1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001, 2002 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.1 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
70 When you invoke GCC, it normally does preprocessing, compilation,
71 assembly and linking. The ``overall options'' allow you to stop this
72 process at an intermediate stage. For example, the @option{-c} option
73 says not to run the linker. Then the output consists of object files
74 output by the assembler.
76 Other options are passed on to one stage of processing. Some options
77 control the preprocessor and others the compiler itself. Yet other
78 options control the assembler and linker; most of these are not
79 documented here, since you rarely need to use any of them.
81 @cindex C compilation options
82 Most of the command line options that you can use with GCC are useful
83 for C programs; when an option is only useful with another language
84 (usually C++), the explanation says so explicitly. If the description
85 for a particular option does not mention a source language, you can use
86 that option with all supported languages.
88 @cindex C++ compilation options
89 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
90 options for compiling C++ programs.
92 @cindex grouping options
93 @cindex options, grouping
94 The @command{gcc} program accepts options and file names as operands. Many
95 options have multi-letter names; therefore multiple single-letter options
96 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
99 @cindex order of options
100 @cindex options, order
101 You can mix options and other arguments. For the most part, the order
102 you use doesn't matter. Order does matter when you use several options
103 of the same kind; for example, if you specify @option{-L} more than once,
104 the directories are searched in the order specified.
106 Many options have long names starting with @samp{-f} or with
107 @samp{-W}---for example, @option{-fforce-mem},
108 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
109 these have both positive and negative forms; the negative form of
110 @option{-ffoo} would be @option{-fno-foo}. This manual documents
111 only one of these two forms, whichever one is not the default.
115 @xref{Option Index}, for an index to GCC's options.
118 * Option Summary:: Brief list of all options, without explanations.
119 * Overall Options:: Controlling the kind of output:
120 an executable, object files, assembler files,
121 or preprocessed source.
122 * Invoking G++:: Compiling C++ programs.
123 * C Dialect Options:: Controlling the variant of C language compiled.
124 * C++ Dialect Options:: Variations on C++.
125 * Objective-C Dialect Options:: Variations on Objective-C.
126 * Language Independent Options:: Controlling how diagnostics should be
128 * Warning Options:: How picky should the compiler be?
129 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
130 * Optimize Options:: How much optimization?
131 * Preprocessor Options:: Controlling header files and macro definitions.
132 Also, getting dependency information for Make.
133 * Assembler Options:: Passing options to the assembler.
134 * Link Options:: Specifying libraries and so on.
135 * Directory Options:: Where to find header files and libraries.
136 Where to find the compiler executable files.
137 * Spec Files:: How to pass switches to sub-processes.
138 * Target Options:: Running a cross-compiler, or an old version of GCC.
139 * Submodel Options:: Specifying minor hardware or convention variations,
140 such as 68010 vs 68020.
141 * Code Gen Options:: Specifying conventions for function calls, data layout
143 * Environment Variables:: Env vars that affect GCC.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
159 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language} @gol
160 -v -### --help --target-help --version}
162 @item C Language Options
163 @xref{C Dialect Options,,Options Controlling C Dialect}.
165 -ansi -std=@var{standard} -aux-info @var{filename} @gol
166 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
167 -fhosted -ffreestanding -fms-extensions @gol
168 -trigraphs -traditional -traditional-cpp @gol
169 -fallow-single-precision -fcond-mismatch @gol
170 -fsigned-bitfields -fsigned-char @gol
171 -funsigned-bitfields -funsigned-char @gol
174 @item C++ Language Options
175 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
177 -fabi-version=@var{n} -fno-access-control -fcheck-new @gol
178 -fconserve-space -fno-const-strings -fdollars-in-identifiers @gol
179 -fno-elide-constructors @gol
180 -fno-enforce-eh-specs -fexternal-templates @gol
181 -falt-external-templates @gol
182 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
183 -fno-implicit-templates @gol
184 -fno-implicit-inline-templates @gol
185 -fno-implement-inlines -fms-extensions @gol
186 -fno-nonansi-builtins -fno-operator-names @gol
187 -fno-optional-diags -fpermissive @gol
188 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
189 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
190 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
191 -Wnon-virtual-dtor -Wreorder @gol
192 -Weffc++ -Wno-deprecated @gol
193 -Wno-non-template-friend -Wold-style-cast @gol
194 -Woverloaded-virtual -Wno-pmf-conversions @gol
195 -Wsign-promo -Wsynth}
197 @item Objective-C Language Options
198 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
200 -fconstant-string-class=@var{class-name} @gol
201 -fgnu-runtime -fnext-runtime -gen-decls @gol
202 -Wno-protocol -Wselector -Wundeclared-selector}
204 @item Language Independent Options
205 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
207 -fmessage-length=@var{n} @gol
208 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
210 @item Warning Options
211 @xref{Warning Options,,Options to Request or Suppress Warnings}.
213 -fsyntax-only -pedantic -pedantic-errors @gol
214 -w -W -Wall -Waggregate-return @gol
215 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
216 -Wconversion -Wno-deprecated-declarations @gol
217 -Wdisabled-optimization -Wno-div-by-zero -Werror @gol
218 -Wfloat-equal -Wformat -Wformat=2 @gol
219 -Wformat-nonliteral -Wformat-security @gol
220 -Wimplicit -Wimplicit-int @gol
221 -Wimplicit-function-declaration @gol
222 -Werror-implicit-function-declaration @gol
223 -Wimport -Winline -Wno-endif-labels @gol
224 -Wlarger-than-@var{len} -Wlong-long @gol
225 -Wmain -Wmissing-braces @gol
226 -Wmissing-format-attribute -Wmissing-noreturn @gol
227 -Wno-multichar -Wno-format-extra-args -Wno-format-y2k @gol
228 -Wno-import -Wnonnull -Wpacked -Wpadded @gol
229 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
230 -Wreturn-type -Wsequence-point -Wshadow @gol
231 -Wsign-compare -Wstrict-aliasing @gol
232 -Wswitch -Wswitch-default -Wswitch-enum @gol
233 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
234 -Wunknown-pragmas -Wunreachable-code @gol
235 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
236 -Wunused-value -Wunused-variable -Wwrite-strings}
238 @item C-only Warning Options
240 -Wbad-function-cast -Wmissing-declarations @gol
241 -Wmissing-prototypes -Wnested-externs @gol
242 -Wstrict-prototypes -Wtraditional}
244 @item Debugging Options
245 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
247 -d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
248 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
249 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-original@r{[}-@var{n}@r{]} -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
251 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
252 -feliminate-dwarf2-dups -fmem-report @gol
253 -fprofile-arcs -fsched-verbose=@var{n} @gol
254 -ftest-coverage -ftime-report @gol
255 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
256 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
257 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
258 -print-multi-directory -print-multi-lib @gol
259 -print-prog-name=@var{program} -print-search-dirs -Q @gol
262 @item Optimization Options
263 @xref{Optimize Options,,Options that Control Optimization}.
265 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
266 -falign-labels=@var{n} -falign-loops=@var{n} @gol
268 -fbranch-probabilities -fcaller-saves -fcprop-registers @gol
269 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
270 -fdelayed-branch -fdelete-null-pointer-checks @gol
271 -fexpensive-optimizations -ffast-math -ffloat-store @gol
272 -fforce-addr -fforce-mem -ffunction-sections @gol
273 -fgcse -fgcse-lm -fgcse-sm -floop-optimize -fcrossjumping @gol
274 -fif-conversion -fif-conversion2 @gol
275 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
276 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
277 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
278 -fno-default-inline -fno-defer-pop @gol
279 -fno-function-cse -fno-guess-branch-probability @gol
280 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
281 -funsafe-math-optimizations -ffinite-math-only @gol
282 -fno-trapping-math -fno-zero-initialized-in-bss @gol
283 -fomit-frame-pointer -foptimize-register-move @gol
284 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
285 -freduce-all-givs -fregmove -frename-registers @gol
286 -freorder-blocks -freorder-functions @gol
287 -frerun-cse-after-loop -frerun-loop-opt @gol
288 -fschedule-insns -fschedule-insns2 @gol
289 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
290 -fsched-spec-load-dangerous -fsignaling-nans @gol
291 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
292 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
293 -ftrapv -funroll-all-loops -funroll-loops @gol
294 --param @var{name}=@var{value}
295 -O -O0 -O1 -O2 -O3 -Os}
297 @item Preprocessor Options
298 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
300 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
301 -C -dD -dI -dM -dN @gol
302 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
303 -idirafter @var{dir} @gol
304 -include @var{file} -imacros @var{file} @gol
305 -iprefix @var{file} -iwithprefix @var{dir} @gol
306 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
307 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
308 -trigraphs -undef -U@var{macro} -Wp,@var{option}}
310 @item Assembler Option
311 @xref{Assembler Options,,Passing Options to the Assembler}.
316 @xref{Link Options,,Options for Linking}.
318 @var{object-file-name} -l@var{library} @gol
319 -nostartfiles -nodefaultlibs -nostdlib @gol
320 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
321 -Wl,@var{option} -Xlinker @var{option} @gol
324 @item Directory Options
325 @xref{Directory Options,,Options for Directory Search}.
327 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
330 @c I wrote this xref this way to avoid overfull hbox. -- rms
331 @xref{Target Options}.
333 -V @var{version} -b @var{machine}}
335 @item Machine Dependent Options
336 @xref{Submodel Options,,Hardware Models and Configurations}.
338 @emph{M680x0 Options}
340 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
341 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
342 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
343 -malign-int -mstrict-align}
345 @emph{M68hc1x Options}
347 -m6811 -m6812 -m68hc11 -m68hc12 @gol
348 -mauto-incdec -minmax -mlong-calls -mshort @gol
349 -msoft-reg-count=@var{count}}
357 -mcpu=@var{cpu-type} @gol
358 -mtune=@var{cpu-type} @gol
359 -mcmodel=@var{code-model} @gol
361 -mapp-regs -mbroken-saverestore -mcypress @gol
362 -mfaster-structs -mflat @gol
363 -mfpu -mhard-float -mhard-quad-float @gol
364 -mimpure-text -mlive-g0 -mno-app-regs @gol
365 -mno-faster-structs -mno-flat -mno-fpu @gol
366 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
367 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
368 -msupersparc -munaligned-doubles -mv8}
370 @emph{Convex Options}
372 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
373 -margcount -mnoargcount @gol
374 -mlong32 -mlong64 @gol
375 -mvolatile-cache -mvolatile-nocache}
377 @emph{AMD29K Options}
379 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
380 -mlarge -mnormal -msmall @gol
381 -mkernel-registers -mno-reuse-arg-regs @gol
382 -mno-stack-check -mno-storem-bug @gol
383 -mreuse-arg-regs -msoft-float -mstack-check @gol
384 -mstorem-bug -muser-registers}
388 -mapcs-frame -mno-apcs-frame @gol
389 -mapcs-26 -mapcs-32 @gol
390 -mapcs-stack-check -mno-apcs-stack-check @gol
391 -mapcs-float -mno-apcs-float @gol
392 -mapcs-reentrant -mno-apcs-reentrant @gol
393 -msched-prolog -mno-sched-prolog @gol
394 -mlittle-endian -mbig-endian -mwords-little-endian @gol
395 -malignment-traps -mno-alignment-traps @gol
396 -msoft-float -mhard-float -mfpe @gol
397 -mthumb-interwork -mno-thumb-interwork @gol
398 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
399 -mstructure-size-boundary=@var{n} @gol
400 -mbsd -mxopen -mno-symrename @gol
401 -mabort-on-noreturn @gol
402 -mlong-calls -mno-long-calls @gol
403 -msingle-pic-base -mno-single-pic-base @gol
404 -mpic-register=@var{reg} @gol
405 -mnop-fun-dllimport @gol
406 -mpoke-function-name @gol
408 -mtpcs-frame -mtpcs-leaf-frame @gol
409 -mcaller-super-interworking -mcallee-super-interworking }
411 @emph{MN10200 Options}
415 @emph{MN10300 Options}
417 -mmult-bug -mno-mult-bug @gol
418 -mam33 -mno-am33 @gol
421 @emph{M32R/D Options}
423 -m32rx -m32r -mcode-model=@var{model-type} -msdata=@var{sdata-type} @gol
428 -m88000 -m88100 -m88110 -mbig-pic @gol
429 -mcheck-zero-division -mhandle-large-shift @gol
430 -midentify-revision -mno-check-zero-division @gol
431 -mno-ocs-debug-info -mno-ocs-frame-position @gol
432 -mno-optimize-arg-area -mno-serialize-volatile @gol
433 -mno-underscores -mocs-debug-info @gol
434 -mocs-frame-position -moptimize-arg-area @gol
435 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
436 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
437 -mversion-03.00 -mwarn-passed-structs}
439 @emph{RS/6000 and PowerPC Options}
441 -mcpu=@var{cpu-type} @gol
442 -mtune=@var{cpu-type} @gol
443 -mpower -mno-power -mpower2 -mno-power2 @gol
444 -mpowerpc -mpowerpc64 -mno-powerpc @gol
445 -maltivec -mno-altivec @gol
446 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
447 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
448 -mnew-mnemonics -mold-mnemonics @gol
449 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
450 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
451 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
452 -mstring -mno-string -mupdate -mno-update @gol
453 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
454 -mstrict-align -mno-strict-align -mrelocatable @gol
455 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
456 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
457 -mcall-aix -mcall-sysv -mcall-netbsd @gol
458 -maix-struct-return -msvr4-struct-return @gol
459 -mabi=altivec -mabi=no-altivec @gol
460 -mabi=spe -mabi=no-spe @gol
461 -misel=yes -misel=no @gol
462 -mprototype -mno-prototype @gol
463 -msim -mmvme -mads -myellowknife -memb -msdata @gol
464 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
466 @emph{Darwin Options}
468 -all_load -allowable_client -arch -arch_errors_fatal @gol
469 -arch_only -bind_at_load -bundle -bundle_loader @gol
470 -client_name -compatibility_version -current_version @gol
471 -dependency-file -dylib_file -dylinker_install_name @gol
472 -dynamic -dynamiclib -exported_symbols_list @gol
473 -filelist -flat_namespace -force_cpusubtype_ALL @gol
474 -force_flat_namespace -headerpad_max_install_names @gol
475 -image_base -init -install_name -keep_private_externs @gol
476 -multi_module -multiply_defined -multiply_defined_unused @gol
477 -noall_load -nomultidefs -noprebind -noseglinkedit @gol
478 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
479 -private_bundle -read_only_relocs -sectalign @gol
480 -sectobjectsymbols -whyload -seg1addr @gol
481 -sectcreate -sectobjectsymbols -sectorder @gol
482 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
483 -segprot -segs_read_only_addr -segs_read_write_addr @gol
484 -single_module -static -sub_library -sub_umbrella @gol
485 -twolevel_namespace -umbrella -undefined @gol
486 -unexported_symbols_list -weak_reference_mismatches -whatsloaded}
490 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
491 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
492 -mminimum-fp-blocks -mnohc-struct-return}
496 -mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
497 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
498 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
499 -mgas -mgp32 -mgp64 @gol
500 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
501 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
502 -mmips-as -mmips-tfile -mno-abicalls @gol
503 -mno-embedded-data -mno-uninit-const-in-rodata @gol
504 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
505 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
506 -mrnames -msoft-float @gol
507 -m4650 -msingle-float -mmad @gol
508 -mstats -EL -EB -G @var{num} -nocpp @gol
509 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
510 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
511 -mbranch-likely -mno-branch-likely}
513 @emph{i386 and x86-64 Options}
515 -mcpu=@var{cpu-type} -march=@var{cpu-type} -mfpmath=@var{unit} @gol
516 -masm=@var{dialect} -mno-fancy-math-387 @gol
517 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
518 -mno-wide-multiply -mrtd -malign-double @gol
519 -mpreferred-stack-boundary=@var{num} @gol
520 -mmmx -msse -msse2 -m3dnow @gol
521 -mthreads -mno-align-stringops -minline-all-stringops @gol
522 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
523 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
525 -mcmodel=@var{code-model} @gol
530 -march=@var{architecture-type} @gol
531 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
532 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
533 -mjump-in-delay -mlinker-opt -mlong-calls @gol
534 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
535 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
536 -mno-jump-in-delay -mno-long-load-store @gol
537 -mno-portable-runtime -mno-soft-float @gol
538 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
539 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
540 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio}
542 @emph{Intel 960 Options}
544 -m@var{cpu-type} -masm-compat -mclean-linkage @gol
545 -mcode-align -mcomplex-addr -mleaf-procedures @gol
546 -mic-compat -mic2.0-compat -mic3.0-compat @gol
547 -mintel-asm -mno-clean-linkage -mno-code-align @gol
548 -mno-complex-addr -mno-leaf-procedures @gol
549 -mno-old-align -mno-strict-align -mno-tail-call @gol
550 -mnumerics -mold-align -msoft-float -mstrict-align @gol
553 @emph{DEC Alpha Options}
555 -mno-fp-regs -msoft-float -malpha-as -mgas @gol
556 -mieee -mieee-with-inexact -mieee-conformant @gol
557 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
558 -mtrap-precision=@var{mode} -mbuild-constants @gol
559 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
560 -mbwx -mmax -mfix -mcix @gol
561 -mfloat-vax -mfloat-ieee @gol
562 -mexplicit-relocs -msmall-data -mlarge-data @gol
563 -mmemory-latency=@var{time}}
565 @emph{DEC Alpha/VMS Options}
569 @emph{Clipper Options}
573 @emph{H8/300 Options}
575 -mrelax -mh -ms -mn -mint32 -malign-300}
579 -m1 -m2 -m3 -m3e @gol
580 -m4-nofpu -m4-single-only -m4-single -m4 @gol
581 -m5-64media -m5-64media-nofpu @gol
582 -m5-32media -m5-32media-nofpu @gol
583 -m5-compact -m5-compact-nofpu @gol
584 -mb -ml -mdalign -mrelax @gol
585 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
586 -mieee -misize -mpadstruct -mspace @gol
587 -mprefergot -musermode}
589 @emph{System V Options}
591 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
596 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
597 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
599 @emph{TMS320C3x/C4x Options}
601 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
602 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
603 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
604 -mparallel-insns -mparallel-mpy -mpreserve-float}
608 -mlong-calls -mno-long-calls -mep -mno-ep @gol
609 -mprolog-function -mno-prolog-function -mspace @gol
610 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
611 -mapp-regs -mno-app-regs @gol
612 -mdisable-callt -mno-disable-callt @gol
618 -m32032 -m32332 -m32532 -m32081 -m32381 @gol
619 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
620 -mregparam -mnoregparam -msb -mnosb @gol
621 -mbitfield -mnobitfield -mhimem -mnohimem}
625 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
626 -mcall-prologues -mno-tablejump -mtiny-stack}
630 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
631 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
632 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
633 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
634 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
638 -mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
639 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
640 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
641 -mno-base-addresses -msingle-exit -mno-single-exit}
645 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
646 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
647 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
648 -minline-float-divide-max-throughput -minline-int-divide-min-latency @gol
649 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
650 -mfixed-range=@var{register-range}}
654 -mextmem -mextmemory -monchip -mno-asm-optimize @gol
655 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
657 @emph{S/390 and zSeries Options}
659 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
660 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
661 -m64 -m31 -mdebug -mno-debug}
665 -mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
666 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
667 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
668 -mstack-align -mdata-align -mconst-align @gol
669 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
670 -melf -maout -melinux -mlinux -sim -sim2}
672 @emph{PDP-11 Options}
674 -mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
675 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
676 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
677 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
678 -mbranch-expensive -mbranch-cheap @gol
679 -msplit -mno-split -munix-asm -mdec-asm}
681 @emph{Xstormy16 Options}
685 @emph{Xtensa Options}
687 -mbig-endian -mlittle-endian @gol
688 -mdensity -mno-density @gol
689 -mmac16 -mno-mac16 @gol
690 -mmul16 -mno-mul16 @gol
691 -mmul32 -mno-mul32 @gol
693 -mminmax -mno-minmax @gol
694 -msext -mno-sext @gol
695 -mbooleans -mno-booleans @gol
696 -mhard-float -msoft-float @gol
697 -mfused-madd -mno-fused-madd @gol
698 -mserialize-volatile -mno-serialize-volatile @gol
699 -mtext-section-literals -mno-text-section-literals @gol
700 -mtarget-align -mno-target-align @gol
701 -mlongcalls -mno-longcalls}
705 -mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 -mhard-float -msoft-float @gol
706 -malloc-cc -mfixed-cc -mdword -mno-dword -mdouble -mno-double @gol
707 -mmedia -mno-media -mmuladd -mno-muladd -mlibrary-pic -macc-4 @gol
708 -macc-8 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
709 -mscc -mno-scc -mcond-exec -mno-cond-exec -mvliw-branch -mno-vliw-branch @gol
710 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
711 -mno-nested-cond-exec -mtomcat-stats @gol
716 @item Code Generation Options
717 @xref{Code Gen Options,,Options for Code Generation Conventions}.
719 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
720 -ffixed-@var{reg} -fexceptions @gol
721 -fnon-call-exceptions -funwind-tables @gol
722 -fasynchronous-unwind-tables @gol
723 -finhibit-size-directive -finstrument-functions @gol
724 -fno-common -fno-ident -fno-gnu-linker @gol
725 -fpcc-struct-return -fpic -fPIC @gol
726 -freg-struct-return -fshared-data -fshort-enums @gol
727 -fshort-double -fshort-wchar -fvolatile @gol
728 -fvolatile-global -fvolatile-static @gol
729 -fverbose-asm -fpack-struct -fstack-check @gol
730 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
731 -fargument-alias -fargument-noalias @gol
732 -fargument-noalias-global -fleading-underscore @gol
733 -ftls-model=@var{model}}
737 * Overall Options:: Controlling the kind of output:
738 an executable, object files, assembler files,
739 or preprocessed source.
740 * C Dialect Options:: Controlling the variant of C language compiled.
741 * C++ Dialect Options:: Variations on C++.
742 * Objective-C Dialect Options:: Variations on Objective-C.
743 * Language Independent Options:: Controlling how diagnostics should be
745 * Warning Options:: How picky should the compiler be?
746 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
747 * Optimize Options:: How much optimization?
748 * Preprocessor Options:: Controlling header files and macro definitions.
749 Also, getting dependency information for Make.
750 * Assembler Options:: Passing options to the assembler.
751 * Link Options:: Specifying libraries and so on.
752 * Directory Options:: Where to find header files and libraries.
753 Where to find the compiler executable files.
754 * Spec Files:: How to pass switches to sub-processes.
755 * Target Options:: Running a cross-compiler, or an old version of GCC.
758 @node Overall Options
759 @section Options Controlling the Kind of Output
761 Compilation can involve up to four stages: preprocessing, compilation
762 proper, assembly and linking, always in that order. The first three
763 stages apply to an individual source file, and end by producing an
764 object file; linking combines all the object files (those newly
765 compiled, and those specified as input) into an executable file.
767 @cindex file name suffix
768 For any given input file, the file name suffix determines what kind of
773 C source code which must be preprocessed.
776 C source code which should not be preprocessed.
779 C++ source code which should not be preprocessed.
782 Objective-C source code. Note that you must link with the library
783 @file{libobjc.a} to make an Objective-C program work.
786 Objective-C source code which should not be preprocessed.
789 C header file (not to be compiled or linked).
793 @itemx @var{file}.cxx
794 @itemx @var{file}.cpp
795 @itemx @var{file}.c++
797 C++ source code which must be preprocessed. Note that in @samp{.cxx},
798 the last two letters must both be literally @samp{x}. Likewise,
799 @samp{.C} refers to a literal capital C@.
802 @itemx @var{file}.for
803 @itemx @var{file}.FOR
804 Fortran source code which should not be preprocessed.
807 @itemx @var{file}.fpp
808 @itemx @var{file}.FPP
809 Fortran source code which must be preprocessed (with the traditional
813 Fortran source code which must be preprocessed with a RATFOR
814 preprocessor (not included with GCC)@.
816 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
817 Using and Porting GNU Fortran}, for more details of the handling of
820 @c FIXME: Descriptions of Java file types.
827 Ada source code file which contains a library unit declaration (a
828 declaration of a package, subprogram, or generic, or a generic
829 instantiation), or a library unit renaming declaration (a package,
830 generic, or subprogram renaming declaration). Such files are also
833 @itemx @var{file}.adb
834 Ada source code file containing a library unit body (a subprogram or
835 package body). Such files are also called @dfn{bodies}.
837 @c GCC also knows about some suffixes for languages not yet included:
846 Assembler code which must be preprocessed.
849 An object file to be fed straight into linking.
850 Any file name with no recognized suffix is treated this way.
854 You can specify the input language explicitly with the @option{-x} option:
857 @item -x @var{language}
858 Specify explicitly the @var{language} for the following input files
859 (rather than letting the compiler choose a default based on the file
860 name suffix). This option applies to all following input files until
861 the next @option{-x} option. Possible values for @var{language} are:
863 c c-header cpp-output
865 objective-c objc-cpp-output
866 assembler assembler-with-cpp
868 f77 f77-cpp-input ratfor
874 Turn off any specification of a language, so that subsequent files are
875 handled according to their file name suffixes (as they are if @option{-x}
876 has not been used at all).
878 @item -pass-exit-codes
879 @opindex pass-exit-codes
880 Normally the @command{gcc} program will exit with the code of 1 if any
881 phase of the compiler returns a non-success return code. If you specify
882 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
883 numerically highest error produced by any phase that returned an error
887 If you only want some of the stages of compilation, you can use
888 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
889 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
890 @command{gcc} is to stop. Note that some combinations (for example,
891 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
896 Compile or assemble the source files, but do not link. The linking
897 stage simply is not done. The ultimate output is in the form of an
898 object file for each source file.
900 By default, the object file name for a source file is made by replacing
901 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
903 Unrecognized input files, not requiring compilation or assembly, are
908 Stop after the stage of compilation proper; do not assemble. The output
909 is in the form of an assembler code file for each non-assembler input
912 By default, the assembler file name for a source file is made by
913 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
915 Input files that don't require compilation are ignored.
919 Stop after the preprocessing stage; do not run the compiler proper. The
920 output is in the form of preprocessed source code, which is sent to the
923 Input files which don't require preprocessing are ignored.
925 @cindex output file option
928 Place output in file @var{file}. This applies regardless to whatever
929 sort of output is being produced, whether it be an executable file,
930 an object file, an assembler file or preprocessed C code.
932 Since only one output file can be specified, it does not make sense to
933 use @option{-o} when compiling more than one input file, unless you are
934 producing an executable file as output.
936 If @option{-o} is not specified, the default is to put an executable file
937 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
938 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
939 all preprocessed C source on standard output.
943 Print (on standard error output) the commands executed to run the stages
944 of compilation. Also print the version number of the compiler driver
945 program and of the preprocessor and the compiler proper.
949 Like @option{-v} except the commands are not executed and all command
950 arguments are quoted. This is useful for shell scripts to capture the
951 driver-generated command lines.
955 Use pipes rather than temporary files for communication between the
956 various stages of compilation. This fails to work on some systems where
957 the assembler is unable to read from a pipe; but the GNU assembler has
962 Print (on the standard output) a description of the command line options
963 understood by @command{gcc}. If the @option{-v} option is also specified
964 then @option{--help} will also be passed on to the various processes
965 invoked by @command{gcc}, so that they can display the command line options
966 they accept. If the @option{-W} option is also specified then command
967 line options which have no documentation associated with them will also
972 Print (on the standard output) a description of target specific command
973 line options for each tool.
977 Display the version number and copyrights of the invoked GCC.
981 @section Compiling C++ Programs
983 @cindex suffixes for C++ source
984 @cindex C++ source file suffixes
985 C++ source files conventionally use one of the suffixes @samp{.C},
986 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
987 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
988 files with these names and compiles them as C++ programs even if you
989 call the compiler the same way as for compiling C programs (usually with
990 the name @command{gcc}).
994 However, C++ programs often require class libraries as well as a
995 compiler that understands the C++ language---and under some
996 circumstances, you might want to compile programs from standard input,
997 or otherwise without a suffix that flags them as C++ programs.
998 @command{g++} is a program that calls GCC with the default language
999 set to C++, and automatically specifies linking against the C++
1000 library. On many systems, @command{g++} is also
1001 installed with the name @command{c++}.
1003 @cindex invoking @command{g++}
1004 When you compile C++ programs, you may specify many of the same
1005 command-line options that you use for compiling programs in any
1006 language; or command-line options meaningful for C and related
1007 languages; or options that are meaningful only for C++ programs.
1008 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1009 explanations of options for languages related to C@.
1010 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1011 explanations of options that are meaningful only for C++ programs.
1013 @node C Dialect Options
1014 @section Options Controlling C Dialect
1015 @cindex dialect options
1016 @cindex language dialect options
1017 @cindex options, dialect
1019 The following options control the dialect of C (or languages derived
1020 from C, such as C++ and Objective-C) that the compiler accepts:
1023 @cindex ANSI support
1027 In C mode, support all ISO C90 programs. In C++ mode,
1028 remove GNU extensions that conflict with ISO C++.
1030 This turns off certain features of GCC that are incompatible with ISO
1031 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1032 such as the @code{asm} and @code{typeof} keywords, and
1033 predefined macros such as @code{unix} and @code{vax} that identify the
1034 type of system you are using. It also enables the undesirable and
1035 rarely used ISO trigraph feature. For the C compiler,
1036 it disables recognition of C++ style @samp{//} comments as well as
1037 the @code{inline} keyword.
1039 The alternate keywords @code{__asm__}, @code{__extension__},
1040 @code{__inline__} and @code{__typeof__} continue to work despite
1041 @option{-ansi}. You would not want to use them in an ISO C program, of
1042 course, but it is useful to put them in header files that might be included
1043 in compilations done with @option{-ansi}. Alternate predefined macros
1044 such as @code{__unix__} and @code{__vax__} are also available, with or
1045 without @option{-ansi}.
1047 The @option{-ansi} option does not cause non-ISO programs to be
1048 rejected gratuitously. For that, @option{-pedantic} is required in
1049 addition to @option{-ansi}. @xref{Warning Options}.
1051 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1052 option is used. Some header files may notice this macro and refrain
1053 from declaring certain functions or defining certain macros that the
1054 ISO standard doesn't call for; this is to avoid interfering with any
1055 programs that might use these names for other things.
1057 Functions which would normally be built in but do not have semantics
1058 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1059 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1060 built-in functions provided by GCC}, for details of the functions
1065 Determine the language standard. This option is currently only
1066 supported when compiling C or C++. A value for this option must be
1067 provided; possible values are
1072 ISO C90 (same as @option{-ansi}).
1074 @item iso9899:199409
1075 ISO C90 as modified in amendment 1.
1081 ISO C99. Note that this standard is not yet fully supported; see
1082 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1083 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1086 Default, ISO C90 plus GNU extensions (including some C99 features).
1090 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1091 this will become the default. The name @samp{gnu9x} is deprecated.
1094 The 1998 ISO C++ standard plus amendments.
1097 The same as @option{-std=c++98} plus GNU extensions. This is the
1098 default for C++ code.
1101 Even when this option is not specified, you can still use some of the
1102 features of newer standards in so far as they do not conflict with
1103 previous C standards. For example, you may use @code{__restrict__} even
1104 when @option{-std=c99} is not specified.
1106 The @option{-std} options specifying some version of ISO C have the same
1107 effects as @option{-ansi}, except that features that were not in ISO C90
1108 but are in the specified version (for example, @samp{//} comments and
1109 the @code{inline} keyword in ISO C99) are not disabled.
1111 @xref{Standards,,Language Standards Supported by GCC}, for details of
1112 these standard versions.
1114 @item -aux-info @var{filename}
1116 Output to the given filename prototyped declarations for all functions
1117 declared and/or defined in a translation unit, including those in header
1118 files. This option is silently ignored in any language other than C@.
1120 Besides declarations, the file indicates, in comments, the origin of
1121 each declaration (source file and line), whether the declaration was
1122 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1123 @samp{O} for old, respectively, in the first character after the line
1124 number and the colon), and whether it came from a declaration or a
1125 definition (@samp{C} or @samp{F}, respectively, in the following
1126 character). In the case of function definitions, a K&R-style list of
1127 arguments followed by their declarations is also provided, inside
1128 comments, after the declaration.
1132 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1133 keyword, so that code can use these words as identifiers. You can use
1134 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1135 instead. @option{-ansi} implies @option{-fno-asm}.
1137 In C++, this switch only affects the @code{typeof} keyword, since
1138 @code{asm} and @code{inline} are standard keywords. You may want to
1139 use the @option{-fno-gnu-keywords} flag instead, which has the same
1140 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1141 switch only affects the @code{asm} and @code{typeof} keywords, since
1142 @code{inline} is a standard keyword in ISO C99.
1145 @itemx -fno-builtin-@var{function}
1146 @opindex fno-builtin
1147 @cindex built-in functions
1148 Don't recognize built-in functions that do not begin with
1149 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1150 functions provided by GCC}, for details of the functions affected,
1151 including those which are not built-in functions when @option{-ansi} or
1152 @option{-std} options for strict ISO C conformance are used because they
1153 do not have an ISO standard meaning.
1155 GCC normally generates special code to handle certain built-in functions
1156 more efficiently; for instance, calls to @code{alloca} may become single
1157 instructions that adjust the stack directly, and calls to @code{memcpy}
1158 may become inline copy loops. The resulting code is often both smaller
1159 and faster, but since the function calls no longer appear as such, you
1160 cannot set a breakpoint on those calls, nor can you change the behavior
1161 of the functions by linking with a different library.
1163 With the @option{-fno-builtin-@var{function}} option
1164 only the built-in function @var{function} is
1165 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1166 function is named this is not built-in in this version of GCC, this
1167 option is ignored. There is no corresponding
1168 @option{-fbuiltin-@var{function}} option; if you wish to enable
1169 built-in functions selectively when using @option{-fno-builtin} or
1170 @option{-ffreestanding}, you may define macros such as:
1173 #define abs(n) __builtin_abs ((n))
1174 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1179 @cindex hosted environment
1181 Assert that compilation takes place in a hosted environment. This implies
1182 @option{-fbuiltin}. A hosted environment is one in which the
1183 entire standard library is available, and in which @code{main} has a return
1184 type of @code{int}. Examples are nearly everything except a kernel.
1185 This is equivalent to @option{-fno-freestanding}.
1187 @item -ffreestanding
1188 @opindex ffreestanding
1189 @cindex hosted environment
1191 Assert that compilation takes place in a freestanding environment. This
1192 implies @option{-fno-builtin}. A freestanding environment
1193 is one in which the standard library may not exist, and program startup may
1194 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1195 This is equivalent to @option{-fno-hosted}.
1197 @xref{Standards,,Language Standards Supported by GCC}, for details of
1198 freestanding and hosted environments.
1200 @item -fms-extensions
1201 @opindex fms-extensions
1202 Accept some non-standard constructs used in Microsoft header files.
1206 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1207 options for strict ISO C conformance) implies @option{-trigraphs}.
1209 @cindex traditional C language
1210 @cindex C language, traditional
1212 @itemx -traditional-cpp
1213 @opindex traditional-cpp
1214 @opindex traditional
1215 Formerly, these options caused GCC to attempt to emulate a pre-standard
1216 C compiler. They are now only supported with the @option{-E} switch.
1217 The preprocessor continues to support a pre-standard mode. See the GNU
1218 CPP manual for details.
1220 @item -fcond-mismatch
1221 @opindex fcond-mismatch
1222 Allow conditional expressions with mismatched types in the second and
1223 third arguments. The value of such an expression is void. This option
1224 is not supported for C++.
1226 @item -funsigned-char
1227 @opindex funsigned-char
1228 Let the type @code{char} be unsigned, like @code{unsigned char}.
1230 Each kind of machine has a default for what @code{char} should
1231 be. It is either like @code{unsigned char} by default or like
1232 @code{signed char} by default.
1234 Ideally, a portable program should always use @code{signed char} or
1235 @code{unsigned char} when it depends on the signedness of an object.
1236 But many programs have been written to use plain @code{char} and
1237 expect it to be signed, or expect it to be unsigned, depending on the
1238 machines they were written for. This option, and its inverse, let you
1239 make such a program work with the opposite default.
1241 The type @code{char} is always a distinct type from each of
1242 @code{signed char} or @code{unsigned char}, even though its behavior
1243 is always just like one of those two.
1246 @opindex fsigned-char
1247 Let the type @code{char} be signed, like @code{signed char}.
1249 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1250 the negative form of @option{-funsigned-char}. Likewise, the option
1251 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1253 @item -fsigned-bitfields
1254 @itemx -funsigned-bitfields
1255 @itemx -fno-signed-bitfields
1256 @itemx -fno-unsigned-bitfields
1257 @opindex fsigned-bitfields
1258 @opindex funsigned-bitfields
1259 @opindex fno-signed-bitfields
1260 @opindex fno-unsigned-bitfields
1261 These options control whether a bit-field is signed or unsigned, when the
1262 declaration does not use either @code{signed} or @code{unsigned}. By
1263 default, such a bit-field is signed, because this is consistent: the
1264 basic integer types such as @code{int} are signed types.
1266 @item -fwritable-strings
1267 @opindex fwritable-strings
1268 Store string constants in the writable data segment and don't uniquize
1269 them. This is for compatibility with old programs which assume they can
1270 write into string constants.
1272 Writing into string constants is a very bad idea; ``constants'' should
1276 @node C++ Dialect Options
1277 @section Options Controlling C++ Dialect
1279 @cindex compiler options, C++
1280 @cindex C++ options, command line
1281 @cindex options, C++
1282 This section describes the command-line options that are only meaningful
1283 for C++ programs; but you can also use most of the GNU compiler options
1284 regardless of what language your program is in. For example, you
1285 might compile a file @code{firstClass.C} like this:
1288 g++ -g -frepo -O -c firstClass.C
1292 In this example, only @option{-frepo} is an option meant
1293 only for C++ programs; you can use the other options with any
1294 language supported by GCC@.
1296 Here is a list of options that are @emph{only} for compiling C++ programs:
1300 @item -fabi-version=@var{n}
1301 @opindex fabi-version
1302 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1303 ABI that first appeared in G++ 3.2. Version 0 will always be the
1304 version that conforms most closely to the C++ ABI specification.
1305 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1308 The default is version 1.
1310 @item -fno-access-control
1311 @opindex fno-access-control
1312 Turn off all access checking. This switch is mainly useful for working
1313 around bugs in the access control code.
1317 Check that the pointer returned by @code{operator new} is non-null
1318 before attempting to modify the storage allocated. The current Working
1319 Paper requires that @code{operator new} never return a null pointer, so
1320 this check is normally unnecessary.
1322 An alternative to using this option is to specify that your
1323 @code{operator new} does not throw any exceptions; if you declare it
1324 @samp{throw()}, G++ will check the return value. See also @samp{new
1327 @item -fconserve-space
1328 @opindex fconserve-space
1329 Put uninitialized or runtime-initialized global variables into the
1330 common segment, as C does. This saves space in the executable at the
1331 cost of not diagnosing duplicate definitions. If you compile with this
1332 flag and your program mysteriously crashes after @code{main()} has
1333 completed, you may have an object that is being destroyed twice because
1334 two definitions were merged.
1336 This option is no longer useful on most targets, now that support has
1337 been added for putting variables into BSS without making them common.
1339 @item -fno-const-strings
1340 @opindex fno-const-strings
1341 Give string constants type @code{char *} instead of type @code{const
1342 char *}. By default, G++ uses type @code{const char *} as required by
1343 the standard. Even if you use @option{-fno-const-strings}, you cannot
1344 actually modify the value of a string constant, unless you also use
1345 @option{-fwritable-strings}.
1347 This option might be removed in a future release of G++. For maximum
1348 portability, you should structure your code so that it works with
1349 string constants that have type @code{const char *}.
1351 @item -fdollars-in-identifiers
1352 @opindex fdollars-in-identifiers
1353 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1354 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1355 @samp{$} by default on most target systems, but there are a few exceptions.)
1356 Traditional C allowed the character @samp{$} to form part of
1357 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1359 @item -fno-elide-constructors
1360 @opindex fno-elide-constructors
1361 The C++ standard allows an implementation to omit creating a temporary
1362 which is only used to initialize another object of the same type.
1363 Specifying this option disables that optimization, and forces G++ to
1364 call the copy constructor in all cases.
1366 @item -fno-enforce-eh-specs
1367 @opindex fno-enforce-eh-specs
1368 Don't check for violation of exception specifications at runtime. This
1369 option violates the C++ standard, but may be useful for reducing code
1370 size in production builds, much like defining @samp{NDEBUG}. The compiler
1371 will still optimize based on the exception specifications.
1373 @item -fexternal-templates
1374 @opindex fexternal-templates
1376 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1377 template instantiation; template instances are emitted or not according
1378 to the location of the template definition. @xref{Template
1379 Instantiation}, for more information.
1381 This option is deprecated.
1383 @item -falt-external-templates
1384 @opindex falt-external-templates
1385 Similar to @option{-fexternal-templates}, but template instances are
1386 emitted or not according to the place where they are first instantiated.
1387 @xref{Template Instantiation}, for more information.
1389 This option is deprecated.
1392 @itemx -fno-for-scope
1394 @opindex fno-for-scope
1395 If @option{-ffor-scope} is specified, the scope of variables declared in
1396 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1397 as specified by the C++ standard.
