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
252 -fdump-ipa-all -fdump-ipa-cgraph @gol
254 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
255 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
256 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
257 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
259 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
260 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
261 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
262 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
263 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
264 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
265 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
266 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
267 -fdump-tree-nrv -fdump-tree-vect @gol
268 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
269 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
270 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
271 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs -ftree-based-profiling @gol
272 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
273 -ftest-coverage -ftime-report -fvar-tracking @gol
274 -g -g@var{level} -gcoff -gdwarf-2 @gol
275 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
276 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
277 -print-multi-directory -print-multi-lib @gol
278 -print-prog-name=@var{program} -print-search-dirs -Q @gol
281 @item Optimization Options
282 @xref{Optimize Options,,Options that Control Optimization}.
283 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
284 -falign-labels=@var{n} -falign-loops=@var{n} @gol
285 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
286 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
287 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
288 -fcaller-saves -fcprop-registers @gol
289 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
290 -fdelayed-branch -fdelete-null-pointer-checks @gol
291 -fexpensive-optimizations -ffast-math -ffloat-store @gol
292 -fforce-addr -fforce-mem -ffunction-sections @gol
293 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
294 -floop-optimize -fcrossjumping -fif-conversion -fif-conversion2 @gol
295 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
296 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
297 -fmodulo-sched -fnew-ra -fno-branch-count-reg @gol
298 -fno-default-inline -fno-defer-pop -floop-optimize2 -fmove-loop-invariants @gol
299 -fno-function-cse -fno-guess-branch-probability @gol
300 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
301 -funsafe-math-optimizations -ffinite-math-only @gol
302 -fno-trapping-math -fno-zero-initialized-in-bss @gol
303 -fomit-frame-pointer -foptimize-register-move @gol
304 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
305 -fprofile-generate -fprofile-use @gol
306 -fregmove -frename-registers @gol
307 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
308 -frerun-cse-after-loop -frerun-loop-opt @gol
309 -frounding-math -fschedule-insns -fschedule-insns2 @gol
310 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
311 -fsched-spec-load-dangerous @gol
312 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
313 -fsched2-use-superblocks @gol
314 -fsched2-use-traces -freschedule-modulo-scheduled-loops @gol
315 -fsignaling-nans -fsingle-precision-constant -fspeculative-prefetching @gol
316 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
317 -funroll-all-loops -funroll-loops -fpeel-loops @gol
318 -fsplit-ivs-in-unroller -funswitch-loops @gol
319 -fvariable-expansion-in-unroller @gol
320 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
321 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
322 -ftree-dominator-opts -ftree-dse -ftree-copyrename @gol
323 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
324 --param @var{name}=@var{value}
325 -O -O0 -O1 -O2 -O3 -Os}
327 @item Preprocessor Options
328 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
329 @gccoptlist{-A@var{question}=@var{answer} @gol
330 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
331 -C -dD -dI -dM -dN @gol
332 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
333 -idirafter @var{dir} @gol
334 -include @var{file} -imacros @var{file} @gol
335 -iprefix @var{file} -iwithprefix @var{dir} @gol
336 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
337 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
338 -P -fworking-directory -remap @gol
339 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
340 -Xpreprocessor @var{option}}
342 @item Assembler Option
343 @xref{Assembler Options,,Passing Options to the Assembler}.
344 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
347 @xref{Link Options,,Options for Linking}.
348 @gccoptlist{@var{object-file-name} -l@var{library} @gol
349 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
350 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
351 -Wl,@var{option} -Xlinker @var{option} @gol
354 @item Directory Options
355 @xref{Directory Options,,Options for Directory Search}.
356 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-}
359 @c I wrote this xref this way to avoid overfull hbox. -- rms
360 @xref{Target Options}.
361 @gccoptlist{-V @var{version} -b @var{machine}}
363 @item Machine Dependent Options
364 @xref{Submodel Options,,Hardware Models and Configurations}.
365 @c This list is ordered alphanumerically by subsection name.
366 @c Try and put the significant identifier (CPU or system) first,
367 @c so users have a clue at guessing where the ones they want will be.
370 @gccoptlist{-EB -EL @gol
371 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
372 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
375 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
376 -mabi=@var{name} @gol
377 -mapcs-stack-check -mno-apcs-stack-check @gol
378 -mapcs-float -mno-apcs-float @gol
379 -mapcs-reentrant -mno-apcs-reentrant @gol
380 -msched-prolog -mno-sched-prolog @gol
381 -mlittle-endian -mbig-endian -mwords-little-endian @gol
382 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
383 -mthumb-interwork -mno-thumb-interwork @gol
384 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
385 -mstructure-size-boundary=@var{n} @gol
386 -mabort-on-noreturn @gol
387 -mlong-calls -mno-long-calls @gol
388 -msingle-pic-base -mno-single-pic-base @gol
389 -mpic-register=@var{reg} @gol
390 -mnop-fun-dllimport @gol
391 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
392 -mpoke-function-name @gol
394 -mtpcs-frame -mtpcs-leaf-frame @gol
395 -mcaller-super-interworking -mcallee-super-interworking}
398 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
399 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
402 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
403 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
404 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
405 -mstack-align -mdata-align -mconst-align @gol
406 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
407 -melf -maout -melinux -mlinux -sim -sim2 @gol
408 -mmul-bug-workaround -mno-mul-bug-workaround}
410 @emph{Darwin Options}
411 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
412 -arch_only -bind_at_load -bundle -bundle_loader @gol
413 -client_name -compatibility_version -current_version @gol
415 -dependency-file -dylib_file -dylinker_install_name @gol
416 -dynamic -dynamiclib -exported_symbols_list @gol
417 -filelist -flat_namespace -force_cpusubtype_ALL @gol
418 -force_flat_namespace -headerpad_max_install_names @gol
419 -image_base -init -install_name -keep_private_externs @gol
420 -multi_module -multiply_defined -multiply_defined_unused @gol
421 -noall_load -no_dead_strip_inits_and_terms @gol
422 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
423 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
424 -private_bundle -read_only_relocs -sectalign @gol
425 -sectobjectsymbols -whyload -seg1addr @gol
426 -sectcreate -sectobjectsymbols -sectorder @gol
427 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
428 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
429 -segprot -segs_read_only_addr -segs_read_write_addr @gol
430 -single_module -static -sub_library -sub_umbrella @gol
431 -twolevel_namespace -umbrella -undefined @gol
432 -unexported_symbols_list -weak_reference_mismatches @gol
433 -whatsloaded -F -gused -gfull -mone-byte-bool}
435 @emph{DEC Alpha Options}
436 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
437 -mieee -mieee-with-inexact -mieee-conformant @gol
438 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
439 -mtrap-precision=@var{mode} -mbuild-constants @gol
440 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
441 -mbwx -mmax -mfix -mcix @gol
442 -mfloat-vax -mfloat-ieee @gol
443 -mexplicit-relocs -msmall-data -mlarge-data @gol
444 -msmall-text -mlarge-text @gol
445 -mmemory-latency=@var{time}}
447 @emph{DEC Alpha/VMS Options}
448 @gccoptlist{-mvms-return-codes}
451 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
452 -mhard-float -msoft-float @gol
453 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
454 -mdouble -mno-double @gol
455 -mmedia -mno-media -mmuladd -mno-muladd @gol
456 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
457 -mlinked-fp -mlong-calls -malign-labels @gol
458 -mlibrary-pic -macc-4 -macc-8 @gol
459 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
460 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
461 -mvliw-branch -mno-vliw-branch @gol
462 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
463 -mno-nested-cond-exec -mtomcat-stats @gol
466 @emph{H8/300 Options}
467 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
470 @gccoptlist{-march=@var{architecture-type} @gol
471 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
472 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
473 -mfixed-range=@var{register-range} @gol
474 -mjump-in-delay -mlinker-opt -mlong-calls @gol
