4468 With @samp{-mflat}, the compiler does not generate save/restore instructions
4469 and will use a "flat" or single register window calling convention.
4470 This model uses %i7 as the frame pointer and is compatible with the normal
4471 register window model. Code from either may be intermixed.
4472 The local registers and the input registers (0-5) are still treated as
4473 "call saved" registers and will be saved on the stack as necessary.
4475 With @samp{-mno-flat} (the default), the compiler emits save/restore
4476 instructions (except for leaf functions) and is the normal mode of operation.
4478 @item -mno-unaligned-doubles
4479 @itemx -munaligned-doubles
4480 Assume that doubles have 8 byte alignment. This is the default.
4482 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
4483 alignment only if they are contained in another type, or if they have an
4484 absolute address. Otherwise, it assumes they have 4 byte alignment.
4485 Specifying this option avoids some rare compatibility problems with code
4486 generated by other compilers. It is not the default because it results
4487 in a performance loss, especially for floating point code.
4489 @item -mno-faster-structs
4490 @itemx -mfaster-structs
4491 With @samp{-mfaster-structs}, the compiler assumes that structures
4492 should have 8 byte alignment. This enables the use of pairs of
4493 @code{ldd} and @code{std} instructions for copies in structure
4494 assignment, in place of twice as many @code{ld} and @code{st} pairs.
4495 However, the use of this changed alignment directly violates the Sparc
4496 ABI. Thus, it's intended only for use on targets where the developer
4497 acknowledges that their resulting code will not be directly in line with
4498 the rules of the ABI.
4502 These two options select variations on the SPARC architecture.
4504 By default (unless specifically configured for the Fujitsu SPARClite),
4505 GCC generates code for the v7 variant of the SPARC architecture.
4507 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
4508 code is that the compiler emits the integer multiply and integer
4509 divide instructions which exist in SPARC v8 but not in SPARC v7.
4511 @samp{-msparclite} will give you SPARClite code. This adds the integer
4512 multiply, integer divide step and scan (@code{ffs}) instructions which
4513 exist in SPARClite but not in SPARC v7.
4515 These options are deprecated and will be deleted in a future GCC release.
4516 They have been replaced with @samp{-mcpu=xxx}.
4520 These two options select the processor for which the code is optimised.
4522 With @samp{-mcypress} (the default), the compiler optimizes code for the
4523 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
4524 This is also appropriate for the older SparcStation 1, 2, IPX etc.
4526 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
4527 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
4528 of the full SPARC v8 instruction set.
4530 These options are deprecated and will be deleted in a future GCC release.
4531 They have been replaced with @samp{-mcpu=xxx}.
4533 @item -mcpu=@var{cpu_type}
4534 Set the instruction set, register set, and instruction scheduling parameters
4535 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
4536 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
4537 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
4538 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
4540 Default instruction scheduling parameters are used for values that select
4541 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
4542 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
4544 Here is a list of each supported architecture and their supported
4549 v8: supersparc, hypersparc
4550 sparclite: f930, f934, sparclite86x
4555 @item -mtune=@var{cpu_type}
4556 Set the instruction scheduling parameters for machine type
4557 @var{cpu_type}, but do not set the instruction set or register set that the
4558 option @samp{-mcpu=}@var{cpu_type} would.
4560 The same values for @samp{-mcpu=}@var{cpu_type} are used for
4561 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
4562 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
4563 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
4564 @samp{tsc701}, @samp{ultrasparc}.
4568 These @samp{-m} switches are supported in addition to the above
4569 on the SPARCLET processor.
4572 @item -mlittle-endian
4573 Generate code for a processor running in little-endian mode.
4576 Treat register @code{%g0} as a normal register.
4577 GCC will continue to clobber it as necessary but will not assume
4578 it always reads as 0.
4580 @item -mbroken-saverestore
4581 Generate code that does not use non-trivial forms of the @code{save} and
4582 @code{restore} instructions. Early versions of the SPARCLET processor do
4583 not correctly handle @code{save} and @code{restore} instructions used with
4584 arguments. They correctly handle them used without arguments. A @code{save}
4585 instruction used without arguments increments the current window pointer
4586 but does not allocate a new stack frame. It is assumed that the window
4587 overflow trap handler will properly handle this case as will interrupt
4591 These @samp{-m} switches are supported in addition to the above
4592 on SPARC V9 processors in 64 bit environments.
4595 @item -mlittle-endian
4596 Generate code for a processor running in little-endian mode.
4600 Generate code for a 32 bit or 64 bit environment.
4601 The 32 bit environment sets int, long and pointer to 32 bits.
4602 The 64 bit environment sets int to 32 bits and long and pointer
4605 @item -mcmodel=medlow
4606 Generate code for the Medium/Low code model: the program must be linked
4607 in the low 32 bits of the address space. Pointers are 64 bits.
4608 Programs can be statically or dynamically linked.
4610 @item -mcmodel=medmid
4611 Generate code for the Medium/Middle code model: the program must be linked
4612 in the low 44 bits of the address space, the text segment must be less than
4613 2G bytes, and data segment must be within 2G of the text segment.
4614 Pointers are 64 bits.
4616 @item -mcmodel=medany
4617 Generate code for the Medium/Anywhere code model: the program may be linked
4618 anywhere in the address space, the text segment must be less than
4619 2G bytes, and data segment must be within 2G of the text segment.
4620 Pointers are 64 bits.
4622 @item -mcmodel=embmedany
4623 Generate code for the Medium/Anywhere code model for embedded systems:
4624 assume a 32 bit text and a 32 bit data segment, both starting anywhere
4625 (determined at link time). Register %g4 points to the base of the
4626 data segment. Pointers still 64 bits.
4627 Programs are statically linked, PIC is not supported.
4630 @itemx -mno-stack-bias
4631 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
4632 frame pointer if present, are offset by -2047 which must be added back
4633 when making stack frame references.
4634 Otherwise, assume no such offset is present.
4637 @node Convex Options
4638 @subsection Convex Options
4639 @cindex Convex options
4641 These @samp{-m} options are defined for Convex:
4645 Generate output for C1. The code will run on any Convex machine.
4646 The preprocessor symbol @code{__convex__c1__} is defined.
4649 Generate output for C2. Uses instructions not available on C1.
4650 Scheduling and other optimizations are chosen for max performance on C2.
4651 The preprocessor symbol @code{__convex_c2__} is defined.
4654 Generate output for C32xx. Uses instructions not available on C1.
4655 Scheduling and other optimizations are chosen for max performance on C32.
4656 The preprocessor symbol @code{__convex_c32__} is defined.
4659 Generate output for C34xx. Uses instructions not available on C1.
4660 Scheduling and other optimizations are chosen for max performance on C34.
4661 The preprocessor symbol @code{__convex_c34__} is defined.
4664 Generate output for C38xx. Uses instructions not available on C1.
4665 Scheduling and other optimizations are chosen for max performance on C38.
4666 The preprocessor symbol @code{__convex_c38__} is defined.
4669 Generate code which puts an argument count in the word preceding each
4670 argument list. This is compatible with regular CC, and a few programs
4671 may need the argument count word. GDB and other source-level debuggers
4672 do not need it; this info is in the symbol table.
4675 Omit the argument count word. This is the default.
4677 @item -mvolatile-cache
4678 Allow volatile references to be cached. This is the default.
4680 @item -mvolatile-nocache
4681 Volatile references bypass the data cache, going all the way to memory.
4682 This is only needed for multi-processor code that does not use standard
4683 synchronization instructions. Making non-volatile references to volatile
4684 locations will not necessarily work.
4687 Type long is 32 bits, the same as type int. This is the default.
4690 Type long is 64 bits, the same as type long long. This option is useless,
4691 because no library support exists for it.
4694 @node AMD29K Options
4695 @subsection AMD29K Options
4696 @cindex AMD29K options
4698 These @samp{-m} options are defined for the AMD Am29000:
4703 @cindex DW bit (29k)
4704 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
4705 halfword operations are directly supported by the hardware. This is the
4710 Generate code that assumes the @code{DW} bit is not set.
4714 @cindex byte writes (29k)
4715 Generate code that assumes the system supports byte and halfword write
4716 operations. This is the default.
4720 Generate code that assumes the systems does not support byte and
4721 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
4725 @cindex memory model (29k)
4726 Use a small memory model that assumes that all function addresses are
4727 either within a single 256 KB segment or at an absolute address of less
4728 than 256k. This allows the @code{call} instruction to be used instead
4729 of a @code{const}, @code{consth}, @code{calli} sequence.
4733 Use the normal memory model: Generate @code{call} instructions only when
4734 calling functions in the same file and @code{calli} instructions
4735 otherwise. This works if each file occupies less than 256 KB but allows
4736 the entire executable to be larger than 256 KB. This is the default.
4739 Always use @code{calli} instructions. Specify this option if you expect
4740 a single file to compile into more than 256 KB of code.
4744 @cindex processor selection (29k)
4745 Generate code for the Am29050.
4749 Generate code for the Am29000. This is the default.
4751 @item -mkernel-registers
4752 @kindex -mkernel-registers
4753 @cindex kernel and user registers (29k)
4754 Generate references to registers @code{gr64-gr95} instead of to
4755 registers @code{gr96-gr127}. This option can be used when compiling
4756 kernel code that wants a set of global registers disjoint from that used
4759 Note that when this option is used, register names in @samp{-f} flags
4760 must use the normal, user-mode, names.
4762 @item -muser-registers
4763 @kindex -muser-registers
4764 Use the normal set of global registers, @code{gr96-gr127}. This is the
4768 @itemx -mno-stack-check
4769 @kindex -mstack-check
4770 @cindex stack checks (29k)
4771 Insert (or do not insert) a call to @code{__msp_check} after each stack
4772 adjustment. This is often used for kernel code.
4775 @itemx -mno-storem-bug
4776 @kindex -mstorem-bug
4777 @cindex storem bug (29k)
4778 @samp{-mstorem-bug} handles 29k processors which cannot handle the
4779 separation of a mtsrim insn and a storem instruction (most 29000 chips
4780 to date, but not the 29050).
4782 @item -mno-reuse-arg-regs
4783 @itemx -mreuse-arg-regs
4784 @kindex -mreuse-arg-regs
4785 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
4786 registers for copying out arguments. This helps detect calling a function
4787 with fewer arguments than it was declared with.
4789 @item -mno-impure-text
4790 @itemx -mimpure-text
4791 @kindex -mimpure-text
4792 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
4793 not pass @samp{-assert pure-text} to the linker when linking a shared object.
4796 @kindex -msoft-float
4797 Generate output containing library calls for floating point.
4798 @strong{Warning:} the requisite libraries are not part of GCC.
4799 Normally the facilities of the machine's usual C compiler are used, but
4800 this can't be done directly in cross-compilation. You must make your
4801 own arrangements to provide suitable library functions for
4806 Do not generate multm or multmu instructions. This is useful for some embedded
4807 systems which do not have trap handlers for these instructions.
4811 @subsection ARM Options
4814 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
4819 @kindex -mapcs-frame
4820 Generate a stack frame that is compliant with the ARM Procedure Call
4821 Standard for all functions, even if this is not strictly necessary for
4822 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
4823 with this option will cause the stack frames not to be generated for
4824 leaf functions. The default is @samp{-mno-apcs-frame}.
4828 This is a synonym for @samp{-mapcs-frame}.
4832 Generate code for a processor running with a 26-bit program counter,
4833 and conforming to the function calling standards for the APCS 26-bit
4834 option. This option replaces the @samp{-m2} and @samp{-m3} options
4835 of previous releases of the compiler.
4839 Generate code for a processor running with a 32-bit program counter,
4840 and conforming to the function calling standards for the APCS 32-bit
4841 option. This option replaces the @samp{-m6} option of previous releases
4844 @item -mapcs-stack-check
4845 @kindex -mapcs-stack-check
4846 @kindex -mno-apcs-stack-check
4847 Generate code to check the amount of stack space available upon entry to
4848 every function (that actually uses some stack space). If there is
4849 insufficient space available then either the function
4850 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
4851 called, depending upon the amount of stack space required. The run time
4852 system is required to provide these functions. The default is
4853 @samp{-mno-apcs-stack-check}, since this produces smaller code.
4856 @kindex -mapcs-float
4857 @kindex -mno-apcs-float
4858 Pass floating point arguments using the float point registers. This is
4859 one of the variants of the APCS. This option is recommended if the
4860 target hardware has a floating point unit or if a lot of floating point
4861 arithmetic is going to be performed by the code. The default is
4862 @samp{-mno-apcs-float}, since integer only code is slightly increased in
4863 size if @samp{-mapcs-float} is used.
4865 @item -mapcs-reentrant
4866 @kindex -mapcs-reentrant
4867 @kindex -mno-apcs-reentrant
4868 Generate reentrant, position independent code. This is the equivalent
4869 to specifying the @samp{-fpic} option. The default is
4870 @samp{-mno-apcs-reentrant}.
4872 @item -mthumb-interwork
4873 @kindex -mthumb-interwork
4874 @kindex -mno-thumb-interwork
4875 Generate code which supports calling between the ARM and THUMB
4876 instruction sets. Without this option the two instruction sets cannot
4877 be reliably used inside one program. The default is
4878 @samp{-mno-thumb-interwork}, since slightly larger code is generated
4879 when @samp{-mthumb-interwork} is specified.
4881 @item -mno-sched-prolog
4882 @kindex -mno-sched-prolog
4883 @kindex -msched-prolog
4884 Prevent the reordering of instructions in the function prolog, or the
4885 merging of those instruction with the instructions in the function's
4886 body. This means that all functions will start with a recognizable set
4887 of instructions (or in fact one of a choice from a small set of
4888 different function prologues), and this information can be used to
4889 locate the start if functions inside an executable piece of code. The
4890 default is @samp{-msched-prolog}.
4893 Generate output containing floating point instructions. This is the
4897 Generate output containing library calls for floating point.
4898 @strong{Warning:} the requisite libraries are not available for all ARM
4899 targets. Normally the facilities of the machine's usual C compiler are
4900 used, but this cannot be done directly in cross-compilation. You must make
4901 your own arrangements to provide suitable library functions for
4904 @samp{-msoft-float} changes the calling convention in the output file;
4905 therefore, it is only useful if you compile @emph{all} of a program with
4906 this option. In particular, you need to compile @file{libgcc.a}, the
4907 library that comes with GCC, with @samp{-msoft-float} in order for
4910 @item -mlittle-endian
4911 Generate code for a processor running in little-endian mode. This is
4912 the default for all standard configurations.
4915 Generate code for a processor running in big-endian mode; the default is
4916 to compile code for a little-endian processor.
4918 @item -mwords-little-endian
4919 This option only applies when generating code for big-endian processors.
4920 Generate code for a little-endian word order but a big-endian byte
4921 order. That is, a byte order of the form @samp{32107654}. Note: this
4922 option should only be used if you require compatibility with code for
4923 big-endian ARM processors generated by versions of the compiler prior to
4926 @item -malignment-traps
4927 @kindex -malignment-traps
4928 Generate code that will not trap if the MMU has alignment traps enabled.
4929 On ARM architectures prior to ARMv4, there were no instructions to
4930 access half-word objects stored in memory. However, when reading from
4931 memory a feature of the ARM architecture allows a word load to be used,
4932 even if the address is unaligned, and the processor core will rotate the
4933 data as it is being loaded. This option tells the compiler that such
4934 misaligned accesses will cause a MMU trap and that it should instead
4935 synthesise the access as a series of byte accesses. The compiler can
4936 still use word accesses to load half-word data if it knows that the
4937 address is aligned to a word boundary.