1398 If @option{-fno-for-scope} is specified, the scope of variables declared in
1399 a @i{for-init-statement} extends to the end of the enclosing scope,
1400 as was the case in old versions of G++, and other (traditional)
1401 implementations of C++.
1403 The default if neither flag is given to follow the standard,
1404 but to allow and give a warning for old-style code that would
1405 otherwise be invalid, or have different behavior.
1407 @item -fno-gnu-keywords
1408 @opindex fno-gnu-keywords
1409 Do not recognize @code{typeof} as a keyword, so that code can use this
1410 word as an identifier. You can use the keyword @code{__typeof__} instead.
1411 @option{-ansi} implies @option{-fno-gnu-keywords}.
1413 @item -fno-implicit-templates
1414 @opindex fno-implicit-templates
1415 Never emit code for non-inline templates which are instantiated
1416 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1417 @xref{Template Instantiation}, for more information.
1419 @item -fno-implicit-inline-templates
1420 @opindex fno-implicit-inline-templates
1421 Don't emit code for implicit instantiations of inline templates, either.
1422 The default is to handle inlines differently so that compiles with and
1423 without optimization will need the same set of explicit instantiations.
1425 @item -fno-implement-inlines
1426 @opindex fno-implement-inlines
1427 To save space, do not emit out-of-line copies of inline functions
1428 controlled by @samp{#pragma implementation}. This will cause linker
1429 errors if these functions are not inlined everywhere they are called.
1431 @item -fms-extensions
1432 @opindex fms-extensions
1433 Disable pedantic warnings about constructs used in MFC, such as implicit
1434 int and getting a pointer to member function via non-standard syntax.
1436 @item -fno-nonansi-builtins
1437 @opindex fno-nonansi-builtins
1438 Disable built-in declarations of functions that are not mandated by
1439 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1440 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1442 @item -fno-operator-names
1443 @opindex fno-operator-names
1444 Do not treat the operator name keywords @code{and}, @code{bitand},
1445 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1446 synonyms as keywords.
1448 @item -fno-optional-diags
1449 @opindex fno-optional-diags
1450 Disable diagnostics that the standard says a compiler does not need to
1451 issue. Currently, the only such diagnostic issued by G++ is the one for
1452 a name having multiple meanings within a class.
1455 @opindex fpermissive
1456 Downgrade messages about nonconformant code from errors to warnings. By
1457 default, G++ effectively sets @option{-pedantic-errors} without
1458 @option{-pedantic}; this option reverses that. This behavior and this
1459 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1463 Enable automatic template instantiation at link time. This option also
1464 implies @option{-fno-implicit-templates}. @xref{Template
1465 Instantiation}, for more information.
1469 Disable generation of information about every class with virtual
1470 functions for use by the C++ runtime type identification features
1471 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1472 of the language, you can save some space by using this flag. Note that
1473 exception handling uses the same information, but it will generate it as
1478 Emit statistics about front-end processing at the end of the compilation.
1479 This information is generally only useful to the G++ development team.
1481 @item -ftemplate-depth-@var{n}
1482 @opindex ftemplate-depth
1483 Set the maximum instantiation depth for template classes to @var{n}.
1484 A limit on the template instantiation depth is needed to detect
1485 endless recursions during template class instantiation. ANSI/ISO C++
1486 conforming programs must not rely on a maximum depth greater than 17.
1488 @item -fuse-cxa-atexit
1489 @opindex fuse-cxa-atexit
1490 Register destructors for objects with static storage duration with the
1491 @code{__cxa_atexit} function rather than the @code{atexit} function.
1492 This option is required for fully standards-compliant handling of static
1493 destructors, but will only work if your C library supports
1494 @code{__cxa_atexit}.
1498 Emit special relocations for vtables and virtual function references
1499 so that the linker can identify unused virtual functions and zero out
1500 vtable slots that refer to them. This is most useful with
1501 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1502 also discard the functions themselves.
1504 This optimization requires GNU as and GNU ld. Not all systems support
1505 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1509 Do not use weak symbol support, even if it is provided by the linker.
1510 By default, G++ will use weak symbols if they are available. This
1511 option exists only for testing, and should not be used by end-users;
1512 it will result in inferior code and has no benefits. This option may
1513 be removed in a future release of G++.
1517 Do not search for header files in the standard directories specific to
1518 C++, but do still search the other standard directories. (This option
1519 is used when building the C++ library.)
1522 In addition, these optimization, warning, and code generation options
1523 have meanings only for C++ programs:
1526 @item -fno-default-inline
1527 @opindex fno-default-inline
1528 Do not assume @samp{inline} for functions defined inside a class scope.
1529 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1530 functions will have linkage like inline functions; they just won't be
1533 @item -Wabi @r{(C++ only)}
1535 Warn when G++ generates code that is probably not compatible with the
1536 vendor-neutral C++ ABI. Although an effort has been made to warn about
1537 all such cases, there are probably some cases that are not warned about,
1538 even though G++ is generating incompatible code. There may also be
1539 cases where warnings are emitted even though the code that is generated
1542 You should rewrite your code to avoid these warnings if you are
1543 concerned about the fact that code generated by G++ may not be binary
1544 compatible with code generated by other compilers.
1546 The known incompatibilites at this point include:
1551 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1552 pack data into the same byte as a base class. For example:
1555 struct A @{ virtual void f(); int f1 : 1; @};
1556 struct B : public A @{ int f2 : 1; @};
1560 In this case, G++ will place @code{B::f2} into the same byte
1561 as@code{A::f1}; other compilers will not. You can avoid this problem
1562 by explicitly padding @code{A} so that its size is a multiple of the
1563 byte size on your platform; that will cause G++ and other compilers to
1564 layout @code{B} identically.
1567 Incorrect handling of tail-padding for virtual bases. G++ does not use
1568 tail padding when laying out virtual bases. For example:
1571 struct A @{ virtual void f(); char c1; @};
1572 struct B @{ B(); char c2; @};
1573 struct C : public A, public virtual B @{@};
1577 In this case, G++ will not place @code{B} into the tail-padding for
1578 @code{A}; other compilers will. You can avoid this problem by
1579 explicitly padding @code{A} so that its size is a multiple of its
1580 alignment (ignoring virtual base classes); that will cause G++ and other
1581 compilers to layout @code{C} identically.
1584 Incorrect handling of bit-fields with declared widths greater than that
1585 of their underlying types, when the bit-fields appear in a union. For
1589 union U @{ int i : 4096; @};
1593 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1594 union too small by the number of bits in an @code{int}.
1597 Empty classes can be placed at incorrect offsets. For example:
1607 struct C : public B, public A @{@};
1611 G++ will place the @code{A} base class of @code{C} at a non-zero offset;
1612 it should be placed at offset zero. G++ mistakenly believes that the
1613 @code{A} data member of @code{B} is already at offset zero.
1616 Names of template functions whose types involve @code{typename} or
1617 template template parameters can be mangled incorrectly.
1620 template <typename Q>
1621 void f(typename Q::X) @{@}
1623 template <template <typename> class Q>
1624 void f(typename Q<int>::X) @{@}
1628 Instantions of these templates may be mangled incorrectly.
1632 @item -Wctor-dtor-privacy @r{(C++ only)}
1633 @opindex Wctor-dtor-privacy
1634 Warn when a class seems unusable, because all the constructors or
1635 destructors in a class are private and the class has no friends or
1636 public static member functions. This warning is enabled by default.
1638 @item -Wnon-virtual-dtor @r{(C++ only)}
1639 @opindex Wnon-virtual-dtor
1640 Warn when a class declares a non-virtual destructor that should probably
1641 be virtual, because it looks like the class will be used polymorphically.
1642 This warning is enabled by @option{-Wall}.
1644 @item -Wreorder @r{(C++ only)}
1646 @cindex reordering, warning
1647 @cindex warning for reordering of member initializers
1648 Warn when the order of member initializers given in the code does not
1649 match the order in which they must be executed. For instance:
1655 A(): j (0), i (1) @{ @}
1659 Here the compiler will warn that the member initializers for @samp{i}
1660 and @samp{j} will be rearranged to match the declaration order of the
1661 members. This warning is enabled by @option{-Wall}.
1664 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1667 @item -Weffc++ @r{(C++ only)}
1669 Warn about violations of the following style guidelines from Scott Meyers'
1670 @cite{Effective C++} book:
1674 Item 11: Define a copy constructor and an assignment operator for classes
1675 with dynamically allocated memory.
1678 Item 12: Prefer initialization to assignment in constructors.
1681 Item 14: Make destructors virtual in base classes.
1684 Item 15: Have @code{operator=} return a reference to @code{*this}.
1687 Item 23: Don't try to return a reference when you must return an object.
1691 and about violations of the following style guidelines from Scott Meyers'
1692 @cite{More Effective C++} book:
1696 Item 6: Distinguish between prefix and postfix forms of increment and
1697 decrement operators.
1700 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1704 If you use this option, you should be aware that the standard library
1705 headers do not obey all of these guidelines; you can use @samp{grep -v}
1706 to filter out those warnings.
1708 @item -Wno-deprecated @r{(C++ only)}
1709 @opindex Wno-deprecated
1710 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1712 @item -Wno-non-template-friend @r{(C++ only)}
1713 @opindex Wno-non-template-friend
1714 Disable warnings when non-templatized friend functions are declared
1715 within a template. With the advent of explicit template specification
1716 support in G++, if the name of the friend is an unqualified-id (i.e.,
1717 @samp{friend foo(int)}), the C++ language specification demands that the
1718 friend declare or define an ordinary, nontemplate function. (Section
1719 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1720 could be interpreted as a particular specialization of a templatized
1721 function. Because this non-conforming behavior is no longer the default
1722 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1723 check existing code for potential trouble spots, and is on by default.
1724 This new compiler behavior can be turned off with
1725 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1726 but disables the helpful warning.
1728 @item -Wold-style-cast @r{(C++ only)}
1729 @opindex Wold-style-cast
1730 Warn if an old-style (C-style) cast to a non-void type is used within
1731 a C++ program. The new-style casts (@samp{static_cast},
1732 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1733 unintended effects, and much easier to grep for.
1735 @item -Woverloaded-virtual @r{(C++ only)}
1736 @opindex Woverloaded-virtual
1737 @cindex overloaded virtual fn, warning
1738 @cindex warning for overloaded virtual fn
1739 Warn when a function declaration hides virtual functions from a
1740 base class. For example, in:
1747 struct B: public A @{
1752 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1760 will fail to compile.
1762 @item -Wno-pmf-conversions @r{(C++ only)}
1763 @opindex Wno-pmf-conversions
1764 Disable the diagnostic for converting a bound pointer to member function
1767 @item -Wsign-promo @r{(C++ only)}
1768 @opindex Wsign-promo
1769 Warn when overload resolution chooses a promotion from unsigned or
1770 enumeral type to a signed type over a conversion to an unsigned type of
1771 the same size. Previous versions of G++ would try to preserve
1772 unsignedness, but the standard mandates the current behavior.
1774 @item -Wsynth @r{(C++ only)}
1776 @cindex warning for synthesized methods
1777 @cindex synthesized methods, warning
1778 Warn when G++'s synthesis behavior does not match that of cfront. For
1784 A& operator = (int);
1794 In this example, G++ will synthesize a default @samp{A& operator =
1795 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1798 @node Objective-C Dialect Options
1799 @section Options Controlling Objective-C Dialect
1801 @cindex compiler options, Objective-C
1802 @cindex Objective-C options, command line
1803 @cindex options, Objective-C
1804 This section describes the command-line options that are only meaningful
1805 for Objective-C programs; but you can also use most of the GNU compiler
1806 options regardless of what language your program is in. For example,
1807 you might compile a file @code{some_class.m} like this:
1810 gcc -g -fgnu-runtime -O -c some_class.m
1814 In this example, only @option{-fgnu-runtime} is an option meant only for
1815 Objective-C programs; you can use the other options with any language
1818 Here is a list of options that are @emph{only} for compiling Objective-C
1822 @item -fconstant-string-class=@var{class-name}
1823 @opindex fconstant-string-class
1824 Use @var{class-name} as the name of the class to instantiate for each
1825 literal string specified with the syntax @code{@@"@dots{}"}. The default
1826 class name is @code{NXConstantString}.
1829 @opindex fgnu-runtime
1830 Generate object code compatible with the standard GNU Objective-C
1831 runtime. This is the default for most types of systems.
1833 @item -fnext-runtime
1834 @opindex fnext-runtime
1835 Generate output compatible with the NeXT runtime. This is the default
1836 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1837 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1842 Dump interface declarations for all classes seen in the source file to a
1843 file named @file{@var{sourcename}.decl}.
1846 @opindex Wno-protocol
1847 If a class is declared to implement a protocol, a warning is issued for
1848 every method in the protocol that is not implemented by the class. The
1849 default behavior is to issue a warning for every method not explicitly
1850 implemented in the class, even if a method implementation is inherited
1851 from the superclass. If you use the @code{-Wno-protocol} option, then
1852 methods inherited from the superclass are considered to be implemented,
1853 and no warning is issued for them.
1857 Warn if multiple methods of different types for the same selector are
1858 found during compilation. The check is performed on the list of methods
1859 in the final stage of compilation. Additionally, a check is performed
1860 that for each selector appearing in a @code{@@selector(@dots{})}
1861 expression, a corresponding method with that selector has been found
1862 during compilation. Because these checks scan the method table only at
1863 the end of compilation, these warnings are not produced if the final
1864 stage of compilation is not reached, for example because an error is
1865 found during compilation, or because the @code{-fsyntax-only} option is
1868 @item -Wundeclared-selector
1869 @opindex Wundeclared-selector
1870 Warn if a @code{@@selector(@dots{})} expression referring to an
1871 undeclared selector is found. A selector is considered undeclared if no
1872 method with that name has been declared (explicitly, in an
1873 @code{@@interface} or @code{@@protocol} declaration, or implicitly, in
1874 an @code{@@implementation} section) before the
1875 @code{@@selector(@dots{})} expression. This option always performs its
1876 checks as soon as a @code{@@selector(@dots{})} expression is found
1877 (while @code{-Wselector} only performs its checks in the final stage of
1878 compilation), and so additionally enforces the coding style convention
1879 that methods and selectors must be declared before being used.
1881 @c not documented because only avail via -Wp
1882 @c @item -print-objc-runtime-info
1886 @node Language Independent Options
1887 @section Options to Control Diagnostic Messages Formatting
1888 @cindex options to control diagnostics formatting
1889 @cindex diagnostic messages
1890 @cindex message formatting
1892 Traditionally, diagnostic messages have been formatted irrespective of
1893 the output device's aspect (e.g.@: its width, @dots{}). The options described
1894 below can be used to control the diagnostic messages formatting
1895 algorithm, e.g.@: how many characters per line, how often source location
1896 information should be reported. Right now, only the C++ front end can
1897 honor these options. However it is expected, in the near future, that
1898 the remaining front ends would be able to digest them correctly.
1901 @item -fmessage-length=@var{n}
1902 @opindex fmessage-length
1903 Try to format error messages so that they fit on lines of about @var{n}
1904 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1905 the front ends supported by GCC@. If @var{n} is zero, then no
1906 line-wrapping will be done; each error message will appear on a single
1909 @opindex fdiagnostics-show-location
1910 @item -fdiagnostics-show-location=once
1911 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1912 reporter to emit @emph{once} source location information; that is, in
1913 case the message is too long to fit on a single physical line and has to
1914 be wrapped, the source location won't be emitted (as prefix) again,
1915 over and over, in subsequent continuation lines. This is the default
1918 @item -fdiagnostics-show-location=every-line
1919 Only meaningful in line-wrapping mode. Instructs the diagnostic
1920 messages reporter to emit the same source location information (as
1921 prefix) for physical lines that result from the process of breaking
1922 a message which is too long to fit on a single line.
1926 @node Warning Options
1927 @section Options to Request or Suppress Warnings
1928 @cindex options to control warnings
1929 @cindex warning messages
1930 @cindex messages, warning
1931 @cindex suppressing warnings
1933 Warnings are diagnostic messages that report constructions which
1934 are not inherently erroneous but which are risky or suggest there
1935 may have been an error.
1937 You can request many specific warnings with options beginning @samp{-W},
1938 for example @option{-Wimplicit} to request warnings on implicit
1939 declarations. Each of these specific warning options also has a
1940 negative form beginning @samp{-Wno-} to turn off warnings;
1941 for example, @option{-Wno-implicit}. This manual lists only one of the
1942 two forms, whichever is not the default.
1944 The following options control the amount and kinds of warnings produced
1945 by GCC; for further, language-specific options also refer to
1946 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1949 @cindex syntax checking
1951 @opindex fsyntax-only
1952 Check the code for syntax errors, but don't do anything beyond that.
1956 Issue all the warnings demanded by strict ISO C and ISO C++;
1957 reject all programs that use forbidden extensions, and some other
1958 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1959 version of the ISO C standard specified by any @option{-std} option used.
1961 Valid ISO C and ISO C++ programs should compile properly with or without
1962 this option (though a rare few will require @option{-ansi} or a
1963 @option{-std} option specifying the required version of ISO C)@. However,
1964 without this option, certain GNU extensions and traditional C and C++
1965 features are supported as well. With this option, they are rejected.
1967 @option{-pedantic} does not cause warning messages for use of the
1968 alternate keywords whose names begin and end with @samp{__}. Pedantic
1969 warnings are also disabled in the expression that follows
1970 @code{__extension__}. However, only system header files should use
1971 these escape routes; application programs should avoid them.
1972 @xref{Alternate Keywords}.
1974 Some users try to use @option{-pedantic} to check programs for strict ISO
1975 C conformance. They soon find that it does not do quite what they want:
1976 it finds some non-ISO practices, but not all---only those for which
1977 ISO C @emph{requires} a diagnostic, and some others for which
1978 diagnostics have been added.
1980 A feature to report any failure to conform to ISO C might be useful in
1981 some instances, but would require considerable additional work and would
1982 be quite different from @option{-pedantic}. We don't have plans to
1983 support such a feature in the near future.
1985 Where the standard specified with @option{-std} represents a GNU
1986 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1987 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1988 extended dialect is based. Warnings from @option{-pedantic} are given
1989 where they are required by the base standard. (It would not make sense
1990 for such warnings to be given only for features not in the specified GNU
1991 C dialect, since by definition the GNU dialects of C include all
1992 features the compiler supports with the given option, and there would be
1993 nothing to warn about.)
1995 @item -pedantic-errors
1996 @opindex pedantic-errors
1997 Like @option{-pedantic}, except that errors are produced rather than
2002 Inhibit all warning messages.
2006 Inhibit warning messages about the use of @samp{#import}.
2008 @item -Wchar-subscripts
2009 @opindex Wchar-subscripts
2010 Warn if an array subscript has type @code{char}. This is a common cause
2011 of error, as programmers often forget that this type is signed on some
2016 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2017 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2021 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2022 the arguments supplied have types appropriate to the format string
2023 specified, and that the conversions specified in the format string make
2024 sense. This includes standard functions, and others specified by format
2025 attributes (@pxref{Function Attributes}), in the @code{printf},
2026 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2027 not in the C standard) families.
2029 The formats are checked against the format features supported by GNU
2030 libc version 2.2. These include all ISO C90 and C99 features, as well
2031 as features from the Single Unix Specification and some BSD and GNU
2032 extensions. Other library implementations may not support all these
2033 features; GCC does not support warning about features that go beyond a
2034 particular library's limitations. However, if @option{-pedantic} is used
2035 with @option{-Wformat}, warnings will be given about format features not
2036 in the selected standard version (but not for @code{strfmon} formats,
2037 since those are not in any version of the C standard). @xref{C Dialect
2038 Options,,Options Controlling C Dialect}.
2040 Since @option{-Wformat} also checks for null format arguments for
2041 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2043 @option{-Wformat} is included in @option{-Wall}. For more control over some
2044 aspects of format checking, the options @option{-Wno-format-y2k},
2045 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2046 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2047 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2049 @item -Wno-format-y2k
2050 @opindex Wno-format-y2k
2051 If @option{-Wformat} is specified, do not warn about @code{strftime}
2052 formats which may yield only a two-digit year.
2054 @item -Wno-format-extra-args
2055 @opindex Wno-format-extra-args
2056 If @option{-Wformat} is specified, do not warn about excess arguments to a
2057 @code{printf} or @code{scanf} format function. The C standard specifies
2058 that such arguments are ignored.
2060 Where the unused arguments lie between used arguments that are
2061 specified with @samp{$} operand number specifications, normally
2062 warnings are still given, since the implementation could not know what
2063 type to pass to @code{va_arg} to skip the unused arguments. However,
2064 in the case of @code{scanf} formats, this option will suppress the
2065 warning if the unused arguments are all pointers, since the Single
2066 Unix Specification says that such unused arguments are allowed.
2068 @item -Wno-format-zero-length
2069 @opindex Wno-format-zero-length
2070 If @option{-Wformat} is specified, do not warn about zero-length formats.
2071 The C standard specifies that zero-length formats are allowed.
2073 @item -Wformat-nonliteral
2074 @opindex Wformat-nonliteral
2075 If @option{-Wformat} is specified, also warn if the format string is not a
2076 string literal and so cannot be checked, unless the format function
2077 takes its format arguments as a @code{va_list}.
2079 @item -Wformat-security
2080 @opindex Wformat-security
2081 If @option{-Wformat} is specified, also warn about uses of format
2082 functions that represent possible security problems. At present, this
2083 warns about calls to @code{printf} and @code{scanf} functions where the
2084 format string is not a string literal and there are no format arguments,
2085 as in @code{printf (foo);}. This may be a security hole if the format
2086 string came from untrusted input and contains @samp{%n}. (This is
2087 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2088 in future warnings may be added to @option{-Wformat-security} that are not
2089 included in @option{-Wformat-nonliteral}.)
2093 Enable @option{-Wformat} plus format checks not included in
2094 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2095 -Wformat-nonliteral -Wformat-security}.
2099 Enable warning about passing a null pointer for arguments marked as
2100 requiring a non-null value by the @code{nonnull} function attribute.
2102 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2103 can be disabled with the @option{-Wno-nonnull} option.
2105 @item -Wimplicit-int
2106 @opindex Wimplicit-int
2107 Warn when a declaration does not specify a type.
2109 @item -Wimplicit-function-declaration
2110 @itemx -Werror-implicit-function-declaration
2111 @opindex Wimplicit-function-declaration
2112 @opindex Werror-implicit-function-declaration
2113 Give a warning (or error) whenever a function is used before being
2118 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2122 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2123 function with external linkage, returning int, taking either zero
2124 arguments, two, or three arguments of appropriate types.
2126 @item -Wmissing-braces
2127 @opindex Wmissing-braces
2128 Warn if an aggregate or union initializer is not fully bracketed. In
2129 the following example, the initializer for @samp{a} is not fully
2130 bracketed, but that for @samp{b} is fully bracketed.
2133 int a[2][2] = @{ 0, 1, 2, 3 @};
2134 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2138 @opindex Wparentheses
2139 Warn if parentheses are omitted in certain contexts, such
2140 as when there is an assignment in a context where a truth value
2141 is expected, or when operators are nested whose precedence people
2142 often get confused about.
2144 Also warn about constructions where there may be confusion to which
2145 @code{if} statement an @code{else} branch belongs. Here is an example of
2160 In C, every @code{else} branch belongs to the innermost possible @code{if}
2161 statement, which in this example is @code{if (b)}. This is often not
2162 what the programmer expected, as illustrated in the above example by
2163 indentation the programmer chose. When there is the potential for this
2164 confusion, GCC will issue a warning when this flag is specified.
2165 To eliminate the warning, add explicit braces around the innermost
2166 @code{if} statement so there is no way the @code{else} could belong to
2167 the enclosing @code{if}. The resulting code would look like this:
2183 @item -Wsequence-point
2184 @opindex Wsequence-point
2185 Warn about code that may have undefined semantics because of violations
2186 of sequence point rules in the C standard.
2188 The C standard defines the order in which expressions in a C program are
2189 evaluated in terms of @dfn{sequence points}, which represent a partial
2190 ordering between the execution of parts of the program: those executed
2191 before the sequence point, and those executed after it. These occur
2192 after the evaluation of a full expression (one which is not part of a
2193 larger expression), after the evaluation of the first operand of a
2194 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2195 function is called (but after the evaluation of its arguments and the
2196 expression denoting the called function), and in certain other places.
2197 Other than as expressed by the sequence point rules, the order of
2198 evaluation of subexpressions of an expression is not specified. All
2199 these rules describe only a partial order rather than a total order,
2200 since, for example, if two functions are called within one expression
2201 with no sequence point between them, the order in which the functions
2202 are called is not specified. However, the standards committee have
2203 ruled that function calls do not overlap.
2205 It is not specified when between sequence points modifications to the
2206 values of objects take effect. Programs whose behavior depends on this
2207 have undefined behavior; the C standard specifies that ``Between the
2208 previous and next sequence point an object shall have its stored value
2209 modified at most once by the evaluation of an expression. Furthermore,
2210 the prior value shall be read only to determine the value to be
2211 stored.''. If a program breaks these rules, the results on any
2212 particular implementation are entirely unpredictable.
2214 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2215 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2216 diagnosed by this option, and it may give an occasional false positive
2217 result, but in general it has been found fairly effective at detecting
2218 this sort of problem in programs.
2220 The present implementation of this option only works for C programs. A
2221 future implementation may also work for C++ programs.
2223 The C standard is worded confusingly, therefore there is some debate
2224 over the precise meaning of the sequence point rules in subtle cases.
2225 Links to discussions of the problem, including proposed formal
2226 definitions, may be found on our readings page, at
2227 @w{@uref{http://gcc.gnu.org/readings.html}}.
2230 @opindex Wreturn-type
2231 Warn whenever a function is defined with a return-type that defaults to
2232 @code{int}. Also warn about any @code{return} statement with no
2233 return-value in a function whose return-type is not @code{void}.
2235 For C++, a function without return type always produces a diagnostic
2236 message, even when @option{-Wno-return-type} is specified. The only
2237 exceptions are @samp{main} and functions defined in system headers.
2241 Warn whenever a @code{switch} statement has an index of enumeral type
2242 and lacks a @code{case} for one or more of the named codes of that
2243 enumeration. (The presence of a @code{default} label prevents this
2244 warning.) @code{case} labels outside the enumeration range also
2245 provoke warnings when this option is used.
2247 @item -Wswitch-default
2248 @opindex Wswitch-switch
2249 Warn whenever a @code{switch} statement does not have a @code{default}
2253 @opindex Wswitch-enum
2254 Warn whenever a @code{switch} statement has an index of enumeral type
2255 and lacks a @code{case} for one or more of the named codes of that
2256 enumeration. @code{case} labels outside the enumeration range also
2257 provoke warnings when this option is used.
2261 Warn if any trigraphs are encountered that might change the meaning of
2262 the program (trigraphs within comments are not warned about).
2264 @item -Wunused-function
2265 @opindex Wunused-function
2266 Warn whenever a static function is declared but not defined or a
2267 non\-inline static function is unused.
2269 @item -Wunused-label
2270 @opindex Wunused-label
2271 Warn whenever a label is declared but not used.
2273 To suppress this warning use the @samp{unused} attribute
2274 (@pxref{Variable Attributes}).
2276 @item -Wunused-parameter
2277 @opindex Wunused-parameter
2278 Warn whenever a function parameter is unused aside from its declaration.
2280 To suppress this warning use the @samp{unused} attribute
2281 (@pxref{Variable Attributes}).
2283 @item -Wunused-variable
2284 @opindex Wunused-variable
2285 Warn whenever a local variable or non-constant static variable is unused
2286 aside from its declaration
2288 To suppress this warning use the @samp{unused} attribute
2289 (@pxref{Variable Attributes}).
2291 @item -Wunused-value
2292 @opindex Wunused-value
2293 Warn whenever a statement computes a result that is explicitly not used.
2295 To suppress this warning cast the expression to @samp{void}.
2299 All the above @option{-Wunused} options combined.
2301 In order to get a warning about an unused function parameter, you must
2302 either specify @samp{-W -Wunused} or separately specify
2303 @option{-Wunused-parameter}.
2305 @item -Wuninitialized
2306 @opindex Wuninitialized
2307 Warn if an automatic variable is used without first being initialized or
2308 if a variable may be clobbered by a @code{setjmp} call.
2310 These warnings are possible only in optimizing compilation,
2311 because they require data flow information that is computed only
2312 when optimizing. If you don't specify @option{-O}, you simply won't
2315 These warnings occur only for variables that are candidates for
2316 register allocation. Therefore, they do not occur for a variable that
2317 is declared @code{volatile}, or whose address is taken, or whose size
2318 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2319 structures, unions or arrays, even when they are in registers.
2321 Note that there may be no warning about a variable that is used only
2322 to compute a value that itself is never used, because such
2323 computations may be deleted by data flow analysis before the warnings
2326 These warnings are made optional because GCC is not smart
2327 enough to see all the reasons why the code might be correct
2328 despite appearing to have an error. Here is one example of how
2349 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2350 always initialized, but GCC doesn't know this. Here is
2351 another common case:
2356 if (change_y) save_y = y, y = new_y;
2358 if (change_y) y = save_y;
2363 This has no bug because @code{save_y} is used only if it is set.
2365 @cindex @code{longjmp} warnings
2366 This option also warns when a non-volatile automatic variable might be
2367 changed by a call to @code{longjmp}. These warnings as well are possible
2368 only in optimizing compilation.
2370 The compiler sees only the calls to @code{setjmp}. It cannot know
2371 where @code{longjmp} will be called; in fact, a signal handler could
2372 call it at any point in the code. As a result, you may get a warning
2373 even when there is in fact no problem because @code{longjmp} cannot
2374 in fact be called at the place which would cause a problem.
2376 Some spurious warnings can be avoided if you declare all the functions
2377 you use that never return as @code{noreturn}. @xref{Function
2380 @item -Wunknown-pragmas
2381 @opindex Wunknown-pragmas
2382 @cindex warning for unknown pragmas
2383 @cindex unknown pragmas, warning
2384 @cindex pragmas, warning of unknown
2385 Warn when a #pragma directive is encountered which is not understood by
2386 GCC@. If this command line option is used, warnings will even be issued
2387 for unknown pragmas in system header files. This is not the case if
2388 the warnings were only enabled by the @option{-Wall} command line option.
2390 @item -Wstrict-aliasing
2391 @opindex Wstrict-aliasing
2392 This option is only active when @option{-fstrict-aliasing} is active.
2393 It warns about code which might break the strict aliasing rules that the
2394 compiler is using for optimization. The warning does not catch all
2395 cases, but does attempt to catch the more common pitfalls. It is
2396 included in @option{-Wall}.
2400 All of the above @samp{-W} options combined. This enables all the
2401 warnings about constructions that some users consider questionable, and
2402 that are easy to avoid (or modify to prevent the warning), even in
2403 conjunction with macros. This also enables some language-specific
2404 warnings described in @ref{C++ Dialect Options} and
2405 @ref{Objective-C Dialect Options}.
2408 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2409 Some of them warn about constructions that users generally do not
2410 consider questionable, but which occasionally you might wish to check
2411 for; others warn about constructions that are necessary or hard to avoid
2412 in some cases, and there is no simple way to modify the code to suppress
2418 Print extra warning messages for these events:
2422 A function can return either with or without a value. (Falling
2423 off the end of the function body is considered returning without
2424 a value.) For example, this function would evoke such a
2438 An expression-statement or the left-hand side of a comma expression
2439 contains no side effects.
2440 To suppress the warning, cast the unused expression to void.
2441 For example, an expression such as @samp{x[i,j]} will cause a warning,
2442 but @samp{x[(void)i,j]} will not.
2445 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2448 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2449 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2450 that of ordinary mathematical notation.
2453 Storage-class specifiers like @code{static} are not the first things in
2454 a declaration. According to the C Standard, this usage is obsolescent.
2457 The return type of a function has a type qualifier such as @code{const}.
2458 Such a type qualifier has no effect, since the value returned by a
2459 function is not an lvalue. (But don't warn about the GNU extension of
2460 @code{volatile void} return types. That extension will be warned about
2461 if @option{-pedantic} is specified.)
2464 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2468 A comparison between signed and unsigned values could produce an
2469 incorrect result when the signed value is converted to unsigned.
2470 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2473 An aggregate has a partly bracketed initializer.
2474 For example, the following code would evoke such a warning,
2475 because braces are missing around the initializer for @code{x.h}:
2478 struct s @{ int f, g; @};
2479 struct t @{ struct s h; int i; @};
2480 struct t x = @{ 1, 2, 3 @};
2484 An aggregate has an initializer which does not initialize all members.
2485 For example, the following code would cause such a warning, because
2486 @code{x.h} would be implicitly initialized to zero:
2489 struct s @{ int f, g, h; @};
2490 struct s x = @{ 3, 4 @};
2494 @item -Wno-div-by-zero
2495 @opindex Wno-div-by-zero
2496 @opindex Wdiv-by-zero
2497 Do not warn about compile-time integer division by zero. Floating point
2498 division by zero is not warned about, as it can be a legitimate way of
2499 obtaining infinities and NaNs.
2501 @item -Wsystem-headers
2502 @opindex Wsystem-headers
2503 @cindex warnings from system headers
2504 @cindex system headers, warnings from
2505 Print warning messages for constructs found in system header files.
2506 Warnings from system headers are normally suppressed, on the assumption
2507 that they usually do not indicate real problems and would only make the
2508 compiler output harder to read. Using this command line option tells
2509 GCC to emit warnings from system headers as if they occurred in user
2510 code. However, note that using @option{-Wall} in conjunction with this
2511 option will @emph{not} warn about unknown pragmas in system
2512 headers---for that, @option{-Wunknown-pragmas} must also be used.
2515 @opindex Wfloat-equal
2516 Warn if floating point values are used in equality comparisons.
2518 The idea behind this is that sometimes it is convenient (for the
2519 programmer) to consider floating-point values as approximations to
2520 infinitely precise real numbers. If you are doing this, then you need
2521 to compute (by analysing the code, or in some other way) the maximum or
2522 likely maximum error that the computation introduces, and allow for it
2523 when performing comparisons (and when producing output, but that's a
2524 different problem). In particular, instead of testing for equality, you
2525 would check to see whether the two values have ranges that overlap; and
2526 this is done with the relational operators, so equality comparisons are
2529 @item -Wtraditional @r{(C only)}
2530 @opindex Wtraditional
2531 Warn about certain constructs that behave differently in traditional and
2532 ISO C@. Also warn about ISO C constructs that have no traditional C
2533 equivalent, and/or problematic constructs which should be avoided.
2537 Macro parameters that appear within string literals in the macro body.
2538 In traditional C macro replacement takes place within string literals,
2539 but does not in ISO C@.
2542 In traditional C, some preprocessor directives did not exist.
2543 Traditional preprocessors would only consider a line to be a directive
2544 if the @samp{#} appeared in column 1 on the line. Therefore
2545 @option{-Wtraditional} warns about directives that traditional C
2546 understands but would ignore because the @samp{#} does not appear as the
2547 first character on the line. It also suggests you hide directives like
2548 @samp{#pragma} not understood by traditional C by indenting them. Some
2549 traditional implementations would not recognize @samp{#elif}, so it
2550 suggests avoiding it altogether.
2553 A function-like macro that appears without arguments.
2556 The unary plus operator.
2559 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2560 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2561 constants.) Note, these suffixes appear in macros defined in the system
2562 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2563 Use of these macros in user code might normally lead to spurious
2564 warnings, however gcc's integrated preprocessor has enough context to
2565 avoid warning in these cases.
2568 A function declared external in one block and then used after the end of
2572 A @code{switch} statement has an operand of type @code{long}.
2575 A non-@code{static} function declaration follows a @code{static} one.
2576 This construct is not accepted by some traditional C compilers.
2579 The ISO type of an integer constant has a different width or
2580 signedness from its traditional type. This warning is only issued if
2581 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2582 typically represent bit patterns, are not warned about.
2585 Usage of ISO string concatenation is detected.
2588 Initialization of automatic aggregates.
2591 Identifier conflicts with labels. Traditional C lacks a separate
2592 namespace for labels.
2595 Initialization of unions. If the initializer is zero, the warning is
2596 omitted. This is done under the assumption that the zero initializer in
2597 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2598 initializer warnings and relies on default initialization to zero in the
2602 Conversions by prototypes between fixed/floating point values and vice
2603 versa. The absence of these prototypes when compiling with traditional
2604 C would cause serious problems. This is a subset of the possible
2605 conversion warnings, for the full set use @option{-Wconversion}.
2608 Use of ISO C style function definitions. This warning intentionally is
2609 @emph{not} issued for prototype declarations or variadic functions
2610 because these ISO C features will appear in your code when using
2611 libiberty's traditional C compatibility macros, @code{PARAMS} and
2612 @code{VPARAMS}. This warning is also bypassed for nested functions
2613 because that feature is already a gcc extension and thus not relevant to
2614 traditional C compatibility.
2619 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2621 @item -Wendif-labels
2622 @opindex Wendif-labels
2623 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2627 Warn whenever a local variable shadows another local variable, parameter or
2628 global variable or whenever a built-in function is shadowed.
2630 @item -Wlarger-than-@var{len}
2631 @opindex Wlarger-than
2632 Warn whenever an object of larger than @var{len} bytes is defined.
2634 @item -Wpointer-arith
2635 @opindex Wpointer-arith
2636 Warn about anything that depends on the ``size of'' a function type or
2637 of @code{void}. GNU C assigns these types a size of 1, for
2638 convenience in calculations with @code{void *} pointers and pointers
2641 @item -Wbad-function-cast @r{(C only)}
2642 @opindex Wbad-function-cast
2643 Warn whenever a function call is cast to a non-matching type.
2644 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2648 Warn whenever a pointer is cast so as to remove a type qualifier from
2649 the target type. For example, warn if a @code{const char *} is cast
2650 to an ordinary @code{char *}.
2653 @opindex Wcast-align
2654 Warn whenever a pointer is cast such that the required alignment of the
2655 target is increased. For example, warn if a @code{char *} is cast to
2656 an @code{int *} on machines where integers can only be accessed at
2657 two- or four-byte boundaries.
2659 @item -Wwrite-strings
2660 @opindex Wwrite-strings
2661 When compiling C, give string constants the type @code{const
2662 char[@var{length}]} so that
2663 copying the address of one into a non-@code{const} @code{char *}
2664 pointer will get a warning; when compiling C++, warn about the
2665 deprecated conversion from string constants to @code{char *}.
2666 These warnings will help you find at
2667 compile time code that can try to write into a string constant, but
2668 only if you have been very careful about using @code{const} in
2669 declarations and prototypes. Otherwise, it will just be a nuisance;
2670 this is why we did not make @option{-Wall} request these warnings.
2673 @opindex Wconversion
2674 Warn if a prototype causes a type conversion that is different from what
2675 would happen to the same argument in the absence of a prototype. This
2676 includes conversions of fixed point to floating and vice versa, and
2677 conversions changing the width or signedness of a fixed point argument
2678 except when the same as the default promotion.
2680 Also, warn if a negative integer constant expression is implicitly
2681 converted to an unsigned type. For example, warn about the assignment
2682 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2683 casts like @code{(unsigned) -1}.
2685 @item -Wsign-compare
2686 @opindex Wsign-compare
2687 @cindex warning for comparison of signed and unsigned values
2688 @cindex comparison of signed and unsigned values, warning
2689 @cindex signed and unsigned values, comparison warning
2690 Warn when a comparison between signed and unsigned values could produce
2691 an incorrect result when the signed value is converted to unsigned.
2692 This warning is also enabled by @option{-W}; to get the other warnings
2693 of @option{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2695 @item -Waggregate-return
2696 @opindex Waggregate-return
2697 Warn if any functions that return structures or unions are defined or
2698 called. (In languages where you can return an array, this also elicits
2701 @item -Wstrict-prototypes @r{(C only)}
2702 @opindex Wstrict-prototypes
2703 Warn if a function is declared or defined without specifying the
2704 argument types. (An old-style function definition is permitted without
2705 a warning if preceded by a declaration which specifies the argument
2708 @item -Wmissing-prototypes @r{(C only)}
2709 @opindex Wmissing-prototypes
2710 Warn if a global function is defined without a previous prototype
2711 declaration. This warning is issued even if the definition itself
2712 provides a prototype. The aim is to detect global functions that fail
2713 to be declared in header files.
2715 @item -Wmissing-declarations
2716 @opindex Wmissing-declarations
2717 Warn if a global function is defined without a previous declaration.
2718 Do so even if the definition itself provides a prototype.
2719 Use this option to detect global functions that are not declared in
2722 @item -Wmissing-noreturn
2723 @opindex Wmissing-noreturn
2724 Warn about functions which might be candidates for attribute @code{noreturn}.
2725 Note these are only possible candidates, not absolute ones. Care should
2726 be taken to manually verify functions actually do not ever return before
2727 adding the @code{noreturn} attribute, otherwise subtle code generation
2728 bugs could be introduced. You will not get a warning for @code{main} in
2729 hosted C environments.