475 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
476 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
477 -mno-jump-in-delay -mno-long-load-store @gol
478 -mno-portable-runtime -mno-soft-float @gol
479 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
480 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
481 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
482 -munix=@var{unix-std} -nolibdld -static -threads}
484 @emph{i386 and x86-64 Options}
485 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
486 -mfpmath=@var{unit} @gol
487 -masm=@var{dialect} -mno-fancy-math-387 @gol
488 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
489 -mno-wide-multiply -mrtd -malign-double @gol
490 -mpreferred-stack-boundary=@var{num} @gol
491 -mmmx -msse -msse2 -msse3 -m3dnow @gol
492 -mthreads -mno-align-stringops -minline-all-stringops @gol
493 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
494 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
495 -mno-red-zone -mno-tls-direct-seg-refs @gol
496 -mcmodel=@var{code-model} @gol
500 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
501 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
502 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
503 -minline-float-divide-max-throughput @gol
504 -minline-int-divide-min-latency @gol
505 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
506 -mfixed-range=@var{register-range}}
508 @emph{M32R/D Options}
509 @gccoptlist{-m32r2 -m32rx -m32r @gol
511 -malign-loops -mno-align-loops @gol
512 -missue-rate=@var{number} @gol
513 -mbranch-cost=@var{number} @gol
514 -mmodel=@var{code-size-model-type} @gol
515 -msdata=@var{sdata-type} @gol
516 -mno-flush-func -mflush-func=@var{name} @gol
517 -mno-flush-trap -mflush-trap=@var{number} @gol
520 @emph{M680x0 Options}
521 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
522 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
523 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
524 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
525 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
527 @emph{M68hc1x Options}
528 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
529 -mauto-incdec -minmax -mlong-calls -mshort @gol
530 -msoft-reg-count=@var{count}}
533 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
534 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
535 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
536 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
537 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
540 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
541 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
542 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
543 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
544 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
545 -mpaired-single -mips3d @gol
546 -mint64 -mlong64 -mlong32 @gol
547 -G@var{num} -membedded-data -mno-embedded-data @gol
548 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
549 -msplit-addresses -mno-split-addresses @gol
550 -mexplicit-relocs -mno-explicit-relocs @gol
551 -mcheck-zero-division -mno-check-zero-division @gol
552 -mdivide-traps -mdivide-breaks @gol
553 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
554 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
555 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
556 -mfix-vr4120 -mno-fix-vr4120 -mfix-sb1 -mno-fix-sb1 @gol
557 -mflush-func=@var{func} -mno-flush-func @gol
558 -mbranch-likely -mno-branch-likely @gol
559 -mfp-exceptions -mno-fp-exceptions @gol
560 -mvr4130-align -mno-vr4130-align}
563 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
564 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
565 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
566 -mno-base-addresses -msingle-exit -mno-single-exit}
568 @emph{MN10300 Options}
569 @gccoptlist{-mmult-bug -mno-mult-bug @gol
570 -mam33 -mno-am33 @gol
571 -mam33-2 -mno-am33-2 @gol
575 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
576 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
577 -mregparam -mnoregparam -msb -mnosb @gol
578 -mbitfield -mnobitfield -mhimem -mnohimem}
580 @emph{PDP-11 Options}
581 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
582 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
583 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
584 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
585 -mbranch-expensive -mbranch-cheap @gol
586 -msplit -mno-split -munix-asm -mdec-asm}
588 @emph{PowerPC Options}
589 See RS/6000 and PowerPC Options.
591 @emph{RS/6000 and PowerPC Options}
592 @gccoptlist{-mcpu=@var{cpu-type} @gol
593 -mtune=@var{cpu-type} @gol
594 -mpower -mno-power -mpower2 -mno-power2 @gol
595 -mpowerpc -mpowerpc64 -mno-powerpc @gol
596 -maltivec -mno-altivec @gol
597 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
598 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
599 -mnew-mnemonics -mold-mnemonics @gol
600 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
601 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
602 -malign-power -malign-natural @gol
603 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
604 -mstring -mno-string -mupdate -mno-update @gol
605 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
606 -mstrict-align -mno-strict-align -mrelocatable @gol
607 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
608 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
609 -mdynamic-no-pic @gol
610 -mprioritize-restricted-insns=@var{priority} @gol
611 -msched-costly-dep=@var{dependence_type} @gol
612 -minsert-sched-nops=@var{scheme} @gol
613 -mcall-sysv -mcall-netbsd @gol
614 -maix-struct-return -msvr4-struct-return @gol
615 -mabi=altivec -mabi=no-altivec @gol
616 -mabi=spe -mabi=no-spe @gol
617 -misel=yes -misel=no @gol
618 -mspe=yes -mspe=no @gol
619 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
620 -mprototype -mno-prototype @gol
621 -msim -mmvme -mads -myellowknife -memb -msdata @gol
622 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
624 @emph{S/390 and zSeries Options}
625 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
626 -mhard-float -msoft-float -mbackchain -mno-backchain -mkernel-backchain @gol
627 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
628 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
629 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
630 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
633 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
634 -m4-nofpu -m4-single-only -m4-single -m4 @gol
635 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
636 -m5-64media -m5-64media-nofpu @gol
637 -m5-32media -m5-32media-nofpu @gol
638 -m5-compact -m5-compact-nofpu @gol
639 -mb -ml -mdalign -mrelax @gol
640 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
641 -mieee -misize -mpadstruct -mspace @gol
642 -mprefergot -musermode}
645 @gccoptlist{-mcpu=@var{cpu-type} @gol
646 -mtune=@var{cpu-type} @gol
647 -mcmodel=@var{code-model} @gol
648 -m32 -m64 -mapp-regs -mno-app-regs @gol
649 -mfaster-structs -mno-faster-structs @gol
650 -mfpu -mno-fpu -mhard-float -msoft-float @gol
651 -mhard-quad-float -msoft-quad-float @gol
652 -mimpure-text -mno-impure-text -mlittle-endian @gol
653 -mstack-bias -mno-stack-bias @gol
654 -munaligned-doubles -mno-unaligned-doubles @gol
655 -mv8plus -mno-v8plus -mvis -mno-vis
658 @emph{System V Options}
659 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
661 @emph{TMS320C3x/C4x Options}
662 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
663 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
664 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
665 -mparallel-insns -mparallel-mpy -mpreserve-float}
668 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
669 -mprolog-function -mno-prolog-function -mspace @gol
670 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
671 -mapp-regs -mno-app-regs @gol
672 -mdisable-callt -mno-disable-callt @gol
678 @gccoptlist{-mg -mgnu -munix}
680 @emph{x86-64 Options}
681 See i386 and x86-64 Options.
683 @emph{Xstormy16 Options}
686 @emph{Xtensa Options}
687 @gccoptlist{-mconst16 -mno-const16 @gol
688 -mfused-madd -mno-fused-madd @gol
689 -mtext-section-literals -mno-text-section-literals @gol
690 -mtarget-align -mno-target-align @gol
691 -mlongcalls -mno-longcalls}
693 @emph{zSeries Options}
694 See S/390 and zSeries Options.
696 @item Code Generation Options
697 @xref{Code Gen Options,,Options for Code Generation Conventions}.
698 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
699 -ffixed-@var{reg} -fexceptions @gol
700 -fnon-call-exceptions -funwind-tables @gol
701 -fasynchronous-unwind-tables @gol
702 -finhibit-size-directive -finstrument-functions @gol
703 -fno-common -fno-ident @gol
704 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
705 -freg-struct-return -fshared-data -fshort-enums @gol
706 -fshort-double -fshort-wchar @gol
707 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
708 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
709 -fargument-alias -fargument-noalias @gol
710 -fargument-noalias-global -fleading-underscore @gol
711 -ftls-model=@var{model} @gol
712 -ftrapv -fwrapv -fbounds-check @gol
717 * Overall Options:: Controlling the kind of output:
718 an executable, object files, assembler files,
719 or preprocessed source.
720 * C Dialect Options:: Controlling the variant of C language compiled.
721 * C++ Dialect Options:: Variations on C++.
722 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
724 * Language Independent Options:: Controlling how diagnostics should be
726 * Warning Options:: How picky should the compiler be?
727 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
728 * Optimize Options:: How much optimization?
729 * Preprocessor Options:: Controlling header files and macro definitions.
730 Also, getting dependency information for Make.
731 * Assembler Options:: Passing options to the assembler.
