1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004 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, 2004 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), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @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 and Objective-C++ Dialect Options:: Variations on Objective-C
126 * Language Independent Options:: Controlling how diagnostics should be
128 * Warning Options:: How picky should the compiler be?
129 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
130 * Optimize Options:: How much optimization?
131 * Preprocessor Options:: Controlling header files and macro definitions.
132 Also, getting dependency information for Make.
133 * Assembler Options:: Passing options to the assembler.
134 * Link Options:: Specifying libraries and so on.
135 * Directory Options:: Where to find header files and libraries.
136 Where to find the compiler executable files.
137 * Spec Files:: How to pass switches to sub-processes.
138 * Target Options:: Running a cross-compiler, or an old version of GCC.
139 * Submodel Options:: Specifying minor hardware or convention variations,
140 such as 68010 vs 68020.
141 * Code Gen Options:: Specifying conventions for function calls, data layout
143 * Environment Variables:: Env vars that affect GCC.
144 * Precompiled Headers:: Compiling a header once, and using it many times.
145 * Running Protoize:: Automatically adding or removing function prototypes.
151 @section Option Summary
153 Here is a summary of all the options, grouped by type. Explanations are
154 in the following sections.
157 @item Overall Options
158 @xref{Overall Options,,Options Controlling the Kind of Output}.
159 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
160 -x @var{language} -v -### --help --target-help --version}
162 @item C Language Options
163 @xref{C Dialect Options,,Options Controlling C Dialect}.
164 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
165 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
166 -fhosted -ffreestanding -fms-extensions @gol
167 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
168 -fallow-single-precision -fcond-mismatch @gol
169 -fsigned-bitfields -fsigned-char @gol
170 -funsigned-bitfields -funsigned-char}
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 @gol
178 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
179 -fno-implicit-templates @gol
180 -fno-implicit-inline-templates @gol
181 -fno-implement-inlines -fms-extensions @gol
182 -fno-nonansi-builtins -fno-operator-names @gol
183 -fno-optional-diags -fpermissive @gol
184 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
185 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
186 -fno-default-inline -fvisibility-inlines-hidden @gol
187 -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 and Objective-C++ Language Options
195 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
196 Objective-C and Objective-C++ Dialects}.
198 -fconstant-string-class=@var{class-name} @gol
199 -fgnu-runtime -fnext-runtime @gol
200 -fno-nil-receivers @gol
201 -fobjc-exceptions @gol
202 -freplace-objc-classes @gol
205 -Wno-protocol -Wselector -Wundeclared-selector}
207 @item Language Independent Options
208 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
209 @gccoptlist{-fmessage-length=@var{n} @gol
210 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
212 @item Warning Options
213 @xref{Warning Options,,Options to Request or Suppress Warnings}.
214 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
215 -w -Wextra -Wall -Waggregate-return @gol
216 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
217 -Wconversion -Wno-deprecated-declarations @gol
218 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
219 -Werror -Werror-implicit-function-declaration @gol
220 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
221 -Wno-format-extra-args -Wformat-nonliteral @gol
222 -Wformat-security -Wformat-y2k @gol
223 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
224 -Wimport -Wno-import -Winit-self -Winline @gol
225 -Wno-invalid-offsetof -Winvalid-pch @gol
226 -Wlarger-than-@var{len} -Wlong-long @gol
227 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
228 -Wmissing-format-attribute -Wmissing-include-dirs @gol
229 -Wmissing-noreturn @gol
230 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
231 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
232 -Wreturn-type -Wsequence-point -Wshadow @gol
233 -Wsign-compare -Wstrict-aliasing -Wstrict-aliasing=2 @gol
234 -Wswitch -Wswitch-default -Wswitch-enum @gol
235 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
236 -Wunknown-pragmas -Wunreachable-code @gol
237 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
238 -Wunused-value -Wunused-variable -Wwrite-strings @gol
241 @item C-only Warning Options
242 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
243 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
244 -Wstrict-prototypes -Wtraditional @gol
245 -Wdeclaration-after-statement}
247 @item Debugging Options
248 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
249 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
250 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
251 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
253 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
254 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
255 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
256 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
258 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
259 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
260 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
261 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
262 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
263 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
264 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
265 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
266 -fdump-tree-nrv -fdump-tree-vect @gol
267 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
268 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
269 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
270 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs -ftree-based-profiling @gol
271 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
272 -ftest-coverage -ftime-report -fvar-tracking @gol
273 -g -g@var{level} -gcoff -gdwarf-2 @gol
274 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
275 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
276 -print-multi-directory -print-multi-lib @gol
277 -print-prog-name=@var{program} -print-search-dirs -Q @gol
280 @item Optimization Options
281 @xref{Optimize Options,,Options that Control Optimization}.
282 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
283 -falign-labels=@var{n} -falign-loops=@var{n} @gol
284 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
285 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
286 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
287 -fcaller-saves -fcprop-registers @gol
288 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
289 -fdelayed-branch -fdelete-null-pointer-checks @gol
290 -fexpensive-optimizations -ffast-math -ffloat-store @gol
291 -fforce-addr -fforce-mem -ffunction-sections @gol
292 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
293 -floop-optimize -fcrossjumping -fif-conversion -fif-conversion2 @gol
294 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
295 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
296 -fmodulo-sched -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
297 -fno-default-inline -fno-defer-pop -floop-optimize2 -fmove-loop-invariants @gol
298 -fno-function-cse -fno-guess-branch-probability @gol
299 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
300 -funsafe-math-optimizations -ffinite-math-only @gol
301 -fno-trapping-math -fno-zero-initialized-in-bss @gol
302 -fomit-frame-pointer -foptimize-register-move @gol
303 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
304 -fprofile-generate -fprofile-use @gol
305 -freduce-all-givs -fregmove -frename-registers @gol
306 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
307 -frerun-cse-after-loop -frerun-loop-opt @gol
308 -frounding-math -fschedule-insns -fschedule-insns2 @gol
309 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
310 -fsched-spec-load-dangerous @gol
311 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
312 -fsched2-use-superblocks @gol
313 -fsched2-use-traces -freschedule-modulo-scheduled-loops @gol
314 -fsignaling-nans -fsingle-precision-constant -fspeculative-prefetching @gol
315 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
316 -funroll-all-loops -funroll-loops -fpeel-loops @gol
317 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
318 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
319 -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
320 -ftree-dominator-opts -ftree-dse -ftree-copyrename @gol
321 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
322 --param @var{name}=@var{value}
323 -O -O0 -O1 -O2 -O3 -Os}
325 @item Preprocessor Options
326 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
327 @gccoptlist{-A@var{question}=@var{answer} @gol
328 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
329 -C -dD -dI -dM -dN @gol
330 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
331 -idirafter @var{dir} @gol
332 -include @var{file} -imacros @var{file} @gol
333 -iprefix @var{file} -iwithprefix @var{dir} @gol
334 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
335 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
336 -P -fworking-directory -remap @gol
337 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
338 -Xpreprocessor @var{option}}
340 @item Assembler Option
341 @xref{Assembler Options,,Passing Options to the Assembler}.
342 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
345 @xref{Link Options,,Options for Linking}.
346 @gccoptlist{@var{object-file-name} -l@var{library} @gol
347 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
348 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
349 -Wl,@var{option} -Xlinker @var{option} @gol
352 @item Directory Options
353 @xref{Directory Options,,Options for Directory Search}.
354 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-}
357 @c I wrote this xref this way to avoid overfull hbox. -- rms
358 @xref{Target Options}.
359 @gccoptlist{-V @var{version} -b @var{machine}}
361 @item Machine Dependent Options
362 @xref{Submodel Options,,Hardware Models and Configurations}.
363 @c This list is ordered alphanumerically by subsection name.
364 @c Try and put the significant identifier (CPU or system) first,
365 @c so users have a clue at guessing where the ones they want will be.
368 @gccoptlist{-EB -EL @gol
369 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
370 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
373 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
374 -mabi=@var{name} @gol
375 -mapcs-stack-check -mno-apcs-stack-check @gol
376 -mapcs-float -mno-apcs-float @gol
377 -mapcs-reentrant -mno-apcs-reentrant @gol
378 -msched-prolog -mno-sched-prolog @gol
379 -mlittle-endian -mbig-endian -mwords-little-endian @gol
380 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
381 -mthumb-interwork -mno-thumb-interwork @gol
382 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
383 -mstructure-size-boundary=@var{n} @gol
384 -mabort-on-noreturn @gol
385 -mlong-calls -mno-long-calls @gol
386 -msingle-pic-base -mno-single-pic-base @gol
387 -mpic-register=@var{reg} @gol
388 -mnop-fun-dllimport @gol
389 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
390 -mpoke-function-name @gol
392 -mtpcs-frame -mtpcs-leaf-frame @gol
393 -mcaller-super-interworking -mcallee-super-interworking}
396 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
397 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
400 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
401 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
402 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
403 -mstack-align -mdata-align -mconst-align @gol
404 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
405 -melf -maout -melinux -mlinux -sim -sim2 @gol
406 -mmul-bug-workaround -mno-mul-bug-workaround}
408 @emph{Darwin Options}
409 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
410 -arch_only -bind_at_load -bundle -bundle_loader @gol
411 -client_name -compatibility_version -current_version @gol
413 -dependency-file -dylib_file -dylinker_install_name @gol
414 -dynamic -dynamiclib -exported_symbols_list @gol
415 -filelist -flat_namespace -force_cpusubtype_ALL @gol
416 -force_flat_namespace -headerpad_max_install_names @gol
417 -image_base -init -install_name -keep_private_externs @gol
418 -multi_module -multiply_defined -multiply_defined_unused @gol
419 -noall_load -no_dead_strip_inits_and_terms @gol
420 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
421 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
422 -private_bundle -read_only_relocs -sectalign @gol
423 -sectobjectsymbols -whyload -seg1addr @gol
424 -sectcreate -sectobjectsymbols -sectorder @gol
425 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
426 -segprot -segs_read_only_addr -segs_read_write_addr @gol
427 -single_module -static -sub_library -sub_umbrella @gol
428 -twolevel_namespace -umbrella -undefined @gol
429 -unexported_symbols_list -weak_reference_mismatches @gol
430 -whatsloaded -F -gused -gfull -mone-byte-bool}
432 @emph{DEC Alpha Options}
433 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
434 -mieee -mieee-with-inexact -mieee-conformant @gol
435 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
436 -mtrap-precision=@var{mode} -mbuild-constants @gol
437 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
438 -mbwx -mmax -mfix -mcix @gol
439 -mfloat-vax -mfloat-ieee @gol
440 -mexplicit-relocs -msmall-data -mlarge-data @gol
441 -msmall-text -mlarge-text @gol
442 -mmemory-latency=@var{time}}
444 @emph{DEC Alpha/VMS Options}
445 @gccoptlist{-mvms-return-codes}
448 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
449 -mhard-float -msoft-float @gol
450 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
451 -mdouble -mno-double @gol
452 -mmedia -mno-media -mmuladd -mno-muladd @gol
453 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic -mlinked-fp @gol
454 -mlibrary-pic -macc-4 -macc-8 @gol
455 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
456 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
457 -mvliw-branch -mno-vliw-branch @gol
458 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
459 -mno-nested-cond-exec -mtomcat-stats @gol
462 @emph{H8/300 Options}
463 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
466 @gccoptlist{-march=@var{architecture-type} @gol
467 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
468 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
469 -mfixed-range=@var{register-range} @gol
470 -mjump-in-delay -mlinker-opt -mlong-calls @gol
471 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
472 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
473 -mno-jump-in-delay -mno-long-load-store @gol
474 -mno-portable-runtime -mno-soft-float @gol
475 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
476 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
477 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
478 -munix=@var{unix-std} -nolibdld -static -threads}
480 @emph{i386 and x86-64 Options}
481 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
482 -mfpmath=@var{unit} @gol
483 -masm=@var{dialect} -mno-fancy-math-387 @gol
484 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
485 -mno-wide-multiply -mrtd -malign-double @gol
486 -mpreferred-stack-boundary=@var{num} @gol
487 -mmmx -msse -msse2 -msse3 -m3dnow @gol
488 -mthreads -mno-align-stringops -minline-all-stringops @gol
489 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
490 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
491 -mno-red-zone -mno-tls-direct-seg-refs @gol
492 -mcmodel=@var{code-model} @gol
496 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
497 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
498 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
499 -minline-float-divide-max-throughput @gol
500 -minline-int-divide-min-latency @gol
501 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
502 -mfixed-range=@var{register-range}}
504 @emph{M32R/D Options}
505 @gccoptlist{-m32r2 -m32rx -m32r @gol
507 -malign-loops -mno-align-loops @gol
508 -missue-rate=@var{number} @gol
509 -mbranch-cost=@var{number} @gol
510 -mmodel=@var{code-size-model-type} @gol
511 -msdata=@var{sdata-type} @gol
512 -mno-flush-func -mflush-func=@var{name} @gol
513 -mno-flush-trap -mflush-trap=@var{number} @gol
516 @emph{M680x0 Options}
517 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
518 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
519 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
520 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
521 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
523 @emph{M68hc1x Options}
524 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
525 -mauto-incdec -minmax -mlong-calls -mshort @gol
526 -msoft-reg-count=@var{count}}
529 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
530 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
531 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
532 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
533 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
536 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
537 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
538 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
539 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
540 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
541 -mpaired-single -mips3d @gol
542 -mint64 -mlong64 -mlong32 @gol
543 -G@var{num} -membedded-data -mno-embedded-data @gol
544 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
545 -msplit-addresses -mno-split-addresses @gol
546 -mexplicit-relocs -mno-explicit-relocs @gol
547 -mcheck-zero-division -mno-check-zero-division @gol
548 -mdivide-traps -mdivide-breaks @gol
549 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
550 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
551 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
552 -mfix-vr4120 -mno-fix-vr4120 -mfix-sb1 -mno-fix-sb1 @gol
553 -mflush-func=@var{func} -mno-flush-func @gol
554 -mbranch-likely -mno-branch-likely @gol
555 -mfp-exceptions -mno-fp-exceptions @gol
556 -mvr4130-align -mno-vr4130-align}
559 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
560 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
561 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
562 -mno-base-addresses -msingle-exit -mno-single-exit}
564 @emph{MN10300 Options}
565 @gccoptlist{-mmult-bug -mno-mult-bug @gol
566 -mam33 -mno-am33 @gol
567 -mam33-2 -mno-am33-2 @gol
571 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
572 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
573 -mregparam -mnoregparam -msb -mnosb @gol
574 -mbitfield -mnobitfield -mhimem -mnohimem}
576 @emph{PDP-11 Options}
577 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
578 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
579 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
580 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
581 -mbranch-expensive -mbranch-cheap @gol
582 -msplit -mno-split -munix-asm -mdec-asm}
584 @emph{PowerPC Options}
585 See RS/6000 and PowerPC Options.
587 @emph{RS/6000 and PowerPC Options}
588 @gccoptlist{-mcpu=@var{cpu-type} @gol
589 -mtune=@var{cpu-type} @gol
590 -mpower -mno-power -mpower2 -mno-power2 @gol
591 -mpowerpc -mpowerpc64 -mno-powerpc @gol
592 -maltivec -mno-altivec @gol
593 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
594 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
595 -mnew-mnemonics -mold-mnemonics @gol
596 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
597 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
598 -malign-power -malign-natural @gol
599 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
600 -mstring -mno-string -mupdate -mno-update @gol
601 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
602 -mstrict-align -mno-strict-align -mrelocatable @gol
603 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
604 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
605 -mdynamic-no-pic @gol
606 -mprioritize-restricted-insns=@var{priority} @gol
607 -msched-costly-dep=@var{dependence_type} @gol
608 -minsert-sched-nops=@var{scheme} @gol
609 -mcall-sysv -mcall-netbsd @gol
610 -maix-struct-return -msvr4-struct-return @gol
611 -mabi=altivec -mabi=no-altivec @gol
612 -mabi=spe -mabi=no-spe @gol
613 -misel=yes -misel=no @gol
614 -mspe=yes -mspe=no @gol
615 -mfloat-gprs=yes -mfloat-gprs=no @gol
616 -mprototype -mno-prototype @gol
617 -msim -mmvme -mads -myellowknife -memb -msdata @gol
618 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
620 @emph{S/390 and zSeries Options}
621 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
622 -mhard-float -msoft-float -mbackchain -mno-backchain -mkernel-backchain @gol
623 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
624 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
625 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
626 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
629 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
630 -m4-nofpu -m4-single-only -m4-single -m4 @gol
631 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
632 -m5-64media -m5-64media-nofpu @gol
633 -m5-32media -m5-32media-nofpu @gol
634 -m5-compact -m5-compact-nofpu @gol
635 -mb -ml -mdalign -mrelax @gol
636 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
637 -mieee -misize -mpadstruct -mspace @gol
638 -mprefergot -musermode}
641 @gccoptlist{-mcpu=@var{cpu-type} @gol
642 -mtune=@var{cpu-type} @gol
643 -mcmodel=@var{code-model} @gol
644 -m32 -m64 -mapp-regs -mno-app-regs @gol
645 -mfaster-structs -mno-faster-structs @gol
646 -mfpu -mno-fpu -mhard-float -msoft-float @gol
647 -mhard-quad-float -msoft-quad-float @gol
648 -mimpure-text -mno-impure-text -mlittle-endian @gol
649 -mstack-bias -mno-stack-bias @gol
650 -munaligned-doubles -mno-unaligned-doubles @gol
651 -mv8plus -mno-v8plus -mvis -mno-vis
654 @emph{System V Options}
655 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
657 @emph{TMS320C3x/C4x Options}
658 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
659 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
660 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
661 -mparallel-insns -mparallel-mpy -mpreserve-float}
664 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
665 -mprolog-function -mno-prolog-function -mspace @gol
666 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
667 -mapp-regs -mno-app-regs @gol
668 -mdisable-callt -mno-disable-callt @gol
674 @gccoptlist{-mg -mgnu -munix}
676 @emph{x86-64 Options}
677 See i386 and x86-64 Options.
679 @emph{Xstormy16 Options}
682 @emph{Xtensa Options}
683 @gccoptlist{-mconst16 -mno-const16 @gol
684 -mfused-madd -mno-fused-madd @gol
685 -mtext-section-literals -mno-text-section-literals @gol
686 -mtarget-align -mno-target-align @gol
687 -mlongcalls -mno-longcalls}
689 @emph{zSeries Options}
690 See S/390 and zSeries Options.
692 @item Code Generation Options
693 @xref{Code Gen Options,,Options for Code Generation Conventions}.
694 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
695 -ffixed-@var{reg} -fexceptions @gol
696 -fnon-call-exceptions -funwind-tables @gol
697 -fasynchronous-unwind-tables @gol
698 -finhibit-size-directive -finstrument-functions @gol
699 -fno-common -fno-ident @gol
700 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
701 -freg-struct-return -fshared-data -fshort-enums @gol
702 -fshort-double -fshort-wchar @gol
703 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
704 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
705 -fargument-alias -fargument-noalias @gol
706 -fargument-noalias-global -fleading-underscore @gol
707 -ftls-model=@var{model} @gol
708 -ftrapv -fwrapv -fbounds-check @gol
713 * Overall Options:: Controlling the kind of output:
714 an executable, object files, assembler files,
715 or preprocessed source.
716 * C Dialect Options:: Controlling the variant of C language compiled.
717 * C++ Dialect Options:: Variations on C++.
718 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
720 * Language Independent Options:: Controlling how diagnostics should be
722 * Warning Options:: How picky should the compiler be?
723 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
724 * Optimize Options:: How much optimization?
725 * Preprocessor Options:: Controlling header files and macro definitions.
726 Also, getting dependency information for Make.
727 * Assembler Options:: Passing options to the assembler.
728 * Link Options:: Specifying libraries and so on.
729 * Directory Options:: Where to find header files and libraries.
730 Where to find the compiler executable files.
731 * Spec Files:: How to pass switches to sub-processes.
732 * Target Options:: Running a cross-compiler, or an old version of GCC.
735 @node Overall Options
736 @section Options Controlling the Kind of Output
738 Compilation can involve up to four stages: preprocessing, compilation
739 proper, assembly and linking, always in that order. GCC is capable of
740 preprocessing and compiling several files either into several
741 assembler input files, or into one assembler input file; then each
742 assembler input file produces an object file, and linking combines all
743 the object files (those newly compiled, and those specified as input)
744 into an executable file.
746 @cindex file name suffix
747 For any given input file, the file name suffix determines what kind of
752 C source code which must be preprocessed.
755 C source code which should not be preprocessed.
758 C++ source code which should not be preprocessed.
761 Objective-C source code. Note that you must link with the @file{libobjc}
762 library to make an Objective-C program work.
765 Objective-C source code which should not be preprocessed.
769 Objective-C++ source code. Note that you must link with the @file{libobjc}
770 library to make an Objective-C++ program work. Note that @samp{.M} refers
771 to a literal capital M@.
774 Objective-C++ source code which should not be preprocessed.
777 C, C++, Objective-C or Objective-C++ header file to be turned into a
782 @itemx @var{file}.cxx
783 @itemx @var{file}.cpp
784 @itemx @var{file}.CPP
785 @itemx @var{file}.c++
787 C++ source code which must be preprocessed. Note that in @samp{.cxx},
788 the last two letters must both be literally @samp{x}. Likewise,
789 @samp{.C} refers to a literal capital C@.
793 C++ header file to be turned into a precompiled header.
796 @itemx @var{file}.for
797 @itemx @var{file}.FOR
798 Fortran source code which should not be preprocessed.
801 @itemx @var{file}.fpp
802 @itemx @var{file}.FPP
803 Fortran source code which must be preprocessed (with the traditional
807 Fortran source code which must be preprocessed with a RATFOR
808 preprocessor (not included with GCC)@.
811 @itemx @var{file}.f95
812 Fortran 90/95 source code which should not be preprocessed.
814 @c FIXME: Descriptions of Java file types.
821 Ada source code file which contains a library unit declaration (a
822 declaration of a package, subprogram, or generic, or a generic
823 instantiation), or a library unit renaming declaration (a package,
824 generic, or subprogram renaming declaration). Such files are also
827 @itemx @var{file}.adb
828 Ada source code file containing a library unit body (a subprogram or
829 package body). Such files are also called @dfn{bodies}.
831 @c GCC also knows about some suffixes for languages not yet included:
840 Assembler code which must be preprocessed.
843 An object file to be fed straight into linking.
844 Any file name with no recognized suffix is treated this way.
848 You can specify the input language explicitly with the @option{-x} option:
851 @item -x @var{language}
852 Specify explicitly the @var{language} for the following input files
853 (rather than letting the compiler choose a default based on the file
854 name suffix). This option applies to all following input files until
855 the next @option{-x} option. Possible values for @var{language} are:
857 c c-header c-cpp-output
858 c++ c++-header c++-cpp-output
859 objective-c objective-c-header objective-c-cpp-output
860 objective-c++ objective-c++-header objective-c++-cpp-output
861 assembler assembler-with-cpp
863 f77 f77-cpp-input ratfor
870 Turn off any specification of a language, so that subsequent files are
871 handled according to their file name suffixes (as they are if @option{-x}
872 has not been used at all).
874 @item -pass-exit-codes
875 @opindex pass-exit-codes
876 Normally the @command{gcc} program will exit with the code of 1 if any
877 phase of the compiler returns a non-success return code. If you specify
878 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
879 numerically highest error produced by any phase that returned an error
883 If you only want some of the stages of compilation, you can use
884 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
885 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
886 @command{gcc} is to stop. Note that some combinations (for example,
887 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
892 Compile or assemble the source files, but do not link. The linking
893 stage simply is not done. The ultimate output is in the form of an
894 object file for each source file.
896 By default, the object file name for a source file is made by replacing
897 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
899 Unrecognized input files, not requiring compilation or assembly, are
904 Stop after the stage of compilation proper; do not assemble. The output
905 is in the form of an assembler code file for each non-assembler input
908 By default, the assembler file name for a source file is made by
909 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
911 Input files that don't require compilation are ignored.
915 Stop after the preprocessing stage; do not run the compiler proper. The
916 output is in the form of preprocessed source code, which is sent to the
919 Input files which don't require preprocessing are ignored.
921 @cindex output file option
924 Place output in file @var{file}. This applies regardless to whatever
925 sort of output is being produced, whether it be an executable file,
926 an object file, an assembler file or preprocessed C code.
928 If @option{-o} is not specified, the default is to put an executable
929 file in @file{a.out}, the object file for
930 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
931 assembler file in @file{@var{source}.s}, a precompiled header file in
932 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
937 Print (on standard error output) the commands executed to run the stages
938 of compilation. Also print the version number of the compiler driver
939 program and of the preprocessor and the compiler proper.
943 Like @option{-v} except the commands are not executed and all command
944 arguments are quoted. This is useful for shell scripts to capture the
945 driver-generated command lines.
949 Use pipes rather than temporary files for communication between the
950 various stages of compilation. This fails to work on some systems where
951 the assembler is unable to read from a pipe; but the GNU assembler has
956 If you are compiling multiple source files, this option tells the driver
957 to pass all the source files to the compiler at once (for those
958 languages for which the compiler can handle this). This will allow
959 intermodule analysis (IMA) to be performed by the compiler. Currently the only
960 language for which this is supported is C. If you pass source files for
961 multiple languages to the driver, using this option, the driver will invoke
962 the compiler(s) that support IMA once each, passing each compiler all the
963 source files appropriate for it. For those languages that do not support
964 IMA this option will be ignored, and the compiler will be invoked once for
965 each source file in that language. If you use this option in conjunction
966 with -save-temps, the compiler will generate multiple pre-processed files
967 (one for each source file), but only one (combined) .o or .s file.
971 Print (on the standard output) a description of the command line options
972 understood by @command{gcc}. If the @option{-v} option is also specified
973 then @option{--help} will also be passed on to the various processes
974 invoked by @command{gcc}, so that they can display the command line options
975 they accept. If the @option{-Wextra} option is also specified then command
976 line options which have no documentation associated with them will also
981 Print (on the standard output) a description of target specific command
982 line options for each tool.
986 Display the version number and copyrights of the invoked GCC.
990 @section Compiling C++ Programs
992 @cindex suffixes for C++ source
993 @cindex C++ source file suffixes
994 C++ source files conventionally use one of the suffixes @samp{.C},
995 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
996 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
997 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
998 files with these names and compiles them as C++ programs even if you
999 call the compiler the same way as for compiling C programs (usually
1000 with the name @command{gcc}).
1004 However, C++ programs often require class libraries as well as a
1005 compiler that understands the C++ language---and under some
1006 circumstances, you might want to compile programs or header files from
1007 standard input, or otherwise without a suffix that flags them as C++
1008 programs. You might also like to precompile a C header file with a
1009 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1010 program that calls GCC with the default language set to C++, and
1011 automatically specifies linking against the C++ library. On many
1012 systems, @command{g++} is also installed with the name @command{c++}.
1014 @cindex invoking @command{g++}
1015 When you compile C++ programs, you may specify many of the same
1016 command-line options that you use for compiling programs in any
1017 language; or command-line options meaningful for C and related
1018 languages; or options that are meaningful only for C++ programs.
1019 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1020 explanations of options for languages related to C@.
1021 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1022 explanations of options that are meaningful only for C++ programs.
1024 @node C Dialect Options
1025 @section Options Controlling C Dialect
1026 @cindex dialect options
1027 @cindex language dialect options
1028 @cindex options, dialect
1030 The following options control the dialect of C (or languages derived
1031 from C, such as C++, Objective-C and Objective-C++) that the compiler
1035 @cindex ANSI support
1039 In C mode, support all ISO C90 programs. In C++ mode,
1040 remove GNU extensions that conflict with ISO C++.
1042 This turns off certain features of GCC that are incompatible with ISO
1043 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1044 such as the @code{asm} and @code{typeof} keywords, and
1045 predefined macros such as @code{unix} and @code{vax} that identify the
1046 type of system you are using. It also enables the undesirable and
1047 rarely used ISO trigraph feature. For the C compiler,
1048 it disables recognition of C++ style @samp{//} comments as well as
1049 the @code{inline} keyword.
1051 The alternate keywords @code{__asm__}, @code{__extension__},
1052 @code{__inline__} and @code{__typeof__} continue to work despite
1053 @option{-ansi}. You would not want to use them in an ISO C program, of
1054 course, but it is useful to put them in header files that might be included
1055 in compilations done with @option{-ansi}. Alternate predefined macros
1056 such as @code{__unix__} and @code{__vax__} are also available, with or
1057 without @option{-ansi}.
1059 The @option{-ansi} option does not cause non-ISO programs to be
1060 rejected gratuitously. For that, @option{-pedantic} is required in
1061 addition to @option{-ansi}. @xref{Warning Options}.
1063 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1064 option is used. Some header files may notice this macro and refrain
1065 from declaring certain functions or defining certain macros that the
1066 ISO standard doesn't call for; this is to avoid interfering with any
1067 programs that might use these names for other things.
1069 Functions which would normally be built in but do not have semantics
1070 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1071 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1072 built-in functions provided by GCC}, for details of the functions
1077 Determine the language standard. This option is currently only
1078 supported when compiling C or C++. A value for this option must be
1079 provided; possible values are
1084 ISO C90 (same as @option{-ansi}).
1086 @item iso9899:199409
1087 ISO C90 as modified in amendment 1.
1093 ISO C99. Note that this standard is not yet fully supported; see
1094 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1095 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1098 Default, ISO C90 plus GNU extensions (including some C99 features).
1102 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1103 this will become the default. The name @samp{gnu9x} is deprecated.
1106 The 1998 ISO C++ standard plus amendments.
1109 The same as @option{-std=c++98} plus GNU extensions. This is the
1110 default for C++ code.
1113 Even when this option is not specified, you can still use some of the
1114 features of newer standards in so far as they do not conflict with
1115 previous C standards. For example, you may use @code{__restrict__} even
1116 when @option{-std=c99} is not specified.
1118 The @option{-std} options specifying some version of ISO C have the same
1119 effects as @option{-ansi}, except that features that were not in ISO C90
1120 but are in the specified version (for example, @samp{//} comments and
1121 the @code{inline} keyword in ISO C99) are not disabled.
1123 @xref{Standards,,Language Standards Supported by GCC}, for details of
1124 these standard versions.
1126 @item -aux-info @var{filename}
1128 Output to the given filename prototyped declarations for all functions
1129 declared and/or defined in a translation unit, including those in header
1130 files. This option is silently ignored in any language other than C@.
1132 Besides declarations, the file indicates, in comments, the origin of
1133 each declaration (source file and line), whether the declaration was
1134 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1135 @samp{O} for old, respectively, in the first character after the line
1136 number and the colon), and whether it came from a declaration or a
1137 definition (@samp{C} or @samp{F}, respectively, in the following
1138 character). In the case of function definitions, a K&R-style list of
1139 arguments followed by their declarations is also provided, inside
1140 comments, after the declaration.
1144 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1145 keyword, so that code can use these words as identifiers. You can use
1146 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1147 instead. @option{-ansi} implies @option{-fno-asm}.
1149 In C++, this switch only affects the @code{typeof} keyword, since
1150 @code{asm} and @code{inline} are standard keywords. You may want to
1151 use the @option{-fno-gnu-keywords} flag instead, which has the same
1152 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1153 switch only affects the @code{asm} and @code{typeof} keywords, since
1154 @code{inline} is a standard keyword in ISO C99.
1157 @itemx -fno-builtin-@var{function}
1158 @opindex fno-builtin
1159 @cindex built-in functions
1160 Don't recognize built-in functions that do not begin with
1161 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1162 functions provided by GCC}, for details of the functions affected,
1163 including those which are not built-in functions when @option{-ansi} or
1164 @option{-std} options for strict ISO C conformance are used because they
1165 do not have an ISO standard meaning.
1167 GCC normally generates special code to handle certain built-in functions
1168 more efficiently; for instance, calls to @code{alloca} may become single
1169 instructions that adjust the stack directly, and calls to @code{memcpy}
1170 may become inline copy loops. The resulting code is often both smaller
1171 and faster, but since the function calls no longer appear as such, you
1172 cannot set a breakpoint on those calls, nor can you change the behavior
1173 of the functions by linking with a different library.
1175 With the @option{-fno-builtin-@var{function}} option
1176 only the built-in function @var{function} is
1177 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1178 function is named this is not built-in in this version of GCC, this
1179 option is ignored. There is no corresponding
1180 @option{-fbuiltin-@var{function}} option; if you wish to enable
1181 built-in functions selectively when using @option{-fno-builtin} or
1182 @option{-ffreestanding}, you may define macros such as:
1185 #define abs(n) __builtin_abs ((n))
1186 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1191 @cindex hosted environment
1193 Assert that compilation takes place in a hosted environment. This implies
1194 @option{-fbuiltin}. A hosted environment is one in which the
1195 entire standard library is available, and in which @code{main} has a return
1196 type of @code{int}. Examples are nearly everything except a kernel.
1197 This is equivalent to @option{-fno-freestanding}.
1199 @item -ffreestanding
1200 @opindex ffreestanding
1201 @cindex hosted environment
1203 Assert that compilation takes place in a freestanding environment. This
1204 implies @option{-fno-builtin}. A freestanding environment
1205 is one in which the standard library may not exist, and program startup may
1206 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1207 This is equivalent to @option{-fno-hosted}.
1209 @xref{Standards,,Language Standards Supported by GCC}, for details of
1210 freestanding and hosted environments.
1212 @item -fms-extensions
1213 @opindex fms-extensions
1214 Accept some non-standard constructs used in Microsoft header files.
1218 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1219 options for strict ISO C conformance) implies @option{-trigraphs}.
1221 @item -no-integrated-cpp
1222 @opindex no-integrated-cpp
1223 Performs a compilation in two passes: preprocessing and compiling. This
1224 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1225 @option{-B} option. The user supplied compilation step can then add in
1226 an additional preprocessing step after normal preprocessing but before
1227 compiling. The default is to use the integrated cpp (internal cpp)
1229 The semantics of this option will change if "cc1", "cc1plus", and
1230 "cc1obj" are merged.
1232 @cindex traditional C language
1233 @cindex C language, traditional
1235 @itemx -traditional-cpp
1236 @opindex traditional-cpp
1237 @opindex traditional
1238 Formerly, these options caused GCC to attempt to emulate a pre-standard
1239 C compiler. They are now only supported with the @option{-E} switch.
1240 The preprocessor continues to support a pre-standard mode. See the GNU
1241 CPP manual for details.
1243 @item -fcond-mismatch
1244 @opindex fcond-mismatch
1245 Allow conditional expressions with mismatched types in the second and
1246 third arguments. The value of such an expression is void. This option
1247 is not supported for C++.
1249 @item -funsigned-char
1250 @opindex funsigned-char
1251 Let the type @code{char} be unsigned, like @code{unsigned char}.
1253 Each kind of machine has a default for what @code{char} should
1254 be. It is either like @code{unsigned char} by default or like
1255 @code{signed char} by default.
1257 Ideally, a portable program should always use @code{signed char} or
1258 @code{unsigned char} when it depends on the signedness of an object.
1259 But many programs have been written to use plain @code{char} and
1260 expect it to be signed, or expect it to be unsigned, depending on the
1261 machines they were written for. This option, and its inverse, let you
1262 make such a program work with the opposite default.
1264 The type @code{char} is always a distinct type from each of
1265 @code{signed char} or @code{unsigned char}, even though its behavior
1266 is always just like one of those two.
1269 @opindex fsigned-char
1270 Let the type @code{char} be signed, like @code{signed char}.
1272 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1273 the negative form of @option{-funsigned-char}. Likewise, the option
1274 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1276 @item -fsigned-bitfields
1277 @itemx -funsigned-bitfields
1278 @itemx -fno-signed-bitfields
1279 @itemx -fno-unsigned-bitfields
1280 @opindex fsigned-bitfields
1281 @opindex funsigned-bitfields
1282 @opindex fno-signed-bitfields
1283 @opindex fno-unsigned-bitfields
1284 These options control whether a bit-field is signed or unsigned, when the
1285 declaration does not use either @code{signed} or @code{unsigned}. By
1286 default, such a bit-field is signed, because this is consistent: the
1287 basic integer types such as @code{int} are signed types.
1290 @node C++ Dialect Options
1291 @section Options Controlling C++ Dialect
1293 @cindex compiler options, C++
1294 @cindex C++ options, command line
1295 @cindex options, C++
1296 This section describes the command-line options that are only meaningful
1297 for C++ programs; but you can also use most of the GNU compiler options
1298 regardless of what language your program is in. For example, you
1299 might compile a file @code{firstClass.C} like this:
1302 g++ -g -frepo -O -c firstClass.C
1306 In this example, only @option{-frepo} is an option meant
1307 only for C++ programs; you can use the other options with any
1308 language supported by GCC@.
1310 Here is a list of options that are @emph{only} for compiling C++ programs:
1314 @item -fabi-version=@var{n}
1315 @opindex fabi-version
1316 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1317 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1318 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1319 the version that conforms most closely to the C++ ABI specification.
1320 Therefore, the ABI obtained using version 0 will change as ABI bugs
1323 The default is version 2.
1325 @item -fno-access-control
1326 @opindex fno-access-control
1327 Turn off all access checking. This switch is mainly useful for working
1328 around bugs in the access control code.
1332 Check that the pointer returned by @code{operator new} is non-null
1333 before attempting to modify the storage allocated. This check is
1334 normally unnecessary because the C++ standard specifies that
1335 @code{operator new} will only return @code{0} if it is declared
1336 @samp{throw()}, in which case the compiler will always check the
1337 return value even without this option. In all other cases, when
1338 @code{operator new} has a non-empty exception specification, memory
1339 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1340 @samp{new (nothrow)}.
1342 @item -fconserve-space
1343 @opindex fconserve-space
1344 Put uninitialized or runtime-initialized global variables into the
1345 common segment, as C does. This saves space in the executable at the
1346 cost of not diagnosing duplicate definitions. If you compile with this
1347 flag and your program mysteriously crashes after @code{main()} has
1348 completed, you may have an object that is being destroyed twice because
1349 two definitions were merged.
1351 This option is no longer useful on most targets, now that support has
1352 been added for putting variables into BSS without making them common.
1354 @item -fno-const-strings
1355 @opindex fno-const-strings
1356 Give string constants type @code{char *} instead of type @code{const
1357 char *}. By default, G++ uses type @code{const char *} as required by
1358 the standard. Even if you use @option{-fno-const-strings}, you cannot
1359 actually modify the value of a string constant.
1361 This option might be removed in a future release of G++. For maximum
1362 portability, you should structure your code so that it works with
1363 string constants that have type @code{const char *}.
1365 @item -fno-elide-constructors
1366 @opindex fno-elide-constructors
1367 The C++ standard allows an implementation to omit creating a temporary
1368 which is only used to initialize another object of the same type.
1369 Specifying this option disables that optimization, and forces G++ to
1370 call the copy constructor in all cases.
1372 @item -fno-enforce-eh-specs
1373 @opindex fno-enforce-eh-specs
1374 Don't check for violation of exception specifications at runtime. This
1375 option violates the C++ standard, but may be useful for reducing code
1376 size in production builds, much like defining @samp{NDEBUG}. The compiler
1377 will still optimize based on the exception specifications.
1380 @itemx -fno-for-scope
1382 @opindex fno-for-scope
1383 If @option{-ffor-scope} is specified, the scope of variables declared in
1384 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1385 as specified by the C++ standard.
1386 If @option{-fno-for-scope} is specified, the scope of variables declared in
1387 a @i{for-init-statement} extends to the end of the enclosing scope,
1388 as was the case in old versions of G++, and other (traditional)
1389 implementations of C++.
1391 The default if neither flag is given to follow the standard,
1392 but to allow and give a warning for old-style code that would
1393 otherwise be invalid, or have different behavior.
1395 @item -fno-gnu-keywords
1396 @opindex fno-gnu-keywords
1397 Do not recognize @code{typeof} as a keyword, so that code can use this
1398 word as an identifier. You can use the keyword @code{__typeof__} instead.
1399 @option{-ansi} implies @option{-fno-gnu-keywords}.
1401 @item -fno-implicit-templates
1402 @opindex fno-implicit-templates
1403 Never emit code for non-inline templates which are instantiated
1404 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1405 @xref{Template Instantiation}, for more information.
1407 @item -fno-implicit-inline-templates
1408 @opindex fno-implicit-inline-templates
1409 Don't emit code for implicit instantiations of inline templates, either.
1410 The default is to handle inlines differently so that compiles with and
1411 without optimization will need the same set of explicit instantiations.
1413 @item -fno-implement-inlines
1414 @opindex fno-implement-inlines
1415 To save space, do not emit out-of-line copies of inline functions
1416 controlled by @samp{#pragma implementation}. This will cause linker
1417 errors if these functions are not inlined everywhere they are called.
1419 @item -fms-extensions
1420 @opindex fms-extensions
1421 Disable pedantic warnings about constructs used in MFC, such as implicit
1422 int and getting a pointer to member function via non-standard syntax.
1424 @item -fno-nonansi-builtins
1425 @opindex fno-nonansi-builtins
1426 Disable built-in declarations of functions that are not mandated by
1427 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1428 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1430 @item -fno-operator-names
1431 @opindex fno-operator-names
1432 Do not treat the operator name keywords @code{and}, @code{bitand},
1433 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1434 synonyms as keywords.
1436 @item -fno-optional-diags
1437 @opindex fno-optional-diags
1438 Disable diagnostics that the standard says a compiler does not need to
1439 issue. Currently, the only such diagnostic issued by G++ is the one for
1440 a name having multiple meanings within a class.
1443 @opindex fpermissive
1444 Downgrade some diagnostics about nonconformant code from errors to
1445 warnings. Thus, using @option{-fpermissive} will allow some
1446 nonconforming code to compile.
1450 Enable automatic template instantiation at link time. This option also
1451 implies @option{-fno-implicit-templates}. @xref{Template
1452 Instantiation}, for more information.
1456 Disable generation of information about every class with virtual
1457 functions for use by the C++ runtime type identification features
1458 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1459 of the language, you can save some space by using this flag. Note that
1460 exception handling uses the same information, but it will generate it as
1465 Emit statistics about front-end processing at the end of the compilation.
1466 This information is generally only useful to the G++ development team.
1468 @item -ftemplate-depth-@var{n}
1469 @opindex ftemplate-depth
1470 Set the maximum instantiation depth for template classes to @var{n}.
1471 A limit on the template instantiation depth is needed to detect
1472 endless recursions during template class instantiation. ANSI/ISO C++
1473 conforming programs must not rely on a maximum depth greater than 17.
1475 @item -fno-threadsafe-statics
1476 @opindex fno-threadsafe-statics
1477 Do not emit the extra code to use the routines specified in the C++
1478 ABI for thread-safe initialization of local statics. You can use this
1479 option to reduce code size slightly in code that doesn't need to be
1482 @item -fuse-cxa-atexit
1483 @opindex fuse-cxa-atexit
1484 Register destructors for objects with static storage duration with the
1485 @code{__cxa_atexit} function rather than the @code{atexit} function.
1486 This option is required for fully standards-compliant handling of static
1487 destructors, but will only work if your C library supports
1488 @code{__cxa_atexit}.
1490 @item -fvisibility-inlines-hidden
1491 @opindex fvisibility-inlines-hidden
1492 Causes all inlined methods to be marked with
1493 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1494 appear in the export table of a DSO and do not require a PLT indirection
1495 when used within the DSO. Enabling this option can have a dramatic effect
1496 on load and link times of a DSO as it massively reduces the size of the
1497 dynamic export table when the library makes heavy use of templates. While
1498 it can cause bloating through duplication of code within each DSO where
1499 it is used, often the wastage is less than the considerable space occupied
1500 by a long symbol name in the export table which is typical when using
1501 templates and namespaces. For even more savings, combine with the
1502 @code{-fvisibility=hidden} switch.
1506 Do not use weak symbol support, even if it is provided by the linker.
1507 By default, G++ will use weak symbols if they are available. This
1508 option exists only for testing, and should not be used by end-users;
1509 it will result in inferior code and has no benefits. This option may
1510 be removed in a future release of G++.
1514 Do not search for header files in the standard directories specific to
1515 C++, but do still search the other standard directories. (This option
1516 is used when building the C++ library.)
1519 In addition, these optimization, warning, and code generation options
1520 have meanings only for C++ programs:
1523 @item -fno-default-inline
1524 @opindex fno-default-inline
1525 Do not assume @samp{inline} for functions defined inside a class scope.
1526 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1527 functions will have linkage like inline functions; they just won't be
1530 @item -Wabi @r{(C++ only)}
1532 Warn when G++ generates code that is probably not compatible with the
1533 vendor-neutral C++ ABI. Although an effort has been made to warn about
1534 all such cases, there are probably some cases that are not warned about,
1535 even though G++ is generating incompatible code. There may also be
1536 cases where warnings are emitted even though the code that is generated
1539 You should rewrite your code to avoid these warnings if you are
1540 concerned about the fact that code generated by G++ may not be binary
1541 compatible with code generated by other compilers.
1543 The known incompatibilities at this point include:
1548 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1549 pack data into the same byte as a base class. For example:
1552 struct A @{ virtual void f(); int f1 : 1; @};
1553 struct B : public A @{ int f2 : 1; @};
1557 In this case, G++ will place @code{B::f2} into the same byte
1558 as@code{A::f1}; other compilers will not. You can avoid this problem
1559 by explicitly padding @code{A} so that its size is a multiple of the
1560 byte size on your platform; that will cause G++ and other compilers to
1561 layout @code{B} identically.
1564 Incorrect handling of tail-padding for virtual bases. G++ does not use
1565 tail padding when laying out virtual bases. For example:
1568 struct A @{ virtual void f(); char c1; @};
1569 struct B @{ B(); char c2; @};
1570 struct C : public A, public virtual B @{@};
1574 In this case, G++ will not place @code{B} into the tail-padding for
1575 @code{A}; other compilers will. You can avoid this problem by
1576 explicitly padding @code{A} so that its size is a multiple of its
1577 alignment (ignoring virtual base classes); that will cause G++ and other
1578 compilers to layout @code{C} identically.
1581 Incorrect handling of bit-fields with declared widths greater than that
1582 of their underlying types, when the bit-fields appear in a union. For
1586 union U @{ int i : 4096; @};
1590 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1591 union too small by the number of bits in an @code{int}.
1594 Empty classes can be placed at incorrect offsets. For example:
1604 struct C : public B, public A @{@};
1608 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1609 it should be placed at offset zero. G++ mistakenly believes that the
1610 @code{A} data member of @code{B} is already at offset zero.
1613 Names of template functions whose types involve @code{typename} or
1614 template template parameters can be mangled incorrectly.
1617 template <typename Q>
1618 void f(typename Q::X) @{@}
1620 template <template <typename> class Q>
1621 void f(typename Q<int>::X) @{@}
1625 Instantiations of these templates may be mangled incorrectly.
1629 @item -Wctor-dtor-privacy @r{(C++ only)}
1630 @opindex Wctor-dtor-privacy
1631 Warn when a class seems unusable because all the constructors or
1632 destructors in that class are private, and it has neither friends nor
1633 public static member functions.