4939 This option is ignored when compiling for ARM architecture 4 or later,
4940 since these processors have instructions to directly access half-word
4943 @item -mno-alignment-traps
4944 @kindex -mno-alignment-traps
4945 Generate code that assumes that the MMU will not trap unaligned
4946 accesses. This produces better code when the target instruction set
4947 does not have half-word memory operations (implementations prior to
4950 Note that you cannot use this option to access unaligned word objects,
4951 since the processor will only fetch one 32-bit aligned object from
4954 The default setting for most targets is -mno-alignment-traps, since
4955 this produces better code when there are no half-word memory
4956 instructions available.
4958 @item -mshort-load-bytes
4959 @kindex -mshort-load-bytes
4960 This is a deprecated alias for @samp{-malignment-traps}.
4962 @item -mno-short-load-bytes
4963 @kindex -mno-short-load-bytes
4964 This is a deprecated alias for @samp{-mno-alignment-traps}.
4966 @item -mshort-load-words
4967 @kindex -mshort-load-words
4968 This is a deprecated alias for @samp{-mno-alignment-traps}.
4970 @item -mno-short-load-words
4971 @kindex -mno-short-load-words
4972 This is a deprecated alias for @samp{-malignment-traps}.
4976 This option only applies to RISC iX. Emulate the native BSD-mode
4977 compiler. This is the default if @samp{-ansi} is not specified.
4981 This option only applies to RISC iX. Emulate the native X/Open-mode
4984 @item -mno-symrename
4985 @kindex -mno-symrename
4986 This option only applies to RISC iX. Do not run the assembler
4987 post-processor, @samp{symrename}, after code has been assembled.
4988 Normally it is necessary to modify some of the standard symbols in
4989 preparation for linking with the RISC iX C library; this option
4990 suppresses this pass. The post-processor is never run when the
4991 compiler is built for cross-compilation.
4995 This specifies the name of the target ARM processor. GCC uses this name
4996 to determine what kind of instructions it can use when generating
4997 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
4998 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
4999 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
5000 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
5001 arm9, arm920, arm920t, arm9tdmi.
5003 @itemx -mtune=<name>
5005 This option is very similar to the @samp{-mcpu=} option, except that
5006 instead of specifying the actual target processor type, and hence
5007 restricting which instructions can be used, it specifies that GCC should
5008 tune the performance of the code as if the target were of the type
5009 specified in this option, but still choosing the instructions that it
5010 will generate based on the cpu specified by a @samp{-mcpu=} option.
5011 For some arm implementations better performance can be obtained by using
5016 This specifies the name of the target ARM architecture. GCC uses this
5017 name to determine what kind of instructions it can use when generating
5018 assembly code. This option can be used in conjunction with or instead
5019 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
5020 armv3, armv3m, armv4, armv4t, armv5.
5022 @item -mfpe=<number>
5023 @itemx -mfp=<number>
5026 This specifies the version of the floating point emulation available on
5027 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
5028 for @samp{-mfpe=} to support older versions of GCC.
5030 @item -mstructure-size-boundary=<n>
5031 @kindex -mstructure-size-boundary
5032 The size of all structures and unions will be rounded up to a multiple
5033 of the number of bits set by this option. Permissible values are 8 and
5034 32. The default value varies for different toolchains. For the COFF
5035 targeted toolchain the default value is 8. Specifying the larger number
5036 can produce faster, more efficient code, but can also increase the size
5037 of the program. The two values are potentially incompatible. Code
5038 compiled with one value cannot necessarily expect to work with code or
5039 libraries compiled with the other value, if they exchange information
5040 using structures or unions. Programmers are encouraged to use the 32
5041 value as future versions of the toolchain may default to this value.
5043 @item -mabort-on-noreturn
5044 @kindex -mabort-on-noreturn
5045 @kindex -mnoabort-on-noreturn
5046 Generate a call to the function abort at the end of a noreturn function.
5047 It will be executed if the function tries to return.
5050 @itemx -mno-long-calls
5051 Tells the compiler to perform function calls by first loading the
5052 address of the function into a register and then performing a subroutine
5053 call on this register. This switch is needed if the target function
5054 will lie outside of the 64 megabyte addressing range of the offset based
5055 version of subroutine call instruction.
5057 Even if this switch is enabled, not all function calls will be turned
5058 into long calls. The heuristic is that static functions, functions
5059 which have the @samp{short-call} attribute, functions that are inside
5060 the scope of a @samp{#pragma no_long_calls} directive and functions whose
5061 definitions have already been compiled within the current compilation
5062 unit, will not be turned into long calls. The exception to this rule is
5063 that weak function definitions, functions with the @samp{long-call}
5064 attribute or the @samp{section} attribute, and functions that are within
5065 the scope of a @samp{#pragma long_calls} directive, will always be
5066 turned into long calls.
5068 This feature is not enabled by default. Specifying
5069 @samp{--no-long-calls} will restore the default behaviour, as will
5070 placing the function calls within the scope of a @samp{#pragma
5071 long_calls_off} directive. Note these switches have no effect on how
5072 the compiler generates code to handle function calls via function
5075 @item -mnop-fun-dllimport
5076 @kindex -mnop-fun-dllimport
5077 Disable the support for the @emph{dllimport} attribute.
5079 @item -msingle-pic-base
5080 @kindex -msingle-pic-base
5081 Treat the register used for PIC addressing as read-only, rather than
5082 loading it in the prologue for each function. The run-time system is
5083 responsible for initialising this register with an appropriate value
5084 before execution begins.
5086 @item -mpic-register=<reg>
5087 @kindex -mpic-register=
5088 Specify the register to be used for PIC addressing. The default is R10
5089 unless stack-checking is enabled, when R9 is used.
5094 @subsection Thumb Options
5095 @cindex Thumb Options
5099 @item -mthumb-interwork
5100 @kindex -mthumb-interwork
5101 @kindex -mno-thumb-interwork
5102 Generate code which supports calling between the THUMB and ARM
5103 instruction sets. Without this option the two instruction sets cannot
5104 be reliably used inside one program. The default is
5105 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
5109 @kindex -mtpcs-frame
5110 @kindex -mno-tpcs-frame
5111 Generate a stack frame that is compliant with the Thumb Procedure Call
5112 Standard for all non-leaf functions. (A leaf function is one that does
5113 not call any other functions). The default is @samp{-mno-apcs-frame}.
5115 @item -mtpcs-leaf-frame
5116 @kindex -mtpcs-leaf-frame
5117 @kindex -mno-tpcs-leaf-frame
5118 Generate a stack frame that is compliant with the Thumb Procedure Call
5119 Standard for all leaf functions. (A leaf function is one that does
5120 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
5122 @item -mlittle-endian
5123 @kindex -mlittle-endian
5124 Generate code for a processor running in little-endian mode. This is
5125 the default for all standard configurations.
5128 @kindex -mbig-endian
5129 Generate code for a processor running in big-endian mode.
5131 @item -mstructure-size-boundary=<n>
5132 @kindex -mstructure-size-boundary
5133 The size of all structures and unions will be rounded up to a multiple
5134 of the number of bits set by this option. Permissible values are 8 and
5135 32. The default value varies for different toolchains. For the COFF
5136 targeted toolchain the default value is 8. Specifying the larger number
5137 can produced faster, more efficient code, but can also increase the size
5138 of the program. The two values are potentially incompatible. Code
5139 compiled with one value cannot necessarily expect to work with code or
5140 libraries compiled with the other value, if they exchange information
5141 using structures or unions. Programmers are encouraged to use the 32
5142 value as future versions of the toolchain may default to this value.
5144 @item -mnop-fun-dllimport
5145 @kindex -mnop-fun-dllimport
5146 Disable the support for the @emph{dllimport} attribute.
5148 @item -mcallee-super-interworking
5149 @kindex -mcallee-super-interworking
5150 Gives all externally visible functions in the file being compiled an ARM
5151 instruction set header which switches to Thumb mode before executing the
5152 rest of the function. This allows these functions to be called from
5153 non-interworking code.
5155 @item -mcaller-super-interworking
5156 @kindex -mcaller-super-interworking
5157 Allows calls via function pointers (including virtual functions) to
5158 execute correctly regardless of whether the target code has been
5159 compiled for interworking or not. There is a small overhead in the cost
5160 of executing a function pointer if this option is enabled.
5162 @item -msingle-pic-base
5163 @kindex -msingle-pic-base
5164 Treat the register used for PIC addressing as read-only, rather than
5165 loading it in the prologue for each function. The run-time system is
5166 responsible for initialising this register with an appropriate value
5167 before execution begins.
5169 @item -mpic-register=<reg>
5170 @kindex -mpic-register=
5171 Specify the register to be used for PIC addressing. The default is R10.
5175 @node MN10200 Options
5176 @subsection MN10200 Options
5177 @cindex MN10200 options
5178 These @samp{-m} options are defined for Matsushita MN10200 architectures:
5182 Indicate to the linker that it should perform a relaxation optimization pass
5183 to shorten branches, calls and absolute memory addresses. This option only
5184 has an effect when used on the command line for the final link step.
5186 This option makes symbolic debugging impossible.
5189 @node MN10300 Options
5190 @subsection MN10300 Options
5191 @cindex MN10300 options
5192 These @samp{-m} options are defined for Matsushita MN10300 architectures:
5196 Generate code to avoid bugs in the multiply instructions for the MN10300
5197 processors. This is the default.
5200 Do not generate code to avoid bugs in the multiply instructions for the
5204 Generate code which uses features specific to the AM33 processor.
5207 Do not generate code which uses features specific to the AM33 processor. This
5211 Indicate to the linker that it should perform a relaxation optimization pass
5212 to shorten branches, calls and absolute memory addresses. This option only
5213 has an effect when used on the command line for the final link step.
5215 This option makes symbolic debugging impossible.
5219 @node M32R/D Options
5220 @subsection M32R/D Options
5221 @cindex M32R/D options
5223 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
5226 @item -mcode-model=small
5227 Assume all objects live in the lower 16MB of memory (so that their addresses
5228 can be loaded with the @code{ld24} instruction), and assume all subroutines
5229 are reachable with the @code{bl} instruction.
5230 This is the default.
5232 The addressability of a particular object can be set with the
5233 @code{model} attribute.
5235 @item -mcode-model=medium
5236 Assume objects may be anywhere in the 32 bit address space (the compiler
5237 will generate @code{seth/add3} instructions to load their addresses), and
5238 assume all subroutines are reachable with the @code{bl} instruction.
5240 @item -mcode-model=large
5241 Assume objects may be anywhere in the 32 bit address space (the compiler
5242 will generate @code{seth/add3} instructions to load their addresses), and
5243 assume subroutines may not be reachable with the @code{bl} instruction
5244 (the compiler will generate the much slower @code{seth/add3/jl}
5245 instruction sequence).
5248 Disable use of the small data area. Variables will be put into
5249 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
5250 @code{section} attribute has been specified).
5251 This is the default.
5253 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
5254 Objects may be explicitly put in the small data area with the
5255 @code{section} attribute using one of these sections.
5258 Put small global and static data in the small data area, but do not
5259 generate special code to reference them.
5262 Put small global and static data in the small data area, and generate
5263 special instructions to reference them.
5266 @cindex smaller data references
5267 Put global and static objects less than or equal to @var{num} bytes
5268 into the small data or bss sections instead of the normal data or bss
5269 sections. The default value of @var{num} is 8.
5270 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
5271 for this option to have any effect.
5273 All modules should be compiled with the same @samp{-G @var{num}} value.
5274 Compiling with different values of @var{num} may or may not work; if it
5275 doesn't the linker will give an error message - incorrect code will not be
5281 @subsection M88K Options
5282 @cindex M88k options
5284 These @samp{-m} options are defined for Motorola 88k architectures:
5289 Generate code that works well on both the m88100 and the
5294 Generate code that works best for the m88100, but that also
5299 Generate code that works best for the m88110, and may not run
5304 Obsolete option to be removed from the next revision.
5307 @item -midentify-revision
5308 @kindex -midentify-revision
5310 @cindex identifying source, compiler (88k)
5311 Include an @code{ident} directive in the assembler output recording the
5312 source file name, compiler name and version, timestamp, and compilation
5315 @item -mno-underscores
5316 @kindex -mno-underscores
5317 @cindex underscores, avoiding (88k)
5318 In assembler output, emit symbol names without adding an underscore
5319 character at the beginning of each name. The default is to use an
5320 underscore as prefix on each name.
5322 @item -mocs-debug-info
5323 @itemx -mno-ocs-debug-info
5324 @kindex -mocs-debug-info
5325 @kindex -mno-ocs-debug-info
5327 @cindex debugging, 88k OCS
5328 Include (or omit) additional debugging information (about registers used
5329 in each stack frame) as specified in the 88open Object Compatibility
5330 Standard, ``OCS''. This extra information allows debugging of code that
5331 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
5332 Delta 88 SVr3.2 is to include this information; other 88k configurations
5333 omit this information by default.
5335 @item -mocs-frame-position
5336 @kindex -mocs-frame-position
5337 @cindex register positions in frame (88k)
5338 When emitting COFF debugging information for automatic variables and
5339 parameters stored on the stack, use the offset from the canonical frame
5340 address, which is the stack pointer (register 31) on entry to the
5341 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
5342 @samp{-mocs-frame-position}; other 88k configurations have the default
5343 @samp{-mno-ocs-frame-position}.
5345 @item -mno-ocs-frame-position
5346 @kindex -mno-ocs-frame-position
5347 @cindex register positions in frame (88k)
5348 When emitting COFF debugging information for automatic variables and
5349 parameters stored on the stack, use the offset from the frame pointer
5350 register (register 30). When this option is in effect, the frame
5351 pointer is not eliminated when debugging information is selected by the
5354 @item -moptimize-arg-area
5355 @itemx -mno-optimize-arg-area
5356 @kindex -moptimize-arg-area
5357 @kindex -mno-optimize-arg-area
5358 @cindex arguments in frame (88k)
5359 Control how function arguments are stored in stack frames.
5360 @samp{-moptimize-arg-area} saves space by optimizing them, but this
5361 conflicts with the 88open specifications. The opposite alternative,
5362 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
5363 GCC does not optimize the argument area.
5365 @item -mshort-data-@var{num}
5366 @kindex -mshort-data-@var{num}
5367 @cindex smaller data references (88k)
5368 @cindex r0-relative references (88k)
5369 Generate smaller data references by making them relative to @code{r0},
5370 which allows loading a value using a single instruction (rather than the
5371 usual two). You control which data references are affected by
5372 specifying @var{num} with this option. For example, if you specify
5373 @samp{-mshort-data-512}, then the data references affected are those
5374 involving displacements of less than 512 bytes.
