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 -Wold-style-definition -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 -nofixprebinding -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{-mconst16 -mno-const16 @gol
641 -mfused-madd -mno-fused-madd @gol
642 -mtext-section-literals -mno-text-section-literals @gol
643 -mtarget-align -mno-target-align @gol
644 -mlongcalls -mno-longcalls}
647 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
648 -mhard-float -msoft-float @gol
649 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
650 -mdouble -mno-double @gol
651 -mmedia -mno-media -mmuladd -mno-muladd @gol
652 -mlibrary-pic -macc-4 -macc-8 @gol
653 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
654 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
655 -mvliw-branch -mno-vliw-branch @gol
656 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
657 -mno-nested-cond-exec -mtomcat-stats @gol
660 @item Code Generation Options
661 @xref{Code Gen Options,,Options for Code Generation Conventions}.
662 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
663 -ffixed-@var{reg} -fexceptions @gol
664 -fnon-call-exceptions -funwind-tables @gol
665 -fasynchronous-unwind-tables @gol
666 -finhibit-size-directive -finstrument-functions @gol
667 -fno-common -fno-ident -fno-gnu-linker @gol
668 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
669 -freg-struct-return -fshared-data -fshort-enums @gol
670 -fshort-double -fshort-wchar @gol
671 -fverbose-asm -fpack-struct -fstack-check @gol
672 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
673 -fargument-alias -fargument-noalias @gol
674 -fargument-noalias-global -fleading-underscore @gol
675 -ftls-model=@var{model} @gol
676 -ftrapv -fwrapv -fbounds-check}
680 * Overall Options:: Controlling the kind of output:
681 an executable, object files, assembler files,
682 or preprocessed source.
683 * C Dialect Options:: Controlling the variant of C language compiled.
684 * C++ Dialect Options:: Variations on C++.
685 * Objective-C Dialect Options:: Variations on Objective-C.
686 * Language Independent Options:: Controlling how diagnostics should be
688 * Warning Options:: How picky should the compiler be?
689 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
690 * Optimize Options:: How much optimization?
691 * Preprocessor Options:: Controlling header files and macro definitions.
692 Also, getting dependency information for Make.
693 * Assembler Options:: Passing options to the assembler.
694 * Link Options:: Specifying libraries and so on.
695 * Directory Options:: Where to find header files and libraries.
696 Where to find the compiler executable files.
697 * Spec Files:: How to pass switches to sub-processes.
698 * Target Options:: Running a cross-compiler, or an old version of GCC.
701 @node Overall Options
702 @section Options Controlling the Kind of Output
704 Compilation can involve up to four stages: preprocessing, compilation
705 proper, assembly and linking, always in that order. GCC is capable of
706 preprocessing and compiling several files either into several
707 assembler input files, or into one assembler input file; then each
708 assembler input file produces an object file, and linking combines all
709 the object files (those newly compiled, and those specified as input)
710 into an executable file.
712 @cindex file name suffix
713 For any given input file, the file name suffix determines what kind of
718 C source code which must be preprocessed.
721 C source code which should not be preprocessed.
724 C++ source code which should not be preprocessed.
727 Objective-C source code. Note that you must link with the library
728 @file{libobjc.a} to make an Objective-C program work.
731 Objective-C source code which should not be preprocessed.
734 C or C++ header file to be turned into a precompiled header.
738 @itemx @var{file}.cxx
739 @itemx @var{file}.cpp
740 @itemx @var{file}.CPP
741 @itemx @var{file}.c++
743 C++ source code which must be preprocessed. Note that in @samp{.cxx},
744 the last two letters must both be literally @samp{x}. Likewise,
745 @samp{.C} refers to a literal capital C@.
749 C++ header file to be turned into a precompiled header.
752 @itemx @var{file}.for
753 @itemx @var{file}.FOR
754 Fortran source code which should not be preprocessed.
757 @itemx @var{file}.fpp
758 @itemx @var{file}.FPP
759 Fortran source code which must be preprocessed (with the traditional
763 Fortran source code which must be preprocessed with a RATFOR
764 preprocessor (not included with GCC)@.
766 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
767 Using and Porting GNU Fortran}, for more details of the handling of
770 @c FIXME: Descriptions of Java file types.
777 Ada source code file which contains a library unit declaration (a
778 declaration of a package, subprogram, or generic, or a generic
779 instantiation), or a library unit renaming declaration (a package,
780 generic, or subprogram renaming declaration). Such files are also
783 @itemx @var{file}.adb
784 Ada source code file containing a library unit body (a subprogram or
785 package body). Such files are also called @dfn{bodies}.
787 @c GCC also knows about some suffixes for languages not yet included:
796 Assembler code which must be preprocessed.
799 An object file to be fed straight into linking.
800 Any file name with no recognized suffix is treated this way.
804 You can specify the input language explicitly with the @option{-x} option:
807 @item -x @var{language}
808 Specify explicitly the @var{language} for the following input files
809 (rather than letting the compiler choose a default based on the file
810 name suffix). This option applies to all following input files until
811 the next @option{-x} option. Possible values for @var{language} are:
813 c c-header cpp-output
814 c++ c++-header c++-cpp-output
815 objective-c objective-c-header objc-cpp-output
816 assembler assembler-with-cpp
818 f77 f77-cpp-input ratfor
824 Turn off any specification of a language, so that subsequent files are
825 handled according to their file name suffixes (as they are if @option{-x}
826 has not been used at all).
828 @item -pass-exit-codes
829 @opindex pass-exit-codes
830 Normally the @command{gcc} program will exit with the code of 1 if any
831 phase of the compiler returns a non-success return code. If you specify
832 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
833 numerically highest error produced by any phase that returned an error
837 If you only want some of the stages of compilation, you can use
838 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
839 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
840 @command{gcc} is to stop. Note that some combinations (for example,
841 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
846 Compile or assemble the source files, but do not link. The linking
847 stage simply is not done. The ultimate output is in the form of an
848 object file for each source file.
850 By default, the object file name for a source file is made by replacing
851 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
853 Unrecognized input files, not requiring compilation or assembly, are
858 Stop after the stage of compilation proper; do not assemble. The output
859 is in the form of an assembler code file for each non-assembler input
862 By default, the assembler file name for a source file is made by
863 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
865 Input files that don't require compilation are ignored.
869 Stop after the preprocessing stage; do not run the compiler proper. The
870 output is in the form of preprocessed source code, which is sent to the
873 Input files which don't require preprocessing are ignored.
875 @cindex output file option
878 Place output in file @var{file}. This applies regardless to whatever
879 sort of output is being produced, whether it be an executable file,
880 an object file, an assembler file or preprocessed C code.
882 If you specify @option{-o} when compiling more than one input file, or
883 you are producing an executable file as output, all the source files
884 on the command line will be compiled at once.
886 If @option{-o} is not specified, the default is to put an executable file
887 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
888 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
889 all preprocessed C source on standard output.
893 Print (on standard error output) the commands executed to run the stages
894 of compilation. Also print the version number of the compiler driver
895 program and of the preprocessor and the compiler proper.
899 Like @option{-v} except the commands are not executed and all command
900 arguments are quoted. This is useful for shell scripts to capture the
901 driver-generated command lines.
905 Use pipes rather than temporary files for communication between the
906 various stages of compilation. This fails to work on some systems where
907 the assembler is unable to read from a pipe; but the GNU assembler has
912 Print (on the standard output) a description of the command line options
913 understood by @command{gcc}. If the @option{-v} option is also specified
914 then @option{--help} will also be passed on to the various processes
915 invoked by @command{gcc}, so that they can display the command line options
916 they accept. If the @option{-Wextra} option is also specified then command
917 line options which have no documentation associated with them will also
922 Print (on the standard output) a description of target specific command
923 line options for each tool.
927 Display the version number and copyrights of the invoked GCC.
931 @section Compiling C++ Programs
933 @cindex suffixes for C++ source
934 @cindex C++ source file suffixes
935 C++ source files conventionally use one of the suffixes @samp{.C},
936 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
937 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
938 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
939 files with these names and compiles them as C++ programs even if you
940 call the compiler the same way as for compiling C programs (usually
941 with the name @command{gcc}).
945 However, C++ programs often require class libraries as well as a
946 compiler that understands the C++ language---and under some
947 circumstances, you might want to compile programs or header files from
948 standard input, or otherwise without a suffix that flags them as C++
949 programs. You might also like to precompile a C header file with a
950 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
951 program that calls GCC with the default language set to C++, and
952 automatically specifies linking against the C++ library. On many
953 systems, @command{g++} is also installed with the name @command{c++}.
955 @cindex invoking @command{g++}
956 When you compile C++ programs, you may specify many of the same
957 command-line options that you use for compiling programs in any
958 language; or command-line options meaningful for C and related
959 languages; or options that are meaningful only for C++ programs.
960 @xref{C Dialect Options,,Options Controlling C Dialect}, for
961 explanations of options for languages related to C@.
962 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
963 explanations of options that are meaningful only for C++ programs.
965 @node C Dialect Options
966 @section Options Controlling C Dialect
967 @cindex dialect options
968 @cindex language dialect options
969 @cindex options, dialect
971 The following options control the dialect of C (or languages derived
972 from C, such as C++ and Objective-C) that the compiler accepts:
979 In C mode, support all ISO C90 programs. In C++ mode,
980 remove GNU extensions that conflict with ISO C++.
982 This turns off certain features of GCC that are incompatible with ISO
983 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
984 such as the @code{asm} and @code{typeof} keywords, and
985 predefined macros such as @code{unix} and @code{vax} that identify the
986 type of system you are using. It also enables the undesirable and
987 rarely used ISO trigraph feature. For the C compiler,
988 it disables recognition of C++ style @samp{//} comments as well as
989 the @code{inline} keyword.
991 The alternate keywords @code{__asm__}, @code{__extension__},
992 @code{__inline__} and @code{__typeof__} continue to work despite
993 @option{-ansi}. You would not want to use them in an ISO C program, of
994 course, but it is useful to put them in header files that might be included
995 in compilations done with @option{-ansi}. Alternate predefined macros
996 such as @code{__unix__} and @code{__vax__} are also available, with or
997 without @option{-ansi}.
999 The @option{-ansi} option does not cause non-ISO programs to be
1000 rejected gratuitously. For that, @option{-pedantic} is required in
1001 addition to @option{-ansi}. @xref{Warning Options}.
1003 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1004 option is used. Some header files may notice this macro and refrain
1005 from declaring certain functions or defining certain macros that the
1006 ISO standard doesn't call for; this is to avoid interfering with any
1007 programs that might use these names for other things.
1009 Functions which would normally be built in but do not have semantics
1010 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1011 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1012 built-in functions provided by GCC}, for details of the functions
1017 Determine the language standard. This option is currently only
1018 supported when compiling C or C++. A value for this option must be
1019 provided; possible values are
1024 ISO C90 (same as @option{-ansi}).
1026 @item iso9899:199409
1027 ISO C90 as modified in amendment 1.
1033 ISO C99. Note that this standard is not yet fully supported; see
1034 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1035 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1038 Default, ISO C90 plus GNU extensions (including some C99 features).
1042 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1043 this will become the default. The name @samp{gnu9x} is deprecated.
1046 The 1998 ISO C++ standard plus amendments.
1049 The same as @option{-std=c++98} plus GNU extensions. This is the
1050 default for C++ code.
1053 Even when this option is not specified, you can still use some of the
1054 features of newer standards in so far as they do not conflict with
1055 previous C standards. For example, you may use @code{__restrict__} even
1056 when @option{-std=c99} is not specified.
1058 The @option{-std} options specifying some version of ISO C have the same
1059 effects as @option{-ansi}, except that features that were not in ISO C90
1060 but are in the specified version (for example, @samp{//} comments and
1061 the @code{inline} keyword in ISO C99) are not disabled.
1063 @xref{Standards,,Language Standards Supported by GCC}, for details of
1064 these standard versions.
1066 @item -aux-info @var{filename}
1068 Output to the given filename prototyped declarations for all functions
1069 declared and/or defined in a translation unit, including those in header
1070 files. This option is silently ignored in any language other than C@.
1072 Besides declarations, the file indicates, in comments, the origin of
1073 each declaration (source file and line), whether the declaration was
1074 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1075 @samp{O} for old, respectively, in the first character after the line
1076 number and the colon), and whether it came from a declaration or a
1077 definition (@samp{C} or @samp{F}, respectively, in the following
1078 character). In the case of function definitions, a K&R-style list of
1079 arguments followed by their declarations is also provided, inside
1080 comments, after the declaration.
1084 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1085 keyword, so that code can use these words as identifiers. You can use
1086 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1087 instead. @option{-ansi} implies @option{-fno-asm}.
1089 In C++, this switch only affects the @code{typeof} keyword, since
1090 @code{asm} and @code{inline} are standard keywords. You may want to
1091 use the @option{-fno-gnu-keywords} flag instead, which has the same
1092 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1093 switch only affects the @code{asm} and @code{typeof} keywords, since
1094 @code{inline} is a standard keyword in ISO C99.
1097 @itemx -fno-builtin-@var{function}
1098 @opindex fno-builtin
1099 @cindex built-in functions
1100 Don't recognize built-in functions that do not begin with
1101 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1102 functions provided by GCC}, for details of the functions affected,
1103 including those which are not built-in functions when @option{-ansi} or
1104 @option{-std} options for strict ISO C conformance are used because they
1105 do not have an ISO standard meaning.
1107 GCC normally generates special code to handle certain built-in functions
1108 more efficiently; for instance, calls to @code{alloca} may become single
1109 instructions that adjust the stack directly, and calls to @code{memcpy}
1110 may become inline copy loops. The resulting code is often both smaller
1111 and faster, but since the function calls no longer appear as such, you
1112 cannot set a breakpoint on those calls, nor can you change the behavior
1113 of the functions by linking with a different library.
1115 With the @option{-fno-builtin-@var{function}} option
1116 only the built-in function @var{function} is
1117 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1118 function is named this is not built-in in this version of GCC, this
1119 option is ignored. There is no corresponding
1120 @option{-fbuiltin-@var{function}} option; if you wish to enable
1121 built-in functions selectively when using @option{-fno-builtin} or
1122 @option{-ffreestanding}, you may define macros such as:
1125 #define abs(n) __builtin_abs ((n))
1126 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1131 @cindex hosted environment
1133 Assert that compilation takes place in a hosted environment. This implies
1134 @option{-fbuiltin}. A hosted environment is one in which the
1135 entire standard library is available, and in which @code{main} has a return
1136 type of @code{int}. Examples are nearly everything except a kernel.
1137 This is equivalent to @option{-fno-freestanding}.
1139 @item -ffreestanding
1140 @opindex ffreestanding
1141 @cindex hosted environment
1143 Assert that compilation takes place in a freestanding environment. This
1144 implies @option{-fno-builtin}. A freestanding environment
1145 is one in which the standard library may not exist, and program startup may
1146 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1147 This is equivalent to @option{-fno-hosted}.
1149 @xref{Standards,,Language Standards Supported by GCC}, for details of
1150 freestanding and hosted environments.
1152 @item -fms-extensions
1153 @opindex fms-extensions
1154 Accept some non-standard constructs used in Microsoft header files.
1158 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1159 options for strict ISO C conformance) implies @option{-trigraphs}.
1161 @item -no-integrated-cpp
1162 @opindex no-integrated-cpp
1163 Performs a compilation in two passes: preprocessing and compiling. This
1164 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1165 @option{-B} option. The user supplied compilation step can then add in
1166 an additional preprocessing step after normal preprocessing but before
1167 compiling. The default is to use the integrated cpp (internal cpp)
1169 The semantics of this option will change if "cc1", "cc1plus", and
1170 "cc1obj" are merged.
1172 @cindex traditional C language
1173 @cindex C language, traditional
1175 @itemx -traditional-cpp
1176 @opindex traditional-cpp
1177 @opindex traditional
1178 Formerly, these options caused GCC to attempt to emulate a pre-standard
1179 C compiler. They are now only supported with the @option{-E} switch.
1180 The preprocessor continues to support a pre-standard mode. See the GNU
1181 CPP manual for details.
1183 @item -fcond-mismatch
1184 @opindex fcond-mismatch
1185 Allow conditional expressions with mismatched types in the second and
1186 third arguments. The value of such an expression is void. This option
1187 is not supported for C++.
1189 @item -funsigned-char
1190 @opindex funsigned-char
1191 Let the type @code{char} be unsigned, like @code{unsigned char}.
1193 Each kind of machine has a default for what @code{char} should
1194 be. It is either like @code{unsigned char} by default or like
1195 @code{signed char} by default.
1197 Ideally, a portable program should always use @code{signed char} or
1198 @code{unsigned char} when it depends on the signedness of an object.
1199 But many programs have been written to use plain @code{char} and
1200 expect it to be signed, or expect it to be unsigned, depending on the
1201 machines they were written for. This option, and its inverse, let you
1202 make such a program work with the opposite default.
1204 The type @code{char} is always a distinct type from each of
1205 @code{signed char} or @code{unsigned char}, even though its behavior
1206 is always just like one of those two.
1209 @opindex fsigned-char
1210 Let the type @code{char} be signed, like @code{signed char}.
1212 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1213 the negative form of @option{-funsigned-char}. Likewise, the option
1214 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1216 @item -fsigned-bitfields
1217 @itemx -funsigned-bitfields
1218 @itemx -fno-signed-bitfields
1219 @itemx -fno-unsigned-bitfields
1220 @opindex fsigned-bitfields
1221 @opindex funsigned-bitfields
1222 @opindex fno-signed-bitfields
1223 @opindex fno-unsigned-bitfields
1224 These options control whether a bit-field is signed or unsigned, when the
1225 declaration does not use either @code{signed} or @code{unsigned}. By
1226 default, such a bit-field is signed, because this is consistent: the
1227 basic integer types such as @code{int} are signed types.
1229 @item -fwritable-strings
1230 @opindex fwritable-strings
1231 Store string constants in the writable data segment and don't uniquize
1232 them. This is for compatibility with old programs which assume they can
1233 write into string constants.
1235 Writing into string constants is a very bad idea; ``constants'' should
1239 @node C++ Dialect Options
1240 @section Options Controlling C++ Dialect
1242 @cindex compiler options, C++
1243 @cindex C++ options, command line
1244 @cindex options, C++
1245 This section describes the command-line options that are only meaningful
1246 for C++ programs; but you can also use most of the GNU compiler options
1247 regardless of what language your program is in. For example, you
1248 might compile a file @code{firstClass.C} like this:
1251 g++ -g -frepo -O -c firstClass.C
1255 In this example, only @option{-frepo} is an option meant
1256 only for C++ programs; you can use the other options with any
1257 language supported by GCC@.
1259 Here is a list of options that are @emph{only} for compiling C++ programs:
1263 @item -fabi-version=@var{n}
1264 @opindex fabi-version
1265 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1266 ABI that first appeared in G++ 3.2. Version 0 will always be the
1267 version that conforms most closely to the C++ ABI specification.
1268 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1271 The default is version 1.
1273 @item -fno-access-control
1274 @opindex fno-access-control
1275 Turn off all access checking. This switch is mainly useful for working
1276 around bugs in the access control code.
1280 Check that the pointer returned by @code{operator new} is non-null
1281 before attempting to modify the storage allocated. This check is
1282 normally unnecessary because the C++ standard specifies that
1283 @code{operator new} will only return @code{0} if it is declared
1284 @samp{throw()}, in which case the compiler will always check the
1285 return value even without this option. In all other cases, when
1286 @code{operator new} has a non-empty exception specification, memory
1287 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1288 @samp{new (nothrow)}.
1290 @item -fconserve-space
1291 @opindex fconserve-space
1292 Put uninitialized or runtime-initialized global variables into the
1293 common segment, as C does. This saves space in the executable at the
1294 cost of not diagnosing duplicate definitions. If you compile with this
1295 flag and your program mysteriously crashes after @code{main()} has
1296 completed, you may have an object that is being destroyed twice because
1297 two definitions were merged.
1299 This option is no longer useful on most targets, now that support has
1300 been added for putting variables into BSS without making them common.
1302 @item -fno-const-strings
1303 @opindex fno-const-strings
1304 Give string constants type @code{char *} instead of type @code{const
1305 char *}. By default, G++ uses type @code{const char *} as required by
1306 the standard. Even if you use @option{-fno-const-strings}, you cannot
1307 actually modify the value of a string constant, unless you also use
1308 @option{-fwritable-strings}.
1310 This option might be removed in a future release of G++. For maximum
1311 portability, you should structure your code so that it works with
1312 string constants that have type @code{const char *}.
1314 @item -fno-elide-constructors
1315 @opindex fno-elide-constructors
1316 The C++ standard allows an implementation to omit creating a temporary
1317 which is only used to initialize another object of the same type.
1318 Specifying this option disables that optimization, and forces G++ to
1319 call the copy constructor in all cases.
1321 @item -fno-enforce-eh-specs
1322 @opindex fno-enforce-eh-specs
1323 Don't check for violation of exception specifications at runtime. This
1324 option violates the C++ standard, but may be useful for reducing code
1325 size in production builds, much like defining @samp{NDEBUG}. The compiler
1326 will still optimize based on the exception specifications.
1328 @item -fexternal-templates
1329 @opindex fexternal-templates
1331 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1332 template instantiation; template instances are emitted or not according
1333 to the location of the template definition. @xref{Template
1334 Instantiation}, for more information.
1336 This option is deprecated.
1338 @item -falt-external-templates
1339 @opindex falt-external-templates
1340 Similar to @option{-fexternal-templates}, but template instances are
1341 emitted or not according to the place where they are first instantiated.
1342 @xref{Template Instantiation}, for more information.
1344 This option is deprecated.
1347 @itemx -fno-for-scope
1349 @opindex fno-for-scope
1350 If @option{-ffor-scope} is specified, the scope of variables declared in
1351 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1352 as specified by the C++ standard.
1353 If @option{-fno-for-scope} is specified, the scope of variables declared in
1354 a @i{for-init-statement} extends to the end of the enclosing scope,
1355 as was the case in old versions of G++, and other (traditional)
1356 implementations of C++.
1358 The default if neither flag is given to follow the standard,
1359 but to allow and give a warning for old-style code that would
1360 otherwise be invalid, or have different behavior.
1362 @item -fno-gnu-keywords
1363 @opindex fno-gnu-keywords
1364 Do not recognize @code{typeof} as a keyword, so that code can use this
1365 word as an identifier. You can use the keyword @code{__typeof__} instead.
1366 @option{-ansi} implies @option{-fno-gnu-keywords}.
1368 @item -fno-implicit-templates
1369 @opindex fno-implicit-templates
1370 Never emit code for non-inline templates which are instantiated
1371 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1372 @xref{Template Instantiation}, for more information.
1374 @item -fno-implicit-inline-templates
1375 @opindex fno-implicit-inline-templates
1376 Don't emit code for implicit instantiations of inline templates, either.
1377 The default is to handle inlines differently so that compiles with and
1378 without optimization will need the same set of explicit instantiations.
1380 @item -fno-implement-inlines
1381 @opindex fno-implement-inlines
1382 To save space, do not emit out-of-line copies of inline functions
1383 controlled by @samp{#pragma implementation}. This will cause linker
1384 errors if these functions are not inlined everywhere they are called.
1386 @item -fms-extensions
1387 @opindex fms-extensions
1388 Disable pedantic warnings about constructs used in MFC, such as implicit
1389 int and getting a pointer to member function via non-standard syntax.
1391 @item -fno-nonansi-builtins
1392 @opindex fno-nonansi-builtins
1393 Disable built-in declarations of functions that are not mandated by
1394 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1395 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1397 @item -fno-operator-names
1398 @opindex fno-operator-names
1399 Do not treat the operator name keywords @code{and}, @code{bitand},
1400 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1401 synonyms as keywords.
1403 @item -fno-optional-diags
1404 @opindex fno-optional-diags
1405 Disable diagnostics that the standard says a compiler does not need to
1406 issue. Currently, the only such diagnostic issued by G++ is the one for
1407 a name having multiple meanings within a class.
1410 @opindex fpermissive
1411 Downgrade some diagnostics about nonconformant code from errors to
1412 warnings. Thus, using @option{-fpermissive} will allow some
1413 nonconforming code to compile.
1417 Enable automatic template instantiation at link time. This option also
1418 implies @option{-fno-implicit-templates}. @xref{Template
1419 Instantiation}, for more information.
1423 Disable generation of information about every class with virtual
1424 functions for use by the C++ runtime type identification features
1425 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1426 of the language, you can save some space by using this flag. Note that
1427 exception handling uses the same information, but it will generate it as
1432 Emit statistics about front-end processing at the end of the compilation.
1433 This information is generally only useful to the G++ development team.
1435 @item -ftemplate-depth-@var{n}
1436 @opindex ftemplate-depth
1437 Set the maximum instantiation depth for template classes to @var{n}.
1438 A limit on the template instantiation depth is needed to detect
1439 endless recursions during template class instantiation. ANSI/ISO C++
1440 conforming programs must not rely on a maximum depth greater than 17.
1442 @item -fuse-cxa-atexit
1443 @opindex fuse-cxa-atexit
1444 Register destructors for objects with static storage duration with the
1445 @code{__cxa_atexit} function rather than the @code{atexit} function.
1446 This option is required for fully standards-compliant handling of static
1447 destructors, but will only work if your C library supports
1448 @code{__cxa_atexit}.
1452 Do not use weak symbol support, even if it is provided by the linker.
1453 By default, G++ will use weak symbols if they are available. This
1454 option exists only for testing, and should not be used by end-users;
1455 it will result in inferior code and has no benefits. This option may
1456 be removed in a future release of G++.
1460 Do not search for header files in the standard directories specific to
1461 C++, but do still search the other standard directories. (This option
1462 is used when building the C++ library.)
1465 In addition, these optimization, warning, and code generation options
1466 have meanings only for C++ programs:
1469 @item -fno-default-inline
1470 @opindex fno-default-inline
1471 Do not assume @samp{inline} for functions defined inside a class scope.
1472 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1473 functions will have linkage like inline functions; they just won't be
1476 @item -Wabi @r{(C++ only)}
1478 Warn when G++ generates code that is probably not compatible with the
1479 vendor-neutral C++ ABI. Although an effort has been made to warn about
1480 all such cases, there are probably some cases that are not warned about,
1481 even though G++ is generating incompatible code. There may also be
1482 cases where warnings are emitted even though the code that is generated
1485 You should rewrite your code to avoid these warnings if you are
1486 concerned about the fact that code generated by G++ may not be binary
1487 compatible with code generated by other compilers.
1489 The known incompatibilities at this point include:
1494 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1495 pack data into the same byte as a base class. For example:
1498 struct A @{ virtual void f(); int f1 : 1; @};
1499 struct B : public A @{ int f2 : 1; @};
1503 In this case, G++ will place @code{B::f2} into the same byte
1504 as@code{A::f1}; other compilers will not. You can avoid this problem
1505 by explicitly padding @code{A} so that its size is a multiple of the
1506 byte size on your platform; that will cause G++ and other compilers to
1507 layout @code{B} identically.
1510 Incorrect handling of tail-padding for virtual bases. G++ does not use
1511 tail padding when laying out virtual bases. For example:
1514 struct A @{ virtual void f(); char c1; @};
1515 struct B @{ B(); char c2; @};
1516 struct C : public A, public virtual B @{@};
1520 In this case, G++ will not place @code{B} into the tail-padding for
1521 @code{A}; other compilers will. You can avoid this problem by
1522 explicitly padding @code{A} so that its size is a multiple of its
1523 alignment (ignoring virtual base classes); that will cause G++ and other
1524 compilers to layout @code{C} identically.
1527 Incorrect handling of bit-fields with declared widths greater than that
1528 of their underlying types, when the bit-fields appear in a union. For
1532 union U @{ int i : 4096; @};
1536 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1537 union too small by the number of bits in an @code{int}.
1540 Empty classes can be placed at incorrect offsets. For example:
1550 struct C : public B, public A @{@};
1554 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1555 it should be placed at offset zero. G++ mistakenly believes that the
1556 @code{A} data member of @code{B} is already at offset zero.
1559 Names of template functions whose types involve @code{typename} or
1560 template template parameters can be mangled incorrectly.
1563 template <typename Q>
1564 void f(typename Q::X) @{@}
1566 template <template <typename> class Q>
1567 void f(typename Q<int>::X) @{@}
1571 Instantiations of these templates may be mangled incorrectly.
1575 @item -Wctor-dtor-privacy @r{(C++ only)}
1576 @opindex Wctor-dtor-privacy
1577 Warn when a class seems unusable because all the constructors or
1578 destructors in that class are private, and it has neither friends nor
1579 public static member functions.
1581 @item -Wnon-virtual-dtor @r{(C++ only)}
1582 @opindex Wnon-virtual-dtor
1583 Warn when a class appears to be polymorphic, thereby requiring a virtual
1584 destructor, yet it declares a non-virtual one.
1585 This warning is enabled by @option{-Wall}.
1587 @item -Wreorder @r{(C++ only)}
1589 @cindex reordering, warning
1590 @cindex warning for reordering of member initializers
1591 Warn when the order of member initializers given in the code does not
1592 match the order in which they must be executed. For instance:
1598 A(): j (0), i (1) @{ @}
1602 The compiler will rearrange the member initializers for @samp{i}
1603 and @samp{j} to match the declaration order of the members, emitting
1604 a warning to that effect. This warning is enabled by @option{-Wall}.
1607 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1610 @item -Weffc++ @r{(C++ only)}
1612 Warn about violations of the following style guidelines from Scott Meyers'
1613 @cite{Effective C++} book:
1617 Item 11: Define a copy constructor and an assignment operator for classes
1618 with dynamically allocated memory.
1621 Item 12: Prefer initialization to assignment in constructors.
1624 Item 14: Make destructors virtual in base classes.
1627 Item 15: Have @code{operator=} return a reference to @code{*this}.
1630 Item 23: Don't try to return a reference when you must return an object.
1634 Also warn about violations of the following style guidelines from
1635 Scott Meyers' @cite{More Effective C++} book:
1639 Item 6: Distinguish between prefix and postfix forms of increment and
1640 decrement operators.
1643 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1647 When selecting this option, be aware that the standard library
1648 headers do not obey all of these guidelines; use @samp{grep -v}
1649 to filter out those warnings.
1651 @item -Wno-deprecated @r{(C++ only)}
1652 @opindex Wno-deprecated
1653 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1655 @item -Wno-non-template-friend @r{(C++ only)}
1656 @opindex Wno-non-template-friend
1657 Disable warnings when non-templatized friend functions are declared
1658 within a template. Since the advent of explicit template specification
1659 support in G++, if the name of the friend is an unqualified-id (i.e.,
1660 @samp{friend foo(int)}), the C++ language specification demands that the
1661 friend declare or define an ordinary, nontemplate function. (Section
1662 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1663 could be interpreted as a particular specialization of a templatized
1664 function. Because this non-conforming behavior is no longer the default
1665 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1666 check existing code for potential trouble spots and is on by default.
1667 This new compiler behavior can be turned off with
1668 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1669 but disables the helpful warning.
1671 @item -Wold-style-cast @r{(C++ only)}
1672 @opindex Wold-style-cast
1673 Warn if an old-style (C-style) cast to a non-void type is used within
1674 a C++ program. The new-style casts (@samp{static_cast},
1675 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1676 unintended effects and much easier to search for.
1678 @item -Woverloaded-virtual @r{(C++ only)}
1679 @opindex Woverloaded-virtual
1680 @cindex overloaded virtual fn, warning
1681 @cindex warning for overloaded virtual fn
1682 Warn when a function declaration hides virtual functions from a
1683 base class. For example, in:
1690 struct B: public A @{
1695 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1703 will fail to compile.
1705 @item -Wno-pmf-conversions @r{(C++ only)}
1706 @opindex Wno-pmf-conversions
1707 Disable the diagnostic for converting a bound pointer to member function
1710 @item -Wsign-promo @r{(C++ only)}
1711 @opindex Wsign-promo
1712 Warn when overload resolution chooses a promotion from unsigned or
1713 enumeral type to a signed type, over a conversion to an unsigned type of
1714 the same size. Previous versions of G++ would try to preserve
1715 unsignedness, but the standard mandates the current behavior.
1717 @item -Wsynth @r{(C++ only)}
1719 @cindex warning for synthesized methods
1720 @cindex synthesized methods, warning
1721 Warn when G++'s synthesis behavior does not match that of cfront. For
1727 A& operator = (int);
1737 In this example, G++ will synthesize a default @samp{A& operator =
1738 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1741 @node Objective-C Dialect Options
1742 @section Options Controlling Objective-C Dialect
1744 @cindex compiler options, Objective-C
1745 @cindex Objective-C options, command line
1746 @cindex options, Objective-C
1747 This section describes the command-line options that are only meaningful
1748 for Objective-C programs, but you can also use most of the GNU compiler
1749 options regardless of what language your program is in. For example,
1750 you might compile a file @code{some_class.m} like this:
1753 gcc -g -fgnu-runtime -O -c some_class.m
1757 In this example, @option{-fgnu-runtime} is an option meant only for
1758 Objective-C programs; you can use the other options with any language
1761 Here is a list of options that are @emph{only} for compiling Objective-C
1765 @item -fconstant-string-class=@var{class-name}
1766 @opindex fconstant-string-class
1767 Use @var{class-name} as the name of the class to instantiate for each
1768 literal string specified with the syntax @code{@@"@dots{}"}. The default
1769 class name is @code{NXConstantString}.
1772 @opindex fgnu-runtime
1773 Generate object code compatible with the standard GNU Objective-C
1774 runtime. This is the default for most types of systems.
1776 @item -fnext-runtime
1777 @opindex fnext-runtime
1778 Generate output compatible with the NeXT runtime. This is the default
1779 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1780 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1785 Dump interface declarations for all classes seen in the source file to a
1786 file named @file{@var{sourcename}.decl}.
1789 @opindex Wno-protocol
1790 If a class is declared to implement a protocol, a warning is issued for
1791 every method in the protocol that is not implemented by the class. The
1792 default behavior is to issue a warning for every method not explicitly
1793 implemented in the class, even if a method implementation is inherited
1794 from the superclass. If you use the @code{-Wno-protocol} option, then
1795 methods inherited from the superclass are considered to be implemented,
1796 and no warning is issued for them.
1800 Warn if multiple methods of different types for the same selector are
1801 found during compilation. The check is performed on the list of methods
1802 in the final stage of compilation. Additionally, a check is performed
1803 for each selector appearing in a @code{@@selector(@dots{})}
1804 expression, and a corresponding method for that selector has been found
1805 during compilation. Because these checks scan the method table only at
1806 the end of compilation, these warnings are not produced if the final
1807 stage of compilation is not reached, for example because an error is
1808 found during compilation, or because the @code{-fsyntax-only} option is
1811 @item -Wundeclared-selector
1812 @opindex Wundeclared-selector
1813 Warn if a @code{@@selector(@dots{})} expression referring to an
1814 undeclared selector is found. A selector is considered undeclared if no
1815 method with that name has been declared before the
1816 @code{@@selector(@dots{})} expression, either explicitly in an
1817 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1818 an @code{@@implementation} section. This option always performs its
1819 checks as soon as a @code{@@selector(@dots{})} expression is found,
1820 while @code{-Wselector} only performs its checks in the final stage of
1821 compilation. This also enforces the coding style convention
1822 that methods and selectors must be declared before being used.
1824 @c not documented because only avail via -Wp
1825 @c @item -print-objc-runtime-info
1829 @node Language Independent Options
1830 @section Options to Control Diagnostic Messages Formatting
1831 @cindex options to control diagnostics formatting
1832 @cindex diagnostic messages
1833 @cindex message formatting
1835 Traditionally, diagnostic messages have been formatted irrespective of
1836 the output device's aspect (e.g.@: its width, @dots{}). The options described
1837 below can be used to control the diagnostic messages formatting
1838 algorithm, e.g.@: how many characters per line, how often source location
1839 information should be reported. Right now, only the C++ front end can
1840 honor these options. However it is expected, in the near future, that
1841 the remaining front ends would be able to digest them correctly.
1844 @item -fmessage-length=@var{n}
1845 @opindex fmessage-length
1846 Try to format error messages so that they fit on lines of about @var{n}
1847 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1848 the front ends supported by GCC@. If @var{n} is zero, then no
1849 line-wrapping will be done; each error message will appear on a single
1852 @opindex fdiagnostics-show-location
1853 @item -fdiagnostics-show-location=once
1854 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1855 reporter to emit @emph{once} source location information; that is, in
1856 case the message is too long to fit on a single physical line and has to
1857 be wrapped, the source location won't be emitted (as prefix) again,
1858 over and over, in subsequent continuation lines. This is the default
1861 @item -fdiagnostics-show-location=every-line
1862 Only meaningful in line-wrapping mode. Instructs the diagnostic
1863 messages reporter to emit the same source location information (as
1864 prefix) for physical lines that result from the process of breaking
1865 a message which is too long to fit on a single line.
1869 @node Warning Options
1870 @section Options to Request or Suppress Warnings
1871 @cindex options to control warnings
1872 @cindex warning messages
1873 @cindex messages, warning
1874 @cindex suppressing warnings
1876 Warnings are diagnostic messages that report constructions which
1877 are not inherently erroneous but which are risky or suggest there
1878 may have been an error.
1880 You can request many specific warnings with options beginning @samp{-W},
1881 for example @option{-Wimplicit} to request warnings on implicit
1882 declarations. Each of these specific warning options also has a
1883 negative form beginning @samp{-Wno-} to turn off warnings;
1884 for example, @option{-Wno-implicit}. This manual lists only one of the
1885 two forms, whichever is not the default.
1887 The following options control the amount and kinds of warnings produced
1888 by GCC; for further, language-specific options also refer to
1889 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1892 @cindex syntax checking
1894 @opindex fsyntax-only
1895 Check the code for syntax errors, but don't do anything beyond that.
1899 Issue all the warnings demanded by strict ISO C and ISO C++;
1900 reject all programs that use forbidden extensions, and some other
1901 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1902 version of the ISO C standard specified by any @option{-std} option used.
1904 Valid ISO C and ISO C++ programs should compile properly with or without
1905 this option (though a rare few will require @option{-ansi} or a
1906 @option{-std} option specifying the required version of ISO C)@. However,
1907 without this option, certain GNU extensions and traditional C and C++
1908 features are supported as well. With this option, they are rejected.
1910 @option{-pedantic} does not cause warning messages for use of the
1911 alternate keywords whose names begin and end with @samp{__}. Pedantic
1912 warnings are also disabled in the expression that follows
1913 @code{__extension__}. However, only system header files should use
1914 these escape routes; application programs should avoid them.
1915 @xref{Alternate Keywords}.
1917 Some users try to use @option{-pedantic} to check programs for strict ISO
1918 C conformance. They soon find that it does not do quite what they want:
1919 it finds some non-ISO practices, but not all---only those for which
1920 ISO C @emph{requires} a diagnostic, and some others for which
1921 diagnostics have been added.
1923 A feature to report any failure to conform to ISO C might be useful in
1924 some instances, but would require considerable additional work and would
1925 be quite different from @option{-pedantic}. We don't have plans to
1926 support such a feature in the near future.
1928 Where the standard specified with @option{-std} represents a GNU
1929 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1930 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1931 extended dialect is based. Warnings from @option{-pedantic} are given
1932 where they are required by the base standard. (It would not make sense
1933 for such warnings to be given only for features not in the specified GNU
1934 C dialect, since by definition the GNU dialects of C include all
1935 features the compiler supports with the given option, and there would be
1936 nothing to warn about.)
1938 @item -pedantic-errors
1939 @opindex pedantic-errors
1940 Like @option{-pedantic}, except that errors are produced rather than
1945 Inhibit all warning messages.
1949 Inhibit warning messages about the use of @samp{#import}.
1951 @item -Wchar-subscripts
1952 @opindex Wchar-subscripts
1953 Warn if an array subscript has type @code{char}. This is a common cause
1954 of error, as programmers often forget that this type is signed on some
1959 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1960 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1964 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1965 the arguments supplied have types appropriate to the format string
1966 specified, and that the conversions specified in the format string make
1967 sense. This includes standard functions, and others specified by format
1968 attributes (@pxref{Function Attributes}), in the @code{printf},
1969 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1970 not in the C standard) families.
1972 The formats are checked against the format features supported by GNU
1973 libc version 2.2. These include all ISO C90 and C99 features, as well
1974 as features from the Single Unix Specification and some BSD and GNU
1975 extensions. Other library implementations may not support all these
1976 features; GCC does not support warning about features that go beyond a
1977 particular library's limitations. However, if @option{-pedantic} is used
1978 with @option{-Wformat}, warnings will be given about format features not
1979 in the selected standard version (but not for @code{strfmon} formats,
1980 since those are not in any version of the C standard). @xref{C Dialect
1981 Options,,Options Controlling C Dialect}.
1983 Since @option{-Wformat} also checks for null format arguments for
1984 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
1986 @option{-Wformat} is included in @option{-Wall}. For more control over some
1987 aspects of format checking, the options @option{-Wno-format-y2k},
1988 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
1989 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
1990 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
1992 @item -Wno-format-y2k
1993 @opindex Wno-format-y2k
1994 If @option{-Wformat} is specified, do not warn about @code{strftime}
1995 formats which may yield only a two-digit year.
