1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003 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, 2003 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 1.2 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
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C Dialect Options:: Variations on Objective-C.
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
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}.
158 @gccoptlist{-c -S -E -o @var{file} -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char @gol
172 @item C++ Language Options
173 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
174 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
175 -fconserve-space -fno-const-strings @gol
176 -fno-elide-constructors @gol
177 -fno-enforce-eh-specs -fexternal-templates @gol
178 -falt-external-templates @gol
179 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
180 -fno-implicit-templates @gol
181 -fno-implicit-inline-templates @gol
182 -fno-implement-inlines -fms-extensions @gol
183 -fno-nonansi-builtins -fno-operator-names @gol
184 -fno-optional-diags -fpermissive @gol
185 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
186 -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
187 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
188 -Wnon-virtual-dtor -Wreorder @gol
189 -Weffc++ -Wno-deprecated @gol
190 -Wno-non-template-friend -Wold-style-cast @gol
191 -Woverloaded-virtual -Wno-pmf-conversions @gol
192 -Wsign-promo -Wsynth}
194 @item Objective-C Language Options
195 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
196 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
197 -fgnu-runtime -fnext-runtime -gen-decls @gol
198 -Wno-protocol -Wselector -Wundeclared-selector}
200 @item Language Independent Options
201 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
202 @gccoptlist{-fmessage-length=@var{n} @gol
203 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
205 @item Warning Options
206 @xref{Warning Options,,Options to Request or Suppress Warnings}.
207 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
208 -w -Wextra -Wall -Waggregate-return @gol
209 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
210 -Wconversion -Wno-deprecated-declarations @gol
211 -Wdisabled-optimization -Wno-div-by-zero -Werror @gol
212 -Wfloat-equal -Wformat -Wformat=2 @gol
213 -Wformat-nonliteral -Wformat-security @gol
214 -Wimplicit -Wimplicit-int @gol
215 -Wimplicit-function-declaration @gol
216 -Werror-implicit-function-declaration @gol
217 -Winit-self Wimport -Winline -Winvalid-pch -Wno-endif-labels @gol
218 -Wno-invalid-offsetof @gol
219 -Wlarger-than-@var{len} -Wlong-long @gol
220 -Wmain -Wmissing-braces @gol
221 -Wmissing-format-attribute -Wmissing-noreturn @gol
222 -Wno-multichar -Wno-format-extra-args -Wno-format-y2k @gol
223 -Wno-import -Wnonnull -Wpacked -Wpadded @gol
224 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
225 -Wreturn-type -Wsequence-point -Wshadow @gol
226 -Wsign-compare -Wstrict-aliasing @gol
227 -Wswitch -Wswitch-default -Wswitch-enum @gol
228 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
229 -Wunknown-pragmas -Wunreachable-code @gol
230 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
231 -Wunused-value -Wunused-variable -Wwrite-strings}
233 @item C-only Warning Options
234 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
235 -Wmissing-prototypes -Wnested-externs @gol
236 -Wstrict-prototypes -Wtraditional @gol
237 -Wdeclaration-after-statement}
239 @item Debugging Options
240 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
241 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
242 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
243 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
244 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
245 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
246 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
247 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
248 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
249 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
250 -ftest-coverage -ftime-report @gol
251 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
252 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
253 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
254 -print-multi-directory -print-multi-lib @gol
255 -print-prog-name=@var{program} -print-search-dirs -Q @gol
258 @item Optimization Options
259 @xref{Optimize Options,,Options that Control Optimization}.
260 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
261 -falign-labels=@var{n} -falign-loops=@var{n} @gol
262 -fbranch-probabilities -fprofile-values -fbranch-target-load-optimize @gol
263 -fbranch-target-load-optimize2 -fcaller-saves -fcprop-registers @gol
264 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
265 -fdelayed-branch -fdelete-null-pointer-checks @gol
266 -fexpensive-optimizations -ffast-math -ffloat-store @gol
267 -fforce-addr -fforce-mem -ffunction-sections @gol
268 -fgcse -fgcse-lm -fgcse-sm -floop-optimize -fcrossjumping @gol
269 -fif-conversion -fif-conversion2 @gol
270 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
271 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
272 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
273 -fno-default-inline -fno-defer-pop @gol
274 -fno-function-cse -fno-guess-branch-probability @gol
275 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
276 -funsafe-math-optimizations -ffinite-math-only @gol
277 -fno-trapping-math -fno-zero-initialized-in-bss @gol
278 -fomit-frame-pointer -foptimize-register-move @gol
279 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
280 -freduce-all-givs -fregmove -frename-registers @gol
281 -freorder-blocks -freorder-functions @gol
282 -frerun-cse-after-loop -frerun-loop-opt @gol
283 -frounding-math -fschedule-insns -fschedule-insns2 @gol
284 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
285 -fsched-spec-load-dangerous -fsched2-use-superblocks @gol
286 -fsched2-use-traces -fsignaling-nans @gol
287 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
288 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
289 -funroll-all-loops -funroll-loops -fpeel-loops @gol
290 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
291 --param @var{name}=@var{value}
292 -O -O0 -O1 -O2 -O3 -Os}
294 @item Preprocessor Options
295 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
296 @gccoptlist{-A@var{question}=@var{answer} @gol
297 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
298 -C -dD -dI -dM -dN @gol
299 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
300 -idirafter @var{dir} @gol
301 -include @var{file} -imacros @var{file} @gol
302 -iprefix @var{file} -iwithprefix @var{dir} @gol
303 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
304 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
305 -P -fworking-directory -remap @gol
306 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
307 -Xpreprocessor @var{option}}
309 @item Assembler Option
310 @xref{Assembler Options,,Passing Options to the Assembler}.
311 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
314 @xref{Link Options,,Options for Linking}.
315 @gccoptlist{@var{object-file-name} -l@var{library} @gol
316 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
317 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
318 -Wl,@var{option} -Xlinker @var{option} @gol
321 @item Directory Options
322 @xref{Directory Options,,Options for Directory Search}.
323 @gccoptlist{-B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
326 @c I wrote this xref this way to avoid overfull hbox. -- rms
327 @xref{Target Options}.
328 @gccoptlist{-V @var{version} -b @var{machine}}
330 @item Machine Dependent Options
331 @xref{Submodel Options,,Hardware Models and Configurations}.
333 @emph{M680x0 Options}
334 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
335 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
336 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
337 -malign-int -mstrict-align}
339 @emph{M68hc1x Options}
340 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
341 -mauto-incdec -minmax -mlong-calls -mshort @gol
342 -msoft-reg-count=@var{count}}
345 @gccoptlist{-mg -mgnu -munix}
348 @gccoptlist{-mcpu=@var{cpu-type} @gol
349 -mtune=@var{cpu-type} @gol
350 -mcmodel=@var{code-model} @gol
352 -mapp-regs -mbroken-saverestore -mcypress @gol
353 -mfaster-structs -mflat @gol
354 -mfpu -mhard-float -mhard-quad-float @gol
355 -mimpure-text -mlittle-endian -mlive-g0 -mno-app-regs @gol
356 -mno-faster-structs -mno-flat -mno-fpu @gol
357 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
358 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
359 -msupersparc -munaligned-doubles -mv8}
362 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
363 -mapcs-26 -mapcs-32 @gol
364 -mapcs-stack-check -mno-apcs-stack-check @gol
365 -mapcs-float -mno-apcs-float @gol
366 -mapcs-reentrant -mno-apcs-reentrant @gol
367 -msched-prolog -mno-sched-prolog @gol
368 -mlittle-endian -mbig-endian -mwords-little-endian @gol
369 -malignment-traps -mno-alignment-traps @gol
370 -msoft-float -mhard-float -mfpe @gol
371 -mthumb-interwork -mno-thumb-interwork @gol
372 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
373 -mstructure-size-boundary=@var{n} @gol
374 -mabort-on-noreturn @gol
375 -mlong-calls -mno-long-calls @gol
376 -msingle-pic-base -mno-single-pic-base @gol
377 -mpic-register=@var{reg} @gol
378 -mnop-fun-dllimport @gol
379 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
380 -mpoke-function-name @gol
382 -mtpcs-frame -mtpcs-leaf-frame @gol
383 -mcaller-super-interworking -mcallee-super-interworking}
385 @emph{MN10200 Options}
388 @emph{MN10300 Options}
389 @gccoptlist{-mmult-bug -mno-mult-bug @gol
390 -mam33 -mno-am33 @gol
391 -mam33-2 -mno-am33-2 @gol
394 @emph{M32R/D Options}
395 @gccoptlist{-m32rx -m32r -mcode-model=@var{model-type} @gol
396 -msdata=@var{sdata-type} -G @var{num}}
399 @gccoptlist{-m88000 -m88100 -m88110 -mbig-pic @gol
400 -mcheck-zero-division -mhandle-large-shift @gol
401 -midentify-revision -mno-check-zero-division @gol
402 -mno-ocs-debug-info -mno-ocs-frame-position @gol
403 -mno-optimize-arg-area -mno-serialize-volatile @gol
404 -mno-underscores -mocs-debug-info @gol
405 -mocs-frame-position -moptimize-arg-area @gol
406 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
407 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
408 -mversion-03.00 -mwarn-passed-structs}
410 @emph{RS/6000 and PowerPC Options}
411 @gccoptlist{-mcpu=@var{cpu-type} @gol
412 -mtune=@var{cpu-type} @gol
413 -mpower -mno-power -mpower2 -mno-power2 @gol
414 -mpowerpc -mpowerpc64 -mno-powerpc @gol
415 -maltivec -mno-altivec @gol
416 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
417 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
418 -mnew-mnemonics -mold-mnemonics @gol
419 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
420 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
421 -malign-power -malign-natural @gol
422 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
423 -mstring -mno-string -mupdate -mno-update @gol
424 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
425 -mstrict-align -mno-strict-align -mrelocatable @gol
426 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
427 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
428 -mdynamic-no-pic @gol
429 -mcall-sysv -mcall-netbsd @gol
430 -maix-struct-return -msvr4-struct-return @gol
431 -mabi=altivec -mabi=no-altivec @gol
432 -mabi=spe -mabi=no-spe @gol
433 -misel=yes -misel=no @gol
434 -mspe=yes -mspe=no @gol
435 -mfloat-gprs=yes -mfloat-gprs=no @gol
436 -mprototype -mno-prototype @gol
437 -msim -mmvme -mads -myellowknife -memb -msdata @gol
438 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
440 @emph{Darwin Options}
441 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
442 -arch_only -bind_at_load -bundle -bundle_loader @gol
443 -client_name -compatibility_version -current_version @gol
444 -dependency-file -dylib_file -dylinker_install_name @gol
445 -dynamic -dynamiclib -exported_symbols_list @gol
446 -filelist -flat_namespace -force_cpusubtype_ALL @gol
447 -force_flat_namespace -headerpad_max_install_names @gol
448 -image_base -init -install_name -keep_private_externs @gol
449 -multi_module -multiply_defined -multiply_defined_unused @gol
450 -noall_load -nomultidefs -noprebind -noseglinkedit @gol
451 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
452 -private_bundle -read_only_relocs -sectalign @gol
453 -sectobjectsymbols -whyload -seg1addr @gol
454 -sectcreate -sectobjectsymbols -sectorder @gol
455 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
456 -segprot -segs_read_only_addr -segs_read_write_addr @gol
457 -single_module -static -sub_library -sub_umbrella @gol
458 -twolevel_namespace -umbrella -undefined @gol
459 -unexported_symbols_list -weak_reference_mismatches @gol
463 @gccoptlist{-mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
464 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
465 -mminimum-fp-blocks -mnohc-struct-return}
468 @gccoptlist{-mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
469 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
470 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
471 -mgas -mgp32 -mgp64 -mhard-float -mint64 -mips1 @gol
472 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
473 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
474 -mmips-as -mmips-tfile -mno-abicalls -mxgot @gol
475 -mno-embedded-data -mno-uninit-const-in-rodata @gol
476 -mno-embedded-pic -mno-long-calls @gol
477 -mno-memcpy -mno-mips-tfile -mno-rnames @gol
478 -mrnames -msoft-float @gol
479 -m4650 -msingle-float -mmad @gol
480 -EL -EB -G @var{num} -nocpp @gol
481 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi -mabi-fake-default @gol
482 -mfix7000 -mfix-sb1 -mno-fix-sb1 @gol
483 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
484 -mbranch-likely -mno-branch-likely}
486 @emph{i386 and x86-64 Options}
487 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
488 -mfpmath=@var{unit} @gol
489 -masm=@var{dialect} -mno-fancy-math-387 @gol
490 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
491 -mno-wide-multiply -mrtd -malign-double @gol
492 -mpreferred-stack-boundary=@var{num} @gol
493 -mmmx -msse -msse2 -mpni -m3dnow @gol
494 -mthreads -mno-align-stringops -minline-all-stringops @gol
495 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
496 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
497 -mno-red-zone -mno-tls-direct-seg-refs @gol
498 -mcmodel=@var{code-model} @gol
502 @gccoptlist{-march=@var{architecture-type} @gol
503 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
504 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
505 -mjump-in-delay -mlinker-opt -mlong-calls @gol
506 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
507 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
508 -mno-jump-in-delay -mno-long-load-store @gol
509 -mno-portable-runtime -mno-soft-float @gol
510 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
511 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
512 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
513 -nolibdld -static -threads}
515 @emph{Intel 960 Options}
516 @gccoptlist{-m@var{cpu-type} -masm-compat -mclean-linkage @gol
517 -mcode-align -mcomplex-addr -mleaf-procedures @gol
518 -mic-compat -mic2.0-compat -mic3.0-compat @gol
519 -mintel-asm -mno-clean-linkage -mno-code-align @gol
520 -mno-complex-addr -mno-leaf-procedures @gol
521 -mno-old-align -mno-strict-align -mno-tail-call @gol
522 -mnumerics -mold-align -msoft-float -mstrict-align @gol
525 @emph{DEC Alpha Options}
526 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
527 -mieee -mieee-with-inexact -mieee-conformant @gol
528 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
529 -mtrap-precision=@var{mode} -mbuild-constants @gol
530 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
531 -mbwx -mmax -mfix -mcix @gol
532 -mfloat-vax -mfloat-ieee @gol
533 -mexplicit-relocs -msmall-data -mlarge-data @gol
534 -msmall-text -mlarge-text @gol
535 -mmemory-latency=@var{time}}
537 @emph{DEC Alpha/VMS Options}
538 @gccoptlist{-mvms-return-codes}
540 @emph{H8/300 Options}
541 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
544 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
545 -m4-nofpu -m4-single-only -m4-single -m4 @gol
546 -m5-64media -m5-64media-nofpu @gol
547 -m5-32media -m5-32media-nofpu @gol
548 -m5-compact -m5-compact-nofpu @gol
549 -mb -ml -mdalign -mrelax @gol
550 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
551 -mieee -misize -mpadstruct -mspace @gol
552 -mprefergot -musermode}
554 @emph{System V Options}
555 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
558 @gccoptlist{-EB -EL @gol
559 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
560 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
562 @emph{TMS320C3x/C4x Options}
563 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
564 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
565 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
566 -mparallel-insns -mparallel-mpy -mpreserve-float}
569 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
570 -mprolog-function -mno-prolog-function -mspace @gol
571 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
572 -mapp-regs -mno-app-regs @gol
573 -mdisable-callt -mno-disable-callt @gol
579 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
580 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
581 -mregparam -mnoregparam -msb -mnosb @gol
582 -mbitfield -mnobitfield -mhimem -mnohimem}
585 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
586 -mcall-prologues -mno-tablejump -mtiny-stack}
589 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
590 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
591 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
592 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
593 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
596 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
597 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
598 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
599 -mno-base-addresses -msingle-exit -mno-single-exit}
602 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
603 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
604 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
605 -minline-float-divide-max-throughput @gol
606 -minline-int-divide-min-latency @gol
607 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
608 -mfixed-range=@var{register-range}}
611 @gccoptlist{-mextmem -mextmemory -monchip -mno-asm-optimize @gol
612 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
614 @emph{S/390 and zSeries Options}
615 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
616 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
617 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
618 -m64 -m31 -mdebug -mno-debug -mesa -mzarch}
621 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
622 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
623 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
624 -mstack-align -mdata-align -mconst-align @gol
625 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
626 -melf -maout -melinux -mlinux -sim -sim2}
628 @emph{PDP-11 Options}
629 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
630 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
631 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
632 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
633 -mbranch-expensive -mbranch-cheap @gol
634 -msplit -mno-split -munix-asm -mdec-asm}
636 @emph{Xstormy16 Options}
639 @emph{Xtensa Options}
640 @gccoptlist{-mbig-endian -mlittle-endian @gol
641 -mdensity -mno-density @gol
642 -mconst16 -mno-const16 @gol
644 -maddx -mno-addx @gol
645 -mmac16 -mno-mac16 @gol
646 -mmul16 -mno-mul16 @gol
647 -mmul32 -mno-mul32 @gol
649 -mminmax -mno-minmax @gol
650 -msext -mno-sext @gol
651 -mbooleans -mno-booleans @gol
652 -mhard-float -msoft-float @gol
653 -mfused-madd -mno-fused-madd @gol
654 -mtext-section-literals -mno-text-section-literals @gol
655 -mtarget-align -mno-target-align @gol
656 -mlongcalls -mno-longcalls}
659 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
660 -mhard-float -msoft-float @gol
661 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
662 -mdouble -mno-double @gol
663 -mmedia -mno-media -mmuladd -mno-muladd @gol
664 -mlibrary-pic -macc-4 -macc-8 @gol
665 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
666 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
667 -mvliw-branch -mno-vliw-branch @gol
668 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
669 -mno-nested-cond-exec -mtomcat-stats @gol
672 @item Code Generation Options
673 @xref{Code Gen Options,,Options for Code Generation Conventions}.
674 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
675 -ffixed-@var{reg} -fexceptions @gol
676 -fnon-call-exceptions -funwind-tables @gol
677 -fasynchronous-unwind-tables @gol
678 -finhibit-size-directive -finstrument-functions @gol
679 -fno-common -fno-ident -fno-gnu-linker @gol
680 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
681 -freg-struct-return -fshared-data -fshort-enums @gol
682 -fshort-double -fshort-wchar @gol
683 -fverbose-asm -fpack-struct -fstack-check @gol
684 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
685 -fargument-alias -fargument-noalias @gol
686 -fargument-noalias-global -fleading-underscore @gol
687 -ftls-model=@var{model} @gol
688 -ftrapv -fwrapv -fbounds-check}
692 * Overall Options:: Controlling the kind of output:
693 an executable, object files, assembler files,
694 or preprocessed source.
695 * C Dialect Options:: Controlling the variant of C language compiled.
696 * C++ Dialect Options:: Variations on C++.
697 * Objective-C Dialect Options:: Variations on Objective-C.
698 * Language Independent Options:: Controlling how diagnostics should be
700 * Warning Options:: How picky should the compiler be?
701 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
702 * Optimize Options:: How much optimization?
703 * Preprocessor Options:: Controlling header files and macro definitions.
704 Also, getting dependency information for Make.
705 * Assembler Options:: Passing options to the assembler.
706 * Link Options:: Specifying libraries and so on.
707 * Directory Options:: Where to find header files and libraries.
708 Where to find the compiler executable files.
709 * Spec Files:: How to pass switches to sub-processes.
710 * Target Options:: Running a cross-compiler, or an old version of GCC.
713 @node Overall Options
714 @section Options Controlling the Kind of Output
716 Compilation can involve up to four stages: preprocessing, compilation
717 proper, assembly and linking, always in that order. GCC is capable of
718 preprocessing and compiling several files either into several
719 assembler input files, or into one assembler input file; then each
720 assembler input file produces an object file, and linking combines all
721 the object files (those newly compiled, and those specified as input)
722 into an executable file.
724 @cindex file name suffix
725 For any given input file, the file name suffix determines what kind of
730 C source code which must be preprocessed.
733 C source code which should not be preprocessed.
736 C++ source code which should not be preprocessed.
739 Objective-C source code. Note that you must link with the library
740 @file{libobjc.a} to make an Objective-C program work.
743 Objective-C source code which should not be preprocessed.
746 C or C++ header file to be turned into a precompiled header.
750 @itemx @var{file}.cxx
751 @itemx @var{file}.cpp
752 @itemx @var{file}.CPP
753 @itemx @var{file}.c++
755 C++ source code which must be preprocessed. Note that in @samp{.cxx},
756 the last two letters must both be literally @samp{x}. Likewise,
757 @samp{.C} refers to a literal capital C@.
761 C++ header file to be turned into a precompiled header.
764 @itemx @var{file}.for
765 @itemx @var{file}.FOR
766 Fortran source code which should not be preprocessed.
769 @itemx @var{file}.fpp
770 @itemx @var{file}.FPP
771 Fortran source code which must be preprocessed (with the traditional
775 Fortran source code which must be preprocessed with a RATFOR
776 preprocessor (not included with GCC)@.
778 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
779 Using and Porting GNU Fortran}, for more details of the handling of
782 @c FIXME: Descriptions of Java file types.
789 Ada source code file which contains a library unit declaration (a
790 declaration of a package, subprogram, or generic, or a generic
791 instantiation), or a library unit renaming declaration (a package,
792 generic, or subprogram renaming declaration). Such files are also
795 @itemx @var{file}.adb
796 Ada source code file containing a library unit body (a subprogram or
797 package body). Such files are also called @dfn{bodies}.
799 @c GCC also knows about some suffixes for languages not yet included:
808 Assembler code which must be preprocessed.
811 An object file to be fed straight into linking.
812 Any file name with no recognized suffix is treated this way.
816 You can specify the input language explicitly with the @option{-x} option:
819 @item -x @var{language}
820 Specify explicitly the @var{language} for the following input files
821 (rather than letting the compiler choose a default based on the file
822 name suffix). This option applies to all following input files until
823 the next @option{-x} option. Possible values for @var{language} are:
825 c c-header cpp-output
826 c++ c++-header c++-cpp-output
827 objective-c objective-c-header objc-cpp-output
828 assembler assembler-with-cpp
830 f77 f77-cpp-input ratfor
836 Turn off any specification of a language, so that subsequent files are
837 handled according to their file name suffixes (as they are if @option{-x}
838 has not been used at all).
840 @item -pass-exit-codes
841 @opindex pass-exit-codes
842 Normally the @command{gcc} program will exit with the code of 1 if any
843 phase of the compiler returns a non-success return code. If you specify
844 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
845 numerically highest error produced by any phase that returned an error
849 If you only want some of the stages of compilation, you can use
850 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
851 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
852 @command{gcc} is to stop. Note that some combinations (for example,
853 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
858 Compile or assemble the source files, but do not link. The linking
859 stage simply is not done. The ultimate output is in the form of an
860 object file for each source file.
862 By default, the object file name for a source file is made by replacing
863 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
865 Unrecognized input files, not requiring compilation or assembly, are
870 Stop after the stage of compilation proper; do not assemble. The output
871 is in the form of an assembler code file for each non-assembler input
874 By default, the assembler file name for a source file is made by
875 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
877 Input files that don't require compilation are ignored.
881 Stop after the preprocessing stage; do not run the compiler proper. The
882 output is in the form of preprocessed source code, which is sent to the
885 Input files which don't require preprocessing are ignored.
887 @cindex output file option
890 Place output in file @var{file}. This applies regardless to whatever
891 sort of output is being produced, whether it be an executable file,
892 an object file, an assembler file or preprocessed C code.
894 If you specify @option{-o} when compiling more than one input file, or
895 you are producing an executable file as output, all the source files
896 on the command line will be compiled at once.
898 If @option{-o} is not specified, the default is to put an executable file
899 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
900 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
901 all preprocessed C source on standard output.
905 Print (on standard error output) the commands executed to run the stages
906 of compilation. Also print the version number of the compiler driver
907 program and of the preprocessor and the compiler proper.
911 Like @option{-v} except the commands are not executed and all command
912 arguments are quoted. This is useful for shell scripts to capture the
913 driver-generated command lines.
917 Use pipes rather than temporary files for communication between the
918 various stages of compilation. This fails to work on some systems where
919 the assembler is unable to read from a pipe; but the GNU assembler has
924 Print (on the standard output) a description of the command line options
925 understood by @command{gcc}. If the @option{-v} option is also specified
926 then @option{--help} will also be passed on to the various processes
927 invoked by @command{gcc}, so that they can display the command line options
928 they accept. If the @option{-Wextra} option is also specified then command
929 line options which have no documentation associated with them will also
934 Print (on the standard output) a description of target specific command
935 line options for each tool.
939 Display the version number and copyrights of the invoked GCC.
943 @section Compiling C++ Programs
945 @cindex suffixes for C++ source
946 @cindex C++ source file suffixes
947 C++ source files conventionally use one of the suffixes @samp{.C},
948 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
949 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
950 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
951 files with these names and compiles them as C++ programs even if you
952 call the compiler the same way as for compiling C programs (usually
953 with the name @command{gcc}).
957 However, C++ programs often require class libraries as well as a
958 compiler that understands the C++ language---and under some
959 circumstances, you might want to compile programs or header files from
960 standard input, or otherwise without a suffix that flags them as C++
961 programs. You might also like to precompile a C header file with a
962 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
963 program that calls GCC with the default language set to C++, and
964 automatically specifies linking against the C++ library. On many
965 systems, @command{g++} is also installed with the name @command{c++}.
967 @cindex invoking @command{g++}
968 When you compile C++ programs, you may specify many of the same
969 command-line options that you use for compiling programs in any
970 language; or command-line options meaningful for C and related
971 languages; or options that are meaningful only for C++ programs.
972 @xref{C Dialect Options,,Options Controlling C Dialect}, for
973 explanations of options for languages related to C@.
974 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
975 explanations of options that are meaningful only for C++ programs.
977 @node C Dialect Options
978 @section Options Controlling C Dialect
979 @cindex dialect options
980 @cindex language dialect options
981 @cindex options, dialect
983 The following options control the dialect of C (or languages derived
984 from C, such as C++ and Objective-C) that the compiler accepts:
991 In C mode, support all ISO C90 programs. In C++ mode,
992 remove GNU extensions that conflict with ISO C++.
994 This turns off certain features of GCC that are incompatible with ISO
995 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
996 such as the @code{asm} and @code{typeof} keywords, and
997 predefined macros such as @code{unix} and @code{vax} that identify the
998 type of system you are using. It also enables the undesirable and
999 rarely used ISO trigraph feature. For the C compiler,
1000 it disables recognition of C++ style @samp{//} comments as well as
1001 the @code{inline} keyword.
1003 The alternate keywords @code{__asm__}, @code{__extension__},
1004 @code{__inline__} and @code{__typeof__} continue to work despite
1005 @option{-ansi}. You would not want to use them in an ISO C program, of
1006 course, but it is useful to put them in header files that might be included
1007 in compilations done with @option{-ansi}. Alternate predefined macros
1008 such as @code{__unix__} and @code{__vax__} are also available, with or
1009 without @option{-ansi}.
1011 The @option{-ansi} option does not cause non-ISO programs to be
1012 rejected gratuitously. For that, @option{-pedantic} is required in
1013 addition to @option{-ansi}. @xref{Warning Options}.
1015 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1016 option is used. Some header files may notice this macro and refrain
1017 from declaring certain functions or defining certain macros that the
1018 ISO standard doesn't call for; this is to avoid interfering with any
1019 programs that might use these names for other things.
1021 Functions which would normally be built in but do not have semantics
1022 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1023 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1024 built-in functions provided by GCC}, for details of the functions
1029 Determine the language standard. This option is currently only
1030 supported when compiling C or C++. A value for this option must be
1031 provided; possible values are
1036 ISO C90 (same as @option{-ansi}).
1038 @item iso9899:199409
1039 ISO C90 as modified in amendment 1.
1045 ISO C99. Note that this standard is not yet fully supported; see
1046 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1047 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1050 Default, ISO C90 plus GNU extensions (including some C99 features).
1054 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1055 this will become the default. The name @samp{gnu9x} is deprecated.
1058 The 1998 ISO C++ standard plus amendments.
1061 The same as @option{-std=c++98} plus GNU extensions. This is the
1062 default for C++ code.
1065 Even when this option is not specified, you can still use some of the
1066 features of newer standards in so far as they do not conflict with
1067 previous C standards. For example, you may use @code{__restrict__} even
1068 when @option{-std=c99} is not specified.
1070 The @option{-std} options specifying some version of ISO C have the same
1071 effects as @option{-ansi}, except that features that were not in ISO C90
1072 but are in the specified version (for example, @samp{//} comments and
1073 the @code{inline} keyword in ISO C99) are not disabled.
1075 @xref{Standards,,Language Standards Supported by GCC}, for details of
1076 these standard versions.
1078 @item -aux-info @var{filename}
1080 Output to the given filename prototyped declarations for all functions
1081 declared and/or defined in a translation unit, including those in header
1082 files. This option is silently ignored in any language other than C@.
1084 Besides declarations, the file indicates, in comments, the origin of
1085 each declaration (source file and line), whether the declaration was
1086 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1087 @samp{O} for old, respectively, in the first character after the line
1088 number and the colon), and whether it came from a declaration or a
1089 definition (@samp{C} or @samp{F}, respectively, in the following
1090 character). In the case of function definitions, a K&R-style list of
1091 arguments followed by their declarations is also provided, inside
1092 comments, after the declaration.
1096 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1097 keyword, so that code can use these words as identifiers. You can use
1098 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1099 instead. @option{-ansi} implies @option{-fno-asm}.
1101 In C++, this switch only affects the @code{typeof} keyword, since
1102 @code{asm} and @code{inline} are standard keywords. You may want to
1103 use the @option{-fno-gnu-keywords} flag instead, which has the same
1104 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1105 switch only affects the @code{asm} and @code{typeof} keywords, since
1106 @code{inline} is a standard keyword in ISO C99.
1109 @itemx -fno-builtin-@var{function}
1110 @opindex fno-builtin
1111 @cindex built-in functions
1112 Don't recognize built-in functions that do not begin with
1113 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1114 functions provided by GCC}, for details of the functions affected,
1115 including those which are not built-in functions when @option{-ansi} or
1116 @option{-std} options for strict ISO C conformance are used because they
1117 do not have an ISO standard meaning.
1119 GCC normally generates special code to handle certain built-in functions
1120 more efficiently; for instance, calls to @code{alloca} may become single
1121 instructions that adjust the stack directly, and calls to @code{memcpy}
1122 may become inline copy loops. The resulting code is often both smaller
1123 and faster, but since the function calls no longer appear as such, you
1124 cannot set a breakpoint on those calls, nor can you change the behavior
1125 of the functions by linking with a different library.
1127 With the @option{-fno-builtin-@var{function}} option
1128 only the built-in function @var{function} is
1129 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1130 function is named this is not built-in in this version of GCC, this
1131 option is ignored. There is no corresponding
1132 @option{-fbuiltin-@var{function}} option; if you wish to enable
1133 built-in functions selectively when using @option{-fno-builtin} or
1134 @option{-ffreestanding}, you may define macros such as:
1137 #define abs(n) __builtin_abs ((n))
1138 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1143 @cindex hosted environment
1145 Assert that compilation takes place in a hosted environment. This implies
1146 @option{-fbuiltin}. A hosted environment is one in which the
1147 entire standard library is available, and in which @code{main} has a return
1148 type of @code{int}. Examples are nearly everything except a kernel.
1149 This is equivalent to @option{-fno-freestanding}.
1151 @item -ffreestanding
1152 @opindex ffreestanding
1153 @cindex hosted environment
1155 Assert that compilation takes place in a freestanding environment. This
1156 implies @option{-fno-builtin}. A freestanding environment
1157 is one in which the standard library may not exist, and program startup may
1158 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1159 This is equivalent to @option{-fno-hosted}.
1161 @xref{Standards,,Language Standards Supported by GCC}, for details of
1162 freestanding and hosted environments.
1164 @item -fms-extensions
1165 @opindex fms-extensions
1166 Accept some non-standard constructs used in Microsoft header files.
1170 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1171 options for strict ISO C conformance) implies @option{-trigraphs}.
1173 @item -no-integrated-cpp
1174 @opindex no-integrated-cpp
1175 Performs a compilation in two passes: preprocessing and compiling. This
1176 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1177 @option{-B} option. The user supplied compilation step can then add in
1178 an additional preprocessing step after normal preprocessing but before
1179 compiling. The default is to use the integrated cpp (internal cpp)
1181 The semantics of this option will change if "cc1", "cc1plus", and
1182 "cc1obj" are merged.
1184 @cindex traditional C language
1185 @cindex C language, traditional
1187 @itemx -traditional-cpp
1188 @opindex traditional-cpp
1189 @opindex traditional
1190 Formerly, these options caused GCC to attempt to emulate a pre-standard
1191 C compiler. They are now only supported with the @option{-E} switch.
1192 The preprocessor continues to support a pre-standard mode. See the GNU
1193 CPP manual for details.
1195 @item -fcond-mismatch
1196 @opindex fcond-mismatch
1197 Allow conditional expressions with mismatched types in the second and
1198 third arguments. The value of such an expression is void. This option
1199 is not supported for C++.
1201 @item -funsigned-char
1202 @opindex funsigned-char
1203 Let the type @code{char} be unsigned, like @code{unsigned char}.
1205 Each kind of machine has a default for what @code{char} should
1206 be. It is either like @code{unsigned char} by default or like
1207 @code{signed char} by default.
1209 Ideally, a portable program should always use @code{signed char} or
1210 @code{unsigned char} when it depends on the signedness of an object.
1211 But many programs have been written to use plain @code{char} and
1212 expect it to be signed, or expect it to be unsigned, depending on the
1213 machines they were written for. This option, and its inverse, let you
1214 make such a program work with the opposite default.
1216 The type @code{char} is always a distinct type from each of
1217 @code{signed char} or @code{unsigned char}, even though its behavior
1218 is always just like one of those two.
1221 @opindex fsigned-char
1222 Let the type @code{char} be signed, like @code{signed char}.
1224 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1225 the negative form of @option{-funsigned-char}. Likewise, the option
1226 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1228 @item -fsigned-bitfields
1229 @itemx -funsigned-bitfields
1230 @itemx -fno-signed-bitfields
1231 @itemx -fno-unsigned-bitfields
1232 @opindex fsigned-bitfields
1233 @opindex funsigned-bitfields
1234 @opindex fno-signed-bitfields
1235 @opindex fno-unsigned-bitfields
1236 These options control whether a bit-field is signed or unsigned, when the
1237 declaration does not use either @code{signed} or @code{unsigned}. By
1238 default, such a bit-field is signed, because this is consistent: the
1239 basic integer types such as @code{int} are signed types.
1241 @item -fwritable-strings
1242 @opindex fwritable-strings
1243 Store string constants in the writable data segment and don't uniquize
1244 them. This is for compatibility with old programs which assume they can
1245 write into string constants.
1247 Writing into string constants is a very bad idea; ``constants'' should
1251 @node C++ Dialect Options
1252 @section Options Controlling C++ Dialect
1254 @cindex compiler options, C++
1255 @cindex C++ options, command line
1256 @cindex options, C++
1257 This section describes the command-line options that are only meaningful
1258 for C++ programs; but you can also use most of the GNU compiler options
1259 regardless of what language your program is in. For example, you
1260 might compile a file @code{firstClass.C} like this:
1263 g++ -g -frepo -O -c firstClass.C
1267 In this example, only @option{-frepo} is an option meant
1268 only for C++ programs; you can use the other options with any
1269 language supported by GCC@.
1271 Here is a list of options that are @emph{only} for compiling C++ programs:
1275 @item -fabi-version=@var{n}
1276 @opindex fabi-version
1277 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1278 ABI that first appeared in G++ 3.2. Version 0 will always be the
1279 version that conforms most closely to the C++ ABI specification.
1280 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1283 The default is version 1.
1285 @item -fno-access-control
1286 @opindex fno-access-control
1287 Turn off all access checking. This switch is mainly useful for working
1288 around bugs in the access control code.
1292 Check that the pointer returned by @code{operator new} is non-null
1293 before attempting to modify the storage allocated. This check is
1294 normally unnecessary because the C++ standard specifies that
1295 @code{operator new} will only return @code{0} if it is declared
1296 @samp{throw()}, in which case the compiler will always check the
1297 return value even without this option. In all other cases, when
1298 @code{operator new} has a non-empty exception specification, memory
1299 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1300 @samp{new (nothrow)}.
1302 @item -fconserve-space
1303 @opindex fconserve-space
1304 Put uninitialized or runtime-initialized global variables into the
1305 common segment, as C does. This saves space in the executable at the
1306 cost of not diagnosing duplicate definitions. If you compile with this
1307 flag and your program mysteriously crashes after @code{main()} has
1308 completed, you may have an object that is being destroyed twice because
1309 two definitions were merged.
1311 This option is no longer useful on most targets, now that support has
1312 been added for putting variables into BSS without making them common.
1314 @item -fno-const-strings
1315 @opindex fno-const-strings
1316 Give string constants type @code{char *} instead of type @code{const
1317 char *}. By default, G++ uses type @code{const char *} as required by
1318 the standard. Even if you use @option{-fno-const-strings}, you cannot
1319 actually modify the value of a string constant, unless you also use
1320 @option{-fwritable-strings}.
1322 This option might be removed in a future release of G++. For maximum
1323 portability, you should structure your code so that it works with
1324 string constants that have type @code{const char *}.
1326 @item -fno-elide-constructors
1327 @opindex fno-elide-constructors
1328 The C++ standard allows an implementation to omit creating a temporary
1329 which is only used to initialize another object of the same type.
1330 Specifying this option disables that optimization, and forces G++ to
1331 call the copy constructor in all cases.
1333 @item -fno-enforce-eh-specs
1334 @opindex fno-enforce-eh-specs
1335 Don't check for violation of exception specifications at runtime. This
1336 option violates the C++ standard, but may be useful for reducing code
1337 size in production builds, much like defining @samp{NDEBUG}. The compiler
1338 will still optimize based on the exception specifications.
1340 @item -fexternal-templates
1341 @opindex fexternal-templates
1343 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1344 template instantiation; template instances are emitted or not according
1345 to the location of the template definition. @xref{Template
1346 Instantiation}, for more information.
1348 This option is deprecated.
1350 @item -falt-external-templates
1351 @opindex falt-external-templates
1352 Similar to @option{-fexternal-templates}, but template instances are
1353 emitted or not according to the place where they are first instantiated.
1354 @xref{Template Instantiation}, for more information.
1356 This option is deprecated.
1359 @itemx -fno-for-scope
1361 @opindex fno-for-scope
1362 If @option{-ffor-scope} is specified, the scope of variables declared in
1363 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1364 as specified by the C++ standard.
1365 If @option{-fno-for-scope} is specified, the scope of variables declared in
1366 a @i{for-init-statement} extends to the end of the enclosing scope,
1367 as was the case in old versions of G++, and other (traditional)
1368 implementations of C++.
1370 The default if neither flag is given to follow the standard,
1371 but to allow and give a warning for old-style code that would
1372 otherwise be invalid, or have different behavior.
1374 @item -fno-gnu-keywords
1375 @opindex fno-gnu-keywords
1376 Do not recognize @code{typeof} as a keyword, so that code can use this
1377 word as an identifier. You can use the keyword @code{__typeof__} instead.
1378 @option{-ansi} implies @option{-fno-gnu-keywords}.
1380 @item -fno-implicit-templates
1381 @opindex fno-implicit-templates
1382 Never emit code for non-inline templates which are instantiated
1383 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1384 @xref{Template Instantiation}, for more information.
1386 @item -fno-implicit-inline-templates
1387 @opindex fno-implicit-inline-templates
1388 Don't emit code for implicit instantiations of inline templates, either.
1389 The default is to handle inlines differently so that compiles with and
1390 without optimization will need the same set of explicit instantiations.
1392 @item -fno-implement-inlines
1393 @opindex fno-implement-inlines
1394 To save space, do not emit out-of-line copies of inline functions
1395 controlled by @samp{#pragma implementation}. This will cause linker
1396 errors if these functions are not inlined everywhere they are called.
1398 @item -fms-extensions
1399 @opindex fms-extensions
1400 Disable pedantic warnings about constructs used in MFC, such as implicit
1401 int and getting a pointer to member function via non-standard syntax.
1403 @item -fno-nonansi-builtins
1404 @opindex fno-nonansi-builtins
1405 Disable built-in declarations of functions that are not mandated by
1406 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1407 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1409 @item -fno-operator-names
1410 @opindex fno-operator-names
1411 Do not treat the operator name keywords @code{and}, @code{bitand},
1412 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1413 synonyms as keywords.
1415 @item -fno-optional-diags
1416 @opindex fno-optional-diags
1417 Disable diagnostics that the standard says a compiler does not need to
1418 issue. Currently, the only such diagnostic issued by G++ is the one for
1419 a name having multiple meanings within a class.
1422 @opindex fpermissive
1423 Downgrade some diagnostics about nonconformant code from errors to
1424 warnings. Thus, using @option{-fpermissive} will allow some
1425 nonconforming code to compile.
1429 Enable automatic template instantiation at link time. This option also
1430 implies @option{-fno-implicit-templates}. @xref{Template
1431 Instantiation}, for more information.
1435 Disable generation of information about every class with virtual
1436 functions for use by the C++ runtime type identification features
1437 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1438 of the language, you can save some space by using this flag. Note that
1439 exception handling uses the same information, but it will generate it as
1444 Emit statistics about front-end processing at the end of the compilation.
1445 This information is generally only useful to the G++ development team.
1447 @item -ftemplate-depth-@var{n}
1448 @opindex ftemplate-depth
1449 Set the maximum instantiation depth for template classes to @var{n}.
1450 A limit on the template instantiation depth is needed to detect
1451 endless recursions during template class instantiation. ANSI/ISO C++
1452 conforming programs must not rely on a maximum depth greater than 17.
1454 @item -fuse-cxa-atexit
1455 @opindex fuse-cxa-atexit
1456 Register destructors for objects with static storage duration with the
1457 @code{__cxa_atexit} function rather than the @code{atexit} function.
1458 This option is required for fully standards-compliant handling of static
1459 destructors, but will only work if your C library supports
1460 @code{__cxa_atexit}.
1464 Do not use weak symbol support, even if it is provided by the linker.
1465 By default, G++ will use weak symbols if they are available. This
1466 option exists only for testing, and should not be used by end-users;
1467 it will result in inferior code and has no benefits. This option may
1468 be removed in a future release of G++.
1472 Do not search for header files in the standard directories specific to
1473 C++, but do still search the other standard directories. (This option
1474 is used when building the C++ library.)
1477 In addition, these optimization, warning, and code generation options
1478 have meanings only for C++ programs:
1481 @item -fno-default-inline
1482 @opindex fno-default-inline
1483 Do not assume @samp{inline} for functions defined inside a class scope.
1484 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1485 functions will have linkage like inline functions; they just won't be
1488 @item -Wabi @r{(C++ only)}
1490 Warn when G++ generates code that is probably not compatible with the
1491 vendor-neutral C++ ABI. Although an effort has been made to warn about
1492 all such cases, there are probably some cases that are not warned about,
1493 even though G++ is generating incompatible code. There may also be
1494 cases where warnings are emitted even though the code that is generated
1497 You should rewrite your code to avoid these warnings if you are
1498 concerned about the fact that code generated by G++ may not be binary
1499 compatible with code generated by other compilers.
1501 The known incompatibilities at this point include:
1506 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1507 pack data into the same byte as a base class. For example:
1510 struct A @{ virtual void f(); int f1 : 1; @};
1511 struct B : public A @{ int f2 : 1; @};
1515 In this case, G++ will place @code{B::f2} into the same byte
1516 as@code{A::f1}; other compilers will not. You can avoid this problem
1517 by explicitly padding @code{A} so that its size is a multiple of the
1518 byte size on your platform; that will cause G++ and other compilers to
1519 layout @code{B} identically.