5375 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
5378 @item -mserialize-volatile
5379 @kindex -mserialize-volatile
5380 @itemx -mno-serialize-volatile
5381 @kindex -mno-serialize-volatile
5382 @cindex sequential consistency on 88k
5383 Do, or don't, generate code to guarantee sequential consistency
5384 of volatile memory references. By default, consistency is
5387 The order of memory references made by the MC88110 processor does
5388 not always match the order of the instructions requesting those
5389 references. In particular, a load instruction may execute before
5390 a preceding store instruction. Such reordering violates
5391 sequential consistency of volatile memory references, when there
5392 are multiple processors. When consistency must be guaranteed,
5393 GNU C generates special instructions, as needed, to force
5394 execution in the proper order.
5396 The MC88100 processor does not reorder memory references and so
5397 always provides sequential consistency. However, by default, GNU
5398 C generates the special instructions to guarantee consistency
5399 even when you use @samp{-m88100}, so that the code may be run on an
5400 MC88110 processor. If you intend to run your code only on the
5401 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
5403 The extra code generated to guarantee consistency may affect the
5404 performance of your application. If you know that you can safely
5405 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
5411 @cindex assembler syntax, 88k
5413 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
5414 related to System V release 4 (SVr4). This controls the following:
5418 Which variant of the assembler syntax to emit.
5420 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
5421 that is used on System V release 4.
5423 @samp{-msvr4} makes GCC issue additional declaration directives used in
5427 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
5428 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
5429 other m88k configurations.
5431 @item -mversion-03.00
5432 @kindex -mversion-03.00
5433 This option is obsolete, and is ignored.
5434 @c ??? which asm syntax better for GAS? option there too?
5436 @item -mno-check-zero-division
5437 @itemx -mcheck-zero-division
5438 @kindex -mno-check-zero-division
5439 @kindex -mcheck-zero-division
5440 @cindex zero division on 88k
5441 Do, or don't, generate code to guarantee that integer division by
5442 zero will be detected. By default, detection is guaranteed.
5444 Some models of the MC88100 processor fail to trap upon integer
5445 division by zero under certain conditions. By default, when
5446 compiling code that might be run on such a processor, GNU C
5447 generates code that explicitly checks for zero-valued divisors
5448 and traps with exception number 503 when one is detected. Use of
5449 mno-check-zero-division suppresses such checking for code
5450 generated to run on an MC88100 processor.
5452 GNU C assumes that the MC88110 processor correctly detects all
5453 instances of integer division by zero. When @samp{-m88110} is
5454 specified, both @samp{-mcheck-zero-division} and
5455 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
5456 zero-valued divisors are generated.
5458 @item -muse-div-instruction
5459 @kindex -muse-div-instruction
5460 @cindex divide instruction, 88k
5461 Use the div instruction for signed integer division on the
5462 MC88100 processor. By default, the div instruction is not used.
5464 On the MC88100 processor the signed integer division instruction
5465 div) traps to the operating system on a negative operand. The
5466 operating system transparently completes the operation, but at a
5467 large cost in execution time. By default, when compiling code
5468 that might be run on an MC88100 processor, GNU C emulates signed
5469 integer division using the unsigned integer division instruction
5470 divu), thereby avoiding the large penalty of a trap to the
5471 operating system. Such emulation has its own, smaller, execution
5472 cost in both time and space. To the extent that your code's
5473 important signed integer division operations are performed on two
5474 nonnegative operands, it may be desirable to use the div
5475 instruction directly.
5477 On the MC88110 processor the div instruction (also known as the
5478 divs instruction) processes negative operands without trapping to
5479 the operating system. When @samp{-m88110} is specified,
5480 @samp{-muse-div-instruction} is ignored, and the div instruction is used
5481 for signed integer division.
5483 Note that the result of dividing INT_MIN by -1 is undefined. In
5484 particular, the behavior of such a division with and without
5485 @samp{-muse-div-instruction} may differ.
5487 @item -mtrap-large-shift
5488 @itemx -mhandle-large-shift
5489 @kindex -mtrap-large-shift
5490 @kindex -mhandle-large-shift
5491 @cindex bit shift overflow (88k)
5492 @cindex large bit shifts (88k)
5493 Include code to detect bit-shifts of more than 31 bits; respectively,
5494 trap such shifts or emit code to handle them properly. By default GCC
5495 makes no special provision for large bit shifts.
5497 @item -mwarn-passed-structs
5498 @kindex -mwarn-passed-structs
5499 @cindex structure passing (88k)
5500 Warn when a function passes a struct as an argument or result.
5501 Structure-passing conventions have changed during the evolution of the C
5502 language, and are often the source of portability problems. By default,
5503 GCC issues no such warning.
5506 @node RS/6000 and PowerPC Options
5507 @subsection IBM RS/6000 and PowerPC Options
5508 @cindex RS/6000 and PowerPC Options
5509 @cindex IBM RS/6000 and PowerPC Options
5511 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
5519 @itemx -mpowerpc-gpopt
5520 @itemx -mno-powerpc-gpopt
5521 @itemx -mpowerpc-gfxopt
5522 @itemx -mno-powerpc-gfxopt
5524 @itemx -mno-powerpc64
5528 @kindex -mpowerpc-gpopt
5529 @kindex -mpowerpc-gfxopt
5531 GCC supports two related instruction set architectures for the
5532 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
5533 instructions supported by the @samp{rios} chip set used in the original
5534 RS/6000 systems and the @dfn{PowerPC} instruction set is the
5535 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
5536 the IBM 4xx microprocessors.
5538 Neither architecture is a subset of the other. However there is a
5539 large common subset of instructions supported by both. An MQ
5540 register is included in processors supporting the POWER architecture.
5542 You use these options to specify which instructions are available on the
5543 processor you are using. The default value of these options is
5544 determined when configuring GCC. Specifying the
5545 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
5546 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
5547 rather than the options listed above.
5549 The @samp{-mpower} option allows GCC to generate instructions that
5550 are found only in the POWER architecture and to use the MQ register.
5551 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
5552 to generate instructions that are present in the POWER2 architecture but
5553 not the original POWER architecture.
5555 The @samp{-mpowerpc} option allows GCC to generate instructions that
5556 are found only in the 32-bit subset of the PowerPC architecture.
5557 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
5558 GCC to use the optional PowerPC architecture instructions in the
5559 General Purpose group, including floating-point square root. Specifying
5560 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
5561 use the optional PowerPC architecture instructions in the Graphics
5562 group, including floating-point select.
5564 The @samp{-mpowerpc64} option allows GCC to generate the additional
5565 64-bit instructions that are found in the full PowerPC64 architecture
5566 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
5567 @samp{-mno-powerpc64}.
5569 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
5570 will use only the instructions in the common subset of both
5571 architectures plus some special AIX common-mode calls, and will not use
5572 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
5573 permits GCC to use any instruction from either architecture and to
5574 allow use of the MQ register; specify this for the Motorola MPC601.
5576 @item -mnew-mnemonics
5577 @itemx -mold-mnemonics
5578 @kindex -mnew-mnemonics
5579 @kindex -mold-mnemonics
5580 Select which mnemonics to use in the generated assembler code.
5581 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
5582 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
5583 requests the assembler mnemonics defined for the POWER architecture.
5584 Instructions defined in only one architecture have only one mnemonic;
5585 GCC uses that mnemonic irrespective of which of these options is
5588 GCC defaults to the mnemonics appropriate for the architecture in
5589 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
5590 value of these option. Unless you are building a cross-compiler, you
5591 should normally not specify either @samp{-mnew-mnemonics} or
5592 @samp{-mold-mnemonics}, but should instead accept the default.
5594 @item -mcpu=@var{cpu_type}
5596 Set architecture type, register usage, choice of mnemonics, and
5597 instruction scheduling parameters for machine type @var{cpu_type}.
5598 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
5599 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
5600 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
5601 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
5602 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
5603 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
5604 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
5605 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
5606 and 64-bit PowerPC architecture machine types, with an appropriate,
5607 generic processor model assumed for scheduling purposes.@refill
5609 Specifying any of the following options:
5610 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
5611 @samp{-mcpu=power}, or @samp{-mcpu=power2}
5612 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
5613 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
5614 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
5615 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
5616 @samp{-mcpu=740}, and @samp{-mcpu=750}
5617 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5618 Exactly similarly, all of @samp{-mcpu=403},
5619 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
5620 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5621 @samp{-mcpu=common} disables both the
5622 @samp{-mpower} and @samp{-mpowerpc} options.@refill
5624 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
5625 that code will operate on all members of the RS/6000 POWER and PowerPC
5626 families. In that case, GCC will use only the instructions in the
5627 common subset of both architectures plus some special AIX common-mode
5628 calls, and will not use the MQ register. GCC assumes a generic
5629 processor model for scheduling purposes.
5631 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
5632 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
5633 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
5634 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
5635 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
5636 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
5637 the @samp{new-mnemonics} option.@refill
5639 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
5640 enables the @samp{-msoft-float} option.
5642 @item -mtune=@var{cpu_type}
5643 Set the instruction scheduling parameters for machine type
5644 @var{cpu_type}, but do not set the architecture type, register usage,
5645 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
5646 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
5647 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
5648 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
5649 instruction scheduling parameters.
5652 @itemx -mno-fp-in-toc
5653 @itemx -mno-sum-in-toc
5654 @itemx -mminimal-toc
5655 @kindex -mminimal-toc
5656 Modify generation of the TOC (Table Of Contents), which is created for
5657 every executable file. The @samp{-mfull-toc} option is selected by
5658 default. In that case, GCC will allocate at least one TOC entry for
5659 each unique non-automatic variable reference in your program. GCC
5660 will also place floating-point constants in the TOC. However, only
5661 16,384 entries are available in the TOC.
5663 If you receive a linker error message that saying you have overflowed
5664 the available TOC space, you can reduce the amount of TOC space used
5665 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
5666 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
5667 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
5668 generate code to calculate the sum of an address and a constant at
5669 run-time instead of putting that sum into the TOC. You may specify one
5670 or both of these options. Each causes GCC to produce very slightly
5671 slower and larger code at the expense of conserving TOC space.
5673 If you still run out of space in the TOC even when you specify both of
5674 these options, specify @samp{-mminimal-toc} instead. This option causes
5675 GCC to make only one TOC entry for every file. When you specify this
5676 option, GCC will produce code that is slower and larger but which
5677 uses extremely little TOC space. You may wish to use this option
5678 only on files that contain less frequently executed code. @refill
5684 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
5685 @code{long} type, and the infrastructure needed to support them.
5686 Specifying @samp{-maix64} implies @samp{-mpowerpc64} and
5687 @samp{-mpowerpc}, while @samp{-maix32} disables the 64-bit ABI and
5688 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-maix32}.
5693 On AIX, pass floating-point arguments to prototyped functions beyond the
5694 register save area (RSA) on the stack in addition to argument FPRs. The
5695 AIX calling convention was extended but not initially documented to
5696 handle an obscure K&R C case of calling a function that takes the
5697 address of its arguments with fewer arguments than declared. AIX XL
5698 compilers access floating point arguments which do not fit in the
5699 RSA from the stack when a subroutine is compiled without
5700 optimization. Because always storing floating-point arguments on the
5701 stack is inefficient and rarely needed, this option is not enabled by
5702 default and only is necessary when calling subroutines compiled by AIX
5703 XL compilers without optimization.
5707 Support @dfn{AIX Threads}. Link an application written to use
5708 @dfn{pthreads} with special libraries and startup code to enable the
5713 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
5714 application written to use message passing with special startup code to
5715 enable the application to run. The system must have PE installed in the
5716 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
5717 must be overridden with the @samp{-specs=} option to specify the
5718 appropriate directory location. The Parallel Environment does not
5719 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
5720 option are incompatible.
5724 @kindex -msoft-float
5725 Generate code that does not use (uses) the floating-point register set.
5726 Software floating point emulation is provided if you use the
5727 @samp{-msoft-float} option, and pass the option to GCC when linking.
5730 @itemx -mno-multiple
5731 Generate code that uses (does not use) the load multiple word
5732 instructions and the store multiple word instructions. These
5733 instructions are generated by default on POWER systems, and not
5734 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
5735 endian PowerPC systems, since those instructions do not work when the
5736 processor is in little endian mode. The exceptions are PPC740 and
5737 PPC750 which permit the instructions usage in little endian mode.
5742 Generate code that uses (does not use) the load string instructions
5743 and the store string word instructions to save multiple registers and
5744 do small block moves. These instructions are generated by default on
5745 POWER systems, and not generated on PowerPC systems. Do not use
5746 @samp{-mstring} on little endian PowerPC systems, since those
5747 instructions do not work when the processor is in little endian mode.
5748 The exceptions are PPC740 and PPC750 which permit the instructions
5749 usage in little endian mode.
5754 Generate code that uses (does not use) the load or store instructions
5755 that update the base register to the address of the calculated memory
5756 location. These instructions are generated by default. If you use
5757 @samp{-mno-update}, there is a small window between the time that the
5758 stack pointer is updated and the address of the previous frame is
5759 stored, which means code that walks the stack frame across interrupts or
5760 signals may get corrupted data.
5763 @itemx -mno-fused-madd
5764 @kindex -mfused-madd
5765 Generate code that uses (does not use) the floating point multiply and
5766 accumulate instructions. These instructions are generated by default if
5767 hardware floating is used.
5769 @item -mno-bit-align
5772 On System V.4 and embedded PowerPC systems do not (do) force structures
5773 and unions that contain bit fields to be aligned to the base type of the
5776 For example, by default a structure containing nothing but 8
5777 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
5778 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
5779 the structure would be aligned to a 1 byte boundary and be one byte in
5782 @item -mno-strict-align
5783 @itemx -mstrict-align
5784 @kindex -mstrict-align
5785 On System V.4 and embedded PowerPC systems do not (do) assume that
5786 unaligned memory references will be handled by the system.
5789 @itemx -mno-relocatable
5790 @kindex -mrelocatable
5791 On embedded PowerPC systems generate code that allows (does not allow)
5792 the program to be relocated to a different address at runtime. If you
5793 use @samp{-mrelocatable} on any module, all objects linked together must
5794 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
5796 @item -mrelocatable-lib
5797 @itemx -mno-relocatable-lib
5798 On embedded PowerPC systems generate code that allows (does not allow)
5799 the program to be relocated to a different address at runtime. Modules
5800 compiled with @samp{-mrelocatable-lib} can be linked with either modules
5801 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
5802 with modules compiled with the @samp{-mrelocatable} options.
5806 On System V.4 and embedded PowerPC systems do not (do) assume that
5807 register 2 contains a pointer to a global area pointing to the addresses
5808 used in the program.
5811 @itemx -mlittle-endian
5812 On System V.4 and embedded PowerPC systems compile code for the
5813 processor in little endian mode. The @samp{-mlittle-endian} option is
5814 the same as @samp{-mlittle}.
5818 On System V.4 and embedded PowerPC systems compile code for the
5819 processor in big endian mode. The @samp{-mbig-endian} option is
5820 the same as @samp{-mbig}.
5823 On System V.4 and embedded PowerPC systems compile code using calling
5824 conventions that adheres to the March 1995 draft of the System V
5825 Application Binary Interface, PowerPC processor supplement. This is the
5826 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
5828 @item -mcall-sysv-eabi
5829 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
5831 @item -mcall-sysv-noeabi
5832 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
5835 On System V.4 and embedded PowerPC systems compile code using calling
5836 conventions that are similar to those used on AIX. This is the
5837 default if you configured GCC using @samp{powerpc-*-eabiaix}.