1635 @item -Wnon-virtual-dtor @r{(C++ only)}
1636 @opindex Wnon-virtual-dtor
1637 Warn when a class appears to be polymorphic, thereby requiring a virtual
1638 destructor, yet it declares a non-virtual one.
1639 This warning is enabled by @option{-Wall}.
1641 @item -Wreorder @r{(C++ only)}
1643 @cindex reordering, warning
1644 @cindex warning for reordering of member initializers
1645 Warn when the order of member initializers given in the code does not
1646 match the order in which they must be executed. For instance:
1652 A(): j (0), i (1) @{ @}
1656 The compiler will rearrange the member initializers for @samp{i}
1657 and @samp{j} to match the declaration order of the members, emitting
1658 a warning to that effect. This warning is enabled by @option{-Wall}.
1661 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1664 @item -Weffc++ @r{(C++ only)}
1666 Warn about violations of the following style guidelines from Scott Meyers'
1667 @cite{Effective C++} book:
1671 Item 11: Define a copy constructor and an assignment operator for classes
1672 with dynamically allocated memory.
1675 Item 12: Prefer initialization to assignment in constructors.
1678 Item 14: Make destructors virtual in base classes.
1681 Item 15: Have @code{operator=} return a reference to @code{*this}.
1684 Item 23: Don't try to return a reference when you must return an object.
1688 Also warn about violations of the following style guidelines from
1689 Scott Meyers' @cite{More Effective C++} book:
1693 Item 6: Distinguish between prefix and postfix forms of increment and
1694 decrement operators.
1697 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1701 When selecting this option, be aware that the standard library
1702 headers do not obey all of these guidelines; use @samp{grep -v}
1703 to filter out those warnings.
1705 @item -Wno-deprecated @r{(C++ only)}
1706 @opindex Wno-deprecated
1707 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1709 @item -Wno-non-template-friend @r{(C++ only)}
1710 @opindex Wno-non-template-friend
1711 Disable warnings when non-templatized friend functions are declared
1712 within a template. Since the advent of explicit template specification
1713 support in G++, if the name of the friend is an unqualified-id (i.e.,
1714 @samp{friend foo(int)}), the C++ language specification demands that the
1715 friend declare or define an ordinary, nontemplate function. (Section
1716 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1717 could be interpreted as a particular specialization of a templatized
1718 function. Because this non-conforming behavior is no longer the default
1719 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1720 check existing code for potential trouble spots and is on by default.
1721 This new compiler behavior can be turned off with
1722 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1723 but disables the helpful warning.
1725 @item -Wold-style-cast @r{(C++ only)}
1726 @opindex Wold-style-cast
1727 Warn if an old-style (C-style) cast to a non-void type is used within
1728 a C++ program. The new-style casts (@samp{static_cast},
1729 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1730 unintended effects and much easier to search for.
1732 @item -Woverloaded-virtual @r{(C++ only)}
1733 @opindex Woverloaded-virtual
1734 @cindex overloaded virtual fn, warning
1735 @cindex warning for overloaded virtual fn
1736 Warn when a function declaration hides virtual functions from a
1737 base class. For example, in:
1744 struct B: public A @{
1749 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1757 will fail to compile.
1759 @item -Wno-pmf-conversions @r{(C++ only)}
1760 @opindex Wno-pmf-conversions
1761 Disable the diagnostic for converting a bound pointer to member function
1764 @item -Wsign-promo @r{(C++ only)}
1765 @opindex Wsign-promo
1766 Warn when overload resolution chooses a promotion from unsigned or
1767 enumerated type to a signed type, over a conversion to an unsigned type of
1768 the same size. Previous versions of G++ would try to preserve
1769 unsignedness, but the standard mandates the current behavior.
1771 @item -Wsynth @r{(C++ only)}
1773 @cindex warning for synthesized methods
1774 @cindex synthesized methods, warning
1775 Warn when G++'s synthesis behavior does not match that of cfront. For
1781 A& operator = (int);
1791 In this example, G++ will synthesize a default @samp{A& operator =
1792 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1795 @node Objective-C and Objective-C++ Dialect Options
1796 @section Options Controlling Objective-C and Objective-C++ Dialects
1798 @cindex compiler options, Objective-C and Objective-C++
1799 @cindex Objective-C and Objective-C++ options, command line
1800 @cindex options, Objective-C and Objective-C++
1801 (NOTE: This manual does not describe the Objective-C and Objective-C++
1802 languages themselves. See @xref{Standards,,Language Standards
1803 Supported by GCC}, for references.)
1805 This section describes the command-line options that are only meaningful
1806 for Objective-C and Objective-C++ programs, but you can also use most of
1807 the language-independent GNU compiler options.
1808 For example, you might compile a file @code{some_class.m} like this:
1811 gcc -g -fgnu-runtime -O -c some_class.m
1815 In this example, @option{-fgnu-runtime} is an option meant only for
1816 Objective-C and Objective-C++ programs; you can use the other options with
1817 any language supported by GCC@.
1819 Note that since Objective-C is an extension of the C language, Objective-C
1820 compilations may also use options specific to the C front-end (e.g.,
1821 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1822 C++-specific options (e.g., @option{-Wabi}).
1824 Here is a list of options that are @emph{only} for compiling Objective-C
1825 and Objective-C++ programs:
1828 @item -fconstant-string-class=@var{class-name}
1829 @opindex fconstant-string-class
1830 Use @var{class-name} as the name of the class to instantiate for each
1831 literal string specified with the syntax @code{@@"@dots{}"}. The default
1832 class name is @code{NXConstantString} if the GNU runtime is being used, and
1833 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1834 @option{-fconstant-cfstrings} option, if also present, will override the
1835 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1836 to be laid out as constant CoreFoundation strings.
1839 @opindex fgnu-runtime
1840 Generate object code compatible with the standard GNU Objective-C
1841 runtime. This is the default for most types of systems.
1843 @item -fnext-runtime
1844 @opindex fnext-runtime
1845 Generate output compatible with the NeXT runtime. This is the default
1846 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1847 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1850 @item -fno-nil-receivers
1851 @opindex fno-nil-receivers
1852 Assume that all Objective-C message dispatches (e.g.,
1853 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1854 is not @code{nil}. This allows for more efficient entry points in the runtime
1855 to be used. Currently, this option is only available in conjunction with
1856 the NeXT runtime on Mac OS X 10.3 and later.
1858 @item -fobjc-exceptions
1859 @opindex fobjc-exceptions
1860 Enable syntactic support for structured exception handling in Objective-C,
1861 similar to what is offered by C++ and Java. Currently, this option is only
1862 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1870 @@catch (AnObjCClass *exc) @{
1877 @@catch (AnotherClass *exc) @{
1880 @@catch (id allOthers) @{
1890 The @code{@@throw} statement may appear anywhere in an Objective-C or
1891 Objective-C++ program; when used inside of a @code{@@catch} block, the
1892 @code{@@throw} may appear without an argument (as shown above), in which case
1893 the object caught by the @code{@@catch} will be rethrown.
1895 Note that only (pointers to) Objective-C objects may be thrown and
1896 caught using this scheme. When an object is thrown, it will be caught
1897 by the nearest @code{@@catch} clause capable of handling objects of that type,
1898 analogously to how @code{catch} blocks work in C++ and Java. A
1899 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1900 any and all Objective-C exceptions not caught by previous @code{@@catch}
1903 The @code{@@finally} clause, if present, will be executed upon exit from the
1904 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1905 regardless of whether any exceptions are thrown, caught or rethrown
1906 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1907 of the @code{finally} clause in Java.
1909 There are several caveats to using the new exception mechanism:
1913 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1914 idioms provided by the @code{NSException} class, the new
1915 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1916 systems, due to additional functionality needed in the (NeXT) Objective-C
1920 As mentioned above, the new exceptions do not support handling
1921 types other than Objective-C objects. Furthermore, when used from
1922 Objective-C++, the Objective-C exception model does not interoperate with C++
1923 exceptions at this time. This means you cannot @code{@@throw} an exception
1924 from Objective-C and @code{catch} it in C++, or vice versa
1925 (i.e., @code{throw @dots{} @@catch}).
1928 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1929 blocks for thread-safe execution:
1932 @@synchronized (ObjCClass *guard) @{
1937 Upon entering the @code{@@synchronized} block, a thread of execution shall
1938 first check whether a lock has been placed on the corresponding @code{guard}
1939 object by another thread. If it has, the current thread shall wait until
1940 the other thread relinquishes its lock. Once @code{guard} becomes available,
1941 the current thread will place its own lock on it, execute the code contained in
1942 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1943 making @code{guard} available to other threads).
1945 Unlike Java, Objective-C does not allow for entire methods to be marked
1946 @code{@@synchronized}. Note that throwing exceptions out of
1947 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1948 to be unlocked properly.
1950 @item -freplace-objc-classes
1951 @opindex freplace-objc-classes
1952 Emit a special marker instructing @command{ld(1)} not to statically link in
1953 the resulting object file, and allow @command{dyld(1)} to load it in at
1954 run time instead. This is used in conjunction with the Fix-and-Continue
1955 debugging mode, where the object file in question may be recompiled and
1956 dynamically reloaded in the course of program execution, without the need
1957 to restart the program itself. Currently, Fix-and-Continue functionality
1958 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1963 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1964 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1965 compile time) with static class references that get initialized at load time,
1966 which improves run-time performance. Specifying the @option{-fzero-link} flag
1967 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1968 to be retained. This is useful in Zero-Link debugging mode, since it allows
1969 for individual class implementations to be modified during program execution.
1973 Dump interface declarations for all classes seen in the source file to a
1974 file named @file{@var{sourcename}.decl}.
1977 @opindex Wno-protocol
1978 If a class is declared to implement a protocol, a warning is issued for
1979 every method in the protocol that is not implemented by the class. The
1980 default behavior is to issue a warning for every method not explicitly
1981 implemented in the class, even if a method implementation is inherited
1982 from the superclass. If you use the @code{-Wno-protocol} option, then
1983 methods inherited from the superclass are considered to be implemented,
1984 and no warning is issued for them.
1988 Warn if multiple methods of different types for the same selector are
1989 found during compilation. The check is performed on the list of methods
1990 in the final stage of compilation. Additionally, a check is performed
1991 for each selector appearing in a @code{@@selector(@dots{})}
1992 expression, and a corresponding method for that selector has been found
1993 during compilation. Because these checks scan the method table only at
1994 the end of compilation, these warnings are not produced if the final
1995 stage of compilation is not reached, for example because an error is
1996 found during compilation, or because the @code{-fsyntax-only} option is
1999 @item -Wundeclared-selector
2000 @opindex Wundeclared-selector
2001 Warn if a @code{@@selector(@dots{})} expression referring to an
2002 undeclared selector is found. A selector is considered undeclared if no
2003 method with that name has been declared before the
2004 @code{@@selector(@dots{})} expression, either explicitly in an
2005 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2006 an @code{@@implementation} section. This option always performs its
2007 checks as soon as a @code{@@selector(@dots{})} expression is found,
2008 while @code{-Wselector} only performs its checks in the final stage of
2009 compilation. This also enforces the coding style convention
2010 that methods and selectors must be declared before being used.
2012 @item -print-objc-runtime-info
2013 @opindex print-objc-runtime-info
2014 Generate C header describing the largest structure that is passed by
2019 @node Language Independent Options
2020 @section Options to Control Diagnostic Messages Formatting
2021 @cindex options to control diagnostics formatting
2022 @cindex diagnostic messages
2023 @cindex message formatting
2025 Traditionally, diagnostic messages have been formatted irrespective of
2026 the output device's aspect (e.g.@: its width, @dots{}). The options described
2027 below can be used to control the diagnostic messages formatting
2028 algorithm, e.g.@: how many characters per line, how often source location
2029 information should be reported. Right now, only the C++ front end can
2030 honor these options. However it is expected, in the near future, that
2031 the remaining front ends would be able to digest them correctly.
2034 @item -fmessage-length=@var{n}
2035 @opindex fmessage-length
2036 Try to format error messages so that they fit on lines of about @var{n}
2037 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2038 the front ends supported by GCC@. If @var{n} is zero, then no
2039 line-wrapping will be done; each error message will appear on a single
2042 @opindex fdiagnostics-show-location
2043 @item -fdiagnostics-show-location=once
2044 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2045 reporter to emit @emph{once} source location information; that is, in
2046 case the message is too long to fit on a single physical line and has to
2047 be wrapped, the source location won't be emitted (as prefix) again,
2048 over and over, in subsequent continuation lines. This is the default
2051 @item -fdiagnostics-show-location=every-line
2052 Only meaningful in line-wrapping mode. Instructs the diagnostic
2053 messages reporter to emit the same source location information (as
2054 prefix) for physical lines that result from the process of breaking
2055 a message which is too long to fit on a single line.
2059 @node Warning Options
2060 @section Options to Request or Suppress Warnings
2061 @cindex options to control warnings
2062 @cindex warning messages
2063 @cindex messages, warning
2064 @cindex suppressing warnings
2066 Warnings are diagnostic messages that report constructions which
2067 are not inherently erroneous but which are risky or suggest there
2068 may have been an error.
2070 You can request many specific warnings with options beginning @samp{-W},
2071 for example @option{-Wimplicit} to request warnings on implicit
2072 declarations. Each of these specific warning options also has a
2073 negative form beginning @samp{-Wno-} to turn off warnings;
2074 for example, @option{-Wno-implicit}. This manual lists only one of the
2075 two forms, whichever is not the default.
2077 The following options control the amount and kinds of warnings produced
2078 by GCC; for further, language-specific options also refer to
2079 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2083 @cindex syntax checking
2085 @opindex fsyntax-only
2086 Check the code for syntax errors, but don't do anything beyond that.
2090 Issue all the warnings demanded by strict ISO C and ISO C++;
2091 reject all programs that use forbidden extensions, and some other
2092 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2093 version of the ISO C standard specified by any @option{-std} option used.
2095 Valid ISO C and ISO C++ programs should compile properly with or without
2096 this option (though a rare few will require @option{-ansi} or a
2097 @option{-std} option specifying the required version of ISO C)@. However,
2098 without this option, certain GNU extensions and traditional C and C++
2099 features are supported as well. With this option, they are rejected.
2101 @option{-pedantic} does not cause warning messages for use of the
2102 alternate keywords whose names begin and end with @samp{__}. Pedantic
2103 warnings are also disabled in the expression that follows
2104 @code{__extension__}. However, only system header files should use
2105 these escape routes; application programs should avoid them.
2106 @xref{Alternate Keywords}.
2108 Some users try to use @option{-pedantic} to check programs for strict ISO
2109 C conformance. They soon find that it does not do quite what they want:
2110 it finds some non-ISO practices, but not all---only those for which
2111 ISO C @emph{requires} a diagnostic, and some others for which
2112 diagnostics have been added.
2114 A feature to report any failure to conform to ISO C might be useful in
2115 some instances, but would require considerable additional work and would
2116 be quite different from @option{-pedantic}. We don't have plans to
2117 support such a feature in the near future.
2119 Where the standard specified with @option{-std} represents a GNU
2120 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2121 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2122 extended dialect is based. Warnings from @option{-pedantic} are given
2123 where they are required by the base standard. (It would not make sense
2124 for such warnings to be given only for features not in the specified GNU
2125 C dialect, since by definition the GNU dialects of C include all
2126 features the compiler supports with the given option, and there would be
2127 nothing to warn about.)
2129 @item -pedantic-errors
2130 @opindex pedantic-errors
2131 Like @option{-pedantic}, except that errors are produced rather than
2136 Inhibit all warning messages.
2140 Inhibit warning messages about the use of @samp{#import}.
2142 @item -Wchar-subscripts
2143 @opindex Wchar-subscripts
2144 Warn if an array subscript has type @code{char}. This is a common cause
2145 of error, as programmers often forget that this type is signed on some
2150 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2151 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2153 @item -Wfatal-errors
2154 @opindex Wfatal-errors
2155 This option causes the compiler to abort compilation on the first error
2156 occurred rather than trying to keep going and printing further error
2161 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2162 the arguments supplied have types appropriate to the format string
2163 specified, and that the conversions specified in the format string make
2164 sense. This includes standard functions, and others specified by format
2165 attributes (@pxref{Function Attributes}), in the @code{printf},
2166 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2167 not in the C standard) families (or other target-specific families).
2169 The formats are checked against the format features supported by GNU
2170 libc version 2.2. These include all ISO C90 and C99 features, as well
2171 as features from the Single Unix Specification and some BSD and GNU
2172 extensions. Other library implementations may not support all these
2173 features; GCC does not support warning about features that go beyond a
2174 particular library's limitations. However, if @option{-pedantic} is used
2175 with @option{-Wformat}, warnings will be given about format features not
2176 in the selected standard version (but not for @code{strfmon} formats,
2177 since those are not in any version of the C standard). @xref{C Dialect
2178 Options,,Options Controlling C Dialect}.
2180 Since @option{-Wformat} also checks for null format arguments for
2181 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2183 @option{-Wformat} is included in @option{-Wall}. For more control over some
2184 aspects of format checking, the options @option{-Wformat-y2k},
2185 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2186 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2187 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2190 @opindex Wformat-y2k
2191 If @option{-Wformat} is specified, also warn about @code{strftime}
2192 formats which may yield only a two-digit year.
2194 @item -Wno-format-extra-args
2195 @opindex Wno-format-extra-args
2196 If @option{-Wformat} is specified, do not warn about excess arguments to a
2197 @code{printf} or @code{scanf} format function. The C standard specifies
2198 that such arguments are ignored.
2200 Where the unused arguments lie between used arguments that are
2201 specified with @samp{$} operand number specifications, normally
2202 warnings are still given, since the implementation could not know what
2203 type to pass to @code{va_arg} to skip the unused arguments. However,
2204 in the case of @code{scanf} formats, this option will suppress the
2205 warning if the unused arguments are all pointers, since the Single
2206 Unix Specification says that such unused arguments are allowed.
2208 @item -Wno-format-zero-length
2209 @opindex Wno-format-zero-length
2210 If @option{-Wformat} is specified, do not warn about zero-length formats.
2211 The C standard specifies that zero-length formats are allowed.
2213 @item -Wformat-nonliteral
2214 @opindex Wformat-nonliteral
2215 If @option{-Wformat} is specified, also warn if the format string is not a
2216 string literal and so cannot be checked, unless the format function
2217 takes its format arguments as a @code{va_list}.
2219 @item -Wformat-security
2220 @opindex Wformat-security
2221 If @option{-Wformat} is specified, also warn about uses of format
2222 functions that represent possible security problems. At present, this
2223 warns about calls to @code{printf} and @code{scanf} functions where the
2224 format string is not a string literal and there are no format arguments,
2225 as in @code{printf (foo);}. This may be a security hole if the format
2226 string came from untrusted input and contains @samp{%n}. (This is
2227 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2228 in future warnings may be added to @option{-Wformat-security} that are not
2229 included in @option{-Wformat-nonliteral}.)
2233 Enable @option{-Wformat} plus format checks not included in
2234 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2235 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2239 Warn about passing a null pointer for arguments marked as
2240 requiring a non-null value by the @code{nonnull} function attribute.
2242 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2243 can be disabled with the @option{-Wno-nonnull} option.
2245 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2247 Warn about uninitialized variables which are initialized with themselves.
2248 Note this option can only be used with the @option{-Wuninitialized} option,
2249 which in turn only works with @option{-O1} and above.
2251 For example, GCC will warn about @code{i} being uninitialized in the
2252 following snippet only when @option{-Winit-self} has been specified:
2263 @item -Wimplicit-int
2264 @opindex Wimplicit-int
2265 Warn when a declaration does not specify a type.
2267 @item -Wimplicit-function-declaration
2268 @itemx -Werror-implicit-function-declaration
2269 @opindex Wimplicit-function-declaration
2270 @opindex Werror-implicit-function-declaration
2271 Give a warning (or error) whenever a function is used before being
2276 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2280 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2281 function with external linkage, returning int, taking either zero
2282 arguments, two, or three arguments of appropriate types.
2284 @item -Wmissing-braces
2285 @opindex Wmissing-braces
2286 Warn if an aggregate or union initializer is not fully bracketed. In
2287 the following example, the initializer for @samp{a} is not fully
2288 bracketed, but that for @samp{b} is fully bracketed.
2291 int a[2][2] = @{ 0, 1, 2, 3 @};
2292 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2295 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2296 @opindex Wmissing-include-dirs
2297 Warn if a user-supplied include directory does not exist.
2300 @opindex Wparentheses
2301 Warn if parentheses are omitted in certain contexts, such
2302 as when there is an assignment in a context where a truth value
2303 is expected, or when operators are nested whose precedence people
2304 often get confused about. Only the warning for an assignment used as
2305 a truth value is supported when compiling C++; the other warnings are
2306 only supported when compiling C@.
2308 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2309 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2310 interpretation from that of ordinary mathematical notation.
2312 Also warn about constructions where there may be confusion to which
2313 @code{if} statement an @code{else} branch belongs. Here is an example of
2328 In C, every @code{else} branch belongs to the innermost possible @code{if}
2329 statement, which in this example is @code{if (b)}. This is often not
2330 what the programmer expected, as illustrated in the above example by
2331 indentation the programmer chose. When there is the potential for this
2332 confusion, GCC will issue a warning when this flag is specified.
2333 To eliminate the warning, add explicit braces around the innermost
2334 @code{if} statement so there is no way the @code{else} could belong to
2335 the enclosing @code{if}. The resulting code would look like this:
2351 @item -Wsequence-point
2352 @opindex Wsequence-point
2353 Warn about code that may have undefined semantics because of violations
2354 of sequence point rules in the C standard.
2356 The C standard defines the order in which expressions in a C program are
2357 evaluated in terms of @dfn{sequence points}, which represent a partial
2358 ordering between the execution of parts of the program: those executed
2359 before the sequence point, and those executed after it. These occur
2360 after the evaluation of a full expression (one which is not part of a
2361 larger expression), after the evaluation of the first operand of a
2362 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2363 function is called (but after the evaluation of its arguments and the
2364 expression denoting the called function), and in certain other places.
2365 Other than as expressed by the sequence point rules, the order of
2366 evaluation of subexpressions of an expression is not specified. All
2367 these rules describe only a partial order rather than a total order,
2368 since, for example, if two functions are called within one expression
2369 with no sequence point between them, the order in which the functions
2370 are called is not specified. However, the standards committee have
2371 ruled that function calls do not overlap.
2373 It is not specified when between sequence points modifications to the
2374 values of objects take effect. Programs whose behavior depends on this
2375 have undefined behavior; the C standard specifies that ``Between the
2376 previous and next sequence point an object shall have its stored value
2377 modified at most once by the evaluation of an expression. Furthermore,
2378 the prior value shall be read only to determine the value to be
2379 stored.''. If a program breaks these rules, the results on any
2380 particular implementation are entirely unpredictable.
2382 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2383 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2384 diagnosed by this option, and it may give an occasional false positive
2385 result, but in general it has been found fairly effective at detecting
2386 this sort of problem in programs.
2388 The present implementation of this option only works for C programs. A
2389 future implementation may also work for C++ programs.
2391 The C standard is worded confusingly, therefore there is some debate
2392 over the precise meaning of the sequence point rules in subtle cases.
2393 Links to discussions of the problem, including proposed formal
2394 definitions, may be found on the GCC readings page, at
2395 @w{@uref{http://gcc.gnu.org/readings.html}}.
2398 @opindex Wreturn-type
2399 Warn whenever a function is defined with a return-type that defaults to
2400 @code{int}. Also warn about any @code{return} statement with no
2401 return-value in a function whose return-type is not @code{void}.
2403 For C, also warn if the return type of a function has a type qualifier
2404 such as @code{const}. Such a type qualifier has no effect, since the
2405 value returned by a function is not an lvalue. ISO C prohibits
2406 qualified @code{void} return types on function definitions, so such
2407 return types always receive a warning even without this option.
2409 For C++, a function without return type always produces a diagnostic
2410 message, even when @option{-Wno-return-type} is specified. The only
2411 exceptions are @samp{main} and functions defined in system headers.
2415 Warn whenever a @code{switch} statement has an index of enumerated type
2416 and lacks a @code{case} for one or more of the named codes of that
2417 enumeration. (The presence of a @code{default} label prevents this
2418 warning.) @code{case} labels outside the enumeration range also
2419 provoke warnings when this option is used.
2421 @item -Wswitch-default
2422 @opindex Wswitch-switch
2423 Warn whenever a @code{switch} statement does not have a @code{default}
2427 @opindex Wswitch-enum
2428 Warn whenever a @code{switch} statement has an index of enumerated type
2429 and lacks a @code{case} for one or more of the named codes of that
2430 enumeration. @code{case} labels outside the enumeration range also
2431 provoke warnings when this option is used.
2435 Warn if any trigraphs are encountered that might change the meaning of
2436 the program (trigraphs within comments are not warned about).
2438 @item -Wunused-function
2439 @opindex Wunused-function
2440 Warn whenever a static function is declared but not defined or a
2441 non\-inline static function is unused.
2443 @item -Wunused-label
2444 @opindex Wunused-label
2445 Warn whenever a label is declared but not used.
2447 To suppress this warning use the @samp{unused} attribute
2448 (@pxref{Variable Attributes}).
2450 @item -Wunused-parameter
2451 @opindex Wunused-parameter
2452 Warn whenever a function parameter is unused aside from its declaration.
2454 To suppress this warning use the @samp{unused} attribute
2455 (@pxref{Variable Attributes}).
2457 @item -Wunused-variable
2458 @opindex Wunused-variable
2459 Warn whenever a local variable or non-constant static variable is unused
2460 aside from its declaration
2462 To suppress this warning use the @samp{unused} attribute
2463 (@pxref{Variable Attributes}).
2465 @item -Wunused-value
2466 @opindex Wunused-value
2467 Warn whenever a statement computes a result that is explicitly not used.
2469 To suppress this warning cast the expression to @samp{void}.
2473 All the above @option{-Wunused} options combined.
2475 In order to get a warning about an unused function parameter, you must
2476 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2477 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2479 @item -Wuninitialized
2480 @opindex Wuninitialized
2481 Warn if an automatic variable is used without first being initialized or
2482 if a variable may be clobbered by a @code{setjmp} call.
2484 These warnings are possible only in optimizing compilation,
2485 because they require data flow information that is computed only
2486 when optimizing. If you don't specify @option{-O}, you simply won't
2489 If you want to warn about code which uses the uninitialized value of the
2490 variable in its own initializer, use the @option{-Winit-self} option.
2492 These warnings occur only for variables that are candidates for
2493 register allocation. Therefore, they do not occur for a variable that
2494 is declared @code{volatile}, or whose address is taken, or whose size
2495 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2496 structures, unions or arrays, even when they are in registers.
2498 Note that there may be no warning about a variable that is used only
2499 to compute a value that itself is never used, because such
2500 computations may be deleted by data flow analysis before the warnings
2503 These warnings are made optional because GCC is not smart
2504 enough to see all the reasons why the code might be correct
2505 despite appearing to have an error. Here is one example of how
2526 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2527 always initialized, but GCC doesn't know this. Here is
2528 another common case:
2533 if (change_y) save_y = y, y = new_y;
2535 if (change_y) y = save_y;
2540 This has no bug because @code{save_y} is used only if it is set.
2542 @cindex @code{longjmp} warnings
2543 This option also warns when a non-volatile automatic variable might be
2544 changed by a call to @code{longjmp}. These warnings as well are possible
2545 only in optimizing compilation.
2547 The compiler sees only the calls to @code{setjmp}. It cannot know
2548 where @code{longjmp} will be called; in fact, a signal handler could
2549 call it at any point in the code. As a result, you may get a warning
2550 even when there is in fact no problem because @code{longjmp} cannot
2551 in fact be called at the place which would cause a problem.
2553 Some spurious warnings can be avoided if you declare all the functions
2554 you use that never return as @code{noreturn}. @xref{Function
2557 @item -Wunknown-pragmas
2558 @opindex Wunknown-pragmas
2559 @cindex warning for unknown pragmas
2560 @cindex unknown pragmas, warning
2561 @cindex pragmas, warning of unknown
2562 Warn when a #pragma directive is encountered which is not understood by
2563 GCC@. If this command line option is used, warnings will even be issued
2564 for unknown pragmas in system header files. This is not the case if
2565 the warnings were only enabled by the @option{-Wall} command line option.
2567 @item -Wstrict-aliasing
2568 @opindex Wstrict-aliasing
2569 This option is only active when @option{-fstrict-aliasing} is active.
2570 It warns about code which might break the strict aliasing rules that the
2571 compiler is using for optimization. The warning does not catch all
2572 cases, but does attempt to catch the more common pitfalls. It is
2573 included in @option{-Wall}.
2575 @item -Wstrict-aliasing=2
2576 @opindex Wstrict-aliasing=2
2577 This option is only active when @option{-fstrict-aliasing} is active.
2578 It warns about all code which might break the strict aliasing rules that the
2579 compiler is using for optimization. This warning catches all cases, but
2580 it will also give a warning for some ambiguous cases that are safe.
2584 All of the above @samp{-W} options combined. This enables all the
2585 warnings about constructions that some users consider questionable, and
2586 that are easy to avoid (or modify to prevent the warning), even in
2587 conjunction with macros. This also enables some language-specific
2588 warnings described in @ref{C++ Dialect Options} and
2589 @ref{Objective-C and Objective-C++ Dialect Options}.
2592 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2593 Some of them warn about constructions that users generally do not
2594 consider questionable, but which occasionally you might wish to check
2595 for; others warn about constructions that are necessary or hard to avoid
2596 in some cases, and there is no simple way to modify the code to suppress
2603 (This option used to be called @option{-W}. The older name is still
2604 supported, but the newer name is more descriptive.) Print extra warning
2605 messages for these events:
2609 A function can return either with or without a value. (Falling
2610 off the end of the function body is considered returning without
2611 a value.) For example, this function would evoke such a
2625 An expression-statement or the left-hand side of a comma expression
2626 contains no side effects.
2627 To suppress the warning, cast the unused expression to void.
2628 For example, an expression such as @samp{x[i,j]} will cause a warning,
2629 but @samp{x[(void)i,j]} will not.
2632 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2635 Storage-class specifiers like @code{static} are not the first things in
2636 a declaration. According to the C Standard, this usage is obsolescent.
2639 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2643 A comparison between signed and unsigned values could produce an
2644 incorrect result when the signed value is converted to unsigned.
2645 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2648 An aggregate has an initializer which does not initialize all members.
2649 This warning can be independently controlled by
2650 @option{-Wmissing-field-initializers}.
2653 A function parameter is declared without a type specifier in K&R-style
2661 An empty body occurs in an @samp{if} or @samp{else} statement.
2664 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2665 @samp{>}, or @samp{>=}.
2668 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2671 Any of several floating-point events that often indicate errors, such as
2672 overflow, underflow, loss of precision, etc.
2674 @item @r{(C++ only)}
2675 An enumerator and a non-enumerator both appear in a conditional expression.
2677 @item @r{(C++ only)}
2678 A non-static reference or non-static @samp{const} member appears in a
2679 class without constructors.
2681 @item @r{(C++ only)}
2682 Ambiguous virtual bases.
2684 @item @r{(C++ only)}
2685 Subscripting an array which has been declared @samp{register}.
2687 @item @r{(C++ only)}
2688 Taking the address of a variable which has been declared @samp{register}.
2690 @item @r{(C++ only)}
2691 A base class is not initialized in a derived class' copy constructor.
2694 @item -Wno-div-by-zero
2695 @opindex Wno-div-by-zero
2696 @opindex Wdiv-by-zero
2697 Do not warn about compile-time integer division by zero. Floating point
2698 division by zero is not warned about, as it can be a legitimate way of
2699 obtaining infinities and NaNs.
2701 @item -Wsystem-headers
2702 @opindex Wsystem-headers
2703 @cindex warnings from system headers
2704 @cindex system headers, warnings from
2705 Print warning messages for constructs found in system header files.
2706 Warnings from system headers are normally suppressed, on the assumption
2707 that they usually do not indicate real problems and would only make the
2708 compiler output harder to read. Using this command line option tells
2709 GCC to emit warnings from system headers as if they occurred in user
2710 code. However, note that using @option{-Wall} in conjunction with this
2711 option will @emph{not} warn about unknown pragmas in system
2712 headers---for that, @option{-Wunknown-pragmas} must also be used.
2715 @opindex Wfloat-equal
2716 Warn if floating point values are used in equality comparisons.
2718 The idea behind this is that sometimes it is convenient (for the
2719 programmer) to consider floating-point values as approximations to
2720 infinitely precise real numbers. If you are doing this, then you need
2721 to compute (by analyzing the code, or in some other way) the maximum or
2722 likely maximum error that the computation introduces, and allow for it
2723 when performing comparisons (and when producing output, but that's a
2724 different problem). In particular, instead of testing for equality, you
2725 would check to see whether the two values have ranges that overlap; and
2726 this is done with the relational operators, so equality comparisons are
2729 @item -Wtraditional @r{(C only)}
2730 @opindex Wtraditional
2731 Warn about certain constructs that behave differently in traditional and
2732 ISO C@. Also warn about ISO C constructs that have no traditional C
2733 equivalent, and/or problematic constructs which should be avoided.
2737 Macro parameters that appear within string literals in the macro body.
2738 In traditional C macro replacement takes place within string literals,
2739 but does not in ISO C@.
2742 In traditional C, some preprocessor directives did not exist.
2743 Traditional preprocessors would only consider a line to be a directive
2744 if the @samp{#} appeared in column 1 on the line. Therefore
2745 @option{-Wtraditional} warns about directives that traditional C
2746 understands but would ignore because the @samp{#} does not appear as the
2747 first character on the line. It also suggests you hide directives like
2748 @samp{#pragma} not understood by traditional C by indenting them. Some
2749 traditional implementations would not recognize @samp{#elif}, so it
2750 suggests avoiding it altogether.
2753 A function-like macro that appears without arguments.
2756 The unary plus operator.
2759 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2760 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2761 constants.) Note, these suffixes appear in macros defined in the system
2762 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2763 Use of these macros in user code might normally lead to spurious
2764 warnings, however GCC's integrated preprocessor has enough context to
2765 avoid warning in these cases.
2768 A function declared external in one block and then used after the end of
2772 A @code{switch} statement has an operand of type @code{long}.
2775 A non-@code{static} function declaration follows a @code{static} one.
2776 This construct is not accepted by some traditional C compilers.
2779 The ISO type of an integer constant has a different width or
2780 signedness from its traditional type. This warning is only issued if
2781 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2782 typically represent bit patterns, are not warned about.
2785 Usage of ISO string concatenation is detected.
2788 Initialization of automatic aggregates.
2791 Identifier conflicts with labels. Traditional C lacks a separate
2792 namespace for labels.
2795 Initialization of unions. If the initializer is zero, the warning is
2796 omitted. This is done under the assumption that the zero initializer in
2797 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2798 initializer warnings and relies on default initialization to zero in the
2802 Conversions by prototypes between fixed/floating point values and vice
2803 versa. The absence of these prototypes when compiling with traditional
2804 C would cause serious problems. This is a subset of the possible
2805 conversion warnings, for the full set use @option{-Wconversion}.
2808 Use of ISO C style function definitions. This warning intentionally is
2809 @emph{not} issued for prototype declarations or variadic functions
2810 because these ISO C features will appear in your code when using
2811 libiberty's traditional C compatibility macros, @code{PARAMS} and
2812 @code{VPARAMS}. This warning is also bypassed for nested functions
2813 because that feature is already a GCC extension and thus not relevant to
2814 traditional C compatibility.
2817 @item -Wdeclaration-after-statement @r{(C only)}
2818 @opindex Wdeclaration-after-statement
2819 Warn when a declaration is found after a statement in a block. This
2820 construct, known from C++, was introduced with ISO C99 and is by default
2821 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2822 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2826 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2828 @item -Wendif-labels
2829 @opindex Wendif-labels
2830 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2834 Warn whenever a local variable shadows another local variable, parameter or
2835 global variable or whenever a built-in function is shadowed.
2837 @item -Wlarger-than-@var{len}
2838 @opindex Wlarger-than
2839 Warn whenever an object of larger than @var{len} bytes is defined.
2841 @item -Wpointer-arith
2842 @opindex Wpointer-arith
2843 Warn about anything that depends on the ``size of'' a function type or
2844 of @code{void}. GNU C assigns these types a size of 1, for
2845 convenience in calculations with @code{void *} pointers and pointers
2848 @item -Wbad-function-cast @r{(C only)}
2849 @opindex Wbad-function-cast
2850 Warn whenever a function call is cast to a non-matching type.
2851 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2855 Warn whenever a pointer is cast so as to remove a type qualifier from
2856 the target type. For example, warn if a @code{const char *} is cast
2857 to an ordinary @code{char *}.
2860 @opindex Wcast-align
2861 Warn whenever a pointer is cast such that the required alignment of the
2862 target is increased. For example, warn if a @code{char *} is cast to
2863 an @code{int *} on machines where integers can only be accessed at
2864 two- or four-byte boundaries.
2866 @item -Wwrite-strings
2867 @opindex Wwrite-strings
2868 When compiling C, give string constants the type @code{const
2869 char[@var{length}]} so that
2870 copying the address of one into a non-@code{const} @code{char *}
2871 pointer will get a warning; when compiling C++, warn about the
2872 deprecated conversion from string constants to @code{char *}.
2873 These warnings will help you find at
2874 compile time code that can try to write into a string constant, but
2875 only if you have been very careful about using @code{const} in
2876 declarations and prototypes. Otherwise, it will just be a nuisance;
2877 this is why we did not make @option{-Wall} request these warnings.
2880 @opindex Wconversion
2881 Warn if a prototype causes a type conversion that is different from what
2882 would happen to the same argument in the absence of a prototype. This
2883 includes conversions of fixed point to floating and vice versa, and
2884 conversions changing the width or signedness of a fixed point argument
2885 except when the same as the default promotion.
2887 Also, warn if a negative integer constant expression is implicitly
2888 converted to an unsigned type. For example, warn about the assignment
2889 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2890 casts like @code{(unsigned) -1}.
2892 @item -Wsign-compare
2893 @opindex Wsign-compare
2894 @cindex warning for comparison of signed and unsigned values
2895 @cindex comparison of signed and unsigned values, warning
2896 @cindex signed and unsigned values, comparison warning
2897 Warn when a comparison between signed and unsigned values could produce
2898 an incorrect result when the signed value is converted to unsigned.
2899 This warning is also enabled by @option{-Wextra}; to get the other warnings
2900 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2902 @item -Waggregate-return
2903 @opindex Waggregate-return
2904 Warn if any functions that return structures or unions are defined or
2905 called. (In languages where you can return an array, this also elicits
2908 @item -Wstrict-prototypes @r{(C only)}
2909 @opindex Wstrict-prototypes
2910 Warn if a function is declared or defined without specifying the
2911 argument types. (An old-style function definition is permitted without
2912 a warning if preceded by a declaration which specifies the argument
2915 @item -Wold-style-definition @r{(C only)}
2916 @opindex Wold-style-definition
2917 Warn if an old-style function definition is used. A warning is given
2918 even if there is a previous prototype.
2920 @item -Wmissing-prototypes @r{(C only)}
2921 @opindex Wmissing-prototypes
2922 Warn if a global function is defined without a previous prototype
2923 declaration. This warning is issued even if the definition itself
2924 provides a prototype. The aim is to detect global functions that fail
2925 to be declared in header files.
2927 @item -Wmissing-declarations @r{(C only)}
2928 @opindex Wmissing-declarations
2929 Warn if a global function is defined without a previous declaration.
2930 Do so even if the definition itself provides a prototype.
2931 Use this option to detect global functions that are not declared in
2934 @item -Wmissing-field-initializers
2935 @opindex Wmissing-field-initializers
2938 Warn if a structure's initializer has some fields missing. For
2939 example, the following code would cause such a warning, because
2940 @code{x.h} is implicitly zero:
2943 struct s @{ int f, g, h; @};
2944 struct s x = @{ 3, 4 @};
2947 This option does not warn about designated initializers, so the following
2948 modification would not trigger a warning:
2951 struct s @{ int f, g, h; @};
2952 struct s x = @{ .f = 3, .g = 4 @};
2955 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
2956 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
2958 @item -Wmissing-noreturn
2959 @opindex Wmissing-noreturn
2960 Warn about functions which might be candidates for attribute @code{noreturn}.
2961 Note these are only possible candidates, not absolute ones. Care should
2962 be taken to manually verify functions actually do not ever return before
2963 adding the @code{noreturn} attribute, otherwise subtle code generation
2964 bugs could be introduced. You will not get a warning for @code{main} in
2965 hosted C environments.
2967 @item -Wmissing-format-attribute
2968 @opindex Wmissing-format-attribute
2970 If @option{-Wformat} is enabled, also warn about functions which might be
2971 candidates for @code{format} attributes. Note these are only possible
2972 candidates, not absolute ones. GCC will guess that @code{format}
2973 attributes might be appropriate for any function that calls a function
2974 like @code{vprintf} or @code{vscanf}, but this might not always be the
2975 case, and some functions for which @code{format} attributes are
2976 appropriate may not be detected. This option has no effect unless
2977 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2979 @item -Wno-multichar
2980 @opindex Wno-multichar
2982 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2983 Usually they indicate a typo in the user's code, as they have
2984 implementation-defined values, and should not be used in portable code.
2986 @item -Wno-deprecated-declarations
2987 @opindex Wno-deprecated-declarations
2988 Do not warn about uses of functions, variables, and types marked as
2989 deprecated by using the @code{deprecated} attribute.
2990 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2991 @pxref{Type Attributes}.)
2995 Warn if a structure is given the packed attribute, but the packed
2996 attribute has no effect on the layout or size of the structure.
2997 Such structures may be mis-aligned for little benefit. For
2998 instance, in this code, the variable @code{f.x} in @code{struct bar}
2999 will be misaligned even though @code{struct bar} does not itself
3000 have the packed attribute:
3007 @} __attribute__((packed));
3017 Warn if padding is included in a structure, either to align an element
3018 of the structure or to align the whole structure. Sometimes when this
3019 happens it is possible to rearrange the fields of the structure to
3020 reduce the padding and so make the structure smaller.
3022 @item -Wredundant-decls
3023 @opindex Wredundant-decls
3024 Warn if anything is declared more than once in the same scope, even in
3025 cases where multiple declaration is valid and changes nothing.
3027 @item -Wnested-externs @r{(C only)}
3028 @opindex Wnested-externs
3029 Warn if an @code{extern} declaration is encountered within a function.
3031 @item -Wunreachable-code
3032 @opindex Wunreachable-code
3033 Warn if the compiler detects that code will never be executed.
3035 This option is intended to warn when the compiler detects that at
3036 least a whole line of source code will never be executed, because
3037 some condition is never satisfied or because it is after a
3038 procedure that never returns.
3040 It is possible for this option to produce a warning even though there
3041 are circumstances under which part of the affected line can be executed,
3042 so care should be taken when removing apparently-unreachable code.
3044 For instance, when a function is inlined, a warning may mean that the
3045 line is unreachable in only one inlined copy of the function.
3047 This option is not made part of @option{-Wall} because in a debugging
3048 version of a program there is often substantial code which checks
3049 correct functioning of the program and is, hopefully, unreachable
3050 because the program does work. Another common use of unreachable
3051 code is to provide behavior which is selectable at compile-time.
3055 Warn if a function can not be inlined and it was declared as inline.
3056 Even with this option, the compiler will not warn about failures to
3057 inline functions declared in system headers.
3059 The compiler uses a variety of heuristics to determine whether or not
3060 to inline a function. For example, the compiler takes into account
3061 the size of the function being inlined and the the amount of inlining
3062 that has already been done in the current function. Therefore,
3063 seemingly insignificant changes in the source program can cause the
3064 warnings produced by @option{-Winline} to appear or disappear.
3066 @item -Wno-invalid-offsetof @r{(C++ only)}
3067 @opindex Wno-invalid-offsetof
3068 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3069 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3070 to a non-POD type is undefined. In existing C++ implementations,
3071 however, @samp{offsetof} typically gives meaningful results even when
3072 applied to certain kinds of non-POD types. (Such as a simple
3073 @samp{struct} that fails to be a POD type only by virtue of having a
3074 constructor.) This flag is for users who are aware that they are
3075 writing nonportable code and who have deliberately chosen to ignore the
3078 The restrictions on @samp{offsetof} may be relaxed in a future version
3079 of the C++ standard.
3082 @opindex Winvalid-pch
3083 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3084 the search path but can't be used.
3088 @opindex Wno-long-long
3089 Warn if @samp{long long} type is used. This is default. To inhibit
3090 the warning messages, use @option{-Wno-long-long}. Flags
3091 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3092 only when @option{-pedantic} flag is used.
3094 @item -Wvariadic-macros
3095 @opindex Wvariadic-macros
3096 @opindex Wno-variadic-macros
3097 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3098 alternate syntax when in pedantic ISO C99 mode. This is default.
3099 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3101 @item -Wdisabled-optimization
3102 @opindex Wdisabled-optimization
3103 Warn if a requested optimization pass is disabled. This warning does
3104 not generally indicate that there is anything wrong with your code; it
3105 merely indicates that GCC's optimizers were unable to handle the code
3106 effectively. Often, the problem is that your code is too big or too
3107 complex; GCC will refuse to optimize programs when the optimization
3108 itself is likely to take inordinate amounts of time.
3112 Make all warnings into errors.
3115 @node Debugging Options
3116 @section Options for Debugging Your Program or GCC
3117 @cindex options, debugging
3118 @cindex debugging information options
3120 GCC has various special options that are used for debugging
3121 either your program or GCC:
3126 Produce debugging information in the operating system's native format
3127 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3130 On most systems that use stabs format, @option{-g} enables use of extra
3131 debugging information that only GDB can use; this extra information
3132 makes debugging work better in GDB but will probably make other debuggers
3134 refuse to read the program. If you want to control for certain whether
3135 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3136 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3138 GCC allows you to use @option{-g} with
3139 @option{-O}. The shortcuts taken by optimized code may occasionally
3140 produce surprising results: some variables you declared may not exist
3141 at all; flow of control may briefly move where you did not expect it;
3142 some statements may not be executed because they compute constant
3143 results or their values were already at hand; some statements may
3144 execute in different places because they were moved out of loops.
3146 Nevertheless it proves possible to debug optimized output. This makes
3147 it reasonable to use the optimizer for programs that might have bugs.
3149 The following options are useful when GCC is generated with the
3150 capability for more than one debugging format.
3154 Produce debugging information for use by GDB@. This means to use the
3155 most expressive format available (DWARF 2, stabs, or the native format
3156 if neither of those are supported), including GDB extensions if at all
3161 Produce debugging information in stabs format (if that is supported),
3162 without GDB extensions. This is the format used by DBX on most BSD
3163 systems. On MIPS, Alpha and System V Release 4 systems this option
3164 produces stabs debugging output which is not understood by DBX or SDB@.
3165 On System V Release 4 systems this option requires the GNU assembler.
3167 @item -feliminate-unused-debug-symbols
3168 @opindex feliminate-unused-debug-symbols
3169 Produce debugging information in stabs format (if that is supported),
3170 for only symbols that are actually used.
3174 Produce debugging information in stabs format (if that is supported),
3175 using GNU extensions understood only by the GNU debugger (GDB)@. The
3176 use of these extensions is likely to make other debuggers crash or
3177 refuse to read the program.