1522 Incorrect handling of tail-padding for virtual bases. G++ does not use
1523 tail padding when laying out virtual bases. For example:
1526 struct A @{ virtual void f(); char c1; @};
1527 struct B @{ B(); char c2; @};
1528 struct C : public A, public virtual B @{@};
1532 In this case, G++ will not place @code{B} into the tail-padding for
1533 @code{A}; other compilers will. You can avoid this problem by
1534 explicitly padding @code{A} so that its size is a multiple of its
1535 alignment (ignoring virtual base classes); that will cause G++ and other
1536 compilers to layout @code{C} identically.
1539 Incorrect handling of bit-fields with declared widths greater than that
1540 of their underlying types, when the bit-fields appear in a union. For
1544 union U @{ int i : 4096; @};
1548 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1549 union too small by the number of bits in an @code{int}.
1552 Empty classes can be placed at incorrect offsets. For example:
1562 struct C : public B, public A @{@};
1566 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1567 it should be placed at offset zero. G++ mistakenly believes that the
1568 @code{A} data member of @code{B} is already at offset zero.
1571 Names of template functions whose types involve @code{typename} or
1572 template template parameters can be mangled incorrectly.
1575 template <typename Q>
1576 void f(typename Q::X) @{@}
1578 template <template <typename> class Q>
1579 void f(typename Q<int>::X) @{@}
1583 Instantiations of these templates may be mangled incorrectly.
1587 @item -Wctor-dtor-privacy @r{(C++ only)}
1588 @opindex Wctor-dtor-privacy
1589 Warn when a class seems unusable because all the constructors or
1590 destructors in that class are private, and it has neither friends nor
1591 public static member functions.
1593 @item -Wnon-virtual-dtor @r{(C++ only)}
1594 @opindex Wnon-virtual-dtor
1595 Warn when a class appears to be polymorphic, thereby requiring a virtual
1596 destructor, yet it declares a non-virtual one.
1597 This warning is enabled by @option{-Wall}.
1599 @item -Wreorder @r{(C++ only)}
1601 @cindex reordering, warning
1602 @cindex warning for reordering of member initializers
1603 Warn when the order of member initializers given in the code does not
1604 match the order in which they must be executed. For instance:
1610 A(): j (0), i (1) @{ @}
1614 The compiler will rearrange the member initializers for @samp{i}
1615 and @samp{j} to match the declaration order of the members, emitting
1616 a warning to that effect. This warning is enabled by @option{-Wall}.
1619 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1622 @item -Weffc++ @r{(C++ only)}
1624 Warn about violations of the following style guidelines from Scott Meyers'
1625 @cite{Effective C++} book:
1629 Item 11: Define a copy constructor and an assignment operator for classes
1630 with dynamically allocated memory.
1633 Item 12: Prefer initialization to assignment in constructors.
1636 Item 14: Make destructors virtual in base classes.
1639 Item 15: Have @code{operator=} return a reference to @code{*this}.
1642 Item 23: Don't try to return a reference when you must return an object.
1646 Also warn about violations of the following style guidelines from
1647 Scott Meyers' @cite{More Effective C++} book:
1651 Item 6: Distinguish between prefix and postfix forms of increment and
1652 decrement operators.
1655 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1659 When selecting this option, be aware that the standard library
1660 headers do not obey all of these guidelines; use @samp{grep -v}
1661 to filter out those warnings.
1663 @item -Wno-deprecated @r{(C++ only)}
1664 @opindex Wno-deprecated
1665 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1667 @item -Wno-non-template-friend @r{(C++ only)}
1668 @opindex Wno-non-template-friend
1669 Disable warnings when non-templatized friend functions are declared
1670 within a template. Since the advent of explicit template specification
1671 support in G++, if the name of the friend is an unqualified-id (i.e.,
1672 @samp{friend foo(int)}), the C++ language specification demands that the
1673 friend declare or define an ordinary, nontemplate function. (Section
1674 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1675 could be interpreted as a particular specialization of a templatized
1676 function. Because this non-conforming behavior is no longer the default
1677 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1678 check existing code for potential trouble spots and is on by default.
1679 This new compiler behavior can be turned off with
1680 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1681 but disables the helpful warning.
1683 @item -Wold-style-cast @r{(C++ only)}
1684 @opindex Wold-style-cast
1685 Warn if an old-style (C-style) cast to a non-void type is used within
1686 a C++ program. The new-style casts (@samp{static_cast},
1687 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1688 unintended effects and much easier to search for.
1690 @item -Woverloaded-virtual @r{(C++ only)}
1691 @opindex Woverloaded-virtual
1692 @cindex overloaded virtual fn, warning
1693 @cindex warning for overloaded virtual fn
1694 Warn when a function declaration hides virtual functions from a
1695 base class. For example, in:
1702 struct B: public A @{
1707 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1715 will fail to compile.
1717 @item -Wno-pmf-conversions @r{(C++ only)}
1718 @opindex Wno-pmf-conversions
1719 Disable the diagnostic for converting a bound pointer to member function
1722 @item -Wsign-promo @r{(C++ only)}
1723 @opindex Wsign-promo
1724 Warn when overload resolution chooses a promotion from unsigned or
1725 enumeral type to a signed type, over a conversion to an unsigned type of
1726 the same size. Previous versions of G++ would try to preserve
1727 unsignedness, but the standard mandates the current behavior.
1729 @item -Wsynth @r{(C++ only)}
1731 @cindex warning for synthesized methods
1732 @cindex synthesized methods, warning
1733 Warn when G++'s synthesis behavior does not match that of cfront. For
1739 A& operator = (int);
1749 In this example, G++ will synthesize a default @samp{A& operator =
1750 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1753 @node Objective-C Dialect Options
1754 @section Options Controlling Objective-C Dialect
1756 @cindex compiler options, Objective-C
1757 @cindex Objective-C options, command line
1758 @cindex options, Objective-C
1759 This section describes the command-line options that are only meaningful
1760 for Objective-C programs, but you can also use most of the GNU compiler
1761 options regardless of what language your program is in. For example,
1762 you might compile a file @code{some_class.m} like this:
1765 gcc -g -fgnu-runtime -O -c some_class.m
1769 In this example, @option{-fgnu-runtime} is an option meant only for
1770 Objective-C programs; you can use the other options with any language
1773 Here is a list of options that are @emph{only} for compiling Objective-C
1777 @item -fconstant-string-class=@var{class-name}
1778 @opindex fconstant-string-class
1779 Use @var{class-name} as the name of the class to instantiate for each
1780 literal string specified with the syntax @code{@@"@dots{}"}. The default
1781 class name is @code{NXConstantString}.
1784 @opindex fgnu-runtime
1785 Generate object code compatible with the standard GNU Objective-C
1786 runtime. This is the default for most types of systems.
1788 @item -fnext-runtime
1789 @opindex fnext-runtime
1790 Generate output compatible with the NeXT runtime. This is the default
1791 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1792 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1797 Dump interface declarations for all classes seen in the source file to a
1798 file named @file{@var{sourcename}.decl}.
1801 @opindex Wno-protocol
1802 If a class is declared to implement a protocol, a warning is issued for
1803 every method in the protocol that is not implemented by the class. The
1804 default behavior is to issue a warning for every method not explicitly
1805 implemented in the class, even if a method implementation is inherited
1806 from the superclass. If you use the @code{-Wno-protocol} option, then
1807 methods inherited from the superclass are considered to be implemented,
1808 and no warning is issued for them.
1812 Warn if multiple methods of different types for the same selector are
1813 found during compilation. The check is performed on the list of methods
1814 in the final stage of compilation. Additionally, a check is performed
1815 for each selector appearing in a @code{@@selector(@dots{})}
1816 expression, and a corresponding method for that selector has been found
1817 during compilation. Because these checks scan the method table only at
1818 the end of compilation, these warnings are not produced if the final
1819 stage of compilation is not reached, for example because an error is
1820 found during compilation, or because the @code{-fsyntax-only} option is
1823 @item -Wundeclared-selector
1824 @opindex Wundeclared-selector
1825 Warn if a @code{@@selector(@dots{})} expression referring to an
1826 undeclared selector is found. A selector is considered undeclared if no
1827 method with that name has been declared before the
1828 @code{@@selector(@dots{})} expression, either explicitly in an
1829 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1830 an @code{@@implementation} section. This option always performs its
1831 checks as soon as a @code{@@selector(@dots{})} expression is found,
1832 while @code{-Wselector} only performs its checks in the final stage of
1833 compilation. This also enforces the coding style convention
1834 that methods and selectors must be declared before being used.
1836 @c not documented because only avail via -Wp
1837 @c @item -print-objc-runtime-info
1841 @node Language Independent Options
1842 @section Options to Control Diagnostic Messages Formatting
1843 @cindex options to control diagnostics formatting
1844 @cindex diagnostic messages
1845 @cindex message formatting
1847 Traditionally, diagnostic messages have been formatted irrespective of
1848 the output device's aspect (e.g.@: its width, @dots{}). The options described
1849 below can be used to control the diagnostic messages formatting
1850 algorithm, e.g.@: how many characters per line, how often source location
1851 information should be reported. Right now, only the C++ front end can
1852 honor these options. However it is expected, in the near future, that
1853 the remaining front ends would be able to digest them correctly.
1856 @item -fmessage-length=@var{n}
1857 @opindex fmessage-length
1858 Try to format error messages so that they fit on lines of about @var{n}
1859 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1860 the front ends supported by GCC@. If @var{n} is zero, then no
1861 line-wrapping will be done; each error message will appear on a single
1864 @opindex fdiagnostics-show-location
1865 @item -fdiagnostics-show-location=once
1866 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1867 reporter to emit @emph{once} source location information; that is, in
1868 case the message is too long to fit on a single physical line and has to
1869 be wrapped, the source location won't be emitted (as prefix) again,
1870 over and over, in subsequent continuation lines. This is the default
1873 @item -fdiagnostics-show-location=every-line
1874 Only meaningful in line-wrapping mode. Instructs the diagnostic
1875 messages reporter to emit the same source location information (as
1876 prefix) for physical lines that result from the process of breaking
1877 a message which is too long to fit on a single line.
1881 @node Warning Options
1882 @section Options to Request or Suppress Warnings
1883 @cindex options to control warnings
1884 @cindex warning messages
1885 @cindex messages, warning
1886 @cindex suppressing warnings
1888 Warnings are diagnostic messages that report constructions which
1889 are not inherently erroneous but which are risky or suggest there
1890 may have been an error.
1892 You can request many specific warnings with options beginning @samp{-W},
1893 for example @option{-Wimplicit} to request warnings on implicit
1894 declarations. Each of these specific warning options also has a
1895 negative form beginning @samp{-Wno-} to turn off warnings;
1896 for example, @option{-Wno-implicit}. This manual lists only one of the
1897 two forms, whichever is not the default.
1899 The following options control the amount and kinds of warnings produced
1900 by GCC; for further, language-specific options also refer to
1901 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1904 @cindex syntax checking
1906 @opindex fsyntax-only
1907 Check the code for syntax errors, but don't do anything beyond that.
1911 Issue all the warnings demanded by strict ISO C and ISO C++;
1912 reject all programs that use forbidden extensions, and some other
1913 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1914 version of the ISO C standard specified by any @option{-std} option used.
1916 Valid ISO C and ISO C++ programs should compile properly with or without
1917 this option (though a rare few will require @option{-ansi} or a
1918 @option{-std} option specifying the required version of ISO C)@. However,
1919 without this option, certain GNU extensions and traditional C and C++
1920 features are supported as well. With this option, they are rejected.
1922 @option{-pedantic} does not cause warning messages for use of the
1923 alternate keywords whose names begin and end with @samp{__}. Pedantic
1924 warnings are also disabled in the expression that follows
1925 @code{__extension__}. However, only system header files should use
1926 these escape routes; application programs should avoid them.
1927 @xref{Alternate Keywords}.
1929 Some users try to use @option{-pedantic} to check programs for strict ISO
1930 C conformance. They soon find that it does not do quite what they want:
1931 it finds some non-ISO practices, but not all---only those for which
1932 ISO C @emph{requires} a diagnostic, and some others for which
1933 diagnostics have been added.
1935 A feature to report any failure to conform to ISO C might be useful in
1936 some instances, but would require considerable additional work and would
1937 be quite different from @option{-pedantic}. We don't have plans to
1938 support such a feature in the near future.
1940 Where the standard specified with @option{-std} represents a GNU
1941 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1942 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1943 extended dialect is based. Warnings from @option{-pedantic} are given
1944 where they are required by the base standard. (It would not make sense
1945 for such warnings to be given only for features not in the specified GNU
1946 C dialect, since by definition the GNU dialects of C include all
1947 features the compiler supports with the given option, and there would be
1948 nothing to warn about.)
1950 @item -pedantic-errors
1951 @opindex pedantic-errors
1952 Like @option{-pedantic}, except that errors are produced rather than
1957 Inhibit all warning messages.
1961 Inhibit warning messages about the use of @samp{#import}.
1963 @item -Wchar-subscripts
1964 @opindex Wchar-subscripts
1965 Warn if an array subscript has type @code{char}. This is a common cause
1966 of error, as programmers often forget that this type is signed on some
1971 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1972 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1976 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1977 the arguments supplied have types appropriate to the format string
1978 specified, and that the conversions specified in the format string make
1979 sense. This includes standard functions, and others specified by format
1980 attributes (@pxref{Function Attributes}), in the @code{printf},
1981 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1982 not in the C standard) families.
1984 The formats are checked against the format features supported by GNU
1985 libc version 2.2. These include all ISO C90 and C99 features, as well
1986 as features from the Single Unix Specification and some BSD and GNU
1987 extensions. Other library implementations may not support all these
1988 features; GCC does not support warning about features that go beyond a
1989 particular library's limitations. However, if @option{-pedantic} is used
1990 with @option{-Wformat}, warnings will be given about format features not
1991 in the selected standard version (but not for @code{strfmon} formats,
1992 since those are not in any version of the C standard). @xref{C Dialect
1993 Options,,Options Controlling C Dialect}.
1995 Since @option{-Wformat} also checks for null format arguments for
1996 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
1998 @option{-Wformat} is included in @option{-Wall}. For more control over some
1999 aspects of format checking, the options @option{-Wno-format-y2k},
2000 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2001 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2002 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2004 @item -Wno-format-y2k
2005 @opindex Wno-format-y2k
2006 If @option{-Wformat} is specified, do not warn about @code{strftime}
2007 formats which may yield only a two-digit year.
2009 @item -Wno-format-extra-args
2010 @opindex Wno-format-extra-args
2011 If @option{-Wformat} is specified, do not warn about excess arguments to a
2012 @code{printf} or @code{scanf} format function. The C standard specifies
2013 that such arguments are ignored.
2015 Where the unused arguments lie between used arguments that are
2016 specified with @samp{$} operand number specifications, normally
2017 warnings are still given, since the implementation could not know what
2018 type to pass to @code{va_arg} to skip the unused arguments. However,
2019 in the case of @code{scanf} formats, this option will suppress the
2020 warning if the unused arguments are all pointers, since the Single
2021 Unix Specification says that such unused arguments are allowed.
2023 @item -Wno-format-zero-length
2024 @opindex Wno-format-zero-length
2025 If @option{-Wformat} is specified, do not warn about zero-length formats.
2026 The C standard specifies that zero-length formats are allowed.
2028 @item -Wformat-nonliteral
2029 @opindex Wformat-nonliteral
2030 If @option{-Wformat} is specified, also warn if the format string is not a
2031 string literal and so cannot be checked, unless the format function
2032 takes its format arguments as a @code{va_list}.
2034 @item -Wformat-security
2035 @opindex Wformat-security
2036 If @option{-Wformat} is specified, also warn about uses of format
2037 functions that represent possible security problems. At present, this
2038 warns about calls to @code{printf} and @code{scanf} functions where the
2039 format string is not a string literal and there are no format arguments,
2040 as in @code{printf (foo);}. This may be a security hole if the format
2041 string came from untrusted input and contains @samp{%n}. (This is
2042 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2043 in future warnings may be added to @option{-Wformat-security} that are not
2044 included in @option{-Wformat-nonliteral}.)
2048 Enable @option{-Wformat} plus format checks not included in
2049 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2050 -Wformat-nonliteral -Wformat-security}.
2054 Enable warning about passing a null pointer for arguments marked as
2055 requiring a non-null value by the @code{nonnull} function attribute.
2057 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2058 can be disabled with the @option{-Wno-nonnull} option.
2060 @item -Winit-self @r{(C, C++, and Objective-C only)}
2062 Enable warning about uninitialized variables which are initalized with themselves.
2063 Note this option can only be used with the @option{-Wuninitialized} option and
2064 that only works with @option{-O}.
2066 For an example, the following code will not warn about i being uninitialized
2067 without this option:
2078 @item -Wimplicit-int
2079 @opindex Wimplicit-int
2080 Warn when a declaration does not specify a type.
2082 @item -Wimplicit-function-declaration
2083 @itemx -Werror-implicit-function-declaration
2084 @opindex Wimplicit-function-declaration
2085 @opindex Werror-implicit-function-declaration
2086 Give a warning (or error) whenever a function is used before being
2091 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2095 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2096 function with external linkage, returning int, taking either zero
2097 arguments, two, or three arguments of appropriate types.
2099 @item -Wmissing-braces
2100 @opindex Wmissing-braces
2101 Warn if an aggregate or union initializer is not fully bracketed. In
2102 the following example, the initializer for @samp{a} is not fully
2103 bracketed, but that for @samp{b} is fully bracketed.
2106 int a[2][2] = @{ 0, 1, 2, 3 @};
2107 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2111 @opindex Wparentheses
2112 Warn if parentheses are omitted in certain contexts, such
2113 as when there is an assignment in a context where a truth value
2114 is expected, or when operators are nested whose precedence people
2115 often get confused about.
2117 Also warn about constructions where there may be confusion to which
2118 @code{if} statement an @code{else} branch belongs. Here is an example of
2133 In C, every @code{else} branch belongs to the innermost possible @code{if}
2134 statement, which in this example is @code{if (b)}. This is often not
2135 what the programmer expected, as illustrated in the above example by
2136 indentation the programmer chose. When there is the potential for this
2137 confusion, GCC will issue a warning when this flag is specified.
2138 To eliminate the warning, add explicit braces around the innermost
2139 @code{if} statement so there is no way the @code{else} could belong to
2140 the enclosing @code{if}. The resulting code would look like this:
2156 @item -Wsequence-point
2157 @opindex Wsequence-point
2158 Warn about code that may have undefined semantics because of violations
2159 of sequence point rules in the C standard.
2161 The C standard defines the order in which expressions in a C program are
2162 evaluated in terms of @dfn{sequence points}, which represent a partial
2163 ordering between the execution of parts of the program: those executed
2164 before the sequence point, and those executed after it. These occur
2165 after the evaluation of a full expression (one which is not part of a
2166 larger expression), after the evaluation of the first operand of a
2167 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2168 function is called (but after the evaluation of its arguments and the
2169 expression denoting the called function), and in certain other places.
2170 Other than as expressed by the sequence point rules, the order of
2171 evaluation of subexpressions of an expression is not specified. All
2172 these rules describe only a partial order rather than a total order,
2173 since, for example, if two functions are called within one expression
2174 with no sequence point between them, the order in which the functions
2175 are called is not specified. However, the standards committee have
2176 ruled that function calls do not overlap.
2178 It is not specified when between sequence points modifications to the
2179 values of objects take effect. Programs whose behavior depends on this
2180 have undefined behavior; the C standard specifies that ``Between the
2181 previous and next sequence point an object shall have its stored value
2182 modified at most once by the evaluation of an expression. Furthermore,
2183 the prior value shall be read only to determine the value to be
2184 stored.''. If a program breaks these rules, the results on any
2185 particular implementation are entirely unpredictable.
2187 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2188 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2189 diagnosed by this option, and it may give an occasional false positive
2190 result, but in general it has been found fairly effective at detecting
2191 this sort of problem in programs.
2193 The present implementation of this option only works for C programs. A
2194 future implementation may also work for C++ programs.
2196 The C standard is worded confusingly, therefore there is some debate
2197 over the precise meaning of the sequence point rules in subtle cases.
2198 Links to discussions of the problem, including proposed formal
2199 definitions, may be found on our readings page, at
2200 @w{@uref{http://gcc.gnu.org/readings.html}}.
2203 @opindex Wreturn-type
2204 Warn whenever a function is defined with a return-type that defaults to
2205 @code{int}. Also warn about any @code{return} statement with no
2206 return-value in a function whose return-type is not @code{void}.
2208 For C++, a function without return type always produces a diagnostic
2209 message, even when @option{-Wno-return-type} is specified. The only
2210 exceptions are @samp{main} and functions defined in system headers.
2214 Warn whenever a @code{switch} statement has an index of enumeral type
2215 and lacks a @code{case} for one or more of the named codes of that
2216 enumeration. (The presence of a @code{default} label prevents this
2217 warning.) @code{case} labels outside the enumeration range also
2218 provoke warnings when this option is used.
2220 @item -Wswitch-default
2221 @opindex Wswitch-switch
2222 Warn whenever a @code{switch} statement does not have a @code{default}
2226 @opindex Wswitch-enum
2227 Warn whenever a @code{switch} statement has an index of enumeral type
2228 and lacks a @code{case} for one or more of the named codes of that
2229 enumeration. @code{case} labels outside the enumeration range also
2230 provoke warnings when this option is used.
2234 Warn if any trigraphs are encountered that might change the meaning of
2235 the program (trigraphs within comments are not warned about).
2237 @item -Wunused-function
2238 @opindex Wunused-function
2239 Warn whenever a static function is declared but not defined or a
2240 non\-inline static function is unused.
2242 @item -Wunused-label
2243 @opindex Wunused-label
2244 Warn whenever a label is declared but not used.
2246 To suppress this warning use the @samp{unused} attribute
2247 (@pxref{Variable Attributes}).
2249 @item -Wunused-parameter
2250 @opindex Wunused-parameter
2251 Warn whenever a function parameter is unused aside from its declaration.
2253 To suppress this warning use the @samp{unused} attribute
2254 (@pxref{Variable Attributes}).
2256 @item -Wunused-variable
2257 @opindex Wunused-variable
2258 Warn whenever a local variable or non-constant static variable is unused
2259 aside from its declaration
2261 To suppress this warning use the @samp{unused} attribute
2262 (@pxref{Variable Attributes}).
2264 @item -Wunused-value
2265 @opindex Wunused-value
2266 Warn whenever a statement computes a result that is explicitly not used.
2268 To suppress this warning cast the expression to @samp{void}.
2272 All the above @option{-Wunused} options combined.
2274 In order to get a warning about an unused function parameter, you must
2275 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2276 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2278 @item -Wuninitialized
2279 @opindex Wuninitialized
2280 Warn if an automatic variable is used without first being initialized or
2281 if a variable may be clobbered by a @code{setjmp} call.
2283 These warnings are possible only in optimizing compilation,
2284 because they require data flow information that is computed only
2285 when optimizing. If you don't specify @option{-O}, you simply won't
2288 If you want to warn about code which uses the uninitialized value of the
2289 variable in its own initializer, use the @option{-Winit-self} option.
2291 These warnings occur only for variables that are candidates for
2292 register allocation. Therefore, they do not occur for a variable that
2293 is declared @code{volatile}, or whose address is taken, or whose size
2294 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2295 structures, unions or arrays, even when they are in registers.
2297 Note that there may be no warning about a variable that is used only
2298 to compute a value that itself is never used, because such
2299 computations may be deleted by data flow analysis before the warnings
2302 These warnings are made optional because GCC is not smart
2303 enough to see all the reasons why the code might be correct
2304 despite appearing to have an error. Here is one example of how
2325 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2326 always initialized, but GCC doesn't know this. Here is
2327 another common case:
2332 if (change_y) save_y = y, y = new_y;
2334 if (change_y) y = save_y;
2339 This has no bug because @code{save_y} is used only if it is set.
2341 @cindex @code{longjmp} warnings
2342 This option also warns when a non-volatile automatic variable might be
2343 changed by a call to @code{longjmp}. These warnings as well are possible
2344 only in optimizing compilation.
2346 The compiler sees only the calls to @code{setjmp}. It cannot know
2347 where @code{longjmp} will be called; in fact, a signal handler could
2348 call it at any point in the code. As a result, you may get a warning
2349 even when there is in fact no problem because @code{longjmp} cannot
2350 in fact be called at the place which would cause a problem.
2352 Some spurious warnings can be avoided if you declare all the functions
2353 you use that never return as @code{noreturn}. @xref{Function
2356 @item -Wunknown-pragmas
2357 @opindex Wunknown-pragmas
2358 @cindex warning for unknown pragmas
2359 @cindex unknown pragmas, warning
2360 @cindex pragmas, warning of unknown
2361 Warn when a #pragma directive is encountered which is not understood by
2362 GCC@. If this command line option is used, warnings will even be issued
2363 for unknown pragmas in system header files. This is not the case if
2364 the warnings were only enabled by the @option{-Wall} command line option.
2366 @item -Wstrict-aliasing
2367 @opindex Wstrict-aliasing
2368 This option is only active when @option{-fstrict-aliasing} is active.
2369 It warns about code which might break the strict aliasing rules that the
2370 compiler is using for optimization. The warning does not catch all
2371 cases, but does attempt to catch the more common pitfalls. It is
2372 included in @option{-Wall}.
2376 All of the above @samp{-W} options combined. This enables all the
2377 warnings about constructions that some users consider questionable, and
2378 that are easy to avoid (or modify to prevent the warning), even in
2379 conjunction with macros. This also enables some language-specific
2380 warnings described in @ref{C++ Dialect Options} and
2381 @ref{Objective-C Dialect Options}.
2384 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2385 Some of them warn about constructions that users generally do not
2386 consider questionable, but which occasionally you might wish to check
2387 for; others warn about constructions that are necessary or hard to avoid
2388 in some cases, and there is no simple way to modify the code to suppress
2395 (This option used to be called @option{-W}. The older name is still
2396 supported, but the newer name is more descriptive.) Print extra warning
2397 messages for these events:
2401 A function can return either with or without a value. (Falling
2402 off the end of the function body is considered returning without
2403 a value.) For example, this function would evoke such a
2417 An expression-statement or the left-hand side of a comma expression
2418 contains no side effects.
2419 To suppress the warning, cast the unused expression to void.
2420 For example, an expression such as @samp{x[i,j]} will cause a warning,
2421 but @samp{x[(void)i,j]} will not.
2424 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2427 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2428 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2429 that of ordinary mathematical notation.
2432 Storage-class specifiers like @code{static} are not the first things in
2433 a declaration. According to the C Standard, this usage is obsolescent.
2436 The return type of a function has a type qualifier such as @code{const}.
2437 Such a type qualifier has no effect, since the value returned by a
2438 function is not an lvalue. (But don't warn about the GNU extension of
2439 @code{volatile void} return types. That extension will be warned about
2440 if @option{-pedantic} is specified.)
2443 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2447 A comparison between signed and unsigned values could produce an
2448 incorrect result when the signed value is converted to unsigned.
2449 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2452 An aggregate has an initializer which does not initialize all members.
2453 For example, the following code would cause such a warning, because
2454 @code{x.h} would be implicitly initialized to zero:
2457 struct s @{ int f, g, h; @};
2458 struct s x = @{ 3, 4 @};
2462 A function parameter is declared without a type specifier in K&R-style
2470 An empty body occurs in an @samp{if} or @samp{else} statement.
2473 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2474 @samp{>}, or @samp{>=}.
2477 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2480 Any of several floating-point events that often indicate errors, such as
2481 overflow, underflow, loss of precision, etc.
2483 @item @r{(C++ only)}
2484 An enumerator and a non-enumerator both appear in a conditional expression.
2486 @item @r{(C++ only)}
2487 A non-static reference or non-static @samp{const} member appears in a
2488 class without constructors.
2490 @item @r{(C++ only)}
2491 Ambiguous virtual bases.
2493 @item @r{(C++ only)}
2494 Subscripting an array which has been declared @samp{register}.
2496 @item @r{(C++ only)}
2497 Taking the address of a variable which has been declared @samp{register}.
2499 @item @r{(C++ only)}
2500 A base class is not initialized in a derived class' copy constructor.
2503 @item -Wno-div-by-zero
2504 @opindex Wno-div-by-zero
2505 @opindex Wdiv-by-zero
2506 Do not warn about compile-time integer division by zero. Floating point
2507 division by zero is not warned about, as it can be a legitimate way of
2508 obtaining infinities and NaNs.
2510 @item -Wsystem-headers
2511 @opindex Wsystem-headers
2512 @cindex warnings from system headers
2513 @cindex system headers, warnings from
2514 Print warning messages for constructs found in system header files.
2515 Warnings from system headers are normally suppressed, on the assumption
2516 that they usually do not indicate real problems and would only make the
2517 compiler output harder to read. Using this command line option tells
2518 GCC to emit warnings from system headers as if they occurred in user
2519 code. However, note that using @option{-Wall} in conjunction with this
2520 option will @emph{not} warn about unknown pragmas in system
2521 headers---for that, @option{-Wunknown-pragmas} must also be used.
2524 @opindex Wfloat-equal
2525 Warn if floating point values are used in equality comparisons.
2527 The idea behind this is that sometimes it is convenient (for the
2528 programmer) to consider floating-point values as approximations to
2529 infinitely precise real numbers. If you are doing this, then you need
2530 to compute (by analyzing the code, or in some other way) the maximum or
2531 likely maximum error that the computation introduces, and allow for it
2532 when performing comparisons (and when producing output, but that's a
2533 different problem). In particular, instead of testing for equality, you
2534 would check to see whether the two values have ranges that overlap; and
2535 this is done with the relational operators, so equality comparisons are
2538 @item -Wtraditional @r{(C only)}
2539 @opindex Wtraditional
2540 Warn about certain constructs that behave differently in traditional and
2541 ISO C@. Also warn about ISO C constructs that have no traditional C
2542 equivalent, and/or problematic constructs which should be avoided.
2546 Macro parameters that appear within string literals in the macro body.
2547 In traditional C macro replacement takes place within string literals,
2548 but does not in ISO C@.
2551 In traditional C, some preprocessor directives did not exist.
2552 Traditional preprocessors would only consider a line to be a directive
2553 if the @samp{#} appeared in column 1 on the line. Therefore
2554 @option{-Wtraditional} warns about directives that traditional C
2555 understands but would ignore because the @samp{#} does not appear as the
2556 first character on the line. It also suggests you hide directives like
2557 @samp{#pragma} not understood by traditional C by indenting them. Some
2558 traditional implementations would not recognize @samp{#elif}, so it
2559 suggests avoiding it altogether.
2562 A function-like macro that appears without arguments.
2565 The unary plus operator.
2568 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2569 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2570 constants.) Note, these suffixes appear in macros defined in the system
2571 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2572 Use of these macros in user code might normally lead to spurious
2573 warnings, however gcc's integrated preprocessor has enough context to
2574 avoid warning in these cases.
2577 A function declared external in one block and then used after the end of
2581 A @code{switch} statement has an operand of type @code{long}.
2584 A non-@code{static} function declaration follows a @code{static} one.
2585 This construct is not accepted by some traditional C compilers.
2588 The ISO type of an integer constant has a different width or
2589 signedness from its traditional type. This warning is only issued if
2590 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2591 typically represent bit patterns, are not warned about.
2594 Usage of ISO string concatenation is detected.
2597 Initialization of automatic aggregates.
2600 Identifier conflicts with labels. Traditional C lacks a separate
2601 namespace for labels.
2604 Initialization of unions. If the initializer is zero, the warning is
2605 omitted. This is done under the assumption that the zero initializer in
2606 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2607 initializer warnings and relies on default initialization to zero in the
2611 Conversions by prototypes between fixed/floating point values and vice
2612 versa. The absence of these prototypes when compiling with traditional
2613 C would cause serious problems. This is a subset of the possible
2614 conversion warnings, for the full set use @option{-Wconversion}.
2617 Use of ISO C style function definitions. This warning intentionally is
2618 @emph{not} issued for prototype declarations or variadic functions
2619 because these ISO C features will appear in your code when using
2620 libiberty's traditional C compatibility macros, @code{PARAMS} and
2621 @code{VPARAMS}. This warning is also bypassed for nested functions
2622 because that feature is already a gcc extension and thus not relevant to
2623 traditional C compatibility.
2626 @item -Wdeclaration-after-statement @r{(C only)}
2627 @opindex Wdeclaration-after-statement
2628 Warn when a declaration is found after a statement in a block. This
2629 construct, known from C++, was introduced with ISO C99 and is by default
2630 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2631 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2635 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2637 @item -Wendif-labels
2638 @opindex Wendif-labels
2639 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2643 Warn whenever a local variable shadows another local variable, parameter or
2644 global variable or whenever a built-in function is shadowed.
2646 @item -Wlarger-than-@var{len}
2647 @opindex Wlarger-than
2648 Warn whenever an object of larger than @var{len} bytes is defined.
2650 @item -Wpointer-arith
2651 @opindex Wpointer-arith
2652 Warn about anything that depends on the ``size of'' a function type or
2653 of @code{void}. GNU C assigns these types a size of 1, for
2654 convenience in calculations with @code{void *} pointers and pointers
2657 @item -Wbad-function-cast @r{(C only)}
2658 @opindex Wbad-function-cast
2659 Warn whenever a function call is cast to a non-matching type.
2660 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2664 Warn whenever a pointer is cast so as to remove a type qualifier from
2665 the target type. For example, warn if a @code{const char *} is cast
2666 to an ordinary @code{char *}.
2669 @opindex Wcast-align
2670 Warn whenever a pointer is cast such that the required alignment of the
2671 target is increased. For example, warn if a @code{char *} is cast to
2672 an @code{int *} on machines where integers can only be accessed at
2673 two- or four-byte boundaries.
2675 @item -Wwrite-strings
2676 @opindex Wwrite-strings
2677 When compiling C, give string constants the type @code{const
2678 char[@var{length}]} so that
2679 copying the address of one into a non-@code{const} @code{char *}
2680 pointer will get a warning; when compiling C++, warn about the
2681 deprecated conversion from string constants to @code{char *}.
2682 These warnings will help you find at
2683 compile time code that can try to write into a string constant, but
2684 only if you have been very careful about using @code{const} in
2685 declarations and prototypes. Otherwise, it will just be a nuisance;
2686 this is why we did not make @option{-Wall} request these warnings.
2689 @opindex Wconversion
2690 Warn if a prototype causes a type conversion that is different from what
2691 would happen to the same argument in the absence of a prototype. This
2692 includes conversions of fixed point to floating and vice versa, and
2693 conversions changing the width or signedness of a fixed point argument
2694 except when the same as the default promotion.
2696 Also, warn if a negative integer constant expression is implicitly
2697 converted to an unsigned type. For example, warn about the assignment
2698 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2699 casts like @code{(unsigned) -1}.
2701 @item -Wsign-compare
2702 @opindex Wsign-compare
2703 @cindex warning for comparison of signed and unsigned values
2704 @cindex comparison of signed and unsigned values, warning
2705 @cindex signed and unsigned values, comparison warning
2706 Warn when a comparison between signed and unsigned values could produce
2707 an incorrect result when the signed value is converted to unsigned.
2708 This warning is also enabled by @option{-Wextra}; to get the other warnings
2709 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2711 @item -Waggregate-return
2712 @opindex Waggregate-return
2713 Warn if any functions that return structures or unions are defined or
2714 called. (In languages where you can return an array, this also elicits
2717 @item -Wstrict-prototypes @r{(C only)}
2718 @opindex Wstrict-prototypes
2719 Warn if a function is declared or defined without specifying the
2720 argument types. (An old-style function definition is permitted without
2721 a warning if preceded by a declaration which specifies the argument
2724 @item -Wmissing-prototypes @r{(C only)}
2725 @opindex Wmissing-prototypes
2726 Warn if a global function is defined without a previous prototype
2727 declaration. This warning is issued even if the definition itself
2728 provides a prototype. The aim is to detect global functions that fail
2729 to be declared in header files.
2731 @item -Wmissing-declarations @r{(C only)}
2732 @opindex Wmissing-declarations
2733 Warn if a global function is defined without a previous declaration.
2734 Do so even if the definition itself provides a prototype.
2735 Use this option to detect global functions that are not declared in
2738 @item -Wmissing-noreturn
2739 @opindex Wmissing-noreturn
2740 Warn about functions which might be candidates for attribute @code{noreturn}.
2741 Note these are only possible candidates, not absolute ones. Care should
2742 be taken to manually verify functions actually do not ever return before
2743 adding the @code{noreturn} attribute, otherwise subtle code generation
2744 bugs could be introduced. You will not get a warning for @code{main} in
2745 hosted C environments.
2747 @item -Wmissing-format-attribute
2748 @opindex Wmissing-format-attribute
2750 If @option{-Wformat} is enabled, also warn about functions which might be
2751 candidates for @code{format} attributes. Note these are only possible
2752 candidates, not absolute ones. GCC will guess that @code{format}
2753 attributes might be appropriate for any function that calls a function
2754 like @code{vprintf} or @code{vscanf}, but this might not always be the
2755 case, and some functions for which @code{format} attributes are
2756 appropriate may not be detected. This option has no effect unless
2757 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2759 @item -Wno-multichar
2760 @opindex Wno-multichar
2762 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2763 Usually they indicate a typo in the user's code, as they have
2764 implementation-defined values, and should not be used in portable code.
2766 @item -Wno-deprecated-declarations
2767 @opindex Wno-deprecated-declarations
2768 Do not warn about uses of functions, variables, and types marked as
2769 deprecated by using the @code{deprecated} attribute.
2770 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2771 @pxref{Type Attributes}.)
2775 Warn if a structure is given the packed attribute, but the packed
2776 attribute has no effect on the layout or size of the structure.
2777 Such structures may be mis-aligned for little benefit. For
2778 instance, in this code, the variable @code{f.x} in @code{struct bar}
2779 will be misaligned even though @code{struct bar} does not itself
2780 have the packed attribute:
2787 @} __attribute__((packed));
2797 Warn if padding is included in a structure, either to align an element
2798 of the structure or to align the whole structure. Sometimes when this
2799 happens it is possible to rearrange the fields of the structure to
2800 reduce the padding and so make the structure smaller.
2802 @item -Wredundant-decls
2803 @opindex Wredundant-decls
2804 Warn if anything is declared more than once in the same scope, even in
2805 cases where multiple declaration is valid and changes nothing.
2807 @item -Wnested-externs @r{(C only)}
2808 @opindex Wnested-externs
2809 Warn if an @code{extern} declaration is encountered within a function.
2811 @item -Wunreachable-code
2812 @opindex Wunreachable-code
2813 Warn if the compiler detects that code will never be executed.
2815 This option is intended to warn when the compiler detects that at
2816 least a whole line of source code will never be executed, because
2817 some condition is never satisfied or because it is after a
2818 procedure that never returns.
2820 It is possible for this option to produce a warning even though there
2821 are circumstances under which part of the affected line can be executed,
2822 so care should be taken when removing apparently-unreachable code.
2824 For instance, when a function is inlined, a warning may mean that the
2825 line is unreachable in only one inlined copy of the function.
2827 This option is not made part of @option{-Wall} because in a debugging
2828 version of a program there is often substantial code which checks
2829 correct functioning of the program and is, hopefully, unreachable
2830 because the program does work. Another common use of unreachable
2831 code is to provide behavior which is selectable at compile-time.
2835 Warn if a function can not be inlined and it was declared as inline.
2836 Even with this option, the compiler will not warn about failures to
2837 inline functions declared in system headers.
2839 The compiler uses a variety of heuristics to determine whether or not
2840 to inline a function. For example, the compiler takes into account
2841 the size of the function being inlined and the the amount of inlining
2842 that has already been done in the current function. Therefore,
2843 seemingly insignificant changes in the source program can cause the
2844 warnings produced by @option{-Winline} to appear or disappear.
2846 @item -Wno-invalid-offsetof @r{(C++ only)}
2847 @opindex Wno-invalid-offsetof
2848 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2849 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2850 to a non-POD type is undefined. In existing C++ implementations,
2851 however, @samp{offsetof} typically gives meaningful results even when
2852 applied to certain kinds of non-POD types. (Such as a simple
2853 @samp{struct} that fails to be a POD type only by virtue of having a
2854 constructor.) This flag is for users who are aware that they are
2855 writing nonportable code and who have deliberately chosen to ignore the
2858 The restrictions on @samp{offsetof} may be relaxed in a future version
2859 of the C++ standard.
2862 @opindex Winvalid-pch
2863 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2864 the search path but can't be used.
2868 @opindex Wno-long-long
2869 Warn if @samp{long long} type is used. This is default. To inhibit
2870 the warning messages, use @option{-Wno-long-long}. Flags
2871 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2872 only when @option{-pedantic} flag is used.
2874 @item -Wdisabled-optimization
2875 @opindex Wdisabled-optimization
2876 Warn if a requested optimization pass is disabled. This warning does
2877 not generally indicate that there is anything wrong with your code; it
2878 merely indicates that GCC's optimizers were unable to handle the code
2879 effectively. Often, the problem is that your code is too big or too
2880 complex; GCC will refuse to optimize programs when the optimization
2881 itself is likely to take inordinate amounts of time.
2885 Make all warnings into errors.
2888 @node Debugging Options
2889 @section Options for Debugging Your Program or GCC
2890 @cindex options, debugging
2891 @cindex debugging information options
2893 GCC has various special options that are used for debugging
2894 either your program or GCC:
2899 Produce debugging information in the operating system's native format
2900 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2903 On most systems that use stabs format, @option{-g} enables use of extra
2904 debugging information that only GDB can use; this extra information
2905 makes debugging work better in GDB but will probably make other debuggers
2907 refuse to read the program. If you want to control for certain whether
2908 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2909 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
2910 or @option{-gvms} (see below).
2912 Unlike most other C compilers, GCC allows you to use @option{-g} with
2913 @option{-O}. The shortcuts taken by optimized code may occasionally
2914 produce surprising results: some variables you declared may not exist
2915 at all; flow of control may briefly move where you did not expect it;
2916 some statements may not be executed because they compute constant
2917 results or their values were already at hand; some statements may
2918 execute in different places because they were moved out of loops.
2920 Nevertheless it proves possible to debug optimized output. This makes
2921 it reasonable to use the optimizer for programs that might have bugs.
2923 The following options are useful when GCC is generated with the
2924 capability for more than one debugging format.
2928 Produce debugging information for use by GDB@. This means to use the
2929 most expressive format available (DWARF 2, stabs, or the native format
2930 if neither of those are supported), including GDB extensions if at all
2935 Produce debugging information in stabs format (if that is supported),
2936 without GDB extensions. This is the format used by DBX on most BSD
2937 systems. On MIPS, Alpha and System V Release 4 systems this option
2938 produces stabs debugging output which is not understood by DBX or SDB@.
2939 On System V Release 4 systems this option requires the GNU assembler.
2941 @item -feliminate-unused-debug-symbols
2942 @opindex feliminate-unused-debug-symbols
2943 Produce debugging information in stabs format (if that is supported),
2944 for only symbols that are actually used.
2948 Produce debugging information in stabs format (if that is supported),
2949 using GNU extensions understood only by the GNU debugger (GDB)@. The
2950 use of these extensions is likely to make other debuggers crash or
2951 refuse to read the program.