5839 @item -mcall-solaris
5840 On System V.4 and embedded PowerPC systems compile code for the Solaris
5844 On System V.4 and embedded PowerPC systems compile code for the
5845 Linux-based GNU system.
5848 @itemx -mno-prototype
5849 On System V.4 and embedded PowerPC systems assume that all calls to
5850 variable argument functions are properly prototyped. Otherwise, the
5851 compiler must insert an instruction before every non prototyped call to
5852 set or clear bit 6 of the condition code register (@var{CR}) to
5853 indicate whether floating point values were passed in the floating point
5854 registers in case the function takes a variable arguments. With
5855 @samp{-mprototype}, only calls to prototyped variable argument functions
5856 will set or clear the bit.
5859 On embedded PowerPC systems, assume that the startup module is called
5860 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
5861 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
5865 On embedded PowerPC systems, assume that the startup module is called
5866 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
5870 On embedded PowerPC systems, assume that the startup module is called
5871 @file{crt0.o} and the standard C libraries are @file{libads.a} and
5875 On embedded PowerPC systems, assume that the startup module is called
5876 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
5880 On System V.4 and embedded PowerPC systems, specify that you are
5881 compiling for a VxWorks system.
5884 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
5885 header to indicate that @samp{eabi} extended relocations are used.
5889 On System V.4 and embedded PowerPC systems do (do not) adhere to the
5890 Embedded Applications Binary Interface (eabi) which is a set of
5891 modifications to the System V.4 specifications. Selecting @code{-meabi}
5892 means that the stack is aligned to an 8 byte boundary, a function
5893 @code{__eabi} is called to from @code{main} to set up the eabi
5894 environment, and the @samp{-msdata} option can use both @code{r2} and
5895 @code{r13} to point to two separate small data areas. Selecting
5896 @code{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
5897 do not call an initialization function from @code{main}, and the
5898 @samp{-msdata} option will only use @code{r13} to point to a single
5899 small data area. The @samp{-meabi} option is on by default if you
5900 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
5903 On System V.4 and embedded PowerPC systems, put small initialized
5904 @code{const} global and static data in the @samp{.sdata2} section, which
5905 is pointed to by register @code{r2}. Put small initialized
5906 non-@code{const} global and static data in the @samp{.sdata} section,
5907 which is pointed to by register @code{r13}. Put small uninitialized
5908 global and static data in the @samp{.sbss} section, which is adjacent to
5909 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
5910 incompatible with the @samp{-mrelocatable} option. The
5911 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
5914 On System V.4 and embedded PowerPC systems, put small global and static
5915 data in the @samp{.sdata} section, which is pointed to by register
5916 @code{r13}. Put small uninitialized global and static data in the
5917 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
5918 The @samp{-msdata=sysv} option is incompatible with the
5919 @samp{-mrelocatable} option.
5921 @item -msdata=default
5923 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
5924 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
5925 same as @samp{-msdata=sysv}.
5928 On System V.4 and embedded PowerPC systems, put small global and static
5929 data in the @samp{.sdata} section. Put small uninitialized global and
5930 static data in the @samp{.sbss} section. Do not use register @code{r13}
5931 to address small data however. This is the default behavior unless
5932 other @samp{-msdata} options are used.
5936 On embedded PowerPC systems, put all initialized global and static data
5937 in the @samp{.data} section, and all uninitialized data in the
5938 @samp{.bss} section.
5941 @cindex smaller data references (PowerPC)
5942 @cindex .sdata/.sdata2 references (PowerPC)
5943 On embedded PowerPC systems, put global and static items less than or
5944 equal to @var{num} bytes into the small data or bss sections instead of
5945 the normal data or bss section. By default, @var{num} is 8. The
5946 @samp{-G @var{num}} switch is also passed to the linker.
5947 All modules should be compiled with the same @samp{-G @var{num}} value.
5950 @itemx -mno-regnames
5951 On System V.4 and embedded PowerPC systems do (do not) emit register
5952 names in the assembly language output using symbolic forms.
5957 @subsection IBM RT Options
5959 @cindex IBM RT options
5961 These @samp{-m} options are defined for the IBM RT PC:
5965 Use an in-line code sequence for integer multiplies. This is the
5968 @item -mcall-lib-mul
5969 Call @code{lmul$$} for integer multiples.
5971 @item -mfull-fp-blocks
5972 Generate full-size floating point data blocks, including the minimum
5973 amount of scratch space recommended by IBM. This is the default.
5975 @item -mminimum-fp-blocks
5976 Do not include extra scratch space in floating point data blocks. This
5977 results in smaller code, but slower execution, since scratch space must
5978 be allocated dynamically.
5980 @cindex @file{varargs.h} and RT PC
5981 @cindex @file{stdarg.h} and RT PC
5982 @item -mfp-arg-in-fpregs
5983 Use a calling sequence incompatible with the IBM calling convention in
5984 which floating point arguments are passed in floating point registers.
5985 Note that @code{varargs.h} and @code{stdargs.h} will not work with
5986 floating point operands if this option is specified.
5988 @item -mfp-arg-in-gregs
5989 Use the normal calling convention for floating point arguments. This is
5992 @item -mhc-struct-return
5993 Return structures of more than one word in memory, rather than in a
5994 register. This provides compatibility with the MetaWare HighC (hc)
5995 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
5996 with the Portable C Compiler (pcc).
5998 @item -mnohc-struct-return
5999 Return some structures of more than one word in registers, when
6000 convenient. This is the default. For compatibility with the
6001 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
6002 option @samp{-mhc-struct-return}.
6006 @subsection MIPS Options
6007 @cindex MIPS options
6009 These @samp{-m} options are defined for the MIPS family of computers:
6012 @item -mcpu=@var{cpu type}
6013 Assume the defaults for the machine type @var{cpu type} when scheduling
6014 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
6015 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
6016 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
6017 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
6018 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
6019 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
6020 @var{cpu type} will schedule things appropriately for that particular
6021 chip, the compiler will not generate any code that does not meet level 1
6022 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
6023 or @samp{-mabi} switch being used.
6026 Issue instructions from level 1 of the MIPS ISA. This is the default.
6027 @samp{r3000} is the default @var{cpu type} at this ISA level.
6030 Issue instructions from level 2 of the MIPS ISA (branch likely, square
6031 root instructions). @samp{r6000} is the default @var{cpu type} at this
6035 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
6036 @samp{r4000} is the default @var{cpu type} at this ISA level.
6039 Issue instructions from level 4 of the MIPS ISA (conditional move,
6040 prefetch, enhanced FPU instructions). @samp{r8000} is the default
6041 @var{cpu type} at this ISA level.
6044 Assume that 32 32-bit floating point registers are available. This is
6048 Assume that 32 64-bit floating point registers are available. This is
6049 the default when the @samp{-mips3} option is used.
6052 Assume that 32 32-bit general purpose registers are available. This is
6056 Assume that 32 64-bit general purpose registers are available. This is
6057 the default when the @samp{-mips3} option is used.
6060 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
6061 explanation of the default, and the width of pointers.
6064 Force long types to be 64 bits wide. See @samp{-mlong32} for an
6065 explanation of the default, and the width of pointers.
6068 Force long, int, and pointer types to be 32 bits wide.
6070 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
6071 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
6072 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
6073 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
6074 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
6075 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
6076 are 32 bits, and longs are 64 bits wide. The width of pointer types is
6077 the smaller of the width of longs or the width of general purpose
6078 registers (which in turn depends on the ISA).
6085 Generate code for the indicated ABI. The default instruction level is
6086 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
6087 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
6088 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
6092 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
6093 add normal debug information. This is the default for all
6094 platforms except for the OSF/1 reference platform, using the OSF/rose
6095 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
6096 switches are used, the @file{mips-tfile} program will encapsulate the
6097 stabs within MIPS ECOFF.
6100 Generate code for the GNU assembler. This is the default on the OSF/1
6101 reference platform, using the OSF/rose object format. Also, this is
6102 the default if the configure option @samp{--with-gnu-as} is used.
6104 @item -msplit-addresses
6105 @itemx -mno-split-addresses
6106 Generate code to load the high and low parts of address constants separately.
6107 This allows @code{gcc} to optimize away redundant loads of the high order
6108 bits of addresses. This optimization requires GNU as and GNU ld.
6109 This optimization is enabled by default for some embedded targets where
6110 GNU as and GNU ld are standard.
6114 The @samp{-mrnames} switch says to output code using the MIPS software
6115 names for the registers, instead of the hardware names (ie, @var{a0}
6116 instead of @var{$4}). The only known assembler that supports this option
6117 is the Algorithmics assembler.
6121 The @samp{-mgpopt} switch says to write all of the data declarations
6122 before the instructions in the text section, this allows the MIPS
6123 assembler to generate one word memory references instead of using two
6124 words for short global or static data items. This is on by default if
6125 optimization is selected.
6129 For each non-inline function processed, the @samp{-mstats} switch
6130 causes the compiler to emit one line to the standard error file to
6131 print statistics about the program (number of registers saved, stack
6136 The @samp{-mmemcpy} switch makes all block moves call the appropriate
6137 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
6138 generating inline code.
6141 @itemx -mno-mips-tfile
6142 The @samp{-mno-mips-tfile} switch causes the compiler not
6143 postprocess the object file with the @file{mips-tfile} program,
6144 after the MIPS assembler has generated it to add debug support. If
6145 @file{mips-tfile} is not run, then no local variables will be
6146 available to the debugger. In addition, @file{stage2} and
6147 @file{stage3} objects will have the temporary file names passed to the
6148 assembler embedded in the object file, which means the objects will
6149 not compare the same. The @samp{-mno-mips-tfile} switch should only
6150 be used when there are bugs in the @file{mips-tfile} program that
6151 prevents compilation.
6154 Generate output containing library calls for floating point.
6155 @strong{Warning:} the requisite libraries are not part of GCC.
6156 Normally the facilities of the machine's usual C compiler are used, but
6157 this can't be done directly in cross-compilation. You must make your
6158 own arrangements to provide suitable library functions for
6162 Generate output containing floating point instructions. This is the
6163 default if you use the unmodified sources.
6166 @itemx -mno-abicalls
6167 Emit (or do not emit) the pseudo operations @samp{.abicalls},
6168 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
6169 position independent code.
6172 @itemx -mno-long-calls
6173 Do all calls with the @samp{JALR} instruction, which requires
6174 loading up a function's address into a register before the call.
6175 You need to use this switch, if you call outside of the current
6176 512 megabyte segment to functions that are not through pointers.
6179 @itemx -mno-half-pic
6180 Put pointers to extern references into the data section and load them
6181 up, rather than put the references in the text section.
6183 @item -membedded-pic
6184 @itemx -mno-embedded-pic
6185 Generate PIC code suitable for some embedded systems. All calls are
6186 made using PC relative address, and all data is addressed using the $gp
6187 register. No more than 65536 bytes of global data may be used. This
6188 requires GNU as and GNU ld which do most of the work. This currently
6189 only works on targets which use ECOFF; it does not work with ELF.
6191 @item -membedded-data
6192 @itemx -mno-embedded-data
6193 Allocate variables to the read-only data section first if possible, then
6194 next in the small data section if possible, otherwise in data. This gives
6195 slightly slower code than the default, but reduces the amount of RAM required
6196 when executing, and thus may be preferred for some embedded systems.
6198 @item -muninit-const-in-rodata
6199 @itemx -mno-uninit-const-in-rodata
6200 When used together with -membedded-data, it will always store uninitialized
6201 const variables in the read-only data section.
6203 @item -msingle-float
6204 @itemx -mdouble-float
6205 The @samp{-msingle-float} switch tells gcc to assume that the floating
6206 point coprocessor only supports single precision operations, as on the
6207 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
6208 double precision operations. This is the default.
6212 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
6213 as on the @samp{r4650} chip.
6216 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
6221 Enable 16-bit instructions.
6224 Use the entry and exit pseudo ops. This option can only be used with
6228 Compile code for the processor in little endian mode.
6229 The requisite libraries are assumed to exist.
6232 Compile code for the processor in big endian mode.
6233 The requisite libraries are assumed to exist.
6236 @cindex smaller data references (MIPS)
6237 @cindex gp-relative references (MIPS)
6238 Put global and static items less than or equal to @var{num} bytes into
6239 the small data or bss sections instead of the normal data or bss
6240 section. This allows the assembler to emit one word memory reference
6241 instructions based on the global pointer (@var{gp} or @var{$28}),
6242 instead of the normal two words used. By default, @var{num} is 8 when
6243 the MIPS assembler is used, and 0 when the GNU assembler is used. The
6244 @samp{-G @var{num}} switch is also passed to the assembler and linker.
6245 All modules should be compiled with the same @samp{-G @var{num}}
6249 Tell the MIPS assembler to not run its preprocessor over user
6250 assembler files (with a @samp{.s} suffix) when assembling them.
6253 Pass an option to gas which will cause nops to be inserted if
6254 the read of the destination register of an mfhi or mflo instruction
6255 occurs in the following two instructions.
6258 Do not include the default crt0.
6262 These options are defined by the macro
6263 @code{TARGET_SWITCHES} in the machine description. The default for the
6264 options is also defined by that macro, which enables you to change the
6269 @subsection Intel 386 Options
6270 @cindex i386 Options
6271 @cindex Intel 386 Options
6273 These @samp{-m} options are defined for the i386 family of computers:
6276 @item -mcpu=@var{cpu type}
6277 Assume the defaults for the machine type @var{cpu type} when scheduling
6278 instructions. The choices for @var{cpu type} are @samp{i386},
6279 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
6280 @samp{pentiumpro}, @samp{k6}, and @samp{athlon}
6282 While picking a specific @var{cpu type} will schedule things appropriately
6283 for that particular chip, the compiler will not generate any code that
6284 does not run on the i386 without the @samp{-march=@var{cpu type}} option
6285 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
6286 is equivalent to @samp{pentiumpro}. @samp{k6} is the AMD chip as
6287 opposed to the Intel ones.
6289 @item -march=@var{cpu type}
6290 Generate instructions for the machine type @var{cpu type}. The choices
6291 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
6292 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
6298 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
6299 respectively. These synonyms are deprecated.
6301 @item -mintel-syntax
6302 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
6306 Control whether or not the compiler uses IEEE floating point
6307 comparisons. These handle correctly the case where the result of a
6308 comparison is unordered.
6311 Generate output containing library calls for floating point.
6312 @strong{Warning:} the requisite libraries are not part of GCC.
6313 Normally the facilities of the machine's usual C compiler are used, but
6314 this can't be done directly in cross-compilation. You must make your
6315 own arrangements to provide suitable library functions for
6318 On machines where a function returns floating point results in the 80387
6319 register stack, some floating point opcodes may be emitted even if
6320 @samp{-msoft-float} is used.
6322 @item -mno-fp-ret-in-387
6323 Do not use the FPU registers for return values of functions.
6325 The usual calling convention has functions return values of types
6326 @code{float} and @code{double} in an FPU register, even if there
6327 is no FPU. The idea is that the operating system should emulate
6330 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
6331 in ordinary CPU registers instead.
6333 @item -mno-fancy-math-387
6334 Some 387 emulators do not support the @code{sin}, @code{cos} and
6335 @code{sqrt} instructions for the 387. Specify this option to avoid
6336 generating those instructions. This option is the default on FreeBSD.