3181 Produce debugging information in COFF format (if that is supported).
3182 This is the format used by SDB on most System V systems prior to
3187 Produce debugging information in XCOFF format (if that is supported).
3188 This is the format used by the DBX debugger on IBM RS/6000 systems.
3192 Produce debugging information in XCOFF format (if that is supported),
3193 using GNU extensions understood only by the GNU debugger (GDB)@. The
3194 use of these extensions is likely to make other debuggers crash or
3195 refuse to read the program, and may cause assemblers other than the GNU
3196 assembler (GAS) to fail with an error.
3200 Produce debugging information in DWARF version 2 format (if that is
3201 supported). This is the format used by DBX on IRIX 6. With this
3202 option, GCC uses features of DWARF version 3 when they are useful;
3203 version 3 is upward compatible with version 2, but may still cause
3204 problems for older debuggers.
3208 Produce debugging information in VMS debug format (if that is
3209 supported). This is the format used by DEBUG on VMS systems.
3212 @itemx -ggdb@var{level}
3213 @itemx -gstabs@var{level}
3214 @itemx -gcoff@var{level}
3215 @itemx -gxcoff@var{level}
3216 @itemx -gvms@var{level}
3217 Request debugging information and also use @var{level} to specify how
3218 much information. The default level is 2.
3220 Level 1 produces minimal information, enough for making backtraces in
3221 parts of the program that you don't plan to debug. This includes
3222 descriptions of functions and external variables, but no information
3223 about local variables and no line numbers.
3225 Level 3 includes extra information, such as all the macro definitions
3226 present in the program. Some debuggers support macro expansion when
3227 you use @option{-g3}.
3229 @option{-gdwarf-2} does not accept a concatenated debug level, because
3230 GCC used to support an option @option{-gdwarf} that meant to generate
3231 debug information in version 1 of the DWARF format (which is very
3232 different from version 2), and it would have been too confusing. That
3233 debug format is long obsolete, but the option cannot be changed now.
3234 Instead use an additional @option{-g@var{level}} option to change the
3235 debug level for DWARF2.
3237 @item -feliminate-dwarf2-dups
3238 @opindex feliminate-dwarf2-dups
3239 Compress DWARF2 debugging information by eliminating duplicated
3240 information about each symbol. This option only makes sense when
3241 generating DWARF2 debugging information with @option{-gdwarf-2}.
3243 @cindex @command{prof}
3246 Generate extra code to write profile information suitable for the
3247 analysis program @command{prof}. You must use this option when compiling
3248 the source files you want data about, and you must also use it when
3251 @cindex @command{gprof}
3254 Generate extra code to write profile information suitable for the
3255 analysis program @command{gprof}. You must use this option when compiling
3256 the source files you want data about, and you must also use it when
3261 Makes the compiler print out each function name as it is compiled, and
3262 print some statistics about each pass when it finishes.
3265 @opindex ftime-report
3266 Makes the compiler print some statistics about the time consumed by each
3267 pass when it finishes.
3270 @opindex fmem-report
3271 Makes the compiler print some statistics about permanent memory
3272 allocation when it finishes.
3274 @item -fprofile-arcs
3275 @opindex fprofile-arcs
3276 Add code so that program flow @dfn{arcs} are instrumented. During
3277 execution the program records how many times each branch and call is
3278 executed and how many times it is taken or returns. When the compiled
3279 program exits it saves this data to a file called
3280 @file{@var{auxname}.gcda} for each source file. The data may be used for
3281 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3282 test coverage analysis (@option{-ftest-coverage}). Each object file's
3283 @var{auxname} is generated from the name of the output file, if
3284 explicitly specified and it is not the final executable, otherwise it is
3285 the basename of the source file. In both cases any suffix is removed
3286 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3287 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3292 Compile the source files with @option{-fprofile-arcs} plus optimization
3293 and code generation options. For test coverage analysis, use the
3294 additional @option{-ftest-coverage} option. You do not need to profile
3295 every source file in a program.
3298 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3299 (the latter implies the former).
3302 Run the program on a representative workload to generate the arc profile
3303 information. This may be repeated any number of times. You can run
3304 concurrent instances of your program, and provided that the file system
3305 supports locking, the data files will be correctly updated. Also
3306 @code{fork} calls are detected and correctly handled (double counting
3310 For profile-directed optimizations, compile the source files again with
3311 the same optimization and code generation options plus
3312 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3313 Control Optimization}).
3316 For test coverage analysis, use @command{gcov} to produce human readable
3317 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3318 @command{gcov} documentation for further information.
3322 With @option{-fprofile-arcs}, for each function of your program GCC
3323 creates a program flow graph, then finds a spanning tree for the graph.
3324 Only arcs that are not on the spanning tree have to be instrumented: the
3325 compiler adds code to count the number of times that these arcs are
3326 executed. When an arc is the only exit or only entrance to a block, the
3327 instrumentation code can be added to the block; otherwise, a new basic
3328 block must be created to hold the instrumentation code.
3330 @item -ftree-based-profiling
3331 @opindex ftree-based-profiling
3332 This option is used in addition to @option{-fprofile-arcs} or
3333 @option{-fbranch-probabilities} to control whether those optimizations
3334 are performed on a tree-based or rtl-based internal representation.
3335 If you use this option when compiling with @option{-fprofile-arcs},
3336 you must also use it when compiling later with @option{-fbranch-probabilities}.
3337 Currently the tree-based optimization is in an early stage of
3338 development, and this option is recommended only for those people
3339 working on improving it.
3342 @item -ftest-coverage
3343 @opindex ftest-coverage
3344 Produce a notes file that the @command{gcov} code-coverage utility
3345 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3346 show program coverage. Each source file's note file is called
3347 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3348 above for a description of @var{auxname} and instructions on how to
3349 generate test coverage data. Coverage data will match the source files
3350 more closely, if you do not optimize.
3352 @item -d@var{letters}
3353 @item -fdump-rtl-@var{pass}
3355 Says to make debugging dumps during compilation at times specified by
3356 @var{letters}. This is used for debugging the RTL-based passes of the
3357 compiler. The file names for most of the dumps are made by appending a
3358 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3359 from the name of the output file, if explicitly specified and it is not
3360 an executable, otherwise it is the basename of the source file.
3362 Most debug dumps can be enabled either passing a letter to the @option{-d}
3363 option, or with a long @option{-fdump-rtl} switch; here are the possible
3364 letters for use in @var{letters} and @var{pass}, and their meanings:
3369 Annotate the assembler output with miscellaneous debugging information.
3372 @itemx -fdump-rtl-bp
3374 @opindex fdump-rtl-bp
3375 Dump after computing branch probabilities, to @file{@var{file}.09.bp}.
3378 @itemx -fdump-rtl-bbro
3380 @opindex fdump-rtl-bbro
3381 Dump after block reordering, to @file{@var{file}.30.bbro}.
3384 @itemx -fdump-rtl-combine
3386 @opindex fdump-rtl-combine
3387 Dump after instruction combination, to the file @file{@var{file}.17.combine}.
3390 @itemx -fdump-rtl-ce1
3391 @itemx -fdump-rtl-ce2
3393 @opindex fdump-rtl-ce1
3394 @opindex fdump-rtl-ce2
3395 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3396 first if conversion, to the file @file{@var{file}.11.ce1}. @option{-dC}
3397 and @option{-fdump-rtl-ce2} enable dumping after the second if
3398 conversion, to the file @file{@var{file}.18.ce2}.
3401 @itemx -fdump-rtl-btl
3402 @itemx -fdump-rtl-dbr
3404 @opindex fdump-rtl-btl
3405 @opindex fdump-rtl-dbr
3406 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3407 target load optimization, to to @file{@var{file}.31.btl}. @option{-dd}
3408 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3409 scheduling, to @file{@var{file}.36.dbr}.
3413 Dump all macro definitions, at the end of preprocessing, in addition to
3417 @itemx -fdump-rtl-ce3
3419 @opindex fdump-rtl-ce3
3420 Dump after the third if conversion, to @file{@var{file}.28.ce3}.
3423 @itemx -fdump-rtl-cfg
3424 @itemx -fdump-rtl-life
3426 @opindex fdump-rtl-cfg
3427 @opindex fdump-rtl-life
3428 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3429 and data flow analysis, to @file{@var{file}.08.cfg}. @option{-df}
3430 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3431 to @file{@var{file}.16.life}.
3434 @itemx -fdump-rtl-greg
3436 @opindex fdump-rtl-greg
3437 Dump after global register allocation, to @file{@var{file}.23.greg}.
3440 @itemx -fdump-rtl-gcse
3441 @itemx -fdump-rtl-bypass
3443 @opindex fdump-rtl-gcse
3444 @opindex fdump-rtl-bypass
3445 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3446 @file{@var{file}.05.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3447 enable dumping after jump bypassing and control flow optimizations, to
3448 @file{@var{file}.07.bypass}.
3451 @itemx -fdump-rtl-eh
3453 @opindex fdump-rtl-eh
3454 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3457 @itemx -fdump-rtl-sibling
3459 @opindex fdump-rtl-sibling
3460 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3463 @itemx -fdump-rtl-jump
3465 @opindex fdump-rtl-jump
3466 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3469 @itemx -fdump-rtl-stack
3471 @opindex fdump-rtl-stack
3472 Dump after conversion from registers to stack, to @file{@var{file}.33.stack}.
3475 @itemx -fdump-rtl-lreg
3477 @opindex fdump-rtl-lreg
3478 Dump after local register allocation, to @file{@var{file}.22.lreg}.
3481 @itemx -fdump-rtl-loop
3482 @itemx -fdump-rtl-loop2
3484 @opindex fdump-rtl-loop
3485 @opindex fdump-rtl-loop2
3486 @option{-dL} and @option{-fdump-rtl-loop} enable dumping after the first
3487 loop optimization pass, to @file{@var{file}.06.loop}. @option{-dL} and
3488 @option{-fdump-rtl-loop2} enable dumping after the second pass, to
3489 @file{@var{file}.13.loop2}.
3492 @itemx -fdump-rtl-sms
3494 @opindex fdump-rtl-sms
3495 Dump after modulo scheduling, to @file{@var{file}.20.sms}.
3498 @itemx -fdump-rtl-mach
3500 @opindex fdump-rtl-mach
3501 Dump after performing the machine dependent reorganization pass, to
3502 @file{@var{file}.35.mach}.
3505 @itemx -fdump-rtl-rnreg
3507 @opindex fdump-rtl-rnreg
3508 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3511 @itemx -fdump-rtl-regmove
3513 @opindex fdump-rtl-regmove
3514 Dump after the register move pass, to @file{@var{file}.19.regmove}.
3517 @itemx -fdump-rtl-postreload
3519 @opindex fdump-rtl-postreload
3520 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3523 @itemx -fdump-rtl-expand
3525 @opindex fdump-rtl-expand
3526 Dump after RTL generation, to @file{@var{file}.00.expand}.
3529 @itemx -fdump-rtl-sched2
3531 @opindex fdump-rtl-sched2
3532 Dump after the second scheduling pass, to @file{@var{file}.32.sched2}.
3535 @itemx -fdump-rtl-cse
3537 @opindex fdump-rtl-cse
3538 Dump after CSE (including the jump optimization that sometimes follows
3539 CSE), to @file{@var{file}.04.cse}.
3542 @itemx -fdump-rtl-sched
3544 @opindex fdump-rtl-sched
3545 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3548 @itemx -fdump-rtl-cse2
3550 @opindex fdump-rtl-cse2
3551 Dump after the second CSE pass (including the jump optimization that
3552 sometimes follows CSE), to @file{@var{file}.15.cse2}.
3555 @itemx -fdump-rtl-tracer
3557 @opindex fdump-rtl-tracer
3558 Dump after running tracer, to @file{@var{file}.12.tracer}.
3561 @itemx -fdump-rtl-vpt
3562 @itemx -fdump-rtl-vartrack
3564 @opindex fdump-rtl-vpt
3565 @opindex fdump-rtl-vartrack
3566 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3567 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3568 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3569 to @file{@var{file}.34.vartrack}.
3572 @itemx -fdump-rtl-flow2
3574 @opindex fdump-rtl-flow2
3575 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3578 @itemx -fdump-rtl-peephole2
3580 @opindex fdump-rtl-peephole2
3581 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3584 @itemx -fdump-rtl-web
3586 @opindex fdump-rtl-web
3587 Dump after live range splitting, to @file{@var{file}.14.web}.
3590 @itemx -fdump-rtl-all
3592 @opindex fdump-rtl-all
3593 Produce all the dumps listed above.
3597 Produce a core dump whenever an error occurs.
3601 Print statistics on memory usage, at the end of the run, to
3606 Annotate the assembler output with a comment indicating which
3607 pattern and alternative was used. The length of each instruction is
3612 Dump the RTL in the assembler output as a comment before each instruction.
3613 Also turns on @option{-dp} annotation.
3617 For each of the other indicated dump files (either with @option{-d} or
3618 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3619 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3623 Just generate RTL for a function instead of compiling it. Usually used
3624 with @samp{r} (@option{-fdump-rtl-expand}).
3628 Dump debugging information during parsing, to standard error.
3631 @item -fdump-unnumbered
3632 @opindex fdump-unnumbered
3633 When doing debugging dumps (see @option{-d} option above), suppress instruction
3634 numbers and line number note output. This makes it more feasible to
3635 use diff on debugging dumps for compiler invocations with different
3636 options, in particular with and without @option{-g}.
3638 @item -fdump-translation-unit @r{(C and C++ only)}
3639 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3640 @opindex fdump-translation-unit
3641 Dump a representation of the tree structure for the entire translation
3642 unit to a file. The file name is made by appending @file{.tu} to the
3643 source file name. If the @samp{-@var{options}} form is used, @var{options}
3644 controls the details of the dump as described for the
3645 @option{-fdump-tree} options.
3647 @item -fdump-class-hierarchy @r{(C++ only)}
3648 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3649 @opindex fdump-class-hierarchy
3650 Dump a representation of each class's hierarchy and virtual function
3651 table layout to a file. The file name is made by appending @file{.class}
3652 to the source file name. If the @samp{-@var{options}} form is used,
3653 @var{options} controls the details of the dump as described for the
3654 @option{-fdump-tree} options.
3656 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3657 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3659 Control the dumping at various stages of processing the intermediate
3660 language tree to a file. The file name is generated by appending a switch
3661 specific suffix to the source file name. If the @samp{-@var{options}}
3662 form is used, @var{options} is a list of @samp{-} separated options that
3663 control the details of the dump. Not all options are applicable to all
3664 dumps, those which are not meaningful will be ignored. The following
3665 options are available
3669 Print the address of each node. Usually this is not meaningful as it
3670 changes according to the environment and source file. Its primary use
3671 is for tying up a dump file with a debug environment.
3673 Inhibit dumping of members of a scope or body of a function merely
3674 because that scope has been reached. Only dump such items when they
3675 are directly reachable by some other path. When dumping pretty-printed
3676 trees, this option inhibits dumping the bodies of control structures.
3678 Print a raw representation of the tree. By default, trees are
3679 pretty-printed into a C-like representation.
3681 Enable more detailed dumps (not honored by every dump option).
3683 Enable dumping various statistics about the pass (not honored by every dump
3686 Enable showing basic block boundaries (disabled in raw dumps).
3688 Enable showing virtual operands for every statement.
3690 Enable showing line numbers for statements.
3692 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3694 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3697 The following tree dumps are possible:
3701 Dump before any tree based optimization, to @file{@var{file}.original}.
3704 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3707 Dump after function inlining, to @file{@var{file}.inlined}.
3710 @opindex fdump-tree-gimple
3711 Dump each function before and after the gimplification pass to a file. The
3712 file name is made by appending @file{.gimple} to the source file name.
3715 @opindex fdump-tree-cfg
3716 Dump the control flow graph of each function to a file. The file name is
3717 made by appending @file{.cfg} to the source file name.
3720 @opindex fdump-tree-vcg
3721 Dump the control flow graph of each function to a file in VCG format. The
3722 file name is made by appending @file{.vcg} to the source file name. Note
3723 that if the file contains more than one function, the generated file cannot
3724 be used directly by VCG. You will need to cut and paste each function's
3725 graph into its own separate file first.
3728 @opindex fdump-tree-ch
3729 Dump each function after copying loop headers. The file name is made by
3730 appending @file{.ch} to the source file name.
3733 @opindex fdump-tree-ssa
3734 Dump SSA related information to a file. The file name is made by appending
3735 @file{.ssa} to the source file name.
3738 @opindex fdump-tree-alias
3739 Dump aliasing information for each function. The file name is made by
3740 appending @file{.alias} to the source file name.
3743 @opindex fdump-tree-ccp
3744 Dump each function after CCP. The file name is made by appending
3745 @file{.ccp} to the source file name.
3748 @opindex fdump-tree-pre
3749 Dump trees after partial redundancy elimination. The file name is made
3750 by appending @file{.pre} to the source file name.
3753 @opindex fdump-tree-fre
3754 Dump trees after full redundancy elimination. The file name is made
3755 by appending @file{.fre} to the source file name.
3758 @opindex fdump-tree-dce
3759 Dump each function after dead code elimination. The file name is made by
3760 appending @file{.dce} to the source file name.
3763 @opindex fdump-tree-mudflap
3764 Dump each function after adding mudflap instrumentation. The file name is
3765 made by appending @file{.mudflap} to the source file name.
3768 @opindex fdump-tree-sra
3769 Dump each function after performing scalar replacement of aggregates. The
3770 file name is made by appending @file{.sra} to the source file name.
3773 @opindex fdump-tree-dom
3774 Dump each function after applying dominator tree optimizations. The file
3775 name is made by appending @file{.dom} to the source file name.
3778 @opindex fdump-tree-dse
3779 Dump each function after applying dead store elimination. The file
3780 name is made by appending @file{.dse} to the source file name.
3783 @opindex fdump-tree-phiopt
3784 Dump each function after optimizing PHI nodes into straightline code. The file
3785 name is made by appending @file{.phiopt} to the source file name.
3788 @opindex fdump-tree-forwprop
3789 Dump each function after forward propagating single use variables. The file
3790 name is made by appending @file{.forwprop} to the source file name.
3793 @opindex fdump-tree-copyrename
3794 Dump each function after applying the copy rename optimization. The file
3795 name is made by appending @file{.copyrename} to the source file name.
3798 @opindex fdump-tree-nrv
3799 Dump each function after applying the named return value optimization on
3800 generic trees. The file name is made by appending @file{.nrv} to the source
3804 @opindex fdump-tree-vect
3805 Dump each function after applying vectorization of loops. The file name is
3806 made by appending @file{.vect} to the source file name.
3809 @opindex fdump-tree-all
3810 Enable all the available tree dumps with the flags provided in this option.
3813 @item -frandom-seed=@var{string}
3814 @opindex frandom-string
3815 This option provides a seed that GCC uses when it would otherwise use
3816 random numbers. It is used to generate certain symbol names
3817 that have to be different in every compiled file. It is also used to
3818 place unique stamps in coverage data files and the object files that
3819 produce them. You can use the @option{-frandom-seed} option to produce
3820 reproducibly identical object files.
3822 The @var{string} should be different for every file you compile.
3824 @item -fsched-verbose=@var{n}
3825 @opindex fsched-verbose
3826 On targets that use instruction scheduling, this option controls the
3827 amount of debugging output the scheduler prints. This information is
3828 written to standard error, unless @option{-dS} or @option{-dR} is
3829 specified, in which case it is output to the usual dump
3830 listing file, @file{.sched} or @file{.sched2} respectively. However
3831 for @var{n} greater than nine, the output is always printed to standard
3834 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3835 same information as @option{-dRS}. For @var{n} greater than one, it
3836 also output basic block probabilities, detailed ready list information
3837 and unit/insn info. For @var{n} greater than two, it includes RTL
3838 at abort point, control-flow and regions info. And for @var{n} over
3839 four, @option{-fsched-verbose} also includes dependence info.
3843 Store the usual ``temporary'' intermediate files permanently; place them
3844 in the current directory and name them based on the source file. Thus,
3845 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3846 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3847 preprocessed @file{foo.i} output file even though the compiler now
3848 normally uses an integrated preprocessor.
3852 Report the CPU time taken by each subprocess in the compilation
3853 sequence. For C source files, this is the compiler proper and assembler
3854 (plus the linker if linking is done). The output looks like this:
3861 The first number on each line is the ``user time,'' that is time spent
3862 executing the program itself. The second number is ``system time,''
3863 time spent executing operating system routines on behalf of the program.
3864 Both numbers are in seconds.
3866 @item -fvar-tracking
3867 @opindex fvar-tracking
3868 Run variable tracking pass. It computes where variables are stored at each
3869 position in code. Better debugging information is then generated
3870 (if the debugging information format supports this information).
3872 It is enabled by default when compiling with optimization (@option{-Os},
3873 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3874 the debug info format supports it.
3876 @item -print-file-name=@var{library}
3877 @opindex print-file-name
3878 Print the full absolute name of the library file @var{library} that
3879 would be used when linking---and don't do anything else. With this
3880 option, GCC does not compile or link anything; it just prints the
3883 @item -print-multi-directory
3884 @opindex print-multi-directory
3885 Print the directory name corresponding to the multilib selected by any
3886 other switches present in the command line. This directory is supposed
3887 to exist in @env{GCC_EXEC_PREFIX}.
3889 @item -print-multi-lib
3890 @opindex print-multi-lib
3891 Print the mapping from multilib directory names to compiler switches
3892 that enable them. The directory name is separated from the switches by
3893 @samp{;}, and each switch starts with an @samp{@@} instead of the
3894 @samp{-}, without spaces between multiple switches. This is supposed to
3895 ease shell-processing.
3897 @item -print-prog-name=@var{program}
3898 @opindex print-prog-name
3899 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3901 @item -print-libgcc-file-name
3902 @opindex print-libgcc-file-name
3903 Same as @option{-print-file-name=libgcc.a}.
3905 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3906 but you do want to link with @file{libgcc.a}. You can do
3909 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3912 @item -print-search-dirs
3913 @opindex print-search-dirs
3914 Print the name of the configured installation directory and a list of
3915 program and library directories @command{gcc} will search---and don't do anything else.
3917 This is useful when @command{gcc} prints the error message
3918 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3919 To resolve this you either need to put @file{cpp0} and the other compiler
3920 components where @command{gcc} expects to find them, or you can set the environment
3921 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3922 Don't forget the trailing '/'.
3923 @xref{Environment Variables}.
3926 @opindex dumpmachine
3927 Print the compiler's target machine (for example,
3928 @samp{i686-pc-linux-gnu})---and don't do anything else.
3931 @opindex dumpversion
3932 Print the compiler version (for example, @samp{3.0})---and don't do
3937 Print the compiler's built-in specs---and don't do anything else. (This
3938 is used when GCC itself is being built.) @xref{Spec Files}.
3940 @item -feliminate-unused-debug-types
3941 @opindex feliminate-unused-debug-types
3942 Normally, when producing DWARF2 output, GCC will emit debugging
3943 information for all types declared in a compilation
3944 unit, regardless of whether or not they are actually used
3945 in that compilation unit. Sometimes this is useful, such as
3946 if, in the debugger, you want to cast a value to a type that is
3947 not actually used in your program (but is declared). More often,
3948 however, this results in a significant amount of wasted space.
3949 With this option, GCC will avoid producing debug symbol output
3950 for types that are nowhere used in the source file being compiled.
3953 @node Optimize Options
3954 @section Options That Control Optimization
3955 @cindex optimize options
3956 @cindex options, optimization
3958 These options control various sorts of optimizations.
3960 Without any optimization option, the compiler's goal is to reduce the
3961 cost of compilation and to make debugging produce the expected
3962 results. Statements are independent: if you stop the program with a
3963 breakpoint between statements, you can then assign a new value to any
3964 variable or change the program counter to any other statement in the
3965 function and get exactly the results you would expect from the source
3968 Turning on optimization flags makes the compiler attempt to improve
3969 the performance and/or code size at the expense of compilation time
3970 and possibly the ability to debug the program.
3972 The compiler performs optimization based on the knowledge it has of
3973 the program. Optimization levels @option{-O2} and above, in
3974 particular, enable @emph{unit-at-a-time} mode, which allows the
3975 compiler to consider information gained from later functions in
3976 the file when compiling a function. Compiling multiple files at
3977 once to a single output file in @emph{unit-at-a-time} mode allows
3978 the compiler to use information gained from all of the files when
3979 compiling each of them.
3981 Not all optimizations are controlled directly by a flag. Only
3982 optimizations that have a flag are listed.
3989 Optimize. Optimizing compilation takes somewhat more time, and a lot
3990 more memory for a large function.
3992 With @option{-O}, the compiler tries to reduce code size and execution
3993 time, without performing any optimizations that take a great deal of
3996 @option{-O} turns on the following optimization flags:
3997 @gccoptlist{-fdefer-pop @gol
3998 -fmerge-constants @gol
4000 -floop-optimize @gol
4001 -fif-conversion @gol
4002 -fif-conversion2 @gol
4003 -fdelayed-branch @gol
4004 -fguess-branch-probability @gol
4007 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4008 where doing so does not interfere with debugging.
4012 Optimize even more. GCC performs nearly all supported optimizations
4013 that do not involve a space-speed tradeoff. The compiler does not
4014 perform loop unrolling or function inlining when you specify @option{-O2}.
4015 As compared to @option{-O}, this option increases both compilation time
4016 and the performance of the generated code.
4018 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4019 also turns on the following optimization flags:
4020 @gccoptlist{-fforce-mem @gol
4021 -foptimize-sibling-calls @gol
4022 -fstrength-reduce @gol
4023 -fcse-follow-jumps -fcse-skip-blocks @gol
4024 -frerun-cse-after-loop -frerun-loop-opt @gol
4025 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
4026 -fdelete-null-pointer-checks @gol
4027 -fexpensive-optimizations @gol
4029 -fschedule-insns -fschedule-insns2 @gol
4030 -fsched-interblock -fsched-spec @gol
4033 -freorder-blocks -freorder-functions @gol
4034 -fstrict-aliasing @gol
4035 -funit-at-a-time @gol
4036 -falign-functions -falign-jumps @gol
4037 -falign-loops -falign-labels @gol
4040 Please note the warning under @option{-fgcse} about
4041 invoking @option{-O2} on programs that use computed gotos.
4045 Optimize yet more. @option{-O3} turns on all optimizations specified by
4046 @option{-O2} and also turns on the @option{-finline-functions},
4047 @option{-fweb} and @option{-fgcse-after-reload} options.
4051 Do not optimize. This is the default.
4055 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4056 do not typically increase code size. It also performs further
4057 optimizations designed to reduce code size.
4059 @option{-Os} disables the following optimization flags:
4060 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4061 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
4063 If you use multiple @option{-O} options, with or without level numbers,
4064 the last such option is the one that is effective.
4067 Options of the form @option{-f@var{flag}} specify machine-independent
4068 flags. Most flags have both positive and negative forms; the negative
4069 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4070 below, only one of the forms is listed---the one you typically will
4071 use. You can figure out the other form by either removing @samp{no-}
4074 The following options control specific optimizations. They are either
4075 activated by @option{-O} options or are related to ones that are. You
4076 can use the following flags in the rare cases when ``fine-tuning'' of
4077 optimizations to be performed is desired.
4080 @item -fno-default-inline
4081 @opindex fno-default-inline
4082 Do not make member functions inline by default merely because they are
4083 defined inside the class scope (C++ only). Otherwise, when you specify
4084 @w{@option{-O}}, member functions defined inside class scope are compiled
4085 inline by default; i.e., you don't need to add @samp{inline} in front of
4086 the member function name.
4088 @item -fno-defer-pop
4089 @opindex fno-defer-pop
4090 Always pop the arguments to each function call as soon as that function
4091 returns. For machines which must pop arguments after a function call,
4092 the compiler normally lets arguments accumulate on the stack for several
4093 function calls and pops them all at once.
4095 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4099 Force memory operands to be copied into registers before doing
4100 arithmetic on them. This produces better code by making all memory
4101 references potential common subexpressions. When they are not common
4102 subexpressions, instruction combination should eliminate the separate
4105 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4108 @opindex fforce-addr
4109 Force memory address constants to be copied into registers before
4110 doing arithmetic on them. This may produce better code just as
4111 @option{-fforce-mem} may.
4113 @item -fomit-frame-pointer
4114 @opindex fomit-frame-pointer
4115 Don't keep the frame pointer in a register for functions that
4116 don't need one. This avoids the instructions to save, set up and
4117 restore frame pointers; it also makes an extra register available
4118 in many functions. @strong{It also makes debugging impossible on
4121 On some machines, such as the VAX, this flag has no effect, because
4122 the standard calling sequence automatically handles the frame pointer
4123 and nothing is saved by pretending it doesn't exist. The
4124 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4125 whether a target machine supports this flag. @xref{Registers,,Register
4126 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4128 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4130 @item -foptimize-sibling-calls
4131 @opindex foptimize-sibling-calls
4132 Optimize sibling and tail recursive calls.
4134 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4138 Don't pay attention to the @code{inline} keyword. Normally this option
4139 is used to keep the compiler from expanding any functions inline.
4140 Note that if you are not optimizing, no functions can be expanded inline.
4142 @item -finline-functions
4143 @opindex finline-functions
4144 Integrate all simple functions into their callers. The compiler
4145 heuristically decides which functions are simple enough to be worth
4146 integrating in this way.
4148 If all calls to a given function are integrated, and the function is
4149 declared @code{static}, then the function is normally not output as
4150 assembler code in its own right.
4152 Enabled at level @option{-O3}.
4154 @item -finline-limit=@var{n}
4155 @opindex finline-limit
4156 By default, GCC limits the size of functions that can be inlined. This flag
4157 allows the control of this limit for functions that are explicitly marked as
4158 inline (i.e., marked with the inline keyword or defined within the class
4159 definition in c++). @var{n} is the size of functions that can be inlined in
4160 number of pseudo instructions (not counting parameter handling). The default
4161 value of @var{n} is 600.
4162 Increasing this value can result in more inlined code at
4163 the cost of compilation time and memory consumption. Decreasing usually makes
4164 the compilation faster and less code will be inlined (which presumably
4165 means slower programs). This option is particularly useful for programs that
4166 use inlining heavily such as those based on recursive templates with C++.
4168 Inlining is actually controlled by a number of parameters, which may be
4169 specified individually by using @option{--param @var{name}=@var{value}}.
4170 The @option{-finline-limit=@var{n}} option sets some of these parameters
4174 @item max-inline-insns-single
4175 is set to @var{n}/2.
4176 @item max-inline-insns-auto
4177 is set to @var{n}/2.
4178 @item min-inline-insns
4179 is set to 130 or @var{n}/4, whichever is smaller.
4180 @item max-inline-insns-rtl
4184 See below for a documentation of the individual
4185 parameters controlling inlining.
4187 @emph{Note:} pseudo instruction represents, in this particular context, an
4188 abstract measurement of function's size. In no way, it represents a count
4189 of assembly instructions and as such its exact meaning might change from one
4190 release to an another.
4192 @item -fkeep-inline-functions
4193 @opindex fkeep-inline-functions
4194 In C, emit @code{static} functions that are declared @code{inline}
4195 into the object file, even if the function has been inlined into all
4196 of its callers. This switch does not affect functions using the
4197 @code{extern inline} extension in GNU C. In C++, emit any and all
4198 inline functions into the object file.
4200 @item -fkeep-static-consts
4201 @opindex fkeep-static-consts
4202 Emit variables declared @code{static const} when optimization isn't turned
4203 on, even if the variables aren't referenced.
4205 GCC enables this option by default. If you want to force the compiler to
4206 check if the variable was referenced, regardless of whether or not
4207 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4209 @item -fmerge-constants
4210 Attempt to merge identical constants (string constants and floating point
4211 constants) across compilation units.
4213 This option is the default for optimized compilation if the assembler and
4214 linker support it. Use @option{-fno-merge-constants} to inhibit this
4217 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4219 @item -fmerge-all-constants
4220 Attempt to merge identical constants and identical variables.
4222 This option implies @option{-fmerge-constants}. In addition to
4223 @option{-fmerge-constants} this considers e.g. even constant initialized
4224 arrays or initialized constant variables with integral or floating point
4225 types. Languages like C or C++ require each non-automatic variable to
4226 have distinct location, so using this option will result in non-conforming
4229 @item -fmodulo-sched
4230 @opindex fmodulo-sched
4231 Perform swing modulo scheduling immediately before the first scheduling
4232 pass. This pass looks at innermost loops and reorders their
4233 instructions by overlapping different iterations.
4237 Use a graph coloring register allocator. Currently this option is meant
4238 only for testing. Users should not specify this option, since it is not
4239 yet ready for production use.
4241 @item -fno-branch-count-reg
4242 @opindex fno-branch-count-reg
4243 Do not use ``decrement and branch'' instructions on a count register,
4244 but instead generate a sequence of instructions that decrement a
4245 register, compare it against zero, then branch based upon the result.
4246 This option is only meaningful on architectures that support such
4247 instructions, which include x86, PowerPC, IA-64 and S/390.
4249 The default is @option{-fbranch-count-reg}, enabled when
4250 @option{-fstrength-reduce} is enabled.
4252 @item -fno-function-cse
4253 @opindex fno-function-cse
4254 Do not put function addresses in registers; make each instruction that
4255 calls a constant function contain the function's address explicitly.
4257 This option results in less efficient code, but some strange hacks
4258 that alter the assembler output may be confused by the optimizations
4259 performed when this option is not used.
4261 The default is @option{-ffunction-cse}
4263 @item -fno-zero-initialized-in-bss
4264 @opindex fno-zero-initialized-in-bss
4265 If the target supports a BSS section, GCC by default puts variables that
4266 are initialized to zero into BSS@. This can save space in the resulting
4269 This option turns off this behavior because some programs explicitly
4270 rely on variables going to the data section. E.g., so that the
4271 resulting executable can find the beginning of that section and/or make
4272 assumptions based on that.
4274 The default is @option{-fzero-initialized-in-bss}.
4276 @item -fbounds-check
4277 @opindex fbounds-check
4278 For front-ends that support it, generate additional code to check that
4279 indices used to access arrays are within the declared range. This is
4280 currently only supported by the Java and Fortran front-ends, where
4281 this option defaults to true and false respectively.
4283 @item -fmudflap -fmudflapth -fmudflapir
4287 @cindex bounds checking
4289 For front-ends that support it (C and C++), instrument all risky
4290 pointer/array dereferencing operations, some standard library
4291 string/heap functions, and some other associated constructs with
4292 range/validity tests. Modules so instrumented should be immune to
4293 buffer overflows, invalid heap use, and some other classes of C/C++
4294 programming errors. The instrumentation relies on a separate runtime
4295 library (@file{libmudflap}), which will be linked into a program if
4296 @option{-fmudflap} is given at link time. Run-time behavior of the
4297 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4298 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4301 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4302 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4303 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4304 instrumentation should ignore pointer reads. This produces less
4305 instrumentation (and therefore faster execution) and still provides
4306 some protection against outright memory corrupting writes, but allows
4307 erroneously read data to propagate within a program.
4309 @item -fstrength-reduce
4310 @opindex fstrength-reduce
4311 Perform the optimizations of loop strength reduction and
4312 elimination of iteration variables.
4314 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4316 @item -fthread-jumps
4317 @opindex fthread-jumps
4318 Perform optimizations where we check to see if a jump branches to a
4319 location where another comparison subsumed by the first is found. If
4320 so, the first branch is redirected to either the destination of the
4321 second branch or a point immediately following it, depending on whether
4322 the condition is known to be true or false.
4324 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4326 @item -fcse-follow-jumps
4327 @opindex fcse-follow-jumps
4328 In common subexpression elimination, scan through jump instructions
4329 when the target of the jump is not reached by any other path. For
4330 example, when CSE encounters an @code{if} statement with an
4331 @code{else} clause, CSE will follow the jump when the condition
4334 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4336 @item -fcse-skip-blocks
4337 @opindex fcse-skip-blocks
4338 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4339 follow jumps which conditionally skip over blocks. When CSE
4340 encounters a simple @code{if} statement with no else clause,
4341 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4342 body of the @code{if}.
4344 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4346 @item -frerun-cse-after-loop
4347 @opindex frerun-cse-after-loop
4348 Re-run common subexpression elimination after loop optimizations has been
4351 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4353 @item -frerun-loop-opt
4354 @opindex frerun-loop-opt
4355 Run the loop optimizer twice.
4357 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4361 Perform a global common subexpression elimination pass.
4362 This pass also performs global constant and copy propagation.
4364 @emph{Note:} When compiling a program using computed gotos, a GCC
4365 extension, you may get better runtime performance if you disable
4366 the global common subexpression elimination pass by adding
4367 @option{-fno-gcse} to the command line.
4369 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4373 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4374 attempt to move loads which are only killed by stores into themselves. This
4375 allows a loop containing a load/store sequence to be changed to a load outside
4376 the loop, and a copy/store within the loop.
4378 Enabled by default when gcse is enabled.
4382 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4383 global common subexpression elimination. This pass will attempt to move
4384 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4385 loops containing a load/store sequence can be changed to a load before
4386 the loop and a store after the loop.
4388 Enabled by default when gcse is enabled.
4392 When @option{-fgcse-las} is enabled, the global common subexpression
4393 elimination pass eliminates redundant loads that come after stores to the
4394 same memory location (both partial and full redundancies).
4396 Enabled by default when gcse is enabled.
4398 @item -fgcse-after-reload
4399 @opindex fgcse-after-reload
4400 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4401 pass is performed after reload. The purpose of this pass is to cleanup
4404 @item -floop-optimize
4405 @opindex floop-optimize
4406 Perform loop optimizations: move constant expressions out of loops, simplify
4407 exit test conditions and optionally do strength-reduction and loop unrolling as
4410 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4412 @item -floop-optimize2
4413 @opindex floop-optimize2
4414 Perform loop optimizations using the new loop optimizer. The optimizations
4415 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4418 @item -fcrossjumping
4419 @opindex crossjumping
4420 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4421 resulting code may or may not perform better than without cross-jumping.
4423 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4425 @item -fif-conversion
4426 @opindex if-conversion
4427 Attempt to transform conditional jumps into branch-less equivalents. This
4428 include use of conditional moves, min, max, set flags and abs instructions, and
4429 some tricks doable by standard arithmetics. The use of conditional execution
4430 on chips where it is available is controlled by @code{if-conversion2}.
4432 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4434 @item -fif-conversion2
4435 @opindex if-conversion2
4436 Use conditional execution (where available) to transform conditional jumps into
4437 branch-less equivalents.
4439 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4441 @item -fdelete-null-pointer-checks
4442 @opindex fdelete-null-pointer-checks
4443 Use global dataflow analysis to identify and eliminate useless checks
4444 for null pointers. The compiler assumes that dereferencing a null
4445 pointer would have halted the program. If a pointer is checked after
4446 it has already been dereferenced, it cannot be null.
4448 In some environments, this assumption is not true, and programs can
4449 safely dereference null pointers. Use
4450 @option{-fno-delete-null-pointer-checks} to disable this optimization
4451 for programs which depend on that behavior.
4453 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4455 @item -fexpensive-optimizations
4456 @opindex fexpensive-optimizations
4457 Perform a number of minor optimizations that are relatively expensive.
4459 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4461 @item -foptimize-register-move
4463 @opindex foptimize-register-move
4465 Attempt to reassign register numbers in move instructions and as
4466 operands of other simple instructions in order to maximize the amount of
4467 register tying. This is especially helpful on machines with two-operand
4470 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4473 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4475 @item -fdelayed-branch
4476 @opindex fdelayed-branch
4477 If supported for the target machine, attempt to reorder instructions
4478 to exploit instruction slots available after delayed branch
4481 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4483 @item -fschedule-insns
4484 @opindex fschedule-insns
4485 If supported for the target machine, attempt to reorder instructions to
4486 eliminate execution stalls due to required data being unavailable. This
4487 helps machines that have slow floating point or memory load instructions
4488 by allowing other instructions to be issued until the result of the load
4489 or floating point instruction is required.
4491 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4493 @item -fschedule-insns2
4494 @opindex fschedule-insns2
4495 Similar to @option{-fschedule-insns}, but requests an additional pass of
4496 instruction scheduling after register allocation has been done. This is
4497 especially useful on machines with a relatively small number of
4498 registers and where memory load instructions take more than one cycle.
4500 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4502 @item -fno-sched-interblock
4503 @opindex fno-sched-interblock
4504 Don't schedule instructions across basic blocks. This is normally
4505 enabled by default when scheduling before register allocation, i.e.@:
4506 with @option{-fschedule-insns} or at @option{-O2} or higher.
4508 @item -fno-sched-spec
4509 @opindex fno-sched-spec
4510 Don't allow speculative motion of non-load instructions. This is normally
4511 enabled by default when scheduling before register allocation, i.e.@:
4512 with @option{-fschedule-insns} or at @option{-O2} or higher.
4514 @item -fsched-spec-load
4515 @opindex fsched-spec-load
4516 Allow speculative motion of some load instructions. This only makes
4517 sense when scheduling before register allocation, i.e.@: with
4518 @option{-fschedule-insns} or at @option{-O2} or higher.
4520 @item -fsched-spec-load-dangerous
4521 @opindex fsched-spec-load-dangerous
4522 Allow speculative motion of more load instructions. This only makes
4523 sense when scheduling before register allocation, i.e.@: with
4524 @option{-fschedule-insns} or at @option{-O2} or higher.
4526 @item -fsched-stalled-insns=@var{n}
4527 @opindex fsched-stalled-insns
4528 Define how many insns (if any) can be moved prematurely from the queue
4529 of stalled insns into the ready list, during the second scheduling pass.
4531 @item -fsched-stalled-insns-dep=@var{n}
4532 @opindex fsched-stalled-insns-dep
4533 Define how many insn groups (cycles) will be examined for a dependency
4534 on a stalled insn that is candidate for premature removal from the queue
4535 of stalled insns. Has an effect only during the second scheduling pass,
4536 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4538 @item -fsched2-use-superblocks
4539 @opindex fsched2-use-superblocks
4540 When scheduling after register allocation, do use superblock scheduling
4541 algorithm. Superblock scheduling allows motion across basic block boundaries
4542 resulting on faster schedules. This option is experimental, as not all machine
4543 descriptions used by GCC model the CPU closely enough to avoid unreliable
4544 results from the algorithm.
4546 This only makes sense when scheduling after register allocation, i.e.@: with
4547 @option{-fschedule-insns2} or at @option{-O2} or higher.
4549 @item -fsched2-use-traces
4550 @opindex fsched2-use-traces
4551 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4552 allocation and additionally perform code duplication in order to increase the
4553 size of superblocks using tracer pass. See @option{-ftracer} for details on
4556 This mode should produce faster but significantly longer programs. Also
4557 without @code{-fbranch-probabilities} the traces constructed may not match the
4558 reality and hurt the performance. This only makes
4559 sense when scheduling after register allocation, i.e.@: with
4560 @option{-fschedule-insns2} or at @option{-O2} or higher.
4562 @item -freschedule-modulo-scheduled-loops
4563 @opindex fscheduling-in-modulo-scheduled-loops
4564 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4565 we may want to prevent the later scheduling passes from changing its schedule, we use this
4566 option to control that.
4568 @item -fcaller-saves
4569 @opindex fcaller-saves
4570 Enable values to be allocated in registers that will be clobbered by
4571 function calls, by emitting extra instructions to save and restore the
4572 registers around such calls. Such allocation is done only when it
4573 seems to result in better code than would otherwise be produced.
4575 This option is always enabled by default on certain machines, usually
4576 those which have no call-preserved registers to use instead.
4578 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4581 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4582 enabled by default at -O and higher.
4585 Perform Full Redundancy Elimination (FRE) on trees. The difference
4586 between FRE and PRE is that FRE only considers expressions
4587 that are computed on all paths leading to the redundant computation.
4588 This analysis faster than PRE, though it exposes fewer redundancies.
4589 This flag is enabled by default at -O and higher.
4592 Perform sparse conditional constant propagation (CCP) on trees. This flag
4593 is enabled by default at -O and higher.
4596 Perform dead code elimination (DCE) on trees. This flag is enabled by
4597 default at -O and higher.
4599 @item -ftree-dominator-opts
4600 Perform dead code elimination (DCE) on trees. This flag is enabled by
4601 default at -O and higher.
4604 Perform loop header copying on trees. This is beneficial since it increases
4605 effectivity of code motion optimizations. It also saves one jump. This flag
4606 is enabled by default at -O and higher. It is not enabled for -Os, since it
4607 usually increases code size.
4609 @item -ftree-loop-optimize
4610 Perform loop optimizations on trees. This flag is enabled by default at -O
4614 Perform loop invariant motion on trees. This pass moves only invartiants that
4615 would be hard to handle on rtl level (function calls, operations that expand to
4616 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
4617 operands of conditions that are invariant out of the loop, so that we can use
4618 just trivial invariantness analysis in loop unswitching. The pass also includes
4622 Create a canonical counter for number of iterations in the loop for that
4623 determining number of iterations requires complicated analysis. Later
4624 optimizations then may determine the number easily. Useful especially
4625 in connection with unrolling.
4628 Perform induction variable optimizations (strength reduction, induction
4629 variable merging and induction variable elimination) on trees.
4632 Perform scalar replacement of aggregates. This pass replaces structure
4633 references with scalars to prevent committing structures to memory too
4634 early. This flag is enabled by default at -O and higher.
4636 @item -ftree-copyrename
4637 Perform copy renaming on trees. This pass attempts to rename compiler
4638 temporaries to other variables at copy locations, usually resulting in
4639 variable names which more closely resemble the original variables. This flag
4640 is enabled by default at -O and higher.
4643 Perform temporary expression replacement during the SSA->normal phase. Single
4644 use/single def temporaries are replaced at their use location with their
4645 defining expression. This results in non-GIMPLE code, but gives the expanders
4646 much more complex trees to work on resulting in better RTL generation. This is
4647 enabled by default at -O and higher.
4650 Perform live range splitting during the SSA->normal phase. Distinct live
4651 ranges of a variable are split into unique variables, allowing for better
4652 optimization later. This is enabled by default at -O and higher.
4654 @item -ftree-vectorize
4655 Perform loop vectorization on trees.
4659 Perform tail duplication to enlarge superblock size. This transformation
4660 simplifies the control flow of the function allowing other optimizations to do
4663 @item -funroll-loops
4664 @opindex funroll-loops
4665 Unroll loops whose number of iterations can be determined at compile
4666 time or upon entry to the loop. @option{-funroll-loops} implies both
4667 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4668 option makes code larger, and may or may not make it run faster.
4670 @item -funroll-all-loops
4671 @opindex funroll-all-loops
4672 Unroll all loops, even if their number of iterations is uncertain when
4673 the loop is entered. This usually makes programs run more slowly.
4674 @option{-funroll-all-loops} implies the same options as
4675 @option{-funroll-loops},
4677 @item -fprefetch-loop-arrays
4678 @opindex fprefetch-loop-arrays
4679 If supported by the target machine, generate instructions to prefetch
4680 memory to improve the performance of loops that access large arrays.