2955 Produce debugging information in COFF format (if that is supported).
2956 This is the format used by SDB on most System V systems prior to
2961 Produce debugging information in XCOFF format (if that is supported).
2962 This is the format used by the DBX debugger on IBM RS/6000 systems.
2966 Produce debugging information in XCOFF format (if that is supported),
2967 using GNU extensions understood only by the GNU debugger (GDB)@. The
2968 use of these extensions is likely to make other debuggers crash or
2969 refuse to read the program, and may cause assemblers other than the GNU
2970 assembler (GAS) to fail with an error.
2974 Produce debugging information in DWARF version 1 format (if that is
2975 supported). This is the format used by SDB on most System V Release 4
2978 This option is deprecated.
2982 Produce debugging information in DWARF version 1 format (if that is
2983 supported), using GNU extensions understood only by the GNU debugger
2984 (GDB)@. The use of these extensions is likely to make other debuggers
2985 crash or refuse to read the program.
2987 This option is deprecated.
2991 Produce debugging information in DWARF version 2 format (if that is
2992 supported). This is the format used by DBX on IRIX 6.
2996 Produce debugging information in VMS debug format (if that is
2997 supported). This is the format used by DEBUG on VMS systems.
3000 @itemx -ggdb@var{level}
3001 @itemx -gstabs@var{level}
3002 @itemx -gcoff@var{level}
3003 @itemx -gxcoff@var{level}
3004 @itemx -gvms@var{level}
3005 Request debugging information and also use @var{level} to specify how
3006 much information. The default level is 2.
3008 Level 1 produces minimal information, enough for making backtraces in
3009 parts of the program that you don't plan to debug. This includes
3010 descriptions of functions and external variables, but no information
3011 about local variables and no line numbers.
3013 Level 3 includes extra information, such as all the macro definitions
3014 present in the program. Some debuggers support macro expansion when
3015 you use @option{-g3}.
3017 Note that in order to avoid confusion between DWARF1 debug level 2,
3018 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
3019 a concatenated debug level. Instead use an additional @option{-g@var{level}}
3020 option to change the debug level for DWARF1 or DWARF2.
3022 @item -feliminate-dwarf2-dups
3023 @opindex feliminate-dwarf2-dups
3024 Compress DWARF2 debugging information by eliminating duplicated
3025 information about each symbol. This option only makes sense when
3026 generating DWARF2 debugging information with @option{-gdwarf-2}.
3028 @cindex @command{prof}
3031 Generate extra code to write profile information suitable for the
3032 analysis program @command{prof}. You must use this option when compiling
3033 the source files you want data about, and you must also use it when
3036 @cindex @command{gprof}
3039 Generate extra code to write profile information suitable for the
3040 analysis program @command{gprof}. You must use this option when compiling
3041 the source files you want data about, and you must also use it when
3046 Makes the compiler print out each function name as it is compiled, and
3047 print some statistics about each pass when it finishes.
3050 @opindex ftime-report
3051 Makes the compiler print some statistics about the time consumed by each
3052 pass when it finishes.
3055 @opindex fmem-report
3056 Makes the compiler print some statistics about permanent memory
3057 allocation when it finishes.
3059 @item -fprofile-arcs
3060 @opindex fprofile-arcs
3061 Add code so that program flow @dfn{arcs} are instrumented. During
3062 execution the program records how many times each branch and call is
3063 executed and how many times it is taken or returns. When the compiled
3064 program exits it saves this data to a file called
3065 @file{@var{auxname}.gcda} for each source file. The data may be used for
3066 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3067 test coverage analysis (@option{-ftest-coverage}). Each object file's
3068 @var{auxname} is generated from the name of the output file, if
3069 explicitly specified and it is not the final executable, otherwise it is
3070 the basename of the source file. In both cases any suffix is removed
3071 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3072 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3077 Compile the source files with @option{-fprofile-arcs} plus optimization
3078 and code generation options. For test coverage analysis, use the
3079 additional @option{-ftest-coverage} option. You do not need to profile
3080 every source file in a program.
3083 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3084 (the latter implies the former).
3087 Run the program on a representative workload to generate the arc profile
3088 information. This may be repeated any number of times. You can run
3089 concurrent instances of your program, and provided that the file system
3090 supports locking, the data files will be correctly updated. Also
3091 @code{fork} calls are detected and correctly handled (double counting
3095 For profile-directed optimizations, compile the source files again with
3096 the same optimization and code generation options plus
3097 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3098 Control Optimization}).
3101 For test coverage analysis, use @command{gcov} to produce human readable
3102 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3103 @command{gcov} documentation for further information.
3107 With @option{-fprofile-arcs}, for each function of your program GCC
3108 creates a program flow graph, then finds a spanning tree for the graph.
3109 Only arcs that are not on the spanning tree have to be instrumented: the
3110 compiler adds code to count the number of times that these arcs are
3111 executed. When an arc is the only exit or only entrance to a block, the
3112 instrumentation code can be added to the block; otherwise, a new basic
3113 block must be created to hold the instrumentation code.
3116 @item -ftest-coverage
3117 @opindex ftest-coverage
3118 Produce a notes file that the @command{gcov} code-coverage utility
3119 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3120 show program coverage. Each source file's note file is called
3121 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3122 above for a description of @var{auxname} and instructions on how to
3123 generate test coverage data. Coverage data will match the source files
3124 more closely, if you do not optimize.
3126 @item -d@var{letters}
3128 Says to make debugging dumps during compilation at times specified by
3129 @var{letters}. This is used for debugging the compiler. The file names
3130 for most of the dumps are made by appending a pass number and a word to
3131 the @var{dumpname}. @var{dumpname} is generated from the name of the
3132 output file, if explicitly specified and it is not an executable,
3133 otherwise it is the basename of the source file. In both cases any
3134 suffix is removed (e.g. @file{foo.00.rtl} or @file{foo.01.sibling}).
3135 Here are the possible letters for use in @var{letters}, and their
3141 Annotate the assembler output with miscellaneous debugging information.
3144 Dump after computing branch probabilities, to @file{@var{file}.16.bp}.
3147 Dump after block reordering, to @file{@var{file}.32.bbro}.
3150 Dump after instruction combination, to the file @file{@var{file}.22.combine}.
3153 Dump after the first if conversion, to the file @file{@var{file}.17.ce1}.
3154 Also dump after the second if conversion, to the file @file{@var{file}.23.ce2}.
3157 Dump after branch target load optimization, to to @file{@var{file}.34.btl}.
3158 Also dump after delayed branch scheduling, to @file{@var{file}.37.dbr}.
3161 Dump all macro definitions, at the end of preprocessing, in addition to
3165 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
3166 @file{@var{file}.010.ussa}.
3169 Dump after the second if conversion, to @file{@var{file}.33.ce3}.
3172 Dump after control and data flow analysis, to @file{@var{file}.15.cfg}.
3173 Also dump after life analysis, to @file{@var{file}.21.life}.
3176 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.11.addressof}.
3179 Dump after global register allocation, to @file{@var{file}.27.greg}.
3182 Dump after GCSE, to @file{@var{file}.12.gcse}.
3183 Also dump after jump bypassing and control flow optimizations, to
3184 @file{@var{file}.14.bypass}.
3187 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3190 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3193 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3196 Dump after conversion from registers to stack, to @file{@var{file}.36.stack}.
3199 Dump after local register allocation, to @file{@var{file}.26.lreg}.
3202 Dump after loop optimization passes, to @file{@var{file}.13.loop} and
3203 @file{@var{file}.19.loop2}.
3206 Dump after performing the machine dependent reorganization pass, to
3207 @file{@var{file}.37.mach}.
3210 Dump after register renumbering, to @file{@var{file}.31.rnreg}.
3213 Dump after the register move pass, to @file{@var{file}.24.regmove}.
3216 Dump after post-reload optimizations, to @file{@var{file}.28.postreload}.
3219 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3222 Dump after the second scheduling pass, to @file{@var{file}.35.sched2}.
3225 Dump after CSE (including the jump optimization that sometimes follows
3226 CSE), to @file{@var{file}.019.cse}.
3229 Dump after the first scheduling pass, to @file{@var{file}.25.sched}.
3232 Dump after the second CSE pass (including the jump optimization that
3233 sometimes follows CSE), to @file{@var{file}.20.cse2}.
3236 Dump after running tracer, to @file{@var{file}.18.tracer}.
3239 Dump after null pointer elimination pass to @file{@var{file}.018.null}.
3242 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3245 Dump after the second flow pass, to @file{@var{file}.29.flow2}.
3248 Dump after SSA conditional constant propagation, to
3249 @file{@var{file}.06.ssaccp}.
3252 Dump after SSA dead code elimination, to @file{@var{file}.07.ssadce}.
3255 Dump after the peephole pass, to @file{@var{file}.30.peephole2}.
3258 Produce all the dumps listed above.
3261 Produce a core dump whenever an error occurs.
3264 Print statistics on memory usage, at the end of the run, to
3268 Annotate the assembler output with a comment indicating which
3269 pattern and alternative was used. The length of each instruction is
3273 Dump the RTL in the assembler output as a comment before each instruction.
3274 Also turns on @option{-dp} annotation.
3277 For each of the other indicated dump files (except for
3278 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3279 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3282 Just generate RTL for a function instead of compiling it. Usually used
3286 Dump debugging information during parsing, to standard error.
3289 @item -fdump-unnumbered
3290 @opindex fdump-unnumbered
3291 When doing debugging dumps (see @option{-d} option above), suppress instruction
3292 numbers and line number note output. This makes it more feasible to
3293 use diff on debugging dumps for compiler invocations with different
3294 options, in particular with and without @option{-g}.
3296 @item -fdump-translation-unit @r{(C and C++ only)}
3297 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3298 @opindex fdump-translation-unit
3299 Dump a representation of the tree structure for the entire translation
3300 unit to a file. The file name is made by appending @file{.tu} to the
3301 source file name. If the @samp{-@var{options}} form is used, @var{options}
3302 controls the details of the dump as described for the
3303 @option{-fdump-tree} options.
3305 @item -fdump-class-hierarchy @r{(C++ only)}
3306 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3307 @opindex fdump-class-hierarchy
3308 Dump a representation of each class's hierarchy and virtual function
3309 table layout to a file. The file name is made by appending @file{.class}
3310 to the source file name. If the @samp{-@var{options}} form is used,
3311 @var{options} controls the details of the dump as described for the
3312 @option{-fdump-tree} options.
3314 @item -fdump-tree-@var{switch} @r{(C++ only)}
3315 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3317 Control the dumping at various stages of processing the intermediate
3318 language tree to a file. The file name is generated by appending a switch
3319 specific suffix to the source file name. If the @samp{-@var{options}}
3320 form is used, @var{options} is a list of @samp{-} separated options that
3321 control the details of the dump. Not all options are applicable to all
3322 dumps, those which are not meaningful will be ignored. The following
3323 options are available
3327 Print the address of each node. Usually this is not meaningful as it
3328 changes according to the environment and source file. Its primary use
3329 is for tying up a dump file with a debug environment.
3331 Inhibit dumping of members of a scope or body of a function merely
3332 because that scope has been reached. Only dump such items when they
3333 are directly reachable by some other path.
3335 Turn on all options.
3338 The following tree dumps are possible:
3341 Dump before any tree based optimization, to @file{@var{file}.original}.
3343 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3345 Dump after function inlining, to @file{@var{file}.inlined}.
3348 @item -frandom-seed=@var{string}
3349 @opindex frandom-string
3350 This option provides a seed that GCC uses when it would otherwise use
3351 random numbers. It is used to generate certain symbol names
3352 that have to be different in every compiled file. It is also used to
3353 place unique stamps in coverage data files and the object files that
3354 produce them. You can use the @option{-frandom-seed} option to produce
3355 reproducibly identical object files.
3357 The @var{string} should be different for every file you compile.
3359 @item -fsched-verbose=@var{n}
3360 @opindex fsched-verbose
3361 On targets that use instruction scheduling, this option controls the
3362 amount of debugging output the scheduler prints. This information is
3363 written to standard error, unless @option{-dS} or @option{-dR} is
3364 specified, in which case it is output to the usual dump
3365 listing file, @file{.sched} or @file{.sched2} respectively. However
3366 for @var{n} greater than nine, the output is always printed to standard
3369 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3370 same information as @option{-dRS}. For @var{n} greater than one, it
3371 also output basic block probabilities, detailed ready list information
3372 and unit/insn info. For @var{n} greater than two, it includes RTL
3373 at abort point, control-flow and regions info. And for @var{n} over
3374 four, @option{-fsched-verbose} also includes dependence info.
3378 Store the usual ``temporary'' intermediate files permanently; place them
3379 in the current directory and name them based on the source file. Thus,
3380 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3381 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3382 preprocessed @file{foo.i} output file even though the compiler now
3383 normally uses an integrated preprocessor.
3387 Report the CPU time taken by each subprocess in the compilation
3388 sequence. For C source files, this is the compiler proper and assembler
3389 (plus the linker if linking is done). The output looks like this:
3396 The first number on each line is the ``user time,'' that is time spent
3397 executing the program itself. The second number is ``system time,''
3398 time spent executing operating system routines on behalf of the program.
3399 Both numbers are in seconds.
3401 @item -print-file-name=@var{library}
3402 @opindex print-file-name
3403 Print the full absolute name of the library file @var{library} that
3404 would be used when linking---and don't do anything else. With this
3405 option, GCC does not compile or link anything; it just prints the
3408 @item -print-multi-directory
3409 @opindex print-multi-directory
3410 Print the directory name corresponding to the multilib selected by any
3411 other switches present in the command line. This directory is supposed
3412 to exist in @env{GCC_EXEC_PREFIX}.
3414 @item -print-multi-lib
3415 @opindex print-multi-lib
3416 Print the mapping from multilib directory names to compiler switches
3417 that enable them. The directory name is separated from the switches by
3418 @samp{;}, and each switch starts with an @samp{@@} instead of the
3419 @samp{-}, without spaces between multiple switches. This is supposed to
3420 ease shell-processing.
3422 @item -print-prog-name=@var{program}
3423 @opindex print-prog-name
3424 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3426 @item -print-libgcc-file-name
3427 @opindex print-libgcc-file-name
3428 Same as @option{-print-file-name=libgcc.a}.
3430 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3431 but you do want to link with @file{libgcc.a}. You can do
3434 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3437 @item -print-search-dirs
3438 @opindex print-search-dirs
3439 Print the name of the configured installation directory and a list of
3440 program and library directories gcc will search---and don't do anything else.
3442 This is useful when gcc prints the error message
3443 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3444 To resolve this you either need to put @file{cpp0} and the other compiler
3445 components where gcc expects to find them, or you can set the environment
3446 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3447 Don't forget the trailing '/'.
3448 @xref{Environment Variables}.
3451 @opindex dumpmachine
3452 Print the compiler's target machine (for example,
3453 @samp{i686-pc-linux-gnu})---and don't do anything else.
3456 @opindex dumpversion
3457 Print the compiler version (for example, @samp{3.0})---and don't do
3462 Print the compiler's built-in specs---and don't do anything else. (This
3463 is used when GCC itself is being built.) @xref{Spec Files}.
3465 @item -feliminate-unused-debug-types
3466 @opindex feliminate-unused-debug-types
3467 Normally, when producing DWARF2 output, GCC will emit debugging
3468 information for all types declared in a compilation
3469 unit, regardless of whether or not they are actually used
3470 in that compilation unit. Sometimes this is useful, such as
3471 if, in the debugger, you want to cast a value to a type that is
3472 not actually used in your program (but is declared). More often,
3473 however, this results in a significant amount of wasted space.
3474 With this option, GCC will avoid producing debug symbol output
3475 for types that are nowhere used in the source file being compiled.
3478 @node Optimize Options
3479 @section Options That Control Optimization
3480 @cindex optimize options
3481 @cindex options, optimization
3483 These options control various sorts of optimizations.
3485 Without any optimization option, the compiler's goal is to reduce the
3486 cost of compilation and to make debugging produce the expected
3487 results. Statements are independent: if you stop the program with a
3488 breakpoint between statements, you can then assign a new value to any
3489 variable or change the program counter to any other statement in the
3490 function and get exactly the results you would expect from the source
3493 Turning on optimization flags makes the compiler attempt to improve
3494 the performance and/or code size at the expense of compilation time
3495 and possibly the ability to debug the program.
3497 The compiler performs optimisation based on the knowledge it has of
3498 the program. Using the @option{-funit-at-a-time} flag will allow the
3499 compiler to consider information gained from later functions in the
3500 file when compiling a function. Compiling multiple files at once to a
3501 single output file (and using @option{-funit-at-a-time}) will allow
3502 the compiler to use information gained from all of the files when
3503 compiling each of them.
3505 Not all optimizations are controlled directly by a flag. Only
3506 optimizations that have a flag are listed.
3513 Optimize. Optimizing compilation takes somewhat more time, and a lot
3514 more memory for a large function.
3516 With @option{-O}, the compiler tries to reduce code size and execution
3517 time, without performing any optimizations that take a great deal of
3520 @option{-O} turns on the following optimization flags:
3521 @gccoptlist{-fdefer-pop @gol
3522 -fmerge-constants @gol
3524 -floop-optimize @gol
3526 -fif-conversion @gol
3527 -fif-conversion2 @gol
3528 -fdelayed-branch @gol
3529 -fguess-branch-probability @gol
3532 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3533 where doing so does not interfere with debugging.
3537 Optimize even more. GCC performs nearly all supported optimizations
3538 that do not involve a space-speed tradeoff. The compiler does not
3539 perform loop unrolling or function inlining when you specify @option{-O2}.
3540 As compared to @option{-O}, this option increases both compilation time
3541 and the performance of the generated code.
3543 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3544 also turns on the following optimization flags:
3545 @gccoptlist{-fforce-mem @gol
3546 -foptimize-sibling-calls @gol
3547 -fstrength-reduce @gol
3548 -fcse-follow-jumps -fcse-skip-blocks @gol
3549 -frerun-cse-after-loop -frerun-loop-opt @gol
3550 -fgcse -fgcse-lm -fgcse-sm @gol
3551 -fdelete-null-pointer-checks @gol
3552 -fexpensive-optimizations @gol
3554 -fschedule-insns -fschedule-insns2 @gol
3555 -fsched-interblock -fsched-spec @gol
3558 -freorder-blocks -freorder-functions @gol
3559 -fstrict-aliasing @gol
3560 -falign-functions -falign-jumps @gol
3561 -falign-loops -falign-labels}
3563 Please note the warning under @option{-fgcse} about
3564 invoking @option{-O2} on programs that use computed gotos.
3568 Optimize yet more. @option{-O3} turns on all optimizations specified by
3569 @option{-O2} and also turns on the @option{-finline-functions},
3570 @option{-funit-at-a-time} and @option{-frename-registers} options.
3574 Do not optimize. This is the default.
3578 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3579 do not typically increase code size. It also performs further
3580 optimizations designed to reduce code size.
3582 @option{-Os} disables the following optimization flags:
3583 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3584 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3586 If you use multiple @option{-O} options, with or without level numbers,
3587 the last such option is the one that is effective.
3590 Options of the form @option{-f@var{flag}} specify machine-independent
3591 flags. Most flags have both positive and negative forms; the negative
3592 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3593 below, only one of the forms is listed---the one you typically will
3594 use. You can figure out the other form by either removing @samp{no-}
3597 The following options control specific optimizations. They are either
3598 activated by @option{-O} options or are related to ones that are. You
3599 can use the following flags in the rare cases when ``fine-tuning'' of
3600 optimizations to be performed is desired.
3603 @item -fno-default-inline
3604 @opindex fno-default-inline
3605 Do not make member functions inline by default merely because they are
3606 defined inside the class scope (C++ only). Otherwise, when you specify
3607 @w{@option{-O}}, member functions defined inside class scope are compiled
3608 inline by default; i.e., you don't need to add @samp{inline} in front of
3609 the member function name.
3611 @item -fno-defer-pop
3612 @opindex fno-defer-pop
3613 Always pop the arguments to each function call as soon as that function
3614 returns. For machines which must pop arguments after a function call,
3615 the compiler normally lets arguments accumulate on the stack for several
3616 function calls and pops them all at once.
3618 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3622 Force memory operands to be copied into registers before doing
3623 arithmetic on them. This produces better code by making all memory
3624 references potential common subexpressions. When they are not common
3625 subexpressions, instruction combination should eliminate the separate
3628 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3631 @opindex fforce-addr
3632 Force memory address constants to be copied into registers before
3633 doing arithmetic on them. This may produce better code just as
3634 @option{-fforce-mem} may.
3636 @item -fomit-frame-pointer
3637 @opindex fomit-frame-pointer
3638 Don't keep the frame pointer in a register for functions that
3639 don't need one. This avoids the instructions to save, set up and
3640 restore frame pointers; it also makes an extra register available
3641 in many functions. @strong{It also makes debugging impossible on
3644 On some machines, such as the VAX, this flag has no effect, because
3645 the standard calling sequence automatically handles the frame pointer
3646 and nothing is saved by pretending it doesn't exist. The
3647 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3648 whether a target machine supports this flag. @xref{Registers,,Register
3649 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3651 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3653 @item -foptimize-sibling-calls
3654 @opindex foptimize-sibling-calls
3655 Optimize sibling and tail recursive calls.
3657 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3661 Don't pay attention to the @code{inline} keyword. Normally this option
3662 is used to keep the compiler from expanding any functions inline.
3663 Note that if you are not optimizing, no functions can be expanded inline.
3665 @item -finline-functions
3666 @opindex finline-functions
3667 Integrate all simple functions into their callers. The compiler
3668 heuristically decides which functions are simple enough to be worth
3669 integrating in this way.
3671 If all calls to a given function are integrated, and the function is
3672 declared @code{static}, then the function is normally not output as
3673 assembler code in its own right.
3675 Enabled at level @option{-O3}.
3677 @item -finline-limit=@var{n}
3678 @opindex finline-limit
3679 By default, gcc limits the size of functions that can be inlined. This flag
3680 allows the control of this limit for functions that are explicitly marked as
3681 inline (i.e., marked with the inline keyword or defined within the class
3682 definition in c++). @var{n} is the size of functions that can be inlined in
3683 number of pseudo instructions (not counting parameter handling). The default
3684 value of @var{n} is 600.
3685 Increasing this value can result in more inlined code at
3686 the cost of compilation time and memory consumption. Decreasing usually makes
3687 the compilation faster and less code will be inlined (which presumably
3688 means slower programs). This option is particularly useful for programs that
3689 use inlining heavily such as those based on recursive templates with C++.
3691 Inlining is actually controlled by a number of parameters, which may be
3692 specified individually by using @option{--param @var{name}=@var{value}}.
3693 The @option{-finline-limit=@var{n}} option sets some of these parameters
3697 @item max-inline-insns
3699 @item max-inline-insns-single
3700 is set to @var{n}/2.
3701 @item max-inline-insns-auto
3702 is set to @var{n}/2.
3703 @item min-inline-insns
3704 is set to 130 or @var{n}/4, whichever is smaller.
3705 @item max-inline-insns-rtl
3709 Using @option{-finline-limit=600} thus results in the default settings
3710 for these parameters. See below for a documentation of the individual
3711 parameters controlling inlining.
3713 @emph{Note:} pseudo instruction represents, in this particular context, an
3714 abstract measurement of function's size. In no way, it represents a count
3715 of assembly instructions and as such its exact meaning might change from one
3716 release to an another.
3718 @item -fkeep-inline-functions
3719 @opindex fkeep-inline-functions
3720 Even if all calls to a given function are integrated, and the function
3721 is declared @code{static}, nevertheless output a separate run-time
3722 callable version of the function. This switch does not affect
3723 @code{extern inline} functions.
3725 @item -fkeep-static-consts
3726 @opindex fkeep-static-consts
3727 Emit variables declared @code{static const} when optimization isn't turned
3728 on, even if the variables aren't referenced.
3730 GCC enables this option by default. If you want to force the compiler to
3731 check if the variable was referenced, regardless of whether or not
3732 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3734 @item -fmerge-constants
3735 Attempt to merge identical constants (string constants and floating point
3736 constants) across compilation units.
3738 This option is the default for optimized compilation if the assembler and
3739 linker support it. Use @option{-fno-merge-constants} to inhibit this
3742 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3744 @item -fmerge-all-constants
3745 Attempt to merge identical constants and identical variables.
3747 This option implies @option{-fmerge-constants}. In addition to
3748 @option{-fmerge-constants} this considers e.g. even constant initialized
3749 arrays or initialized constant variables with integral or floating point
3750 types. Languages like C or C++ require each non-automatic variable to
3751 have distinct location, so using this option will result in non-conforming
3756 Use a graph coloring register allocator. Currently this option is meant
3757 for testing, so we are interested to hear about miscompilations with
3760 @item -fno-branch-count-reg
3761 @opindex fno-branch-count-reg
3762 Do not use ``decrement and branch'' instructions on a count register,
3763 but instead generate a sequence of instructions that decrement a
3764 register, compare it against zero, then branch based upon the result.
3765 This option is only meaningful on architectures that support such
3766 instructions, which include x86, PowerPC, IA-64 and S/390.
3768 The default is @option{-fbranch-count-reg}, enabled when
3769 @option{-fstrength-reduce} is enabled.
3771 @item -fno-function-cse
3772 @opindex fno-function-cse
3773 Do not put function addresses in registers; make each instruction that
3774 calls a constant function contain the function's address explicitly.
3776 This option results in less efficient code, but some strange hacks
3777 that alter the assembler output may be confused by the optimizations
3778 performed when this option is not used.
3780 The default is @option{-ffunction-cse}
3782 @item -fno-zero-initialized-in-bss
3783 @opindex fno-zero-initialized-in-bss
3784 If the target supports a BSS section, GCC by default puts variables that
3785 are initialized to zero into BSS@. This can save space in the resulting
3788 This option turns off this behavior because some programs explicitly
3789 rely on variables going to the data section. E.g., so that the
3790 resulting executable can find the beginning of that section and/or make
3791 assumptions based on that.
3793 The default is @option{-fzero-initialized-in-bss}.
3795 @item -fstrength-reduce
3796 @opindex fstrength-reduce
3797 Perform the optimizations of loop strength reduction and
3798 elimination of iteration variables.
3800 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3802 @item -fthread-jumps
3803 @opindex fthread-jumps
3804 Perform optimizations where we check to see if a jump branches to a
3805 location where another comparison subsumed by the first is found. If
3806 so, the first branch is redirected to either the destination of the
3807 second branch or a point immediately following it, depending on whether
3808 the condition is known to be true or false.
3810 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3812 @item -fcse-follow-jumps
3813 @opindex fcse-follow-jumps
3814 In common subexpression elimination, scan through jump instructions
3815 when the target of the jump is not reached by any other path. For
3816 example, when CSE encounters an @code{if} statement with an
3817 @code{else} clause, CSE will follow the jump when the condition
3820 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3822 @item -fcse-skip-blocks
3823 @opindex fcse-skip-blocks
3824 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3825 follow jumps which conditionally skip over blocks. When CSE
3826 encounters a simple @code{if} statement with no else clause,
3827 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3828 body of the @code{if}.
3830 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3832 @item -frerun-cse-after-loop
3833 @opindex frerun-cse-after-loop
3834 Re-run common subexpression elimination after loop optimizations has been
3837 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3839 @item -frerun-loop-opt
3840 @opindex frerun-loop-opt
3841 Run the loop optimizer twice.
3843 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3847 Perform a global common subexpression elimination pass.
3848 This pass also performs global constant and copy propagation.
3850 @emph{Note:} When compiling a program using computed gotos, a GCC
3851 extension, you may get better runtime performance if you disable
3852 the global common subexpression elimination pass by adding
3853 @option{-fno-gcse} to the command line.
3855 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3859 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3860 attempt to move loads which are only killed by stores into themselves. This
3861 allows a loop containing a load/store sequence to be changed to a load outside
3862 the loop, and a copy/store within the loop.
3864 Enabled by default when gcse is enabled.
3868 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3869 subexpression elimination. This pass will attempt to move stores out of loops.
3870 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3871 can be changed to a load before the loop and a store after the loop.
3873 Enabled by default when gcse is enabled.
3875 @item -floop-optimize
3876 @opindex floop-optimize
3877 Perform loop optimizations: move constant expressions out of loops, simplify
3878 exit test conditions and optionally do strength-reduction and loop unrolling as
3881 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3883 @item -fcrossjumping
3884 @opindex crossjumping
3885 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3886 resulting code may or may not perform better than without cross-jumping.
3888 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3890 @item -fif-conversion
3891 @opindex if-conversion
3892 Attempt to transform conditional jumps into branch-less equivalents. This
3893 include use of conditional moves, min, max, set flags and abs instructions, and
3894 some tricks doable by standard arithmetics. The use of conditional execution
3895 on chips where it is available is controlled by @code{if-conversion2}.
3897 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3899 @item -fif-conversion2
3900 @opindex if-conversion2
3901 Use conditional execution (where available) to transform conditional jumps into
3902 branch-less equivalents.
3904 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3906 @item -fdelete-null-pointer-checks
3907 @opindex fdelete-null-pointer-checks
3908 Use global dataflow analysis to identify and eliminate useless checks
3909 for null pointers. The compiler assumes that dereferencing a null
3910 pointer would have halted the program. If a pointer is checked after
3911 it has already been dereferenced, it cannot be null.
3913 In some environments, this assumption is not true, and programs can
3914 safely dereference null pointers. Use
3915 @option{-fno-delete-null-pointer-checks} to disable this optimization
3916 for programs which depend on that behavior.
3918 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3920 @item -fexpensive-optimizations
3921 @opindex fexpensive-optimizations
3922 Perform a number of minor optimizations that are relatively expensive.
3924 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3926 @item -foptimize-register-move
3928 @opindex foptimize-register-move
3930 Attempt to reassign register numbers in move instructions and as
3931 operands of other simple instructions in order to maximize the amount of
3932 register tying. This is especially helpful on machines with two-operand
3935 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3938 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3940 @item -fdelayed-branch
3941 @opindex fdelayed-branch
3942 If supported for the target machine, attempt to reorder instructions
3943 to exploit instruction slots available after delayed branch
3946 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3948 @item -fschedule-insns
3949 @opindex fschedule-insns
3950 If supported for the target machine, attempt to reorder instructions to
3951 eliminate execution stalls due to required data being unavailable. This
3952 helps machines that have slow floating point or memory load instructions
3953 by allowing other instructions to be issued until the result of the load
3954 or floating point instruction is required.
3956 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3958 @item -fschedule-insns2
3959 @opindex fschedule-insns2
3960 Similar to @option{-fschedule-insns}, but requests an additional pass of
3961 instruction scheduling after register allocation has been done. This is
3962 especially useful on machines with a relatively small number of
3963 registers and where memory load instructions take more than one cycle.
3965 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3967 @item -fno-sched-interblock
3968 @opindex fno-sched-interblock
3969 Don't schedule instructions across basic blocks. This is normally
3970 enabled by default when scheduling before register allocation, i.e.@:
3971 with @option{-fschedule-insns} or at @option{-O2} or higher.
3973 @item -fno-sched-spec
3974 @opindex fno-sched-spec
3975 Don't allow speculative motion of non-load instructions. This is normally
3976 enabled by default when scheduling before register allocation, i.e.@:
3977 with @option{-fschedule-insns} or at @option{-O2} or higher.
3979 @item -fsched-spec-load
3980 @opindex fsched-spec-load
3981 Allow speculative motion of some load instructions. This only makes
3982 sense when scheduling before register allocation, i.e.@: with
3983 @option{-fschedule-insns} or at @option{-O2} or higher.
3985 @item -fsched-spec-load-dangerous
3986 @opindex fsched-spec-load-dangerous
3987 Allow speculative motion of more load instructions. This only makes
3988 sense when scheduling before register allocation, i.e.@: with
3989 @option{-fschedule-insns} or at @option{-O2} or higher.
3991 @item -fsched2-use-superblocks
3992 @opindex fsched2-use-superblocks
3993 When scheduling after register allocation, do use superblock scheduling
3994 algorithm. Superblock scheduling allows motion across basic block boundaries
3995 resulting on faster schedules. This option is experimental, as not all machine
3996 descriptions used by GCC model the CPU closely enough to avoid unreliable
3997 results from the algorithm.
3999 This only makes sense when scheduling after register allocation, i.e.@: with
4000 @option{-fschedule-insns2} or at @option{-O2} or higher.
4002 @item -fsched2-use-traces
4003 @opindex fsched2-use-traces
4004 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4005 allocation and additionally perform code duplication in order to increase the
4006 size of superblocks using tracer pass. See @option{-ftracer} for details on
4009 This mode should produce faster but significantly longer programs. Also
4010 without @code{-fbranch-probabilities} the traces constructed may not match the
4011 reality and hurt the performance. This only makes
4012 sense when scheduling after register allocation, i.e.@: with
4013 @option{-fschedule-insns2} or at @option{-O2} or higher.
4015 @item -fcaller-saves
4016 @opindex fcaller-saves
4017 Enable values to be allocated in registers that will be clobbered by
4018 function calls, by emitting extra instructions to save and restore the
4019 registers around such calls. Such allocation is done only when it
4020 seems to result in better code than would otherwise be produced.
4022 This option is always enabled by default on certain machines, usually
4023 those which have no call-preserved registers to use instead.
4025 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4027 @item -fmove-all-movables
4028 @opindex fmove-all-movables
4029 Forces all invariant computations in loops to be moved
4032 @item -freduce-all-givs
4033 @opindex freduce-all-givs
4034 Forces all general-induction variables in loops to be
4037 @emph{Note:} When compiling programs written in Fortran,
4038 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4039 by default when you use the optimizer.
4041 These options may generate better or worse code; results are highly
4042 dependent on the structure of loops within the source code.
4044 These two options are intended to be removed someday, once
4045 they have helped determine the efficacy of various
4046 approaches to improving loop optimizations.
4048 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4049 know how use of these options affects
4050 the performance of your production code.
4051 We're very interested in code that runs @emph{slower}
4052 when these options are @emph{enabled}.
4055 @itemx -fno-peephole2
4056 @opindex fno-peephole
4057 @opindex fno-peephole2
4058 Disable any machine-specific peephole optimizations. The difference
4059 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4060 are implemented in the compiler; some targets use one, some use the
4061 other, a few use both.
4063 @option{-fpeephole} is enabled by default.
4064 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4066 @item -fno-guess-branch-probability
4067 @opindex fno-guess-branch-probability
4068 Do not guess branch probabilities using a randomized model.
4070 Sometimes gcc will opt to use a randomized model to guess branch
4071 probabilities, when none are available from either profiling feedback
4072 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4073 different runs of the compiler on the same program may produce different
4076 In a hard real-time system, people don't want different runs of the
4077 compiler to produce code that has different behavior; minimizing
4078 non-determinism is of paramount import. This switch allows users to
4079 reduce non-determinism, possibly at the expense of inferior
4082 The default is @option{-fguess-branch-probability} at levels
4083 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4085 @item -freorder-blocks
4086 @opindex freorder-blocks
4087 Reorder basic blocks in the compiled function in order to reduce number of
4088 taken branches and improve code locality.
4090 Enabled at levels @option{-O2}, @option{-O3}.
4092 @item -freorder-functions
4093 @opindex freorder-functions
4094 Reorder basic blocks in the compiled function in order to reduce number of
4095 taken branches and improve code locality. This is implemented by using special
4096 subsections @code{text.hot} for most frequently executed functions and
4097 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4098 the linker so object file format must support named sections and linker must
4099 place them in a reasonable way.
4101 Also profile feedback must be available in to make this option effective. See
4102 @option{-fprofile-arcs} for details.
4104 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4106 @item -fstrict-aliasing
4107 @opindex fstrict-aliasing
4108 Allows the compiler to assume the strictest aliasing rules applicable to
4109 the language being compiled. For C (and C++), this activates
4110 optimizations based on the type of expressions. In particular, an
4111 object of one type is assumed never to reside at the same address as an
4112 object of a different type, unless the types are almost the same. For
4113 example, an @code{unsigned int} can alias an @code{int}, but not a
4114 @code{void*} or a @code{double}. A character type may alias any other
4117 Pay special attention to code like this:
4130 The practice of reading from a different union member than the one most
4131 recently written to (called ``type-punning'') is common. Even with
4132 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4133 is accessed through the union type. So, the code above will work as
4134 expected. However, this code might not:
4145 Every language that wishes to perform language-specific alias analysis
4146 should define a function that computes, given an @code{tree}
4147 node, an alias set for the node. Nodes in different alias sets are not
4148 allowed to alias. For an example, see the C front-end function
4149 @code{c_get_alias_set}.
4151 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4153 @item -falign-functions
4154 @itemx -falign-functions=@var{n}
4155 @opindex falign-functions
4156 Align the start of functions to the next power-of-two greater than
4157 @var{n}, skipping up to @var{n} bytes. For instance,
4158 @option{-falign-functions=32} aligns functions to the next 32-byte
4159 boundary, but @option{-falign-functions=24} would align to the next
4160 32-byte boundary only if this can be done by skipping 23 bytes or less.
4162 @option{-fno-align-functions} and @option{-falign-functions=1} are
4163 equivalent and mean that functions will not be aligned.
4165 Some assemblers only support this flag when @var{n} is a power of two;
4166 in that case, it is rounded up.
4168 If @var{n} is not specified or is zero, use a machine-dependent default.
4170 Enabled at levels @option{-O2}, @option{-O3}.
4172 @item -falign-labels
4173 @itemx -falign-labels=@var{n}
4174 @opindex falign-labels
4175 Align all branch targets to a power-of-two boundary, skipping up to
4176 @var{n} bytes like @option{-falign-functions}. This option can easily
4177 make code slower, because it must insert dummy operations for when the
4178 branch target is reached in the usual flow of the code.
4180 @option{-fno-align-labels} and @option{-falign-labels=1} are
4181 equivalent and mean that labels will not be aligned.
4183 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4184 are greater than this value, then their values are used instead.
4186 If @var{n} is not specified or is zero, use a machine-dependent default
4187 which is very likely to be @samp{1}, meaning no alignment.
4189 Enabled at levels @option{-O2}, @option{-O3}.
4192 @itemx -falign-loops=@var{n}
4193 @opindex falign-loops
4194 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4195 like @option{-falign-functions}. The hope is that the loop will be
4196 executed many times, which will make up for any execution of the dummy
4199 @option{-fno-align-loops} and @option{-falign-loops=1} are
4200 equivalent and mean that loops will not be aligned.
4202 If @var{n} is not specified or is zero, use a machine-dependent default.
4204 Enabled at levels @option{-O2}, @option{-O3}.
4207 @itemx -falign-jumps=@var{n}
4208 @opindex falign-jumps
4209 Align branch targets to a power-of-two boundary, for branch targets
4210 where the targets can only be reached by jumping, skipping up to @var{n}
4211 bytes like @option{-falign-functions}. In this case, no dummy operations
4214 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4215 equivalent and mean that loops will not be aligned.
4217 If @var{n} is not specified or is zero, use a machine-dependent default.
4219 Enabled at levels @option{-O2}, @option{-O3}.
4221 @item -frename-registers
4222 @opindex frename-registers
4223 Attempt to avoid false dependencies in scheduled code by making use
4224 of registers left over after register allocation. This optimization
4225 will most benefit processors with lots of registers. It can, however,
4226 make debugging impossible, since variables will no longer stay in
4227 a ``home register''.
4229 Enabled at levels @option{-O3}.
4231 @item -fno-cprop-registers
4232 @opindex fno-cprop-registers
4233 After register allocation and post-register allocation instruction splitting,
4234 we perform a copy-propagation pass to try to reduce scheduling dependencies
4235 and occasionally eliminate the copy.
4237 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4241 The following options control compiler behavior regarding floating
4242 point arithmetic. These options trade off between speed and
4243 correctness. All must be specifically enabled.
4247 @opindex ffloat-store
4248 Do not store floating point variables in registers, and inhibit other
4249 options that might change whether a floating point value is taken from a
4252 @cindex floating point precision
4253 This option prevents undesirable excess precision on machines such as
4254 the 68000 where the floating registers (of the 68881) keep more
4255 precision than a @code{double} is supposed to have. Similarly for the
4256 x86 architecture. For most programs, the excess precision does only
4257 good, but a few programs rely on the precise definition of IEEE floating
4258 point. Use @option{-ffloat-store} for such programs, after modifying
4259 them to store all pertinent intermediate computations into variables.
4263 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4264 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4265 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4267 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4269 This option should never be turned on by any @option{-O} option since
4270 it can result in incorrect output for programs which depend on
4271 an exact implementation of IEEE or ISO rules/specifications for
4274 @item -fno-math-errno
4275 @opindex fno-math-errno
4276 Do not set ERRNO after calling math functions that are executed
4277 with a single instruction, e.g., sqrt. A program that relies on
4278 IEEE exceptions for math error handling may want to use this flag
4279 for speed while maintaining IEEE arithmetic compatibility.
4281 This option should never be turned on by any @option{-O} option since
4282 it can result in incorrect output for programs which depend on
4283 an exact implementation of IEEE or ISO rules/specifications for
4286 The default is @option{-fmath-errno}.
4288 @item -funsafe-math-optimizations
4289 @opindex funsafe-math-optimizations
4290 Allow optimizations for floating-point arithmetic that (a) assume
4291 that arguments and results are valid and (b) may violate IEEE or
4292 ANSI standards. When used at link-time, it may include libraries
4293 or startup files that change the default FPU control word or other
4294 similar optimizations.
4296 This option should never be turned on by any @option{-O} option since
4297 it can result in incorrect output for programs which depend on
4298 an exact implementation of IEEE or ISO rules/specifications for
4301 The default is @option{-fno-unsafe-math-optimizations}.
4303 @item -ffinite-math-only
4304 @opindex ffinite-math-only
4305 Allow optimizations for floating-point arithmetic that assume
4306 that arguments and results are not NaNs or +-Infs.
4308 This option should never be turned on by any @option{-O} option since
4309 it can result in incorrect output for programs which depend on
4310 an exact implementation of IEEE or ISO rules/specifications.
4312 The default is @option{-fno-finite-math-only}.
4314 @item -fno-trapping-math
4315 @opindex fno-trapping-math
4316 Compile code assuming that floating-point operations cannot generate
4317 user-visible traps. These traps include division by zero, overflow,
4318 underflow, inexact result and invalid operation. This option implies
4319 @option{-fno-signaling-nans}. Setting this option may allow faster
4320 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4322 This option should never be turned on by any @option{-O} option since
4323 it can result in incorrect output for programs which depend on
4324 an exact implementation of IEEE or ISO rules/specifications for
4327 The default is @option{-ftrapping-math}.
4329 @item -frounding-math
4330 @opindex frounding-math
4331 Disable transformations and optimizations that assume default floating
4332 point rounding behavior. This is round-to-zero for all floating point
4333 to integer conversions, and round-to-nearest for all other arithmetic
4334 truncations. This option should be specified for programs that change
4335 the FP rounding mode dynamically, or that may be executed with a
4336 non-default rounding mode. This option disables constant folding of
4337 floating point expressions at compile-time (which may be affected by
4338 rounding mode) and arithmetic transformations that are unsafe in the
4339 presence of sign-dependent rounding modes.
4341 The default is @option{-fno-rounding-math}.
4343 This option is experimental and does not currently guarantee to
4344 disable all GCC optimizations that are affected by rounding mode.
4345 Future versions of gcc may provide finer control of this setting
4346 using C99's @code{FENV_ACCESS} pragma. This command line option
4347 will be used to specify the default state for @code{FENV_ACCESS}.
4349 @item -fsignaling-nans
4350 @opindex fsignaling-nans
4351 Compile code assuming that IEEE signaling NaNs may generate user-visible
4352 traps during floating-point operations. Setting this option disables
4353 optimizations that may change the number of exceptions visible with
4354 signaling NaNs. This option implies @option{-ftrapping-math}.
4356 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4359 The default is @option{-fno-signaling-nans}.