732 * Link Options:: Specifying libraries and so on.
733 * Directory Options:: Where to find header files and libraries.
734 Where to find the compiler executable files.
735 * Spec Files:: How to pass switches to sub-processes.
736 * Target Options:: Running a cross-compiler, or an old version of GCC.
739 @node Overall Options
740 @section Options Controlling the Kind of Output
742 Compilation can involve up to four stages: preprocessing, compilation
743 proper, assembly and linking, always in that order. GCC is capable of
744 preprocessing and compiling several files either into several
745 assembler input files, or into one assembler input file; then each
746 assembler input file produces an object file, and linking combines all
747 the object files (those newly compiled, and those specified as input)
748 into an executable file.
750 @cindex file name suffix
751 For any given input file, the file name suffix determines what kind of
756 C source code which must be preprocessed.
759 C source code which should not be preprocessed.
762 C++ source code which should not be preprocessed.
765 Objective-C source code. Note that you must link with the @file{libobjc}
766 library to make an Objective-C program work.
769 Objective-C source code which should not be preprocessed.
773 Objective-C++ source code. Note that you must link with the @file{libobjc}
774 library to make an Objective-C++ program work. Note that @samp{.M} refers
775 to a literal capital M@.
778 Objective-C++ source code which should not be preprocessed.
781 C, C++, Objective-C or Objective-C++ header file to be turned into a
786 @itemx @var{file}.cxx
787 @itemx @var{file}.cpp
788 @itemx @var{file}.CPP
789 @itemx @var{file}.c++
791 C++ source code which must be preprocessed. Note that in @samp{.cxx},
792 the last two letters must both be literally @samp{x}. Likewise,
793 @samp{.C} refers to a literal capital C@.
797 C++ header file to be turned into a precompiled header.
800 @itemx @var{file}.for
801 @itemx @var{file}.FOR
802 Fortran source code which should not be preprocessed.
805 @itemx @var{file}.fpp
806 @itemx @var{file}.FPP
807 Fortran source code which must be preprocessed (with the traditional
811 Fortran source code which must be preprocessed with a RATFOR
812 preprocessor (not included with GCC)@.
815 @itemx @var{file}.f95
816 Fortran 90/95 source code which should not be preprocessed.
818 @c FIXME: Descriptions of Java file types.
825 Ada source code file which contains a library unit declaration (a
826 declaration of a package, subprogram, or generic, or a generic
827 instantiation), or a library unit renaming declaration (a package,
828 generic, or subprogram renaming declaration). Such files are also
831 @itemx @var{file}.adb
832 Ada source code file containing a library unit body (a subprogram or
833 package body). Such files are also called @dfn{bodies}.
835 @c GCC also knows about some suffixes for languages not yet included:
844 Assembler code which must be preprocessed.
847 An object file to be fed straight into linking.
848 Any file name with no recognized suffix is treated this way.
852 You can specify the input language explicitly with the @option{-x} option:
855 @item -x @var{language}
856 Specify explicitly the @var{language} for the following input files
857 (rather than letting the compiler choose a default based on the file
858 name suffix). This option applies to all following input files until
859 the next @option{-x} option. Possible values for @var{language} are:
861 c c-header c-cpp-output
862 c++ c++-header c++-cpp-output
863 objective-c objective-c-header objective-c-cpp-output
864 objective-c++ objective-c++-header objective-c++-cpp-output
865 assembler assembler-with-cpp
867 f77 f77-cpp-input ratfor
874 Turn off any specification of a language, so that subsequent files are
875 handled according to their file name suffixes (as they are if @option{-x}
876 has not been used at all).
878 @item -pass-exit-codes
879 @opindex pass-exit-codes
880 Normally the @command{gcc} program will exit with the code of 1 if any
881 phase of the compiler returns a non-success return code. If you specify
882 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
883 numerically highest error produced by any phase that returned an error
887 If you only want some of the stages of compilation, you can use
888 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
889 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
890 @command{gcc} is to stop. Note that some combinations (for example,
891 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
896 Compile or assemble the source files, but do not link. The linking
897 stage simply is not done. The ultimate output is in the form of an
898 object file for each source file.
900 By default, the object file name for a source file is made by replacing
901 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
903 Unrecognized input files, not requiring compilation or assembly, are
908 Stop after the stage of compilation proper; do not assemble. The output
909 is in the form of an assembler code file for each non-assembler input
912 By default, the assembler file name for a source file is made by
913 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
915 Input files that don't require compilation are ignored.
919 Stop after the preprocessing stage; do not run the compiler proper. The
920 output is in the form of preprocessed source code, which is sent to the
923 Input files which don't require preprocessing are ignored.
925 @cindex output file option
928 Place output in file @var{file}. This applies regardless to whatever
929 sort of output is being produced, whether it be an executable file,
930 an object file, an assembler file or preprocessed C code.
932 If @option{-o} is not specified, the default is to put an executable
933 file in @file{a.out}, the object file for
934 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
935 assembler file in @file{@var{source}.s}, a precompiled header file in
936 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
941 Print (on standard error output) the commands executed to run the stages
942 of compilation. Also print the version number of the compiler driver
943 program and of the preprocessor and the compiler proper.
947 Like @option{-v} except the commands are not executed and all command
948 arguments are quoted. This is useful for shell scripts to capture the
949 driver-generated command lines.
953 Use pipes rather than temporary files for communication between the
954 various stages of compilation. This fails to work on some systems where
955 the assembler is unable to read from a pipe; but the GNU assembler has
960 If you are compiling multiple source files, this option tells the driver
961 to pass all the source files to the compiler at once (for those
962 languages for which the compiler can handle this). This will allow
963 intermodule analysis (IMA) to be performed by the compiler. Currently the only
964 language for which this is supported is C@. If you pass source files for
965 multiple languages to the driver, using this option, the driver will invoke
966 the compiler(s) that support IMA once each, passing each compiler all the
967 source files appropriate for it. For those languages that do not support
968 IMA this option will be ignored, and the compiler will be invoked once for
969 each source file in that language. If you use this option in conjunction
970 with @option{-save-temps}, the compiler will generate multiple
972 (one for each source file), but only one (combined) @file{.o} or
977 Print (on the standard output) a description of the command line options
978 understood by @command{gcc}. If the @option{-v} option is also specified
979 then @option{--help} will also be passed on to the various processes
980 invoked by @command{gcc}, so that they can display the command line options
981 they accept. If the @option{-Wextra} option is also specified then command
982 line options which have no documentation associated with them will also
987 Print (on the standard output) a description of target specific command
988 line options for each tool.
992 Display the version number and copyrights of the invoked GCC.
996 @section Compiling C++ Programs
998 @cindex suffixes for C++ source
999 @cindex C++ source file suffixes
1000 C++ source files conventionally use one of the suffixes @samp{.C},
1001 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1002 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1003 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1004 files with these names and compiles them as C++ programs even if you
1005 call the compiler the same way as for compiling C programs (usually
1006 with the name @command{gcc}).
1010 However, C++ programs often require class libraries as well as a
1011 compiler that understands the C++ language---and under some
1012 circumstances, you might want to compile programs or header files from
1013 standard input, or otherwise without a suffix that flags them as C++
1014 programs. You might also like to precompile a C header file with a
1015 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1016 program that calls GCC with the default language set to C++, and
1017 automatically specifies linking against the C++ library. On many
1018 systems, @command{g++} is also installed with the name @command{c++}.
1020 @cindex invoking @command{g++}
1021 When you compile C++ programs, you may specify many of the same
1022 command-line options that you use for compiling programs in any
1023 language; or command-line options meaningful for C and related
1024 languages; or options that are meaningful only for C++ programs.
1025 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1026 explanations of options for languages related to C@.
1027 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1028 explanations of options that are meaningful only for C++ programs.
1030 @node C Dialect Options
1031 @section Options Controlling C Dialect
1032 @cindex dialect options
1033 @cindex language dialect options
1034 @cindex options, dialect
1036 The following options control the dialect of C (or languages derived
1037 from C, such as C++, Objective-C and Objective-C++) that the compiler
1041 @cindex ANSI support
1045 In C mode, support all ISO C90 programs. In C++ mode,
1046 remove GNU extensions that conflict with ISO C++.