4682 @item -fmove-all-movables
4683 @opindex fmove-all-movables
4684 Forces all invariant computations in loops to be moved
4687 @item -freduce-all-givs
4688 @opindex freduce-all-givs
4689 Forces all general-induction variables in loops to be
4692 @emph{Note:} When compiling programs written in Fortran,
4693 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4694 by default when you use the optimizer.
4696 These options may generate better or worse code; results are highly
4697 dependent on the structure of loops within the source code.
4699 These two options are intended to be removed someday, once
4700 they have helped determine the efficacy of various
4701 approaches to improving loop optimizations.
4703 Please contact @w{@email{gcc@@gcc.gnu.org}}, and describe how use of
4704 these options affects the performance of your production code.
4705 Examples of code that runs @emph{slower} when these options are
4706 @emph{enabled} are very valuable.
4709 @itemx -fno-peephole2
4710 @opindex fno-peephole
4711 @opindex fno-peephole2
4712 Disable any machine-specific peephole optimizations. The difference
4713 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4714 are implemented in the compiler; some targets use one, some use the
4715 other, a few use both.
4717 @option{-fpeephole} is enabled by default.
4718 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4720 @item -fno-guess-branch-probability
4721 @opindex fno-guess-branch-probability
4722 Do not guess branch probabilities using a randomized model.
4724 Sometimes GCC will opt to use a randomized model to guess branch
4725 probabilities, when none are available from either profiling feedback
4726 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4727 different runs of the compiler on the same program may produce different
4730 In a hard real-time system, people don't want different runs of the
4731 compiler to produce code that has different behavior; minimizing
4732 non-determinism is of paramount import. This switch allows users to
4733 reduce non-determinism, possibly at the expense of inferior
4736 The default is @option{-fguess-branch-probability} at levels
4737 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4739 @item -freorder-blocks
4740 @opindex freorder-blocks
4741 Reorder basic blocks in the compiled function in order to reduce number of
4742 taken branches and improve code locality.
4744 Enabled at levels @option{-O2}, @option{-O3}.
4746 @item -freorder-blocks-and-partition
4747 @opindex freorder-blocks-and-partition
4748 In addition to reordering basic blocks in the compiled function, in order
4749 to reduce number of taken branches, partitions hot and cold basic blocks
4750 into separate sections of the assembly and .o files, to improve
4751 paging and cache locality performance.
4753 This optimization is automatically turned off in the presence of
4754 exception handling, for linkonce sections, for functions with a user-defined
4755 section attribute and on any architecture that does not support named
4758 @item -freorder-functions
4759 @opindex freorder-functions
4760 Reorder basic blocks in the compiled function in order to reduce number of
4761 taken branches and improve code locality. This is implemented by using special
4762 subsections @code{.text.hot} for most frequently executed functions and
4763 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4764 the linker so object file format must support named sections and linker must
4765 place them in a reasonable way.
4767 Also profile feedback must be available in to make this option effective. See
4768 @option{-fprofile-arcs} for details.
4770 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4772 @item -fstrict-aliasing
4773 @opindex fstrict-aliasing
4774 Allows the compiler to assume the strictest aliasing rules applicable to
4775 the language being compiled. For C (and C++), this activates
4776 optimizations based on the type of expressions. In particular, an
4777 object of one type is assumed never to reside at the same address as an
4778 object of a different type, unless the types are almost the same. For
4779 example, an @code{unsigned int} can alias an @code{int}, but not a
4780 @code{void*} or a @code{double}. A character type may alias any other
4783 Pay special attention to code like this:
4796 The practice of reading from a different union member than the one most
4797 recently written to (called ``type-punning'') is common. Even with
4798 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4799 is accessed through the union type. So, the code above will work as
4800 expected. However, this code might not:
4811 Every language that wishes to perform language-specific alias analysis
4812 should define a function that computes, given an @code{tree}
4813 node, an alias set for the node. Nodes in different alias sets are not
4814 allowed to alias. For an example, see the C front-end function
4815 @code{c_get_alias_set}.
4817 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4819 @item -falign-functions
4820 @itemx -falign-functions=@var{n}
4821 @opindex falign-functions
4822 Align the start of functions to the next power-of-two greater than
4823 @var{n}, skipping up to @var{n} bytes. For instance,
4824 @option{-falign-functions=32} aligns functions to the next 32-byte
4825 boundary, but @option{-falign-functions=24} would align to the next
4826 32-byte boundary only if this can be done by skipping 23 bytes or less.
4828 @option{-fno-align-functions} and @option{-falign-functions=1} are
4829 equivalent and mean that functions will not be aligned.
4831 Some assemblers only support this flag when @var{n} is a power of two;
4832 in that case, it is rounded up.
4834 If @var{n} is not specified or is zero, use a machine-dependent default.
4836 Enabled at levels @option{-O2}, @option{-O3}.
4838 @item -falign-labels
4839 @itemx -falign-labels=@var{n}
4840 @opindex falign-labels
4841 Align all branch targets to a power-of-two boundary, skipping up to
4842 @var{n} bytes like @option{-falign-functions}. This option can easily
4843 make code slower, because it must insert dummy operations for when the
4844 branch target is reached in the usual flow of the code.
4846 @option{-fno-align-labels} and @option{-falign-labels=1} are
4847 equivalent and mean that labels will not be aligned.
4849 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4850 are greater than this value, then their values are used instead.
4852 If @var{n} is not specified or is zero, use a machine-dependent default
4853 which is very likely to be @samp{1}, meaning no alignment.
4855 Enabled at levels @option{-O2}, @option{-O3}.
4858 @itemx -falign-loops=@var{n}
4859 @opindex falign-loops
4860 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4861 like @option{-falign-functions}. The hope is that the loop will be
4862 executed many times, which will make up for any execution of the dummy
4865 @option{-fno-align-loops} and @option{-falign-loops=1} are
4866 equivalent and mean that loops will not be aligned.
4868 If @var{n} is not specified or is zero, use a machine-dependent default.
4870 Enabled at levels @option{-O2}, @option{-O3}.
4873 @itemx -falign-jumps=@var{n}
4874 @opindex falign-jumps
4875 Align branch targets to a power-of-two boundary, for branch targets
4876 where the targets can only be reached by jumping, skipping up to @var{n}
4877 bytes like @option{-falign-functions}. In this case, no dummy operations
4880 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4881 equivalent and mean that loops will not be aligned.
4883 If @var{n} is not specified or is zero, use a machine-dependent default.
4885 Enabled at levels @option{-O2}, @option{-O3}.
4887 @item -funit-at-a-time
4888 @opindex funit-at-a-time
4889 Parse the whole compilation unit before starting to produce code.
4890 This allows some extra optimizations to take place but consumes
4891 more memory (in general). There are some compatibility issues
4892 with @emph{unit-at-at-time} mode:
4895 enabling @emph{unit-at-a-time} mode may change the order
4896 in which functions, variables, and top-level @code{asm} statements
4897 are emitted, and will likely break code relying on some particular
4898 ordering. The majority of such top-level @code{asm} statements,
4899 though, can be replaced by @code{section} attributes.
4902 @emph{unit-at-a-time} mode removes unreferenced static variables
4903 and functions are removed. This may result in undefined references
4904 when an @code{asm} statement refers directly to variables or functions
4905 that are otherwise unused. In that case either the variable/function
4906 shall be listed as an operand of the @code{asm} statement operand or,
4907 in the case of top-level @code{asm} statements the attribute @code{used}
4908 shall be used on the declaration.
4911 Static functions now can use non-standard passing conventions that
4912 may break @code{asm} statements calling functions directly. Again,
4913 attribute @code{used} will prevent this behavior.
4916 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
4917 but this scheme may not be supported by future releases of GCC.
4919 Enabled at levels @option{-O2}, @option{-O3}.
4923 Constructs webs as commonly used for register allocation purposes and assign
4924 each web individual pseudo register. This allows the register allocation pass
4925 to operate on pseudos directly, but also strengthens several other optimization
4926 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4927 however, make debugging impossible, since variables will no longer stay in a
4930 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
4931 on targets where the default format for debugging information supports
4934 @item -fno-cprop-registers
4935 @opindex fno-cprop-registers
4936 After register allocation and post-register allocation instruction splitting,
4937 we perform a copy-propagation pass to try to reduce scheduling dependencies
4938 and occasionally eliminate the copy.
4940 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4942 @item -fprofile-generate
4943 @opindex fprofile-generate
4945 Enable options usually used for instrumenting application to produce
4946 profile useful for later recompilation with profile feedback based
4947 optimization. You must use @code{-fprofile-generate} both when
4948 compiling and when linking your program.
4950 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
4953 @opindex fprofile-use
4954 Enable profile feedback directed optimizations, and optimizations
4955 generally profitable only with profile feedback available.
4957 The following options are enabled: @code{-fbranch-probabilities},
4958 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4962 The following options control compiler behavior regarding floating
4963 point arithmetic. These options trade off between speed and
4964 correctness. All must be specifically enabled.
4968 @opindex ffloat-store
4969 Do not store floating point variables in registers, and inhibit other
4970 options that might change whether a floating point value is taken from a
4973 @cindex floating point precision
4974 This option prevents undesirable excess precision on machines such as
4975 the 68000 where the floating registers (of the 68881) keep more
4976 precision than a @code{double} is supposed to have. Similarly for the
4977 x86 architecture. For most programs, the excess precision does only
4978 good, but a few programs rely on the precise definition of IEEE floating
4979 point. Use @option{-ffloat-store} for such programs, after modifying
4980 them to store all pertinent intermediate computations into variables.
4984 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4985 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4986 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4988 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4990 This option should never be turned on by any @option{-O} option since
4991 it can result in incorrect output for programs which depend on
4992 an exact implementation of IEEE or ISO rules/specifications for
4995 @item -fno-math-errno
4996 @opindex fno-math-errno
4997 Do not set ERRNO after calling math functions that are executed
4998 with a single instruction, e.g., sqrt. A program that relies on
4999 IEEE exceptions for math error handling may want to use this flag
5000 for speed while maintaining IEEE arithmetic compatibility.
5002 This option should never be turned on by any @option{-O} option since
5003 it can result in incorrect output for programs which depend on
5004 an exact implementation of IEEE or ISO rules/specifications for
5007 The default is @option{-fmath-errno}.
5009 @item -funsafe-math-optimizations
5010 @opindex funsafe-math-optimizations
5011 Allow optimizations for floating-point arithmetic that (a) assume
5012 that arguments and results are valid and (b) may violate IEEE or
5013 ANSI standards. When used at link-time, it may include libraries
5014 or startup files that change the default FPU control word or other
5015 similar optimizations.
5017 This option should never be turned on by any @option{-O} option since
5018 it can result in incorrect output for programs which depend on
5019 an exact implementation of IEEE or ISO rules/specifications for
5022 The default is @option{-fno-unsafe-math-optimizations}.
5024 @item -ffinite-math-only
5025 @opindex ffinite-math-only
5026 Allow optimizations for floating-point arithmetic that assume
5027 that arguments and results are not NaNs or +-Infs.
5029 This option should never be turned on by any @option{-O} option since
5030 it can result in incorrect output for programs which depend on
5031 an exact implementation of IEEE or ISO rules/specifications.
5033 The default is @option{-fno-finite-math-only}.
5035 @item -fno-trapping-math
5036 @opindex fno-trapping-math
5037 Compile code assuming that floating-point operations cannot generate
5038 user-visible traps. These traps include division by zero, overflow,
5039 underflow, inexact result and invalid operation. This option implies
5040 @option{-fno-signaling-nans}. Setting this option may allow faster
5041 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5043 This option should never be turned on by any @option{-O} option since
5044 it can result in incorrect output for programs which depend on
5045 an exact implementation of IEEE or ISO rules/specifications for
5048 The default is @option{-ftrapping-math}.
5050 @item -frounding-math
5051 @opindex frounding-math
5052 Disable transformations and optimizations that assume default floating
5053 point rounding behavior. This is round-to-zero for all floating point
5054 to integer conversions, and round-to-nearest for all other arithmetic
5055 truncations. This option should be specified for programs that change
5056 the FP rounding mode dynamically, or that may be executed with a
5057 non-default rounding mode. This option disables constant folding of
5058 floating point expressions at compile-time (which may be affected by
5059 rounding mode) and arithmetic transformations that are unsafe in the
5060 presence of sign-dependent rounding modes.
5062 The default is @option{-fno-rounding-math}.
5064 This option is experimental and does not currently guarantee to
5065 disable all GCC optimizations that are affected by rounding mode.
5066 Future versions of GCC may provide finer control of this setting
5067 using C99's @code{FENV_ACCESS} pragma. This command line option
5068 will be used to specify the default state for @code{FENV_ACCESS}.
5070 @item -fsignaling-nans
5071 @opindex fsignaling-nans
5072 Compile code assuming that IEEE signaling NaNs may generate user-visible
5073 traps during floating-point operations. Setting this option disables
5074 optimizations that may change the number of exceptions visible with
5075 signaling NaNs. This option implies @option{-ftrapping-math}.
5077 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5080 The default is @option{-fno-signaling-nans}.
5082 This option is experimental and does not currently guarantee to
5083 disable all GCC optimizations that affect signaling NaN behavior.
5085 @item -fsingle-precision-constant
5086 @opindex fsingle-precision-constant
5087 Treat floating point constant as single precision constant instead of
5088 implicitly converting it to double precision constant.
5093 The following options control optimizations that may improve
5094 performance, but are not enabled by any @option{-O} options. This
5095 section includes experimental options that may produce broken code.
5098 @item -fbranch-probabilities
5099 @opindex fbranch-probabilities
5100 After running a program compiled with @option{-fprofile-arcs}
5101 (@pxref{Debugging Options,, Options for Debugging Your Program or
5102 @command{gcc}}), you can compile it a second time using
5103 @option{-fbranch-probabilities}, to improve optimizations based on
5104 the number of times each branch was taken. When the program
5105 compiled with @option{-fprofile-arcs} exits it saves arc execution
5106 counts to a file called @file{@var{sourcename}.gcda} for each source
5107 file The information in this data file is very dependent on the
5108 structure of the generated code, so you must use the same source code
5109 and the same optimization options for both compilations.
5111 With @option{-fbranch-probabilities}, GCC puts a
5112 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5113 These can be used to improve optimization. Currently, they are only
5114 used in one place: in @file{reorg.c}, instead of guessing which path a
5115 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5116 exactly determine which path is taken more often.
5118 @item -fprofile-values
5119 @opindex fprofile-values
5120 If combined with @option{-fprofile-arcs}, it adds code so that some
5121 data about values of expressions in the program is gathered.
5123 With @option{-fbranch-probabilities}, it reads back the data gathered
5124 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5125 notes to instructions for their later usage in optimizations.
5127 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5131 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5132 a code to gather information about values of expressions.
5134 With @option{-fbranch-probabilities}, it reads back the data gathered
5135 and actually performs the optimizations based on them.
5136 Currently the optimizations include specialization of division operation
5137 using the knowledge about the value of the denominator.
5139 @item -fspeculative-prefetching
5140 @opindex fspeculative-prefetching
5141 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5142 a code to gather information about addresses of memory references in the
5145 With @option{-fbranch-probabilities}, it reads back the data gathered
5146 and issues prefetch instructions according to them. In addition to the opportunities
5147 noticed by @option{-fprefetch-loop-arrays}, it also notices more complicated
5148 memory access patterns -- for example accesses to the data stored in linked
5149 list whose elements are usually allocated sequentially.
5151 In order to prevent issuing double prefetches, usage of
5152 @option{-fspeculative-prefetching} implies @option{-fno-prefetch-loop-arrays}.
5154 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5156 @item -frename-registers
5157 @opindex frename-registers
5158 Attempt to avoid false dependencies in scheduled code by making use
5159 of registers left over after register allocation. This optimization
5160 will most benefit processors with lots of registers. Depending on the
5161 debug information format adopted by the target, however, it can
5162 make debugging impossible, since variables will no longer stay in
5163 a ``home register''.
5165 Not enabled by default at any level because it has known bugs.
5169 Use a graph coloring register allocator. Currently this option is meant
5170 for testing, so we are interested to hear about miscompilations with
5175 Perform tail duplication to enlarge superblock size. This transformation
5176 simplifies the control flow of the function allowing other optimizations to do
5179 Enabled with @option{-fprofile-use}.
5181 @item -funroll-loops
5182 @opindex funroll-loops
5183 Unroll loops whose number of iterations can be determined at compile time or
5184 upon entry to the loop. @option{-funroll-loops} implies
5185 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
5186 (i.e. complete removal of loops with small constant number of iterations).
5187 This option makes code larger, and may or may not make it run faster.
5189 Enabled with @option{-fprofile-use}.
5191 @item -funroll-all-loops
5192 @opindex funroll-all-loops
5193 Unroll all loops, even if their number of iterations is uncertain when
5194 the loop is entered. This usually makes programs run more slowly.
5195 @option{-funroll-all-loops} implies the same options as
5196 @option{-funroll-loops}.
5199 @opindex fpeel-loops
5200 Peels the loops for that there is enough information that they do not
5201 roll much (from profile feedback). It also turns on complete loop peeling
5202 (i.e. complete removal of loops with small constant number of iterations).
5204 Enabled with @option{-fprofile-use}.
5206 @item -fmove-loop-invariants
5207 @opindex fmove-loop-invariants
5208 Enables the loop invariant motion pass in the new loop optimizer. Enabled
5209 at level @option{-O1}
5211 @item -funswitch-loops
5212 @opindex funswitch-loops
5213 Move branches with loop invariant conditions out of the loop, with duplicates
5214 of the loop on both branches (modified according to result of the condition).
5216 @item -fold-unroll-loops
5217 @opindex fold-unroll-loops
5218 Unroll loops whose number of iterations can be determined at compile
5219 time or upon entry to the loop, using the old loop unroller whose loop
5220 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
5221 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
5222 option makes code larger, and may or may not make it run faster.
5224 @item -fold-unroll-all-loops
5225 @opindex fold-unroll-all-loops
5226 Unroll all loops, even if their number of iterations is uncertain when
5227 the loop is entered. This is done using the old loop unroller whose loop
5228 recognition is based on notes from frontend. This usually makes programs run more slowly.
5229 @option{-fold-unroll-all-loops} implies the same options as
5230 @option{-fold-unroll-loops}.
5232 @item -fprefetch-loop-arrays
5233 @opindex fprefetch-loop-arrays
5234 If supported by the target machine, generate instructions to prefetch
5235 memory to improve the performance of loops that access large arrays.
5237 Disabled at level @option{-Os}.
5239 @item -ffunction-sections
5240 @itemx -fdata-sections
5241 @opindex ffunction-sections
5242 @opindex fdata-sections
5243 Place each function or data item into its own section in the output
5244 file if the target supports arbitrary sections. The name of the
5245 function or the name of the data item determines the section's name
5248 Use these options on systems where the linker can perform optimizations
5249 to improve locality of reference in the instruction space. Most systems
5250 using the ELF object format and SPARC processors running Solaris 2 have
5251 linkers with such optimizations. AIX may have these optimizations in
5254 Only use these options when there are significant benefits from doing
5255 so. When you specify these options, the assembler and linker will
5256 create larger object and executable files and will also be slower.
5257 You will not be able to use @code{gprof} on all systems if you
5258 specify this option and you may have problems with debugging if
5259 you specify both this option and @option{-g}.
5261 @item -fbranch-target-load-optimize
5262 @opindex fbranch-target-load-optimize
5263 Perform branch target register load optimization before prologue / epilogue
5265 The use of target registers can typically be exposed only during reload,
5266 thus hoisting loads out of loops and doing inter-block scheduling needs
5267 a separate optimization pass.
5269 @item -fbranch-target-load-optimize2
5270 @opindex fbranch-target-load-optimize2
5271 Perform branch target register load optimization after prologue / epilogue
5274 @item -fbtr-bb-exclusive
5275 @opindex fbtr-bb-exclusive
5276 When performing branch target register load optimization, don't reuse
5277 branch target registers in within any basic block.
5279 @item --param @var{name}=@var{value}
5281 In some places, GCC uses various constants to control the amount of
5282 optimization that is done. For example, GCC will not inline functions
5283 that contain more that a certain number of instructions. You can
5284 control some of these constants on the command-line using the
5285 @option{--param} option.
5287 The names of specific parameters, and the meaning of the values, are
5288 tied to the internals of the compiler, and are subject to change
5289 without notice in future releases.
5291 In each case, the @var{value} is an integer. The allowable choices for
5292 @var{name} are given in the following table:
5295 @item max-crossjump-edges
5296 The maximum number of incoming edges to consider for crossjumping.
5297 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5298 the number of edges incoming to each block. Increasing values mean
5299 more aggressive optimization, making the compile time increase with
5300 probably small improvement in executable size.
5302 @item min-crossjump-insns
5303 The minimum number of instructions which must be matched at the end
5304 of two blocks before crossjumping will be performed on them. This
5305 value is ignored in the case where all instructions in the block being
5306 crossjumped from are matched. The default value is 5.
5308 @item max-delay-slot-insn-search
5309 The maximum number of instructions to consider when looking for an
5310 instruction to fill a delay slot. If more than this arbitrary number of
5311 instructions is searched, the time savings from filling the delay slot
5312 will be minimal so stop searching. Increasing values mean more
5313 aggressive optimization, making the compile time increase with probably
5314 small improvement in executable run time.
5316 @item max-delay-slot-live-search
5317 When trying to fill delay slots, the maximum number of instructions to
5318 consider when searching for a block with valid live register
5319 information. Increasing this arbitrarily chosen value means more
5320 aggressive optimization, increasing the compile time. This parameter
5321 should be removed when the delay slot code is rewritten to maintain the
5324 @item max-gcse-memory
5325 The approximate maximum amount of memory that will be allocated in
5326 order to perform the global common subexpression elimination
5327 optimization. If more memory than specified is required, the
5328 optimization will not be done.
5330 @item max-gcse-passes
5331 The maximum number of passes of GCSE to run. The default is 1.
5333 @item max-pending-list-length
5334 The maximum number of pending dependencies scheduling will allow
5335 before flushing the current state and starting over. Large functions
5336 with few branches or calls can create excessively large lists which
5337 needlessly consume memory and resources.
5339 @item max-inline-insns-single
5340 Several parameters control the tree inliner used in gcc.
5341 This number sets the maximum number of instructions (counted in GCC's
5342 internal representation) in a single function that the tree inliner
5343 will consider for inlining. This only affects functions declared
5344 inline and methods implemented in a class declaration (C++).
5345 The default value is 500.
5347 @item max-inline-insns-auto
5348 When you use @option{-finline-functions} (included in @option{-O3}),
5349 a lot of functions that would otherwise not be considered for inlining
5350 by the compiler will be investigated. To those functions, a different
5351 (more restrictive) limit compared to functions declared inline can
5353 The default value is 120.
5355 @item large-function-insns
5356 The limit specifying really large functions. For functions greater than this
5357 limit inlining is constrained by @option{--param large-function-growth}.
5358 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
5359 algorithms used by the backend.
5360 This parameter is ignored when @option{-funit-at-a-time} is not used.
5361 The default value is 3000.
5363 @item large-function-growth
5364 Specifies maximal growth of large function caused by inlining in percents.
5365 This parameter is ignored when @option{-funit-at-a-time} is not used.
5366 The default value is 200.
5368 @item inline-unit-growth
5369 Specifies maximal overall growth of the compilation unit caused by inlining.
5370 This parameter is ignored when @option{-funit-at-a-time} is not used.
5371 The default value is 150.
5373 @item max-inline-insns-recursive
5374 @itemx max-inline-insns-recursive-auto
5375 Specifies maximum number of instructions out-of-line copy of self recursive inline
5376 function can grow into by performing recursive inlining.
5378 For functions declared inline @option{--param max-inline-insns-recursive} is
5379 taken into acount. For function not declared inline, recursive inlining
5380 happens only when @option{-finline-functions} (included in @option{-O3}) is
5381 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5382 default value is 500.
5384 @item max-inline-recursive-depth
5385 @itemx max-inline-recursive-depth-auto
5386 Specifies maximum recursion depth used by the recursive inlining.
5388 For functions declared inline @option{--param max-inline-recursive-depth} is
5389 taken into acount. For function not declared inline, recursive inlining
5390 happens only when @option{-finline-functions} (included in @option{-O3}) is
5391 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5392 default value is 500.
5394 @item max-inline-insns-rtl
5395 For languages that use the RTL inliner (this happens at a later stage
5396 than tree inlining), you can set the maximum allowable size (counted
5397 in RTL instructions) for the RTL inliner with this parameter.
5398 The default value is 600.
5400 @item max-unrolled-insns
5401 The maximum number of instructions that a loop should have if that loop
5402 is unrolled, and if the loop is unrolled, it determines how many times
5403 the loop code is unrolled.
5405 @item max-average-unrolled-insns
5406 The maximum number of instructions biased by probabilities of their execution
5407 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5408 it determines how many times the loop code is unrolled.
5410 @item max-unroll-times
5411 The maximum number of unrollings of a single loop.
5413 @item max-peeled-insns
5414 The maximum number of instructions that a loop should have if that loop
5415 is peeled, and if the loop is peeled, it determines how many times
5416 the loop code is peeled.
5418 @item max-peel-times
5419 The maximum number of peelings of a single loop.
5421 @item max-completely-peeled-insns
5422 The maximum number of insns of a completely peeled loop.
5424 @item max-completely-peel-times
5425 The maximum number of iterations of a loop to be suitable for complete peeling.
5427 @item max-unswitch-insns
5428 The maximum number of insns of an unswitched loop.
5430 @item max-unswitch-level
5431 The maximum number of branches unswitched in a single loop.
5434 The minimum cost of an expensive expression in the loop invariant motion.
5436 @item iv-consider-all-candidates-bound
5437 Bound on number of candidates for induction variables below that
5438 all candidates are considered for each use in induction variable
5439 optimizations. Only the most relevant candidates are considered
5440 if there are more candidates, to avoid quadratic time complexity.
5442 @item iv-max-considered-uses
5443 The induction variable optimizations give up on loops that contain more
5444 induction variable uses.
5446 @item max-iterations-to-track
5448 The maximum number of iterations of a loop the brute force algorithm
5449 for analysis of # of iterations of the loop tries to evaluate.
5451 @item hot-bb-count-fraction
5452 Select fraction of the maximal count of repetitions of basic block in program
5453 given basic block needs to have to be considered hot.
5455 @item hot-bb-frequency-fraction
5456 Select fraction of the maximal frequency of executions of basic block in
5457 function given basic block needs to have to be considered hot
5459 @item tracer-dynamic-coverage
5460 @itemx tracer-dynamic-coverage-feedback
5462 This value is used to limit superblock formation once the given percentage of
5463 executed instructions is covered. This limits unnecessary code size
5466 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5467 feedback is available. The real profiles (as opposed to statically estimated
5468 ones) are much less balanced allowing the threshold to be larger value.
5470 @item tracer-max-code-growth
5471 Stop tail duplication once code growth has reached given percentage. This is
5472 rather hokey argument, as most of the duplicates will be eliminated later in
5473 cross jumping, so it may be set to much higher values than is the desired code
5476 @item tracer-min-branch-ratio
5478 Stop reverse growth when the reverse probability of best edge is less than this
5479 threshold (in percent).
5481 @item tracer-min-branch-ratio
5482 @itemx tracer-min-branch-ratio-feedback
5484 Stop forward growth if the best edge do have probability lower than this
5487 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5488 compilation for profile feedback and one for compilation without. The value
5489 for compilation with profile feedback needs to be more conservative (higher) in
5490 order to make tracer effective.
5492 @item max-cse-path-length
5494 Maximum number of basic blocks on path that cse considers. The default is 10.
5496 @item global-var-threshold
5498 Counts the number of function calls (N) and the number of
5499 call-clobbered variables (V). If NxV is larger than this limit, a
5500 single artificial variable will be created to represent all the
5501 call-clobbered variables at function call sites. This artificial
5502 variable will then be made to alias every call-clobbered variable.
5503 (done as int * size_t on the host machine; beware overflow).
5505 @item max-aliased-vops
5507 Maxiumum number of virtual operands allowed to represent aliases
5508 before triggering the alias grouping heuristic. Alias grouping
5509 reduces compile times and memory consumption needed for aliasing at
5510 the expense of precision loss in alias information.
5512 @item ggc-min-expand
5514 GCC uses a garbage collector to manage its own memory allocation. This
5515 parameter specifies the minimum percentage by which the garbage
5516 collector's heap should be allowed to expand between collections.
5517 Tuning this may improve compilation speed; it has no effect on code
5520 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
5521 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
5522 the smallest of actual RAM and RLIMIT_DATA or RLIMIT_AS. If
5523 GCC is not able to calculate RAM on a particular platform, the lower
5524 bound of 30% is used. Setting this parameter and
5525 @option{ggc-min-heapsize} to zero causes a full collection to occur at
5526 every opportunity. This is extremely slow, but can be useful for
5529 @item ggc-min-heapsize
5531 Minimum size of the garbage collector's heap before it begins bothering
5532 to collect garbage. The first collection occurs after the heap expands
5533 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
5534 tuning this may improve compilation speed, and has no effect on code
5537 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
5538 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
5539 with a lower bound of 4096 (four megabytes) and an upper bound of
5540 131072 (128 megabytes). If GCC is not able to calculate RAM on a
5541 particular platform, the lower bound is used. Setting this parameter
5542 very large effectively disables garbage collection. Setting this
5543 parameter and @option{ggc-min-expand} to zero causes a full collection
5544 to occur at every opportunity.
5546 @item max-reload-search-insns
5547 The maximum number of instruction reload should look backward for equivalent
5548 register. Increasing values mean more aggressive optimization, making the
5549 compile time increase with probably slightly better performance. The default
5552 @item max-cselib-memory-location
5553 The maximum number of memory locations cselib should take into acount.
5554 Increasing values mean more aggressive optimization, making the compile time
5555 increase with probably slightly better performance. The default value is 500.
5557 @item reorder-blocks-duplicate
5558 @itemx reorder-blocks-duplicate-feedback
5560 Used by basic block reordering pass to decide whether to use unconditional
5561 branch or duplicate the code on its destination. Code is duplicated when its
5562 estimated size is smaller than this value multiplied by the estimated size of
5563 unconditional jump in the hot spots of the program.
5565 The @option{reorder-block-duplicate-feedback} is used only when profile
5566 feedback is available and may be set to higher values than
5567 @option{reorder-block-duplicate} since information about the hot spots is more
5570 @item max-sched-region-blocks
5571 The maximum number of blocks in a region to be considered for
5572 interblock scheduling. The default value is 10.
5574 @item max-sched-region-insns
5575 The maximum number of insns in a region to be considered for
5576 interblock scheduling. The default value is 100.
5578 @item integer-share-limit
5579 Small integer constants can use a shared data structure, reducing the
5580 compiler's memory usage and increasing its speed. This sets the maximum
5581 value of a shared integer constant's. The default value is 256.
5586 @node Preprocessor Options
5587 @section Options Controlling the Preprocessor
5588 @cindex preprocessor options
5589 @cindex options, preprocessor
5591 These options control the C preprocessor, which is run on each C source
5592 file before actual compilation.
5594 If you use the @option{-E} option, nothing is done except preprocessing.
5595 Some of these options make sense only together with @option{-E} because
5596 they cause the preprocessor output to be unsuitable for actual
5601 You can use @option{-Wp,@var{option}} to bypass the compiler driver
5602 and pass @var{option} directly through to the preprocessor. If
5603 @var{option} contains commas, it is split into multiple options at the
5604 commas. However, many options are modified, translated or interpreted
5605 by the compiler driver before being passed to the preprocessor, and
5606 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
5607 interface is undocumented and subject to change, so whenever possible
5608 you should avoid using @option{-Wp} and let the driver handle the
5611 @item -Xpreprocessor @var{option}
5612 @opindex preprocessor
5613 Pass @var{option} as an option to the preprocessor. You can use this to
5614 supply system-specific preprocessor options which GCC does not know how to
5617 If you want to pass an option that takes an argument, you must use
5618 @option{-Xpreprocessor} twice, once for the option and once for the argument.
5621 @include cppopts.texi
5623 @node Assembler Options
5624 @section Passing Options to the Assembler
5626 @c prevent bad page break with this line
5627 You can pass options to the assembler.
5630 @item -Wa,@var{option}
5632 Pass @var{option} as an option to the assembler. If @var{option}
5633 contains commas, it is split into multiple options at the commas.
5635 @item -Xassembler @var{option}
5637 Pass @var{option} as an option to the assembler. You can use this to
5638 supply system-specific assembler options which GCC does not know how to
5641 If you want to pass an option that takes an argument, you must use
5642 @option{-Xassembler} twice, once for the option and once for the argument.
5647 @section Options for Linking
5648 @cindex link options
5649 @cindex options, linking
5651 These options come into play when the compiler links object files into
5652 an executable output file. They are meaningless if the compiler is
5653 not doing a link step.
5657 @item @var{object-file-name}
5658 A file name that does not end in a special recognized suffix is
5659 considered to name an object file or library. (Object files are
5660 distinguished from libraries by the linker according to the file
5661 contents.) If linking is done, these object files are used as input
5670 If any of these options is used, then the linker is not run, and
5671 object file names should not be used as arguments. @xref{Overall
5675 @item -l@var{library}
5676 @itemx -l @var{library}
5678 Search the library named @var{library} when linking. (The second
5679 alternative with the library as a separate argument is only for
5680 POSIX compliance and is not recommended.)
5682 It makes a difference where in the command you write this option; the
5683 linker searches and processes libraries and object files in the order they
5684 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5685 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5686 to functions in @samp{z}, those functions may not be loaded.
5688 The linker searches a standard list of directories for the library,
5689 which is actually a file named @file{lib@var{library}.a}. The linker
5690 then uses this file as if it had been specified precisely by name.
5692 The directories searched include several standard system directories
5693 plus any that you specify with @option{-L}.
5695 Normally the files found this way are library files---archive files
5696 whose members are object files. The linker handles an archive file by
5697 scanning through it for members which define symbols that have so far
5698 been referenced but not defined. But if the file that is found is an
5699 ordinary object file, it is linked in the usual fashion. The only
5700 difference between using an @option{-l} option and specifying a file name
5701 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5702 and searches several directories.
5706 You need this special case of the @option{-l} option in order to
5707 link an Objective-C or Objective-C++ program.
5710 @opindex nostartfiles
5711 Do not use the standard system startup files when linking.
5712 The standard system libraries are used normally, unless @option{-nostdlib}
5713 or @option{-nodefaultlibs} is used.
5715 @item -nodefaultlibs
5716 @opindex nodefaultlibs
5717 Do not use the standard system libraries when linking.
5718 Only the libraries you specify will be passed to the linker.
5719 The standard startup files are used normally, unless @option{-nostartfiles}
5720 is used. The compiler may generate calls to @code{memcmp},
5721 @code{memset}, @code{memcpy} and @code{memmove}.
5722 These entries are usually resolved by entries in
5723 libc. These entry points should be supplied through some other
5724 mechanism when this option is specified.
5728 Do not use the standard system startup files or libraries when linking.
5729 No startup files and only the libraries you specify will be passed to
5730 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
5731 @code{memcpy} and @code{memmove}.
5732 These entries are usually resolved by entries in
5733 libc. These entry points should be supplied through some other
5734 mechanism when this option is specified.
5736 @cindex @option{-lgcc}, use with @option{-nostdlib}
5737 @cindex @option{-nostdlib} and unresolved references
5738 @cindex unresolved references and @option{-nostdlib}
5739 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5740 @cindex @option{-nodefaultlibs} and unresolved references
5741 @cindex unresolved references and @option{-nodefaultlibs}
5742 One of the standard libraries bypassed by @option{-nostdlib} and
5743 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5744 that GCC uses to overcome shortcomings of particular machines, or special
5745 needs for some languages.
5746 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5747 Collection (GCC) Internals},
5748 for more discussion of @file{libgcc.a}.)
5749 In most cases, you need @file{libgcc.a} even when you want to avoid
5750 other standard libraries. In other words, when you specify @option{-nostdlib}
5751 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5752 This ensures that you have no unresolved references to internal GCC
5753 library subroutines. (For example, @samp{__main}, used to ensure C++
5754 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5755 GNU Compiler Collection (GCC) Internals}.)
5759 Produce a position independent executable on targets which support it.
5760 For predictable results, you must also specify the same set of options
5761 that were used to generate code (@option{-fpie}, @option{-fPIE},
5762 or model suboptions) when you specify this option.
5766 Remove all symbol table and relocation information from the executable.
5770 On systems that support dynamic linking, this prevents linking with the shared
5771 libraries. On other systems, this option has no effect.
5775 Produce a shared object which can then be linked with other objects to
5776 form an executable. Not all systems support this option. For predictable
5777 results, you must also specify the same set of options that were used to
5778 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5779 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5780 needs to build supplementary stub code for constructors to work. On
5781 multi-libbed systems, @samp{gcc -shared} must select the correct support
5782 libraries to link against. Failing to supply the correct flags may lead
5783 to subtle defects. Supplying them in cases where they are not necessary
5786 @item -shared-libgcc
5787 @itemx -static-libgcc
5788 @opindex shared-libgcc
5789 @opindex static-libgcc
5790 On systems that provide @file{libgcc} as a shared library, these options
5791 force the use of either the shared or static version respectively.
5792 If no shared version of @file{libgcc} was built when the compiler was
5793 configured, these options have no effect.
5795 There are several situations in which an application should use the
5796 shared @file{libgcc} instead of the static version. The most common
5797 of these is when the application wishes to throw and catch exceptions
5798 across different shared libraries. In that case, each of the libraries
5799 as well as the application itself should use the shared @file{libgcc}.
5801 Therefore, the G++ and GCJ drivers automatically add
5802 @option{-shared-libgcc} whenever you build a shared library or a main
5803 executable, because C++ and Java programs typically use exceptions, so
5804 this is the right thing to do.
5806 If, instead, you use the GCC driver to create shared libraries, you may
5807 find that they will not always be linked with the shared @file{libgcc}.
5808 If GCC finds, at its configuration time, that you have a non-GNU linker
5809 or a GNU linker that does not support option @option{--eh-frame-hdr},
5810 it will link the shared version of @file{libgcc} into shared libraries
5811 by default. Otherwise, it will take advantage of the linker and optimize
5812 away the linking with the shared version of @file{libgcc}, linking with
5813 the static version of libgcc by default. This allows exceptions to
5814 propagate through such shared libraries, without incurring relocation
5815 costs at library load time.
5817 However, if a library or main executable is supposed to throw or catch
5818 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5819 for the languages used in the program, or using the option
5820 @option{-shared-libgcc}, such that it is linked with the shared
5825 Bind references to global symbols when building a shared object. Warn
5826 about any unresolved references (unless overridden by the link editor
5827 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5830 @item -Xlinker @var{option}
5832 Pass @var{option} as an option to the linker. You can use this to
5833 supply system-specific linker options which GCC does not know how to
5836 If you want to pass an option that takes an argument, you must use
5837 @option{-Xlinker} twice, once for the option and once for the argument.
5838 For example, to pass @option{-assert definitions}, you must write
5839 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5840 @option{-Xlinker "-assert definitions"}, because this passes the entire
5841 string as a single argument, which is not what the linker expects.
5843 @item -Wl,@var{option}
5845 Pass @var{option} as an option to the linker. If @var{option} contains
5846 commas, it is split into multiple options at the commas.
5848 @item -u @var{symbol}
5850 Pretend the symbol @var{symbol} is undefined, to force linking of
5851 library modules to define it. You can use @option{-u} multiple times with
5852 different symbols to force loading of additional library modules.
5855 @node Directory Options
5856 @section Options for Directory Search
5857 @cindex directory options
5858 @cindex options, directory search
5861 These options specify directories to search for header files, for
5862 libraries and for parts of the compiler:
5867 Add the directory @var{dir} to the head of the list of directories to be
5868 searched for header files. This can be used to override a system header
5869 file, substituting your own version, since these directories are
5870 searched before the system header file directories. However, you should
5871 not use this option to add directories that contain vendor-supplied
5872 system header files (use @option{-isystem} for that). If you use more than
5873 one @option{-I} option, the directories are scanned in left-to-right
5874 order; the standard system directories come after.
5876 If a standard system include directory, or a directory specified with
5877 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5878 option will be ignored. The directory will still be searched but as a
5879 system directory at its normal position in the system include chain.
5880 This is to ensure that GCC's procedure to fix buggy system headers and
5881 the ordering for the include_next directive are not inadvertently changed.
5882 If you really need to change the search order for system directories,
5883 use the @option{-nostdinc} and/or @option{-isystem} options.
5885 @item -iquote@var{dir}
5887 Add the directory @var{dir} to the head of the list of directories to
5888 be searched for header files only for the case of @samp{#include
5889 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
5890 otherwise just like @option{-I}.
5894 Add directory @var{dir} to the list of directories to be searched
5897 @item -B@var{prefix}
5899 This option specifies where to find the executables, libraries,
5900 include files, and data files of the compiler itself.
5902 The compiler driver program runs one or more of the subprograms
5903 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5904 @var{prefix} as a prefix for each program it tries to run, both with and
5905 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5907 For each subprogram to be run, the compiler driver first tries the
5908 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5909 was not specified, the driver tries two standard prefixes, which are
5910 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5911 those results in a file name that is found, the unmodified program
5912 name is searched for using the directories specified in your
5913 @env{PATH} environment variable.
5915 The compiler will check to see if the path provided by the @option{-B}
5916 refers to a directory, and if necessary it will add a directory
5917 separator character at the end of the path.
5919 @option{-B} prefixes that effectively specify directory names also apply
5920 to libraries in the linker, because the compiler translates these
5921 options into @option{-L} options for the linker. They also apply to
5922 includes files in the preprocessor, because the compiler translates these
5923 options into @option{-isystem} options for the preprocessor. In this case,
5924 the compiler appends @samp{include} to the prefix.
5926 The run-time support file @file{libgcc.a} can also be searched for using
5927 the @option{-B} prefix, if needed. If it is not found there, the two
5928 standard prefixes above are tried, and that is all. The file is left
5929 out of the link if it is not found by those means.
5931 Another way to specify a prefix much like the @option{-B} prefix is to use
5932 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5935 As a special kludge, if the path provided by @option{-B} is
5936 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5937 9, then it will be replaced by @file{[dir/]include}. This is to help
5938 with boot-strapping the compiler.
5940 @item -specs=@var{file}
5942 Process @var{file} after the compiler reads in the standard @file{specs}
5943 file, in order to override the defaults that the @file{gcc} driver
5944 program uses when determining what switches to pass to @file{cc1},
5945 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5946 @option{-specs=@var{file}} can be specified on the command line, and they
5947 are processed in order, from left to right.
5951 This option has been deprecated. Please use @option{-iquote} instead for
5952 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
5953 Any directories you specify with @option{-I} options before the @option{-I-}
5954 option are searched only for the case of @samp{#include "@var{file}"};
5955 they are not searched for @samp{#include <@var{file}>}.
5957 If additional directories are specified with @option{-I} options after
5958 the @option{-I-}, these directories are searched for all @samp{#include}
5959 directives. (Ordinarily @emph{all} @option{-I} directories are used
5962 In addition, the @option{-I-} option inhibits the use of the current
5963 directory (where the current input file came from) as the first search
5964 directory for @samp{#include "@var{file}"}. There is no way to
5965 override this effect of @option{-I-}. With @option{-I.} you can specify
5966 searching the directory which was current when the compiler was
5967 invoked. That is not exactly the same as what the preprocessor does
5968 by default, but it is often satisfactory.
5970 @option{-I-} does not inhibit the use of the standard system directories
5971 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5978 @section Specifying subprocesses and the switches to pass to them
5981 @command{gcc} is a driver program. It performs its job by invoking a
5982 sequence of other programs to do the work of compiling, assembling and
5983 linking. GCC interprets its command-line parameters and uses these to
5984 deduce which programs it should invoke, and which command-line options
5985 it ought to place on their command lines. This behavior is controlled
5986 by @dfn{spec strings}. In most cases there is one spec string for each
5987 program that GCC can invoke, but a few programs have multiple spec
5988 strings to control their behavior. The spec strings built into GCC can
5989 be overridden by using the @option{-specs=} command-line switch to specify
5992 @dfn{Spec files} are plaintext files that are used to construct spec
5993 strings. They consist of a sequence of directives separated by blank
5994 lines. The type of directive is determined by the first non-whitespace
5995 character on the line and it can be one of the following:
5998 @item %@var{command}
5999 Issues a @var{command} to the spec file processor. The commands that can
6003 @item %include <@var{file}>
6005 Search for @var{file} and insert its text at the current point in the
6008 @item %include_noerr <@var{file}>
6009 @cindex %include_noerr
6010 Just like @samp{%include}, but do not generate an error message if the include
6011 file cannot be found.
6013 @item %rename @var{old_name} @var{new_name}
6015 Rename the spec string @var{old_name} to @var{new_name}.
6019 @item *[@var{spec_name}]:
6020 This tells the compiler to create, override or delete the named spec
6021 string. All lines after this directive up to the next directive or
6022 blank line are considered to be the text for the spec string. If this
6023 results in an empty string then the spec will be deleted. (Or, if the
6024 spec did not exist, then nothing will happened.) Otherwise, if the spec
6025 does not currently exist a new spec will be created. If the spec does
6026 exist then its contents will be overridden by the text of this
6027 directive, unless the first character of that text is the @samp{+}
6028 character, in which case the text will be appended to the spec.
6030 @item [@var{suffix}]:
6031 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6032 and up to the next directive or blank line are considered to make up the
6033 spec string for the indicated suffix. When the compiler encounters an
6034 input file with the named suffix, it will processes the spec string in
6035 order to work out how to compile that file. For example:
6042 This says that any input file whose name ends in @samp{.ZZ} should be
6043 passed to the program @samp{z-compile}, which should be invoked with the
6044 command-line switch @option{-input} and with the result of performing the
6045 @samp{%i} substitution. (See below.)
6047 As an alternative to providing a spec string, the text that follows a
6048 suffix directive can be one of the following:
6051 @item @@@var{language}
6052 This says that the suffix is an alias for a known @var{language}. This is
6053 similar to using the @option{-x} command-line switch to GCC to specify a
6054 language explicitly. For example:
6061 Says that .ZZ files are, in fact, C++ source files.
6064 This causes an error messages saying:
6067 @var{name} compiler not installed on this system.
6071 GCC already has an extensive list of suffixes built into it.
6072 This directive will add an entry to the end of the list of suffixes, but
6073 since the list is searched from the end backwards, it is effectively
6074 possible to override earlier entries using this technique.
6078 GCC has the following spec strings built into it. Spec files can
6079 override these strings or create their own. Note that individual
6080 targets can also add their own spec strings to this list.
6083 asm Options to pass to the assembler
6084 asm_final Options to pass to the assembler post-processor
6085 cpp Options to pass to the C preprocessor
6086 cc1 Options to pass to the C compiler
6087 cc1plus Options to pass to the C++ compiler
6088 endfile Object files to include at the end of the link
6089 link Options to pass to the linker
6090 lib Libraries to include on the command line to the linker
6091 libgcc Decides which GCC support library to pass to the linker
6092 linker Sets the name of the linker
6093 predefines Defines to be passed to the C preprocessor
6094 signed_char Defines to pass to CPP to say whether @code{char} is signed
6096 startfile Object files to include at the start of the link
6099 Here is a small example of a spec file:
6105 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6108 This example renames the spec called @samp{lib} to @samp{old_lib} and
6109 then overrides the previous definition of @samp{lib} with a new one.