4361 This option is experimental and does not currently guarantee to
4362 disable all GCC optimizations that affect signaling NaN behavior.
4364 @item -fsingle-precision-constant
4365 @opindex fsingle-precision-constant
4366 Treat floating point constant as single precision constant instead of
4367 implicitly converting it to double precision constant.
4372 The following options control optimizations that may improve
4373 performance, but are not enabled by any @option{-O} options. This
4374 section includes experimental options that may produce broken code.
4377 @item -fbranch-probabilities
4378 @opindex fbranch-probabilities
4379 After running a program compiled with @option{-fprofile-arcs}
4380 (@pxref{Debugging Options,, Options for Debugging Your Program or
4381 @command{gcc}}), you can compile it a second time using
4382 @option{-fbranch-probabilities}, to improve optimizations based on
4383 the number of times each branch was taken. When the program
4384 compiled with @option{-fprofile-arcs} exits it saves arc execution
4385 counts to a file called @file{@var{sourcename}.gcda} for each source
4386 file The information in this data file is very dependent on the
4387 structure of the generated code, so you must use the same source code
4388 and the same optimization options for both compilations.
4390 With @option{-fbranch-probabilities}, GCC puts a
4391 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4392 These can be used to improve optimization. Currently, they are only
4393 used in one place: in @file{reorg.c}, instead of guessing which path a
4394 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4395 exactly determine which path is taken more often.
4397 @item -fprofile-values
4398 @opindex fprofile-values
4399 If combined with @option{-fprofile-arcs}, it adds code so that some
4400 data about values of expressions in the program is gathered.
4402 With @option{-fbranch-probabilities}, it reads back the data gathered
4403 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4404 notes to instructions for their later usage in optimizations.
4408 Use a graph coloring register allocator. Currently this option is meant
4409 for testing, so we are interested to hear about miscompilations with
4414 Perform tail duplication to enlarge superblock size. This transformation
4415 simplifies the control flow of the function allowing other optimizations to do
4418 @item -funit-at-a-time
4419 @opindex funit-at-a-time
4420 Parse the whole compilation unit before starting to produce code.
4421 This allows some extra optimizations to take place but consumes more
4424 @item -funroll-loops
4425 @opindex funroll-loops
4426 Unroll loops whose number of iterations can be determined at compile time or
4427 upon entry to the loop. @option{-funroll-loops} implies
4428 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4429 (i.e. complete removal of loops with small constant number of iterations).
4430 This option makes code larger, and may or may not make it run faster.
4432 @item -funroll-all-loops
4433 @opindex funroll-all-loops
4434 Unroll all loops, even if their number of iterations is uncertain when
4435 the loop is entered. This usually makes programs run more slowly.
4436 @option{-funroll-all-loops} implies the same options as
4437 @option{-funroll-loops}.
4440 @opindex fpeel-loops
4441 Peels the loops for that there is enough information that they do not
4442 roll much (from profile feedback). It also turns on complete loop peeling
4443 (i.e. complete removal of loops with small constant number of iterations).
4445 @item -funswitch-loops
4446 @opindex funswitch-loops
4447 Move branches with loop invariant conditions out of the loop, with duplicates
4448 of the loop on both branches (modified according to result of the condition).
4450 @item -fold-unroll-loops
4451 @opindex fold-unroll-loops
4452 Unroll loops whose number of iterations can be determined at compile
4453 time or upon entry to the loop, using the old loop unroller whose loop
4454 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4455 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4456 option makes code larger, and may or may not make it run faster.
4458 @item -fold-unroll-all-loops
4459 @opindex fold-unroll-all-loops
4460 Unroll all loops, even if their number of iterations is uncertain when
4461 the loop is entered. This is done using the old loop unroller whose loop
4462 recognition is based on notes from frontend. This usually makes programs run more slowly.
4463 @option{-fold-unroll-all-loops} implies the same options as
4464 @option{-fold-unroll-loops}.
4466 @item -funswitch-loops
4467 @opindex funswitch-loops
4468 Move branches with loop invariant conditions out of the loop, with duplicates
4469 of the loop on both branches (modified according to result of the condition).
4471 @item -funswitch-loops
4472 @opindex funswitch-loops
4473 Move branches with loop invariant conditions out of the loop, with duplicates
4474 of the loop on both branches (modified according to result of the condition).
4476 @item -fprefetch-loop-arrays
4477 @opindex fprefetch-loop-arrays
4478 If supported by the target machine, generate instructions to prefetch
4479 memory to improve the performance of loops that access large arrays.
4481 Disabled at level @option{-Os}.
4483 @item -ffunction-sections
4484 @itemx -fdata-sections
4485 @opindex ffunction-sections
4486 @opindex fdata-sections
4487 Place each function or data item into its own section in the output
4488 file if the target supports arbitrary sections. The name of the
4489 function or the name of the data item determines the section's name
4492 Use these options on systems where the linker can perform optimizations
4493 to improve locality of reference in the instruction space. Most systems
4494 using the ELF object format and SPARC processors running Solaris 2 have
4495 linkers with such optimizations. AIX may have these optimizations in
4498 Only use these options when there are significant benefits from doing
4499 so. When you specify these options, the assembler and linker will
4500 create larger object and executable files and will also be slower.
4501 You will not be able to use @code{gprof} on all systems if you
4502 specify this option and you may have problems with debugging if
4503 you specify both this option and @option{-g}.
4507 Perform optimizations in static single assignment form. Each function's
4508 flow graph is translated into SSA form, optimizations are performed, and
4509 the flow graph is translated back from SSA form. Users should not
4510 specify this option, since it is not yet ready for production use.
4514 Perform Sparse Conditional Constant Propagation in SSA form. Requires
4515 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
4519 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
4520 Like @option{-fssa}, this is an experimental feature.
4522 @item -fbranch-target-load-optimize
4523 @opindex fbranch-target-load-optimize
4524 Perform branch target register load optimization before prologue / epilogue
4526 The use of target registers can typically be exposed only during reload,
4527 thus hoisting loads out of loops and doing inter-block scheduling needs
4528 a separate optimization pass.
4530 @item -fbranch-target-load-optimize2
4531 @opindex fbranch-target-load-optimize2
4532 Perform branch target register load optimization after prologue / epilogue
4538 @item --param @var{name}=@var{value}
4540 In some places, GCC uses various constants to control the amount of
4541 optimization that is done. For example, GCC will not inline functions
4542 that contain more that a certain number of instructions. You can
4543 control some of these constants on the command-line using the
4544 @option{--param} option.
4546 In each case, the @var{value} is an integer. The allowable choices for
4547 @var{name} are given in the following table:
4550 @item max-crossjump-edges
4551 The maximum number of incoming edges to consider for crossjumping.
4552 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4553 the number of edges incoming to each block. Increasing values mean
4554 more aggressive optimization, making the compile time increase with
4555 probably small improvement in executable size.
4557 @item max-delay-slot-insn-search
4558 The maximum number of instructions to consider when looking for an
4559 instruction to fill a delay slot. If more than this arbitrary number of
4560 instructions is searched, the time savings from filling the delay slot
4561 will be minimal so stop searching. Increasing values mean more
4562 aggressive optimization, making the compile time increase with probably
4563 small improvement in executable run time.
4565 @item max-delay-slot-live-search
4566 When trying to fill delay slots, the maximum number of instructions to
4567 consider when searching for a block with valid live register
4568 information. Increasing this arbitrarily chosen value means more
4569 aggressive optimization, increasing the compile time. This parameter
4570 should be removed when the delay slot code is rewritten to maintain the
4573 @item max-gcse-memory
4574 The approximate maximum amount of memory that will be allocated in
4575 order to perform the global common subexpression elimination
4576 optimization. If more memory than specified is required, the
4577 optimization will not be done.
4579 @item max-gcse-passes
4580 The maximum number of passes of GCSE to run.
4582 @item max-pending-list-length
4583 The maximum number of pending dependencies scheduling will allow
4584 before flushing the current state and starting over. Large functions
4585 with few branches or calls can create excessively large lists which
4586 needlessly consume memory and resources.
4588 @item max-inline-insns-single
4589 Several parameters control the tree inliner used in gcc.
4590 This number sets the maximum number of instructions (counted in gcc's
4591 internal representation) in a single function that the tree inliner
4592 will consider for inlining. This only affects functions declared
4593 inline and methods implemented in a class declaration (C++).
4594 The default value is 500.
4596 @item max-inline-insns-auto
4597 When you use @option{-finline-functions} (included in @option{-O3}),
4598 a lot of functions that would otherwise not be considered for inlining
4599 by the compiler will be investigated. To those functions, a different
4600 (more restrictive) limit compared to functions declared inline can
4602 The default value is 150.
4604 @item max-inline-insns
4605 The tree inliner does decrease the allowable size for single functions
4606 to be inlined after we already inlined the number of instructions
4607 given here by repeated inlining. This number should be a factor of
4608 two or more larger than the single function limit.
4609 Higher numbers result in better runtime performance, but incur higher
4610 compile-time resource (CPU time, memory) requirements and result in
4611 larger binaries. Very high values are not advisable, as too large
4612 binaries may adversely affect runtime performance.
4613 The default value is 200.
4615 @item max-inline-slope
4616 After exceeding the maximum number of inlined instructions by repeated
4617 inlining, a linear function is used to decrease the allowable size
4618 for single functions. The slope of that function is the negative
4619 reciprocal of the number specified here.
4620 This parameter is ignored when @option{-funit-at-a-time} is used.
4621 The default value is 32.
4623 @item min-inline-insns
4624 The repeated inlining is throttled more and more by the linear function
4625 after exceeding the limit. To avoid too much throttling, a minimum for
4626 this function is specified here to allow repeated inlining for very small
4627 functions even when a lot of repeated inlining already has been done.
4628 This parameter is ignored when @option{-funit-at-a-time} is used.
4629 The default value is 10.
4631 @item large-function-insns
4632 The limit specifying really large functions. For functions greater than this
4633 limit inlining is constrained by @option{--param large-function-growth}.
4634 This parameter is usefull primarily to avoid extreme compilation time caused by non-linear
4635 algorithms used by the backend.
4636 This parameter is ignored when @option{-funit-at-a-time} is not used.
4637 The default value is 30000.
4639 @item large-function-growth
4640 Specifies maximal growth of large functtion caused by inlining in percents.
4641 This parameter is ignored when @option{-funit-at-a-time} is not used.
4642 The default value is 200.
4644 @item inline-unit-growth
4645 Specifies maximal overall growth of the compilation unit caused by inlining.
4646 This parameter is ignored when @option{-funit-at-a-time} is not used.
4647 The default value is 150.
4649 @item max-inline-insns-rtl
4650 For languages that use the RTL inliner (this happens at a later stage
4651 than tree inlining), you can set the maximum allowable size (counted
4652 in RTL instructions) for the RTL inliner with this parameter.
4653 The default value is 600.
4656 @item max-unrolled-insns
4657 The maximum number of instructions that a loop should have if that loop
4658 is unrolled, and if the loop is unrolled, it determines how many times
4659 the loop code is unrolled.
4661 @item max-average-unrolled-insns
4662 The maximum number of instructions biased by probabilities of their execution
4663 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4664 it determines how many times the loop code is unrolled.
4666 @item max-unroll-times
4667 The maximum number of unrollings of a single loop.
4669 @item max-peeled-insns
4670 The maximum number of instructions that a loop should have if that loop
4671 is peeled, and if the loop is peeled, it determines how many times
4672 the loop code is peeled.
4674 @item max-peel-times
4675 The maximum number of peelings of a single loop.
4677 @item max-completely-peeled-insns
4678 The maximum number of insns of a completely peeled loop.
4680 @item max-completely-peel-times
4681 The maximum number of iterations of a loop to be suitable for complete peeling.
4683 @item max-unswitch-insns
4684 The maximum number of insns of an unswitched loop.
4686 @item max-unswitch-level
4687 The maximum number of branches unswitched in a single loop.
4689 @item hot-bb-count-fraction
4690 Select fraction of the maximal count of repetitions of basic block in program
4691 given basic block needs to have to be considered hot.
4693 @item hot-bb-frequency-fraction
4694 Select fraction of the maximal frequency of executions of basic block in
4695 function given basic block needs to have to be considered hot
4697 @item tracer-dynamic-coverage
4698 @itemx tracer-dynamic-coverage-feedback
4700 This value is used to limit superblock formation once the given percentage of
4701 executed instructions is covered. This limits unnecessary code size
4704 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4705 feedback is available. The real profiles (as opposed to statically estimated
4706 ones) are much less balanced allowing the threshold to be larger value.
4708 @item tracer-max-code-growth
4709 Stop tail duplication once code growth has reached given percentage. This is
4710 rather hokey argument, as most of the duplicates will be eliminated later in
4711 cross jumping, so it may be set to much higher values than is the desired code
4714 @item tracer-min-branch-ratio
4716 Stop reverse growth when the reverse probability of best edge is less than this
4717 threshold (in percent).
4719 @item tracer-min-branch-ratio
4720 @itemx tracer-min-branch-ratio-feedback
4722 Stop forward growth if the best edge do have probability lower than this
4725 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4726 compilation for profile feedback and one for compilation without. The value
4727 for compilation with profile feedback needs to be more conservative (higher) in
4728 order to make tracer effective.
4730 @item max-cse-path-length
4732 Maximum number of basic blocks on path that cse considers.
4734 @item ggc-min-expand
4736 GCC uses a garbage collector to manage its own memory allocation. This
4737 parameter specifies the minimum percentage by which the garbage
4738 collector's heap should be allowed to expand between collections.
4739 Tuning this may improve compilation speed; it has no effect on code
4742 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4743 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4744 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4745 GCC is not able to calculate RAM on a particular platform, the lower
4746 bound of 30% is used. Setting this parameter and
4747 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4748 every opportunity. This is extremely slow, but can be useful for
4751 @item ggc-min-heapsize
4753 Minimum size of the garbage collector's heap before it begins bothering
4754 to collect garbage. The first collection occurs after the heap expands
4755 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4756 tuning this may improve compilation speed, and has no effect on code
4759 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4760 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4761 available, the notion of "RAM" is the smallest of actual RAM,
4762 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4763 RAM on a particular platform, the lower bound is used. Setting this
4764 parameter very large effectively disables garbage collection. Setting
4765 this parameter and @option{ggc-min-expand} to zero causes a full
4766 collection to occur at every opportunity.
4768 @item reorder-blocks-duplicate
4769 @itemx reorder-blocks-duplicate-feedback
4771 Used by basic block reordering pass to decide whether to use unconditional
4772 branch or duplicate the code on its destination. Code is duplicated when its
4773 estimated size is smaller than this value multiplied by the estimated size of
4774 unconditional jump in the hot spots of the program.
4776 The @option{reorder-block-duplicate-feedback} is used only when profile
4777 feedback is available and may be set to higher values than
4778 @option{reorder-block-duplicate} since information about the hot spots is more
4783 @node Preprocessor Options
4784 @section Options Controlling the Preprocessor
4785 @cindex preprocessor options
4786 @cindex options, preprocessor
4788 These options control the C preprocessor, which is run on each C source
4789 file before actual compilation.
4791 If you use the @option{-E} option, nothing is done except preprocessing.
4792 Some of these options make sense only together with @option{-E} because
4793 they cause the preprocessor output to be unsuitable for actual
4798 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4799 and pass @var{option} directly through to the preprocessor. If
4800 @var{option} contains commas, it is split into multiple options at the
4801 commas. However, many options are modified, translated or interpreted
4802 by the compiler driver before being passed to the preprocessor, and
4803 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4804 interface is undocumented and subject to change, so whenever possible
4805 you should avoid using @option{-Wp} and let the driver handle the
4808 @item -Xpreprocessor @var{option}
4809 @opindex preprocessor
4810 Pass @var{option} as an option to the preprocessor. You can use this to
4811 supply system-specific preprocessor options which GCC does not know how to
4814 If you want to pass an option that takes an argument, you must use
4815 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4818 @include cppopts.texi
4820 @node Assembler Options
4821 @section Passing Options to the Assembler
4823 @c prevent bad page break with this line
4824 You can pass options to the assembler.
4827 @item -Wa,@var{option}
4829 Pass @var{option} as an option to the assembler. If @var{option}
4830 contains commas, it is split into multiple options at the commas.
4832 @item -Xassembler @var{option}
4834 Pass @var{option} as an option to the assembler. You can use this to
4835 supply system-specific assembler options which GCC does not know how to
4838 If you want to pass an option that takes an argument, you must use
4839 @option{-Xassembler} twice, once for the option and once for the argument.
4844 @section Options for Linking
4845 @cindex link options
4846 @cindex options, linking
4848 These options come into play when the compiler links object files into
4849 an executable output file. They are meaningless if the compiler is
4850 not doing a link step.
4854 @item @var{object-file-name}
4855 A file name that does not end in a special recognized suffix is
4856 considered to name an object file or library. (Object files are
4857 distinguished from libraries by the linker according to the file
4858 contents.) If linking is done, these object files are used as input
4867 If any of these options is used, then the linker is not run, and
4868 object file names should not be used as arguments. @xref{Overall
4872 @item -l@var{library}
4873 @itemx -l @var{library}
4875 Search the library named @var{library} when linking. (The second
4876 alternative with the library as a separate argument is only for
4877 POSIX compliance and is not recommended.)
4879 It makes a difference where in the command you write this option; the
4880 linker searches and processes libraries and object files in the order they
4881 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4882 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4883 to functions in @samp{z}, those functions may not be loaded.
4885 The linker searches a standard list of directories for the library,
4886 which is actually a file named @file{lib@var{library}.a}. The linker
4887 then uses this file as if it had been specified precisely by name.
4889 The directories searched include several standard system directories
4890 plus any that you specify with @option{-L}.
4892 Normally the files found this way are library files---archive files
4893 whose members are object files. The linker handles an archive file by
4894 scanning through it for members which define symbols that have so far
4895 been referenced but not defined. But if the file that is found is an
4896 ordinary object file, it is linked in the usual fashion. The only
4897 difference between using an @option{-l} option and specifying a file name
4898 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4899 and searches several directories.
4903 You need this special case of the @option{-l} option in order to
4904 link an Objective-C program.
4907 @opindex nostartfiles
4908 Do not use the standard system startup files when linking.
4909 The standard system libraries are used normally, unless @option{-nostdlib}
4910 or @option{-nodefaultlibs} is used.
4912 @item -nodefaultlibs
4913 @opindex nodefaultlibs
4914 Do not use the standard system libraries when linking.
4915 Only the libraries you specify will be passed to the linker.
4916 The standard startup files are used normally, unless @option{-nostartfiles}
4917 is used. The compiler may generate calls to memcmp, memset, and memcpy
4918 for System V (and ISO C) environments or to bcopy and bzero for
4919 BSD environments. These entries are usually resolved by entries in
4920 libc. These entry points should be supplied through some other
4921 mechanism when this option is specified.
4925 Do not use the standard system startup files or libraries when linking.
4926 No startup files and only the libraries you specify will be passed to
4927 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4928 for System V (and ISO C) environments or to bcopy and bzero for
4929 BSD environments. These entries are usually resolved by entries in
4930 libc. These entry points should be supplied through some other
4931 mechanism when this option is specified.
4933 @cindex @option{-lgcc}, use with @option{-nostdlib}
4934 @cindex @option{-nostdlib} and unresolved references
4935 @cindex unresolved references and @option{-nostdlib}
4936 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4937 @cindex @option{-nodefaultlibs} and unresolved references
4938 @cindex unresolved references and @option{-nodefaultlibs}
4939 One of the standard libraries bypassed by @option{-nostdlib} and
4940 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4941 that GCC uses to overcome shortcomings of particular machines, or special
4942 needs for some languages.
4943 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
4944 Collection (GCC) Internals},
4945 for more discussion of @file{libgcc.a}.)
4946 In most cases, you need @file{libgcc.a} even when you want to avoid
4947 other standard libraries. In other words, when you specify @option{-nostdlib}
4948 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4949 This ensures that you have no unresolved references to internal GCC
4950 library subroutines. (For example, @samp{__main}, used to ensure C++
4951 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
4952 GNU Compiler Collection (GCC) Internals}.)
4956 Produce a position independent executable on targets which support it.
4957 For predictable results, you must also specify the same set of options
4958 that were used to generate code (@option{-fpie}, @option{-fPIE},
4959 or model suboptions) when you specify this option.
4963 Remove all symbol table and relocation information from the executable.
4967 On systems that support dynamic linking, this prevents linking with the shared
4968 libraries. On other systems, this option has no effect.
4972 Produce a shared object which can then be linked with other objects to
4973 form an executable. Not all systems support this option. For predictable
4974 results, you must also specify the same set of options that were used to
4975 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4976 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4977 needs to build supplementary stub code for constructors to work. On
4978 multi-libbed systems, @samp{gcc -shared} must select the correct support
4979 libraries to link against. Failing to supply the correct flags may lead
4980 to subtle defects. Supplying them in cases where they are not necessary
4983 @item -shared-libgcc
4984 @itemx -static-libgcc
4985 @opindex shared-libgcc
4986 @opindex static-libgcc
4987 On systems that provide @file{libgcc} as a shared library, these options
4988 force the use of either the shared or static version respectively.
4989 If no shared version of @file{libgcc} was built when the compiler was
4990 configured, these options have no effect.
4992 There are several situations in which an application should use the
4993 shared @file{libgcc} instead of the static version. The most common
4994 of these is when the application wishes to throw and catch exceptions
4995 across different shared libraries. In that case, each of the libraries
4996 as well as the application itself should use the shared @file{libgcc}.
4998 Therefore, the G++ and GCJ drivers automatically add
4999 @option{-shared-libgcc} whenever you build a shared library or a main
5000 executable, because C++ and Java programs typically use exceptions, so
5001 this is the right thing to do.
5003 If, instead, you use the GCC driver to create shared libraries, you may
5004 find that they will not always be linked with the shared @file{libgcc}.
5005 If GCC finds, at its configuration time, that you have a GNU linker that
5006 does not support option @option{--eh-frame-hdr}, it will link the shared
5007 version of @file{libgcc} into shared libraries by default. Otherwise,
5008 it will take advantage of the linker and optimize away the linking with
5009 the shared version of @file{libgcc}, linking with the static version of
5010 libgcc by default. This allows exceptions to propagate through such
5011 shared libraries, without incurring relocation costs at library load
5014 However, if a library or main executable is supposed to throw or catch
5015 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5016 for the languages used in the program, or using the option
5017 @option{-shared-libgcc}, such that it is linked with the shared
5022 Bind references to global symbols when building a shared object. Warn
5023 about any unresolved references (unless overridden by the link editor
5024 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5027 @item -Xlinker @var{option}
5029 Pass @var{option} as an option to the linker. You can use this to
5030 supply system-specific linker options which GCC does not know how to
5033 If you want to pass an option that takes an argument, you must use
5034 @option{-Xlinker} twice, once for the option and once for the argument.
5035 For example, to pass @option{-assert definitions}, you must write
5036 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5037 @option{-Xlinker "-assert definitions"}, because this passes the entire
5038 string as a single argument, which is not what the linker expects.
5040 @item -Wl,@var{option}
5042 Pass @var{option} as an option to the linker. If @var{option} contains
5043 commas, it is split into multiple options at the commas.
5045 @item -u @var{symbol}
5047 Pretend the symbol @var{symbol} is undefined, to force linking of
5048 library modules to define it. You can use @option{-u} multiple times with
5049 different symbols to force loading of additional library modules.
5052 @node Directory Options
5053 @section Options for Directory Search
5054 @cindex directory options
5055 @cindex options, directory search
5058 These options specify directories to search for header files, for
5059 libraries and for parts of the compiler:
5064 Add the directory @var{dir} to the head of the list of directories to be
5065 searched for header files. This can be used to override a system header
5066 file, substituting your own version, since these directories are
5067 searched before the system header file directories. However, you should
5068 not use this option to add directories that contain vendor-supplied
5069 system header files (use @option{-isystem} for that). If you use more than
5070 one @option{-I} option, the directories are scanned in left-to-right
5071 order; the standard system directories come after.
5073 If a standard system include directory, or a directory specified with
5074 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5075 option will be ignored. The directory will still be searched but as a
5076 system directory at its normal position in the system include chain.
5077 This is to ensure that GCC's procedure to fix buggy system headers and
5078 the ordering for the include_next directive are not inadvertently changed.
5079 If you really need to change the search order for system directories,
5080 use the @option{-nostdinc} and/or @option{-isystem} options.
5084 Any directories you specify with @option{-I} options before the @option{-I-}
5085 option are searched only for the case of @samp{#include "@var{file}"};
5086 they are not searched for @samp{#include <@var{file}>}.
5088 If additional directories are specified with @option{-I} options after
5089 the @option{-I-}, these directories are searched for all @samp{#include}
5090 directives. (Ordinarily @emph{all} @option{-I} directories are used
5093 In addition, the @option{-I-} option inhibits the use of the current
5094 directory (where the current input file came from) as the first search
5095 directory for @samp{#include "@var{file}"}. There is no way to
5096 override this effect of @option{-I-}. With @option{-I.} you can specify
5097 searching the directory which was current when the compiler was
5098 invoked. That is not exactly the same as what the preprocessor does
5099 by default, but it is often satisfactory.
5101 @option{-I-} does not inhibit the use of the standard system directories
5102 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5107 Add directory @var{dir} to the list of directories to be searched
5110 @item -B@var{prefix}
5112 This option specifies where to find the executables, libraries,
5113 include files, and data files of the compiler itself.
5115 The compiler driver program runs one or more of the subprograms
5116 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5117 @var{prefix} as a prefix for each program it tries to run, both with and
5118 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5120 For each subprogram to be run, the compiler driver first tries the
5121 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5122 was not specified, the driver tries two standard prefixes, which are
5123 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5124 those results in a file name that is found, the unmodified program
5125 name is searched for using the directories specified in your
5126 @env{PATH} environment variable.
5128 The compiler will check to see if the path provided by the @option{-B}
5129 refers to a directory, and if necessary it will add a directory
5130 separator character at the end of the path.
5132 @option{-B} prefixes that effectively specify directory names also apply
5133 to libraries in the linker, because the compiler translates these
5134 options into @option{-L} options for the linker. They also apply to
5135 includes files in the preprocessor, because the compiler translates these
5136 options into @option{-isystem} options for the preprocessor. In this case,
5137 the compiler appends @samp{include} to the prefix.
5139 The run-time support file @file{libgcc.a} can also be searched for using
5140 the @option{-B} prefix, if needed. If it is not found there, the two
5141 standard prefixes above are tried, and that is all. The file is left
5142 out of the link if it is not found by those means.
5144 Another way to specify a prefix much like the @option{-B} prefix is to use
5145 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5148 As a special kludge, if the path provided by @option{-B} is
5149 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5150 9, then it will be replaced by @file{[dir/]include}. This is to help
5151 with boot-strapping the compiler.
5153 @item -specs=@var{file}
5155 Process @var{file} after the compiler reads in the standard @file{specs}
5156 file, in order to override the defaults that the @file{gcc} driver
5157 program uses when determining what switches to pass to @file{cc1},
5158 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5159 @option{-specs=@var{file}} can be specified on the command line, and they
5160 are processed in order, from left to right.
5166 @section Specifying subprocesses and the switches to pass to them
5169 @command{gcc} is a driver program. It performs its job by invoking a
5170 sequence of other programs to do the work of compiling, assembling and
5171 linking. GCC interprets its command-line parameters and uses these to
5172 deduce which programs it should invoke, and which command-line options
5173 it ought to place on their command lines. This behavior is controlled
5174 by @dfn{spec strings}. In most cases there is one spec string for each
5175 program that GCC can invoke, but a few programs have multiple spec
5176 strings to control their behavior. The spec strings built into GCC can
5177 be overridden by using the @option{-specs=} command-line switch to specify
5180 @dfn{Spec files} are plaintext files that are used to construct spec
5181 strings. They consist of a sequence of directives separated by blank
5182 lines. The type of directive is determined by the first non-whitespace
5183 character on the line and it can be one of the following:
5186 @item %@var{command}
5187 Issues a @var{command} to the spec file processor. The commands that can
5191 @item %include <@var{file}>
5193 Search for @var{file} and insert its text at the current point in the
5196 @item %include_noerr <@var{file}>
5197 @cindex %include_noerr
5198 Just like @samp{%include}, but do not generate an error message if the include
5199 file cannot be found.
5201 @item %rename @var{old_name} @var{new_name}
5203 Rename the spec string @var{old_name} to @var{new_name}.
5207 @item *[@var{spec_name}]:
5208 This tells the compiler to create, override or delete the named spec
5209 string. All lines after this directive up to the next directive or
5210 blank line are considered to be the text for the spec string. If this
5211 results in an empty string then the spec will be deleted. (Or, if the
5212 spec did not exist, then nothing will happened.) Otherwise, if the spec
5213 does not currently exist a new spec will be created. If the spec does
5214 exist then its contents will be overridden by the text of this
5215 directive, unless the first character of that text is the @samp{+}
5216 character, in which case the text will be appended to the spec.
5218 @item [@var{suffix}]:
5219 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5220 and up to the next directive or blank line are considered to make up the
5221 spec string for the indicated suffix. When the compiler encounters an
5222 input file with the named suffix, it will processes the spec string in
5223 order to work out how to compile that file. For example:
5230 This says that any input file whose name ends in @samp{.ZZ} should be
5231 passed to the program @samp{z-compile}, which should be invoked with the
5232 command-line switch @option{-input} and with the result of performing the
5233 @samp{%i} substitution. (See below.)
5235 As an alternative to providing a spec string, the text that follows a
5236 suffix directive can be one of the following:
5239 @item @@@var{language}
5240 This says that the suffix is an alias for a known @var{language}. This is
5241 similar to using the @option{-x} command-line switch to GCC to specify a
5242 language explicitly. For example:
5249 Says that .ZZ files are, in fact, C++ source files.
5252 This causes an error messages saying:
5255 @var{name} compiler not installed on this system.
5259 GCC already has an extensive list of suffixes built into it.
5260 This directive will add an entry to the end of the list of suffixes, but
5261 since the list is searched from the end backwards, it is effectively
5262 possible to override earlier entries using this technique.
5266 GCC has the following spec strings built into it. Spec files can
5267 override these strings or create their own. Note that individual
5268 targets can also add their own spec strings to this list.
5271 asm Options to pass to the assembler
5272 asm_final Options to pass to the assembler post-processor
5273 cpp Options to pass to the C preprocessor
5274 cc1 Options to pass to the C compiler
5275 cc1plus Options to pass to the C++ compiler
5276 endfile Object files to include at the end of the link
5277 link Options to pass to the linker
5278 lib Libraries to include on the command line to the linker
5279 libgcc Decides which GCC support library to pass to the linker
5280 linker Sets the name of the linker
5281 predefines Defines to be passed to the C preprocessor
5282 signed_char Defines to pass to CPP to say whether @code{char} is signed
5284 startfile Object files to include at the start of the link
5287 Here is a small example of a spec file:
5293 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5296 This example renames the spec called @samp{lib} to @samp{old_lib} and
5297 then overrides the previous definition of @samp{lib} with a new one.
5298 The new definition adds in some extra command-line options before
5299 including the text of the old definition.
5301 @dfn{Spec strings} are a list of command-line options to be passed to their
5302 corresponding program. In addition, the spec strings can contain
5303 @samp{%}-prefixed sequences to substitute variable text or to
5304 conditionally insert text into the command line. Using these constructs
5305 it is possible to generate quite complex command lines.
5307 Here is a table of all defined @samp{%}-sequences for spec
5308 strings. Note that spaces are not generated automatically around the
5309 results of expanding these sequences. Therefore you can concatenate them
5310 together or combine them with constant text in a single argument.
5314 Substitute one @samp{%} into the program name or argument.
5317 Substitute the name of the input file being processed.
5320 Substitute the basename of the input file being processed.
5321 This is the substring up to (and not including) the last period
5322 and not including the directory.
5325 This is the same as @samp{%b}, but include the file suffix (text after
5329 Marks the argument containing or following the @samp{%d} as a
5330 temporary file name, so that that file will be deleted if GCC exits
5331 successfully. Unlike @samp{%g}, this contributes no text to the
5334 @item %g@var{suffix}
5335 Substitute a file name that has suffix @var{suffix} and is chosen
5336 once per compilation, and mark the argument in the same way as
5337 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5338 name is now chosen in a way that is hard to predict even when previously
5339 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5340 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5341 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5342 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5343 was simply substituted with a file name chosen once per compilation,
5344 without regard to any appended suffix (which was therefore treated
5345 just like ordinary text), making such attacks more likely to succeed.
5347 @item %u@var{suffix}
5348 Like @samp{%g}, but generates a new temporary file name even if
5349 @samp{%u@var{suffix}} was already seen.
5351 @item %U@var{suffix}
5352 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5353 new one if there is no such last file name. In the absence of any
5354 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5355 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5356 would involve the generation of two distinct file names, one
5357 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5358 simply substituted with a file name chosen for the previous @samp{%u},
5359 without regard to any appended suffix.
5361 @item %j@var{suffix}
5362 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5363 writable, and if save-temps is off; otherwise, substitute the name
5364 of a temporary file, just like @samp{%u}. This temporary file is not
5365 meant for communication between processes, but rather as a junk
5368 @item %|@var{suffix}
5369 @itemx %m@var{suffix}
5370 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5371 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5372 all. These are the two most common ways to instruct a program that it
5373 should read from standard input or write to standard output. If you
5374 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5375 construct: see for example @file{f/lang-specs.h}.
5377 @item %.@var{SUFFIX}
5378 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5379 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5380 terminated by the next space or %.
5383 Marks the argument containing or following the @samp{%w} as the
5384 designated output file of this compilation. This puts the argument
5385 into the sequence of arguments that @samp{%o} will substitute later.
5388 Substitutes the names of all the output files, with spaces
5389 automatically placed around them. You should write spaces
5390 around the @samp{%o} as well or the results are undefined.
5391 @samp{%o} is for use in the specs for running the linker.
5392 Input files whose names have no recognized suffix are not compiled
5393 at all, but they are included among the output files, so they will
5397 Substitutes the suffix for object files. Note that this is
5398 handled specially when it immediately follows @samp{%g, %u, or %U},
5399 because of the need for those to form complete file names. The
5400 handling is such that @samp{%O} is treated exactly as if it had already
5401 been substituted, except that @samp{%g, %u, and %U} do not currently
5402 support additional @var{suffix} characters following @samp{%O} as they would
5403 following, for example, @samp{.o}.
5406 Substitutes the standard macro predefinitions for the
5407 current target machine. Use this when running @code{cpp}.
5410 Like @samp{%p}, but puts @samp{__} before and after the name of each
5411 predefined macro, except for macros that start with @samp{__} or with
5412 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5416 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5417 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5418 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5422 Current argument is the name of a library or startup file of some sort.
5423 Search for that file in a standard list of directories and substitute
5424 the full name found.
5427 Print @var{str} as an error message. @var{str} is terminated by a newline.
5428 Use this when inconsistent options are detected.
5431 Substitute the contents of spec string @var{name} at this point.
5434 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5436 @item %x@{@var{option}@}
5437 Accumulate an option for @samp{%X}.
5440 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5444 Output the accumulated assembler options specified by @option{-Wa}.
5447 Output the accumulated preprocessor options specified by @option{-Wp}.
5450 Process the @code{asm} spec. This is used to compute the
5451 switches to be passed to the assembler.
5454 Process the @code{asm_final} spec. This is a spec string for
5455 passing switches to an assembler post-processor, if such a program is
5459 Process the @code{link} spec. This is the spec for computing the
5460 command line passed to the linker. Typically it will make use of the
5461 @samp{%L %G %S %D and %E} sequences.
5464 Dump out a @option{-L} option for each directory that GCC believes might
5465 contain startup files. If the target supports multilibs then the
5466 current multilib directory will be prepended to each of these paths.
5469 Output the multilib directory with directory separators replaced with
5470 @samp{_}. If multilib directories are not set, or the multilib directory is
5471 @file{.} then this option emits nothing.
5474 Process the @code{lib} spec. This is a spec string for deciding which
5475 libraries should be included on the command line to the linker.
5478 Process the @code{libgcc} spec. This is a spec string for deciding
5479 which GCC support library should be included on the command line to the linker.
5482 Process the @code{startfile} spec. This is a spec for deciding which
5483 object files should be the first ones passed to the linker. Typically
5484 this might be a file named @file{crt0.o}.
5487 Process the @code{endfile} spec. This is a spec string that specifies
5488 the last object files that will be passed to the linker.
5491 Process the @code{cpp} spec. This is used to construct the arguments
5492 to be passed to the C preprocessor.
5495 Process the @code{signed_char} spec. This is intended to be used
5496 to tell cpp whether a char is signed. It typically has the definition:
5498 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5502 Process the @code{cc1} spec. This is used to construct the options to be
5503 passed to the actual C compiler (@samp{cc1}).
5506 Process the @code{cc1plus} spec. This is used to construct the options to be
5507 passed to the actual C++ compiler (@samp{cc1plus}).
5510 Substitute the variable part of a matched option. See below.
5511 Note that each comma in the substituted string is replaced by
5515 Remove all occurrences of @code{-S} from the command line. Note---this
5516 command is position dependent. @samp{%} commands in the spec string
5517 before this one will see @code{-S}, @samp{%} commands in the spec string
5518 after this one will not.
5520 @item %:@var{function}(@var{args})
5521 Call the named function @var{function}, passing it @var{args}.
5522 @var{args} is first processed as a nested spec string, then split
5523 into an argument vector in the usual fashion. The function returns
5524 a string which is processed as if it had appeared literally as part
5525 of the current spec.
5527 The following built-in spec functions are provided:
5530 @item @code{if-exists}
5531 The @code{if-exists} spec function takes one argument, an absolute
5532 pathname to a file. If the file exists, @code{if-exists} returns the
5533 pathname. Here is a small example of its usage:
5537 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5540 @item @code{if-exists-else}
5541 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5542 spec function, except that it takes two arguments. The first argument is
5543 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5544 returns the pathname. If it does not exist, it returns the second argument.
5545 This way, @code{if-exists-else} can be used to select one file or another,
5546 based on the existence of the first. Here is a small example of its usage:
5550 crt0%O%s %:if-exists(crti%O%s) \
5551 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5556 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5557 If that switch was not specified, this substitutes nothing. Note that
5558 the leading dash is omitted when specifying this option, and it is
5559 automatically inserted if the substitution is performed. Thus the spec
5560 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5561 and would output the command line option @option{-foo}.
5563 @item %W@{@code{S}@}
5564 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5567 @item %@{@code{S}*@}
5568 Substitutes all the switches specified to GCC whose names start
5569 with @code{-S}, but which also take an argument. This is used for
5570 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5571 GCC considers @option{-o foo} as being
5572 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5573 text, including the space. Thus two arguments would be generated.
5575 @item %@{@code{S}*&@code{T}*@}
5576 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5577 (the order of @code{S} and @code{T} in the spec is not significant).
5578 There can be any number of ampersand-separated variables; for each the
5579 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5581 @item %@{@code{S}:@code{X}@}
5582 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5584 @item %@{!@code{S}:@code{X}@}
5585 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5587 @item %@{@code{S}*:@code{X}@}
5588 Substitutes @code{X} if one or more switches whose names start with
5589 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5590 once, no matter how many such switches appeared. However, if @code{%*}
5591 appears somewhere in @code{X}, then @code{X} will be substituted once
5592 for each matching switch, with the @code{%*} replaced by the part of
5593 that switch that matched the @code{*}.
5595 @item %@{.@code{S}:@code{X}@}
5596 Substitutes @code{X}, if processing a file with suffix @code{S}.
5598 @item %@{!.@code{S}:@code{X}@}
5599 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5601 @item %@{@code{S}|@code{P}:@code{X}@}
5602 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5603 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5604 although they have a stronger binding than the @samp{|}. If @code{%*}
5605 appears in @code{X}, all of the alternatives must be starred, and only
5606 the first matching alternative is substituted.
5608 For example, a spec string like this:
5611 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5614 will output the following command-line options from the following input
5615 command-line options:
5620 -d fred.c -foo -baz -boggle
5621 -d jim.d -bar -baz -boggle
5624 @item %@{S:X; T:Y; :D@}
5626 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5627 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5628 be as many clauses as you need. This may be combined with @code{.},
5629 @code{!}, @code{|}, and @code{*} as needed.
5634 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5635 construct may contain other nested @samp{%} constructs or spaces, or
5636 even newlines. They are processed as usual, as described above.
5637 Trailing white space in @code{X} is ignored. White space may also
5638 appear anywhere on the left side of the colon in these constructs,
5639 except between @code{.} or @code{*} and the corresponding word.
5641 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5642 handled specifically in these constructs. If another value of
5643 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5644 @option{-W} switch is found later in the command line, the earlier
5645 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5646 just one letter, which passes all matching options.
5648 The character @samp{|} at the beginning of the predicate text is used to
5649 indicate that a command should be piped to the following command, but
5650 only if @option{-pipe} is specified.
5652 It is built into GCC which switches take arguments and which do not.
5653 (You might think it would be useful to generalize this to allow each
5654 compiler's spec to say which switches take arguments. But this cannot
5655 be done in a consistent fashion. GCC cannot even decide which input
5656 files have been specified without knowing which switches take arguments,
5657 and it must know which input files to compile in order to tell which
5660 GCC also knows implicitly that arguments starting in @option{-l} are to be
5661 treated as compiler output files, and passed to the linker in their
5662 proper position among the other output files.
5664 @c man begin OPTIONS
5666 @node Target Options
5667 @section Specifying Target Machine and Compiler Version
5668 @cindex target options
5669 @cindex cross compiling
5670 @cindex specifying machine version
5671 @cindex specifying compiler version and target machine
5672 @cindex compiler version, specifying
5673 @cindex target machine, specifying
5675 The usual way to run GCC is to run the executable called @file{gcc}, or
5676 @file{<machine>-gcc} when cross-compiling, or
5677 @file{<machine>-gcc-<version>} to run a version other than the one that
5678 was installed last. Sometimes this is inconvenient, so GCC provides
5679 options that will switch to another cross-compiler or version.
5682 @item -b @var{machine}
5684 The argument @var{machine} specifies the target machine for compilation.
5686 The value to use for @var{machine} is the same as was specified as the
5687 machine type when configuring GCC as a cross-compiler. For
5688 example, if a cross-compiler was configured with @samp{configure
5689 i386v}, meaning to compile for an 80386 running System V, then you
5690 would specify @option{-b i386v} to run that cross compiler.
5692 @item -V @var{version}
5694 The argument @var{version} specifies which version of GCC to run.
5695 This is useful when multiple versions are installed. For example,
5696 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5699 The @option{-V} and @option{-b} options work by running the
5700 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5701 use them if you can just run that directly.
5703 @node Submodel Options
5704 @section Hardware Models and Configurations
5705 @cindex submodel options
5706 @cindex specifying hardware config
5707 @cindex hardware models and configurations, specifying
5708 @cindex machine dependent options
5710 Earlier we discussed the standard option @option{-b} which chooses among
5711 different installed compilers for completely different target
5712 machines, such as VAX vs.@: 68000 vs.@: 80386.
5714 In addition, each of these target machine types can have its own
5715 special options, starting with @samp{-m}, to choose among various
5716 hardware models or configurations---for example, 68010 vs 68020,
5717 floating coprocessor or none. A single installed version of the
5718 compiler can compile for any model or configuration, according to the
5721 Some configurations of the compiler also support additional special
5722 options, usually for compatibility with other compilers on the same
5725 These options are defined by the macro @code{TARGET_SWITCHES} in the
5726 machine description. The default for the options is also defined by
5727 that macro, which enables you to change the defaults.