6337 As of revision 2.6.1, these instructions are not generated unless you
6338 also use the @samp{-ffast-math} switch.
6340 @item -malign-double
6341 @itemx -mno-align-double
6342 Control whether GCC aligns @code{double}, @code{long double}, and
6343 @code{long long} variables on a two word boundary or a one word
6344 boundary. Aligning @code{double} variables on a two word boundary will
6345 produce code that runs somewhat faster on a @samp{Pentium} at the
6346 expense of more memory.
6348 @item -m128bit-long-double
6349 @itemx -m128bit-long-double
6350 Control the size of @code{long double} type. i386 application binary interface
6351 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
6352 preffer @code{long double} aligned to 8 or 16 byte boundary. This is
6353 impossible to reach with 12 byte long doubles in the array accesses.
6355 @strong{Warning:} if you use the @samp{-m128bit-long-double} switch, the
6356 structures and arrays containing @code{long double} will change their size as
6357 well as function calling convention for function taking @code{long double}
6360 @item -m96bit-long-double
6361 @itemx -m96bit-long-double
6362 Set the size of @code{long double} to 96 bytes as requires by the i386
6363 application binary interface. This is the default.
6366 @itemx -mno-svr3-shlib
6367 Control whether GCC places uninitialized locals into @code{bss} or
6368 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
6369 These options are meaningful only on System V Release 3.
6371 @item -mno-wide-multiply
6372 @itemx -mwide-multiply
6373 Control whether GCC uses the @code{mul} and @code{imul} that produce
6374 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
6375 long} multiplies and 32-bit division by constants.
6378 Use a different function-calling convention, in which functions that
6379 take a fixed number of arguments return with the @code{ret} @var{num}
6380 instruction, which pops their arguments while returning. This saves one
6381 instruction in the caller since there is no need to pop the arguments
6384 You can specify that an individual function is called with this calling
6385 sequence with the function attribute @samp{stdcall}. You can also
6386 override the @samp{-mrtd} option by using the function attribute
6387 @samp{cdecl}. @xref{Function Attributes}.
6389 @strong{Warning:} this calling convention is incompatible with the one
6390 normally used on Unix, so you cannot use it if you need to call
6391 libraries compiled with the Unix compiler.
6393 Also, you must provide function prototypes for all functions that
6394 take variable numbers of arguments (including @code{printf});
6395 otherwise incorrect code will be generated for calls to those
6398 In addition, seriously incorrect code will result if you call a
6399 function with too many arguments. (Normally, extra arguments are
6400 harmlessly ignored.)
6402 @item -mreg-alloc=@var{regs}
6403 Control the default allocation order of integer registers. The
6404 string @var{regs} is a series of letters specifying a register. The
6405 supported letters are: @code{a} allocate EAX; @code{b} allocate EBX;
6406 @code{c} allocate ECX; @code{d} allocate EDX; @code{S} allocate ESI;
6407 @code{D} allocate EDI; @code{B} allocate EBP.
6409 @item -mregparm=@var{num}
6410 Control how many registers are used to pass integer arguments. By
6411 default, no registers are used to pass arguments, and at most 3
6412 registers can be used. You can control this behavior for a specific
6413 function by using the function attribute @samp{regparm}.
6414 @xref{Function Attributes}.
6416 @strong{Warning:} if you use this switch, and
6417 @var{num} is nonzero, then you must build all modules with the same
6418 value, including any libraries. This includes the system libraries and
6421 @item -malign-loops=@var{num}
6422 Align loops to a 2 raised to a @var{num} byte boundary. If
6423 @samp{-malign-loops} is not specified, the default is 2 unless
6424 gas 2.8 (or later) is being used in which case the default is
6425 to align the loop on a 16 byte boundary if it is less than 8
6428 @item -malign-jumps=@var{num}
6429 Align instructions that are only jumped to to a 2 raised to a @var{num}
6430 byte boundary. If @samp{-malign-jumps} is not specified, the default is
6431 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
6432 gas 2.8 (or later) is being used in which case the default is
6433 to align the instruction on a 16 byte boundary if it is less
6436 @item -malign-functions=@var{num}
6437 Align the start of functions to a 2 raised to @var{num} byte boundary.
6438 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
6439 for a 386, and 4 if optimizing for a 486.
6441 @item -mpreferred-stack-boundary=@var{num}
6442 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
6443 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
6444 the default is 4 (16 bytes or 128 bits).
6446 The stack is required to be aligned on a 4 byte boundary. On Pentium
6447 and PentiumPro, @code{double} and @code{long double} values should be
6448 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
6449 significant run time performance penalties. On Pentium III, the
6450 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
6451 penalties if it is not 16 byte aligned.
6453 To ensure proper alignment of this values on the stack, the stack boundary
6454 must be as aligned as that required by any value stored on the stack.
6455 Further, every function must be generated such that it keeps the stack
6456 aligned. Thus calling a function compiled with a higher preferred
6457 stack boundary from a function compiled with a lower preferred stack
6458 boundary will most likely misalign the stack. It is recommended that
6459 libraries that use callbacks always use the default setting.
6461 This extra alignment does consume extra stack space. Code that is sensitive
6462 to stack space usage, such as embedded systems and operating system kernels,
6463 may want to reduce the preferred alignment to
6464 @samp{-mpreferred-stack-boundary=2}.
6468 Use PUSH operations to store outgoing parameters. This method is shorter
6469 and usually equally fast as method using SUB/MOV operations and is enabled
6470 by default. In some cases disabling it may improve performance because of
6471 improved scheduling and reduced dependencies.
6473 @item -maccumulate-outgoing-args
6474 @kindex -maccumulate-outgoing-args
6475 If enabled, the maximum amount of space required for outgoing arguments will be
6476 computed in the function prologue. This in faster on most modern CPUs
6477 because of reduced dependencies, improved scheduling and reduced stack usage
6478 when preferred stack boundary is not equal to 2. The drawback is a notable
6479 increase in code size. This switch implies -mno-push-args.
6483 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
6484 on thread-safe exception handling must compile and link all code with the
6485 @samp{-mthreads} option. When compiling, @samp{-mthreads} defines
6486 @samp{-D_MT}; when linking, it links in a special thread helper library
6487 @samp{-lmingwthrd} which cleans up per thread exception handling data.
6489 @item -mno-align-stringops
6490 @kindex -mno-align-stringops
6491 Do not align destination of inlined string operations. This switch reduces
6492 code size and improves performance in case the destination is already aligned,
6493 but gcc don't know about it.
6495 @item -minline-all-stringops
6496 @kindex -minline-all-stringops
6497 By default GCC inlines string operations only when destination is known to be
6498 aligned at least to 4 byte boundary. This enables more inlining, increase code
6499 size, but may improve performance of code that depends on fast memcpy, strlen
6500 and memset for short lengths.
6504 @subsection HPPA Options
6505 @cindex HPPA Options
6507 These @samp{-m} options are defined for the HPPA family of computers:
6510 @item -march=@var{architecture type}
6511 Generate code for the specified architecture. The choices for
6512 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
6513 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
6514 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
6515 architecture option for your machine. Code compiled for lower numbered
6516 architectures will run on higher numbered architectures, but not the
6519 PA 2.0 support currently requires gas snapshot 19990413 or later. The
6520 next release of binutils (current is 2.9.1) will probably contain PA 2.0
6524 @itemx -mpa-risc-1-1
6525 @itemx -mpa-risc-2-0
6526 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
6529 Generate code suitable for big switch tables. Use this option only if
6530 the assembler/linker complain about out of range branches within a switch
6533 @item -mjump-in-delay
6534 Fill delay slots of function calls with unconditional jump instructions
6535 by modifying the return pointer for the function call to be the target
6536 of the conditional jump.
6538 @item -mdisable-fpregs
6539 Prevent floating point registers from being used in any manner. This is
6540 necessary for compiling kernels which perform lazy context switching of
6541 floating point registers. If you use this option and attempt to perform
6542 floating point operations, the compiler will abort.
6544 @item -mdisable-indexing
6545 Prevent the compiler from using indexing address modes. This avoids some
6546 rather obscure problems when compiling MIG generated code under MACH.
6548 @item -mno-space-regs
6549 Generate code that assumes the target has no space registers. This allows
6550 GCC to generate faster indirect calls and use unscaled index address modes.
6552 Such code is suitable for level 0 PA systems and kernels.
6554 @item -mfast-indirect-calls
6555 Generate code that assumes calls never cross space boundaries. This
6556 allows GCC to emit code which performs faster indirect calls.
6558 This option will not work in the presence of shared libraries or nested
6561 @item -mlong-load-store
6562 Generate 3-instruction load and store sequences as sometimes required by
6563 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
6566 @item -mportable-runtime
6567 Use the portable calling conventions proposed by HP for ELF systems.
6570 Enable the use of assembler directives only GAS understands.
6572 @item -mschedule=@var{cpu type}
6573 Schedule code according to the constraints for the machine type
6574 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
6575 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
6576 @file{/usr/lib/sched.models} on an HP-UX system to determine the
6577 proper scheduling option for your machine.
6580 Enable the optimization pass in the HPUX linker. Note this makes symbolic
6581 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
6582 in which they give bogus error messages when linking some programs.
6585 Generate output containing library calls for floating point.
6586 @strong{Warning:} the requisite libraries are not available for all HPPA
6587 targets. Normally the facilities of the machine's usual C compiler are
6588 used, but this cannot be done directly in cross-compilation. You must make
6589 your own arrangements to provide suitable library functions for
6590 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
6591 does provide software floating point support.
6593 @samp{-msoft-float} changes the calling convention in the output file;
6594 therefore, it is only useful if you compile @emph{all} of a program with
6595 this option. In particular, you need to compile @file{libgcc.a}, the
6596 library that comes with GCC, with @samp{-msoft-float} in order for
6600 @node Intel 960 Options
6601 @subsection Intel 960 Options
6603 These @samp{-m} options are defined for the Intel 960 implementations:
6606 @item -m@var{cpu type}
6607 Assume the defaults for the machine type @var{cpu type} for some of
6608 the other options, including instruction scheduling, floating point
6609 support, and addressing modes. The choices for @var{cpu type} are
6610 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
6611 @samp{sa}, and @samp{sb}.
6617 The @samp{-mnumerics} option indicates that the processor does support
6618 floating-point instructions. The @samp{-msoft-float} option indicates
6619 that floating-point support should not be assumed.
6621 @item -mleaf-procedures
6622 @itemx -mno-leaf-procedures
6623 Do (or do not) attempt to alter leaf procedures to be callable with the
6624 @code{bal} instruction as well as @code{call}. This will result in more
6625 efficient code for explicit calls when the @code{bal} instruction can be
6626 substituted by the assembler or linker, but less efficient code in other
6627 cases, such as calls via function pointers, or using a linker that doesn't
6628 support this optimization.
6631 @itemx -mno-tail-call
6632 Do (or do not) make additional attempts (beyond those of the
6633 machine-independent portions of the compiler) to optimize tail-recursive
6634 calls into branches. You may not want to do this because the detection of
6635 cases where this is not valid is not totally complete. The default is
6636 @samp{-mno-tail-call}.
6638 @item -mcomplex-addr
6639 @itemx -mno-complex-addr
6640 Assume (or do not assume) that the use of a complex addressing mode is a
6641 win on this implementation of the i960. Complex addressing modes may not
6642 be worthwhile on the K-series, but they definitely are on the C-series.
6643 The default is currently @samp{-mcomplex-addr} for all processors except
6647 @itemx -mno-code-align
6648 Align code to 8-byte boundaries for faster fetching (or don't bother).
6649 Currently turned on by default for C-series implementations only.
6652 @item -mclean-linkage
6653 @itemx -mno-clean-linkage
6654 These options are not fully implemented.
6658 @itemx -mic2.0-compat
6659 @itemx -mic3.0-compat
6660 Enable compatibility with iC960 v2.0 or v3.0.
6664 Enable compatibility with the iC960 assembler.
6666 @item -mstrict-align
6667 @itemx -mno-strict-align
6668 Do not permit (do permit) unaligned accesses.
6671 Enable structure-alignment compatibility with Intel's gcc release version
6672 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
6674 @item -mlong-double-64
6675 Implement type @samp{long double} as 64-bit floating point numbers.
6676 Without the option @samp{long double} is implemented by 80-bit
6677 floating point numbers. The only reason we have it because there is
6678 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
6679 is only useful for people using soft-float targets. Otherwise, we
6680 should recommend against use of it.
6684 @node DEC Alpha Options
6685 @subsection DEC Alpha Options
6687 These @samp{-m} options are defined for the DEC Alpha implementations:
6690 @item -mno-soft-float
6692 Use (do not use) the hardware floating-point instructions for
6693 floating-point operations. When @code{-msoft-float} is specified,
6694 functions in @file{libgcc1.c} will be used to perform floating-point
6695 operations. Unless they are replaced by routines that emulate the
6696 floating-point operations, or compiled in such a way as to call such
6697 emulations routines, these routines will issue floating-point
6698 operations. If you are compiling for an Alpha without floating-point
6699 operations, you must ensure that the library is built so as not to call
6702 Note that Alpha implementations without floating-point operations are
6703 required to have floating-point registers.
6707 Generate code that uses (does not use) the floating-point register set.
6708 @code{-mno-fp-regs} implies @code{-msoft-float}. If the floating-point
6709 register set is not used, floating point operands are passed in integer
6710 registers as if they were integers and floating-point results are passed
6711 in $0 instead of $f0. This is a non-standard calling sequence, so any
6712 function with a floating-point argument or return value called by code
6713 compiled with @code{-mno-fp-regs} must also be compiled with that
6716 A typical use of this option is building a kernel that does not use,
6717 and hence need not save and restore, any floating-point registers.
6720 The Alpha architecture implements floating-point hardware optimized for
6721 maximum performance. It is mostly compliant with the IEEE floating
6722 point standard. However, for full compliance, software assistance is
6723 required. This option generates code fully IEEE compliant code
6724 @emph{except} that the @var{inexact flag} is not maintained (see below).
6725 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
6726 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
6727 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
6728 code is less efficient but is able to correctly support denormalized
6729 numbers and exceptional IEEE values such as not-a-number and plus/minus
6730 infinity. Other Alpha compilers call this option
6731 @code{-ieee_with_no_inexact}.
6733 @item -mieee-with-inexact
6734 @c overfull hbox here --bob 22 jul96
6735 @c original text between ignore ... end ignore
6737 This is like @samp{-mieee} except the generated code also maintains the
6738 IEEE @var{inexact flag}. Turning on this option causes the generated
6739 code to implement fully-compliant IEEE math. The option is a shorthand
6740 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
6741 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
6742 implementations the resulting code may execute significantly slower than
6743 the code generated by default. Since there is very little code that
6744 depends on the @var{inexact flag}, you should normally not specify this
6745 option. Other Alpha compilers call this option
6746 @samp{-ieee_with_inexact}.
6748 @c changed paragraph
6749 This is like @samp{-mieee} except the generated code also maintains the
6750 IEEE @var{inexact flag}. Turning on this option causes the generated
6751 code to implement fully-compliant IEEE math. The option is a shorthand
6752 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
6753 @samp{-mieee-conformant},
6754 @samp{-mfp-trap-mode=sui},
6755 and @samp{-mtrap-precision=i}.