1997 @item -Wno-format-extra-args
1998 @opindex Wno-format-extra-args
1999 If @option{-Wformat} is specified, do not warn about excess arguments to a
2000 @code{printf} or @code{scanf} format function. The C standard specifies
2001 that such arguments are ignored.
2003 Where the unused arguments lie between used arguments that are
2004 specified with @samp{$} operand number specifications, normally
2005 warnings are still given, since the implementation could not know what
2006 type to pass to @code{va_arg} to skip the unused arguments. However,
2007 in the case of @code{scanf} formats, this option will suppress the
2008 warning if the unused arguments are all pointers, since the Single
2009 Unix Specification says that such unused arguments are allowed.
2011 @item -Wno-format-zero-length
2012 @opindex Wno-format-zero-length
2013 If @option{-Wformat} is specified, do not warn about zero-length formats.
2014 The C standard specifies that zero-length formats are allowed.
2016 @item -Wformat-nonliteral
2017 @opindex Wformat-nonliteral
2018 If @option{-Wformat} is specified, also warn if the format string is not a
2019 string literal and so cannot be checked, unless the format function
2020 takes its format arguments as a @code{va_list}.
2022 @item -Wformat-security
2023 @opindex Wformat-security
2024 If @option{-Wformat} is specified, also warn about uses of format
2025 functions that represent possible security problems. At present, this
2026 warns about calls to @code{printf} and @code{scanf} functions where the
2027 format string is not a string literal and there are no format arguments,
2028 as in @code{printf (foo);}. This may be a security hole if the format
2029 string came from untrusted input and contains @samp{%n}. (This is
2030 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2031 in future warnings may be added to @option{-Wformat-security} that are not
2032 included in @option{-Wformat-nonliteral}.)
2036 Enable @option{-Wformat} plus format checks not included in
2037 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2038 -Wformat-nonliteral -Wformat-security}.
2042 Enable warning about passing a null pointer for arguments marked as
2043 requiring a non-null value by the @code{nonnull} function attribute.
2045 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2046 can be disabled with the @option{-Wno-nonnull} option.
2048 @item -Winit-self @r{(C, C++, and Objective-C only)}
2050 Enable warning about uninitialized variables which are initalized with themselves.
2051 Note this option can only be used with the @option{-Wuninitialized} option and
2052 that only works with @option{-O}.
2054 For an example, the following code will not warn about i being uninitialized
2055 without this option:
2066 @item -Wimplicit-int
2067 @opindex Wimplicit-int
2068 Warn when a declaration does not specify a type.
2070 @item -Wimplicit-function-declaration
2071 @itemx -Werror-implicit-function-declaration
2072 @opindex Wimplicit-function-declaration
2073 @opindex Werror-implicit-function-declaration
2074 Give a warning (or error) whenever a function is used before being
2079 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2083 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2084 function with external linkage, returning int, taking either zero
2085 arguments, two, or three arguments of appropriate types.
2087 @item -Wmissing-braces
2088 @opindex Wmissing-braces
2089 Warn if an aggregate or union initializer is not fully bracketed. In
2090 the following example, the initializer for @samp{a} is not fully
2091 bracketed, but that for @samp{b} is fully bracketed.
2094 int a[2][2] = @{ 0, 1, 2, 3 @};
2095 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2099 @opindex Wparentheses
2100 Warn if parentheses are omitted in certain contexts, such
2101 as when there is an assignment in a context where a truth value
2102 is expected, or when operators are nested whose precedence people
2103 often get confused about.
2105 Also warn about constructions where there may be confusion to which
2106 @code{if} statement an @code{else} branch belongs. Here is an example of
2121 In C, every @code{else} branch belongs to the innermost possible @code{if}
2122 statement, which in this example is @code{if (b)}. This is often not
2123 what the programmer expected, as illustrated in the above example by
2124 indentation the programmer chose. When there is the potential for this
2125 confusion, GCC will issue a warning when this flag is specified.
2126 To eliminate the warning, add explicit braces around the innermost
2127 @code{if} statement so there is no way the @code{else} could belong to
2128 the enclosing @code{if}. The resulting code would look like this:
2144 @item -Wsequence-point
2145 @opindex Wsequence-point
2146 Warn about code that may have undefined semantics because of violations
2147 of sequence point rules in the C standard.
2149 The C standard defines the order in which expressions in a C program are
2150 evaluated in terms of @dfn{sequence points}, which represent a partial
2151 ordering between the execution of parts of the program: those executed
2152 before the sequence point, and those executed after it. These occur
2153 after the evaluation of a full expression (one which is not part of a
2154 larger expression), after the evaluation of the first operand of a
2155 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2156 function is called (but after the evaluation of its arguments and the
2157 expression denoting the called function), and in certain other places.
2158 Other than as expressed by the sequence point rules, the order of
2159 evaluation of subexpressions of an expression is not specified. All
2160 these rules describe only a partial order rather than a total order,
2161 since, for example, if two functions are called within one expression
2162 with no sequence point between them, the order in which the functions
2163 are called is not specified. However, the standards committee have
2164 ruled that function calls do not overlap.
2166 It is not specified when between sequence points modifications to the
2167 values of objects take effect. Programs whose behavior depends on this
2168 have undefined behavior; the C standard specifies that ``Between the
2169 previous and next sequence point an object shall have its stored value
2170 modified at most once by the evaluation of an expression. Furthermore,
2171 the prior value shall be read only to determine the value to be
2172 stored.''. If a program breaks these rules, the results on any
2173 particular implementation are entirely unpredictable.
2175 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2176 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2177 diagnosed by this option, and it may give an occasional false positive
2178 result, but in general it has been found fairly effective at detecting
2179 this sort of problem in programs.
2181 The present implementation of this option only works for C programs. A
2182 future implementation may also work for C++ programs.
2184 The C standard is worded confusingly, therefore there is some debate
2185 over the precise meaning of the sequence point rules in subtle cases.
2186 Links to discussions of the problem, including proposed formal
2187 definitions, may be found on our readings page, at
2188 @w{@uref{http://gcc.gnu.org/readings.html}}.
2191 @opindex Wreturn-type
2192 Warn whenever a function is defined with a return-type that defaults to
2193 @code{int}. Also warn about any @code{return} statement with no
2194 return-value in a function whose return-type is not @code{void}.
2196 For C++, a function without return type always produces a diagnostic
2197 message, even when @option{-Wno-return-type} is specified. The only
2198 exceptions are @samp{main} and functions defined in system headers.
2202 Warn whenever a @code{switch} statement has an index of enumeral type
2203 and lacks a @code{case} for one or more of the named codes of that
2204 enumeration. (The presence of a @code{default} label prevents this
2205 warning.) @code{case} labels outside the enumeration range also
2206 provoke warnings when this option is used.
2208 @item -Wswitch-default
2209 @opindex Wswitch-switch
2210 Warn whenever a @code{switch} statement does not have a @code{default}
2214 @opindex Wswitch-enum
2215 Warn whenever a @code{switch} statement has an index of enumeral type
2216 and lacks a @code{case} for one or more of the named codes of that
2217 enumeration. @code{case} labels outside the enumeration range also
2218 provoke warnings when this option is used.
2222 Warn if any trigraphs are encountered that might change the meaning of
2223 the program (trigraphs within comments are not warned about).
2225 @item -Wunused-function
2226 @opindex Wunused-function
2227 Warn whenever a static function is declared but not defined or a
2228 non\-inline static function is unused.
2230 @item -Wunused-label
2231 @opindex Wunused-label
2232 Warn whenever a label is declared but not used.
2234 To suppress this warning use the @samp{unused} attribute
2235 (@pxref{Variable Attributes}).
2237 @item -Wunused-parameter
2238 @opindex Wunused-parameter
2239 Warn whenever a function parameter is unused aside from its declaration.
2241 To suppress this warning use the @samp{unused} attribute
2242 (@pxref{Variable Attributes}).
2244 @item -Wunused-variable
2245 @opindex Wunused-variable
2246 Warn whenever a local variable or non-constant static variable is unused
2247 aside from its declaration
2249 To suppress this warning use the @samp{unused} attribute
2250 (@pxref{Variable Attributes}).
2252 @item -Wunused-value
2253 @opindex Wunused-value
2254 Warn whenever a statement computes a result that is explicitly not used.
2256 To suppress this warning cast the expression to @samp{void}.
2260 All the above @option{-Wunused} options combined.
2262 In order to get a warning about an unused function parameter, you must
2263 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2264 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2266 @item -Wuninitialized
2267 @opindex Wuninitialized
2268 Warn if an automatic variable is used without first being initialized or
2269 if a variable may be clobbered by a @code{setjmp} call.
2271 These warnings are possible only in optimizing compilation,
2272 because they require data flow information that is computed only
2273 when optimizing. If you don't specify @option{-O}, you simply won't
2276 If you want to warn about code which uses the uninitialized value of the
2277 variable in its own initializer, use the @option{-Winit-self} option.
2279 These warnings occur only for variables that are candidates for
2280 register allocation. Therefore, they do not occur for a variable that
2281 is declared @code{volatile}, or whose address is taken, or whose size
2282 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2283 structures, unions or arrays, even when they are in registers.
2285 Note that there may be no warning about a variable that is used only
2286 to compute a value that itself is never used, because such
2287 computations may be deleted by data flow analysis before the warnings
2290 These warnings are made optional because GCC is not smart
2291 enough to see all the reasons why the code might be correct
2292 despite appearing to have an error. Here is one example of how
2313 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2314 always initialized, but GCC doesn't know this. Here is
2315 another common case:
2320 if (change_y) save_y = y, y = new_y;
2322 if (change_y) y = save_y;
2327 This has no bug because @code{save_y} is used only if it is set.
2329 @cindex @code{longjmp} warnings
2330 This option also warns when a non-volatile automatic variable might be
2331 changed by a call to @code{longjmp}. These warnings as well are possible
2332 only in optimizing compilation.
2334 The compiler sees only the calls to @code{setjmp}. It cannot know
2335 where @code{longjmp} will be called; in fact, a signal handler could
2336 call it at any point in the code. As a result, you may get a warning
2337 even when there is in fact no problem because @code{longjmp} cannot
2338 in fact be called at the place which would cause a problem.
2340 Some spurious warnings can be avoided if you declare all the functions
2341 you use that never return as @code{noreturn}. @xref{Function
2344 @item -Wunknown-pragmas
2345 @opindex Wunknown-pragmas
2346 @cindex warning for unknown pragmas
2347 @cindex unknown pragmas, warning
2348 @cindex pragmas, warning of unknown
2349 Warn when a #pragma directive is encountered which is not understood by
2350 GCC@. If this command line option is used, warnings will even be issued
2351 for unknown pragmas in system header files. This is not the case if
2352 the warnings were only enabled by the @option{-Wall} command line option.
2354 @item -Wstrict-aliasing
2355 @opindex Wstrict-aliasing
2356 This option is only active when @option{-fstrict-aliasing} is active.
2357 It warns about code which might break the strict aliasing rules that the
2358 compiler is using for optimization. The warning does not catch all
2359 cases, but does attempt to catch the more common pitfalls. It is
2360 included in @option{-Wall}.
2364 All of the above @samp{-W} options combined. This enables all the
2365 warnings about constructions that some users consider questionable, and
2366 that are easy to avoid (or modify to prevent the warning), even in
2367 conjunction with macros. This also enables some language-specific
2368 warnings described in @ref{C++ Dialect Options} and
2369 @ref{Objective-C Dialect Options}.
2372 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2373 Some of them warn about constructions that users generally do not
2374 consider questionable, but which occasionally you might wish to check
2375 for; others warn about constructions that are necessary or hard to avoid
2376 in some cases, and there is no simple way to modify the code to suppress
2383 (This option used to be called @option{-W}. The older name is still
2384 supported, but the newer name is more descriptive.) Print extra warning
2385 messages for these events:
2389 A function can return either with or without a value. (Falling
2390 off the end of the function body is considered returning without
2391 a value.) For example, this function would evoke such a
2405 An expression-statement or the left-hand side of a comma expression
2406 contains no side effects.
2407 To suppress the warning, cast the unused expression to void.
2408 For example, an expression such as @samp{x[i,j]} will cause a warning,
2409 but @samp{x[(void)i,j]} will not.
2412 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2415 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2416 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2417 that of ordinary mathematical notation.
2420 Storage-class specifiers like @code{static} are not the first things in
2421 a declaration. According to the C Standard, this usage is obsolescent.
2424 The return type of a function has a type qualifier such as @code{const}.
2425 Such a type qualifier has no effect, since the value returned by a
2426 function is not an lvalue. (But don't warn about the GNU extension of
2427 @code{volatile void} return types. That extension will be warned about
2428 if @option{-pedantic} is specified.)
2431 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2435 A comparison between signed and unsigned values could produce an
2436 incorrect result when the signed value is converted to unsigned.
2437 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2440 An aggregate has an initializer which does not initialize all members.
2441 For example, the following code would cause such a warning, because
2442 @code{x.h} would be implicitly initialized to zero:
2445 struct s @{ int f, g, h; @};
2446 struct s x = @{ 3, 4 @};
2450 A function parameter is declared without a type specifier in K&R-style
2458 An empty body occurs in an @samp{if} or @samp{else} statement.
2461 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2462 @samp{>}, or @samp{>=}.
2465 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2468 Any of several floating-point events that often indicate errors, such as
2469 overflow, underflow, loss of precision, etc.
2471 @item @r{(C++ only)}
2472 An enumerator and a non-enumerator both appear in a conditional expression.
2474 @item @r{(C++ only)}
2475 A non-static reference or non-static @samp{const} member appears in a
2476 class without constructors.
2478 @item @r{(C++ only)}
2479 Ambiguous virtual bases.
2481 @item @r{(C++ only)}
2482 Subscripting an array which has been declared @samp{register}.
2484 @item @r{(C++ only)}
2485 Taking the address of a variable which has been declared @samp{register}.
2487 @item @r{(C++ only)}
2488 A base class is not initialized in a derived class' copy constructor.
2491 @item -Wno-div-by-zero
2492 @opindex Wno-div-by-zero
2493 @opindex Wdiv-by-zero
2494 Do not warn about compile-time integer division by zero. Floating point
2495 division by zero is not warned about, as it can be a legitimate way of
2496 obtaining infinities and NaNs.
2498 @item -Wsystem-headers
2499 @opindex Wsystem-headers
2500 @cindex warnings from system headers
2501 @cindex system headers, warnings from
2502 Print warning messages for constructs found in system header files.
2503 Warnings from system headers are normally suppressed, on the assumption
2504 that they usually do not indicate real problems and would only make the
2505 compiler output harder to read. Using this command line option tells
2506 GCC to emit warnings from system headers as if they occurred in user
2507 code. However, note that using @option{-Wall} in conjunction with this
2508 option will @emph{not} warn about unknown pragmas in system
2509 headers---for that, @option{-Wunknown-pragmas} must also be used.
2512 @opindex Wfloat-equal
2513 Warn if floating point values are used in equality comparisons.
2515 The idea behind this is that sometimes it is convenient (for the
2516 programmer) to consider floating-point values as approximations to
2517 infinitely precise real numbers. If you are doing this, then you need
2518 to compute (by analyzing the code, or in some other way) the maximum or
2519 likely maximum error that the computation introduces, and allow for it
2520 when performing comparisons (and when producing output, but that's a
2521 different problem). In particular, instead of testing for equality, you
2522 would check to see whether the two values have ranges that overlap; and
2523 this is done with the relational operators, so equality comparisons are
2526 @item -Wtraditional @r{(C only)}
2527 @opindex Wtraditional
2528 Warn about certain constructs that behave differently in traditional and
2529 ISO C@. Also warn about ISO C constructs that have no traditional C
2530 equivalent, and/or problematic constructs which should be avoided.
2534 Macro parameters that appear within string literals in the macro body.
2535 In traditional C macro replacement takes place within string literals,
2536 but does not in ISO C@.
2539 In traditional C, some preprocessor directives did not exist.
2540 Traditional preprocessors would only consider a line to be a directive
2541 if the @samp{#} appeared in column 1 on the line. Therefore
2542 @option{-Wtraditional} warns about directives that traditional C
2543 understands but would ignore because the @samp{#} does not appear as the
2544 first character on the line. It also suggests you hide directives like
2545 @samp{#pragma} not understood by traditional C by indenting them. Some
2546 traditional implementations would not recognize @samp{#elif}, so it
2547 suggests avoiding it altogether.
2550 A function-like macro that appears without arguments.
2553 The unary plus operator.
2556 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2557 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2558 constants.) Note, these suffixes appear in macros defined in the system
2559 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2560 Use of these macros in user code might normally lead to spurious
2561 warnings, however gcc's integrated preprocessor has enough context to
2562 avoid warning in these cases.
2565 A function declared external in one block and then used after the end of
2569 A @code{switch} statement has an operand of type @code{long}.
2572 A non-@code{static} function declaration follows a @code{static} one.
2573 This construct is not accepted by some traditional C compilers.
2576 The ISO type of an integer constant has a different width or
2577 signedness from its traditional type. This warning is only issued if
2578 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2579 typically represent bit patterns, are not warned about.
2582 Usage of ISO string concatenation is detected.
2585 Initialization of automatic aggregates.
2588 Identifier conflicts with labels. Traditional C lacks a separate
2589 namespace for labels.
2592 Initialization of unions. If the initializer is zero, the warning is
2593 omitted. This is done under the assumption that the zero initializer in
2594 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2595 initializer warnings and relies on default initialization to zero in the
2599 Conversions by prototypes between fixed/floating point values and vice
2600 versa. The absence of these prototypes when compiling with traditional
2601 C would cause serious problems. This is a subset of the possible
2602 conversion warnings, for the full set use @option{-Wconversion}.
2605 Use of ISO C style function definitions. This warning intentionally is
2606 @emph{not} issued for prototype declarations or variadic functions
2607 because these ISO C features will appear in your code when using
2608 libiberty's traditional C compatibility macros, @code{PARAMS} and
2609 @code{VPARAMS}. This warning is also bypassed for nested functions
2610 because that feature is already a gcc extension and thus not relevant to
2611 traditional C compatibility.
2614 @item -Wdeclaration-after-statement @r{(C only)}
2615 @opindex Wdeclaration-after-statement
2616 Warn when a declaration is found after a statement in a block. This
2617 construct, known from C++, was introduced with ISO C99 and is by default
2618 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2619 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2623 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2625 @item -Wendif-labels
2626 @opindex Wendif-labels
2627 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2631 Warn whenever a local variable shadows another local variable, parameter or
2632 global variable or whenever a built-in function is shadowed.
2634 @item -Wlarger-than-@var{len}
2635 @opindex Wlarger-than
2636 Warn whenever an object of larger than @var{len} bytes is defined.
2638 @item -Wpointer-arith
2639 @opindex Wpointer-arith
2640 Warn about anything that depends on the ``size of'' a function type or
2641 of @code{void}. GNU C assigns these types a size of 1, for
2642 convenience in calculations with @code{void *} pointers and pointers
2645 @item -Wbad-function-cast @r{(C only)}
2646 @opindex Wbad-function-cast
2647 Warn whenever a function call is cast to a non-matching type.
2648 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2652 Warn whenever a pointer is cast so as to remove a type qualifier from
2653 the target type. For example, warn if a @code{const char *} is cast
2654 to an ordinary @code{char *}.
2657 @opindex Wcast-align
2658 Warn whenever a pointer is cast such that the required alignment of the
2659 target is increased. For example, warn if a @code{char *} is cast to
2660 an @code{int *} on machines where integers can only be accessed at
2661 two- or four-byte boundaries.
2663 @item -Wwrite-strings
2664 @opindex Wwrite-strings
2665 When compiling C, give string constants the type @code{const
2666 char[@var{length}]} so that
2667 copying the address of one into a non-@code{const} @code{char *}
2668 pointer will get a warning; when compiling C++, warn about the
2669 deprecated conversion from string constants to @code{char *}.
2670 These warnings will help you find at
2671 compile time code that can try to write into a string constant, but
2672 only if you have been very careful about using @code{const} in
2673 declarations and prototypes. Otherwise, it will just be a nuisance;
2674 this is why we did not make @option{-Wall} request these warnings.
2677 @opindex Wconversion
2678 Warn if a prototype causes a type conversion that is different from what
2679 would happen to the same argument in the absence of a prototype. This
2680 includes conversions of fixed point to floating and vice versa, and
2681 conversions changing the width or signedness of a fixed point argument
2682 except when the same as the default promotion.
2684 Also, warn if a negative integer constant expression is implicitly
2685 converted to an unsigned type. For example, warn about the assignment
2686 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2687 casts like @code{(unsigned) -1}.
2689 @item -Wsign-compare
2690 @opindex Wsign-compare
2691 @cindex warning for comparison of signed and unsigned values
2692 @cindex comparison of signed and unsigned values, warning
2693 @cindex signed and unsigned values, comparison warning
2694 Warn when a comparison between signed and unsigned values could produce
2695 an incorrect result when the signed value is converted to unsigned.
2696 This warning is also enabled by @option{-Wextra}; to get the other warnings
2697 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2699 @item -Waggregate-return
2700 @opindex Waggregate-return
2701 Warn if any functions that return structures or unions are defined or
2702 called. (In languages where you can return an array, this also elicits
2705 @item -Wstrict-prototypes @r{(C only)}
2706 @opindex Wstrict-prototypes
2707 Warn if a function is declared or defined without specifying the
2708 argument types. (An old-style function definition is permitted without
2709 a warning if preceded by a declaration which specifies the argument
2712 @item -Wold-style-definition @r{(C only)}
2713 @opindex Wold-style-definition
2714 Warn if an old-style function definition is used. A warning is given
2715 even if there is a previous prototype.
2717 @item -Wmissing-prototypes @r{(C only)}
2718 @opindex Wmissing-prototypes
2719 Warn if a global function is defined without a previous prototype
2720 declaration. This warning is issued even if the definition itself
2721 provides a prototype. The aim is to detect global functions that fail
2722 to be declared in header files.
2724 @item -Wmissing-declarations @r{(C only)}
2725 @opindex Wmissing-declarations
2726 Warn if a global function is defined without a previous declaration.
2727 Do so even if the definition itself provides a prototype.
2728 Use this option to detect global functions that are not declared in
2731 @item -Wmissing-noreturn
2732 @opindex Wmissing-noreturn
2733 Warn about functions which might be candidates for attribute @code{noreturn}.
2734 Note these are only possible candidates, not absolute ones. Care should
2735 be taken to manually verify functions actually do not ever return before
2736 adding the @code{noreturn} attribute, otherwise subtle code generation
2737 bugs could be introduced. You will not get a warning for @code{main} in
2738 hosted C environments.
2740 @item -Wmissing-format-attribute
2741 @opindex Wmissing-format-attribute
2743 If @option{-Wformat} is enabled, also warn about functions which might be
2744 candidates for @code{format} attributes. Note these are only possible
2745 candidates, not absolute ones. GCC will guess that @code{format}
2746 attributes might be appropriate for any function that calls a function
2747 like @code{vprintf} or @code{vscanf}, but this might not always be the
2748 case, and some functions for which @code{format} attributes are
2749 appropriate may not be detected. This option has no effect unless
2750 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2752 @item -Wno-multichar
2753 @opindex Wno-multichar
2755 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2756 Usually they indicate a typo in the user's code, as they have
2757 implementation-defined values, and should not be used in portable code.
2759 @item -Wno-deprecated-declarations
2760 @opindex Wno-deprecated-declarations
2761 Do not warn about uses of functions, variables, and types marked as
2762 deprecated by using the @code{deprecated} attribute.
2763 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2764 @pxref{Type Attributes}.)
2768 Warn if a structure is given the packed attribute, but the packed
2769 attribute has no effect on the layout or size of the structure.
2770 Such structures may be mis-aligned for little benefit. For
2771 instance, in this code, the variable @code{f.x} in @code{struct bar}
2772 will be misaligned even though @code{struct bar} does not itself
2773 have the packed attribute:
2780 @} __attribute__((packed));
2790 Warn if padding is included in a structure, either to align an element
2791 of the structure or to align the whole structure. Sometimes when this
2792 happens it is possible to rearrange the fields of the structure to
2793 reduce the padding and so make the structure smaller.
2795 @item -Wredundant-decls
2796 @opindex Wredundant-decls
2797 Warn if anything is declared more than once in the same scope, even in
2798 cases where multiple declaration is valid and changes nothing.
2800 @item -Wnested-externs @r{(C only)}
2801 @opindex Wnested-externs
2802 Warn if an @code{extern} declaration is encountered within a function.
2804 @item -Wunreachable-code
2805 @opindex Wunreachable-code
2806 Warn if the compiler detects that code will never be executed.
2808 This option is intended to warn when the compiler detects that at
2809 least a whole line of source code will never be executed, because
2810 some condition is never satisfied or because it is after a
2811 procedure that never returns.
2813 It is possible for this option to produce a warning even though there
2814 are circumstances under which part of the affected line can be executed,
2815 so care should be taken when removing apparently-unreachable code.
2817 For instance, when a function is inlined, a warning may mean that the
2818 line is unreachable in only one inlined copy of the function.
2820 This option is not made part of @option{-Wall} because in a debugging
2821 version of a program there is often substantial code which checks
2822 correct functioning of the program and is, hopefully, unreachable
2823 because the program does work. Another common use of unreachable
2824 code is to provide behavior which is selectable at compile-time.
2828 Warn if a function can not be inlined and it was declared as inline.
2829 Even with this option, the compiler will not warn about failures to
2830 inline functions declared in system headers.
2832 The compiler uses a variety of heuristics to determine whether or not
2833 to inline a function. For example, the compiler takes into account
2834 the size of the function being inlined and the the amount of inlining
2835 that has already been done in the current function. Therefore,
2836 seemingly insignificant changes in the source program can cause the
2837 warnings produced by @option{-Winline} to appear or disappear.
2839 @item -Wno-invalid-offsetof @r{(C++ only)}
2840 @opindex Wno-invalid-offsetof
2841 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2842 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2843 to a non-POD type is undefined. In existing C++ implementations,
2844 however, @samp{offsetof} typically gives meaningful results even when
2845 applied to certain kinds of non-POD types. (Such as a simple
2846 @samp{struct} that fails to be a POD type only by virtue of having a
2847 constructor.) This flag is for users who are aware that they are
2848 writing nonportable code and who have deliberately chosen to ignore the
2851 The restrictions on @samp{offsetof} may be relaxed in a future version
2852 of the C++ standard.
2855 @opindex Winvalid-pch
2856 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2857 the search path but can't be used.
2861 @opindex Wno-long-long
2862 Warn if @samp{long long} type is used. This is default. To inhibit
2863 the warning messages, use @option{-Wno-long-long}. Flags
2864 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2865 only when @option{-pedantic} flag is used.
2867 @item -Wdisabled-optimization
2868 @opindex Wdisabled-optimization
2869 Warn if a requested optimization pass is disabled. This warning does
2870 not generally indicate that there is anything wrong with your code; it
2871 merely indicates that GCC's optimizers were unable to handle the code
2872 effectively. Often, the problem is that your code is too big or too
2873 complex; GCC will refuse to optimize programs when the optimization
2874 itself is likely to take inordinate amounts of time.
2878 Make all warnings into errors.
2881 @node Debugging Options
2882 @section Options for Debugging Your Program or GCC
2883 @cindex options, debugging
2884 @cindex debugging information options
2886 GCC has various special options that are used for debugging
2887 either your program or GCC:
2892 Produce debugging information in the operating system's native format
2893 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2896 On most systems that use stabs format, @option{-g} enables use of extra
2897 debugging information that only GDB can use; this extra information
2898 makes debugging work better in GDB but will probably make other debuggers
2900 refuse to read the program. If you want to control for certain whether
2901 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2902 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
2903 or @option{-gvms} (see below).
2905 Unlike most other C compilers, GCC allows you to use @option{-g} with
2906 @option{-O}. The shortcuts taken by optimized code may occasionally
2907 produce surprising results: some variables you declared may not exist
2908 at all; flow of control may briefly move where you did not expect it;
2909 some statements may not be executed because they compute constant
2910 results or their values were already at hand; some statements may
2911 execute in different places because they were moved out of loops.
2913 Nevertheless it proves possible to debug optimized output. This makes
2914 it reasonable to use the optimizer for programs that might have bugs.
2916 The following options are useful when GCC is generated with the
2917 capability for more than one debugging format.
2921 Produce debugging information for use by GDB@. This means to use the
2922 most expressive format available (DWARF 2, stabs, or the native format
2923 if neither of those are supported), including GDB extensions if at all
2928 Produce debugging information in stabs format (if that is supported),
2929 without GDB extensions. This is the format used by DBX on most BSD
2930 systems. On MIPS, Alpha and System V Release 4 systems this option
2931 produces stabs debugging output which is not understood by DBX or SDB@.
2932 On System V Release 4 systems this option requires the GNU assembler.
2934 @item -feliminate-unused-debug-symbols
2935 @opindex feliminate-unused-debug-symbols
2936 Produce debugging information in stabs format (if that is supported),
2937 for only symbols that are actually used.
2941 Produce debugging information in stabs format (if that is supported),
2942 using GNU extensions understood only by the GNU debugger (GDB)@. The
2943 use of these extensions is likely to make other debuggers crash or
2944 refuse to read the program.
2948 Produce debugging information in COFF format (if that is supported).
2949 This is the format used by SDB on most System V systems prior to
2954 Produce debugging information in XCOFF format (if that is supported).
2955 This is the format used by the DBX debugger on IBM RS/6000 systems.
2959 Produce debugging information in XCOFF format (if that is supported),
2960 using GNU extensions understood only by the GNU debugger (GDB)@. The
2961 use of these extensions is likely to make other debuggers crash or
2962 refuse to read the program, and may cause assemblers other than the GNU
2963 assembler (GAS) to fail with an error.
2967 Produce debugging information in DWARF version 1 format (if that is
2968 supported). This is the format used by SDB on most System V Release 4
2971 This option is deprecated.
2975 Produce debugging information in DWARF version 1 format (if that is
2976 supported), using GNU extensions understood only by the GNU debugger
2977 (GDB)@. The use of these extensions is likely to make other debuggers
2978 crash or refuse to read the program.
2980 This option is deprecated.
2984 Produce debugging information in DWARF version 2 format (if that is
2985 supported). This is the format used by DBX on IRIX 6.
2989 Produce debugging information in VMS debug format (if that is
2990 supported). This is the format used by DEBUG on VMS systems.
2993 @itemx -ggdb@var{level}
2994 @itemx -gstabs@var{level}
2995 @itemx -gcoff@var{level}
2996 @itemx -gxcoff@var{level}
2997 @itemx -gvms@var{level}
2998 Request debugging information and also use @var{level} to specify how
2999 much information. The default level is 2.
3001 Level 1 produces minimal information, enough for making backtraces in
3002 parts of the program that you don't plan to debug. This includes
3003 descriptions of functions and external variables, but no information
3004 about local variables and no line numbers.
3006 Level 3 includes extra information, such as all the macro definitions
3007 present in the program. Some debuggers support macro expansion when
3008 you use @option{-g3}.
3010 Note that in order to avoid confusion between DWARF1 debug level 2,
3011 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
3012 a concatenated debug level. Instead use an additional @option{-g@var{level}}
3013 option to change the debug level for DWARF1 or DWARF2.
3015 @item -feliminate-dwarf2-dups
3016 @opindex feliminate-dwarf2-dups
3017 Compress DWARF2 debugging information by eliminating duplicated
3018 information about each symbol. This option only makes sense when
3019 generating DWARF2 debugging information with @option{-gdwarf-2}.
3021 @cindex @command{prof}
3024 Generate extra code to write profile information suitable for the
3025 analysis program @command{prof}. You must use this option when compiling
3026 the source files you want data about, and you must also use it when
3029 @cindex @command{gprof}
3032 Generate extra code to write profile information suitable for the
3033 analysis program @command{gprof}. You must use this option when compiling
3034 the source files you want data about, and you must also use it when
3039 Makes the compiler print out each function name as it is compiled, and
3040 print some statistics about each pass when it finishes.
3043 @opindex ftime-report
3044 Makes the compiler print some statistics about the time consumed by each
3045 pass when it finishes.
3048 @opindex fmem-report
3049 Makes the compiler print some statistics about permanent memory
3050 allocation when it finishes.
3052 @item -fprofile-arcs
3053 @opindex fprofile-arcs
3054 Add code so that program flow @dfn{arcs} are instrumented. During
3055 execution the program records how many times each branch and call is
3056 executed and how many times it is taken or returns. When the compiled
3057 program exits it saves this data to a file called
3058 @file{@var{auxname}.gcda} for each source file. The data may be used for
3059 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3060 test coverage analysis (@option{-ftest-coverage}). Each object file's
3061 @var{auxname} is generated from the name of the output file, if
3062 explicitly specified and it is not the final executable, otherwise it is
3063 the basename of the source file. In both cases any suffix is removed
3064 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3065 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3070 Compile the source files with @option{-fprofile-arcs} plus optimization
3071 and code generation options. For test coverage analysis, use the
3072 additional @option{-ftest-coverage} option. You do not need to profile
3073 every source file in a program.
3076 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3077 (the latter implies the former).
3080 Run the program on a representative workload to generate the arc profile
3081 information. This may be repeated any number of times. You can run
3082 concurrent instances of your program, and provided that the file system
3083 supports locking, the data files will be correctly updated. Also
3084 @code{fork} calls are detected and correctly handled (double counting
3088 For profile-directed optimizations, compile the source files again with
3089 the same optimization and code generation options plus
3090 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3091 Control Optimization}).
3094 For test coverage analysis, use @command{gcov} to produce human readable
3095 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3096 @command{gcov} documentation for further information.
3100 With @option{-fprofile-arcs}, for each function of your program GCC
3101 creates a program flow graph, then finds a spanning tree for the graph.
3102 Only arcs that are not on the spanning tree have to be instrumented: the
3103 compiler adds code to count the number of times that these arcs are
3104 executed. When an arc is the only exit or only entrance to a block, the
3105 instrumentation code can be added to the block; otherwise, a new basic
3106 block must be created to hold the instrumentation code.
3109 @item -ftest-coverage
3110 @opindex ftest-coverage
3111 Produce a notes file that the @command{gcov} code-coverage utility
3112 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3113 show program coverage. Each source file's note file is called
3114 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3115 above for a description of @var{auxname} and instructions on how to
3116 generate test coverage data. Coverage data will match the source files
3117 more closely, if you do not optimize.
3119 @item -d@var{letters}
3121 Says to make debugging dumps during compilation at times specified by
3122 @var{letters}. This is used for debugging the compiler. The file names
3123 for most of the dumps are made by appending a pass number and a word to
3124 the @var{dumpname}. @var{dumpname} is generated from the name of the
3125 output file, if explicitly specified and it is not an executable,
3126 otherwise it is the basename of the source file. In both cases any
3127 suffix is removed (e.g. @file{foo.00.rtl} or @file{foo.01.sibling}).
3128 Here are the possible letters for use in @var{letters}, and their
3134 Annotate the assembler output with miscellaneous debugging information.
3137 Dump after computing branch probabilities, to @file{@var{file}.16.bp}.
3140 Dump after block reordering, to @file{@var{file}.32.bbro}.
3143 Dump after instruction combination, to the file @file{@var{file}.22.combine}.
3146 Dump after the first if conversion, to the file @file{@var{file}.17.ce1}.
3147 Also dump after the second if conversion, to the file @file{@var{file}.23.ce2}.
3150 Dump after branch target load optimization, to to @file{@var{file}.34.btl}.
3151 Also dump after delayed branch scheduling, to @file{@var{file}.37.dbr}.
3154 Dump all macro definitions, at the end of preprocessing, in addition to
3158 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
3159 @file{@var{file}.010.ussa}.
3162 Dump after the second if conversion, to @file{@var{file}.33.ce3}.
3165 Dump after control and data flow analysis, to @file{@var{file}.15.cfg}.
3166 Also dump after life analysis, to @file{@var{file}.21.life}.
3169 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.11.addressof}.
3172 Dump after global register allocation, to @file{@var{file}.27.greg}.
3175 Dump after GCSE, to @file{@var{file}.12.gcse}.
3176 Also dump after jump bypassing and control flow optimizations, to
3177 @file{@var{file}.14.bypass}.
3180 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3183 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3186 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3189 Dump after conversion from registers to stack, to @file{@var{file}.36.stack}.
3192 Dump after local register allocation, to @file{@var{file}.26.lreg}.
3195 Dump after loop optimization passes, to @file{@var{file}.13.loop} and
3196 @file{@var{file}.19.loop2}.
3199 Dump after performing the machine dependent reorganization pass, to
3200 @file{@var{file}.37.mach}.
3203 Dump after register renumbering, to @file{@var{file}.31.rnreg}.
3206 Dump after the register move pass, to @file{@var{file}.24.regmove}.
3209 Dump after post-reload optimizations, to @file{@var{file}.28.postreload}.
3212 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3215 Dump after the second scheduling pass, to @file{@var{file}.35.sched2}.
3218 Dump after CSE (including the jump optimization that sometimes follows
3219 CSE), to @file{@var{file}.019.cse}.
3222 Dump after the first scheduling pass, to @file{@var{file}.25.sched}.
3225 Dump after the second CSE pass (including the jump optimization that
3226 sometimes follows CSE), to @file{@var{file}.20.cse2}.
3229 Dump after running tracer, to @file{@var{file}.18.tracer}.
3232 Dump after null pointer elimination pass to @file{@var{file}.018.null}.
3235 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3238 Dump after the second flow pass, to @file{@var{file}.29.flow2}.
3241 Dump after SSA conditional constant propagation, to
3242 @file{@var{file}.06.ssaccp}.
3245 Dump after SSA dead code elimination, to @file{@var{file}.07.ssadce}.
3248 Dump after the peephole pass, to @file{@var{file}.30.peephole2}.
3251 Produce all the dumps listed above.
3254 Produce a core dump whenever an error occurs.
3257 Print statistics on memory usage, at the end of the run, to
3261 Annotate the assembler output with a comment indicating which
3262 pattern and alternative was used. The length of each instruction is
3266 Dump the RTL in the assembler output as a comment before each instruction.
3267 Also turns on @option{-dp} annotation.
3270 For each of the other indicated dump files (except for
3271 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3272 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3275 Just generate RTL for a function instead of compiling it. Usually used
3279 Dump debugging information during parsing, to standard error.
3282 @item -fdump-unnumbered
3283 @opindex fdump-unnumbered
3284 When doing debugging dumps (see @option{-d} option above), suppress instruction
3285 numbers and line number note output. This makes it more feasible to
3286 use diff on debugging dumps for compiler invocations with different
3287 options, in particular with and without @option{-g}.
3289 @item -fdump-translation-unit @r{(C and C++ only)}
3290 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3291 @opindex fdump-translation-unit
3292 Dump a representation of the tree structure for the entire translation
3293 unit to a file. The file name is made by appending @file{.tu} to the
3294 source file name. If the @samp{-@var{options}} form is used, @var{options}
3295 controls the details of the dump as described for the
3296 @option{-fdump-tree} options.
3298 @item -fdump-class-hierarchy @r{(C++ only)}
3299 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3300 @opindex fdump-class-hierarchy
3301 Dump a representation of each class's hierarchy and virtual function
3302 table layout to a file. The file name is made by appending @file{.class}
3303 to the source file name. If the @samp{-@var{options}} form is used,
3304 @var{options} controls the details of the dump as described for the
3305 @option{-fdump-tree} options.
3307 @item -fdump-tree-@var{switch} @r{(C++ only)}
3308 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3310 Control the dumping at various stages of processing the intermediate
3311 language tree to a file. The file name is generated by appending a switch
3312 specific suffix to the source file name. If the @samp{-@var{options}}
3313 form is used, @var{options} is a list of @samp{-} separated options that
3314 control the details of the dump. Not all options are applicable to all
3315 dumps, those which are not meaningful will be ignored. The following
3316 options are available
3320 Print the address of each node. Usually this is not meaningful as it
3321 changes according to the environment and source file. Its primary use
3322 is for tying up a dump file with a debug environment.
3324 Inhibit dumping of members of a scope or body of a function merely
3325 because that scope has been reached. Only dump such items when they
3326 are directly reachable by some other path.
3328 Turn on all options.
3331 The following tree dumps are possible:
3334 Dump before any tree based optimization, to @file{@var{file}.original}.
3336 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3338 Dump after function inlining, to @file{@var{file}.inlined}.
3341 @item -frandom-seed=@var{string}
3342 @opindex frandom-string
3343 This option provides a seed that GCC uses when it would otherwise use
3344 random numbers. It is used to generate certain symbol names
3345 that have to be different in every compiled file. It is also used to
3346 place unique stamps in coverage data files and the object files that
3347 produce them. You can use the @option{-frandom-seed} option to produce
3348 reproducibly identical object files.