5739 * RS/6000 and PowerPC Options::
5743 * i386 and x86-64 Options::
5745 * Intel 960 Options::
5746 * DEC Alpha Options::
5747 * DEC Alpha/VMS Options::
5750 * System V Options::
5751 * TMS320C3x/C4x Options::
5759 * S/390 and zSeries Options::
5763 * Xstormy16 Options::
5768 @node M680x0 Options
5769 @subsection M680x0 Options
5770 @cindex M680x0 options
5772 These are the @samp{-m} options defined for the 68000 series. The default
5773 values for these options depends on which style of 68000 was selected when
5774 the compiler was configured; the defaults for the most common choices are
5782 Generate output for a 68000. This is the default
5783 when the compiler is configured for 68000-based systems.
5785 Use this option for microcontrollers with a 68000 or EC000 core,
5786 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5792 Generate output for a 68020. This is the default
5793 when the compiler is configured for 68020-based systems.
5797 Generate output containing 68881 instructions for floating point.
5798 This is the default for most 68020 systems unless @option{--nfp} was
5799 specified when the compiler was configured.
5803 Generate output for a 68030. This is the default when the compiler is
5804 configured for 68030-based systems.
5808 Generate output for a 68040. This is the default when the compiler is
5809 configured for 68040-based systems.
5811 This option inhibits the use of 68881/68882 instructions that have to be
5812 emulated by software on the 68040. Use this option if your 68040 does not
5813 have code to emulate those instructions.
5817 Generate output for a 68060. This is the default when the compiler is
5818 configured for 68060-based systems.
5820 This option inhibits the use of 68020 and 68881/68882 instructions that
5821 have to be emulated by software on the 68060. Use this option if your 68060
5822 does not have code to emulate those instructions.
5826 Generate output for a CPU32. This is the default
5827 when the compiler is configured for CPU32-based systems.
5829 Use this option for microcontrollers with a
5830 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5831 68336, 68340, 68341, 68349 and 68360.
5835 Generate output for a 520X ``coldfire'' family cpu. This is the default
5836 when the compiler is configured for 520X-based systems.
5838 Use this option for microcontroller with a 5200 core, including
5839 the MCF5202, MCF5203, MCF5204 and MCF5202.
5844 Generate output for a 68040, without using any of the new instructions.
5845 This results in code which can run relatively efficiently on either a
5846 68020/68881 or a 68030 or a 68040. The generated code does use the
5847 68881 instructions that are emulated on the 68040.
5851 Generate output for a 68060, without using any of the new instructions.
5852 This results in code which can run relatively efficiently on either a
5853 68020/68881 or a 68030 or a 68040. The generated code does use the
5854 68881 instructions that are emulated on the 68060.
5857 @opindex msoft-float
5858 Generate output containing library calls for floating point.
5859 @strong{Warning:} the requisite libraries are not available for all m68k
5860 targets. Normally the facilities of the machine's usual C compiler are
5861 used, but this can't be done directly in cross-compilation. You must
5862 make your own arrangements to provide suitable library functions for
5863 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5864 @samp{m68k-*-coff} do provide software floating point support.
5868 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5871 @opindex mnobitfield
5872 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5873 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5877 Do use the bit-field instructions. The @option{-m68020} option implies
5878 @option{-mbitfield}. This is the default if you use a configuration
5879 designed for a 68020.
5883 Use a different function-calling convention, in which functions
5884 that take a fixed number of arguments return with the @code{rtd}
5885 instruction, which pops their arguments while returning. This
5886 saves one instruction in the caller since there is no need to pop
5887 the arguments there.
5889 This calling convention is incompatible with the one normally
5890 used on Unix, so you cannot use it if you need to call libraries
5891 compiled with the Unix compiler.
5893 Also, you must provide function prototypes for all functions that
5894 take variable numbers of arguments (including @code{printf});
5895 otherwise incorrect code will be generated for calls to those
5898 In addition, seriously incorrect code will result if you call a
5899 function with too many arguments. (Normally, extra arguments are
5900 harmlessly ignored.)
5902 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5903 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5906 @itemx -mno-align-int
5908 @opindex mno-align-int
5909 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5910 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5911 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5912 Aligning variables on 32-bit boundaries produces code that runs somewhat
5913 faster on processors with 32-bit busses at the expense of more memory.
5915 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5916 align structures containing the above types differently than
5917 most published application binary interface specifications for the m68k.
5921 Use the pc-relative addressing mode of the 68000 directly, instead of
5922 using a global offset table. At present, this option implies @option{-fpic},
5923 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5924 not presently supported with @option{-mpcrel}, though this could be supported for
5925 68020 and higher processors.
5927 @item -mno-strict-align
5928 @itemx -mstrict-align
5929 @opindex mno-strict-align
5930 @opindex mstrict-align
5931 Do not (do) assume that unaligned memory references will be handled by
5936 @node M68hc1x Options
5937 @subsection M68hc1x Options
5938 @cindex M68hc1x options
5940 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5941 microcontrollers. The default values for these options depends on
5942 which style of microcontroller was selected when the compiler was configured;
5943 the defaults for the most common choices are given below.
5950 Generate output for a 68HC11. This is the default
5951 when the compiler is configured for 68HC11-based systems.
5957 Generate output for a 68HC12. This is the default
5958 when the compiler is configured for 68HC12-based systems.
5964 Generate output for a 68HCS12.
5967 @opindex mauto-incdec
5968 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5975 Enable the use of 68HC12 min and max instructions.
5978 @itemx -mno-long-calls
5979 @opindex mlong-calls
5980 @opindex mno-long-calls
5981 Treat all calls as being far away (near). If calls are assumed to be
5982 far away, the compiler will use the @code{call} instruction to
5983 call a function and the @code{rtc} instruction for returning.
5987 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5989 @item -msoft-reg-count=@var{count}
5990 @opindex msoft-reg-count
5991 Specify the number of pseudo-soft registers which are used for the
5992 code generation. The maximum number is 32. Using more pseudo-soft
5993 register may or may not result in better code depending on the program.
5994 The default is 4 for 68HC11 and 2 for 68HC12.
5999 @subsection VAX Options
6002 These @samp{-m} options are defined for the VAX:
6007 Do not output certain jump instructions (@code{aobleq} and so on)
6008 that the Unix assembler for the VAX cannot handle across long
6013 Do output those jump instructions, on the assumption that you
6014 will assemble with the GNU assembler.
6018 Output code for g-format floating point numbers instead of d-format.
6022 @subsection SPARC Options
6023 @cindex SPARC options
6025 These @samp{-m} switches are supported on the SPARC:
6030 @opindex mno-app-regs
6032 Specify @option{-mapp-regs} to generate output using the global registers
6033 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6036 To be fully SVR4 ABI compliant at the cost of some performance loss,
6037 specify @option{-mno-app-regs}. You should compile libraries and system
6038 software with this option.
6043 @opindex mhard-float
6044 Generate output containing floating point instructions. This is the
6050 @opindex msoft-float
6051 Generate output containing library calls for floating point.
6052 @strong{Warning:} the requisite libraries are not available for all SPARC
6053 targets. Normally the facilities of the machine's usual C compiler are
6054 used, but this cannot be done directly in cross-compilation. You must make
6055 your own arrangements to provide suitable library functions for
6056 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6057 @samp{sparclite-*-*} do provide software floating point support.
6059 @option{-msoft-float} changes the calling convention in the output file;
6060 therefore, it is only useful if you compile @emph{all} of a program with
6061 this option. In particular, you need to compile @file{libgcc.a}, the
6062 library that comes with GCC, with @option{-msoft-float} in order for
6065 @item -mhard-quad-float
6066 @opindex mhard-quad-float
6067 Generate output containing quad-word (long double) floating point
6070 @item -msoft-quad-float
6071 @opindex msoft-quad-float
6072 Generate output containing library calls for quad-word (long double)
6073 floating point instructions. The functions called are those specified
6074 in the SPARC ABI@. This is the default.
6076 As of this writing, there are no sparc implementations that have hardware
6077 support for the quad-word floating point instructions. They all invoke
6078 a trap handler for one of these instructions, and then the trap handler
6079 emulates the effect of the instruction. Because of the trap handler overhead,
6080 this is much slower than calling the ABI library routines. Thus the
6081 @option{-msoft-quad-float} option is the default.
6087 With @option{-mflat}, the compiler does not generate save/restore instructions
6088 and will use a ``flat'' or single register window calling convention.
6089 This model uses %i7 as the frame pointer and is compatible with the normal
6090 register window model. Code from either may be intermixed.
6091 The local registers and the input registers (0--5) are still treated as
6092 ``call saved'' registers and will be saved on the stack as necessary.
6094 With @option{-mno-flat} (the default), the compiler emits save/restore
6095 instructions (except for leaf functions) and is the normal mode of operation.
6097 @item -mno-unaligned-doubles
6098 @itemx -munaligned-doubles
6099 @opindex mno-unaligned-doubles
6100 @opindex munaligned-doubles
6101 Assume that doubles have 8 byte alignment. This is the default.
6103 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6104 alignment only if they are contained in another type, or if they have an
6105 absolute address. Otherwise, it assumes they have 4 byte alignment.
6106 Specifying this option avoids some rare compatibility problems with code
6107 generated by other compilers. It is not the default because it results
6108 in a performance loss, especially for floating point code.
6110 @item -mno-faster-structs
6111 @itemx -mfaster-structs
6112 @opindex mno-faster-structs
6113 @opindex mfaster-structs
6114 With @option{-mfaster-structs}, the compiler assumes that structures
6115 should have 8 byte alignment. This enables the use of pairs of
6116 @code{ldd} and @code{std} instructions for copies in structure
6117 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6118 However, the use of this changed alignment directly violates the SPARC
6119 ABI@. Thus, it's intended only for use on targets where the developer
6120 acknowledges that their resulting code will not be directly in line with
6121 the rules of the ABI@.
6124 @opindex mimpure-text
6125 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6126 the compiler to not pass @option{-z text} to the linker when linking a
6127 shared object. Using this option, you can link position-dependent
6128 code into a shared object.
6130 @option{-mimpure-text} suppresses the ``relocations remain against
6131 allocatable but non-writable sections'' linker error message.
6132 However, the necessary relocations will trigger copy-on-write, and the
6133 shared object is not actually shared across processes. Instead of
6134 using @option{-mimpure-text}, you should compile all source code with
6135 @option{-fpic} or @option{-fPIC}.
6137 This option is only available on SunOS and Solaris.
6143 These two options select variations on the SPARC architecture.
6145 By default (unless specifically configured for the Fujitsu SPARClite),
6146 GCC generates code for the v7 variant of the SPARC architecture.
6148 @option{-mv8} will give you SPARC v8 code. The only difference from v7
6149 code is that the compiler emits the integer multiply and integer
6150 divide instructions which exist in SPARC v8 but not in SPARC v7.
6152 @option{-msparclite} will give you SPARClite code. This adds the integer
6153 multiply, integer divide step and scan (@code{ffs}) instructions which
6154 exist in SPARClite but not in SPARC v7.
6156 These options are deprecated and will be deleted in a future GCC release.
6157 They have been replaced with @option{-mcpu=xxx}.
6162 @opindex msupersparc
6163 These two options select the processor for which the code is optimized.
6165 With @option{-mcypress} (the default), the compiler optimizes code for the
6166 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
6167 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6169 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6170 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6171 of the full SPARC v8 instruction set.
6173 These options are deprecated and will be deleted in a future GCC release.
6174 They have been replaced with @option{-mcpu=xxx}.
6176 @item -mcpu=@var{cpu_type}
6178 Set the instruction set, register set, and instruction scheduling parameters
6179 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6180 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6181 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
6182 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6185 Default instruction scheduling parameters are used for values that select
6186 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6187 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6189 Here is a list of each supported architecture and their supported
6194 v8: supersparc, hypersparc
6195 sparclite: f930, f934, sparclite86x
6197 v9: ultrasparc, ultrasparc3
6200 @item -mtune=@var{cpu_type}
6202 Set the instruction scheduling parameters for machine type
6203 @var{cpu_type}, but do not set the instruction set or register set that the
6204 option @option{-mcpu=@var{cpu_type}} would.
6206 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6207 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6208 that select a particular cpu implementation. Those are @samp{cypress},
6209 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6210 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6215 These @samp{-m} switches are supported in addition to the above
6216 on the SPARCLET processor.
6219 @item -mlittle-endian
6220 @opindex mlittle-endian
6221 Generate code for a processor running in little-endian mode.
6225 Treat register @code{%g0} as a normal register.
6226 GCC will continue to clobber it as necessary but will not assume
6227 it always reads as 0.
6229 @item -mbroken-saverestore
6230 @opindex mbroken-saverestore
6231 Generate code that does not use non-trivial forms of the @code{save} and
6232 @code{restore} instructions. Early versions of the SPARCLET processor do
6233 not correctly handle @code{save} and @code{restore} instructions used with
6234 arguments. They correctly handle them used without arguments. A @code{save}
6235 instruction used without arguments increments the current window pointer
6236 but does not allocate a new stack frame. It is assumed that the window
6237 overflow trap handler will properly handle this case as will interrupt
6241 These @samp{-m} switches are supported in addition to the above
6242 on SPARC V9 processors in 64-bit environments.
6245 @item -mlittle-endian
6246 @opindex mlittle-endian
6247 Generate code for a processor running in little-endian mode. It is only
6248 available for a few configurations and most notably not on Solaris.
6254 Generate code for a 32-bit or 64-bit environment.
6255 The 32-bit environment sets int, long and pointer to 32 bits.
6256 The 64-bit environment sets int to 32 bits and long and pointer
6259 @item -mcmodel=medlow
6260 @opindex mcmodel=medlow
6261 Generate code for the Medium/Low code model: the program must be linked
6262 in the low 32 bits of the address space. Pointers are 64 bits.
6263 Programs can be statically or dynamically linked.
6265 @item -mcmodel=medmid
6266 @opindex mcmodel=medmid
6267 Generate code for the Medium/Middle code model: the program must be linked
6268 in the low 44 bits of the address space, the text segment must be less than
6269 2G bytes, and data segment must be within 2G of the text segment.
6270 Pointers are 64 bits.
6272 @item -mcmodel=medany
6273 @opindex mcmodel=medany
6274 Generate code for the Medium/Anywhere code model: the program may be linked
6275 anywhere in the address space, the text segment must be less than
6276 2G bytes, and data segment must be within 2G of the text segment.
6277 Pointers are 64 bits.
6279 @item -mcmodel=embmedany
6280 @opindex mcmodel=embmedany
6281 Generate code for the Medium/Anywhere code model for embedded systems:
6282 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6283 (determined at link time). Register %g4 points to the base of the
6284 data segment. Pointers are still 64 bits.
6285 Programs are statically linked, PIC is not supported.
6288 @itemx -mno-stack-bias
6289 @opindex mstack-bias
6290 @opindex mno-stack-bias
6291 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6292 frame pointer if present, are offset by @minus{}2047 which must be added back
6293 when making stack frame references.
6294 Otherwise, assume no such offset is present.
6298 @subsection ARM Options
6301 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6306 @opindex mapcs-frame
6307 Generate a stack frame that is compliant with the ARM Procedure Call
6308 Standard for all functions, even if this is not strictly necessary for
6309 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6310 with this option will cause the stack frames not to be generated for
6311 leaf functions. The default is @option{-mno-apcs-frame}.
6315 This is a synonym for @option{-mapcs-frame}.
6319 Generate code for a processor running with a 26-bit program counter,
6320 and conforming to the function calling standards for the APCS 26-bit
6321 option. This option replaces the @option{-m2} and @option{-m3} options
6322 of previous releases of the compiler.
6326 Generate code for a processor running with a 32-bit program counter,
6327 and conforming to the function calling standards for the APCS 32-bit
6328 option. This option replaces the @option{-m6} option of previous releases
6332 @c not currently implemented
6333 @item -mapcs-stack-check
6334 @opindex mapcs-stack-check
6335 Generate code to check the amount of stack space available upon entry to
6336 every function (that actually uses some stack space). If there is
6337 insufficient space available then either the function
6338 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6339 called, depending upon the amount of stack space required. The run time
6340 system is required to provide these functions. The default is
6341 @option{-mno-apcs-stack-check}, since this produces smaller code.
6343 @c not currently implemented
6345 @opindex mapcs-float
6346 Pass floating point arguments using the float point registers. This is
6347 one of the variants of the APCS@. This option is recommended if the
6348 target hardware has a floating point unit or if a lot of floating point
6349 arithmetic is going to be performed by the code. The default is
6350 @option{-mno-apcs-float}, since integer only code is slightly increased in
6351 size if @option{-mapcs-float} is used.
6353 @c not currently implemented
6354 @item -mapcs-reentrant
6355 @opindex mapcs-reentrant
6356 Generate reentrant, position independent code. The default is
6357 @option{-mno-apcs-reentrant}.
6360 @item -mthumb-interwork
6361 @opindex mthumb-interwork
6362 Generate code which supports calling between the ARM and Thumb
6363 instruction sets. Without this option the two instruction sets cannot
6364 be reliably used inside one program. The default is
6365 @option{-mno-thumb-interwork}, since slightly larger code is generated
6366 when @option{-mthumb-interwork} is specified.
6368 @item -mno-sched-prolog
6369 @opindex mno-sched-prolog
6370 Prevent the reordering of instructions in the function prolog, or the
6371 merging of those instruction with the instructions in the function's
6372 body. This means that all functions will start with a recognizable set
6373 of instructions (or in fact one of a choice from a small set of
6374 different function prologues), and this information can be used to
6375 locate the start if functions inside an executable piece of code. The
6376 default is @option{-msched-prolog}.
6379 @opindex mhard-float
6380 Generate output containing floating point instructions. This is the
6384 @opindex msoft-float
6385 Generate output containing library calls for floating point.
6386 @strong{Warning:} the requisite libraries are not available for all ARM
6387 targets. Normally the facilities of the machine's usual C compiler are
6388 used, but this cannot be done directly in cross-compilation. You must make
6389 your own arrangements to provide suitable library functions for
6392 @option{-msoft-float} changes the calling convention in the output file;
6393 therefore, it is only useful if you compile @emph{all} of a program with
6394 this option. In particular, you need to compile @file{libgcc.a}, the
6395 library that comes with GCC, with @option{-msoft-float} in order for
6398 @item -mlittle-endian
6399 @opindex mlittle-endian
6400 Generate code for a processor running in little-endian mode. This is
6401 the default for all standard configurations.
6404 @opindex mbig-endian
6405 Generate code for a processor running in big-endian mode; the default is
6406 to compile code for a little-endian processor.
6408 @item -mwords-little-endian
6409 @opindex mwords-little-endian
6410 This option only applies when generating code for big-endian processors.
6411 Generate code for a little-endian word order but a big-endian byte
6412 order. That is, a byte order of the form @samp{32107654}. Note: this
6413 option should only be used if you require compatibility with code for
6414 big-endian ARM processors generated by versions of the compiler prior to
6417 @item -malignment-traps
6418 @opindex malignment-traps
6419 Generate code that will not trap if the MMU has alignment traps enabled.
6420 On ARM architectures prior to ARMv4, there were no instructions to
6421 access half-word objects stored in memory. However, when reading from
6422 memory a feature of the ARM architecture allows a word load to be used,
6423 even if the address is unaligned, and the processor core will rotate the
6424 data as it is being loaded. This option tells the compiler that such
6425 misaligned accesses will cause a MMU trap and that it should instead
6426 synthesize the access as a series of byte accesses. The compiler can
6427 still use word accesses to load half-word data if it knows that the
6428 address is aligned to a word boundary.
6430 This option is ignored when compiling for ARM architecture 4 or later,
6431 since these processors have instructions to directly access half-word
6434 @item -mno-alignment-traps
6435 @opindex mno-alignment-traps
6436 Generate code that assumes that the MMU will not trap unaligned
6437 accesses. This produces better code when the target instruction set
6438 does not have half-word memory operations (i.e.@: implementations prior to
6441 Note that you cannot use this option to access unaligned word objects,
6442 since the processor will only fetch one 32-bit aligned object from
6445 The default setting for most targets is @option{-mno-alignment-traps}, since
6446 this produces better code when there are no half-word memory
6447 instructions available.
6449 @item -mshort-load-bytes
6450 @itemx -mno-short-load-words
6451 @opindex mshort-load-bytes
6452 @opindex mno-short-load-words
6453 These are deprecated aliases for @option{-malignment-traps}.
6455 @item -mno-short-load-bytes
6456 @itemx -mshort-load-words
6457 @opindex mno-short-load-bytes
6458 @opindex mshort-load-words
6459 This are deprecated aliases for @option{-mno-alignment-traps}.
6461 @item -mcpu=@var{name}
6463 This specifies the name of the target ARM processor. GCC uses this name
6464 to determine what kind of instructions it can emit when generating
6465 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6466 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6467 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6468 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6469 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6470 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6471 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6472 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6473 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6474 @samp{arm1020t}, @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
6476 @itemx -mtune=@var{name}
6478 This option is very similar to the @option{-mcpu=} option, except that
6479 instead of specifying the actual target processor type, and hence
6480 restricting which instructions can be used, it specifies that GCC should
6481 tune the performance of the code as if the target were of the type
6482 specified in this option, but still choosing the instructions that it
6483 will generate based on the cpu specified by a @option{-mcpu=} option.
6484 For some ARM implementations better performance can be obtained by using
6487 @item -march=@var{name}
6489 This specifies the name of the target ARM architecture. GCC uses this
6490 name to determine what kind of instructions it can emit when generating
6491 assembly code. This option can be used in conjunction with or instead
6492 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6493 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6494 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{iwmmxt}, @samp{ep9312}.
6496 @item -mfpe=@var{number}
6497 @itemx -mfp=@var{number}
6500 This specifies the version of the floating point emulation available on
6501 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6502 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6504 @item -mstructure-size-boundary=@var{n}
6505 @opindex mstructure-size-boundary
6506 The size of all structures and unions will be rounded up to a multiple
6507 of the number of bits set by this option. Permissible values are 8 and
6508 32. The default value varies for different toolchains. For the COFF
6509 targeted toolchain the default value is 8. Specifying the larger number
6510 can produce faster, more efficient code, but can also increase the size
6511 of the program. The two values are potentially incompatible. Code
6512 compiled with one value cannot necessarily expect to work with code or
6513 libraries compiled with the other value, if they exchange information
6514 using structures or unions.
6516 @item -mabort-on-noreturn
6517 @opindex mabort-on-noreturn
6518 Generate a call to the function @code{abort} at the end of a
6519 @code{noreturn} function. It will be executed if the function tries to
6523 @itemx -mno-long-calls
6524 @opindex mlong-calls
6525 @opindex mno-long-calls
6526 Tells the compiler to perform function calls by first loading the
6527 address of the function into a register and then performing a subroutine
6528 call on this register. This switch is needed if the target function
6529 will lie outside of the 64 megabyte addressing range of the offset based
6530 version of subroutine call instruction.
6532 Even if this switch is enabled, not all function calls will be turned
6533 into long calls. The heuristic is that static functions, functions
6534 which have the @samp{short-call} attribute, functions that are inside
6535 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6536 definitions have already been compiled within the current compilation
6537 unit, will not be turned into long calls. The exception to this rule is
6538 that weak function definitions, functions with the @samp{long-call}
6539 attribute or the @samp{section} attribute, and functions that are within
6540 the scope of a @samp{#pragma long_calls} directive, will always be
6541 turned into long calls.
6543 This feature is not enabled by default. Specifying
6544 @option{-mno-long-calls} will restore the default behavior, as will
6545 placing the function calls within the scope of a @samp{#pragma
6546 long_calls_off} directive. Note these switches have no effect on how
6547 the compiler generates code to handle function calls via function
6550 @item -mnop-fun-dllimport
6551 @opindex mnop-fun-dllimport
6552 Disable support for the @code{dllimport} attribute.
6554 @item -msingle-pic-base
6555 @opindex msingle-pic-base
6556 Treat the register used for PIC addressing as read-only, rather than
6557 loading it in the prologue for each function. The run-time system is
6558 responsible for initializing this register with an appropriate value
6559 before execution begins.
6561 @item -mpic-register=@var{reg}
6562 @opindex mpic-register
6563 Specify the register to be used for PIC addressing. The default is R10
6564 unless stack-checking is enabled, when R9 is used.
6566 @item -mcirrus-fix-invalid-insns
6567 @opindex mcirrus-fix-invalid-insns
6568 @opindex mno-cirrus-fix-invalid-insns
6569 Insert NOPs into the instruction stream to in order to work around
6570 problems with invalid Maverick instruction combinations. This option
6571 is only valid if the @option{-mcpu=ep9312} option has been used to
6572 enable generation of instructions for the Cirrus Maverick floating
6573 point co-processor. This option is not enabled by default, since the
6574 problem is only present in older Maverick implementations. The default
6575 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6578 @item -mpoke-function-name
6579 @opindex mpoke-function-name
6580 Write the name of each function into the text section, directly
6581 preceding the function prologue. The generated code is similar to this:
6585 .ascii "arm_poke_function_name", 0
6588 .word 0xff000000 + (t1 - t0)
6589 arm_poke_function_name
6591 stmfd sp!, @{fp, ip, lr, pc@}
6595 When performing a stack backtrace, code can inspect the value of
6596 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6597 location @code{pc - 12} and the top 8 bits are set, then we know that
6598 there is a function name embedded immediately preceding this location
6599 and has length @code{((pc[-3]) & 0xff000000)}.
6603 Generate code for the 16-bit Thumb instruction set. The default is to
6604 use the 32-bit ARM instruction set.
6607 @opindex mtpcs-frame
6608 Generate a stack frame that is compliant with the Thumb Procedure Call
6609 Standard for all non-leaf functions. (A leaf function is one that does
6610 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6612 @item -mtpcs-leaf-frame
6613 @opindex mtpcs-leaf-frame
6614 Generate a stack frame that is compliant with the Thumb Procedure Call
6615 Standard for all leaf functions. (A leaf function is one that does
6616 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6618 @item -mcallee-super-interworking
6619 @opindex mcallee-super-interworking
6620 Gives all externally visible functions in the file being compiled an ARM
6621 instruction set header which switches to Thumb mode before executing the
6622 rest of the function. This allows these functions to be called from
6623 non-interworking code.
6625 @item -mcaller-super-interworking
6626 @opindex mcaller-super-interworking
6627 Allows calls via function pointers (including virtual functions) to
6628 execute correctly regardless of whether the target code has been
6629 compiled for interworking or not. There is a small overhead in the cost
6630 of executing a function pointer if this option is enabled.
6634 @node MN10200 Options
6635 @subsection MN10200 Options
6636 @cindex MN10200 options
6638 These @option{-m} options are defined for Matsushita MN10200 architectures:
6643 Indicate to the linker that it should perform a relaxation optimization pass
6644 to shorten branches, calls and absolute memory addresses. This option only
6645 has an effect when used on the command line for the final link step.
6647 This option makes symbolic debugging impossible.
6650 @node MN10300 Options
6651 @subsection MN10300 Options
6652 @cindex MN10300 options
6654 These @option{-m} options are defined for Matsushita MN10300 architectures:
6659 Generate code to avoid bugs in the multiply instructions for the MN10300
6660 processors. This is the default.
6663 @opindex mno-mult-bug
6664 Do not generate code to avoid bugs in the multiply instructions for the
6669 Generate code which uses features specific to the AM33 processor.
6673 Do not generate code which uses features specific to the AM33 processor. This
6678 Do not link in the C run-time initialization object file.
6682 Indicate to the linker that it should perform a relaxation optimization pass
6683 to shorten branches, calls and absolute memory addresses. This option only
6684 has an effect when used on the command line for the final link step.
6686 This option makes symbolic debugging impossible.
6690 @node M32R/D Options
6691 @subsection M32R/D Options
6692 @cindex M32R/D options
6694 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6699 Generate code for the M32R/X@.
6703 Generate code for the M32R@. This is the default.
6705 @item -mcode-model=small
6706 @opindex mcode-model=small
6707 Assume all objects live in the lower 16MB of memory (so that their addresses
6708 can be loaded with the @code{ld24} instruction), and assume all subroutines
6709 are reachable with the @code{bl} instruction.
6710 This is the default.
6712 The addressability of a particular object can be set with the
6713 @code{model} attribute.
6715 @item -mcode-model=medium
6716 @opindex mcode-model=medium
6717 Assume objects may be anywhere in the 32-bit address space (the compiler
6718 will generate @code{seth/add3} instructions to load their addresses), and
6719 assume all subroutines are reachable with the @code{bl} instruction.
6721 @item -mcode-model=large
6722 @opindex mcode-model=large
6723 Assume objects may be anywhere in the 32-bit address space (the compiler
6724 will generate @code{seth/add3} instructions to load their addresses), and
6725 assume subroutines may not be reachable with the @code{bl} instruction
6726 (the compiler will generate the much slower @code{seth/add3/jl}
6727 instruction sequence).
6730 @opindex msdata=none
6731 Disable use of the small data area. Variables will be put into
6732 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6733 @code{section} attribute has been specified).
6734 This is the default.
6736 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6737 Objects may be explicitly put in the small data area with the
6738 @code{section} attribute using one of these sections.
6741 @opindex msdata=sdata
6742 Put small global and static data in the small data area, but do not
6743 generate special code to reference them.
6747 Put small global and static data in the small data area, and generate
6748 special instructions to reference them.
6752 @cindex smaller data references
6753 Put global and static objects less than or equal to @var{num} bytes
6754 into the small data or bss sections instead of the normal data or bss
6755 sections. The default value of @var{num} is 8.
6756 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6757 for this option to have any effect.
6759 All modules should be compiled with the same @option{-G @var{num}} value.
6760 Compiling with different values of @var{num} may or may not work; if it
6761 doesn't the linker will give an error message---incorrect code will not be
6767 @subsection M88K Options
6768 @cindex M88k options
6770 These @samp{-m} options are defined for Motorola 88k architectures:
6775 Generate code that works well on both the m88100 and the
6780 Generate code that works best for the m88100, but that also
6785 Generate code that works best for the m88110, and may not run
6790 Obsolete option to be removed from the next revision.
6793 @item -midentify-revision
6794 @opindex midentify-revision
6795 @cindex identifying source, compiler (88k)
6796 Include an @code{ident} directive in the assembler output recording the
6797 source file name, compiler name and version, timestamp, and compilation
6800 @item -mno-underscores
6801 @opindex mno-underscores
6802 @cindex underscores, avoiding (88k)
6803 In assembler output, emit symbol names without adding an underscore
6804 character at the beginning of each name. The default is to use an
6805 underscore as prefix on each name.
6807 @item -mocs-debug-info
6808 @itemx -mno-ocs-debug-info
6809 @opindex mocs-debug-info
6810 @opindex mno-ocs-debug-info
6812 @cindex debugging, 88k OCS
6813 Include (or omit) additional debugging information (about registers used
6814 in each stack frame) as specified in the 88open Object Compatibility
6815 Standard, ``OCS''@. This extra information allows debugging of code that
6816 has had the frame pointer eliminated. The default for SVr4 and Delta 88
6817 SVr3.2 is to include this information; other 88k configurations omit this
6818 information by default.
6820 @item -mocs-frame-position
6821 @opindex mocs-frame-position
6822 @cindex register positions in frame (88k)
6823 When emitting COFF debugging information for automatic variables and
6824 parameters stored on the stack, use the offset from the canonical frame
6825 address, which is the stack pointer (register 31) on entry to the
6826 function. The SVr4 and Delta88 SVr3.2, and BCS configurations use
6827 @option{-mocs-frame-position}; other 88k configurations have the default
6828 @option{-mno-ocs-frame-position}.
6830 @item -mno-ocs-frame-position
6831 @opindex mno-ocs-frame-position
6832 @cindex register positions in frame (88k)
6833 When emitting COFF debugging information for automatic variables and
6834 parameters stored on the stack, use the offset from the frame pointer
6835 register (register 30). When this option is in effect, the frame
6836 pointer is not eliminated when debugging information is selected by the
6839 @item -moptimize-arg-area
6840 @opindex moptimize-arg-area
6841 @cindex arguments in frame (88k)
6842 Save space by reorganizing the stack frame. This option generates code
6843 that does not agree with the 88open specifications, but uses less
6846 @itemx -mno-optimize-arg-area
6847 @opindex mno-optimize-arg-area
6848 Do not reorganize the stack frame to save space. This is the default.
6849 The generated conforms to the specification, but uses more memory.
6851 @item -mshort-data-@var{num}
6852 @opindex mshort-data
6853 @cindex smaller data references (88k)
6854 @cindex r0-relative references (88k)
6855 Generate smaller data references by making them relative to @code{r0},
6856 which allows loading a value using a single instruction (rather than the
6857 usual two). You control which data references are affected by
6858 specifying @var{num} with this option. For example, if you specify
6859 @option{-mshort-data-512}, then the data references affected are those
6860 involving displacements of less than 512 bytes.
6861 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6864 @item -mserialize-volatile
6865 @opindex mserialize-volatile
6866 @itemx -mno-serialize-volatile
6867 @opindex mno-serialize-volatile
6868 @cindex sequential consistency on 88k
6869 Do, or don't, generate code to guarantee sequential consistency
6870 of volatile memory references. By default, consistency is
6873 The order of memory references made by the MC88110 processor does
6874 not always match the order of the instructions requesting those
6875 references. In particular, a load instruction may execute before
6876 a preceding store instruction. Such reordering violates
6877 sequential consistency of volatile memory references, when there
6878 are multiple processors. When consistency must be guaranteed,
6879 GCC generates special instructions, as needed, to force
6880 execution in the proper order.
6882 The MC88100 processor does not reorder memory references and so
6883 always provides sequential consistency. However, by default, GCC
6884 generates the special instructions to guarantee consistency
6885 even when you use @option{-m88100}, so that the code may be run on an
6886 MC88110 processor. If you intend to run your code only on the
6887 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6889 The extra code generated to guarantee consistency may affect the
6890 performance of your application. If you know that you can safely
6891 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6897 @cindex assembler syntax, 88k
6899 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6900 related to System V release 4 (SVr4). This controls the following:
6904 Which variant of the assembler syntax to emit.
6906 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6907 that is used on System V release 4.
6909 @option{-msvr4} makes GCC issue additional declaration directives used in
6913 @option{-msvr4} is the default for the m88k-motorola-sysv4 configuration.
6914 @option{-msvr3} is the default for all other m88k configurations.
6916 @item -mversion-03.00
6917 @opindex mversion-03.00
6918 This option is obsolete, and is ignored.
6919 @c ??? which asm syntax better for GAS? option there too?
6921 @item -mno-check-zero-division
6922 @itemx -mcheck-zero-division
6923 @opindex mno-check-zero-division
6924 @opindex mcheck-zero-division
6925 @cindex zero division on 88k
6926 Do, or don't, generate code to guarantee that integer division by
6927 zero will be detected. By default, detection is guaranteed.
6929 Some models of the MC88100 processor fail to trap upon integer
6930 division by zero under certain conditions. By default, when
6931 compiling code that might be run on such a processor, GCC
6932 generates code that explicitly checks for zero-valued divisors
6933 and traps with exception number 503 when one is detected. Use of
6934 @option{-mno-check-zero-division} suppresses such checking for code
6935 generated to run on an MC88100 processor.
6937 GCC assumes that the MC88110 processor correctly detects all instances
6938 of integer division by zero. When @option{-m88110} is specified, no
6939 explicit checks for zero-valued divisors are generated, and both
6940 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6943 @item -muse-div-instruction
6944 @opindex muse-div-instruction
6945 @cindex divide instruction, 88k
6946 Use the div instruction for signed integer division on the
6947 MC88100 processor. By default, the div instruction is not used.
6949 On the MC88100 processor the signed integer division instruction
6950 div) traps to the operating system on a negative operand. The
6951 operating system transparently completes the operation, but at a
6952 large cost in execution time. By default, when compiling code
6953 that might be run on an MC88100 processor, GCC emulates signed
6954 integer division using the unsigned integer division instruction
6955 divu), thereby avoiding the large penalty of a trap to the
6956 operating system. Such emulation has its own, smaller, execution
6957 cost in both time and space. To the extent that your code's
6958 important signed integer division operations are performed on two
6959 nonnegative operands, it may be desirable to use the div
6960 instruction directly.
6962 On the MC88110 processor the div instruction (also known as the
6963 divs instruction) processes negative operands without trapping to
6964 the operating system. When @option{-m88110} is specified,
6965 @option{-muse-div-instruction} is ignored, and the div instruction is used
6966 for signed integer division.
6968 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6969 particular, the behavior of such a division with and without
6970 @option{-muse-div-instruction} may differ.
6972 @item -mtrap-large-shift
6973 @itemx -mhandle-large-shift
6974 @opindex mtrap-large-shift
6975 @opindex mhandle-large-shift
6976 @cindex bit shift overflow (88k)
6977 @cindex large bit shifts (88k)
6978 Include code to detect bit-shifts of more than 31 bits; respectively,
6979 trap such shifts or emit code to handle them properly. By default GCC
6980 makes no special provision for large bit shifts.
6982 @item -mwarn-passed-structs
6983 @opindex mwarn-passed-structs
6984 @cindex structure passing (88k)
6985 Warn when a function passes a struct as an argument or result.
6986 Structure-passing conventions have changed during the evolution of the C
6987 language, and are often the source of portability problems. By default,
6988 GCC issues no such warning.
6991 @c break page here to avoid unsightly interparagraph stretch.
6995 @node RS/6000 and PowerPC Options
6996 @subsection IBM RS/6000 and PowerPC Options
6997 @cindex RS/6000 and PowerPC Options
6998 @cindex IBM RS/6000 and PowerPC Options
7000 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
7008 @itemx -mpowerpc-gpopt
7009 @itemx -mno-powerpc-gpopt
7010 @itemx -mpowerpc-gfxopt
7011 @itemx -mno-powerpc-gfxopt
7013 @itemx -mno-powerpc64
7019 @opindex mno-powerpc
7020 @opindex mpowerpc-gpopt
7021 @opindex mno-powerpc-gpopt
7022 @opindex mpowerpc-gfxopt
7023 @opindex mno-powerpc-gfxopt
7025 @opindex mno-powerpc64
7026 GCC supports two related instruction set architectures for the
7027 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
7028 instructions supported by the @samp{rios} chip set used in the original
7029 RS/6000 systems and the @dfn{PowerPC} instruction set is the
7030 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
7031 the IBM 4xx microprocessors.
7033 Neither architecture is a subset of the other. However there is a
7034 large common subset of instructions supported by both. An MQ
7035 register is included in processors supporting the POWER architecture.
7037 You use these options to specify which instructions are available on the
7038 processor you are using. The default value of these options is
7039 determined when configuring GCC@. Specifying the
7040 @option{-mcpu=@var{cpu_type}} overrides the specification of these
7041 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
7042 rather than the options listed above.
7044 The @option{-mpower} option allows GCC to generate instructions that
7045 are found only in the POWER architecture and to use the MQ register.
7046 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
7047 to generate instructions that are present in the POWER2 architecture but
7048 not the original POWER architecture.
7050 The @option{-mpowerpc} option allows GCC to generate instructions that
7051 are found only in the 32-bit subset of the PowerPC architecture.
7052 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
7053 GCC to use the optional PowerPC architecture instructions in the
7054 General Purpose group, including floating-point square root. Specifying
7055 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
7056 use the optional PowerPC architecture instructions in the Graphics
7057 group, including floating-point select.
7059 The @option{-mpowerpc64} option allows GCC to generate the additional
7060 64-bit instructions that are found in the full PowerPC64 architecture
7061 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
7062 @option{-mno-powerpc64}.
7064 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7065 will use only the instructions in the common subset of both
7066 architectures plus some special AIX common-mode calls, and will not use
7067 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7068 permits GCC to use any instruction from either architecture and to
7069 allow use of the MQ register; specify this for the Motorola MPC601.
7071 @item -mnew-mnemonics
7072 @itemx -mold-mnemonics
7073 @opindex mnew-mnemonics
7074 @opindex mold-mnemonics
7075 Select which mnemonics to use in the generated assembler code. With
7076 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7077 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7078 assembler mnemonics defined for the POWER architecture. Instructions
7079 defined in only one architecture have only one mnemonic; GCC uses that
7080 mnemonic irrespective of which of these options is specified.
7082 GCC defaults to the mnemonics appropriate for the architecture in
7083 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7084 value of these option. Unless you are building a cross-compiler, you
7085 should normally not specify either @option{-mnew-mnemonics} or
7086 @option{-mold-mnemonics}, but should instead accept the default.
7088 @item -mcpu=@var{cpu_type}
7090 Set architecture type, register usage, choice of mnemonics, and
7091 instruction scheduling parameters for machine type @var{cpu_type}.
7092 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
7093 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
7094 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
7095 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
7096 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
7097 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
7099 @option{-mcpu=common} selects a completely generic processor. Code
7100 generated under this option will run on any POWER or PowerPC processor.
7101 GCC will use only the instructions in the common subset of both
7102 architectures, and will not use the MQ register. GCC assumes a generic
7103 processor model for scheduling purposes.
7105 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7106 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7107 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7108 types, with an appropriate, generic processor model assumed for
7109 scheduling purposes.
7111 The other options specify a specific processor. Code generated under
7112 those options will run best on that processor, and may not run at all on
7115 The @option{-mcpu} options automatically enable or disable other
7116 @option{-m} options as follows:
7120 @option{-mno-power}, @option{-mno-powerpc}
7127 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
7142 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7145 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7150 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
7153 @item -mtune=@var{cpu_type}
7155 Set the instruction scheduling parameters for machine type
7156 @var{cpu_type}, but do not set the architecture type, register usage, or
7157 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7158 values for @var{cpu_type} are used for @option{-mtune} as for
7159 @option{-mcpu}. If both are specified, the code generated will use the
7160 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7161 scheduling parameters set by @option{-mtune}.
7166 @opindex mno-altivec
7167 These switches enable or disable the use of built-in functions that
7168 allow access to the AltiVec instruction set. You may also need to set
7169 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7174 Extend the current ABI with SPE ABI extensions. This does not change
7175 the default ABI, instead it adds the SPE ABI extensions to the current
7179 @opindex mabi=no-spe
7180 Disable Booke SPE ABI extensions for the current ABI.
7182 @item -misel=@var{yes/no}
7185 This switch enables or disables the generation of ISEL instructions.
7187 @item -mspe=@var{yes/no}
7190 This switch enables or disables the generation of SPE simd
7193 @item -mfloat-gprs=@var{yes/no}
7195 @opindex mfloat-gprs
7196 This switch enables or disables the generation of floating point
7197 operations on the general purpose registers for architectures that
7198 support it. This option is currently only available on the MPC8540.
7201 @itemx -mno-fp-in-toc
7202 @itemx -mno-sum-in-toc
7203 @itemx -mminimal-toc
7205 @opindex mno-fp-in-toc
7206 @opindex mno-sum-in-toc
7207 @opindex mminimal-toc
7208 Modify generation of the TOC (Table Of Contents), which is created for
7209 every executable file. The @option{-mfull-toc} option is selected by
7210 default. In that case, GCC will allocate at least one TOC entry for
7211 each unique non-automatic variable reference in your program. GCC
7212 will also place floating-point constants in the TOC@. However, only
7213 16,384 entries are available in the TOC@.
7215 If you receive a linker error message that saying you have overflowed
7216 the available TOC space, you can reduce the amount of TOC space used
7217 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7218 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7219 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7220 generate code to calculate the sum of an address and a constant at
7221 run-time instead of putting that sum into the TOC@. You may specify one
7222 or both of these options. Each causes GCC to produce very slightly
7223 slower and larger code at the expense of conserving TOC space.