1048 This turns off certain features of GCC that are incompatible with ISO
1049 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1050 such as the @code{asm} and @code{typeof} keywords, and
1051 predefined macros such as @code{unix} and @code{vax} that identify the
1052 type of system you are using. It also enables the undesirable and
1053 rarely used ISO trigraph feature. For the C compiler,
1054 it disables recognition of C++ style @samp{//} comments as well as
1055 the @code{inline} keyword.
1057 The alternate keywords @code{__asm__}, @code{__extension__},
1058 @code{__inline__} and @code{__typeof__} continue to work despite
1059 @option{-ansi}. You would not want to use them in an ISO C program, of
1060 course, but it is useful to put them in header files that might be included
1061 in compilations done with @option{-ansi}. Alternate predefined macros
1062 such as @code{__unix__} and @code{__vax__} are also available, with or
1063 without @option{-ansi}.
1065 The @option{-ansi} option does not cause non-ISO programs to be
1066 rejected gratuitously. For that, @option{-pedantic} is required in
1067 addition to @option{-ansi}. @xref{Warning Options}.
1069 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1070 option is used. Some header files may notice this macro and refrain
1071 from declaring certain functions or defining certain macros that the
1072 ISO standard doesn't call for; this is to avoid interfering with any
1073 programs that might use these names for other things.
1075 Functions which would normally be built in but do not have semantics
1076 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1077 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1078 built-in functions provided by GCC}, for details of the functions
1083 Determine the language standard. This option is currently only
1084 supported when compiling C or C++. A value for this option must be
1085 provided; possible values are
1090 ISO C90 (same as @option{-ansi}).
1092 @item iso9899:199409
1093 ISO C90 as modified in amendment 1.
1099 ISO C99. Note that this standard is not yet fully supported; see
1100 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1101 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1104 Default, ISO C90 plus GNU extensions (including some C99 features).
1108 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1109 this will become the default. The name @samp{gnu9x} is deprecated.
1112 The 1998 ISO C++ standard plus amendments.
1115 The same as @option{-std=c++98} plus GNU extensions. This is the
1116 default for C++ code.
1119 Even when this option is not specified, you can still use some of the
1120 features of newer standards in so far as they do not conflict with
1121 previous C standards. For example, you may use @code{__restrict__} even
1122 when @option{-std=c99} is not specified.
1124 The @option{-std} options specifying some version of ISO C have the same
1125 effects as @option{-ansi}, except that features that were not in ISO C90
1126 but are in the specified version (for example, @samp{//} comments and
1127 the @code{inline} keyword in ISO C99) are not disabled.
1129 @xref{Standards,,Language Standards Supported by GCC}, for details of
1130 these standard versions.
1132 @item -aux-info @var{filename}
1134 Output to the given filename prototyped declarations for all functions
1135 declared and/or defined in a translation unit, including those in header
1136 files. This option is silently ignored in any language other than C@.
1138 Besides declarations, the file indicates, in comments, the origin of
1139 each declaration (source file and line), whether the declaration was
1140 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1141 @samp{O} for old, respectively, in the first character after the line
1142 number and the colon), and whether it came from a declaration or a
1143 definition (@samp{C} or @samp{F}, respectively, in the following
1144 character). In the case of function definitions, a K&R-style list of
1145 arguments followed by their declarations is also provided, inside
1146 comments, after the declaration.
1150 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1151 keyword, so that code can use these words as identifiers. You can use
1152 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1153 instead. @option{-ansi} implies @option{-fno-asm}.
1155 In C++, this switch only affects the @code{typeof} keyword, since
1156 @code{asm} and @code{inline} are standard keywords. You may want to
1157 use the @option{-fno-gnu-keywords} flag instead, which has the same
1158 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1159 switch only affects the @code{asm} and @code{typeof} keywords, since
1160 @code{inline} is a standard keyword in ISO C99.
1163 @itemx -fno-builtin-@var{function}
1164 @opindex fno-builtin
1165 @cindex built-in functions
1166 Don't recognize built-in functions that do not begin with
1167 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1168 functions provided by GCC}, for details of the functions affected,
1169 including those which are not built-in functions when @option{-ansi} or
1170 @option{-std} options for strict ISO C conformance are used because they
1171 do not have an ISO standard meaning.
1173 GCC normally generates special code to handle certain built-in functions
1174 more efficiently; for instance, calls to @code{alloca} may become single
1175 instructions that adjust the stack directly, and calls to @code{memcpy}
1176 may become inline copy loops. The resulting code is often both smaller
1177 and faster, but since the function calls no longer appear as such, you
1178 cannot set a breakpoint on those calls, nor can you change the behavior
1179 of the functions by linking with a different library. In addition,
1180 when a function is recognized as a built-in function, GCC may use
1181 information about that function to warn about problems with calls to
1182 that function, or to generate more efficient code, even if the
1183 resulting code still contains calls to that function. For example,
1184 warnings are given with @option{-Wformat} for bad calls to
1185 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1186 known not to modify global memory.
1188 With the @option{-fno-builtin-@var{function}} option
1189 only the built-in function @var{function} is
1190 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1191 function is named this is not built-in in this version of GCC, this
1192 option is ignored. There is no corresponding
1193 @option{-fbuiltin-@var{function}} option; if you wish to enable
1194 built-in functions selectively when using @option{-fno-builtin} or
1195 @option{-ffreestanding}, you may define macros such as:
1198 #define abs(n) __builtin_abs ((n))
1199 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1204 @cindex hosted environment
1206 Assert that compilation takes place in a hosted environment. This implies
1207 @option{-fbuiltin}. A hosted environment is one in which the
1208 entire standard library is available, and in which @code{main} has a return
1209 type of @code{int}. Examples are nearly everything except a kernel.
1210 This is equivalent to @option{-fno-freestanding}.
1212 @item -ffreestanding
1213 @opindex ffreestanding
1214 @cindex hosted environment
1216 Assert that compilation takes place in a freestanding environment. This
1217 implies @option{-fno-builtin}. A freestanding environment
1218 is one in which the standard library may not exist, and program startup may
1219 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1220 This is equivalent to @option{-fno-hosted}.
1222 @xref{Standards,,Language Standards Supported by GCC}, for details of
1223 freestanding and hosted environments.
1225 @item -fms-extensions
1226 @opindex fms-extensions
1227 Accept some non-standard constructs used in Microsoft header files.
1229 Some cases of unnamed fields in structures and unions are only
1230 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1231 fields within structs/unions}, for details.
1235 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1236 options for strict ISO C conformance) implies @option{-trigraphs}.
1238 @item -no-integrated-cpp
1239 @opindex no-integrated-cpp
1240 Performs a compilation in two passes: preprocessing and compiling. This
1241 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1242 @option{-B} option. The user supplied compilation step can then add in
1243 an additional preprocessing step after normal preprocessing but before
1244 compiling. The default is to use the integrated cpp (internal cpp)
1246 The semantics of this option will change if "cc1", "cc1plus", and
1247 "cc1obj" are merged.
1249 @cindex traditional C language
1250 @cindex C language, traditional
1252 @itemx -traditional-cpp
1253 @opindex traditional-cpp
1254 @opindex traditional
1255 Formerly, these options caused GCC to attempt to emulate a pre-standard
1256 C compiler. They are now only supported with the @option{-E} switch.
1257 The preprocessor continues to support a pre-standard mode. See the GNU
1258 CPP manual for details.
1260 @item -fcond-mismatch
1261 @opindex fcond-mismatch
1262 Allow conditional expressions with mismatched types in the second and
1263 third arguments. The value of such an expression is void. This option
1264 is not supported for C++.
1266 @item -funsigned-char
1267 @opindex funsigned-char
1268 Let the type @code{char} be unsigned, like @code{unsigned char}.
1270 Each kind of machine has a default for what @code{char} should
1271 be. It is either like @code{unsigned char} by default or like
1272 @code{signed char} by default.
1274 Ideally, a portable program should always use @code{signed char} or
1275 @code{unsigned char} when it depends on the signedness of an object.