3350 The @var{string} should be different for every file you compile.
3352 @item -fsched-verbose=@var{n}
3353 @opindex fsched-verbose
3354 On targets that use instruction scheduling, this option controls the
3355 amount of debugging output the scheduler prints. This information is
3356 written to standard error, unless @option{-dS} or @option{-dR} is
3357 specified, in which case it is output to the usual dump
3358 listing file, @file{.sched} or @file{.sched2} respectively. However
3359 for @var{n} greater than nine, the output is always printed to standard
3362 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3363 same information as @option{-dRS}. For @var{n} greater than one, it
3364 also output basic block probabilities, detailed ready list information
3365 and unit/insn info. For @var{n} greater than two, it includes RTL
3366 at abort point, control-flow and regions info. And for @var{n} over
3367 four, @option{-fsched-verbose} also includes dependence info.
3371 Store the usual ``temporary'' intermediate files permanently; place them
3372 in the current directory and name them based on the source file. Thus,
3373 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3374 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3375 preprocessed @file{foo.i} output file even though the compiler now
3376 normally uses an integrated preprocessor.
3380 Report the CPU time taken by each subprocess in the compilation
3381 sequence. For C source files, this is the compiler proper and assembler
3382 (plus the linker if linking is done). The output looks like this:
3389 The first number on each line is the ``user time,'' that is time spent
3390 executing the program itself. The second number is ``system time,''
3391 time spent executing operating system routines on behalf of the program.
3392 Both numbers are in seconds.
3394 @item -print-file-name=@var{library}
3395 @opindex print-file-name
3396 Print the full absolute name of the library file @var{library} that
3397 would be used when linking---and don't do anything else. With this
3398 option, GCC does not compile or link anything; it just prints the
3401 @item -print-multi-directory
3402 @opindex print-multi-directory
3403 Print the directory name corresponding to the multilib selected by any
3404 other switches present in the command line. This directory is supposed
3405 to exist in @env{GCC_EXEC_PREFIX}.
3407 @item -print-multi-lib
3408 @opindex print-multi-lib
3409 Print the mapping from multilib directory names to compiler switches
3410 that enable them. The directory name is separated from the switches by
3411 @samp{;}, and each switch starts with an @samp{@@} instead of the
3412 @samp{-}, without spaces between multiple switches. This is supposed to
3413 ease shell-processing.
3415 @item -print-prog-name=@var{program}
3416 @opindex print-prog-name
3417 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3419 @item -print-libgcc-file-name
3420 @opindex print-libgcc-file-name
3421 Same as @option{-print-file-name=libgcc.a}.
3423 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3424 but you do want to link with @file{libgcc.a}. You can do
3427 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3430 @item -print-search-dirs
3431 @opindex print-search-dirs
3432 Print the name of the configured installation directory and a list of
3433 program and library directories gcc will search---and don't do anything else.
3435 This is useful when gcc prints the error message
3436 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3437 To resolve this you either need to put @file{cpp0} and the other compiler
3438 components where gcc expects to find them, or you can set the environment
3439 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3440 Don't forget the trailing '/'.
3441 @xref{Environment Variables}.
3444 @opindex dumpmachine
3445 Print the compiler's target machine (for example,
3446 @samp{i686-pc-linux-gnu})---and don't do anything else.
3449 @opindex dumpversion
3450 Print the compiler version (for example, @samp{3.0})---and don't do
3455 Print the compiler's built-in specs---and don't do anything else. (This
3456 is used when GCC itself is being built.) @xref{Spec Files}.
3458 @item -feliminate-unused-debug-types
3459 @opindex feliminate-unused-debug-types
3460 Normally, when producing DWARF2 output, GCC will emit debugging
3461 information for all types declared in a compilation
3462 unit, regardless of whether or not they are actually used
3463 in that compilation unit. Sometimes this is useful, such as
3464 if, in the debugger, you want to cast a value to a type that is
3465 not actually used in your program (but is declared). More often,
3466 however, this results in a significant amount of wasted space.
3467 With this option, GCC will avoid producing debug symbol output
3468 for types that are nowhere used in the source file being compiled.
3471 @node Optimize Options
3472 @section Options That Control Optimization
3473 @cindex optimize options
3474 @cindex options, optimization
3476 These options control various sorts of optimizations.
3478 Without any optimization option, the compiler's goal is to reduce the
3479 cost of compilation and to make debugging produce the expected
3480 results. Statements are independent: if you stop the program with a
3481 breakpoint between statements, you can then assign a new value to any
3482 variable or change the program counter to any other statement in the
3483 function and get exactly the results you would expect from the source
3486 Turning on optimization flags makes the compiler attempt to improve
3487 the performance and/or code size at the expense of compilation time
3488 and possibly the ability to debug the program.
3490 The compiler performs optimisation based on the knowledge it has of
3491 the program. Using the @option{-funit-at-a-time} flag will allow the
3492 compiler to consider information gained from later functions in the
3493 file when compiling a function. Compiling multiple files at once to a
3494 single output file (and using @option{-funit-at-a-time}) will allow
3495 the compiler to use information gained from all of the files when
3496 compiling each of them.
3498 Not all optimizations are controlled directly by a flag. Only
3499 optimizations that have a flag are listed.
3506 Optimize. Optimizing compilation takes somewhat more time, and a lot
3507 more memory for a large function.
3509 With @option{-O}, the compiler tries to reduce code size and execution
3510 time, without performing any optimizations that take a great deal of
3513 @option{-O} turns on the following optimization flags:
3514 @gccoptlist{-fdefer-pop @gol
3515 -fmerge-constants @gol
3517 -floop-optimize @gol
3519 -fif-conversion @gol
3520 -fif-conversion2 @gol
3521 -fdelayed-branch @gol
3522 -fguess-branch-probability @gol
3525 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3526 where doing so does not interfere with debugging.
3530 Optimize even more. GCC performs nearly all supported optimizations
3531 that do not involve a space-speed tradeoff. The compiler does not
3532 perform loop unrolling or function inlining when you specify @option{-O2}.
3533 As compared to @option{-O}, this option increases both compilation time
3534 and the performance of the generated code.
3536 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3537 also turns on the following optimization flags:
3538 @gccoptlist{-fforce-mem @gol
3539 -foptimize-sibling-calls @gol
3540 -fstrength-reduce @gol
3541 -fcse-follow-jumps -fcse-skip-blocks @gol
3542 -frerun-cse-after-loop -frerun-loop-opt @gol
3543 -fgcse -fgcse-lm -fgcse-sm @gol
3544 -fdelete-null-pointer-checks @gol
3545 -fexpensive-optimizations @gol
3547 -fschedule-insns -fschedule-insns2 @gol
3548 -fsched-interblock -fsched-spec @gol
3551 -freorder-blocks -freorder-functions @gol
3552 -fstrict-aliasing @gol
3553 -falign-functions -falign-jumps @gol
3554 -falign-loops -falign-labels}
3556 Please note the warning under @option{-fgcse} about
3557 invoking @option{-O2} on programs that use computed gotos.
3561 Optimize yet more. @option{-O3} turns on all optimizations specified by
3562 @option{-O2} and also turns on the @option{-finline-functions},
3563 @option{-funit-at-a-time} and @option{-frename-registers} options.
3567 Do not optimize. This is the default.
3571 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3572 do not typically increase code size. It also performs further
3573 optimizations designed to reduce code size.
3575 @option{-Os} disables the following optimization flags:
3576 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3577 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3579 If you use multiple @option{-O} options, with or without level numbers,
3580 the last such option is the one that is effective.
3583 Options of the form @option{-f@var{flag}} specify machine-independent
3584 flags. Most flags have both positive and negative forms; the negative
3585 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3586 below, only one of the forms is listed---the one you typically will
3587 use. You can figure out the other form by either removing @samp{no-}
3590 The following options control specific optimizations. They are either
3591 activated by @option{-O} options or are related to ones that are. You
3592 can use the following flags in the rare cases when ``fine-tuning'' of
3593 optimizations to be performed is desired.
3596 @item -fno-default-inline
3597 @opindex fno-default-inline
3598 Do not make member functions inline by default merely because they are
3599 defined inside the class scope (C++ only). Otherwise, when you specify
3600 @w{@option{-O}}, member functions defined inside class scope are compiled
3601 inline by default; i.e., you don't need to add @samp{inline} in front of
3602 the member function name.
3604 @item -fno-defer-pop
3605 @opindex fno-defer-pop
3606 Always pop the arguments to each function call as soon as that function
3607 returns. For machines which must pop arguments after a function call,
3608 the compiler normally lets arguments accumulate on the stack for several
3609 function calls and pops them all at once.
3611 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3615 Force memory operands to be copied into registers before doing
3616 arithmetic on them. This produces better code by making all memory
3617 references potential common subexpressions. When they are not common
3618 subexpressions, instruction combination should eliminate the separate
3621 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3624 @opindex fforce-addr
3625 Force memory address constants to be copied into registers before
3626 doing arithmetic on them. This may produce better code just as
3627 @option{-fforce-mem} may.
3629 @item -fomit-frame-pointer
3630 @opindex fomit-frame-pointer
3631 Don't keep the frame pointer in a register for functions that
3632 don't need one. This avoids the instructions to save, set up and
3633 restore frame pointers; it also makes an extra register available
3634 in many functions. @strong{It also makes debugging impossible on
3637 On some machines, such as the VAX, this flag has no effect, because
3638 the standard calling sequence automatically handles the frame pointer
3639 and nothing is saved by pretending it doesn't exist. The
3640 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3641 whether a target machine supports this flag. @xref{Registers,,Register
3642 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3644 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3646 @item -foptimize-sibling-calls
3647 @opindex foptimize-sibling-calls
3648 Optimize sibling and tail recursive calls.
3650 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3654 Don't pay attention to the @code{inline} keyword. Normally this option
3655 is used to keep the compiler from expanding any functions inline.
3656 Note that if you are not optimizing, no functions can be expanded inline.
3658 @item -finline-functions
3659 @opindex finline-functions
3660 Integrate all simple functions into their callers. The compiler
3661 heuristically decides which functions are simple enough to be worth
3662 integrating in this way.
3664 If all calls to a given function are integrated, and the function is
3665 declared @code{static}, then the function is normally not output as
3666 assembler code in its own right.
3668 Enabled at level @option{-O3}.
3670 @item -finline-limit=@var{n}
3671 @opindex finline-limit
3672 By default, gcc limits the size of functions that can be inlined. This flag
3673 allows the control of this limit for functions that are explicitly marked as
3674 inline (i.e., marked with the inline keyword or defined within the class
3675 definition in c++). @var{n} is the size of functions that can be inlined in
3676 number of pseudo instructions (not counting parameter handling). The default
3677 value of @var{n} is 600.
3678 Increasing this value can result in more inlined code at
3679 the cost of compilation time and memory consumption. Decreasing usually makes
3680 the compilation faster and less code will be inlined (which presumably
3681 means slower programs). This option is particularly useful for programs that
3682 use inlining heavily such as those based on recursive templates with C++.
3684 Inlining is actually controlled by a number of parameters, which may be
3685 specified individually by using @option{--param @var{name}=@var{value}}.
3686 The @option{-finline-limit=@var{n}} option sets some of these parameters
3690 @item max-inline-insns
3692 @item max-inline-insns-single
3693 is set to @var{n}/2.
3694 @item max-inline-insns-auto
3695 is set to @var{n}/2.
3696 @item min-inline-insns
3697 is set to 130 or @var{n}/4, whichever is smaller.
3698 @item max-inline-insns-rtl
3702 Using @option{-finline-limit=600} thus results in the default settings
3703 for these parameters. See below for a documentation of the individual
3704 parameters controlling inlining.
3706 @emph{Note:} pseudo instruction represents, in this particular context, an
3707 abstract measurement of function's size. In no way, it represents a count
3708 of assembly instructions and as such its exact meaning might change from one
3709 release to an another.
3711 @item -fkeep-inline-functions
3712 @opindex fkeep-inline-functions
3713 Even if all calls to a given function are integrated, and the function
3714 is declared @code{static}, nevertheless output a separate run-time
3715 callable version of the function. This switch does not affect
3716 @code{extern inline} functions.
3718 @item -fkeep-static-consts
3719 @opindex fkeep-static-consts
3720 Emit variables declared @code{static const} when optimization isn't turned
3721 on, even if the variables aren't referenced.
3723 GCC enables this option by default. If you want to force the compiler to
3724 check if the variable was referenced, regardless of whether or not
3725 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3727 @item -fmerge-constants
3728 Attempt to merge identical constants (string constants and floating point
3729 constants) across compilation units.
3731 This option is the default for optimized compilation if the assembler and
3732 linker support it. Use @option{-fno-merge-constants} to inhibit this
3735 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3737 @item -fmerge-all-constants
3738 Attempt to merge identical constants and identical variables.
3740 This option implies @option{-fmerge-constants}. In addition to
3741 @option{-fmerge-constants} this considers e.g. even constant initialized
3742 arrays or initialized constant variables with integral or floating point
3743 types. Languages like C or C++ require each non-automatic variable to
3744 have distinct location, so using this option will result in non-conforming
3749 Use a graph coloring register allocator. Currently this option is meant
3750 for testing, so we are interested to hear about miscompilations with
3753 @item -fno-branch-count-reg
3754 @opindex fno-branch-count-reg
3755 Do not use ``decrement and branch'' instructions on a count register,
3756 but instead generate a sequence of instructions that decrement a
3757 register, compare it against zero, then branch based upon the result.
3758 This option is only meaningful on architectures that support such
3759 instructions, which include x86, PowerPC, IA-64 and S/390.
3761 The default is @option{-fbranch-count-reg}, enabled when
3762 @option{-fstrength-reduce} is enabled.
3764 @item -fno-function-cse
3765 @opindex fno-function-cse
3766 Do not put function addresses in registers; make each instruction that
3767 calls a constant function contain the function's address explicitly.
3769 This option results in less efficient code, but some strange hacks
3770 that alter the assembler output may be confused by the optimizations
3771 performed when this option is not used.
3773 The default is @option{-ffunction-cse}
3775 @item -fno-zero-initialized-in-bss
3776 @opindex fno-zero-initialized-in-bss
3777 If the target supports a BSS section, GCC by default puts variables that
3778 are initialized to zero into BSS@. This can save space in the resulting
3781 This option turns off this behavior because some programs explicitly
3782 rely on variables going to the data section. E.g., so that the
3783 resulting executable can find the beginning of that section and/or make
3784 assumptions based on that.
3786 The default is @option{-fzero-initialized-in-bss}.
3788 @item -fstrength-reduce
3789 @opindex fstrength-reduce
3790 Perform the optimizations of loop strength reduction and
3791 elimination of iteration variables.
3793 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3795 @item -fthread-jumps
3796 @opindex fthread-jumps
3797 Perform optimizations where we check to see if a jump branches to a
3798 location where another comparison subsumed by the first is found. If
3799 so, the first branch is redirected to either the destination of the
3800 second branch or a point immediately following it, depending on whether
3801 the condition is known to be true or false.
3803 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3805 @item -fcse-follow-jumps
3806 @opindex fcse-follow-jumps
3807 In common subexpression elimination, scan through jump instructions
3808 when the target of the jump is not reached by any other path. For
3809 example, when CSE encounters an @code{if} statement with an
3810 @code{else} clause, CSE will follow the jump when the condition
3813 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3815 @item -fcse-skip-blocks
3816 @opindex fcse-skip-blocks
3817 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3818 follow jumps which conditionally skip over blocks. When CSE
3819 encounters a simple @code{if} statement with no else clause,
3820 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3821 body of the @code{if}.
3823 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3825 @item -frerun-cse-after-loop
3826 @opindex frerun-cse-after-loop
3827 Re-run common subexpression elimination after loop optimizations has been
3830 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3832 @item -frerun-loop-opt
3833 @opindex frerun-loop-opt
3834 Run the loop optimizer twice.
3836 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3840 Perform a global common subexpression elimination pass.
3841 This pass also performs global constant and copy propagation.
3843 @emph{Note:} When compiling a program using computed gotos, a GCC
3844 extension, you may get better runtime performance if you disable
3845 the global common subexpression elimination pass by adding
3846 @option{-fno-gcse} to the command line.
3848 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3852 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3853 attempt to move loads which are only killed by stores into themselves. This
3854 allows a loop containing a load/store sequence to be changed to a load outside
3855 the loop, and a copy/store within the loop.
3857 Enabled by default when gcse is enabled.
3861 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3862 subexpression elimination. This pass will attempt to move stores out of loops.
3863 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3864 can be changed to a load before the loop and a store after the loop.
3866 Enabled by default when gcse is enabled.
3868 @item -floop-optimize
3869 @opindex floop-optimize
3870 Perform loop optimizations: move constant expressions out of loops, simplify
3871 exit test conditions and optionally do strength-reduction and loop unrolling as
3874 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3876 @item -fcrossjumping
3877 @opindex crossjumping
3878 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3879 resulting code may or may not perform better than without cross-jumping.
3881 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3883 @item -fif-conversion
3884 @opindex if-conversion
3885 Attempt to transform conditional jumps into branch-less equivalents. This
3886 include use of conditional moves, min, max, set flags and abs instructions, and
3887 some tricks doable by standard arithmetics. The use of conditional execution
3888 on chips where it is available is controlled by @code{if-conversion2}.
3890 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3892 @item -fif-conversion2
3893 @opindex if-conversion2
3894 Use conditional execution (where available) to transform conditional jumps into
3895 branch-less equivalents.
3897 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3899 @item -fdelete-null-pointer-checks
3900 @opindex fdelete-null-pointer-checks
3901 Use global dataflow analysis to identify and eliminate useless checks
3902 for null pointers. The compiler assumes that dereferencing a null
3903 pointer would have halted the program. If a pointer is checked after
3904 it has already been dereferenced, it cannot be null.
3906 In some environments, this assumption is not true, and programs can
3907 safely dereference null pointers. Use
3908 @option{-fno-delete-null-pointer-checks} to disable this optimization
3909 for programs which depend on that behavior.
3911 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3913 @item -fexpensive-optimizations
3914 @opindex fexpensive-optimizations
3915 Perform a number of minor optimizations that are relatively expensive.
3917 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3919 @item -foptimize-register-move
3921 @opindex foptimize-register-move
3923 Attempt to reassign register numbers in move instructions and as
3924 operands of other simple instructions in order to maximize the amount of
3925 register tying. This is especially helpful on machines with two-operand
3928 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3931 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3933 @item -fdelayed-branch
3934 @opindex fdelayed-branch
3935 If supported for the target machine, attempt to reorder instructions
3936 to exploit instruction slots available after delayed branch
3939 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3941 @item -fschedule-insns
3942 @opindex fschedule-insns
3943 If supported for the target machine, attempt to reorder instructions to
3944 eliminate execution stalls due to required data being unavailable. This
3945 helps machines that have slow floating point or memory load instructions
3946 by allowing other instructions to be issued until the result of the load
3947 or floating point instruction is required.
3949 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3951 @item -fschedule-insns2
3952 @opindex fschedule-insns2
3953 Similar to @option{-fschedule-insns}, but requests an additional pass of
3954 instruction scheduling after register allocation has been done. This is
3955 especially useful on machines with a relatively small number of
3956 registers and where memory load instructions take more than one cycle.
3958 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3960 @item -fno-sched-interblock
3961 @opindex fno-sched-interblock
3962 Don't schedule instructions across basic blocks. This is normally
3963 enabled by default when scheduling before register allocation, i.e.@:
3964 with @option{-fschedule-insns} or at @option{-O2} or higher.
3966 @item -fno-sched-spec
3967 @opindex fno-sched-spec
3968 Don't allow speculative motion of non-load instructions. This is normally
3969 enabled by default when scheduling before register allocation, i.e.@:
3970 with @option{-fschedule-insns} or at @option{-O2} or higher.
3972 @item -fsched-spec-load
3973 @opindex fsched-spec-load
3974 Allow speculative motion of some load instructions. This only makes
3975 sense when scheduling before register allocation, i.e.@: with
3976 @option{-fschedule-insns} or at @option{-O2} or higher.
3978 @item -fsched-spec-load-dangerous
3979 @opindex fsched-spec-load-dangerous
3980 Allow speculative motion of more load instructions. This only makes
3981 sense when scheduling before register allocation, i.e.@: with
3982 @option{-fschedule-insns} or at @option{-O2} or higher.
3984 @item -fsched2-use-superblocks
3985 @opindex fsched2-use-superblocks
3986 When scheduling after register allocation, do use superblock scheduling
3987 algorithm. Superblock scheduling allows motion across basic block boundaries
3988 resulting on faster schedules. This option is experimental, as not all machine
3989 descriptions used by GCC model the CPU closely enough to avoid unreliable
3990 results from the algorithm.
3992 This only makes sense when scheduling after register allocation, i.e.@: with
3993 @option{-fschedule-insns2} or at @option{-O2} or higher.
3995 @item -fsched2-use-traces
3996 @opindex fsched2-use-traces
3997 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
3998 allocation and additionally perform code duplication in order to increase the
3999 size of superblocks using tracer pass. See @option{-ftracer} for details on
4002 This mode should produce faster but significantly longer programs. Also
4003 without @code{-fbranch-probabilities} the traces constructed may not match the
4004 reality and hurt the performance. This only makes
4005 sense when scheduling after register allocation, i.e.@: with
4006 @option{-fschedule-insns2} or at @option{-O2} or higher.
4008 @item -fcaller-saves
4009 @opindex fcaller-saves
4010 Enable values to be allocated in registers that will be clobbered by
4011 function calls, by emitting extra instructions to save and restore the
4012 registers around such calls. Such allocation is done only when it
4013 seems to result in better code than would otherwise be produced.
4015 This option is always enabled by default on certain machines, usually
4016 those which have no call-preserved registers to use instead.
4018 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4020 @item -fmove-all-movables
4021 @opindex fmove-all-movables
4022 Forces all invariant computations in loops to be moved
4025 @item -freduce-all-givs
4026 @opindex freduce-all-givs
4027 Forces all general-induction variables in loops to be
4030 @emph{Note:} When compiling programs written in Fortran,
4031 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4032 by default when you use the optimizer.
4034 These options may generate better or worse code; results are highly
4035 dependent on the structure of loops within the source code.
4037 These two options are intended to be removed someday, once
4038 they have helped determine the efficacy of various
4039 approaches to improving loop optimizations.
4041 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4042 know how use of these options affects
4043 the performance of your production code.
4044 We're very interested in code that runs @emph{slower}
4045 when these options are @emph{enabled}.
4048 @itemx -fno-peephole2
4049 @opindex fno-peephole
4050 @opindex fno-peephole2
4051 Disable any machine-specific peephole optimizations. The difference
4052 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4053 are implemented in the compiler; some targets use one, some use the
4054 other, a few use both.
4056 @option{-fpeephole} is enabled by default.
4057 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4059 @item -fno-guess-branch-probability
4060 @opindex fno-guess-branch-probability
4061 Do not guess branch probabilities using a randomized model.
4063 Sometimes gcc will opt to use a randomized model to guess branch
4064 probabilities, when none are available from either profiling feedback
4065 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4066 different runs of the compiler on the same program may produce different
4069 In a hard real-time system, people don't want different runs of the
4070 compiler to produce code that has different behavior; minimizing
4071 non-determinism is of paramount import. This switch allows users to
4072 reduce non-determinism, possibly at the expense of inferior
4075 The default is @option{-fguess-branch-probability} at levels
4076 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4078 @item -freorder-blocks
4079 @opindex freorder-blocks
4080 Reorder basic blocks in the compiled function in order to reduce number of
4081 taken branches and improve code locality.
4083 Enabled at levels @option{-O2}, @option{-O3}.
4085 @item -freorder-functions
4086 @opindex freorder-functions
4087 Reorder basic blocks in the compiled function in order to reduce number of
4088 taken branches and improve code locality. This is implemented by using special
4089 subsections @code{text.hot} for most frequently executed functions and
4090 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4091 the linker so object file format must support named sections and linker must
4092 place them in a reasonable way.
4094 Also profile feedback must be available in to make this option effective. See
4095 @option{-fprofile-arcs} for details.
4097 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4099 @item -fstrict-aliasing
4100 @opindex fstrict-aliasing
4101 Allows the compiler to assume the strictest aliasing rules applicable to
4102 the language being compiled. For C (and C++), this activates
4103 optimizations based on the type of expressions. In particular, an
4104 object of one type is assumed never to reside at the same address as an
4105 object of a different type, unless the types are almost the same. For
4106 example, an @code{unsigned int} can alias an @code{int}, but not a
4107 @code{void*} or a @code{double}. A character type may alias any other
4110 Pay special attention to code like this:
4123 The practice of reading from a different union member than the one most
4124 recently written to (called ``type-punning'') is common. Even with
4125 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4126 is accessed through the union type. So, the code above will work as
4127 expected. However, this code might not:
4138 Every language that wishes to perform language-specific alias analysis
4139 should define a function that computes, given an @code{tree}
4140 node, an alias set for the node. Nodes in different alias sets are not
4141 allowed to alias. For an example, see the C front-end function
4142 @code{c_get_alias_set}.
4144 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4146 @item -falign-functions
4147 @itemx -falign-functions=@var{n}
4148 @opindex falign-functions
4149 Align the start of functions to the next power-of-two greater than
4150 @var{n}, skipping up to @var{n} bytes. For instance,
4151 @option{-falign-functions=32} aligns functions to the next 32-byte
4152 boundary, but @option{-falign-functions=24} would align to the next
4153 32-byte boundary only if this can be done by skipping 23 bytes or less.
4155 @option{-fno-align-functions} and @option{-falign-functions=1} are
4156 equivalent and mean that functions will not be aligned.
4158 Some assemblers only support this flag when @var{n} is a power of two;
4159 in that case, it is rounded up.
4161 If @var{n} is not specified or is zero, use a machine-dependent default.
4163 Enabled at levels @option{-O2}, @option{-O3}.
4165 @item -falign-labels
4166 @itemx -falign-labels=@var{n}
4167 @opindex falign-labels
4168 Align all branch targets to a power-of-two boundary, skipping up to
4169 @var{n} bytes like @option{-falign-functions}. This option can easily
4170 make code slower, because it must insert dummy operations for when the
4171 branch target is reached in the usual flow of the code.
4173 @option{-fno-align-labels} and @option{-falign-labels=1} are
4174 equivalent and mean that labels will not be aligned.
4176 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4177 are greater than this value, then their values are used instead.
4179 If @var{n} is not specified or is zero, use a machine-dependent default
4180 which is very likely to be @samp{1}, meaning no alignment.
4182 Enabled at levels @option{-O2}, @option{-O3}.
4185 @itemx -falign-loops=@var{n}
4186 @opindex falign-loops
4187 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4188 like @option{-falign-functions}. The hope is that the loop will be
4189 executed many times, which will make up for any execution of the dummy
4192 @option{-fno-align-loops} and @option{-falign-loops=1} are
4193 equivalent and mean that loops will not be aligned.
4195 If @var{n} is not specified or is zero, use a machine-dependent default.
4197 Enabled at levels @option{-O2}, @option{-O3}.
4200 @itemx -falign-jumps=@var{n}
4201 @opindex falign-jumps
4202 Align branch targets to a power-of-two boundary, for branch targets
4203 where the targets can only be reached by jumping, skipping up to @var{n}
4204 bytes like @option{-falign-functions}. In this case, no dummy operations
4207 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4208 equivalent and mean that loops will not be aligned.
4210 If @var{n} is not specified or is zero, use a machine-dependent default.
4212 Enabled at levels @option{-O2}, @option{-O3}.
4214 @item -frename-registers
4215 @opindex frename-registers
4216 Attempt to avoid false dependencies in scheduled code by making use
4217 of registers left over after register allocation. This optimization
4218 will most benefit processors with lots of registers. It can, however,
4219 make debugging impossible, since variables will no longer stay in
4220 a ``home register''.
4222 Enabled at levels @option{-O3}.
4224 @item -fno-cprop-registers
4225 @opindex fno-cprop-registers
4226 After register allocation and post-register allocation instruction splitting,
4227 we perform a copy-propagation pass to try to reduce scheduling dependencies
4228 and occasionally eliminate the copy.
4230 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4234 The following options control compiler behavior regarding floating
4235 point arithmetic. These options trade off between speed and
4236 correctness. All must be specifically enabled.
4240 @opindex ffloat-store
4241 Do not store floating point variables in registers, and inhibit other
4242 options that might change whether a floating point value is taken from a
4245 @cindex floating point precision
4246 This option prevents undesirable excess precision on machines such as
4247 the 68000 where the floating registers (of the 68881) keep more
4248 precision than a @code{double} is supposed to have. Similarly for the
4249 x86 architecture. For most programs, the excess precision does only
4250 good, but a few programs rely on the precise definition of IEEE floating
4251 point. Use @option{-ffloat-store} for such programs, after modifying
4252 them to store all pertinent intermediate computations into variables.
4256 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4257 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4258 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4260 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4262 This option should never be turned on by any @option{-O} option since
4263 it can result in incorrect output for programs which depend on
4264 an exact implementation of IEEE or ISO rules/specifications for
4267 @item -fno-math-errno
4268 @opindex fno-math-errno
4269 Do not set ERRNO after calling math functions that are executed
4270 with a single instruction, e.g., sqrt. A program that relies on
4271 IEEE exceptions for math error handling may want to use this flag
4272 for speed while maintaining IEEE arithmetic compatibility.
4274 This option should never be turned on by any @option{-O} option since
4275 it can result in incorrect output for programs which depend on
4276 an exact implementation of IEEE or ISO rules/specifications for
4279 The default is @option{-fmath-errno}.
4281 @item -funsafe-math-optimizations
4282 @opindex funsafe-math-optimizations
4283 Allow optimizations for floating-point arithmetic that (a) assume
4284 that arguments and results are valid and (b) may violate IEEE or
4285 ANSI standards. When used at link-time, it may include libraries
4286 or startup files that change the default FPU control word or other
4287 similar optimizations.
4289 This option should never be turned on by any @option{-O} option since
4290 it can result in incorrect output for programs which depend on
4291 an exact implementation of IEEE or ISO rules/specifications for
4294 The default is @option{-fno-unsafe-math-optimizations}.
4296 @item -ffinite-math-only
4297 @opindex ffinite-math-only
4298 Allow optimizations for floating-point arithmetic that assume
4299 that arguments and results are not NaNs or +-Infs.
4301 This option should never be turned on by any @option{-O} option since
4302 it can result in incorrect output for programs which depend on
4303 an exact implementation of IEEE or ISO rules/specifications.
4305 The default is @option{-fno-finite-math-only}.
4307 @item -fno-trapping-math
4308 @opindex fno-trapping-math
4309 Compile code assuming that floating-point operations cannot generate
4310 user-visible traps. These traps include division by zero, overflow,
4311 underflow, inexact result and invalid operation. This option implies
4312 @option{-fno-signaling-nans}. Setting this option may allow faster
4313 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4315 This option should never be turned on by any @option{-O} option since
4316 it can result in incorrect output for programs which depend on
4317 an exact implementation of IEEE or ISO rules/specifications for
4320 The default is @option{-ftrapping-math}.
4322 @item -frounding-math
4323 @opindex frounding-math
4324 Disable transformations and optimizations that assume default floating
4325 point rounding behavior. This is round-to-zero for all floating point
4326 to integer conversions, and round-to-nearest for all other arithmetic
4327 truncations. This option should be specified for programs that change
4328 the FP rounding mode dynamically, or that may be executed with a
4329 non-default rounding mode. This option disables constant folding of
4330 floating point expressions at compile-time (which may be affected by
4331 rounding mode) and arithmetic transformations that are unsafe in the
4332 presence of sign-dependent rounding modes.
4334 The default is @option{-fno-rounding-math}.
4336 This option is experimental and does not currently guarantee to
4337 disable all GCC optimizations that are affected by rounding mode.
4338 Future versions of gcc may provide finer control of this setting
4339 using C99's @code{FENV_ACCESS} pragma. This command line option
4340 will be used to specify the default state for @code{FENV_ACCESS}.
4342 @item -fsignaling-nans
4343 @opindex fsignaling-nans
4344 Compile code assuming that IEEE signaling NaNs may generate user-visible
4345 traps during floating-point operations. Setting this option disables
4346 optimizations that may change the number of exceptions visible with
4347 signaling NaNs. This option implies @option{-ftrapping-math}.
4349 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4352 The default is @option{-fno-signaling-nans}.
4354 This option is experimental and does not currently guarantee to
4355 disable all GCC optimizations that affect signaling NaN behavior.
4357 @item -fsingle-precision-constant
4358 @opindex fsingle-precision-constant
4359 Treat floating point constant as single precision constant instead of
4360 implicitly converting it to double precision constant.
4365 The following options control optimizations that may improve
4366 performance, but are not enabled by any @option{-O} options. This
4367 section includes experimental options that may produce broken code.
4370 @item -fbranch-probabilities
4371 @opindex fbranch-probabilities
4372 After running a program compiled with @option{-fprofile-arcs}
4373 (@pxref{Debugging Options,, Options for Debugging Your Program or
4374 @command{gcc}}), you can compile it a second time using
4375 @option{-fbranch-probabilities}, to improve optimizations based on
4376 the number of times each branch was taken. When the program
4377 compiled with @option{-fprofile-arcs} exits it saves arc execution
4378 counts to a file called @file{@var{sourcename}.gcda} for each source
4379 file The information in this data file is very dependent on the
4380 structure of the generated code, so you must use the same source code
4381 and the same optimization options for both compilations.
4383 With @option{-fbranch-probabilities}, GCC puts a
4384 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4385 These can be used to improve optimization. Currently, they are only
4386 used in one place: in @file{reorg.c}, instead of guessing which path a
4387 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4388 exactly determine which path is taken more often.
4390 @item -fprofile-values
4391 @opindex fprofile-values
4392 If combined with @option{-fprofile-arcs}, it adds code so that some
4393 data about values of expressions in the program is gathered.
4395 With @option{-fbranch-probabilities}, it reads back the data gathered
4396 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4397 notes to instructions for their later usage in optimizations.
4401 Use a graph coloring register allocator. Currently this option is meant
4402 for testing, so we are interested to hear about miscompilations with
4407 Perform tail duplication to enlarge superblock size. This transformation
4408 simplifies the control flow of the function allowing other optimizations to do
4411 @item -funit-at-a-time
4412 @opindex funit-at-a-time
4413 Parse the whole compilation unit before starting to produce code.
4414 This allows some extra optimizations to take place but consumes more
4417 @item -funroll-loops
4418 @opindex funroll-loops
4419 Unroll loops whose number of iterations can be determined at compile time or
4420 upon entry to the loop. @option{-funroll-loops} implies
4421 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4422 (i.e. complete removal of loops with small constant number of iterations).
4423 This option makes code larger, and may or may not make it run faster.
4425 @item -funroll-all-loops
4426 @opindex funroll-all-loops
4427 Unroll all loops, even if their number of iterations is uncertain when
4428 the loop is entered. This usually makes programs run more slowly.
4429 @option{-funroll-all-loops} implies the same options as
4430 @option{-funroll-loops}.
4433 @opindex fpeel-loops
4434 Peels the loops for that there is enough information that they do not
4435 roll much (from profile feedback). It also turns on complete loop peeling
4436 (i.e. complete removal of loops with small constant number of iterations).
4438 @item -funswitch-loops
4439 @opindex funswitch-loops
4440 Move branches with loop invariant conditions out of the loop, with duplicates
4441 of the loop on both branches (modified according to result of the condition).
4443 @item -fold-unroll-loops
4444 @opindex fold-unroll-loops
4445 Unroll loops whose number of iterations can be determined at compile
4446 time or upon entry to the loop, using the old loop unroller whose loop
4447 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4448 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4449 option makes code larger, and may or may not make it run faster.
4451 @item -fold-unroll-all-loops
4452 @opindex fold-unroll-all-loops
4453 Unroll all loops, even if their number of iterations is uncertain when
4454 the loop is entered. This is done using the old loop unroller whose loop
4455 recognition is based on notes from frontend. This usually makes programs run more slowly.
4456 @option{-fold-unroll-all-loops} implies the same options as
4457 @option{-fold-unroll-loops}.
4459 @item -funswitch-loops
4460 @opindex funswitch-loops
4461 Move branches with loop invariant conditions out of the loop, with duplicates
4462 of the loop on both branches (modified according to result of the condition).
4464 @item -funswitch-loops
4465 @opindex funswitch-loops
4466 Move branches with loop invariant conditions out of the loop, with duplicates
4467 of the loop on both branches (modified according to result of the condition).
4469 @item -fprefetch-loop-arrays
4470 @opindex fprefetch-loop-arrays
4471 If supported by the target machine, generate instructions to prefetch
4472 memory to improve the performance of loops that access large arrays.
4474 Disabled at level @option{-Os}.
4476 @item -ffunction-sections
4477 @itemx -fdata-sections
4478 @opindex ffunction-sections
4479 @opindex fdata-sections
4480 Place each function or data item into its own section in the output
4481 file if the target supports arbitrary sections. The name of the
4482 function or the name of the data item determines the section's name
4485 Use these options on systems where the linker can perform optimizations
4486 to improve locality of reference in the instruction space. Most systems
4487 using the ELF object format and SPARC processors running Solaris 2 have
4488 linkers with such optimizations. AIX may have these optimizations in
4491 Only use these options when there are significant benefits from doing
4492 so. When you specify these options, the assembler and linker will
4493 create larger object and executable files and will also be slower.
4494 You will not be able to use @code{gprof} on all systems if you
4495 specify this option and you may have problems with debugging if
4496 you specify both this option and @option{-g}.
4500 Perform optimizations in static single assignment form. Each function's
4501 flow graph is translated into SSA form, optimizations are performed, and
4502 the flow graph is translated back from SSA form. Users should not
4503 specify this option, since it is not yet ready for production use.
4507 Perform Sparse Conditional Constant Propagation in SSA form. Requires
4508 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
4512 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
4513 Like @option{-fssa}, this is an experimental feature.
4515 @item -fbranch-target-load-optimize
4516 @opindex fbranch-target-load-optimize
4517 Perform branch target register load optimization before prologue / epilogue
4519 The use of target registers can typically be exposed only during reload,
4520 thus hoisting loads out of loops and doing inter-block scheduling needs
4521 a separate optimization pass.
4523 @item -fbranch-target-load-optimize2
4524 @opindex fbranch-target-load-optimize2
4525 Perform branch target register load optimization after prologue / epilogue
4531 @item --param @var{name}=@var{value}
4533 In some places, GCC uses various constants to control the amount of
4534 optimization that is done. For example, GCC will not inline functions
4535 that contain more that a certain number of instructions. You can
4536 control some of these constants on the command-line using the
4537 @option{--param} option.
4539 In each case, the @var{value} is an integer. The allowable choices for
4540 @var{name} are given in the following table:
4543 @item max-crossjump-edges
4544 The maximum number of incoming edges to consider for crossjumping.
4545 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4546 the number of edges incoming to each block. Increasing values mean
4547 more aggressive optimization, making the compile time increase with
4548 probably small improvement in executable size.
4550 @item max-delay-slot-insn-search
4551 The maximum number of instructions to consider when looking for an
4552 instruction to fill a delay slot. If more than this arbitrary number of
4553 instructions is searched, the time savings from filling the delay slot
4554 will be minimal so stop searching. Increasing values mean more
4555 aggressive optimization, making the compile time increase with probably
4556 small improvement in executable run time.
4558 @item max-delay-slot-live-search
4559 When trying to fill delay slots, the maximum number of instructions to
4560 consider when searching for a block with valid live register
4561 information. Increasing this arbitrarily chosen value means more
4562 aggressive optimization, increasing the compile time. This parameter
4563 should be removed when the delay slot code is rewritten to maintain the
4566 @item max-gcse-memory
4567 The approximate maximum amount of memory that will be allocated in
4568 order to perform the global common subexpression elimination
4569 optimization. If more memory than specified is required, the
4570 optimization will not be done.
4572 @item max-gcse-passes
4573 The maximum number of passes of GCSE to run.
4575 @item max-pending-list-length
4576 The maximum number of pending dependencies scheduling will allow
4577 before flushing the current state and starting over. Large functions
4578 with few branches or calls can create excessively large lists which
4579 needlessly consume memory and resources.
4581 @item max-inline-insns-single
4582 Several parameters control the tree inliner used in gcc.
4583 This number sets the maximum number of instructions (counted in gcc's
4584 internal representation) in a single function that the tree inliner
4585 will consider for inlining. This only affects functions declared
4586 inline and methods implemented in a class declaration (C++).
4587 The default value is 500.
4589 @item max-inline-insns-auto
4590 When you use @option{-finline-functions} (included in @option{-O3}),
4591 a lot of functions that would otherwise not be considered for inlining
4592 by the compiler will be investigated. To those functions, a different
4593 (more restrictive) limit compared to functions declared inline can
4595 The default value is 150.
4597 @item max-inline-insns
4598 The tree inliner does decrease the allowable size for single functions
4599 to be inlined after we already inlined the number of instructions
4600 given here by repeated inlining. This number should be a factor of
4601 two or more larger than the single function limit.
4602 Higher numbers result in better runtime performance, but incur higher
4603 compile-time resource (CPU time, memory) requirements and result in
4604 larger binaries. Very high values are not advisable, as too large
4605 binaries may adversely affect runtime performance.
4606 The default value is 200.
4608 @item max-inline-slope
4609 After exceeding the maximum number of inlined instructions by repeated
4610 inlining, a linear function is used to decrease the allowable size
4611 for single functions. The slope of that function is the negative
4612 reciprocal of the number specified here.