7225 If you still run out of space in the TOC even when you specify both of
7226 these options, specify @option{-mminimal-toc} instead. This option causes
7227 GCC to make only one TOC entry for every file. When you specify this
7228 option, GCC will produce code that is slower and larger but which
7229 uses extremely little TOC space. You may wish to use this option
7230 only on files that contain less frequently executed code.
7236 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7237 @code{long} type, and the infrastructure needed to support them.
7238 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7239 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7240 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7245 @opindex mno-xl-call
7246 On AIX, pass floating-point arguments to prototyped functions beyond the
7247 register save area (RSA) on the stack in addition to argument FPRs. The
7248 AIX calling convention was extended but not initially documented to
7249 handle an obscure K&R C case of calling a function that takes the
7250 address of its arguments with fewer arguments than declared. AIX XL
7251 compilers access floating point arguments which do not fit in the
7252 RSA from the stack when a subroutine is compiled without
7253 optimization. Because always storing floating-point arguments on the
7254 stack is inefficient and rarely needed, this option is not enabled by
7255 default and only is necessary when calling subroutines compiled by AIX
7256 XL compilers without optimization.
7260 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7261 application written to use message passing with special startup code to
7262 enable the application to run. The system must have PE installed in the
7263 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7264 must be overridden with the @option{-specs=} option to specify the
7265 appropriate directory location. The Parallel Environment does not
7266 support threads, so the @option{-mpe} option and the @option{-pthread}
7267 option are incompatible.
7269 @item -malign-natural
7270 @itemx -malign-power
7271 @opindex malign-natural
7272 @opindex malign-power
7273 On AIX, Darwin, and 64-bit PowerPC Linux, the option
7274 @option{-malign-natural} overrides the ABI-defined alignment of larger
7275 types, such as floating-point doubles, on their natural size-based boundary.
7276 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7277 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7281 @opindex msoft-float
7282 @opindex mhard-float
7283 Generate code that does not use (uses) the floating-point register set.
7284 Software floating point emulation is provided if you use the
7285 @option{-msoft-float} option, and pass the option to GCC when linking.
7288 @itemx -mno-multiple
7290 @opindex mno-multiple
7291 Generate code that uses (does not use) the load multiple word
7292 instructions and the store multiple word instructions. These
7293 instructions are generated by default on POWER systems, and not
7294 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7295 endian PowerPC systems, since those instructions do not work when the
7296 processor is in little endian mode. The exceptions are PPC740 and
7297 PPC750 which permit the instructions usage in little endian mode.
7303 Generate code that uses (does not use) the load string instructions
7304 and the store string word instructions to save multiple registers and
7305 do small block moves. These instructions are generated by default on
7306 POWER systems, and not generated on PowerPC systems. Do not use
7307 @option{-mstring} on little endian PowerPC systems, since those
7308 instructions do not work when the processor is in little endian mode.
7309 The exceptions are PPC740 and PPC750 which permit the instructions
7310 usage in little endian mode.
7316 Generate code that uses (does not use) the load or store instructions
7317 that update the base register to the address of the calculated memory
7318 location. These instructions are generated by default. If you use
7319 @option{-mno-update}, there is a small window between the time that the
7320 stack pointer is updated and the address of the previous frame is
7321 stored, which means code that walks the stack frame across interrupts or
7322 signals may get corrupted data.
7325 @itemx -mno-fused-madd
7326 @opindex mfused-madd
7327 @opindex mno-fused-madd
7328 Generate code that uses (does not use) the floating point multiply and
7329 accumulate instructions. These instructions are generated by default if
7330 hardware floating is used.
7332 @item -mno-bit-align
7334 @opindex mno-bit-align
7336 On System V.4 and embedded PowerPC systems do not (do) force structures
7337 and unions that contain bit-fields to be aligned to the base type of the
7340 For example, by default a structure containing nothing but 8
7341 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7342 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7343 the structure would be aligned to a 1 byte boundary and be one byte in
7346 @item -mno-strict-align
7347 @itemx -mstrict-align
7348 @opindex mno-strict-align
7349 @opindex mstrict-align
7350 On System V.4 and embedded PowerPC systems do not (do) assume that
7351 unaligned memory references will be handled by the system.
7354 @itemx -mno-relocatable
7355 @opindex mrelocatable
7356 @opindex mno-relocatable
7357 On embedded PowerPC systems generate code that allows (does not allow)
7358 the program to be relocated to a different address at runtime. If you
7359 use @option{-mrelocatable} on any module, all objects linked together must
7360 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7362 @item -mrelocatable-lib
7363 @itemx -mno-relocatable-lib
7364 @opindex mrelocatable-lib
7365 @opindex mno-relocatable-lib
7366 On embedded PowerPC systems generate code that allows (does not allow)
7367 the program to be relocated to a different address at runtime. Modules
7368 compiled with @option{-mrelocatable-lib} can be linked with either modules
7369 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7370 with modules compiled with the @option{-mrelocatable} options.
7376 On System V.4 and embedded PowerPC systems do not (do) assume that
7377 register 2 contains a pointer to a global area pointing to the addresses
7378 used in the program.
7381 @itemx -mlittle-endian
7383 @opindex mlittle-endian
7384 On System V.4 and embedded PowerPC systems compile code for the
7385 processor in little endian mode. The @option{-mlittle-endian} option is
7386 the same as @option{-mlittle}.
7391 @opindex mbig-endian
7392 On System V.4 and embedded PowerPC systems compile code for the
7393 processor in big endian mode. The @option{-mbig-endian} option is
7394 the same as @option{-mbig}.
7396 @item -mdynamic-no-pic
7397 @opindex mdynamic-no-pic
7398 On Darwin and Mac OS X systems, compile code so that it is not
7399 relocatable, but that its external references are relocatable. The
7400 resulting code is suitable for applications, but not shared
7405 On System V.4 and embedded PowerPC systems compile code using calling
7406 conventions that adheres to the March 1995 draft of the System V
7407 Application Binary Interface, PowerPC processor supplement. This is the
7408 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7410 @item -mcall-sysv-eabi
7411 @opindex mcall-sysv-eabi
7412 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7414 @item -mcall-sysv-noeabi
7415 @opindex mcall-sysv-noeabi
7416 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7418 @item -mcall-solaris
7419 @opindex mcall-solaris
7420 On System V.4 and embedded PowerPC systems compile code for the Solaris
7424 @opindex mcall-linux
7425 On System V.4 and embedded PowerPC systems compile code for the
7426 Linux-based GNU system.
7430 On System V.4 and embedded PowerPC systems compile code for the
7431 Hurd-based GNU system.
7434 @opindex mcall-netbsd
7435 On System V.4 and embedded PowerPC systems compile code for the
7436 NetBSD operating system.
7438 @item -maix-struct-return
7439 @opindex maix-struct-return
7440 Return all structures in memory (as specified by the AIX ABI)@.
7442 @item -msvr4-struct-return
7443 @opindex msvr4-struct-return
7444 Return structures smaller than 8 bytes in registers (as specified by the
7448 @opindex mabi=altivec
7449 Extend the current ABI with AltiVec ABI extensions. This does not
7450 change the default ABI, instead it adds the AltiVec ABI extensions to
7453 @item -mabi=no-altivec
7454 @opindex mabi=no-altivec
7455 Disable AltiVec ABI extensions for the current ABI.
7458 @itemx -mno-prototype
7460 @opindex mno-prototype
7461 On System V.4 and embedded PowerPC systems assume that all calls to
7462 variable argument functions are properly prototyped. Otherwise, the
7463 compiler must insert an instruction before every non prototyped call to
7464 set or clear bit 6 of the condition code register (@var{CR}) to
7465 indicate whether floating point values were passed in the floating point
7466 registers in case the function takes a variable arguments. With
7467 @option{-mprototype}, only calls to prototyped variable argument functions
7468 will set or clear the bit.
7472 On embedded PowerPC systems, assume that the startup module is called
7473 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7474 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7479 On embedded PowerPC systems, assume that the startup module is called
7480 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7485 On embedded PowerPC systems, assume that the startup module is called
7486 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7490 @opindex myellowknife
7491 On embedded PowerPC systems, assume that the startup module is called
7492 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7497 On System V.4 and embedded PowerPC systems, specify that you are
7498 compiling for a VxWorks system.
7502 Specify that you are compiling for the WindISS simulation environment.
7506 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7507 header to indicate that @samp{eabi} extended relocations are used.
7513 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7514 Embedded Applications Binary Interface (eabi) which is a set of
7515 modifications to the System V.4 specifications. Selecting @option{-meabi}
7516 means that the stack is aligned to an 8 byte boundary, a function
7517 @code{__eabi} is called to from @code{main} to set up the eabi
7518 environment, and the @option{-msdata} option can use both @code{r2} and
7519 @code{r13} to point to two separate small data areas. Selecting
7520 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7521 do not call an initialization function from @code{main}, and the
7522 @option{-msdata} option will only use @code{r13} to point to a single
7523 small data area. The @option{-meabi} option is on by default if you
7524 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7527 @opindex msdata=eabi
7528 On System V.4 and embedded PowerPC systems, put small initialized
7529 @code{const} global and static data in the @samp{.sdata2} section, which
7530 is pointed to by register @code{r2}. Put small initialized
7531 non-@code{const} global and static data in the @samp{.sdata} section,
7532 which is pointed to by register @code{r13}. Put small uninitialized
7533 global and static data in the @samp{.sbss} section, which is adjacent to
7534 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7535 incompatible with the @option{-mrelocatable} option. The
7536 @option{-msdata=eabi} option also sets the @option{-memb} option.
7539 @opindex msdata=sysv
7540 On System V.4 and embedded PowerPC systems, put small global and static
7541 data in the @samp{.sdata} section, which is pointed to by register
7542 @code{r13}. Put small uninitialized global and static data in the
7543 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7544 The @option{-msdata=sysv} option is incompatible with the
7545 @option{-mrelocatable} option.
7547 @item -msdata=default
7549 @opindex msdata=default
7551 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7552 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7553 same as @option{-msdata=sysv}.
7556 @opindex msdata-data
7557 On System V.4 and embedded PowerPC systems, put small global and static
7558 data in the @samp{.sdata} section. Put small uninitialized global and
7559 static data in the @samp{.sbss} section. Do not use register @code{r13}
7560 to address small data however. This is the default behavior unless
7561 other @option{-msdata} options are used.
7565 @opindex msdata=none
7567 On embedded PowerPC systems, put all initialized global and static data
7568 in the @samp{.data} section, and all uninitialized data in the
7569 @samp{.bss} section.
7573 @cindex smaller data references (PowerPC)
7574 @cindex .sdata/.sdata2 references (PowerPC)
7575 On embedded PowerPC systems, put global and static items less than or
7576 equal to @var{num} bytes into the small data or bss sections instead of
7577 the normal data or bss section. By default, @var{num} is 8. The
7578 @option{-G @var{num}} switch is also passed to the linker.
7579 All modules should be compiled with the same @option{-G @var{num}} value.
7582 @itemx -mno-regnames
7584 @opindex mno-regnames
7585 On System V.4 and embedded PowerPC systems do (do not) emit register
7586 names in the assembly language output using symbolic forms.
7589 @itemx -mno-longcall
7591 @opindex mno-longcall
7592 Default to making all function calls via pointers, so that functions
7593 which reside further than 64 megabytes (67,108,864 bytes) from the
7594 current location can be called. This setting can be overridden by the
7595 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7597 Some linkers are capable of detecting out-of-range calls and generating
7598 glue code on the fly. On these systems, long calls are unnecessary and
7599 generate slower code. As of this writing, the AIX linker can do this,
7600 as can the GNU linker for PowerPC/64. It is planned to add this feature
7601 to the GNU linker for 32-bit PowerPC systems as well.
7603 In the future, we may cause GCC to ignore all longcall specifications
7604 when the linker is known to generate glue.
7608 Adds support for multithreading with the @dfn{pthreads} library.
7609 This option sets flags for both the preprocessor and linker.
7613 @node Darwin Options
7614 @subsection Darwin Options
7615 @cindex Darwin options
7617 These options are defined for all architectures running the Darwin operating
7618 system. They are useful for compatibility with other Mac OS compilers.
7623 Loads all members of static archive libraries.
7624 See man ld(1) for more information.
7626 @item -arch_errors_fatal
7627 @opindex arch_errors_fatal
7628 Cause the errors having to do with files that have the wrong architecture
7632 @opindex bind_at_load
7633 Causes the output file to be marked such that the dynamic linker will
7634 bind all undefined references when the file is loaded or launched.
7638 Produce a Mach-o bundle format file.
7639 See man ld(1) for more information.
7641 @item -bundle_loader @var{executable}
7642 @opindex bundle_loader
7643 This specifies the @var{executable} that will be loading the build
7644 output file being linked. See man ld(1) for more information.
7646 @item -allowable_client @var{client_name}
7650 @item -compatibility_version
7651 @item -current_version
7652 @item -dependency-file
7654 @item -dylinker_install_name
7657 @item -exported_symbols_list
7659 @item -flat_namespace
7660 @item -force_cpusubtype_ALL
7661 @item -force_flat_namespace
7662 @item -headerpad_max_install_names
7666 @item -keep_private_externs
7668 @item -multiply_defined
7669 @item -multiply_defined_unused
7673 @item -noseglinkedit
7674 @item -pagezero_size
7676 @item -prebind_all_twolevel_modules
7677 @item -private_bundle
7678 @item -read_only_relocs
7680 @item -sectobjectsymbols
7684 @item -sectobjectsymbols
7686 @item -seg_addr_table
7687 @item -seg_addr_table_filename
7690 @item -segs_read_only_addr
7691 @item -segs_read_write_addr
7692 @item -single_module
7696 @item -twolevel_namespace
7699 @item -unexported_symbols_list
7700 @item -weak_reference_mismatches
7703 @opindex allowable_client
7705 @opindex client_name
7706 @opindex compatibility_version
7707 @opindex current_version
7708 @opindex dependency-file
7710 @opindex dylinker_install_name
7713 @opindex exported_symbols_list
7715 @opindex flat_namespace
7716 @opindex force_cpusubtype_ALL
7717 @opindex force_flat_namespace
7718 @opindex headerpad_max_install_names
7721 @opindex install_name
7722 @opindex keep_private_externs
7723 @opindex multi_module
7724 @opindex multiply_defined
7725 @opindex multiply_defined_unused
7727 @opindex nomultidefs
7729 @opindex noseglinkedit
7730 @opindex pagezero_size
7732 @opindex prebind_all_twolevel_modules
7733 @opindex private_bundle
7734 @opindex read_only_relocs
7736 @opindex sectobjectsymbols
7740 @opindex sectobjectsymbols
7742 @opindex seg_addr_table
7743 @opindex seg_addr_table_filename
7744 @opindex seglinkedit
7746 @opindex segs_read_only_addr
7747 @opindex segs_read_write_addr
7748 @opindex single_module
7750 @opindex sub_library
7751 @opindex sub_umbrella
7752 @opindex twolevel_namespace
7755 @opindex unexported_symbols_list
7756 @opindex weak_reference_mismatches
7757 @opindex whatsloaded
7759 This options are available for Darwin linker. Darwin linker man page
7760 describes them in detail.
7765 @subsection IBM RT Options
7767 @cindex IBM RT options
7769 These @samp{-m} options are defined for the IBM RT PC:
7773 @opindex min-line-mul
7774 Use an in-line code sequence for integer multiplies. This is the
7777 @item -mcall-lib-mul
7778 @opindex mcall-lib-mul
7779 Call @code{lmul$$} for integer multiples.
7781 @item -mfull-fp-blocks
7782 @opindex mfull-fp-blocks
7783 Generate full-size floating point data blocks, including the minimum
7784 amount of scratch space recommended by IBM@. This is the default.
7786 @item -mminimum-fp-blocks
7787 @opindex mminimum-fp-blocks
7788 Do not include extra scratch space in floating point data blocks. This
7789 results in smaller code, but slower execution, since scratch space must
7790 be allocated dynamically.
7792 @cindex @file{stdarg.h} and RT PC
7793 @item -mfp-arg-in-fpregs
7794 @opindex mfp-arg-in-fpregs
7795 Use a calling sequence incompatible with the IBM calling convention in
7796 which floating point arguments are passed in floating point registers.
7797 Note that @code{stdarg.h} will not work with floating point operands
7798 if this option is specified.
7800 @item -mfp-arg-in-gregs
7801 @opindex mfp-arg-in-gregs
7802 Use the normal calling convention for floating point arguments. This is
7805 @item -mhc-struct-return
7806 @opindex mhc-struct-return
7807 Return structures of more than one word in memory, rather than in a
7808 register. This provides compatibility with the MetaWare HighC (hc)
7809 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7810 with the Portable C Compiler (pcc).
7812 @item -mnohc-struct-return
7813 @opindex mnohc-struct-return
7814 Return some structures of more than one word in registers, when
7815 convenient. This is the default. For compatibility with the
7816 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7817 option @option{-mhc-struct-return}.
7821 @subsection MIPS Options
7822 @cindex MIPS options
7824 These @samp{-m} options are defined for the MIPS family of computers:
7828 @item -march=@var{arch}
7830 Generate code that will run on @var{arch}, which can be the name of a
7831 generic MIPS ISA, or the name of a particular processor.
7833 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7834 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7835 The processor names are:
7836 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7838 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7839 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
7843 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4300},
7844 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
7845 The special value @samp{from-abi} selects the
7846 most compatible architecture for the selected ABI (that is,
7847 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7849 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7850 (for example, @samp{-march=r2k}). Prefixes are optional, and
7851 @samp{vr} may be written @samp{r}.
7853 GCC defines two macros based on the value of this option. The first
7854 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7855 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7856 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7857 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7858 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7860 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7861 above. In other words, it will have the full prefix and will not
7862 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7863 the macro names the resolved architecture (either @samp{"mips1"} or
7864 @samp{"mips3"}). It names the default architecture when no
7865 @option{-march} option is given.
7867 @item -mtune=@var{arch}
7869 Optimize for @var{arch}. Among other things, this option controls
7870 the way instructions are scheduled, and the perceived cost of arithmetic
7871 operations. The list of @var{arch} values is the same as for
7874 When this option is not used, GCC will optimize for the processor
7875 specified by @option{-march}. By using @option{-march} and
7876 @option{-mtune} together, it is possible to generate code that will
7877 run on a family of processors, but optimize the code for one
7878 particular member of that family.
7880 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7881 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7882 @samp{-march} ones described above.
7886 Equivalent to @samp{-march=mips1}.
7890 Equivalent to @samp{-march=mips2}.
7894 Equivalent to @samp{-march=mips3}.
7898 Equivalent to @samp{-march=mips4}.
7902 Equivalent to @samp{-march=mips32}.
7906 Equivalent to @samp{-march=mips32r2}.
7910 Equivalent to @samp{-march=mips64}.
7913 @itemx -mno-fused-madd
7914 @opindex mfused-madd
7915 @opindex mno-fused-madd
7916 Generate code that uses (does not use) the floating point multiply and
7917 accumulate instructions, when they are available. These instructions
7918 are generated by default if they are available, but this may be
7919 undesirable if the extra precision causes problems or on certain chips
7920 in the mode where denormals are rounded to zero where denormals
7921 generated by multiply and accumulate instructions cause exceptions
7926 Assume that floating point registers are 32 bits wide.
7930 Assume that floating point registers are 64 bits wide.
7934 Assume that general purpose registers are 32 bits wide.
7938 Assume that general purpose registers are 64 bits wide.
7942 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7943 explanation of the default, and the width of pointers.
7947 Force long types to be 64 bits wide. See @option{-mlong32} for an
7948 explanation of the default, and the width of pointers.
7952 Force long, int, and pointer types to be 32 bits wide.
7954 The default size of ints, longs and pointers depends on the ABI@. All
7955 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
7956 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
7957 are the same size as longs, or the same size as integer registers,
7958 whichever is smaller.
7972 Generate code for the given ABI@.
7974 Note that there are two embedded ABIs: @option{-mabi=eabi}
7975 selects the one defined by Cygnus while @option{-meabi=meabi}
7976 selects the one defined by MIPS@. Both these ABIs have
7977 32-bit and 64-bit variants. Normally, GCC will generate
7978 64-bit code when you select a 64-bit architecture, but you
7979 can use @option{-mgp32} to get 32-bit code instead.
7981 @item -mabi-fake-default
7982 @opindex mabi-fake-default
7983 You don't want to know what this option does. No, really. I mean
7984 it. Move on to the next option.
7986 What? You're still here? Oh, well@enddots{} Ok, here's the deal. GCC
7987 wants the default set of options to get the root of the multilib tree,
7988 and the shared library SONAMEs without any multilib-indicating
7989 suffixes. This is not convenience for @samp{mips64-linux-gnu}, since
7990 we want to default to the N32 ABI, while still being binary-compatible
7991 with @samp{mips-linux-gnu} if you stick to the O32 ABI@. Being
7992 binary-compatible means shared libraries should have the same SONAMEs,
7993 and libraries should live in the same location. Having O32 libraries
7994 in a sub-directory named say @file{o32} is not acceptable.
7996 So we trick GCC into believing that O32 is the default ABI, except
7997 that we override the default with some internal command-line
7998 processing magic. Problem is, if we stopped at that, and you then
7999 created a multilib-aware package that used the output of @command{gcc
8000 -print-multi-lib} to decide which multilibs to build, and how, and
8001 you'd find yourself in an awkward situation when you found out that
8002 some of the options listed ended up mapping to the same multilib, and
8003 none of your libraries was actually built for the multilib that
8004 @option{-print-multi-lib} claims to be the default. So we added this
8005 option that disables the default switcher, falling back to GCC's
8006 original notion of the default library. Confused yet?
8008 For short: don't ever use this option, unless you find it in the list
8009 of additional options to be used when building for multilibs, in the
8010 output of @option{gcc -print-multi-lib}.
8014 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
8015 add normal debug information. This is the default for all
8016 platforms except for the OSF/1 reference platform, using the OSF/rose
8017 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
8018 switches are used, the @file{mips-tfile} program will encapsulate the
8019 stabs within MIPS ECOFF@.
8023 Generate code for the GNU assembler. This is the default on the OSF/1
8024 reference platform, using the OSF/rose object format. Also, this is
8025 the default if the configure option @option{--with-gnu-as} is used.
8027 @item -msplit-addresses
8028 @itemx -mno-split-addresses
8029 @opindex msplit-addresses
8030 @opindex mno-split-addresses
8031 Generate code to load the high and low parts of address constants separately.
8032 This allows GCC to optimize away redundant loads of the high order
8033 bits of addresses. This optimization requires GNU as and GNU ld.
8034 This optimization is enabled by default for some embedded targets where
8035 GNU as and GNU ld are standard.
8041 The @option{-mrnames} switch says to output code using the MIPS software
8042 names for the registers, instead of the hardware names (ie, @var{a0}
8043 instead of @var{$4}). The only known assembler that supports this option
8044 is the Algorithmics assembler.
8050 The @option{-mmemcpy} switch makes all block moves call the appropriate
8051 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
8052 generating inline code.
8055 @itemx -mno-mips-tfile
8056 @opindex mmips-tfile
8057 @opindex mno-mips-tfile
8058 The @option{-mno-mips-tfile} switch causes the compiler not
8059 postprocess the object file with the @file{mips-tfile} program,
8060 after the MIPS assembler has generated it to add debug support. If
8061 @file{mips-tfile} is not run, then no local variables will be
8062 available to the debugger. In addition, @file{stage2} and
8063 @file{stage3} objects will have the temporary file names passed to the
8064 assembler embedded in the object file, which means the objects will
8065 not compare the same. The @option{-mno-mips-tfile} switch should only
8066 be used when there are bugs in the @file{mips-tfile} program that
8067 prevents compilation.
8070 @opindex msoft-float
8071 Generate output containing library calls for floating point.
8072 @strong{Warning:} the requisite libraries are not part of GCC@.
8073 Normally the facilities of the machine's usual C compiler are used, but
8074 this can't be done directly in cross-compilation. You must make your
8075 own arrangements to provide suitable library functions for
8079 @opindex mhard-float
8080 Generate output containing floating point instructions. This is the
8081 default if you use the unmodified sources.
8084 @itemx -mno-abicalls
8086 @opindex mno-abicalls
8087 Emit (or do not emit) the pseudo operations @samp{.abicalls},
8088 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
8089 position independent code.
8095 Lift (or do not lift) the usual restrictions on the size of the global
8098 GCC normally uses a single instruction to load values from the GOT.
8099 While this is relatively efficient, it will only work if the GOT
8100 is smaller than about 64k. Anything larger will cause the linker
8101 to report an error such as:
8103 @cindex relocation truncated to fit (MIPS)
8105 relocation truncated to fit: R_MIPS_GOT16 foobar
8108 If this happens, you should recompile your code with @option{-mxgot}.
8109 It should then work with very large GOTs, although it will also be
8110 less efficient, since it will take three instructions to fetch the
8111 value of a global symbol.
8113 Note that some linkers can create multiple GOTs. If you have such a
8114 linker, you should only need to use @option{-mxgot} when a single object
8115 file accesses more than 64k's worth of GOT entries. Very few do.
8117 These options have no effect unless GCC is generating position
8121 @itemx -mno-long-calls
8122 @opindex mlong-calls
8123 @opindex mno-long-calls
8124 Do all calls with the @samp{JALR} instruction, which requires
8125 loading up a function's address into a register before the call.
8126 You need to use this switch, if you call outside of the current
8127 512 megabyte segment to functions that are not through pointers.
8129 @item -membedded-pic
8130 @itemx -mno-embedded-pic
8131 @opindex membedded-pic
8132 @opindex mno-embedded-pic
8133 Generate PIC code suitable for some embedded systems. All calls are
8134 made using PC relative address, and all data is addressed using the $gp
8135 register. No more than 65536 bytes of global data may be used. This
8136 requires GNU as and GNU ld which do most of the work. This currently
8137 only works on targets which use ECOFF; it does not work with ELF@.
8139 @item -membedded-data
8140 @itemx -mno-embedded-data
8141 @opindex membedded-data
8142 @opindex mno-embedded-data
8143 Allocate variables to the read-only data section first if possible, then
8144 next in the small data section if possible, otherwise in data. This gives
8145 slightly slower code than the default, but reduces the amount of RAM required
8146 when executing, and thus may be preferred for some embedded systems.
8148 @item -muninit-const-in-rodata
8149 @itemx -mno-uninit-const-in-rodata
8150 @opindex muninit-const-in-rodata
8151 @opindex mno-uninit-const-in-rodata
8152 When used together with @option{-membedded-data}, it will always store uninitialized
8153 const variables in the read-only data section.
8155 @item -msingle-float
8156 @itemx -mdouble-float
8157 @opindex msingle-float
8158 @opindex mdouble-float
8159 The @option{-msingle-float} switch tells gcc to assume that the floating
8160 point coprocessor only supports single precision operations, as on the
8161 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
8162 double precision operations. This is the default.
8168 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
8169 as on the @samp{r4650} chip.
8173 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
8174 @option{-mcpu=r4650}.
8180 Enable 16-bit instructions.
8184 Use the entry and exit pseudo ops. This option can only be used with
8189 Compile code for the processor in little endian mode.
8190 The requisite libraries are assumed to exist.
8194 Compile code for the processor in big endian mode.
8195 The requisite libraries are assumed to exist.
8199 @cindex smaller data references (MIPS)
8200 @cindex gp-relative references (MIPS)
8201 Put global and static items less than or equal to @var{num} bytes into
8202 the small data or bss sections instead of the normal data or bss
8203 section. This allows the assembler to emit one word memory reference
8204 instructions based on the global pointer (@var{gp} or @var{$28}),
8205 instead of the normal two words used. By default, @var{num} is 8 when
8206 the MIPS assembler is used, and 0 when the GNU assembler is used. The
8207 @option{-G @var{num}} switch is also passed to the assembler and linker.
8208 All modules should be compiled with the same @option{-G @var{num}}
8213 Tell the MIPS assembler to not run its preprocessor over user
8214 assembler files (with a @samp{.s} suffix) when assembling them.
8218 Pass an option to gas which will cause nops to be inserted if
8219 the read of the destination register of an mfhi or mflo instruction
8220 occurs in the following two instructions.
8225 Work around certain SB-1 CPU core errata.
8226 (This flag currently works around the SB-1 revision 2
8227 ``F1'' and ``F2'' floating point errata.)
8231 Do not include the default crt0.
8233 @item -mflush-func=@var{func}
8234 @itemx -mno-flush-func
8235 @opindex mflush-func
8236 Specifies the function to call to flush the I and D caches, or to not
8237 call any such function. If called, the function must take the same
8238 arguments as the common @code{_flush_func()}, that is, the address of the
8239 memory range for which the cache is being flushed, the size of the
8240 memory range, and the number 3 (to flush both caches). The default
8241 depends on the target gcc was configured for, but commonly is either
8242 @samp{_flush_func} or @samp{__cpu_flush}.
8244 @item -mbranch-likely
8245 @itemx -mno-branch-likely
8246 @opindex mbranch-likely
8247 @opindex mno-branch-likely
8248 Enable or disable use of Branch Likely instructions, regardless of the
8249 default for the selected architecture. By default, Branch Likely
8250 instructions may be generated if they are supported by the selected
8251 architecture. An exception is for the MIPS32 and MIPS64 architectures
8252 and processors which implement those architectures; for those, Branch
8253 Likely instructions will not be generated by default because the MIPS32
8254 and MIPS64 architectures specifically deprecate their use.
8257 @node i386 and x86-64 Options
8258 @subsection Intel 386 and AMD x86-64 Options
8259 @cindex i386 Options
8260 @cindex x86-64 Options
8261 @cindex Intel 386 Options
8262 @cindex AMD x86-64 Options
8264 These @samp{-m} options are defined for the i386 and x86-64 family of
8268 @item -mtune=@var{cpu-type}
8270 Tune to @var{cpu-type} everything applicable about the generated code, except
8271 for the ABI and the set of available instructions. The choices for
8272 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
8273 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
8274 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
8275 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
8276 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8}, @samp{c3}
8279 While picking a specific @var{cpu-type} will schedule things appropriately
8280 for that particular chip, the compiler will not generate any code that
8281 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8282 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
8283 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
8284 AMD chips as opposed to the Intel ones.
8286 @item -march=@var{cpu-type}
8288 Generate instructions for the machine type @var{cpu-type}. The choices
8289 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8290 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8292 @item -mcpu=@var{cpu-type}
8294 A deprecated synonym for @option{-mtune}.
8303 @opindex mpentiumpro
8304 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8305 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8306 These synonyms are deprecated.
8308 @item -mfpmath=@var{unit}
8310 generate floating point arithmetics for selected unit @var{unit}. the choices
8315 Use the standard 387 floating point coprocessor present majority of chips and
8316 emulated otherwise. Code compiled with this option will run almost everywhere.
8317 The temporary results are computed in 80bit precision instead of precision
8318 specified by the type resulting in slightly different results compared to most
8319 of other chips. See @option{-ffloat-store} for more detailed description.
8321 This is the default choice for i386 compiler.
8324 Use scalar floating point instructions present in the SSE instruction set.
8325 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8326 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8327 instruction set supports only single precision arithmetics, thus the double and
8328 extended precision arithmetics is still done using 387. Later version, present
8329 only in Pentium4 and the future AMD x86-64 chips supports double precision
8332 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8333 @option{-msse2} switches to enable SSE extensions and make this option
8334 effective. For x86-64 compiler, these extensions are enabled by default.
8336 The resulting code should be considerably faster in majority of cases and avoid
8337 the numerical instability problems of 387 code, but may break some existing
8338 code that expects temporaries to be 80bit.
8340 This is the default choice for x86-64 compiler.
8343 Use all SSE extensions enabled by @option{-msse2} as well as the new
8344 SSE extensions in Prescott New Instructions. @option{-mpni} also
8345 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8346 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8350 Attempt to utilize both instruction sets at once. This effectively double the
8351 amount of available registers and on chips with separate execution units for
8352 387 and SSE the execution resources too. Use this option with care, as it is
8353 still experimental, because gcc register allocator does not model separate
8354 functional units well resulting in instable performance.
8357 @item -masm=@var{dialect}
8358 @opindex masm=@var{dialect}
8359 Output asm instructions using selected @var{dialect}. Supported choices are
8360 @samp{intel} or @samp{att} (the default one).
8365 @opindex mno-ieee-fp
8366 Control whether or not the compiler uses IEEE floating point
8367 comparisons. These handle correctly the case where the result of a
8368 comparison is unordered.
8371 @opindex msoft-float
8372 Generate output containing library calls for floating point.
8373 @strong{Warning:} the requisite libraries are not part of GCC@.
8374 Normally the facilities of the machine's usual C compiler are used, but
8375 this can't be done directly in cross-compilation. You must make your
8376 own arrangements to provide suitable library functions for
8379 On machines where a function returns floating point results in the 80387
8380 register stack, some floating point opcodes may be emitted even if
8381 @option{-msoft-float} is used.
8383 @item -mno-fp-ret-in-387
8384 @opindex mno-fp-ret-in-387
8385 Do not use the FPU registers for return values of functions.
8387 The usual calling convention has functions return values of types
8388 @code{float} and @code{double} in an FPU register, even if there
8389 is no FPU@. The idea is that the operating system should emulate
8392 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8393 in ordinary CPU registers instead.
8395 @item -mno-fancy-math-387
8396 @opindex mno-fancy-math-387
8397 Some 387 emulators do not support the @code{sin}, @code{cos} and
8398 @code{sqrt} instructions for the 387. Specify this option to avoid
8399 generating those instructions. This option is the default on FreeBSD,
8400 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8401 indicates that the target cpu will always have an FPU and so the
8402 instruction will not need emulation. As of revision 2.6.1, these
8403 instructions are not generated unless you also use the
8404 @option{-funsafe-math-optimizations} switch.
8406 @item -malign-double
8407 @itemx -mno-align-double
8408 @opindex malign-double
8409 @opindex mno-align-double
8410 Control whether GCC aligns @code{double}, @code{long double}, and
8411 @code{long long} variables on a two word boundary or a one word
8412 boundary. Aligning @code{double} variables on a two word boundary will
8413 produce code that runs somewhat faster on a @samp{Pentium} at the
8414 expense of more memory.
8416 @strong{Warning:} if you use the @option{-malign-double} switch,
8417 structures containing the above types will be aligned differently than
8418 the published application binary interface specifications for the 386
8419 and will not be binary compatible with structures in code compiled
8420 without that switch.
8422 @item -m96bit-long-double
8423 @item -m128bit-long-double
8424 @opindex m96bit-long-double
8425 @opindex m128bit-long-double
8426 These switches control the size of @code{long double} type. The i386
8427 application binary interface specifies the size to be 96 bits,
8428 so @option{-m96bit-long-double} is the default in 32 bit mode.
8430 Modern architectures (Pentium and newer) would prefer @code{long double}
8431 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8432 conforming to the ABI, this would not be possible. So specifying a
8433 @option{-m128bit-long-double} will align @code{long double}
8434 to a 16 byte boundary by padding the @code{long double} with an additional
8437 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8438 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8440 Notice that neither of these options enable any extra precision over the x87
8441 standard of 80 bits for a @code{long double}.
8443 @strong{Warning:} if you override the default value for your target ABI, the
8444 structures and arrays containing @code{long double} will change their size as
8445 well as function calling convention for function taking @code{long double}
8446 will be modified. Hence they will not be binary compatible with arrays or
8447 structures in code compiled without that switch.
8451 @itemx -mno-svr3-shlib
8452 @opindex msvr3-shlib
8453 @opindex mno-svr3-shlib
8454 Control whether GCC places uninitialized local variables into the
8455 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8456 into @code{bss}. These options are meaningful only on System V Release 3.
8460 Use a different function-calling convention, in which functions that
8461 take a fixed number of arguments return with the @code{ret} @var{num}
8462 instruction, which pops their arguments while returning. This saves one
8463 instruction in the caller since there is no need to pop the arguments
8466 You can specify that an individual function is called with this calling
8467 sequence with the function attribute @samp{stdcall}. You can also
8468 override the @option{-mrtd} option by using the function attribute
8469 @samp{cdecl}. @xref{Function Attributes}.
8471 @strong{Warning:} this calling convention is incompatible with the one
8472 normally used on Unix, so you cannot use it if you need to call
8473 libraries compiled with the Unix compiler.
8475 Also, you must provide function prototypes for all functions that
8476 take variable numbers of arguments (including @code{printf});
8477 otherwise incorrect code will be generated for calls to those
8480 In addition, seriously incorrect code will result if you call a
8481 function with too many arguments. (Normally, extra arguments are
8482 harmlessly ignored.)
8484 @item -mregparm=@var{num}
8486 Control how many registers are used to pass integer arguments. By
8487 default, no registers are used to pass arguments, and at most 3
8488 registers can be used. You can control this behavior for a specific
8489 function by using the function attribute @samp{regparm}.
8490 @xref{Function Attributes}.
8492 @strong{Warning:} if you use this switch, and
8493 @var{num} is nonzero, then you must build all modules with the same
8494 value, including any libraries. This includes the system libraries and
8497 @item -mpreferred-stack-boundary=@var{num}
8498 @opindex mpreferred-stack-boundary
8499 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8500 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8501 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8502 size (@option{-Os}), in which case the default is the minimum correct
8503 alignment (4 bytes for x86, and 8 bytes for x86-64).
8505 On Pentium and PentiumPro, @code{double} and @code{long double} values
8506 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8507 suffer significant run time performance penalties. On Pentium III, the
8508 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8509 penalties if it is not 16 byte aligned.
8511 To ensure proper alignment of this values on the stack, the stack boundary
8512 must be as aligned as that required by any value stored on the stack.
8513 Further, every function must be generated such that it keeps the stack
8514 aligned. Thus calling a function compiled with a higher preferred
8515 stack boundary from a function compiled with a lower preferred stack
8516 boundary will most likely misalign the stack. It is recommended that
8517 libraries that use callbacks always use the default setting.
8519 This extra alignment does consume extra stack space, and generally
8520 increases code size. Code that is sensitive to stack space usage, such
8521 as embedded systems and operating system kernels, may want to reduce the
8522 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8540 These switches enable or disable the use of built-in functions that allow
8541 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8543 @xref{X86 Built-in Functions}, for details of the functions enabled
8544 and disabled by these switches.
8546 To have SSE/SSE2 instructions generated automatically from floating-point
8547 code, see @option{-mfpmath=sse}.
8550 @itemx -mno-push-args
8552 @opindex mno-push-args
8553 Use PUSH operations to store outgoing parameters. This method is shorter
8554 and usually equally fast as method using SUB/MOV operations and is enabled
8555 by default. In some cases disabling it may improve performance because of
8556 improved scheduling and reduced dependencies.
8558 @item -maccumulate-outgoing-args
8559 @opindex maccumulate-outgoing-args
8560 If enabled, the maximum amount of space required for outgoing arguments will be
8561 computed in the function prologue. This is faster on most modern CPUs
8562 because of reduced dependencies, improved scheduling and reduced stack usage
8563 when preferred stack boundary is not equal to 2. The drawback is a notable
8564 increase in code size. This switch implies @option{-mno-push-args}.
8568 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8569 on thread-safe exception handling must compile and link all code with the
8570 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8571 @option{-D_MT}; when linking, it links in a special thread helper library
8572 @option{-lmingwthrd} which cleans up per thread exception handling data.
8574 @item -mno-align-stringops
8575 @opindex mno-align-stringops
8576 Do not align destination of inlined string operations. This switch reduces
8577 code size and improves performance in case the destination is already aligned,
8578 but gcc don't know about it.
8580 @item -minline-all-stringops
8581 @opindex minline-all-stringops
8582 By default GCC inlines string operations only when destination is known to be
8583 aligned at least to 4 byte boundary. This enables more inlining, increase code
8584 size, but may improve performance of code that depends on fast memcpy, strlen
8585 and memset for short lengths.
8587 @item -momit-leaf-frame-pointer
8588 @opindex momit-leaf-frame-pointer
8589 Don't keep the frame pointer in a register for leaf functions. This
8590 avoids the instructions to save, set up and restore frame pointers and
8591 makes an extra register available in leaf functions. The option
8592 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8593 which might make debugging harder.
8595 @item -mtls-direct-seg-refs
8596 @itemx -mno-tls-direct-seg-refs
8597 @opindex mtls-direct-seg-refs
8598 Controls whether TLS variables may be accessed with offsets from the
8599 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8600 or whether the thread base pointer must be added. Whether or not this
8601 is legal depends on the operating system, and whether it maps the
8602 segment to cover the entire TLS area.
8604 For systems that use GNU libc, the default is on.
8607 These @samp{-m} switches are supported in addition to the above
8608 on AMD x86-64 processors in 64-bit environments.
8615 Generate code for a 32-bit or 64-bit environment.
8616 The 32-bit environment sets int, long and pointer to 32 bits and
8617 generates code that runs on any i386 system.
8618 The 64-bit environment sets int to 32 bits and long and pointer
8619 to 64 bits and generates code for AMD's x86-64 architecture.
8622 @opindex no-red-zone
8623 Do not use a so called red zone for x86-64 code. The red zone is mandated
8624 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8625 stack pointer that will not be modified by signal or interrupt handlers
8626 and therefore can be used for temporary data without adjusting the stack
8627 pointer. The flag @option{-mno-red-zone} disables this red zone.
8629 @item -mcmodel=small
8630 @opindex mcmodel=small
8631 Generate code for the small code model: the program and its symbols must
8632 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8633 Programs can be statically or dynamically linked. This is the default
8636 @item -mcmodel=kernel
8637 @opindex mcmodel=kernel
8638 Generate code for the kernel code model. The kernel runs in the
8639 negative 2 GB of the address space.
8640 This model has to be used for Linux kernel code.
8642 @item -mcmodel=medium
8643 @opindex mcmodel=medium
8644 Generate code for the medium model: The program is linked in the lower 2
8645 GB of the address space but symbols can be located anywhere in the
8646 address space. Programs can be statically or dynamically linked, but
8647 building of shared libraries are not supported with the medium model.
8649 @item -mcmodel=large
8650 @opindex mcmodel=large
8651 Generate code for the large model: This model makes no assumptions
8652 about addresses and sizes of sections. Currently GCC does not implement
8657 @subsection HPPA Options
8658 @cindex HPPA Options
8660 These @samp{-m} options are defined for the HPPA family of computers:
8663 @item -march=@var{architecture-type}
8665 Generate code for the specified architecture. The choices for
8666 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8667 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8668 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8669 architecture option for your machine. Code compiled for lower numbered
8670 architectures will run on higher numbered architectures, but not the
8673 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8674 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8678 @itemx -mpa-risc-1-1
8679 @itemx -mpa-risc-2-0
8680 @opindex mpa-risc-1-0
8681 @opindex mpa-risc-1-1
8682 @opindex mpa-risc-2-0
8683 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8686 @opindex mbig-switch
8687 Generate code suitable for big switch tables. Use this option only if
8688 the assembler/linker complain about out of range branches within a switch
8691 @item -mjump-in-delay
8692 @opindex mjump-in-delay
8693 Fill delay slots of function calls with unconditional jump instructions
8694 by modifying the return pointer for the function call to be the target
8695 of the conditional jump.
8697 @item -mdisable-fpregs
8698 @opindex mdisable-fpregs
8699 Prevent floating point registers from being used in any manner. This is
8700 necessary for compiling kernels which perform lazy context switching of
8701 floating point registers. If you use this option and attempt to perform
8702 floating point operations, the compiler will abort.
8704 @item -mdisable-indexing
8705 @opindex mdisable-indexing
8706 Prevent the compiler from using indexing address modes. This avoids some
8707 rather obscure problems when compiling MIG generated code under MACH@.