2731 @item -Wmissing-format-attribute
2732 @opindex Wmissing-format-attribute
2734 If @option{-Wformat} is enabled, also warn about functions which might be
2735 candidates for @code{format} attributes. Note these are only possible
2736 candidates, not absolute ones. GCC will guess that @code{format}
2737 attributes might be appropriate for any function that calls a function
2738 like @code{vprintf} or @code{vscanf}, but this might not always be the
2739 case, and some functions for which @code{format} attributes are
2740 appropriate may not be detected. This option has no effect unless
2741 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2743 @item -Wno-multichar
2744 @opindex Wno-multichar
2746 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2747 Usually they indicate a typo in the user's code, as they have
2748 implementation-defined values, and should not be used in portable code.
2750 @item -Wno-deprecated-declarations
2751 @opindex Wno-deprecated-declarations
2752 Do not warn about uses of functions, variables, and types marked as
2753 deprecated by using the @code{deprecated} attribute.
2754 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2755 @pxref{Type Attributes}.)
2759 Warn if a structure is given the packed attribute, but the packed
2760 attribute has no effect on the layout or size of the structure.
2761 Such structures may be mis-aligned for little benefit. For
2762 instance, in this code, the variable @code{f.x} in @code{struct bar}
2763 will be misaligned even though @code{struct bar} does not itself
2764 have the packed attribute:
2771 @} __attribute__((packed));
2781 Warn if padding is included in a structure, either to align an element
2782 of the structure or to align the whole structure. Sometimes when this
2783 happens it is possible to rearrange the fields of the structure to
2784 reduce the padding and so make the structure smaller.
2786 @item -Wredundant-decls
2787 @opindex Wredundant-decls
2788 Warn if anything is declared more than once in the same scope, even in
2789 cases where multiple declaration is valid and changes nothing.
2791 @item -Wnested-externs @r{(C only)}
2792 @opindex Wnested-externs
2793 Warn if an @code{extern} declaration is encountered within a function.
2795 @item -Wunreachable-code
2796 @opindex Wunreachable-code
2797 Warn if the compiler detects that code will never be executed.
2799 This option is intended to warn when the compiler detects that at
2800 least a whole line of source code will never be executed, because
2801 some condition is never satisfied or because it is after a
2802 procedure that never returns.
2804 It is possible for this option to produce a warning even though there
2805 are circumstances under which part of the affected line can be executed,
2806 so care should be taken when removing apparently-unreachable code.
2808 For instance, when a function is inlined, a warning may mean that the
2809 line is unreachable in only one inlined copy of the function.
2811 This option is not made part of @option{-Wall} because in a debugging
2812 version of a program there is often substantial code which checks
2813 correct functioning of the program and is, hopefully, unreachable
2814 because the program does work. Another common use of unreachable
2815 code is to provide behavior which is selectable at compile-time.
2819 Warn if a function can not be inlined and it was declared as inline.
2823 @opindex Wno-long-long
2824 Warn if @samp{long long} type is used. This is default. To inhibit
2825 the warning messages, use @option{-Wno-long-long}. Flags
2826 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2827 only when @option{-pedantic} flag is used.
2829 @item -Wdisabled-optimization
2830 @opindex Wdisabled-optimization
2831 Warn if a requested optimization pass is disabled. This warning does
2832 not generally indicate that there is anything wrong with your code; it
2833 merely indicates that GCC's optimizers were unable to handle the code
2834 effectively. Often, the problem is that your code is too big or too
2835 complex; GCC will refuse to optimize programs when the optimization
2836 itself is likely to take inordinate amounts of time.
2840 Make all warnings into errors.
2843 @node Debugging Options
2844 @section Options for Debugging Your Program or GCC
2845 @cindex options, debugging
2846 @cindex debugging information options
2848 GCC has various special options that are used for debugging
2849 either your program or GCC:
2854 Produce debugging information in the operating system's native format
2855 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2858 On most systems that use stabs format, @option{-g} enables use of extra
2859 debugging information that only GDB can use; this extra information
2860 makes debugging work better in GDB but will probably make other debuggers
2862 refuse to read the program. If you want to control for certain whether
2863 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2864 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
2865 or @option{-gvms} (see below).
2867 Unlike most other C compilers, GCC allows you to use @option{-g} with
2868 @option{-O}. The shortcuts taken by optimized code may occasionally
2869 produce surprising results: some variables you declared may not exist
2870 at all; flow of control may briefly move where you did not expect it;
2871 some statements may not be executed because they compute constant
2872 results or their values were already at hand; some statements may
2873 execute in different places because they were moved out of loops.
2875 Nevertheless it proves possible to debug optimized output. This makes
2876 it reasonable to use the optimizer for programs that might have bugs.
2878 The following options are useful when GCC is generated with the
2879 capability for more than one debugging format.
2883 Produce debugging information for use by GDB@. This means to use the
2884 most expressive format available (DWARF 2, stabs, or the native format
2885 if neither of those are supported), including GDB extensions if at all
2890 Produce debugging information in stabs format (if that is supported),
2891 without GDB extensions. This is the format used by DBX on most BSD
2892 systems. On MIPS, Alpha and System V Release 4 systems this option
2893 produces stabs debugging output which is not understood by DBX or SDB@.
2894 On System V Release 4 systems this option requires the GNU assembler.
2898 Produce debugging information in stabs format (if that is supported),
2899 using GNU extensions understood only by the GNU debugger (GDB)@. The
2900 use of these extensions is likely to make other debuggers crash or
2901 refuse to read the program.
2905 Produce debugging information in COFF format (if that is supported).
2906 This is the format used by SDB on most System V systems prior to
2911 Produce debugging information in XCOFF format (if that is supported).
2912 This is the format used by the DBX debugger on IBM RS/6000 systems.
2916 Produce debugging information in XCOFF format (if that is supported),
2917 using GNU extensions understood only by the GNU debugger (GDB)@. The
2918 use of these extensions is likely to make other debuggers crash or
2919 refuse to read the program, and may cause assemblers other than the GNU
2920 assembler (GAS) to fail with an error.
2924 Produce debugging information in DWARF version 1 format (if that is
2925 supported). This is the format used by SDB on most System V Release 4
2928 This option is deprecated.
2932 Produce debugging information in DWARF version 1 format (if that is
2933 supported), using GNU extensions understood only by the GNU debugger
2934 (GDB)@. The use of these extensions is likely to make other debuggers
2935 crash or refuse to read the program.
2937 This option is deprecated.
2941 Produce debugging information in DWARF version 2 format (if that is
2942 supported). This is the format used by DBX on IRIX 6.
2946 Produce debugging information in VMS debug format (if that is
2947 supported). This is the format used by DEBUG on VMS systems.
2950 @itemx -ggdb@var{level}
2951 @itemx -gstabs@var{level}
2952 @itemx -gcoff@var{level}
2953 @itemx -gxcoff@var{level}
2954 @itemx -gvms@var{level}
2955 Request debugging information and also use @var{level} to specify how
2956 much information. The default level is 2.
2958 Level 1 produces minimal information, enough for making backtraces in
2959 parts of the program that you don't plan to debug. This includes
2960 descriptions of functions and external variables, but no information
2961 about local variables and no line numbers.
2963 Level 3 includes extra information, such as all the macro definitions
2964 present in the program. Some debuggers support macro expansion when
2965 you use @option{-g3}.
2967 Note that in order to avoid confusion between DWARF1 debug level 2,
2968 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
2969 a concatenated debug level. Instead use an additional @option{-g@var{level}}
2970 option to change the debug level for DWARF1 or DWARF2.
2972 @item -feliminate-dwarf2-dups
2973 @opindex feliminate-dwarf2-dups
2974 Compress DWARF2 debugging information by eliminating duplicated
2975 information about each symbol. This option only makes sense when
2976 generating DWARF2 debugging information with @option{-gdwarf-2}.
2981 Generate extra code to write profile information suitable for the
2982 analysis program @code{prof}. You must use this option when compiling
2983 the source files you want data about, and you must also use it when
2986 @cindex @code{gprof}
2989 Generate extra code to write profile information suitable for the
2990 analysis program @code{gprof}. You must use this option when compiling
2991 the source files you want data about, and you must also use it when
2996 Makes the compiler print out each function name as it is compiled, and
2997 print some statistics about each pass when it finishes.
3000 @opindex ftime-report
3001 Makes the compiler print some statistics about the time consumed by each
3002 pass when it finishes.
3005 @opindex fmem-report
3006 Makes the compiler print some statistics about permanent memory
3007 allocation when it finishes.
3009 @item -fprofile-arcs
3010 @opindex fprofile-arcs
3011 Instrument @dfn{arcs} during compilation to generate coverage data or
3012 for profile-directed block ordering. During execution the program
3013 records how many times each branch is executed and how many times it is
3014 taken. When the compiled program exits it saves this data to a file
3015 called @file{@var{auxname}.da} for each source file. @var{auxname} is
3016 generated from the name of the output file, if explicitly specified and
3017 it is not the final executable, otherwise it is the basename of the
3018 source file. In both cases any suffix is removed (e.g. @file{foo.da}
3019 for input file @file{dir/foo.c}, or @file{dir/foo.da} for output file
3020 specified as @option{-o dir/foo.o}).
3022 For profile-directed block ordering, compile the program with
3023 @option{-fprofile-arcs} plus optimization and code generation options,
3024 generate the arc profile information by running the program on a
3025 selected workload, and then compile the program again with the same
3026 optimization and code generation options plus
3027 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3028 Control Optimization}).
3030 The other use of @option{-fprofile-arcs} is for use with @code{gcov},
3031 when it is used with the @option{-ftest-coverage} option.
3033 With @option{-fprofile-arcs}, for each function of your program GCC
3034 creates a program flow graph, then finds a spanning tree for the graph.
3035 Only arcs that are not on the spanning tree have to be instrumented: the
3036 compiler adds code to count the number of times that these arcs are
3037 executed. When an arc is the only exit or only entrance to a block, the
3038 instrumentation code can be added to the block; otherwise, a new basic
3039 block must be created to hold the instrumentation code.
3042 @item -ftest-coverage
3043 @opindex ftest-coverage
3044 Create data files for the @code{gcov} code-coverage utility
3045 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}). See
3046 @option{-fprofile-arcs} option above for a description of @var{auxname}.
3049 @item @var{auxname}.bb
3050 A mapping from basic blocks to line numbers, which @code{gcov} uses to
3051 associate basic block execution counts with line numbers.
3053 @item @var{auxname}.bbg
3054 A list of all arcs in the program flow graph. This allows @code{gcov}
3055 to reconstruct the program flow graph, so that it can compute all basic
3056 block and arc execution counts from the information in the
3057 @file{@var{auxname}.da} file.
3060 Use @option{-ftest-coverage} with @option{-fprofile-arcs}; the latter
3061 option adds instrumentation to the program, which then writes
3062 execution counts to another data file:
3065 @item @var{auxname}.da
3066 Runtime arc execution counts, used in conjunction with the arc
3067 information in the file @file{@var{auxname}.bbg}.
3070 Coverage data will map better to the source files if
3071 @option{-ftest-coverage} is used without optimization.
3073 @item -d@var{letters}
3075 Says to make debugging dumps during compilation at times specified by
3076 @var{letters}. This is used for debugging the compiler. The file names
3077 for most of the dumps are made by appending a pass number and a word to
3078 the @var{dumpname}. @var{dumpname} is generated from the name of the
3079 output file, if explicitly specified and it is not an executable,
3080 otherwise it is the basename of the source file. In both cases any
3081 suffix is removed (e.g. @file{foo.00.rtl} or @file{foo.01.sibling}).
3082 Here are the possible letters for use in @var{letters}, and their
3088 Annotate the assembler output with miscellaneous debugging information.
3091 Dump after computing branch probabilities, to @file{@var{file}.15.bp}.
3094 Dump after block reordering, to @file{@var{file}.30.bbro}.
3097 Dump after instruction combination, to the file @file{@var{file}.17.combine}.
3100 Dump after the first if conversion, to the file @file{@var{file}.18.ce}.
3103 Dump after delayed branch scheduling, to @file{@var{file}.32.dbr}.
3106 Dump all macro definitions, at the end of preprocessing, in addition to
3110 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
3111 @file{@var{file}.07.ussa}.
3114 Dump after the second if conversion, to @file{@var{file}.27.ce2}.
3117 Dump after life analysis, to @file{@var{file}.16.life}.
3120 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.10.addressof}.
3123 Dump after global register allocation, to @file{@var{file}.22.greg}.
3126 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3129 Dump after reg-to-stack conversion, to @file{@var{file}.29.stack}.
3132 Dump after post-reload optimizations, to @file{@var{file}.23.postreload}.
3135 Dump after GCSE, to @file{@var{file}.11.gcse}.
3138 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3141 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3144 Dump after conversion from registers to stack, to @file{@var{file}.33.stack}.
3147 Dump after local register allocation, to @file{@var{file}.21.lreg}.
3150 Dump after loop optimization, to @file{@var{file}.12.loop}.
3153 Dump after performing the machine dependent reorganisation pass, to
3154 @file{@var{file}.31.mach}.
3157 Dump after register renumbering, to @file{@var{file}.26.rnreg}.
3160 Dump after the register move pass, to @file{@var{file}.19.regmove}.
3163 Dump after RTL generation, to @file{@var{file}.00.rtl}.
3166 Dump after the second scheduling pass, to @file{@var{file}.28.sched2}.
3169 Dump after CSE (including the jump optimization that sometimes follows
3170 CSE), to @file{@var{file}.09.cse}.
3173 Dump after the first scheduling pass, to @file{@var{file}.20.sched}.
3176 Dump after the second CSE pass (including the jump optimization that
3177 sometimes follows CSE), to @file{@var{file}.13.cse2}.
3180 Dump after null pointer ellimination pass ti @file{@var{file}.08.null}.
3183 Dump after the second flow pass, to @file{@var{file}.24.flow2}.
3186 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
3189 Dump after the peephole pass, to @file{@var{file}.25.peephole2}.
3192 Produce all the dumps listed above.
3195 Print statistics on memory usage, at the end of the run, to
3199 Annotate the assembler output with a comment indicating which
3200 pattern and alternative was used. The length of each instruction is
3204 Dump the RTL in the assembler output as a comment before each instruction.
3205 Also turns on @option{-dp} annotation.
3208 For each of the other indicated dump files (except for
3209 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
3210 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3213 Just generate RTL for a function instead of compiling it. Usually used
3217 Dump debugging information during parsing, to standard error.
3220 @item -fdump-unnumbered
3221 @opindex fdump-unnumbered
3222 When doing debugging dumps (see @option{-d} option above), suppress instruction
3223 numbers and line number note output. This makes it more feasible to
3224 use diff on debugging dumps for compiler invocations with different
3225 options, in particular with and without @option{-g}.
3227 @item -fdump-translation-unit @r{(C and C++ only)}
3228 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3229 @opindex fdump-translation-unit
3230 Dump a representation of the tree structure for the entire translation
3231 unit to a file. The file name is made by appending @file{.tu} to the
3232 source file name. If the @samp{-@var{options}} form is used, @var{options}
3233 controls the details of the dump as described for the
3234 @option{-fdump-tree} options.
3236 @item -fdump-class-hierarchy @r{(C++ only)}
3237 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3238 @opindex fdump-class-hierarchy
3239 Dump a representation of each class's hierarchy and virtual function
3240 table layout to a file. The file name is made by appending @file{.class}
3241 to the source file name. If the @samp{-@var{options}} form is used,
3242 @var{options} controls the details of the dump as described for the
3243 @option{-fdump-tree} options.
3245 @item -fdump-tree-@var{switch} @r{(C++ only)}
3246 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3248 Control the dumping at various stages of processing the intermediate
3249 language tree to a file. The file name is generated by appending a switch
3250 specific suffix to the source file name. If the @samp{-@var{options}}
3251 form is used, @var{options} is a list of @samp{-} separated options that
3252 control the details of the dump. Not all options are applicable to all
3253 dumps, those which are not meaningful will be ignored. The following
3254 options are available
3258 Print the address of each node. Usually this is not meaningful as it
3259 changes according to the environment and source file. Its primary use
3260 is for tying up a dump file with a debug environment.
3262 Inhibit dumping of members of a scope or body of a function merely
3263 because that scope has been reached. Only dump such items when they
3264 are directly reachable by some other path.
3266 Turn on all options.
3269 The following tree dumps are possible:
3272 Dump before any tree based optimization, to @file{@var{file}.original}.
3274 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3276 Dump after function inlining, to @file{@var{file}.inlined}.
3279 @item -fsched-verbose=@var{n}
3280 @opindex fsched-verbose
3281 On targets that use instruction scheduling, this option controls the
3282 amount of debugging output the scheduler prints. This information is
3283 written to standard error, unless @option{-dS} or @option{-dR} is
3284 specified, in which case it is output to the usual dump
3285 listing file, @file{.sched} or @file{.sched2} respectively. However
3286 for @var{n} greater than nine, the output is always printed to standard
3289 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3290 same information as @option{-dRS}. For @var{n} greater than one, it
3291 also output basic block probabilities, detailed ready list information
3292 and unit/insn info. For @var{n} greater than two, it includes RTL
3293 at abort point, control-flow and regions info. And for @var{n} over
3294 four, @option{-fsched-verbose} also includes dependence info.
3298 Store the usual ``temporary'' intermediate files permanently; place them
3299 in the current directory and name them based on the source file. Thus,
3300 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3301 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3302 preprocessed @file{foo.i} output file even though the compiler now
3303 normally uses an integrated preprocessor.
3307 Report the CPU time taken by each subprocess in the compilation
3308 sequence. For C source files, this is the compiler proper and assembler
3309 (plus the linker if linking is done). The output looks like this:
3316 The first number on each line is the ``user time,'' that is time spent
3317 executing the program itself. The second number is ``system time,''
3318 time spent executing operating system routines on behalf of the program.
3319 Both numbers are in seconds.
3321 @item -print-file-name=@var{library}
3322 @opindex print-file-name
3323 Print the full absolute name of the library file @var{library} that
3324 would be used when linking---and don't do anything else. With this
3325 option, GCC does not compile or link anything; it just prints the
3328 @item -print-multi-directory
3329 @opindex print-multi-directory
3330 Print the directory name corresponding to the multilib selected by any
3331 other switches present in the command line. This directory is supposed
3332 to exist in @env{GCC_EXEC_PREFIX}.
3334 @item -print-multi-lib
3335 @opindex print-multi-lib
3336 Print the mapping from multilib directory names to compiler switches
3337 that enable them. The directory name is separated from the switches by
3338 @samp{;}, and each switch starts with an @samp{@@} instead of the
3339 @samp{-}, without spaces between multiple switches. This is supposed to
3340 ease shell-processing.
3342 @item -print-prog-name=@var{program}
3343 @opindex print-prog-name
3344 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3346 @item -print-libgcc-file-name
3347 @opindex print-libgcc-file-name
3348 Same as @option{-print-file-name=libgcc.a}.
3350 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3351 but you do want to link with @file{libgcc.a}. You can do
3354 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3357 @item -print-search-dirs
3358 @opindex print-search-dirs
3359 Print the name of the configured installation directory and a list of
3360 program and library directories gcc will search---and don't do anything else.
3362 This is useful when gcc prints the error message
3363 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3364 To resolve this you either need to put @file{cpp0} and the other compiler
3365 components where gcc expects to find them, or you can set the environment
3366 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3367 Don't forget the trailing '/'.
3368 @xref{Environment Variables}.
3371 @opindex dumpmachine
3372 Print the compiler's target machine (for example,
3373 @samp{i686-pc-linux-gnu})---and don't do anything else.
3376 @opindex dumpversion
3377 Print the compiler version (for example, @samp{3.0})---and don't do
3382 Print the compiler's built-in specs---and don't do anything else. (This
3383 is used when GCC itself is being built.) @xref{Spec Files}.
3386 @node Optimize Options
3387 @section Options That Control Optimization
3388 @cindex optimize options
3389 @cindex options, optimization
3391 These options control various sorts of optimizations:
3398 Optimize. Optimizing compilation takes somewhat more time, and a lot
3399 more memory for a large function.
3401 Without @option{-O}, the compiler's goal is to reduce the cost of
3402 compilation and to make debugging produce the expected results.
3403 Statements are independent: if you stop the program with a breakpoint
3404 between statements, you can then assign a new value to any variable or
3405 change the program counter to any other statement in the function and
3406 get exactly the results you would expect from the source code.
3408 With @option{-O}, the compiler tries to reduce code size and execution
3409 time, without performing any optimizations that take a great deal of
3414 Optimize even more. GCC performs nearly all supported optimizations
3415 that do not involve a space-speed tradeoff. The compiler does not
3416 perform loop unrolling or function inlining when you specify @option{-O2}.
3417 As compared to @option{-O}, this option increases both compilation time
3418 and the performance of the generated code.
3420 @option{-O2} turns on all optional optimizations except for loop
3421 unrolling, function inlining, and register renaming. It also turns on
3422 the @option{-fforce-mem} and @option{-fstrict-aliasing} option on all
3423 machines and frame pointer elimination on machines where doing so does
3424 not interfere with debugging.
3426 Please note the warning under @option{-fgcse} about
3427 invoking @option{-O2} on programs that use computed gotos.
3431 Optimize yet more. @option{-O3} turns on all optimizations specified by
3432 @option{-O2} and also turns on the @option{-finline-functions} and
3433 @option{-frename-registers} options.
3441 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3442 do not typically increase code size. It also performs further
3443 optimizations designed to reduce code size.
3445 If you use multiple @option{-O} options, with or without level numbers,
3446 the last such option is the one that is effective.
3449 Options of the form @option{-f@var{flag}} specify machine-independent
3450 flags. Most flags have both positive and negative forms; the negative
3451 form of @option{-ffoo} would be @option{-fno-foo}. In the table below,
3452 only one of the forms is listed---the one which is not the default.
3453 You can figure out the other form by either removing @samp{no-} or
3458 @opindex ffloat-store
3459 Do not store floating point variables in registers, and inhibit other
3460 options that might change whether a floating point value is taken from a
3463 @cindex floating point precision
3464 This option prevents undesirable excess precision on machines such as
3465 the 68000 where the floating registers (of the 68881) keep more
3466 precision than a @code{double} is supposed to have. Similarly for the
3467 x86 architecture. For most programs, the excess precision does only
3468 good, but a few programs rely on the precise definition of IEEE floating
3469 point. Use @option{-ffloat-store} for such programs, after modifying
3470 them to store all pertinent intermediate computations into variables.
3472 @item -fno-default-inline
3473 @opindex fno-default-inline
3474 Do not make member functions inline by default merely because they are
3475 defined inside the class scope (C++ only). Otherwise, when you specify
3476 @w{@option{-O}}, member functions defined inside class scope are compiled
3477 inline by default; i.e., you don't need to add @samp{inline} in front of
3478 the member function name.
3480 @item -fno-defer-pop
3481 @opindex fno-defer-pop
3482 Always pop the arguments to each function call as soon as that function
3483 returns. For machines which must pop arguments after a function call,
3484 the compiler normally lets arguments accumulate on the stack for several
3485 function calls and pops them all at once.
3489 Force memory operands to be copied into registers before doing
3490 arithmetic on them. This produces better code by making all memory
3491 references potential common subexpressions. When they are not common
3492 subexpressions, instruction combination should eliminate the separate
3493 register-load. The @option{-O2} option turns on this option.
3496 @opindex fforce-addr
3497 Force memory address constants to be copied into registers before
3498 doing arithmetic on them. This may produce better code just as
3499 @option{-fforce-mem} may.
3501 @item -fomit-frame-pointer
3502 @opindex fomit-frame-pointer
3503 Don't keep the frame pointer in a register for functions that
3504 don't need one. This avoids the instructions to save, set up and
3505 restore frame pointers; it also makes an extra register available
3506 in many functions. @strong{It also makes debugging impossible on
3509 On some machines, such as the VAX, this flag has no effect, because
3510 the standard calling sequence automatically handles the frame pointer
3511 and nothing is saved by pretending it doesn't exist. The
3512 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3513 whether a target machine supports this flag. @xref{Registers,,Register
3514 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3516 @item -foptimize-sibling-calls
3517 @opindex foptimize-sibling-calls
3518 Optimize sibling and tail recursive calls.
3522 This option generates traps for signed overflow on addition, subtraction,
3523 multiplication operations.
3527 Don't pay attention to the @code{inline} keyword. Normally this option
3528 is used to keep the compiler from expanding any functions inline.
3529 Note that if you are not optimizing, no functions can be expanded inline.
3531 @item -finline-functions
3532 @opindex finline-functions
3533 Integrate all simple functions into their callers. The compiler
3534 heuristically decides which functions are simple enough to be worth
3535 integrating in this way.
3537 If all calls to a given function are integrated, and the function is
3538 declared @code{static}, then the function is normally not output as
3539 assembler code in its own right.
3541 @item -finline-limit=@var{n}
3542 @opindex finline-limit
3543 By default, gcc limits the size of functions that can be inlined. This flag
3544 allows the control of this limit for functions that are explicitly marked as
3545 inline (ie marked with the inline keyword or defined within the class
3546 definition in c++). @var{n} is the size of functions that can be inlined in
3547 number of pseudo instructions (not counting parameter handling). The default
3548 value of @var{n} is 600.
3549 Increasing this value can result in more inlined code at
3550 the cost of compilation time and memory consumption. Decreasing usually makes
3551 the compilation faster and less code will be inlined (which presumably
3552 means slower programs). This option is particularly useful for programs that
3553 use inlining heavily such as those based on recursive templates with C++.
3555 @emph{Note:} pseudo instruction represents, in this particular context, an
3556 abstract measurement of function's size. In no way, it represents a count
3557 of assembly instructions and as such its exact meaning might change from one
3558 release to an another.
3560 @item -fkeep-inline-functions
3561 @opindex fkeep-inline-functions
3562 Even if all calls to a given function are integrated, and the function
3563 is declared @code{static}, nevertheless output a separate run-time
3564 callable version of the function. This switch does not affect
3565 @code{extern inline} functions.
3567 @item -fkeep-static-consts
3568 @opindex fkeep-static-consts
3569 Emit variables declared @code{static const} when optimization isn't turned
3570 on, even if the variables aren't referenced.
3572 GCC enables this option by default. If you want to force the compiler to
3573 check if the variable was referenced, regardless of whether or not
3574 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3576 @item -fmerge-constants
3577 Attempt to merge identical constants (string constants and floating point
3578 constants) accross compilation units.
3580 This option is default for optimized compilation if assembler and linker
3581 support it. Use @option{-fno-merge-constants} to inhibit this behavior.
3583 @item -fmerge-all-constants
3584 Attempt to merge identical constants and identical variables.
3586 This option implies @option{-fmerge-constants}. In addition to
3587 @option{-fmerge-constants} this considers e.g. even constant initialized
3588 arrays or initialized constant variables with integral or floating point
3589 types. Languages like C or C++ require each non-automatic variable to
3590 have distinct location, so using this option will result in non-conforming
3595 Use a graph coloring register allocator. Currently this option is meant
3596 for testing, so we are interested to hear about miscompilations with
3599 @item -fno-branch-count-reg
3600 @opindex fno-branch-count-reg
3601 Do not use ``decrement and branch'' instructions on a count register,
3602 but instead generate a sequence of instructions that decrement a
3603 register, compare it against zero, then branch based upon the result.
3604 This option is only meaningful on architectures that support such
3605 instructions, which include x86, PowerPC, IA-64 and S/390.
3607 @item -fno-function-cse
3608 @opindex fno-function-cse
3609 Do not put function addresses in registers; make each instruction that
3610 calls a constant function contain the function's address explicitly.
3612 This option results in less efficient code, but some strange hacks
3613 that alter the assembler output may be confused by the optimizations
3614 performed when this option is not used.
3618 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
3619 @option{-fno-trapping-math}, @option{-ffinite-math-only} and @*
3620 @option{-fno-signaling-nans}.
3622 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
3624 This option should never be turned on by any @option{-O} option since
3625 it can result in incorrect output for programs which depend on
3626 an exact implementation of IEEE or ISO rules/specifications for
3629 @item -fno-math-errno
3630 @opindex fno-math-errno
3631 Do not set ERRNO after calling math functions that are executed
3632 with a single instruction, e.g., sqrt. A program that relies on
3633 IEEE exceptions for math error handling may want to use this flag
3634 for speed while maintaining IEEE arithmetic compatibility.
3636 This option should never be turned on by any @option{-O} option since
3637 it can result in incorrect output for programs which depend on
3638 an exact implementation of IEEE or ISO rules/specifications for
3641 The default is @option{-fmath-errno}.
3643 @item -funsafe-math-optimizations
3644 @opindex funsafe-math-optimizations
3645 Allow optimizations for floating-point arithmetic that (a) assume
3646 that arguments and results are valid and (b) may violate IEEE or
3647 ANSI standards. When used at link-time, it may include libraries
3648 or startup files that change the default FPU control word or other
3649 similar optimizations.
3651 This option should never be turned on by any @option{-O} option since
3652 it can result in incorrect output for programs which depend on
3653 an exact implementation of IEEE or ISO rules/specifications for
3656 The default is @option{-fno-unsafe-math-optimizations}.
3658 @item -ffinite-math-only
3659 @opindex ffinite-math-only
3660 Allow optimizations for floating-point arithmetic that assume
3661 that arguments and results are not NaNs or +-Infs.
3663 This option should never be turned on by any @option{-O} option since
3664 it can result in incorrect output for programs which depend on
3665 an exact implementation of IEEE or ISO rules/specifications.
3667 The default is @option{-fno-finite-math-only}.
3669 @item -fno-trapping-math
3670 @opindex fno-trapping-math
3671 Compile code assuming that floating-point operations cannot generate
3672 user-visible traps. These traps include division by zero, overflow,
3673 underflow, inexact result and invalid operation. This option implies
3674 @option{-fno-signaling-nans}. Setting this option may allow faster
3675 code if one relies on ``non-stop'' IEEE arithmetic, for example.
3677 This option should never be turned on by any @option{-O} option since
3678 it can result in incorrect output for programs which depend on
3679 an exact implementation of IEEE or ISO rules/specifications for
3682 The default is @option{-ftrapping-math}.
3684 @item -fsignaling-nans
3685 @opindex fsignaling-nans
3686 Compile code assuming that IEEE signaling NaNs may generate user-visible
3687 traps during floating-point operations. Setting this option disables
3688 optimizations that may change the number of exceptions visible with
3689 signaling NaNs. This option implies @option{-ftrapping-math}.
3691 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
3694 The default is @option{-fno-signaling-nans}.
3696 This option is experimental and does not currently guarantee to
3697 disable all GCC optimizations that affect signaling NaN behavior.
3699 @item -fno-zero-initialized-in-bss
3700 @opindex fno-zero-initialized-in-bss
3701 If the target supports a BSS section, GCC by default puts variables that
3702 are initialized to zero into BSS@. This can save space in the resulting
3705 This option turns off this behavior because some programs explicitly
3706 rely on variables going to the data section. E.g., so that the
3707 resulting executable can find the beginning of that section and/or make
3708 assumptions based on that.
3710 The default is @option{-fzero-initialized-in-bss}.
3712 @item -fbounds-check
3713 @opindex fbounds-check
3714 For front-ends that support it, generate additional code to check that
3715 indices used to access arrays are within the declared range. This is
3716 currenly only supported by the Java and Fortran 77 front-ends, where
3717 this option defaults to true and false respectively.
3721 The following options control specific optimizations. The @option{-O2}
3722 option turns on all of these optimizations except @option{-funroll-loops}
3723 and @option{-funroll-all-loops}. On most machines, the @option{-O} option
3724 turns on the @option{-fthread-jumps} and @option{-fdelayed-branch} options,
3725 but specific machines may handle it differently.
3727 You can use the following flags in the rare cases when ``fine-tuning''
3728 of optimizations to be performed is desired.
3730 Not all of the optimizations performed by GCC have @option{-f} options
3734 @item -fstrength-reduce
3735 @opindex fstrength-reduce
3736 Perform the optimizations of loop strength reduction and
3737 elimination of iteration variables.
3739 @item -fthread-jumps
3740 @opindex fthread-jumps
3741 Perform optimizations where we check to see if a jump branches to a
3742 location where another comparison subsumed by the first is found. If
3743 so, the first branch is redirected to either the destination of the
3744 second branch or a point immediately following it, depending on whether
3745 the condition is known to be true or false.
3747 @item -fcse-follow-jumps
3748 @opindex fcse-follow-jumps
3749 In common subexpression elimination, scan through jump instructions
3750 when the target of the jump is not reached by any other path. For
3751 example, when CSE encounters an @code{if} statement with an
3752 @code{else} clause, CSE will follow the jump when the condition
3755 @item -fcse-skip-blocks
3756 @opindex fcse-skip-blocks
3757 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3758 follow jumps which conditionally skip over blocks. When CSE
3759 encounters a simple @code{if} statement with no else clause,
3760 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3761 body of the @code{if}.
3763 @item -frerun-cse-after-loop
3764 @opindex frerun-cse-after-loop
3765 Re-run common subexpression elimination after loop optimizations has been
3768 @item -frerun-loop-opt
3769 @opindex frerun-loop-opt
3770 Run the loop optimizer twice.
3774 Perform a global common subexpression elimination pass.
3775 This pass also performs global constant and copy propagation.
3777 @emph{Note:} When compiling a program using computed gotos, a GCC
3778 extension, you may get better runtime performance if you disable
3779 the global common subexpression elmination pass by adding
3780 @option{-fno-gcse} to the command line.
3784 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3785 attempt to move loads which are only killed by stores into themselves. This
3786 allows a loop containing a load/store sequence to be changed to a load outside
3787 the loop, and a copy/store within the loop.
3791 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3792 subexpression elimination. This pass will attempt to move stores out of loops.
3793 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3794 can be changed to a load before the loop and a store after the loop.
3796 @item -floop-optimize
3797 @opindex floop-optimize
3798 Perform loop optimizations: move constant expressions out of loops, simplify
3799 exit test conditions and optionally do strength-reduction and loop unrolling as
3802 @item -fcrossjumping
3803 @opindex crossjumping
3804 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3805 resulting code may or may not perform better than without cross-jumping.
3807 @item -fif-conversion
3808 @opindex if-conversion
3809 Attempt to transform conditional jumps into branch-less equivalents. This
3810 include use of conditional moves, min, max, set flags and abs instructions, and
3811 some tricks doable by standard arithmetics. The use of conditional execution
3812 on chips where it is available is controlled by @code{if-conversion2}.
3814 @item -fif-conversion2
3815 @opindex if-conversion2
3816 Use conditional execution (where available) to transform conditional jumps into
3817 branch-less equivalents.
3819 @item -fdelete-null-pointer-checks
3820 @opindex fdelete-null-pointer-checks
3821 Use global dataflow analysis to identify and eliminate useless checks
3822 for null pointers. The compiler assumes that dereferencing a null
3823 pointer would have halted the program. If a pointer is checked after
3824 it has already been dereferenced, it cannot be null.
3826 In some environments, this assumption is not true, and programs can
3827 safely dereference null pointers. Use
3828 @option{-fno-delete-null-pointer-checks} to disable this optimization
3829 for programs which depend on that behavior.
3831 @item -fexpensive-optimizations
3832 @opindex fexpensive-optimizations
3833 Perform a number of minor optimizations that are relatively expensive.
3835 @item -foptimize-register-move
3837 @opindex foptimize-register-move
3839 Attempt to reassign register numbers in move instructions and as
3840 operands of other simple instructions in order to maximize the amount of
3841 register tying. This is especially helpful on machines with two-operand
3842 instructions. GCC enables this optimization by default with @option{-O2}
3845 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3848 @item -fdelayed-branch
3849 @opindex fdelayed-branch
3850 If supported for the target machine, attempt to reorder instructions
3851 to exploit instruction slots available after delayed branch
3854 @item -fschedule-insns
3855 @opindex fschedule-insns
3856 If supported for the target machine, attempt to reorder instructions to
3857 eliminate execution stalls due to required data being unavailable. This
3858 helps machines that have slow floating point or memory load instructions
3859 by allowing other instructions to be issued until the result of the load
3860 or floating point instruction is required.
3862 @item -fschedule-insns2
3863 @opindex fschedule-insns2
3864 Similar to @option{-fschedule-insns}, but requests an additional pass of
3865 instruction scheduling after register allocation has been done. This is
3866 especially useful on machines with a relatively small number of
3867 registers and where memory load instructions take more than one cycle.
3869 @item -fno-sched-interblock
3870 @opindex fno-sched-interblock
3871 Don't schedule instructions across basic blocks. This is normally
3872 enabled by default when scheduling before register allocation, i.e.@:
3873 with @option{-fschedule-insns} or at @option{-O2} or higher.
3875 @item -fno-sched-spec
3876 @opindex fno-sched-spec
3877 Don't allow speculative motion of non-load instructions. This is normally
3878 enabled by default when scheduling before register allocation, i.e.@:
3879 with @option{-fschedule-insns} or at @option{-O2} or higher.
3881 @item -fsched-spec-load
3882 @opindex fsched-spec-load
3883 Allow speculative motion of some load instructions. This only makes
3884 sense when scheduling before register allocation, i.e.@: with
3885 @option{-fschedule-insns} or at @option{-O2} or higher.
3887 @item -fsched-spec-load-dangerous
3888 @opindex fsched-spec-load-dangerous
3889 Allow speculative motion of more load instructions. This only makes
3890 sense when scheduling before register allocation, i.e.@: with
3891 @option{-fschedule-insns} or at @option{-O2} or higher.
3893 @item -ffunction-sections
3894 @itemx -fdata-sections
3895 @opindex ffunction-sections
3896 @opindex fdata-sections
3897 Place each function or data item into its own section in the output
3898 file if the target supports arbitrary sections. The name of the
3899 function or the name of the data item determines the section's name
3902 Use these options on systems where the linker can perform optimizations
3903 to improve locality of reference in the instruction space. HPPA
3904 processors running HP-UX and SPARC processors running Solaris 2 have
3905 linkers with such optimizations. Other systems using the ELF object format
3906 as well as AIX may have these optimizations in the future.
3908 Only use these options when there are significant benefits from doing
3909 so. When you specify these options, the assembler and linker will
3910 create larger object and executable files and will also be slower.
3911 You will not be able to use @code{gprof} on all systems if you
3912 specify this option and you may have problems with debugging if
3913 you specify both this option and @option{-g}.
3915 @item -fcaller-saves
3916 @opindex fcaller-saves
3917 Enable values to be allocated in registers that will be clobbered by
3918 function calls, by emitting extra instructions to save and restore the
3919 registers around such calls. Such allocation is done only when it
3920 seems to result in better code than would otherwise be produced.
3922 This option is always enabled by default on certain machines, usually
3923 those which have no call-preserved registers to use instead.
3925 For all machines, optimization level 2 and higher enables this flag by
3930 Perform tail duplication to enlarge superblock size. This transformation
3931 simplifies the control flow of the function allowing other optimizations to do
3934 @item -funroll-loops
3935 @opindex funroll-loops
3936 Unroll loops whose number of iterations can be determined at compile
3937 time or upon entry to the loop. @option{-funroll-loops} implies both
3938 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
3939 option makes code larger, and may or may not make it run faster.
3941 @item -funroll-all-loops
3942 @opindex funroll-all-loops
3943 Unroll all loops, even if their number of iterations is uncertain when
3944 the loop is entered. This usually makes programs run more slowly.
3945 @option{-funroll-all-loops} implies the same options as
3946 @option{-funroll-loops},
3948 @item -fprefetch-loop-arrays
3949 @opindex fprefetch-loop-arrays
3950 If supported by the target machine, generate instructions to prefetch
3951 memory to improve the performance of loops that access large arrays.
3953 @item -fmove-all-movables
3954 @opindex fmove-all-movables
3955 Forces all invariant computations in loops to be moved
3958 @item -freduce-all-givs
3959 @opindex freduce-all-givs
3960 Forces all general-induction variables in loops to be
3963 @emph{Note:} When compiling programs written in Fortran,
3964 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3965 by default when you use the optimizer.
3967 These options may generate better or worse code; results are highly
3968 dependent on the structure of loops within the source code.
3970 These two options are intended to be removed someday, once
3971 they have helped determine the efficacy of various
3972 approaches to improving loop optimizations.
3974 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3975 know how use of these options affects
3976 the performance of your production code.
3977 We're very interested in code that runs @emph{slower}
3978 when these options are @emph{enabled}.
3981 @itemx -fno-peephole2
3982 @opindex fno-peephole
3983 @opindex fno-peephole2
3984 Disable any machine-specific peephole optimizations. The difference
3985 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3986 are implemented in the compiler; some targets use one, some use the
3987 other, a few use both.
3989 @item -fbranch-probabilities
3990 @opindex fbranch-probabilities
3991 After running a program compiled with @option{-fprofile-arcs}
3992 (@pxref{Debugging Options,, Options for Debugging Your Program or
3993 @command{gcc}}), you can compile it a second time using
3994 @option{-fbranch-probabilities}, to improve optimizations based on
3995 the number of times each branch was taken. When the program
3996 compiled with @option{-fprofile-arcs} exits it saves arc execution
3997 counts to a file called @file{@var{sourcename}.da} for each source
3998 file The information in this data file is very dependent on the
3999 structure of the generated code, so you must use the same source code
4000 and the same optimization options for both compilations.
4002 With @option{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
4003 note on the first instruction of each basic block, and a
4004 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4005 These can be used to improve optimization. Currently, they are only
4006 used in one place: in @file{reorg.c}, instead of guessing which path a
4007 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4008 exactly determine which path is taken more often.