4613 This parameter is ignored when @option{-funit-at-a-time} is used.
4614 The default value is 32.
4616 @item min-inline-insns
4617 The repeated inlining is throttled more and more by the linear function
4618 after exceeding the limit. To avoid too much throttling, a minimum for
4619 this function is specified here to allow repeated inlining for very small
4620 functions even when a lot of repeated inlining already has been done.
4621 This parameter is ignored when @option{-funit-at-a-time} is used.
4622 The default value is 10.
4624 @item large-function-insns
4625 The limit specifying really large functions. For functions greater than this
4626 limit inlining is constrained by @option{--param large-function-growth}.
4627 This parameter is usefull primarily to avoid extreme compilation time caused by non-linear
4628 algorithms used by the backend.
4629 This parameter is ignored when @option{-funit-at-a-time} is not used.
4630 The default value is 30000.
4632 @item large-function-growth
4633 Specifies maximal growth of large functtion caused by inlining in percents.
4634 This parameter is ignored when @option{-funit-at-a-time} is not used.
4635 The default value is 200.
4637 @item inline-unit-growth
4638 Specifies maximal overall growth of the compilation unit caused by inlining.
4639 This parameter is ignored when @option{-funit-at-a-time} is not used.
4640 The default value is 150.
4642 @item max-inline-insns-rtl
4643 For languages that use the RTL inliner (this happens at a later stage
4644 than tree inlining), you can set the maximum allowable size (counted
4645 in RTL instructions) for the RTL inliner with this parameter.
4646 The default value is 600.
4649 @item max-unrolled-insns
4650 The maximum number of instructions that a loop should have if that loop
4651 is unrolled, and if the loop is unrolled, it determines how many times
4652 the loop code is unrolled.
4654 @item max-average-unrolled-insns
4655 The maximum number of instructions biased by probabilities of their execution
4656 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4657 it determines how many times the loop code is unrolled.
4659 @item max-unroll-times
4660 The maximum number of unrollings of a single loop.
4662 @item max-peeled-insns
4663 The maximum number of instructions that a loop should have if that loop
4664 is peeled, and if the loop is peeled, it determines how many times
4665 the loop code is peeled.
4667 @item max-peel-times
4668 The maximum number of peelings of a single loop.
4670 @item max-completely-peeled-insns
4671 The maximum number of insns of a completely peeled loop.
4673 @item max-completely-peel-times
4674 The maximum number of iterations of a loop to be suitable for complete peeling.
4676 @item max-unswitch-insns
4677 The maximum number of insns of an unswitched loop.
4679 @item max-unswitch-level
4680 The maximum number of branches unswitched in a single loop.
4682 @item hot-bb-count-fraction
4683 Select fraction of the maximal count of repetitions of basic block in program
4684 given basic block needs to have to be considered hot.
4686 @item hot-bb-frequency-fraction
4687 Select fraction of the maximal frequency of executions of basic block in
4688 function given basic block needs to have to be considered hot
4690 @item tracer-dynamic-coverage
4691 @itemx tracer-dynamic-coverage-feedback
4693 This value is used to limit superblock formation once the given percentage of
4694 executed instructions is covered. This limits unnecessary code size
4697 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4698 feedback is available. The real profiles (as opposed to statically estimated
4699 ones) are much less balanced allowing the threshold to be larger value.
4701 @item tracer-max-code-growth
4702 Stop tail duplication once code growth has reached given percentage. This is
4703 rather hokey argument, as most of the duplicates will be eliminated later in
4704 cross jumping, so it may be set to much higher values than is the desired code
4707 @item tracer-min-branch-ratio
4709 Stop reverse growth when the reverse probability of best edge is less than this
4710 threshold (in percent).
4712 @item tracer-min-branch-ratio
4713 @itemx tracer-min-branch-ratio-feedback
4715 Stop forward growth if the best edge do have probability lower than this
4718 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4719 compilation for profile feedback and one for compilation without. The value
4720 for compilation with profile feedback needs to be more conservative (higher) in
4721 order to make tracer effective.
4723 @item max-cse-path-length
4725 Maximum number of basic blocks on path that cse considers.
4727 @item ggc-min-expand
4729 GCC uses a garbage collector to manage its own memory allocation. This
4730 parameter specifies the minimum percentage by which the garbage
4731 collector's heap should be allowed to expand between collections.
4732 Tuning this may improve compilation speed; it has no effect on code
4735 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4736 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4737 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4738 GCC is not able to calculate RAM on a particular platform, the lower
4739 bound of 30% is used. Setting this parameter and
4740 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4741 every opportunity. This is extremely slow, but can be useful for
4744 @item ggc-min-heapsize
4746 Minimum size of the garbage collector's heap before it begins bothering
4747 to collect garbage. The first collection occurs after the heap expands
4748 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4749 tuning this may improve compilation speed, and has no effect on code
4752 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4753 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4754 available, the notion of "RAM" is the smallest of actual RAM,
4755 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4756 RAM on a particular platform, the lower bound is used. Setting this
4757 parameter very large effectively disables garbage collection. Setting
4758 this parameter and @option{ggc-min-expand} to zero causes a full
4759 collection to occur at every opportunity.
4761 @item reorder-blocks-duplicate
4762 @itemx reorder-blocks-duplicate-feedback
4764 Used by basic block reordering pass to decide whether to use unconditional
4765 branch or duplicate the code on its destination. Code is duplicated when its
4766 estimated size is smaller than this value multiplied by the estimated size of
4767 unconditional jump in the hot spots of the program.
4769 The @option{reorder-block-duplicate-feedback} is used only when profile
4770 feedback is available and may be set to higher values than
4771 @option{reorder-block-duplicate} since information about the hot spots is more
4776 @node Preprocessor Options
4777 @section Options Controlling the Preprocessor
4778 @cindex preprocessor options
4779 @cindex options, preprocessor
4781 These options control the C preprocessor, which is run on each C source
4782 file before actual compilation.
4784 If you use the @option{-E} option, nothing is done except preprocessing.
4785 Some of these options make sense only together with @option{-E} because
4786 they cause the preprocessor output to be unsuitable for actual
4791 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4792 and pass @var{option} directly through to the preprocessor. If
4793 @var{option} contains commas, it is split into multiple options at the
4794 commas. However, many options are modified, translated or interpreted
4795 by the compiler driver before being passed to the preprocessor, and
4796 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4797 interface is undocumented and subject to change, so whenever possible
4798 you should avoid using @option{-Wp} and let the driver handle the
4801 @item -Xpreprocessor @var{option}
4802 @opindex preprocessor
4803 Pass @var{option} as an option to the preprocessor. You can use this to
4804 supply system-specific preprocessor options which GCC does not know how to
4807 If you want to pass an option that takes an argument, you must use
4808 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4811 @include cppopts.texi
4813 @node Assembler Options
4814 @section Passing Options to the Assembler
4816 @c prevent bad page break with this line
4817 You can pass options to the assembler.
4820 @item -Wa,@var{option}
4822 Pass @var{option} as an option to the assembler. If @var{option}
4823 contains commas, it is split into multiple options at the commas.
4825 @item -Xassembler @var{option}
4827 Pass @var{option} as an option to the assembler. You can use this to
4828 supply system-specific assembler options which GCC does not know how to
4831 If you want to pass an option that takes an argument, you must use
4832 @option{-Xassembler} twice, once for the option and once for the argument.
4837 @section Options for Linking
4838 @cindex link options
4839 @cindex options, linking
4841 These options come into play when the compiler links object files into
4842 an executable output file. They are meaningless if the compiler is
4843 not doing a link step.
4847 @item @var{object-file-name}
4848 A file name that does not end in a special recognized suffix is
4849 considered to name an object file or library. (Object files are
4850 distinguished from libraries by the linker according to the file
4851 contents.) If linking is done, these object files are used as input
4860 If any of these options is used, then the linker is not run, and
4861 object file names should not be used as arguments. @xref{Overall
4865 @item -l@var{library}
4866 @itemx -l @var{library}
4868 Search the library named @var{library} when linking. (The second
4869 alternative with the library as a separate argument is only for
4870 POSIX compliance and is not recommended.)
4872 It makes a difference where in the command you write this option; the
4873 linker searches and processes libraries and object files in the order they
4874 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4875 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4876 to functions in @samp{z}, those functions may not be loaded.
4878 The linker searches a standard list of directories for the library,
4879 which is actually a file named @file{lib@var{library}.a}. The linker
4880 then uses this file as if it had been specified precisely by name.
4882 The directories searched include several standard system directories
4883 plus any that you specify with @option{-L}.
4885 Normally the files found this way are library files---archive files
4886 whose members are object files. The linker handles an archive file by
4887 scanning through it for members which define symbols that have so far
4888 been referenced but not defined. But if the file that is found is an
4889 ordinary object file, it is linked in the usual fashion. The only
4890 difference between using an @option{-l} option and specifying a file name
4891 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4892 and searches several directories.
4896 You need this special case of the @option{-l} option in order to
4897 link an Objective-C program.
4900 @opindex nostartfiles
4901 Do not use the standard system startup files when linking.
4902 The standard system libraries are used normally, unless @option{-nostdlib}
4903 or @option{-nodefaultlibs} is used.
4905 @item -nodefaultlibs
4906 @opindex nodefaultlibs
4907 Do not use the standard system libraries when linking.
4908 Only the libraries you specify will be passed to the linker.
4909 The standard startup files are used normally, unless @option{-nostartfiles}
4910 is used. The compiler may generate calls to memcmp, memset, and memcpy
4911 for System V (and ISO C) environments or to bcopy and bzero for
4912 BSD environments. These entries are usually resolved by entries in
4913 libc. These entry points should be supplied through some other
4914 mechanism when this option is specified.
4918 Do not use the standard system startup files or libraries when linking.
4919 No startup files and only the libraries you specify will be passed to
4920 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4921 for System V (and ISO C) environments or to bcopy and bzero for
4922 BSD environments. These entries are usually resolved by entries in
4923 libc. These entry points should be supplied through some other
4924 mechanism when this option is specified.
4926 @cindex @option{-lgcc}, use with @option{-nostdlib}
4927 @cindex @option{-nostdlib} and unresolved references
4928 @cindex unresolved references and @option{-nostdlib}
4929 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4930 @cindex @option{-nodefaultlibs} and unresolved references
4931 @cindex unresolved references and @option{-nodefaultlibs}
4932 One of the standard libraries bypassed by @option{-nostdlib} and
4933 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4934 that GCC uses to overcome shortcomings of particular machines, or special
4935 needs for some languages.
4936 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
4937 Collection (GCC) Internals},
4938 for more discussion of @file{libgcc.a}.)
4939 In most cases, you need @file{libgcc.a} even when you want to avoid
4940 other standard libraries. In other words, when you specify @option{-nostdlib}
4941 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4942 This ensures that you have no unresolved references to internal GCC
4943 library subroutines. (For example, @samp{__main}, used to ensure C++
4944 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
4945 GNU Compiler Collection (GCC) Internals}.)
4949 Produce a position independent executable on targets which support it.
4950 For predictable results, you must also specify the same set of options
4951 that were used to generate code (@option{-fpie}, @option{-fPIE},
4952 or model suboptions) when you specify this option.
4956 Remove all symbol table and relocation information from the executable.
4960 On systems that support dynamic linking, this prevents linking with the shared
4961 libraries. On other systems, this option has no effect.
4965 Produce a shared object which can then be linked with other objects to
4966 form an executable. Not all systems support this option. For predictable
4967 results, you must also specify the same set of options that were used to
4968 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4969 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4970 needs to build supplementary stub code for constructors to work. On
4971 multi-libbed systems, @samp{gcc -shared} must select the correct support
4972 libraries to link against. Failing to supply the correct flags may lead
4973 to subtle defects. Supplying them in cases where they are not necessary
4976 @item -shared-libgcc
4977 @itemx -static-libgcc
4978 @opindex shared-libgcc
4979 @opindex static-libgcc
4980 On systems that provide @file{libgcc} as a shared library, these options
4981 force the use of either the shared or static version respectively.
4982 If no shared version of @file{libgcc} was built when the compiler was
4983 configured, these options have no effect.
4985 There are several situations in which an application should use the
4986 shared @file{libgcc} instead of the static version. The most common
4987 of these is when the application wishes to throw and catch exceptions
4988 across different shared libraries. In that case, each of the libraries
4989 as well as the application itself should use the shared @file{libgcc}.
4991 Therefore, the G++ and GCJ drivers automatically add
4992 @option{-shared-libgcc} whenever you build a shared library or a main
4993 executable, because C++ and Java programs typically use exceptions, so
4994 this is the right thing to do.
4996 If, instead, you use the GCC driver to create shared libraries, you may
4997 find that they will not always be linked with the shared @file{libgcc}.
4998 If GCC finds, at its configuration time, that you have a GNU linker that
4999 does not support option @option{--eh-frame-hdr}, it will link the shared
5000 version of @file{libgcc} into shared libraries by default. Otherwise,
5001 it will take advantage of the linker and optimize away the linking with
5002 the shared version of @file{libgcc}, linking with the static version of
5003 libgcc by default. This allows exceptions to propagate through such
5004 shared libraries, without incurring relocation costs at library load
5007 However, if a library or main executable is supposed to throw or catch
5008 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5009 for the languages used in the program, or using the option
5010 @option{-shared-libgcc}, such that it is linked with the shared
5015 Bind references to global symbols when building a shared object. Warn
5016 about any unresolved references (unless overridden by the link editor
5017 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5020 @item -Xlinker @var{option}
5022 Pass @var{option} as an option to the linker. You can use this to
5023 supply system-specific linker options which GCC does not know how to
5026 If you want to pass an option that takes an argument, you must use
5027 @option{-Xlinker} twice, once for the option and once for the argument.
5028 For example, to pass @option{-assert definitions}, you must write
5029 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5030 @option{-Xlinker "-assert definitions"}, because this passes the entire
5031 string as a single argument, which is not what the linker expects.
5033 @item -Wl,@var{option}
5035 Pass @var{option} as an option to the linker. If @var{option} contains
5036 commas, it is split into multiple options at the commas.
5038 @item -u @var{symbol}
5040 Pretend the symbol @var{symbol} is undefined, to force linking of
5041 library modules to define it. You can use @option{-u} multiple times with
5042 different symbols to force loading of additional library modules.
5045 @node Directory Options
5046 @section Options for Directory Search
5047 @cindex directory options
5048 @cindex options, directory search
5051 These options specify directories to search for header files, for
5052 libraries and for parts of the compiler:
5057 Add the directory @var{dir} to the head of the list of directories to be
5058 searched for header files. This can be used to override a system header
5059 file, substituting your own version, since these directories are
5060 searched before the system header file directories. However, you should
5061 not use this option to add directories that contain vendor-supplied
5062 system header files (use @option{-isystem} for that). If you use more than
5063 one @option{-I} option, the directories are scanned in left-to-right
5064 order; the standard system directories come after.
5066 If a standard system include directory, or a directory specified with
5067 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5068 option will be ignored. The directory will still be searched but as a
5069 system directory at its normal position in the system include chain.
5070 This is to ensure that GCC's procedure to fix buggy system headers and
5071 the ordering for the include_next directive are not inadvertently changed.
5072 If you really need to change the search order for system directories,
5073 use the @option{-nostdinc} and/or @option{-isystem} options.
5077 Any directories you specify with @option{-I} options before the @option{-I-}
5078 option are searched only for the case of @samp{#include "@var{file}"};
5079 they are not searched for @samp{#include <@var{file}>}.
5081 If additional directories are specified with @option{-I} options after
5082 the @option{-I-}, these directories are searched for all @samp{#include}
5083 directives. (Ordinarily @emph{all} @option{-I} directories are used
5086 In addition, the @option{-I-} option inhibits the use of the current
5087 directory (where the current input file came from) as the first search
5088 directory for @samp{#include "@var{file}"}. There is no way to
5089 override this effect of @option{-I-}. With @option{-I.} you can specify
5090 searching the directory which was current when the compiler was
5091 invoked. That is not exactly the same as what the preprocessor does
5092 by default, but it is often satisfactory.
5094 @option{-I-} does not inhibit the use of the standard system directories
5095 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5100 Add directory @var{dir} to the list of directories to be searched
5103 @item -B@var{prefix}
5105 This option specifies where to find the executables, libraries,
5106 include files, and data files of the compiler itself.
5108 The compiler driver program runs one or more of the subprograms
5109 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5110 @var{prefix} as a prefix for each program it tries to run, both with and
5111 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5113 For each subprogram to be run, the compiler driver first tries the
5114 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5115 was not specified, the driver tries two standard prefixes, which are
5116 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5117 those results in a file name that is found, the unmodified program
5118 name is searched for using the directories specified in your
5119 @env{PATH} environment variable.
5121 The compiler will check to see if the path provided by the @option{-B}
5122 refers to a directory, and if necessary it will add a directory
5123 separator character at the end of the path.
5125 @option{-B} prefixes that effectively specify directory names also apply
5126 to libraries in the linker, because the compiler translates these
5127 options into @option{-L} options for the linker. They also apply to
5128 includes files in the preprocessor, because the compiler translates these
5129 options into @option{-isystem} options for the preprocessor. In this case,
5130 the compiler appends @samp{include} to the prefix.
5132 The run-time support file @file{libgcc.a} can also be searched for using
5133 the @option{-B} prefix, if needed. If it is not found there, the two
5134 standard prefixes above are tried, and that is all. The file is left
5135 out of the link if it is not found by those means.
5137 Another way to specify a prefix much like the @option{-B} prefix is to use
5138 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5141 As a special kludge, if the path provided by @option{-B} is
5142 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5143 9, then it will be replaced by @file{[dir/]include}. This is to help
5144 with boot-strapping the compiler.
5146 @item -specs=@var{file}
5148 Process @var{file} after the compiler reads in the standard @file{specs}
5149 file, in order to override the defaults that the @file{gcc} driver
5150 program uses when determining what switches to pass to @file{cc1},
5151 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5152 @option{-specs=@var{file}} can be specified on the command line, and they
5153 are processed in order, from left to right.
5159 @section Specifying subprocesses and the switches to pass to them
5162 @command{gcc} is a driver program. It performs its job by invoking a
5163 sequence of other programs to do the work of compiling, assembling and
5164 linking. GCC interprets its command-line parameters and uses these to
5165 deduce which programs it should invoke, and which command-line options
5166 it ought to place on their command lines. This behavior is controlled
5167 by @dfn{spec strings}. In most cases there is one spec string for each
5168 program that GCC can invoke, but a few programs have multiple spec
5169 strings to control their behavior. The spec strings built into GCC can
5170 be overridden by using the @option{-specs=} command-line switch to specify
5173 @dfn{Spec files} are plaintext files that are used to construct spec
5174 strings. They consist of a sequence of directives separated by blank
5175 lines. The type of directive is determined by the first non-whitespace
5176 character on the line and it can be one of the following:
5179 @item %@var{command}
5180 Issues a @var{command} to the spec file processor. The commands that can
5184 @item %include <@var{file}>
5186 Search for @var{file} and insert its text at the current point in the
5189 @item %include_noerr <@var{file}>
5190 @cindex %include_noerr
5191 Just like @samp{%include}, but do not generate an error message if the include
5192 file cannot be found.
5194 @item %rename @var{old_name} @var{new_name}
5196 Rename the spec string @var{old_name} to @var{new_name}.
5200 @item *[@var{spec_name}]:
5201 This tells the compiler to create, override or delete the named spec
5202 string. All lines after this directive up to the next directive or
5203 blank line are considered to be the text for the spec string. If this
5204 results in an empty string then the spec will be deleted. (Or, if the
5205 spec did not exist, then nothing will happened.) Otherwise, if the spec
5206 does not currently exist a new spec will be created. If the spec does
5207 exist then its contents will be overridden by the text of this
5208 directive, unless the first character of that text is the @samp{+}
5209 character, in which case the text will be appended to the spec.
5211 @item [@var{suffix}]:
5212 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5213 and up to the next directive or blank line are considered to make up the
5214 spec string for the indicated suffix. When the compiler encounters an
5215 input file with the named suffix, it will processes the spec string in
5216 order to work out how to compile that file. For example:
5223 This says that any input file whose name ends in @samp{.ZZ} should be
5224 passed to the program @samp{z-compile}, which should be invoked with the
5225 command-line switch @option{-input} and with the result of performing the
5226 @samp{%i} substitution. (See below.)
5228 As an alternative to providing a spec string, the text that follows a
5229 suffix directive can be one of the following:
5232 @item @@@var{language}
5233 This says that the suffix is an alias for a known @var{language}. This is
5234 similar to using the @option{-x} command-line switch to GCC to specify a
5235 language explicitly. For example:
5242 Says that .ZZ files are, in fact, C++ source files.
5245 This causes an error messages saying:
5248 @var{name} compiler not installed on this system.
5252 GCC already has an extensive list of suffixes built into it.
5253 This directive will add an entry to the end of the list of suffixes, but
5254 since the list is searched from the end backwards, it is effectively
5255 possible to override earlier entries using this technique.
5259 GCC has the following spec strings built into it. Spec files can
5260 override these strings or create their own. Note that individual
5261 targets can also add their own spec strings to this list.
5264 asm Options to pass to the assembler
5265 asm_final Options to pass to the assembler post-processor
5266 cpp Options to pass to the C preprocessor
5267 cc1 Options to pass to the C compiler
5268 cc1plus Options to pass to the C++ compiler
5269 endfile Object files to include at the end of the link
5270 link Options to pass to the linker
5271 lib Libraries to include on the command line to the linker
5272 libgcc Decides which GCC support library to pass to the linker
5273 linker Sets the name of the linker
5274 predefines Defines to be passed to the C preprocessor
5275 signed_char Defines to pass to CPP to say whether @code{char} is signed
5277 startfile Object files to include at the start of the link
5280 Here is a small example of a spec file:
5286 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5289 This example renames the spec called @samp{lib} to @samp{old_lib} and
5290 then overrides the previous definition of @samp{lib} with a new one.
5291 The new definition adds in some extra command-line options before
5292 including the text of the old definition.
5294 @dfn{Spec strings} are a list of command-line options to be passed to their
5295 corresponding program. In addition, the spec strings can contain
5296 @samp{%}-prefixed sequences to substitute variable text or to
5297 conditionally insert text into the command line. Using these constructs
5298 it is possible to generate quite complex command lines.
5300 Here is a table of all defined @samp{%}-sequences for spec
5301 strings. Note that spaces are not generated automatically around the
5302 results of expanding these sequences. Therefore you can concatenate them
5303 together or combine them with constant text in a single argument.
5307 Substitute one @samp{%} into the program name or argument.
5310 Substitute the name of the input file being processed.
5313 Substitute the basename of the input file being processed.
5314 This is the substring up to (and not including) the last period
5315 and not including the directory.
5318 This is the same as @samp{%b}, but include the file suffix (text after
5322 Marks the argument containing or following the @samp{%d} as a
5323 temporary file name, so that that file will be deleted if GCC exits
5324 successfully. Unlike @samp{%g}, this contributes no text to the
5327 @item %g@var{suffix}
5328 Substitute a file name that has suffix @var{suffix} and is chosen
5329 once per compilation, and mark the argument in the same way as
5330 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5331 name is now chosen in a way that is hard to predict even when previously
5332 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5333 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5334 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5335 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5336 was simply substituted with a file name chosen once per compilation,
5337 without regard to any appended suffix (which was therefore treated
5338 just like ordinary text), making such attacks more likely to succeed.
5340 @item %u@var{suffix}
5341 Like @samp{%g}, but generates a new temporary file name even if
5342 @samp{%u@var{suffix}} was already seen.
5344 @item %U@var{suffix}
5345 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5346 new one if there is no such last file name. In the absence of any
5347 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5348 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5349 would involve the generation of two distinct file names, one
5350 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5351 simply substituted with a file name chosen for the previous @samp{%u},
5352 without regard to any appended suffix.
5354 @item %j@var{suffix}
5355 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5356 writable, and if save-temps is off; otherwise, substitute the name
5357 of a temporary file, just like @samp{%u}. This temporary file is not
5358 meant for communication between processes, but rather as a junk
5361 @item %|@var{suffix}
5362 @itemx %m@var{suffix}
5363 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5364 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5365 all. These are the two most common ways to instruct a program that it
5366 should read from standard input or write to standard output. If you
5367 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5368 construct: see for example @file{f/lang-specs.h}.
5370 @item %.@var{SUFFIX}
5371 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5372 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5373 terminated by the next space or %.
5376 Marks the argument containing or following the @samp{%w} as the
5377 designated output file of this compilation. This puts the argument
5378 into the sequence of arguments that @samp{%o} will substitute later.
5381 Substitutes the names of all the output files, with spaces
5382 automatically placed around them. You should write spaces
5383 around the @samp{%o} as well or the results are undefined.
5384 @samp{%o} is for use in the specs for running the linker.
5385 Input files whose names have no recognized suffix are not compiled
5386 at all, but they are included among the output files, so they will
5390 Substitutes the suffix for object files. Note that this is
5391 handled specially when it immediately follows @samp{%g, %u, or %U},
5392 because of the need for those to form complete file names. The
5393 handling is such that @samp{%O} is treated exactly as if it had already
5394 been substituted, except that @samp{%g, %u, and %U} do not currently
5395 support additional @var{suffix} characters following @samp{%O} as they would
5396 following, for example, @samp{.o}.
5399 Substitutes the standard macro predefinitions for the
5400 current target machine. Use this when running @code{cpp}.
5403 Like @samp{%p}, but puts @samp{__} before and after the name of each
5404 predefined macro, except for macros that start with @samp{__} or with
5405 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5409 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5410 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5411 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5415 Current argument is the name of a library or startup file of some sort.
5416 Search for that file in a standard list of directories and substitute
5417 the full name found.
5420 Print @var{str} as an error message. @var{str} is terminated by a newline.
5421 Use this when inconsistent options are detected.
5424 Substitute the contents of spec string @var{name} at this point.
5427 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5429 @item %x@{@var{option}@}
5430 Accumulate an option for @samp{%X}.
5433 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5437 Output the accumulated assembler options specified by @option{-Wa}.
5440 Output the accumulated preprocessor options specified by @option{-Wp}.
5443 Process the @code{asm} spec. This is used to compute the
5444 switches to be passed to the assembler.
5447 Process the @code{asm_final} spec. This is a spec string for
5448 passing switches to an assembler post-processor, if such a program is
5452 Process the @code{link} spec. This is the spec for computing the
5453 command line passed to the linker. Typically it will make use of the
5454 @samp{%L %G %S %D and %E} sequences.
5457 Dump out a @option{-L} option for each directory that GCC believes might
5458 contain startup files. If the target supports multilibs then the
5459 current multilib directory will be prepended to each of these paths.
5462 Output the multilib directory with directory separators replaced with
5463 @samp{_}. If multilib directories are not set, or the multilib directory is
5464 @file{.} then this option emits nothing.
5467 Process the @code{lib} spec. This is a spec string for deciding which
5468 libraries should be included on the command line to the linker.
5471 Process the @code{libgcc} spec. This is a spec string for deciding
5472 which GCC support library should be included on the command line to the linker.
5475 Process the @code{startfile} spec. This is a spec for deciding which
5476 object files should be the first ones passed to the linker. Typically
5477 this might be a file named @file{crt0.o}.
5480 Process the @code{endfile} spec. This is a spec string that specifies
5481 the last object files that will be passed to the linker.
5484 Process the @code{cpp} spec. This is used to construct the arguments
5485 to be passed to the C preprocessor.
5488 Process the @code{signed_char} spec. This is intended to be used
5489 to tell cpp whether a char is signed. It typically has the definition:
5491 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5495 Process the @code{cc1} spec. This is used to construct the options to be
5496 passed to the actual C compiler (@samp{cc1}).
5499 Process the @code{cc1plus} spec. This is used to construct the options to be
5500 passed to the actual C++ compiler (@samp{cc1plus}).
5503 Substitute the variable part of a matched option. See below.
5504 Note that each comma in the substituted string is replaced by
5508 Remove all occurrences of @code{-S} from the command line. Note---this
5509 command is position dependent. @samp{%} commands in the spec string
5510 before this one will see @code{-S}, @samp{%} commands in the spec string
5511 after this one will not.
5513 @item %:@var{function}(@var{args})
5514 Call the named function @var{function}, passing it @var{args}.
5515 @var{args} is first processed as a nested spec string, then split
5516 into an argument vector in the usual fashion. The function returns
5517 a string which is processed as if it had appeared literally as part
5518 of the current spec.
5520 The following built-in spec functions are provided:
5523 @item @code{if-exists}
5524 The @code{if-exists} spec function takes one argument, an absolute
5525 pathname to a file. If the file exists, @code{if-exists} returns the
5526 pathname. Here is a small example of its usage:
5530 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5533 @item @code{if-exists-else}
5534 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5535 spec function, except that it takes two arguments. The first argument is
5536 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5537 returns the pathname. If it does not exist, it returns the second argument.
5538 This way, @code{if-exists-else} can be used to select one file or another,
5539 based on the existence of the first. Here is a small example of its usage:
5543 crt0%O%s %:if-exists(crti%O%s) \
5544 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5549 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5550 If that switch was not specified, this substitutes nothing. Note that
5551 the leading dash is omitted when specifying this option, and it is
5552 automatically inserted if the substitution is performed. Thus the spec
5553 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5554 and would output the command line option @option{-foo}.
5556 @item %W@{@code{S}@}
5557 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5560 @item %@{@code{S}*@}
5561 Substitutes all the switches specified to GCC whose names start
5562 with @code{-S}, but which also take an argument. This is used for
5563 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5564 GCC considers @option{-o foo} as being
5565 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5566 text, including the space. Thus two arguments would be generated.
5568 @item %@{@code{S}*&@code{T}*@}
5569 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5570 (the order of @code{S} and @code{T} in the spec is not significant).
5571 There can be any number of ampersand-separated variables; for each the
5572 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5574 @item %@{@code{S}:@code{X}@}
5575 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5577 @item %@{!@code{S}:@code{X}@}
5578 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5580 @item %@{@code{S}*:@code{X}@}
5581 Substitutes @code{X} if one or more switches whose names start with
5582 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5583 once, no matter how many such switches appeared. However, if @code{%*}
5584 appears somewhere in @code{X}, then @code{X} will be substituted once
5585 for each matching switch, with the @code{%*} replaced by the part of
5586 that switch that matched the @code{*}.
5588 @item %@{.@code{S}:@code{X}@}
5589 Substitutes @code{X}, if processing a file with suffix @code{S}.
5591 @item %@{!.@code{S}:@code{X}@}
5592 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5594 @item %@{@code{S}|@code{P}:@code{X}@}
5595 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5596 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5597 although they have a stronger binding than the @samp{|}. If @code{%*}
5598 appears in @code{X}, all of the alternatives must be starred, and only
5599 the first matching alternative is substituted.
5601 For example, a spec string like this:
5604 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5607 will output the following command-line options from the following input
5608 command-line options:
5613 -d fred.c -foo -baz -boggle
5614 -d jim.d -bar -baz -boggle
5617 @item %@{S:X; T:Y; :D@}
5619 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5620 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5621 be as many clauses as you need. This may be combined with @code{.},
5622 @code{!}, @code{|}, and @code{*} as needed.
5627 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5628 construct may contain other nested @samp{%} constructs or spaces, or
5629 even newlines. They are processed as usual, as described above.
5630 Trailing white space in @code{X} is ignored. White space may also
5631 appear anywhere on the left side of the colon in these constructs,
5632 except between @code{.} or @code{*} and the corresponding word.
5634 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5635 handled specifically in these constructs. If another value of
5636 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5637 @option{-W} switch is found later in the command line, the earlier
5638 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5639 just one letter, which passes all matching options.
5641 The character @samp{|} at the beginning of the predicate text is used to
5642 indicate that a command should be piped to the following command, but
5643 only if @option{-pipe} is specified.
5645 It is built into GCC which switches take arguments and which do not.
5646 (You might think it would be useful to generalize this to allow each
5647 compiler's spec to say which switches take arguments. But this cannot
5648 be done in a consistent fashion. GCC cannot even decide which input
5649 files have been specified without knowing which switches take arguments,
5650 and it must know which input files to compile in order to tell which
5653 GCC also knows implicitly that arguments starting in @option{-l} are to be
5654 treated as compiler output files, and passed to the linker in their
5655 proper position among the other output files.
5657 @c man begin OPTIONS
5659 @node Target Options
5660 @section Specifying Target Machine and Compiler Version
5661 @cindex target options
5662 @cindex cross compiling
5663 @cindex specifying machine version
5664 @cindex specifying compiler version and target machine
5665 @cindex compiler version, specifying
5666 @cindex target machine, specifying
5668 The usual way to run GCC is to run the executable called @file{gcc}, or
5669 @file{<machine>-gcc} when cross-compiling, or
5670 @file{<machine>-gcc-<version>} to run a version other than the one that
5671 was installed last. Sometimes this is inconvenient, so GCC provides
5672 options that will switch to another cross-compiler or version.
5675 @item -b @var{machine}
5677 The argument @var{machine} specifies the target machine for compilation.
5679 The value to use for @var{machine} is the same as was specified as the
5680 machine type when configuring GCC as a cross-compiler. For
5681 example, if a cross-compiler was configured with @samp{configure
5682 i386v}, meaning to compile for an 80386 running System V, then you
5683 would specify @option{-b i386v} to run that cross compiler.
5685 @item -V @var{version}
5687 The argument @var{version} specifies which version of GCC to run.
5688 This is useful when multiple versions are installed. For example,
5689 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5692 The @option{-V} and @option{-b} options work by running the
5693 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5694 use them if you can just run that directly.
5696 @node Submodel Options
5697 @section Hardware Models and Configurations
5698 @cindex submodel options
5699 @cindex specifying hardware config
5700 @cindex hardware models and configurations, specifying
5701 @cindex machine dependent options
5703 Earlier we discussed the standard option @option{-b} which chooses among
5704 different installed compilers for completely different target
5705 machines, such as VAX vs.@: 68000 vs.@: 80386.
5707 In addition, each of these target machine types can have its own
5708 special options, starting with @samp{-m}, to choose among various
5709 hardware models or configurations---for example, 68010 vs 68020,
5710 floating coprocessor or none. A single installed version of the
5711 compiler can compile for any model or configuration, according to the
5714 Some configurations of the compiler also support additional special
5715 options, usually for compatibility with other compilers on the same
5718 These options are defined by the macro @code{TARGET_SWITCHES} in the
5719 machine description. The default for the options is also defined by
5720 that macro, which enables you to change the defaults.
5732 * RS/6000 and PowerPC Options::
5736 * i386 and x86-64 Options::
5738 * Intel 960 Options::
5739 * DEC Alpha Options::
5740 * DEC Alpha/VMS Options::
5743 * System V Options::
5744 * TMS320C3x/C4x Options::
5752 * S/390 and zSeries Options::
5756 * Xstormy16 Options::
5761 @node M680x0 Options
5762 @subsection M680x0 Options
5763 @cindex M680x0 options
5765 These are the @samp{-m} options defined for the 68000 series. The default
5766 values for these options depends on which style of 68000 was selected when
5767 the compiler was configured; the defaults for the most common choices are
5775 Generate output for a 68000. This is the default
5776 when the compiler is configured for 68000-based systems.
5778 Use this option for microcontrollers with a 68000 or EC000 core,
5779 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5785 Generate output for a 68020. This is the default
5786 when the compiler is configured for 68020-based systems.
5790 Generate output containing 68881 instructions for floating point.
5791 This is the default for most 68020 systems unless @option{--nfp} was
5792 specified when the compiler was configured.
5796 Generate output for a 68030. This is the default when the compiler is
5797 configured for 68030-based systems.
5801 Generate output for a 68040. This is the default when the compiler is
5802 configured for 68040-based systems.
5804 This option inhibits the use of 68881/68882 instructions that have to be
5805 emulated by software on the 68040. Use this option if your 68040 does not
5806 have code to emulate those instructions.
5810 Generate output for a 68060. This is the default when the compiler is
5811 configured for 68060-based systems.
5813 This option inhibits the use of 68020 and 68881/68882 instructions that
5814 have to be emulated by software on the 68060. Use this option if your 68060
5815 does not have code to emulate those instructions.
5819 Generate output for a CPU32. This is the default
5820 when the compiler is configured for CPU32-based systems.
5822 Use this option for microcontrollers with a
5823 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5824 68336, 68340, 68341, 68349 and 68360.
5828 Generate output for a 520X ``coldfire'' family cpu. This is the default
5829 when the compiler is configured for 520X-based systems.
5831 Use this option for microcontroller with a 5200 core, including
5832 the MCF5202, MCF5203, MCF5204 and MCF5202.
5837 Generate output for a 68040, without using any of the new instructions.
5838 This results in code which can run relatively efficiently on either a
5839 68020/68881 or a 68030 or a 68040. The generated code does use the
5840 68881 instructions that are emulated on the 68040.
5844 Generate output for a 68060, 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 68060.
5850 @opindex msoft-float
5851 Generate output containing library calls for floating point.
5852 @strong{Warning:} the requisite libraries are not available for all m68k
5853 targets. Normally the facilities of the machine's usual C compiler are
5854 used, but this can't be done directly in cross-compilation. You must
5855 make your own arrangements to provide suitable library functions for
5856 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5857 @samp{m68k-*-coff} do provide software floating point support.
5861 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5864 @opindex mnobitfield
5865 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5866 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5870 Do use the bit-field instructions. The @option{-m68020} option implies
5871 @option{-mbitfield}. This is the default if you use a configuration
5872 designed for a 68020.
5876 Use a different function-calling convention, in which functions
5877 that take a fixed number of arguments return with the @code{rtd}
5878 instruction, which pops their arguments while returning. This
5879 saves one instruction in the caller since there is no need to pop
5880 the arguments there.
5882 This calling convention is incompatible with the one normally
5883 used on Unix, so you cannot use it if you need to call libraries
5884 compiled with the Unix compiler.
5886 Also, you must provide function prototypes for all functions that
5887 take variable numbers of arguments (including @code{printf});
5888 otherwise incorrect code will be generated for calls to those
5891 In addition, seriously incorrect code will result if you call a
5892 function with too many arguments. (Normally, extra arguments are
5893 harmlessly ignored.)
5895 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5896 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5899 @itemx -mno-align-int
5901 @opindex mno-align-int
5902 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5903 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5904 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5905 Aligning variables on 32-bit boundaries produces code that runs somewhat
5906 faster on processors with 32-bit busses at the expense of more memory.
5908 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5909 align structures containing the above types differently than
5910 most published application binary interface specifications for the m68k.
5914 Use the pc-relative addressing mode of the 68000 directly, instead of
5915 using a global offset table. At present, this option implies @option{-fpic},
5916 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5917 not presently supported with @option{-mpcrel}, though this could be supported for
5918 68020 and higher processors.
5920 @item -mno-strict-align
5921 @itemx -mstrict-align
5922 @opindex mno-strict-align
5923 @opindex mstrict-align
5924 Do not (do) assume that unaligned memory references will be handled by
5929 @node M68hc1x Options
5930 @subsection M68hc1x Options
5931 @cindex M68hc1x options
5933 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5934 microcontrollers. The default values for these options depends on
5935 which style of microcontroller was selected when the compiler was configured;
5936 the defaults for the most common choices are given below.
5943 Generate output for a 68HC11. This is the default
5944 when the compiler is configured for 68HC11-based systems.
5950 Generate output for a 68HC12. This is the default
5951 when the compiler is configured for 68HC12-based systems.
5957 Generate output for a 68HCS12.
5960 @opindex mauto-incdec
5961 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5968 Enable the use of 68HC12 min and max instructions.
5971 @itemx -mno-long-calls
5972 @opindex mlong-calls
5973 @opindex mno-long-calls
5974 Treat all calls as being far away (near). If calls are assumed to be
5975 far away, the compiler will use the @code{call} instruction to
5976 call a function and the @code{rtc} instruction for returning.
5980 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5982 @item -msoft-reg-count=@var{count}
5983 @opindex msoft-reg-count
5984 Specify the number of pseudo-soft registers which are used for the
5985 code generation. The maximum number is 32. Using more pseudo-soft
5986 register may or may not result in better code depending on the program.
5987 The default is 4 for 68HC11 and 2 for 68HC12.
5992 @subsection VAX Options
5995 These @samp{-m} options are defined for the VAX:
6000 Do not output certain jump instructions (@code{aobleq} and so on)
6001 that the Unix assembler for the VAX cannot handle across long
6006 Do output those jump instructions, on the assumption that you
6007 will assemble with the GNU assembler.
6011 Output code for g-format floating point numbers instead of d-format.
6015 @subsection SPARC Options
6016 @cindex SPARC options
6018 These @samp{-m} switches are supported on the SPARC:
6023 @opindex mno-app-regs
6025 Specify @option{-mapp-regs} to generate output using the global registers
6026 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6029 To be fully SVR4 ABI compliant at the cost of some performance loss,
6030 specify @option{-mno-app-regs}. You should compile libraries and system
6031 software with this option.
6036 @opindex mhard-float
6037 Generate output containing floating point instructions. This is the
6043 @opindex msoft-float
6044 Generate output containing library calls for floating point.
6045 @strong{Warning:} the requisite libraries are not available for all SPARC
6046 targets. Normally the facilities of the machine's usual C compiler are
6047 used, but this cannot be done directly in cross-compilation. You must make
6048 your own arrangements to provide suitable library functions for
6049 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6050 @samp{sparclite-*-*} do provide software floating point support.