1276 But many programs have been written to use plain @code{char} and
1277 expect it to be signed, or expect it to be unsigned, depending on the
1278 machines they were written for. This option, and its inverse, let you
1279 make such a program work with the opposite default.
1281 The type @code{char} is always a distinct type from each of
1282 @code{signed char} or @code{unsigned char}, even though its behavior
1283 is always just like one of those two.
1286 @opindex fsigned-char
1287 Let the type @code{char} be signed, like @code{signed char}.
1289 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1290 the negative form of @option{-funsigned-char}. Likewise, the option
1291 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1293 @item -fsigned-bitfields
1294 @itemx -funsigned-bitfields
1295 @itemx -fno-signed-bitfields
1296 @itemx -fno-unsigned-bitfields
1297 @opindex fsigned-bitfields
1298 @opindex funsigned-bitfields
1299 @opindex fno-signed-bitfields
1300 @opindex fno-unsigned-bitfields
1301 These options control whether a bit-field is signed or unsigned, when the
1302 declaration does not use either @code{signed} or @code{unsigned}. By
1303 default, such a bit-field is signed, because this is consistent: the
1304 basic integer types such as @code{int} are signed types.
1307 @node C++ Dialect Options
1308 @section Options Controlling C++ Dialect
1310 @cindex compiler options, C++
1311 @cindex C++ options, command line
1312 @cindex options, C++
1313 This section describes the command-line options that are only meaningful
1314 for C++ programs; but you can also use most of the GNU compiler options
1315 regardless of what language your program is in. For example, you
1316 might compile a file @code{firstClass.C} like this:
1319 g++ -g -frepo -O -c firstClass.C
1323 In this example, only @option{-frepo} is an option meant
1324 only for C++ programs; you can use the other options with any
1325 language supported by GCC@.
1327 Here is a list of options that are @emph{only} for compiling C++ programs:
1331 @item -fabi-version=@var{n}
1332 @opindex fabi-version
1333 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1334 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1335 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1336 the version that conforms most closely to the C++ ABI specification.
1337 Therefore, the ABI obtained using version 0 will change as ABI bugs
1340 The default is version 2.
1342 @item -fno-access-control
1343 @opindex fno-access-control
1344 Turn off all access checking. This switch is mainly useful for working
1345 around bugs in the access control code.
1349 Check that the pointer returned by @code{operator new} is non-null
1350 before attempting to modify the storage allocated. This check is
1351 normally unnecessary because the C++ standard specifies that
1352 @code{operator new} will only return @code{0} if it is declared
1353 @samp{throw()}, in which case the compiler will always check the
1354 return value even without this option. In all other cases, when
1355 @code{operator new} has a non-empty exception specification, memory
1356 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1357 @samp{new (nothrow)}.
1359 @item -fconserve-space
1360 @opindex fconserve-space
1361 Put uninitialized or runtime-initialized global variables into the
1362 common segment, as C does. This saves space in the executable at the
1363 cost of not diagnosing duplicate definitions. If you compile with this
1364 flag and your program mysteriously crashes after @code{main()} has
1365 completed, you may have an object that is being destroyed twice because
1366 two definitions were merged.
1368 This option is no longer useful on most targets, now that support has
1369 been added for putting variables into BSS without making them common.
1371 @item -fno-const-strings
1372 @opindex fno-const-strings
1373 Give string constants type @code{char *} instead of type @code{const
1374 char *}. By default, G++ uses type @code{const char *} as required by
1375 the standard. Even if you use @option{-fno-const-strings}, you cannot
1376 actually modify the value of a string constant.
1378 This option might be removed in a future release of G++. For maximum
1379 portability, you should structure your code so that it works with
1380 string constants that have type @code{const char *}.
1382 @item -fno-elide-constructors
1383 @opindex fno-elide-constructors
1384 The C++ standard allows an implementation to omit creating a temporary
1385 which is only used to initialize another object of the same type.
1386 Specifying this option disables that optimization, and forces G++ to
1387 call the copy constructor in all cases.
1389 @item -fno-enforce-eh-specs
1390 @opindex fno-enforce-eh-specs
1391 Don't check for violation of exception specifications at runtime. This
1392 option violates the C++ standard, but may be useful for reducing code
1393 size in production builds, much like defining @samp{NDEBUG}. The compiler
1394 will still optimize based on the exception specifications.
1397 @itemx -fno-for-scope
1399 @opindex fno-for-scope
1400 If @option{-ffor-scope} is specified, the scope of variables declared in
1401 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1402 as specified by the C++ standard.
1403 If @option{-fno-for-scope} is specified, the scope of variables declared in
1404 a @i{for-init-statement} extends to the end of the enclosing scope,
1405 as was the case in old versions of G++, and other (traditional)
1406 implementations of C++.
1408 The default if neither flag is given to follow the standard,
1409 but to allow and give a warning for old-style code that would
1410 otherwise be invalid, or have different behavior.
1412 @item -fno-gnu-keywords
1413 @opindex fno-gnu-keywords
1414 Do not recognize @code{typeof} as a keyword, so that code can use this
1415 word as an identifier. You can use the keyword @code{__typeof__} instead.
1416 @option{-ansi} implies @option{-fno-gnu-keywords}.
1418 @item -fno-implicit-templates
1419 @opindex fno-implicit-templates
1420 Never emit code for non-inline templates which are instantiated
1421 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1422 @xref{Template Instantiation}, for more information.
1424 @item -fno-implicit-inline-templates
1425 @opindex fno-implicit-inline-templates
1426 Don't emit code for implicit instantiations of inline templates, either.
1427 The default is to handle inlines differently so that compiles with and
1428 without optimization will need the same set of explicit instantiations.
1430 @item -fno-implement-inlines
1431 @opindex fno-implement-inlines
1432 To save space, do not emit out-of-line copies of inline functions
1433 controlled by @samp{#pragma implementation}. This will cause linker
1434 errors if these functions are not inlined everywhere they are called.
1436 @item -fms-extensions
1437 @opindex fms-extensions
1438 Disable pedantic warnings about constructs used in MFC, such as implicit
1439 int and getting a pointer to member function via non-standard syntax.
1441 @item -fno-nonansi-builtins
1442 @opindex fno-nonansi-builtins
1443 Disable built-in declarations of functions that are not mandated by
1444 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1445 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1447 @item -fno-operator-names
1448 @opindex fno-operator-names
1449 Do not treat the operator name keywords @code{and}, @code{bitand},
1450 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1451 synonyms as keywords.
1453 @item -fno-optional-diags
1454 @opindex fno-optional-diags
1455 Disable diagnostics that the standard says a compiler does not need to
1456 issue. Currently, the only such diagnostic issued by G++ is the one for
1457 a name having multiple meanings within a class.
1460 @opindex fpermissive
1461 Downgrade some diagnostics about nonconformant code from errors to
1462 warnings. Thus, using @option{-fpermissive} will allow some
1463 nonconforming code to compile.
1467 Enable automatic template instantiation at link time. This option also
1468 implies @option{-fno-implicit-templates}. @xref{Template
1469 Instantiation}, for more information.
1473 Disable generation of information about every class with virtual
1474 functions for use by the C++ runtime type identification features
1475 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1476 of the language, you can save some space by using this flag. Note that
1477 exception handling uses the same information, but it will generate it as
1482 Emit statistics about front-end processing at the end of the compilation.
1483 This information is generally only useful to the G++ development team.
1485 @item -ftemplate-depth-@var{n}
1486 @opindex ftemplate-depth
1487 Set the maximum instantiation depth for template classes to @var{n}.
1488 A limit on the template instantiation depth is needed to detect
1489 endless recursions during template class instantiation. ANSI/ISO C++
1490 conforming programs must not rely on a maximum depth greater than 17.
1492 @item -fno-threadsafe-statics
1493 @opindex fno-threadsafe-statics
1494 Do not emit the extra code to use the routines specified in the C++
1495 ABI for thread-safe initialization of local statics. You can use this
1496 option to reduce code size slightly in code that doesn't need to be
1499 @item -fuse-cxa-atexit
1500 @opindex fuse-cxa-atexit
1501 Register destructors for objects with static storage duration with the
1502 @code{__cxa_atexit} function rather than the @code{atexit} function.