6756 On some Alpha implementations the resulting code may execute
6757 significantly slower than the code generated by default. Since there
6758 is very little code that depends on the @var{inexact flag}, you should
6759 normally not specify this option. Other Alpha compilers call this
6760 option @samp{-ieee_with_inexact}.
6761 @c end changes to prevent overfull hboxes
6763 @item -mfp-trap-mode=@var{trap mode}
6764 This option controls what floating-point related traps are enabled.
6765 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
6766 The trap mode can be set to one of four values:
6770 This is the default (normal) setting. The only traps that are enabled
6771 are the ones that cannot be disabled in software (e.g., division by zero
6775 In addition to the traps enabled by @samp{n}, underflow traps are enabled
6779 Like @samp{su}, but the instructions are marked to be safe for software
6780 completion (see Alpha architecture manual for details).
6783 Like @samp{su}, but inexact traps are enabled as well.
6786 @item -mfp-rounding-mode=@var{rounding mode}
6787 Selects the IEEE rounding mode. Other Alpha compilers call this option
6788 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
6793 Normal IEEE rounding mode. Floating point numbers are rounded towards
6794 the nearest machine number or towards the even machine number in case
6798 Round towards minus infinity.
6801 Chopped rounding mode. Floating point numbers are rounded towards zero.
6804 Dynamic rounding mode. A field in the floating point control register
6805 (@var{fpcr}, see Alpha architecture reference manual) controls the
6806 rounding mode in effect. The C library initializes this register for
6807 rounding towards plus infinity. Thus, unless your program modifies the
6808 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
6811 @item -mtrap-precision=@var{trap precision}
6812 In the Alpha architecture, floating point traps are imprecise. This
6813 means without software assistance it is impossible to recover from a
6814 floating trap and program execution normally needs to be terminated.
6815 GCC can generate code that can assist operating system trap handlers
6816 in determining the exact location that caused a floating point trap.
6817 Depending on the requirements of an application, different levels of
6818 precisions can be selected:
6822 Program precision. This option is the default and means a trap handler
6823 can only identify which program caused a floating point exception.
6826 Function precision. The trap handler can determine the function that
6827 caused a floating point exception.
6830 Instruction precision. The trap handler can determine the exact
6831 instruction that caused a floating point exception.
6834 Other Alpha compilers provide the equivalent options called
6835 @samp{-scope_safe} and @samp{-resumption_safe}.
6837 @item -mieee-conformant
6838 This option marks the generated code as IEEE conformant. You must not
6839 use this option unless you also specify @samp{-mtrap-precision=i} and either
6840 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
6841 is to emit the line @samp{.eflag 48} in the function prologue of the
6842 generated assembly file. Under DEC Unix, this has the effect that
6843 IEEE-conformant math library routines will be linked in.
6845 @item -mbuild-constants
6846 Normally GCC examines a 32- or 64-bit integer constant to
6847 see if it can construct it from smaller constants in two or three
6848 instructions. If it cannot, it will output the constant as a literal and
6849 generate code to load it from the data segment at runtime.
6851 Use this option to require GCC to construct @emph{all} integer constants
6852 using code, even if it takes more instructions (the maximum is six).
6854 You would typically use this option to build a shared library dynamic
6855 loader. Itself a shared library, it must relocate itself in memory
6856 before it can find the variables and constants in its own data segment.
6860 Select whether to generate code to be assembled by the vendor-supplied
6861 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
6869 Indicate whether GCC should generate code to use the optional BWX,
6870 CIX, and MAX instruction sets. The default is to use the instruction sets
6871 supported by the CPU type specified via @samp{-mcpu=} option or that
6872 of the CPU on which GCC was built if none was specified.
6874 @item -mcpu=@var{cpu_type}
6875 Set the instruction set, register set, and instruction scheduling
6876 parameters for machine type @var{cpu_type}. You can specify either the
6877 @samp{EV} style name or the corresponding chip number. GCC
6878 supports scheduling parameters for the EV4 and EV5 family of processors
6879 and will choose the default values for the instruction set from
6880 the processor you specify. If you do not specify a processor type,
6881 GCC will default to the processor on which the compiler was built.
6883 Supported values for @var{cpu_type} are
6888 Schedules as an EV4 and has no instruction set extensions.
6892 Schedules as an EV5 and has no instruction set extensions.
6896 Schedules as an EV5 and supports the BWX extension.
6901 Schedules as an EV5 and supports the BWX and MAX extensions.
6905 Schedules as an EV5 (until Digital releases the scheduling parameters
6906 for the EV6) and supports the BWX, CIX, and MAX extensions.
6909 @item -mmemory-latency=@var{time}
6910 Sets the latency the scheduler should assume for typical memory
6911 references as seen by the application. This number is highly
6912 dependent on the memory access patterns used by the application
6913 and the size of the external cache on the machine.
6915 Valid options for @var{time} are
6919 A decimal number representing clock cycles.
6925 The compiler contains estimates of the number of clock cycles for
6926 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
6927 (also called Dcache, Scache, and Bcache), as well as to main memory.
6928 Note that L3 is only valid for EV5.
6933 @node Clipper Options
6934 @subsection Clipper Options
6936 These @samp{-m} options are defined for the Clipper implementations:
6940 Produce code for a C300 Clipper processor. This is the default.
6943 Produce code for a C400 Clipper processor i.e. use floating point
6947 @node H8/300 Options
6948 @subsection H8/300 Options
6950 These @samp{-m} options are defined for the H8/300 implementations:
6954 Shorten some address references at link time, when possible; uses the
6955 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
6956 ld.info, Using ld}, for a fuller description.
6959 Generate code for the H8/300H.
6962 Generate code for the H8/S.
6965 Generate code for the H8/S2600. This switch must be used with -ms.
6968 Make @code{int} data 32 bits by default.
6971 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
6972 The default for the H8/300H and H8/S is to align longs and floats on 4
6974 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
6975 This option has no effect on the H8/300.
6979 @subsection SH Options
6981 These @samp{-m} options are defined for the SH implementations:
6985 Generate code for the SH1.
6988 Generate code for the SH2.
6991 Generate code for the SH3.
6994 Generate code for the SH3e.
6997 Generate code for the SH4 without a floating-point unit.
6999 @item -m4-single-only
7000 Generate code for the SH4 with a floating-point unit that only
7001 supports single-precision arithmentic.
7004 Generate code for the SH4 assuming the floating-point unit is in
7005 single-precision mode by default.
7008 Generate code for the SH4.
7011 Compile code for the processor in big endian mode.
7014 Compile code for the processor in little endian mode.
7017 Align doubles at 64 bit boundaries. Note that this changes the calling
7018 conventions, and thus some functions from the standard C library will
7019 not work unless you recompile it first with -mdalign.
7022 Shorten some address references at link time, when possible; uses the
7023 linker option @samp{-relax}.
7026 Use 32-bit offsets in @code{switch} tables. The default is to use
7030 Enable the use of the instruction @code{fmovd}.
7033 Comply with the calling conventions defined by Hitachi.
7036 Mark the @code{MAC} register as call-clobbered, even if
7037 @code{-mhitachi} is given.
7040 Dump instruction size and location in the assembly code.
7043 This option is deprecated. It pads structures to multiple of 4 bytes,
7044 which is incompatible with the SH ABI.
7047 Optimize for space instead of speed. Implied by @code{-Os}.
7050 When generating position-independent code, emit function calls using
7051 the Global Offset Table instead of the Procedure Linkage Table.
7054 Generate a library function call to invalidate instruction cache
7055 entries, after fixing up a trampoline. This library function call
7056 doesn't assume it can write to the whole memory address space. This
7057 is the default when the target is @code{sh-*-linux*}.
7060 @node System V Options
7061 @subsection Options for System V
7063 These additional options are available on System V Release 4 for
7064 compatibility with other compilers on those systems:
7068 Create a shared object.
7069 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
7072 Identify the versions of each tool used by the compiler, in a
7073 @code{.ident} assembler directive in the output.
7076 Refrain from adding @code{.ident} directives to the output file (this is
7079 @item -YP,@var{dirs}
7080 Search the directories @var{dirs}, and no others, for libraries
7081 specified with @samp{-l}.
7084 Look in the directory @var{dir} to find the M4 preprocessor.
7085 The assembler uses this option.
7086 @c This is supposed to go with a -Yd for predefined M4 macro files, but
7087 @c the generic assembler that comes with Solaris takes just -Ym.
7090 @node TMS320C3x/C4x Options
7091 @subsection TMS320C3x/C4x Options
7092 @cindex TMS320C3x/C4x Options
7094 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
7098 @item -mcpu=@var{cpu_type}
7099 Set the instruction set, register set, and instruction scheduling
7100 parameters for machine type @var{cpu_type}. Supported values for
7101 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
7102 @samp{c44}. The default is @samp{c40} to generate code for the
7107 @itemx -msmall-memory
7109 Generates code for the big or small memory model. The small memory
7110 model assumed that all data fits into one 64K word page. At run-time
7111 the data page (DP) register must be set to point to the 64K page
7112 containing the .bss and .data program sections. The big memory model is
7113 the default and requires reloading of the DP register for every direct
7118 Allow (disallow) allocation of general integer operands into the block
7123 Enable (disable) generation of code using decrement and branch,
7124 DBcond(D), instructions. This is enabled by default for the C4x. To be
7125 on the safe side, this is disabled for the C3x, since the maximum
7126 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
7127 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
7128 that it can utilise the decrement and branch instruction, but will give
7129 up if there is more than one memory reference in the loop. Thus a loop
7130 where the loop counter is decremented can generate slightly more
7131 efficient code, in cases where the RPTB instruction cannot be utilised.
7133 @item -mdp-isr-reload
7135 Force the DP register to be saved on entry to an interrupt service
7136 routine (ISR), reloaded to point to the data section, and restored on
7137 exit from the ISR. This should not be required unless someone has
7138 violated the small memory model by modifying the DP register, say within
7143 For the C3x use the 24-bit MPYI instruction for integer multiplies
7144 instead of a library call to guarantee 32-bit results. Note that if one
7145 of the operands is a constant, then the multiplication will be performed
7146 using shifts and adds. If the -mmpyi option is not specified for the C3x,
7147 then squaring operations are performed inline instead of a library call.
7150 @itemx -mno-fast-fix
7151 The C3x/C4x FIX instruction to convert a floating point value to an
7152 integer value chooses the nearest integer less than or equal to the
7153 floating point value rather than to the nearest integer. Thus if the
7154 floating point number is negative, the result will be incorrectly
7155 truncated an additional code is necessary to detect and correct this
7156 case. This option can be used to disable generation of the additional
7157 code required to correct the result.
7161 Enable (disable) generation of repeat block sequences using the RPTB
7162 instruction for zero overhead looping. The RPTB construct is only used
7163 for innermost loops that do not call functions or jump across the loop
7164 boundaries. There is no advantage having nested RPTB loops due to the
7165 overhead required to save and restore the RC, RS, and RE registers.
7166 This is enabled by default with -O2.
7168 @item -mrpts=@var{count}
7170 Enable (disable) the use of the single instruction repeat instruction
7171 RPTS. If a repeat block contains a single instruction, and the loop
7172 count can be guaranteed to be less than the value @var{count}, GCC will
7173 emit a RPTS instruction instead of a RPTB. If no value is specified,
7174 then a RPTS will be emitted even if the loop count cannot be determined
7175 at compile time. Note that the repeated instruction following RPTS does
7176 not have to be reloaded from memory each iteration, thus freeing up the
7177 CPU buses for operands. However, since interrupts are blocked by this
7178 instruction, it is disabled by default.
7180 @item -mloop-unsigned
7181 @itemx -mno-loop-unsigned
7182 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
7183 is 2^31 + 1 since these instructions test if the iteration count is
7184 negative to terminate the loop. If the iteration count is unsigned
7185 there is a possibility than the 2^31 + 1 maximum iteration count may be
7186 exceeded. This switch allows an unsigned iteration count.
7189 Try to emit an assembler syntax that the TI assembler (asm30) is happy
7190 with. This also enforces compatibility with the API employed by the TI
7191 C3x C compiler. For example, long doubles are passed as structures
7192 rather than in floating point registers.
7196 Generate code that uses registers (stack) for passing arguments to functions.
7197 By default, arguments are passed in registers where possible rather
7198 than by pushing arguments on to the stack.
7200 @item -mparallel-insns
7201 @itemx -mno-parallel-insns
7202 Allow the generation of parallel instructions. This is enabled by
7205 @item -mparallel-mpy
7206 @itemx -mno-parallel-mpy
7207 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
7208 provided -mparallel-insns is also specified. These instructions have
7209 tight register constraints which can pessimize the code generation
7215 @subsection V850 Options
7216 @cindex V850 Options
7218 These @samp{-m} options are defined for V850 implementations:
7222 @itemx -mno-long-calls
7223 Treat all calls as being far away (near). If calls are assumed to be
7224 far away, the compiler will always load the functions address up into a
7225 register, and call indirect through the pointer.
7229 Do not optimize (do optimize) basic blocks that use the same index
7230 pointer 4 or more times to copy pointer into the @code{ep} register, and
7231 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
7232 option is on by default if you optimize.
7234 @item -mno-prolog-function
7235 @itemx -mprolog-function
7236 Do not use (do use) external functions to save and restore registers at
7237 the prolog and epilog of a function. The external functions are slower,
7238 but use less code space if more than one function saves the same number
7239 of registers. The @samp{-mprolog-function} option is on by default if
7243 Try to make the code as small as possible. At present, this just turns
7244 on the @samp{-mep} and @samp{-mprolog-function} options.
7247 Put static or global variables whose size is @var{n} bytes or less into
7248 the tiny data area that register @code{ep} points to. The tiny data
7249 area can hold up to 256 bytes in total (128 bytes for byte references).
7252 Put static or global variables whose size is @var{n} bytes or less into
7253 the small data area that register @code{gp} points to. The small data
7254 area can hold up to 64 kilobytes.
7257 Put static or global variables whose size is @var{n} bytes or less into
7258 the first 32 kilobytes of memory.
7261 Specify that the target processor is the V850.
7264 Generate code suitable for big switch tables. Use this option only if
7265 the assembler/linker complain about out of range branches within a switch
7270 @subsection ARC Options
7273 These options are defined for ARC implementations:
7277 Compile code for little endian mode. This is the default.
7280 Compile code for big endian mode.
7283 Prepend the name of the cpu to all public symbol names.
7284 In multiple-processor systems, there are many ARC variants with different
7285 instruction and register set characteristics. This flag prevents code
7286 compiled for one cpu to be linked with code compiled for another.
7287 No facility exists for handling variants that are "almost identical".
7288 This is an all or nothing option.
7290 @item -mcpu=@var{cpu}
7291 Compile code for ARC variant @var{cpu}.
7292 Which variants are supported depend on the configuration.
7293 All variants support @samp{-mcpu=base}, this is the default.
7295 @item -mtext=@var{text section}
7296 @itemx -mdata=@var{data section}
7297 @itemx -mrodata=@var{readonly data section}
7298 Put functions, data, and readonly data in @var{text section},
7299 @var{data section}, and @var{readonly data section} respectively
7300 by default. This can be overridden with the @code{section} attribute.
7301 @xref{Variable Attributes}.