4010 @item -fno-guess-branch-probability
4011 @opindex fno-guess-branch-probability
4012 Do not guess branch probabilities using a randomized model.
4014 Sometimes gcc will opt to use a randomized model to guess branch
4015 probabilities, when none are available from either profiling feedback
4016 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4017 different runs of the compiler on the same program may produce different
4020 In a hard real-time system, people don't want different runs of the
4021 compiler to produce code that has different behavior; minimizing
4022 non-determinism is of paramount import. This switch allows users to
4023 reduce non-determinism, possibly at the expense of inferior
4026 @item -freorder-blocks
4027 @opindex freorder-blocks
4028 Reorder basic blocks in the compiled function in order to reduce number of
4029 taken branches and improve code locality.
4031 @item -freorder-functions
4032 @opindex freorder-functions
4033 Reorder basic blocks in the compiled function in order to reduce number of
4034 taken branches and improve code locality. This is implemented by using special
4035 subsections @code{text.hot} for most frequently executed functions and
4036 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4037 the linker so object file format must support named sections and linker must
4038 place them in resonable way.
4040 Also profile feedback must be available in to make this option effective. See
4041 @option{-fprofile-arcs} for details.
4043 @item -fstrict-aliasing
4044 @opindex fstrict-aliasing
4045 Allows the compiler to assume the strictest aliasing rules applicable to
4046 the language being compiled. For C (and C++), this activates
4047 optimizations based on the type of expressions. In particular, an
4048 object of one type is assumed never to reside at the same address as an
4049 object of a different type, unless the types are almost the same. For
4050 example, an @code{unsigned int} can alias an @code{int}, but not a
4051 @code{void*} or a @code{double}. A character type may alias any other
4054 Pay special attention to code like this:
4067 The practice of reading from a different union member than the one most
4068 recently written to (called ``type-punning'') is common. Even with
4069 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4070 is accessed through the union type. So, the code above will work as
4071 expected. However, this code might not:
4082 Every language that wishes to perform language-specific alias analysis
4083 should define a function that computes, given an @code{tree}
4084 node, an alias set for the node. Nodes in different alias sets are not
4085 allowed to alias. For an example, see the C front-end function
4086 @code{c_get_alias_set}.
4088 For all machines, optimization level 2 and higher enables this flag by
4091 @item -falign-functions
4092 @itemx -falign-functions=@var{n}
4093 @opindex falign-functions
4094 Align the start of functions to the next power-of-two greater than
4095 @var{n}, skipping up to @var{n} bytes. For instance,
4096 @option{-falign-functions=32} aligns functions to the next 32-byte
4097 boundary, but @option{-falign-functions=24} would align to the next
4098 32-byte boundary only if this can be done by skipping 23 bytes or less.
4100 @option{-fno-align-functions} and @option{-falign-functions=1} are
4101 equivalent and mean that functions will not be aligned.
4103 Some assemblers only support this flag when @var{n} is a power of two;
4104 in that case, it is rounded up.
4106 If @var{n} is not specified, use a machine-dependent default.
4108 @item -falign-labels
4109 @itemx -falign-labels=@var{n}
4110 @opindex falign-labels
4111 Align all branch targets to a power-of-two boundary, skipping up to
4112 @var{n} bytes like @option{-falign-functions}. This option can easily
4113 make code slower, because it must insert dummy operations for when the
4114 branch target is reached in the usual flow of the code.
4116 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4117 are greater than this value, then their values are used instead.
4119 If @var{n} is not specified, use a machine-dependent default which is
4120 very likely to be @samp{1}, meaning no alignment.
4123 @itemx -falign-loops=@var{n}
4124 @opindex falign-loops
4125 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4126 like @option{-falign-functions}. The hope is that the loop will be
4127 executed many times, which will make up for any execution of the dummy
4130 If @var{n} is not specified, use a machine-dependent default.
4133 @itemx -falign-jumps=@var{n}
4134 @opindex falign-jumps
4135 Align branch targets to a power-of-two boundary, for branch targets
4136 where the targets can only be reached by jumping, skipping up to @var{n}
4137 bytes like @option{-falign-functions}. In this case, no dummy operations
4140 If @var{n} is not specified, use a machine-dependent default.
4144 Perform optimizations in static single assignment form. Each function's
4145 flow graph is translated into SSA form, optimizations are performed, and
4146 the flow graph is translated back from SSA form. Users should not
4147 specify this option, since it is not yet ready for production use.
4151 Perform Sparse Conditional Constant Propagation in SSA form. Requires
4152 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
4156 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
4157 Like @option{-fssa}, this is an experimental feature.
4159 @item -fsingle-precision-constant
4160 @opindex fsingle-precision-constant
4161 Treat floating point constant as single precision constant instead of
4162 implicitly converting it to double precision constant.
4164 @item -frename-registers
4165 @opindex frename-registers
4166 Attempt to avoid false dependencies in scheduled code by making use
4167 of registers left over after register allocation. This optimization
4168 will most benefit processors with lots of registers. It can, however,
4169 make debugging impossible, since variables will no longer stay in
4170 a ``home register''.
4172 @item -fno-cprop-registers
4173 @opindex fno-cprop-registers
4174 After register allocation and post-register allocation instruction splitting,
4175 we perform a copy-propagation pass to try to reduce scheduling dependencies
4176 and occasionally eliminate the copy.
4178 @item --param @var{name}=@var{value}
4180 In some places, GCC uses various constants to control the amount of
4181 optimization that is done. For example, GCC will not inline functions
4182 that contain more that a certain number of instructions. You can
4183 control some of these constants on the command-line using the
4184 @option{--param} option.
4186 In each case, the @var{value} is an integer. The allowable choices for
4187 @var{name} are given in the following table:
4190 @item max-delay-slot-insn-search
4191 The maximum number of instructions to consider when looking for an
4192 instruction to fill a delay slot. If more than this arbitrary number of
4193 instructions is searched, the time savings from filling the delay slot
4194 will be minimal so stop searching. Increasing values mean more
4195 aggressive optimization, making the compile time increase with probably
4196 small improvement in executable run time.
4198 @item max-delay-slot-live-search
4199 When trying to fill delay slots, the maximum number of instructions to
4200 consider when searching for a block with valid live register
4201 information. Increasing this arbitrarily chosen value means more
4202 aggressive optimization, increasing the compile time. This parameter
4203 should be removed when the delay slot code is rewritten to maintain the
4206 @item max-gcse-memory
4207 The approximate maximum amount of memory that will be allocated in
4208 order to perform the global common subexpression elimination
4209 optimization. If more memory than specified is required, the
4210 optimization will not be done.
4212 @item max-gcse-passes
4213 The maximum number of passes of GCSE to run.
4215 @item max-pending-list-length
4216 The maximum number of pending dependencies scheduling will allow
4217 before flushing the current state and starting over. Large functions
4218 with few branches or calls can create excessively large lists which
4219 needlessly consume memory and resources.
4221 @item max-inline-insns
4222 If an function contains more than this many instructions, it
4223 will not be inlined. This option is precisely equivalent to
4224 @option{-finline-limit}.
4226 @item max-unrolled-insns
4227 The maximum number of instructions that a loop should have if that loop
4228 is unrolled, and if the loop is unrolled, it determines how many times
4229 the loop code is unrolled.
4231 @item hot-bb-count-fraction
4232 Select fraction of the maximal count of repetitions of basic block in program
4233 given basic block needs to have to be considered hot.
4235 @item hot-bb-frequency-fraction
4236 Select fraction of the maximal frequency of executions of basic block in
4237 function given basic block needs to have to be considered hot
4239 @item tracer-dynamic-coverage
4240 @itemx tracer-dynamic-coverage-feedback
4242 This value is used to limit superblock formation once given percentage of
4243 executed instructions is covered. This limits unnecesary code size expansion.
4245 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4246 feedback is available. The real profiles (as opposed to statically estimated
4247 ones) are much less balanced allowing the threshold to be larger value.
4249 @item tracer-max-code-growth
4250 Stop tail duplication once code growth has reached given percentage. This is
4251 rather hokey argument, as most of the duplicates will be elliminated later in
4252 cross jumping, so it may be set to much higher values than is the desired code
4255 @item tracer-min-branch-ratio
4257 Stop reverse growth when the reverse probability of best edge is less than this
4258 threshold (in percent).
4260 @item tracer-min-branch-ratio
4261 @itemx tracer-min-branch-ratio-feedback
4263 Stop forward growth if the best edge do have probability lower than this
4266 Similary to @option{tracer-dynamic-coverage} two values are present, one for
4267 compilation for profile feedback and one for compilation without. The value
4268 for compilation with profile feedback needs to be more conservative (higher) in
4269 order to make tracer effective.
4271 @item ggc-min-expand
4273 GCC uses a garbage collector to manage its own memory allocation. This
4274 parameter specifies the minimum percentage by which the garbage
4275 collector's heap should be allowed to expand between collections.
4276 Tuning this may improve compilation speed; it has no effect on code
4279 The default is 30%. Setting this parameter and
4280 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4281 every opportunity. This is extremely slow, but can be useful for
4284 @item ggc-min-heapsize
4286 Minimum size of the garbage collector's heap before it begins bothering
4287 to collect garbage. The first collection occurs after the heap expands
4288 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4289 tuning this may improve compilation speed, and has no effect on code
4292 The default is 4096 (four megabytes). Setting this parameter very large
4293 effectively disables garbage collection. Setting this parameter and
4294 @option{ggc-min-expand} to zero causes a full collection to occur at
4300 @node Preprocessor Options
4301 @section Options Controlling the Preprocessor
4302 @cindex preprocessor options
4303 @cindex options, preprocessor
4305 These options control the C preprocessor, which is run on each C source
4306 file before actual compilation.
4308 If you use the @option{-E} option, nothing is done except preprocessing.
4309 Some of these options make sense only together with @option{-E} because
4310 they cause the preprocessor output to be unsuitable for actual
4314 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4315 and pass @var{option} directly through to the preprocessor. If
4316 @var{option} contains commas, it is split into multiple options at the
4317 commas. However, many options are modified, translated or interpreted
4318 by the compiler driver before being passed to the preprocessor, and
4319 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4320 interface is undocumented and subject to change, so whenever possible
4321 you should avoid using @option{-Wp} and let the driver handle the
4324 @include cppopts.texi
4326 @node Assembler Options
4327 @section Passing Options to the Assembler
4329 @c prevent bad page break with this line
4330 You can pass options to the assembler.
4333 @item -Wa,@var{option}
4335 Pass @var{option} as an option to the assembler. If @var{option}
4336 contains commas, it is split into multiple options at the commas.
4340 @section Options for Linking
4341 @cindex link options
4342 @cindex options, linking
4344 These options come into play when the compiler links object files into
4345 an executable output file. They are meaningless if the compiler is
4346 not doing a link step.
4350 @item @var{object-file-name}
4351 A file name that does not end in a special recognized suffix is
4352 considered to name an object file or library. (Object files are
4353 distinguished from libraries by the linker according to the file
4354 contents.) If linking is done, these object files are used as input
4363 If any of these options is used, then the linker is not run, and
4364 object file names should not be used as arguments. @xref{Overall
4368 @item -l@var{library}
4369 @itemx -l @var{library}
4371 Search the library named @var{library} when linking. (The second
4372 alternative with the library as a separate argument is only for
4373 POSIX compliance and is not recommended.)
4375 It makes a difference where in the command you write this option; the
4376 linker searches and processes libraries and object files in the order they
4377 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4378 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4379 to functions in @samp{z}, those functions may not be loaded.
4381 The linker searches a standard list of directories for the library,
4382 which is actually a file named @file{lib@var{library}.a}. The linker
4383 then uses this file as if it had been specified precisely by name.
4385 The directories searched include several standard system directories
4386 plus any that you specify with @option{-L}.
4388 Normally the files found this way are library files---archive files
4389 whose members are object files. The linker handles an archive file by
4390 scanning through it for members which define symbols that have so far
4391 been referenced but not defined. But if the file that is found is an
4392 ordinary object file, it is linked in the usual fashion. The only
4393 difference between using an @option{-l} option and specifying a file name
4394 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4395 and searches several directories.
4399 You need this special case of the @option{-l} option in order to
4400 link an Objective-C program.
4403 @opindex nostartfiles
4404 Do not use the standard system startup files when linking.
4405 The standard system libraries are used normally, unless @option{-nostdlib}
4406 or @option{-nodefaultlibs} is used.
4408 @item -nodefaultlibs
4409 @opindex nodefaultlibs
4410 Do not use the standard system libraries when linking.
4411 Only the libraries you specify will be passed to the linker.
4412 The standard startup files are used normally, unless @option{-nostartfiles}
4413 is used. The compiler may generate calls to memcmp, memset, and memcpy
4414 for System V (and ISO C) environments or to bcopy and bzero for
4415 BSD environments. These entries are usually resolved by entries in
4416 libc. These entry points should be supplied through some other
4417 mechanism when this option is specified.
4421 Do not use the standard system startup files or libraries when linking.
4422 No startup files and only the libraries you specify will be passed to
4423 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4424 for System V (and ISO C) environments or to bcopy and bzero for
4425 BSD environments. These entries are usually resolved by entries in
4426 libc. These entry points should be supplied through some other
4427 mechanism when this option is specified.
4429 @cindex @option{-lgcc}, use with @option{-nostdlib}
4430 @cindex @option{-nostdlib} and unresolved references
4431 @cindex unresolved references and @option{-nostdlib}
4432 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4433 @cindex @option{-nodefaultlibs} and unresolved references
4434 @cindex unresolved references and @option{-nodefaultlibs}
4435 One of the standard libraries bypassed by @option{-nostdlib} and
4436 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4437 that GCC uses to overcome shortcomings of particular machines, or special
4438 needs for some languages.
4439 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
4440 Collection (GCC) Internals},
4441 for more discussion of @file{libgcc.a}.)
4442 In most cases, you need @file{libgcc.a} even when you want to avoid
4443 other standard libraries. In other words, when you specify @option{-nostdlib}
4444 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4445 This ensures that you have no unresolved references to internal GCC
4446 library subroutines. (For example, @samp{__main}, used to ensure C++
4447 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
4448 GNU Compiler Collection (GCC) Internals}.)
4452 Remove all symbol table and relocation information from the executable.
4456 On systems that support dynamic linking, this prevents linking with the shared
4457 libraries. On other systems, this option has no effect.
4461 Produce a shared object which can then be linked with other objects to
4462 form an executable. Not all systems support this option. For predictable
4463 results, you must also specify the same set of options that were used to
4464 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4465 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4466 needs to build supplementary stub code for constructors to work. On
4467 multi-libbed systems, @samp{gcc -shared} must select the correct support
4468 libraries to link against. Failing to supply the correct flags may lead
4469 to subtle defects. Supplying them in cases where they are not necessary
4472 @item -shared-libgcc
4473 @itemx -static-libgcc
4474 @opindex shared-libgcc
4475 @opindex static-libgcc
4476 On systems that provide @file{libgcc} as a shared library, these options
4477 force the use of either the shared or static version respectively.
4478 If no shared version of @file{libgcc} was built when the compiler was
4479 configured, these options have no effect.
4481 There are several situations in which an application should use the
4482 shared @file{libgcc} instead of the static version. The most common
4483 of these is when the application wishes to throw and catch exceptions
4484 across different shared libraries. In that case, each of the libraries
4485 as well as the application itself should use the shared @file{libgcc}.
4487 Therefore, the G++ and GCJ drivers automatically add
4488 @option{-shared-libgcc} whenever you build a shared library or a main
4489 executable, because C++ and Java programs typically use exceptions, so
4490 this is the right thing to do.
4492 If, instead, you use the GCC driver to create shared libraries, you may
4493 find that they will not always be linked with the shared @file{libgcc}.
4494 If GCC finds, at its configuration time, that you have a GNU linker that
4495 does not support option @option{--eh-frame-hdr}, it will link the shared
4496 version of @file{libgcc} into shared libraries by default. Otherwise,
4497 it will take advantage of the linker and optimize away the linking with
4498 the shared version of @file{libgcc}, linking with the static version of
4499 libgcc by default. This allows exceptions to propagate through such
4500 shared libraries, without incurring relocation costs at library load
4503 However, if a library or main executable is supposed to throw or catch
4504 exceptions, you must link it using the G++ or GCJ driver, as appropriate
4505 for the languages used in the program, or using the option
4506 @option{-shared-libgcc}, such that it is linked with the shared
4511 Bind references to global symbols when building a shared object. Warn
4512 about any unresolved references (unless overridden by the link editor
4513 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4516 @item -Xlinker @var{option}
4518 Pass @var{option} as an option to the linker. You can use this to
4519 supply system-specific linker options which GCC does not know how to
4522 If you want to pass an option that takes an argument, you must use
4523 @option{-Xlinker} twice, once for the option and once for the argument.
4524 For example, to pass @option{-assert definitions}, you must write
4525 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4526 @option{-Xlinker "-assert definitions"}, because this passes the entire
4527 string as a single argument, which is not what the linker expects.
4529 @item -Wl,@var{option}
4531 Pass @var{option} as an option to the linker. If @var{option} contains
4532 commas, it is split into multiple options at the commas.
4534 @item -u @var{symbol}
4536 Pretend the symbol @var{symbol} is undefined, to force linking of
4537 library modules to define it. You can use @option{-u} multiple times with
4538 different symbols to force loading of additional library modules.
4541 @node Directory Options
4542 @section Options for Directory Search
4543 @cindex directory options
4544 @cindex options, directory search
4547 These options specify directories to search for header files, for
4548 libraries and for parts of the compiler:
4553 Add the directory @var{dir} to the head of the list of directories to be
4554 searched for header files. This can be used to override a system header
4555 file, substituting your own version, since these directories are
4556 searched before the system header file directories. However, you should
4557 not use this option to add directories that contain vendor-supplied
4558 system header files (use @option{-isystem} for that). If you use more than
4559 one @option{-I} option, the directories are scanned in left-to-right
4560 order; the standard system directories come after.
4562 If a standard system include directory, or a directory specified with
4563 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
4564 option will be ignored. The directory will still be searched but as a
4565 system directory at its normal position in the system include chain.
4566 This is to ensure that GCC's procedure to fix buggy system headers and
4567 the ordering for the include_next directive are not inadvertantly changed.
4568 If you really need to change the search order for system directories,
4569 use the @option{-nostdinc} and/or @option{-isystem} options.
4573 Any directories you specify with @option{-I} options before the @option{-I-}
4574 option are searched only for the case of @samp{#include "@var{file}"};
4575 they are not searched for @samp{#include <@var{file}>}.
4577 If additional directories are specified with @option{-I} options after
4578 the @option{-I-}, these directories are searched for all @samp{#include}
4579 directives. (Ordinarily @emph{all} @option{-I} directories are used
4582 In addition, the @option{-I-} option inhibits the use of the current
4583 directory (where the current input file came from) as the first search
4584 directory for @samp{#include "@var{file}"}. There is no way to
4585 override this effect of @option{-I-}. With @option{-I.} you can specify
4586 searching the directory which was current when the compiler was
4587 invoked. That is not exactly the same as what the preprocessor does
4588 by default, but it is often satisfactory.
4590 @option{-I-} does not inhibit the use of the standard system directories
4591 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4596 Add directory @var{dir} to the list of directories to be searched
4599 @item -B@var{prefix}
4601 This option specifies where to find the executables, libraries,
4602 include files, and data files of the compiler itself.
4604 The compiler driver program runs one or more of the subprograms
4605 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4606 @var{prefix} as a prefix for each program it tries to run, both with and
4607 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4609 For each subprogram to be run, the compiler driver first tries the
4610 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4611 was not specified, the driver tries two standard prefixes, which are
4612 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4613 those results in a file name that is found, the unmodified program
4614 name is searched for using the directories specified in your
4615 @env{PATH} environment variable.
4617 The compiler will check to see if the path provided by the @option{-B}
4618 refers to a directory, and if necessary it will add a directory
4619 separator character at the end of the path.
4621 @option{-B} prefixes that effectively specify directory names also apply
4622 to libraries in the linker, because the compiler translates these
4623 options into @option{-L} options for the linker. They also apply to
4624 includes files in the preprocessor, because the compiler translates these
4625 options into @option{-isystem} options for the preprocessor. In this case,
4626 the compiler appends @samp{include} to the prefix.
4628 The run-time support file @file{libgcc.a} can also be searched for using
4629 the @option{-B} prefix, if needed. If it is not found there, the two
4630 standard prefixes above are tried, and that is all. The file is left
4631 out of the link if it is not found by those means.
4633 Another way to specify a prefix much like the @option{-B} prefix is to use
4634 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4637 As a special kludge, if the path provided by @option{-B} is
4638 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4639 9, then it will be replaced by @file{[dir/]include}. This is to help
4640 with boot-strapping the compiler.
4642 @item -specs=@var{file}
4644 Process @var{file} after the compiler reads in the standard @file{specs}
4645 file, in order to override the defaults that the @file{gcc} driver
4646 program uses when determining what switches to pass to @file{cc1},
4647 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4648 @option{-specs=@var{file}} can be specified on the command line, and they
4649 are processed in order, from left to right.
4655 @section Specifying subprocesses and the switches to pass to them
4657 @command{gcc} is a driver program. It performs its job by invoking a
4658 sequence of other programs to do the work of compiling, assembling and
4659 linking. GCC interprets its command-line parameters and uses these to
4660 deduce which programs it should invoke, and which command-line options
4661 it ought to place on their command lines. This behavior is controlled
4662 by @dfn{spec strings}. In most cases there is one spec string for each
4663 program that GCC can invoke, but a few programs have multiple spec
4664 strings to control their behavior. The spec strings built into GCC can
4665 be overridden by using the @option{-specs=} command-line switch to specify
4668 @dfn{Spec files} are plaintext files that are used to construct spec
4669 strings. They consist of a sequence of directives separated by blank
4670 lines. The type of directive is determined by the first non-whitespace
4671 character on the line and it can be one of the following:
4674 @item %@var{command}
4675 Issues a @var{command} to the spec file processor. The commands that can
4679 @item %include <@var{file}>
4681 Search for @var{file} and insert its text at the current point in the
4684 @item %include_noerr <@var{file}>
4685 @cindex %include_noerr
4686 Just like @samp{%include}, but do not generate an error message if the include
4687 file cannot be found.
4689 @item %rename @var{old_name} @var{new_name}
4691 Rename the spec string @var{old_name} to @var{new_name}.
4695 @item *[@var{spec_name}]:
4696 This tells the compiler to create, override or delete the named spec
4697 string. All lines after this directive up to the next directive or
4698 blank line are considered to be the text for the spec string. If this
4699 results in an empty string then the spec will be deleted. (Or, if the
4700 spec did not exist, then nothing will happened.) Otherwise, if the spec
4701 does not currently exist a new spec will be created. If the spec does
4702 exist then its contents will be overridden by the text of this
4703 directive, unless the first character of that text is the @samp{+}
4704 character, in which case the text will be appended to the spec.
4706 @item [@var{suffix}]:
4707 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4708 and up to the next directive or blank line are considered to make up the
4709 spec string for the indicated suffix. When the compiler encounters an
4710 input file with the named suffix, it will processes the spec string in
4711 order to work out how to compile that file. For example:
4718 This says that any input file whose name ends in @samp{.ZZ} should be
4719 passed to the program @samp{z-compile}, which should be invoked with the
4720 command-line switch @option{-input} and with the result of performing the
4721 @samp{%i} substitution. (See below.)
4723 As an alternative to providing a spec string, the text that follows a
4724 suffix directive can be one of the following:
4727 @item @@@var{language}
4728 This says that the suffix is an alias for a known @var{language}. This is
4729 similar to using the @option{-x} command-line switch to GCC to specify a
4730 language explicitly. For example:
4737 Says that .ZZ files are, in fact, C++ source files.
4740 This causes an error messages saying:
4743 @var{name} compiler not installed on this system.
4747 GCC already has an extensive list of suffixes built into it.
4748 This directive will add an entry to the end of the list of suffixes, but
4749 since the list is searched from the end backwards, it is effectively
4750 possible to override earlier entries using this technique.
4754 GCC has the following spec strings built into it. Spec files can
4755 override these strings or create their own. Note that individual
4756 targets can also add their own spec strings to this list.
4759 asm Options to pass to the assembler
4760 asm_final Options to pass to the assembler post-processor
4761 cpp Options to pass to the C preprocessor
4762 cc1 Options to pass to the C compiler
4763 cc1plus Options to pass to the C++ compiler
4764 endfile Object files to include at the end of the link
4765 link Options to pass to the linker
4766 lib Libraries to include on the command line to the linker
4767 libgcc Decides which GCC support library to pass to the linker
4768 linker Sets the name of the linker
4769 predefines Defines to be passed to the C preprocessor
4770 signed_char Defines to pass to CPP to say whether @code{char} is signed
4772 startfile Object files to include at the start of the link
4775 Here is a small example of a spec file:
4781 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4784 This example renames the spec called @samp{lib} to @samp{old_lib} and
4785 then overrides the previous definition of @samp{lib} with a new one.
4786 The new definition adds in some extra command-line options before
4787 including the text of the old definition.
4789 @dfn{Spec strings} are a list of command-line options to be passed to their
4790 corresponding program. In addition, the spec strings can contain
4791 @samp{%}-prefixed sequences to substitute variable text or to
4792 conditionally insert text into the command line. Using these constructs
4793 it is possible to generate quite complex command lines.
4795 Here is a table of all defined @samp{%}-sequences for spec
4796 strings. Note that spaces are not generated automatically around the
4797 results of expanding these sequences. Therefore you can concatenate them
4798 together or combine them with constant text in a single argument.
4802 Substitute one @samp{%} into the program name or argument.
4805 Substitute the name of the input file being processed.
4808 Substitute the basename of the input file being processed.
4809 This is the substring up to (and not including) the last period
4810 and not including the directory.
4813 This is the same as @samp{%b}, but include the file suffix (text after
4817 Marks the argument containing or following the @samp{%d} as a
4818 temporary file name, so that that file will be deleted if GCC exits
4819 successfully. Unlike @samp{%g}, this contributes no text to the
4822 @item %g@var{suffix}
4823 Substitute a file name that has suffix @var{suffix} and is chosen
4824 once per compilation, and mark the argument in the same way as
4825 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4826 name is now chosen in a way that is hard to predict even when previously
4827 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
4828 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4829 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4830 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4831 was simply substituted with a file name chosen once per compilation,
4832 without regard to any appended suffix (which was therefore treated
4833 just like ordinary text), making such attacks more likely to succeed.
4835 @item %u@var{suffix}
4836 Like @samp{%g}, but generates a new temporary file name even if
4837 @samp{%u@var{suffix}} was already seen.
4839 @item %U@var{suffix}
4840 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4841 new one if there is no such last file name. In the absence of any
4842 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4843 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
4844 would involve the generation of two distinct file names, one
4845 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4846 simply substituted with a file name chosen for the previous @samp{%u},
4847 without regard to any appended suffix.
4849 @item %j@var{SUFFIX}
4850 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
4851 writable, and if save-temps is off; otherwise, substitute the name
4852 of a temporary file, just like @samp{%u}. This temporary file is not
4853 meant for communication between processes, but rather as a junk
4856 @item %.@var{SUFFIX}
4857 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4858 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4859 terminated by the next space or %.
4862 Marks the argument containing or following the @samp{%w} as the
4863 designated output file of this compilation. This puts the argument
4864 into the sequence of arguments that @samp{%o} will substitute later.
4867 Substitutes the names of all the output files, with spaces
4868 automatically placed around them. You should write spaces
4869 around the @samp{%o} as well or the results are undefined.
4870 @samp{%o} is for use in the specs for running the linker.
4871 Input files whose names have no recognized suffix are not compiled
4872 at all, but they are included among the output files, so they will
4876 Substitutes the suffix for object files. Note that this is
4877 handled specially when it immediately follows @samp{%g, %u, or %U},
4878 because of the need for those to form complete file names. The
4879 handling is such that @samp{%O} is treated exactly as if it had already
4880 been substituted, except that @samp{%g, %u, and %U} do not currently
4881 support additional @var{suffix} characters following @samp{%O} as they would
4882 following, for example, @samp{.o}.
4885 Substitutes the standard macro predefinitions for the
4886 current target machine. Use this when running @code{cpp}.
4889 Like @samp{%p}, but puts @samp{__} before and after the name of each
4890 predefined macro, except for macros that start with @samp{__} or with
4891 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4895 Substitute a @option{-iprefix} option made from @env{GCC_EXEC_PREFIX}.
4898 Current argument is the name of a library or startup file of some sort.
4899 Search for that file in a standard list of directories and substitute
4900 the full name found.
4903 Print @var{str} as an error message. @var{str} is terminated by a newline.
4904 Use this when inconsistent options are detected.
4907 Output @samp{-} if the input for the current command is coming from a pipe.
4910 Substitute the contents of spec string @var{name} at this point.
4913 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
4915 @item %x@{@var{option}@}
4916 Accumulate an option for @samp{%X}.
4919 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
4923 Output the accumulated assembler options specified by @option{-Wa}.
4926 Output the accumulated preprocessor options specified by @option{-Wp}.
4929 Substitute the major version number of GCC@.
4930 (For version 2.9.5, this is 2.)
4933 Substitute the minor version number of GCC@.
4934 (For version 2.9.5, this is 9.)
4937 Substitute the patch level number of GCC@.
4938 (For version 2.9.5, this is 5.)
4941 Process the @code{asm} spec. This is used to compute the
4942 switches to be passed to the assembler.
4945 Process the @code{asm_final} spec. This is a spec string for
4946 passing switches to an assembler post-processor, if such a program is
4950 Process the @code{link} spec. This is the spec for computing the
4951 command line passed to the linker. Typically it will make use of the
4952 @samp{%L %G %S %D and %E} sequences.
4955 Dump out a @option{-L} option for each directory that GCC believes might
4956 contain startup files. If the target supports multilibs then the
4957 current multilib directory will be prepended to each of these paths.
4960 Output the multilib directory with directory separators replaced with
4961 @samp{_}. If multilib directories are not set, or the multilib directory is
4962 @file{.} then this option emits nothing.
4965 Process the @code{lib} spec. This is a spec string for deciding which
4966 libraries should be included on the command line to the linker.
4969 Process the @code{libgcc} spec. This is a spec string for deciding
4970 which GCC support library should be included on the command line to the linker.
4973 Process the @code{startfile} spec. This is a spec for deciding which
4974 object files should be the first ones passed to the linker. Typically
4975 this might be a file named @file{crt0.o}.
4978 Process the @code{endfile} spec. This is a spec string that specifies
4979 the last object files that will be passed to the linker.
4982 Process the @code{cpp} spec. This is used to construct the arguments
4983 to be passed to the C preprocessor.
4986 Process the @code{signed_char} spec. This is intended to be used
4987 to tell cpp whether a char is signed. It typically has the definition:
4989 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4993 Process the @code{cc1} spec. This is used to construct the options to be
4994 passed to the actual C compiler (@samp{cc1}).
4997 Process the @code{cc1plus} spec. This is used to construct the options to be
4998 passed to the actual C++ compiler (@samp{cc1plus}).
5001 Substitute the variable part of a matched option. See below.
5002 Note that each comma in the substituted string is replaced by
5005 @item %:@var{function}(@var{args})
5006 Call the named function @var{function}, passing it @var{args}.
5007 @var{args} is first processed as a nested spec string, then split
5008 into an argument vector in the usual fashion. The function returns
5009 a string which is processed as if it had appeared literally as part
5010 of the current spec.
5012 The following built-in spec functions are provided:
5015 @item @code{if-exists}
5016 The @code{if-exists} spec function takes one argument, an absolute
5017 pathname to a file. If the file exists, @code{if-exists} returns the
5018 pathname. Here is a small example of its usage:
5022 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5025 @item @code{if-exists-else}
5026 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5027 spec function, except that it takes two arguments. The first argument is
5028 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5029 returns the pathname. If it does not exist, it returns the second argument.
5030 This way, @code{if-exists-else} can be used to select one file or another,
5031 based on the existence of the first. Here is a small example of its usage:
5035 crt0%O%s %:if-exists(crti%O%s) %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5040 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5041 If that switch was not specified, this substitutes nothing. Note that
5042 the leading dash is omitted when specifying this option, and it is
5043 automatically inserted if the substitution is performed. Thus the spec
5044 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5045 and would output the command line option @option{-foo}.
5047 @item %W@{@code{S}@}
5048 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5051 @item %@{@code{S}*@}
5052 Substitutes all the switches specified to GCC whose names start
5053 with @code{-S}, but which also take an argument. This is used for
5054 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5055 GCC considers @option{-o foo} as being
5056 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5057 text, including the space. Thus two arguments would be generated.
5059 @item %@{^@code{S}*@}
5060 Like %@{@code{S}*@}, but don't put a blank between a switch and its
5061 argument. Thus %@{^o*@} would only generate one argument, not two.
5063 @item %@{@code{S}*&@code{T}*@}
5064 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5065 (the order of @code{S} and @code{T} in the spec is not significant).
5066 There can be any number of ampersand-separated variables; for each the
5067 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5069 @item %@{<@code{S}@}
5070 Remove all occurrences of @code{-S} from the command line. Note---this
5071 command is position dependent. @samp{%} commands in the spec string
5072 before this option will see @code{-S}, @samp{%} commands in the spec
5073 string after this option will not.
5075 @item %@{@code{S}*:@code{X}@}
5076 Substitutes @code{X} if one or more switches whose names start with
5077 @code{-S} are specified to GCC@. Note that the tail part of the
5078 @code{-S} option (i.e.@: the part matched by the @samp{*}) will be substituted
5079 for each occurrence of @samp{%*} within @code{X}.
5081 @item %@{@code{S}:@code{X}@}
5082 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC@.
5084 @item %@{!@code{S}:@code{X}@}
5085 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC@.
5087 @item %@{|@code{S}:@code{X}@}
5088 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
5090 @item %@{|!@code{S}:@code{X}@}
5091 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
5093 @item %@{.@code{S}:@code{X}@}
5094 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
5096 @item %@{!.@code{S}:@code{X}@}
5097 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
5099 @item %@{@code{S}|@code{P}:@code{X}@}
5100 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@. This may be
5101 combined with @samp{!} and @samp{.} sequences as well, although they
5102 have a stronger binding than the @samp{|}. For example a spec string
5106 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5109 will output the following command-line options from the following input
5110 command-line options:
5115 -d fred.c -foo -baz -boggle
5116 -d jim.d -bar -baz -boggle
5121 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
5122 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
5123 or spaces, or even newlines. They are processed as usual, as described
5126 The @option{-O}, @option{-f}, @option{-m}, and @option{-W}
5127 switches are handled specifically in these
5128 constructs. If another value of @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5129 @option{-W} switch is found later in the command line, the earlier switch
5130 value is ignored, except with @{@code{S}*@} where @code{S} is just one
5131 letter, which passes all matching options.
5133 The character @samp{|} at the beginning of the predicate text is used to indicate
5134 that a command should be piped to the following command, but only if @option{-pipe}
5137 It is built into GCC which switches take arguments and which do not.
5138 (You might think it would be useful to generalize this to allow each
5139 compiler's spec to say which switches take arguments. But this cannot
5140 be done in a consistent fashion. GCC cannot even decide which input
5141 files have been specified without knowing which switches take arguments,
5142 and it must know which input files to compile in order to tell which
5145 GCC also knows implicitly that arguments starting in @option{-l} are to be
5146 treated as compiler output files, and passed to the linker in their
5147 proper position among the other output files.
5149 @c man begin OPTIONS
5151 @node Target Options
5152 @section Specifying Target Machine and Compiler Version
5153 @cindex target options
5154 @cindex cross compiling
5155 @cindex specifying machine version
5156 @cindex specifying compiler version and target machine
5157 @cindex compiler version, specifying
5158 @cindex target machine, specifying
5160 The usual way to run GCC is to run the executable called @file{gcc}, or
5161 @file{<machine>-gcc} when cross-compiling, or
5162 @file{<machine>-gcc-<version>} to run a version other than the one that
5163 was installed last. Sometimes this is inconvenient, so GCC provides
5164 options that will switch to another cross-compiler or version.
5167 @item -b @var{machine}
5169 The argument @var{machine} specifies the target machine for compilation.
5171 The value to use for @var{machine} is the same as was specified as the
5172 machine type when configuring GCC as a cross-compiler. For
5173 example, if a cross-compiler was configured with @samp{configure
5174 i386v}, meaning to compile for an 80386 running System V, then you
5175 would specify @option{-b i386v} to run that cross compiler.
5177 @item -V @var{version}
5179 The argument @var{version} specifies which version of GCC to run.
5180 This is useful when multiple versions are installed. For example,
5181 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5184 The @option{-V} and @option{-b} options work by running the
5185 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5186 use them if you can just run that directly.
5188 @node Submodel Options
5189 @section Hardware Models and Configurations
5190 @cindex submodel options
5191 @cindex specifying hardware config
5192 @cindex hardware models and configurations, specifying
5193 @cindex machine dependent options
5195 Earlier we discussed the standard option @option{-b} which chooses among
5196 different installed compilers for completely different target
5197 machines, such as VAX vs.@: 68000 vs.@: 80386.
5199 In addition, each of these target machine types can have its own
5200 special options, starting with @samp{-m}, to choose among various
5201 hardware models or configurations---for example, 68010 vs 68020,
5202 floating coprocessor or none. A single installed version of the
5203 compiler can compile for any model or configuration, according to the
5206 Some configurations of the compiler also support additional special
5207 options, usually for compatibility with other compilers on the same
5210 These options are defined by the macro @code{TARGET_SWITCHES} in the
5211 machine description. The default for the options is also defined by
5212 that macro, which enables you to change the defaults.
5226 * RS/6000 and PowerPC Options::
5230 * i386 and x86-64 Options::
5232 * Intel 960 Options::
5233 * DEC Alpha Options::
5234 * DEC Alpha/VMS Options::
5238 * System V Options::
5239 * TMS320C3x/C4x Options::
5247 * S/390 and zSeries Options::
5251 * Xstormy16 Options::
5256 @node M680x0 Options
5257 @subsection M680x0 Options
5258 @cindex M680x0 options
5260 These are the @samp{-m} options defined for the 68000 series. The default
5261 values for these options depends on which style of 68000 was selected when
5262 the compiler was configured; the defaults for the most common choices are
5270 Generate output for a 68000. This is the default
5271 when the compiler is configured for 68000-based systems.
5273 Use this option for microcontrollers with a 68000 or EC000 core,
5274 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5280 Generate output for a 68020. This is the default
5281 when the compiler is configured for 68020-based systems.
5285 Generate output containing 68881 instructions for floating point.
5286 This is the default for most 68020 systems unless @option{--nfp} was
5287 specified when the compiler was configured.
5291 Generate output for a 68030. This is the default when the compiler is
5292 configured for 68030-based systems.
5296 Generate output for a 68040. This is the default when the compiler is
5297 configured for 68040-based systems.
5299 This option inhibits the use of 68881/68882 instructions that have to be
5300 emulated by software on the 68040. Use this option if your 68040 does not
5301 have code to emulate those instructions.
5305 Generate output for a 68060. This is the default when the compiler is
5306 configured for 68060-based systems.
5308 This option inhibits the use of 68020 and 68881/68882 instructions that
5309 have to be emulated by software on the 68060. Use this option if your 68060
5310 does not have code to emulate those instructions.
5314 Generate output for a CPU32. This is the default
5315 when the compiler is configured for CPU32-based systems.
5317 Use this option for microcontrollers with a
5318 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5319 68336, 68340, 68341, 68349 and 68360.
5323 Generate output for a 520X ``coldfire'' family cpu. This is the default
5324 when the compiler is configured for 520X-based systems.
5326 Use this option for microcontroller with a 5200 core, including
5327 the MCF5202, MCF5203, MCF5204 and MCF5202.
5332 Generate output for a 68040, without using any of the new instructions.
5333 This results in code which can run relatively efficiently on either a
5334 68020/68881 or a 68030 or a 68040. The generated code does use the
5335 68881 instructions that are emulated on the 68040.
5339 Generate output for a 68060, without using any of the new instructions.
5340 This results in code which can run relatively efficiently on either a
5341 68020/68881 or a 68030 or a 68040. The generated code does use the
5342 68881 instructions that are emulated on the 68060.
5346 Generate output containing Sun FPA instructions for floating point.
5349 @opindex msoft-float
5350 Generate output containing library calls for floating point.
5351 @strong{Warning:} the requisite libraries are not available for all m68k
5352 targets. Normally the facilities of the machine's usual C compiler are
5353 used, but this can't be done directly in cross-compilation. You must
5354 make your own arrangements to provide suitable library functions for
5355 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5356 @samp{m68k-*-coff} do provide software floating point support.
5360 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5363 @opindex mnobitfield
5364 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5365 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5369 Do use the bit-field instructions. The @option{-m68020} option implies
5370 @option{-mbitfield}. This is the default if you use a configuration
5371 designed for a 68020.
5375 Use a different function-calling convention, in which functions
5376 that take a fixed number of arguments return with the @code{rtd}
5377 instruction, which pops their arguments while returning. This
5378 saves one instruction in the caller since there is no need to pop
5379 the arguments there.
5381 This calling convention is incompatible with the one normally
5382 used on Unix, so you cannot use it if you need to call libraries
5383 compiled with the Unix compiler.