1503 This option is required for fully standards-compliant handling of static
1504 destructors, but will only work if your C library supports
1505 @code{__cxa_atexit}.
1507 @item -fvisibility-inlines-hidden
1508 @opindex fvisibility-inlines-hidden
1509 Causes all inlined methods to be marked with
1510 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1511 appear in the export table of a DSO and do not require a PLT indirection
1512 when used within the DSO. Enabling this option can have a dramatic effect
1513 on load and link times of a DSO as it massively reduces the size of the
1514 dynamic export table when the library makes heavy use of templates. While
1515 it can cause bloating through duplication of code within each DSO where
1516 it is used, often the wastage is less than the considerable space occupied
1517 by a long symbol name in the export table which is typical when using
1518 templates and namespaces. For even more savings, combine with the
1519 @option{-fvisibility=hidden} switch.
1523 Do not use weak symbol support, even if it is provided by the linker.
1524 By default, G++ will use weak symbols if they are available. This
1525 option exists only for testing, and should not be used by end-users;
1526 it will result in inferior code and has no benefits. This option may
1527 be removed in a future release of G++.
1531 Do not search for header files in the standard directories specific to
1532 C++, but do still search the other standard directories. (This option
1533 is used when building the C++ library.)
1536 In addition, these optimization, warning, and code generation options
1537 have meanings only for C++ programs:
1540 @item -fno-default-inline
1541 @opindex fno-default-inline
1542 Do not assume @samp{inline} for functions defined inside a class scope.
1543 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1544 functions will have linkage like inline functions; they just won't be
1547 @item -Wabi @r{(C++ only)}
1549 Warn when G++ generates code that is probably not compatible with the
1550 vendor-neutral C++ ABI. Although an effort has been made to warn about
1551 all such cases, there are probably some cases that are not warned about,
1552 even though G++ is generating incompatible code. There may also be
1553 cases where warnings are emitted even though the code that is generated
1556 You should rewrite your code to avoid these warnings if you are
1557 concerned about the fact that code generated by G++ may not be binary
1558 compatible with code generated by other compilers.
1560 The known incompatibilities at this point include:
1565 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1566 pack data into the same byte as a base class. For example:
1569 struct A @{ virtual void f(); int f1 : 1; @};
1570 struct B : public A @{ int f2 : 1; @};
1574 In this case, G++ will place @code{B::f2} into the same byte
1575 as@code{A::f1}; other compilers will not. You can avoid this problem
1576 by explicitly padding @code{A} so that its size is a multiple of the
1577 byte size on your platform; that will cause G++ and other compilers to
1578 layout @code{B} identically.
1581 Incorrect handling of tail-padding for virtual bases. G++ does not use
1582 tail padding when laying out virtual bases. For example:
1585 struct A @{ virtual void f(); char c1; @};
1586 struct B @{ B(); char c2; @};
1587 struct C : public A, public virtual B @{@};
1591 In this case, G++ will not place @code{B} into the tail-padding for
1592 @code{A}; other compilers will. You can avoid this problem by
1593 explicitly padding @code{A} so that its size is a multiple of its
1594 alignment (ignoring virtual base classes); that will cause G++ and other
1595 compilers to layout @code{C} identically.
1598 Incorrect handling of bit-fields with declared widths greater than that
1599 of their underlying types, when the bit-fields appear in a union. For
1603 union U @{ int i : 4096; @};
1607 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1608 union too small by the number of bits in an @code{int}.
1611 Empty classes can be placed at incorrect offsets. For example:
1621 struct C : public B, public A @{@};
1625 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1626 it should be placed at offset zero. G++ mistakenly believes that the
1627 @code{A} data member of @code{B} is already at offset zero.
1630 Names of template functions whose types involve @code{typename} or
1631 template template parameters can be mangled incorrectly.
1634 template <typename Q>
1635 void f(typename Q::X) @{@}
1637 template <template <typename> class Q>
1638 void f(typename Q<int>::X) @{@}
1642 Instantiations of these templates may be mangled incorrectly.
1646 @item -Wctor-dtor-privacy @r{(C++ only)}
1647 @opindex Wctor-dtor-privacy
1648 Warn when a class seems unusable because all the constructors or
1649 destructors in that class are private, and it has neither friends nor
1650 public static member functions.
1652 @item -Wnon-virtual-dtor @r{(C++ only)}
1653 @opindex Wnon-virtual-dtor
1654 Warn when a class appears to be polymorphic, thereby requiring a virtual
1655 destructor, yet it declares a non-virtual one.
1656 This warning is enabled by @option{-Wall}.
1658 @item -Wreorder @r{(C++ only)}
1660 @cindex reordering, warning
1661 @cindex warning for reordering of member initializers
1662 Warn when the order of member initializers given in the code does not
1663 match the order in which they must be executed. For instance:
1669 A(): j (0), i (1) @{ @}
1673 The compiler will rearrange the member initializers for @samp{i}
1674 and @samp{j} to match the declaration order of the members, emitting
1675 a warning to that effect. This warning is enabled by @option{-Wall}.
1678 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1681 @item -Weffc++ @r{(C++ only)}
1683 Warn about violations of the following style guidelines from Scott Meyers'
1684 @cite{Effective C++} book:
1688 Item 11: Define a copy constructor and an assignment operator for classes
1689 with dynamically allocated memory.
1692 Item 12: Prefer initialization to assignment in constructors.
1695 Item 14: Make destructors virtual in base classes.
1698 Item 15: Have @code{operator=} return a reference to @code{*this}.
1701 Item 23: Don't try to return a reference when you must return an object.
1705 Also warn about violations of the following style guidelines from
1706 Scott Meyers' @cite{More Effective C++} book:
1710 Item 6: Distinguish between prefix and postfix forms of increment and
1711 decrement operators.
1714 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1718 When selecting this option, be aware that the standard library
1719 headers do not obey all of these guidelines; use @samp{grep -v}
1720 to filter out those warnings.
1722 @item -Wno-deprecated @r{(C++ only)}
1723 @opindex Wno-deprecated
1724 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1726 @item -Wno-non-template-friend @r{(C++ only)}
1727 @opindex Wno-non-template-friend
1728 Disable warnings when non-templatized friend functions are declared
1729 within a template. Since the advent of explicit template specification
1730 support in G++, if the name of the friend is an unqualified-id (i.e.,
1731 @samp{friend foo(int)}), the C++ language specification demands that the
1732 friend declare or define an ordinary, nontemplate function. (Section
1733 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1734 could be interpreted as a particular specialization of a templatized
1735 function. Because this non-conforming behavior is no longer the default
1736 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1737 check existing code for potential trouble spots and is on by default.
1738 This new compiler behavior can be turned off with
1739 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1740 but disables the helpful warning.
1742 @item -Wold-style-cast @r{(C++ only)}
1743 @opindex Wold-style-cast
1744 Warn if an old-style (C-style) cast to a non-void type is used within
1745 a C++ program. The new-style casts (@samp{static_cast},
1746 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1747 unintended effects and much easier to search for.
1749 @item -Woverloaded-virtual @r{(C++ only)}
1750 @opindex Woverloaded-virtual
1751 @cindex overloaded virtual fn, warning
1752 @cindex warning for overloaded virtual fn
1753 Warn when a function declaration hides virtual functions from a
1754 base class. For example, in:
1761 struct B: public A @{
1766 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1774 will fail to compile.
1776 @item -Wno-pmf-conversions @r{(C++ only)}
1777 @opindex Wno-pmf-conversions
1778 Disable the diagnostic for converting a bound pointer to member function
1781 @item -Wsign-promo @r{(C++ only)}
1782 @opindex Wsign-promo
1783 Warn when overload resolution chooses a promotion from unsigned or
1784 enumerated type to a signed type, over a conversion to an unsigned type of
1785 the same size. Previous versions of G++ would try to preserve
1786 unsignedness, but the standard mandates the current behavior.