8709 @item -mno-space-regs
8710 @opindex mno-space-regs
8711 Generate code that assumes the target has no space registers. This allows
8712 GCC to generate faster indirect calls and use unscaled index address modes.
8714 Such code is suitable for level 0 PA systems and kernels.
8716 @item -mfast-indirect-calls
8717 @opindex mfast-indirect-calls
8718 Generate code that assumes calls never cross space boundaries. This
8719 allows GCC to emit code which performs faster indirect calls.
8721 This option will not work in the presence of shared libraries or nested
8724 @item -mlong-load-store
8725 @opindex mlong-load-store
8726 Generate 3-instruction load and store sequences as sometimes required by
8727 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8730 @item -mportable-runtime
8731 @opindex mportable-runtime
8732 Use the portable calling conventions proposed by HP for ELF systems.
8736 Enable the use of assembler directives only GAS understands.
8738 @item -mschedule=@var{cpu-type}
8740 Schedule code according to the constraints for the machine type
8741 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8742 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8743 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8744 proper scheduling option for your machine. The default scheduling is
8748 @opindex mlinker-opt
8749 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8750 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8751 linkers in which they give bogus error messages when linking some programs.
8754 @opindex msoft-float
8755 Generate output containing library calls for floating point.
8756 @strong{Warning:} the requisite libraries are not available for all HPPA
8757 targets. Normally the facilities of the machine's usual C compiler are
8758 used, but this cannot be done directly in cross-compilation. You must make
8759 your own arrangements to provide suitable library functions for
8760 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8761 does provide software floating point support.
8763 @option{-msoft-float} changes the calling convention in the output file;
8764 therefore, it is only useful if you compile @emph{all} of a program with
8765 this option. In particular, you need to compile @file{libgcc.a}, the
8766 library that comes with GCC, with @option{-msoft-float} in order for
8771 Generate the predefine, @code{_SIO}, for server IO. The default is
8772 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8773 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8774 options are available under HP-UX and HI-UX.
8778 Use GNU ld specific options. This passes @option{-shared} to ld when
8779 building a shared library. It is the default when GCC is configured,
8780 explicitly or implicitly, with the GNU linker. This option does not
8781 have any affect on which ld is called, it only changes what parameters
8782 are passed to that ld. The ld that is called is determined by the
8783 @option{--with-ld} configure option, gcc's program search path, and
8784 finally by the user's @env{PATH}. The linker used by GCC can be printed
8785 using @samp{which `gcc -print-prog-name=ld`}.
8789 Use HP ld specific options. This passes @option{-b} to ld when building
8790 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8791 links. It is the default when GCC is configured, explicitly or
8792 implicitly, with the HP linker. This option does not have any affect on
8793 which ld is called, it only changes what parameters are passed to that
8794 ld. The ld that is called is determined by the @option{--with-ld}
8795 configure option, gcc's program search path, and finally by the user's
8796 @env{PATH}. The linker used by GCC can be printed using @samp{which
8797 `gcc -print-prog-name=ld`}.
8800 @opindex mno-long-calls
8801 Generate code that uses long call sequences. This ensures that a call
8802 is always able to reach linker generated stubs. The default is to generate
8803 long calls only when the distance from the call site to the beginning
8804 of the function or translation unit, as the case may be, exceeds a
8805 predefined limit set by the branch type being used. The limits for
8806 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8807 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8810 Distances are measured from the beginning of functions when using the
8811 @option{-ffunction-sections} option, or when using the @option{-mgas}
8812 and @option{-mno-portable-runtime} options together under HP-UX with
8815 It is normally not desirable to use this option as it will degrade
8816 performance. However, it may be useful in large applications,
8817 particularly when partial linking is used to build the application.
8819 The types of long calls used depends on the capabilities of the
8820 assembler and linker, and the type of code being generated. The
8821 impact on systems that support long absolute calls, and long pic
8822 symbol-difference or pc-relative calls should be relatively small.
8823 However, an indirect call is used on 32-bit ELF systems in pic code
8824 and it is quite long.
8828 Suppress the generation of link options to search libdld.sl when the
8829 @option{-static} option is specified on HP-UX 10 and later.
8833 The HP-UX implementation of setlocale in libc has a dependency on
8834 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8835 when the @option{-static} option is specified, special link options
8836 are needed to resolve this dependency.
8838 On HP-UX 10 and later, the GCC driver adds the necessary options to
8839 link with libdld.sl when the @option{-static} option is specified.
8840 This causes the resulting binary to be dynamic. On the 64-bit port,
8841 the linkers generate dynamic binaries by default in any case. The
8842 @option{-nolibdld} option can be used to prevent the GCC driver from
8843 adding these link options.
8847 Add support for multithreading with the @dfn{dce thread} library
8848 under HP-UX. This option sets flags for both the preprocessor and
8852 @node Intel 960 Options
8853 @subsection Intel 960 Options
8855 These @samp{-m} options are defined for the Intel 960 implementations:
8858 @item -m@var{cpu-type}
8866 Assume the defaults for the machine type @var{cpu-type} for some of
8867 the other options, including instruction scheduling, floating point
8868 support, and addressing modes. The choices for @var{cpu-type} are
8869 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8870 @samp{sa}, and @samp{sb}.
8877 @opindex msoft-float
8878 The @option{-mnumerics} option indicates that the processor does support
8879 floating-point instructions. The @option{-msoft-float} option indicates
8880 that floating-point support should not be assumed.
8882 @item -mleaf-procedures
8883 @itemx -mno-leaf-procedures
8884 @opindex mleaf-procedures
8885 @opindex mno-leaf-procedures
8886 Do (or do not) attempt to alter leaf procedures to be callable with the
8887 @code{bal} instruction as well as @code{call}. This will result in more
8888 efficient code for explicit calls when the @code{bal} instruction can be
8889 substituted by the assembler or linker, but less efficient code in other
8890 cases, such as calls via function pointers, or using a linker that doesn't
8891 support this optimization.
8894 @itemx -mno-tail-call
8896 @opindex mno-tail-call
8897 Do (or do not) make additional attempts (beyond those of the
8898 machine-independent portions of the compiler) to optimize tail-recursive
8899 calls into branches. You may not want to do this because the detection of
8900 cases where this is not valid is not totally complete. The default is
8901 @option{-mno-tail-call}.
8903 @item -mcomplex-addr
8904 @itemx -mno-complex-addr
8905 @opindex mcomplex-addr
8906 @opindex mno-complex-addr
8907 Assume (or do not assume) that the use of a complex addressing mode is a
8908 win on this implementation of the i960. Complex addressing modes may not
8909 be worthwhile on the K-series, but they definitely are on the C-series.
8910 The default is currently @option{-mcomplex-addr} for all processors except
8914 @itemx -mno-code-align
8915 @opindex mcode-align
8916 @opindex mno-code-align
8917 Align code to 8-byte boundaries for faster fetching (or don't bother).
8918 Currently turned on by default for C-series implementations only.
8921 @item -mclean-linkage
8922 @itemx -mno-clean-linkage
8923 @opindex mclean-linkage
8924 @opindex mno-clean-linkage
8925 These options are not fully implemented.
8929 @itemx -mic2.0-compat
8930 @itemx -mic3.0-compat
8932 @opindex mic2.0-compat
8933 @opindex mic3.0-compat
8934 Enable compatibility with iC960 v2.0 or v3.0.
8938 @opindex masm-compat
8940 Enable compatibility with the iC960 assembler.
8942 @item -mstrict-align
8943 @itemx -mno-strict-align
8944 @opindex mstrict-align
8945 @opindex mno-strict-align
8946 Do not permit (do permit) unaligned accesses.
8950 Enable structure-alignment compatibility with Intel's gcc release version
8951 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8953 @item -mlong-double-64
8954 @opindex mlong-double-64
8955 Implement type @samp{long double} as 64-bit floating point numbers.
8956 Without the option @samp{long double} is implemented by 80-bit
8957 floating point numbers. The only reason we have it because there is
8958 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8959 is only useful for people using soft-float targets. Otherwise, we
8960 should recommend against use of it.
8964 @node DEC Alpha Options
8965 @subsection DEC Alpha Options
8967 These @samp{-m} options are defined for the DEC Alpha implementations:
8970 @item -mno-soft-float
8972 @opindex mno-soft-float
8973 @opindex msoft-float
8974 Use (do not use) the hardware floating-point instructions for
8975 floating-point operations. When @option{-msoft-float} is specified,
8976 functions in @file{libgcc.a} will be used to perform floating-point
8977 operations. Unless they are replaced by routines that emulate the
8978 floating-point operations, or compiled in such a way as to call such
8979 emulations routines, these routines will issue floating-point
8980 operations. If you are compiling for an Alpha without floating-point
8981 operations, you must ensure that the library is built so as not to call
8984 Note that Alpha implementations without floating-point operations are
8985 required to have floating-point registers.
8990 @opindex mno-fp-regs
8991 Generate code that uses (does not use) the floating-point register set.
8992 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8993 register set is not used, floating point operands are passed in integer
8994 registers as if they were integers and floating-point results are passed
8995 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8996 so any function with a floating-point argument or return value called by code
8997 compiled with @option{-mno-fp-regs} must also be compiled with that
9000 A typical use of this option is building a kernel that does not use,
9001 and hence need not save and restore, any floating-point registers.
9005 The Alpha architecture implements floating-point hardware optimized for
9006 maximum performance. It is mostly compliant with the IEEE floating
9007 point standard. However, for full compliance, software assistance is
9008 required. This option generates code fully IEEE compliant code
9009 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9010 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9011 defined during compilation. The resulting code is less efficient but is
9012 able to correctly support denormalized numbers and exceptional IEEE
9013 values such as not-a-number and plus/minus infinity. Other Alpha
9014 compilers call this option @option{-ieee_with_no_inexact}.
9016 @item -mieee-with-inexact
9017 @opindex mieee-with-inexact
9018 This is like @option{-mieee} except the generated code also maintains
9019 the IEEE @var{inexact-flag}. Turning on this option causes the
9020 generated code to implement fully-compliant IEEE math. In addition to
9021 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9022 macro. On some Alpha implementations the resulting code may execute
9023 significantly slower than the code generated by default. Since there is
9024 very little code that depends on the @var{inexact-flag}, you should
9025 normally not specify this option. Other Alpha compilers call this
9026 option @option{-ieee_with_inexact}.
9028 @item -mfp-trap-mode=@var{trap-mode}
9029 @opindex mfp-trap-mode
9030 This option controls what floating-point related traps are enabled.
9031 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9032 The trap mode can be set to one of four values:
9036 This is the default (normal) setting. The only traps that are enabled
9037 are the ones that cannot be disabled in software (e.g., division by zero
9041 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9045 Like @samp{su}, but the instructions are marked to be safe for software
9046 completion (see Alpha architecture manual for details).
9049 Like @samp{su}, but inexact traps are enabled as well.
9052 @item -mfp-rounding-mode=@var{rounding-mode}
9053 @opindex mfp-rounding-mode
9054 Selects the IEEE rounding mode. Other Alpha compilers call this option
9055 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9060 Normal IEEE rounding mode. Floating point numbers are rounded towards
9061 the nearest machine number or towards the even machine number in case
9065 Round towards minus infinity.
9068 Chopped rounding mode. Floating point numbers are rounded towards zero.
9071 Dynamic rounding mode. A field in the floating point control register
9072 (@var{fpcr}, see Alpha architecture reference manual) controls the
9073 rounding mode in effect. The C library initializes this register for
9074 rounding towards plus infinity. Thus, unless your program modifies the
9075 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9078 @item -mtrap-precision=@var{trap-precision}
9079 @opindex mtrap-precision
9080 In the Alpha architecture, floating point traps are imprecise. This
9081 means without software assistance it is impossible to recover from a
9082 floating trap and program execution normally needs to be terminated.
9083 GCC can generate code that can assist operating system trap handlers
9084 in determining the exact location that caused a floating point trap.
9085 Depending on the requirements of an application, different levels of
9086 precisions can be selected:
9090 Program precision. This option is the default and means a trap handler
9091 can only identify which program caused a floating point exception.
9094 Function precision. The trap handler can determine the function that
9095 caused a floating point exception.
9098 Instruction precision. The trap handler can determine the exact
9099 instruction that caused a floating point exception.
9102 Other Alpha compilers provide the equivalent options called
9103 @option{-scope_safe} and @option{-resumption_safe}.
9105 @item -mieee-conformant
9106 @opindex mieee-conformant
9107 This option marks the generated code as IEEE conformant. You must not
9108 use this option unless you also specify @option{-mtrap-precision=i} and either
9109 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9110 is to emit the line @samp{.eflag 48} in the function prologue of the
9111 generated assembly file. Under DEC Unix, this has the effect that
9112 IEEE-conformant math library routines will be linked in.
9114 @item -mbuild-constants
9115 @opindex mbuild-constants
9116 Normally GCC examines a 32- or 64-bit integer constant to
9117 see if it can construct it from smaller constants in two or three
9118 instructions. If it cannot, it will output the constant as a literal and
9119 generate code to load it from the data segment at runtime.
9121 Use this option to require GCC to construct @emph{all} integer constants
9122 using code, even if it takes more instructions (the maximum is six).
9124 You would typically use this option to build a shared library dynamic
9125 loader. Itself a shared library, it must relocate itself in memory
9126 before it can find the variables and constants in its own data segment.
9132 Select whether to generate code to be assembled by the vendor-supplied
9133 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9151 Indicate whether GCC should generate code to use the optional BWX,
9152 CIX, FIX and MAX instruction sets. The default is to use the instruction
9153 sets supported by the CPU type specified via @option{-mcpu=} option or that
9154 of the CPU on which GCC was built if none was specified.
9159 @opindex mfloat-ieee
9160 Generate code that uses (does not use) VAX F and G floating point
9161 arithmetic instead of IEEE single and double precision.
9163 @item -mexplicit-relocs
9164 @itemx -mno-explicit-relocs
9165 @opindex mexplicit-relocs
9166 @opindex mno-explicit-relocs
9167 Older Alpha assemblers provided no way to generate symbol relocations
9168 except via assembler macros. Use of these macros does not allow
9169 optimal instruction scheduling. GNU binutils as of version 2.12
9170 supports a new syntax that allows the compiler to explicitly mark
9171 which relocations should apply to which instructions. This option
9172 is mostly useful for debugging, as GCC detects the capabilities of
9173 the assembler when it is built and sets the default accordingly.
9177 @opindex msmall-data
9178 @opindex mlarge-data
9179 When @option{-mexplicit-relocs} is in effect, static data is
9180 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9181 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9182 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9183 16-bit relocations off of the @code{$gp} register. This limits the
9184 size of the small data area to 64KB, but allows the variables to be
9185 directly accessed via a single instruction.
9187 The default is @option{-mlarge-data}. With this option the data area
9188 is limited to just below 2GB. Programs that require more than 2GB of
9189 data must use @code{malloc} or @code{mmap} to allocate the data in the
9190 heap instead of in the program's data segment.
9192 When generating code for shared libraries, @option{-fpic} implies
9193 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9197 @opindex msmall-text
9198 @opindex mlarge-text
9199 When @option{-msmall-text} is used, the compiler assumes that the
9200 code of the entire program (or shared library) fits in 4MB, and is
9201 thus reachable with a branch instruction. When @option{-msmall-data}
9202 is used, the compiler can assume that all local symbols share the
9203 same @code{$gp} value, and thus reduce the number of instructions
9204 required for a function call from 4 to 1.
9206 The default is @option{-mlarge-text}.
9208 @item -mcpu=@var{cpu_type}
9210 Set the instruction set and instruction scheduling parameters for
9211 machine type @var{cpu_type}. You can specify either the @samp{EV}
9212 style name or the corresponding chip number. GCC supports scheduling
9213 parameters for the EV4, EV5 and EV6 family of processors and will
9214 choose the default values for the instruction set from the processor
9215 you specify. If you do not specify a processor type, GCC will default
9216 to the processor on which the compiler was built.
9218 Supported values for @var{cpu_type} are
9224 Schedules as an EV4 and has no instruction set extensions.
9228 Schedules as an EV5 and has no instruction set extensions.
9232 Schedules as an EV5 and supports the BWX extension.
9237 Schedules as an EV5 and supports the BWX and MAX extensions.
9241 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9245 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9248 @item -mtune=@var{cpu_type}
9250 Set only the instruction scheduling parameters for machine type
9251 @var{cpu_type}. The instruction set is not changed.
9253 @item -mmemory-latency=@var{time}
9254 @opindex mmemory-latency
9255 Sets the latency the scheduler should assume for typical memory
9256 references as seen by the application. This number is highly
9257 dependent on the memory access patterns used by the application
9258 and the size of the external cache on the machine.
9260 Valid options for @var{time} are
9264 A decimal number representing clock cycles.
9270 The compiler contains estimates of the number of clock cycles for
9271 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9272 (also called Dcache, Scache, and Bcache), as well as to main memory.
9273 Note that L3 is only valid for EV5.
9278 @node DEC Alpha/VMS Options
9279 @subsection DEC Alpha/VMS Options
9281 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9284 @item -mvms-return-codes
9285 @opindex mvms-return-codes
9286 Return VMS condition codes from main. The default is to return POSIX
9287 style condition (e.g.@ error) codes.
9290 @node H8/300 Options
9291 @subsection H8/300 Options
9293 These @samp{-m} options are defined for the H8/300 implementations:
9298 Shorten some address references at link time, when possible; uses the
9299 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9300 ld.info, Using ld}, for a fuller description.
9304 Generate code for the H8/300H@.
9308 Generate code for the H8S@.
9312 Generate code for the H8S and H8/300H in the normal mode. This switch
9313 must be used either with -mh or -ms.
9317 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9321 Make @code{int} data 32 bits by default.
9325 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9326 The default for the H8/300H and H8S is to align longs and floats on 4
9328 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9329 This option has no effect on the H8/300.
9333 @subsection SH Options
9335 These @samp{-m} options are defined for the SH implementations:
9340 Generate code for the SH1.
9344 Generate code for the SH2.
9347 Generate code for the SH2e.
9351 Generate code for the SH3.
9355 Generate code for the SH3e.
9359 Generate code for the SH4 without a floating-point unit.
9361 @item -m4-single-only
9362 @opindex m4-single-only
9363 Generate code for the SH4 with a floating-point unit that only
9364 supports single-precision arithmetic.
9368 Generate code for the SH4 assuming the floating-point unit is in
9369 single-precision mode by default.
9373 Generate code for the SH4.
9377 Compile code for the processor in big endian mode.
9381 Compile code for the processor in little endian mode.
9385 Align doubles at 64-bit boundaries. Note that this changes the calling
9386 conventions, and thus some functions from the standard C library will
9387 not work unless you recompile it first with @option{-mdalign}.
9391 Shorten some address references at link time, when possible; uses the
9392 linker option @option{-relax}.
9396 Use 32-bit offsets in @code{switch} tables. The default is to use
9401 Enable the use of the instruction @code{fmovd}.
9405 Comply with the calling conventions defined by Renesas.
9409 Mark the @code{MAC} register as call-clobbered, even if
9410 @option{-mhitachi} is given.
9414 Increase IEEE-compliance of floating-point code.
9418 Dump instruction size and location in the assembly code.
9422 This option is deprecated. It pads structures to multiple of 4 bytes,
9423 which is incompatible with the SH ABI@.
9427 Optimize for space instead of speed. Implied by @option{-Os}.
9431 When generating position-independent code, emit function calls using
9432 the Global Offset Table instead of the Procedure Linkage Table.
9436 Generate a library function call to invalidate instruction cache
9437 entries, after fixing up a trampoline. This library function call
9438 doesn't assume it can write to the whole memory address space. This
9439 is the default when the target is @code{sh-*-linux*}.
9442 @node System V Options
9443 @subsection Options for System V
9445 These additional options are available on System V Release 4 for
9446 compatibility with other compilers on those systems:
9451 Create a shared object.
9452 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9456 Identify the versions of each tool used by the compiler, in a
9457 @code{.ident} assembler directive in the output.
9461 Refrain from adding @code{.ident} directives to the output file (this is
9464 @item -YP,@var{dirs}
9466 Search the directories @var{dirs}, and no others, for libraries
9467 specified with @option{-l}.
9471 Look in the directory @var{dir} to find the M4 preprocessor.
9472 The assembler uses this option.
9473 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9474 @c the generic assembler that comes with Solaris takes just -Ym.
9477 @node TMS320C3x/C4x Options
9478 @subsection TMS320C3x/C4x Options
9479 @cindex TMS320C3x/C4x Options
9481 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9485 @item -mcpu=@var{cpu_type}
9487 Set the instruction set, register set, and instruction scheduling
9488 parameters for machine type @var{cpu_type}. Supported values for
9489 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9490 @samp{c44}. The default is @samp{c40} to generate code for the
9495 @itemx -msmall-memory
9497 @opindex mbig-memory
9499 @opindex msmall-memory
9501 Generates code for the big or small memory model. The small memory
9502 model assumed that all data fits into one 64K word page. At run-time
9503 the data page (DP) register must be set to point to the 64K page
9504 containing the .bss and .data program sections. The big memory model is
9505 the default and requires reloading of the DP register for every direct
9512 Allow (disallow) allocation of general integer operands into the block
9519 Enable (disable) generation of code using decrement and branch,
9520 DBcond(D), instructions. This is enabled by default for the C4x. To be
9521 on the safe side, this is disabled for the C3x, since the maximum
9522 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9523 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9524 that it can utilize the decrement and branch instruction, but will give
9525 up if there is more than one memory reference in the loop. Thus a loop
9526 where the loop counter is decremented can generate slightly more
9527 efficient code, in cases where the RPTB instruction cannot be utilized.
9529 @item -mdp-isr-reload
9531 @opindex mdp-isr-reload
9533 Force the DP register to be saved on entry to an interrupt service
9534 routine (ISR), reloaded to point to the data section, and restored on
9535 exit from the ISR@. This should not be required unless someone has
9536 violated the small memory model by modifying the DP register, say within
9543 For the C3x use the 24-bit MPYI instruction for integer multiplies
9544 instead of a library call to guarantee 32-bit results. Note that if one
9545 of the operands is a constant, then the multiplication will be performed
9546 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9547 then squaring operations are performed inline instead of a library call.
9550 @itemx -mno-fast-fix
9552 @opindex mno-fast-fix
9553 The C3x/C4x FIX instruction to convert a floating point value to an
9554 integer value chooses the nearest integer less than or equal to the
9555 floating point value rather than to the nearest integer. Thus if the
9556 floating point number is negative, the result will be incorrectly
9557 truncated an additional code is necessary to detect and correct this
9558 case. This option can be used to disable generation of the additional
9559 code required to correct the result.
9565 Enable (disable) generation of repeat block sequences using the RPTB
9566 instruction for zero overhead looping. The RPTB construct is only used
9567 for innermost loops that do not call functions or jump across the loop
9568 boundaries. There is no advantage having nested RPTB loops due to the
9569 overhead required to save and restore the RC, RS, and RE registers.
9570 This is enabled by default with @option{-O2}.
9572 @item -mrpts=@var{count}
9576 Enable (disable) the use of the single instruction repeat instruction
9577 RPTS@. If a repeat block contains a single instruction, and the loop
9578 count can be guaranteed to be less than the value @var{count}, GCC will
9579 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9580 then a RPTS will be emitted even if the loop count cannot be determined
9581 at compile time. Note that the repeated instruction following RPTS does
9582 not have to be reloaded from memory each iteration, thus freeing up the
9583 CPU buses for operands. However, since interrupts are blocked by this
9584 instruction, it is disabled by default.
9586 @item -mloop-unsigned
9587 @itemx -mno-loop-unsigned
9588 @opindex mloop-unsigned
9589 @opindex mno-loop-unsigned
9590 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9591 is @math{2^{31} + 1} since these instructions test if the iteration count is
9592 negative to terminate the loop. If the iteration count is unsigned
9593 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9594 exceeded. This switch allows an unsigned iteration count.
9598 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9599 with. This also enforces compatibility with the API employed by the TI
9600 C3x C compiler. For example, long doubles are passed as structures
9601 rather than in floating point registers.
9607 Generate code that uses registers (stack) for passing arguments to functions.
9608 By default, arguments are passed in registers where possible rather
9609 than by pushing arguments on to the stack.
9611 @item -mparallel-insns
9612 @itemx -mno-parallel-insns
9613 @opindex mparallel-insns
9614 @opindex mno-parallel-insns
9615 Allow the generation of parallel instructions. This is enabled by
9616 default with @option{-O2}.
9618 @item -mparallel-mpy
9619 @itemx -mno-parallel-mpy
9620 @opindex mparallel-mpy
9621 @opindex mno-parallel-mpy
9622 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9623 provided @option{-mparallel-insns} is also specified. These instructions have
9624 tight register constraints which can pessimize the code generation
9630 @subsection V850 Options
9631 @cindex V850 Options
9633 These @samp{-m} options are defined for V850 implementations:
9637 @itemx -mno-long-calls
9638 @opindex mlong-calls
9639 @opindex mno-long-calls
9640 Treat all calls as being far away (near). If calls are assumed to be
9641 far away, the compiler will always load the functions address up into a
9642 register, and call indirect through the pointer.
9648 Do not optimize (do optimize) basic blocks that use the same index
9649 pointer 4 or more times to copy pointer into the @code{ep} register, and
9650 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9651 option is on by default if you optimize.
9653 @item -mno-prolog-function
9654 @itemx -mprolog-function
9655 @opindex mno-prolog-function
9656 @opindex mprolog-function
9657 Do not use (do use) external functions to save and restore registers
9658 at the prologue and epilogue of a function. The external functions
9659 are slower, but use less code space if more than one function saves
9660 the same number of registers. The @option{-mprolog-function} option
9661 is on by default if you optimize.
9665 Try to make the code as small as possible. At present, this just turns
9666 on the @option{-mep} and @option{-mprolog-function} options.
9670 Put static or global variables whose size is @var{n} bytes or less into
9671 the tiny data area that register @code{ep} points to. The tiny data
9672 area can hold up to 256 bytes in total (128 bytes for byte references).
9676 Put static or global variables whose size is @var{n} bytes or less into
9677 the small data area that register @code{gp} points to. The small data
9678 area can hold up to 64 kilobytes.
9682 Put static or global variables whose size is @var{n} bytes or less into
9683 the first 32 kilobytes of memory.
9687 Specify that the target processor is the V850.
9690 @opindex mbig-switch
9691 Generate code suitable for big switch tables. Use this option only if
9692 the assembler/linker complain about out of range branches within a switch
9697 This option will cause r2 and r5 to be used in the code generated by
9698 the compiler. This setting is the default.
9701 @opindex mno-app-regs
9702 This option will cause r2 and r5 to be treated as fixed registers.
9706 Specify that the target processor is the V850E1. The preprocessor
9707 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9708 this option is used.
9712 Specify that the target processor is the V850E. The preprocessor
9713 constant @samp{__v850e__} will be defined if this option is used.
9715 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9716 are defined then a default target processor will be chosen and the
9717 relevant @samp{__v850*__} preprocessor constant will be defined.
9719 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9720 defined, regardless of which processor variant is the target.
9722 @item -mdisable-callt
9723 @opindex mdisable-callt
9724 This option will suppress generation of the CALLT instruction for the
9725 v850e and v850e1 flavors of the v850 architecture. The default is
9726 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9731 @subsection ARC Options
9734 These options are defined for ARC implementations:
9739 Compile code for little endian mode. This is the default.
9743 Compile code for big endian mode.
9746 @opindex mmangle-cpu
9747 Prepend the name of the cpu to all public symbol names.
9748 In multiple-processor systems, there are many ARC variants with different
9749 instruction and register set characteristics. This flag prevents code
9750 compiled for one cpu to be linked with code compiled for another.
9751 No facility exists for handling variants that are ``almost identical''.
9752 This is an all or nothing option.
9754 @item -mcpu=@var{cpu}
9756 Compile code for ARC variant @var{cpu}.
9757 Which variants are supported depend on the configuration.
9758 All variants support @option{-mcpu=base}, this is the default.
9760 @item -mtext=@var{text-section}
9761 @itemx -mdata=@var{data-section}
9762 @itemx -mrodata=@var{readonly-data-section}
9766 Put functions, data, and readonly data in @var{text-section},
9767 @var{data-section}, and @var{readonly-data-section} respectively
9768 by default. This can be overridden with the @code{section} attribute.
9769 @xref{Variable Attributes}.
9774 @subsection NS32K Options
9775 @cindex NS32K options
9777 These are the @samp{-m} options defined for the 32000 series. The default
9778 values for these options depends on which style of 32000 was selected when
9779 the compiler was configured; the defaults for the most common choices are
9787 Generate output for a 32032. This is the default
9788 when the compiler is configured for 32032 and 32016 based systems.
9794 Generate output for a 32332. This is the default
9795 when the compiler is configured for 32332-based systems.
9801 Generate output for a 32532. This is the default
9802 when the compiler is configured for 32532-based systems.
9806 Generate output containing 32081 instructions for floating point.
9807 This is the default for all systems.
9811 Generate output containing 32381 instructions for floating point. This
9812 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9813 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9817 Try and generate multiply-add floating point instructions @code{polyF}
9818 and @code{dotF}. This option is only available if the @option{-m32381}
9819 option is in effect. Using these instructions requires changes to
9820 register allocation which generally has a negative impact on
9821 performance. This option should only be enabled when compiling code
9822 particularly likely to make heavy use of multiply-add instructions.
9825 @opindex mnomulti-add
9826 Do not try and generate multiply-add floating point instructions
9827 @code{polyF} and @code{dotF}. This is the default on all platforms.
9830 @opindex msoft-float
9831 Generate output containing library calls for floating point.
9832 @strong{Warning:} the requisite libraries may not be available.
9834 @item -mieee-compare
9835 @itemx -mno-ieee-compare
9836 @opindex mieee-compare
9837 @opindex mno-ieee-compare
9838 Control whether or not the compiler uses IEEE floating point
9839 comparisons. These handle correctly the case where the result of a
9840 comparison is unordered.
9841 @strong{Warning:} the requisite kernel support may not be available.
9844 @opindex mnobitfield
9845 Do not use the bit-field instructions. On some machines it is faster to
9846 use shifting and masking operations. This is the default for the pc532.
9850 Do use the bit-field instructions. This is the default for all platforms
9855 Use a different function-calling convention, in which functions
9856 that take a fixed number of arguments return pop their
9857 arguments on return with the @code{ret} instruction.
9859 This calling convention is incompatible with the one normally
9860 used on Unix, so you cannot use it if you need to call libraries
9861 compiled with the Unix compiler.
9863 Also, you must provide function prototypes for all functions that
9864 take variable numbers of arguments (including @code{printf});
9865 otherwise incorrect code will be generated for calls to those
9868 In addition, seriously incorrect code will result if you call a
9869 function with too many arguments. (Normally, extra arguments are
9870 harmlessly ignored.)
9872 This option takes its name from the 680x0 @code{rtd} instruction.
9877 Use a different function-calling convention where the first two arguments
9878 are passed in registers.
9880 This calling convention is incompatible with the one normally
9881 used on Unix, so you cannot use it if you need to call libraries
9882 compiled with the Unix compiler.
9885 @opindex mnoregparam
9886 Do not pass any arguments in registers. This is the default for all
9891 It is OK to use the sb as an index register which is always loaded with
9892 zero. This is the default for the pc532-netbsd target.
9896 The sb register is not available for use or has not been initialized to
9897 zero by the run time system. This is the default for all targets except
9898 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9899 @option{-fpic} is set.
9903 Many ns32000 series addressing modes use displacements of up to 512MB@.
9904 If an address is above 512MB then displacements from zero can not be used.
9905 This option causes code to be generated which can be loaded above 512MB@.
9906 This may be useful for operating systems or ROM code.
9910 Assume code will be loaded in the first 512MB of virtual address space.
9911 This is the default for all platforms.
9917 @subsection AVR Options
9920 These options are defined for AVR implementations:
9923 @item -mmcu=@var{mcu}
9925 Specify ATMEL AVR instruction set or MCU type.
9927 Instruction set avr1 is for the minimal AVR core, not supported by the C
9928 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9929 attiny11, attiny12, attiny15, attiny28).
9931 Instruction set avr2 (default) is for the classic AVR core with up to
9932 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9933 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9934 at90c8534, at90s8535).
9936 Instruction set avr3 is for the classic AVR core with up to 128K program
9937 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9939 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9940 memory space (MCU types: atmega8, atmega83, atmega85).
9942 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9943 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9944 atmega64, atmega128, at43usb355, at94k).
9948 Output instruction sizes to the asm file.
9950 @item -minit-stack=@var{N}
9951 @opindex minit-stack
9952 Specify the initial stack address, which may be a symbol or numeric value,
9953 @samp{__stack} is the default.
9955 @item -mno-interrupts
9956 @opindex mno-interrupts
9957 Generated code is not compatible with hardware interrupts.
9958 Code size will be smaller.
9960 @item -mcall-prologues
9961 @opindex mcall-prologues
9962 Functions prologues/epilogues expanded as call to appropriate
9963 subroutines. Code size will be smaller.
9965 @item -mno-tablejump
9966 @opindex mno-tablejump
9967 Do not generate tablejump insns which sometimes increase code size.
9970 @opindex mtiny-stack
9971 Change only the low 8 bits of the stack pointer.
9975 @subsection MCore Options
9976 @cindex MCore options
9978 These are the @samp{-m} options defined for the Motorola M*Core
9986 @opindex mno-hardlit
9987 Inline constants into the code stream if it can be done in two
9988 instructions or less.
9994 Use the divide instruction. (Enabled by default).
9996 @item -mrelax-immediate
9997 @itemx -mno-relax-immediate
9998 @opindex mrelax-immediate
9999 @opindex mno-relax-immediate
10000 Allow arbitrary sized immediates in bit operations.
10002 @item -mwide-bitfields
10003 @itemx -mno-wide-bitfields
10004 @opindex mwide-bitfields
10005 @opindex mno-wide-bitfields
10006 Always treat bit-fields as int-sized.
10008 @item -m4byte-functions
10009 @itemx -mno-4byte-functions
10010 @opindex m4byte-functions
10011 @opindex mno-4byte-functions
10012 Force all functions to be aligned to a four byte boundary.
10014 @item -mcallgraph-data
10015 @itemx -mno-callgraph-data
10016 @opindex mcallgraph-data
10017 @opindex mno-callgraph-data
10018 Emit callgraph information.
10021 @itemx -mno-slow-bytes
10022 @opindex mslow-bytes
10023 @opindex mno-slow-bytes
10024 Prefer word access when reading byte quantities.
10026 @item -mlittle-endian
10027 @itemx -mbig-endian
10028 @opindex mlittle-endian
10029 @opindex mbig-endian
10030 Generate code for a little endian target.
10036 Generate code for the 210 processor.
10039 @node IA-64 Options
10040 @subsection IA-64 Options
10041 @cindex IA-64 Options
10043 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10047 @opindex mbig-endian
10048 Generate code for a big endian target. This is the default for HP-UX@.
10050 @item -mlittle-endian
10051 @opindex mlittle-endian
10052 Generate code for a little endian target. This is the default for AIX5
10058 @opindex mno-gnu-as
10059 Generate (or don't) code for the GNU assembler. This is the default.
10060 @c Also, this is the default if the configure option @option{--with-gnu-as}
10066 @opindex mno-gnu-ld
10067 Generate (or don't) code for the GNU linker. This is the default.
10068 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10073 Generate code that does not use a global pointer register. The result
10074 is not position independent code, and violates the IA-64 ABI@.
10076 @item -mvolatile-asm-stop
10077 @itemx -mno-volatile-asm-stop
10078 @opindex mvolatile-asm-stop
10079 @opindex mno-volatile-asm-stop
10080 Generate (or don't) a stop bit immediately before and after volatile asm
10085 Generate code that works around Itanium B step errata.
10087 @item -mregister-names
10088 @itemx -mno-register-names
10089 @opindex mregister-names
10090 @opindex mno-register-names
10091 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10092 the stacked registers. This may make assembler output more readable.
10098 Disable (or enable) optimizations that use the small data section. This may
10099 be useful for working around optimizer bugs.
10101 @item -mconstant-gp
10102 @opindex mconstant-gp
10103 Generate code that uses a single constant global pointer value. This is
10104 useful when compiling kernel code.
10108 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10109 This is useful when compiling firmware code.
10111 @item -minline-float-divide-min-latency
10112 @opindex minline-float-divide-min-latency
10113 Generate code for inline divides of floating point values
10114 using the minimum latency algorithm.
10116 @item -minline-float-divide-max-throughput
10117 @opindex minline-float-divide-max-throughput
10118 Generate code for inline divides of floating point values
10119 using the maximum throughput algorithm.
10121 @item -minline-int-divide-min-latency
10122 @opindex minline-int-divide-min-latency
10123 Generate code for inline divides of integer values
10124 using the minimum latency algorithm.
10126 @item -minline-int-divide-max-throughput
10127 @opindex minline-int-divide-max-throughput
10128 Generate code for inline divides of integer values
10129 using the maximum throughput algorithm.
10131 @item -mno-dwarf2-asm
10132 @itemx -mdwarf2-asm
10133 @opindex mno-dwarf2-asm
10134 @opindex mdwarf2-asm
10135 Don't (or do) generate assembler code for the DWARF2 line number debugging
10136 info. This may be useful when not using the GNU assembler.
10138 @item -mfixed-range=@var{register-range}
10139 @opindex mfixed-range
10140 Generate code treating the given register range as fixed registers.
10141 A fixed register is one that the register allocator can not use. This is
10142 useful when compiling kernel code. A register range is specified as
10143 two registers separated by a dash. Multiple register ranges can be
10144 specified separated by a comma.
10146 @item -mearly-stop-bits
10147 @itemx -mno-early-stop-bits
10148 @opindex mearly-stop-bits
10149 @opindex mno-early-stop-bits
10150 Allow stop bits to be placed earlier than immediately preceding the
10151 instruction that triggered the stop bit. This can improve instruction
10152 scheduling, but does not always do so.
10156 @subsection D30V Options
10157 @cindex D30V Options
10159 These @samp{-m} options are defined for D30V implementations:
10164 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
10165 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
10166 memory, which starts at location @code{0x80000000}.
10169 @opindex mextmemory
10170 Same as the @option{-mextmem} switch.
10174 Link the @samp{.text} section into onchip text memory, which starts at
10175 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
10176 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
10177 into onchip data memory, which starts at location @code{0x20000000}.
10179 @item -mno-asm-optimize
10180 @itemx -masm-optimize
10181 @opindex mno-asm-optimize
10182 @opindex masm-optimize
10183 Disable (enable) passing @option{-O} to the assembler when optimizing.
10184 The assembler uses the @option{-O} option to automatically parallelize
10185 adjacent short instructions where possible.
10187 @item -mbranch-cost=@var{n}
10188 @opindex mbranch-cost
10189 Increase the internal costs of branches to @var{n}. Higher costs means
10190 that the compiler will issue more instructions to avoid doing a branch.
10193 @item -mcond-exec=@var{n}
10194 @opindex mcond-exec
10195 Specify the maximum number of conditionally executed instructions that
10196 replace a branch. The default is 4.
10199 @node S/390 and zSeries Options
10200 @subsection S/390 and zSeries Options
10201 @cindex S/390 and zSeries Options
10203 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10207 @itemx -msoft-float
10208 @opindex mhard-float
10209 @opindex msoft-float
10210 Use (do not use) the hardware floating-point instructions and registers
10211 for floating-point operations. When @option{-msoft-float} is specified,
10212 functions in @file{libgcc.a} will be used to perform floating-point
10213 operations. When @option{-mhard-float} is specified, the compiler
10214 generates IEEE floating-point instructions. This is the default.
10217 @itemx -mno-backchain
10218 @opindex mbackchain
10219 @opindex mno-backchain
10220 Generate (or do not generate) code which maintains an explicit
10221 backchain within the stack frame that points to the caller's frame.
10222 This is currently needed to allow debugging. The default is to
10223 generate the backchain.
10226 @itemx -mno-small-exec
10227 @opindex msmall-exec
10228 @opindex mno-small-exec
10229 Generate (or do not generate) code using the @code{bras} instruction
10230 to do subroutine calls.
10231 This only works reliably if the total executable size does not
10232 exceed 64k. The default is to use the @code{basr} instruction instead,
10233 which does not have this limitation.
10239 When @option{-m31} is specified, generate code compliant to the
10240 Linux for S/390 ABI@. When @option{-m64} is specified, generate
10241 code compliant to the Linux for zSeries ABI@. This allows GCC in
10242 particular to generate 64-bit instructions. For the @samp{s390}
10243 targets, the default is @option{-m31}, while the @samp{s390x}
10244 targets default to @option{-m64}.
10250 When @option{-mzarch} is specified, generate code using the
10251 instructions available on z/Architecture.
10252 When @option{-mesa} is specified, generate code using the
10253 instructions available on ESA/390. Note that @option{-mesa} is
10254 not possible with @option{-m64}.
10255 When generating code compliant to the Linux for S/390 ABI,
10256 the default is @option{-mesa}. When generating code compliant
10257 to the Linux for zSeries ABI, the default is @option{-mzarch}.
10263 Generate (or do not generate) code using the @code{mvcle} instruction
10264 to perform block moves. When @option{-mno-mvcle} is specified,
10265 use a @code{mvc} loop instead. This is the default.
10271 Print (or do not print) additional debug information when compiling.
10272 The default is to not print debug information.
10274 @item -march=@var{cpu-type}
10276 Generate code that will run on @var{cpu-type}, which is the name of a system
10277 representing a certain processor type. Possible values for
10278 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10279 When generating code using the instructions available on z/Architecture,
10280 the default is @option{-march=z900}. Otherwise, the default is
10281 @option{-march=g5}.
10283 @item -mtune=@var{cpu-type}
10285 Tune to @var{cpu-type} everything applicable about the generated code,
10286 except for the ABI and the set of available instructions.
10287 The list of @var{cpu-type} values is the same as for @option{-march}.
10288 The default is the value used for @option{-march}.
10293 @subsection CRIS Options
10294 @cindex CRIS Options
10296 These options are defined specifically for the CRIS ports.
10299 @item -march=@var{architecture-type}
10300 @itemx -mcpu=@var{architecture-type}
10303 Generate code for the specified architecture. The choices for
10304 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10305 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10306 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10309 @item -mtune=@var{architecture-type}
10311 Tune to @var{architecture-type} everything applicable about the generated
10312 code, except for the ABI and the set of available instructions. The
10313 choices for @var{architecture-type} are the same as for
10314 @option{-march=@var{architecture-type}}.
10316 @item -mmax-stack-frame=@var{n}
10317 @opindex mmax-stack-frame
10318 Warn when the stack frame of a function exceeds @var{n} bytes.
10320 @item -melinux-stacksize=@var{n}
10321 @opindex melinux-stacksize
10322 Only available with the @samp{cris-axis-aout} target. Arranges for
10323 indications in the program to the kernel loader that the stack of the
10324 program should be set to @var{n} bytes.
10330 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10331 @option{-march=v3} and @option{-march=v8} respectively.
10335 Enable CRIS-specific verbose debug-related information in the assembly
10336 code. This option also has the effect to turn off the @samp{#NO_APP}
10337 formatted-code indicator to the assembler at the beginning of the
10342 Do not use condition-code results from previous instruction; always emit
10343 compare and test instructions before use of condition codes.