6110 The new definition adds in some extra command-line options before
6111 including the text of the old definition.
6113 @dfn{Spec strings} are a list of command-line options to be passed to their
6114 corresponding program. In addition, the spec strings can contain
6115 @samp{%}-prefixed sequences to substitute variable text or to
6116 conditionally insert text into the command line. Using these constructs
6117 it is possible to generate quite complex command lines.
6119 Here is a table of all defined @samp{%}-sequences for spec
6120 strings. Note that spaces are not generated automatically around the
6121 results of expanding these sequences. Therefore you can concatenate them
6122 together or combine them with constant text in a single argument.
6126 Substitute one @samp{%} into the program name or argument.
6129 Substitute the name of the input file being processed.
6132 Substitute the basename of the input file being processed.
6133 This is the substring up to (and not including) the last period
6134 and not including the directory.
6137 This is the same as @samp{%b}, but include the file suffix (text after
6141 Marks the argument containing or following the @samp{%d} as a
6142 temporary file name, so that that file will be deleted if GCC exits
6143 successfully. Unlike @samp{%g}, this contributes no text to the
6146 @item %g@var{suffix}
6147 Substitute a file name that has suffix @var{suffix} and is chosen
6148 once per compilation, and mark the argument in the same way as
6149 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6150 name is now chosen in a way that is hard to predict even when previously
6151 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6152 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6153 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6154 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6155 was simply substituted with a file name chosen once per compilation,
6156 without regard to any appended suffix (which was therefore treated
6157 just like ordinary text), making such attacks more likely to succeed.
6159 @item %u@var{suffix}
6160 Like @samp{%g}, but generates a new temporary file name even if
6161 @samp{%u@var{suffix}} was already seen.
6163 @item %U@var{suffix}
6164 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6165 new one if there is no such last file name. In the absence of any
6166 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6167 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6168 would involve the generation of two distinct file names, one
6169 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6170 simply substituted with a file name chosen for the previous @samp{%u},
6171 without regard to any appended suffix.
6173 @item %j@var{suffix}
6174 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6175 writable, and if save-temps is off; otherwise, substitute the name
6176 of a temporary file, just like @samp{%u}. This temporary file is not
6177 meant for communication between processes, but rather as a junk
6180 @item %|@var{suffix}
6181 @itemx %m@var{suffix}
6182 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6183 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6184 all. These are the two most common ways to instruct a program that it
6185 should read from standard input or write to standard output. If you
6186 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6187 construct: see for example @file{f/lang-specs.h}.
6189 @item %.@var{SUFFIX}
6190 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6191 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6192 terminated by the next space or %.
6195 Marks the argument containing or following the @samp{%w} as the
6196 designated output file of this compilation. This puts the argument
6197 into the sequence of arguments that @samp{%o} will substitute later.
6200 Substitutes the names of all the output files, with spaces
6201 automatically placed around them. You should write spaces
6202 around the @samp{%o} as well or the results are undefined.
6203 @samp{%o} is for use in the specs for running the linker.
6204 Input files whose names have no recognized suffix are not compiled
6205 at all, but they are included among the output files, so they will
6209 Substitutes the suffix for object files. Note that this is
6210 handled specially when it immediately follows @samp{%g, %u, or %U},
6211 because of the need for those to form complete file names. The
6212 handling is such that @samp{%O} is treated exactly as if it had already
6213 been substituted, except that @samp{%g, %u, and %U} do not currently
6214 support additional @var{suffix} characters following @samp{%O} as they would
6215 following, for example, @samp{.o}.
6218 Substitutes the standard macro predefinitions for the
6219 current target machine. Use this when running @code{cpp}.
6222 Like @samp{%p}, but puts @samp{__} before and after the name of each
6223 predefined macro, except for macros that start with @samp{__} or with
6224 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6228 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6229 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
6230 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6234 Current argument is the name of a library or startup file of some sort.
6235 Search for that file in a standard list of directories and substitute
6236 the full name found.
6239 Print @var{str} as an error message. @var{str} is terminated by a newline.
6240 Use this when inconsistent options are detected.
6243 Substitute the contents of spec string @var{name} at this point.
6246 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6248 @item %x@{@var{option}@}
6249 Accumulate an option for @samp{%X}.
6252 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6256 Output the accumulated assembler options specified by @option{-Wa}.
6259 Output the accumulated preprocessor options specified by @option{-Wp}.
6262 Process the @code{asm} spec. This is used to compute the
6263 switches to be passed to the assembler.
6266 Process the @code{asm_final} spec. This is a spec string for
6267 passing switches to an assembler post-processor, if such a program is
6271 Process the @code{link} spec. This is the spec for computing the
6272 command line passed to the linker. Typically it will make use of the
6273 @samp{%L %G %S %D and %E} sequences.
6276 Dump out a @option{-L} option for each directory that GCC believes might
6277 contain startup files. If the target supports multilibs then the
6278 current multilib directory will be prepended to each of these paths.
6281 Output the multilib directory with directory separators replaced with
6282 @samp{_}. If multilib directories are not set, or the multilib directory is
6283 @file{.} then this option emits nothing.
6286 Process the @code{lib} spec. This is a spec string for deciding which
6287 libraries should be included on the command line to the linker.
6290 Process the @code{libgcc} spec. This is a spec string for deciding
6291 which GCC support library should be included on the command line to the linker.
6294 Process the @code{startfile} spec. This is a spec for deciding which
6295 object files should be the first ones passed to the linker. Typically
6296 this might be a file named @file{crt0.o}.
6299 Process the @code{endfile} spec. This is a spec string that specifies
6300 the last object files that will be passed to the linker.
6303 Process the @code{cpp} spec. This is used to construct the arguments
6304 to be passed to the C preprocessor.
6307 Process the @code{cc1} spec. This is used to construct the options to be
6308 passed to the actual C compiler (@samp{cc1}).
6311 Process the @code{cc1plus} spec. This is used to construct the options to be
6312 passed to the actual C++ compiler (@samp{cc1plus}).
6315 Substitute the variable part of a matched option. See below.
6316 Note that each comma in the substituted string is replaced by
6320 Remove all occurrences of @code{-S} from the command line. Note---this
6321 command is position dependent. @samp{%} commands in the spec string
6322 before this one will see @code{-S}, @samp{%} commands in the spec string
6323 after this one will not.
6325 @item %:@var{function}(@var{args})
6326 Call the named function @var{function}, passing it @var{args}.
6327 @var{args} is first processed as a nested spec string, then split
6328 into an argument vector in the usual fashion. The function returns
6329 a string which is processed as if it had appeared literally as part
6330 of the current spec.
6332 The following built-in spec functions are provided:
6335 @item @code{if-exists}
6336 The @code{if-exists} spec function takes one argument, an absolute
6337 pathname to a file. If the file exists, @code{if-exists} returns the
6338 pathname. Here is a small example of its usage:
6342 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6345 @item @code{if-exists-else}
6346 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6347 spec function, except that it takes two arguments. The first argument is
6348 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6349 returns the pathname. If it does not exist, it returns the second argument.
6350 This way, @code{if-exists-else} can be used to select one file or another,
6351 based on the existence of the first. Here is a small example of its usage:
6355 crt0%O%s %:if-exists(crti%O%s) \
6356 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6361 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6362 If that switch was not specified, this substitutes nothing. Note that
6363 the leading dash is omitted when specifying this option, and it is
6364 automatically inserted if the substitution is performed. Thus the spec
6365 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6366 and would output the command line option @option{-foo}.
6368 @item %W@{@code{S}@}
6369 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6372 @item %@{@code{S}*@}
6373 Substitutes all the switches specified to GCC whose names start
6374 with @code{-S}, but which also take an argument. This is used for
6375 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6376 GCC considers @option{-o foo} as being
6377 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6378 text, including the space. Thus two arguments would be generated.
6380 @item %@{@code{S}*&@code{T}*@}
6381 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6382 (the order of @code{S} and @code{T} in the spec is not significant).
6383 There can be any number of ampersand-separated variables; for each the
6384 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6386 @item %@{@code{S}:@code{X}@}
6387 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6389 @item %@{!@code{S}:@code{X}@}
6390 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6392 @item %@{@code{S}*:@code{X}@}
6393 Substitutes @code{X} if one or more switches whose names start with
6394 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6395 once, no matter how many such switches appeared. However, if @code{%*}
6396 appears somewhere in @code{X}, then @code{X} will be substituted once
6397 for each matching switch, with the @code{%*} replaced by the part of
6398 that switch that matched the @code{*}.
6400 @item %@{.@code{S}:@code{X}@}
6401 Substitutes @code{X}, if processing a file with suffix @code{S}.
6403 @item %@{!.@code{S}:@code{X}@}
6404 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6406 @item %@{@code{S}|@code{P}:@code{X}@}
6407 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6408 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6409 although they have a stronger binding than the @samp{|}. If @code{%*}
6410 appears in @code{X}, all of the alternatives must be starred, and only
6411 the first matching alternative is substituted.
6413 For example, a spec string like this:
6416 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6419 will output the following command-line options from the following input
6420 command-line options:
6425 -d fred.c -foo -baz -boggle
6426 -d jim.d -bar -baz -boggle
6429 @item %@{S:X; T:Y; :D@}
6431 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6432 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6433 be as many clauses as you need. This may be combined with @code{.},
6434 @code{!}, @code{|}, and @code{*} as needed.
6439 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6440 construct may contain other nested @samp{%} constructs or spaces, or
6441 even newlines. They are processed as usual, as described above.
6442 Trailing white space in @code{X} is ignored. White space may also
6443 appear anywhere on the left side of the colon in these constructs,
6444 except between @code{.} or @code{*} and the corresponding word.
6446 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6447 handled specifically in these constructs. If another value of
6448 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
6449 @option{-W} switch is found later in the command line, the earlier
6450 switch value is ignored, except with @{@code{S}*@} where @code{S} is
6451 just one letter, which passes all matching options.
6453 The character @samp{|} at the beginning of the predicate text is used to
6454 indicate that a command should be piped to the following command, but
6455 only if @option{-pipe} is specified.
6457 It is built into GCC which switches take arguments and which do not.
6458 (You might think it would be useful to generalize this to allow each
6459 compiler's spec to say which switches take arguments. But this cannot
6460 be done in a consistent fashion. GCC cannot even decide which input
6461 files have been specified without knowing which switches take arguments,
6462 and it must know which input files to compile in order to tell which
6465 GCC also knows implicitly that arguments starting in @option{-l} are to be
6466 treated as compiler output files, and passed to the linker in their
6467 proper position among the other output files.
6469 @c man begin OPTIONS
6471 @node Target Options
6472 @section Specifying Target Machine and Compiler Version
6473 @cindex target options
6474 @cindex cross compiling
6475 @cindex specifying machine version
6476 @cindex specifying compiler version and target machine
6477 @cindex compiler version, specifying
6478 @cindex target machine, specifying
6480 The usual way to run GCC is to run the executable called @file{gcc}, or
6481 @file{<machine>-gcc} when cross-compiling, or
6482 @file{<machine>-gcc-<version>} to run a version other than the one that
6483 was installed last. Sometimes this is inconvenient, so GCC provides
6484 options that will switch to another cross-compiler or version.
6487 @item -b @var{machine}
6489 The argument @var{machine} specifies the target machine for compilation.
6491 The value to use for @var{machine} is the same as was specified as the
6492 machine type when configuring GCC as a cross-compiler. For
6493 example, if a cross-compiler was configured with @samp{configure
6494 i386v}, meaning to compile for an 80386 running System V, then you
6495 would specify @option{-b i386v} to run that cross compiler.
6497 @item -V @var{version}
6499 The argument @var{version} specifies which version of GCC to run.
6500 This is useful when multiple versions are installed. For example,
6501 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
6504 The @option{-V} and @option{-b} options work by running the
6505 @file{<machine>-gcc-<version>} executable, so there's no real reason to
6506 use them if you can just run that directly.
6508 @node Submodel Options
6509 @section Hardware Models and Configurations
6510 @cindex submodel options
6511 @cindex specifying hardware config
6512 @cindex hardware models and configurations, specifying
6513 @cindex machine dependent options
6515 Earlier we discussed the standard option @option{-b} which chooses among
6516 different installed compilers for completely different target
6517 machines, such as VAX vs.@: 68000 vs.@: 80386.
6519 In addition, each of these target machine types can have its own
6520 special options, starting with @samp{-m}, to choose among various
6521 hardware models or configurations---for example, 68010 vs 68020,
6522 floating coprocessor or none. A single installed version of the
6523 compiler can compile for any model or configuration, according to the
6526 Some configurations of the compiler also support additional special
6527 options, usually for compatibility with other compilers on the same
6530 These options are defined by the macro @code{TARGET_SWITCHES} in the
6531 machine description. The default for the options is also defined by
6532 that macro, which enables you to change the defaults.
6534 @c This list is ordered alphanumerically by subsection name.
6535 @c It should be the same order and spelling as these options are listed
6536 @c in Machine Dependent Options
6544 * DEC Alpha Options::
6545 * DEC Alpha/VMS Options::
6549 * i386 and x86-64 Options::
6561 * RS/6000 and PowerPC Options::
6562 * S/390 and zSeries Options::
6565 * System V Options::
6566 * TMS320C3x/C4x Options::
6570 * Xstormy16 Options::
6576 @subsection ARC Options
6579 These options are defined for ARC implementations:
6584 Compile code for little endian mode. This is the default.
6588 Compile code for big endian mode.
6591 @opindex mmangle-cpu
6592 Prepend the name of the cpu to all public symbol names.
6593 In multiple-processor systems, there are many ARC variants with different
6594 instruction and register set characteristics. This flag prevents code
6595 compiled for one cpu to be linked with code compiled for another.
6596 No facility exists for handling variants that are ``almost identical''.
6597 This is an all or nothing option.
6599 @item -mcpu=@var{cpu}
6601 Compile code for ARC variant @var{cpu}.
6602 Which variants are supported depend on the configuration.
6603 All variants support @option{-mcpu=base}, this is the default.
6605 @item -mtext=@var{text-section}
6606 @itemx -mdata=@var{data-section}
6607 @itemx -mrodata=@var{readonly-data-section}
6611 Put functions, data, and readonly data in @var{text-section},
6612 @var{data-section}, and @var{readonly-data-section} respectively
6613 by default. This can be overridden with the @code{section} attribute.
6614 @xref{Variable Attributes}.
6619 @subsection ARM Options
6622 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6626 @item -mabi=@var{name}
6628 Generate code for the specified ABI. Permissible values are: @samp{apcs-gnu},
6629 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6632 @opindex mapcs-frame
6633 Generate a stack frame that is compliant with the ARM Procedure Call
6634 Standard for all functions, even if this is not strictly necessary for
6635 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6636 with this option will cause the stack frames not to be generated for
6637 leaf functions. The default is @option{-mno-apcs-frame}.
6641 This is a synonym for @option{-mapcs-frame}.
6644 @c not currently implemented
6645 @item -mapcs-stack-check
6646 @opindex mapcs-stack-check
6647 Generate code to check the amount of stack space available upon entry to
6648 every function (that actually uses some stack space). If there is
6649 insufficient space available then either the function
6650 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6651 called, depending upon the amount of stack space required. The run time
6652 system is required to provide these functions. The default is
6653 @option{-mno-apcs-stack-check}, since this produces smaller code.
6655 @c not currently implemented
6657 @opindex mapcs-float
6658 Pass floating point arguments using the float point registers. This is
6659 one of the variants of the APCS@. This option is recommended if the
6660 target hardware has a floating point unit or if a lot of floating point
6661 arithmetic is going to be performed by the code. The default is
6662 @option{-mno-apcs-float}, since integer only code is slightly increased in
6663 size if @option{-mapcs-float} is used.
6665 @c not currently implemented
6666 @item -mapcs-reentrant
6667 @opindex mapcs-reentrant
6668 Generate reentrant, position independent code. The default is
6669 @option{-mno-apcs-reentrant}.
6672 @item -mthumb-interwork
6673 @opindex mthumb-interwork
6674 Generate code which supports calling between the ARM and Thumb
6675 instruction sets. Without this option the two instruction sets cannot
6676 be reliably used inside one program. The default is
6677 @option{-mno-thumb-interwork}, since slightly larger code is generated
6678 when @option{-mthumb-interwork} is specified.
6680 @item -mno-sched-prolog
6681 @opindex mno-sched-prolog
6682 Prevent the reordering of instructions in the function prolog, or the
6683 merging of those instruction with the instructions in the function's
6684 body. This means that all functions will start with a recognizable set
6685 of instructions (or in fact one of a choice from a small set of
6686 different function prologues), and this information can be used to
6687 locate the start if functions inside an executable piece of code. The
6688 default is @option{-msched-prolog}.
6691 @opindex mhard-float
6692 Generate output containing floating point instructions. This is the
6696 @opindex msoft-float
6697 Generate output containing library calls for floating point.
6698 @strong{Warning:} the requisite libraries are not available for all ARM
6699 targets. Normally the facilities of the machine's usual C compiler are
6700 used, but this cannot be done directly in cross-compilation. You must make
6701 your own arrangements to provide suitable library functions for
6704 @option{-msoft-float} changes the calling convention in the output file;
6705 therefore, it is only useful if you compile @emph{all} of a program with
6706 this option. In particular, you need to compile @file{libgcc.a}, the
6707 library that comes with GCC, with @option{-msoft-float} in order for
6710 @item -mfloat-abi=@var{name}
6712 Specifies which ABI to use for floating point values. Permissible values
6713 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6715 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6716 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6717 of floating point instructions, but still uses the soft-float calling
6720 @item -mlittle-endian
6721 @opindex mlittle-endian
6722 Generate code for a processor running in little-endian mode. This is
6723 the default for all standard configurations.
6726 @opindex mbig-endian
6727 Generate code for a processor running in big-endian mode; the default is
6728 to compile code for a little-endian processor.
6730 @item -mwords-little-endian
6731 @opindex mwords-little-endian
6732 This option only applies when generating code for big-endian processors.
6733 Generate code for a little-endian word order but a big-endian byte
6734 order. That is, a byte order of the form @samp{32107654}. Note: this
6735 option should only be used if you require compatibility with code for
6736 big-endian ARM processors generated by versions of the compiler prior to
6739 @item -mcpu=@var{name}
6741 This specifies the name of the target ARM processor. GCC uses this name
6742 to determine what kind of instructions it can emit when generating
6743 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6744 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6745 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6746 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6747 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6748 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
6749 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6750 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6751 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
6752 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
6753 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
6754 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
6755 @samp{arm1136j-s}, @samp{arm1136jf-s} ,@samp{xscale}, @samp{iwmmxt},
6758 @itemx -mtune=@var{name}
6760 This option is very similar to the @option{-mcpu=} option, except that
6761 instead of specifying the actual target processor type, and hence
6762 restricting which instructions can be used, it specifies that GCC should
6763 tune the performance of the code as if the target were of the type
6764 specified in this option, but still choosing the instructions that it
6765 will generate based on the cpu specified by a @option{-mcpu=} option.
6766 For some ARM implementations better performance can be obtained by using
6769 @item -march=@var{name}
6771 This specifies the name of the target ARM architecture. GCC uses this
6772 name to determine what kind of instructions it can emit when generating
6773 assembly code. This option can be used in conjunction with or instead
6774 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6775 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6776 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6777 @samp{iwmmxt}, @samp{ep9312}.
6779 @item -mfpu=@var{name}
6780 @itemx -mfpe=@var{number}
6781 @itemx -mfp=@var{number}
6785 This specifies what floating point hardware (or hardware emulation) is
6786 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6787 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6788 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6789 with older versions of GCC@.
6791 If @option{-msoft-float} is specified this specifies the format of
6792 floating point values.
6794 @item -mstructure-size-boundary=@var{n}
6795 @opindex mstructure-size-boundary
6796 The size of all structures and unions will be rounded up to a multiple
6797 of the number of bits set by this option. Permissible values are 8, 32
6798 and 64. The default value varies for different toolchains. For the COFF
6799 targeted toolchain the default value is 8. A value of 64 is only allowed
6800 if the underlying ABI supports it.
6802 Specifying the larger number can produce faster, more efficient code, but
6803 can also increase the size of the program. Different values are potentially
6804 incompatible. Code compiled with one value cannot necessarily expect to
6805 work with code or libraries compiled with another value, if they exchange
6806 information using structures or unions.
6808 @item -mabort-on-noreturn
6809 @opindex mabort-on-noreturn
6810 Generate a call to the function @code{abort} at the end of a
6811 @code{noreturn} function. It will be executed if the function tries to
6815 @itemx -mno-long-calls
6816 @opindex mlong-calls
6817 @opindex mno-long-calls
6818 Tells the compiler to perform function calls by first loading the
6819 address of the function into a register and then performing a subroutine
6820 call on this register. This switch is needed if the target function
6821 will lie outside of the 64 megabyte addressing range of the offset based
6822 version of subroutine call instruction.
6824 Even if this switch is enabled, not all function calls will be turned
6825 into long calls. The heuristic is that static functions, functions
6826 which have the @samp{short-call} attribute, functions that are inside
6827 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6828 definitions have already been compiled within the current compilation
6829 unit, will not be turned into long calls. The exception to this rule is
6830 that weak function definitions, functions with the @samp{long-call}
6831 attribute or the @samp{section} attribute, and functions that are within
6832 the scope of a @samp{#pragma long_calls} directive, will always be
6833 turned into long calls.
6835 This feature is not enabled by default. Specifying
6836 @option{-mno-long-calls} will restore the default behavior, as will
6837 placing the function calls within the scope of a @samp{#pragma
6838 long_calls_off} directive. Note these switches have no effect on how
6839 the compiler generates code to handle function calls via function
6842 @item -mnop-fun-dllimport
6843 @opindex mnop-fun-dllimport
6844 Disable support for the @code{dllimport} attribute.
6846 @item -msingle-pic-base
6847 @opindex msingle-pic-base
6848 Treat the register used for PIC addressing as read-only, rather than
6849 loading it in the prologue for each function. The run-time system is
6850 responsible for initializing this register with an appropriate value
6851 before execution begins.
6853 @item -mpic-register=@var{reg}
6854 @opindex mpic-register
6855 Specify the register to be used for PIC addressing. The default is R10
6856 unless stack-checking is enabled, when R9 is used.
6858 @item -mcirrus-fix-invalid-insns
6859 @opindex mcirrus-fix-invalid-insns
6860 @opindex mno-cirrus-fix-invalid-insns
6861 Insert NOPs into the instruction stream to in order to work around
6862 problems with invalid Maverick instruction combinations. This option
6863 is only valid if the @option{-mcpu=ep9312} option has been used to
6864 enable generation of instructions for the Cirrus Maverick floating
6865 point co-processor. This option is not enabled by default, since the
6866 problem is only present in older Maverick implementations. The default
6867 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6870 @item -mpoke-function-name
6871 @opindex mpoke-function-name
6872 Write the name of each function into the text section, directly
6873 preceding the function prologue. The generated code is similar to this:
6877 .ascii "arm_poke_function_name", 0
6880 .word 0xff000000 + (t1 - t0)
6881 arm_poke_function_name
6883 stmfd sp!, @{fp, ip, lr, pc@}
6887 When performing a stack backtrace, code can inspect the value of
6888 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6889 location @code{pc - 12} and the top 8 bits are set, then we know that
6890 there is a function name embedded immediately preceding this location
6891 and has length @code{((pc[-3]) & 0xff000000)}.
6895 Generate code for the 16-bit Thumb instruction set. The default is to
6896 use the 32-bit ARM instruction set.
6899 @opindex mtpcs-frame
6900 Generate a stack frame that is compliant with the Thumb Procedure Call
6901 Standard for all non-leaf functions. (A leaf function is one that does
6902 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6904 @item -mtpcs-leaf-frame
6905 @opindex mtpcs-leaf-frame
6906 Generate a stack frame that is compliant with the Thumb Procedure Call
6907 Standard for all leaf functions. (A leaf function is one that does
6908 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6910 @item -mcallee-super-interworking
6911 @opindex mcallee-super-interworking
6912 Gives all externally visible functions in the file being compiled an ARM
6913 instruction set header which switches to Thumb mode before executing the
6914 rest of the function. This allows these functions to be called from
6915 non-interworking code.
6917 @item -mcaller-super-interworking
6918 @opindex mcaller-super-interworking
6919 Allows calls via function pointers (including virtual functions) to
6920 execute correctly regardless of whether the target code has been
6921 compiled for interworking or not. There is a small overhead in the cost
6922 of executing a function pointer if this option is enabled.
6927 @subsection AVR Options
6930 These options are defined for AVR implementations:
6933 @item -mmcu=@var{mcu}
6935 Specify ATMEL AVR instruction set or MCU type.
6937 Instruction set avr1 is for the minimal AVR core, not supported by the C
6938 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
6939 attiny11, attiny12, attiny15, attiny28).
6941 Instruction set avr2 (default) is for the classic AVR core with up to
6942 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
6943 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
6944 at90c8534, at90s8535).
6946 Instruction set avr3 is for the classic AVR core with up to 128K program
6947 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
6949 Instruction set avr4 is for the enhanced AVR core with up to 8K program
6950 memory space (MCU types: atmega8, atmega83, atmega85).
6952 Instruction set avr5 is for the enhanced AVR core with up to 128K program
6953 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
6954 atmega64, atmega128, at43usb355, at94k).
6958 Output instruction sizes to the asm file.
6960 @item -minit-stack=@var{N}
6961 @opindex minit-stack
6962 Specify the initial stack address, which may be a symbol or numeric value,
6963 @samp{__stack} is the default.
6965 @item -mno-interrupts
6966 @opindex mno-interrupts
6967 Generated code is not compatible with hardware interrupts.
6968 Code size will be smaller.
6970 @item -mcall-prologues
6971 @opindex mcall-prologues
6972 Functions prologues/epilogues expanded as call to appropriate
6973 subroutines. Code size will be smaller.
6975 @item -mno-tablejump
6976 @opindex mno-tablejump
6977 Do not generate tablejump insns which sometimes increase code size.
6980 @opindex mtiny-stack
6981 Change only the low 8 bits of the stack pointer.
6985 Assume int to be 8 bit integer. This affects the sizes of all types: A
6986 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
6987 and long long will be 4 bytes. Please note that this option does not
6988 comply to the C standards, but it will provide you with smaller code
6993 @subsection CRIS Options
6994 @cindex CRIS Options
6996 These options are defined specifically for the CRIS ports.
6999 @item -march=@var{architecture-type}
7000 @itemx -mcpu=@var{architecture-type}
7003 Generate code for the specified architecture. The choices for
7004 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7005 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
7006 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7009 @item -mtune=@var{architecture-type}
7011 Tune to @var{architecture-type} everything applicable about the generated
7012 code, except for the ABI and the set of available instructions. The
7013 choices for @var{architecture-type} are the same as for
7014 @option{-march=@var{architecture-type}}.
7016 @item -mmax-stack-frame=@var{n}
7017 @opindex mmax-stack-frame
7018 Warn when the stack frame of a function exceeds @var{n} bytes.
7020 @item -melinux-stacksize=@var{n}
7021 @opindex melinux-stacksize
7022 Only available with the @samp{cris-axis-aout} target. Arranges for
7023 indications in the program to the kernel loader that the stack of the
7024 program should be set to @var{n} bytes.
7030 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7031 @option{-march=v3} and @option{-march=v8} respectively.
7033 @item -mmul-bug-workaround
7034 @itemx -mno-mul-bug-workaround
7035 @opindex mmul-bug-workaround
7036 @opindex mno-mul-bug-workaround
7037 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7038 models where it applies. This option is active by default.
7042 Enable CRIS-specific verbose debug-related information in the assembly
7043 code. This option also has the effect to turn off the @samp{#NO_APP}
7044 formatted-code indicator to the assembler at the beginning of the
7049 Do not use condition-code results from previous instruction; always emit
7050 compare and test instructions before use of condition codes.
7052 @item -mno-side-effects
7053 @opindex mno-side-effects
7054 Do not emit instructions with side-effects in addressing modes other than
7058 @itemx -mno-stack-align
7060 @itemx -mno-data-align
7061 @itemx -mconst-align
7062 @itemx -mno-const-align
7063 @opindex mstack-align
7064 @opindex mno-stack-align
7065 @opindex mdata-align
7066 @opindex mno-data-align
7067 @opindex mconst-align
7068 @opindex mno-const-align
7069 These options (no-options) arranges (eliminate arrangements) for the
7070 stack-frame, individual data and constants to be aligned for the maximum
7071 single data access size for the chosen CPU model. The default is to
7072 arrange for 32-bit alignment. ABI details such as structure layout are
7073 not affected by these options.
7081 Similar to the stack- data- and const-align options above, these options
7082 arrange for stack-frame, writable data and constants to all be 32-bit,
7083 16-bit or 8-bit aligned. The default is 32-bit alignment.
7085 @item -mno-prologue-epilogue
7086 @itemx -mprologue-epilogue
7087 @opindex mno-prologue-epilogue
7088 @opindex mprologue-epilogue
7089 With @option{-mno-prologue-epilogue}, the normal function prologue and
7090 epilogue that sets up the stack-frame are omitted and no return
7091 instructions or return sequences are generated in the code. Use this
7092 option only together with visual inspection of the compiled code: no
7093 warnings or errors are generated when call-saved registers must be saved,
7094 or storage for local variable needs to be allocated.
7100 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7101 instruction sequences that load addresses for functions from the PLT part
7102 of the GOT rather than (traditional on other architectures) calls to the
7103 PLT. The default is @option{-mgotplt}.
7107 Legacy no-op option only recognized with the cris-axis-aout target.
7111 Legacy no-op option only recognized with the cris-axis-elf and
7112 cris-axis-linux-gnu targets.
7116 Only recognized with the cris-axis-aout target, where it selects a
7117 GNU/linux-like multilib, include files and instruction set for
7122 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7126 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7127 to link with input-output functions from a simulator library. Code,
7128 initialized data and zero-initialized data are allocated consecutively.
7132 Like @option{-sim}, but pass linker options to locate initialized data at
7133 0x40000000 and zero-initialized data at 0x80000000.
7136 @node Darwin Options
7137 @subsection Darwin Options
7138 @cindex Darwin options
7140 These options are defined for all architectures running the Darwin operating
7141 system. They are useful for compatibility with other Mac OS compilers.
7146 Add the framework directory @var{dir} to the head of the list of
7147 directories to be searched for header files. These directories are
7148 interleaved with those specified by @option{-I} options and are
7149 scanned in a left-to-right order.
7151 A framework directory is a directory with frameworks in it. A
7152 framework is a directory with a @samp{"Headers"} and/or
7153 @samp{"PrivateHeaders"} directory contained directly in it that ends
7154 in @samp{".framework"}. The name of a framework is the name of this
7155 directory excluding the @samp{".framework"}. Headers associated with
7156 the framework are found in one of those two directories, with
7157 @samp{"Headers"} being searched first. A subframework is a framework
7158 directory that is in a framework's @samp{"Frameworks"} directory.
7159 Includes of subframework headers can only appear in a header of a
7160 framework that contains the subframework, or in a sibling subframework
7161 header. Two subframeworks are siblings if they occur in the same
7162 framework. A subframework should not have the same name as a
7163 framework, a warning will be issued if this is violated. Currently a
7164 subframework cannot have subframeworks, in the future, the mechanism
7165 may be extended to support this. The standard frameworks can be found
7166 in @samp{"/System/Library/Frameworks"} and
7167 @samp{"/Library/Frameworks"}. An example include looks like
7168 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7169 the name of the framework and header.h is found in the
7170 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7174 Emit debugging information for symbols that are used. For STABS
7175 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7176 This is by default ON.
7180 Emit debugging information for all symbols and types.
7182 @item -mone-byte-bool
7183 @opindex -mone-byte-bool
7184 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
7185 By default @samp{sizeof(bool)} is @samp{4} when compiling for
7186 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
7187 option has no effect on x86.
7189 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
7190 to generate code that is not binary compatible with code generated
7191 without that switch. Using this switch may require recompiling all
7192 other modules in a program, including system libraries. Use this
7193 switch to conform to a non-default data model.
7195 @item -mfix-and-continue
7196 @itemx -ffix-and-continue
7197 @itemx -findirect-data
7198 @opindex mfix-and-continue
7199 @opindex ffix-and-continue
7200 @opindex findirect-data
7201 Generate code suitable for fast turn around development. Needed to
7202 enable gdb to dynamically load @code{.o} files into already running
7203 programs. @option{-findirect-data} and @option{-ffix-and-continue}
7204 are provided for backwards compatibility.
7208 Loads all members of static archive libraries.
7209 See man ld(1) for more information.
7211 @item -arch_errors_fatal
7212 @opindex arch_errors_fatal
7213 Cause the errors having to do with files that have the wrong architecture
7217 @opindex bind_at_load
7218 Causes the output file to be marked such that the dynamic linker will
7219 bind all undefined references when the file is loaded or launched.
7223 Produce a Mach-o bundle format file.
7224 See man ld(1) for more information.
7226 @item -bundle_loader @var{executable}
7227 @opindex bundle_loader
7228 This specifies the @var{executable} that will be loading the build
7229 output file being linked. See man ld(1) for more information.
7231 @item -allowable_client @var{client_name}
7235 @itemx -compatibility_version
7236 @itemx -current_version
7238 @itemx -dependency-file
7240 @itemx -dylinker_install_name
7243 @itemx -exported_symbols_list
7245 @itemx -flat_namespace
7246 @itemx -force_cpusubtype_ALL
7247 @itemx -force_flat_namespace
7248 @itemx -headerpad_max_install_names
7251 @itemx -install_name
7252 @itemx -keep_private_externs
7253 @itemx -multi_module
7254 @itemx -multiply_defined
7255 @itemx -multiply_defined_unused
7257 @itemx -no_dead_strip_inits_and_terms
7258 @itemx -nofixprebinding
7261 @itemx -noseglinkedit
7262 @itemx -pagezero_size
7264 @itemx -prebind_all_twolevel_modules
7265 @itemx -private_bundle
7266 @itemx -read_only_relocs
7268 @itemx -sectobjectsymbols
7272 @itemx -sectobjectsymbols
7274 @itemx -seg_addr_table
7275 @itemx -seg_addr_table_filename
7278 @itemx -segs_read_only_addr
7279 @itemx -segs_read_write_addr
7280 @itemx -single_module
7283 @itemx -sub_umbrella
7284 @itemx -twolevel_namespace
7287 @itemx -unexported_symbols_list
7288 @itemx -weak_reference_mismatches
7291 @opindex allowable_client
7293 @opindex client_name
7294 @opindex compatibility_version
7295 @opindex current_version
7297 @opindex dependency-file
7299 @opindex dylinker_install_name
7302 @opindex exported_symbols_list
7304 @opindex flat_namespace
7305 @opindex force_cpusubtype_ALL
7306 @opindex force_flat_namespace
7307 @opindex headerpad_max_install_names
7310 @opindex install_name
7311 @opindex keep_private_externs
7312 @opindex multi_module
7313 @opindex multiply_defined
7314 @opindex multiply_defined_unused
7316 @opindex no_dead_strip_inits_and_terms
7317 @opindex nofixprebinding
7318 @opindex nomultidefs
7320 @opindex noseglinkedit
7321 @opindex pagezero_size
7323 @opindex prebind_all_twolevel_modules
7324 @opindex private_bundle
7325 @opindex read_only_relocs
7327 @opindex sectobjectsymbols
7331 @opindex sectobjectsymbols
7333 @opindex seg_addr_table
7334 @opindex seg_addr_table_filename
7335 @opindex seglinkedit
7337 @opindex segs_read_only_addr
7338 @opindex segs_read_write_addr
7339 @opindex single_module
7341 @opindex sub_library
7342 @opindex sub_umbrella
7343 @opindex twolevel_namespace
7346 @opindex unexported_symbols_list
7347 @opindex weak_reference_mismatches
7348 @opindex whatsloaded
7350 These options are available for Darwin linker. Darwin linker man page
7351 describes them in detail.
7354 @node DEC Alpha Options
7355 @subsection DEC Alpha Options
7357 These @samp{-m} options are defined for the DEC Alpha implementations:
7360 @item -mno-soft-float
7362 @opindex mno-soft-float
7363 @opindex msoft-float
7364 Use (do not use) the hardware floating-point instructions for
7365 floating-point operations. When @option{-msoft-float} is specified,
7366 functions in @file{libgcc.a} will be used to perform floating-point
7367 operations. Unless they are replaced by routines that emulate the
7368 floating-point operations, or compiled in such a way as to call such
7369 emulations routines, these routines will issue floating-point
7370 operations. If you are compiling for an Alpha without floating-point
7371 operations, you must ensure that the library is built so as not to call
7374 Note that Alpha implementations without floating-point operations are
7375 required to have floating-point registers.
7380 @opindex mno-fp-regs
7381 Generate code that uses (does not use) the floating-point register set.
7382 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7383 register set is not used, floating point operands are passed in integer
7384 registers as if they were integers and floating-point results are passed
7385 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
7386 so any function with a floating-point argument or return value called by code
7387 compiled with @option{-mno-fp-regs} must also be compiled with that
7390 A typical use of this option is building a kernel that does not use,
7391 and hence need not save and restore, any floating-point registers.
7395 The Alpha architecture implements floating-point hardware optimized for
7396 maximum performance. It is mostly compliant with the IEEE floating
7397 point standard. However, for full compliance, software assistance is
7398 required. This option generates code fully IEEE compliant code
7399 @emph{except} that the @var{inexact-flag} is not maintained (see below).
7400 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
7401 defined during compilation. The resulting code is less efficient but is
7402 able to correctly support denormalized numbers and exceptional IEEE
7403 values such as not-a-number and plus/minus infinity. Other Alpha
7404 compilers call this option @option{-ieee_with_no_inexact}.
7406 @item -mieee-with-inexact
7407 @opindex mieee-with-inexact
7408 This is like @option{-mieee} except the generated code also maintains
7409 the IEEE @var{inexact-flag}. Turning on this option causes the
7410 generated code to implement fully-compliant IEEE math. In addition to
7411 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
7412 macro. On some Alpha implementations the resulting code may execute
7413 significantly slower than the code generated by default. Since there is
7414 very little code that depends on the @var{inexact-flag}, you should
7415 normally not specify this option. Other Alpha compilers call this
7416 option @option{-ieee_with_inexact}.
7418 @item -mfp-trap-mode=@var{trap-mode}
7419 @opindex mfp-trap-mode
7420 This option controls what floating-point related traps are enabled.
7421 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
7422 The trap mode can be set to one of four values:
7426 This is the default (normal) setting. The only traps that are enabled
7427 are the ones that cannot be disabled in software (e.g., division by zero
7431 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7435 Like @samp{su}, but the instructions are marked to be safe for software
7436 completion (see Alpha architecture manual for details).
7439 Like @samp{su}, but inexact traps are enabled as well.
7442 @item -mfp-rounding-mode=@var{rounding-mode}
7443 @opindex mfp-rounding-mode
7444 Selects the IEEE rounding mode. Other Alpha compilers call this option
7445 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
7450 Normal IEEE rounding mode. Floating point numbers are rounded towards
7451 the nearest machine number or towards the even machine number in case
7455 Round towards minus infinity.
7458 Chopped rounding mode. Floating point numbers are rounded towards zero.
7461 Dynamic rounding mode. A field in the floating point control register
7462 (@var{fpcr}, see Alpha architecture reference manual) controls the
7463 rounding mode in effect. The C library initializes this register for
7464 rounding towards plus infinity. Thus, unless your program modifies the
7465 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7468 @item -mtrap-precision=@var{trap-precision}
7469 @opindex mtrap-precision
7470 In the Alpha architecture, floating point traps are imprecise. This
7471 means without software assistance it is impossible to recover from a
7472 floating trap and program execution normally needs to be terminated.
7473 GCC can generate code that can assist operating system trap handlers
7474 in determining the exact location that caused a floating point trap.
7475 Depending on the requirements of an application, different levels of
7476 precisions can be selected:
7480 Program precision. This option is the default and means a trap handler
7481 can only identify which program caused a floating point exception.
7484 Function precision. The trap handler can determine the function that
7485 caused a floating point exception.
7488 Instruction precision. The trap handler can determine the exact
7489 instruction that caused a floating point exception.
7492 Other Alpha compilers provide the equivalent options called
7493 @option{-scope_safe} and @option{-resumption_safe}.
7495 @item -mieee-conformant
7496 @opindex mieee-conformant
7497 This option marks the generated code as IEEE conformant. You must not
7498 use this option unless you also specify @option{-mtrap-precision=i} and either
7499 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
7500 is to emit the line @samp{.eflag 48} in the function prologue of the
7501 generated assembly file. Under DEC Unix, this has the effect that
7502 IEEE-conformant math library routines will be linked in.
7504 @item -mbuild-constants
7505 @opindex mbuild-constants
7506 Normally GCC examines a 32- or 64-bit integer constant to
7507 see if it can construct it from smaller constants in two or three
7508 instructions. If it cannot, it will output the constant as a literal and
7509 generate code to load it from the data segment at runtime.
7511 Use this option to require GCC to construct @emph{all} integer constants
7512 using code, even if it takes more instructions (the maximum is six).
7514 You would typically use this option to build a shared library dynamic
7515 loader. Itself a shared library, it must relocate itself in memory
7516 before it can find the variables and constants in its own data segment.
7522 Select whether to generate code to be assembled by the vendor-supplied
7523 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
7541 Indicate whether GCC should generate code to use the optional BWX,
7542 CIX, FIX and MAX instruction sets. The default is to use the instruction
7543 sets supported by the CPU type specified via @option{-mcpu=} option or that
7544 of the CPU on which GCC was built if none was specified.
7549 @opindex mfloat-ieee
7550 Generate code that uses (does not use) VAX F and G floating point
7551 arithmetic instead of IEEE single and double precision.
7553 @item -mexplicit-relocs
7554 @itemx -mno-explicit-relocs
7555 @opindex mexplicit-relocs
7556 @opindex mno-explicit-relocs
7557 Older Alpha assemblers provided no way to generate symbol relocations
7558 except via assembler macros. Use of these macros does not allow
7559 optimal instruction scheduling. GNU binutils as of version 2.12
7560 supports a new syntax that allows the compiler to explicitly mark
7561 which relocations should apply to which instructions. This option
7562 is mostly useful for debugging, as GCC detects the capabilities of
7563 the assembler when it is built and sets the default accordingly.
7567 @opindex msmall-data
7568 @opindex mlarge-data
7569 When @option{-mexplicit-relocs} is in effect, static data is
7570 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
7571 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
7572 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
7573 16-bit relocations off of the @code{$gp} register. This limits the
7574 size of the small data area to 64KB, but allows the variables to be
7575 directly accessed via a single instruction.
7577 The default is @option{-mlarge-data}. With this option the data area
7578 is limited to just below 2GB. Programs that require more than 2GB of
7579 data must use @code{malloc} or @code{mmap} to allocate the data in the
7580 heap instead of in the program's data segment.
7582 When generating code for shared libraries, @option{-fpic} implies
7583 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
7587 @opindex msmall-text
7588 @opindex mlarge-text
7589 When @option{-msmall-text} is used, the compiler assumes that the
7590 code of the entire program (or shared library) fits in 4MB, and is
7591 thus reachable with a branch instruction. When @option{-msmall-data}
7592 is used, the compiler can assume that all local symbols share the
7593 same @code{$gp} value, and thus reduce the number of instructions
7594 required for a function call from 4 to 1.
7596 The default is @option{-mlarge-text}.
7598 @item -mcpu=@var{cpu_type}
7600 Set the instruction set and instruction scheduling parameters for
7601 machine type @var{cpu_type}. You can specify either the @samp{EV}
7602 style name or the corresponding chip number. GCC supports scheduling
7603 parameters for the EV4, EV5 and EV6 family of processors and will
7604 choose the default values for the instruction set from the processor
7605 you specify. If you do not specify a processor type, GCC will default
7606 to the processor on which the compiler was built.
7608 Supported values for @var{cpu_type} are
7614 Schedules as an EV4 and has no instruction set extensions.
7618 Schedules as an EV5 and has no instruction set extensions.
7622 Schedules as an EV5 and supports the BWX extension.
7627 Schedules as an EV5 and supports the BWX and MAX extensions.
7631 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
7635 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
7638 @item -mtune=@var{cpu_type}
7640 Set only the instruction scheduling parameters for machine type
7641 @var{cpu_type}. The instruction set is not changed.
7643 @item -mmemory-latency=@var{time}
7644 @opindex mmemory-latency
7645 Sets the latency the scheduler should assume for typical memory
7646 references as seen by the application. This number is highly
7647 dependent on the memory access patterns used by the application
7648 and the size of the external cache on the machine.
7650 Valid options for @var{time} are
7654 A decimal number representing clock cycles.
7660 The compiler contains estimates of the number of clock cycles for
7661 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7662 (also called Dcache, Scache, and Bcache), as well as to main memory.
7663 Note that L3 is only valid for EV5.
7668 @node DEC Alpha/VMS Options
7669 @subsection DEC Alpha/VMS Options
7671 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
7674 @item -mvms-return-codes
7675 @opindex mvms-return-codes
7676 Return VMS condition codes from main. The default is to return POSIX
7677 style condition (e.g.@ error) codes.
7681 @subsection FRV Options
7688 Only use the first 32 general purpose registers.
7693 Use all 64 general purpose registers.
7698 Use only the first 32 floating point registers.
7703 Use all 64 floating point registers
7706 @opindex mhard-float
7708 Use hardware instructions for floating point operations.
7711 @opindex msoft-float
7713 Use library routines for floating point operations.
7718 Dynamically allocate condition code registers.
7723 Do not try to dynamically allocate condition code registers, only
7724 use @code{icc0} and @code{fcc0}.
7729 Change ABI to use double word insns.
7734 Do not use double word instructions.
7739 Use floating point double instructions.
7744 Do not use floating point double instructions.
7749 Use media instructions.
7754 Do not use media instructions.
7759 Use multiply and add/subtract instructions.
7764 Do not use multiply and add/subtract instructions.
7769 Select the FDPIC ABI, that uses function descriptors to represent
7770 pointers to functions. Without any PIC/PIE-related options, it
7771 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
7772 assumes GOT entries and small data are within a 12-bit range from the
7773 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
7774 are computed with 32 bits.
7777 @opindex minline-plt
7779 Enable inlining of PLT entries in function calls to functions that are
7780 not known to bind locally. It has no effect without @option{-mfdpic}.
7781 It's enabled by default if optimizing for speed and compiling for
7782 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
7783 optimization option such as @option{-O3} or above is present in the
7789 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
7790 that is known to be in read-only sections. It's enabled by default,
7791 except for @option{-fpic} or @option{-fpie}: even though it may help
7792 make the global offset table smaller, it trades 1 instruction for 4.