1788 @item -Wsynth @r{(C++ only)}
1790 @cindex warning for synthesized methods
1791 @cindex synthesized methods, warning
1792 Warn when G++'s synthesis behavior does not match that of cfront. For
1798 A& operator = (int);
1808 In this example, G++ will synthesize a default @samp{A& operator =
1809 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1812 @node Objective-C and Objective-C++ Dialect Options
1813 @section Options Controlling Objective-C and Objective-C++ Dialects
1815 @cindex compiler options, Objective-C and Objective-C++
1816 @cindex Objective-C and Objective-C++ options, command line
1817 @cindex options, Objective-C and Objective-C++
1818 (NOTE: This manual does not describe the Objective-C and Objective-C++
1819 languages themselves. See @xref{Standards,,Language Standards
1820 Supported by GCC}, for references.)
1822 This section describes the command-line options that are only meaningful
1823 for Objective-C and Objective-C++ programs, but you can also use most of
1824 the language-independent GNU compiler options.
1825 For example, you might compile a file @code{some_class.m} like this:
1828 gcc -g -fgnu-runtime -O -c some_class.m
1832 In this example, @option{-fgnu-runtime} is an option meant only for
1833 Objective-C and Objective-C++ programs; you can use the other options with
1834 any language supported by GCC@.
1836 Note that since Objective-C is an extension of the C language, Objective-C
1837 compilations may also use options specific to the C front-end (e.g.,
1838 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1839 C++-specific options (e.g., @option{-Wabi}).
1841 Here is a list of options that are @emph{only} for compiling Objective-C
1842 and Objective-C++ programs:
1845 @item -fconstant-string-class=@var{class-name}
1846 @opindex fconstant-string-class
1847 Use @var{class-name} as the name of the class to instantiate for each
1848 literal string specified with the syntax @code{@@"@dots{}"}. The default
1849 class name is @code{NXConstantString} if the GNU runtime is being used, and
1850 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1851 @option{-fconstant-cfstrings} option, if also present, will override the
1852 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1853 to be laid out as constant CoreFoundation strings.
1856 @opindex fgnu-runtime
1857 Generate object code compatible with the standard GNU Objective-C
1858 runtime. This is the default for most types of systems.
1860 @item -fnext-runtime
1861 @opindex fnext-runtime
1862 Generate output compatible with the NeXT runtime. This is the default
1863 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1864 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1867 @item -fno-nil-receivers
1868 @opindex fno-nil-receivers
1869 Assume that all Objective-C message dispatches (e.g.,
1870 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1871 is not @code{nil}. This allows for more efficient entry points in the runtime
1872 to be used. Currently, this option is only available in conjunction with
1873 the NeXT runtime on Mac OS X 10.3 and later.
1875 @item -fobjc-exceptions
1876 @opindex fobjc-exceptions
1877 Enable syntactic support for structured exception handling in Objective-C,
1878 similar to what is offered by C++ and Java. Currently, this option is only
1879 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1887 @@catch (AnObjCClass *exc) @{
1894 @@catch (AnotherClass *exc) @{
1897 @@catch (id allOthers) @{
1907 The @code{@@throw} statement may appear anywhere in an Objective-C or
1908 Objective-C++ program; when used inside of a @code{@@catch} block, the
1909 @code{@@throw} may appear without an argument (as shown above), in which case
1910 the object caught by the @code{@@catch} will be rethrown.
1912 Note that only (pointers to) Objective-C objects may be thrown and
1913 caught using this scheme. When an object is thrown, it will be caught
1914 by the nearest @code{@@catch} clause capable of handling objects of that type,
1915 analogously to how @code{catch} blocks work in C++ and Java. A
1916 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1917 any and all Objective-C exceptions not caught by previous @code{@@catch}
1920 The @code{@@finally} clause, if present, will be executed upon exit from the
1921 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1922 regardless of whether any exceptions are thrown, caught or rethrown
1923 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1924 of the @code{finally} clause in Java.
1926 There are several caveats to using the new exception mechanism:
1930 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1931 idioms provided by the @code{NSException} class, the new
1932 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1933 systems, due to additional functionality needed in the (NeXT) Objective-C
1937 As mentioned above, the new exceptions do not support handling
1938 types other than Objective-C objects. Furthermore, when used from
1939 Objective-C++, the Objective-C exception model does not interoperate with C++
1940 exceptions at this time. This means you cannot @code{@@throw} an exception
1941 from Objective-C and @code{catch} it in C++, or vice versa
1942 (i.e., @code{throw @dots{} @@catch}).
1945 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1946 blocks for thread-safe execution:
1949 @@synchronized (ObjCClass *guard) @{
1954 Upon entering the @code{@@synchronized} block, a thread of execution shall
1955 first check whether a lock has been placed on the corresponding @code{guard}
1956 object by another thread. If it has, the current thread shall wait until
1957 the other thread relinquishes its lock. Once @code{guard} becomes available,
1958 the current thread will place its own lock on it, execute the code contained in
1959 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1960 making @code{guard} available to other threads).
1962 Unlike Java, Objective-C does not allow for entire methods to be marked
1963 @code{@@synchronized}. Note that throwing exceptions out of
1964 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1965 to be unlocked properly.
1967 @item -freplace-objc-classes
1968 @opindex freplace-objc-classes
1969 Emit a special marker instructing @command{ld(1)} not to statically link in
1970 the resulting object file, and allow @command{dyld(1)} to load it in at
1971 run time instead. This is used in conjunction with the Fix-and-Continue
1972 debugging mode, where the object file in question may be recompiled and
1973 dynamically reloaded in the course of program execution, without the need
1974 to restart the program itself. Currently, Fix-and-Continue functionality
1975 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1980 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1981 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1982 compile time) with static class references that get initialized at load time,
1983 which improves run-time performance. Specifying the @option{-fzero-link} flag
1984 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1985 to be retained. This is useful in Zero-Link debugging mode, since it allows
1986 for individual class implementations to be modified during program execution.
1990 Dump interface declarations for all classes seen in the source file to a
1991 file named @file{@var{sourcename}.decl}.
1994 @opindex Wno-protocol
1995 If a class is declared to implement a protocol, a warning is issued for
1996 every method in the protocol that is not implemented by the class. The
1997 default behavior is to issue a warning for every method not explicitly
1998 implemented in the class, even if a method implementation is inherited
1999 from the superclass. If you use the @option{-Wno-protocol} option, then
2000 methods inherited from the superclass are considered to be implemented,
2001 and no warning is issued for them.
2005 Warn if multiple methods of different types for the same selector are
2006 found during compilation. The check is performed on the list of methods
2007 in the final stage of compilation. Additionally, a check is performed
2008 for each selector appearing in a @code{@@selector(@dots{})}
2009 expression, and a corresponding method for that selector has been found
2010 during compilation. Because these checks scan the method table only at
2011 the end of compilation, these warnings are not produced if the final
2012 stage of compilation is not reached, for example because an error is
2013 found during compilation, or because the @option{-fsyntax-only} option is
2016 @item -Wundeclared-selector
2017 @opindex Wundeclared-selector
2018 Warn if a @code{@@selector(@dots{})} expression referring to an
2019 undeclared selector is found. A selector is considered undeclared if no
2020 method with that name has been declared before the
2021 @code{@@selector(@dots{})} expression, either explicitly in an
2022 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2023 an @code{@@implementation} section. This option always performs its
2024 checks as soon as a @code{@@selector(@dots{})} expression is found,
2025 while @option{-Wselector} only performs its checks in the final stage of
2026 compilation. This also enforces the coding style convention
2027 that methods and selectors must be declared before being used.
2029 @item -print-objc-runtime-info
2030 @opindex print-objc-runtime-info
2031 Generate C header describing the largest structure that is passed by
2036 @node Language Independent Options
2037 @section Options to Control Diagnostic Messages Formatting
2038 @cindex options to control diagnostics formatting
2039 @cindex diagnostic messages
2040 @cindex message formatting
2042 Traditionally, diagnostic messages have been formatted irrespective of
2043 the output device's aspect (e.g.@: its width, @dots{}). The options described
2044 below can be used to control the diagnostic messages formatting
2045 algorithm, e.g.@: how many characters per line, how often source location
2046 information should be reported. Right now, only the C++ front end can
2047 honor these options. However it is expected, in the near future, that
2048 the remaining front ends would be able to digest them correctly.