5385 Also, you must provide function prototypes for all functions that
5386 take variable numbers of arguments (including @code{printf});
5387 otherwise incorrect code will be generated for calls to those
5390 In addition, seriously incorrect code will result if you call a
5391 function with too many arguments. (Normally, extra arguments are
5392 harmlessly ignored.)
5394 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5395 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5398 @itemx -mno-align-int
5400 @opindex mno-align-int
5401 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5402 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5403 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5404 Aligning variables on 32-bit boundaries produces code that runs somewhat
5405 faster on processors with 32-bit busses at the expense of more memory.
5407 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5408 align structures containing the above types differently than
5409 most published application binary interface specifications for the m68k.
5413 Use the pc-relative addressing mode of the 68000 directly, instead of
5414 using a global offset table. At present, this option implies @option{-fpic},
5415 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5416 not presently supported with @option{-mpcrel}, though this could be supported for
5417 68020 and higher processors.
5419 @item -mno-strict-align
5420 @itemx -mstrict-align
5421 @opindex mno-strict-align
5422 @opindex mstrict-align
5423 Do not (do) assume that unaligned memory references will be handled by
5428 @node M68hc1x Options
5429 @subsection M68hc1x Options
5430 @cindex M68hc1x options
5432 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5433 microcontrollers. The default values for these options depends on
5434 which style of microcontroller was selected when the compiler was configured;
5435 the defaults for the most common choices are given below.
5442 Generate output for a 68HC11. This is the default
5443 when the compiler is configured for 68HC11-based systems.
5449 Generate output for a 68HC12. This is the default
5450 when the compiler is configured for 68HC12-based systems.
5453 @opindex mauto-incdec
5454 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5461 Enable the use of 68HC12 min and max instructions.
5464 @itemx -mno-long-calls
5465 @opindex mlong-calls
5466 @opindex mno-long-calls
5467 Treat all calls as being far away (near). If calls are assumed to be
5468 far away, the compiler will use the @code{call} instruction to
5469 call a function and the @code{rtc} instruction for returning.
5473 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5475 @item -msoft-reg-count=@var{count}
5476 @opindex msoft-reg-count
5477 Specify the number of pseudo-soft registers which are used for the
5478 code generation. The maximum number is 32. Using more pseudo-soft
5479 register may or may not result in better code depending on the program.
5480 The default is 4 for 68HC11 and 2 for 68HC12.
5485 @subsection VAX Options
5488 These @samp{-m} options are defined for the VAX:
5493 Do not output certain jump instructions (@code{aobleq} and so on)
5494 that the Unix assembler for the VAX cannot handle across long
5499 Do output those jump instructions, on the assumption that you
5500 will assemble with the GNU assembler.
5504 Output code for g-format floating point numbers instead of d-format.
5508 @subsection SPARC Options
5509 @cindex SPARC options
5511 These @samp{-m} switches are supported on the SPARC:
5516 @opindex mno-app-regs
5518 Specify @option{-mapp-regs} to generate output using the global registers
5519 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5522 To be fully SVR4 ABI compliant at the cost of some performance loss,
5523 specify @option{-mno-app-regs}. You should compile libraries and system
5524 software with this option.
5529 @opindex mhard-float
5530 Generate output containing floating point instructions. This is the
5536 @opindex msoft-float
5537 Generate output containing library calls for floating point.
5538 @strong{Warning:} the requisite libraries are not available for all SPARC
5539 targets. Normally the facilities of the machine's usual C compiler are
5540 used, but this cannot be done directly in cross-compilation. You must make
5541 your own arrangements to provide suitable library functions for
5542 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5543 @samp{sparclite-*-*} do provide software floating point support.
5545 @option{-msoft-float} changes the calling convention in the output file;
5546 therefore, it is only useful if you compile @emph{all} of a program with
5547 this option. In particular, you need to compile @file{libgcc.a}, the
5548 library that comes with GCC, with @option{-msoft-float} in order for
5551 @item -mhard-quad-float
5552 @opindex mhard-quad-float
5553 Generate output containing quad-word (long double) floating point
5556 @item -msoft-quad-float
5557 @opindex msoft-quad-float
5558 Generate output containing library calls for quad-word (long double)
5559 floating point instructions. The functions called are those specified
5560 in the SPARC ABI@. This is the default.
5562 As of this writing, there are no sparc implementations that have hardware
5563 support for the quad-word floating point instructions. They all invoke
5564 a trap handler for one of these instructions, and then the trap handler
5565 emulates the effect of the instruction. Because of the trap handler overhead,
5566 this is much slower than calling the ABI library routines. Thus the
5567 @option{-msoft-quad-float} option is the default.
5573 With @option{-mflat}, the compiler does not generate save/restore instructions
5574 and will use a ``flat'' or single register window calling convention.
5575 This model uses %i7 as the frame pointer and is compatible with the normal
5576 register window model. Code from either may be intermixed.
5577 The local registers and the input registers (0--5) are still treated as
5578 ``call saved'' registers and will be saved on the stack as necessary.
5580 With @option{-mno-flat} (the default), the compiler emits save/restore
5581 instructions (except for leaf functions) and is the normal mode of operation.
5583 @item -mno-unaligned-doubles
5584 @itemx -munaligned-doubles
5585 @opindex mno-unaligned-doubles
5586 @opindex munaligned-doubles
5587 Assume that doubles have 8 byte alignment. This is the default.
5589 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5590 alignment only if they are contained in another type, or if they have an
5591 absolute address. Otherwise, it assumes they have 4 byte alignment.
5592 Specifying this option avoids some rare compatibility problems with code
5593 generated by other compilers. It is not the default because it results
5594 in a performance loss, especially for floating point code.
5596 @item -mno-faster-structs
5597 @itemx -mfaster-structs
5598 @opindex mno-faster-structs
5599 @opindex mfaster-structs
5600 With @option{-mfaster-structs}, the compiler assumes that structures
5601 should have 8 byte alignment. This enables the use of pairs of
5602 @code{ldd} and @code{std} instructions for copies in structure
5603 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5604 However, the use of this changed alignment directly violates the SPARC
5605 ABI@. Thus, it's intended only for use on targets where the developer
5606 acknowledges that their resulting code will not be directly in line with
5607 the rules of the ABI@.
5613 These two options select variations on the SPARC architecture.
5615 By default (unless specifically configured for the Fujitsu SPARClite),
5616 GCC generates code for the v7 variant of the SPARC architecture.
5618 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5619 code is that the compiler emits the integer multiply and integer
5620 divide instructions which exist in SPARC v8 but not in SPARC v7.
5622 @option{-msparclite} will give you SPARClite code. This adds the integer
5623 multiply, integer divide step and scan (@code{ffs}) instructions which
5624 exist in SPARClite but not in SPARC v7.
5626 These options are deprecated and will be deleted in a future GCC release.
5627 They have been replaced with @option{-mcpu=xxx}.
5632 @opindex msupersparc
5633 These two options select the processor for which the code is optimized.
5635 With @option{-mcypress} (the default), the compiler optimizes code for the
5636 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
5637 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
5639 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
5640 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
5641 of the full SPARC v8 instruction set.
5643 These options are deprecated and will be deleted in a future GCC release.
5644 They have been replaced with @option{-mcpu=xxx}.
5646 @item -mcpu=@var{cpu_type}
5648 Set the instruction set, register set, and instruction scheduling parameters
5649 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5650 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5651 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5652 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
5655 Default instruction scheduling parameters are used for values that select
5656 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5657 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5659 Here is a list of each supported architecture and their supported
5664 v8: supersparc, hypersparc
5665 sparclite: f930, f934, sparclite86x
5667 v9: ultrasparc, ultrasparc3
5670 @item -mtune=@var{cpu_type}
5672 Set the instruction scheduling parameters for machine type
5673 @var{cpu_type}, but do not set the instruction set or register set that the
5674 option @option{-mcpu=@var{cpu_type}} would.
5676 The same values for @option{-mcpu=@var{cpu_type}} can be used for
5677 @option{-mtune=@var{cpu_type}}, but the only useful values are those
5678 that select a particular cpu implementation. Those are @samp{cypress},
5679 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
5680 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
5685 These @samp{-m} switches are supported in addition to the above
5686 on the SPARCLET processor.
5689 @item -mlittle-endian
5690 @opindex mlittle-endian
5691 Generate code for a processor running in little-endian mode.
5695 Treat register @code{%g0} as a normal register.
5696 GCC will continue to clobber it as necessary but will not assume
5697 it always reads as 0.
5699 @item -mbroken-saverestore
5700 @opindex mbroken-saverestore
5701 Generate code that does not use non-trivial forms of the @code{save} and
5702 @code{restore} instructions. Early versions of the SPARCLET processor do
5703 not correctly handle @code{save} and @code{restore} instructions used with
5704 arguments. They correctly handle them used without arguments. A @code{save}
5705 instruction used without arguments increments the current window pointer
5706 but does not allocate a new stack frame. It is assumed that the window
5707 overflow trap handler will properly handle this case as will interrupt
5711 These @samp{-m} switches are supported in addition to the above
5712 on SPARC V9 processors in 64-bit environments.
5715 @item -mlittle-endian
5716 @opindex mlittle-endian
5717 Generate code for a processor running in little-endian mode.
5723 Generate code for a 32-bit or 64-bit environment.
5724 The 32-bit environment sets int, long and pointer to 32 bits.
5725 The 64-bit environment sets int to 32 bits and long and pointer
5728 @item -mcmodel=medlow
5729 @opindex mcmodel=medlow
5730 Generate code for the Medium/Low code model: the program must be linked
5731 in the low 32 bits of the address space. Pointers are 64 bits.
5732 Programs can be statically or dynamically linked.
5734 @item -mcmodel=medmid
5735 @opindex mcmodel=medmid
5736 Generate code for the Medium/Middle code model: the program must be linked
5737 in the low 44 bits of the address space, the text segment must be less than
5738 2G bytes, and data segment must be within 2G of the text segment.
5739 Pointers are 64 bits.
5741 @item -mcmodel=medany
5742 @opindex mcmodel=medany
5743 Generate code for the Medium/Anywhere code model: the program may be linked
5744 anywhere in the address space, the text segment must be less than
5745 2G bytes, and data segment must be within 2G of the text segment.
5746 Pointers are 64 bits.
5748 @item -mcmodel=embmedany
5749 @opindex mcmodel=embmedany
5750 Generate code for the Medium/Anywhere code model for embedded systems:
5751 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5752 (determined at link time). Register %g4 points to the base of the
5753 data segment. Pointers are still 64 bits.
5754 Programs are statically linked, PIC is not supported.
5757 @itemx -mno-stack-bias
5758 @opindex mstack-bias
5759 @opindex mno-stack-bias
5760 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
5761 frame pointer if present, are offset by @minus{}2047 which must be added back
5762 when making stack frame references.
5763 Otherwise, assume no such offset is present.
5766 @node Convex Options
5767 @subsection Convex Options
5768 @cindex Convex options
5770 These @samp{-m} options are defined for Convex:
5775 Generate output for C1. The code will run on any Convex machine.
5776 The preprocessor symbol @code{__convex__c1__} is defined.
5780 Generate output for C2. Uses instructions not available on C1.
5781 Scheduling and other optimizations are chosen for max performance on C2.
5782 The preprocessor symbol @code{__convex_c2__} is defined.
5786 Generate output for C32xx. Uses instructions not available on C1.
5787 Scheduling and other optimizations are chosen for max performance on C32.
5788 The preprocessor symbol @code{__convex_c32__} is defined.
5792 Generate output for C34xx. Uses instructions not available on C1.
5793 Scheduling and other optimizations are chosen for max performance on C34.
5794 The preprocessor symbol @code{__convex_c34__} is defined.
5798 Generate output for C38xx. Uses instructions not available on C1.
5799 Scheduling and other optimizations are chosen for max performance on C38.
5800 The preprocessor symbol @code{__convex_c38__} is defined.
5804 Generate code which puts an argument count in the word preceding each
5805 argument list. This is compatible with regular CC, and a few programs
5806 may need the argument count word. GDB and other source-level debuggers
5807 do not need it; this info is in the symbol table.
5810 @opindex mnoargcount
5811 Omit the argument count word. This is the default.
5813 @item -mvolatile-cache
5814 @opindex mvolatile-cache
5815 Allow volatile references to be cached. This is the default.
5817 @item -mvolatile-nocache
5818 @opindex mvolatile-nocache
5819 Volatile references bypass the data cache, going all the way to memory.
5820 This is only needed for multi-processor code that does not use standard
5821 synchronization instructions. Making non-volatile references to volatile
5822 locations will not necessarily work.
5826 Type long is 32 bits, the same as type int. This is the default.
5830 Type long is 64 bits, the same as type long long. This option is useless,
5831 because no library support exists for it.
5834 @node AMD29K Options
5835 @subsection AMD29K Options
5836 @cindex AMD29K options
5838 These @samp{-m} options are defined for the AMD Am29000:
5843 @cindex DW bit (29k)
5844 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5845 halfword operations are directly supported by the hardware. This is the
5850 Generate code that assumes the @code{DW} bit is not set.
5854 @cindex byte writes (29k)
5855 Generate code that assumes the system supports byte and halfword write
5856 operations. This is the default.
5860 Generate code that assumes the systems does not support byte and
5861 halfword write operations. @option{-mnbw} implies @option{-mndw}.
5865 @cindex memory model (29k)
5866 Use a small memory model that assumes that all function addresses are
5867 either within a single 256 KB segment or at an absolute address of less
5868 than 256k. This allows the @code{call} instruction to be used instead
5869 of a @code{const}, @code{consth}, @code{calli} sequence.
5873 Use the normal memory model: Generate @code{call} instructions only when
5874 calling functions in the same file and @code{calli} instructions
5875 otherwise. This works if each file occupies less than 256 KB but allows
5876 the entire executable to be larger than 256 KB@. This is the default.
5880 Always use @code{calli} instructions. Specify this option if you expect
5881 a single file to compile into more than 256 KB of code.
5885 @cindex processor selection (29k)
5886 Generate code for the Am29050.
5890 Generate code for the Am29000. This is the default.
5892 @item -mkernel-registers
5893 @opindex mkernel-registers
5894 @cindex kernel and user registers (29k)
5895 Generate references to registers @code{gr64-gr95} instead of to
5896 registers @code{gr96-gr127}. This option can be used when compiling
5897 kernel code that wants a set of global registers disjoint from that used
5900 Note that when this option is used, register names in @samp{-f} flags
5901 must use the normal, user-mode, names.
5903 @item -muser-registers
5904 @opindex muser-registers
5905 Use the normal set of global registers, @code{gr96-gr127}. This is the
5909 @itemx -mno-stack-check
5910 @opindex mstack-check
5911 @opindex mno-stack-check
5912 @cindex stack checks (29k)
5913 Insert (or do not insert) a call to @code{__msp_check} after each stack
5914 adjustment. This is often used for kernel code.
5917 @itemx -mno-storem-bug
5918 @opindex mstorem-bug
5919 @opindex mno-storem-bug
5920 @cindex storem bug (29k)
5921 @option{-mstorem-bug} handles 29k processors which cannot handle the
5922 separation of a mtsrim insn and a storem instruction (most 29000 chips
5923 to date, but not the 29050).
5925 @item -mno-reuse-arg-regs
5926 @itemx -mreuse-arg-regs
5927 @opindex mno-reuse-arg-regs
5928 @opindex mreuse-arg-regs
5929 @option{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5930 registers for copying out arguments. This helps detect calling a function
5931 with fewer arguments than it was declared with.
5933 @item -mno-impure-text
5934 @itemx -mimpure-text
5935 @opindex mno-impure-text
5936 @opindex mimpure-text
5937 @option{-mimpure-text}, used in addition to @option{-shared}, tells the compiler to
5938 not pass @option{-assert pure-text} to the linker when linking a shared object.
5941 @opindex msoft-float
5942 Generate output containing library calls for floating point.
5943 @strong{Warning:} the requisite libraries are not part of GCC@.
5944 Normally the facilities of the machine's usual C compiler are used, but
5945 this can't be done directly in cross-compilation. You must make your
5946 own arrangements to provide suitable library functions for
5951 Do not generate multm or multmu instructions. This is useful for some embedded
5952 systems which do not have trap handlers for these instructions.
5956 @subsection ARM Options
5959 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5964 @opindex mapcs-frame
5965 Generate a stack frame that is compliant with the ARM Procedure Call
5966 Standard for all functions, even if this is not strictly necessary for
5967 correct execution of the code. Specifying @option{-fomit-frame-pointer}
5968 with this option will cause the stack frames not to be generated for
5969 leaf functions. The default is @option{-mno-apcs-frame}.
5973 This is a synonym for @option{-mapcs-frame}.
5977 Generate code for a processor running with a 26-bit program counter,
5978 and conforming to the function calling standards for the APCS 26-bit
5979 option. This option replaces the @option{-m2} and @option{-m3} options
5980 of previous releases of the compiler.
5984 Generate code for a processor running with a 32-bit program counter,
5985 and conforming to the function calling standards for the APCS 32-bit
5986 option. This option replaces the @option{-m6} option of previous releases
5990 @c not currently implemented
5991 @item -mapcs-stack-check
5992 @opindex mapcs-stack-check
5993 Generate code to check the amount of stack space available upon entry to
5994 every function (that actually uses some stack space). If there is
5995 insufficient space available then either the function
5996 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5997 called, depending upon the amount of stack space required. The run time
5998 system is required to provide these functions. The default is
5999 @option{-mno-apcs-stack-check}, since this produces smaller code.
6001 @c not currently implemented
6003 @opindex mapcs-float
6004 Pass floating point arguments using the float point registers. This is
6005 one of the variants of the APCS@. This option is recommended if the
6006 target hardware has a floating point unit or if a lot of floating point
6007 arithmetic is going to be performed by the code. The default is
6008 @option{-mno-apcs-float}, since integer only code is slightly increased in
6009 size if @option{-mapcs-float} is used.
6011 @c not currently implemented
6012 @item -mapcs-reentrant
6013 @opindex mapcs-reentrant
6014 Generate reentrant, position independent code. The default is
6015 @option{-mno-apcs-reentrant}.
6018 @item -mthumb-interwork
6019 @opindex mthumb-interwork
6020 Generate code which supports calling between the ARM and Thumb
6021 instruction sets. Without this option the two instruction sets cannot
6022 be reliably used inside one program. The default is
6023 @option{-mno-thumb-interwork}, since slightly larger code is generated
6024 when @option{-mthumb-interwork} is specified.
6026 @item -mno-sched-prolog
6027 @opindex mno-sched-prolog
6028 Prevent the reordering of instructions in the function prolog, or the
6029 merging of those instruction with the instructions in the function's
6030 body. This means that all functions will start with a recognizable set
6031 of instructions (or in fact one of a choice from a small set of
6032 different function prologues), and this information can be used to
6033 locate the start if functions inside an executable piece of code. The
6034 default is @option{-msched-prolog}.
6037 @opindex mhard-float
6038 Generate output containing floating point instructions. This is the
6042 @opindex msoft-float
6043 Generate output containing library calls for floating point.
6044 @strong{Warning:} the requisite libraries are not available for all ARM
6045 targets. Normally the facilities of the machine's usual C compiler are
6046 used, but this cannot be done directly in cross-compilation. You must make
6047 your own arrangements to provide suitable library functions for
6050 @option{-msoft-float} changes the calling convention in the output file;
6051 therefore, it is only useful if you compile @emph{all} of a program with
6052 this option. In particular, you need to compile @file{libgcc.a}, the
6053 library that comes with GCC, with @option{-msoft-float} in order for
6056 @item -mlittle-endian
6057 @opindex mlittle-endian
6058 Generate code for a processor running in little-endian mode. This is
6059 the default for all standard configurations.
6062 @opindex mbig-endian
6063 Generate code for a processor running in big-endian mode; the default is
6064 to compile code for a little-endian processor.
6066 @item -mwords-little-endian
6067 @opindex mwords-little-endian
6068 This option only applies when generating code for big-endian processors.
6069 Generate code for a little-endian word order but a big-endian byte
6070 order. That is, a byte order of the form @samp{32107654}. Note: this
6071 option should only be used if you require compatibility with code for
6072 big-endian ARM processors generated by versions of the compiler prior to
6075 @item -malignment-traps
6076 @opindex malignment-traps
6077 Generate code that will not trap if the MMU has alignment traps enabled.
6078 On ARM architectures prior to ARMv4, there were no instructions to
6079 access half-word objects stored in memory. However, when reading from
6080 memory a feature of the ARM architecture allows a word load to be used,
6081 even if the address is unaligned, and the processor core will rotate the
6082 data as it is being loaded. This option tells the compiler that such
6083 misaligned accesses will cause a MMU trap and that it should instead
6084 synthesize the access as a series of byte accesses. The compiler can
6085 still use word accesses to load half-word data if it knows that the
6086 address is aligned to a word boundary.
6088 This option is ignored when compiling for ARM architecture 4 or later,
6089 since these processors have instructions to directly access half-word
6092 @item -mno-alignment-traps
6093 @opindex mno-alignment-traps
6094 Generate code that assumes that the MMU will not trap unaligned
6095 accesses. This produces better code when the target instruction set
6096 does not have half-word memory operations (i.e.@: implementations prior to
6099 Note that you cannot use this option to access unaligned word objects,
6100 since the processor will only fetch one 32-bit aligned object from
6103 The default setting for most targets is @option{-mno-alignment-traps}, since
6104 this produces better code when there are no half-word memory
6105 instructions available.
6107 @item -mshort-load-bytes
6108 @itemx -mno-short-load-words
6109 @opindex mshort-load-bytes
6110 @opindex mno-short-load-words
6111 These are deprecated aliases for @option{-malignment-traps}.
6113 @item -mno-short-load-bytes
6114 @itemx -mshort-load-words
6115 @opindex mno-short-load-bytes
6116 @opindex mshort-load-words
6117 This are deprecated aliases for @option{-mno-alignment-traps}.
6121 This option only applies to RISC iX@. Emulate the native BSD-mode
6122 compiler. This is the default if @option{-ansi} is not specified.
6126 This option only applies to RISC iX@. Emulate the native X/Open-mode
6129 @item -mno-symrename
6130 @opindex mno-symrename
6131 This option only applies to RISC iX@. Do not run the assembler
6132 post-processor, @samp{symrename}, after code has been assembled.
6133 Normally it is necessary to modify some of the standard symbols in
6134 preparation for linking with the RISC iX C library; this option
6135 suppresses this pass. The post-processor is never run when the
6136 compiler is built for cross-compilation.
6138 @item -mcpu=@var{name}
6140 This specifies the name of the target ARM processor. GCC uses this name
6141 to determine what kind of instructions it can emit when generating
6142 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6143 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6144 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6145 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6146 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6147 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6148 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6149 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6150 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6151 @samp{arm1020t}, @samp{xscale}.
6153 @itemx -mtune=@var{name}
6155 This option is very similar to the @option{-mcpu=} option, except that
6156 instead of specifying the actual target processor type, and hence
6157 restricting which instructions can be used, it specifies that GCC should
6158 tune the performance of the code as if the target were of the type
6159 specified in this option, but still choosing the instructions that it
6160 will generate based on the cpu specified by a @option{-mcpu=} option.
6161 For some ARM implementations better performance can be obtained by using
6164 @item -march=@var{name}
6166 This specifies the name of the target ARM architecture. GCC uses this
6167 name to determine what kind of instructions it can emit when generating
6168 assembly code. This option can be used in conjunction with or instead
6169 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6170 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6171 @samp{armv5}, @samp{armv5t}, @samp{armv5te}.
6173 @item -mfpe=@var{number}
6174 @itemx -mfp=@var{number}
6177 This specifies the version of the floating point emulation available on
6178 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6179 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6181 @item -mstructure-size-boundary=@var{n}
6182 @opindex mstructure-size-boundary
6183 The size of all structures and unions will be rounded up to a multiple
6184 of the number of bits set by this option. Permissible values are 8 and
6185 32. The default value varies for different toolchains. For the COFF
6186 targeted toolchain the default value is 8. Specifying the larger number
6187 can produce faster, more efficient code, but can also increase the size
6188 of the program. The two values are potentially incompatible. Code
6189 compiled with one value cannot necessarily expect to work with code or
6190 libraries compiled with the other value, if they exchange information
6191 using structures or unions.
6193 @item -mabort-on-noreturn
6194 @opindex mabort-on-noreturn
6195 Generate a call to the function @code{abort} at the end of a
6196 @code{noreturn} function. It will be executed if the function tries to
6200 @itemx -mno-long-calls
6201 @opindex mlong-calls
6202 @opindex mno-long-calls
6203 Tells the compiler to perform function calls by first loading the
6204 address of the function into a register and then performing a subroutine
6205 call on this register. This switch is needed if the target function
6206 will lie outside of the 64 megabyte addressing range of the offset based
6207 version of subroutine call instruction.
6209 Even if this switch is enabled, not all function calls will be turned
6210 into long calls. The heuristic is that static functions, functions
6211 which have the @samp{short-call} attribute, functions that are inside
6212 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6213 definitions have already been compiled within the current compilation
6214 unit, will not be turned into long calls. The exception to this rule is
6215 that weak function definitions, functions with the @samp{long-call}
6216 attribute or the @samp{section} attribute, and functions that are within
6217 the scope of a @samp{#pragma long_calls} directive, will always be
6218 turned into long calls.
6220 This feature is not enabled by default. Specifying
6221 @option{-mno-long-calls} will restore the default behavior, as will
6222 placing the function calls within the scope of a @samp{#pragma
6223 long_calls_off} directive. Note these switches have no effect on how
6224 the compiler generates code to handle function calls via function
6227 @item -mnop-fun-dllimport
6228 @opindex mnop-fun-dllimport
6229 Disable support for the @code{dllimport} attribute.
6231 @item -msingle-pic-base
6232 @opindex msingle-pic-base
6233 Treat the register used for PIC addressing as read-only, rather than
6234 loading it in the prologue for each function. The run-time system is
6235 responsible for initializing this register with an appropriate value
6236 before execution begins.
6238 @item -mpic-register=@var{reg}
6239 @opindex mpic-register
6240 Specify the register to be used for PIC addressing. The default is R10
6241 unless stack-checking is enabled, when R9 is used.
6243 @item -mpoke-function-name
6244 @opindex mpoke-function-name
6245 Write the name of each function into the text section, directly
6246 preceding the function prologue. The generated code is similar to this:
6250 .ascii "arm_poke_function_name", 0
6253 .word 0xff000000 + (t1 - t0)
6254 arm_poke_function_name
6256 stmfd sp!, @{fp, ip, lr, pc@}
6260 When performing a stack backtrace, code can inspect the value of
6261 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6262 location @code{pc - 12} and the top 8 bits are set, then we know that
6263 there is a function name embedded immediately preceding this location
6264 and has length @code{((pc[-3]) & 0xff000000)}.
6268 Generate code for the 16-bit Thumb instruction set. The default is to
6269 use the 32-bit ARM instruction set.
6272 @opindex mtpcs-frame
6273 Generate a stack frame that is compliant with the Thumb Procedure Call
6274 Standard for all non-leaf functions. (A leaf function is one that does
6275 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6277 @item -mtpcs-leaf-frame
6278 @opindex mtpcs-leaf-frame
6279 Generate a stack frame that is compliant with the Thumb Procedure Call
6280 Standard for all leaf functions. (A leaf function is one that does
6281 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6283 @item -mcallee-super-interworking
6284 @opindex mcallee-super-interworking
6285 Gives all externally visible functions in the file being compiled an ARM
6286 instruction set header which switches to Thumb mode before executing the
6287 rest of the function. This allows these functions to be called from
6288 non-interworking code.
6290 @item -mcaller-super-interworking
6291 @opindex mcaller-super-interworking
6292 Allows calls via function pointers (including virtual functions) to
6293 execute correctly regardless of whether the target code has been
6294 compiled for interworking or not. There is a small overhead in the cost
6295 of executing a function pointer if this option is enabled.
6299 @node MN10200 Options
6300 @subsection MN10200 Options
6301 @cindex MN10200 options
6302 These @option{-m} options are defined for Matsushita MN10200 architectures:
6307 Indicate to the linker that it should perform a relaxation optimization pass
6308 to shorten branches, calls and absolute memory addresses. This option only
6309 has an effect when used on the command line for the final link step.
6311 This option makes symbolic debugging impossible.
6314 @node MN10300 Options
6315 @subsection MN10300 Options
6316 @cindex MN10300 options
6317 These @option{-m} options are defined for Matsushita MN10300 architectures:
6322 Generate code to avoid bugs in the multiply instructions for the MN10300
6323 processors. This is the default.
6326 @opindex mno-mult-bug
6327 Do not generate code to avoid bugs in the multiply instructions for the
6332 Generate code which uses features specific to the AM33 processor.
6336 Do not generate code which uses features specific to the AM33 processor. This
6341 Do not link in the C run-time initialization object file.
6345 Indicate to the linker that it should perform a relaxation optimization pass
6346 to shorten branches, calls and absolute memory addresses. This option only
6347 has an effect when used on the command line for the final link step.
6349 This option makes symbolic debugging impossible.
6353 @node M32R/D Options
6354 @subsection M32R/D Options
6355 @cindex M32R/D options
6357 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6362 Generate code for the M32R/X@.
6366 Generate code for the M32R@. This is the default.
6368 @item -mcode-model=small
6369 @opindex mcode-model=small
6370 Assume all objects live in the lower 16MB of memory (so that their addresses
6371 can be loaded with the @code{ld24} instruction), and assume all subroutines
6372 are reachable with the @code{bl} instruction.
6373 This is the default.
6375 The addressability of a particular object can be set with the
6376 @code{model} attribute.
6378 @item -mcode-model=medium
6379 @opindex mcode-model=medium
6380 Assume objects may be anywhere in the 32-bit address space (the compiler
6381 will generate @code{seth/add3} instructions to load their addresses), and
6382 assume all subroutines are reachable with the @code{bl} instruction.
6384 @item -mcode-model=large
6385 @opindex mcode-model=large
6386 Assume objects may be anywhere in the 32-bit address space (the compiler
6387 will generate @code{seth/add3} instructions to load their addresses), and
6388 assume subroutines may not be reachable with the @code{bl} instruction
6389 (the compiler will generate the much slower @code{seth/add3/jl}
6390 instruction sequence).
6393 @opindex msdata=none
6394 Disable use of the small data area. Variables will be put into
6395 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6396 @code{section} attribute has been specified).
6397 This is the default.
6399 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6400 Objects may be explicitly put in the small data area with the
6401 @code{section} attribute using one of these sections.
6404 @opindex msdata=sdata
6405 Put small global and static data in the small data area, but do not
6406 generate special code to reference them.
6410 Put small global and static data in the small data area, and generate
6411 special instructions to reference them.
6415 @cindex smaller data references
6416 Put global and static objects less than or equal to @var{num} bytes
6417 into the small data or bss sections instead of the normal data or bss
6418 sections. The default value of @var{num} is 8.
6419 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6420 for this option to have any effect.
6422 All modules should be compiled with the same @option{-G @var{num}} value.
6423 Compiling with different values of @var{num} may or may not work; if it
6424 doesn't the linker will give an error message---incorrect code will not be
6430 @subsection M88K Options
6431 @cindex M88k options
6433 These @samp{-m} options are defined for Motorola 88k architectures:
6438 Generate code that works well on both the m88100 and the
6443 Generate code that works best for the m88100, but that also
6448 Generate code that works best for the m88110, and may not run
6453 Obsolete option to be removed from the next revision.
6456 @item -midentify-revision
6457 @opindex midentify-revision
6458 @cindex identifying source, compiler (88k)
6459 Include an @code{ident} directive in the assembler output recording the
6460 source file name, compiler name and version, timestamp, and compilation
6463 @item -mno-underscores
6464 @opindex mno-underscores
6465 @cindex underscores, avoiding (88k)
6466 In assembler output, emit symbol names without adding an underscore
6467 character at the beginning of each name. The default is to use an
6468 underscore as prefix on each name.
6470 @item -mocs-debug-info
6471 @itemx -mno-ocs-debug-info
6472 @opindex mocs-debug-info
6473 @opindex mno-ocs-debug-info
6475 @cindex debugging, 88k OCS
6476 Include (or omit) additional debugging information (about registers used
6477 in each stack frame) as specified in the 88open Object Compatibility
6478 Standard, ``OCS''@. This extra information allows debugging of code that
6479 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6480 Delta 88 SVr3.2 is to include this information; other 88k configurations
6481 omit this information by default.
6483 @item -mocs-frame-position
6484 @opindex mocs-frame-position
6485 @cindex register positions in frame (88k)
6486 When emitting COFF debugging information for automatic variables and
6487 parameters stored on the stack, use the offset from the canonical frame
6488 address, which is the stack pointer (register 31) on entry to the
6489 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6490 @option{-mocs-frame-position}; other 88k configurations have the default
6491 @option{-mno-ocs-frame-position}.
6493 @item -mno-ocs-frame-position
6494 @opindex mno-ocs-frame-position
6495 @cindex register positions in frame (88k)
6496 When emitting COFF debugging information for automatic variables and
6497 parameters stored on the stack, use the offset from the frame pointer
6498 register (register 30). When this option is in effect, the frame
6499 pointer is not eliminated when debugging information is selected by the
6502 @item -moptimize-arg-area
6503 @opindex moptimize-arg-area
6504 @cindex arguments in frame (88k)
6505 Save space by reorganizing the stack frame. This option generates code
6506 that does not agree with the 88open specifications, but uses less
6509 @itemx -mno-optimize-arg-area
6510 @opindex mno-optimize-arg-area
6511 Do not reorganize the stack frame to save space. This is the default.
6512 The generated conforms to the specification, but uses more memory.
6514 @item -mshort-data-@var{num}
6515 @opindex mshort-data
6516 @cindex smaller data references (88k)
6517 @cindex r0-relative references (88k)
6518 Generate smaller data references by making them relative to @code{r0},
6519 which allows loading a value using a single instruction (rather than the
6520 usual two). You control which data references are affected by
6521 specifying @var{num} with this option. For example, if you specify
6522 @option{-mshort-data-512}, then the data references affected are those
6523 involving displacements of less than 512 bytes.
6524 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6527 @item -mserialize-volatile
6528 @opindex mserialize-volatile
6529 @itemx -mno-serialize-volatile
6530 @opindex mno-serialize-volatile
6531 @cindex sequential consistency on 88k
6532 Do, or don't, generate code to guarantee sequential consistency
6533 of volatile memory references. By default, consistency is
6536 The order of memory references made by the MC88110 processor does
6537 not always match the order of the instructions requesting those
6538 references. In particular, a load instruction may execute before
6539 a preceding store instruction. Such reordering violates
6540 sequential consistency of volatile memory references, when there
6541 are multiple processors. When consistency must be guaranteed,
6542 GCC generates special instructions, as needed, to force
6543 execution in the proper order.
6545 The MC88100 processor does not reorder memory references and so
6546 always provides sequential consistency. However, by default, GCC
6547 generates the special instructions to guarantee consistency
6548 even when you use @option{-m88100}, so that the code may be run on an
6549 MC88110 processor. If you intend to run your code only on the
6550 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6552 The extra code generated to guarantee consistency may affect the
6553 performance of your application. If you know that you can safely
6554 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6560 @cindex assembler syntax, 88k
6562 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6563 related to System V release 4 (SVr4). This controls the following:
6567 Which variant of the assembler syntax to emit.
6569 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6570 that is used on System V release 4.
6572 @option{-msvr4} makes GCC issue additional declaration directives used in
6576 @option{-msvr4} is the default for the m88k-motorola-sysv4 and
6577 m88k-dg-dgux m88k configurations. @option{-msvr3} is the default for all
6578 other m88k configurations.
6580 @item -mversion-03.00
6581 @opindex mversion-03.00
6582 This option is obsolete, and is ignored.
6583 @c ??? which asm syntax better for GAS? option there too?
6585 @item -mno-check-zero-division
6586 @itemx -mcheck-zero-division
6587 @opindex mno-check-zero-division
6588 @opindex mcheck-zero-division
6589 @cindex zero division on 88k
6590 Do, or don't, generate code to guarantee that integer division by
6591 zero will be detected. By default, detection is guaranteed.
6593 Some models of the MC88100 processor fail to trap upon integer
6594 division by zero under certain conditions. By default, when
6595 compiling code that might be run on such a processor, GCC
6596 generates code that explicitly checks for zero-valued divisors
6597 and traps with exception number 503 when one is detected. Use of
6598 @option{-mno-check-zero-division} suppresses such checking for code
6599 generated to run on an MC88100 processor.
6601 GCC assumes that the MC88110 processor correctly detects all instances
6602 of integer division by zero. When @option{-m88110} is specified, no
6603 explicit checks for zero-valued divisors are generated, and both
6604 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6607 @item -muse-div-instruction
6608 @opindex muse-div-instruction
6609 @cindex divide instruction, 88k
6610 Use the div instruction for signed integer division on the
6611 MC88100 processor. By default, the div instruction is not used.
6613 On the MC88100 processor the signed integer division instruction
6614 div) traps to the operating system on a negative operand. The
6615 operating system transparently completes the operation, but at a
6616 large cost in execution time. By default, when compiling code
6617 that might be run on an MC88100 processor, GCC emulates signed
6618 integer division using the unsigned integer division instruction
6619 divu), thereby avoiding the large penalty of a trap to the
6620 operating system. Such emulation has its own, smaller, execution
6621 cost in both time and space. To the extent that your code's
6622 important signed integer division operations are performed on two
6623 nonnegative operands, it may be desirable to use the div
6624 instruction directly.
6626 On the MC88110 processor the div instruction (also known as the
6627 divs instruction) processes negative operands without trapping to
6628 the operating system. When @option{-m88110} is specified,
6629 @option{-muse-div-instruction} is ignored, and the div instruction is used
6630 for signed integer division.
6632 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6633 particular, the behavior of such a division with and without
6634 @option{-muse-div-instruction} may differ.
6636 @item -mtrap-large-shift
6637 @itemx -mhandle-large-shift
6638 @opindex mtrap-large-shift
6639 @opindex mhandle-large-shift
6640 @cindex bit shift overflow (88k)
6641 @cindex large bit shifts (88k)
6642 Include code to detect bit-shifts of more than 31 bits; respectively,
6643 trap such shifts or emit code to handle them properly. By default GCC
6644 makes no special provision for large bit shifts.
6646 @item -mwarn-passed-structs
6647 @opindex mwarn-passed-structs
6648 @cindex structure passing (88k)
6649 Warn when a function passes a struct as an argument or result.
6650 Structure-passing conventions have changed during the evolution of the C
6651 language, and are often the source of portability problems. By default,
6652 GCC issues no such warning.
6655 @c break page here to avoid unsightly interparagraph stretch.
6659 @node RS/6000 and PowerPC Options
6660 @subsection IBM RS/6000 and PowerPC Options
6661 @cindex RS/6000 and PowerPC Options
6662 @cindex IBM RS/6000 and PowerPC Options
6664 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6672 @itemx -mpowerpc-gpopt
6673 @itemx -mno-powerpc-gpopt
6674 @itemx -mpowerpc-gfxopt
6675 @itemx -mno-powerpc-gfxopt
6677 @itemx -mno-powerpc64
6683 @opindex mno-powerpc
6684 @opindex mpowerpc-gpopt
6685 @opindex mno-powerpc-gpopt
6686 @opindex mpowerpc-gfxopt
6687 @opindex mno-powerpc-gfxopt
6689 @opindex mno-powerpc64
6690 GCC supports two related instruction set architectures for the
6691 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6692 instructions supported by the @samp{rios} chip set used in the original
6693 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6694 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6695 the IBM 4xx microprocessors.
6697 Neither architecture is a subset of the other. However there is a
6698 large common subset of instructions supported by both. An MQ
6699 register is included in processors supporting the POWER architecture.
6701 You use these options to specify which instructions are available on the
6702 processor you are using. The default value of these options is
6703 determined when configuring GCC@. Specifying the
6704 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6705 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6706 rather than the options listed above.
6708 The @option{-mpower} option allows GCC to generate instructions that
6709 are found only in the POWER architecture and to use the MQ register.
6710 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6711 to generate instructions that are present in the POWER2 architecture but
6712 not the original POWER architecture.
6714 The @option{-mpowerpc} option allows GCC to generate instructions that
6715 are found only in the 32-bit subset of the PowerPC architecture.
6716 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6717 GCC to use the optional PowerPC architecture instructions in the
6718 General Purpose group, including floating-point square root. Specifying
6719 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6720 use the optional PowerPC architecture instructions in the Graphics
6721 group, including floating-point select.
6723 The @option{-mpowerpc64} option allows GCC to generate the additional
6724 64-bit instructions that are found in the full PowerPC64 architecture
6725 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6726 @option{-mno-powerpc64}.
6728 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6729 will use only the instructions in the common subset of both
6730 architectures plus some special AIX common-mode calls, and will not use
6731 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6732 permits GCC to use any instruction from either architecture and to
6733 allow use of the MQ register; specify this for the Motorola MPC601.
6735 @item -mnew-mnemonics
6736 @itemx -mold-mnemonics
6737 @opindex mnew-mnemonics
6738 @opindex mold-mnemonics
6739 Select which mnemonics to use in the generated assembler code. With
6740 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6741 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6742 assembler mnemonics defined for the POWER architecture. Instructions
6743 defined in only one architecture have only one mnemonic; GCC uses that
6744 mnemonic irrespective of which of these options is specified.