7793 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
7794 one of which may be shared by multiple symbols, and it avoids the need
7795 for a GOT entry for the referenced symbol, so it's more likely to be a
7796 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
7798 @item -multilib-library-pic
7799 @opindex multilib-library-pic
7801 Link with the (library, not FD) pic libraries. It's implied by
7802 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
7803 @option{-fpic} without @option{-mfdpic}. You should never have to use
7809 Follow the EABI requirement of always creating a frame pointer whenever
7810 a stack frame is allocated. This option is enabled by default and can
7811 be disabled with @option{-mno-linked-fp}.
7814 @opindex mlibrary-pic
7816 Generate position-independent EABI code.
7821 Use only the first four media accumulator registers.
7826 Use all eight media accumulator registers.
7831 Pack VLIW instructions.
7836 Do not pack VLIW instructions.
7841 Do not mark ABI switches in e_flags.
7846 Enable the use of conditional-move instructions (default).
7848 This switch is mainly for debugging the compiler and will likely be removed
7849 in a future version.
7851 @item -mno-cond-move
7852 @opindex mno-cond-move
7854 Disable the use of conditional-move instructions.
7856 This switch is mainly for debugging the compiler and will likely be removed
7857 in a future version.
7862 Enable the use of conditional set instructions (default).
7864 This switch is mainly for debugging the compiler and will likely be removed
7865 in a future version.
7870 Disable the use of conditional set instructions.
7872 This switch is mainly for debugging the compiler and will likely be removed
7873 in a future version.
7878 Enable the use of conditional execution (default).
7880 This switch is mainly for debugging the compiler and will likely be removed
7881 in a future version.
7883 @item -mno-cond-exec
7884 @opindex mno-cond-exec
7886 Disable the use of conditional execution.
7888 This switch is mainly for debugging the compiler and will likely be removed
7889 in a future version.
7892 @opindex mvliw-branch
7894 Run a pass to pack branches into VLIW instructions (default).
7896 This switch is mainly for debugging the compiler and will likely be removed
7897 in a future version.
7899 @item -mno-vliw-branch
7900 @opindex mno-vliw-branch
7902 Do not run a pass to pack branches into VLIW instructions.
7904 This switch is mainly for debugging the compiler and will likely be removed
7905 in a future version.
7907 @item -mmulti-cond-exec
7908 @opindex mmulti-cond-exec
7910 Enable optimization of @code{&&} and @code{||} in conditional execution
7913 This switch is mainly for debugging the compiler and will likely be removed
7914 in a future version.
7916 @item -mno-multi-cond-exec
7917 @opindex mno-multi-cond-exec
7919 Disable optimization of @code{&&} and @code{||} in conditional execution.
7921 This switch is mainly for debugging the compiler and will likely be removed
7922 in a future version.
7924 @item -mnested-cond-exec
7925 @opindex mnested-cond-exec
7927 Enable nested conditional execution optimizations (default).
7929 This switch is mainly for debugging the compiler and will likely be removed
7930 in a future version.
7932 @item -mno-nested-cond-exec
7933 @opindex mno-nested-cond-exec
7935 Disable nested conditional execution optimizations.
7937 This switch is mainly for debugging the compiler and will likely be removed
7938 in a future version.
7940 @item -mtomcat-stats
7941 @opindex mtomcat-stats
7943 Cause gas to print out tomcat statistics.
7945 @item -mcpu=@var{cpu}
7948 Select the processor type for which to generate code. Possible values are
7949 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
7954 @node H8/300 Options
7955 @subsection H8/300 Options
7957 These @samp{-m} options are defined for the H8/300 implementations:
7962 Shorten some address references at link time, when possible; uses the
7963 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
7964 ld, Using ld}, for a fuller description.
7968 Generate code for the H8/300H@.
7972 Generate code for the H8S@.
7976 Generate code for the H8S and H8/300H in the normal mode. This switch
7977 must be used either with -mh or -ms.
7981 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
7985 Make @code{int} data 32 bits by default.
7989 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
7990 The default for the H8/300H and H8S is to align longs and floats on 4
7992 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
7993 This option has no effect on the H8/300.
7997 @subsection HPPA Options
7998 @cindex HPPA Options
8000 These @samp{-m} options are defined for the HPPA family of computers:
8003 @item -march=@var{architecture-type}
8005 Generate code for the specified architecture. The choices for
8006 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8007 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8008 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8009 architecture option for your machine. Code compiled for lower numbered
8010 architectures will run on higher numbered architectures, but not the
8014 @itemx -mpa-risc-1-1
8015 @itemx -mpa-risc-2-0
8016 @opindex mpa-risc-1-0
8017 @opindex mpa-risc-1-1
8018 @opindex mpa-risc-2-0
8019 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8022 @opindex mbig-switch
8023 Generate code suitable for big switch tables. Use this option only if
8024 the assembler/linker complain about out of range branches within a switch
8027 @item -mjump-in-delay
8028 @opindex mjump-in-delay
8029 Fill delay slots of function calls with unconditional jump instructions
8030 by modifying the return pointer for the function call to be the target
8031 of the conditional jump.
8033 @item -mdisable-fpregs
8034 @opindex mdisable-fpregs
8035 Prevent floating point registers from being used in any manner. This is
8036 necessary for compiling kernels which perform lazy context switching of
8037 floating point registers. If you use this option and attempt to perform
8038 floating point operations, the compiler will abort.
8040 @item -mdisable-indexing
8041 @opindex mdisable-indexing
8042 Prevent the compiler from using indexing address modes. This avoids some
8043 rather obscure problems when compiling MIG generated code under MACH@.
8045 @item -mno-space-regs
8046 @opindex mno-space-regs
8047 Generate code that assumes the target has no space registers. This allows
8048 GCC to generate faster indirect calls and use unscaled index address modes.
8050 Such code is suitable for level 0 PA systems and kernels.
8052 @item -mfast-indirect-calls
8053 @opindex mfast-indirect-calls
8054 Generate code that assumes calls never cross space boundaries. This
8055 allows GCC to emit code which performs faster indirect calls.
8057 This option will not work in the presence of shared libraries or nested
8060 @item -mfixed-range=@var{register-range}
8061 @opindex mfixed-range
8062 Generate code treating the given register range as fixed registers.
8063 A fixed register is one that the register allocator can not use. This is
8064 useful when compiling kernel code. A register range is specified as
8065 two registers separated by a dash. Multiple register ranges can be
8066 specified separated by a comma.
8068 @item -mlong-load-store
8069 @opindex mlong-load-store
8070 Generate 3-instruction load and store sequences as sometimes required by
8071 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8074 @item -mportable-runtime
8075 @opindex mportable-runtime
8076 Use the portable calling conventions proposed by HP for ELF systems.
8080 Enable the use of assembler directives only GAS understands.
8082 @item -mschedule=@var{cpu-type}
8084 Schedule code according to the constraints for the machine type
8085 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8086 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8087 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8088 proper scheduling option for your machine. The default scheduling is
8092 @opindex mlinker-opt
8093 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8094 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8095 linkers in which they give bogus error messages when linking some programs.
8098 @opindex msoft-float
8099 Generate output containing library calls for floating point.
8100 @strong{Warning:} the requisite libraries are not available for all HPPA
8101 targets. Normally the facilities of the machine's usual C compiler are
8102 used, but this cannot be done directly in cross-compilation. You must make
8103 your own arrangements to provide suitable library functions for
8104 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8105 does provide software floating point support.
8107 @option{-msoft-float} changes the calling convention in the output file;
8108 therefore, it is only useful if you compile @emph{all} of a program with
8109 this option. In particular, you need to compile @file{libgcc.a}, the
8110 library that comes with GCC, with @option{-msoft-float} in order for
8115 Generate the predefine, @code{_SIO}, for server IO. The default is
8116 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8117 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8118 options are available under HP-UX and HI-UX.
8122 Use GNU ld specific options. This passes @option{-shared} to ld when
8123 building a shared library. It is the default when GCC is configured,
8124 explicitly or implicitly, with the GNU linker. This option does not
8125 have any affect on which ld is called, it only changes what parameters
8126 are passed to that ld. The ld that is called is determined by the
8127 @option{--with-ld} configure option, GCC's program search path, and
8128 finally by the user's @env{PATH}. The linker used by GCC can be printed
8129 using @samp{which `gcc -print-prog-name=ld`}.
8133 Use HP ld specific options. This passes @option{-b} to ld when building
8134 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8135 links. It is the default when GCC is configured, explicitly or
8136 implicitly, with the HP linker. This option does not have any affect on
8137 which ld is called, it only changes what parameters are passed to that
8138 ld. The ld that is called is determined by the @option{--with-ld}
8139 configure option, GCC's program search path, and finally by the user's
8140 @env{PATH}. The linker used by GCC can be printed using @samp{which
8141 `gcc -print-prog-name=ld`}.
8144 @opindex mno-long-calls
8145 Generate code that uses long call sequences. This ensures that a call
8146 is always able to reach linker generated stubs. The default is to generate
8147 long calls only when the distance from the call site to the beginning
8148 of the function or translation unit, as the case may be, exceeds a
8149 predefined limit set by the branch type being used. The limits for
8150 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8151 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8154 Distances are measured from the beginning of functions when using the
8155 @option{-ffunction-sections} option, or when using the @option{-mgas}
8156 and @option{-mno-portable-runtime} options together under HP-UX with
8159 It is normally not desirable to use this option as it will degrade
8160 performance. However, it may be useful in large applications,
8161 particularly when partial linking is used to build the application.
8163 The types of long calls used depends on the capabilities of the
8164 assembler and linker, and the type of code being generated. The
8165 impact on systems that support long absolute calls, and long pic
8166 symbol-difference or pc-relative calls should be relatively small.
8167 However, an indirect call is used on 32-bit ELF systems in pic code
8168 and it is quite long.
8170 @item -munix=@var{unix-std}
8172 Generate compiler predefines and select a startfile for the specified
8173 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
8174 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
8175 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
8176 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
8177 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
8180 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
8181 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
8182 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
8183 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
8184 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
8185 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
8187 It is @emph{important} to note that this option changes the interfaces
8188 for various library routines. It also affects the operational behavior
8189 of the C library. Thus, @emph{extreme} care is needed in using this
8192 Library code that is intended to operate with more than one UNIX
8193 standard must test, set and restore the variable @var{__xpg4_extended_mask}
8194 as appropriate. Most GNU software doesn't provide this capability.
8198 Suppress the generation of link options to search libdld.sl when the
8199 @option{-static} option is specified on HP-UX 10 and later.
8203 The HP-UX implementation of setlocale in libc has a dependency on
8204 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8205 when the @option{-static} option is specified, special link options
8206 are needed to resolve this dependency.
8208 On HP-UX 10 and later, the GCC driver adds the necessary options to
8209 link with libdld.sl when the @option{-static} option is specified.
8210 This causes the resulting binary to be dynamic. On the 64-bit port,
8211 the linkers generate dynamic binaries by default in any case. The
8212 @option{-nolibdld} option can be used to prevent the GCC driver from
8213 adding these link options.
8217 Add support for multithreading with the @dfn{dce thread} library
8218 under HP-UX. This option sets flags for both the preprocessor and
8222 @node i386 and x86-64 Options
8223 @subsection Intel 386 and AMD x86-64 Options
8224 @cindex i386 Options
8225 @cindex x86-64 Options
8226 @cindex Intel 386 Options
8227 @cindex AMD x86-64 Options
8229 These @samp{-m} options are defined for the i386 and x86-64 family of
8233 @item -mtune=@var{cpu-type}
8235 Tune to @var{cpu-type} everything applicable about the generated code, except
8236 for the ABI and the set of available instructions. The choices for
8240 Original Intel's i386 CPU.
8242 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8244 Intel Pentium CPU with no MMX support.
8246 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8247 @item i686, pentiumpro
8248 Intel PentiumPro CPU.
8250 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8251 @item pentium3, pentium3m
8252 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8255 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8256 support. Used by Centrino notebooks.
8257 @item pentium4, pentium4m
8258 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8260 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8263 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8264 SSE2 and SSE3 instruction set support.
8266 AMD K6 CPU with MMX instruction set support.
8268 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8269 @item athlon, athlon-tbird
8270 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8272 @item athlon-4, athlon-xp, athlon-mp
8273 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8274 instruction set support.
8275 @item k8, opteron, athlon64, athlon-fx
8276 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8277 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8279 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8282 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8283 instruction set support.
8285 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8286 implemented for this chip.)
8288 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8289 implemented for this chip.)
8292 While picking a specific @var{cpu-type} will schedule things appropriately
8293 for that particular chip, the compiler will not generate any code that
8294 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8297 @item -march=@var{cpu-type}
8299 Generate instructions for the machine type @var{cpu-type}. The choices
8300 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8301 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8303 @item -mcpu=@var{cpu-type}
8305 A deprecated synonym for @option{-mtune}.
8314 @opindex mpentiumpro
8315 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8316 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8317 These synonyms are deprecated.
8319 @item -mfpmath=@var{unit}
8321 Generate floating point arithmetics for selected unit @var{unit}. The choices
8326 Use the standard 387 floating point coprocessor present majority of chips and
8327 emulated otherwise. Code compiled with this option will run almost everywhere.
8328 The temporary results are computed in 80bit precision instead of precision
8329 specified by the type resulting in slightly different results compared to most
8330 of other chips. See @option{-ffloat-store} for more detailed description.
8332 This is the default choice for i386 compiler.
8335 Use scalar floating point instructions present in the SSE instruction set.
8336 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8337 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8338 instruction set supports only single precision arithmetics, thus the double and
8339 extended precision arithmetics is still done using 387. Later version, present
8340 only in Pentium4 and the future AMD x86-64 chips supports double precision
8343 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8344 @option{-msse2} switches to enable SSE extensions and make this option
8345 effective. For x86-64 compiler, these extensions are enabled by default.
8347 The resulting code should be considerably faster in the majority of cases and avoid
8348 the numerical instability problems of 387 code, but may break some existing
8349 code that expects temporaries to be 80bit.
8351 This is the default choice for the x86-64 compiler.
8354 Attempt to utilize both instruction sets at once. This effectively double the
8355 amount of available registers and on chips with separate execution units for
8356 387 and SSE the execution resources too. Use this option with care, as it is
8357 still experimental, because the GCC register allocator does not model separate
8358 functional units well resulting in instable performance.
8361 @item -masm=@var{dialect}
8362 @opindex masm=@var{dialect}
8363 Output asm instructions using selected @var{dialect}. Supported choices are
8364 @samp{intel} or @samp{att} (the default one).
8369 @opindex mno-ieee-fp
8370 Control whether or not the compiler uses IEEE floating point
8371 comparisons. These handle correctly the case where the result of a
8372 comparison is unordered.
8375 @opindex msoft-float
8376 Generate output containing library calls for floating point.
8377 @strong{Warning:} the requisite libraries are not part of GCC@.
8378 Normally the facilities of the machine's usual C compiler are used, but
8379 this can't be done directly in cross-compilation. You must make your
8380 own arrangements to provide suitable library functions for
8383 On machines where a function returns floating point results in the 80387
8384 register stack, some floating point opcodes may be emitted even if
8385 @option{-msoft-float} is used.
8387 @item -mno-fp-ret-in-387
8388 @opindex mno-fp-ret-in-387
8389 Do not use the FPU registers for return values of functions.
8391 The usual calling convention has functions return values of types
8392 @code{float} and @code{double} in an FPU register, even if there
8393 is no FPU@. The idea is that the operating system should emulate
8396 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8397 in ordinary CPU registers instead.
8399 @item -mno-fancy-math-387
8400 @opindex mno-fancy-math-387
8401 Some 387 emulators do not support the @code{sin}, @code{cos} and
8402 @code{sqrt} instructions for the 387. Specify this option to avoid
8403 generating those instructions. This option is the default on FreeBSD,
8404 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8405 indicates that the target cpu will always have an FPU and so the
8406 instruction will not need emulation. As of revision 2.6.1, these
8407 instructions are not generated unless you also use the
8408 @option{-funsafe-math-optimizations} switch.
8410 @item -malign-double
8411 @itemx -mno-align-double
8412 @opindex malign-double
8413 @opindex mno-align-double
8414 Control whether GCC aligns @code{double}, @code{long double}, and
8415 @code{long long} variables on a two word boundary or a one word
8416 boundary. Aligning @code{double} variables on a two word boundary will
8417 produce code that runs somewhat faster on a @samp{Pentium} at the
8418 expense of more memory.
8420 @strong{Warning:} if you use the @option{-malign-double} switch,
8421 structures containing the above types will be aligned differently than
8422 the published application binary interface specifications for the 386
8423 and will not be binary compatible with structures in code compiled
8424 without that switch.
8426 @item -m96bit-long-double
8427 @itemx -m128bit-long-double
8428 @opindex m96bit-long-double
8429 @opindex m128bit-long-double
8430 These switches control the size of @code{long double} type. The i386
8431 application binary interface specifies the size to be 96 bits,
8432 so @option{-m96bit-long-double} is the default in 32 bit mode.
8434 Modern architectures (Pentium and newer) would prefer @code{long double}
8435 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8436 conforming to the ABI, this would not be possible. So specifying a
8437 @option{-m128bit-long-double} will align @code{long double}
8438 to a 16 byte boundary by padding the @code{long double} with an additional
8441 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8442 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8444 Notice that neither of these options enable any extra precision over the x87
8445 standard of 80 bits for a @code{long double}.
8447 @strong{Warning:} if you override the default value for your target ABI, the
8448 structures and arrays containing @code{long double} variables will change
8449 their size as well as function calling convention for function taking
8450 @code{long double} will be modified. Hence they will not be binary
8451 compatible with arrays or structures in code compiled without that switch.
8455 @itemx -mno-svr3-shlib
8456 @opindex msvr3-shlib
8457 @opindex mno-svr3-shlib
8458 Control whether GCC places uninitialized local variables into the
8459 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8460 into @code{bss}. These options are meaningful only on System V Release 3.
8464 Use a different function-calling convention, in which functions that
8465 take a fixed number of arguments return with the @code{ret} @var{num}
8466 instruction, which pops their arguments while returning. This saves one
8467 instruction in the caller since there is no need to pop the arguments
8470 You can specify that an individual function is called with this calling
8471 sequence with the function attribute @samp{stdcall}. You can also
8472 override the @option{-mrtd} option by using the function attribute
8473 @samp{cdecl}. @xref{Function Attributes}.
8475 @strong{Warning:} this calling convention is incompatible with the one
8476 normally used on Unix, so you cannot use it if you need to call
8477 libraries compiled with the Unix compiler.
8479 Also, you must provide function prototypes for all functions that
8480 take variable numbers of arguments (including @code{printf});
8481 otherwise incorrect code will be generated for calls to those
8484 In addition, seriously incorrect code will result if you call a
8485 function with too many arguments. (Normally, extra arguments are
8486 harmlessly ignored.)
8488 @item -mregparm=@var{num}
8490 Control how many registers are used to pass integer arguments. By
8491 default, no registers are used to pass arguments, and at most 3
8492 registers can be used. You can control this behavior for a specific
8493 function by using the function attribute @samp{regparm}.
8494 @xref{Function Attributes}.
8496 @strong{Warning:} if you use this switch, and
8497 @var{num} is nonzero, then you must build all modules with the same
8498 value, including any libraries. This includes the system libraries and
8501 @item -mpreferred-stack-boundary=@var{num}
8502 @opindex mpreferred-stack-boundary
8503 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8504 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8505 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8506 size (@option{-Os}), in which case the default is the minimum correct
8507 alignment (4 bytes for x86, and 8 bytes for x86-64).
8509 On Pentium and PentiumPro, @code{double} and @code{long double} values
8510 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8511 suffer significant run time performance penalties. On Pentium III, the
8512 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8513 penalties if it is not 16 byte aligned.
8515 To ensure proper alignment of this values on the stack, the stack boundary
8516 must be as aligned as that required by any value stored on the stack.
8517 Further, every function must be generated such that it keeps the stack
8518 aligned. Thus calling a function compiled with a higher preferred
8519 stack boundary from a function compiled with a lower preferred stack
8520 boundary will most likely misalign the stack. It is recommended that
8521 libraries that use callbacks always use the default setting.
8523 This extra alignment does consume extra stack space, and generally
8524 increases code size. Code that is sensitive to stack space usage, such
8525 as embedded systems and operating system kernels, may want to reduce the
8526 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8544 These switches enable or disable the use of built-in functions that allow
8545 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8548 @xref{X86 Built-in Functions}, for details of the functions enabled
8549 and disabled by these switches.
8551 To have SSE/SSE2 instructions generated automatically from floating-point
8552 code, see @option{-mfpmath=sse}.
8555 @itemx -mno-push-args
8557 @opindex mno-push-args
8558 Use PUSH operations to store outgoing parameters. This method is shorter
8559 and usually equally fast as method using SUB/MOV operations and is enabled
8560 by default. In some cases disabling it may improve performance because of
8561 improved scheduling and reduced dependencies.
8563 @item -maccumulate-outgoing-args
8564 @opindex maccumulate-outgoing-args
8565 If enabled, the maximum amount of space required for outgoing arguments will be
8566 computed in the function prologue. This is faster on most modern CPUs
8567 because of reduced dependencies, improved scheduling and reduced stack usage
8568 when preferred stack boundary is not equal to 2. The drawback is a notable
8569 increase in code size. This switch implies @option{-mno-push-args}.
8573 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8574 on thread-safe exception handling must compile and link all code with the
8575 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8576 @option{-D_MT}; when linking, it links in a special thread helper library
8577 @option{-lmingwthrd} which cleans up per thread exception handling data.
8579 @item -mno-align-stringops
8580 @opindex mno-align-stringops
8581 Do not align destination of inlined string operations. This switch reduces
8582 code size and improves performance in case the destination is already aligned,
8583 but GCC doesn't know about it.
8585 @item -minline-all-stringops
8586 @opindex minline-all-stringops
8587 By default GCC inlines string operations only when destination is known to be
8588 aligned at least to 4 byte boundary. This enables more inlining, increase code
8589 size, but may improve performance of code that depends on fast memcpy, strlen
8590 and memset for short lengths.
8592 @item -momit-leaf-frame-pointer
8593 @opindex momit-leaf-frame-pointer
8594 Don't keep the frame pointer in a register for leaf functions. This
8595 avoids the instructions to save, set up and restore frame pointers and
8596 makes an extra register available in leaf functions. The option
8597 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8598 which might make debugging harder.
8600 @item -mtls-direct-seg-refs
8601 @itemx -mno-tls-direct-seg-refs
8602 @opindex mtls-direct-seg-refs
8603 Controls whether TLS variables may be accessed with offsets from the
8604 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8605 or whether the thread base pointer must be added. Whether or not this
8606 is legal depends on the operating system, and whether it maps the
8607 segment to cover the entire TLS area.
8609 For systems that use GNU libc, the default is on.
8612 These @samp{-m} switches are supported in addition to the above
8613 on AMD x86-64 processors in 64-bit environments.
8620 Generate code for a 32-bit or 64-bit environment.
8621 The 32-bit environment sets int, long and pointer to 32 bits and
8622 generates code that runs on any i386 system.
8623 The 64-bit environment sets int to 32 bits and long and pointer
8624 to 64 bits and generates code for AMD's x86-64 architecture.
8627 @opindex no-red-zone
8628 Do not use a so called red zone for x86-64 code. The red zone is mandated
8629 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8630 stack pointer that will not be modified by signal or interrupt handlers
8631 and therefore can be used for temporary data without adjusting the stack
8632 pointer. The flag @option{-mno-red-zone} disables this red zone.
8634 @item -mcmodel=small
8635 @opindex mcmodel=small
8636 Generate code for the small code model: the program and its symbols must
8637 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8638 Programs can be statically or dynamically linked. This is the default
8641 @item -mcmodel=kernel
8642 @opindex mcmodel=kernel
8643 Generate code for the kernel code model. The kernel runs in the
8644 negative 2 GB of the address space.
8645 This model has to be used for Linux kernel code.
8647 @item -mcmodel=medium
8648 @opindex mcmodel=medium
8649 Generate code for the medium model: The program is linked in the lower 2
8650 GB of the address space but symbols can be located anywhere in the
8651 address space. Programs can be statically or dynamically linked, but
8652 building of shared libraries are not supported with the medium model.
8654 @item -mcmodel=large
8655 @opindex mcmodel=large
8656 Generate code for the large model: This model makes no assumptions
8657 about addresses and sizes of sections. Currently GCC does not implement
8662 @subsection IA-64 Options
8663 @cindex IA-64 Options
8665 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8669 @opindex mbig-endian
8670 Generate code for a big endian target. This is the default for HP-UX@.
8672 @item -mlittle-endian
8673 @opindex mlittle-endian
8674 Generate code for a little endian target. This is the default for AIX5
8681 Generate (or don't) code for the GNU assembler. This is the default.
8682 @c Also, this is the default if the configure option @option{--with-gnu-as}
8689 Generate (or don't) code for the GNU linker. This is the default.
8690 @c Also, this is the default if the configure option @option{--with-gnu-ld}
8695 Generate code that does not use a global pointer register. The result
8696 is not position independent code, and violates the IA-64 ABI@.
8698 @item -mvolatile-asm-stop
8699 @itemx -mno-volatile-asm-stop
8700 @opindex mvolatile-asm-stop
8701 @opindex mno-volatile-asm-stop
8702 Generate (or don't) a stop bit immediately before and after volatile asm
8707 Generate code that works around Itanium B step errata.
8709 @item -mregister-names
8710 @itemx -mno-register-names
8711 @opindex mregister-names
8712 @opindex mno-register-names
8713 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8714 the stacked registers. This may make assembler output more readable.
8720 Disable (or enable) optimizations that use the small data section. This may
8721 be useful for working around optimizer bugs.
8724 @opindex mconstant-gp
8725 Generate code that uses a single constant global pointer value. This is
8726 useful when compiling kernel code.
8730 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
8731 This is useful when compiling firmware code.
8733 @item -minline-float-divide-min-latency
8734 @opindex minline-float-divide-min-latency
8735 Generate code for inline divides of floating point values
8736 using the minimum latency algorithm.
8738 @item -minline-float-divide-max-throughput
8739 @opindex minline-float-divide-max-throughput
8740 Generate code for inline divides of floating point values
8741 using the maximum throughput algorithm.
8743 @item -minline-int-divide-min-latency
8744 @opindex minline-int-divide-min-latency
8745 Generate code for inline divides of integer values
8746 using the minimum latency algorithm.
8748 @item -minline-int-divide-max-throughput
8749 @opindex minline-int-divide-max-throughput
8750 Generate code for inline divides of integer values
8751 using the maximum throughput algorithm.
8753 @item -mno-dwarf2-asm
8755 @opindex mno-dwarf2-asm
8756 @opindex mdwarf2-asm
8757 Don't (or do) generate assembler code for the DWARF2 line number debugging
8758 info. This may be useful when not using the GNU assembler.
8760 @item -mfixed-range=@var{register-range}
8761 @opindex mfixed-range
8762 Generate code treating the given register range as fixed registers.
8763 A fixed register is one that the register allocator can not use. This is
8764 useful when compiling kernel code. A register range is specified as
8765 two registers separated by a dash. Multiple register ranges can be
8766 specified separated by a comma.
8768 @item -mearly-stop-bits
8769 @itemx -mno-early-stop-bits
8770 @opindex mearly-stop-bits
8771 @opindex mno-early-stop-bits
8772 Allow stop bits to be placed earlier than immediately preceding the
8773 instruction that triggered the stop bit. This can improve instruction
8774 scheduling, but does not always do so.
8777 @node M32R/D Options
8778 @subsection M32R/D Options
8779 @cindex M32R/D options
8781 These @option{-m} options are defined for Renesas M32R/D architectures:
8786 Generate code for the M32R/2@.
8790 Generate code for the M32R/X@.
8794 Generate code for the M32R@. This is the default.
8797 @opindex mmodel=small
8798 Assume all objects live in the lower 16MB of memory (so that their addresses
8799 can be loaded with the @code{ld24} instruction), and assume all subroutines
8800 are reachable with the @code{bl} instruction.
8801 This is the default.
8803 The addressability of a particular object can be set with the
8804 @code{model} attribute.
8806 @item -mmodel=medium
8807 @opindex mmodel=medium
8808 Assume objects may be anywhere in the 32-bit address space (the compiler
8809 will generate @code{seth/add3} instructions to load their addresses), and
8810 assume all subroutines are reachable with the @code{bl} instruction.
8813 @opindex mmodel=large
8814 Assume objects may be anywhere in the 32-bit address space (the compiler
8815 will generate @code{seth/add3} instructions to load their addresses), and
8816 assume subroutines may not be reachable with the @code{bl} instruction
8817 (the compiler will generate the much slower @code{seth/add3/jl}
8818 instruction sequence).
8821 @opindex msdata=none
8822 Disable use of the small data area. Variables will be put into
8823 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
8824 @code{section} attribute has been specified).
8825 This is the default.
8827 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
8828 Objects may be explicitly put in the small data area with the
8829 @code{section} attribute using one of these sections.
8832 @opindex msdata=sdata
8833 Put small global and static data in the small data area, but do not
8834 generate special code to reference them.
8838 Put small global and static data in the small data area, and generate
8839 special instructions to reference them.
8843 @cindex smaller data references
8844 Put global and static objects less than or equal to @var{num} bytes
8845 into the small data or bss sections instead of the normal data or bss
8846 sections. The default value of @var{num} is 8.
8847 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
8848 for this option to have any effect.
8850 All modules should be compiled with the same @option{-G @var{num}} value.
8851 Compiling with different values of @var{num} may or may not work; if it
8852 doesn't the linker will give an error message---incorrect code will not be
8857 Makes the M32R specific code in the compiler display some statistics
8858 that might help in debugging programs.
8861 @opindex malign-loops
8862 Align all loops to a 32-byte boundary.
8864 @item -mno-align-loops
8865 @opindex mno-align-loops
8866 Do not enforce a 32-byte alignment for loops. This is the default.
8868 @item -missue-rate=@var{number}
8869 @opindex missue-rate=@var{number}
8870 Issue @var{number} instructions per cycle. @var{number} can only be 1
8873 @item -mbranch-cost=@var{number}
8874 @opindex mbranch-cost=@var{number}
8875 @var{number} can only be 1 or 2. If it is 1 then branches will be
8876 preferred over conditional code, if it is 2, then the opposite will
8879 @item -mflush-trap=@var{number}
8880 @opindex mflush-trap=@var{number}
8881 Specifies the trap number to use to flush the cache. The default is
8882 12. Valid numbers are between 0 and 15 inclusive.
8884 @item -mno-flush-trap
8885 @opindex mno-flush-trap
8886 Specifies that the cache cannot be flushed by using a trap.
8888 @item -mflush-func=@var{name}
8889 @opindex mflush-func=@var{name}
8890 Specifies the name of the operating system function to call to flush
8891 the cache. The default is @emph{_flush_cache}, but a function call
8892 will only be used if a trap is not available.
8894 @item -mno-flush-func
8895 @opindex mno-flush-func
8896 Indicates that there is no OS function for flushing the cache.
8900 @node M680x0 Options
8901 @subsection M680x0 Options
8902 @cindex M680x0 options
8904 These are the @samp{-m} options defined for the 68000 series. The default
8905 values for these options depends on which style of 68000 was selected when
8906 the compiler was configured; the defaults for the most common choices are
8914 Generate output for a 68000. This is the default
8915 when the compiler is configured for 68000-based systems.
8917 Use this option for microcontrollers with a 68000 or EC000 core,
8918 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
8924 Generate output for a 68020. This is the default
8925 when the compiler is configured for 68020-based systems.
8929 Generate output containing 68881 instructions for floating point.
8930 This is the default for most 68020 systems unless @option{--nfp} was
8931 specified when the compiler was configured.
8935 Generate output for a 68030. This is the default when the compiler is
8936 configured for 68030-based systems.
8940 Generate output for a 68040. This is the default when the compiler is
8941 configured for 68040-based systems.
8943 This option inhibits the use of 68881/68882 instructions that have to be
8944 emulated by software on the 68040. Use this option if your 68040 does not
8945 have code to emulate those instructions.
8949 Generate output for a 68060. This is the default when the compiler is
8950 configured for 68060-based systems.
8952 This option inhibits the use of 68020 and 68881/68882 instructions that
8953 have to be emulated by software on the 68060. Use this option if your 68060
8954 does not have code to emulate those instructions.
8958 Generate output for a CPU32. This is the default
8959 when the compiler is configured for CPU32-based systems.
8961 Use this option for microcontrollers with a
8962 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
8963 68336, 68340, 68341, 68349 and 68360.
8967 Generate output for a 520X ``coldfire'' family cpu. This is the default
8968 when the compiler is configured for 520X-based systems.
8970 Use this option for microcontroller with a 5200 core, including
8971 the MCF5202, MCF5203, MCF5204 and MCF5202.
8976 Generate output for a 68040, without using any of the new instructions.
8977 This results in code which can run relatively efficiently on either a
8978 68020/68881 or a 68030 or a 68040. The generated code does use the
8979 68881 instructions that are emulated on the 68040.
8983 Generate output for a 68060, without using any of the new instructions.
8984 This results in code which can run relatively efficiently on either a
8985 68020/68881 or a 68030 or a 68040. The generated code does use the
8986 68881 instructions that are emulated on the 68060.
8989 @opindex msoft-float
8990 Generate output containing library calls for floating point.
8991 @strong{Warning:} the requisite libraries are not available for all m68k
8992 targets. Normally the facilities of the machine's usual C compiler are
8993 used, but this can't be done directly in cross-compilation. You must
8994 make your own arrangements to provide suitable library functions for
8995 cross-compilation. The embedded targets @samp{m68k-*-aout} and
8996 @samp{m68k-*-coff} do provide software floating point support.
9000 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9001 Additionally, parameters passed on the stack are also aligned to a
9002 16-bit boundary even on targets whose API mandates promotion to 32-bit.
9005 @opindex mnobitfield
9006 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
9007 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
9011 Do use the bit-field instructions. The @option{-m68020} option implies
9012 @option{-mbitfield}. This is the default if you use a configuration
9013 designed for a 68020.
9017 Use a different function-calling convention, in which functions
9018 that take a fixed number of arguments return with the @code{rtd}
9019 instruction, which pops their arguments while returning. This
9020 saves one instruction in the caller since there is no need to pop
9021 the arguments there.
9023 This calling convention is incompatible with the one normally
9024 used on Unix, so you cannot use it if you need to call libraries
9025 compiled with the Unix compiler.
9027 Also, you must provide function prototypes for all functions that
9028 take variable numbers of arguments (including @code{printf});
9029 otherwise incorrect code will be generated for calls to those
9032 In addition, seriously incorrect code will result if you call a
9033 function with too many arguments. (Normally, extra arguments are
9034 harmlessly ignored.)
9036 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
9037 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
9040 @itemx -mno-align-int
9042 @opindex mno-align-int
9043 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
9044 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
9045 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
9046 Aligning variables on 32-bit boundaries produces code that runs somewhat
9047 faster on processors with 32-bit busses at the expense of more memory.
9049 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
9050 align structures containing the above types differently than
9051 most published application binary interface specifications for the m68k.
9055 Use the pc-relative addressing mode of the 68000 directly, instead of
9056 using a global offset table. At present, this option implies @option{-fpic},
9057 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
9058 not presently supported with @option{-mpcrel}, though this could be supported for
9059 68020 and higher processors.
9061 @item -mno-strict-align
9062 @itemx -mstrict-align
9063 @opindex mno-strict-align
9064 @opindex mstrict-align
9065 Do not (do) assume that unaligned memory references will be handled by
9069 Generate code that allows the data segment to be located in a different
9070 area of memory from the text segment. This allows for execute in place in
9071 an environment without virtual memory management. This option implies -fPIC.
9074 Generate code that assumes that the data segment follows the text segment.
9075 This is the default.
9077 @item -mid-shared-library
9078 Generate code that supports shared libraries via the library ID method.
9079 This allows for execute in place and shared libraries in an environment
9080 without virtual memory management. This option implies -fPIC.
9082 @item -mno-id-shared-library
9083 Generate code that doesn't assume ID based shared libraries are being used.
9084 This is the default.
9086 @item -mshared-library-id=n
9087 Specified the identification number of the ID based shared library being
9088 compiled. Specifying a value of 0 will generate more compact code, specifying
9089 other values will force the allocation of that number to the current
9090 library but is no more space or time efficient than omitting this option.
9094 @node M68hc1x Options
9095 @subsection M68hc1x Options
9096 @cindex M68hc1x options
9098 These are the @samp{-m} options defined for the 68hc11 and 68hc12
9099 microcontrollers. The default values for these options depends on
9100 which style of microcontroller was selected when the compiler was configured;
9101 the defaults for the most common choices are given below.
9108 Generate output for a 68HC11. This is the default
9109 when the compiler is configured for 68HC11-based systems.
9115 Generate output for a 68HC12. This is the default
9116 when the compiler is configured for 68HC12-based systems.
9122 Generate output for a 68HCS12.
9125 @opindex mauto-incdec
9126 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
9133 Enable the use of 68HC12 min and max instructions.
9136 @itemx -mno-long-calls
9137 @opindex mlong-calls
9138 @opindex mno-long-calls
9139 Treat all calls as being far away (near). If calls are assumed to be
9140 far away, the compiler will use the @code{call} instruction to
9141 call a function and the @code{rtc} instruction for returning.
9145 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9147 @item -msoft-reg-count=@var{count}
9148 @opindex msoft-reg-count
9149 Specify the number of pseudo-soft registers which are used for the
9150 code generation. The maximum number is 32. Using more pseudo-soft
9151 register may or may not result in better code depending on the program.
9152 The default is 4 for 68HC11 and 2 for 68HC12.
9157 @subsection MCore Options
9158 @cindex MCore options
9160 These are the @samp{-m} options defined for the Motorola M*Core
9168 @opindex mno-hardlit
9169 Inline constants into the code stream if it can be done in two
9170 instructions or less.
9176 Use the divide instruction. (Enabled by default).
9178 @item -mrelax-immediate
9179 @itemx -mno-relax-immediate
9180 @opindex mrelax-immediate
9181 @opindex mno-relax-immediate
9182 Allow arbitrary sized immediates in bit operations.
9184 @item -mwide-bitfields
9185 @itemx -mno-wide-bitfields
9186 @opindex mwide-bitfields
9187 @opindex mno-wide-bitfields
9188 Always treat bit-fields as int-sized.
9190 @item -m4byte-functions
9191 @itemx -mno-4byte-functions
9192 @opindex m4byte-functions
9193 @opindex mno-4byte-functions
9194 Force all functions to be aligned to a four byte boundary.
9196 @item -mcallgraph-data
9197 @itemx -mno-callgraph-data
9198 @opindex mcallgraph-data
9199 @opindex mno-callgraph-data
9200 Emit callgraph information.
9203 @itemx -mno-slow-bytes
9204 @opindex mslow-bytes
9205 @opindex mno-slow-bytes
9206 Prefer word access when reading byte quantities.
9208 @item -mlittle-endian
9210 @opindex mlittle-endian
9211 @opindex mbig-endian
9212 Generate code for a little endian target.
9218 Generate code for the 210 processor.
9222 @subsection MIPS Options
9223 @cindex MIPS options
9229 Generate big-endian code.
9233 Generate little-endian code. This is the default for @samp{mips*el-*-*}
9236 @item -march=@var{arch}
9238 Generate code that will run on @var{arch}, which can be the name of a
9239 generic MIPS ISA, or the name of a particular processor.
9241 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
9242 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
9243 The processor names are:
9244 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
9246 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
9247 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
9251 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
9252 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
9253 The special value @samp{from-abi} selects the
9254 most compatible architecture for the selected ABI (that is,
9255 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
9257 In processor names, a final @samp{000} can be abbreviated as @samp{k}
9258 (for example, @samp{-march=r2k}). Prefixes are optional, and
9259 @samp{vr} may be written @samp{r}.
9261 GCC defines two macros based on the value of this option. The first
9262 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
9263 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
9264 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
9265 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
9266 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
9268 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
9269 above. In other words, it will have the full prefix and will not
9270 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
9271 the macro names the resolved architecture (either @samp{"mips1"} or
9272 @samp{"mips3"}). It names the default architecture when no
9273 @option{-march} option is given.
9275 @item -mtune=@var{arch}
9277 Optimize for @var{arch}. Among other things, this option controls
9278 the way instructions are scheduled, and the perceived cost of arithmetic
9279 operations. The list of @var{arch} values is the same as for
9282 When this option is not used, GCC will optimize for the processor
9283 specified by @option{-march}. By using @option{-march} and
9284 @option{-mtune} together, it is possible to generate code that will
9285 run on a family of processors, but optimize the code for one
9286 particular member of that family.
9288 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
9289 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
9290 @samp{-march} ones described above.
9294 Equivalent to @samp{-march=mips1}.
9298 Equivalent to @samp{-march=mips2}.
9302 Equivalent to @samp{-march=mips3}.
9306 Equivalent to @samp{-march=mips4}.
9310 Equivalent to @samp{-march=mips32}.
9314 Equivalent to @samp{-march=mips32r2}.
9318 Equivalent to @samp{-march=mips64}.
9324 Use (do not use) the MIPS16 ISA.
9336 Generate code for the given ABI@.
9338 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
9339 generates 64-bit code when you select a 64-bit architecture, but you
9340 can use @option{-mgp32} to get 32-bit code instead.
9342 For information about the O64 ABI, see
9343 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
9346 @itemx -mno-abicalls
9348 @opindex mno-abicalls
9349 Generate (do not generate) SVR4-style position-independent code.
9350 @option{-mabicalls} is the default for SVR4-based systems.
9356 Lift (do not lift) the usual restrictions on the size of the global
9359 GCC normally uses a single instruction to load values from the GOT.
9360 While this is relatively efficient, it will only work if the GOT
9361 is smaller than about 64k. Anything larger will cause the linker
9362 to report an error such as:
9364 @cindex relocation truncated to fit (MIPS)
9366 relocation truncated to fit: R_MIPS_GOT16 foobar
9369 If this happens, you should recompile your code with @option{-mxgot}.
9370 It should then work with very large GOTs, although it will also be
9371 less efficient, since it will take three instructions to fetch the
9372 value of a global symbol.
9374 Note that some linkers can create multiple GOTs. If you have such a
9375 linker, you should only need to use @option{-mxgot} when a single object
9376 file accesses more than 64k's worth of GOT entries. Very few do.
9378 These options have no effect unless GCC is generating position
9383 Assume that general-purpose registers are 32 bits wide.
9387 Assume that general-purpose registers are 64 bits wide.
9391 Assume that floating-point registers are 32 bits wide.
9395 Assume that floating-point registers are 64 bits wide.
9398 @opindex mhard-float
9399 Use floating-point coprocessor instructions.
9402 @opindex msoft-float
9403 Do not use floating-point coprocessor instructions. Implement
9404 floating-point calculations using library calls instead.
9406 @item -msingle-float
9407 @opindex msingle-float
9408 Assume that the floating-point coprocessor only supports single-precision
9411 @itemx -mdouble-float
9412 @opindex mdouble-float
9413 Assume that the floating-point coprocessor supports double-precision
9414 operations. This is the default.
9416 @itemx -mpaired-single
9417 @itemx -mno-paired-single
9418 @opindex mpaired-single
9419 @opindex mno-paired-single
9420 Use (do not use) the paired single instructions.
9426 Use (do not use) the MIPS-3D ASE. The option @option{-mips3d} implies
9427 @option{-mpaired-single}.
9431 Force @code{int} and @code{long} types to be 64 bits wide. See
9432 @option{-mlong32} for an explanation of the default and the way
9433 that the pointer size is determined.
9437 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
9438 an explanation of the default and the way that the pointer size is
9443 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
9445 The default size of @code{int}s, @code{long}s and pointers depends on
9446 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
9447 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
9448 32-bit @code{long}s. Pointers are the same size as @code{long}s,
9449 or the same size as integer registers, whichever is smaller.
9453 @cindex smaller data references (MIPS)
9454 @cindex gp-relative references (MIPS)
9455 Put global and static items less than or equal to @var{num} bytes into
9456 the small data or bss section instead of the normal data or bss section.
9457 This allows the data to be accessed using a single instruction.
9459 All modules should be compiled with the same @option{-G @var{num}}
9462 @item -membedded-data
9463 @itemx -mno-embedded-data
9464 @opindex membedded-data
9465 @opindex mno-embedded-data
9466 Allocate variables to the read-only data section first if possible, then
9467 next in the small data section if possible, otherwise in data. This gives
9468 slightly slower code than the default, but reduces the amount of RAM required
9469 when executing, and thus may be preferred for some embedded systems.
9471 @item -muninit-const-in-rodata
9472 @itemx -mno-uninit-const-in-rodata
9473 @opindex muninit-const-in-rodata
9474 @opindex mno-uninit-const-in-rodata
9475 Put uninitialized @code{const} variables in the read-only data section.
9476 This option is only meaningful in conjunction with @option{-membedded-data}.
9478 @item -msplit-addresses
9479 @itemx -mno-split-addresses
9480 @opindex msplit-addresses
9481 @opindex mno-split-addresses
9482 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
9483 relocation operators. This option has been superceded by
9484 @option{-mexplicit-relocs} but is retained for backwards compatibility.
9486 @item -mexplicit-relocs
9487 @itemx -mno-explicit-relocs
9488 @opindex mexplicit-relocs
9489 @opindex mno-explicit-relocs
9490 Use (do not use) assembler relocation operators when dealing with symbolic
9491 addresses. The alternative, selected by @option{-mno-explicit-relocs},
9492 is to use assembler macros instead.
9494 @option{-mexplicit-relocs} is the default if GCC was configured
9495 to use an assembler that supports relocation operators.
9497 @item -mcheck-zero-division
9498 @itemx -mno-check-zero-division
9499 @opindex mcheck-zero-division
9500 @opindex mno-check-zero-division
9501 Trap (do not trap) on integer division by zero. The default is
9502 @option{-mcheck-zero-division}.
9504 @item -mdivide-traps
9505 @itemx -mdivide-breaks
9506 @opindex mdivide-traps
9507 @opindex mdivide-breaks
9508 MIPS systems check for division by zero by generating either a
9509 conditional trap or a break instruction. Using traps results in
9510 smaller code, but is only supported on MIPS II and later. Also, some
9511 versions of the Linux kernel have a bug that prevents trap from
9512 generating the proper signal (SIGFPE). Use @option{-mdivide-traps} to
9513 allow conditional traps on architectures that support them and
9514 @option{-mdivide-breaks} to force the use of breaks.
9516 The default is usually @option{-mdivide-traps}, but this can be
9517 overridden at configure time using @option{--with-divide=breaks}.
9518 Divide-by-zero checks can be completely disabled using
9519 @option{-mno-check-zero-division}.
9525 Force (do not force) the use of @code{memcpy()} for non-trivial block
9526 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
9527 most constant-sized copies.
9530 @itemx -mno-long-calls
9531 @opindex mlong-calls
9532 @opindex mno-long-calls
9533 Disable (do not disable) use of the @code{jal} instruction. Calling
9534 functions using @code{jal} is more efficient but requires the caller
9535 and callee to be in the same 256 megabyte segment.
9537 This option has no effect on abicalls code. The default is
9538 @option{-mno-long-calls}.
9544 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
9545 instructions, as provided by the R4650 ISA.