6052 @option{-msoft-float} changes the calling convention in the output file;
6053 therefore, it is only useful if you compile @emph{all} of a program with
6054 this option. In particular, you need to compile @file{libgcc.a}, the
6055 library that comes with GCC, with @option{-msoft-float} in order for
6058 @item -mhard-quad-float
6059 @opindex mhard-quad-float
6060 Generate output containing quad-word (long double) floating point
6063 @item -msoft-quad-float
6064 @opindex msoft-quad-float
6065 Generate output containing library calls for quad-word (long double)
6066 floating point instructions. The functions called are those specified
6067 in the SPARC ABI@. This is the default.
6069 As of this writing, there are no sparc implementations that have hardware
6070 support for the quad-word floating point instructions. They all invoke
6071 a trap handler for one of these instructions, and then the trap handler
6072 emulates the effect of the instruction. Because of the trap handler overhead,
6073 this is much slower than calling the ABI library routines. Thus the
6074 @option{-msoft-quad-float} option is the default.
6080 With @option{-mflat}, the compiler does not generate save/restore instructions
6081 and will use a ``flat'' or single register window calling convention.
6082 This model uses %i7 as the frame pointer and is compatible with the normal
6083 register window model. Code from either may be intermixed.
6084 The local registers and the input registers (0--5) are still treated as
6085 ``call saved'' registers and will be saved on the stack as necessary.
6087 With @option{-mno-flat} (the default), the compiler emits save/restore
6088 instructions (except for leaf functions) and is the normal mode of operation.
6090 @item -mno-unaligned-doubles
6091 @itemx -munaligned-doubles
6092 @opindex mno-unaligned-doubles
6093 @opindex munaligned-doubles
6094 Assume that doubles have 8 byte alignment. This is the default.
6096 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6097 alignment only if they are contained in another type, or if they have an
6098 absolute address. Otherwise, it assumes they have 4 byte alignment.
6099 Specifying this option avoids some rare compatibility problems with code
6100 generated by other compilers. It is not the default because it results
6101 in a performance loss, especially for floating point code.
6103 @item -mno-faster-structs
6104 @itemx -mfaster-structs
6105 @opindex mno-faster-structs
6106 @opindex mfaster-structs
6107 With @option{-mfaster-structs}, the compiler assumes that structures
6108 should have 8 byte alignment. This enables the use of pairs of
6109 @code{ldd} and @code{std} instructions for copies in structure
6110 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6111 However, the use of this changed alignment directly violates the SPARC
6112 ABI@. Thus, it's intended only for use on targets where the developer
6113 acknowledges that their resulting code will not be directly in line with
6114 the rules of the ABI@.
6117 @opindex mimpure-text
6118 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6119 the compiler to not pass @option{-z text} to the linker when linking a
6120 shared object. Using this option, you can link position-dependent
6121 code into a shared object.
6123 @option{-mimpure-text} suppresses the ``relocations remain against
6124 allocatable but non-writable sections'' linker error message.
6125 However, the necessary relocations will trigger copy-on-write, and the
6126 shared object is not actually shared across processes. Instead of
6127 using @option{-mimpure-text}, you should compile all source code with
6128 @option{-fpic} or @option{-fPIC}.
6130 This option is only available on SunOS and Solaris.
6136 These two options select variations on the SPARC architecture.
6138 By default (unless specifically configured for the Fujitsu SPARClite),
6139 GCC generates code for the v7 variant of the SPARC architecture.
6141 @option{-mv8} will give you SPARC v8 code. The only difference from v7
6142 code is that the compiler emits the integer multiply and integer
6143 divide instructions which exist in SPARC v8 but not in SPARC v7.
6145 @option{-msparclite} will give you SPARClite code. This adds the integer
6146 multiply, integer divide step and scan (@code{ffs}) instructions which
6147 exist in SPARClite but not in SPARC v7.
6149 These options are deprecated and will be deleted in a future GCC release.
6150 They have been replaced with @option{-mcpu=xxx}.
6155 @opindex msupersparc
6156 These two options select the processor for which the code is optimized.
6158 With @option{-mcypress} (the default), the compiler optimizes code for the
6159 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
6160 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6162 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6163 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6164 of the full SPARC v8 instruction set.
6166 These options are deprecated and will be deleted in a future GCC release.
6167 They have been replaced with @option{-mcpu=xxx}.
6169 @item -mcpu=@var{cpu_type}
6171 Set the instruction set, register set, and instruction scheduling parameters
6172 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6173 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6174 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
6175 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6178 Default instruction scheduling parameters are used for values that select
6179 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6180 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6182 Here is a list of each supported architecture and their supported
6187 v8: supersparc, hypersparc
6188 sparclite: f930, f934, sparclite86x
6190 v9: ultrasparc, ultrasparc3
6193 @item -mtune=@var{cpu_type}
6195 Set the instruction scheduling parameters for machine type
6196 @var{cpu_type}, but do not set the instruction set or register set that the
6197 option @option{-mcpu=@var{cpu_type}} would.
6199 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6200 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6201 that select a particular cpu implementation. Those are @samp{cypress},
6202 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6203 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6208 These @samp{-m} switches are supported in addition to the above
6209 on the SPARCLET processor.
6212 @item -mlittle-endian
6213 @opindex mlittle-endian
6214 Generate code for a processor running in little-endian mode.
6218 Treat register @code{%g0} as a normal register.
6219 GCC will continue to clobber it as necessary but will not assume
6220 it always reads as 0.
6222 @item -mbroken-saverestore
6223 @opindex mbroken-saverestore
6224 Generate code that does not use non-trivial forms of the @code{save} and
6225 @code{restore} instructions. Early versions of the SPARCLET processor do
6226 not correctly handle @code{save} and @code{restore} instructions used with
6227 arguments. They correctly handle them used without arguments. A @code{save}
6228 instruction used without arguments increments the current window pointer
6229 but does not allocate a new stack frame. It is assumed that the window
6230 overflow trap handler will properly handle this case as will interrupt
6234 These @samp{-m} switches are supported in addition to the above
6235 on SPARC V9 processors in 64-bit environments.
6238 @item -mlittle-endian
6239 @opindex mlittle-endian
6240 Generate code for a processor running in little-endian mode. It is only
6241 available for a few configurations and most notably not on Solaris.
6247 Generate code for a 32-bit or 64-bit environment.
6248 The 32-bit environment sets int, long and pointer to 32 bits.
6249 The 64-bit environment sets int to 32 bits and long and pointer
6252 @item -mcmodel=medlow
6253 @opindex mcmodel=medlow
6254 Generate code for the Medium/Low code model: the program must be linked
6255 in the low 32 bits of the address space. Pointers are 64 bits.
6256 Programs can be statically or dynamically linked.
6258 @item -mcmodel=medmid
6259 @opindex mcmodel=medmid
6260 Generate code for the Medium/Middle code model: the program must be linked
6261 in the low 44 bits of the address space, the text segment must be less than
6262 2G bytes, and data segment must be within 2G of the text segment.
6263 Pointers are 64 bits.
6265 @item -mcmodel=medany
6266 @opindex mcmodel=medany
6267 Generate code for the Medium/Anywhere code model: the program may be linked
6268 anywhere in 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=embmedany
6273 @opindex mcmodel=embmedany
6274 Generate code for the Medium/Anywhere code model for embedded systems:
6275 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6276 (determined at link time). Register %g4 points to the base of the
6277 data segment. Pointers are still 64 bits.
6278 Programs are statically linked, PIC is not supported.
6281 @itemx -mno-stack-bias
6282 @opindex mstack-bias
6283 @opindex mno-stack-bias
6284 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6285 frame pointer if present, are offset by @minus{}2047 which must be added back
6286 when making stack frame references.
6287 Otherwise, assume no such offset is present.
6291 @subsection ARM Options
6294 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6299 @opindex mapcs-frame
6300 Generate a stack frame that is compliant with the ARM Procedure Call
6301 Standard for all functions, even if this is not strictly necessary for
6302 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6303 with this option will cause the stack frames not to be generated for
6304 leaf functions. The default is @option{-mno-apcs-frame}.
6308 This is a synonym for @option{-mapcs-frame}.
6312 Generate code for a processor running with a 26-bit program counter,
6313 and conforming to the function calling standards for the APCS 26-bit
6314 option. This option replaces the @option{-m2} and @option{-m3} options
6315 of previous releases of the compiler.
6319 Generate code for a processor running with a 32-bit program counter,
6320 and conforming to the function calling standards for the APCS 32-bit
6321 option. This option replaces the @option{-m6} option of previous releases
6325 @c not currently implemented
6326 @item -mapcs-stack-check
6327 @opindex mapcs-stack-check
6328 Generate code to check the amount of stack space available upon entry to
6329 every function (that actually uses some stack space). If there is
6330 insufficient space available then either the function
6331 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6332 called, depending upon the amount of stack space required. The run time
6333 system is required to provide these functions. The default is
6334 @option{-mno-apcs-stack-check}, since this produces smaller code.
6336 @c not currently implemented
6338 @opindex mapcs-float
6339 Pass floating point arguments using the float point registers. This is
6340 one of the variants of the APCS@. This option is recommended if the
6341 target hardware has a floating point unit or if a lot of floating point
6342 arithmetic is going to be performed by the code. The default is
6343 @option{-mno-apcs-float}, since integer only code is slightly increased in
6344 size if @option{-mapcs-float} is used.
6346 @c not currently implemented
6347 @item -mapcs-reentrant
6348 @opindex mapcs-reentrant
6349 Generate reentrant, position independent code. The default is
6350 @option{-mno-apcs-reentrant}.
6353 @item -mthumb-interwork
6354 @opindex mthumb-interwork
6355 Generate code which supports calling between the ARM and Thumb
6356 instruction sets. Without this option the two instruction sets cannot
6357 be reliably used inside one program. The default is
6358 @option{-mno-thumb-interwork}, since slightly larger code is generated
6359 when @option{-mthumb-interwork} is specified.
6361 @item -mno-sched-prolog
6362 @opindex mno-sched-prolog
6363 Prevent the reordering of instructions in the function prolog, or the
6364 merging of those instruction with the instructions in the function's
6365 body. This means that all functions will start with a recognizable set
6366 of instructions (or in fact one of a choice from a small set of
6367 different function prologues), and this information can be used to
6368 locate the start if functions inside an executable piece of code. The
6369 default is @option{-msched-prolog}.
6372 @opindex mhard-float
6373 Generate output containing floating point instructions. This is the
6377 @opindex msoft-float
6378 Generate output containing library calls for floating point.
6379 @strong{Warning:} the requisite libraries are not available for all ARM
6380 targets. Normally the facilities of the machine's usual C compiler are
6381 used, but this cannot be done directly in cross-compilation. You must make
6382 your own arrangements to provide suitable library functions for
6385 @option{-msoft-float} changes the calling convention in the output file;
6386 therefore, it is only useful if you compile @emph{all} of a program with
6387 this option. In particular, you need to compile @file{libgcc.a}, the
6388 library that comes with GCC, with @option{-msoft-float} in order for
6391 @item -mlittle-endian
6392 @opindex mlittle-endian
6393 Generate code for a processor running in little-endian mode. This is
6394 the default for all standard configurations.
6397 @opindex mbig-endian
6398 Generate code for a processor running in big-endian mode; the default is
6399 to compile code for a little-endian processor.
6401 @item -mwords-little-endian
6402 @opindex mwords-little-endian
6403 This option only applies when generating code for big-endian processors.
6404 Generate code for a little-endian word order but a big-endian byte
6405 order. That is, a byte order of the form @samp{32107654}. Note: this
6406 option should only be used if you require compatibility with code for
6407 big-endian ARM processors generated by versions of the compiler prior to
6410 @item -malignment-traps
6411 @opindex malignment-traps
6412 Generate code that will not trap if the MMU has alignment traps enabled.
6413 On ARM architectures prior to ARMv4, there were no instructions to
6414 access half-word objects stored in memory. However, when reading from
6415 memory a feature of the ARM architecture allows a word load to be used,
6416 even if the address is unaligned, and the processor core will rotate the
6417 data as it is being loaded. This option tells the compiler that such
6418 misaligned accesses will cause a MMU trap and that it should instead
6419 synthesize the access as a series of byte accesses. The compiler can
6420 still use word accesses to load half-word data if it knows that the
6421 address is aligned to a word boundary.
6423 This option is ignored when compiling for ARM architecture 4 or later,
6424 since these processors have instructions to directly access half-word
6427 @item -mno-alignment-traps
6428 @opindex mno-alignment-traps
6429 Generate code that assumes that the MMU will not trap unaligned
6430 accesses. This produces better code when the target instruction set
6431 does not have half-word memory operations (i.e.@: implementations prior to
6434 Note that you cannot use this option to access unaligned word objects,
6435 since the processor will only fetch one 32-bit aligned object from
6438 The default setting for most targets is @option{-mno-alignment-traps}, since
6439 this produces better code when there are no half-word memory
6440 instructions available.
6442 @item -mshort-load-bytes
6443 @itemx -mno-short-load-words
6444 @opindex mshort-load-bytes
6445 @opindex mno-short-load-words
6446 These are deprecated aliases for @option{-malignment-traps}.
6448 @item -mno-short-load-bytes
6449 @itemx -mshort-load-words
6450 @opindex mno-short-load-bytes
6451 @opindex mshort-load-words
6452 This are deprecated aliases for @option{-mno-alignment-traps}.
6454 @item -mcpu=@var{name}
6456 This specifies the name of the target ARM processor. GCC uses this name
6457 to determine what kind of instructions it can emit when generating
6458 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6459 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6460 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6461 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6462 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6463 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6464 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6465 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6466 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6467 @samp{arm1020t}, @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
6469 @itemx -mtune=@var{name}
6471 This option is very similar to the @option{-mcpu=} option, except that
6472 instead of specifying the actual target processor type, and hence
6473 restricting which instructions can be used, it specifies that GCC should
6474 tune the performance of the code as if the target were of the type
6475 specified in this option, but still choosing the instructions that it
6476 will generate based on the cpu specified by a @option{-mcpu=} option.
6477 For some ARM implementations better performance can be obtained by using
6480 @item -march=@var{name}
6482 This specifies the name of the target ARM architecture. GCC uses this
6483 name to determine what kind of instructions it can emit when generating
6484 assembly code. This option can be used in conjunction with or instead
6485 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6486 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6487 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{iwmmxt}, @samp{ep9312}.
6489 @item -mfpe=@var{number}
6490 @itemx -mfp=@var{number}
6493 This specifies the version of the floating point emulation available on
6494 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6495 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6497 @item -mstructure-size-boundary=@var{n}
6498 @opindex mstructure-size-boundary
6499 The size of all structures and unions will be rounded up to a multiple
6500 of the number of bits set by this option. Permissible values are 8 and
6501 32. The default value varies for different toolchains. For the COFF
6502 targeted toolchain the default value is 8. Specifying the larger number
6503 can produce faster, more efficient code, but can also increase the size
6504 of the program. The two values are potentially incompatible. Code
6505 compiled with one value cannot necessarily expect to work with code or
6506 libraries compiled with the other value, if they exchange information
6507 using structures or unions.
6509 @item -mabort-on-noreturn
6510 @opindex mabort-on-noreturn
6511 Generate a call to the function @code{abort} at the end of a
6512 @code{noreturn} function. It will be executed if the function tries to
6516 @itemx -mno-long-calls
6517 @opindex mlong-calls
6518 @opindex mno-long-calls
6519 Tells the compiler to perform function calls by first loading the
6520 address of the function into a register and then performing a subroutine
6521 call on this register. This switch is needed if the target function
6522 will lie outside of the 64 megabyte addressing range of the offset based
6523 version of subroutine call instruction.
6525 Even if this switch is enabled, not all function calls will be turned
6526 into long calls. The heuristic is that static functions, functions
6527 which have the @samp{short-call} attribute, functions that are inside
6528 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6529 definitions have already been compiled within the current compilation
6530 unit, will not be turned into long calls. The exception to this rule is
6531 that weak function definitions, functions with the @samp{long-call}
6532 attribute or the @samp{section} attribute, and functions that are within
6533 the scope of a @samp{#pragma long_calls} directive, will always be
6534 turned into long calls.
6536 This feature is not enabled by default. Specifying
6537 @option{-mno-long-calls} will restore the default behavior, as will
6538 placing the function calls within the scope of a @samp{#pragma
6539 long_calls_off} directive. Note these switches have no effect on how
6540 the compiler generates code to handle function calls via function
6543 @item -mnop-fun-dllimport
6544 @opindex mnop-fun-dllimport
6545 Disable support for the @code{dllimport} attribute.
6547 @item -msingle-pic-base
6548 @opindex msingle-pic-base
6549 Treat the register used for PIC addressing as read-only, rather than
6550 loading it in the prologue for each function. The run-time system is
6551 responsible for initializing this register with an appropriate value
6552 before execution begins.
6554 @item -mpic-register=@var{reg}
6555 @opindex mpic-register
6556 Specify the register to be used for PIC addressing. The default is R10
6557 unless stack-checking is enabled, when R9 is used.
6559 @item -mcirrus-fix-invalid-insns
6560 @opindex mcirrus-fix-invalid-insns
6561 @opindex mno-cirrus-fix-invalid-insns
6562 Insert NOPs into the instruction stream to in order to work around
6563 problems with invalid Maverick instruction combinations. This option
6564 is only valid if the @option{-mcpu=ep9312} option has been used to
6565 enable generation of instructions for the Cirrus Maverick floating
6566 point co-processor. This option is not enabled by default, since the
6567 problem is only present in older Maverick implementations. The default
6568 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6571 @item -mpoke-function-name
6572 @opindex mpoke-function-name
6573 Write the name of each function into the text section, directly
6574 preceding the function prologue. The generated code is similar to this:
6578 .ascii "arm_poke_function_name", 0
6581 .word 0xff000000 + (t1 - t0)
6582 arm_poke_function_name
6584 stmfd sp!, @{fp, ip, lr, pc@}
6588 When performing a stack backtrace, code can inspect the value of
6589 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6590 location @code{pc - 12} and the top 8 bits are set, then we know that
6591 there is a function name embedded immediately preceding this location
6592 and has length @code{((pc[-3]) & 0xff000000)}.
6596 Generate code for the 16-bit Thumb instruction set. The default is to
6597 use the 32-bit ARM instruction set.
6600 @opindex mtpcs-frame
6601 Generate a stack frame that is compliant with the Thumb Procedure Call
6602 Standard for all non-leaf functions. (A leaf function is one that does
6603 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6605 @item -mtpcs-leaf-frame
6606 @opindex mtpcs-leaf-frame
6607 Generate a stack frame that is compliant with the Thumb Procedure Call
6608 Standard for all leaf functions. (A leaf function is one that does
6609 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6611 @item -mcallee-super-interworking
6612 @opindex mcallee-super-interworking
6613 Gives all externally visible functions in the file being compiled an ARM
6614 instruction set header which switches to Thumb mode before executing the
6615 rest of the function. This allows these functions to be called from
6616 non-interworking code.
6618 @item -mcaller-super-interworking
6619 @opindex mcaller-super-interworking
6620 Allows calls via function pointers (including virtual functions) to
6621 execute correctly regardless of whether the target code has been
6622 compiled for interworking or not. There is a small overhead in the cost
6623 of executing a function pointer if this option is enabled.
6627 @node MN10200 Options
6628 @subsection MN10200 Options
6629 @cindex MN10200 options
6631 These @option{-m} options are defined for Matsushita MN10200 architectures:
6636 Indicate to the linker that it should perform a relaxation optimization pass
6637 to shorten branches, calls and absolute memory addresses. This option only
6638 has an effect when used on the command line for the final link step.
6640 This option makes symbolic debugging impossible.
6643 @node MN10300 Options
6644 @subsection MN10300 Options
6645 @cindex MN10300 options
6647 These @option{-m} options are defined for Matsushita MN10300 architectures:
6652 Generate code to avoid bugs in the multiply instructions for the MN10300
6653 processors. This is the default.
6656 @opindex mno-mult-bug
6657 Do not generate code to avoid bugs in the multiply instructions for the
6662 Generate code which uses features specific to the AM33 processor.
6666 Do not generate code which uses features specific to the AM33 processor. This
6671 Do not link in the C run-time initialization object file.
6675 Indicate to the linker that it should perform a relaxation optimization pass
6676 to shorten branches, calls and absolute memory addresses. This option only
6677 has an effect when used on the command line for the final link step.
6679 This option makes symbolic debugging impossible.
6683 @node M32R/D Options
6684 @subsection M32R/D Options
6685 @cindex M32R/D options
6687 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6692 Generate code for the M32R/X@.
6696 Generate code for the M32R@. This is the default.
6698 @item -mcode-model=small
6699 @opindex mcode-model=small
6700 Assume all objects live in the lower 16MB of memory (so that their addresses
6701 can be loaded with the @code{ld24} instruction), and assume all subroutines
6702 are reachable with the @code{bl} instruction.
6703 This is the default.
6705 The addressability of a particular object can be set with the
6706 @code{model} attribute.
6708 @item -mcode-model=medium
6709 @opindex mcode-model=medium
6710 Assume objects may be anywhere in the 32-bit address space (the compiler
6711 will generate @code{seth/add3} instructions to load their addresses), and
6712 assume all subroutines are reachable with the @code{bl} instruction.
6714 @item -mcode-model=large
6715 @opindex mcode-model=large
6716 Assume objects may be anywhere in the 32-bit address space (the compiler
6717 will generate @code{seth/add3} instructions to load their addresses), and
6718 assume subroutines may not be reachable with the @code{bl} instruction
6719 (the compiler will generate the much slower @code{seth/add3/jl}
6720 instruction sequence).
6723 @opindex msdata=none
6724 Disable use of the small data area. Variables will be put into
6725 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6726 @code{section} attribute has been specified).
6727 This is the default.
6729 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6730 Objects may be explicitly put in the small data area with the
6731 @code{section} attribute using one of these sections.
6734 @opindex msdata=sdata
6735 Put small global and static data in the small data area, but do not
6736 generate special code to reference them.
6740 Put small global and static data in the small data area, and generate
6741 special instructions to reference them.
6745 @cindex smaller data references
6746 Put global and static objects less than or equal to @var{num} bytes
6747 into the small data or bss sections instead of the normal data or bss
6748 sections. The default value of @var{num} is 8.
6749 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6750 for this option to have any effect.
6752 All modules should be compiled with the same @option{-G @var{num}} value.
6753 Compiling with different values of @var{num} may or may not work; if it
6754 doesn't the linker will give an error message---incorrect code will not be
6760 @subsection M88K Options
6761 @cindex M88k options
6763 These @samp{-m} options are defined for Motorola 88k architectures:
6768 Generate code that works well on both the m88100 and the
6773 Generate code that works best for the m88100, but that also
6778 Generate code that works best for the m88110, and may not run
6783 Obsolete option to be removed from the next revision.
6786 @item -midentify-revision
6787 @opindex midentify-revision
6788 @cindex identifying source, compiler (88k)
6789 Include an @code{ident} directive in the assembler output recording the
6790 source file name, compiler name and version, timestamp, and compilation
6793 @item -mno-underscores
6794 @opindex mno-underscores
6795 @cindex underscores, avoiding (88k)
6796 In assembler output, emit symbol names without adding an underscore
6797 character at the beginning of each name. The default is to use an
6798 underscore as prefix on each name.
6800 @item -mocs-debug-info
6801 @itemx -mno-ocs-debug-info
6802 @opindex mocs-debug-info
6803 @opindex mno-ocs-debug-info
6805 @cindex debugging, 88k OCS
6806 Include (or omit) additional debugging information (about registers used
6807 in each stack frame) as specified in the 88open Object Compatibility
6808 Standard, ``OCS''@. This extra information allows debugging of code that
6809 has had the frame pointer eliminated. The default for SVr4 and Delta 88
6810 SVr3.2 is to include this information; other 88k configurations omit this
6811 information by default.
6813 @item -mocs-frame-position
6814 @opindex mocs-frame-position
6815 @cindex register positions in frame (88k)
6816 When emitting COFF debugging information for automatic variables and
6817 parameters stored on the stack, use the offset from the canonical frame
6818 address, which is the stack pointer (register 31) on entry to the
6819 function. The SVr4 and Delta88 SVr3.2, and BCS configurations use
6820 @option{-mocs-frame-position}; other 88k configurations have the default
6821 @option{-mno-ocs-frame-position}.
6823 @item -mno-ocs-frame-position
6824 @opindex mno-ocs-frame-position
6825 @cindex register positions in frame (88k)
6826 When emitting COFF debugging information for automatic variables and
6827 parameters stored on the stack, use the offset from the frame pointer
6828 register (register 30). When this option is in effect, the frame
6829 pointer is not eliminated when debugging information is selected by the
6832 @item -moptimize-arg-area
6833 @opindex moptimize-arg-area
6834 @cindex arguments in frame (88k)
6835 Save space by reorganizing the stack frame. This option generates code
6836 that does not agree with the 88open specifications, but uses less
6839 @itemx -mno-optimize-arg-area
6840 @opindex mno-optimize-arg-area
6841 Do not reorganize the stack frame to save space. This is the default.
6842 The generated conforms to the specification, but uses more memory.
6844 @item -mshort-data-@var{num}
6845 @opindex mshort-data
6846 @cindex smaller data references (88k)
6847 @cindex r0-relative references (88k)
6848 Generate smaller data references by making them relative to @code{r0},
6849 which allows loading a value using a single instruction (rather than the
6850 usual two). You control which data references are affected by
6851 specifying @var{num} with this option. For example, if you specify
6852 @option{-mshort-data-512}, then the data references affected are those
6853 involving displacements of less than 512 bytes.
6854 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6857 @item -mserialize-volatile
6858 @opindex mserialize-volatile
6859 @itemx -mno-serialize-volatile
6860 @opindex mno-serialize-volatile
6861 @cindex sequential consistency on 88k
6862 Do, or don't, generate code to guarantee sequential consistency
6863 of volatile memory references. By default, consistency is
6866 The order of memory references made by the MC88110 processor does
6867 not always match the order of the instructions requesting those
6868 references. In particular, a load instruction may execute before
6869 a preceding store instruction. Such reordering violates
6870 sequential consistency of volatile memory references, when there
6871 are multiple processors. When consistency must be guaranteed,
6872 GCC generates special instructions, as needed, to force
6873 execution in the proper order.
6875 The MC88100 processor does not reorder memory references and so
6876 always provides sequential consistency. However, by default, GCC
6877 generates the special instructions to guarantee consistency
6878 even when you use @option{-m88100}, so that the code may be run on an
6879 MC88110 processor. If you intend to run your code only on the
6880 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6882 The extra code generated to guarantee consistency may affect the
6883 performance of your application. If you know that you can safely
6884 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6890 @cindex assembler syntax, 88k
6892 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6893 related to System V release 4 (SVr4). This controls the following:
6897 Which variant of the assembler syntax to emit.
6899 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6900 that is used on System V release 4.
6902 @option{-msvr4} makes GCC issue additional declaration directives used in
6906 @option{-msvr4} is the default for the m88k-motorola-sysv4 configuration.
6907 @option{-msvr3} is the default for all other m88k configurations.
6909 @item -mversion-03.00
6910 @opindex mversion-03.00
6911 This option is obsolete, and is ignored.
6912 @c ??? which asm syntax better for GAS? option there too?
6914 @item -mno-check-zero-division
6915 @itemx -mcheck-zero-division
6916 @opindex mno-check-zero-division
6917 @opindex mcheck-zero-division
6918 @cindex zero division on 88k
6919 Do, or don't, generate code to guarantee that integer division by
6920 zero will be detected. By default, detection is guaranteed.
6922 Some models of the MC88100 processor fail to trap upon integer
6923 division by zero under certain conditions. By default, when
6924 compiling code that might be run on such a processor, GCC
6925 generates code that explicitly checks for zero-valued divisors
6926 and traps with exception number 503 when one is detected. Use of
6927 @option{-mno-check-zero-division} suppresses such checking for code
6928 generated to run on an MC88100 processor.
6930 GCC assumes that the MC88110 processor correctly detects all instances
6931 of integer division by zero. When @option{-m88110} is specified, no
6932 explicit checks for zero-valued divisors are generated, and both
6933 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6936 @item -muse-div-instruction
6937 @opindex muse-div-instruction
6938 @cindex divide instruction, 88k
6939 Use the div instruction for signed integer division on the
6940 MC88100 processor. By default, the div instruction is not used.
6942 On the MC88100 processor the signed integer division instruction
6943 div) traps to the operating system on a negative operand. The
6944 operating system transparently completes the operation, but at a
6945 large cost in execution time. By default, when compiling code
6946 that might be run on an MC88100 processor, GCC emulates signed
6947 integer division using the unsigned integer division instruction
6948 divu), thereby avoiding the large penalty of a trap to the
6949 operating system. Such emulation has its own, smaller, execution
6950 cost in both time and space. To the extent that your code's
6951 important signed integer division operations are performed on two
6952 nonnegative operands, it may be desirable to use the div
6953 instruction directly.
6955 On the MC88110 processor the div instruction (also known as the
6956 divs instruction) processes negative operands without trapping to
6957 the operating system. When @option{-m88110} is specified,
6958 @option{-muse-div-instruction} is ignored, and the div instruction is used
6959 for signed integer division.
6961 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6962 particular, the behavior of such a division with and without
6963 @option{-muse-div-instruction} may differ.
6965 @item -mtrap-large-shift
6966 @itemx -mhandle-large-shift
6967 @opindex mtrap-large-shift
6968 @opindex mhandle-large-shift
6969 @cindex bit shift overflow (88k)
6970 @cindex large bit shifts (88k)
6971 Include code to detect bit-shifts of more than 31 bits; respectively,
6972 trap such shifts or emit code to handle them properly. By default GCC
6973 makes no special provision for large bit shifts.
6975 @item -mwarn-passed-structs
6976 @opindex mwarn-passed-structs
6977 @cindex structure passing (88k)
6978 Warn when a function passes a struct as an argument or result.
6979 Structure-passing conventions have changed during the evolution of the C
6980 language, and are often the source of portability problems. By default,
6981 GCC issues no such warning.
6984 @c break page here to avoid unsightly interparagraph stretch.
6988 @node RS/6000 and PowerPC Options
6989 @subsection IBM RS/6000 and PowerPC Options
6990 @cindex RS/6000 and PowerPC Options
6991 @cindex IBM RS/6000 and PowerPC Options
6993 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
7001 @itemx -mpowerpc-gpopt
7002 @itemx -mno-powerpc-gpopt
7003 @itemx -mpowerpc-gfxopt
7004 @itemx -mno-powerpc-gfxopt
7006 @itemx -mno-powerpc64
7012 @opindex mno-powerpc
7013 @opindex mpowerpc-gpopt
7014 @opindex mno-powerpc-gpopt
7015 @opindex mpowerpc-gfxopt
7016 @opindex mno-powerpc-gfxopt
7018 @opindex mno-powerpc64
7019 GCC supports two related instruction set architectures for the
7020 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
7021 instructions supported by the @samp{rios} chip set used in the original
7022 RS/6000 systems and the @dfn{PowerPC} instruction set is the
7023 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
7024 the IBM 4xx microprocessors.
7026 Neither architecture is a subset of the other. However there is a
7027 large common subset of instructions supported by both. An MQ
7028 register is included in processors supporting the POWER architecture.
7030 You use these options to specify which instructions are available on the
7031 processor you are using. The default value of these options is
7032 determined when configuring GCC@. Specifying the
7033 @option{-mcpu=@var{cpu_type}} overrides the specification of these
7034 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
7035 rather than the options listed above.
7037 The @option{-mpower} option allows GCC to generate instructions that
7038 are found only in the POWER architecture and to use the MQ register.
7039 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
7040 to generate instructions that are present in the POWER2 architecture but
7041 not the original POWER architecture.
7043 The @option{-mpowerpc} option allows GCC to generate instructions that
7044 are found only in the 32-bit subset of the PowerPC architecture.
7045 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
7046 GCC to use the optional PowerPC architecture instructions in the
7047 General Purpose group, including floating-point square root. Specifying
7048 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
7049 use the optional PowerPC architecture instructions in the Graphics
7050 group, including floating-point select.
7052 The @option{-mpowerpc64} option allows GCC to generate the additional
7053 64-bit instructions that are found in the full PowerPC64 architecture
7054 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
7055 @option{-mno-powerpc64}.
7057 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7058 will use only the instructions in the common subset of both
7059 architectures plus some special AIX common-mode calls, and will not use
7060 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7061 permits GCC to use any instruction from either architecture and to
7062 allow use of the MQ register; specify this for the Motorola MPC601.
7064 @item -mnew-mnemonics
7065 @itemx -mold-mnemonics
7066 @opindex mnew-mnemonics
7067 @opindex mold-mnemonics
7068 Select which mnemonics to use in the generated assembler code. With
7069 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7070 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7071 assembler mnemonics defined for the POWER architecture. Instructions
7072 defined in only one architecture have only one mnemonic; GCC uses that
7073 mnemonic irrespective of which of these options is specified.
7075 GCC defaults to the mnemonics appropriate for the architecture in
7076 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7077 value of these option. Unless you are building a cross-compiler, you
7078 should normally not specify either @option{-mnew-mnemonics} or
7079 @option{-mold-mnemonics}, but should instead accept the default.
7081 @item -mcpu=@var{cpu_type}
7083 Set architecture type, register usage, choice of mnemonics, and
7084 instruction scheduling parameters for machine type @var{cpu_type}.
7085 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
7086 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
7087 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
7088 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
7089 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
7090 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
7092 @option{-mcpu=common} selects a completely generic processor. Code
7093 generated under this option will run on any POWER or PowerPC processor.
7094 GCC will use only the instructions in the common subset of both
7095 architectures, and will not use the MQ register. GCC assumes a generic
7096 processor model for scheduling purposes.
7098 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7099 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7100 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7101 types, with an appropriate, generic processor model assumed for
7102 scheduling purposes.
7104 The other options specify a specific processor. Code generated under
7105 those options will run best on that processor, and may not run at all on
7108 The @option{-mcpu} options automatically enable or disable other
7109 @option{-m} options as follows:
7113 @option{-mno-power}, @option{-mno-powerpc}
7120 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
7135 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7138 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7143 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
7146 @item -mtune=@var{cpu_type}
7148 Set the instruction scheduling parameters for machine type
7149 @var{cpu_type}, but do not set the architecture type, register usage, or
7150 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7151 values for @var{cpu_type} are used for @option{-mtune} as for
7152 @option{-mcpu}. If both are specified, the code generated will use the
7153 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7154 scheduling parameters set by @option{-mtune}.
7159 @opindex mno-altivec
7160 These switches enable or disable the use of built-in functions that
7161 allow access to the AltiVec instruction set. You may also need to set
7162 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7167 Extend the current ABI with SPE ABI extensions. This does not change
7168 the default ABI, instead it adds the SPE ABI extensions to the current
7172 @opindex mabi=no-spe
7173 Disable Booke SPE ABI extensions for the current ABI.
7175 @item -misel=@var{yes/no}
7178 This switch enables or disables the generation of ISEL instructions.
7180 @item -mspe=@var{yes/no}
7183 This switch enables or disables the generation of SPE simd
7186 @item -mfloat-gprs=@var{yes/no}
7188 @opindex mfloat-gprs
7189 This switch enables or disables the generation of floating point
7190 operations on the general purpose registers for architectures that
7191 support it. This option is currently only available on the MPC8540.
7194 @itemx -mno-fp-in-toc
7195 @itemx -mno-sum-in-toc
7196 @itemx -mminimal-toc
7198 @opindex mno-fp-in-toc
7199 @opindex mno-sum-in-toc
7200 @opindex mminimal-toc
7201 Modify generation of the TOC (Table Of Contents), which is created for
7202 every executable file. The @option{-mfull-toc} option is selected by
7203 default. In that case, GCC will allocate at least one TOC entry for
7204 each unique non-automatic variable reference in your program. GCC
7205 will also place floating-point constants in the TOC@. However, only
7206 16,384 entries are available in the TOC@.
7208 If you receive a linker error message that saying you have overflowed
7209 the available TOC space, you can reduce the amount of TOC space used
7210 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7211 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7212 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7213 generate code to calculate the sum of an address and a constant at
7214 run-time instead of putting that sum into the TOC@. You may specify one
7215 or both of these options. Each causes GCC to produce very slightly
7216 slower and larger code at the expense of conserving TOC space.
7218 If you still run out of space in the TOC even when you specify both of
7219 these options, specify @option{-mminimal-toc} instead. This option causes
7220 GCC to make only one TOC entry for every file. When you specify this
7221 option, GCC will produce code that is slower and larger but which
7222 uses extremely little TOC space. You may wish to use this option
7223 only on files that contain less frequently executed code.
7229 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7230 @code{long} type, and the infrastructure needed to support them.
7231 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7232 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7233 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7238 @opindex mno-xl-call
7239 On AIX, pass floating-point arguments to prototyped functions beyond the
7240 register save area (RSA) on the stack in addition to argument FPRs. The
7241 AIX calling convention was extended but not initially documented to
7242 handle an obscure K&R C case of calling a function that takes the
7243 address of its arguments with fewer arguments than declared. AIX XL
7244 compilers access floating point arguments which do not fit in the
7245 RSA from the stack when a subroutine is compiled without
7246 optimization. Because always storing floating-point arguments on the
7247 stack is inefficient and rarely needed, this option is not enabled by
7248 default and only is necessary when calling subroutines compiled by AIX
7249 XL compilers without optimization.
7253 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7254 application written to use message passing with special startup code to
7255 enable the application to run. The system must have PE installed in the
7256 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7257 must be overridden with the @option{-specs=} option to specify the
7258 appropriate directory location. The Parallel Environment does not
7259 support threads, so the @option{-mpe} option and the @option{-pthread}
7260 option are incompatible.
7262 @item -malign-natural
7263 @itemx -malign-power
7264 @opindex malign-natural
7265 @opindex malign-power
7266 On AIX, Darwin, and 64-bit PowerPC Linux, the option
7267 @option{-malign-natural} overrides the ABI-defined alignment of larger
7268 types, such as floating-point doubles, on their natural size-based boundary.
7269 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7270 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7274 @opindex msoft-float
7275 @opindex mhard-float
7276 Generate code that does not use (uses) the floating-point register set.
7277 Software floating point emulation is provided if you use the
7278 @option{-msoft-float} option, and pass the option to GCC when linking.
7281 @itemx -mno-multiple
7283 @opindex mno-multiple
7284 Generate code that uses (does not use) the load multiple word
7285 instructions and the store multiple word instructions. These
7286 instructions are generated by default on POWER systems, and not
7287 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7288 endian PowerPC systems, since those instructions do not work when the
7289 processor is in little endian mode. The exceptions are PPC740 and
7290 PPC750 which permit the instructions usage in little endian mode.
7296 Generate code that uses (does not use) the load string instructions
7297 and the store string word instructions to save multiple registers and
7298 do small block moves. These instructions are generated by default on
7299 POWER systems, and not generated on PowerPC systems. Do not use
7300 @option{-mstring} on little endian PowerPC systems, since those
7301 instructions do not work when the processor is in little endian mode.
7302 The exceptions are PPC740 and PPC750 which permit the instructions
7303 usage in little endian mode.
7309 Generate code that uses (does not use) the load or store instructions
7310 that update the base register to the address of the calculated memory
7311 location. These instructions are generated by default. If you use
7312 @option{-mno-update}, there is a small window between the time that the
7313 stack pointer is updated and the address of the previous frame is
7314 stored, which means code that walks the stack frame across interrupts or
7315 signals may get corrupted data.
7318 @itemx -mno-fused-madd
7319 @opindex mfused-madd
7320 @opindex mno-fused-madd
7321 Generate code that uses (does not use) the floating point multiply and
7322 accumulate instructions. These instructions are generated by default if
7323 hardware floating is used.
7325 @item -mno-bit-align
7327 @opindex mno-bit-align
7329 On System V.4 and embedded PowerPC systems do not (do) force structures
7330 and unions that contain bit-fields to be aligned to the base type of the
7333 For example, by default a structure containing nothing but 8
7334 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7335 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7336 the structure would be aligned to a 1 byte boundary and be one byte in
7339 @item -mno-strict-align
7340 @itemx -mstrict-align
7341 @opindex mno-strict-align
7342 @opindex mstrict-align
7343 On System V.4 and embedded PowerPC systems do not (do) assume that
7344 unaligned memory references will be handled by the system.
7347 @itemx -mno-relocatable
7348 @opindex mrelocatable
7349 @opindex mno-relocatable
7350 On embedded PowerPC systems generate code that allows (does not allow)
7351 the program to be relocated to a different address at runtime. If you
7352 use @option{-mrelocatable} on any module, all objects linked together must
7353 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7355 @item -mrelocatable-lib
7356 @itemx -mno-relocatable-lib
7357 @opindex mrelocatable-lib
7358 @opindex mno-relocatable-lib
7359 On embedded PowerPC systems generate code that allows (does not allow)
7360 the program to be relocated to a different address at runtime. Modules
7361 compiled with @option{-mrelocatable-lib} can be linked with either modules
7362 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7363 with modules compiled with the @option{-mrelocatable} options.