6746 GCC defaults to the mnemonics appropriate for the architecture in
6747 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6748 value of these option. Unless you are building a cross-compiler, you
6749 should normally not specify either @option{-mnew-mnemonics} or
6750 @option{-mold-mnemonics}, but should instead accept the default.
6752 @item -mcpu=@var{cpu_type}
6754 Set architecture type, register usage, choice of mnemonics, and
6755 instruction scheduling parameters for machine type @var{cpu_type}.
6756 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6757 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6758 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6759 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
6760 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
6761 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
6763 @option{-mcpu=common} selects a completely generic processor. Code
6764 generated under this option will run on any POWER or PowerPC processor.
6765 GCC will use only the instructions in the common subset of both
6766 architectures, and will not use the MQ register. GCC assumes a generic
6767 processor model for scheduling purposes.
6769 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
6770 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
6771 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
6772 types, with an appropriate, generic processor model assumed for
6773 scheduling purposes.
6775 The other options specify a specific processor. Code generated under
6776 those options will run best on that processor, and may not run at all on
6779 The @option{-mcpu} options automatically enable or disable other
6780 @option{-m} options as follows:
6784 @option{-mno-power}, @option{-mno-powerc}
6791 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
6806 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6809 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6814 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
6817 @item -mtune=@var{cpu_type}
6819 Set the instruction scheduling parameters for machine type
6820 @var{cpu_type}, but do not set the architecture type, register usage, or
6821 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
6822 values for @var{cpu_type} are used for @option{-mtune} as for
6823 @option{-mcpu}. If both are specified, the code generated will use the
6824 architecture, registers, and mnemonics set by @option{-mcpu}, but the
6825 scheduling parameters set by @option{-mtune}.
6830 @opindex mno-altivec
6831 These switches enable or disable the use of built-in functions that
6832 allow access to the AltiVec instruction set. You may also need to set
6833 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
6838 Extend the current ABI with SPE ABI extensions. This does not change
6839 the default ABI, instead it adds the SPE ABI extensions to the current
6843 @opindex mabi=no-spe
6844 Disable Booke SPE ABI extensions for the current ABI.
6846 @item -misel=@var{yes/no}
6849 This switch enables or disables the generation of ISEL instructions.
6852 @itemx -mno-fp-in-toc
6853 @itemx -mno-sum-in-toc
6854 @itemx -mminimal-toc
6856 @opindex mno-fp-in-toc
6857 @opindex mno-sum-in-toc
6858 @opindex mminimal-toc
6859 Modify generation of the TOC (Table Of Contents), which is created for
6860 every executable file. The @option{-mfull-toc} option is selected by
6861 default. In that case, GCC will allocate at least one TOC entry for
6862 each unique non-automatic variable reference in your program. GCC
6863 will also place floating-point constants in the TOC@. However, only
6864 16,384 entries are available in the TOC@.
6866 If you receive a linker error message that saying you have overflowed
6867 the available TOC space, you can reduce the amount of TOC space used
6868 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
6869 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
6870 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
6871 generate code to calculate the sum of an address and a constant at
6872 run-time instead of putting that sum into the TOC@. You may specify one
6873 or both of these options. Each causes GCC to produce very slightly
6874 slower and larger code at the expense of conserving TOC space.
6876 If you still run out of space in the TOC even when you specify both of
6877 these options, specify @option{-mminimal-toc} instead. This option causes
6878 GCC to make only one TOC entry for every file. When you specify this
6879 option, GCC will produce code that is slower and larger but which
6880 uses extremely little TOC space. You may wish to use this option
6881 only on files that contain less frequently executed code.
6887 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6888 @code{long} type, and the infrastructure needed to support them.
6889 Specifying @option{-maix64} implies @option{-mpowerpc64} and
6890 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
6891 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
6896 @opindex mno-xl-call
6897 On AIX, pass floating-point arguments to prototyped functions beyond the
6898 register save area (RSA) on the stack in addition to argument FPRs. The
6899 AIX calling convention was extended but not initially documented to
6900 handle an obscure K&R C case of calling a function that takes the
6901 address of its arguments with fewer arguments than declared. AIX XL
6902 compilers access floating point arguments which do not fit in the
6903 RSA from the stack when a subroutine is compiled without
6904 optimization. Because always storing floating-point arguments on the
6905 stack is inefficient and rarely needed, this option is not enabled by
6906 default and only is necessary when calling subroutines compiled by AIX
6907 XL compilers without optimization.
6911 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
6912 application written to use message passing with special startup code to
6913 enable the application to run. The system must have PE installed in the
6914 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6915 must be overridden with the @option{-specs=} option to specify the
6916 appropriate directory location. The Parallel Environment does not
6917 support threads, so the @option{-mpe} option and the @option{-pthread}
6918 option are incompatible.
6922 @opindex msoft-float
6923 @opindex mhard-float
6924 Generate code that does not use (uses) the floating-point register set.
6925 Software floating point emulation is provided if you use the
6926 @option{-msoft-float} option, and pass the option to GCC when linking.
6929 @itemx -mno-multiple
6931 @opindex mno-multiple
6932 Generate code that uses (does not use) the load multiple word
6933 instructions and the store multiple word instructions. These
6934 instructions are generated by default on POWER systems, and not
6935 generated on PowerPC systems. Do not use @option{-mmultiple} on little
6936 endian PowerPC systems, since those instructions do not work when the
6937 processor is in little endian mode. The exceptions are PPC740 and
6938 PPC750 which permit the instructions usage in little endian mode.
6944 Generate code that uses (does not use) the load string instructions
6945 and the store string word instructions to save multiple registers and
6946 do small block moves. These instructions are generated by default on
6947 POWER systems, and not generated on PowerPC systems. Do not use
6948 @option{-mstring} on little endian PowerPC systems, since those
6949 instructions do not work when the processor is in little endian mode.
6950 The exceptions are PPC740 and PPC750 which permit the instructions
6951 usage in little endian mode.
6957 Generate code that uses (does not use) the load or store instructions
6958 that update the base register to the address of the calculated memory
6959 location. These instructions are generated by default. If you use
6960 @option{-mno-update}, there is a small window between the time that the
6961 stack pointer is updated and the address of the previous frame is
6962 stored, which means code that walks the stack frame across interrupts or
6963 signals may get corrupted data.
6966 @itemx -mno-fused-madd
6967 @opindex mfused-madd
6968 @opindex mno-fused-madd
6969 Generate code that uses (does not use) the floating point multiply and
6970 accumulate instructions. These instructions are generated by default if
6971 hardware floating is used.
6973 @item -mno-bit-align
6975 @opindex mno-bit-align
6977 On System V.4 and embedded PowerPC systems do not (do) force structures
6978 and unions that contain bit-fields to be aligned to the base type of the
6981 For example, by default a structure containing nothing but 8
6982 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
6983 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
6984 the structure would be aligned to a 1 byte boundary and be one byte in
6987 @item -mno-strict-align
6988 @itemx -mstrict-align
6989 @opindex mno-strict-align
6990 @opindex mstrict-align
6991 On System V.4 and embedded PowerPC systems do not (do) assume that
6992 unaligned memory references will be handled by the system.
6995 @itemx -mno-relocatable
6996 @opindex mrelocatable
6997 @opindex mno-relocatable
6998 On embedded PowerPC systems generate code that allows (does not allow)
6999 the program to be relocated to a different address at runtime. If you
7000 use @option{-mrelocatable} on any module, all objects linked together must
7001 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7003 @item -mrelocatable-lib
7004 @itemx -mno-relocatable-lib
7005 @opindex mrelocatable-lib
7006 @opindex mno-relocatable-lib
7007 On embedded PowerPC systems generate code that allows (does not allow)
7008 the program to be relocated to a different address at runtime. Modules
7009 compiled with @option{-mrelocatable-lib} can be linked with either modules
7010 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7011 with modules compiled with the @option{-mrelocatable} options.
7017 On System V.4 and embedded PowerPC systems do not (do) assume that
7018 register 2 contains a pointer to a global area pointing to the addresses
7019 used in the program.
7022 @itemx -mlittle-endian
7024 @opindex mlittle-endian
7025 On System V.4 and embedded PowerPC systems compile code for the
7026 processor in little endian mode. The @option{-mlittle-endian} option is
7027 the same as @option{-mlittle}.
7032 @opindex mbig-endian
7033 On System V.4 and embedded PowerPC systems compile code for the
7034 processor in big endian mode. The @option{-mbig-endian} option is
7035 the same as @option{-mbig}.
7039 On System V.4 and embedded PowerPC systems compile code using calling
7040 conventions that adheres to the March 1995 draft of the System V
7041 Application Binary Interface, PowerPC processor supplement. This is the
7042 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7044 @item -mcall-sysv-eabi
7045 @opindex mcall-sysv-eabi
7046 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7048 @item -mcall-sysv-noeabi
7049 @opindex mcall-sysv-noeabi
7050 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7054 On System V.4 and embedded PowerPC systems compile code using calling
7055 conventions that are similar to those used on AIX@. This is the
7056 default if you configured GCC using @samp{powerpc-*-eabiaix}.
7058 @item -mcall-solaris
7059 @opindex mcall-solaris
7060 On System V.4 and embedded PowerPC systems compile code for the Solaris
7064 @opindex mcall-linux
7065 On System V.4 and embedded PowerPC systems compile code for the
7066 Linux-based GNU system.
7070 On System V.4 and embedded PowerPC systems compile code for the
7071 Hurd-based GNU system.
7074 @opindex mcall-netbsd
7075 On System V.4 and embedded PowerPC systems compile code for the
7076 NetBSD operating system.
7078 @item -maix-struct-return
7079 @opindex maix-struct-return
7080 Return all structures in memory (as specified by the AIX ABI)@.
7082 @item -msvr4-struct-return
7083 @opindex msvr4-struct-return
7084 Return structures smaller than 8 bytes in registers (as specified by the
7088 @opindex mabi=altivec
7089 Extend the current ABI with AltiVec ABI extensions. This does not
7090 change the default ABI, instead it adds the AltiVec ABI extensions to
7093 @item -mabi=no-altivec
7094 @opindex mabi=no-altivec
7095 Disable AltiVec ABI extensions for the current ABI.
7098 @itemx -mno-prototype
7100 @opindex mno-prototype
7101 On System V.4 and embedded PowerPC systems assume that all calls to
7102 variable argument functions are properly prototyped. Otherwise, the
7103 compiler must insert an instruction before every non prototyped call to
7104 set or clear bit 6 of the condition code register (@var{CR}) to
7105 indicate whether floating point values were passed in the floating point
7106 registers in case the function takes a variable arguments. With
7107 @option{-mprototype}, only calls to prototyped variable argument functions
7108 will set or clear the bit.
7112 On embedded PowerPC systems, assume that the startup module is called
7113 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7114 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7119 On embedded PowerPC systems, assume that the startup module is called
7120 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7125 On embedded PowerPC systems, assume that the startup module is called
7126 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7130 @opindex myellowknife
7131 On embedded PowerPC systems, assume that the startup module is called
7132 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7137 On System V.4 and embedded PowerPC systems, specify that you are
7138 compiling for a VxWorks system.
7142 Specify that you are compiling for the WindISS simulation environment.
7146 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7147 header to indicate that @samp{eabi} extended relocations are used.
7153 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7154 Embedded Applications Binary Interface (eabi) which is a set of
7155 modifications to the System V.4 specifications. Selecting @option{-meabi}
7156 means that the stack is aligned to an 8 byte boundary, a function
7157 @code{__eabi} is called to from @code{main} to set up the eabi
7158 environment, and the @option{-msdata} option can use both @code{r2} and
7159 @code{r13} to point to two separate small data areas. Selecting
7160 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7161 do not call an initialization function from @code{main}, and the
7162 @option{-msdata} option will only use @code{r13} to point to a single
7163 small data area. The @option{-meabi} option is on by default if you
7164 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7167 @opindex msdata=eabi
7168 On System V.4 and embedded PowerPC systems, put small initialized
7169 @code{const} global and static data in the @samp{.sdata2} section, which
7170 is pointed to by register @code{r2}. Put small initialized
7171 non-@code{const} global and static data in the @samp{.sdata} section,
7172 which is pointed to by register @code{r13}. Put small uninitialized
7173 global and static data in the @samp{.sbss} section, which is adjacent to
7174 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7175 incompatible with the @option{-mrelocatable} option. The
7176 @option{-msdata=eabi} option also sets the @option{-memb} option.
7179 @opindex msdata=sysv
7180 On System V.4 and embedded PowerPC systems, put small global and static
7181 data in the @samp{.sdata} section, which is pointed to by register
7182 @code{r13}. Put small uninitialized global and static data in the
7183 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7184 The @option{-msdata=sysv} option is incompatible with the
7185 @option{-mrelocatable} option.
7187 @item -msdata=default
7189 @opindex msdata=default
7191 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7192 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7193 same as @option{-msdata=sysv}.
7196 @opindex msdata-data
7197 On System V.4 and embedded PowerPC systems, put small global and static
7198 data in the @samp{.sdata} section. Put small uninitialized global and
7199 static data in the @samp{.sbss} section. Do not use register @code{r13}
7200 to address small data however. This is the default behavior unless
7201 other @option{-msdata} options are used.
7205 @opindex msdata=none
7207 On embedded PowerPC systems, put all initialized global and static data
7208 in the @samp{.data} section, and all uninitialized data in the
7209 @samp{.bss} section.
7213 @cindex smaller data references (PowerPC)
7214 @cindex .sdata/.sdata2 references (PowerPC)
7215 On embedded PowerPC systems, put global and static items less than or
7216 equal to @var{num} bytes into the small data or bss sections instead of
7217 the normal data or bss section. By default, @var{num} is 8. The
7218 @option{-G @var{num}} switch is also passed to the linker.
7219 All modules should be compiled with the same @option{-G @var{num}} value.
7222 @itemx -mno-regnames
7224 @opindex mno-regnames
7225 On System V.4 and embedded PowerPC systems do (do not) emit register
7226 names in the assembly language output using symbolic forms.
7229 @itemx -mno-longcall
7231 @opindex mno-longcall
7232 Default to making all function calls via pointers, so that functions
7233 which reside further than 64 megabytes (67,108,864 bytes) from the
7234 current location can be called. This setting can be overridden by the
7235 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7237 Some linkers are capable of detecting out-of-range calls and generating
7238 glue code on the fly. On these systems, long calls are unnecessary and
7239 generate slower code. As of this writing, the AIX linker can do this,
7240 as can the GNU linker for PowerPC/64. It is planned to add this feature
7241 to the GNU linker for 32-bit PowerPC systems as well.
7243 In the future, we may cause GCC to ignore all longcall specifications
7244 when the linker is known to generate glue.
7248 Adds support for multithreading with the @dfn{pthreads} library.
7249 This option sets flags for both the preprocessor and linker.
7253 @node Darwin Options
7254 @subsection Darwin Options
7255 @cindex Darwin options
7258 @item -allowable_client
7260 @item -arch_errors_fatal
7264 @item -bundle_loader
7266 @item -compatibility_version
7267 @item -current_version
7268 @item -dependency-file
7270 @item -dylinker_install_name
7273 @item -exported_symbols_list
7275 @item -flat_namespace
7276 @item -force_cpusubtype_ALL
7277 @item -force_flat_namespace
7278 @item -headerpad_max_install_names
7282 @item -keep_private_externs
7284 @item -multiply_defined
7285 @item -multiply_defined_unused
7289 @item -noseglinkedit
7290 @item -pagezero_size
7292 @item -prebind_all_twolevel_modules
7293 @item -private_bundle
7294 @item -read_only_relocs
7296 @item -sectobjectsymbols
7300 @item -sectobjectsymbols
7302 @item -seg_addr_table
7303 @item -seg_addr_table_filename
7306 @item -segs_read_only_addr
7307 @item -segs_read_write_addr
7308 @item -single_module
7312 @item -twolevel_namespace
7315 @item -unexported_symbols_list
7316 @item -weak_reference_mismatches
7319 @opindex allowable_client
7321 @opindex arch_errors_fatal
7323 @opindex bind_at_load
7325 @opindex bundle_loader
7326 @opindex client_name
7327 @opindex compatibility_version
7328 @opindex current_version
7329 @opindex dependency-file
7331 @opindex dylinker_install_name
7334 @opindex exported_symbols_list
7336 @opindex flat_namespace
7337 @opindex force_cpusubtype_ALL
7338 @opindex force_flat_namespace
7339 @opindex headerpad_max_install_names
7342 @opindex install_name
7343 @opindex keep_private_externs
7344 @opindex multi_module
7345 @opindex multiply_defined
7346 @opindex multiply_defined_unused
7348 @opindex nomultidefs
7350 @opindex noseglinkedit
7351 @opindex pagezero_size
7353 @opindex prebind_all_twolevel_modules
7354 @opindex private_bundle
7355 @opindex read_only_relocs
7357 @opindex sectobjectsymbols
7361 @opindex sectobjectsymbols
7363 @opindex seg_addr_table
7364 @opindex seg_addr_table_filename
7365 @opindex seglinkedit
7367 @opindex segs_read_only_addr
7368 @opindex segs_read_write_addr
7369 @opindex single_module
7371 @opindex sub_library
7372 @opindex sub_umbrella
7373 @opindex twolevel_namespace
7376 @opindex unexported_symbols_list
7377 @opindex weak_reference_mismatches
7378 @opindex whatsloaded
7380 This options are available for Darwin linker. Darwin linker man page
7381 describes them in detail.
7386 @subsection IBM RT Options
7388 @cindex IBM RT options
7390 These @samp{-m} options are defined for the IBM RT PC:
7394 @opindex min-line-mul
7395 Use an in-line code sequence for integer multiplies. This is the
7398 @item -mcall-lib-mul
7399 @opindex mcall-lib-mul
7400 Call @code{lmul$$} for integer multiples.
7402 @item -mfull-fp-blocks
7403 @opindex mfull-fp-blocks
7404 Generate full-size floating point data blocks, including the minimum
7405 amount of scratch space recommended by IBM@. This is the default.
7407 @item -mminimum-fp-blocks
7408 @opindex mminimum-fp-blocks
7409 Do not include extra scratch space in floating point data blocks. This
7410 results in smaller code, but slower execution, since scratch space must
7411 be allocated dynamically.
7413 @cindex @file{stdarg.h} and RT PC
7414 @item -mfp-arg-in-fpregs
7415 @opindex mfp-arg-in-fpregs
7416 Use a calling sequence incompatible with the IBM calling convention in
7417 which floating point arguments are passed in floating point registers.
7418 Note that @code{stdarg.h} will not work with floating point operands
7419 if this option is specified.
7421 @item -mfp-arg-in-gregs
7422 @opindex mfp-arg-in-gregs
7423 Use the normal calling convention for floating point arguments. This is
7426 @item -mhc-struct-return
7427 @opindex mhc-struct-return
7428 Return structures of more than one word in memory, rather than in a
7429 register. This provides compatibility with the MetaWare HighC (hc)
7430 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7431 with the Portable C Compiler (pcc).
7433 @item -mnohc-struct-return
7434 @opindex mnohc-struct-return
7435 Return some structures of more than one word in registers, when
7436 convenient. This is the default. For compatibility with the
7437 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7438 option @option{-mhc-struct-return}.
7442 @subsection MIPS Options
7443 @cindex MIPS options
7445 These @samp{-m} options are defined for the MIPS family of computers:
7449 @item -march=@var{arch}
7451 Generate code that will run on @var{arch}, which can be the name of a
7452 generic MIPS ISA, or the name of a particular processor. The ISA names
7453 are: @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4}, @samp{mips32}
7454 and @samp{mips64}. The processor names are: @samp{r2000},
7455 @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{vr4100}, @samp{vr4300},
7456 @samp{r4400}, @samp{r4600}, @samp{r4650}, @samp{vr5000}, @samp{r6000},
7457 @samp{r8000}, @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7458 @samp{orion}, and @samp{sb1}. The special value @samp{from-abi} selects the
7459 most compatible architecture for the selected ABI (that is,
7460 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7462 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7463 (for example, @samp{-march=r2k}). Prefixes are optional, and
7464 @samp{vr} may be written @samp{r}.
7466 GCC defines two macros based on the value of this option. The first
7467 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7468 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7469 where @var{foo} is the capitialized value of @samp{_MIPS_ARCH}@.
7470 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7471 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7473 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7474 above. In other words, it will have the full prefix and will not
7475 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7476 the macro names the resolved architecture (either @samp{"mips1"} or
7477 @samp{"mips3"}). It names the default architecture when no
7478 @option{-march} option is given.
7480 @item -mtune=@var{arch}
7482 Optimize for @var{arch}. Among other things, this option controls
7483 the way instructions are scheduled, and the perceived cost of arithmetic
7484 operations. The list of @var{arch} values is the same as for
7487 When this option is not used, GCC will optimize for the processor
7488 specified by @option{-march}. By using @option{-march} and
7489 @option{-mtune} together, it is possible to generate code that will
7490 run on a family of processors, but optimize the code for one
7491 particular member of that family.
7493 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7494 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7495 @samp{-march} ones described above.
7499 Equivalent to @samp{-march=mips1}.
7503 Equivalent to @samp{-march=mips2}.
7507 Equivalent to @samp{-march=mips3}.
7511 Equivalent to @samp{-march=mips4}.
7515 Equivalent to @samp{-march=mips32}.
7519 Equivalent to @samp{-march=mips64}.
7522 @itemx -mno-fused-madd
7523 @opindex mfused-madd
7524 @opindex mno-fused-madd
7525 Generate code that uses (does not use) the floating point multiply and
7526 accumulate instructions, when they are available. These instructions
7527 are generated by default if they are available, but this may be
7528 undesirable if the extra precision causes problems or on certain chips
7529 in the mode where denormals are rounded to zero where denormals
7530 generated by multiply and accumulate instructions cause exceptions
7535 Assume that floating point registers are 32 bits wide.
7539 Assume that floating point registers are 64 bits wide.
7543 Assume that general purpose registers are 32 bits wide.
7547 Assume that general purpose registers are 64 bits wide.
7551 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7552 explanation of the default, and the width of pointers.
7556 Force long types to be 64 bits wide. See @option{-mlong32} for an
7557 explanation of the default, and the width of pointers.
7561 Force long, int, and pointer types to be 32 bits wide.
7563 The default size of ints, longs and pointers depends on the ABI@. All
7564 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
7565 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
7566 are the same size as longs, or the same size as integer registers,
7567 whichever is smaller.
7581 Generate code for the given ABI@.
7583 Note that there are two embedded ABIs: @option{-mabi=eabi}
7584 selects the one defined by Cygnus while @option{-meabi=meabi}
7585 selects the one defined by MIPS@. Both these ABIs have
7586 32-bit and 64-bit variants. Normally, GCC will generate
7587 64-bit code when you select a 64-bit architecture, but you
7588 can use @option{-mgp32} to get 32-bit code instead.
7592 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7593 add normal debug information. This is the default for all
7594 platforms except for the OSF/1 reference platform, using the OSF/rose
7595 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7596 switches are used, the @file{mips-tfile} program will encapsulate the
7597 stabs within MIPS ECOFF@.
7601 Generate code for the GNU assembler. This is the default on the OSF/1
7602 reference platform, using the OSF/rose object format. Also, this is
7603 the default if the configure option @option{--with-gnu-as} is used.
7605 @item -msplit-addresses
7606 @itemx -mno-split-addresses
7607 @opindex msplit-addresses
7608 @opindex mno-split-addresses
7609 Generate code to load the high and low parts of address constants separately.
7610 This allows GCC to optimize away redundant loads of the high order
7611 bits of addresses. This optimization requires GNU as and GNU ld.
7612 This optimization is enabled by default for some embedded targets where
7613 GNU as and GNU ld are standard.
7619 The @option{-mrnames} switch says to output code using the MIPS software
7620 names for the registers, instead of the hardware names (ie, @var{a0}
7621 instead of @var{$4}). The only known assembler that supports this option
7622 is the Algorithmics assembler.
7628 The @option{-mgpopt} switch says to write all of the data declarations
7629 before the instructions in the text section, this allows the MIPS
7630 assembler to generate one word memory references instead of using two
7631 words for short global or static data items. This is on by default if
7632 optimization is selected.
7638 For each non-inline function processed, the @option{-mstats} switch
7639 causes the compiler to emit one line to the standard error file to
7640 print statistics about the program (number of registers saved, stack
7647 The @option{-mmemcpy} switch makes all block moves call the appropriate
7648 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7649 generating inline code.
7652 @itemx -mno-mips-tfile
7653 @opindex mmips-tfile
7654 @opindex mno-mips-tfile
7655 The @option{-mno-mips-tfile} switch causes the compiler not
7656 postprocess the object file with the @file{mips-tfile} program,
7657 after the MIPS assembler has generated it to add debug support. If
7658 @file{mips-tfile} is not run, then no local variables will be
7659 available to the debugger. In addition, @file{stage2} and
7660 @file{stage3} objects will have the temporary file names passed to the
7661 assembler embedded in the object file, which means the objects will
7662 not compare the same. The @option{-mno-mips-tfile} switch should only
7663 be used when there are bugs in the @file{mips-tfile} program that
7664 prevents compilation.
7667 @opindex msoft-float
7668 Generate output containing library calls for floating point.
7669 @strong{Warning:} the requisite libraries are not part of GCC@.
7670 Normally the facilities of the machine's usual C compiler are used, but
7671 this can't be done directly in cross-compilation. You must make your
7672 own arrangements to provide suitable library functions for
7676 @opindex mhard-float
7677 Generate output containing floating point instructions. This is the
7678 default if you use the unmodified sources.
7681 @itemx -mno-abicalls
7683 @opindex mno-abicalls
7684 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7685 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7686 position independent code.
7689 @itemx -mno-long-calls
7690 @opindex mlong-calls
7691 @opindex mno-long-calls
7692 Do all calls with the @samp{JALR} instruction, which requires
7693 loading up a function's address into a register before the call.
7694 You need to use this switch, if you call outside of the current
7695 512 megabyte segment to functions that are not through pointers.
7698 @itemx -mno-half-pic
7700 @opindex mno-half-pic
7701 Put pointers to extern references into the data section and load them
7702 up, rather than put the references in the text section.
7704 @item -membedded-pic
7705 @itemx -mno-embedded-pic
7706 @opindex membedded-pic
7707 @opindex mno-embedded-pic
7708 Generate PIC code suitable for some embedded systems. All calls are
7709 made using PC relative address, and all data is addressed using the $gp
7710 register. No more than 65536 bytes of global data may be used. This
7711 requires GNU as and GNU ld which do most of the work. This currently
7712 only works on targets which use ECOFF; it does not work with ELF@.
7714 @item -membedded-data
7715 @itemx -mno-embedded-data
7716 @opindex membedded-data
7717 @opindex mno-embedded-data
7718 Allocate variables to the read-only data section first if possible, then
7719 next in the small data section if possible, otherwise in data. This gives
7720 slightly slower code than the default, but reduces the amount of RAM required
7721 when executing, and thus may be preferred for some embedded systems.
7723 @item -muninit-const-in-rodata
7724 @itemx -mno-uninit-const-in-rodata
7725 @opindex muninit-const-in-rodata
7726 @opindex mno-uninit-const-in-rodata
7727 When used together with @option{-membedded-data}, it will always store uninitialized
7728 const variables in the read-only data section.
7730 @item -msingle-float
7731 @itemx -mdouble-float
7732 @opindex msingle-float
7733 @opindex mdouble-float
7734 The @option{-msingle-float} switch tells gcc to assume that the floating
7735 point coprocessor only supports single precision operations, as on the
7736 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7737 double precision operations. This is the default.
7743 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7744 as on the @samp{r4650} chip.
7748 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7749 @option{-mcpu=r4650}.
7755 Enable 16-bit instructions.
7759 Use the entry and exit pseudo ops. This option can only be used with
7764 Compile code for the processor in little endian mode.
7765 The requisite libraries are assumed to exist.
7769 Compile code for the processor in big endian mode.
7770 The requisite libraries are assumed to exist.
7774 @cindex smaller data references (MIPS)
7775 @cindex gp-relative references (MIPS)
7776 Put global and static items less than or equal to @var{num} bytes into
7777 the small data or bss sections instead of the normal data or bss
7778 section. This allows the assembler to emit one word memory reference
7779 instructions based on the global pointer (@var{gp} or @var{$28}),
7780 instead of the normal two words used. By default, @var{num} is 8 when
7781 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7782 @option{-G @var{num}} switch is also passed to the assembler and linker.
7783 All modules should be compiled with the same @option{-G @var{num}}
7788 Tell the MIPS assembler to not run its preprocessor over user
7789 assembler files (with a @samp{.s} suffix) when assembling them.
7793 Pass an option to gas which will cause nops to be inserted if
7794 the read of the destination register of an mfhi or mflo instruction
7795 occurs in the following two instructions.
7799 Do not include the default crt0.
7801 @item -mflush-func=@var{func}
7802 @itemx -mno-flush-func
7803 @opindex mflush-func
7804 Specifies the function to call to flush the I and D caches, or to not
7805 call any such function. If called, the function must take the same
7806 arguments as the common @code{_flush_func()}, that is, the address of the
7807 memory range for which the cache is being flushed, the size of the
7808 memory range, and the number 3 (to flush both caches). The default
7809 depends on the target gcc was configured for, but commonly is either
7810 @samp{_flush_func} or @samp{__cpu_flush}.
7812 @item -mbranch-likely
7813 @itemx -mno-branch-likely
7814 @opindex mbranch-likely
7815 @opindex mno-branch-likely
7816 Enable or disable use of Branch Likely instructions, regardless of the
7817 default for the selected architecture. By default, Branch Likely
7818 instructions may be generated if they are supported by the selected
7819 architecture. An exception is for the MIPS32 and MIPS64 architectures
7820 and processors which implement those architectures; for those, Branch
7821 Likely instructions will not be generated by default because the MIPS32
7822 and MIPS64 architectures specifically deprecate their use.
7825 @node i386 and x86-64 Options
7826 @subsection Intel 386 and AMD x86-64 Options
7827 @cindex i386 Options
7828 @cindex x86-64 Options
7829 @cindex Intel 386 Options
7830 @cindex AMD x86-64 Options
7832 These @samp{-m} options are defined for the i386 and x86-64 family of
7836 @item -mcpu=@var{cpu-type}
7838 Tune to @var{cpu-type} everything applicable about the generated code, except
7839 for the ABI and the set of available instructions. The choices for
7840 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
7841 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
7842 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
7843 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
7844 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2} and @samp{c3}.
7846 While picking a specific @var{cpu-type} will schedule things appropriately
7847 for that particular chip, the compiler will not generate any code that
7848 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7849 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7850 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7851 AMD chips as opposed to the Intel ones.
7853 @item -march=@var{cpu-type}
7855 Generate instructions for the machine type @var{cpu-type}. The choices
7856 for @var{cpu-type} are the same as for @option{-mcpu}. Moreover,
7857 specifying @option{-march=@var{cpu-type}} implies @option{-mcpu=@var{cpu-type}}.
7866 @opindex mpentiumpro
7867 These options are synonyms for @option{-mcpu=i386}, @option{-mcpu=i486},
7868 @option{-mcpu=pentium}, and @option{-mcpu=pentiumpro} respectively.
7869 These synonyms are deprecated.
7871 @item -mfpmath=@var{unit}
7873 generate floating point arithmetics for selected unit @var{unit}. the choices
7878 Use the standard 387 floating point coprocessor present majority of chips and
7879 emulated otherwise. Code compiled with this option will run almost everywhere.
7880 The temporary results are computed in 80bit precesion instead of precision
7881 specified by the type resulting in slightly different results compared to most
7882 of other chips. See @option{-ffloat-store} for more detailed description.
7884 This is the default choice for i386 compiler.
7887 Use scalar floating point instructions present in the SSE instruction set.
7888 This instruction set is supported by Pentium3 and newer chips, in the AMD line
7889 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
7890 instruction set supports only single precision arithmetics, thus the double and
7891 extended precision arithmetics is still done using 387. Later version, present
7892 only in Pentium4 and the future AMD x86-64 chips supports double precision
7895 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
7896 @option{-msse2} switches to enable SSE extensions and make this option
7897 effective. For x86-64 compiler, these extensions are enabled by default.
7899 The resulting code should be considerably faster in majority of cases and avoid
7900 the numerical instability problems of 387 code, but may break some existing
7901 code that expects temporaries to be 80bit.
7903 This is the default choice for x86-64 compiler.
7906 Attempt to utilize both instruction sets at once. This effectivly double the
7907 amount of available registers and on chips with separate execution units for
7908 387 and SSE the execution resources too. Use this option with care, as it is
7909 still experimental, because gcc register allocator does not model separate
7910 functional units well resulting in instable performance.
7913 @item -masm=@var{dialect}
7914 @opindex masm=@var{dialect}
7915 Output asm instructions using selected @var{dialect}. Supported choices are
7916 @samp{intel} or @samp{att} (the default one).
7921 @opindex mno-ieee-fp
7922 Control whether or not the compiler uses IEEE floating point
7923 comparisons. These handle correctly the case where the result of a
7924 comparison is unordered.
7927 @opindex msoft-float
7928 Generate output containing library calls for floating point.
7929 @strong{Warning:} the requisite libraries are not part of GCC@.
7930 Normally the facilities of the machine's usual C compiler are used, but
7931 this can't be done directly in cross-compilation. You must make your
7932 own arrangements to provide suitable library functions for
7935 On machines where a function returns floating point results in the 80387
7936 register stack, some floating point opcodes may be emitted even if
7937 @option{-msoft-float} is used.
7939 @item -mno-fp-ret-in-387
7940 @opindex mno-fp-ret-in-387
7941 Do not use the FPU registers for return values of functions.
7943 The usual calling convention has functions return values of types
7944 @code{float} and @code{double} in an FPU register, even if there
7945 is no FPU@. The idea is that the operating system should emulate
7948 The option @option{-mno-fp-ret-in-387} causes such values to be returned
7949 in ordinary CPU registers instead.
7951 @item -mno-fancy-math-387
7952 @opindex mno-fancy-math-387
7953 Some 387 emulators do not support the @code{sin}, @code{cos} and
7954 @code{sqrt} instructions for the 387. Specify this option to avoid
7955 generating those instructions. This option is the default on FreeBSD,
7956 OpenBSD and NetBSD@. This option is overridden when @option{-march}
7957 indicates that the target cpu will always have an FPU and so the
7958 instruction will not need emulation. As of revision 2.6.1, these
7959 instructions are not generated unless you also use the
7960 @option{-funsafe-math-optimizations} switch.
7962 @item -malign-double
7963 @itemx -mno-align-double
7964 @opindex malign-double
7965 @opindex mno-align-double
7966 Control whether GCC aligns @code{double}, @code{long double}, and
7967 @code{long long} variables on a two word boundary or a one word
7968 boundary. Aligning @code{double} variables on a two word boundary will
7969 produce code that runs somewhat faster on a @samp{Pentium} at the
7970 expense of more memory.
7972 @strong{Warning:} if you use the @samp{-malign-double} switch,
7973 structures containing the above types will be aligned differently than
7974 the published application binary interface specifications for the 386.
7976 @item -m128bit-long-double
7977 @opindex m128bit-long-double
7978 Control the size of @code{long double} type. i386 application binary interface
7979 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
7980 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
7981 impossible to reach with 12 byte long doubles in the array accesses.
7983 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
7984 structures and arrays containing @code{long double} will change their size as
7985 well as function calling convention for function taking @code{long double}
7988 @item -m96bit-long-double
7989 @opindex m96bit-long-double
7990 Set the size of @code{long double} to 96 bits as required by the i386
7991 application binary interface. This is the default.
7994 @itemx -mno-svr3-shlib
7995 @opindex msvr3-shlib
7996 @opindex mno-svr3-shlib
7997 Control whether GCC places uninitialized local variables into the
7998 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
7999 into @code{bss}. These options are meaningful only on System V Release 3.
8003 Use a different function-calling convention, in which functions that
8004 take a fixed number of arguments return with the @code{ret} @var{num}
8005 instruction, which pops their arguments while returning. This saves one
8006 instruction in the caller since there is no need to pop the arguments
8009 You can specify that an individual function is called with this calling
8010 sequence with the function attribute @samp{stdcall}. You can also
8011 override the @option{-mrtd} option by using the function attribute
8012 @samp{cdecl}. @xref{Function Attributes}.
8014 @strong{Warning:} this calling convention is incompatible with the one
8015 normally used on Unix, so you cannot use it if you need to call
8016 libraries compiled with the Unix compiler.
8018 Also, you must provide function prototypes for all functions that
8019 take variable numbers of arguments (including @code{printf});
8020 otherwise incorrect code will be generated for calls to those
8023 In addition, seriously incorrect code will result if you call a
8024 function with too many arguments. (Normally, extra arguments are
8025 harmlessly ignored.)
8027 @item -mregparm=@var{num}
8029 Control how many registers are used to pass integer arguments. By
8030 default, no registers are used to pass arguments, and at most 3
8031 registers can be used. You can control this behavior for a specific
8032 function by using the function attribute @samp{regparm}.
8033 @xref{Function Attributes}.
8035 @strong{Warning:} if you use this switch, and
8036 @var{num} is nonzero, then you must build all modules with the same
8037 value, including any libraries. This includes the system libraries and
8040 @item -mpreferred-stack-boundary=@var{num}
8041 @opindex mpreferred-stack-boundary
8042 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8043 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8044 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8045 size (@option{-Os}), in which case the default is the minimum correct
8046 alignment (4 bytes for x86, and 8 bytes for x86-64).
8048 On Pentium and PentiumPro, @code{double} and @code{long double} values
8049 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8050 suffer significant run time performance penalties. On Pentium III, the
8051 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8052 penalties if it is not 16 byte aligned.
8054 To ensure proper alignment of this values on the stack, the stack boundary
8055 must be as aligned as that required by any value stored on the stack.
8056 Further, every function must be generated such that it keeps the stack
8057 aligned. Thus calling a function compiled with a higher preferred
8058 stack boundary from a function compiled with a lower preferred stack
8059 boundary will most likely misalign the stack. It is recommended that
8060 libraries that use callbacks always use the default setting.
8062 This extra alignment does consume extra stack space, and generally
8063 increases code size. Code that is sensitive to stack space usage, such
8064 as embedded systems and operating system kernels, may want to reduce the
8065 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8081 These switches enable or disable the use of built-in functions that allow
8082 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8084 @xref{X86 Built-in Functions}, for details of the functions enabled
8085 and disabled by these switches.
8087 To have SSE/SSE2 instructions generated automatically from floating-point
8088 code, see @option{-mfpmath=sse}.
8091 @itemx -mno-push-args
8093 @opindex mno-push-args
8094 Use PUSH operations to store outgoing parameters. This method is shorter
8095 and usually equally fast as method using SUB/MOV operations and is enabled
8096 by default. In some cases disabling it may improve performance because of
8097 improved scheduling and reduced dependencies.
8099 @item -maccumulate-outgoing-args
8100 @opindex maccumulate-outgoing-args
8101 If enabled, the maximum amount of space required for outgoing arguments will be
8102 computed in the function prologue. This is faster on most modern CPUs
8103 because of reduced dependencies, improved scheduling and reduced stack usage
8104 when preferred stack boundary is not equal to 2. The drawback is a notable
8105 increase in code size. This switch implies @option{-mno-push-args}.
8109 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8110 on thread-safe exception handling must compile and link all code with the
8111 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8112 @option{-D_MT}; when linking, it links in a special thread helper library
8113 @option{-lmingwthrd} which cleans up per thread exception handling data.
8115 @item -mno-align-stringops
8116 @opindex mno-align-stringops
8117 Do not align destination of inlined string operations. This switch reduces
8118 code size and improves performance in case the destination is already aligned,
8119 but gcc don't know about it.
8121 @item -minline-all-stringops
8122 @opindex minline-all-stringops
8123 By default GCC inlines string operations only when destination is known to be
8124 aligned at least to 4 byte boundary. This enables more inlining, increase code
8125 size, but may improve performance of code that depends on fast memcpy, strlen
8126 and memset for short lengths.
8128 @item -momit-leaf-frame-pointer
8129 @opindex momit-leaf-frame-pointer
8130 Don't keep the frame pointer in a register for leaf functions. This
8131 avoids the instructions to save, set up and restore frame pointers and
8132 makes an extra register available in leaf functions. The option
8133 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8134 which might make debugging harder.
8137 These @samp{-m} switches are supported in addition to the above
8138 on AMD x86-64 processors in 64-bit environments.
8145 Generate code for a 32-bit or 64-bit environment.
8146 The 32-bit environment sets int, long and pointer to 32 bits and
8147 generates code that runs on any i386 system.
8148 The 64-bit environment sets int to 32 bits and long and pointer
8149 to 64 bits and generates code for AMD's x86-64 architecture.
8152 @opindex no-red-zone
8153 Do not use a so called red zone for x86-64 code. The red zone is mandated
8154 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8155 stack pointer that will not be modified by signal or interrupt handlers
8156 and therefore can be used for temporary data without adjusting the stack
8157 pointer. The flag @option{-mno-red-zone} disables this red zone.
8159 @item -mcmodel=small
8160 @opindex mcmodel=small
8161 Generate code for the small code model: the program and its symbols must
8162 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8163 Programs can be statically or dynamically linked. This is the default
8166 @item -mcmodel=kernel
8167 @opindex mcmodel=kernel
8168 Generate code for the kernel code model. The kernel runs in the
8169 negative 2 GB of the address space.