9548 @itemx -mno-fused-madd
9549 @opindex mfused-madd
9550 @opindex mno-fused-madd
9551 Enable (disable) use of the floating point multiply-accumulate
9552 instructions, when they are available. The default is
9553 @option{-mfused-madd}.
9555 When multiply-accumulate instructions are used, the intermediate
9556 product is calculated to infinite precision and is not subject to
9557 the FCSR Flush to Zero bit. This may be undesirable in some
9562 Tell the MIPS assembler to not run its preprocessor over user
9563 assembler files (with a @samp{.s} suffix) when assembling them.
9566 @itemx -mno-fix-r4000
9568 @opindex mno-fix-r4000
9569 Work around certain R4000 CPU errata:
9572 A double-word or a variable shift may give an incorrect result if executed
9573 immediately after starting an integer division.
9575 A double-word or a variable shift may give an incorrect result if executed
9576 while an integer multiplication is in progress.
9578 An integer division may give an incorrect result if started in a delay slot
9579 of a taken branch or a jump.
9583 @itemx -mno-fix-r4400
9585 @opindex mno-fix-r4400
9586 Work around certain R4400 CPU errata:
9589 A double-word or a variable shift may give an incorrect result if executed
9590 immediately after starting an integer division.
9594 @itemx -mno-fix-vr4120
9595 @opindex mfix-vr4120
9596 Work around certain VR4120 errata:
9599 @code{dmultu} does not always produce the correct result.
9601 @code{div} and @code{ddiv} do not always produce the correct result if one
9602 of the operands is negative.
9604 The workarounds for the division errata rely on special functions in
9605 @file{libgcc.a}. At present, these functions are only provided by
9606 the @code{mips64vr*-elf} configurations.
9608 Other VR4120 errata require a nop to be inserted between certain pairs of
9609 instructions. These errata are handled by the assembler, not by GCC itself.
9614 Work around certain SB-1 CPU core errata.
9615 (This flag currently works around the SB-1 revision 2
9616 ``F1'' and ``F2'' floating point errata.)
9618 @item -mflush-func=@var{func}
9619 @itemx -mno-flush-func
9620 @opindex mflush-func
9621 Specifies the function to call to flush the I and D caches, or to not
9622 call any such function. If called, the function must take the same
9623 arguments as the common @code{_flush_func()}, that is, the address of the
9624 memory range for which the cache is being flushed, the size of the
9625 memory range, and the number 3 (to flush both caches). The default
9626 depends on the target GCC was configured for, but commonly is either
9627 @samp{_flush_func} or @samp{__cpu_flush}.
9629 @item -mbranch-likely
9630 @itemx -mno-branch-likely
9631 @opindex mbranch-likely
9632 @opindex mno-branch-likely
9633 Enable or disable use of Branch Likely instructions, regardless of the
9634 default for the selected architecture. By default, Branch Likely
9635 instructions may be generated if they are supported by the selected
9636 architecture. An exception is for the MIPS32 and MIPS64 architectures
9637 and processors which implement those architectures; for those, Branch
9638 Likely instructions will not be generated by default because the MIPS32
9639 and MIPS64 architectures specifically deprecate their use.
9641 @item -mfp-exceptions
9642 @itemx -mno-fp-exceptions
9643 @opindex mfp-exceptions
9644 Specifies whether FP exceptions are enabled. This affects how we schedule
9645 FP instructions for some processors. The default is that FP exceptions are
9648 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
9649 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
9652 @item -mvr4130-align
9653 @itemx -mno-vr4130-align
9654 @opindex mvr4130-align
9655 The VR4130 pipeline is two-way superscalar, but can only issue two
9656 instructions together if the first one is 8-byte aligned. When this
9657 option is enabled, GCC will align pairs of instructions that it
9658 thinks should execute in parallel.
9660 This option only has an effect when optimizing for the VR4130.
9661 It normally makes code faster, but at the expense of making it bigger.
9662 It is enabled by default at optimization level @option{-O3}.
9666 @subsection MMIX Options
9667 @cindex MMIX Options
9669 These options are defined for the MMIX:
9673 @itemx -mno-libfuncs
9675 @opindex mno-libfuncs
9676 Specify that intrinsic library functions are being compiled, passing all
9677 values in registers, no matter the size.
9682 @opindex mno-epsilon
9683 Generate floating-point comparison instructions that compare with respect
9684 to the @code{rE} epsilon register.
9686 @item -mabi=mmixware
9688 @opindex mabi-mmixware
9690 Generate code that passes function parameters and return values that (in
9691 the called function) are seen as registers @code{$0} and up, as opposed to
9692 the GNU ABI which uses global registers @code{$231} and up.
9695 @itemx -mno-zero-extend
9696 @opindex mzero-extend
9697 @opindex mno-zero-extend
9698 When reading data from memory in sizes shorter than 64 bits, use (do not
9699 use) zero-extending load instructions by default, rather than
9700 sign-extending ones.
9703 @itemx -mno-knuthdiv
9705 @opindex mno-knuthdiv
9706 Make the result of a division yielding a remainder have the same sign as
9707 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9708 remainder follows the sign of the dividend. Both methods are
9709 arithmetically valid, the latter being almost exclusively used.
9711 @item -mtoplevel-symbols
9712 @itemx -mno-toplevel-symbols
9713 @opindex mtoplevel-symbols
9714 @opindex mno-toplevel-symbols
9715 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9716 code can be used with the @code{PREFIX} assembly directive.
9720 Generate an executable in the ELF format, rather than the default
9721 @samp{mmo} format used by the @command{mmix} simulator.
9723 @item -mbranch-predict
9724 @itemx -mno-branch-predict
9725 @opindex mbranch-predict
9726 @opindex mno-branch-predict
9727 Use (do not use) the probable-branch instructions, when static branch
9728 prediction indicates a probable branch.
9730 @item -mbase-addresses
9731 @itemx -mno-base-addresses
9732 @opindex mbase-addresses
9733 @opindex mno-base-addresses
9734 Generate (do not generate) code that uses @emph{base addresses}. Using a
9735 base address automatically generates a request (handled by the assembler
9736 and the linker) for a constant to be set up in a global register. The
9737 register is used for one or more base address requests within the range 0
9738 to 255 from the value held in the register. The generally leads to short
9739 and fast code, but the number of different data items that can be
9740 addressed is limited. This means that a program that uses lots of static
9741 data may require @option{-mno-base-addresses}.
9744 @itemx -mno-single-exit
9745 @opindex msingle-exit
9746 @opindex mno-single-exit
9747 Force (do not force) generated code to have a single exit point in each
9751 @node MN10300 Options
9752 @subsection MN10300 Options
9753 @cindex MN10300 options
9755 These @option{-m} options are defined for Matsushita MN10300 architectures:
9760 Generate code to avoid bugs in the multiply instructions for the MN10300
9761 processors. This is the default.
9764 @opindex mno-mult-bug
9765 Do not generate code to avoid bugs in the multiply instructions for the
9770 Generate code which uses features specific to the AM33 processor.
9774 Do not generate code which uses features specific to the AM33 processor. This
9779 Do not link in the C run-time initialization object file.
9783 Indicate to the linker that it should perform a relaxation optimization pass
9784 to shorten branches, calls and absolute memory addresses. This option only
9785 has an effect when used on the command line for the final link step.
9787 This option makes symbolic debugging impossible.
9791 @subsection NS32K Options
9792 @cindex NS32K options
9794 These are the @samp{-m} options defined for the 32000 series. The default
9795 values for these options depends on which style of 32000 was selected when
9796 the compiler was configured; the defaults for the most common choices are
9804 Generate output for a 32032. This is the default
9805 when the compiler is configured for 32032 and 32016 based systems.
9811 Generate output for a 32332. This is the default
9812 when the compiler is configured for 32332-based systems.
9818 Generate output for a 32532. This is the default
9819 when the compiler is configured for 32532-based systems.
9823 Generate output containing 32081 instructions for floating point.
9824 This is the default for all systems.
9828 Generate output containing 32381 instructions for floating point. This
9829 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9830 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9834 Try and generate multiply-add floating point instructions @code{polyF}
9835 and @code{dotF}. This option is only available if the @option{-m32381}
9836 option is in effect. Using these instructions requires changes to
9837 register allocation which generally has a negative impact on
9838 performance. This option should only be enabled when compiling code
9839 particularly likely to make heavy use of multiply-add instructions.
9842 @opindex mnomulti-add
9843 Do not try and generate multiply-add floating point instructions
9844 @code{polyF} and @code{dotF}. This is the default on all platforms.
9847 @opindex msoft-float
9848 Generate output containing library calls for floating point.
9849 @strong{Warning:} the requisite libraries may not be available.
9851 @item -mieee-compare
9852 @itemx -mno-ieee-compare
9853 @opindex mieee-compare
9854 @opindex mno-ieee-compare
9855 Control whether or not the compiler uses IEEE floating point
9856 comparisons. These handle correctly the case where the result of a
9857 comparison is unordered.
9858 @strong{Warning:} the requisite kernel support may not be available.
9861 @opindex mnobitfield
9862 Do not use the bit-field instructions. On some machines it is faster to
9863 use shifting and masking operations. This is the default for the pc532.
9867 Do use the bit-field instructions. This is the default for all platforms
9872 Use a different function-calling convention, in which functions
9873 that take a fixed number of arguments return pop their
9874 arguments on return with the @code{ret} instruction.
9876 This calling convention is incompatible with the one normally
9877 used on Unix, so you cannot use it if you need to call libraries
9878 compiled with the Unix compiler.
9880 Also, you must provide function prototypes for all functions that
9881 take variable numbers of arguments (including @code{printf});
9882 otherwise incorrect code will be generated for calls to those
9885 In addition, seriously incorrect code will result if you call a
9886 function with too many arguments. (Normally, extra arguments are
9887 harmlessly ignored.)
9889 This option takes its name from the 680x0 @code{rtd} instruction.
9894 Use a different function-calling convention where the first two arguments
9895 are passed in registers.
9897 This calling convention is incompatible with the one normally
9898 used on Unix, so you cannot use it if you need to call libraries
9899 compiled with the Unix compiler.
9902 @opindex mnoregparam
9903 Do not pass any arguments in registers. This is the default for all
9908 It is OK to use the sb as an index register which is always loaded with
9909 zero. This is the default for the pc532-netbsd target.
9913 The sb register is not available for use or has not been initialized to
9914 zero by the run time system. This is the default for all targets except
9915 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9916 @option{-fpic} is set.
9920 Many ns32000 series addressing modes use displacements of up to 512MB@.
9921 If an address is above 512MB then displacements from zero can not be used.
9922 This option causes code to be generated which can be loaded above 512MB@.
9923 This may be useful for operating systems or ROM code.
9927 Assume code will be loaded in the first 512MB of virtual address space.
9928 This is the default for all platforms.
9932 @node PDP-11 Options
9933 @subsection PDP-11 Options
9934 @cindex PDP-11 Options
9936 These options are defined for the PDP-11:
9941 Use hardware FPP floating point. This is the default. (FIS floating
9942 point on the PDP-11/40 is not supported.)
9945 @opindex msoft-float
9946 Do not use hardware floating point.
9950 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
9954 Return floating-point results in memory. This is the default.
9958 Generate code for a PDP-11/40.
9962 Generate code for a PDP-11/45. This is the default.
9966 Generate code for a PDP-11/10.
9968 @item -mbcopy-builtin
9969 @opindex bcopy-builtin
9970 Use inline @code{movmemhi} patterns for copying memory. This is the
9975 Do not use inline @code{movmemhi} patterns for copying memory.
9981 Use 16-bit @code{int}. This is the default.
9987 Use 32-bit @code{int}.
9992 @opindex mno-float32
9993 Use 64-bit @code{float}. This is the default.
9998 @opindex mno-float64
9999 Use 32-bit @code{float}.
10003 Use @code{abshi2} pattern. This is the default.
10007 Do not use @code{abshi2} pattern.
10009 @item -mbranch-expensive
10010 @opindex mbranch-expensive
10011 Pretend that branches are expensive. This is for experimenting with
10012 code generation only.
10014 @item -mbranch-cheap
10015 @opindex mbranch-cheap
10016 Do not pretend that branches are expensive. This is the default.
10020 Generate code for a system with split I&D.
10024 Generate code for a system without split I&D. This is the default.
10028 Use Unix assembler syntax. This is the default when configured for
10029 @samp{pdp11-*-bsd}.
10033 Use DEC assembler syntax. This is the default when configured for any
10034 PDP-11 target other than @samp{pdp11-*-bsd}.
10037 @node PowerPC Options
10038 @subsection PowerPC Options
10039 @cindex PowerPC options
10041 These are listed under @xref{RS/6000 and PowerPC Options}.
10043 @node RS/6000 and PowerPC Options
10044 @subsection IBM RS/6000 and PowerPC Options
10045 @cindex RS/6000 and PowerPC Options
10046 @cindex IBM RS/6000 and PowerPC Options
10048 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
10055 @itemx -mno-powerpc
10056 @itemx -mpowerpc-gpopt
10057 @itemx -mno-powerpc-gpopt
10058 @itemx -mpowerpc-gfxopt
10059 @itemx -mno-powerpc-gfxopt
10061 @itemx -mno-powerpc64
10065 @opindex mno-power2
10067 @opindex mno-powerpc
10068 @opindex mpowerpc-gpopt
10069 @opindex mno-powerpc-gpopt
10070 @opindex mpowerpc-gfxopt
10071 @opindex mno-powerpc-gfxopt
10072 @opindex mpowerpc64
10073 @opindex mno-powerpc64
10074 GCC supports two related instruction set architectures for the
10075 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
10076 instructions supported by the @samp{rios} chip set used in the original
10077 RS/6000 systems and the @dfn{PowerPC} instruction set is the
10078 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
10079 the IBM 4xx microprocessors.
10081 Neither architecture is a subset of the other. However there is a
10082 large common subset of instructions supported by both. An MQ
10083 register is included in processors supporting the POWER architecture.
10085 You use these options to specify which instructions are available on the
10086 processor you are using. The default value of these options is
10087 determined when configuring GCC@. Specifying the
10088 @option{-mcpu=@var{cpu_type}} overrides the specification of these
10089 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
10090 rather than the options listed above.
10092 The @option{-mpower} option allows GCC to generate instructions that
10093 are found only in the POWER architecture and to use the MQ register.
10094 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
10095 to generate instructions that are present in the POWER2 architecture but
10096 not the original POWER architecture.
10098 The @option{-mpowerpc} option allows GCC to generate instructions that
10099 are found only in the 32-bit subset of the PowerPC architecture.
10100 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
10101 GCC to use the optional PowerPC architecture instructions in the
10102 General Purpose group, including floating-point square root. Specifying
10103 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
10104 use the optional PowerPC architecture instructions in the Graphics
10105 group, including floating-point select.
10107 The @option{-mpowerpc64} option allows GCC to generate the additional
10108 64-bit instructions that are found in the full PowerPC64 architecture
10109 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
10110 @option{-mno-powerpc64}.
10112 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
10113 will use only the instructions in the common subset of both
10114 architectures plus some special AIX common-mode calls, and will not use
10115 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
10116 permits GCC to use any instruction from either architecture and to
10117 allow use of the MQ register; specify this for the Motorola MPC601.
10119 @item -mnew-mnemonics
10120 @itemx -mold-mnemonics
10121 @opindex mnew-mnemonics
10122 @opindex mold-mnemonics
10123 Select which mnemonics to use in the generated assembler code. With
10124 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
10125 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
10126 assembler mnemonics defined for the POWER architecture. Instructions
10127 defined in only one architecture have only one mnemonic; GCC uses that
10128 mnemonic irrespective of which of these options is specified.
10130 GCC defaults to the mnemonics appropriate for the architecture in
10131 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
10132 value of these option. Unless you are building a cross-compiler, you
10133 should normally not specify either @option{-mnew-mnemonics} or
10134 @option{-mold-mnemonics}, but should instead accept the default.
10136 @item -mcpu=@var{cpu_type}
10138 Set architecture type, register usage, choice of mnemonics, and
10139 instruction scheduling parameters for machine type @var{cpu_type}.
10140 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
10141 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
10142 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
10143 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
10144 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
10145 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
10146 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
10147 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
10148 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
10150 @option{-mcpu=common} selects a completely generic processor. Code
10151 generated under this option will run on any POWER or PowerPC processor.
10152 GCC will use only the instructions in the common subset of both
10153 architectures, and will not use the MQ register. GCC assumes a generic
10154 processor model for scheduling purposes.
10156 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
10157 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
10158 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
10159 types, with an appropriate, generic processor model assumed for
10160 scheduling purposes.
10162 The other options specify a specific processor. Code generated under
10163 those options will run best on that processor, and may not run at all on
10166 The @option{-mcpu} options automatically enable or disable the
10167 following options: @option{-maltivec}, @option{-mhard-float},
10168 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
10169 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
10170 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
10171 @option{-mstring}. The particular options set for any particular CPU
10172 will vary between compiler versions, depending on what setting seems
10173 to produce optimal code for that CPU; it doesn't necessarily reflect
10174 the actual hardware's capabilities. If you wish to set an individual
10175 option to a particular value, you may specify it after the
10176 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
10178 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
10179 not enabled or disabled by the @option{-mcpu} option at present, since
10180 AIX does not have full support for these options. You may still
10181 enable or disable them individually if you're sure it'll work in your
10184 @item -mtune=@var{cpu_type}
10186 Set the instruction scheduling parameters for machine type
10187 @var{cpu_type}, but do not set the architecture type, register usage, or
10188 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
10189 values for @var{cpu_type} are used for @option{-mtune} as for
10190 @option{-mcpu}. If both are specified, the code generated will use the
10191 architecture, registers, and mnemonics set by @option{-mcpu}, but the
10192 scheduling parameters set by @option{-mtune}.
10195 @itemx -mno-altivec
10197 @opindex mno-altivec
10198 These switches enable or disable the use of built-in functions that
10199 allow access to the AltiVec instruction set. You may also need to set
10200 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
10205 Extend the current ABI with SPE ABI extensions. This does not change
10206 the default ABI, instead it adds the SPE ABI extensions to the current
10210 @opindex mabi=no-spe
10211 Disable Booke SPE ABI extensions for the current ABI.
10213 @item -misel=@var{yes/no}
10216 This switch enables or disables the generation of ISEL instructions.
10218 @item -mspe=@var{yes/no}
10221 This switch enables or disables the generation of SPE simd
10224 @item -mfloat-gprs=@var{yes/no}
10225 @itemx -mfloat-gprs
10226 @opindex mfloat-gprs
10227 This switch enables or disables the generation of floating point
10228 operations on the general purpose registers for architectures that
10229 support it. This option is currently only available on the MPC8540.
10235 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
10236 targets (including GNU/Linux). The 32-bit environment sets int, long
10237 and pointer to 32 bits and generates code that runs on any PowerPC
10238 variant. The 64-bit environment sets int to 32 bits and long and
10239 pointer to 64 bits, and generates code for PowerPC64, as for
10240 @option{-mpowerpc64}.
10243 @itemx -mno-fp-in-toc
10244 @itemx -mno-sum-in-toc
10245 @itemx -mminimal-toc
10247 @opindex mno-fp-in-toc
10248 @opindex mno-sum-in-toc
10249 @opindex mminimal-toc
10250 Modify generation of the TOC (Table Of Contents), which is created for
10251 every executable file. The @option{-mfull-toc} option is selected by
10252 default. In that case, GCC will allocate at least one TOC entry for
10253 each unique non-automatic variable reference in your program. GCC
10254 will also place floating-point constants in the TOC@. However, only
10255 16,384 entries are available in the TOC@.
10257 If you receive a linker error message that saying you have overflowed
10258 the available TOC space, you can reduce the amount of TOC space used
10259 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
10260 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
10261 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
10262 generate code to calculate the sum of an address and a constant at
10263 run-time instead of putting that sum into the TOC@. You may specify one
10264 or both of these options. Each causes GCC to produce very slightly
10265 slower and larger code at the expense of conserving TOC space.
10267 If you still run out of space in the TOC even when you specify both of
10268 these options, specify @option{-mminimal-toc} instead. This option causes
10269 GCC to make only one TOC entry for every file. When you specify this
10270 option, GCC will produce code that is slower and larger but which
10271 uses extremely little TOC space. You may wish to use this option
10272 only on files that contain less frequently executed code.
10278 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
10279 @code{long} type, and the infrastructure needed to support them.
10280 Specifying @option{-maix64} implies @option{-mpowerpc64} and
10281 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
10282 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
10285 @itemx -mno-xl-call
10287 @opindex mno-xl-call
10288 On AIX, pass floating-point arguments to prototyped functions beyond the
10289 register save area (RSA) on the stack in addition to argument FPRs. The
10290 AIX calling convention was extended but not initially documented to
10291 handle an obscure K&R C case of calling a function that takes the
10292 address of its arguments with fewer arguments than declared. AIX XL
10293 compilers access floating point arguments which do not fit in the
10294 RSA from the stack when a subroutine is compiled without
10295 optimization. Because always storing floating-point arguments on the
10296 stack is inefficient and rarely needed, this option is not enabled by
10297 default and only is necessary when calling subroutines compiled by AIX
10298 XL compilers without optimization.
10302 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
10303 application written to use message passing with special startup code to
10304 enable the application to run. The system must have PE installed in the
10305 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
10306 must be overridden with the @option{-specs=} option to specify the
10307 appropriate directory location. The Parallel Environment does not
10308 support threads, so the @option{-mpe} option and the @option{-pthread}
10309 option are incompatible.
10311 @item -malign-natural
10312 @itemx -malign-power
10313 @opindex malign-natural
10314 @opindex malign-power
10315 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
10316 @option{-malign-natural} overrides the ABI-defined alignment of larger
10317 types, such as floating-point doubles, on their natural size-based boundary.
10318 The option @option{-malign-power} instructs GCC to follow the ABI-specified
10319 alignment rules. GCC defaults to the standard alignment defined in the ABI.
10322 @itemx -mhard-float
10323 @opindex msoft-float
10324 @opindex mhard-float
10325 Generate code that does not use (uses) the floating-point register set.
10326 Software floating point emulation is provided if you use the
10327 @option{-msoft-float} option, and pass the option to GCC when linking.
10330 @itemx -mno-multiple
10332 @opindex mno-multiple
10333 Generate code that uses (does not use) the load multiple word
10334 instructions and the store multiple word instructions. These
10335 instructions are generated by default on POWER systems, and not
10336 generated on PowerPC systems. Do not use @option{-mmultiple} on little
10337 endian PowerPC systems, since those instructions do not work when the
10338 processor is in little endian mode. The exceptions are PPC740 and
10339 PPC750 which permit the instructions usage in little endian mode.
10344 @opindex mno-string
10345 Generate code that uses (does not use) the load string instructions
10346 and the store string word instructions to save multiple registers and
10347 do small block moves. These instructions are generated by default on
10348 POWER systems, and not generated on PowerPC systems. Do not use
10349 @option{-mstring} on little endian PowerPC systems, since those
10350 instructions do not work when the processor is in little endian mode.
10351 The exceptions are PPC740 and PPC750 which permit the instructions
10352 usage in little endian mode.
10357 @opindex mno-update
10358 Generate code that uses (does not use) the load or store instructions
10359 that update the base register to the address of the calculated memory
10360 location. These instructions are generated by default. If you use
10361 @option{-mno-update}, there is a small window between the time that the
10362 stack pointer is updated and the address of the previous frame is
10363 stored, which means code that walks the stack frame across interrupts or
10364 signals may get corrupted data.
10367 @itemx -mno-fused-madd
10368 @opindex mfused-madd
10369 @opindex mno-fused-madd
10370 Generate code that uses (does not use) the floating point multiply and
10371 accumulate instructions. These instructions are generated by default if
10372 hardware floating is used.
10374 @item -mno-bit-align
10376 @opindex mno-bit-align
10377 @opindex mbit-align
10378 On System V.4 and embedded PowerPC systems do not (do) force structures
10379 and unions that contain bit-fields to be aligned to the base type of the
10382 For example, by default a structure containing nothing but 8
10383 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
10384 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
10385 the structure would be aligned to a 1 byte boundary and be one byte in
10388 @item -mno-strict-align
10389 @itemx -mstrict-align
10390 @opindex mno-strict-align
10391 @opindex mstrict-align
10392 On System V.4 and embedded PowerPC systems do not (do) assume that
10393 unaligned memory references will be handled by the system.
10395 @item -mrelocatable
10396 @itemx -mno-relocatable
10397 @opindex mrelocatable
10398 @opindex mno-relocatable
10399 On embedded PowerPC systems generate code that allows (does not allow)
10400 the program to be relocated to a different address at runtime. If you
10401 use @option{-mrelocatable} on any module, all objects linked together must
10402 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
10404 @item -mrelocatable-lib
10405 @itemx -mno-relocatable-lib
10406 @opindex mrelocatable-lib
10407 @opindex mno-relocatable-lib
10408 On embedded PowerPC systems generate code that allows (does not allow)
10409 the program to be relocated to a different address at runtime. Modules
10410 compiled with @option{-mrelocatable-lib} can be linked with either modules
10411 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
10412 with modules compiled with the @option{-mrelocatable} options.
10418 On System V.4 and embedded PowerPC systems do not (do) assume that
10419 register 2 contains a pointer to a global area pointing to the addresses
10420 used in the program.
10423 @itemx -mlittle-endian
10425 @opindex mlittle-endian
10426 On System V.4 and embedded PowerPC systems compile code for the
10427 processor in little endian mode. The @option{-mlittle-endian} option is
10428 the same as @option{-mlittle}.
10431 @itemx -mbig-endian
10433 @opindex mbig-endian
10434 On System V.4 and embedded PowerPC systems compile code for the
10435 processor in big endian mode. The @option{-mbig-endian} option is
10436 the same as @option{-mbig}.
10438 @item -mdynamic-no-pic
10439 @opindex mdynamic-no-pic
10440 On Darwin and Mac OS X systems, compile code so that it is not
10441 relocatable, but that its external references are relocatable. The
10442 resulting code is suitable for applications, but not shared
10445 @item -mprioritize-restricted-insns=@var{priority}
10446 @opindex mprioritize-restricted-insns
10447 This option controls the priority that is assigned to
10448 dispatch-slot restricted instructions during the second scheduling
10449 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
10450 @var{no/highest/second-highest} priority to dispatch slot restricted
10453 @item -msched-costly-dep=@var{dependence_type}
10454 @opindex msched-costly-dep
10455 This option controls which dependences are considered costly
10456 by the target during instruction scheduling. The argument
10457 @var{dependence_type} takes one of the following values:
10458 @var{no}: no dependence is costly,
10459 @var{all}: all dependences are costly,
10460 @var{true_store_to_load}: a true dependence from store to load is costly,
10461 @var{store_to_load}: any dependence from store to load is costly,
10462 @var{number}: any dependence which latency >= @var{number} is costly.
10464 @item -minsert-sched-nops=@var{scheme}
10465 @opindex minsert-sched-nops
10466 This option controls which nop insertion scheme will be used during
10467 the second scheduling pass. The argument @var{scheme} takes one of the
10469 @var{no}: Don't insert nops.
10470 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
10471 according to the scheduler's grouping.
10472 @var{regroup_exact}: Insert nops to force costly dependent insns into
10473 separate groups. Insert exactly as many nops as needed to force an insn
10474 to a new group, according to the estimated processor grouping.
10475 @var{number}: Insert nops to force costly dependent insns into
10476 separate groups. Insert @var{number} nops to force an insn to a new group.
10479 @opindex mcall-sysv
10480 On System V.4 and embedded PowerPC systems compile code using calling
10481 conventions that adheres to the March 1995 draft of the System V
10482 Application Binary Interface, PowerPC processor supplement. This is the
10483 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
10485 @item -mcall-sysv-eabi
10486 @opindex mcall-sysv-eabi
10487 Specify both @option{-mcall-sysv} and @option{-meabi} options.
10489 @item -mcall-sysv-noeabi
10490 @opindex mcall-sysv-noeabi
10491 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
10493 @item -mcall-solaris
10494 @opindex mcall-solaris
10495 On System V.4 and embedded PowerPC systems compile code for the Solaris
10499 @opindex mcall-linux
10500 On System V.4 and embedded PowerPC systems compile code for the
10501 Linux-based GNU system.
10505 On System V.4 and embedded PowerPC systems compile code for the
10506 Hurd-based GNU system.
10508 @item -mcall-netbsd
10509 @opindex mcall-netbsd
10510 On System V.4 and embedded PowerPC systems compile code for the
10511 NetBSD operating system.
10513 @item -maix-struct-return
10514 @opindex maix-struct-return
10515 Return all structures in memory (as specified by the AIX ABI)@.
10517 @item -msvr4-struct-return
10518 @opindex msvr4-struct-return
10519 Return structures smaller than 8 bytes in registers (as specified by the
10522 @item -mabi=altivec
10523 @opindex mabi=altivec
10524 Extend the current ABI with AltiVec ABI extensions. This does not
10525 change the default ABI, instead it adds the AltiVec ABI extensions to
10528 @item -mabi=no-altivec
10529 @opindex mabi=no-altivec
10530 Disable AltiVec ABI extensions for the current ABI.
10533 @itemx -mno-prototype
10534 @opindex mprototype
10535 @opindex mno-prototype
10536 On System V.4 and embedded PowerPC systems assume that all calls to
10537 variable argument functions are properly prototyped. Otherwise, the
10538 compiler must insert an instruction before every non prototyped call to
10539 set or clear bit 6 of the condition code register (@var{CR}) to
10540 indicate whether floating point values were passed in the floating point
10541 registers in case the function takes a variable arguments. With
10542 @option{-mprototype}, only calls to prototyped variable argument functions
10543 will set or clear the bit.
10547 On embedded PowerPC systems, assume that the startup module is called
10548 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
10549 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
10554 On embedded PowerPC systems, assume that the startup module is called
10555 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
10560 On embedded PowerPC systems, assume that the startup module is called
10561 @file{crt0.o} and the standard C libraries are @file{libads.a} and
10564 @item -myellowknife
10565 @opindex myellowknife
10566 On embedded PowerPC systems, assume that the startup module is called
10567 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
10572 On System V.4 and embedded PowerPC systems, specify that you are
10573 compiling for a VxWorks system.
10577 Specify that you are compiling for the WindISS simulation environment.
10581 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
10582 header to indicate that @samp{eabi} extended relocations are used.
10588 On System V.4 and embedded PowerPC systems do (do not) adhere to the
10589 Embedded Applications Binary Interface (eabi) which is a set of
10590 modifications to the System V.4 specifications. Selecting @option{-meabi}
10591 means that the stack is aligned to an 8 byte boundary, a function
10592 @code{__eabi} is called to from @code{main} to set up the eabi
10593 environment, and the @option{-msdata} option can use both @code{r2} and
10594 @code{r13} to point to two separate small data areas. Selecting
10595 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
10596 do not call an initialization function from @code{main}, and the
10597 @option{-msdata} option will only use @code{r13} to point to a single
10598 small data area. The @option{-meabi} option is on by default if you
10599 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
10602 @opindex msdata=eabi
10603 On System V.4 and embedded PowerPC systems, put small initialized
10604 @code{const} global and static data in the @samp{.sdata2} section, which
10605 is pointed to by register @code{r2}. Put small initialized
10606 non-@code{const} global and static data in the @samp{.sdata} section,
10607 which is pointed to by register @code{r13}. Put small uninitialized
10608 global and static data in the @samp{.sbss} section, which is adjacent to
10609 the @samp{.sdata} section. The @option{-msdata=eabi} option is
10610 incompatible with the @option{-mrelocatable} option. The
10611 @option{-msdata=eabi} option also sets the @option{-memb} option.
10614 @opindex msdata=sysv
10615 On System V.4 and embedded PowerPC systems, put small global and static
10616 data in the @samp{.sdata} section, which is pointed to by register
10617 @code{r13}. Put small uninitialized global and static data in the
10618 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
10619 The @option{-msdata=sysv} option is incompatible with the
10620 @option{-mrelocatable} option.
10622 @item -msdata=default
10624 @opindex msdata=default
10626 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
10627 compile code the same as @option{-msdata=eabi}, otherwise compile code the
10628 same as @option{-msdata=sysv}.
10631 @opindex msdata-data
10632 On System V.4 and embedded PowerPC systems, put small global and static
10633 data in the @samp{.sdata} section. Put small uninitialized global and
10634 static data in the @samp{.sbss} section. Do not use register @code{r13}
10635 to address small data however. This is the default behavior unless
10636 other @option{-msdata} options are used.
10640 @opindex msdata=none
10642 On embedded PowerPC systems, put all initialized global and static data
10643 in the @samp{.data} section, and all uninitialized data in the
10644 @samp{.bss} section.
10648 @cindex smaller data references (PowerPC)
10649 @cindex .sdata/.sdata2 references (PowerPC)
10650 On embedded PowerPC systems, put global and static items less than or
10651 equal to @var{num} bytes into the small data or bss sections instead of
10652 the normal data or bss section. By default, @var{num} is 8. The
10653 @option{-G @var{num}} switch is also passed to the linker.
10654 All modules should be compiled with the same @option{-G @var{num}} value.
10657 @itemx -mno-regnames
10659 @opindex mno-regnames
10660 On System V.4 and embedded PowerPC systems do (do not) emit register
10661 names in the assembly language output using symbolic forms.
10664 @itemx -mno-longcall
10666 @opindex mno-longcall
10667 Default to making all function calls indirectly, using a register, so
10668 that functions which reside further than 32 megabytes (33,554,432
10669 bytes) from the current location can be called. This setting can be
10670 overridden by the @code{shortcall} function attribute, or by
10671 @code{#pragma longcall(0)}.
10673 Some linkers are capable of detecting out-of-range calls and generating
10674 glue code on the fly. On these systems, long calls are unnecessary and
10675 generate slower code. As of this writing, the AIX linker can do this,
10676 as can the GNU linker for PowerPC/64. It is planned to add this feature
10677 to the GNU linker for 32-bit PowerPC systems as well.
10679 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
10680 callee, L42'', plus a ``branch island'' (glue code). The two target
10681 addresses represent the callee and the ``branch island.'' The
10682 Darwin/PPC linker will prefer the first address and generate a ``bl
10683 callee'' if the PPC ``bl'' instruction will reach the callee directly;
10684 otherwise, the linker will generate ``bl L42'' to call the ``branch
10685 island.'' The ``branch island'' is appended to the body of the
10686 calling function; it computes the full 32-bit address of the callee
10689 On Mach-O (Darwin) systems, this option directs the compiler emit to
10690 the glue for every direct call, and the Darwin linker decides whether
10691 to use or discard it.
10693 In the future, we may cause GCC to ignore all longcall specifications
10694 when the linker is known to generate glue.
10698 Adds support for multithreading with the @dfn{pthreads} library.
10699 This option sets flags for both the preprocessor and linker.
10703 @node S/390 and zSeries Options
10704 @subsection S/390 and zSeries Options
10705 @cindex S/390 and zSeries Options
10707 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10711 @itemx -msoft-float
10712 @opindex mhard-float
10713 @opindex msoft-float
10714 Use (do not use) the hardware floating-point instructions and registers
10715 for floating-point operations. When @option{-msoft-float} is specified,
10716 functions in @file{libgcc.a} will be used to perform floating-point
10717 operations. When @option{-mhard-float} is specified, the compiler
10718 generates IEEE floating-point instructions. This is the default.
10721 @itemx -mno-backchain
10722 @itemx -mkernel-backchain
10723 @opindex mbackchain
10724 @opindex mno-backchain
10725 @opindex mkernel-backchain
10726 In order to provide a backchain the address of the caller's frame
10727 is stored within the callee's stack frame.
10728 A backchain may be needed to allow debugging using tools that do not understand
10729 DWARF-2 call frame information.
10730 For @option{-mno-backchain} no backchain is maintained at all which is the
10732 If one of the other options is present the backchain pointer is placed either
10733 on top of the stack frame (@option{-mkernel-backchain}) or on
10734 the bottom (@option{-mbackchain}).
10735 Beside the different backchain location @option{-mkernel-backchain}
10736 also changes stack frame layout breaking the ABI. This option
10737 is intended to be used for code which internally needs a backchain but has
10738 to get by with a limited stack size e.g. the linux kernel.
10739 Internal unwinding code not using DWARF-2 info has to be able to locate the
10740 return address of a function. That will be eased be the fact that
10741 the return address of a function is placed two words below the backchain
10745 @itemx -mno-small-exec
10746 @opindex msmall-exec
10747 @opindex mno-small-exec
10748 Generate (or do not generate) code using the @code{bras} instruction
10749 to do subroutine calls.
10750 This only works reliably if the total executable size does not
10751 exceed 64k. The default is to use the @code{basr} instruction instead,
10752 which does not have this limitation.
10758 When @option{-m31} is specified, generate code compliant to the
10759 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10760 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10761 particular to generate 64-bit instructions. For the @samp{s390}
10762 targets, the default is @option{-m31}, while the @samp{s390x}
10763 targets default to @option{-m64}.
10769 When @option{-mzarch} is specified, generate code using the
10770 instructions available on z/Architecture.
10771 When @option{-mesa} is specified, generate code using the
10772 instructions available on ESA/390. Note that @option{-mesa} is
10773 not possible with @option{-m64}.
10774 When generating code compliant to the GNU/Linux for S/390 ABI,
10775 the default is @option{-mesa}. When generating code compliant
10776 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10782 Generate (or do not generate) code using the @code{mvcle} instruction
10783 to perform block moves. When @option{-mno-mvcle} is specified,
10784 use a @code{mvc} loop instead. This is the default.
10790 Print (or do not print) additional debug information when compiling.
10791 The default is to not print debug information.
10793 @item -march=@var{cpu-type}
10795 Generate code that will run on @var{cpu-type}, which is the name of a system
10796 representing a certain processor type. Possible values for
10797 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10798 When generating code using the instructions available on z/Architecture,
10799 the default is @option{-march=z900}. Otherwise, the default is
10800 @option{-march=g5}.
10802 @item -mtune=@var{cpu-type}
10804 Tune to @var{cpu-type} everything applicable about the generated code,
10805 except for the ABI and the set of available instructions.
10806 The list of @var{cpu-type} values is the same as for @option{-march}.
10807 The default is the value used for @option{-march}.
10810 @itemx -mno-tpf-trace
10811 @opindex mtpf-trace
10812 @opindex mno-tpf-trace
10813 Generate code that adds (does not add) in TPF OS specific branches to trace
10814 routines in the operating system. This option is off by default, even
10815 when compiling for the TPF OS.
10818 @itemx -mno-fused-madd
10819 @opindex mfused-madd
10820 @opindex mno-fused-madd
10821 Generate code that uses (does not use) the floating point multiply and
10822 accumulate instructions. These instructions are generated by default if
10823 hardware floating point is used.
10825 @item -mwarn-framesize=@var{framesize}
10826 @opindex mwarn-framesize
10827 Emit a warning if the current function exceeds the given frame size. Because
10828 this is a compile time check it doesn't need to be a real problem when the program
10829 runs. It is intended to identify functions which most probably cause
10830 a stack overflow. It is useful to be used in an environment with limited stack
10831 size e.g. the linux kernel.
10833 @item -mwarn-dynamicstack
10834 @opindex mwarn-dynamicstack
10835 Emit a warning if the function calls alloca or uses dynamically
10836 sized arrays. This is generally a bad idea with a limited stack size.
10838 @item -mstack-guard=@var{stack-guard}
10839 @item -mstack-size=@var{stack-size}
10840 @opindex mstack-guard
10841 @opindex mstack-size
10842 These arguments always have to be used in conjunction. If they are present the s390
10843 back end emits additional instructions in the function prologue which trigger a trap
10844 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
10845 (remember that the stack on s390 grows downward). These options are intended to
10846 be used to help debugging stack overflow problems. The additionally emitted code
10847 cause only little overhead and hence can also be used in production like systems
10848 without greater performance degradation. The given values have to be exact
10849 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard}.
10850 In order to be efficient the extra code makes the assumption that the stack starts
10851 at an address aligned to the value given by @var{stack-size}.
10855 @subsection SH Options
10857 These @samp{-m} options are defined for the SH implementations:
10862 Generate code for the SH1.
10866 Generate code for the SH2.
10869 Generate code for the SH2e.
10873 Generate code for the SH3.
10877 Generate code for the SH3e.
10881 Generate code for the SH4 without a floating-point unit.
10883 @item -m4-single-only
10884 @opindex m4-single-only
10885 Generate code for the SH4 with a floating-point unit that only
10886 supports single-precision arithmetic.
10890 Generate code for the SH4 assuming the floating-point unit is in
10891 single-precision mode by default.
10895 Generate code for the SH4.
10899 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
10900 floating-point unit is not used.
10902 @item -m4a-single-only
10903 @opindex m4a-single-only
10904 Generate code for the SH4a, in such a way that no double-precision
10905 floating point operations are used.
10908 @opindex m4a-single
10909 Generate code for the SH4a assuming the floating-point unit is in
10910 single-precision mode by default.
10914 Generate code for the SH4a.
10918 Same as @option{-m4a-nofpu}, except that it implicitly passes
10919 @option{-dsp} to the assembler. GCC doesn't generate any DSP
10920 instructions at the moment.
10924 Compile code for the processor in big endian mode.
10928 Compile code for the processor in little endian mode.
10932 Align doubles at 64-bit boundaries. Note that this changes the calling
10933 conventions, and thus some functions from the standard C library will
10934 not work unless you recompile it first with @option{-mdalign}.
10938 Shorten some address references at link time, when possible; uses the
10939 linker option @option{-relax}.
10943 Use 32-bit offsets in @code{switch} tables. The default is to use
10948 Enable the use of the instruction @code{fmovd}.
10952 Comply with the calling conventions defined by Renesas.
10956 Comply with the calling conventions defined by Renesas.
10960 Comply with the calling conventions defined for GCC before the Renesas
10961 conventions were available. This option is the default for all
10962 targets of the SH toolchain except for @samp{sh-symbianelf}.
10965 @opindex mnomacsave
10966 Mark the @code{MAC} register as call-clobbered, even if
10967 @option{-mhitachi} is given.
10971 Increase IEEE-compliance of floating-point code.
10975 Dump instruction size and location in the assembly code.
10978 @opindex mpadstruct
10979 This option is deprecated. It pads structures to multiple of 4 bytes,
10980 which is incompatible with the SH ABI@.
10984 Optimize for space instead of speed. Implied by @option{-Os}.
10987 @opindex mprefergot
10988 When generating position-independent code, emit function calls using
10989 the Global Offset Table instead of the Procedure Linkage Table.
10993 Generate a library function call to invalidate instruction cache
10994 entries, after fixing up a trampoline. This library function call
10995 doesn't assume it can write to the whole memory address space. This
10996 is the default when the target is @code{sh-*-linux*}.
10999 @node SPARC Options
11000 @subsection SPARC Options
11001 @cindex SPARC options
11003 These @samp{-m} options are supported on the SPARC:
11006 @item -mno-app-regs
11008 @opindex mno-app-regs
11010 Specify @option{-mapp-regs} to generate output using the global registers
11011 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
11014 To be fully SVR4 ABI compliant at the cost of some performance loss,
11015 specify @option{-mno-app-regs}. You should compile libraries and system
11016 software with this option.
11019 @itemx -mhard-float
11021 @opindex mhard-float
11022 Generate output containing floating point instructions. This is the
11026 @itemx -msoft-float
11028 @opindex msoft-float
11029 Generate output containing library calls for floating point.
11030 @strong{Warning:} the requisite libraries are not available for all SPARC
11031 targets. Normally the facilities of the machine's usual C compiler are
11032 used, but this cannot be done directly in cross-compilation. You must make
11033 your own arrangements to provide suitable library functions for
11034 cross-compilation. The embedded targets @samp{sparc-*-aout} and
11035 @samp{sparclite-*-*} do provide software floating point support.
11037 @option{-msoft-float} changes the calling convention in the output file;
11038 therefore, it is only useful if you compile @emph{all} of a program with
11039 this option. In particular, you need to compile @file{libgcc.a}, the
11040 library that comes with GCC, with @option{-msoft-float} in order for
11043 @item -mhard-quad-float
11044 @opindex mhard-quad-float
11045 Generate output containing quad-word (long double) floating point
11048 @item -msoft-quad-float
11049 @opindex msoft-quad-float
11050 Generate output containing library calls for quad-word (long double)
11051 floating point instructions. The functions called are those specified
11052 in the SPARC ABI@. This is the default.
11054 As of this writing, there are no SPARC implementations that have hardware
11055 support for the quad-word floating point instructions. They all invoke
11056 a trap handler for one of these instructions, and then the trap handler
11057 emulates the effect of the instruction. Because of the trap handler overhead,
11058 this is much slower than calling the ABI library routines. Thus the
11059 @option{-msoft-quad-float} option is the default.
11061 @item -mno-unaligned-doubles
11062 @itemx -munaligned-doubles
11063 @opindex mno-unaligned-doubles
11064 @opindex munaligned-doubles
11065 Assume that doubles have 8 byte alignment. This is the default.
11067 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
11068 alignment only if they are contained in another type, or if they have an
11069 absolute address. Otherwise, it assumes they have 4 byte alignment.
11070 Specifying this option avoids some rare compatibility problems with code
11071 generated by other compilers. It is not the default because it results
11072 in a performance loss, especially for floating point code.
11074 @item -mno-faster-structs
11075 @itemx -mfaster-structs
11076 @opindex mno-faster-structs
11077 @opindex mfaster-structs
11078 With @option{-mfaster-structs}, the compiler assumes that structures
11079 should have 8 byte alignment. This enables the use of pairs of
11080 @code{ldd} and @code{std} instructions for copies in structure
11081 assignment, in place of twice as many @code{ld} and @code{st} pairs.
11082 However, the use of this changed alignment directly violates the SPARC
11083 ABI@. Thus, it's intended only for use on targets where the developer
11084 acknowledges that their resulting code will not be directly in line with
11085 the rules of the ABI@.
11087 @item -mimpure-text
11088 @opindex mimpure-text
11089 @option{-mimpure-text}, used in addition to @option{-shared}, tells
11090 the compiler to not pass @option{-z text} to the linker when linking a
11091 shared object. Using this option, you can link position-dependent
11092 code into a shared object.
11094 @option{-mimpure-text} suppresses the ``relocations remain against
11095 allocatable but non-writable sections'' linker error message.
11096 However, the necessary relocations will trigger copy-on-write, and the
11097 shared object is not actually shared across processes. Instead of
11098 using @option{-mimpure-text}, you should compile all source code with
11099 @option{-fpic} or @option{-fPIC}.
11101 This option is only available on SunOS and Solaris.
11103 @item -mcpu=@var{cpu_type}
11105 Set the instruction set, register set, and instruction scheduling parameters
11106 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
11107 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
11108 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
11109 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
11110 @samp{ultrasparc3}.