7369 On System V.4 and embedded PowerPC systems do not (do) assume that
7370 register 2 contains a pointer to a global area pointing to the addresses
7371 used in the program.
7374 @itemx -mlittle-endian
7376 @opindex mlittle-endian
7377 On System V.4 and embedded PowerPC systems compile code for the
7378 processor in little endian mode. The @option{-mlittle-endian} option is
7379 the same as @option{-mlittle}.
7384 @opindex mbig-endian
7385 On System V.4 and embedded PowerPC systems compile code for the
7386 processor in big endian mode. The @option{-mbig-endian} option is
7387 the same as @option{-mbig}.
7389 @item -mdynamic-no-pic
7390 @opindex mdynamic-no-pic
7391 On Darwin and Mac OS X systems, compile code so that it is not
7392 relocatable, but that its external references are relocatable. The
7393 resulting code is suitable for applications, but not shared
7398 On System V.4 and embedded PowerPC systems compile code using calling
7399 conventions that adheres to the March 1995 draft of the System V
7400 Application Binary Interface, PowerPC processor supplement. This is the
7401 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7403 @item -mcall-sysv-eabi
7404 @opindex mcall-sysv-eabi
7405 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7407 @item -mcall-sysv-noeabi
7408 @opindex mcall-sysv-noeabi
7409 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7411 @item -mcall-solaris
7412 @opindex mcall-solaris
7413 On System V.4 and embedded PowerPC systems compile code for the Solaris
7417 @opindex mcall-linux
7418 On System V.4 and embedded PowerPC systems compile code for the
7419 Linux-based GNU system.
7423 On System V.4 and embedded PowerPC systems compile code for the
7424 Hurd-based GNU system.
7427 @opindex mcall-netbsd
7428 On System V.4 and embedded PowerPC systems compile code for the
7429 NetBSD operating system.
7431 @item -maix-struct-return
7432 @opindex maix-struct-return
7433 Return all structures in memory (as specified by the AIX ABI)@.
7435 @item -msvr4-struct-return
7436 @opindex msvr4-struct-return
7437 Return structures smaller than 8 bytes in registers (as specified by the
7441 @opindex mabi=altivec
7442 Extend the current ABI with AltiVec ABI extensions. This does not
7443 change the default ABI, instead it adds the AltiVec ABI extensions to
7446 @item -mabi=no-altivec
7447 @opindex mabi=no-altivec
7448 Disable AltiVec ABI extensions for the current ABI.
7451 @itemx -mno-prototype
7453 @opindex mno-prototype
7454 On System V.4 and embedded PowerPC systems assume that all calls to
7455 variable argument functions are properly prototyped. Otherwise, the
7456 compiler must insert an instruction before every non prototyped call to
7457 set or clear bit 6 of the condition code register (@var{CR}) to
7458 indicate whether floating point values were passed in the floating point
7459 registers in case the function takes a variable arguments. With
7460 @option{-mprototype}, only calls to prototyped variable argument functions
7461 will set or clear the bit.
7465 On embedded PowerPC systems, assume that the startup module is called
7466 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7467 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7472 On embedded PowerPC systems, assume that the startup module is called
7473 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7478 On embedded PowerPC systems, assume that the startup module is called
7479 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7483 @opindex myellowknife
7484 On embedded PowerPC systems, assume that the startup module is called
7485 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7490 On System V.4 and embedded PowerPC systems, specify that you are
7491 compiling for a VxWorks system.
7495 Specify that you are compiling for the WindISS simulation environment.
7499 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7500 header to indicate that @samp{eabi} extended relocations are used.
7506 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7507 Embedded Applications Binary Interface (eabi) which is a set of
7508 modifications to the System V.4 specifications. Selecting @option{-meabi}
7509 means that the stack is aligned to an 8 byte boundary, a function
7510 @code{__eabi} is called to from @code{main} to set up the eabi
7511 environment, and the @option{-msdata} option can use both @code{r2} and
7512 @code{r13} to point to two separate small data areas. Selecting
7513 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7514 do not call an initialization function from @code{main}, and the
7515 @option{-msdata} option will only use @code{r13} to point to a single
7516 small data area. The @option{-meabi} option is on by default if you
7517 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7520 @opindex msdata=eabi
7521 On System V.4 and embedded PowerPC systems, put small initialized
7522 @code{const} global and static data in the @samp{.sdata2} section, which
7523 is pointed to by register @code{r2}. Put small initialized
7524 non-@code{const} global and static data in the @samp{.sdata} section,
7525 which is pointed to by register @code{r13}. Put small uninitialized
7526 global and static data in the @samp{.sbss} section, which is adjacent to
7527 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7528 incompatible with the @option{-mrelocatable} option. The
7529 @option{-msdata=eabi} option also sets the @option{-memb} option.
7532 @opindex msdata=sysv
7533 On System V.4 and embedded PowerPC systems, put small global and static
7534 data in the @samp{.sdata} section, which is pointed to by register
7535 @code{r13}. Put small uninitialized global and static data in the
7536 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7537 The @option{-msdata=sysv} option is incompatible with the
7538 @option{-mrelocatable} option.
7540 @item -msdata=default
7542 @opindex msdata=default
7544 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7545 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7546 same as @option{-msdata=sysv}.
7549 @opindex msdata-data
7550 On System V.4 and embedded PowerPC systems, put small global and static
7551 data in the @samp{.sdata} section. Put small uninitialized global and
7552 static data in the @samp{.sbss} section. Do not use register @code{r13}
7553 to address small data however. This is the default behavior unless
7554 other @option{-msdata} options are used.
7558 @opindex msdata=none
7560 On embedded PowerPC systems, put all initialized global and static data
7561 in the @samp{.data} section, and all uninitialized data in the
7562 @samp{.bss} section.
7566 @cindex smaller data references (PowerPC)
7567 @cindex .sdata/.sdata2 references (PowerPC)
7568 On embedded PowerPC systems, put global and static items less than or
7569 equal to @var{num} bytes into the small data or bss sections instead of
7570 the normal data or bss section. By default, @var{num} is 8. The
7571 @option{-G @var{num}} switch is also passed to the linker.
7572 All modules should be compiled with the same @option{-G @var{num}} value.
7575 @itemx -mno-regnames
7577 @opindex mno-regnames
7578 On System V.4 and embedded PowerPC systems do (do not) emit register
7579 names in the assembly language output using symbolic forms.
7582 @itemx -mno-longcall
7584 @opindex mno-longcall
7585 Default to making all function calls via pointers, so that functions
7586 which reside further than 64 megabytes (67,108,864 bytes) from the
7587 current location can be called. This setting can be overridden by the
7588 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7590 Some linkers are capable of detecting out-of-range calls and generating
7591 glue code on the fly. On these systems, long calls are unnecessary and
7592 generate slower code. As of this writing, the AIX linker can do this,
7593 as can the GNU linker for PowerPC/64. It is planned to add this feature
7594 to the GNU linker for 32-bit PowerPC systems as well.
7596 In the future, we may cause GCC to ignore all longcall specifications
7597 when the linker is known to generate glue.
7601 Adds support for multithreading with the @dfn{pthreads} library.
7602 This option sets flags for both the preprocessor and linker.
7606 @node Darwin Options
7607 @subsection Darwin Options
7608 @cindex Darwin options
7610 These options are defined for all architectures running the Darwin operating
7611 system. They are useful for compatibility with other Mac OS compilers.
7616 Loads all members of static archive libraries.
7617 See man ld(1) for more information.
7619 @item -arch_errors_fatal
7620 @opindex arch_errors_fatal
7621 Cause the errors having to do with files that have the wrong architecture
7625 @opindex bind_at_load
7626 Causes the output file to be marked such that the dynamic linker will
7627 bind all undefined references when the file is loaded or launched.
7631 Produce a Mach-o bundle format file.
7632 See man ld(1) for more information.
7634 @item -bundle_loader @var{executable}
7635 @opindex bundle_loader
7636 This specifies the @var{executable} that will be loading the build
7637 output file being linked. See man ld(1) for more information.
7639 @item -allowable_client @var{client_name}
7643 @item -compatibility_version
7644 @item -current_version
7645 @item -dependency-file
7647 @item -dylinker_install_name
7650 @item -exported_symbols_list
7652 @item -flat_namespace
7653 @item -force_cpusubtype_ALL
7654 @item -force_flat_namespace
7655 @item -headerpad_max_install_names
7659 @item -keep_private_externs
7661 @item -multiply_defined
7662 @item -multiply_defined_unused
7664 @item -nofixprebinding
7667 @item -noseglinkedit
7668 @item -pagezero_size
7670 @item -prebind_all_twolevel_modules
7671 @item -private_bundle
7672 @item -read_only_relocs
7674 @item -sectobjectsymbols
7678 @item -sectobjectsymbols
7680 @item -seg_addr_table
7681 @item -seg_addr_table_filename
7684 @item -segs_read_only_addr
7685 @item -segs_read_write_addr
7686 @item -single_module
7690 @item -twolevel_namespace
7693 @item -unexported_symbols_list
7694 @item -weak_reference_mismatches
7697 @opindex allowable_client
7699 @opindex client_name
7700 @opindex compatibility_version
7701 @opindex current_version
7702 @opindex dependency-file
7704 @opindex dylinker_install_name
7707 @opindex exported_symbols_list
7709 @opindex flat_namespace
7710 @opindex force_cpusubtype_ALL
7711 @opindex force_flat_namespace
7712 @opindex headerpad_max_install_names
7715 @opindex install_name
7716 @opindex keep_private_externs
7717 @opindex multi_module
7718 @opindex multiply_defined
7719 @opindex multiply_defined_unused
7721 @opindex nofixprebinding
7722 @opindex nomultidefs
7724 @opindex noseglinkedit
7725 @opindex pagezero_size
7727 @opindex prebind_all_twolevel_modules
7728 @opindex private_bundle
7729 @opindex read_only_relocs
7731 @opindex sectobjectsymbols
7735 @opindex sectobjectsymbols
7737 @opindex seg_addr_table
7738 @opindex seg_addr_table_filename
7739 @opindex seglinkedit
7741 @opindex segs_read_only_addr
7742 @opindex segs_read_write_addr
7743 @opindex single_module
7745 @opindex sub_library
7746 @opindex sub_umbrella
7747 @opindex twolevel_namespace
7750 @opindex unexported_symbols_list
7751 @opindex weak_reference_mismatches
7752 @opindex whatsloaded
7754 This options are available for Darwin linker. Darwin linker man page
7755 describes them in detail.
7760 @subsection IBM RT Options
7762 @cindex IBM RT options
7764 These @samp{-m} options are defined for the IBM RT PC:
7768 @opindex min-line-mul
7769 Use an in-line code sequence for integer multiplies. This is the
7772 @item -mcall-lib-mul
7773 @opindex mcall-lib-mul
7774 Call @code{lmul$$} for integer multiples.
7776 @item -mfull-fp-blocks
7777 @opindex mfull-fp-blocks
7778 Generate full-size floating point data blocks, including the minimum
7779 amount of scratch space recommended by IBM@. This is the default.
7781 @item -mminimum-fp-blocks
7782 @opindex mminimum-fp-blocks
7783 Do not include extra scratch space in floating point data blocks. This
7784 results in smaller code, but slower execution, since scratch space must
7785 be allocated dynamically.
7787 @cindex @file{stdarg.h} and RT PC
7788 @item -mfp-arg-in-fpregs
7789 @opindex mfp-arg-in-fpregs
7790 Use a calling sequence incompatible with the IBM calling convention in
7791 which floating point arguments are passed in floating point registers.
7792 Note that @code{stdarg.h} will not work with floating point operands
7793 if this option is specified.
7795 @item -mfp-arg-in-gregs
7796 @opindex mfp-arg-in-gregs
7797 Use the normal calling convention for floating point arguments. This is
7800 @item -mhc-struct-return
7801 @opindex mhc-struct-return
7802 Return structures of more than one word in memory, rather than in a
7803 register. This provides compatibility with the MetaWare HighC (hc)
7804 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7805 with the Portable C Compiler (pcc).
7807 @item -mnohc-struct-return
7808 @opindex mnohc-struct-return
7809 Return some structures of more than one word in registers, when
7810 convenient. This is the default. For compatibility with the
7811 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7812 option @option{-mhc-struct-return}.
7816 @subsection MIPS Options
7817 @cindex MIPS options
7819 These @samp{-m} options are defined for the MIPS family of computers:
7823 @item -march=@var{arch}
7825 Generate code that will run on @var{arch}, which can be the name of a
7826 generic MIPS ISA, or the name of a particular processor.
7828 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7829 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7830 The processor names are:
7831 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7833 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7834 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
7838 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4300},
7839 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
7840 The special value @samp{from-abi} selects the
7841 most compatible architecture for the selected ABI (that is,
7842 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7844 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7845 (for example, @samp{-march=r2k}). Prefixes are optional, and
7846 @samp{vr} may be written @samp{r}.
7848 GCC defines two macros based on the value of this option. The first
7849 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7850 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7851 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7852 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7853 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7855 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7856 above. In other words, it will have the full prefix and will not
7857 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7858 the macro names the resolved architecture (either @samp{"mips1"} or
7859 @samp{"mips3"}). It names the default architecture when no
7860 @option{-march} option is given.
7862 @item -mtune=@var{arch}
7864 Optimize for @var{arch}. Among other things, this option controls
7865 the way instructions are scheduled, and the perceived cost of arithmetic
7866 operations. The list of @var{arch} values is the same as for
7869 When this option is not used, GCC will optimize for the processor
7870 specified by @option{-march}. By using @option{-march} and
7871 @option{-mtune} together, it is possible to generate code that will
7872 run on a family of processors, but optimize the code for one
7873 particular member of that family.
7875 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7876 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7877 @samp{-march} ones described above.
7881 Equivalent to @samp{-march=mips1}.
7885 Equivalent to @samp{-march=mips2}.
7889 Equivalent to @samp{-march=mips3}.
7893 Equivalent to @samp{-march=mips4}.
7897 Equivalent to @samp{-march=mips32}.
7901 Equivalent to @samp{-march=mips32r2}.
7905 Equivalent to @samp{-march=mips64}.
7908 @itemx -mno-fused-madd
7909 @opindex mfused-madd
7910 @opindex mno-fused-madd
7911 Generate code that uses (does not use) the floating point multiply and
7912 accumulate instructions, when they are available. These instructions
7913 are generated by default if they are available, but this may be
7914 undesirable if the extra precision causes problems or on certain chips
7915 in the mode where denormals are rounded to zero where denormals
7916 generated by multiply and accumulate instructions cause exceptions
7921 Assume that floating point registers are 32 bits wide.
7925 Assume that floating point registers are 64 bits wide.
7929 Assume that general purpose registers are 32 bits wide.
7933 Assume that general purpose registers are 64 bits wide.
7937 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7938 explanation of the default, and the width of pointers.
7942 Force 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, int, and pointer types to be 32 bits wide.
7949 The default size of ints, longs and pointers depends on the ABI@. All
7950 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
7951 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
7952 are the same size as longs, or the same size as integer registers,
7953 whichever is smaller.
7967 Generate code for the given ABI@.
7969 Note that there are two embedded ABIs: @option{-mabi=eabi}
7970 selects the one defined by Cygnus while @option{-meabi=meabi}
7971 selects the one defined by MIPS@. Both these ABIs have
7972 32-bit and 64-bit variants. Normally, GCC will generate
7973 64-bit code when you select a 64-bit architecture, but you
7974 can use @option{-mgp32} to get 32-bit code instead.
7976 @item -mabi-fake-default
7977 @opindex mabi-fake-default
7978 You don't want to know what this option does. No, really. I mean
7979 it. Move on to the next option.
7981 What? You're still here? Oh, well@enddots{} Ok, here's the deal. GCC
7982 wants the default set of options to get the root of the multilib tree,
7983 and the shared library SONAMEs without any multilib-indicating
7984 suffixes. This is not convenience for @samp{mips64-linux-gnu}, since
7985 we want to default to the N32 ABI, while still being binary-compatible
7986 with @samp{mips-linux-gnu} if you stick to the O32 ABI@. Being
7987 binary-compatible means shared libraries should have the same SONAMEs,
7988 and libraries should live in the same location. Having O32 libraries
7989 in a sub-directory named say @file{o32} is not acceptable.
7991 So we trick GCC into believing that O32 is the default ABI, except
7992 that we override the default with some internal command-line
7993 processing magic. Problem is, if we stopped at that, and you then
7994 created a multilib-aware package that used the output of @command{gcc
7995 -print-multi-lib} to decide which multilibs to build, and how, and
7996 you'd find yourself in an awkward situation when you found out that
7997 some of the options listed ended up mapping to the same multilib, and
7998 none of your libraries was actually built for the multilib that
7999 @option{-print-multi-lib} claims to be the default. So we added this
8000 option that disables the default switcher, falling back to GCC's
8001 original notion of the default library. Confused yet?
8003 For short: don't ever use this option, unless you find it in the list
8004 of additional options to be used when building for multilibs, in the
8005 output of @option{gcc -print-multi-lib}.
8009 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
8010 add normal debug information. This is the default for all
8011 platforms except for the OSF/1 reference platform, using the OSF/rose
8012 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
8013 switches are used, the @file{mips-tfile} program will encapsulate the
8014 stabs within MIPS ECOFF@.
8018 Generate code for the GNU assembler. This is the default on the OSF/1
8019 reference platform, using the OSF/rose object format. Also, this is
8020 the default if the configure option @option{--with-gnu-as} is used.
8022 @item -msplit-addresses
8023 @itemx -mno-split-addresses
8024 @opindex msplit-addresses
8025 @opindex mno-split-addresses
8026 Generate code to load the high and low parts of address constants separately.
8027 This allows GCC to optimize away redundant loads of the high order
8028 bits of addresses. This optimization requires GNU as and GNU ld.
8029 This optimization is enabled by default for some embedded targets where
8030 GNU as and GNU ld are standard.
8036 The @option{-mrnames} switch says to output code using the MIPS software
8037 names for the registers, instead of the hardware names (ie, @var{a0}
8038 instead of @var{$4}). The only known assembler that supports this option
8039 is the Algorithmics assembler.
8045 The @option{-mmemcpy} switch makes all block moves call the appropriate
8046 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
8047 generating inline code.
8050 @itemx -mno-mips-tfile
8051 @opindex mmips-tfile
8052 @opindex mno-mips-tfile
8053 The @option{-mno-mips-tfile} switch causes the compiler not
8054 postprocess the object file with the @file{mips-tfile} program,
8055 after the MIPS assembler has generated it to add debug support. If
8056 @file{mips-tfile} is not run, then no local variables will be
8057 available to the debugger. In addition, @file{stage2} and
8058 @file{stage3} objects will have the temporary file names passed to the
8059 assembler embedded in the object file, which means the objects will
8060 not compare the same. The @option{-mno-mips-tfile} switch should only
8061 be used when there are bugs in the @file{mips-tfile} program that
8062 prevents compilation.
8065 @opindex msoft-float
8066 Generate output containing library calls for floating point.
8067 @strong{Warning:} the requisite libraries are not part of GCC@.
8068 Normally the facilities of the machine's usual C compiler are used, but
8069 this can't be done directly in cross-compilation. You must make your
8070 own arrangements to provide suitable library functions for
8074 @opindex mhard-float
8075 Generate output containing floating point instructions. This is the
8076 default if you use the unmodified sources.
8079 @itemx -mno-abicalls
8081 @opindex mno-abicalls
8082 Emit (or do not emit) the pseudo operations @samp{.abicalls},
8083 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
8084 position independent code.
8090 Lift (or do not lift) the usual restrictions on the size of the global
8093 GCC normally uses a single instruction to load values from the GOT.
8094 While this is relatively efficient, it will only work if the GOT
8095 is smaller than about 64k. Anything larger will cause the linker
8096 to report an error such as:
8098 @cindex relocation truncated to fit (MIPS)
8100 relocation truncated to fit: R_MIPS_GOT16 foobar
8103 If this happens, you should recompile your code with @option{-mxgot}.
8104 It should then work with very large GOTs, although it will also be
8105 less efficient, since it will take three instructions to fetch the
8106 value of a global symbol.
8108 Note that some linkers can create multiple GOTs. If you have such a
8109 linker, you should only need to use @option{-mxgot} when a single object
8110 file accesses more than 64k's worth of GOT entries. Very few do.
8112 These options have no effect unless GCC is generating position
8116 @itemx -mno-long-calls
8117 @opindex mlong-calls
8118 @opindex mno-long-calls
8119 Do all calls with the @samp{JALR} instruction, which requires
8120 loading up a function's address into a register before the call.
8121 You need to use this switch, if you call outside of the current
8122 512 megabyte segment to functions that are not through pointers.
8124 @item -membedded-pic
8125 @itemx -mno-embedded-pic
8126 @opindex membedded-pic
8127 @opindex mno-embedded-pic
8128 Generate PIC code suitable for some embedded systems. All calls are
8129 made using PC relative address, and all data is addressed using the $gp
8130 register. No more than 65536 bytes of global data may be used. This
8131 requires GNU as and GNU ld which do most of the work. This currently
8132 only works on targets which use ECOFF; it does not work with ELF@.
8134 @item -membedded-data
8135 @itemx -mno-embedded-data
8136 @opindex membedded-data
8137 @opindex mno-embedded-data
8138 Allocate variables to the read-only data section first if possible, then
8139 next in the small data section if possible, otherwise in data. This gives
8140 slightly slower code than the default, but reduces the amount of RAM required
8141 when executing, and thus may be preferred for some embedded systems.
8143 @item -muninit-const-in-rodata
8144 @itemx -mno-uninit-const-in-rodata
8145 @opindex muninit-const-in-rodata
8146 @opindex mno-uninit-const-in-rodata
8147 When used together with @option{-membedded-data}, it will always store uninitialized
8148 const variables in the read-only data section.
8150 @item -msingle-float
8151 @itemx -mdouble-float
8152 @opindex msingle-float
8153 @opindex mdouble-float
8154 The @option{-msingle-float} switch tells gcc to assume that the floating
8155 point coprocessor only supports single precision operations, as on the
8156 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
8157 double precision operations. This is the default.
8163 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
8164 as on the @samp{r4650} chip.
8168 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
8169 @option{-mcpu=r4650}.
8175 Enable 16-bit instructions.
8179 Use the entry and exit pseudo ops. This option can only be used with
8184 Compile code for the processor in little endian mode.
8185 The requisite libraries are assumed to exist.
8189 Compile code for the processor in big endian mode.
8190 The requisite libraries are assumed to exist.
8194 @cindex smaller data references (MIPS)
8195 @cindex gp-relative references (MIPS)
8196 Put global and static items less than or equal to @var{num} bytes into
8197 the small data or bss sections instead of the normal data or bss
8198 section. This allows the assembler to emit one word memory reference
8199 instructions based on the global pointer (@var{gp} or @var{$28}),
8200 instead of the normal two words used. By default, @var{num} is 8 when
8201 the MIPS assembler is used, and 0 when the GNU assembler is used. The
8202 @option{-G @var{num}} switch is also passed to the assembler and linker.
8203 All modules should be compiled with the same @option{-G @var{num}}
8208 Tell the MIPS assembler to not run its preprocessor over user
8209 assembler files (with a @samp{.s} suffix) when assembling them.
8213 Pass an option to gas which will cause nops to be inserted if
8214 the read of the destination register of an mfhi or mflo instruction
8215 occurs in the following two instructions.
8220 Work around certain SB-1 CPU core errata.
8221 (This flag currently works around the SB-1 revision 2
8222 ``F1'' and ``F2'' floating point errata.)
8226 Do not include the default crt0.
8228 @item -mflush-func=@var{func}
8229 @itemx -mno-flush-func
8230 @opindex mflush-func
8231 Specifies the function to call to flush the I and D caches, or to not
8232 call any such function. If called, the function must take the same
8233 arguments as the common @code{_flush_func()}, that is, the address of the
8234 memory range for which the cache is being flushed, the size of the
8235 memory range, and the number 3 (to flush both caches). The default
8236 depends on the target gcc was configured for, but commonly is either
8237 @samp{_flush_func} or @samp{__cpu_flush}.
8239 @item -mbranch-likely
8240 @itemx -mno-branch-likely
8241 @opindex mbranch-likely
8242 @opindex mno-branch-likely
8243 Enable or disable use of Branch Likely instructions, regardless of the
8244 default for the selected architecture. By default, Branch Likely
8245 instructions may be generated if they are supported by the selected
8246 architecture. An exception is for the MIPS32 and MIPS64 architectures
8247 and processors which implement those architectures; for those, Branch
8248 Likely instructions will not be generated by default because the MIPS32
8249 and MIPS64 architectures specifically deprecate their use.
8252 @node i386 and x86-64 Options
8253 @subsection Intel 386 and AMD x86-64 Options
8254 @cindex i386 Options
8255 @cindex x86-64 Options
8256 @cindex Intel 386 Options
8257 @cindex AMD x86-64 Options
8259 These @samp{-m} options are defined for the i386 and x86-64 family of
8263 @item -mtune=@var{cpu-type}
8265 Tune to @var{cpu-type} everything applicable about the generated code, except
8266 for the ABI and the set of available instructions. The choices for
8267 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
8268 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
8269 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
8270 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
8271 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8}, @samp{c3}
8274 While picking a specific @var{cpu-type} will schedule things appropriately
8275 for that particular chip, the compiler will not generate any code that
8276 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8277 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
8278 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
8279 AMD chips as opposed to the Intel ones.
8281 @item -march=@var{cpu-type}
8283 Generate instructions for the machine type @var{cpu-type}. The choices
8284 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8285 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8287 @item -mcpu=@var{cpu-type}
8289 A deprecated synonym for @option{-mtune}.
8298 @opindex mpentiumpro
8299 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8300 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8301 These synonyms are deprecated.
8303 @item -mfpmath=@var{unit}
8305 generate floating point arithmetics for selected unit @var{unit}. the choices
8310 Use the standard 387 floating point coprocessor present majority of chips and
8311 emulated otherwise. Code compiled with this option will run almost everywhere.
8312 The temporary results are computed in 80bit precision instead of precision
8313 specified by the type resulting in slightly different results compared to most
8314 of other chips. See @option{-ffloat-store} for more detailed description.
8316 This is the default choice for i386 compiler.
8319 Use scalar floating point instructions present in the SSE instruction set.
8320 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8321 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8322 instruction set supports only single precision arithmetics, thus the double and
8323 extended precision arithmetics is still done using 387. Later version, present
8324 only in Pentium4 and the future AMD x86-64 chips supports double precision
8327 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8328 @option{-msse2} switches to enable SSE extensions and make this option
8329 effective. For x86-64 compiler, these extensions are enabled by default.
8331 The resulting code should be considerably faster in majority of cases and avoid
8332 the numerical instability problems of 387 code, but may break some existing
8333 code that expects temporaries to be 80bit.
8335 This is the default choice for x86-64 compiler.
8338 Use all SSE extensions enabled by @option{-msse2} as well as the new
8339 SSE extensions in Prescott New Instructions. @option{-mpni} also
8340 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8341 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8345 Attempt to utilize both instruction sets at once. This effectively double the
8346 amount of available registers and on chips with separate execution units for
8347 387 and SSE the execution resources too. Use this option with care, as it is
8348 still experimental, because gcc register allocator does not model separate
8349 functional units well resulting in instable performance.
8352 @item -masm=@var{dialect}
8353 @opindex masm=@var{dialect}
8354 Output asm instructions using selected @var{dialect}. Supported choices are
8355 @samp{intel} or @samp{att} (the default one).
8360 @opindex mno-ieee-fp
8361 Control whether or not the compiler uses IEEE floating point
8362 comparisons. These handle correctly the case where the result of a
8363 comparison is unordered.
8366 @opindex msoft-float
8367 Generate output containing library calls for floating point.
8368 @strong{Warning:} the requisite libraries are not part of GCC@.
8369 Normally the facilities of the machine's usual C compiler are used, but
8370 this can't be done directly in cross-compilation. You must make your
8371 own arrangements to provide suitable library functions for
8374 On machines where a function returns floating point results in the 80387
8375 register stack, some floating point opcodes may be emitted even if
8376 @option{-msoft-float} is used.
8378 @item -mno-fp-ret-in-387
8379 @opindex mno-fp-ret-in-387
8380 Do not use the FPU registers for return values of functions.
8382 The usual calling convention has functions return values of types
8383 @code{float} and @code{double} in an FPU register, even if there
8384 is no FPU@. The idea is that the operating system should emulate
8387 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8388 in ordinary CPU registers instead.
8390 @item -mno-fancy-math-387
8391 @opindex mno-fancy-math-387
8392 Some 387 emulators do not support the @code{sin}, @code{cos} and
8393 @code{sqrt} instructions for the 387. Specify this option to avoid
8394 generating those instructions. This option is the default on FreeBSD,
8395 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8396 indicates that the target cpu will always have an FPU and so the
8397 instruction will not need emulation. As of revision 2.6.1, these
8398 instructions are not generated unless you also use the
8399 @option{-funsafe-math-optimizations} switch.
8401 @item -malign-double
8402 @itemx -mno-align-double
8403 @opindex malign-double
8404 @opindex mno-align-double
8405 Control whether GCC aligns @code{double}, @code{long double}, and
8406 @code{long long} variables on a two word boundary or a one word
8407 boundary. Aligning @code{double} variables on a two word boundary will
8408 produce code that runs somewhat faster on a @samp{Pentium} at the
8409 expense of more memory.
8411 @strong{Warning:} if you use the @option{-malign-double} switch,
8412 structures containing the above types will be aligned differently than
8413 the published application binary interface specifications for the 386
8414 and will not be binary compatible with structures in code compiled
8415 without that switch.
8417 @item -m96bit-long-double
8418 @item -m128bit-long-double
8419 @opindex m96bit-long-double
8420 @opindex m128bit-long-double
8421 These switches control the size of @code{long double} type. The i386
8422 application binary interface specifies the size to be 96 bits,
8423 so @option{-m96bit-long-double} is the default in 32 bit mode.
8425 Modern architectures (Pentium and newer) would prefer @code{long double}
8426 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8427 conforming to the ABI, this would not be possible. So specifying a
8428 @option{-m128bit-long-double} will align @code{long double}
8429 to a 16 byte boundary by padding the @code{long double} with an additional
8432 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8433 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8435 Notice that neither of these options enable any extra precision over the x87
8436 standard of 80 bits for a @code{long double}.
8438 @strong{Warning:} if you override the default value for your target ABI, the
8439 structures and arrays containing @code{long double} will change their size as
8440 well as function calling convention for function taking @code{long double}
8441 will be modified. Hence they will not be binary compatible with arrays or
8442 structures in code compiled without that switch.
8446 @itemx -mno-svr3-shlib
8447 @opindex msvr3-shlib
8448 @opindex mno-svr3-shlib
8449 Control whether GCC places uninitialized local variables into the
8450 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8451 into @code{bss}. These options are meaningful only on System V Release 3.
8455 Use a different function-calling convention, in which functions that
8456 take a fixed number of arguments return with the @code{ret} @var{num}
8457 instruction, which pops their arguments while returning. This saves one
8458 instruction in the caller since there is no need to pop the arguments
8461 You can specify that an individual function is called with this calling
8462 sequence with the function attribute @samp{stdcall}. You can also
8463 override the @option{-mrtd} option by using the function attribute
8464 @samp{cdecl}. @xref{Function Attributes}.
8466 @strong{Warning:} this calling convention is incompatible with the one
8467 normally used on Unix, so you cannot use it if you need to call
8468 libraries compiled with the Unix compiler.
8470 Also, you must provide function prototypes for all functions that
8471 take variable numbers of arguments (including @code{printf});
8472 otherwise incorrect code will be generated for calls to those
8475 In addition, seriously incorrect code will result if you call a
8476 function with too many arguments. (Normally, extra arguments are
8477 harmlessly ignored.)
8479 @item -mregparm=@var{num}
8481 Control how many registers are used to pass integer arguments. By
8482 default, no registers are used to pass arguments, and at most 3
8483 registers can be used. You can control this behavior for a specific
8484 function by using the function attribute @samp{regparm}.
8485 @xref{Function Attributes}.
8487 @strong{Warning:} if you use this switch, and
8488 @var{num} is nonzero, then you must build all modules with the same
8489 value, including any libraries. This includes the system libraries and
8492 @item -mpreferred-stack-boundary=@var{num}
8493 @opindex mpreferred-stack-boundary
8494 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8495 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8496 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8497 size (@option{-Os}), in which case the default is the minimum correct
8498 alignment (4 bytes for x86, and 8 bytes for x86-64).
8500 On Pentium and PentiumPro, @code{double} and @code{long double} values
8501 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8502 suffer significant run time performance penalties. On Pentium III, the
8503 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8504 penalties if it is not 16 byte aligned.
8506 To ensure proper alignment of this values on the stack, the stack boundary
8507 must be as aligned as that required by any value stored on the stack.
8508 Further, every function must be generated such that it keeps the stack
8509 aligned. Thus calling a function compiled with a higher preferred
8510 stack boundary from a function compiled with a lower preferred stack
8511 boundary will most likely misalign the stack. It is recommended that
8512 libraries that use callbacks always use the default setting.
8514 This extra alignment does consume extra stack space, and generally
8515 increases code size. Code that is sensitive to stack space usage, such
8516 as embedded systems and operating system kernels, may want to reduce the
8517 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8535 These switches enable or disable the use of built-in functions that allow
8536 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8538 @xref{X86 Built-in Functions}, for details of the functions enabled
8539 and disabled by these switches.
8541 To have SSE/SSE2 instructions generated automatically from floating-point
8542 code, see @option{-mfpmath=sse}.
8545 @itemx -mno-push-args
8547 @opindex mno-push-args
8548 Use PUSH operations to store outgoing parameters. This method is shorter
8549 and usually equally fast as method using SUB/MOV operations and is enabled
8550 by default. In some cases disabling it may improve performance because of
8551 improved scheduling and reduced dependencies.
8553 @item -maccumulate-outgoing-args
8554 @opindex maccumulate-outgoing-args
8555 If enabled, the maximum amount of space required for outgoing arguments will be
8556 computed in the function prologue. This is faster on most modern CPUs
8557 because of reduced dependencies, improved scheduling and reduced stack usage
8558 when preferred stack boundary is not equal to 2. The drawback is a notable
8559 increase in code size. This switch implies @option{-mno-push-args}.
8563 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8564 on thread-safe exception handling must compile and link all code with the
8565 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8566 @option{-D_MT}; when linking, it links in a special thread helper library
8567 @option{-lmingwthrd} which cleans up per thread exception handling data.
8569 @item -mno-align-stringops
8570 @opindex mno-align-stringops
8571 Do not align destination of inlined string operations. This switch reduces
8572 code size and improves performance in case the destination is already aligned,
8573 but gcc don't know about it.
8575 @item -minline-all-stringops
8576 @opindex minline-all-stringops
8577 By default GCC inlines string operations only when destination is known to be
8578 aligned at least to 4 byte boundary. This enables more inlining, increase code
8579 size, but may improve performance of code that depends on fast memcpy, strlen
8580 and memset for short lengths.
8582 @item -momit-leaf-frame-pointer
8583 @opindex momit-leaf-frame-pointer
8584 Don't keep the frame pointer in a register for leaf functions. This
8585 avoids the instructions to save, set up and restore frame pointers and
8586 makes an extra register available in leaf functions. The option
8587 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8588 which might make debugging harder.
8590 @item -mtls-direct-seg-refs
8591 @itemx -mno-tls-direct-seg-refs
8592 @opindex mtls-direct-seg-refs
8593 Controls whether TLS variables may be accessed with offsets from the
8594 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8595 or whether the thread base pointer must be added. Whether or not this
8596 is legal depends on the operating system, and whether it maps the
8597 segment to cover the entire TLS area.
8599 For systems that use GNU libc, the default is on.
8602 These @samp{-m} switches are supported in addition to the above
8603 on AMD x86-64 processors in 64-bit environments.
8610 Generate code for a 32-bit or 64-bit environment.
8611 The 32-bit environment sets int, long and pointer to 32 bits and
8612 generates code that runs on any i386 system.
8613 The 64-bit environment sets int to 32 bits and long and pointer
8614 to 64 bits and generates code for AMD's x86-64 architecture.
8617 @opindex no-red-zone
8618 Do not use a so called red zone for x86-64 code. The red zone is mandated
8619 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8620 stack pointer that will not be modified by signal or interrupt handlers
8621 and therefore can be used for temporary data without adjusting the stack
8622 pointer. The flag @option{-mno-red-zone} disables this red zone.
8624 @item -mcmodel=small
8625 @opindex mcmodel=small
8626 Generate code for the small code model: the program and its symbols must
8627 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8628 Programs can be statically or dynamically linked. This is the default
8631 @item -mcmodel=kernel
8632 @opindex mcmodel=kernel
8633 Generate code for the kernel code model. The kernel runs in the
8634 negative 2 GB of the address space.
8635 This model has to be used for Linux kernel code.
8637 @item -mcmodel=medium
8638 @opindex mcmodel=medium
8639 Generate code for the medium model: The program is linked in the lower 2
8640 GB of the address space but symbols can be located anywhere in the
8641 address space. Programs can be statically or dynamically linked, but
8642 building of shared libraries are not supported with the medium model.
8644 @item -mcmodel=large
8645 @opindex mcmodel=large
8646 Generate code for the large model: This model makes no assumptions
8647 about addresses and sizes of sections. Currently GCC does not implement
8652 @subsection HPPA Options
8653 @cindex HPPA Options
8655 These @samp{-m} options are defined for the HPPA family of computers:
8658 @item -march=@var{architecture-type}
8660 Generate code for the specified architecture. The choices for
8661 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8662 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8663 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8664 architecture option for your machine. Code compiled for lower numbered
8665 architectures will run on higher numbered architectures, but not the
8668 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8669 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8673 @itemx -mpa-risc-1-1
8674 @itemx -mpa-risc-2-0
8675 @opindex mpa-risc-1-0
8676 @opindex mpa-risc-1-1
8677 @opindex mpa-risc-2-0
8678 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8681 @opindex mbig-switch
8682 Generate code suitable for big switch tables. Use this option only if
8683 the assembler/linker complain about out of range branches within a switch
8686 @item -mjump-in-delay
8687 @opindex mjump-in-delay
8688 Fill delay slots of function calls with unconditional jump instructions
8689 by modifying the return pointer for the function call to be the target
8690 of the conditional jump.
8692 @item -mdisable-fpregs
8693 @opindex mdisable-fpregs
8694 Prevent floating point registers from being used in any manner. This is
8695 necessary for compiling kernels which perform lazy context switching of
8696 floating point registers. If you use this option and attempt to perform
8697 floating point operations, the compiler will abort.
8699 @item -mdisable-indexing
8700 @opindex mdisable-indexing
8701 Prevent the compiler from using indexing address modes. This avoids some
8702 rather obscure problems when compiling MIG generated code under MACH@.
8704 @item -mno-space-regs
8705 @opindex mno-space-regs
8706 Generate code that assumes the target has no space registers. This allows
8707 GCC to generate faster indirect calls and use unscaled index address modes.
8709 Such code is suitable for level 0 PA systems and kernels.
8711 @item -mfast-indirect-calls
8712 @opindex mfast-indirect-calls
8713 Generate code that assumes calls never cross space boundaries. This
8714 allows GCC to emit code which performs faster indirect calls.
8716 This option will not work in the presence of shared libraries or nested
8719 @item -mlong-load-store
8720 @opindex mlong-load-store
8721 Generate 3-instruction load and store sequences as sometimes required by
8722 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8725 @item -mportable-runtime
8726 @opindex mportable-runtime
8727 Use the portable calling conventions proposed by HP for ELF systems.
8731 Enable the use of assembler directives only GAS understands.
8733 @item -mschedule=@var{cpu-type}
8735 Schedule code according to the constraints for the machine type
8736 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8737 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8738 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8739 proper scheduling option for your machine. The default scheduling is
8743 @opindex mlinker-opt
8744 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8745 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8746 linkers in which they give bogus error messages when linking some programs.
8749 @opindex msoft-float
8750 Generate output containing library calls for floating point.
8751 @strong{Warning:} the requisite libraries are not available for all HPPA
8752 targets. Normally the facilities of the machine's usual C compiler are
8753 used, but this cannot be done directly in cross-compilation. You must make
8754 your own arrangements to provide suitable library functions for
8755 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8756 does provide software floating point support.
8758 @option{-msoft-float} changes the calling convention in the output file;
8759 therefore, it is only useful if you compile @emph{all} of a program with
8760 this option. In particular, you need to compile @file{libgcc.a}, the
8761 library that comes with GCC, with @option{-msoft-float} in order for
8766 Generate the predefine, @code{_SIO}, for server IO. The default is
8767 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8768 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8769 options are available under HP-UX and HI-UX.
8773 Use GNU ld specific options. This passes @option{-shared} to ld when
8774 building a shared library. It is the default when GCC is configured,
8775 explicitly or implicitly, with the GNU linker. This option does not
8776 have any affect on which ld is called, it only changes what parameters
8777 are passed to that ld. The ld that is called is determined by the
8778 @option{--with-ld} configure option, gcc's program search path, and
8779 finally by the user's @env{PATH}. The linker used by GCC can be printed
8780 using @samp{which `gcc -print-prog-name=ld`}.