8170 This model has to be used for Linux kernel code.
8172 @item -mcmodel=medium
8173 @opindex mcmodel=medium
8174 Generate code for the medium model: The program is linked in the lower 2
8175 GB of the address space but symbols can be located anywhere in the
8176 address space. Programs can be statically or dynamically linked, but
8177 building of shared libraries are not supported with the medium model.
8179 @item -mcmodel=large
8180 @opindex mcmodel=large
8181 Generate code for the large model: This model makes no assumptions
8182 about addresses and sizes of sections. Currently GCC does not implement
8187 @subsection HPPA Options
8188 @cindex HPPA Options
8190 These @samp{-m} options are defined for the HPPA family of computers:
8193 @item -march=@var{architecture-type}
8195 Generate code for the specified architecture. The choices for
8196 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8197 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8198 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8199 architecture option for your machine. Code compiled for lower numbered
8200 architectures will run on higher numbered architectures, but not the
8203 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8204 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8208 @itemx -mpa-risc-1-1
8209 @itemx -mpa-risc-2-0
8210 @opindex mpa-risc-1-0
8211 @opindex mpa-risc-1-1
8212 @opindex mpa-risc-2-0
8213 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8216 @opindex mbig-switch
8217 Generate code suitable for big switch tables. Use this option only if
8218 the assembler/linker complain about out of range branches within a switch
8221 @item -mjump-in-delay
8222 @opindex mjump-in-delay
8223 Fill delay slots of function calls with unconditional jump instructions
8224 by modifying the return pointer for the function call to be the target
8225 of the conditional jump.
8227 @item -mdisable-fpregs
8228 @opindex mdisable-fpregs
8229 Prevent floating point registers from being used in any manner. This is
8230 necessary for compiling kernels which perform lazy context switching of
8231 floating point registers. If you use this option and attempt to perform
8232 floating point operations, the compiler will abort.
8234 @item -mdisable-indexing
8235 @opindex mdisable-indexing
8236 Prevent the compiler from using indexing address modes. This avoids some
8237 rather obscure problems when compiling MIG generated code under MACH@.
8239 @item -mno-space-regs
8240 @opindex mno-space-regs
8241 Generate code that assumes the target has no space registers. This allows
8242 GCC to generate faster indirect calls and use unscaled index address modes.
8244 Such code is suitable for level 0 PA systems and kernels.
8246 @item -mfast-indirect-calls
8247 @opindex mfast-indirect-calls
8248 Generate code that assumes calls never cross space boundaries. This
8249 allows GCC to emit code which performs faster indirect calls.
8251 This option will not work in the presence of shared libraries or nested
8254 @item -mlong-load-store
8255 @opindex mlong-load-store
8256 Generate 3-instruction load and store sequences as sometimes required by
8257 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8260 @item -mportable-runtime
8261 @opindex mportable-runtime
8262 Use the portable calling conventions proposed by HP for ELF systems.
8266 Enable the use of assembler directives only GAS understands.
8268 @item -mschedule=@var{cpu-type}
8270 Schedule code according to the constraints for the machine type
8271 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8272 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8273 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8274 proper scheduling option for your machine. The default scheduling is
8278 @opindex mlinker-opt
8279 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8280 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8281 linkers in which they give bogus error messages when linking some programs.
8284 @opindex msoft-float
8285 Generate output containing library calls for floating point.
8286 @strong{Warning:} the requisite libraries are not available for all HPPA
8287 targets. Normally the facilities of the machine's usual C compiler are
8288 used, but this cannot be done directly in cross-compilation. You must make
8289 your own arrangements to provide suitable library functions for
8290 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8291 does provide software floating point support.
8293 @option{-msoft-float} changes the calling convention in the output file;
8294 therefore, it is only useful if you compile @emph{all} of a program with
8295 this option. In particular, you need to compile @file{libgcc.a}, the
8296 library that comes with GCC, with @option{-msoft-float} in order for
8301 Generate the predefine, @code{_SIO}, for server IO. The default is
8302 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8303 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8304 options are available under HP-UX and HI-UX.
8308 Use GNU ld specific options. This passes @option{-shared} to ld when
8309 building a shared library. It is the default when GCC is configured,
8310 explicitly or implicitly, with the GNU linker. This option does not
8311 have any affect on which ld is called, it only changes what parameters
8312 are passed to that ld. The ld that is called is determined by the
8313 @option{--with-ld} configure option, gcc's program search path, and
8314 finally by the user's @env{PATH}. The linker used by GCC can be printed
8315 using @samp{which `gcc -print-prog-name=ld`}.
8319 Use HP ld specific options. This passes @option{-b} to ld when building
8320 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8321 links. It is the default when GCC is configured, explicitly or
8322 implicitly, with the HP linker. This option does not have any affect on
8323 which ld is called, it only changes what parameters are passed to that
8324 ld. The ld that is called is determined by the @option{--with-ld}
8325 configure option, gcc's program search path, and finally by the user's
8326 @env{PATH}. The linker used by GCC can be printed using @samp{which
8327 `gcc -print-prog-name=ld`}.
8330 @opindex mno-long-calls
8331 Generate code that uses long call sequences. This ensures that a call
8332 is always able to reach linker generated stubs. The default is to generate
8333 long calls only when the distance from the call site to the beginning
8334 of the function or translation unit, as the case may be, exceeds a
8335 predefined limit set by the branch type being used. The limits for
8336 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8337 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8340 Distances are measured from the beginning of functions when using the
8341 @option{-ffunction-sections} option, or when using the @option{-mgas}
8342 and @option{-mno-portable-runtime} options together under HP-UX with
8345 It is normally not desirable to use this option as it will degrade
8346 performance. However, it may be useful in large applications,
8347 particularly when partial linking is used to build the application.
8349 The types of long calls used depends on the capabilities of the
8350 assembler and linker, and the type of code being generated. The
8351 impact on systems that support long absolute calls, and long pic
8352 symbol-difference or pc-relative calls should be relatively small.
8353 However, an indirect call is used on 32-bit ELF systems in pic code
8354 and it is quite long.
8358 @node Intel 960 Options
8359 @subsection Intel 960 Options
8361 These @samp{-m} options are defined for the Intel 960 implementations:
8364 @item -m@var{cpu-type}
8372 Assume the defaults for the machine type @var{cpu-type} for some of
8373 the other options, including instruction scheduling, floating point
8374 support, and addressing modes. The choices for @var{cpu-type} are
8375 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8376 @samp{sa}, and @samp{sb}.
8383 @opindex msoft-float
8384 The @option{-mnumerics} option indicates that the processor does support
8385 floating-point instructions. The @option{-msoft-float} option indicates
8386 that floating-point support should not be assumed.
8388 @item -mleaf-procedures
8389 @itemx -mno-leaf-procedures
8390 @opindex mleaf-procedures
8391 @opindex mno-leaf-procedures
8392 Do (or do not) attempt to alter leaf procedures to be callable with the
8393 @code{bal} instruction as well as @code{call}. This will result in more
8394 efficient code for explicit calls when the @code{bal} instruction can be
8395 substituted by the assembler or linker, but less efficient code in other
8396 cases, such as calls via function pointers, or using a linker that doesn't
8397 support this optimization.
8400 @itemx -mno-tail-call
8402 @opindex mno-tail-call
8403 Do (or do not) make additional attempts (beyond those of the
8404 machine-independent portions of the compiler) to optimize tail-recursive
8405 calls into branches. You may not want to do this because the detection of
8406 cases where this is not valid is not totally complete. The default is
8407 @option{-mno-tail-call}.
8409 @item -mcomplex-addr
8410 @itemx -mno-complex-addr
8411 @opindex mcomplex-addr
8412 @opindex mno-complex-addr
8413 Assume (or do not assume) that the use of a complex addressing mode is a
8414 win on this implementation of the i960. Complex addressing modes may not
8415 be worthwhile on the K-series, but they definitely are on the C-series.
8416 The default is currently @option{-mcomplex-addr} for all processors except
8420 @itemx -mno-code-align
8421 @opindex mcode-align
8422 @opindex mno-code-align
8423 Align code to 8-byte boundaries for faster fetching (or don't bother).
8424 Currently turned on by default for C-series implementations only.
8427 @item -mclean-linkage
8428 @itemx -mno-clean-linkage
8429 @opindex mclean-linkage
8430 @opindex mno-clean-linkage
8431 These options are not fully implemented.
8435 @itemx -mic2.0-compat
8436 @itemx -mic3.0-compat
8438 @opindex mic2.0-compat
8439 @opindex mic3.0-compat
8440 Enable compatibility with iC960 v2.0 or v3.0.
8444 @opindex masm-compat
8446 Enable compatibility with the iC960 assembler.
8448 @item -mstrict-align
8449 @itemx -mno-strict-align
8450 @opindex mstrict-align
8451 @opindex mno-strict-align
8452 Do not permit (do permit) unaligned accesses.
8456 Enable structure-alignment compatibility with Intel's gcc release version
8457 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8459 @item -mlong-double-64
8460 @opindex mlong-double-64
8461 Implement type @samp{long double} as 64-bit floating point numbers.
8462 Without the option @samp{long double} is implemented by 80-bit
8463 floating point numbers. The only reason we have it because there is
8464 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8465 is only useful for people using soft-float targets. Otherwise, we
8466 should recommend against use of it.
8470 @node DEC Alpha Options
8471 @subsection DEC Alpha Options
8473 These @samp{-m} options are defined for the DEC Alpha implementations:
8476 @item -mno-soft-float
8478 @opindex mno-soft-float
8479 @opindex msoft-float
8480 Use (do not use) the hardware floating-point instructions for
8481 floating-point operations. When @option{-msoft-float} is specified,
8482 functions in @file{libgcc.a} will be used to perform floating-point
8483 operations. Unless they are replaced by routines that emulate the
8484 floating-point operations, or compiled in such a way as to call such
8485 emulations routines, these routines will issue floating-point
8486 operations. If you are compiling for an Alpha without floating-point
8487 operations, you must ensure that the library is built so as not to call
8490 Note that Alpha implementations without floating-point operations are
8491 required to have floating-point registers.
8496 @opindex mno-fp-regs
8497 Generate code that uses (does not use) the floating-point register set.
8498 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8499 register set is not used, floating point operands are passed in integer
8500 registers as if they were integers and floating-point results are passed
8501 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8502 so any function with a floating-point argument or return value called by code
8503 compiled with @option{-mno-fp-regs} must also be compiled with that
8506 A typical use of this option is building a kernel that does not use,
8507 and hence need not save and restore, any floating-point registers.
8511 The Alpha architecture implements floating-point hardware optimized for
8512 maximum performance. It is mostly compliant with the IEEE floating
8513 point standard. However, for full compliance, software assistance is
8514 required. This option generates code fully IEEE compliant code
8515 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8516 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8517 defined during compilation. The resulting code is less efficient but is
8518 able to correctly support denormalized numbers and exceptional IEEE
8519 values such as not-a-number and plus/minus infinity. Other Alpha
8520 compilers call this option @option{-ieee_with_no_inexact}.
8522 @item -mieee-with-inexact
8523 @opindex mieee-with-inexact
8524 This is like @option{-mieee} except the generated code also maintains
8525 the IEEE @var{inexact-flag}. Turning on this option causes the
8526 generated code to implement fully-compliant IEEE math. In addition to
8527 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8528 macro. On some Alpha implementations the resulting code may execute
8529 significantly slower than the code generated by default. Since there is
8530 very little code that depends on the @var{inexact-flag}, you should
8531 normally not specify this option. Other Alpha compilers call this
8532 option @option{-ieee_with_inexact}.
8534 @item -mfp-trap-mode=@var{trap-mode}
8535 @opindex mfp-trap-mode
8536 This option controls what floating-point related traps are enabled.
8537 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8538 The trap mode can be set to one of four values:
8542 This is the default (normal) setting. The only traps that are enabled
8543 are the ones that cannot be disabled in software (e.g., division by zero
8547 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8551 Like @samp{su}, but the instructions are marked to be safe for software
8552 completion (see Alpha architecture manual for details).
8555 Like @samp{su}, but inexact traps are enabled as well.
8558 @item -mfp-rounding-mode=@var{rounding-mode}
8559 @opindex mfp-rounding-mode
8560 Selects the IEEE rounding mode. Other Alpha compilers call this option
8561 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8566 Normal IEEE rounding mode. Floating point numbers are rounded towards
8567 the nearest machine number or towards the even machine number in case
8571 Round towards minus infinity.
8574 Chopped rounding mode. Floating point numbers are rounded towards zero.
8577 Dynamic rounding mode. A field in the floating point control register
8578 (@var{fpcr}, see Alpha architecture reference manual) controls the
8579 rounding mode in effect. The C library initializes this register for
8580 rounding towards plus infinity. Thus, unless your program modifies the
8581 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8584 @item -mtrap-precision=@var{trap-precision}
8585 @opindex mtrap-precision
8586 In the Alpha architecture, floating point traps are imprecise. This
8587 means without software assistance it is impossible to recover from a
8588 floating trap and program execution normally needs to be terminated.
8589 GCC can generate code that can assist operating system trap handlers
8590 in determining the exact location that caused a floating point trap.
8591 Depending on the requirements of an application, different levels of
8592 precisions can be selected:
8596 Program precision. This option is the default and means a trap handler
8597 can only identify which program caused a floating point exception.
8600 Function precision. The trap handler can determine the function that
8601 caused a floating point exception.
8604 Instruction precision. The trap handler can determine the exact
8605 instruction that caused a floating point exception.
8608 Other Alpha compilers provide the equivalent options called
8609 @option{-scope_safe} and @option{-resumption_safe}.
8611 @item -mieee-conformant
8612 @opindex mieee-conformant
8613 This option marks the generated code as IEEE conformant. You must not
8614 use this option unless you also specify @option{-mtrap-precision=i} and either
8615 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8616 is to emit the line @samp{.eflag 48} in the function prologue of the
8617 generated assembly file. Under DEC Unix, this has the effect that
8618 IEEE-conformant math library routines will be linked in.
8620 @item -mbuild-constants
8621 @opindex mbuild-constants
8622 Normally GCC examines a 32- or 64-bit integer constant to
8623 see if it can construct it from smaller constants in two or three
8624 instructions. If it cannot, it will output the constant as a literal and
8625 generate code to load it from the data segment at runtime.
8627 Use this option to require GCC to construct @emph{all} integer constants
8628 using code, even if it takes more instructions (the maximum is six).
8630 You would typically use this option to build a shared library dynamic
8631 loader. Itself a shared library, it must relocate itself in memory
8632 before it can find the variables and constants in its own data segment.
8638 Select whether to generate code to be assembled by the vendor-supplied
8639 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8657 Indicate whether GCC should generate code to use the optional BWX,
8658 CIX, FIX and MAX instruction sets. The default is to use the instruction
8659 sets supported by the CPU type specified via @option{-mcpu=} option or that
8660 of the CPU on which GCC was built if none was specified.
8665 @opindex mfloat-ieee
8666 Generate code that uses (does not use) VAX F and G floating point
8667 arithmetic instead of IEEE single and double precision.
8669 @item -mexplicit-relocs
8670 @itemx -mno-explicit-relocs
8671 @opindex mexplicit-relocs
8672 @opindex mno-explicit-relocs
8673 Older Alpha assemblers provided no way to generate symbol relocations
8674 except via assembler macros. Use of these macros does not allow
8675 optimial instruction scheduling. GNU binutils as of version 2.12
8676 supports a new syntax that allows the compiler to explicitly mark
8677 which relocations should apply to which instructions. This option
8678 is mostly useful for debugging, as GCC detects the capabilities of
8679 the assembler when it is built and sets the default accordingly.
8683 @opindex msmall-data
8684 @opindex mlarge-data
8685 When @option{-mexplicit-relocs} is in effect, static data is
8686 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8687 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8688 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8689 16-bit relocations off of the @code{$gp} register. This limits the
8690 size of the small data area to 64KB, but allows the variables to be
8691 directly accessed via a single instruction.
8693 The default is @option{-mlarge-data}. With this option the data area
8694 is limited to just below 2GB. Programs that require more than 2GB of
8695 data must use @code{malloc} or @code{mmap} to allocate the data in the
8696 heap instead of in the program's data segment.
8698 When generating code for shared libraries, @option{-fpic} implies
8699 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8701 @item -mcpu=@var{cpu_type}
8703 Set the instruction set and instruction scheduling parameters for
8704 machine type @var{cpu_type}. You can specify either the @samp{EV}
8705 style name or the corresponding chip number. GCC supports scheduling
8706 parameters for the EV4, EV5 and EV6 family of processors and will
8707 choose the default values for the instruction set from the processor
8708 you specify. If you do not specify a processor type, GCC will default
8709 to the processor on which the compiler was built.
8711 Supported values for @var{cpu_type} are
8717 Schedules as an EV4 and has no instruction set extensions.
8721 Schedules as an EV5 and has no instruction set extensions.
8725 Schedules as an EV5 and supports the BWX extension.
8730 Schedules as an EV5 and supports the BWX and MAX extensions.
8734 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8738 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8741 @item -mtune=@var{cpu_type}
8743 Set only the instruction scheduling parameters for machine type
8744 @var{cpu_type}. The instruction set is not changed.
8746 @item -mmemory-latency=@var{time}
8747 @opindex mmemory-latency
8748 Sets the latency the scheduler should assume for typical memory
8749 references as seen by the application. This number is highly
8750 dependent on the memory access patterns used by the application
8751 and the size of the external cache on the machine.
8753 Valid options for @var{time} are
8757 A decimal number representing clock cycles.
8763 The compiler contains estimates of the number of clock cycles for
8764 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8765 (also called Dcache, Scache, and Bcache), as well as to main memory.
8766 Note that L3 is only valid for EV5.
8771 @node DEC Alpha/VMS Options
8772 @subsection DEC Alpha/VMS Options
8774 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8777 @item -mvms-return-codes
8778 @opindex mvms-return-codes
8779 Return VMS condition codes from main. The default is to return POSIX
8780 style condition (e.g.@ error) codes.
8783 @node Clipper Options
8784 @subsection Clipper Options
8786 These @samp{-m} options are defined for the Clipper implementations:
8791 Produce code for a C300 Clipper processor. This is the default.
8795 Produce code for a C400 Clipper processor, i.e.@: use floating point
8799 @node H8/300 Options
8800 @subsection H8/300 Options
8802 These @samp{-m} options are defined for the H8/300 implementations:
8807 Shorten some address references at link time, when possible; uses the
8808 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8809 ld.info, Using ld}, for a fuller description.
8813 Generate code for the H8/300H@.
8817 Generate code for the H8S@.
8821 Generate code for the H8S and H8/300H in the normal mode. This switch
8822 must be used either with -mh or -ms.
8826 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8830 Make @code{int} data 32 bits by default.
8834 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8835 The default for the H8/300H and H8S is to align longs and floats on 4
8837 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8838 This option has no effect on the H8/300.
8842 @subsection SH Options
8844 These @samp{-m} options are defined for the SH implementations:
8849 Generate code for the SH1.
8853 Generate code for the SH2.
8857 Generate code for the SH3.
8861 Generate code for the SH3e.
8865 Generate code for the SH4 without a floating-point unit.
8867 @item -m4-single-only
8868 @opindex m4-single-only
8869 Generate code for the SH4 with a floating-point unit that only
8870 supports single-precision arithmetic.
8874 Generate code for the SH4 assuming the floating-point unit is in
8875 single-precision mode by default.
8879 Generate code for the SH4.
8883 Compile code for the processor in big endian mode.
8887 Compile code for the processor in little endian mode.
8891 Align doubles at 64-bit boundaries. Note that this changes the calling
8892 conventions, and thus some functions from the standard C library will
8893 not work unless you recompile it first with @option{-mdalign}.
8897 Shorten some address references at link time, when possible; uses the
8898 linker option @option{-relax}.
8902 Use 32-bit offsets in @code{switch} tables. The default is to use
8907 Enable the use of the instruction @code{fmovd}.
8911 Comply with the calling conventions defined by Hitachi.
8915 Mark the @code{MAC} register as call-clobbered, even if
8916 @option{-mhitachi} is given.
8920 Increase IEEE-compliance of floating-point code.
8924 Dump instruction size and location in the assembly code.
8928 This option is deprecated. It pads structures to multiple of 4 bytes,
8929 which is incompatible with the SH ABI@.
8933 Optimize for space instead of speed. Implied by @option{-Os}.
8937 When generating position-independent code, emit function calls using
8938 the Global Offset Table instead of the Procedure Linkage Table.
8942 Generate a library function call to invalidate instruction cache
8943 entries, after fixing up a trampoline. This library function call
8944 doesn't assume it can write to the whole memory address space. This
8945 is the default when the target is @code{sh-*-linux*}.
8948 @node System V Options
8949 @subsection Options for System V
8951 These additional options are available on System V Release 4 for
8952 compatibility with other compilers on those systems:
8957 Create a shared object.
8958 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
8962 Identify the versions of each tool used by the compiler, in a
8963 @code{.ident} assembler directive in the output.
8967 Refrain from adding @code{.ident} directives to the output file (this is
8970 @item -YP,@var{dirs}
8972 Search the directories @var{dirs}, and no others, for libraries
8973 specified with @option{-l}.
8977 Look in the directory @var{dir} to find the M4 preprocessor.
8978 The assembler uses this option.
8979 @c This is supposed to go with a -Yd for predefined M4 macro files, but
8980 @c the generic assembler that comes with Solaris takes just -Ym.
8983 @node TMS320C3x/C4x Options
8984 @subsection TMS320C3x/C4x Options
8985 @cindex TMS320C3x/C4x Options
8987 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
8991 @item -mcpu=@var{cpu_type}
8993 Set the instruction set, register set, and instruction scheduling
8994 parameters for machine type @var{cpu_type}. Supported values for
8995 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
8996 @samp{c44}. The default is @samp{c40} to generate code for the
9001 @itemx -msmall-memory
9003 @opindex mbig-memory
9005 @opindex msmall-memory
9007 Generates code for the big or small memory model. The small memory
9008 model assumed that all data fits into one 64K word page. At run-time
9009 the data page (DP) register must be set to point to the 64K page
9010 containing the .bss and .data program sections. The big memory model is
9011 the default and requires reloading of the DP register for every direct
9018 Allow (disallow) allocation of general integer operands into the block
9025 Enable (disable) generation of code using decrement and branch,
9026 DBcond(D), instructions. This is enabled by default for the C4x. To be
9027 on the safe side, this is disabled for the C3x, since the maximum
9028 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9029 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9030 that it can utilize the decrement and branch instruction, but will give
9031 up if there is more than one memory reference in the loop. Thus a loop
9032 where the loop counter is decremented can generate slightly more
9033 efficient code, in cases where the RPTB instruction cannot be utilized.
9035 @item -mdp-isr-reload
9037 @opindex mdp-isr-reload
9039 Force the DP register to be saved on entry to an interrupt service
9040 routine (ISR), reloaded to point to the data section, and restored on
9041 exit from the ISR@. This should not be required unless someone has
9042 violated the small memory model by modifying the DP register, say within
9049 For the C3x use the 24-bit MPYI instruction for integer multiplies
9050 instead of a library call to guarantee 32-bit results. Note that if one
9051 of the operands is a constant, then the multiplication will be performed
9052 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9053 then squaring operations are performed inline instead of a library call.
9056 @itemx -mno-fast-fix
9058 @opindex mno-fast-fix
9059 The C3x/C4x FIX instruction to convert a floating point value to an
9060 integer value chooses the nearest integer less than or equal to the
9061 floating point value rather than to the nearest integer. Thus if the
9062 floating point number is negative, the result will be incorrectly
9063 truncated an additional code is necessary to detect and correct this
9064 case. This option can be used to disable generation of the additional
9065 code required to correct the result.
9071 Enable (disable) generation of repeat block sequences using the RPTB
9072 instruction for zero overhead looping. The RPTB construct is only used
9073 for innermost loops that do not call functions or jump across the loop
9074 boundaries. There is no advantage having nested RPTB loops due to the
9075 overhead required to save and restore the RC, RS, and RE registers.
9076 This is enabled by default with @option{-O2}.
9078 @item -mrpts=@var{count}
9082 Enable (disable) the use of the single instruction repeat instruction
9083 RPTS@. If a repeat block contains a single instruction, and the loop
9084 count can be guaranteed to be less than the value @var{count}, GCC will
9085 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9086 then a RPTS will be emitted even if the loop count cannot be determined
9087 at compile time. Note that the repeated instruction following RPTS does
9088 not have to be reloaded from memory each iteration, thus freeing up the
9089 CPU buses for operands. However, since interrupts are blocked by this
9090 instruction, it is disabled by default.
9092 @item -mloop-unsigned
9093 @itemx -mno-loop-unsigned
9094 @opindex mloop-unsigned
9095 @opindex mno-loop-unsigned
9096 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9097 is @math{2^{31} + 1} since these instructions test if the iteration count is
9098 negative to terminate the loop. If the iteration count is unsigned
9099 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9100 exceeded. This switch allows an unsigned iteration count.
9104 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9105 with. This also enforces compatibility with the API employed by the TI
9106 C3x C compiler. For example, long doubles are passed as structures
9107 rather than in floating point registers.
9113 Generate code that uses registers (stack) for passing arguments to functions.
9114 By default, arguments are passed in registers where possible rather
9115 than by pushing arguments on to the stack.
9117 @item -mparallel-insns
9118 @itemx -mno-parallel-insns
9119 @opindex mparallel-insns
9120 @opindex mno-parallel-insns
9121 Allow the generation of parallel instructions. This is enabled by
9122 default with @option{-O2}.
9124 @item -mparallel-mpy
9125 @itemx -mno-parallel-mpy
9126 @opindex mparallel-mpy
9127 @opindex mno-parallel-mpy
9128 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9129 provided @option{-mparallel-insns} is also specified. These instructions have
9130 tight register constraints which can pessimize the code generation
9136 @subsection V850 Options
9137 @cindex V850 Options
9139 These @samp{-m} options are defined for V850 implementations:
9143 @itemx -mno-long-calls
9144 @opindex mlong-calls
9145 @opindex mno-long-calls
9146 Treat all calls as being far away (near). If calls are assumed to be
9147 far away, the compiler will always load the functions address up into a
9148 register, and call indirect through the pointer.
9154 Do not optimize (do optimize) basic blocks that use the same index
9155 pointer 4 or more times to copy pointer into the @code{ep} register, and
9156 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9157 option is on by default if you optimize.
9159 @item -mno-prolog-function
9160 @itemx -mprolog-function
9161 @opindex mno-prolog-function
9162 @opindex mprolog-function
9163 Do not use (do use) external functions to save and restore registers at
9164 the prolog and epilog of a function. The external functions are slower,
9165 but use less code space if more than one function saves the same number
9166 of registers. The @option{-mprolog-function} option is on by default if
9171 Try to make the code as small as possible. At present, this just turns
9172 on the @option{-mep} and @option{-mprolog-function} options.
9176 Put static or global variables whose size is @var{n} bytes or less into
9177 the tiny data area that register @code{ep} points to. The tiny data
9178 area can hold up to 256 bytes in total (128 bytes for byte references).
9182 Put static or global variables whose size is @var{n} bytes or less into
9183 the small data area that register @code{gp} points to. The small data
9184 area can hold up to 64 kilobytes.
9188 Put static or global variables whose size is @var{n} bytes or less into
9189 the first 32 kilobytes of memory.
9193 Specify that the target processor is the V850.
9196 @opindex mbig-switch
9197 Generate code suitable for big switch tables. Use this option only if
9198 the assembler/linker complain about out of range branches within a switch
9203 This option will cause r2 and r5 to be used in the code generated by
9204 the compiler. This setting is the default.
9207 @opindex -mno-app-regs
9208 This option will cause r2 and r5 to be treated as fixed registers.
9212 Specify that the target processor is the V850E. The preprocessor
9213 constant @samp{__v850e__} will be defined if this option is used.
9215 If neither @option{-mv850} nor @option{-mv850e} are defined
9216 then a default target processor will be chosen and the relevant
9217 @samp{__v850*__} preprocessor constant will be defined.
9219 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9220 defined, regardless of which processor variant is the target.
9222 @item -mdisable-callt
9223 @opindex -mdisable-callt
9224 This option will suppress generation of the CALLT instruction for the
9225 v850e flavors of the v850 architecture. The default is
9226 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9231 @subsection ARC Options
9234 These options are defined for ARC implementations:
9239 Compile code for little endian mode. This is the default.
9243 Compile code for big endian mode.
9246 @opindex mmangle-cpu
9247 Prepend the name of the cpu to all public symbol names.
9248 In multiple-processor systems, there are many ARC variants with different
9249 instruction and register set characteristics. This flag prevents code
9250 compiled for one cpu to be linked with code compiled for another.
9251 No facility exists for handling variants that are ``almost identical''.
9252 This is an all or nothing option.
9254 @item -mcpu=@var{cpu}
9256 Compile code for ARC variant @var{cpu}.
9257 Which variants are supported depend on the configuration.
9258 All variants support @option{-mcpu=base}, this is the default.
9260 @item -mtext=@var{text-section}
9261 @itemx -mdata=@var{data-section}
9262 @itemx -mrodata=@var{readonly-data-section}
9266 Put functions, data, and readonly data in @var{text-section},
9267 @var{data-section}, and @var{readonly-data-section} respectively
9268 by default. This can be overridden with the @code{section} attribute.
9269 @xref{Variable Attributes}.
9274 @subsection NS32K Options
9275 @cindex NS32K options
9277 These are the @samp{-m} options defined for the 32000 series. The default
9278 values for these options depends on which style of 32000 was selected when
9279 the compiler was configured; the defaults for the most common choices are
9287 Generate output for a 32032. This is the default
9288 when the compiler is configured for 32032 and 32016 based systems.
9294 Generate output for a 32332. This is the default
9295 when the compiler is configured for 32332-based systems.
9301 Generate output for a 32532. This is the default
9302 when the compiler is configured for 32532-based systems.
9306 Generate output containing 32081 instructions for floating point.
9307 This is the default for all systems.
9311 Generate output containing 32381 instructions for floating point. This
9312 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9313 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9317 Try and generate multiply-add floating point instructions @code{polyF}
9318 and @code{dotF}. This option is only available if the @option{-m32381}
9319 option is in effect. Using these instructions requires changes to
9320 register allocation which generally has a negative impact on
9321 performance. This option should only be enabled when compiling code
9322 particularly likely to make heavy use of multiply-add instructions.
9325 @opindex mnomulti-add
9326 Do not try and generate multiply-add floating point instructions
9327 @code{polyF} and @code{dotF}. This is the default on all platforms.
9330 @opindex msoft-float
9331 Generate output containing library calls for floating point.
9332 @strong{Warning:} the requisite libraries may not be available.
9334 @item -mieee-compare
9335 @itemx -mno-ieee-compare
9336 @opindex mieee-compare
9337 @opindex mno-ieee-compare
9338 Control whether or not the compiler uses IEEE floating point
9339 comparisons. These handle correctly the case where the result of a
9340 comparison is unordered.
9341 @strong{Warning:} the requisite kernel support may not be available.
9344 @opindex mnobitfield
9345 Do not use the bit-field instructions. On some machines it is faster to
9346 use shifting and masking operations. This is the default for the pc532.
9350 Do use the bit-field instructions. This is the default for all platforms
9355 Use a different function-calling convention, in which functions
9356 that take a fixed number of arguments return pop their
9357 arguments on return with the @code{ret} instruction.
9359 This calling convention is incompatible with the one normally
9360 used on Unix, so you cannot use it if you need to call libraries
9361 compiled with the Unix compiler.
9363 Also, you must provide function prototypes for all functions that
9364 take variable numbers of arguments (including @code{printf});
9365 otherwise incorrect code will be generated for calls to those
9368 In addition, seriously incorrect code will result if you call a
9369 function with too many arguments. (Normally, extra arguments are
9370 harmlessly ignored.)
9372 This option takes its name from the 680x0 @code{rtd} instruction.
9377 Use a different function-calling convention where the first two arguments
9378 are passed in registers.
9380 This calling convention is incompatible with the one normally
9381 used on Unix, so you cannot use it if you need to call libraries
9382 compiled with the Unix compiler.
9385 @opindex mnoregparam
9386 Do not pass any arguments in registers. This is the default for all
9391 It is OK to use the sb as an index register which is always loaded with
9392 zero. This is the default for the pc532-netbsd target.
9396 The sb register is not available for use or has not been initialized to
9397 zero by the run time system. This is the default for all targets except
9398 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9399 @option{-fpic} is set.
9403 Many ns32000 series addressing modes use displacements of up to 512MB@.
9404 If an address is above 512MB then displacements from zero can not be used.
9405 This option causes code to be generated which can be loaded above 512MB@.
9406 This may be useful for operating systems or ROM code.
9410 Assume code will be loaded in the first 512MB of virtual address space.
9411 This is the default for all platforms.
9417 @subsection AVR Options
9420 These options are defined for AVR implementations:
9423 @item -mmcu=@var{mcu}
9425 Specify ATMEL AVR instruction set or MCU type.
9427 Instruction set avr1 is for the minimal AVR core, not supported by the C
9428 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9429 attiny11, attiny12, attiny15, attiny28).
9431 Instruction set avr2 (default) is for the classic AVR core with up to
9432 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9433 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9434 at90c8534, at90s8535).
9436 Instruction set avr3 is for the classic AVR core with up to 128K program
9437 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9439 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9440 memory space (MCU types: atmega8, atmega83, atmega85).
9442 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9443 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9444 atmega64, atmega128, at43usb355, at94k).
9448 Output instruction sizes to the asm file.
9450 @item -minit-stack=@var{N}
9451 @opindex minit-stack
9452 Specify the initial stack address, which may be a symbol or numeric value,
9453 @samp{__stack} is the default.
9455 @item -mno-interrupts
9456 @opindex mno-interrupts
9457 Generated code is not compatible with hardware interrupts.
9458 Code size will be smaller.
9460 @item -mcall-prologues
9461 @opindex mcall-prologues
9462 Functions prologues/epilogues expanded as call to appropriate
9463 subroutines. Code size will be smaller.
9465 @item -mno-tablejump
9466 @opindex mno-tablejump
9467 Do not generate tablejump insns which sometimes increase code size.
9470 @opindex mtiny-stack
9471 Change only the low 8 bits of the stack pointer.
9475 @subsection MCore Options
9476 @cindex MCore options
9478 These are the @samp{-m} options defined for the Motorola M*Core
9488 @opindex mno-hardlit
9489 Inline constants into the code stream if it can be done in two
9490 instructions or less.
9498 Use the divide instruction. (Enabled by default).
9500 @item -mrelax-immediate
9501 @itemx -mrelax-immediate
9502 @itemx -mno-relax-immediate
9503 @opindex mrelax-immediate
9504 @opindex mrelax-immediate
9505 @opindex mno-relax-immediate
9506 Allow arbitrary sized immediates in bit operations.
9508 @item -mwide-bitfields
9509 @itemx -mwide-bitfields
9510 @itemx -mno-wide-bitfields
9511 @opindex mwide-bitfields
9512 @opindex mwide-bitfields
9513 @opindex mno-wide-bitfields
9514 Always treat bit-fields as int-sized.
9516 @item -m4byte-functions
9517 @itemx -m4byte-functions
9518 @itemx -mno-4byte-functions
9519 @opindex m4byte-functions
9520 @opindex m4byte-functions
9521 @opindex mno-4byte-functions
9522 Force all functions to be aligned to a four byte boundary.
9524 @item -mcallgraph-data
9525 @itemx -mcallgraph-data
9526 @itemx -mno-callgraph-data
9527 @opindex mcallgraph-data
9528 @opindex mcallgraph-data
9529 @opindex mno-callgraph-data
9530 Emit callgraph information.
9534 @itemx -mno-slow-bytes
9535 @opindex mslow-bytes
9536 @opindex mslow-bytes
9537 @opindex mno-slow-bytes
9538 Prefer word access when reading byte quantities.
9540 @item -mlittle-endian
9541 @itemx -mlittle-endian
9543 @opindex mlittle-endian
9544 @opindex mlittle-endian
9545 @opindex mbig-endian
9546 Generate code for a little endian target.
9554 Generate code for the 210 processor.
9558 @subsection IA-64 Options
9559 @cindex IA-64 Options
9561 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9565 @opindex mbig-endian
9566 Generate code for a big endian target. This is the default for HP-UX@.
9568 @item -mlittle-endian
9569 @opindex mlittle-endian
9570 Generate code for a little endian target. This is the default for AIX5
9577 Generate (or don't) code for the GNU assembler. This is the default.
9578 @c Also, this is the default if the configure option @option{--with-gnu-as}
9585 Generate (or don't) code for the GNU linker. This is the default.
9586 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9591 Generate code that does not use a global pointer register. The result
9592 is not position independent code, and violates the IA-64 ABI@.
9594 @item -mvolatile-asm-stop
9595 @itemx -mno-volatile-asm-stop
9596 @opindex mvolatile-asm-stop
9597 @opindex mno-volatile-asm-stop
9598 Generate (or don't) a stop bit immediately before and after volatile asm
9603 Generate code that works around Itanium B step errata.
9605 @item -mregister-names
9606 @itemx -mno-register-names
9607 @opindex mregister-names
9608 @opindex mno-register-names
9609 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9610 the stacked registers. This may make assembler output more readable.
9616 Disable (or enable) optimizations that use the small data section. This may
9617 be useful for working around optimizer bugs.
9620 @opindex mconstant-gp
9621 Generate code that uses a single constant global pointer value. This is
9622 useful when compiling kernel code.
9626 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9627 This is useful when compiling firmware code.
9629 @item -minline-float-divide-min-latency
9630 @opindex minline-float-divide-min-latency
9631 Generate code for inline divides of floating point values
9632 using the minimum latency algorithm.
9634 @item -minline-float-divide-max-throughput
9635 @opindex minline-float-divide-max-throughput
9636 Generate code for inline divides of floating point values
9637 using the maximum throughput algorithm.
9639 @item -minline-int-divide-min-latency
9640 @opindex minline-int-divide-min-latency
9641 Generate code for inline divides of integer values
9642 using the minimum latency algorithm.
9644 @item -minline-int-divide-max-throughput
9645 @opindex minline-int-divide-max-throughput
9646 Generate code for inline divides of integer values
9647 using the maximum throughput algorithm.
9649 @item -mno-dwarf2-asm
9651 @opindex mno-dwarf2-asm
9652 @opindex mdwarf2-asm
9653 Don't (or do) generate assembler code for the DWARF2 line number debugging
9654 info. This may be useful when not using the GNU assembler.
9656 @item -mfixed-range=@var{register-range}
9657 @opindex mfixed-range
9658 Generate code treating the given register range as fixed registers.
9659 A fixed register is one that the register allocator can not use. This is
9660 useful when compiling kernel code. A register range is specified as
9661 two registers separated by a dash. Multiple register ranges can be
9662 specified separated by a comma.
9666 @subsection D30V Options
9667 @cindex D30V Options
9669 These @samp{-m} options are defined for D30V implementations:
9674 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
9675 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
9676 memory, which starts at location @code{0x80000000}.
9680 Same as the @option{-mextmem} switch.
9684 Link the @samp{.text} section into onchip text memory, which starts at
9685 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
9686 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
9687 into onchip data memory, which starts at location @code{0x20000000}.
9689 @item -mno-asm-optimize
9690 @itemx -masm-optimize
9691 @opindex mno-asm-optimize
9692 @opindex masm-optimize
9693 Disable (enable) passing @option{-O} to the assembler when optimizing.
9694 The assembler uses the @option{-O} option to automatically parallelize
9695 adjacent short instructions where possible.
9697 @item -mbranch-cost=@var{n}
9698 @opindex mbranch-cost
9699 Increase the internal costs of branches to @var{n}. Higher costs means
9700 that the compiler will issue more instructions to avoid doing a branch.
9703 @item -mcond-exec=@var{n}
9705 Specify the maximum number of conditionally executed instructions that
9706 replace a branch. The default is 4.
9709 @node S/390 and zSeries Options
9710 @subsection S/390 and zSeries Options
9711 @cindex S/390 and zSeries Options
9713 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9718 @opindex mhard-float
9719 @opindex msoft-float
9720 Use (do not use) the hardware floating-point instructions and registers
9721 for floating-point operations. When @option{-msoft-float} is specified,
9722 functions in @file{libgcc.a} will be used to perform floating-point
9723 operations. When @option{-mhard-float} is specified, the compiler
9724 generates IEEE floating-point instructions. This is the default.
9727 @itemx -mno-backchain
9729 @opindex mno-backchain
9730 Generate (or do not generate) code which maintains an explicit
9731 backchain within the stack frame that points to the caller's frame.
9732 This is currently needed to allow debugging. The default is to
9733 generate the backchain.
9736 @itemx -mno-small-exec
9737 @opindex msmall-exec
9738 @opindex mno-small-exec
9739 Generate (or do not generate) code using the @code{bras} instruction
9740 to do subroutine calls.
9741 This only works reliably if the total executable size does not
9742 exceed 64k. The default is to use the @code{basr} instruction instead,
9743 which does not have this limitation.