10345 @item -mno-side-effects
10346 @opindex mno-side-effects
10347 Do not emit instructions with side-effects in addressing modes other than
10350 @item -mstack-align
10351 @itemx -mno-stack-align
10352 @itemx -mdata-align
10353 @itemx -mno-data-align
10354 @itemx -mconst-align
10355 @itemx -mno-const-align
10356 @opindex mstack-align
10357 @opindex mno-stack-align
10358 @opindex mdata-align
10359 @opindex mno-data-align
10360 @opindex mconst-align
10361 @opindex mno-const-align
10362 These options (no-options) arranges (eliminate arrangements) for the
10363 stack-frame, individual data and constants to be aligned for the maximum
10364 single data access size for the chosen CPU model. The default is to
10365 arrange for 32-bit alignment. ABI details such as structure layout are
10366 not affected by these options.
10374 Similar to the stack- data- and const-align options above, these options
10375 arrange for stack-frame, writable data and constants to all be 32-bit,
10376 16-bit or 8-bit aligned. The default is 32-bit alignment.
10378 @item -mno-prologue-epilogue
10379 @itemx -mprologue-epilogue
10380 @opindex mno-prologue-epilogue
10381 @opindex mprologue-epilogue
10382 With @option{-mno-prologue-epilogue}, the normal function prologue and
10383 epilogue that sets up the stack-frame are omitted and no return
10384 instructions or return sequences are generated in the code. Use this
10385 option only together with visual inspection of the compiled code: no
10386 warnings or errors are generated when call-saved registers must be saved,
10387 or storage for local variable needs to be allocated.
10391 @opindex mno-gotplt
10393 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10394 instruction sequences that load addresses for functions from the PLT part
10395 of the GOT rather than (traditional on other architectures) calls to the
10396 PLT. The default is @option{-mgotplt}.
10400 Legacy no-op option only recognized with the cris-axis-aout target.
10404 Legacy no-op option only recognized with the cris-axis-elf and
10405 cris-axis-linux-gnu targets.
10409 Only recognized with the cris-axis-aout target, where it selects a
10410 GNU/linux-like multilib, include files and instruction set for
10411 @option{-march=v8}.
10415 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10419 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10420 to link with input-output functions from a simulator library. Code,
10421 initialized data and zero-initialized data are allocated consecutively.
10425 Like @option{-sim}, but pass linker options to locate initialized data at
10426 0x40000000 and zero-initialized data at 0x80000000.
10430 @subsection MMIX Options
10431 @cindex MMIX Options
10433 These options are defined for the MMIX:
10437 @itemx -mno-libfuncs
10439 @opindex mno-libfuncs
10440 Specify that intrinsic library functions are being compiled, passing all
10441 values in registers, no matter the size.
10444 @itemx -mno-epsilon
10446 @opindex mno-epsilon
10447 Generate floating-point comparison instructions that compare with respect
10448 to the @code{rE} epsilon register.
10450 @item -mabi=mmixware
10452 @opindex mabi-mmixware
10454 Generate code that passes function parameters and return values that (in
10455 the called function) are seen as registers @code{$0} and up, as opposed to
10456 the GNU ABI which uses global registers @code{$231} and up.
10458 @item -mzero-extend
10459 @itemx -mno-zero-extend
10460 @opindex mzero-extend
10461 @opindex mno-zero-extend
10462 When reading data from memory in sizes shorter than 64 bits, use (do not
10463 use) zero-extending load instructions by default, rather than
10464 sign-extending ones.
10467 @itemx -mno-knuthdiv
10469 @opindex mno-knuthdiv
10470 Make the result of a division yielding a remainder have the same sign as
10471 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10472 remainder follows the sign of the dividend. Both methods are
10473 arithmetically valid, the latter being almost exclusively used.
10475 @item -mtoplevel-symbols
10476 @itemx -mno-toplevel-symbols
10477 @opindex mtoplevel-symbols
10478 @opindex mno-toplevel-symbols
10479 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10480 code can be used with the @code{PREFIX} assembly directive.
10484 Generate an executable in the ELF format, rather than the default
10485 @samp{mmo} format used by the @command{mmix} simulator.
10487 @item -mbranch-predict
10488 @itemx -mno-branch-predict
10489 @opindex mbranch-predict
10490 @opindex mno-branch-predict
10491 Use (do not use) the probable-branch instructions, when static branch
10492 prediction indicates a probable branch.
10494 @item -mbase-addresses
10495 @itemx -mno-base-addresses
10496 @opindex mbase-addresses
10497 @opindex mno-base-addresses
10498 Generate (do not generate) code that uses @emph{base addresses}. Using a
10499 base address automatically generates a request (handled by the assembler
10500 and the linker) for a constant to be set up in a global register. The
10501 register is used for one or more base address requests within the range 0
10502 to 255 from the value held in the register. The generally leads to short
10503 and fast code, but the number of different data items that can be
10504 addressed is limited. This means that a program that uses lots of static
10505 data may require @option{-mno-base-addresses}.
10507 @item -msingle-exit
10508 @itemx -mno-single-exit
10509 @opindex msingle-exit
10510 @opindex mno-single-exit
10511 Force (do not force) generated code to have a single exit point in each
10515 @node PDP-11 Options
10516 @subsection PDP-11 Options
10517 @cindex PDP-11 Options
10519 These options are defined for the PDP-11:
10524 Use hardware FPP floating point. This is the default. (FIS floating
10525 point on the PDP-11/40 is not supported.)
10528 @opindex msoft-float
10529 Do not use hardware floating point.
10533 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10537 Return floating-point results in memory. This is the default.
10541 Generate code for a PDP-11/40.
10545 Generate code for a PDP-11/45. This is the default.
10549 Generate code for a PDP-11/10.
10551 @item -mbcopy-builtin
10552 @opindex bcopy-builtin
10553 Use inline @code{movstrhi} patterns for copying memory. This is the
10558 Do not use inline @code{movstrhi} patterns for copying memory.
10564 Use 16-bit @code{int}. This is the default.
10570 Use 32-bit @code{int}.
10573 @itemx -mno-float32
10575 @opindex mno-float32
10576 Use 64-bit @code{float}. This is the default.
10581 @opindex mno-float64
10582 Use 32-bit @code{float}.
10586 Use @code{abshi2} pattern. This is the default.
10590 Do not use @code{abshi2} pattern.
10592 @item -mbranch-expensive
10593 @opindex mbranch-expensive
10594 Pretend that branches are expensive. This is for experimenting with
10595 code generation only.
10597 @item -mbranch-cheap
10598 @opindex mbranch-cheap
10599 Do not pretend that branches are expensive. This is the default.
10603 Generate code for a system with split I&D.
10607 Generate code for a system without split I&D. This is the default.
10611 Use Unix assembler syntax. This is the default when configured for
10612 @samp{pdp11-*-bsd}.
10616 Use DEC assembler syntax. This is the default when configured for any
10617 PDP-11 target other than @samp{pdp11-*-bsd}.
10620 @node Xstormy16 Options
10621 @subsection Xstormy16 Options
10622 @cindex Xstormy16 Options
10624 These options are defined for Xstormy16:
10629 Choose startup files and linker script suitable for the simulator.
10633 @subsection FRV Options
10634 @cindex FRV Options
10640 Only use the first 32 general purpose registers.
10645 Use all 64 general purpose registers.
10650 Use only the first 32 floating point registers.
10655 Use all 64 floating point registers
10658 @opindex mhard-float
10660 Use hardware instructions for floating point operations.
10663 @opindex msoft-float
10665 Use library routines for floating point operations.
10670 Dynamically allocate condition code registers.
10675 Do not try to dynamically allocate condition code registers, only
10676 use @code{icc0} and @code{fcc0}.
10681 Change ABI to use double word insns.
10686 Do not use double word instructions.
10691 Use floating point double instructions.
10694 @opindex mno-double
10696 Do not use floating point double instructions.
10701 Use media instructions.
10706 Do not use media instructions.
10711 Use multiply and add/subtract instructions.
10714 @opindex mno-muladd
10716 Do not use multiply and add/subtract instructions.
10718 @item -mlibrary-pic
10719 @opindex mlibrary-pic
10721 Enable PIC support for building libraries
10726 Use only the first four media accumulator registers.
10731 Use all eight media accumulator registers.
10736 Pack VLIW instructions.
10741 Do not pack VLIW instructions.
10744 @opindex mno-eflags
10746 Do not mark ABI switches in e_flags.
10749 @opindex mcond-move
10751 Enable the use of conditional-move instructions (default).
10753 This switch is mainly for debugging the compiler and will likely be removed
10754 in a future version.
10756 @item -mno-cond-move
10757 @opindex mno-cond-move
10759 Disable the use of conditional-move instructions.
10761 This switch is mainly for debugging the compiler and will likely be removed
10762 in a future version.
10767 Enable the use of conditional set instructions (default).
10769 This switch is mainly for debugging the compiler and will likely be removed
10770 in a future version.
10775 Disable the use of conditional set instructions.
10777 This switch is mainly for debugging the compiler and will likely be removed
10778 in a future version.
10781 @opindex mcond-exec
10783 Enable the use of conditional execution (default).
10785 This switch is mainly for debugging the compiler and will likely be removed
10786 in a future version.
10788 @item -mno-cond-exec
10789 @opindex mno-cond-exec
10791 Disable the use of conditional execution.
10793 This switch is mainly for debugging the compiler and will likely be removed
10794 in a future version.
10796 @item -mvliw-branch
10797 @opindex mvliw-branch
10799 Run a pass to pack branches into VLIW instructions (default).
10801 This switch is mainly for debugging the compiler and will likely be removed
10802 in a future version.
10804 @item -mno-vliw-branch
10805 @opindex mno-vliw-branch
10807 Do not run a pass to pack branches into VLIW instructions.
10809 This switch is mainly for debugging the compiler and will likely be removed
10810 in a future version.
10812 @item -mmulti-cond-exec
10813 @opindex mmulti-cond-exec
10815 Enable optimization of @code{&&} and @code{||} in conditional execution
10818 This switch is mainly for debugging the compiler and will likely be removed
10819 in a future version.
10821 @item -mno-multi-cond-exec
10822 @opindex mno-multi-cond-exec
10824 Disable optimization of @code{&&} and @code{||} in conditional execution.
10826 This switch is mainly for debugging the compiler and will likely be removed
10827 in a future version.
10829 @item -mnested-cond-exec
10830 @opindex mnested-cond-exec
10832 Enable nested conditional execution optimizations (default).
10834 This switch is mainly for debugging the compiler and will likely be removed
10835 in a future version.
10837 @item -mno-nested-cond-exec
10838 @opindex mno-nested-cond-exec
10840 Disable nested conditional execution optimizations.
10842 This switch is mainly for debugging the compiler and will likely be removed
10843 in a future version.
10845 @item -mtomcat-stats
10846 @opindex mtomcat-stats
10848 Cause gas to print out tomcat statistics.
10850 @item -mcpu=@var{cpu}
10853 Select the processor type for which to generate code. Possible values are
10854 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10859 @node Xtensa Options
10860 @subsection Xtensa Options
10861 @cindex Xtensa Options
10863 The Xtensa architecture is designed to support many different
10864 configurations. The compiler's default options can be set to match a
10865 particular Xtensa configuration by copying a configuration file into the
10866 GCC sources when building GCC@. The options below may be used to
10867 override the default options.
10871 @itemx -mlittle-endian
10872 @opindex mbig-endian
10873 @opindex mlittle-endian
10874 Specify big-endian or little-endian byte ordering for the target Xtensa
10878 @itemx -mno-density
10880 @opindex mno-density
10881 Enable or disable use of the optional Xtensa code density instructions.
10884 @itemx -mno-const16
10886 @opindex mno-const16
10887 Enable or disable use of @code{CONST16} instructions for loading
10888 constant values. The @code{CONST16} instruction is currently not a
10889 standard option from Tensilica. When enabled, @code{CONST16}
10890 instructions are always used in place of the standard @code{L32R}
10891 instructions. The use of @code{CONST16} is enabled by default only if
10892 the @code{L32R} instruction is not available.
10898 Enable or disable use of the Xtensa @code{ABS} instruction for absolute
10905 Enable or disable use of the Xtensa @code{ADDX} and @code{SUBX}
10912 Enable or disable use of the Xtensa MAC16 option. When enabled, GCC
10913 will generate MAC16 instructions from standard C code, with the
10914 limitation that it will use neither the MR register file nor any
10915 instruction that operates on the MR registers. When this option is
10916 disabled, GCC will translate 16-bit multiply/accumulate operations to a
10917 combination of core instructions and library calls, depending on whether
10918 any other multiplier options are enabled.
10924 Enable or disable use of the 16-bit integer multiplier option. When
10925 enabled, the compiler will generate 16-bit multiply instructions for
10926 multiplications of 16 bits or smaller in standard C code. When this
10927 option is disabled, the compiler will either use 32-bit multiply or
10928 MAC16 instructions if they are available or generate library calls to
10929 perform the multiply operations using shifts and adds.
10935 Enable or disable use of the 32-bit integer multiplier option. When
10936 enabled, the compiler will generate 32-bit multiply instructions for
10937 multiplications of 32 bits or smaller in standard C code. When this
10938 option is disabled, the compiler will generate library calls to perform
10939 the multiply operations using either shifts and adds or 16-bit multiply
10940 instructions if they are available.
10946 Enable or disable use of the optional normalization shift amount
10947 (@code{NSA}) instructions to implement the built-in @code{ffs} function.
10952 @opindex mno-minmax
10953 Enable or disable use of the optional minimum and maximum value
10960 Enable or disable use of the optional sign extend (@code{SEXT})
10964 @itemx -mno-booleans
10966 @opindex mno-booleans
10967 Enable or disable support for the boolean register file used by Xtensa
10968 coprocessors. This is not typically useful by itself but may be
10969 required for other options that make use of the boolean registers (e.g.,
10970 the floating-point option).
10973 @itemx -msoft-float
10974 @opindex mhard-float
10975 @opindex msoft-float
10976 Enable or disable use of the floating-point option. When enabled, GCC
10977 generates floating-point instructions for 32-bit @code{float}
10978 operations. When this option is disabled, GCC generates library calls
10979 to emulate 32-bit floating-point operations using integer instructions.
10980 Regardless of this option, 64-bit @code{double} operations are always
10981 emulated with calls to library functions.
10984 @itemx -mno-fused-madd
10985 @opindex mfused-madd
10986 @opindex mno-fused-madd
10987 Enable or disable use of fused multiply/add and multiply/subtract
10988 instructions in the floating-point option. This has no effect if the
10989 floating-point option is not also enabled. Disabling fused multiply/add
10990 and multiply/subtract instructions forces the compiler to use separate
10991 instructions for the multiply and add/subtract operations. This may be
10992 desirable in some cases where strict IEEE 754-compliant results are
10993 required: the fused multiply add/subtract instructions do not round the
10994 intermediate result, thereby producing results with @emph{more} bits of
10995 precision than specified by the IEEE standard. Disabling fused multiply
10996 add/subtract instructions also ensures that the program output is not
10997 sensitive to the compiler's ability to combine multiply and add/subtract
11000 @item -mtext-section-literals
11001 @itemx -mno-text-section-literals
11002 @opindex mtext-section-literals
11003 @opindex mno-text-section-literals
11004 Control the treatment of literal pools. The default is
11005 @option{-mno-text-section-literals}, which places literals in a separate
11006 section in the output file. This allows the literal pool to be placed
11007 in a data RAM/ROM, and it also allows the linker to combine literal
11008 pools from separate object files to remove redundant literals and
11009 improve code size. With @option{-mtext-section-literals}, the literals
11010 are interspersed in the text section in order to keep them as close as
11011 possible to their references. This may be necessary for large assembly
11014 @item -mtarget-align
11015 @itemx -mno-target-align
11016 @opindex mtarget-align
11017 @opindex mno-target-align
11018 When this option is enabled, GCC instructs the assembler to
11019 automatically align instructions to reduce branch penalties at the
11020 expense of some code density. The assembler attempts to widen density
11021 instructions to align branch targets and the instructions following call
11022 instructions. If there are not enough preceding safe density
11023 instructions to align a target, no widening will be performed. The
11024 default is @option{-mtarget-align}. These options do not affect the
11025 treatment of auto-aligned instructions like @code{LOOP}, which the
11026 assembler will always align, either by widening density instructions or
11027 by inserting no-op instructions.
11030 @itemx -mno-longcalls
11031 @opindex mlongcalls
11032 @opindex mno-longcalls
11033 When this option is enabled, GCC instructs the assembler to translate
11034 direct calls to indirect calls unless it can determine that the target
11035 of a direct call is in the range allowed by the call instruction. This
11036 translation typically occurs for calls to functions in other source
11037 files. Specifically, the assembler translates a direct @code{CALL}
11038 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11039 The default is @option{-mno-longcalls}. This option should be used in
11040 programs where the call target can potentially be out of range. This
11041 option is implemented in the assembler, not the compiler, so the
11042 assembly code generated by GCC will still show direct call
11043 instructions---look at the disassembled object code to see the actual
11044 instructions. Note that the assembler will use an indirect call for
11045 every cross-file call, not just those that really will be out of range.
11048 @node Code Gen Options
11049 @section Options for Code Generation Conventions
11050 @cindex code generation conventions
11051 @cindex options, code generation
11052 @cindex run-time options
11054 These machine-independent options control the interface conventions
11055 used in code generation.
11057 Most of them have both positive and negative forms; the negative form
11058 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11059 one of the forms is listed---the one which is not the default. You
11060 can figure out the other form by either removing @samp{no-} or adding
11064 @item -fbounds-check
11065 @opindex fbounds-check
11066 For front-ends that support it, generate additional code to check that
11067 indices used to access arrays are within the declared range. This is
11068 currently only supported by the Java and Fortran 77 front-ends, where
11069 this option defaults to true and false respectively.
11073 This option generates traps for signed overflow on addition, subtraction,
11074 multiplication operations.
11078 This option instructs the compiler to assume that signed arithmetic
11079 overflow of addition, subtraction and multiplication wraps around
11080 using twos-complement representation. This flag enables some optimizations
11081 and disables other. This option is enabled by default for the Java
11082 front-end, as required by the Java language specification.
11085 @opindex fexceptions
11086 Enable exception handling. Generates extra code needed to propagate
11087 exceptions. For some targets, this implies GCC will generate frame
11088 unwind information for all functions, which can produce significant data
11089 size overhead, although it does not affect execution. If you do not
11090 specify this option, GCC will enable it by default for languages like
11091 C++ which normally require exception handling, and disable it for
11092 languages like C that do not normally require it. However, you may need
11093 to enable this option when compiling C code that needs to interoperate
11094 properly with exception handlers written in C++. You may also wish to
11095 disable this option if you are compiling older C++ programs that don't
11096 use exception handling.
11098 @item -fnon-call-exceptions
11099 @opindex fnon-call-exceptions
11100 Generate code that allows trapping instructions to throw exceptions.
11101 Note that this requires platform-specific runtime support that does
11102 not exist everywhere. Moreover, it only allows @emph{trapping}
11103 instructions to throw exceptions, i.e.@: memory references or floating
11104 point instructions. It does not allow exceptions to be thrown from
11105 arbitrary signal handlers such as @code{SIGALRM}.
11107 @item -funwind-tables
11108 @opindex funwind-tables
11109 Similar to @option{-fexceptions}, except that it will just generate any needed
11110 static data, but will not affect the generated code in any other way.
11111 You will normally not enable this option; instead, a language processor
11112 that needs this handling would enable it on your behalf.
11114 @item -fasynchronous-unwind-tables
11115 @opindex funwind-tables
11116 Generate unwind table in dwarf2 format, if supported by target machine. The
11117 table is exact at each instruction boundary, so it can be used for stack
11118 unwinding from asynchronous events (such as debugger or garbage collector).
11120 @item -fpcc-struct-return
11121 @opindex fpcc-struct-return
11122 Return ``short'' @code{struct} and @code{union} values in memory like
11123 longer ones, rather than in registers. This convention is less
11124 efficient, but it has the advantage of allowing intercallability between
11125 GCC-compiled files and files compiled with other compilers, particularly
11126 the Portable C Compiler (pcc).
11128 The precise convention for returning structures in memory depends
11129 on the target configuration macros.
11131 Short structures and unions are those whose size and alignment match
11132 that of some integer type.
11134 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11135 switch is not binary compatible with code compiled with the
11136 @option{-freg-struct-return} switch.
11137 Use it to conform to a non-default application binary interface.
11139 @item -freg-struct-return
11140 @opindex freg-struct-return
11141 Return @code{struct} and @code{union} values in registers when possible.
11142 This is more efficient for small structures than
11143 @option{-fpcc-struct-return}.
11145 If you specify neither @option{-fpcc-struct-return} nor
11146 @option{-freg-struct-return}, GCC defaults to whichever convention is
11147 standard for the target. If there is no standard convention, GCC
11148 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11149 the principal compiler. In those cases, we can choose the standard, and
11150 we chose the more efficient register return alternative.
11152 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11153 switch is not binary compatible with code compiled with the
11154 @option{-fpcc-struct-return} switch.
11155 Use it to conform to a non-default application binary interface.
11157 @item -fshort-enums
11158 @opindex fshort-enums
11159 Allocate to an @code{enum} type only as many bytes as it needs for the
11160 declared range of possible values. Specifically, the @code{enum} type
11161 will be equivalent to the smallest integer type which has enough room.
11163 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11164 code that is not binary compatible with code generated without that switch.
11165 Use it to conform to a non-default application binary interface.
11167 @item -fshort-double
11168 @opindex fshort-double
11169 Use the same size for @code{double} as for @code{float}.
11171 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11172 code that is not binary compatible with code generated without that switch.
11173 Use it to conform to a non-default application binary interface.
11175 @item -fshort-wchar
11176 @opindex fshort-wchar
11177 Override the underlying type for @samp{wchar_t} to be @samp{short
11178 unsigned int} instead of the default for the target. This option is
11179 useful for building programs to run under WINE@.
11181 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11182 code that is not binary compatible with code generated without that switch.
11183 Use it to conform to a non-default application binary interface.
11185 @item -fshared-data
11186 @opindex fshared-data
11187 Requests that the data and non-@code{const} variables of this
11188 compilation be shared data rather than private data. The distinction
11189 makes sense only on certain operating systems, where shared data is
11190 shared between processes running the same program, while private data
11191 exists in one copy per process.
11194 @opindex fno-common
11195 In C, allocate even uninitialized global variables in the data section of the
11196 object file, rather than generating them as common blocks. This has the
11197 effect that if the same variable is declared (without @code{extern}) in
11198 two different compilations, you will get an error when you link them.
11199 The only reason this might be useful is if you wish to verify that the
11200 program will work on other systems which always work this way.
11204 Ignore the @samp{#ident} directive.
11206 @item -fno-gnu-linker
11207 @opindex fno-gnu-linker
11208 Do not output global initializations (such as C++ constructors and
11209 destructors) in the form used by the GNU linker (on systems where the GNU
11210 linker is the standard method of handling them). Use this option when
11211 you want to use a non-GNU linker, which also requires using the
11212 @command{collect2} program to make sure the system linker includes
11213 constructors and destructors. (@command{collect2} is included in the GCC
11214 distribution.) For systems which @emph{must} use @command{collect2}, the
11215 compiler driver @command{gcc} is configured to do this automatically.
11217 @item -finhibit-size-directive
11218 @opindex finhibit-size-directive
11219 Don't output a @code{.size} assembler directive, or anything else that
11220 would cause trouble if the function is split in the middle, and the
11221 two halves are placed at locations far apart in memory. This option is
11222 used when compiling @file{crtstuff.c}; you should not need to use it
11225 @item -fverbose-asm
11226 @opindex fverbose-asm
11227 Put extra commentary information in the generated assembly code to
11228 make it more readable. This option is generally only of use to those
11229 who actually need to read the generated assembly code (perhaps while
11230 debugging the compiler itself).
11232 @option{-fno-verbose-asm}, the default, causes the
11233 extra information to be omitted and is useful when comparing two assembler
11238 @cindex global offset table
11240 Generate position-independent code (PIC) suitable for use in a shared
11241 library, if supported for the target machine. Such code accesses all
11242 constant addresses through a global offset table (GOT)@. The dynamic
11243 loader resolves the GOT entries when the program starts (the dynamic
11244 loader is not part of GCC; it is part of the operating system). If
11245 the GOT size for the linked executable exceeds a machine-specific
11246 maximum size, you get an error message from the linker indicating that
11247 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11248 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
11249 on the m68k and RS/6000. The 386 has no such limit.)
11251 Position-independent code requires special support, and therefore works
11252 only on certain machines. For the 386, GCC supports PIC for System V
11253 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11254 position-independent.
11258 If supported for the target machine, emit position-independent code,
11259 suitable for dynamic linking and avoiding any limit on the size of the
11260 global offset table. This option makes a difference on the m68k, m88k,
11263 Position-independent code requires special support, and therefore works
11264 only on certain machines.
11270 These options are similar to @option{-fpic} and @option{-fPIC}, but
11271 generated position independent code can be only linked into executables.
11272 Usually these options are used when @option{-pie} GCC option will be
11273 used during linking.
11275 @item -ffixed-@var{reg}
11277 Treat the register named @var{reg} as a fixed register; generated code
11278 should never refer to it (except perhaps as a stack pointer, frame
11279 pointer or in some other fixed role).
11281 @var{reg} must be the name of a register. The register names accepted
11282 are machine-specific and are defined in the @code{REGISTER_NAMES}
11283 macro in the machine description macro file.
11285 This flag does not have a negative form, because it specifies a
11288 @item -fcall-used-@var{reg}
11289 @opindex fcall-used
11290 Treat the register named @var{reg} as an allocable register that is
11291 clobbered by function calls. It may be allocated for temporaries or
11292 variables that do not live across a call. Functions compiled this way
11293 will not save and restore the register @var{reg}.
11295 It is an error to used this flag with the frame pointer or stack pointer.
11296 Use of this flag for other registers that have fixed pervasive roles in
11297 the machine's execution model will produce disastrous results.
11299 This flag does not have a negative form, because it specifies a
11302 @item -fcall-saved-@var{reg}
11303 @opindex fcall-saved
11304 Treat the register named @var{reg} as an allocable register saved by
11305 functions. It may be allocated even for temporaries or variables that
11306 live across a call. Functions compiled this way will save and restore
11307 the register @var{reg} if they use it.
11309 It is an error to used this flag with the frame pointer or stack pointer.
11310 Use of this flag for other registers that have fixed pervasive roles in
11311 the machine's execution model will produce disastrous results.
11313 A different sort of disaster will result from the use of this flag for
11314 a register in which function values may be returned.
11316 This flag does not have a negative form, because it specifies a
11319 @item -fpack-struct
11320 @opindex fpack-struct
11321 Pack all structure members together without holes.
11323 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11324 code that is not binary compatible with code generated without that switch.
11325 Additionally, it makes the code suboptimal.
11326 Use it to conform to a non-default application binary interface.
11328 @item -finstrument-functions
11329 @opindex finstrument-functions
11330 Generate instrumentation calls for entry and exit to functions. Just
11331 after function entry and just before function exit, the following
11332 profiling functions will be called with the address of the current
11333 function and its call site. (On some platforms,
11334 @code{__builtin_return_address} does not work beyond the current
11335 function, so the call site information may not be available to the
11336 profiling functions otherwise.)
11339 void __cyg_profile_func_enter (void *this_fn,
11341 void __cyg_profile_func_exit (void *this_fn,
11345 The first argument is the address of the start of the current function,
11346 which may be looked up exactly in the symbol table.
11348 This instrumentation is also done for functions expanded inline in other
11349 functions. The profiling calls will indicate where, conceptually, the
11350 inline function is entered and exited. This means that addressable
11351 versions of such functions must be available. If all your uses of a
11352 function are expanded inline, this may mean an additional expansion of
11353 code size. If you use @samp{extern inline} in your C code, an
11354 addressable version of such functions must be provided. (This is
11355 normally the case anyways, but if you get lucky and the optimizer always
11356 expands the functions inline, you might have gotten away without
11357 providing static copies.)
11359 A function may be given the attribute @code{no_instrument_function}, in
11360 which case this instrumentation will not be done. This can be used, for
11361 example, for the profiling functions listed above, high-priority
11362 interrupt routines, and any functions from which the profiling functions
11363 cannot safely be called (perhaps signal handlers, if the profiling
11364 routines generate output or allocate memory).
11366 @item -fstack-check
11367 @opindex fstack-check
11368 Generate code to verify that you do not go beyond the boundary of the
11369 stack. You should specify this flag if you are running in an
11370 environment with multiple threads, but only rarely need to specify it in
11371 a single-threaded environment since stack overflow is automatically
11372 detected on nearly all systems if there is only one stack.
11374 Note that this switch does not actually cause checking to be done; the
11375 operating system must do that. The switch causes generation of code
11376 to ensure that the operating system sees the stack being extended.
11378 @item -fstack-limit-register=@var{reg}
11379 @itemx -fstack-limit-symbol=@var{sym}
11380 @itemx -fno-stack-limit
11381 @opindex fstack-limit-register
11382 @opindex fstack-limit-symbol
11383 @opindex fno-stack-limit
11384 Generate code to ensure that the stack does not grow beyond a certain value,
11385 either the value of a register or the address of a symbol. If the stack
11386 would grow beyond the value, a signal is raised. For most targets,
11387 the signal is raised before the stack overruns the boundary, so
11388 it is possible to catch the signal without taking special precautions.
11390 For instance, if the stack starts at absolute address @samp{0x80000000}
11391 and grows downwards, you can use the flags
11392 @option{-fstack-limit-symbol=__stack_limit} and
11393 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11394 of 128KB@. Note that this may only work with the GNU linker.
11396 @cindex aliasing of parameters
11397 @cindex parameters, aliased
11398 @item -fargument-alias
11399 @itemx -fargument-noalias
11400 @itemx -fargument-noalias-global
11401 @opindex fargument-alias
11402 @opindex fargument-noalias
11403 @opindex fargument-noalias-global
11404 Specify the possible relationships among parameters and between
11405 parameters and global data.
11407 @option{-fargument-alias} specifies that arguments (parameters) may
11408 alias each other and may alias global storage.@*
11409 @option{-fargument-noalias} specifies that arguments do not alias
11410 each other, but may alias global storage.@*
11411 @option{-fargument-noalias-global} specifies that arguments do not
11412 alias each other and do not alias global storage.
11414 Each language will automatically use whatever option is required by
11415 the language standard. You should not need to use these options yourself.
11417 @item -fleading-underscore
11418 @opindex fleading-underscore
11419 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11420 change the way C symbols are represented in the object file. One use
11421 is to help link with legacy assembly code.
11423 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11424 generate code that is not binary compatible with code generated without that
11425 switch. Use it to conform to a non-default application binary interface.
11426 Not all targets provide complete support for this switch.
11428 @item -ftls-model=@var{model}
11429 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11430 The @var{model} argument should be one of @code{global-dynamic},
11431 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11433 The default without @option{-fpic} is @code{initial-exec}; with
11434 @option{-fpic} the default is @code{global-dynamic}.
11439 @node Environment Variables
11440 @section Environment Variables Affecting GCC
11441 @cindex environment variables
11443 @c man begin ENVIRONMENT
11444 This section describes several environment variables that affect how GCC
11445 operates. Some of them work by specifying directories or prefixes to use
11446 when searching for various kinds of files. Some are used to specify other
11447 aspects of the compilation environment.
11449 Note that you can also specify places to search using options such as
11450 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11451 take precedence over places specified using environment variables, which
11452 in turn take precedence over those specified by the configuration of GCC@.
11453 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11454 GNU Compiler Collection (GCC) Internals}.
11459 @c @itemx LC_COLLATE
11461 @c @itemx LC_MONETARY
11462 @c @itemx LC_NUMERIC
11467 @c @findex LC_COLLATE
11468 @findex LC_MESSAGES
11469 @c @findex LC_MONETARY
11470 @c @findex LC_NUMERIC
11474 These environment variables control the way that GCC uses
11475 localization information that allow GCC to work with different
11476 national conventions. GCC inspects the locale categories
11477 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11478 so. These locale categories can be set to any value supported by your
11479 installation. A typical value is @samp{en_UK} for English in the United
11482 The @env{LC_CTYPE} environment variable specifies character
11483 classification. GCC uses it to determine the character boundaries in
11484 a string; this is needed for some multibyte encodings that contain quote
11485 and escape characters that would otherwise be interpreted as a string
11488 The @env{LC_MESSAGES} environment variable specifies the language to
11489 use in diagnostic messages.
11491 If the @env{LC_ALL} environment variable is set, it overrides the value
11492 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11493 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11494 environment variable. If none of these variables are set, GCC
11495 defaults to traditional C English behavior.
11499 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11500 files. GCC uses temporary files to hold the output of one stage of
11501 compilation which is to be used as input to the next stage: for example,
11502 the output of the preprocessor, which is the input to the compiler
11505 @item GCC_EXEC_PREFIX
11506 @findex GCC_EXEC_PREFIX
11507 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11508 names of the subprograms executed by the compiler. No slash is added
11509 when this prefix is combined with the name of a subprogram, but you can
11510 specify a prefix that ends with a slash if you wish.
11512 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11513 an appropriate prefix to use based on the pathname it was invoked with.
11515 If GCC cannot find the subprogram using the specified prefix, it
11516 tries looking in the usual places for the subprogram.
11518 The default value of @env{GCC_EXEC_PREFIX} is
11519 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11520 of @code{prefix} when you ran the @file{configure} script.
11522 Other prefixes specified with @option{-B} take precedence over this prefix.
11524 This prefix is also used for finding files such as @file{crt0.o} that are
11527 In addition, the prefix is used in an unusual way in finding the
11528 directories to search for header files. For each of the standard
11529 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11530 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11531 replacing that beginning with the specified prefix to produce an
11532 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11533 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11534 These alternate directories are searched first; the standard directories
11537 @item COMPILER_PATH
11538 @findex COMPILER_PATH
11539 The value of @env{COMPILER_PATH} is a colon-separated list of
11540 directories, much like @env{PATH}. GCC tries the directories thus
11541 specified when searching for subprograms, if it can't find the
11542 subprograms using @env{GCC_EXEC_PREFIX}.
11545 @findex LIBRARY_PATH
11546 The value of @env{LIBRARY_PATH} is a colon-separated list of
11547 directories, much like @env{PATH}. When configured as a native compiler,
11548 GCC tries the directories thus specified when searching for special
11549 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11550 using GCC also uses these directories when searching for ordinary
11551 libraries for the @option{-l} option (but directories specified with
11552 @option{-L} come first).
11556 @cindex locale definition
11557 This variable is used to pass locale information to the compiler. One way in
11558 which this information is used is to determine the character set to be used
11559 when character literals, string literals and comments are parsed in C and C++.
11560 When the compiler is configured to allow multibyte characters,
11561 the following values for @env{LANG} are recognized:
11565 Recognize JIS characters.
11567 Recognize SJIS characters.
11569 Recognize EUCJP characters.
11572 If @env{LANG} is not defined, or if it has some other value, then the
11573 compiler will use mblen and mbtowc as defined by the default locale to
11574 recognize and translate multibyte characters.
11578 Some additional environments variables affect the behavior of the
11581 @include cppenv.texi
11585 @node Precompiled Headers
11586 @section Using Precompiled Headers
11587 @cindex precompiled headers
11588 @cindex speed of compilation
11590 Often large projects have many header files that are included in every
11591 source file. The time the compiler takes to process these header files
11592 over and over again can account for nearly all of the time required to
11593 build the project. To make builds faster, GCC allows users to
11594 `precompile' a header file; then, if builds can use the precompiled
11595 header file they will be much faster.
11597 To create a precompiled header file, simply compile it as you would any
11598 other file, if necessary using the @option{-x} option to make the driver
11599 treat it as a C or C++ header file. You will probably want to use a
11600 tool like @command{make} to keep the precompiled header up-to-date when
11601 the headers it contains change.
11603 A precompiled header file will be searched for when @code{#include} is
11604 seen in the compilation. As it searches for the included file
11605 (@pxref{Search Path,,Search Path,cpp.info,The C Preprocessor}) the
11606 compiler looks for a precompiled header in each directory just before it
11607 looks for the include file in that directory. The name searched for is
11608 the name specified in the @code{#include} with @samp{.gch} appended. If
11609 the precompiled header file can't be used, it is ignored.
11611 For instance, if you have @code{#include "all.h"}, and you have
11612 @file{all.h.gch} in the same directory as @file{all.h}, then the
11613 precompiled header file will be used if possible, and the original
11614 header will be used otherwise.
11616 Alternatively, you might decide to put the precompiled header file in a
11617 directory and use @option{-I} to ensure that directory is searched
11618 before (or instead of) the directory containing the original header.
11619 Then, if you want to check that the precompiled header file is always
11620 used, you can put a file of the same name as the original header in this
11621 directory containing an @code{#error} command.
11623 This also works with @option{-include}. So yet another way to use
11624 precompiled headers, good for projects not designed with precompiled
11625 header files in mind, is to simply take most of the header files used by
11626 a project, include them from another header file, precompile that header
11627 file, and @option{-include} the precompiled header. If the header files
11628 have guards against multiple inclusion, they will be skipped because
11629 they've already been included (in the precompiled header).
11631 If you need to precompile the same header file for different
11632 languages, targets, or compiler options, you can instead make a
11633 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11634 header in the directory. (It doesn't matter what you call the files
11635 in the directory, every precompiled header in the directory will be
11636 considered.) The first precompiled header encountered in the
11637 directory that is valid for this compilation will be used; they're
11638 searched in no particular order.
11640 There are many other possibilities, limited only by your imagination,
11641 good sense, and the constraints of your build system.
11643 A precompiled header file can be used only when these conditions apply:
11647 Only one precompiled header can be used in a particular compilation.
11649 A precompiled header can't be used once the first C token is seen. You
11650 can have preprocessor directives before a precompiled header; you can
11651 even include a precompiled header from inside another header, so long as
11652 there are no C tokens before the @code{#include}.
11654 The precompiled header file must be produced for the same language as
11655 the current compilation. You can't use a C precompiled header for a C++
11658 The precompiled header file must be produced by the same compiler
11659 version and configuration as the current compilation is using.
11660 The easiest way to guarantee this is to use the same compiler binary
11661 for creating and using precompiled headers.
11663 Any macros defined before the precompiled header (including with
11664 @option{-D}) must either be defined in the same way as when the
11665 precompiled header was generated, or must not affect the precompiled
11666 header, which usually means that the they don't appear in the
11667 precompiled header at all.
11669 Certain command-line options must be defined in the same way as when the
11670 precompiled header was generated. At present, it's not clear which
11671 options are safe to change and which are not; the safest choice is to
11672 use exactly the same options when generating and using the precompiled
11676 For all of these but the last, the compiler will automatically ignore
11677 the precompiled header if the conditions aren't met. For the last item,
11678 some option changes will cause the precompiled header to be rejected,
11679 but not all incompatible option combinations have yet been found. If
11680 you find a new incompatible combination, please consider filing a bug
11681 report, see @ref{Bugs}.
11683 @node Running Protoize
11684 @section Running Protoize
11686 The program @code{protoize} is an optional part of GCC@. You can use
11687 it to add prototypes to a program, thus converting the program to ISO
11688 C in one respect. The companion program @code{unprotoize} does the
11689 reverse: it removes argument types from any prototypes that are found.
11691 When you run these programs, you must specify a set of source files as
11692 command line arguments. The conversion programs start out by compiling
11693 these files to see what functions they define. The information gathered
11694 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11696 After scanning comes actual conversion. The specified files are all
11697 eligible to be converted; any files they include (whether sources or
11698 just headers) are eligible as well.
11700 But not all the eligible files are converted. By default,
11701 @code{protoize} and @code{unprotoize} convert only source and header
11702 files in the current directory. You can specify additional directories
11703 whose files should be converted with the @option{-d @var{directory}}
11704 option. You can also specify particular files to exclude with the
11705 @option{-x @var{file}} option. A file is converted if it is eligible, its
11706 directory name matches one of the specified directory names, and its
11707 name within the directory has not been excluded.
11709 Basic conversion with @code{protoize} consists of rewriting most
11710 function definitions and function declarations to specify the types of
11711 the arguments. The only ones not rewritten are those for varargs
11714 @code{protoize} optionally inserts prototype declarations at the
11715 beginning of the source file, to make them available for any calls that
11716 precede the function's definition. Or it can insert prototype
11717 declarations with block scope in the blocks where undeclared functions
11720 Basic conversion with @code{unprotoize} consists of rewriting most
11721 function declarations to remove any argument types, and rewriting
11722 function definitions to the old-style pre-ISO form.
11724 Both conversion programs print a warning for any function declaration or
11725 definition that they can't convert. You can suppress these warnings
11728 The output from @code{protoize} or @code{unprotoize} replaces the
11729 original source file. The original file is renamed to a name ending
11730 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11731 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11732 for DOS) file already exists, then the source file is simply discarded.
11734 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11735 scan the program and collect information about the functions it uses.
11736 So neither of these programs will work until GCC is installed.
11738 Here is a table of the options you can use with @code{protoize} and
11739 @code{unprotoize}. Each option works with both programs unless
11743 @item -B @var{directory}
11744 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11745 usual directory (normally @file{/usr/local/lib}). This file contains
11746 prototype information about standard system functions. This option
11747 applies only to @code{protoize}.
11749 @item -c @var{compilation-options}
11750 Use @var{compilation-options} as the options when running @command{gcc} to
11751 produce the @samp{.X} files. The special option @option{-aux-info} is
11752 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11754 Note that the compilation options must be given as a single argument to
11755 @code{protoize} or @code{unprotoize}. If you want to specify several
11756 @command{gcc} options, you must quote the entire set of compilation options
11757 to make them a single word in the shell.
11759 There are certain @command{gcc} arguments that you cannot use, because they
11760 would produce the wrong kind of output. These include @option{-g},
11761 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11762 the @var{compilation-options}, they are ignored.
11765 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11766 systems) instead of @samp{.c}. This is convenient if you are converting
11767 a C program to C++. This option applies only to @code{protoize}.
11770 Add explicit global declarations. This means inserting explicit
11771 declarations at the beginning of each source file for each function
11772 that is called in the file and was not declared. These declarations
11773 precede the first function definition that contains a call to an
11774 undeclared function. This option applies only to @code{protoize}.
11776 @item -i @var{string}
11777 Indent old-style parameter declarations with the string @var{string}.
11778 This option applies only to @code{protoize}.
11780 @code{unprotoize} converts prototyped function definitions to old-style
11781 function definitions, where the arguments are declared between the
11782 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11783 uses five spaces as the indentation. If you want to indent with just
11784 one space instead, use @option{-i " "}.
11787 Keep the @samp{.X} files. Normally, they are deleted after conversion
11791 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11792 a prototype declaration for each function in each block which calls the
11793 function without any declaration. This option applies only to
11797 Make no real changes. This mode just prints information about the conversions
11798 that would have been done without @option{-n}.
11801 Make no @samp{.save} files. The original files are simply deleted.
11802 Use this option with caution.
11804 @item -p @var{program}
11805 Use the program @var{program} as the compiler. Normally, the name
11806 @file{gcc} is used.
11809 Work quietly. Most warnings are suppressed.
11812 Print the version number, just like @option{-v} for @command{gcc}.
11815 If you need special compiler options to compile one of your program's
11816 source files, then you should generate that file's @samp{.X} file
11817 specially, by running @command{gcc} on that source file with the
11818 appropriate options and the option @option{-aux-info}. Then run
11819 @code{protoize} on the entire set of files. @code{protoize} will use
11820 the existing @samp{.X} file because it is newer than the source file.
11824 gcc -Dfoo=bar file1.c -aux-info file1.X
11829 You need to include the special files along with the rest in the
11830 @code{protoize} command, even though their @samp{.X} files already
11831 exist, because otherwise they won't get converted.
11833 @xref{Protoize Caveats}, for more information on how to use
11834 @code{protoize} successfully.