7306 @subsection NS32K Options
7307 @cindex NS32K options
7309 These are the @samp{-m} options defined for the 32000 series. The default
7310 values for these options depends on which style of 32000 was selected when
7311 the compiler was configured; the defaults for the most common choices are
7317 Generate output for a 32032. This is the default
7318 when the compiler is configured for 32032 and 32016 based systems.
7322 Generate output for a 32332. This is the default
7323 when the compiler is configured for 32332-based systems.
7327 Generate output for a 32532. This is the default
7328 when the compiler is configured for 32532-based systems.
7331 Generate output containing 32081 instructions for floating point.
7332 This is the default for all systems.
7335 Generate output containing 32381 instructions for floating point. This
7336 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
7337 and 32532 cpus. This is the default for the pc532-netbsd configuration.
7340 Try and generate multiply-add floating point instructions @code{polyF}
7341 and @code{dotF}. This option is only available if the @samp{-m32381}
7342 option is in effect. Using these instructions requires changes to to
7343 register allocation which generally has a negative impact on
7344 performance. This option should only be enabled when compiling code
7345 particularly likely to make heavy use of multiply-add instructions.
7348 Do not try and generate multiply-add floating point instructions
7349 @code{polyF} and @code{dotF}. This is the default on all platforms.
7352 Generate output containing library calls for floating point.
7353 @strong{Warning:} the requisite libraries may not be available.
7356 Do not use the bit-field instructions. On some machines it is faster to
7357 use shifting and masking operations. This is the default for the pc532.
7360 Do use the bit-field instructions. This is the default for all platforms
7364 Use a different function-calling convention, in which functions
7365 that take a fixed number of arguments return pop their
7366 arguments on return with the @code{ret} instruction.
7368 This calling convention is incompatible with the one normally
7369 used on Unix, so you cannot use it if you need to call libraries
7370 compiled with the Unix compiler.
7372 Also, you must provide function prototypes for all functions that
7373 take variable numbers of arguments (including @code{printf});
7374 otherwise incorrect code will be generated for calls to those
7377 In addition, seriously incorrect code will result if you call a
7378 function with too many arguments. (Normally, extra arguments are
7379 harmlessly ignored.)
7381 This option takes its name from the 680x0 @code{rtd} instruction.
7385 Use a different function-calling convention where the first two arguments
7386 are passed in registers.
7388 This calling convention is incompatible with the one normally
7389 used on Unix, so you cannot use it if you need to call libraries
7390 compiled with the Unix compiler.
7393 Do not pass any arguments in registers. This is the default for all
7397 It is OK to use the sb as an index register which is always loaded with
7398 zero. This is the default for the pc532-netbsd target.
7401 The sb register is not available for use or has not been initialized to
7402 zero by the run time system. This is the default for all targets except
7403 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
7404 @samp{-fpic} is set.
7407 Many ns32000 series addressing modes use displacements of up to 512MB.
7408 If an address is above 512MB then displacements from zero can not be used.
7409 This option causes code to be generated which can be loaded above 512MB.
7410 This may be useful for operating systems or ROM code.
7413 Assume code will be loaded in the first 512MB of virtual address space.
7414 This is the default for all platforms.
7420 @subsection AVR Options
7423 These options are defined for AVR implementations:
7426 @item -mmcu=@var{mcu}
7427 Specify ATMEL AVR instruction set or MCU type.
7429 Instruction set avr1 is for the minimal AVR core, not supported by the C
7430 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7431 attiny11, attiny12, attiny15, attiny28).
7433 Instruction set avr2 (default) is for the classic AVR core with up to
7434 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7435 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7436 at90c8534, at90s8535).
7438 Instruction set avr3 is for the classic AVR core with up to 128K program
7439 memory space (MCU types: atmega103, atmega603).
7441 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7442 memory space (MCU types: atmega83, atmega85).
7444 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7445 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
7448 Output instruction sizes to the asm file.
7450 @item -minit-stack=@var{N}
7451 Specify the initial stack address, which may be a symbol or numeric value,
7452 __stack is the default.
7454 @item -mno-interrupts
7455 Generated code is not compatible with hardware interrupts.
7456 Code size will be smaller.
7458 @item -mcall-prologues
7459 Functions prologues/epilogues expanded as call to appropriate
7460 subroutines. Code size will be smaller.
7462 @item -mno-tablejump
7463 Do not generate tablejump insns which sometimes increase code size.
7466 Change only the low 8 bits of the stack pointer.
7470 @subsection MCore Options
7471 @cindex MCore options
7473 These are the @samp{-m} options defined for the Motorola M*Core
7481 Inline constants into the code stream if it can be done in two
7482 instructions or less.
7487 Use the divide instruction. (Enabled by default).
7489 @item -mrelax-immediate
7490 @itemx -mrelax-immediate
7491 @itemx -mno-relax-immediate
7492 Allow arbitrary sized immediates in bit operations.
7494 @item -mwide-bitfields
7495 @itemx -mwide-bitfields
7496 @itemx -mno-wide-bitfields
7497 Always treat bitfields as int-sized.
7499 @item -m4byte-functions
7500 @itemx -m4byte-functions
7501 @itemx -mno-4byte-functions
7502 Force all functions to be aligned to a four byte boundary.
7504 @item -mcallgraph-data
7505 @itemx -mcallgraph-data
7506 @itemx -mno-callgraph-data
7507 Emit callgraph information.
7511 @itemx -mno-slow-bytes
7512 Prefer word access when reading byte quantities.
7514 @item -mlittle-endian
7515 @itemx -mlittle-endian
7517 Generate code for a little endian target.
7522 Generate code for the 210 processor.
7526 @subsection D30V Options
7527 @cindex D30V Options
7529 These @samp{-m} options are defined for D30V implementations:
7533 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
7534 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
7535 memory, which starts at location @code{0x80000000}.
7538 Same as the @samp{-mextmem} switch.
7541 Link the @samp{.text} section into onchip text memory, which starts at
7542 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
7543 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
7544 into onchip data memory, which starts at location @code{0x20000000}.
7546 @item -mno-asm-optimize
7547 @itemx -masm-optimize
7548 Disable (enable) passing @samp{-O} to the assembler when optimizing.
7549 The assembler uses the @samp{-O} option to automatically parallelize
7550 adjacent short instructions where possible.
7552 @item -mbranch-cost=@var{n}
7553 Increase the internal costs of branches to @var{n}. Higher costs means
7554 that the compiler will issue more instructions to avoid doing a branch.
7557 @item -mcond-exec=@var{n}
7558 Specify the maximum number of conditionally executed instructions that
7559 replace a branch. The default is 4.
7562 @node Code Gen Options
7563 @section Options for Code Generation Conventions
7564 @cindex code generation conventions
7565 @cindex options, code generation
7566 @cindex run-time options
7568 These machine-independent options control the interface conventions
7569 used in code generation.
7571 Most of them have both positive and negative forms; the negative form
7572 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
7573 one of the forms is listed---the one which is not the default. You
7574 can figure out the other form by either removing @samp{no-} or adding
7579 Enable exception handling. Generates extra code needed to propagate
7580 exceptions. For some targets, this implies GNU CC will generate frame
7581 unwind information for all functions, which can produce significant data
7582 size overhead, although it does not affect execution. If you do not
7583 specify this option, GNU CC will enable it by default for languages like
7584 C++ which normally require exception handling, and disable itfor
7585 languages like C that do not normally require it. However, you may need
7586 to enable this option when compiling C code that needs to interoperate
7587 properly with exception handlers written in C++. You may also wish to
7588 disable this option if you are compiling older C++ programs that don't
7589 use exception handling.
7591 @item -funwind-tables
7592 Similar to @code{-fexceptions}, except that it will just generate any needed
7593 static data, but will not affect the generated code in any other way.
7594 You will normally not enable this option; instead, a language processor
7595 that needs this handling would enable it on your behalf.
7597 @item -fpcc-struct-return
7598 Return ``short'' @code{struct} and @code{union} values in memory like
7599 longer ones, rather than in registers. This convention is less
7600 efficient, but it has the advantage of allowing intercallability between
7601 GCC-compiled files and files compiled with other compilers.
7603 The precise convention for returning structures in memory depends
7604 on the target configuration macros.
7606 Short structures and unions are those whose size and alignment match
7607 that of some integer type.
7609 @item -freg-struct-return
7610 Use the convention that @code{struct} and @code{union} values are
7611 returned in registers when possible. This is more efficient for small
7612 structures than @samp{-fpcc-struct-return}.
7614 If you specify neither @samp{-fpcc-struct-return} nor its contrary
7615 @samp{-freg-struct-return}, GCC defaults to whichever convention is
7616 standard for the target. If there is no standard convention, GCC
7617 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
7618 is the principal compiler. In those cases, we can choose the standard,
7619 and we chose the more efficient register return alternative.
7622 Allocate to an @code{enum} type only as many bytes as it needs for the
7623 declared range of possible values. Specifically, the @code{enum} type
7624 will be equivalent to the smallest integer type which has enough room.
7626 @item -fshort-double
7627 Use the same size for @code{double} as for @code{float}.
7630 Requests that the data and non-@code{const} variables of this
7631 compilation be shared data rather than private data. The distinction
7632 makes sense only on certain operating systems, where shared data is
7633 shared between processes running the same program, while private data
7634 exists in one copy per process.
7637 Allocate even uninitialized global variables in the data section of the
7638 object file, rather than generating them as common blocks. This has the
7639 effect that if the same variable is declared (without @code{extern}) in
7640 two different compilations, you will get an error when you link them.
7641 The only reason this might be useful is if you wish to verify that the
7642 program will work on other systems which always work this way.
7645 Ignore the @samp{#ident} directive.
7647 @item -fno-gnu-linker
7648 Do not output global initializations (such as C++ constructors and
7649 destructors) in the form used by the GNU linker (on systems where the GNU
7650 linker is the standard method of handling them). Use this option when
7651 you want to use a non-GNU linker, which also requires using the
7652 @code{collect2} program to make sure the system linker includes
7653 constructors and destructors. (@code{collect2} is included in the GCC
7654 distribution.) For systems which @emph{must} use @code{collect2}, the
7655 compiler driver @code{gcc} is configured to do this automatically.
7657 @item -finhibit-size-directive
7658 Don't output a @code{.size} assembler directive, or anything else that
7659 would cause trouble if the function is split in the middle, and the
7660 two halves are placed at locations far apart in memory. This option is
7661 used when compiling @file{crtstuff.c}; you should not need to use it
7665 Put extra commentary information in the generated assembly code to
7666 make it more readable. This option is generally only of use to those
7667 who actually need to read the generated assembly code (perhaps while
7668 debugging the compiler itself).
7670 @samp{-fno-verbose-asm}, the default, causes the
7671 extra information to be omitted and is useful when comparing two assembler
7675 Consider all memory references through pointers to be volatile.
7677 @item -fvolatile-global
7678 Consider all memory references to extern and global data items to
7679 be volatile. GCC does not consider static data items to be volatile
7680 because of this switch.
7682 @item -fvolatile-static
7683 Consider all memory references to static data to be volatile.
7686 @cindex global offset table
7688 Generate position-independent code (PIC) suitable for use in a shared
7689 library, if supported for the target machine. Such code accesses all
7690 constant addresses through a global offset table (GOT). The dynamic
7691 loader resolves the GOT entries when the program starts (the dynamic
7692 loader is not part of GCC; it is part of the operating system). If
7693 the GOT size for the linked executable exceeds a machine-specific
7694 maximum size, you get an error message from the linker indicating that
7695 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
7696 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
7697 on the m68k and RS/6000. The 386 has no such limit.)
7699 Position-independent code requires special support, and therefore works
7700 only on certain machines. For the 386, GCC supports PIC for System V
7701 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
7702 position-independent.
7705 If supported for the target machine, emit position-independent code,
7706 suitable for dynamic linking and avoiding any limit on the size of the
7707 global offset table. This option makes a difference on the m68k, m88k,
7710 Position-independent code requires special support, and therefore works
7711 only on certain machines.
7713 @item -ffixed-@var{reg}
7714 Treat the register named @var{reg} as a fixed register; generated code
7715 should never refer to it (except perhaps as a stack pointer, frame
7716 pointer or in some other fixed role).
7718 @var{reg} must be the name of a register. The register names accepted
7719 are machine-specific and are defined in the @code{REGISTER_NAMES}
7720 macro in the machine description macro file.
7722 This flag does not have a negative form, because it specifies a
7725 @item -fcall-used-@var{reg}
7726 Treat the register named @var{reg} as an allocable register that is
7727 clobbered by function calls. It may be allocated for temporaries or
7728 variables that do not live across a call. Functions compiled this way
7729 will not save and restore the register @var{reg}.
7731 It is an error to used this flag with the frame pointer or stack pointer.
7732 Use of this flag for other registers that have fixed pervasive roles in
7733 the machine's execution model will produce disastrous results.
7735 This flag does not have a negative form, because it specifies a
7738 @item -fcall-saved-@var{reg}
7739 Treat the register named @var{reg} as an allocable register saved by
7740 functions. It may be allocated even for temporaries or variables that
7741 live across a call. Functions compiled this way will save and restore
7742 the register @var{reg} if they use it.
7744 It is an error to used this flag with the frame pointer or stack pointer.
7745 Use of this flag for other registers that have fixed pervasive roles in
7746 the machine's execution model will produce disastrous results.
7748 A different sort of disaster will result from the use of this flag for
7749 a register in which function values may be returned.
7751 This flag does not have a negative form, because it specifies a
7755 Pack all structure members together without holes. Usually you would
7756 not want to use this option, since it makes the code suboptimal, and
7757 the offsets of structure members won't agree with system libraries.
7759 @item -fcheck-memory-usage
7760 Generate extra code to check each memory access. GCC will generate
7761 code that is suitable for a detector of bad memory accesses such as
7764 Normally, you should compile all, or none, of your code with this option.
7766 If you do mix code compiled with and without this option,
7767 you must ensure that all code that has side effects
7768 and that is called by code compiled with this option
7769 is, itself, compiled with this option.
7770 If you do not, you might get erroneous messages from the detector.
7772 If you use functions from a library that have side-effects (such as
7773 @code{read}), you might not be able to recompile the library and
7774 specify this option. In that case, you can enable the
7775 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
7776 your code and make other functions look as if they were compiled with
7777 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
7778 which are provided by the detector. If you cannot find or build
7779 stubs for every function you call, you might have to specify
7780 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
7782 If you specify this option, you can not use the @code{asm} or
7783 @code{__asm__} keywords in functions with memory checking enabled. GNU
7784 CC cannot understand what the @code{asm} statement may do, and therefore
7785 cannot generate the appropriate code, so it will reject it. However, if
7786 you specify the function attribute @code{no_check_memory_usage}
7787 (@pxref{Function Attributes}), GNU CC will disable memory checking within a
7788 function; you may use @code{asm} statements inside such functions. You
7789 may have an inline expansion of a non-checked function within a checked
7790 function; in that case GNU CC will not generate checks for the inlined
7791 function's memory accesses.
7793 If you move your @code{asm} statements to non-checked inline functions
7794 and they do access memory, you can add calls to the support code in your
7795 inline function, to indicate any reads, writes, or copies being done.