8784 Use HP ld specific options. This passes @option{-b} to ld when building
8785 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8786 links. It is the default when GCC is configured, explicitly or
8787 implicitly, with the HP linker. This option does not have any affect on
8788 which ld is called, it only changes what parameters are passed to that
8789 ld. The ld that is called is determined by the @option{--with-ld}
8790 configure option, gcc's program search path, and finally by the user's
8791 @env{PATH}. The linker used by GCC can be printed using @samp{which
8792 `gcc -print-prog-name=ld`}.
8795 @opindex mno-long-calls
8796 Generate code that uses long call sequences. This ensures that a call
8797 is always able to reach linker generated stubs. The default is to generate
8798 long calls only when the distance from the call site to the beginning
8799 of the function or translation unit, as the case may be, exceeds a
8800 predefined limit set by the branch type being used. The limits for
8801 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8802 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8805 Distances are measured from the beginning of functions when using the
8806 @option{-ffunction-sections} option, or when using the @option{-mgas}
8807 and @option{-mno-portable-runtime} options together under HP-UX with
8810 It is normally not desirable to use this option as it will degrade
8811 performance. However, it may be useful in large applications,
8812 particularly when partial linking is used to build the application.
8814 The types of long calls used depends on the capabilities of the
8815 assembler and linker, and the type of code being generated. The
8816 impact on systems that support long absolute calls, and long pic
8817 symbol-difference or pc-relative calls should be relatively small.
8818 However, an indirect call is used on 32-bit ELF systems in pic code
8819 and it is quite long.
8823 Suppress the generation of link options to search libdld.sl when the
8824 @option{-static} option is specified on HP-UX 10 and later.
8828 The HP-UX implementation of setlocale in libc has a dependency on
8829 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8830 when the @option{-static} option is specified, special link options
8831 are needed to resolve this dependency.
8833 On HP-UX 10 and later, the GCC driver adds the necessary options to
8834 link with libdld.sl when the @option{-static} option is specified.
8835 This causes the resulting binary to be dynamic. On the 64-bit port,
8836 the linkers generate dynamic binaries by default in any case. The
8837 @option{-nolibdld} option can be used to prevent the GCC driver from
8838 adding these link options.
8842 Add support for multithreading with the @dfn{dce thread} library
8843 under HP-UX. This option sets flags for both the preprocessor and
8847 @node Intel 960 Options
8848 @subsection Intel 960 Options
8850 These @samp{-m} options are defined for the Intel 960 implementations:
8853 @item -m@var{cpu-type}
8861 Assume the defaults for the machine type @var{cpu-type} for some of
8862 the other options, including instruction scheduling, floating point
8863 support, and addressing modes. The choices for @var{cpu-type} are
8864 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8865 @samp{sa}, and @samp{sb}.
8872 @opindex msoft-float
8873 The @option{-mnumerics} option indicates that the processor does support
8874 floating-point instructions. The @option{-msoft-float} option indicates
8875 that floating-point support should not be assumed.
8877 @item -mleaf-procedures
8878 @itemx -mno-leaf-procedures
8879 @opindex mleaf-procedures
8880 @opindex mno-leaf-procedures
8881 Do (or do not) attempt to alter leaf procedures to be callable with the
8882 @code{bal} instruction as well as @code{call}. This will result in more
8883 efficient code for explicit calls when the @code{bal} instruction can be
8884 substituted by the assembler or linker, but less efficient code in other
8885 cases, such as calls via function pointers, or using a linker that doesn't
8886 support this optimization.
8889 @itemx -mno-tail-call
8891 @opindex mno-tail-call
8892 Do (or do not) make additional attempts (beyond those of the
8893 machine-independent portions of the compiler) to optimize tail-recursive
8894 calls into branches. You may not want to do this because the detection of
8895 cases where this is not valid is not totally complete. The default is
8896 @option{-mno-tail-call}.
8898 @item -mcomplex-addr
8899 @itemx -mno-complex-addr
8900 @opindex mcomplex-addr
8901 @opindex mno-complex-addr
8902 Assume (or do not assume) that the use of a complex addressing mode is a
8903 win on this implementation of the i960. Complex addressing modes may not
8904 be worthwhile on the K-series, but they definitely are on the C-series.
8905 The default is currently @option{-mcomplex-addr} for all processors except
8909 @itemx -mno-code-align
8910 @opindex mcode-align
8911 @opindex mno-code-align
8912 Align code to 8-byte boundaries for faster fetching (or don't bother).
8913 Currently turned on by default for C-series implementations only.
8916 @item -mclean-linkage
8917 @itemx -mno-clean-linkage
8918 @opindex mclean-linkage
8919 @opindex mno-clean-linkage
8920 These options are not fully implemented.
8924 @itemx -mic2.0-compat
8925 @itemx -mic3.0-compat
8927 @opindex mic2.0-compat
8928 @opindex mic3.0-compat
8929 Enable compatibility with iC960 v2.0 or v3.0.
8933 @opindex masm-compat
8935 Enable compatibility with the iC960 assembler.
8937 @item -mstrict-align
8938 @itemx -mno-strict-align
8939 @opindex mstrict-align
8940 @opindex mno-strict-align
8941 Do not permit (do permit) unaligned accesses.
8945 Enable structure-alignment compatibility with Intel's gcc release version
8946 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8948 @item -mlong-double-64
8949 @opindex mlong-double-64
8950 Implement type @samp{long double} as 64-bit floating point numbers.
8951 Without the option @samp{long double} is implemented by 80-bit
8952 floating point numbers. The only reason we have it because there is
8953 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8954 is only useful for people using soft-float targets. Otherwise, we
8955 should recommend against use of it.
8959 @node DEC Alpha Options
8960 @subsection DEC Alpha Options
8962 These @samp{-m} options are defined for the DEC Alpha implementations:
8965 @item -mno-soft-float
8967 @opindex mno-soft-float
8968 @opindex msoft-float
8969 Use (do not use) the hardware floating-point instructions for
8970 floating-point operations. When @option{-msoft-float} is specified,
8971 functions in @file{libgcc.a} will be used to perform floating-point
8972 operations. Unless they are replaced by routines that emulate the
8973 floating-point operations, or compiled in such a way as to call such
8974 emulations routines, these routines will issue floating-point
8975 operations. If you are compiling for an Alpha without floating-point
8976 operations, you must ensure that the library is built so as not to call
8979 Note that Alpha implementations without floating-point operations are
8980 required to have floating-point registers.
8985 @opindex mno-fp-regs
8986 Generate code that uses (does not use) the floating-point register set.
8987 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8988 register set is not used, floating point operands are passed in integer
8989 registers as if they were integers and floating-point results are passed
8990 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8991 so any function with a floating-point argument or return value called by code
8992 compiled with @option{-mno-fp-regs} must also be compiled with that
8995 A typical use of this option is building a kernel that does not use,
8996 and hence need not save and restore, any floating-point registers.
9000 The Alpha architecture implements floating-point hardware optimized for
9001 maximum performance. It is mostly compliant with the IEEE floating
9002 point standard. However, for full compliance, software assistance is
9003 required. This option generates code fully IEEE compliant code
9004 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9005 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9006 defined during compilation. The resulting code is less efficient but is
9007 able to correctly support denormalized numbers and exceptional IEEE
9008 values such as not-a-number and plus/minus infinity. Other Alpha
9009 compilers call this option @option{-ieee_with_no_inexact}.
9011 @item -mieee-with-inexact
9012 @opindex mieee-with-inexact
9013 This is like @option{-mieee} except the generated code also maintains
9014 the IEEE @var{inexact-flag}. Turning on this option causes the
9015 generated code to implement fully-compliant IEEE math. In addition to
9016 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9017 macro. On some Alpha implementations the resulting code may execute
9018 significantly slower than the code generated by default. Since there is
9019 very little code that depends on the @var{inexact-flag}, you should
9020 normally not specify this option. Other Alpha compilers call this
9021 option @option{-ieee_with_inexact}.
9023 @item -mfp-trap-mode=@var{trap-mode}
9024 @opindex mfp-trap-mode
9025 This option controls what floating-point related traps are enabled.
9026 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9027 The trap mode can be set to one of four values:
9031 This is the default (normal) setting. The only traps that are enabled
9032 are the ones that cannot be disabled in software (e.g., division by zero
9036 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9040 Like @samp{su}, but the instructions are marked to be safe for software
9041 completion (see Alpha architecture manual for details).
9044 Like @samp{su}, but inexact traps are enabled as well.
9047 @item -mfp-rounding-mode=@var{rounding-mode}
9048 @opindex mfp-rounding-mode
9049 Selects the IEEE rounding mode. Other Alpha compilers call this option
9050 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9055 Normal IEEE rounding mode. Floating point numbers are rounded towards
9056 the nearest machine number or towards the even machine number in case
9060 Round towards minus infinity.
9063 Chopped rounding mode. Floating point numbers are rounded towards zero.
9066 Dynamic rounding mode. A field in the floating point control register
9067 (@var{fpcr}, see Alpha architecture reference manual) controls the
9068 rounding mode in effect. The C library initializes this register for
9069 rounding towards plus infinity. Thus, unless your program modifies the
9070 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9073 @item -mtrap-precision=@var{trap-precision}
9074 @opindex mtrap-precision
9075 In the Alpha architecture, floating point traps are imprecise. This
9076 means without software assistance it is impossible to recover from a
9077 floating trap and program execution normally needs to be terminated.
9078 GCC can generate code that can assist operating system trap handlers
9079 in determining the exact location that caused a floating point trap.
9080 Depending on the requirements of an application, different levels of
9081 precisions can be selected:
9085 Program precision. This option is the default and means a trap handler
9086 can only identify which program caused a floating point exception.
9089 Function precision. The trap handler can determine the function that
9090 caused a floating point exception.
9093 Instruction precision. The trap handler can determine the exact
9094 instruction that caused a floating point exception.
9097 Other Alpha compilers provide the equivalent options called
9098 @option{-scope_safe} and @option{-resumption_safe}.
9100 @item -mieee-conformant
9101 @opindex mieee-conformant
9102 This option marks the generated code as IEEE conformant. You must not
9103 use this option unless you also specify @option{-mtrap-precision=i} and either
9104 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9105 is to emit the line @samp{.eflag 48} in the function prologue of the
9106 generated assembly file. Under DEC Unix, this has the effect that
9107 IEEE-conformant math library routines will be linked in.
9109 @item -mbuild-constants
9110 @opindex mbuild-constants
9111 Normally GCC examines a 32- or 64-bit integer constant to
9112 see if it can construct it from smaller constants in two or three
9113 instructions. If it cannot, it will output the constant as a literal and
9114 generate code to load it from the data segment at runtime.
9116 Use this option to require GCC to construct @emph{all} integer constants
9117 using code, even if it takes more instructions (the maximum is six).
9119 You would typically use this option to build a shared library dynamic
9120 loader. Itself a shared library, it must relocate itself in memory
9121 before it can find the variables and constants in its own data segment.
9127 Select whether to generate code to be assembled by the vendor-supplied
9128 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9146 Indicate whether GCC should generate code to use the optional BWX,
9147 CIX, FIX and MAX instruction sets. The default is to use the instruction
9148 sets supported by the CPU type specified via @option{-mcpu=} option or that
9149 of the CPU on which GCC was built if none was specified.
9154 @opindex mfloat-ieee
9155 Generate code that uses (does not use) VAX F and G floating point
9156 arithmetic instead of IEEE single and double precision.
9158 @item -mexplicit-relocs
9159 @itemx -mno-explicit-relocs
9160 @opindex mexplicit-relocs
9161 @opindex mno-explicit-relocs
9162 Older Alpha assemblers provided no way to generate symbol relocations
9163 except via assembler macros. Use of these macros does not allow
9164 optimal instruction scheduling. GNU binutils as of version 2.12
9165 supports a new syntax that allows the compiler to explicitly mark
9166 which relocations should apply to which instructions. This option
9167 is mostly useful for debugging, as GCC detects the capabilities of
9168 the assembler when it is built and sets the default accordingly.
9172 @opindex msmall-data
9173 @opindex mlarge-data
9174 When @option{-mexplicit-relocs} is in effect, static data is
9175 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9176 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9177 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9178 16-bit relocations off of the @code{$gp} register. This limits the
9179 size of the small data area to 64KB, but allows the variables to be
9180 directly accessed via a single instruction.
9182 The default is @option{-mlarge-data}. With this option the data area
9183 is limited to just below 2GB. Programs that require more than 2GB of
9184 data must use @code{malloc} or @code{mmap} to allocate the data in the
9185 heap instead of in the program's data segment.
9187 When generating code for shared libraries, @option{-fpic} implies
9188 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9192 @opindex msmall-text
9193 @opindex mlarge-text
9194 When @option{-msmall-text} is used, the compiler assumes that the
9195 code of the entire program (or shared library) fits in 4MB, and is
9196 thus reachable with a branch instruction. When @option{-msmall-data}
9197 is used, the compiler can assume that all local symbols share the
9198 same @code{$gp} value, and thus reduce the number of instructions
9199 required for a function call from 4 to 1.
9201 The default is @option{-mlarge-text}.
9203 @item -mcpu=@var{cpu_type}
9205 Set the instruction set and instruction scheduling parameters for
9206 machine type @var{cpu_type}. You can specify either the @samp{EV}
9207 style name or the corresponding chip number. GCC supports scheduling
9208 parameters for the EV4, EV5 and EV6 family of processors and will
9209 choose the default values for the instruction set from the processor
9210 you specify. If you do not specify a processor type, GCC will default
9211 to the processor on which the compiler was built.
9213 Supported values for @var{cpu_type} are
9219 Schedules as an EV4 and has no instruction set extensions.
9223 Schedules as an EV5 and has no instruction set extensions.
9227 Schedules as an EV5 and supports the BWX extension.
9232 Schedules as an EV5 and supports the BWX and MAX extensions.
9236 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9240 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9243 @item -mtune=@var{cpu_type}
9245 Set only the instruction scheduling parameters for machine type
9246 @var{cpu_type}. The instruction set is not changed.
9248 @item -mmemory-latency=@var{time}
9249 @opindex mmemory-latency
9250 Sets the latency the scheduler should assume for typical memory
9251 references as seen by the application. This number is highly
9252 dependent on the memory access patterns used by the application
9253 and the size of the external cache on the machine.
9255 Valid options for @var{time} are
9259 A decimal number representing clock cycles.
9265 The compiler contains estimates of the number of clock cycles for
9266 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9267 (also called Dcache, Scache, and Bcache), as well as to main memory.
9268 Note that L3 is only valid for EV5.
9273 @node DEC Alpha/VMS Options
9274 @subsection DEC Alpha/VMS Options
9276 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9279 @item -mvms-return-codes
9280 @opindex mvms-return-codes
9281 Return VMS condition codes from main. The default is to return POSIX
9282 style condition (e.g.@ error) codes.
9285 @node H8/300 Options
9286 @subsection H8/300 Options
9288 These @samp{-m} options are defined for the H8/300 implementations:
9293 Shorten some address references at link time, when possible; uses the
9294 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9295 ld.info, Using ld}, for a fuller description.
9299 Generate code for the H8/300H@.
9303 Generate code for the H8S@.
9307 Generate code for the H8S and H8/300H in the normal mode. This switch
9308 must be used either with -mh or -ms.
9312 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9316 Make @code{int} data 32 bits by default.
9320 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9321 The default for the H8/300H and H8S is to align longs and floats on 4
9323 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9324 This option has no effect on the H8/300.
9328 @subsection SH Options
9330 These @samp{-m} options are defined for the SH implementations:
9335 Generate code for the SH1.
9339 Generate code for the SH2.
9342 Generate code for the SH2e.
9346 Generate code for the SH3.
9350 Generate code for the SH3e.
9354 Generate code for the SH4 without a floating-point unit.
9356 @item -m4-single-only
9357 @opindex m4-single-only
9358 Generate code for the SH4 with a floating-point unit that only
9359 supports single-precision arithmetic.
9363 Generate code for the SH4 assuming the floating-point unit is in
9364 single-precision mode by default.
9368 Generate code for the SH4.
9372 Compile code for the processor in big endian mode.
9376 Compile code for the processor in little endian mode.
9380 Align doubles at 64-bit boundaries. Note that this changes the calling
9381 conventions, and thus some functions from the standard C library will
9382 not work unless you recompile it first with @option{-mdalign}.
9386 Shorten some address references at link time, when possible; uses the
9387 linker option @option{-relax}.
9391 Use 32-bit offsets in @code{switch} tables. The default is to use
9396 Enable the use of the instruction @code{fmovd}.
9400 Comply with the calling conventions defined by Renesas.
9404 Mark the @code{MAC} register as call-clobbered, even if
9405 @option{-mhitachi} is given.
9409 Increase IEEE-compliance of floating-point code.
9413 Dump instruction size and location in the assembly code.
9417 This option is deprecated. It pads structures to multiple of 4 bytes,
9418 which is incompatible with the SH ABI@.
9422 Optimize for space instead of speed. Implied by @option{-Os}.
9426 When generating position-independent code, emit function calls using
9427 the Global Offset Table instead of the Procedure Linkage Table.
9431 Generate a library function call to invalidate instruction cache
9432 entries, after fixing up a trampoline. This library function call
9433 doesn't assume it can write to the whole memory address space. This
9434 is the default when the target is @code{sh-*-linux*}.
9437 @node System V Options
9438 @subsection Options for System V
9440 These additional options are available on System V Release 4 for
9441 compatibility with other compilers on those systems:
9446 Create a shared object.
9447 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9451 Identify the versions of each tool used by the compiler, in a
9452 @code{.ident} assembler directive in the output.
9456 Refrain from adding @code{.ident} directives to the output file (this is
9459 @item -YP,@var{dirs}
9461 Search the directories @var{dirs}, and no others, for libraries
9462 specified with @option{-l}.
9466 Look in the directory @var{dir} to find the M4 preprocessor.
9467 The assembler uses this option.
9468 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9469 @c the generic assembler that comes with Solaris takes just -Ym.
9472 @node TMS320C3x/C4x Options
9473 @subsection TMS320C3x/C4x Options
9474 @cindex TMS320C3x/C4x Options
9476 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9480 @item -mcpu=@var{cpu_type}
9482 Set the instruction set, register set, and instruction scheduling
9483 parameters for machine type @var{cpu_type}. Supported values for
9484 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9485 @samp{c44}. The default is @samp{c40} to generate code for the
9490 @itemx -msmall-memory
9492 @opindex mbig-memory
9494 @opindex msmall-memory
9496 Generates code for the big or small memory model. The small memory
9497 model assumed that all data fits into one 64K word page. At run-time
9498 the data page (DP) register must be set to point to the 64K page
9499 containing the .bss and .data program sections. The big memory model is
9500 the default and requires reloading of the DP register for every direct
9507 Allow (disallow) allocation of general integer operands into the block
9514 Enable (disable) generation of code using decrement and branch,
9515 DBcond(D), instructions. This is enabled by default for the C4x. To be
9516 on the safe side, this is disabled for the C3x, since the maximum
9517 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9518 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9519 that it can utilize the decrement and branch instruction, but will give
9520 up if there is more than one memory reference in the loop. Thus a loop
9521 where the loop counter is decremented can generate slightly more
9522 efficient code, in cases where the RPTB instruction cannot be utilized.
9524 @item -mdp-isr-reload
9526 @opindex mdp-isr-reload
9528 Force the DP register to be saved on entry to an interrupt service
9529 routine (ISR), reloaded to point to the data section, and restored on
9530 exit from the ISR@. This should not be required unless someone has
9531 violated the small memory model by modifying the DP register, say within
9538 For the C3x use the 24-bit MPYI instruction for integer multiplies
9539 instead of a library call to guarantee 32-bit results. Note that if one
9540 of the operands is a constant, then the multiplication will be performed
9541 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9542 then squaring operations are performed inline instead of a library call.
9545 @itemx -mno-fast-fix
9547 @opindex mno-fast-fix
9548 The C3x/C4x FIX instruction to convert a floating point value to an
9549 integer value chooses the nearest integer less than or equal to the
9550 floating point value rather than to the nearest integer. Thus if the
9551 floating point number is negative, the result will be incorrectly
9552 truncated an additional code is necessary to detect and correct this
9553 case. This option can be used to disable generation of the additional
9554 code required to correct the result.
9560 Enable (disable) generation of repeat block sequences using the RPTB
9561 instruction for zero overhead looping. The RPTB construct is only used
9562 for innermost loops that do not call functions or jump across the loop
9563 boundaries. There is no advantage having nested RPTB loops due to the
9564 overhead required to save and restore the RC, RS, and RE registers.
9565 This is enabled by default with @option{-O2}.
9567 @item -mrpts=@var{count}
9571 Enable (disable) the use of the single instruction repeat instruction
9572 RPTS@. If a repeat block contains a single instruction, and the loop
9573 count can be guaranteed to be less than the value @var{count}, GCC will
9574 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9575 then a RPTS will be emitted even if the loop count cannot be determined
9576 at compile time. Note that the repeated instruction following RPTS does
9577 not have to be reloaded from memory each iteration, thus freeing up the
9578 CPU buses for operands. However, since interrupts are blocked by this
9579 instruction, it is disabled by default.
9581 @item -mloop-unsigned
9582 @itemx -mno-loop-unsigned
9583 @opindex mloop-unsigned
9584 @opindex mno-loop-unsigned
9585 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9586 is @math{2^{31} + 1} since these instructions test if the iteration count is
9587 negative to terminate the loop. If the iteration count is unsigned
9588 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9589 exceeded. This switch allows an unsigned iteration count.
9593 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9594 with. This also enforces compatibility with the API employed by the TI
9595 C3x C compiler. For example, long doubles are passed as structures
9596 rather than in floating point registers.
9602 Generate code that uses registers (stack) for passing arguments to functions.
9603 By default, arguments are passed in registers where possible rather
9604 than by pushing arguments on to the stack.
9606 @item -mparallel-insns
9607 @itemx -mno-parallel-insns
9608 @opindex mparallel-insns
9609 @opindex mno-parallel-insns
9610 Allow the generation of parallel instructions. This is enabled by
9611 default with @option{-O2}.
9613 @item -mparallel-mpy
9614 @itemx -mno-parallel-mpy
9615 @opindex mparallel-mpy
9616 @opindex mno-parallel-mpy
9617 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9618 provided @option{-mparallel-insns} is also specified. These instructions have
9619 tight register constraints which can pessimize the code generation
9625 @subsection V850 Options
9626 @cindex V850 Options
9628 These @samp{-m} options are defined for V850 implementations:
9632 @itemx -mno-long-calls
9633 @opindex mlong-calls
9634 @opindex mno-long-calls
9635 Treat all calls as being far away (near). If calls are assumed to be
9636 far away, the compiler will always load the functions address up into a
9637 register, and call indirect through the pointer.
9643 Do not optimize (do optimize) basic blocks that use the same index
9644 pointer 4 or more times to copy pointer into the @code{ep} register, and
9645 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9646 option is on by default if you optimize.
9648 @item -mno-prolog-function
9649 @itemx -mprolog-function
9650 @opindex mno-prolog-function
9651 @opindex mprolog-function
9652 Do not use (do use) external functions to save and restore registers
9653 at the prologue and epilogue of a function. The external functions
9654 are slower, but use less code space if more than one function saves
9655 the same number of registers. The @option{-mprolog-function} option
9656 is on by default if you optimize.
9660 Try to make the code as small as possible. At present, this just turns
9661 on the @option{-mep} and @option{-mprolog-function} options.
9665 Put static or global variables whose size is @var{n} bytes or less into
9666 the tiny data area that register @code{ep} points to. The tiny data
9667 area can hold up to 256 bytes in total (128 bytes for byte references).
9671 Put static or global variables whose size is @var{n} bytes or less into
9672 the small data area that register @code{gp} points to. The small data
9673 area can hold up to 64 kilobytes.
9677 Put static or global variables whose size is @var{n} bytes or less into
9678 the first 32 kilobytes of memory.
9682 Specify that the target processor is the V850.
9685 @opindex mbig-switch
9686 Generate code suitable for big switch tables. Use this option only if
9687 the assembler/linker complain about out of range branches within a switch
9692 This option will cause r2 and r5 to be used in the code generated by
9693 the compiler. This setting is the default.
9696 @opindex mno-app-regs
9697 This option will cause r2 and r5 to be treated as fixed registers.
9701 Specify that the target processor is the V850E1. The preprocessor
9702 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9703 this option is used.
9707 Specify that the target processor is the V850E. The preprocessor
9708 constant @samp{__v850e__} will be defined if this option is used.
9710 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9711 are defined then a default target processor will be chosen and the
9712 relevant @samp{__v850*__} preprocessor constant will be defined.
9714 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9715 defined, regardless of which processor variant is the target.
9717 @item -mdisable-callt
9718 @opindex mdisable-callt
9719 This option will suppress generation of the CALLT instruction for the
9720 v850e and v850e1 flavors of the v850 architecture. The default is
9721 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9726 @subsection ARC Options
9729 These options are defined for ARC implementations:
9734 Compile code for little endian mode. This is the default.
9738 Compile code for big endian mode.
9741 @opindex mmangle-cpu
9742 Prepend the name of the cpu to all public symbol names.
9743 In multiple-processor systems, there are many ARC variants with different
9744 instruction and register set characteristics. This flag prevents code
9745 compiled for one cpu to be linked with code compiled for another.
9746 No facility exists for handling variants that are ``almost identical''.
9747 This is an all or nothing option.
9749 @item -mcpu=@var{cpu}
9751 Compile code for ARC variant @var{cpu}.
9752 Which variants are supported depend on the configuration.
9753 All variants support @option{-mcpu=base}, this is the default.
9755 @item -mtext=@var{text-section}
9756 @itemx -mdata=@var{data-section}
9757 @itemx -mrodata=@var{readonly-data-section}
9761 Put functions, data, and readonly data in @var{text-section},
9762 @var{data-section}, and @var{readonly-data-section} respectively
9763 by default. This can be overridden with the @code{section} attribute.
9764 @xref{Variable Attributes}.
9769 @subsection NS32K Options
9770 @cindex NS32K options
9772 These are the @samp{-m} options defined for the 32000 series. The default
9773 values for these options depends on which style of 32000 was selected when
9774 the compiler was configured; the defaults for the most common choices are
9782 Generate output for a 32032. This is the default
9783 when the compiler is configured for 32032 and 32016 based systems.
9789 Generate output for a 32332. This is the default
9790 when the compiler is configured for 32332-based systems.
9796 Generate output for a 32532. This is the default
9797 when the compiler is configured for 32532-based systems.
9801 Generate output containing 32081 instructions for floating point.
9802 This is the default for all systems.
9806 Generate output containing 32381 instructions for floating point. This
9807 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9808 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9812 Try and generate multiply-add floating point instructions @code{polyF}
9813 and @code{dotF}. This option is only available if the @option{-m32381}
9814 option is in effect. Using these instructions requires changes to
9815 register allocation which generally has a negative impact on
9816 performance. This option should only be enabled when compiling code
9817 particularly likely to make heavy use of multiply-add instructions.
9820 @opindex mnomulti-add
9821 Do not try and generate multiply-add floating point instructions
9822 @code{polyF} and @code{dotF}. This is the default on all platforms.
9825 @opindex msoft-float
9826 Generate output containing library calls for floating point.
9827 @strong{Warning:} the requisite libraries may not be available.
9829 @item -mieee-compare
9830 @itemx -mno-ieee-compare
9831 @opindex mieee-compare
9832 @opindex mno-ieee-compare
9833 Control whether or not the compiler uses IEEE floating point
9834 comparisons. These handle correctly the case where the result of a
9835 comparison is unordered.
9836 @strong{Warning:} the requisite kernel support may not be available.
9839 @opindex mnobitfield
9840 Do not use the bit-field instructions. On some machines it is faster to
9841 use shifting and masking operations. This is the default for the pc532.
9845 Do use the bit-field instructions. This is the default for all platforms
9850 Use a different function-calling convention, in which functions
9851 that take a fixed number of arguments return pop their
9852 arguments on return with the @code{ret} instruction.
9854 This calling convention is incompatible with the one normally
9855 used on Unix, so you cannot use it if you need to call libraries
9856 compiled with the Unix compiler.
9858 Also, you must provide function prototypes for all functions that
9859 take variable numbers of arguments (including @code{printf});
9860 otherwise incorrect code will be generated for calls to those
9863 In addition, seriously incorrect code will result if you call a
9864 function with too many arguments. (Normally, extra arguments are
9865 harmlessly ignored.)
9867 This option takes its name from the 680x0 @code{rtd} instruction.
9872 Use a different function-calling convention where the first two arguments
9873 are passed in registers.
9875 This calling convention is incompatible with the one normally
9876 used on Unix, so you cannot use it if you need to call libraries
9877 compiled with the Unix compiler.
9880 @opindex mnoregparam
9881 Do not pass any arguments in registers. This is the default for all
9886 It is OK to use the sb as an index register which is always loaded with
9887 zero. This is the default for the pc532-netbsd target.
9891 The sb register is not available for use or has not been initialized to
9892 zero by the run time system. This is the default for all targets except
9893 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9894 @option{-fpic} is set.
9898 Many ns32000 series addressing modes use displacements of up to 512MB@.
9899 If an address is above 512MB then displacements from zero can not be used.
9900 This option causes code to be generated which can be loaded above 512MB@.
9901 This may be useful for operating systems or ROM code.
9905 Assume code will be loaded in the first 512MB of virtual address space.
9906 This is the default for all platforms.
9912 @subsection AVR Options
9915 These options are defined for AVR implementations:
9918 @item -mmcu=@var{mcu}
9920 Specify ATMEL AVR instruction set or MCU type.
9922 Instruction set avr1 is for the minimal AVR core, not supported by the C
9923 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9924 attiny11, attiny12, attiny15, attiny28).
9926 Instruction set avr2 (default) is for the classic AVR core with up to
9927 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9928 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9929 at90c8534, at90s8535).
9931 Instruction set avr3 is for the classic AVR core with up to 128K program
9932 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9934 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9935 memory space (MCU types: atmega8, atmega83, atmega85).
9937 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9938 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9939 atmega64, atmega128, at43usb355, at94k).
9943 Output instruction sizes to the asm file.
9945 @item -minit-stack=@var{N}
9946 @opindex minit-stack
9947 Specify the initial stack address, which may be a symbol or numeric value,
9948 @samp{__stack} is the default.
9950 @item -mno-interrupts
9951 @opindex mno-interrupts
9952 Generated code is not compatible with hardware interrupts.
9953 Code size will be smaller.
9955 @item -mcall-prologues
9956 @opindex mcall-prologues
9957 Functions prologues/epilogues expanded as call to appropriate
9958 subroutines. Code size will be smaller.
9960 @item -mno-tablejump
9961 @opindex mno-tablejump
9962 Do not generate tablejump insns which sometimes increase code size.
9965 @opindex mtiny-stack
9966 Change only the low 8 bits of the stack pointer.
9970 @subsection MCore Options
9971 @cindex MCore options
9973 These are the @samp{-m} options defined for the Motorola M*Core
9981 @opindex mno-hardlit
9982 Inline constants into the code stream if it can be done in two
9983 instructions or less.
9989 Use the divide instruction. (Enabled by default).
9991 @item -mrelax-immediate
9992 @itemx -mno-relax-immediate
9993 @opindex mrelax-immediate
9994 @opindex mno-relax-immediate
9995 Allow arbitrary sized immediates in bit operations.
9997 @item -mwide-bitfields
9998 @itemx -mno-wide-bitfields
9999 @opindex mwide-bitfields
10000 @opindex mno-wide-bitfields
10001 Always treat bit-fields as int-sized.
10003 @item -m4byte-functions
10004 @itemx -mno-4byte-functions
10005 @opindex m4byte-functions
10006 @opindex mno-4byte-functions
10007 Force all functions to be aligned to a four byte boundary.
10009 @item -mcallgraph-data
10010 @itemx -mno-callgraph-data
10011 @opindex mcallgraph-data
10012 @opindex mno-callgraph-data
10013 Emit callgraph information.
10016 @itemx -mno-slow-bytes
10017 @opindex mslow-bytes
10018 @opindex mno-slow-bytes
10019 Prefer word access when reading byte quantities.
10021 @item -mlittle-endian
10022 @itemx -mbig-endian
10023 @opindex mlittle-endian
10024 @opindex mbig-endian
10025 Generate code for a little endian target.
10031 Generate code for the 210 processor.
10034 @node IA-64 Options
10035 @subsection IA-64 Options
10036 @cindex IA-64 Options
10038 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10042 @opindex mbig-endian
10043 Generate code for a big endian target. This is the default for HP-UX@.
10045 @item -mlittle-endian
10046 @opindex mlittle-endian
10047 Generate code for a little endian target. This is the default for AIX5
10053 @opindex mno-gnu-as
10054 Generate (or don't) code for the GNU assembler. This is the default.
10055 @c Also, this is the default if the configure option @option{--with-gnu-as}
10061 @opindex mno-gnu-ld
10062 Generate (or don't) code for the GNU linker. This is the default.
10063 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10068 Generate code that does not use a global pointer register. The result
10069 is not position independent code, and violates the IA-64 ABI@.
10071 @item -mvolatile-asm-stop
10072 @itemx -mno-volatile-asm-stop
10073 @opindex mvolatile-asm-stop
10074 @opindex mno-volatile-asm-stop
10075 Generate (or don't) a stop bit immediately before and after volatile asm
10080 Generate code that works around Itanium B step errata.
10082 @item -mregister-names
10083 @itemx -mno-register-names
10084 @opindex mregister-names
10085 @opindex mno-register-names
10086 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10087 the stacked registers. This may make assembler output more readable.
10093 Disable (or enable) optimizations that use the small data section. This may
10094 be useful for working around optimizer bugs.
10096 @item -mconstant-gp
10097 @opindex mconstant-gp
10098 Generate code that uses a single constant global pointer value. This is
10099 useful when compiling kernel code.
10103 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10104 This is useful when compiling firmware code.
10106 @item -minline-float-divide-min-latency
10107 @opindex minline-float-divide-min-latency
10108 Generate code for inline divides of floating point values
10109 using the minimum latency algorithm.
10111 @item -minline-float-divide-max-throughput
10112 @opindex minline-float-divide-max-throughput
10113 Generate code for inline divides of floating point values
10114 using the maximum throughput algorithm.
10116 @item -minline-int-divide-min-latency
10117 @opindex minline-int-divide-min-latency
10118 Generate code for inline divides of integer values
10119 using the minimum latency algorithm.
10121 @item -minline-int-divide-max-throughput
10122 @opindex minline-int-divide-max-throughput
10123 Generate code for inline divides of integer values
10124 using the maximum throughput algorithm.
10126 @item -mno-dwarf2-asm
10127 @itemx -mdwarf2-asm
10128 @opindex mno-dwarf2-asm
10129 @opindex mdwarf2-asm
10130 Don't (or do) generate assembler code for the DWARF2 line number debugging
10131 info. This may be useful when not using the GNU assembler.
10133 @item -mfixed-range=@var{register-range}
10134 @opindex mfixed-range
10135 Generate code treating the given register range as fixed registers.
10136 A fixed register is one that the register allocator can not use. This is
10137 useful when compiling kernel code. A register range is specified as
10138 two registers separated by a dash. Multiple register ranges can be
10139 specified separated by a comma.
10141 @item -mearly-stop-bits
10142 @itemx -mno-early-stop-bits
10143 @opindex mearly-stop-bits
10144 @opindex mno-early-stop-bits
10145 Allow stop bits to be placed earlier than immediately preceding the
10146 instruction that triggered the stop bit. This can improve instruction
10147 scheduling, but does not always do so.
10151 @subsection D30V Options
10152 @cindex D30V Options
10154 These @samp{-m} options are defined for D30V implementations:
10159 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
10160 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
10161 memory, which starts at location @code{0x80000000}.
10164 @opindex mextmemory
10165 Same as the @option{-mextmem} switch.
10169 Link the @samp{.text} section into onchip text memory, which starts at
10170 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
10171 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
10172 into onchip data memory, which starts at location @code{0x20000000}.
10174 @item -mno-asm-optimize
10175 @itemx -masm-optimize
10176 @opindex mno-asm-optimize
10177 @opindex masm-optimize
10178 Disable (enable) passing @option{-O} to the assembler when optimizing.
10179 The assembler uses the @option{-O} option to automatically parallelize
10180 adjacent short instructions where possible.
10182 @item -mbranch-cost=@var{n}
10183 @opindex mbranch-cost
10184 Increase the internal costs of branches to @var{n}. Higher costs means
10185 that the compiler will issue more instructions to avoid doing a branch.
10188 @item -mcond-exec=@var{n}
10189 @opindex mcond-exec
10190 Specify the maximum number of conditionally executed instructions that
10191 replace a branch. The default is 4.
10194 @node S/390 and zSeries Options
10195 @subsection S/390 and zSeries Options
10196 @cindex S/390 and zSeries Options
10198 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10202 @itemx -msoft-float
10203 @opindex mhard-float
10204 @opindex msoft-float
10205 Use (do not use) the hardware floating-point instructions and registers
10206 for floating-point operations. When @option{-msoft-float} is specified,
10207 functions in @file{libgcc.a} will be used to perform floating-point
10208 operations. When @option{-mhard-float} is specified, the compiler
10209 generates IEEE floating-point instructions. This is the default.
10212 @itemx -mno-backchain
10213 @opindex mbackchain
10214 @opindex mno-backchain
10215 Generate (or do not generate) code which maintains an explicit
10216 backchain within the stack frame that points to the caller's frame.
10217 This is currently needed to allow debugging. The default is to
10218 generate the backchain.
10221 @itemx -mno-small-exec
10222 @opindex msmall-exec
10223 @opindex mno-small-exec
10224 Generate (or do not generate) code using the @code{bras} instruction
10225 to do subroutine calls.
10226 This only works reliably if the total executable size does not
10227 exceed 64k. The default is to use the @code{basr} instruction instead,
10228 which does not have this limitation.
10234 When @option{-m31} is specified, generate code compliant to the
10235 Linux for S/390 ABI@. When @option{-m64} is specified, generate
10236 code compliant to the Linux for zSeries ABI@. This allows GCC in
10237 particular to generate 64-bit instructions. For the @samp{s390}
10238 targets, the default is @option{-m31}, while the @samp{s390x}
10239 targets default to @option{-m64}.
10245 When @option{-mzarch} is specified, generate code using the
10246 instructions available on z/Architecture.
10247 When @option{-mesa} is specified, generate code using the
10248 instructions available on ESA/390. Note that @option{-mesa} is
10249 not possible with @option{-m64}.
10250 When generating code compliant to the Linux for S/390 ABI,
10251 the default is @option{-mesa}. When generating code compliant
10252 to the Linux for zSeries ABI, the default is @option{-mzarch}.
10258 Generate (or do not generate) code using the @code{mvcle} instruction
10259 to perform block moves. When @option{-mno-mvcle} is specified,
10260 use a @code{mvc} loop instead. This is the default.
10266 Print (or do not print) additional debug information when compiling.
10267 The default is to not print debug information.
10269 @item -march=@var{cpu-type}
10271 Generate code that will run on @var{cpu-type}, which is the name of a system
10272 representing a certain processor type. Possible values for
10273 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10274 When generating code using the instructions available on z/Architecture,
10275 the default is @option{-march=z900}. Otherwise, the default is
10276 @option{-march=g5}.
10278 @item -mtune=@var{cpu-type}
10280 Tune to @var{cpu-type} everything applicable about the generated code,
10281 except for the ABI and the set of available instructions.
10282 The list of @var{cpu-type} values is the same as for @option{-march}.
10283 The default is the value used for @option{-march}.
10288 @subsection CRIS Options
10289 @cindex CRIS Options
10291 These options are defined specifically for the CRIS ports.
10294 @item -march=@var{architecture-type}
10295 @itemx -mcpu=@var{architecture-type}
10298 Generate code for the specified architecture. The choices for
10299 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10300 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10301 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10304 @item -mtune=@var{architecture-type}
10306 Tune to @var{architecture-type} everything applicable about the generated
10307 code, except for the ABI and the set of available instructions. The
10308 choices for @var{architecture-type} are the same as for
10309 @option{-march=@var{architecture-type}}.
10311 @item -mmax-stack-frame=@var{n}
10312 @opindex mmax-stack-frame
10313 Warn when the stack frame of a function exceeds @var{n} bytes.
10315 @item -melinux-stacksize=@var{n}
10316 @opindex melinux-stacksize
10317 Only available with the @samp{cris-axis-aout} target. Arranges for
10318 indications in the program to the kernel loader that the stack of the
10319 program should be set to @var{n} bytes.
10325 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10326 @option{-march=v3} and @option{-march=v8} respectively.
10330 Enable CRIS-specific verbose debug-related information in the assembly
10331 code. This option also has the effect to turn off the @samp{#NO_APP}
10332 formatted-code indicator to the assembler at the beginning of the
10337 Do not use condition-code results from previous instruction; always emit
10338 compare and test instructions before use of condition codes.