9749 When @option{-m31} is specified, generate code compliant to the
9750 Linux for S/390 ABI@. When @option{-m64} is specified, generate
9751 code compliant to the Linux for zSeries ABI@. This allows GCC in
9752 particular to generate 64-bit instructions. For the @samp{s390}
9753 targets, the default is @option{-m31}, while the @samp{s390x}
9754 targets default to @option{-m64}.
9760 Generate (or do not generate) code using the @code{mvcle} instruction
9761 to perform block moves. When @option{-mno-mvcle} is specifed,
9762 use a @code{mvc} loop instead. This is the default.
9768 Print (or do not print) additional debug information when compiling.
9769 The default is to not print debug information.
9774 @subsection CRIS Options
9775 @cindex CRIS Options
9777 These options are defined specifically for the CRIS ports.
9780 @item -march=@var{architecture-type}
9781 @itemx -mcpu=@var{architecture-type}
9784 Generate code for the specified architecture. The choices for
9785 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9786 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
9787 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9790 @item -mtune=@var{architecture-type}
9792 Tune to @var{architecture-type} everything applicable about the generated
9793 code, except for the ABI and the set of available instructions. The
9794 choices for @var{architecture-type} are the same as for
9795 @option{-march=@var{architecture-type}}.
9797 @item -mmax-stack-frame=@var{n}
9798 @opindex mmax-stack-frame
9799 Warn when the stack frame of a function exceeds @var{n} bytes.
9801 @item -melinux-stacksize=@var{n}
9802 @opindex melinux-stacksize
9803 Only available with the @samp{cris-axis-aout} target. Arranges for
9804 indications in the program to the kernel loader that the stack of the
9805 program should be set to @var{n} bytes.
9811 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9812 @option{-march=v3} and @option{-march=v8} respectively.
9816 Enable CRIS-specific verbose debug-related information in the assembly
9817 code. This option also has the effect to turn off the @samp{#NO_APP}
9818 formatted-code indicator to the assembler at the beginning of the
9823 Do not use condition-code results from previous instruction; always emit
9824 compare and test instructions before use of condition codes.
9826 @item -mno-side-effects
9827 @opindex mno-side-effects
9828 Do not emit instructions with side-effects in addressing modes other than
9832 @itemx -mno-stack-align
9834 @itemx -mno-data-align
9835 @itemx -mconst-align
9836 @itemx -mno-const-align
9837 @opindex mstack-align
9838 @opindex mno-stack-align
9839 @opindex mdata-align
9840 @opindex mno-data-align
9841 @opindex mconst-align
9842 @opindex mno-const-align
9843 These options (no-options) arranges (eliminate arrangements) for the
9844 stack-frame, individual data and constants to be aligned for the maximum
9845 single data access size for the chosen CPU model. The default is to
9846 arrange for 32-bit alignment. ABI details such as structure layout are
9847 not affected by these options.
9855 Similar to the stack- data- and const-align options above, these options
9856 arrange for stack-frame, writable data and constants to all be 32-bit,
9857 16-bit or 8-bit aligned. The default is 32-bit alignment.
9859 @item -mno-prologue-epilogue
9860 @itemx -mprologue-epilogue
9861 @opindex mno-prologue-epilogue
9862 @opindex mprologue-epilogue
9863 With @option{-mno-prologue-epilogue}, the normal function prologue and
9864 epilogue that sets up the stack-frame are omitted and no return
9865 instructions or return sequences are generated in the code. Use this
9866 option only together with visual inspection of the compiled code: no
9867 warnings or errors are generated when call-saved registers must be saved,
9868 or storage for local variable needs to be allocated.
9874 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9875 instruction sequences that load addresses for functions from the PLT part
9876 of the GOT rather than (traditional on other architectures) calls to the
9877 PLT. The default is @option{-mgotplt}.
9881 Legacy no-op option only recognized with the cris-axis-aout target.
9885 Legacy no-op option only recognized with the cris-axis-elf and
9886 cris-axis-linux-gnu targets.
9890 Only recognized with the cris-axis-aout target, where it selects a
9891 GNU/linux-like multilib, include files and instruction set for
9896 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9900 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
9901 to link with input-output functions from a simulator library. Code,
9902 initialized data and zero-initialized data are allocated consecutively.
9906 Like @option{-sim}, but pass linker options to locate initialized data at
9907 0x40000000 and zero-initialized data at 0x80000000.
9911 @subsection MMIX Options
9912 @cindex MMIX Options
9914 These options are defined for the MMIX:
9918 @itemx -mno-libfuncs
9920 @opindex mno-libfuncs
9921 Specify that intrinsic library functions are being compiled, passing all
9922 values in registers, no matter the size.
9927 @opindex mno-epsilon
9928 Generate floating-point comparison instructions that compare with respect
9929 to the @code{rE} epsilon register.
9931 @item -mabi=mmixware
9933 @opindex mabi-mmixware
9935 Generate code that passes function parameters and return values that (in
9936 the called function) are seen as registers @code{$0} and up, as opposed to
9937 the GNU ABI which uses global registers @code{$231} and up.
9940 @itemx -mno-zero-extend
9941 @opindex mzero-extend
9942 @opindex mno-zero-extend
9943 When reading data from memory in sizes shorter than 64 bits, use (do not
9944 use) zero-extending load instructions by default, rather than
9945 sign-extending ones.
9948 @itemx -mno-knuthdiv
9950 @opindex mno-knuthdiv
9951 Make the result of a division yielding a remainder have the same sign as
9952 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9953 remainder follows the sign of the dividend. Both methods are
9954 arithmetically valid, the latter being almost exclusively used.
9956 @item -mtoplevel-symbols
9957 @itemx -mno-toplevel-symbols
9958 @opindex mtoplevel-symbols
9959 @opindex mno-toplevel-symbols
9960 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9961 code can be used with the @code{PREFIX} assembly directive.
9965 Generate an executable in the ELF format, rather than the default
9966 @samp{mmo} format used by the @command{mmix} simulator.
9968 @item -mbranch-predict
9969 @itemx -mno-branch-predict
9970 @opindex mbranch-predict
9971 @opindex mno-branch-predict
9972 Use (do not use) the probable-branch instructions, when static branch
9973 prediction indicates a probable branch.
9975 @item -mbase-addresses
9976 @itemx -mno-base-addresses
9977 @opindex mbase-addresses
9978 @opindex mno-base-addresses
9979 Generate (do not generate) code that uses @emph{base addresses}. Using a
9980 base address automatically generates a request (handled by the assembler
9981 and the linker) for a constant to be set up in a global register. The
9982 register is used for one or more base address requests within the range 0
9983 to 255 from the value held in the register. The generally leads to short
9984 and fast code, but the number of different data items that can be
9985 addressed is limited. This means that a program that uses lots of static
9986 data may require @option{-mno-base-addresses}.
9989 @itemx -mno-single-exit
9990 @opindex msingle-exit
9991 @opindex mno-single-exit
9992 Force (do not force) generated code to have a single exit point in each
9996 @node PDP-11 Options
9997 @subsection PDP-11 Options
9998 @cindex PDP-11 Options
10000 These options are defined for the PDP-11:
10005 Use hardware FPP floating point. This is the default. (FIS floating
10006 point on the PDP-11/40 is not supported.)
10009 @opindex msoft-float
10010 Do not use hardware floating point.
10014 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10018 Return floating-point results in memory. This is the default.
10022 Generate code for a PDP-11/40.
10026 Generate code for a PDP-11/45. This is the default.
10030 Generate code for a PDP-11/10.
10032 @item -mbcopy-builtin
10033 @opindex bcopy-builtin
10034 Use inline @code{movstrhi} patterns for copying memory. This is the
10039 Do not use inline @code{movstrhi} patterns for copying memory.
10045 Use 16-bit @code{int}. This is the default.
10051 Use 32-bit @code{int}.
10054 @itemx -mno-float32
10056 @opindex mno-float32
10057 Use 64-bit @code{float}. This is the default.
10062 @opindex mno-float64
10063 Use 32-bit @code{float}.
10067 Use @code{abshi2} pattern. This is the default.
10071 Do not use @code{abshi2} pattern.
10073 @item -mbranch-expensive
10074 @opindex mbranch-expensive
10075 Pretend that branches are expensive. This is for experimenting with
10076 code generation only.
10078 @item -mbranch-cheap
10079 @opindex mbranch-cheap
10080 Do not pretend that branches are expensive. This is the default.
10084 Generate code for a system with split I&D.
10088 Generate code for a system without split I&D. This is the default.
10092 Use Unix assembler syntax. This is the default when configured for
10093 @samp{pdp11-*-bsd}.
10097 Use DEC assembler syntax. This is the default when configured for any
10098 PDP-11 target other than @samp{pdp11-*-bsd}.
10101 @node Xstormy16 Options
10102 @subsection Xstormy16 Options
10103 @cindex Xstormy16 Options
10105 These options are defined for Xstormy16:
10110 Choose startup files and linker script suitable for the simulator.
10114 @subsection FRV Options
10115 @cindex FRV Options
10121 Only use the first 32 general purpose registers.
10126 Use all 64 general purpose registers.
10131 Use only the first 32 floating point registers.
10136 Use all 64 floating point registers
10139 @opindex mhard-float
10141 Use hardware instructions for floating point operations.
10144 @opindex msoft-float
10146 Use library routines for floating point operations.
10151 Dynamically allocate condition code registers.
10156 Do not try to dynamically allocate condition code registers, only
10157 use @code{icc0} and @code{fcc0}.
10162 Change ABI to use double word insns.
10167 Do not use double word instructions.
10172 Use floating point double instructions.
10175 @opindex mno-double
10177 Do not use floating point double instructions.
10182 Use media instructions.
10187 Do not use media instructions.
10192 Use multiply and add/subtract instructions.
10195 @opindex mno-muladd
10197 Do not use multiply and add/subtract instructions.
10199 @item -mlibrary-pic
10200 @opindex mlibrary-pic
10202 Enable PIC support for building libraries
10207 Use only the first four media accumulator registers.
10212 Use all eight media accumulator registers.
10217 Pack VLIW instructions.
10222 Do not pack VLIW instructions.
10225 @opindex mno-eflags
10227 Do not mark ABI switches in e_flags.
10230 @opindex mcond-move
10232 Enable the use of conditional-move instructions (default).
10234 This switch is mainly for debugging the compiler and will likely be removed
10235 in a future version.
10237 @item -mno-cond-move
10238 @opindex mno-cond-move
10240 Disable the use of conditional-move instructions.
10242 This switch is mainly for debugging the compiler and will likely be removed
10243 in a future version.
10248 Enable the use of conditional set instructions (default).
10250 This switch is mainly for debugging the compiler and will likely be removed
10251 in a future version.
10256 Disable the use of conditional set instructions.
10258 This switch is mainly for debugging the compiler and will likely be removed
10259 in a future version.
10262 @opindex mcond-exec
10264 Enable the use of conditional execution (default).
10266 This switch is mainly for debugging the compiler and will likely be removed
10267 in a future version.
10269 @item -mno-cond-exec
10270 @opindex mno-cond-exec
10272 Disable the use of conditional execution.
10274 This switch is mainly for debugging the compiler and will likely be removed
10275 in a future version.
10277 @item -mvliw-branch
10278 @opindex mvliw-branch
10280 Run a pass to pack branches into VLIW instructions (default).
10282 This switch is mainly for debugging the compiler and will likely be removed
10283 in a future version.
10285 @item -mno-vliw-branch
10286 @opindex mno-vliw-branch
10288 Do not run a pass to pack branches into VLIW instructions.
10290 This switch is mainly for debugging the compiler and will likely be removed
10291 in a future version.
10293 @item -mmulti-cond-exec
10294 @opindex mmulti-cond-exec
10296 Enable optimization of @code{&&} and @code{||} in conditional execution
10299 This switch is mainly for debugging the compiler and will likely be removed
10300 in a future version.
10302 @item -mno-multi-cond-exec
10303 @opindex mno-multi-cond-exec
10305 Disable optimization of @code{&&} and @code{||} in conditional execution.
10307 This switch is mainly for debugging the compiler and will likely be removed
10308 in a future version.
10310 @item -mnested-cond-exec
10311 @opindex mnested-cond-exec
10313 Enable nested conditional execution optimizations (default).
10315 This switch is mainly for debugging the compiler and will likely be removed
10316 in a future version.
10318 @item -mno-nested-cond-exec
10319 @opindex mno-nested-cond-exec
10321 Disable nested conditional execution optimizations.
10323 This switch is mainly for debugging the compiler and will likely be removed
10324 in a future version.
10326 @item -mtomcat-stats
10327 @opindex mtomcat-stats
10329 Cause gas to print out tomcat statistics.
10331 @item -mcpu=@var{cpu}
10334 Select the processor type for which to generate code. Possible values are
10335 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10340 @node Xtensa Options
10341 @subsection Xtensa Options
10342 @cindex Xtensa Options
10344 The Xtensa architecture is designed to support many different
10345 configurations. The compiler's default options can be set to match a
10346 particular Xtensa configuration by copying a configuration file into the
10347 GCC sources when building GCC@. The options below may be used to
10348 override the default options.
10352 @itemx -mlittle-endian
10353 @opindex mbig-endian
10354 @opindex mlittle-endian
10355 Specify big-endian or little-endian byte ordering for the target Xtensa
10359 @itemx -mno-density
10361 @opindex mno-density
10362 Enable or disable use of the optional Xtensa code density instructions.
10368 Enable or disable use of the Xtensa MAC16 option. When enabled, GCC
10369 will generate MAC16 instructions from standard C code, with the
10370 limitation that it will use neither the MR register file nor any
10371 instruction that operates on the MR registers. When this option is
10372 disabled, GCC will translate 16-bit multiply/accumulate operations to a
10373 combination of core instructions and library calls, depending on whether
10374 any other multiplier options are enabled.
10380 Enable or disable use of the 16-bit integer multiplier option. When
10381 enabled, the compiler will generate 16-bit multiply instructions for
10382 multiplications of 16 bits or smaller in standard C code. When this
10383 option is disabled, the compiler will either use 32-bit multiply or
10384 MAC16 instructions if they are available or generate library calls to
10385 perform the multiply operations using shifts and adds.
10391 Enable or disable use of the 32-bit integer multiplier option. When
10392 enabled, the compiler will generate 32-bit multiply instructions for
10393 multiplications of 32 bits or smaller in standard C code. When this
10394 option is disabled, the compiler will generate library calls to perform
10395 the multiply operations using either shifts and adds or 16-bit multiply
10396 instructions if they are available.
10402 Enable or disable use of the optional normalization shift amount
10403 (@code{NSA}) instructions to implement the built-in @code{ffs} function.
10408 @opindex mno-minmax
10409 Enable or disable use of the optional minimum and maximum value
10416 Enable or disable use of the optional sign extend (@code{SEXT})
10420 @itemx -mno-booleans
10422 @opindex mno-booleans
10423 Enable or disable support for the boolean register file used by Xtensa
10424 coprocessors. This is not typically useful by itself but may be
10425 required for other options that make use of the boolean registers (e.g.,
10426 the floating-point option).
10429 @itemx -msoft-float
10430 @opindex mhard-float
10431 @opindex msoft-float
10432 Enable or disable use of the floating-point option. When enabled, GCC
10433 generates floating-point instructions for 32-bit @code{float}
10434 operations. When this option is disabled, GCC generates library calls
10435 to emulate 32-bit floating-point operations using integer instructions.
10436 Regardless of this option, 64-bit @code{double} operations are always
10437 emulated with calls to library functions.
10440 @itemx -mno-fused-madd
10441 @opindex mfused-madd
10442 @opindex mno-fused-madd
10443 Enable or disable use of fused multiply/add and multiply/subtract
10444 instructions in the floating-point option. This has no effect if the
10445 floating-point option is not also enabled. Disabling fused multiply/add
10446 and multiply/subtract instructions forces the compiler to use separate
10447 instructions for the multiply and add/subtract operations. This may be
10448 desirable in some cases where strict IEEE 754-compliant results are
10449 required: the fused multiply add/subtract instructions do not round the
10450 intermediate result, thereby producing results with @emph{more} bits of
10451 precision than specified by the IEEE standard. Disabling fused multiply
10452 add/subtract instructions also ensures that the program output is not
10453 sensitive to the compiler's ability to combine multiply and add/subtract
10456 @item -mserialize-volatile
10457 @itemx -mno-serialize-volatile
10458 @opindex mserialize-volatile
10459 @opindex mno-serialize-volatile
10460 When this option is enabled, GCC inserts @code{MEMW} instructions before
10461 @code{volatile} memory references to guarantee sequential consistency.
10462 The default is @option{-mserialize-volatile}. Use
10463 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
10465 @item -mtext-section-literals
10466 @itemx -mno-text-section-literals
10467 @opindex mtext-section-literals
10468 @opindex mno-text-section-literals
10469 Control the treatment of literal pools. The default is
10470 @option{-mno-text-section-literals}, which places literals in a separate
10471 section in the output file. This allows the literal pool to be placed
10472 in a data RAM/ROM, and it also allows the linker to combine literal
10473 pools from separate object files to remove redundant literals and
10474 improve code size. With @option{-mtext-section-literals}, the literals
10475 are interspersed in the text section in order to keep them as close as
10476 possible to their references. This may be necessary for large assembly
10479 @item -mtarget-align
10480 @itemx -mno-target-align
10481 @opindex mtarget-align
10482 @opindex mno-target-align
10483 When this option is enabled, GCC instructs the assembler to
10484 automatically align instructions to reduce branch penalties at the
10485 expense of some code density. The assembler attempts to widen density
10486 instructions to align branch targets and the instructions following call
10487 instructions. If there are not enough preceding safe density
10488 instructions to align a target, no widening will be performed. The
10489 default is @option{-mtarget-align}. These options do not affect the
10490 treatment of auto-aligned instructions like @code{LOOP}, which the
10491 assembler will always align, either by widening density instructions or
10492 by inserting no-op instructions.
10495 @itemx -mno-longcalls
10496 @opindex mlongcalls
10497 @opindex mno-longcalls
10498 When this option is enabled, GCC instructs the assembler to translate
10499 direct calls to indirect calls unless it can determine that the target
10500 of a direct call is in the range allowed by the call instruction. This
10501 translation typically occurs for calls to functions in other source
10502 files. Specifically, the assembler translates a direct @code{CALL}
10503 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10504 The default is @option{-mno-longcalls}. This option should be used in
10505 programs where the call target can potentially be out of range. This
10506 option is implemented in the assembler, not the compiler, so the
10507 assembly code generated by GCC will still show direct call
10508 instructions---look at the disassembled object code to see the actual
10509 instructions. Note that the assembler will use an indirect call for
10510 every cross-file call, not just those that really will be out of range.
10513 @node Code Gen Options
10514 @section Options for Code Generation Conventions
10515 @cindex code generation conventions
10516 @cindex options, code generation
10517 @cindex run-time options
10519 These machine-independent options control the interface conventions
10520 used in code generation.
10522 Most of them have both positive and negative forms; the negative form
10523 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10524 one of the forms is listed---the one which is not the default. You
10525 can figure out the other form by either removing @samp{no-} or adding
10530 @opindex fexceptions
10531 Enable exception handling. Generates extra code needed to propagate
10532 exceptions. For some targets, this implies GCC will generate frame
10533 unwind information for all functions, which can produce significant data
10534 size overhead, although it does not affect execution. If you do not
10535 specify this option, GCC will enable it by default for languages like
10536 C++ which normally require exception handling, and disable it for
10537 languages like C that do not normally require it. However, you may need
10538 to enable this option when compiling C code that needs to interoperate
10539 properly with exception handlers written in C++. You may also wish to
10540 disable this option if you are compiling older C++ programs that don't
10541 use exception handling.
10543 @item -fnon-call-exceptions
10544 @opindex fnon-call-exceptions
10545 Generate code that allows trapping instructions to throw exceptions.
10546 Note that this requires platform-specific runtime support that does
10547 not exist everywhere. Moreover, it only allows @emph{trapping}
10548 instructions to throw exceptions, i.e.@: memory references or floating
10549 point instructions. It does not allow exceptions to be thrown from
10550 arbitrary signal handlers such as @code{SIGALRM}.
10552 @item -funwind-tables
10553 @opindex funwind-tables
10554 Similar to @option{-fexceptions}, except that it will just generate any needed
10555 static data, but will not affect the generated code in any other way.
10556 You will normally not enable this option; instead, a language processor
10557 that needs this handling would enable it on your behalf.
10559 @item -fasynchronous-unwind-tables
10560 @opindex funwind-tables
10561 Generate unwind table in dwarf2 format, if supported by target machine. The
10562 table is exact at each instruction boundary, so it can be used for stack
10563 unwinding from asynchronous events (such as debugger or garbage collector).
10565 @item -fpcc-struct-return
10566 @opindex fpcc-struct-return
10567 Return ``short'' @code{struct} and @code{union} values in memory like
10568 longer ones, rather than in registers. This convention is less
10569 efficient, but it has the advantage of allowing intercallability between
10570 GCC-compiled files and files compiled with other compilers, particularly
10571 the Portable C Compiler (pcc).
10573 The precise convention for returning structures in memory depends
10574 on the target configuration macros.
10576 Short structures and unions are those whose size and alignment match
10577 that of some integer type.
10579 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
10580 switch is not binary compatible with code compiled with the
10581 @option{-freg-struct-return} switch.
10582 Use it to conform to a non-default application binary interface.
10584 @item -freg-struct-return
10585 @opindex freg-struct-return
10586 Return @code{struct} and @code{union} values in registers when possible.
10587 This is more efficient for small structures than
10588 @option{-fpcc-struct-return}.
10590 If you specify neither @option{-fpcc-struct-return} nor
10591 @option{-freg-struct-return}, GCC defaults to whichever convention is
10592 standard for the target. If there is no standard convention, GCC
10593 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
10594 the principal compiler. In those cases, we can choose the standard, and
10595 we chose the more efficient register return alternative.
10597 @strong{Warning:} code compiled with the @option{-freg-struct-return}
10598 switch is not binary compatible with code compiled with the
10599 @option{-fpcc-struct-return} switch.
10600 Use it to conform to a non-default application binary interface.
10602 @item -fshort-enums
10603 @opindex fshort-enums
10604 Allocate to an @code{enum} type only as many bytes as it needs for the
10605 declared range of possible values. Specifically, the @code{enum} type
10606 will be equivalent to the smallest integer type which has enough room.
10608 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
10609 code that is not binary compatible with code generated without that switch.
10610 Use it to conform to a non-default application binary interface.
10612 @item -fshort-double
10613 @opindex fshort-double
10614 Use the same size for @code{double} as for @code{float}.
10616 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
10617 code that is not binary compatible with code generated without that switch.
10618 Use it to conform to a non-default application binary interface.
10620 @item -fshort-wchar
10621 @opindex fshort-wchar
10622 Override the underlying type for @samp{wchar_t} to be @samp{short
10623 unsigned int} instead of the default for the target. This option is
10624 useful for building programs to run under WINE@.
10626 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
10627 code that is not binary compatible with code generated without that switch.
10628 Use it to conform to a non-default application binary interface.
10630 @item -fshared-data
10631 @opindex fshared-data
10632 Requests that the data and non-@code{const} variables of this
10633 compilation be shared data rather than private data. The distinction
10634 makes sense only on certain operating systems, where shared data is
10635 shared between processes running the same program, while private data
10636 exists in one copy per process.
10639 @opindex fno-common
10640 In C, allocate even uninitialized global variables in the data section of the
10641 object file, rather than generating them as common blocks. This has the
10642 effect that if the same variable is declared (without @code{extern}) in
10643 two different compilations, you will get an error when you link them.
10644 The only reason this might be useful is if you wish to verify that the
10645 program will work on other systems which always work this way.
10649 Ignore the @samp{#ident} directive.
10651 @item -fno-gnu-linker
10652 @opindex fno-gnu-linker
10653 Do not output global initializations (such as C++ constructors and
10654 destructors) in the form used by the GNU linker (on systems where the GNU
10655 linker is the standard method of handling them). Use this option when
10656 you want to use a non-GNU linker, which also requires using the
10657 @command{collect2} program to make sure the system linker includes
10658 constructors and destructors. (@command{collect2} is included in the GCC
10659 distribution.) For systems which @emph{must} use @command{collect2}, the
10660 compiler driver @command{gcc} is configured to do this automatically.
10662 @item -finhibit-size-directive
10663 @opindex finhibit-size-directive
10664 Don't output a @code{.size} assembler directive, or anything else that
10665 would cause trouble if the function is split in the middle, and the
10666 two halves are placed at locations far apart in memory. This option is
10667 used when compiling @file{crtstuff.c}; you should not need to use it
10670 @item -fverbose-asm
10671 @opindex fverbose-asm
10672 Put extra commentary information in the generated assembly code to
10673 make it more readable. This option is generally only of use to those
10674 who actually need to read the generated assembly code (perhaps while
10675 debugging the compiler itself).
10677 @option{-fno-verbose-asm}, the default, causes the
10678 extra information to be omitted and is useful when comparing two assembler
10683 Consider all memory references through pointers to be volatile.
10685 @item -fvolatile-global
10686 @opindex fvolatile-global
10687 Consider all memory references to extern and global data items to
10688 be volatile. GCC does not consider static data items to be volatile
10689 because of this switch.
10691 @item -fvolatile-static
10692 @opindex fvolatile-static
10693 Consider all memory references to static data to be volatile.
10697 @cindex global offset table
10699 Generate position-independent code (PIC) suitable for use in a shared
10700 library, if supported for the target machine. Such code accesses all
10701 constant addresses through a global offset table (GOT)@. The dynamic
10702 loader resolves the GOT entries when the program starts (the dynamic
10703 loader is not part of GCC; it is part of the operating system). If
10704 the GOT size for the linked executable exceeds a machine-specific
10705 maximum size, you get an error message from the linker indicating that
10706 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
10707 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
10708 on the m68k and RS/6000. The 386 has no such limit.)
10710 Position-independent code requires special support, and therefore works
10711 only on certain machines. For the 386, GCC supports PIC for System V
10712 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
10713 position-independent.
10717 If supported for the target machine, emit position-independent code,
10718 suitable for dynamic linking and avoiding any limit on the size of the
10719 global offset table. This option makes a difference on the m68k, m88k,
10722 Position-independent code requires special support, and therefore works
10723 only on certain machines.
10725 @item -ffixed-@var{reg}
10727 Treat the register named @var{reg} as a fixed register; generated code
10728 should never refer to it (except perhaps as a stack pointer, frame
10729 pointer or in some other fixed role).
10731 @var{reg} must be the name of a register. The register names accepted
10732 are machine-specific and are defined in the @code{REGISTER_NAMES}
10733 macro in the machine description macro file.
10735 This flag does not have a negative form, because it specifies a
10738 @item -fcall-used-@var{reg}
10739 @opindex fcall-used
10740 Treat the register named @var{reg} as an allocable register that is
10741 clobbered by function calls. It may be allocated for temporaries or
10742 variables that do not live across a call. Functions compiled this way
10743 will not save and restore the register @var{reg}.
10745 It is an error to used this flag with the frame pointer or stack pointer.
10746 Use of this flag for other registers that have fixed pervasive roles in
10747 the machine's execution model will produce disastrous results.
10749 This flag does not have a negative form, because it specifies a
10752 @item -fcall-saved-@var{reg}
10753 @opindex fcall-saved
10754 Treat the register named @var{reg} as an allocable register saved by
10755 functions. It may be allocated even for temporaries or variables that
10756 live across a call. Functions compiled this way will save and restore
10757 the register @var{reg} if they use it.
10759 It is an error to used this flag with the frame pointer or stack pointer.
10760 Use of this flag for other registers that have fixed pervasive roles in
10761 the machine's execution model will produce disastrous results.
10763 A different sort of disaster will result from the use of this flag for
10764 a register in which function values may be returned.
10766 This flag does not have a negative form, because it specifies a
10769 @item -fpack-struct
10770 @opindex fpack-struct
10771 Pack all structure members together without holes.
10773 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
10774 code that is not binary compatible with code generated without that switch.
10775 Additionally, it makes the code suboptimial.
10776 Use it to conform to a non-default application binary interface.
10778 @item -finstrument-functions
10779 @opindex finstrument-functions
10780 Generate instrumentation calls for entry and exit to functions. Just
10781 after function entry and just before function exit, the following
10782 profiling functions will be called with the address of the current
10783 function and its call site. (On some platforms,
10784 @code{__builtin_return_address} does not work beyond the current
10785 function, so the call site information may not be available to the
10786 profiling functions otherwise.)
10789 void __cyg_profile_func_enter (void *this_fn,
10791 void __cyg_profile_func_exit (void *this_fn,
10795 The first argument is the address of the start of the current function,
10796 which may be looked up exactly in the symbol table.
10798 This instrumentation is also done for functions expanded inline in other
10799 functions. The profiling calls will indicate where, conceptually, the
10800 inline function is entered and exited. This means that addressable
10801 versions of such functions must be available. If all your uses of a
10802 function are expanded inline, this may mean an additional expansion of
10803 code size. If you use @samp{extern inline} in your C code, an
10804 addressable version of such functions must be provided. (This is
10805 normally the case anyways, but if you get lucky and the optimizer always
10806 expands the functions inline, you might have gotten away without
10807 providing static copies.)
10809 A function may be given the attribute @code{no_instrument_function}, in
10810 which case this instrumentation will not be done. This can be used, for
10811 example, for the profiling functions listed above, high-priority
10812 interrupt routines, and any functions from which the profiling functions
10813 cannot safely be called (perhaps signal handlers, if the profiling
10814 routines generate output or allocate memory).
10816 @item -fstack-check
10817 @opindex fstack-check
10818 Generate code to verify that you do not go beyond the boundary of the
10819 stack. You should specify this flag if you are running in an
10820 environment with multiple threads, but only rarely need to specify it in
10821 a single-threaded environment since stack overflow is automatically
10822 detected on nearly all systems if there is only one stack.
10824 Note that this switch does not actually cause checking to be done; the
10825 operating system must do that. The switch causes generation of code
10826 to ensure that the operating system sees the stack being extended.
10828 @item -fstack-limit-register=@var{reg}
10829 @itemx -fstack-limit-symbol=@var{sym}
10830 @itemx -fno-stack-limit
10831 @opindex fstack-limit-register
10832 @opindex fstack-limit-symbol
10833 @opindex fno-stack-limit
10834 Generate code to ensure that the stack does not grow beyond a certain value,
10835 either the value of a register or the address of a symbol. If the stack
10836 would grow beyond the value, a signal is raised. For most targets,
10837 the signal is raised before the stack overruns the boundary, so
10838 it is possible to catch the signal without taking special precautions.
10840 For instance, if the stack starts at absolute address @samp{0x80000000}
10841 and grows downwards, you can use the flags
10842 @option{-fstack-limit-symbol=__stack_limit} and
10843 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
10844 of 128KB@. Note that this may only work with the GNU linker.
10846 @cindex aliasing of parameters
10847 @cindex parameters, aliased
10848 @item -fargument-alias
10849 @itemx -fargument-noalias
10850 @itemx -fargument-noalias-global
10851 @opindex fargument-alias
10852 @opindex fargument-noalias
10853 @opindex fargument-noalias-global
10854 Specify the possible relationships among parameters and between
10855 parameters and global data.
10857 @option{-fargument-alias} specifies that arguments (parameters) may
10858 alias each other and may alias global storage.@*
10859 @option{-fargument-noalias} specifies that arguments do not alias
10860 each other, but may alias global storage.@*
10861 @option{-fargument-noalias-global} specifies that arguments do not
10862 alias each other and do not alias global storage.
10864 Each language will automatically use whatever option is required by
10865 the language standard. You should not need to use these options yourself.
10867 @item -fleading-underscore
10868 @opindex fleading-underscore
10869 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
10870 change the way C symbols are represented in the object file. One use
10871 is to help link with legacy assembly code.
10873 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
10874 generate code that is not binary compatible with code generated without that
10875 switch. Use it to conform to a non-default application binary interface.
10876 Not all targets provide complete support for this switch.
10878 @item -ftls-model=@var{model}
10879 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
10880 The @var{model} argument should be one of @code{global-dynamic},
10881 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
10883 The default without @option{-fpic} is @code{initial-exec}; with
10884 @option{-fpic} the default is @code{global-dynamic}.
10889 @node Environment Variables
10890 @section Environment Variables Affecting GCC
10891 @cindex environment variables
10893 @c man begin ENVIRONMENT
10895 This section describes several environment variables that affect how GCC
10896 operates. Some of them work by specifying directories or prefixes to use
10897 when searching for various kinds of files. Some are used to specify other
10898 aspects of the compilation environment.
10900 Note that you can also specify places to search using options such as
10901 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
10902 take precedence over places specified using environment variables, which
10903 in turn take precedence over those specified by the configuration of GCC@.
10904 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
10905 GNU Compiler Collection (GCC) Internals}.
10910 @c @itemx LC_COLLATE
10912 @c @itemx LC_MONETARY
10913 @c @itemx LC_NUMERIC
10918 @c @findex LC_COLLATE
10919 @findex LC_MESSAGES
10920 @c @findex LC_MONETARY
10921 @c @findex LC_NUMERIC
10925 These environment variables control the way that GCC uses
10926 localization information that allow GCC to work with different
10927 national conventions. GCC inspects the locale categories
10928 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
10929 so. These locale categories can be set to any value supported by your
10930 installation. A typical value is @samp{en_UK} for English in the United
10933 The @env{LC_CTYPE} environment variable specifies character
10934 classification. GCC uses it to determine the character boundaries in
10935 a string; this is needed for some multibyte encodings that contain quote
10936 and escape characters that would otherwise be interpreted as a string
10939 The @env{LC_MESSAGES} environment variable specifies the language to
10940 use in diagnostic messages.
10942 If the @env{LC_ALL} environment variable is set, it overrides the value
10943 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
10944 and @env{LC_MESSAGES} default to the value of the @env{LANG}
10945 environment variable. If none of these variables are set, GCC
10946 defaults to traditional C English behavior.
10950 If @env{TMPDIR} is set, it specifies the directory to use for temporary
10951 files. GCC uses temporary files to hold the output of one stage of
10952 compilation which is to be used as input to the next stage: for example,
10953 the output of the preprocessor, which is the input to the compiler
10956 @item GCC_EXEC_PREFIX
10957 @findex GCC_EXEC_PREFIX
10958 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
10959 names of the subprograms executed by the compiler. No slash is added
10960 when this prefix is combined with the name of a subprogram, but you can
10961 specify a prefix that ends with a slash if you wish.
10963 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
10964 an appropriate prefix to use based on the pathname it was invoked with.
10966 If GCC cannot find the subprogram using the specified prefix, it
10967 tries looking in the usual places for the subprogram.
10969 The default value of @env{GCC_EXEC_PREFIX} is
10970 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
10971 of @code{prefix} when you ran the @file{configure} script.
10973 Other prefixes specified with @option{-B} take precedence over this prefix.
10975 This prefix is also used for finding files such as @file{crt0.o} that are
10978 In addition, the prefix is used in an unusual way in finding the
10979 directories to search for header files. For each of the standard
10980 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
10981 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
10982 replacing that beginning with the specified prefix to produce an
10983 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
10984 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
10985 These alternate directories are searched first; the standard directories
10988 @item COMPILER_PATH
10989 @findex COMPILER_PATH
10990 The value of @env{COMPILER_PATH} is a colon-separated list of
10991 directories, much like @env{PATH}. GCC tries the directories thus
10992 specified when searching for subprograms, if it can't find the
10993 subprograms using @env{GCC_EXEC_PREFIX}.
10996 @findex LIBRARY_PATH
10997 The value of @env{LIBRARY_PATH} is a colon-separated list of
10998 directories, much like @env{PATH}. When configured as a native compiler,
10999 GCC tries the directories thus specified when searching for special
11000 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11001 using GCC also uses these directories when searching for ordinary
11002 libraries for the @option{-l} option (but directories specified with
11003 @option{-L} come first).
11007 @cindex locale definition
11008 This variable is used to pass locale information to the compiler. One way in
11009 which this information is used is to determine the character set to be used
11010 when character literals, string literals and comments are parsed in C and C++.
11011 When the compiler is configured to allow multibyte characters,
11012 the following values for @env{LANG} are recognized:
11016 Recognize JIS characters.
11018 Recognize SJIS characters.
11020 Recognize EUCJP characters.
11023 If @env{LANG} is not defined, or if it has some other value, then the
11024 compiler will use mblen and mbtowc as defined by the default locale to
11025 recognize and translate multibyte characters.
11029 Some additional environments variables affect the behavior of the
11032 @include cppenv.texi
11036 @node Running Protoize
11037 @section Running Protoize
11039 The program @code{protoize} is an optional part of GCC@. You can use
11040 it to add prototypes to a program, thus converting the program to ISO
11041 C in one respect. The companion program @code{unprotoize} does the
11042 reverse: it removes argument types from any prototypes that are found.
11044 When you run these programs, you must specify a set of source files as
11045 command line arguments. The conversion programs start out by compiling
11046 these files to see what functions they define. The information gathered
11047 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11049 After scanning comes actual conversion. The specified files are all
11050 eligible to be converted; any files they include (whether sources or
11051 just headers) are eligible as well.
11053 But not all the eligible files are converted. By default,
11054 @code{protoize} and @code{unprotoize} convert only source and header
11055 files in the current directory. You can specify additional directories
11056 whose files should be converted with the @option{-d @var{directory}}
11057 option. You can also specify particular files to exclude with the
11058 @option{-x @var{file}} option. A file is converted if it is eligible, its
11059 directory name matches one of the specified directory names, and its
11060 name within the directory has not been excluded.
11062 Basic conversion with @code{protoize} consists of rewriting most
11063 function definitions and function declarations to specify the types of
11064 the arguments. The only ones not rewritten are those for varargs
11067 @code{protoize} optionally inserts prototype declarations at the
11068 beginning of the source file, to make them available for any calls that
11069 precede the function's definition. Or it can insert prototype
11070 declarations with block scope in the blocks where undeclared functions
11073 Basic conversion with @code{unprotoize} consists of rewriting most
11074 function declarations to remove any argument types, and rewriting
11075 function definitions to the old-style pre-ISO form.
11077 Both conversion programs print a warning for any function declaration or
11078 definition that they can't convert. You can suppress these warnings
11081 The output from @code{protoize} or @code{unprotoize} replaces the
11082 original source file. The original file is renamed to a name ending
11083 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11084 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11085 for DOS) file already exists, then the source file is simply discarded.
11087 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11088 scan the program and collect information about the functions it uses.
11089 So neither of these programs will work until GCC is installed.
11091 Here is a table of the options you can use with @code{protoize} and
11092 @code{unprotoize}. Each option works with both programs unless
11096 @item -B @var{directory}
11097 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11098 usual directory (normally @file{/usr/local/lib}). This file contains
11099 prototype information about standard system functions. This option
11100 applies only to @code{protoize}.
11102 @item -c @var{compilation-options}
11103 Use @var{compilation-options} as the options when running @code{gcc} to
11104 produce the @samp{.X} files. The special option @option{-aux-info} is
11105 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
11107 Note that the compilation options must be given as a single argument to
11108 @code{protoize} or @code{unprotoize}. If you want to specify several
11109 @code{gcc} options, you must quote the entire set of compilation options
11110 to make them a single word in the shell.
11112 There are certain @code{gcc} arguments that you cannot use, because they
11113 would produce the wrong kind of output. These include @option{-g},
11114 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11115 the @var{compilation-options}, they are ignored.
11118 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11119 systems) instead of @samp{.c}. This is convenient if you are converting
11120 a C program to C++. This option applies only to @code{protoize}.
11123 Add explicit global declarations. This means inserting explicit
11124 declarations at the beginning of each source file for each function
11125 that is called in the file and was not declared. These declarations
11126 precede the first function definition that contains a call to an
11127 undeclared function. This option applies only to @code{protoize}.
11129 @item -i @var{string}
11130 Indent old-style parameter declarations with the string @var{string}.
11131 This option applies only to @code{protoize}.
11133 @code{unprotoize} converts prototyped function definitions to old-style
11134 function definitions, where the arguments are declared between the
11135 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11136 uses five spaces as the indentation. If you want to indent with just
11137 one space instead, use @option{-i " "}.
11140 Keep the @samp{.X} files. Normally, they are deleted after conversion
11144 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11145 a prototype declaration for each function in each block which calls the
11146 function without any declaration. This option applies only to
11150 Make no real changes. This mode just prints information about the conversions
11151 that would have been done without @option{-n}.
11154 Make no @samp{.save} files. The original files are simply deleted.
11155 Use this option with caution.
11157 @item -p @var{program}
11158 Use the program @var{program} as the compiler. Normally, the name
11159 @file{gcc} is used.
11162 Work quietly. Most warnings are suppressed.
11165 Print the version number, just like @option{-v} for @code{gcc}.
11168 If you need special compiler options to compile one of your program's
11169 source files, then you should generate that file's @samp{.X} file
11170 specially, by running @code{gcc} on that source file with the
11171 appropriate options and the option @option{-aux-info}. Then run
11172 @code{protoize} on the entire set of files. @code{protoize} will use
11173 the existing @samp{.X} file because it is newer than the source file.
11177 gcc -Dfoo=bar file1.c -aux-info file1.X
11182 You need to include the special files along with the rest in the
11183 @code{protoize} command, even though their @samp{.X} files already
11184 exist, because otherwise they won't get converted.
11186 @xref{Protoize Caveats}, for more information on how to use
11187 @code{protoize} successfully.