7796 These calls would be similar to those done in the stubs described above.
7798 @item -fprefix-function-name
7799 Request GCC to add a prefix to the symbols generated for function names.
7800 GCC adds a prefix to the names of functions defined as well as
7801 functions called. Code compiled with this option and code compiled
7802 without the option can't be linked together, unless stubs are used.
7804 If you compile the following code with @samp{-fprefix-function-name}
7806 extern void bar (int);
7815 GCC will compile the code as if it was written:
7817 extern void prefix_bar (int);
7821 return prefix_bar (a + 5);
7824 This option is designed to be used with @samp{-fcheck-memory-usage}.
7826 @item -finstrument-functions
7827 Generate instrumentation calls for entry and exit to functions. Just
7828 after function entry and just before function exit, the following
7829 profiling functions will be called with the address of the current
7830 function and its call site. (On some platforms,
7831 @code{__builtin_return_address} does not work beyond the current
7832 function, so the call site information may not be available to the
7833 profiling functions otherwise.)
7836 void __cyg_profile_func_enter (void *this_fn, void *call_site);
7837 void __cyg_profile_func_exit (void *this_fn, void *call_site);
7840 The first argument is the address of the start of the current function,
7841 which may be looked up exactly in the symbol table.
7843 This instrumentation is also done for functions expanded inline in other
7844 functions. The profiling calls will indicate where, conceptually, the
7845 inline function is entered and exited. This means that addressable
7846 versions of such functions must be available. If all your uses of a
7847 function are expanded inline, this may mean an additional expansion of
7848 code size. If you use @samp{extern inline} in your C code, an
7849 addressable version of such functions must be provided. (This is
7850 normally the case anyways, but if you get lucky and the optimizer always
7851 expands the functions inline, you might have gotten away without
7852 providing static copies.)
7854 A function may be given the attribute @code{no_instrument_function}, in
7855 which case this instrumentation will not be done. This can be used, for
7856 example, for the profiling functions listed above, high-priority
7857 interrupt routines, and any functions from which the profiling functions
7858 cannot safely be called (perhaps signal handlers, if the profiling
7859 routines generate output or allocate memory).
7862 Generate code to verify that you do not go beyond the boundary of the
7863 stack. You should specify this flag if you are running in an
7864 environment with multiple threads, but only rarely need to specify it in
7865 a single-threaded environment since stack overflow is automatically
7866 detected on nearly all systems if there is only one stack.
7868 Note that this switch does not actually cause checking to be done; the
7869 operating system must do that. The switch causes generation of code
7870 to ensure that the operating system sees the stack being extended.
7872 @item -fstack-limit-register=@var{reg}
7873 @itemx -fstack-limit-symbol=@var{sym}
7874 @itemx -fno-stack-limit
7875 Generate code to ensure that the stack does not grow beyond a certain value,
7876 either the value of a register or the address of a symbol. If the stack
7877 would grow beyond the value, a signal is raised. For most targets,
7878 the signal is raised before the stack overruns the boundary, so
7879 it is possible to catch the signal without taking special precautions.
7881 For instance, if the stack starts at address @samp{0x80000000} and grows
7882 downwards you can use the flags
7883 @samp{-fstack-limit-symbol=__stack_limit}
7884 @samp{-Wl,--defsym,__stack_limit=0x7ffe0000} which will enforce a stack
7887 @cindex aliasing of parameters
7888 @cindex parameters, aliased
7889 @item -fargument-alias
7890 @itemx -fargument-noalias
7891 @itemx -fargument-noalias-global
7892 Specify the possible relationships among parameters and between
7893 parameters and global data.
7895 @samp{-fargument-alias} specifies that arguments (parameters) may
7896 alias each other and may alias global storage.
7897 @samp{-fargument-noalias} specifies that arguments do not alias
7898 each other, but may alias global storage.
7899 @samp{-fargument-noalias-global} specifies that arguments do not
7900 alias each other and do not alias global storage.
7902 Each language will automatically use whatever option is required by
7903 the language standard. You should not need to use these options yourself.
7905 @item -fleading-underscore
7906 This option and its counterpart, -fno-leading-underscore, forcibly
7907 change the way C symbols are represented in the object file. One use
7908 is to help link with legacy assembly code.
7910 Be warned that you should know what you are doing when invoking this
7911 option, and that not all targets provide complete support for it.
7916 @node Environment Variables
7917 @section Environment Variables Affecting GCC
7918 @cindex environment variables
7920 @c man begin ENVIRONMENT
7922 This section describes several environment variables that affect how GCC
7923 operates. Some of them work by specifying directories or prefixes to use
7924 when searching for various kinds of files. Some are used to specify other
7925 aspects of the compilation environment.
7928 Note that you can also specify places to search using options such as
7929 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7930 take precedence over places specified using environment variables, which
7931 in turn take precedence over those specified by the configuration of GCC.
7935 Note that you can also specify places to search using options such as
7936 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
7937 take precedence over places specified using environment variables, which
7938 in turn take precedence over those specified by the configuration of GCC.
7945 @c @itemx LC_COLLATE
7947 @c @itemx LC_MONETARY
7948 @c @itemx LC_NUMERIC
7953 @c @findex LC_COLLATE
7955 @c @findex LC_MONETARY
7956 @c @findex LC_NUMERIC
7960 These environment variables control the way that GCC uses
7961 localization information that allow GCC to work with different
7962 national conventions. GCC inspects the locale categories
7963 @code{LC_CTYPE} and @code{LC_MESSAGES} if it has been configured to do
7964 so. These locale categories can be set to any value supported by your
7965 installation. A typical value is @samp{en_UK} for English in the United
7968 The @code{LC_CTYPE} environment variable specifies character
7969 classification. GCC uses it to determine the character boundaries in
7970 a string; this is needed for some multibyte encodings that contain quote
7971 and escape characters that would otherwise be interpreted as a string
7974 The @code{LC_MESSAGES} environment variable specifies the language to
7975 use in diagnostic messages.
7977 If the @code{LC_ALL} environment variable is set, it overrides the value
7978 of @code{LC_CTYPE} and @code{LC_MESSAGES}; otherwise, @code{LC_CTYPE}
7979 and @code{LC_MESSAGES} default to the value of the @code{LANG}
7980 environment variable. If none of these variables are set, GCC
7981 defaults to traditional C English behavior.
7985 If @code{TMPDIR} is set, it specifies the directory to use for temporary
7986 files. GCC uses temporary files to hold the output of one stage of
7987 compilation which is to be used as input to the next stage: for example,
7988 the output of the preprocessor, which is the input to the compiler
7991 @item GCC_EXEC_PREFIX
7992 @findex GCC_EXEC_PREFIX
7993 If @code{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
7994 names of the subprograms executed by the compiler. No slash is added
7995 when this prefix is combined with the name of a subprogram, but you can
7996 specify a prefix that ends with a slash if you wish.
7998 If @code{GCC_EXEC_PREFIX} is not set, GNU CC will attempt to figure out
7999 an appropriate prefix to use based on the pathname it was invoked with.
8001 If GCC cannot find the subprogram using the specified prefix, it
8002 tries looking in the usual places for the subprogram.
8004 The default value of @code{GCC_EXEC_PREFIX} is
8005 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
8006 of @code{prefix} when you ran the @file{configure} script.
8008 Other prefixes specified with @samp{-B} take precedence over this prefix.
8010 This prefix is also used for finding files such as @file{crt0.o} that are
8013 In addition, the prefix is used in an unusual way in finding the
8014 directories to search for header files. For each of the standard
8015 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
8016 (more precisely, with the value of @code{GCC_INCLUDE_DIR}), GCC tries
8017 replacing that beginning with the specified prefix to produce an
8018 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
8019 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
8020 These alternate directories are searched first; the standard directories
8024 @findex COMPILER_PATH
8025 The value of @code{COMPILER_PATH} is a colon-separated list of
8026 directories, much like @code{PATH}. GCC tries the directories thus
8027 specified when searching for subprograms, if it can't find the
8028 subprograms using @code{GCC_EXEC_PREFIX}.
8031 @findex LIBRARY_PATH
8032 The value of @code{LIBRARY_PATH} is a colon-separated list of
8033 directories, much like @code{PATH}. When configured as a native compiler,
8034 GCC tries the directories thus specified when searching for special
8035 linker files, if it can't find them using @code{GCC_EXEC_PREFIX}. Linking
8036 using GCC also uses these directories when searching for ordinary
8037 libraries for the @samp{-l} option (but directories specified with
8038 @samp{-L} come first).
8040 @item C_INCLUDE_PATH
8041 @itemx CPLUS_INCLUDE_PATH
8042 @itemx OBJC_INCLUDE_PATH
8043 @findex C_INCLUDE_PATH
8044 @findex CPLUS_INCLUDE_PATH
8045 @findex OBJC_INCLUDE_PATH
8046 @c @itemx OBJCPLUS_INCLUDE_PATH
8047 These environment variables pertain to particular languages. Each
8048 variable's value is a colon-separated list of directories, much like
8049 @code{PATH}. When GCC searches for header files, it tries the
8050 directories listed in the variable for the language you are using, after
8051 the directories specified with @samp{-I} but before the standard header
8054 @item DEPENDENCIES_OUTPUT
8055 @findex DEPENDENCIES_OUTPUT
8056 @cindex dependencies for make as output
8057 If this variable is set, its value specifies how to output dependencies
8058 for Make based on the header files processed by the compiler. This
8059 output looks much like the output from the @samp{-M} option
8060 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
8061 in addition to the usual results of compilation.
8063 The value of @code{DEPENDENCIES_OUTPUT} can be just a file name, in
8064 which case the Make rules are written to that file, guessing the target
8065 name from the source file name. Or the value can have the form
8066 @samp{@var{file} @var{target}}, in which case the rules are written to
8067 file @var{file} using @var{target} as the target name.
8071 @cindex locale definition
8072 This variable is used to pass locale information to the compiler. One way in
8073 which this information is used is to determine the character set to be used
8074 when character literals, string literals and comments are parsed in C and C++.
8075 When the compiler is configured to allow multibyte characters,
8076 the following values for @code{LANG} are recognized:
8080 Recognize JIS characters.
8082 Recognize SJIS characters.
8084 Recognize EUCJP characters.
8087 If @code{LANG} is not defined, or if it has some other value, then the
8088 compiler will use mblen and mbtowc as defined by the default locale to
8089 recognize and translate multibyte characters.
8094 @node Running Protoize
8095 @section Running Protoize
8097 The program @code{protoize} is an optional part of GNU C. You can use
8098 it to add prototypes to a program, thus converting the program to ISO
8099 C in one respect. The companion program @code{unprotoize} does the
8100 reverse: it removes argument types from any prototypes that are found.
8102 When you run these programs, you must specify a set of source files as
8103 command line arguments. The conversion programs start out by compiling
8104 these files to see what functions they define. The information gathered
8105 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
8107 After scanning comes actual conversion. The specified files are all
8108 eligible to be converted; any files they include (whether sources or
8109 just headers) are eligible as well.
8111 But not all the eligible files are converted. By default,
8112 @code{protoize} and @code{unprotoize} convert only source and header
8113 files in the current directory. You can specify additional directories
8114 whose files should be converted with the @samp{-d @var{directory}}
8115 option. You can also specify particular files to exclude with the
8116 @samp{-x @var{file}} option. A file is converted if it is eligible, its
8117 directory name matches one of the specified directory names, and its
8118 name within the directory has not been excluded.
8120 Basic conversion with @code{protoize} consists of rewriting most
8121 function definitions and function declarations to specify the types of
8122 the arguments. The only ones not rewritten are those for varargs
8125 @code{protoize} optionally inserts prototype declarations at the
8126 beginning of the source file, to make them available for any calls that
8127 precede the function's definition. Or it can insert prototype
8128 declarations with block scope in the blocks where undeclared functions
8131 Basic conversion with @code{unprotoize} consists of rewriting most
8132 function declarations to remove any argument types, and rewriting
8133 function definitions to the old-style pre-ISO form.
8135 Both conversion programs print a warning for any function declaration or
8136 definition that they can't convert. You can suppress these warnings
8139 The output from @code{protoize} or @code{unprotoize} replaces the
8140 original source file. The original file is renamed to a name ending
8141 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
8142 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
8143 for DOS) file already exists, then the source file is simply discarded.
8145 @code{protoize} and @code{unprotoize} both depend on GCC itself to
8146 scan the program and collect information about the functions it uses.
8147 So neither of these programs will work until GCC is installed.
8149 Here is a table of the options you can use with @code{protoize} and
8150 @code{unprotoize}. Each option works with both programs unless
8154 @item -B @var{directory}
8155 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
8156 usual directory (normally @file{/usr/local/lib}). This file contains
8157 prototype information about standard system functions. This option
8158 applies only to @code{protoize}.
8160 @item -c @var{compilation-options}
8161 Use @var{compilation-options} as the options when running @code{gcc} to
8162 produce the @samp{.X} files. The special option @samp{-aux-info} is
8163 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
8165 Note that the compilation options must be given as a single argument to
8166 @code{protoize} or @code{unprotoize}. If you want to specify several
8167 @code{gcc} options, you must quote the entire set of compilation options
8168 to make them a single word in the shell.
8170 There are certain @code{gcc} arguments that you cannot use, because they
8171 would produce the wrong kind of output. These include @samp{-g},
8172 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
8173 the @var{compilation-options}, they are ignored.
8176 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
8177 systems) instead of @samp{.c}. This is convenient if you are converting
8178 a C program to C++. This option applies only to @code{protoize}.
8181 Add explicit global declarations. This means inserting explicit
8182 declarations at the beginning of each source file for each function
8183 that is called in the file and was not declared. These declarations
8184 precede the first function definition that contains a call to an
8185 undeclared function. This option applies only to @code{protoize}.
8187 @item -i @var{string}
8188 Indent old-style parameter declarations with the string @var{string}.
8189 This option applies only to @code{protoize}.
8191 @code{unprotoize} converts prototyped function definitions to old-style
8192 function definitions, where the arguments are declared between the
8193 argument list and the initial @samp{@{}. By default, @code{unprotoize}
8194 uses five spaces as the indentation. If you want to indent with just
8195 one space instead, use @samp{-i " "}.
8198 Keep the @samp{.X} files. Normally, they are deleted after conversion
8202 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
8203 a prototype declaration for each function in each block which calls the
8204 function without any declaration. This option applies only to
8208 Make no real changes. This mode just prints information about the conversions
8209 that would have been done without @samp{-n}.
8212 Make no @samp{.save} files. The original files are simply deleted.
8213 Use this option with caution.
8215 @item -p @var{program}
8216 Use the program @var{program} as the compiler. Normally, the name
8220 Work quietly. Most warnings are suppressed.
8223 Print the version number, just like @samp{-v} for @code{gcc}.
8226 If you need special compiler options to compile one of your program's
8227 source files, then you should generate that file's @samp{.X} file
8228 specially, by running @code{gcc} on that source file with the
8229 appropriate options and the option @samp{-aux-info}. Then run
8230 @code{protoize} on the entire set of files. @code{protoize} will use
8231 the existing @samp{.X} file because it is newer than the source file.
8235 gcc -Dfoo=bar file1.c -aux-info
8240 You need to include the special files along with the rest in the
8241 @code{protoize} command, even though their @samp{.X} files already
8242 exist, because otherwise they won't get converted.
8244 @xref{Protoize Caveats}, for more information on how to use
8245 @code{protoize} successfully.