11112 Default instruction scheduling parameters are used for values that select
11113 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
11114 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
11116 Here is a list of each supported architecture and their supported
11121 v8: supersparc, hypersparc
11122 sparclite: f930, f934, sparclite86x
11124 v9: ultrasparc, ultrasparc3
11127 By default (unless configured otherwise), GCC generates code for the V7
11128 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
11129 additionally optimizes it for the Cypress CY7C602 chip, as used in the
11130 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
11131 SPARCStation 1, 2, IPX etc.
11133 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
11134 architecture. The only difference from V7 code is that the compiler emits
11135 the integer multiply and integer divide instructions which exist in SPARC-V8
11136 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
11137 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
11140 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
11141 the SPARC architecture. This adds the integer multiply, integer divide step
11142 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
11143 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
11144 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
11145 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
11146 MB86934 chip, which is the more recent SPARClite with FPU.
11148 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
11149 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
11150 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
11151 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
11152 optimizes it for the TEMIC SPARClet chip.
11154 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
11155 architecture. This adds 64-bit integer and floating-point move instructions,
11156 3 additional floating-point condition code registers and conditional move
11157 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
11158 optimizes it for the Sun UltraSPARC I/II chips. With
11159 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
11160 Sun UltraSPARC III chip.
11162 @item -mtune=@var{cpu_type}
11164 Set the instruction scheduling parameters for machine type
11165 @var{cpu_type}, but do not set the instruction set or register set that the
11166 option @option{-mcpu=@var{cpu_type}} would.
11168 The same values for @option{-mcpu=@var{cpu_type}} can be used for
11169 @option{-mtune=@var{cpu_type}}, but the only useful values are those
11170 that select a particular cpu implementation. Those are @samp{cypress},
11171 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
11172 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
11173 @samp{ultrasparc3}.
11178 @opindex mno-v8plus
11179 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
11180 difference from the V8 ABI is that the global and out registers are
11181 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
11182 mode for all SPARC-V9 processors.
11188 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
11189 Visual Instruction Set extensions. The default is @option{-mno-vis}.
11192 These @samp{-m} options are supported in addition to the above
11193 on SPARC-V9 processors in 64-bit environments:
11196 @item -mlittle-endian
11197 @opindex mlittle-endian
11198 Generate code for a processor running in little-endian mode. It is only
11199 available for a few configurations and most notably not on Solaris.
11205 Generate code for a 32-bit or 64-bit environment.
11206 The 32-bit environment sets int, long and pointer to 32 bits.
11207 The 64-bit environment sets int to 32 bits and long and pointer
11210 @item -mcmodel=medlow
11211 @opindex mcmodel=medlow
11212 Generate code for the Medium/Low code model: 64-bit addresses, programs
11213 must be linked in the low 32 bits of memory. Programs can be statically
11214 or dynamically linked.
11216 @item -mcmodel=medmid
11217 @opindex mcmodel=medmid
11218 Generate code for the Medium/Middle code model: 64-bit addresses, programs
11219 must be linked in the low 44 bits of memory, the text and data segments must
11220 be less than 2GB in size and the data segment must be located within 2GB of
11223 @item -mcmodel=medany
11224 @opindex mcmodel=medany
11225 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
11226 may be linked anywhere in memory, the text and data segments must be less
11227 than 2GB in size and the data segment must be located within 2GB of the
11230 @item -mcmodel=embmedany
11231 @opindex mcmodel=embmedany
11232 Generate code for the Medium/Anywhere code model for embedded systems:
11233 64-bit addresses, the text and data segments must be less than 2GB in
11234 size, both starting anywhere in memory (determined at link time). The
11235 global register %g4 points to the base of the data segment. Programs
11236 are statically linked and PIC is not supported.
11239 @itemx -mno-stack-bias
11240 @opindex mstack-bias
11241 @opindex mno-stack-bias
11242 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
11243 frame pointer if present, are offset by @minus{}2047 which must be added back
11244 when making stack frame references. This is the default in 64-bit mode.
11245 Otherwise, assume no such offset is present.
11248 These switches are supported in addition to the above on Solaris:
11253 Add support for multithreading using the Solaris threads library. This
11254 option sets flags for both the preprocessor and linker. This option does
11255 not affect the thread safety of object code produced by the compiler or
11256 that of libraries supplied with it.
11260 Add support for multithreading using the POSIX threads library. This
11261 option sets flags for both the preprocessor and linker. This option does
11262 not affect the thread safety of object code produced by the compiler or
11263 that of libraries supplied with it.
11266 @node System V Options
11267 @subsection Options for System V
11269 These additional options are available on System V Release 4 for
11270 compatibility with other compilers on those systems:
11275 Create a shared object.
11276 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
11280 Identify the versions of each tool used by the compiler, in a
11281 @code{.ident} assembler directive in the output.
11285 Refrain from adding @code{.ident} directives to the output file (this is
11288 @item -YP,@var{dirs}
11290 Search the directories @var{dirs}, and no others, for libraries
11291 specified with @option{-l}.
11293 @item -Ym,@var{dir}
11295 Look in the directory @var{dir} to find the M4 preprocessor.
11296 The assembler uses this option.
11297 @c This is supposed to go with a -Yd for predefined M4 macro files, but
11298 @c the generic assembler that comes with Solaris takes just -Ym.
11301 @node TMS320C3x/C4x Options
11302 @subsection TMS320C3x/C4x Options
11303 @cindex TMS320C3x/C4x Options
11305 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
11309 @item -mcpu=@var{cpu_type}
11311 Set the instruction set, register set, and instruction scheduling
11312 parameters for machine type @var{cpu_type}. Supported values for
11313 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
11314 @samp{c44}. The default is @samp{c40} to generate code for the
11319 @itemx -msmall-memory
11321 @opindex mbig-memory
11323 @opindex msmall-memory
11325 Generates code for the big or small memory model. The small memory
11326 model assumed that all data fits into one 64K word page. At run-time
11327 the data page (DP) register must be set to point to the 64K page
11328 containing the .bss and .data program sections. The big memory model is
11329 the default and requires reloading of the DP register for every direct
11336 Allow (disallow) allocation of general integer operands into the block
11337 count register BK@.
11343 Enable (disable) generation of code using decrement and branch,
11344 DBcond(D), instructions. This is enabled by default for the C4x. To be
11345 on the safe side, this is disabled for the C3x, since the maximum
11346 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
11347 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
11348 that it can utilize the decrement and branch instruction, but will give
11349 up if there is more than one memory reference in the loop. Thus a loop
11350 where the loop counter is decremented can generate slightly more
11351 efficient code, in cases where the RPTB instruction cannot be utilized.
11353 @item -mdp-isr-reload
11355 @opindex mdp-isr-reload
11357 Force the DP register to be saved on entry to an interrupt service
11358 routine (ISR), reloaded to point to the data section, and restored on
11359 exit from the ISR@. This should not be required unless someone has
11360 violated the small memory model by modifying the DP register, say within
11367 For the C3x use the 24-bit MPYI instruction for integer multiplies
11368 instead of a library call to guarantee 32-bit results. Note that if one
11369 of the operands is a constant, then the multiplication will be performed
11370 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
11371 then squaring operations are performed inline instead of a library call.
11374 @itemx -mno-fast-fix
11376 @opindex mno-fast-fix
11377 The C3x/C4x FIX instruction to convert a floating point value to an
11378 integer value chooses the nearest integer less than or equal to the
11379 floating point value rather than to the nearest integer. Thus if the
11380 floating point number is negative, the result will be incorrectly
11381 truncated an additional code is necessary to detect and correct this
11382 case. This option can be used to disable generation of the additional
11383 code required to correct the result.
11389 Enable (disable) generation of repeat block sequences using the RPTB
11390 instruction for zero overhead looping. The RPTB construct is only used
11391 for innermost loops that do not call functions or jump across the loop
11392 boundaries. There is no advantage having nested RPTB loops due to the
11393 overhead required to save and restore the RC, RS, and RE registers.
11394 This is enabled by default with @option{-O2}.
11396 @item -mrpts=@var{count}
11400 Enable (disable) the use of the single instruction repeat instruction
11401 RPTS@. If a repeat block contains a single instruction, and the loop
11402 count can be guaranteed to be less than the value @var{count}, GCC will
11403 emit a RPTS instruction instead of a RPTB@. If no value is specified,
11404 then a RPTS will be emitted even if the loop count cannot be determined
11405 at compile time. Note that the repeated instruction following RPTS does
11406 not have to be reloaded from memory each iteration, thus freeing up the
11407 CPU buses for operands. However, since interrupts are blocked by this
11408 instruction, it is disabled by default.
11410 @item -mloop-unsigned
11411 @itemx -mno-loop-unsigned
11412 @opindex mloop-unsigned
11413 @opindex mno-loop-unsigned
11414 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
11415 is @math{2^{31} + 1} since these instructions test if the iteration count is
11416 negative to terminate the loop. If the iteration count is unsigned
11417 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
11418 exceeded. This switch allows an unsigned iteration count.
11422 Try to emit an assembler syntax that the TI assembler (asm30) is happy
11423 with. This also enforces compatibility with the API employed by the TI
11424 C3x C compiler. For example, long doubles are passed as structures
11425 rather than in floating point registers.
11431 Generate code that uses registers (stack) for passing arguments to functions.
11432 By default, arguments are passed in registers where possible rather
11433 than by pushing arguments on to the stack.
11435 @item -mparallel-insns
11436 @itemx -mno-parallel-insns
11437 @opindex mparallel-insns
11438 @opindex mno-parallel-insns
11439 Allow the generation of parallel instructions. This is enabled by
11440 default with @option{-O2}.
11442 @item -mparallel-mpy
11443 @itemx -mno-parallel-mpy
11444 @opindex mparallel-mpy
11445 @opindex mno-parallel-mpy
11446 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
11447 provided @option{-mparallel-insns} is also specified. These instructions have
11448 tight register constraints which can pessimize the code generation
11449 of large functions.
11454 @subsection V850 Options
11455 @cindex V850 Options
11457 These @samp{-m} options are defined for V850 implementations:
11461 @itemx -mno-long-calls
11462 @opindex mlong-calls
11463 @opindex mno-long-calls
11464 Treat all calls as being far away (near). If calls are assumed to be
11465 far away, the compiler will always load the functions address up into a
11466 register, and call indirect through the pointer.
11472 Do not optimize (do optimize) basic blocks that use the same index
11473 pointer 4 or more times to copy pointer into the @code{ep} register, and
11474 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
11475 option is on by default if you optimize.
11477 @item -mno-prolog-function
11478 @itemx -mprolog-function
11479 @opindex mno-prolog-function
11480 @opindex mprolog-function
11481 Do not use (do use) external functions to save and restore registers
11482 at the prologue and epilogue of a function. The external functions
11483 are slower, but use less code space if more than one function saves
11484 the same number of registers. The @option{-mprolog-function} option
11485 is on by default if you optimize.
11489 Try to make the code as small as possible. At present, this just turns
11490 on the @option{-mep} and @option{-mprolog-function} options.
11492 @item -mtda=@var{n}
11494 Put static or global variables whose size is @var{n} bytes or less into
11495 the tiny data area that register @code{ep} points to. The tiny data
11496 area can hold up to 256 bytes in total (128 bytes for byte references).
11498 @item -msda=@var{n}
11500 Put static or global variables whose size is @var{n} bytes or less into
11501 the small data area that register @code{gp} points to. The small data
11502 area can hold up to 64 kilobytes.
11504 @item -mzda=@var{n}
11506 Put static or global variables whose size is @var{n} bytes or less into
11507 the first 32 kilobytes of memory.
11511 Specify that the target processor is the V850.
11514 @opindex mbig-switch
11515 Generate code suitable for big switch tables. Use this option only if
11516 the assembler/linker complain about out of range branches within a switch
11521 This option will cause r2 and r5 to be used in the code generated by
11522 the compiler. This setting is the default.
11524 @item -mno-app-regs
11525 @opindex mno-app-regs
11526 This option will cause r2 and r5 to be treated as fixed registers.
11530 Specify that the target processor is the V850E1. The preprocessor
11531 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
11532 this option is used.
11536 Specify that the target processor is the V850E. The preprocessor
11537 constant @samp{__v850e__} will be defined if this option is used.
11539 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
11540 are defined then a default target processor will be chosen and the
11541 relevant @samp{__v850*__} preprocessor constant will be defined.
11543 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
11544 defined, regardless of which processor variant is the target.
11546 @item -mdisable-callt
11547 @opindex mdisable-callt
11548 This option will suppress generation of the CALLT instruction for the
11549 v850e and v850e1 flavors of the v850 architecture. The default is
11550 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
11555 @subsection VAX Options
11556 @cindex VAX options
11558 These @samp{-m} options are defined for the VAX:
11563 Do not output certain jump instructions (@code{aobleq} and so on)
11564 that the Unix assembler for the VAX cannot handle across long
11569 Do output those jump instructions, on the assumption that you
11570 will assemble with the GNU assembler.
11574 Output code for g-format floating point numbers instead of d-format.
11577 @node x86-64 Options
11578 @subsection x86-64 Options
11579 @cindex x86-64 options
11581 These are listed under @xref{i386 and x86-64 Options}.
11583 @node Xstormy16 Options
11584 @subsection Xstormy16 Options
11585 @cindex Xstormy16 Options
11587 These options are defined for Xstormy16:
11592 Choose startup files and linker script suitable for the simulator.
11595 @node Xtensa Options
11596 @subsection Xtensa Options
11597 @cindex Xtensa Options
11599 These options are supported for Xtensa targets:
11603 @itemx -mno-const16
11605 @opindex mno-const16
11606 Enable or disable use of @code{CONST16} instructions for loading
11607 constant values. The @code{CONST16} instruction is currently not a
11608 standard option from Tensilica. When enabled, @code{CONST16}
11609 instructions are always used in place of the standard @code{L32R}
11610 instructions. The use of @code{CONST16} is enabled by default only if
11611 the @code{L32R} instruction is not available.
11614 @itemx -mno-fused-madd
11615 @opindex mfused-madd
11616 @opindex mno-fused-madd
11617 Enable or disable use of fused multiply/add and multiply/subtract
11618 instructions in the floating-point option. This has no effect if the
11619 floating-point option is not also enabled. Disabling fused multiply/add
11620 and multiply/subtract instructions forces the compiler to use separate
11621 instructions for the multiply and add/subtract operations. This may be
11622 desirable in some cases where strict IEEE 754-compliant results are
11623 required: the fused multiply add/subtract instructions do not round the
11624 intermediate result, thereby producing results with @emph{more} bits of
11625 precision than specified by the IEEE standard. Disabling fused multiply
11626 add/subtract instructions also ensures that the program output is not
11627 sensitive to the compiler's ability to combine multiply and add/subtract
11630 @item -mtext-section-literals
11631 @itemx -mno-text-section-literals
11632 @opindex mtext-section-literals
11633 @opindex mno-text-section-literals
11634 Control the treatment of literal pools. The default is
11635 @option{-mno-text-section-literals}, which places literals in a separate
11636 section in the output file. This allows the literal pool to be placed
11637 in a data RAM/ROM, and it also allows the linker to combine literal
11638 pools from separate object files to remove redundant literals and
11639 improve code size. With @option{-mtext-section-literals}, the literals
11640 are interspersed in the text section in order to keep them as close as
11641 possible to their references. This may be necessary for large assembly
11644 @item -mtarget-align
11645 @itemx -mno-target-align
11646 @opindex mtarget-align
11647 @opindex mno-target-align
11648 When this option is enabled, GCC instructs the assembler to
11649 automatically align instructions to reduce branch penalties at the
11650 expense of some code density. The assembler attempts to widen density
11651 instructions to align branch targets and the instructions following call
11652 instructions. If there are not enough preceding safe density
11653 instructions to align a target, no widening will be performed. The
11654 default is @option{-mtarget-align}. These options do not affect the
11655 treatment of auto-aligned instructions like @code{LOOP}, which the
11656 assembler will always align, either by widening density instructions or
11657 by inserting no-op instructions.
11660 @itemx -mno-longcalls
11661 @opindex mlongcalls
11662 @opindex mno-longcalls
11663 When this option is enabled, GCC instructs the assembler to translate
11664 direct calls to indirect calls unless it can determine that the target
11665 of a direct call is in the range allowed by the call instruction. This
11666 translation typically occurs for calls to functions in other source
11667 files. Specifically, the assembler translates a direct @code{CALL}
11668 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11669 The default is @option{-mno-longcalls}. This option should be used in
11670 programs where the call target can potentially be out of range. This
11671 option is implemented in the assembler, not the compiler, so the
11672 assembly code generated by GCC will still show direct call
11673 instructions---look at the disassembled object code to see the actual
11674 instructions. Note that the assembler will use an indirect call for
11675 every cross-file call, not just those that really will be out of range.
11678 @node zSeries Options
11679 @subsection zSeries Options
11680 @cindex zSeries options
11682 These are listed under @xref{S/390 and zSeries Options}.
11684 @node Code Gen Options
11685 @section Options for Code Generation Conventions
11686 @cindex code generation conventions
11687 @cindex options, code generation
11688 @cindex run-time options
11690 These machine-independent options control the interface conventions
11691 used in code generation.
11693 Most of them have both positive and negative forms; the negative form
11694 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11695 one of the forms is listed---the one which is not the default. You
11696 can figure out the other form by either removing @samp{no-} or adding
11700 @item -fbounds-check
11701 @opindex fbounds-check
11702 For front-ends that support it, generate additional code to check that
11703 indices used to access arrays are within the declared range. This is
11704 currently only supported by the Java and Fortran 77 front-ends, where
11705 this option defaults to true and false respectively.
11709 This option generates traps for signed overflow on addition, subtraction,
11710 multiplication operations.
11714 This option instructs the compiler to assume that signed arithmetic
11715 overflow of addition, subtraction and multiplication wraps around
11716 using twos-complement representation. This flag enables some optimizations
11717 and disables other. This option is enabled by default for the Java
11718 front-end, as required by the Java language specification.
11721 @opindex fexceptions
11722 Enable exception handling. Generates extra code needed to propagate
11723 exceptions. For some targets, this implies GCC will generate frame
11724 unwind information for all functions, which can produce significant data
11725 size overhead, although it does not affect execution. If you do not
11726 specify this option, GCC will enable it by default for languages like
11727 C++ which normally require exception handling, and disable it for
11728 languages like C that do not normally require it. However, you may need
11729 to enable this option when compiling C code that needs to interoperate
11730 properly with exception handlers written in C++. You may also wish to
11731 disable this option if you are compiling older C++ programs that don't
11732 use exception handling.
11734 @item -fnon-call-exceptions
11735 @opindex fnon-call-exceptions
11736 Generate code that allows trapping instructions to throw exceptions.
11737 Note that this requires platform-specific runtime support that does
11738 not exist everywhere. Moreover, it only allows @emph{trapping}
11739 instructions to throw exceptions, i.e.@: memory references or floating
11740 point instructions. It does not allow exceptions to be thrown from
11741 arbitrary signal handlers such as @code{SIGALRM}.
11743 @item -funwind-tables
11744 @opindex funwind-tables
11745 Similar to @option{-fexceptions}, except that it will just generate any needed
11746 static data, but will not affect the generated code in any other way.
11747 You will normally not enable this option; instead, a language processor
11748 that needs this handling would enable it on your behalf.
11750 @item -fasynchronous-unwind-tables
11751 @opindex fasynchronous-unwind-tables
11752 Generate unwind table in dwarf2 format, if supported by target machine. The
11753 table is exact at each instruction boundary, so it can be used for stack
11754 unwinding from asynchronous events (such as debugger or garbage collector).
11756 @item -fpcc-struct-return
11757 @opindex fpcc-struct-return
11758 Return ``short'' @code{struct} and @code{union} values in memory like
11759 longer ones, rather than in registers. This convention is less
11760 efficient, but it has the advantage of allowing intercallability between
11761 GCC-compiled files and files compiled with other compilers, particularly
11762 the Portable C Compiler (pcc).
11764 The precise convention for returning structures in memory depends
11765 on the target configuration macros.
11767 Short structures and unions are those whose size and alignment match
11768 that of some integer type.
11770 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11771 switch is not binary compatible with code compiled with the
11772 @option{-freg-struct-return} switch.
11773 Use it to conform to a non-default application binary interface.
11775 @item -freg-struct-return
11776 @opindex freg-struct-return
11777 Return @code{struct} and @code{union} values in registers when possible.
11778 This is more efficient for small structures than
11779 @option{-fpcc-struct-return}.
11781 If you specify neither @option{-fpcc-struct-return} nor
11782 @option{-freg-struct-return}, GCC defaults to whichever convention is
11783 standard for the target. If there is no standard convention, GCC
11784 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11785 the principal compiler. In those cases, we can choose the standard, and
11786 we chose the more efficient register return alternative.
11788 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11789 switch is not binary compatible with code compiled with the
11790 @option{-fpcc-struct-return} switch.
11791 Use it to conform to a non-default application binary interface.
11793 @item -fshort-enums
11794 @opindex fshort-enums
11795 Allocate to an @code{enum} type only as many bytes as it needs for the
11796 declared range of possible values. Specifically, the @code{enum} type
11797 will be equivalent to the smallest integer type which has enough room.
11799 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11800 code that is not binary compatible with code generated without that switch.
11801 Use it to conform to a non-default application binary interface.
11803 @item -fshort-double
11804 @opindex fshort-double
11805 Use the same size for @code{double} as for @code{float}.
11807 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11808 code that is not binary compatible with code generated without that switch.
11809 Use it to conform to a non-default application binary interface.
11811 @item -fshort-wchar
11812 @opindex fshort-wchar
11813 Override the underlying type for @samp{wchar_t} to be @samp{short
11814 unsigned int} instead of the default for the target. This option is
11815 useful for building programs to run under WINE@.
11817 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11818 code that is not binary compatible with code generated without that switch.
11819 Use it to conform to a non-default application binary interface.
11821 @item -fshared-data
11822 @opindex fshared-data
11823 Requests that the data and non-@code{const} variables of this
11824 compilation be shared data rather than private data. The distinction
11825 makes sense only on certain operating systems, where shared data is
11826 shared between processes running the same program, while private data
11827 exists in one copy per process.
11830 @opindex fno-common
11831 In C, allocate even uninitialized global variables in the data section of the
11832 object file, rather than generating them as common blocks. This has the
11833 effect that if the same variable is declared (without @code{extern}) in
11834 two different compilations, you will get an error when you link them.
11835 The only reason this might be useful is if you wish to verify that the
11836 program will work on other systems which always work this way.
11840 Ignore the @samp{#ident} directive.
11842 @item -finhibit-size-directive
11843 @opindex finhibit-size-directive
11844 Don't output a @code{.size} assembler directive, or anything else that
11845 would cause trouble if the function is split in the middle, and the
11846 two halves are placed at locations far apart in memory. This option is
11847 used when compiling @file{crtstuff.c}; you should not need to use it
11850 @item -fverbose-asm
11851 @opindex fverbose-asm
11852 Put extra commentary information in the generated assembly code to
11853 make it more readable. This option is generally only of use to those
11854 who actually need to read the generated assembly code (perhaps while
11855 debugging the compiler itself).
11857 @option{-fno-verbose-asm}, the default, causes the
11858 extra information to be omitted and is useful when comparing two assembler
11863 @cindex global offset table
11865 Generate position-independent code (PIC) suitable for use in a shared
11866 library, if supported for the target machine. Such code accesses all
11867 constant addresses through a global offset table (GOT)@. The dynamic
11868 loader resolves the GOT entries when the program starts (the dynamic
11869 loader is not part of GCC; it is part of the operating system). If
11870 the GOT size for the linked executable exceeds a machine-specific
11871 maximum size, you get an error message from the linker indicating that
11872 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11873 instead. (These maximums are 8k on the SPARC and 32k
11874 on the m68k and RS/6000. The 386 has no such limit.)
11876 Position-independent code requires special support, and therefore works
11877 only on certain machines. For the 386, GCC supports PIC for System V
11878 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11879 position-independent.
11883 If supported for the target machine, emit position-independent code,
11884 suitable for dynamic linking and avoiding any limit on the size of the
11885 global offset table. This option makes a difference on the m68k
11888 Position-independent code requires special support, and therefore works
11889 only on certain machines.
11895 These options are similar to @option{-fpic} and @option{-fPIC}, but
11896 generated position independent code can be only linked into executables.
11897 Usually these options are used when @option{-pie} GCC option will be
11898 used during linking.
11900 @item -ffixed-@var{reg}
11902 Treat the register named @var{reg} as a fixed register; generated code
11903 should never refer to it (except perhaps as a stack pointer, frame
11904 pointer or in some other fixed role).
11906 @var{reg} must be the name of a register. The register names accepted
11907 are machine-specific and are defined in the @code{REGISTER_NAMES}
11908 macro in the machine description macro file.
11910 This flag does not have a negative form, because it specifies a
11913 @item -fcall-used-@var{reg}
11914 @opindex fcall-used
11915 Treat the register named @var{reg} as an allocable register that is
11916 clobbered by function calls. It may be allocated for temporaries or
11917 variables that do not live across a call. Functions compiled this way
11918 will not save and restore the register @var{reg}.
11920 It is an error to used this flag with the frame pointer or stack pointer.
11921 Use of this flag for other registers that have fixed pervasive roles in
11922 the machine's execution model will produce disastrous results.
11924 This flag does not have a negative form, because it specifies a
11927 @item -fcall-saved-@var{reg}
11928 @opindex fcall-saved
11929 Treat the register named @var{reg} as an allocable register saved by
11930 functions. It may be allocated even for temporaries or variables that
11931 live across a call. Functions compiled this way will save and restore
11932 the register @var{reg} if they use it.
11934 It is an error to used this flag with the frame pointer or stack pointer.
11935 Use of this flag for other registers that have fixed pervasive roles in
11936 the machine's execution model will produce disastrous results.
11938 A different sort of disaster will result from the use of this flag for
11939 a register in which function values may be returned.
11941 This flag does not have a negative form, because it specifies a
11944 @item -fpack-struct[=@var{n}]
11945 @opindex fpack-struct
11946 Without a value specified, pack all structure members together without
11947 holes. When a value is specified (which must be a small power of two), pack
11948 structure members according to this value, representing the maximum
11949 alignment (that is, objects with default alignment requirements larger than
11950 this will be output potentially unaligned at the next fitting location.
11952 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11953 code that is not binary compatible with code generated without that switch.
11954 Additionally, it makes the code suboptimal.
11955 Use it to conform to a non-default application binary interface.
11957 @item -finstrument-functions
11958 @opindex finstrument-functions
11959 Generate instrumentation calls for entry and exit to functions. Just
11960 after function entry and just before function exit, the following
11961 profiling functions will be called with the address of the current
11962 function and its call site. (On some platforms,
11963 @code{__builtin_return_address} does not work beyond the current
11964 function, so the call site information may not be available to the
11965 profiling functions otherwise.)
11968 void __cyg_profile_func_enter (void *this_fn,
11970 void __cyg_profile_func_exit (void *this_fn,
11974 The first argument is the address of the start of the current function,
11975 which may be looked up exactly in the symbol table.
11977 This instrumentation is also done for functions expanded inline in other
11978 functions. The profiling calls will indicate where, conceptually, the
11979 inline function is entered and exited. This means that addressable
11980 versions of such functions must be available. If all your uses of a
11981 function are expanded inline, this may mean an additional expansion of
11982 code size. If you use @samp{extern inline} in your C code, an
11983 addressable version of such functions must be provided. (This is
11984 normally the case anyways, but if you get lucky and the optimizer always
11985 expands the functions inline, you might have gotten away without
11986 providing static copies.)
11988 A function may be given the attribute @code{no_instrument_function}, in
11989 which case this instrumentation will not be done. This can be used, for
11990 example, for the profiling functions listed above, high-priority
11991 interrupt routines, and any functions from which the profiling functions
11992 cannot safely be called (perhaps signal handlers, if the profiling
11993 routines generate output or allocate memory).
11995 @item -fstack-check
11996 @opindex fstack-check
11997 Generate code to verify that you do not go beyond the boundary of the
11998 stack. You should specify this flag if you are running in an
11999 environment with multiple threads, but only rarely need to specify it in
12000 a single-threaded environment since stack overflow is automatically
12001 detected on nearly all systems if there is only one stack.
12003 Note that this switch does not actually cause checking to be done; the
12004 operating system must do that. The switch causes generation of code
12005 to ensure that the operating system sees the stack being extended.
12007 @item -fstack-limit-register=@var{reg}
12008 @itemx -fstack-limit-symbol=@var{sym}
12009 @itemx -fno-stack-limit
12010 @opindex fstack-limit-register
12011 @opindex fstack-limit-symbol
12012 @opindex fno-stack-limit
12013 Generate code to ensure that the stack does not grow beyond a certain value,
12014 either the value of a register or the address of a symbol. If the stack
12015 would grow beyond the value, a signal is raised. For most targets,
12016 the signal is raised before the stack overruns the boundary, so
12017 it is possible to catch the signal without taking special precautions.
12019 For instance, if the stack starts at absolute address @samp{0x80000000}
12020 and grows downwards, you can use the flags
12021 @option{-fstack-limit-symbol=__stack_limit} and
12022 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
12023 of 128KB@. Note that this may only work with the GNU linker.
12025 @cindex aliasing of parameters
12026 @cindex parameters, aliased
12027 @item -fargument-alias
12028 @itemx -fargument-noalias
12029 @itemx -fargument-noalias-global
12030 @opindex fargument-alias
12031 @opindex fargument-noalias
12032 @opindex fargument-noalias-global
12033 Specify the possible relationships among parameters and between
12034 parameters and global data.
12036 @option{-fargument-alias} specifies that arguments (parameters) may
12037 alias each other and may alias global storage.@*
12038 @option{-fargument-noalias} specifies that arguments do not alias
12039 each other, but may alias global storage.@*
12040 @option{-fargument-noalias-global} specifies that arguments do not
12041 alias each other and do not alias global storage.
12043 Each language will automatically use whatever option is required by
12044 the language standard. You should not need to use these options yourself.
12046 @item -fleading-underscore
12047 @opindex fleading-underscore
12048 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
12049 change the way C symbols are represented in the object file. One use
12050 is to help link with legacy assembly code.
12052 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
12053 generate code that is not binary compatible with code generated without that
12054 switch. Use it to conform to a non-default application binary interface.
12055 Not all targets provide complete support for this switch.
12057 @item -ftls-model=@var{model}
12058 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
12059 The @var{model} argument should be one of @code{global-dynamic},
12060 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
12062 The default without @option{-fpic} is @code{initial-exec}; with
12063 @option{-fpic} the default is @code{global-dynamic}.
12065 @item -fvisibility=@var{default|internal|hidden|protected}
12066 @opindex fvisibility
12067 Set the default ELF image symbol visibility to the specified option - all
12068 symbols will be marked with this unless overrided within the code.
12069 Using this feature can very substantially improve linking and
12070 load times of shared object libraries, produce more optimised
12071 code, provide near-perfect API export and prevent symbol clashes.
12072 It is @strong{strongly} recommended that you use this in any shared objects
12075 Despite the nomenclature, @code{default} always means public ie;
12076 available to be linked against from outside the shared object.
12077 @code{protected} and @code{internal} are pretty useless in real-world
12078 usage so the only other commonly used option will be @code{hidden}.
12079 The default if -fvisibility isn't specified is @code{default} ie; make every
12080 symbol public - this causes the same behaviour as previous versions of
12083 A good explanation of the benefits offered by ensuring ELF
12084 symbols have the correct visibility is given by ``How To Write
12085 Shared Libraries'' by Ulrich Drepper (which can be found at
12086 @w{@uref{http://people.redhat.com/~drepper/}}) - however a superior
12087 solution made possible by this option to marking things hidden when
12088 the default is public is to make the default hidden and mark things
12089 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
12090 and @code{__attribute__ ((visibility("default")))} instead of
12091 @code{__declspec(dllexport)} you get almost identical semantics with
12092 identical syntax. This is a great boon to those working with
12093 cross-platform projects.
12095 For those adding visibility support to existing code, you may find
12096 @samp{#pragma GCC visibility} of use. This works by you enclosing
12097 the declarations you wish to set visibility for with (for example)
12098 @samp{#pragma GCC visibility push(hidden)} and
12099 @samp{#pragma GCC visibility pop}. These can be nested up to sixteen
12100 times. Bear in mind that symbol visibility should be viewed @strong{as
12101 part of the API interface contract} and thus all new code should
12102 always specify visibility when it is not the default ie; declarations
12103 only for use within the local DSO should @strong{always} be marked explicitly
12104 as hidden as so to avoid PLT indirection overheads - making this
12105 abundantly clear also aids readability and self-documentation of the code.
12106 Note that due to ISO C++ specification requirements, operator new and
12107 operator delete must always be of default visibility.
12109 An overview of these techniques, their benefits and how to use them
12110 is at @w{@uref{http://www.nedprod.com/programs/gccvisibility.html}}.
12116 @node Environment Variables
12117 @section Environment Variables Affecting GCC
12118 @cindex environment variables
12120 @c man begin ENVIRONMENT
12121 This section describes several environment variables that affect how GCC
12122 operates. Some of them work by specifying directories or prefixes to use
12123 when searching for various kinds of files. Some are used to specify other
12124 aspects of the compilation environment.
12126 Note that you can also specify places to search using options such as
12127 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
12128 take precedence over places specified using environment variables, which
12129 in turn take precedence over those specified by the configuration of GCC@.
12130 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
12131 GNU Compiler Collection (GCC) Internals}.
12136 @c @itemx LC_COLLATE
12138 @c @itemx LC_MONETARY
12139 @c @itemx LC_NUMERIC
12144 @c @findex LC_COLLATE
12145 @findex LC_MESSAGES
12146 @c @findex LC_MONETARY
12147 @c @findex LC_NUMERIC
12151 These environment variables control the way that GCC uses
12152 localization information that allow GCC to work with different
12153 national conventions. GCC inspects the locale categories
12154 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
12155 so. These locale categories can be set to any value supported by your
12156 installation. A typical value is @samp{en_UK} for English in the United
12159 The @env{LC_CTYPE} environment variable specifies character
12160 classification. GCC uses it to determine the character boundaries in
12161 a string; this is needed for some multibyte encodings that contain quote
12162 and escape characters that would otherwise be interpreted as a string
12165 The @env{LC_MESSAGES} environment variable specifies the language to
12166 use in diagnostic messages.
12168 If the @env{LC_ALL} environment variable is set, it overrides the value
12169 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
12170 and @env{LC_MESSAGES} default to the value of the @env{LANG}
12171 environment variable. If none of these variables are set, GCC
12172 defaults to traditional C English behavior.
12176 If @env{TMPDIR} is set, it specifies the directory to use for temporary
12177 files. GCC uses temporary files to hold the output of one stage of
12178 compilation which is to be used as input to the next stage: for example,
12179 the output of the preprocessor, which is the input to the compiler
12182 @item GCC_EXEC_PREFIX
12183 @findex GCC_EXEC_PREFIX
12184 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
12185 names of the subprograms executed by the compiler. No slash is added
12186 when this prefix is combined with the name of a subprogram, but you can
12187 specify a prefix that ends with a slash if you wish.
12189 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
12190 an appropriate prefix to use based on the pathname it was invoked with.
12192 If GCC cannot find the subprogram using the specified prefix, it
12193 tries looking in the usual places for the subprogram.
12195 The default value of @env{GCC_EXEC_PREFIX} is
12196 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
12197 of @code{prefix} when you ran the @file{configure} script.
12199 Other prefixes specified with @option{-B} take precedence over this prefix.
12201 This prefix is also used for finding files such as @file{crt0.o} that are
12204 In addition, the prefix is used in an unusual way in finding the
12205 directories to search for header files. For each of the standard
12206 directories whose name normally begins with @samp{/usr/local/lib/gcc}
12207 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
12208 replacing that beginning with the specified prefix to produce an
12209 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
12210 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
12211 These alternate directories are searched first; the standard directories
12214 @item COMPILER_PATH
12215 @findex COMPILER_PATH
12216 The value of @env{COMPILER_PATH} is a colon-separated list of
12217 directories, much like @env{PATH}. GCC tries the directories thus
12218 specified when searching for subprograms, if it can't find the
12219 subprograms using @env{GCC_EXEC_PREFIX}.
12222 @findex LIBRARY_PATH
12223 The value of @env{LIBRARY_PATH} is a colon-separated list of
12224 directories, much like @env{PATH}. When configured as a native compiler,
12225 GCC tries the directories thus specified when searching for special
12226 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
12227 using GCC also uses these directories when searching for ordinary
12228 libraries for the @option{-l} option (but directories specified with
12229 @option{-L} come first).
12233 @cindex locale definition
12234 This variable is used to pass locale information to the compiler. One way in
12235 which this information is used is to determine the character set to be used
12236 when character literals, string literals and comments are parsed in C and C++.
12237 When the compiler is configured to allow multibyte characters,
12238 the following values for @env{LANG} are recognized:
12242 Recognize JIS characters.
12244 Recognize SJIS characters.
12246 Recognize EUCJP characters.
12249 If @env{LANG} is not defined, or if it has some other value, then the
12250 compiler will use mblen and mbtowc as defined by the default locale to
12251 recognize and translate multibyte characters.
12255 Some additional environments variables affect the behavior of the
12258 @include cppenv.texi
12262 @node Precompiled Headers
12263 @section Using Precompiled Headers
12264 @cindex precompiled headers
12265 @cindex speed of compilation
12267 Often large projects have many header files that are included in every
12268 source file. The time the compiler takes to process these header files
12269 over and over again can account for nearly all of the time required to
12270 build the project. To make builds faster, GCC allows users to
12271 `precompile' a header file; then, if builds can use the precompiled
12272 header file they will be much faster.
12274 @strong{Caution:} There are a few known situations where GCC will
12275 crash when trying to use a precompiled header. If you have trouble
12276 with a precompiled header, you should remove the precompiled header
12277 and compile without it. In addition, please use GCC's on-line
12278 defect-tracking system to report any problems you encounter with
12279 precompiled headers. @xref{Bugs}.
12281 To create a precompiled header file, simply compile it as you would any
12282 other file, if necessary using the @option{-x} option to make the driver
12283 treat it as a C or C++ header file. You will probably want to use a
12284 tool like @command{make} to keep the precompiled header up-to-date when
12285 the headers it contains change.
12287 A precompiled header file will be searched for when @code{#include} is
12288 seen in the compilation. As it searches for the included file
12289 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
12290 compiler looks for a precompiled header in each directory just before it
12291 looks for the include file in that directory. The name searched for is
12292 the name specified in the @code{#include} with @samp{.gch} appended. If
12293 the precompiled header file can't be used, it is ignored.
12295 For instance, if you have @code{#include "all.h"}, and you have
12296 @file{all.h.gch} in the same directory as @file{all.h}, then the
12297 precompiled header file will be used if possible, and the original
12298 header will be used otherwise.
12300 Alternatively, you might decide to put the precompiled header file in a
12301 directory and use @option{-I} to ensure that directory is searched
12302 before (or instead of) the directory containing the original header.
12303 Then, if you want to check that the precompiled header file is always
12304 used, you can put a file of the same name as the original header in this
12305 directory containing an @code{#error} command.
12307 This also works with @option{-include}. So yet another way to use
12308 precompiled headers, good for projects not designed with precompiled
12309 header files in mind, is to simply take most of the header files used by
12310 a project, include them from another header file, precompile that header
12311 file, and @option{-include} the precompiled header. If the header files
12312 have guards against multiple inclusion, they will be skipped because
12313 they've already been included (in the precompiled header).
12315 If you need to precompile the same header file for different
12316 languages, targets, or compiler options, you can instead make a
12317 @emph{directory} named like @file{all.h.gch}, and put each precompiled
12318 header in the directory, perhaps using @option{-o}. It doesn't matter
12319 what you call the files in the directory, every precompiled header in
12320 the directory will be considered. The first precompiled header
12321 encountered in the directory that is valid for this compilation will
12322 be used; they're searched in no particular order.
12324 There are many other possibilities, limited only by your imagination,
12325 good sense, and the constraints of your build system.
12327 A precompiled header file can be used only when these conditions apply:
12331 Only one precompiled header can be used in a particular compilation.
12334 A precompiled header can't be used once the first C token is seen. You
12335 can have preprocessor directives before a precompiled header; you can
12336 even include a precompiled header from inside another header, so long as
12337 there are no C tokens before the @code{#include}.
12340 The precompiled header file must be produced for the same language as
12341 the current compilation. You can't use a C precompiled header for a C++
12345 The precompiled header file must be produced by the same compiler
12346 version and configuration as the current compilation is using.
12347 The easiest way to guarantee this is to use the same compiler binary
12348 for creating and using precompiled headers.
12351 Any macros defined before the precompiled header is included must
12352 either be defined in the same way as when the precompiled header was
12353 generated, or must not affect the precompiled header, which usually
12354 means that the they don't appear in the precompiled header at all.
12356 The @option{-D} option is one way to define a macro before a
12357 precompiled header is included; using a @code{#define} can also do it.
12358 There are also some options that define macros implicitly, like
12359 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
12362 @item If debugging information is output when using the precompiled
12363 header, using @option{-g} or similar, the same kind of debugging information
12364 must have been output when building the precompiled header. However,
12365 a precompiled header built using @option{-g} can be used in a compilation
12366 when no debugging information is being output.
12368 @item The same @option{-m} options must generally be used when building
12369 and using the precompiled header. @xref{Submodel Options},
12370 for any cases where this rule is relaxed.
12372 @item Each of the following options must be the same when building and using
12373 the precompiled header:
12375 @gccoptlist{-fexceptions -funit-at-a-time}
12378 Some other command-line options starting with @option{-f},
12379 @option{-p}, or @option{-O} must be defined in the same way as when
12380 the precompiled header was generated. At present, it's not clear
12381 which options are safe to change and which are not; the safest choice
12382 is to use exactly the same options when generating and using the
12383 precompiled header. The following are known to be safe:
12385 @gccoptlist{-fpreprocessed -pedantic-errors}
12389 For all of these except the last, the compiler will automatically
12390 ignore the precompiled header if the conditions aren't met. If you
12391 find an option combination that doesn't work and doesn't cause the
12392 precompiled header to be ignored, please consider filing a bug report,
12395 If you do use differing options when generating and using the
12396 precompiled header, the actual behaviour will be a mixture of the
12397 behaviour for the options. For instance, if you use @option{-g} to
12398 generate the precompiled header but not when using it, you may or may
12399 not get debugging information for routines in the precompiled header.
12401 @node Running Protoize
12402 @section Running Protoize
12404 The program @code{protoize} is an optional part of GCC@. You can use
12405 it to add prototypes to a program, thus converting the program to ISO
12406 C in one respect. The companion program @code{unprotoize} does the
12407 reverse: it removes argument types from any prototypes that are found.
12409 When you run these programs, you must specify a set of source files as
12410 command line arguments. The conversion programs start out by compiling
12411 these files to see what functions they define. The information gathered
12412 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
12414 After scanning comes actual conversion. The specified files are all
12415 eligible to be converted; any files they include (whether sources or
12416 just headers) are eligible as well.
12418 But not all the eligible files are converted. By default,
12419 @code{protoize} and @code{unprotoize} convert only source and header
12420 files in the current directory. You can specify additional directories
12421 whose files should be converted with the @option{-d @var{directory}}
12422 option. You can also specify particular files to exclude with the
12423 @option{-x @var{file}} option. A file is converted if it is eligible, its
12424 directory name matches one of the specified directory names, and its
12425 name within the directory has not been excluded.
12427 Basic conversion with @code{protoize} consists of rewriting most
12428 function definitions and function declarations to specify the types of
12429 the arguments. The only ones not rewritten are those for varargs
12432 @code{protoize} optionally inserts prototype declarations at the
12433 beginning of the source file, to make them available for any calls that
12434 precede the function's definition. Or it can insert prototype
12435 declarations with block scope in the blocks where undeclared functions
12438 Basic conversion with @code{unprotoize} consists of rewriting most
12439 function declarations to remove any argument types, and rewriting
12440 function definitions to the old-style pre-ISO form.
12442 Both conversion programs print a warning for any function declaration or
12443 definition that they can't convert. You can suppress these warnings
12446 The output from @code{protoize} or @code{unprotoize} replaces the
12447 original source file. The original file is renamed to a name ending
12448 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
12449 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
12450 for DOS) file already exists, then the source file is simply discarded.
12452 @code{protoize} and @code{unprotoize} both depend on GCC itself to
12453 scan the program and collect information about the functions it uses.
12454 So neither of these programs will work until GCC is installed.
12456 Here is a table of the options you can use with @code{protoize} and
12457 @code{unprotoize}. Each option works with both programs unless
12461 @item -B @var{directory}
12462 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
12463 usual directory (normally @file{/usr/local/lib}). This file contains
12464 prototype information about standard system functions. This option
12465 applies only to @code{protoize}.
12467 @item -c @var{compilation-options}
12468 Use @var{compilation-options} as the options when running @command{gcc} to
12469 produce the @samp{.X} files. The special option @option{-aux-info} is
12470 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
12472 Note that the compilation options must be given as a single argument to
12473 @code{protoize} or @code{unprotoize}. If you want to specify several
12474 @command{gcc} options, you must quote the entire set of compilation options
12475 to make them a single word in the shell.
12477 There are certain @command{gcc} arguments that you cannot use, because they
12478 would produce the wrong kind of output. These include @option{-g},
12479 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
12480 the @var{compilation-options}, they are ignored.
12483 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
12484 systems) instead of @samp{.c}. This is convenient if you are converting
12485 a C program to C++. This option applies only to @code{protoize}.
12488 Add explicit global declarations. This means inserting explicit
12489 declarations at the beginning of each source file for each function
12490 that is called in the file and was not declared. These declarations
12491 precede the first function definition that contains a call to an
12492 undeclared function. This option applies only to @code{protoize}.
12494 @item -i @var{string}
12495 Indent old-style parameter declarations with the string @var{string}.
12496 This option applies only to @code{protoize}.
12498 @code{unprotoize} converts prototyped function definitions to old-style
12499 function definitions, where the arguments are declared between the
12500 argument list and the initial @samp{@{}. By default, @code{unprotoize}
12501 uses five spaces as the indentation. If you want to indent with just
12502 one space instead, use @option{-i " "}.
12505 Keep the @samp{.X} files. Normally, they are deleted after conversion
12509 Add explicit local declarations. @code{protoize} with @option{-l} inserts
12510 a prototype declaration for each function in each block which calls the
12511 function without any declaration. This option applies only to
12515 Make no real changes. This mode just prints information about the conversions
12516 that would have been done without @option{-n}.
12519 Make no @samp{.save} files. The original files are simply deleted.
12520 Use this option with caution.
12522 @item -p @var{program}
12523 Use the program @var{program} as the compiler. Normally, the name
12524 @file{gcc} is used.
12527 Work quietly. Most warnings are suppressed.
12530 Print the version number, just like @option{-v} for @command{gcc}.
12533 If you need special compiler options to compile one of your program's
12534 source files, then you should generate that file's @samp{.X} file
12535 specially, by running @command{gcc} on that source file with the
12536 appropriate options and the option @option{-aux-info}. Then run
12537 @code{protoize} on the entire set of files. @code{protoize} will use
12538 the existing @samp{.X} file because it is newer than the source file.
12542 gcc -Dfoo=bar file1.c -aux-info file1.X
12547 You need to include the special files along with the rest in the
12548 @code{protoize} command, even though their @samp{.X} files already
12549 exist, because otherwise they won't get converted.
12551 @xref{Protoize Caveats}, for more information on how to use
12552 @code{protoize} successfully.