10340 @item -mno-side-effects
10341 @opindex mno-side-effects
10342 Do not emit instructions with side-effects in addressing modes other than
10345 @item -mstack-align
10346 @itemx -mno-stack-align
10347 @itemx -mdata-align
10348 @itemx -mno-data-align
10349 @itemx -mconst-align
10350 @itemx -mno-const-align
10351 @opindex mstack-align
10352 @opindex mno-stack-align
10353 @opindex mdata-align
10354 @opindex mno-data-align
10355 @opindex mconst-align
10356 @opindex mno-const-align
10357 These options (no-options) arranges (eliminate arrangements) for the
10358 stack-frame, individual data and constants to be aligned for the maximum
10359 single data access size for the chosen CPU model. The default is to
10360 arrange for 32-bit alignment. ABI details such as structure layout are
10361 not affected by these options.
10369 Similar to the stack- data- and const-align options above, these options
10370 arrange for stack-frame, writable data and constants to all be 32-bit,
10371 16-bit or 8-bit aligned. The default is 32-bit alignment.
10373 @item -mno-prologue-epilogue
10374 @itemx -mprologue-epilogue
10375 @opindex mno-prologue-epilogue
10376 @opindex mprologue-epilogue
10377 With @option{-mno-prologue-epilogue}, the normal function prologue and
10378 epilogue that sets up the stack-frame are omitted and no return
10379 instructions or return sequences are generated in the code. Use this
10380 option only together with visual inspection of the compiled code: no
10381 warnings or errors are generated when call-saved registers must be saved,
10382 or storage for local variable needs to be allocated.
10386 @opindex mno-gotplt
10388 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10389 instruction sequences that load addresses for functions from the PLT part
10390 of the GOT rather than (traditional on other architectures) calls to the
10391 PLT. The default is @option{-mgotplt}.
10395 Legacy no-op option only recognized with the cris-axis-aout target.
10399 Legacy no-op option only recognized with the cris-axis-elf and
10400 cris-axis-linux-gnu targets.
10404 Only recognized with the cris-axis-aout target, where it selects a
10405 GNU/linux-like multilib, include files and instruction set for
10406 @option{-march=v8}.
10410 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10414 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10415 to link with input-output functions from a simulator library. Code,
10416 initialized data and zero-initialized data are allocated consecutively.
10420 Like @option{-sim}, but pass linker options to locate initialized data at
10421 0x40000000 and zero-initialized data at 0x80000000.
10425 @subsection MMIX Options
10426 @cindex MMIX Options
10428 These options are defined for the MMIX:
10432 @itemx -mno-libfuncs
10434 @opindex mno-libfuncs
10435 Specify that intrinsic library functions are being compiled, passing all
10436 values in registers, no matter the size.
10439 @itemx -mno-epsilon
10441 @opindex mno-epsilon
10442 Generate floating-point comparison instructions that compare with respect
10443 to the @code{rE} epsilon register.
10445 @item -mabi=mmixware
10447 @opindex mabi-mmixware
10449 Generate code that passes function parameters and return values that (in
10450 the called function) are seen as registers @code{$0} and up, as opposed to
10451 the GNU ABI which uses global registers @code{$231} and up.
10453 @item -mzero-extend
10454 @itemx -mno-zero-extend
10455 @opindex mzero-extend
10456 @opindex mno-zero-extend
10457 When reading data from memory in sizes shorter than 64 bits, use (do not
10458 use) zero-extending load instructions by default, rather than
10459 sign-extending ones.
10462 @itemx -mno-knuthdiv
10464 @opindex mno-knuthdiv
10465 Make the result of a division yielding a remainder have the same sign as
10466 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10467 remainder follows the sign of the dividend. Both methods are
10468 arithmetically valid, the latter being almost exclusively used.
10470 @item -mtoplevel-symbols
10471 @itemx -mno-toplevel-symbols
10472 @opindex mtoplevel-symbols
10473 @opindex mno-toplevel-symbols
10474 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10475 code can be used with the @code{PREFIX} assembly directive.
10479 Generate an executable in the ELF format, rather than the default
10480 @samp{mmo} format used by the @command{mmix} simulator.
10482 @item -mbranch-predict
10483 @itemx -mno-branch-predict
10484 @opindex mbranch-predict
10485 @opindex mno-branch-predict
10486 Use (do not use) the probable-branch instructions, when static branch
10487 prediction indicates a probable branch.
10489 @item -mbase-addresses
10490 @itemx -mno-base-addresses
10491 @opindex mbase-addresses
10492 @opindex mno-base-addresses
10493 Generate (do not generate) code that uses @emph{base addresses}. Using a
10494 base address automatically generates a request (handled by the assembler
10495 and the linker) for a constant to be set up in a global register. The
10496 register is used for one or more base address requests within the range 0
10497 to 255 from the value held in the register. The generally leads to short
10498 and fast code, but the number of different data items that can be
10499 addressed is limited. This means that a program that uses lots of static
10500 data may require @option{-mno-base-addresses}.
10502 @item -msingle-exit
10503 @itemx -mno-single-exit
10504 @opindex msingle-exit
10505 @opindex mno-single-exit
10506 Force (do not force) generated code to have a single exit point in each
10510 @node PDP-11 Options
10511 @subsection PDP-11 Options
10512 @cindex PDP-11 Options
10514 These options are defined for the PDP-11:
10519 Use hardware FPP floating point. This is the default. (FIS floating
10520 point on the PDP-11/40 is not supported.)
10523 @opindex msoft-float
10524 Do not use hardware floating point.
10528 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10532 Return floating-point results in memory. This is the default.
10536 Generate code for a PDP-11/40.
10540 Generate code for a PDP-11/45. This is the default.
10544 Generate code for a PDP-11/10.
10546 @item -mbcopy-builtin
10547 @opindex bcopy-builtin
10548 Use inline @code{movstrhi} patterns for copying memory. This is the
10553 Do not use inline @code{movstrhi} patterns for copying memory.
10559 Use 16-bit @code{int}. This is the default.
10565 Use 32-bit @code{int}.
10568 @itemx -mno-float32
10570 @opindex mno-float32
10571 Use 64-bit @code{float}. This is the default.
10576 @opindex mno-float64
10577 Use 32-bit @code{float}.
10581 Use @code{abshi2} pattern. This is the default.
10585 Do not use @code{abshi2} pattern.
10587 @item -mbranch-expensive
10588 @opindex mbranch-expensive
10589 Pretend that branches are expensive. This is for experimenting with
10590 code generation only.
10592 @item -mbranch-cheap
10593 @opindex mbranch-cheap
10594 Do not pretend that branches are expensive. This is the default.
10598 Generate code for a system with split I&D.
10602 Generate code for a system without split I&D. This is the default.
10606 Use Unix assembler syntax. This is the default when configured for
10607 @samp{pdp11-*-bsd}.
10611 Use DEC assembler syntax. This is the default when configured for any
10612 PDP-11 target other than @samp{pdp11-*-bsd}.
10615 @node Xstormy16 Options
10616 @subsection Xstormy16 Options
10617 @cindex Xstormy16 Options
10619 These options are defined for Xstormy16:
10624 Choose startup files and linker script suitable for the simulator.
10628 @subsection FRV Options
10629 @cindex FRV Options
10635 Only use the first 32 general purpose registers.
10640 Use all 64 general purpose registers.
10645 Use only the first 32 floating point registers.
10650 Use all 64 floating point registers
10653 @opindex mhard-float
10655 Use hardware instructions for floating point operations.
10658 @opindex msoft-float
10660 Use library routines for floating point operations.
10665 Dynamically allocate condition code registers.
10670 Do not try to dynamically allocate condition code registers, only
10671 use @code{icc0} and @code{fcc0}.
10676 Change ABI to use double word insns.
10681 Do not use double word instructions.
10686 Use floating point double instructions.
10689 @opindex mno-double
10691 Do not use floating point double instructions.
10696 Use media instructions.
10701 Do not use media instructions.
10706 Use multiply and add/subtract instructions.
10709 @opindex mno-muladd
10711 Do not use multiply and add/subtract instructions.
10713 @item -mlibrary-pic
10714 @opindex mlibrary-pic
10716 Enable PIC support for building libraries
10721 Use only the first four media accumulator registers.
10726 Use all eight media accumulator registers.
10731 Pack VLIW instructions.
10736 Do not pack VLIW instructions.
10739 @opindex mno-eflags
10741 Do not mark ABI switches in e_flags.
10744 @opindex mcond-move
10746 Enable the use of conditional-move instructions (default).
10748 This switch is mainly for debugging the compiler and will likely be removed
10749 in a future version.
10751 @item -mno-cond-move
10752 @opindex mno-cond-move
10754 Disable the use of conditional-move instructions.
10756 This switch is mainly for debugging the compiler and will likely be removed
10757 in a future version.
10762 Enable the use of conditional set instructions (default).
10764 This switch is mainly for debugging the compiler and will likely be removed
10765 in a future version.
10770 Disable the use of conditional set instructions.
10772 This switch is mainly for debugging the compiler and will likely be removed
10773 in a future version.
10776 @opindex mcond-exec
10778 Enable the use of conditional execution (default).
10780 This switch is mainly for debugging the compiler and will likely be removed
10781 in a future version.
10783 @item -mno-cond-exec
10784 @opindex mno-cond-exec
10786 Disable the use of conditional execution.
10788 This switch is mainly for debugging the compiler and will likely be removed
10789 in a future version.
10791 @item -mvliw-branch
10792 @opindex mvliw-branch
10794 Run a pass to pack branches into VLIW instructions (default).
10796 This switch is mainly for debugging the compiler and will likely be removed
10797 in a future version.
10799 @item -mno-vliw-branch
10800 @opindex mno-vliw-branch
10802 Do not run a pass to pack branches into VLIW instructions.
10804 This switch is mainly for debugging the compiler and will likely be removed
10805 in a future version.
10807 @item -mmulti-cond-exec
10808 @opindex mmulti-cond-exec
10810 Enable optimization of @code{&&} and @code{||} in conditional execution
10813 This switch is mainly for debugging the compiler and will likely be removed
10814 in a future version.
10816 @item -mno-multi-cond-exec
10817 @opindex mno-multi-cond-exec
10819 Disable optimization of @code{&&} and @code{||} in conditional execution.
10821 This switch is mainly for debugging the compiler and will likely be removed
10822 in a future version.
10824 @item -mnested-cond-exec
10825 @opindex mnested-cond-exec
10827 Enable nested conditional execution optimizations (default).
10829 This switch is mainly for debugging the compiler and will likely be removed
10830 in a future version.
10832 @item -mno-nested-cond-exec
10833 @opindex mno-nested-cond-exec
10835 Disable nested conditional execution optimizations.
10837 This switch is mainly for debugging the compiler and will likely be removed
10838 in a future version.
10840 @item -mtomcat-stats
10841 @opindex mtomcat-stats
10843 Cause gas to print out tomcat statistics.
10845 @item -mcpu=@var{cpu}
10848 Select the processor type for which to generate code. Possible values are
10849 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10854 @node Xtensa Options
10855 @subsection Xtensa Options
10856 @cindex Xtensa Options
10858 These options are supported for Xtensa targets:
10862 @itemx -mno-const16
10864 @opindex mno-const16
10865 Enable or disable use of @code{CONST16} instructions for loading
10866 constant values. The @code{CONST16} instruction is currently not a
10867 standard option from Tensilica. When enabled, @code{CONST16}
10868 instructions are always used in place of the standard @code{L32R}
10869 instructions. The use of @code{CONST16} is enabled by default only if
10870 the @code{L32R} instruction is not available.
10873 @itemx -mno-fused-madd
10874 @opindex mfused-madd
10875 @opindex mno-fused-madd
10876 Enable or disable use of fused multiply/add and multiply/subtract
10877 instructions in the floating-point option. This has no effect if the
10878 floating-point option is not also enabled. Disabling fused multiply/add
10879 and multiply/subtract instructions forces the compiler to use separate
10880 instructions for the multiply and add/subtract operations. This may be
10881 desirable in some cases where strict IEEE 754-compliant results are
10882 required: the fused multiply add/subtract instructions do not round the
10883 intermediate result, thereby producing results with @emph{more} bits of
10884 precision than specified by the IEEE standard. Disabling fused multiply
10885 add/subtract instructions also ensures that the program output is not
10886 sensitive to the compiler's ability to combine multiply and add/subtract
10889 @item -mtext-section-literals
10890 @itemx -mno-text-section-literals
10891 @opindex mtext-section-literals
10892 @opindex mno-text-section-literals
10893 Control the treatment of literal pools. The default is
10894 @option{-mno-text-section-literals}, which places literals in a separate
10895 section in the output file. This allows the literal pool to be placed
10896 in a data RAM/ROM, and it also allows the linker to combine literal
10897 pools from separate object files to remove redundant literals and
10898 improve code size. With @option{-mtext-section-literals}, the literals
10899 are interspersed in the text section in order to keep them as close as
10900 possible to their references. This may be necessary for large assembly
10903 @item -mtarget-align
10904 @itemx -mno-target-align
10905 @opindex mtarget-align
10906 @opindex mno-target-align
10907 When this option is enabled, GCC instructs the assembler to
10908 automatically align instructions to reduce branch penalties at the
10909 expense of some code density. The assembler attempts to widen density
10910 instructions to align branch targets and the instructions following call
10911 instructions. If there are not enough preceding safe density
10912 instructions to align a target, no widening will be performed. The
10913 default is @option{-mtarget-align}. These options do not affect the
10914 treatment of auto-aligned instructions like @code{LOOP}, which the
10915 assembler will always align, either by widening density instructions or
10916 by inserting no-op instructions.
10919 @itemx -mno-longcalls
10920 @opindex mlongcalls
10921 @opindex mno-longcalls
10922 When this option is enabled, GCC instructs the assembler to translate
10923 direct calls to indirect calls unless it can determine that the target
10924 of a direct call is in the range allowed by the call instruction. This
10925 translation typically occurs for calls to functions in other source
10926 files. Specifically, the assembler translates a direct @code{CALL}
10927 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10928 The default is @option{-mno-longcalls}. This option should be used in
10929 programs where the call target can potentially be out of range. This
10930 option is implemented in the assembler, not the compiler, so the
10931 assembly code generated by GCC will still show direct call
10932 instructions---look at the disassembled object code to see the actual
10933 instructions. Note that the assembler will use an indirect call for
10934 every cross-file call, not just those that really will be out of range.
10937 @node Code Gen Options
10938 @section Options for Code Generation Conventions
10939 @cindex code generation conventions
10940 @cindex options, code generation
10941 @cindex run-time options
10943 These machine-independent options control the interface conventions
10944 used in code generation.
10946 Most of them have both positive and negative forms; the negative form
10947 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10948 one of the forms is listed---the one which is not the default. You
10949 can figure out the other form by either removing @samp{no-} or adding
10953 @item -fbounds-check
10954 @opindex fbounds-check
10955 For front-ends that support it, generate additional code to check that
10956 indices used to access arrays are within the declared range. This is
10957 currently only supported by the Java and Fortran 77 front-ends, where
10958 this option defaults to true and false respectively.
10962 This option generates traps for signed overflow on addition, subtraction,
10963 multiplication operations.
10967 This option instructs the compiler to assume that signed arithmetic
10968 overflow of addition, subtraction and multiplication wraps around
10969 using twos-complement representation. This flag enables some optimizations
10970 and disables other. This option is enabled by default for the Java
10971 front-end, as required by the Java language specification.
10974 @opindex fexceptions
10975 Enable exception handling. Generates extra code needed to propagate
10976 exceptions. For some targets, this implies GCC will generate frame
10977 unwind information for all functions, which can produce significant data
10978 size overhead, although it does not affect execution. If you do not
10979 specify this option, GCC will enable it by default for languages like
10980 C++ which normally require exception handling, and disable it for
10981 languages like C that do not normally require it. However, you may need
10982 to enable this option when compiling C code that needs to interoperate
10983 properly with exception handlers written in C++. You may also wish to
10984 disable this option if you are compiling older C++ programs that don't
10985 use exception handling.
10987 @item -fnon-call-exceptions
10988 @opindex fnon-call-exceptions
10989 Generate code that allows trapping instructions to throw exceptions.
10990 Note that this requires platform-specific runtime support that does
10991 not exist everywhere. Moreover, it only allows @emph{trapping}
10992 instructions to throw exceptions, i.e.@: memory references or floating
10993 point instructions. It does not allow exceptions to be thrown from
10994 arbitrary signal handlers such as @code{SIGALRM}.
10996 @item -funwind-tables
10997 @opindex funwind-tables
10998 Similar to @option{-fexceptions}, except that it will just generate any needed
10999 static data, but will not affect the generated code in any other way.
11000 You will normally not enable this option; instead, a language processor
11001 that needs this handling would enable it on your behalf.
11003 @item -fasynchronous-unwind-tables
11004 @opindex funwind-tables
11005 Generate unwind table in dwarf2 format, if supported by target machine. The
11006 table is exact at each instruction boundary, so it can be used for stack
11007 unwinding from asynchronous events (such as debugger or garbage collector).
11009 @item -fpcc-struct-return
11010 @opindex fpcc-struct-return
11011 Return ``short'' @code{struct} and @code{union} values in memory like
11012 longer ones, rather than in registers. This convention is less
11013 efficient, but it has the advantage of allowing intercallability between
11014 GCC-compiled files and files compiled with other compilers, particularly
11015 the Portable C Compiler (pcc).
11017 The precise convention for returning structures in memory depends
11018 on the target configuration macros.
11020 Short structures and unions are those whose size and alignment match
11021 that of some integer type.
11023 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11024 switch is not binary compatible with code compiled with the
11025 @option{-freg-struct-return} switch.
11026 Use it to conform to a non-default application binary interface.
11028 @item -freg-struct-return
11029 @opindex freg-struct-return
11030 Return @code{struct} and @code{union} values in registers when possible.
11031 This is more efficient for small structures than
11032 @option{-fpcc-struct-return}.
11034 If you specify neither @option{-fpcc-struct-return} nor
11035 @option{-freg-struct-return}, GCC defaults to whichever convention is
11036 standard for the target. If there is no standard convention, GCC
11037 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11038 the principal compiler. In those cases, we can choose the standard, and
11039 we chose the more efficient register return alternative.
11041 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11042 switch is not binary compatible with code compiled with the
11043 @option{-fpcc-struct-return} switch.
11044 Use it to conform to a non-default application binary interface.
11046 @item -fshort-enums
11047 @opindex fshort-enums
11048 Allocate to an @code{enum} type only as many bytes as it needs for the
11049 declared range of possible values. Specifically, the @code{enum} type
11050 will be equivalent to the smallest integer type which has enough room.
11052 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11053 code that is not binary compatible with code generated without that switch.
11054 Use it to conform to a non-default application binary interface.
11056 @item -fshort-double
11057 @opindex fshort-double
11058 Use the same size for @code{double} as for @code{float}.
11060 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11061 code that is not binary compatible with code generated without that switch.
11062 Use it to conform to a non-default application binary interface.
11064 @item -fshort-wchar
11065 @opindex fshort-wchar
11066 Override the underlying type for @samp{wchar_t} to be @samp{short
11067 unsigned int} instead of the default for the target. This option is
11068 useful for building programs to run under WINE@.
11070 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11071 code that is not binary compatible with code generated without that switch.
11072 Use it to conform to a non-default application binary interface.
11074 @item -fshared-data
11075 @opindex fshared-data
11076 Requests that the data and non-@code{const} variables of this
11077 compilation be shared data rather than private data. The distinction
11078 makes sense only on certain operating systems, where shared data is
11079 shared between processes running the same program, while private data
11080 exists in one copy per process.
11083 @opindex fno-common
11084 In C, allocate even uninitialized global variables in the data section of the
11085 object file, rather than generating them as common blocks. This has the
11086 effect that if the same variable is declared (without @code{extern}) in
11087 two different compilations, you will get an error when you link them.
11088 The only reason this might be useful is if you wish to verify that the
11089 program will work on other systems which always work this way.
11093 Ignore the @samp{#ident} directive.
11095 @item -fno-gnu-linker
11096 @opindex fno-gnu-linker
11097 Do not output global initializations (such as C++ constructors and
11098 destructors) in the form used by the GNU linker (on systems where the GNU
11099 linker is the standard method of handling them). Use this option when
11100 you want to use a non-GNU linker, which also requires using the
11101 @command{collect2} program to make sure the system linker includes
11102 constructors and destructors. (@command{collect2} is included in the GCC
11103 distribution.) For systems which @emph{must} use @command{collect2}, the
11104 compiler driver @command{gcc} is configured to do this automatically.
11106 @item -finhibit-size-directive
11107 @opindex finhibit-size-directive
11108 Don't output a @code{.size} assembler directive, or anything else that
11109 would cause trouble if the function is split in the middle, and the
11110 two halves are placed at locations far apart in memory. This option is
11111 used when compiling @file{crtstuff.c}; you should not need to use it
11114 @item -fverbose-asm
11115 @opindex fverbose-asm
11116 Put extra commentary information in the generated assembly code to
11117 make it more readable. This option is generally only of use to those
11118 who actually need to read the generated assembly code (perhaps while
11119 debugging the compiler itself).
11121 @option{-fno-verbose-asm}, the default, causes the
11122 extra information to be omitted and is useful when comparing two assembler
11127 @cindex global offset table
11129 Generate position-independent code (PIC) suitable for use in a shared
11130 library, if supported for the target machine. Such code accesses all
11131 constant addresses through a global offset table (GOT)@. The dynamic
11132 loader resolves the GOT entries when the program starts (the dynamic
11133 loader is not part of GCC; it is part of the operating system). If
11134 the GOT size for the linked executable exceeds a machine-specific
11135 maximum size, you get an error message from the linker indicating that
11136 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11137 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
11138 on the m68k and RS/6000. The 386 has no such limit.)
11140 Position-independent code requires special support, and therefore works
11141 only on certain machines. For the 386, GCC supports PIC for System V
11142 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11143 position-independent.
11147 If supported for the target machine, emit position-independent code,
11148 suitable for dynamic linking and avoiding any limit on the size of the
11149 global offset table. This option makes a difference on the m68k, m88k,
11152 Position-independent code requires special support, and therefore works
11153 only on certain machines.
11159 These options are similar to @option{-fpic} and @option{-fPIC}, but
11160 generated position independent code can be only linked into executables.
11161 Usually these options are used when @option{-pie} GCC option will be
11162 used during linking.
11164 @item -ffixed-@var{reg}
11166 Treat the register named @var{reg} as a fixed register; generated code
11167 should never refer to it (except perhaps as a stack pointer, frame
11168 pointer or in some other fixed role).
11170 @var{reg} must be the name of a register. The register names accepted
11171 are machine-specific and are defined in the @code{REGISTER_NAMES}
11172 macro in the machine description macro file.
11174 This flag does not have a negative form, because it specifies a
11177 @item -fcall-used-@var{reg}
11178 @opindex fcall-used
11179 Treat the register named @var{reg} as an allocable register that is
11180 clobbered by function calls. It may be allocated for temporaries or
11181 variables that do not live across a call. Functions compiled this way
11182 will not save and restore the register @var{reg}.
11184 It is an error to used this flag with the frame pointer or stack pointer.
11185 Use of this flag for other registers that have fixed pervasive roles in
11186 the machine's execution model will produce disastrous results.
11188 This flag does not have a negative form, because it specifies a
11191 @item -fcall-saved-@var{reg}
11192 @opindex fcall-saved
11193 Treat the register named @var{reg} as an allocable register saved by
11194 functions. It may be allocated even for temporaries or variables that
11195 live across a call. Functions compiled this way will save and restore
11196 the register @var{reg} if they use it.
11198 It is an error to used this flag with the frame pointer or stack pointer.
11199 Use of this flag for other registers that have fixed pervasive roles in
11200 the machine's execution model will produce disastrous results.
11202 A different sort of disaster will result from the use of this flag for
11203 a register in which function values may be returned.
11205 This flag does not have a negative form, because it specifies a
11208 @item -fpack-struct
11209 @opindex fpack-struct
11210 Pack all structure members together without holes.
11212 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11213 code that is not binary compatible with code generated without that switch.
11214 Additionally, it makes the code suboptimal.
11215 Use it to conform to a non-default application binary interface.
11217 @item -finstrument-functions
11218 @opindex finstrument-functions
11219 Generate instrumentation calls for entry and exit to functions. Just
11220 after function entry and just before function exit, the following
11221 profiling functions will be called with the address of the current
11222 function and its call site. (On some platforms,
11223 @code{__builtin_return_address} does not work beyond the current
11224 function, so the call site information may not be available to the
11225 profiling functions otherwise.)
11228 void __cyg_profile_func_enter (void *this_fn,
11230 void __cyg_profile_func_exit (void *this_fn,
11234 The first argument is the address of the start of the current function,
11235 which may be looked up exactly in the symbol table.
11237 This instrumentation is also done for functions expanded inline in other
11238 functions. The profiling calls will indicate where, conceptually, the
11239 inline function is entered and exited. This means that addressable
11240 versions of such functions must be available. If all your uses of a
11241 function are expanded inline, this may mean an additional expansion of
11242 code size. If you use @samp{extern inline} in your C code, an
11243 addressable version of such functions must be provided. (This is
11244 normally the case anyways, but if you get lucky and the optimizer always
11245 expands the functions inline, you might have gotten away without
11246 providing static copies.)
11248 A function may be given the attribute @code{no_instrument_function}, in
11249 which case this instrumentation will not be done. This can be used, for
11250 example, for the profiling functions listed above, high-priority
11251 interrupt routines, and any functions from which the profiling functions
11252 cannot safely be called (perhaps signal handlers, if the profiling
11253 routines generate output or allocate memory).
11255 @item -fstack-check
11256 @opindex fstack-check
11257 Generate code to verify that you do not go beyond the boundary of the
11258 stack. You should specify this flag if you are running in an
11259 environment with multiple threads, but only rarely need to specify it in
11260 a single-threaded environment since stack overflow is automatically
11261 detected on nearly all systems if there is only one stack.
11263 Note that this switch does not actually cause checking to be done; the
11264 operating system must do that. The switch causes generation of code
11265 to ensure that the operating system sees the stack being extended.
11267 @item -fstack-limit-register=@var{reg}
11268 @itemx -fstack-limit-symbol=@var{sym}
11269 @itemx -fno-stack-limit
11270 @opindex fstack-limit-register
11271 @opindex fstack-limit-symbol
11272 @opindex fno-stack-limit
11273 Generate code to ensure that the stack does not grow beyond a certain value,
11274 either the value of a register or the address of a symbol. If the stack
11275 would grow beyond the value, a signal is raised. For most targets,
11276 the signal is raised before the stack overruns the boundary, so
11277 it is possible to catch the signal without taking special precautions.
11279 For instance, if the stack starts at absolute address @samp{0x80000000}
11280 and grows downwards, you can use the flags
11281 @option{-fstack-limit-symbol=__stack_limit} and
11282 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11283 of 128KB@. Note that this may only work with the GNU linker.
11285 @cindex aliasing of parameters
11286 @cindex parameters, aliased
11287 @item -fargument-alias
11288 @itemx -fargument-noalias
11289 @itemx -fargument-noalias-global
11290 @opindex fargument-alias
11291 @opindex fargument-noalias
11292 @opindex fargument-noalias-global
11293 Specify the possible relationships among parameters and between
11294 parameters and global data.
11296 @option{-fargument-alias} specifies that arguments (parameters) may
11297 alias each other and may alias global storage.@*
11298 @option{-fargument-noalias} specifies that arguments do not alias
11299 each other, but may alias global storage.@*
11300 @option{-fargument-noalias-global} specifies that arguments do not
11301 alias each other and do not alias global storage.
11303 Each language will automatically use whatever option is required by
11304 the language standard. You should not need to use these options yourself.
11306 @item -fleading-underscore
11307 @opindex fleading-underscore
11308 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11309 change the way C symbols are represented in the object file. One use
11310 is to help link with legacy assembly code.
11312 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11313 generate code that is not binary compatible with code generated without that
11314 switch. Use it to conform to a non-default application binary interface.
11315 Not all targets provide complete support for this switch.
11317 @item -ftls-model=@var{model}
11318 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11319 The @var{model} argument should be one of @code{global-dynamic},
11320 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11322 The default without @option{-fpic} is @code{initial-exec}; with
11323 @option{-fpic} the default is @code{global-dynamic}.
11328 @node Environment Variables
11329 @section Environment Variables Affecting GCC
11330 @cindex environment variables
11332 @c man begin ENVIRONMENT
11333 This section describes several environment variables that affect how GCC
11334 operates. Some of them work by specifying directories or prefixes to use
11335 when searching for various kinds of files. Some are used to specify other
11336 aspects of the compilation environment.
11338 Note that you can also specify places to search using options such as
11339 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11340 take precedence over places specified using environment variables, which
11341 in turn take precedence over those specified by the configuration of GCC@.
11342 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11343 GNU Compiler Collection (GCC) Internals}.
11348 @c @itemx LC_COLLATE
11350 @c @itemx LC_MONETARY
11351 @c @itemx LC_NUMERIC
11356 @c @findex LC_COLLATE
11357 @findex LC_MESSAGES
11358 @c @findex LC_MONETARY
11359 @c @findex LC_NUMERIC
11363 These environment variables control the way that GCC uses
11364 localization information that allow GCC to work with different
11365 national conventions. GCC inspects the locale categories
11366 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11367 so. These locale categories can be set to any value supported by your
11368 installation. A typical value is @samp{en_UK} for English in the United
11371 The @env{LC_CTYPE} environment variable specifies character
11372 classification. GCC uses it to determine the character boundaries in
11373 a string; this is needed for some multibyte encodings that contain quote
11374 and escape characters that would otherwise be interpreted as a string
11377 The @env{LC_MESSAGES} environment variable specifies the language to
11378 use in diagnostic messages.
11380 If the @env{LC_ALL} environment variable is set, it overrides the value
11381 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11382 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11383 environment variable. If none of these variables are set, GCC
11384 defaults to traditional C English behavior.
11388 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11389 files. GCC uses temporary files to hold the output of one stage of
11390 compilation which is to be used as input to the next stage: for example,
11391 the output of the preprocessor, which is the input to the compiler
11394 @item GCC_EXEC_PREFIX
11395 @findex GCC_EXEC_PREFIX
11396 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11397 names of the subprograms executed by the compiler. No slash is added
11398 when this prefix is combined with the name of a subprogram, but you can
11399 specify a prefix that ends with a slash if you wish.
11401 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11402 an appropriate prefix to use based on the pathname it was invoked with.
11404 If GCC cannot find the subprogram using the specified prefix, it
11405 tries looking in the usual places for the subprogram.
11407 The default value of @env{GCC_EXEC_PREFIX} is
11408 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11409 of @code{prefix} when you ran the @file{configure} script.
11411 Other prefixes specified with @option{-B} take precedence over this prefix.
11413 This prefix is also used for finding files such as @file{crt0.o} that are
11416 In addition, the prefix is used in an unusual way in finding the
11417 directories to search for header files. For each of the standard
11418 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11419 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11420 replacing that beginning with the specified prefix to produce an
11421 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11422 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11423 These alternate directories are searched first; the standard directories
11426 @item COMPILER_PATH
11427 @findex COMPILER_PATH
11428 The value of @env{COMPILER_PATH} is a colon-separated list of
11429 directories, much like @env{PATH}. GCC tries the directories thus
11430 specified when searching for subprograms, if it can't find the
11431 subprograms using @env{GCC_EXEC_PREFIX}.
11434 @findex LIBRARY_PATH
11435 The value of @env{LIBRARY_PATH} is a colon-separated list of
11436 directories, much like @env{PATH}. When configured as a native compiler,
11437 GCC tries the directories thus specified when searching for special
11438 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11439 using GCC also uses these directories when searching for ordinary
11440 libraries for the @option{-l} option (but directories specified with
11441 @option{-L} come first).
11445 @cindex locale definition
11446 This variable is used to pass locale information to the compiler. One way in
11447 which this information is used is to determine the character set to be used
11448 when character literals, string literals and comments are parsed in C and C++.
11449 When the compiler is configured to allow multibyte characters,
11450 the following values for @env{LANG} are recognized:
11454 Recognize JIS characters.
11456 Recognize SJIS characters.
11458 Recognize EUCJP characters.
11461 If @env{LANG} is not defined, or if it has some other value, then the
11462 compiler will use mblen and mbtowc as defined by the default locale to
11463 recognize and translate multibyte characters.
11467 Some additional environments variables affect the behavior of the
11470 @include cppenv.texi
11474 @node Precompiled Headers
11475 @section Using Precompiled Headers
11476 @cindex precompiled headers
11477 @cindex speed of compilation
11479 Often large projects have many header files that are included in every
11480 source file. The time the compiler takes to process these header files
11481 over and over again can account for nearly all of the time required to
11482 build the project. To make builds faster, GCC allows users to
11483 `precompile' a header file; then, if builds can use the precompiled
11484 header file they will be much faster.
11486 To create a precompiled header file, simply compile it as you would any
11487 other file, if necessary using the @option{-x} option to make the driver
11488 treat it as a C or C++ header file. You will probably want to use a
11489 tool like @command{make} to keep the precompiled header up-to-date when
11490 the headers it contains change.
11492 A precompiled header file will be searched for when @code{#include} is
11493 seen in the compilation. As it searches for the included file
11494 (@pxref{Search Path,,Search Path,cpp.info,The C Preprocessor}) the
11495 compiler looks for a precompiled header in each directory just before it
11496 looks for the include file in that directory. The name searched for is
11497 the name specified in the @code{#include} with @samp{.gch} appended. If
11498 the precompiled header file can't be used, it is ignored.
11500 For instance, if you have @code{#include "all.h"}, and you have
11501 @file{all.h.gch} in the same directory as @file{all.h}, then the
11502 precompiled header file will be used if possible, and the original
11503 header will be used otherwise.
11505 Alternatively, you might decide to put the precompiled header file in a
11506 directory and use @option{-I} to ensure that directory is searched
11507 before (or instead of) the directory containing the original header.
11508 Then, if you want to check that the precompiled header file is always
11509 used, you can put a file of the same name as the original header in this
11510 directory containing an @code{#error} command.
11512 This also works with @option{-include}. So yet another way to use
11513 precompiled headers, good for projects not designed with precompiled
11514 header files in mind, is to simply take most of the header files used by
11515 a project, include them from another header file, precompile that header
11516 file, and @option{-include} the precompiled header. If the header files
11517 have guards against multiple inclusion, they will be skipped because
11518 they've already been included (in the precompiled header).
11520 If you need to precompile the same header file for different
11521 languages, targets, or compiler options, you can instead make a
11522 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11523 header in the directory. (It doesn't matter what you call the files
11524 in the directory, every precompiled header in the directory will be
11525 considered.) The first precompiled header encountered in the
11526 directory that is valid for this compilation will be used; they're
11527 searched in no particular order.
11529 There are many other possibilities, limited only by your imagination,
11530 good sense, and the constraints of your build system.
11532 A precompiled header file can be used only when these conditions apply:
11536 Only one precompiled header can be used in a particular compilation.
11538 A precompiled header can't be used once the first C token is seen. You
11539 can have preprocessor directives before a precompiled header; you can
11540 even include a precompiled header from inside another header, so long as
11541 there are no C tokens before the @code{#include}.
11543 The precompiled header file must be produced for the same language as
11544 the current compilation. You can't use a C precompiled header for a C++
11547 The precompiled header file must be produced by the same compiler
11548 version and configuration as the current compilation is using.
11549 The easiest way to guarantee this is to use the same compiler binary
11550 for creating and using precompiled headers.
11552 Any macros defined before the precompiled header (including with
11553 @option{-D}) must either be defined in the same way as when the
11554 precompiled header was generated, or must not affect the precompiled
11555 header, which usually means that the they don't appear in the
11556 precompiled header at all.
11558 Certain command-line options must be defined in the same way as when the
11559 precompiled header was generated. At present, it's not clear which
11560 options are safe to change and which are not; the safest choice is to
11561 use exactly the same options when generating and using the precompiled
11565 For all of these but the last, the compiler will automatically ignore
11566 the precompiled header if the conditions aren't met. For the last item,
11567 some option changes will cause the precompiled header to be rejected,
11568 but not all incompatible option combinations have yet been found. If
11569 you find a new incompatible combination, please consider filing a bug
11570 report, see @ref{Bugs}.
11572 @node Running Protoize
11573 @section Running Protoize
11575 The program @code{protoize} is an optional part of GCC@. You can use
11576 it to add prototypes to a program, thus converting the program to ISO
11577 C in one respect. The companion program @code{unprotoize} does the
11578 reverse: it removes argument types from any prototypes that are found.
11580 When you run these programs, you must specify a set of source files as
11581 command line arguments. The conversion programs start out by compiling
11582 these files to see what functions they define. The information gathered
11583 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11585 After scanning comes actual conversion. The specified files are all
11586 eligible to be converted; any files they include (whether sources or
11587 just headers) are eligible as well.
11589 But not all the eligible files are converted. By default,
11590 @code{protoize} and @code{unprotoize} convert only source and header
11591 files in the current directory. You can specify additional directories
11592 whose files should be converted with the @option{-d @var{directory}}
11593 option. You can also specify particular files to exclude with the
11594 @option{-x @var{file}} option. A file is converted if it is eligible, its
11595 directory name matches one of the specified directory names, and its
11596 name within the directory has not been excluded.
11598 Basic conversion with @code{protoize} consists of rewriting most
11599 function definitions and function declarations to specify the types of
11600 the arguments. The only ones not rewritten are those for varargs
11603 @code{protoize} optionally inserts prototype declarations at the
11604 beginning of the source file, to make them available for any calls that
11605 precede the function's definition. Or it can insert prototype
11606 declarations with block scope in the blocks where undeclared functions
11609 Basic conversion with @code{unprotoize} consists of rewriting most
11610 function declarations to remove any argument types, and rewriting
11611 function definitions to the old-style pre-ISO form.
11613 Both conversion programs print a warning for any function declaration or
11614 definition that they can't convert. You can suppress these warnings
11617 The output from @code{protoize} or @code{unprotoize} replaces the
11618 original source file. The original file is renamed to a name ending
11619 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11620 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11621 for DOS) file already exists, then the source file is simply discarded.
11623 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11624 scan the program and collect information about the functions it uses.
11625 So neither of these programs will work until GCC is installed.
11627 Here is a table of the options you can use with @code{protoize} and
11628 @code{unprotoize}. Each option works with both programs unless
11632 @item -B @var{directory}
11633 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11634 usual directory (normally @file{/usr/local/lib}). This file contains
11635 prototype information about standard system functions. This option
11636 applies only to @code{protoize}.
11638 @item -c @var{compilation-options}
11639 Use @var{compilation-options} as the options when running @command{gcc} to
11640 produce the @samp{.X} files. The special option @option{-aux-info} is
11641 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11643 Note that the compilation options must be given as a single argument to
11644 @code{protoize} or @code{unprotoize}. If you want to specify several
11645 @command{gcc} options, you must quote the entire set of compilation options
11646 to make them a single word in the shell.
11648 There are certain @command{gcc} arguments that you cannot use, because they
11649 would produce the wrong kind of output. These include @option{-g},
11650 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11651 the @var{compilation-options}, they are ignored.
11654 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11655 systems) instead of @samp{.c}. This is convenient if you are converting
11656 a C program to C++. This option applies only to @code{protoize}.
11659 Add explicit global declarations. This means inserting explicit
11660 declarations at the beginning of each source file for each function
11661 that is called in the file and was not declared. These declarations
11662 precede the first function definition that contains a call to an
11663 undeclared function. This option applies only to @code{protoize}.
11665 @item -i @var{string}
11666 Indent old-style parameter declarations with the string @var{string}.
11667 This option applies only to @code{protoize}.
11669 @code{unprotoize} converts prototyped function definitions to old-style
11670 function definitions, where the arguments are declared between the
11671 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11672 uses five spaces as the indentation. If you want to indent with just
11673 one space instead, use @option{-i " "}.
11676 Keep the @samp{.X} files. Normally, they are deleted after conversion
11680 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11681 a prototype declaration for each function in each block which calls the
11682 function without any declaration. This option applies only to
11686 Make no real changes. This mode just prints information about the conversions
11687 that would have been done without @option{-n}.
11690 Make no @samp{.save} files. The original files are simply deleted.
11691 Use this option with caution.
11693 @item -p @var{program}
11694 Use the program @var{program} as the compiler. Normally, the name
11695 @file{gcc} is used.
11698 Work quietly. Most warnings are suppressed.
11701 Print the version number, just like @option{-v} for @command{gcc}.
11704 If you need special compiler options to compile one of your program's
11705 source files, then you should generate that file's @samp{.X} file
11706 specially, by running @command{gcc} on that source file with the
11707 appropriate options and the option @option{-aux-info}. Then run
11708 @code{protoize} on the entire set of files. @code{protoize} will use
11709 the existing @samp{.X} file because it is newer than the source file.
11713 gcc -Dfoo=bar file1.c -aux-info file1.X
11718 You need to include the special files along with the rest in the
11719 @code{protoize} command, even though their @samp{.X} files already
11720 exist, because otherwise they won't get converted.
11722 @xref{Protoize Caveats}, for more information on how to use
11723 @code{protoize} successfully.