2051 @item -fmessage-length=@var{n}
2052 @opindex fmessage-length
2053 Try to format error messages so that they fit on lines of about @var{n}
2054 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2055 the front ends supported by GCC@. If @var{n} is zero, then no
2056 line-wrapping will be done; each error message will appear on a single
2059 @opindex fdiagnostics-show-location
2060 @item -fdiagnostics-show-location=once
2061 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2062 reporter to emit @emph{once} source location information; that is, in
2063 case the message is too long to fit on a single physical line and has to
2064 be wrapped, the source location won't be emitted (as prefix) again,
2065 over and over, in subsequent continuation lines. This is the default
2068 @item -fdiagnostics-show-location=every-line
2069 Only meaningful in line-wrapping mode. Instructs the diagnostic
2070 messages reporter to emit the same source location information (as
2071 prefix) for physical lines that result from the process of breaking
2072 a message which is too long to fit on a single line.
2076 @node Warning Options
2077 @section Options to Request or Suppress Warnings
2078 @cindex options to control warnings
2079 @cindex warning messages
2080 @cindex messages, warning
2081 @cindex suppressing warnings
2083 Warnings are diagnostic messages that report constructions which
2084 are not inherently erroneous but which are risky or suggest there
2085 may have been an error.
2087 You can request many specific warnings with options beginning @samp{-W},
2088 for example @option{-Wimplicit} to request warnings on implicit
2089 declarations. Each of these specific warning options also has a
2090 negative form beginning @samp{-Wno-} to turn off warnings;
2091 for example, @option{-Wno-implicit}. This manual lists only one of the
2092 two forms, whichever is not the default.
2094 The following options control the amount and kinds of warnings produced
2095 by GCC; for further, language-specific options also refer to
2096 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2100 @cindex syntax checking
2102 @opindex fsyntax-only
2103 Check the code for syntax errors, but don't do anything beyond that.
2107 Issue all the warnings demanded by strict ISO C and ISO C++;
2108 reject all programs that use forbidden extensions, and some other
2109 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2110 version of the ISO C standard specified by any @option{-std} option used.
2112 Valid ISO C and ISO C++ programs should compile properly with or without
2113 this option (though a rare few will require @option{-ansi} or a
2114 @option{-std} option specifying the required version of ISO C)@. However,
2115 without this option, certain GNU extensions and traditional C and C++
2116 features are supported as well. With this option, they are rejected.
2118 @option{-pedantic} does not cause warning messages for use of the
2119 alternate keywords whose names begin and end with @samp{__}. Pedantic
2120 warnings are also disabled in the expression that follows
2121 @code{__extension__}. However, only system header files should use
2122 these escape routes; application programs should avoid them.
2123 @xref{Alternate Keywords}.
2125 Some users try to use @option{-pedantic} to check programs for strict ISO
2126 C conformance. They soon find that it does not do quite what they want:
2127 it finds some non-ISO practices, but not all---only those for which
2128 ISO C @emph{requires} a diagnostic, and some others for which
2129 diagnostics have been added.
2131 A feature to report any failure to conform to ISO C might be useful in
2132 some instances, but would require considerable additional work and would
2133 be quite different from @option{-pedantic}. We don't have plans to
2134 support such a feature in the near future.
2136 Where the standard specified with @option{-std} represents a GNU
2137 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2138 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2139 extended dialect is based. Warnings from @option{-pedantic} are given
2140 where they are required by the base standard. (It would not make sense
2141 for such warnings to be given only for features not in the specified GNU
2142 C dialect, since by definition the GNU dialects of C include all
2143 features the compiler supports with the given option, and there would be
2144 nothing to warn about.)
2146 @item -pedantic-errors
2147 @opindex pedantic-errors
2148 Like @option{-pedantic}, except that errors are produced rather than
2153 Inhibit all warning messages.
2157 Inhibit warning messages about the use of @samp{#import}.
2159 @item -Wchar-subscripts
2160 @opindex Wchar-subscripts
2161 Warn if an array subscript has type @code{char}. This is a common cause
2162 of error, as programmers often forget that this type is signed on some
2167 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2168 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2170 @item -Wfatal-errors
2171 @opindex Wfatal-errors
2172 This option causes the compiler to abort compilation on the first error
2173 occurred rather than trying to keep going and printing further error
2178 @opindex ffreestanding
2179 @opindex fno-builtin
2180 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2181 the arguments supplied have types appropriate to the format string
2182 specified, and that the conversions specified in the format string make
2183 sense. This includes standard functions, and others specified by format
2184 attributes (@pxref{Function Attributes}), in the @code{printf},
2185 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2186 not in the C standard) families (or other target-specific families).
2187 Which functions are checked without format attributes having been
2188 specified depends on the standard version selected, and such checks of
2189 functions without the attribute specified are disabled by
2190 @option{-ffreestanding} or @option{-fno-builtin}.
2192 The formats are checked against the format features supported by GNU
2193 libc version 2.2. These include all ISO C90 and C99 features, as well
2194 as features from the Single Unix Specification and some BSD and GNU
2195 extensions. Other library implementations may not support all these
2196 features; GCC does not support warning about features that go beyond a
2197 particular library's limitations. However, if @option{-pedantic} is used
2198 with @option{-Wformat}, warnings will be given about format features not
2199 in the selected standard version (but not for @code{strfmon} formats,
2200 since those are not in any version of the C standard). @xref{C Dialect
2201 Options,,Options Controlling C Dialect}.
2203 Since @option{-Wformat} also checks for null format arguments for
2204 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2206 @option{-Wformat} is included in @option{-Wall}. For more control over some
2207 aspects of format checking, the options @option{-Wformat-y2k},
2208 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2209 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2210 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2213 @opindex Wformat-y2k
2214 If @option{-Wformat} is specified, also warn about @code{strftime}
2215 formats which may yield only a two-digit year.
2217 @item -Wno-format-extra-args
2218 @opindex Wno-format-extra-args
2219 If @option{-Wformat} is specified, do not warn about excess arguments to a
2220 @code{printf} or @code{scanf} format function. The C standard specifies
2221 that such arguments are ignored.
2223 Where the unused arguments lie between used arguments that are
2224 specified with @samp{$} operand number specifications, normally
2225 warnings are still given, since the implementation could not know what
2226 type to pass to @code{va_arg} to skip the unused arguments. However,
2227 in the case of @code{scanf} formats, this option will suppress the
2228 warning if the unused arguments are all pointers, since the Single
2229 Unix Specification says that such unused arguments are allowed.
2231 @item -Wno-format-zero-length
2232 @opindex Wno-format-zero-length
2233 If @option{-Wformat} is specified, do not warn about zero-length formats.
2234 The C standard specifies that zero-length formats are allowed.
2236 @item -Wformat-nonliteral
2237 @opindex Wformat-nonliteral
2238 If @option{-Wformat} is specified, also warn if the format string is not a
2239 string literal and so cannot be checked, unless the format function
2240 takes its format arguments as a @code{va_list}.
2242 @item -Wformat-security
2243 @opindex Wformat-security
2244 If @option{-Wformat} is specified, also warn about uses of format
2245 functions that represent possible security problems. At present, this
2246 warns about calls to @code{printf} and @code{scanf} functions where the
2247 format string is not a string literal and there are no format arguments,
2248 as in @code{printf (foo);}. This may be a security hole if the format
2249 string came from untrusted input and contains @samp{%n}. (This is
2250 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2251 in future warnings may be added to @option{-Wformat-security} that are not
2252 included in @option{-Wformat-nonliteral}.)
2256 Enable @option{-Wformat} plus format checks not included in
2257 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2258 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2262 Warn about passing a null pointer for arguments marked as
2263 requiring a non-null value by the @code{nonnull} function attribute.
2265 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2266 can be disabled with the @option{-Wno-nonnull} option.
2268 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2270 Warn about uninitialized variables which are initialized with themselves.
2271 Note this option can only be used with the @option{-Wuninitialized} option,
2272 which in turn only works with @option{-O1} and above.
2274 For example, GCC will warn about @code{i} being uninitialized in the
2275 following snippet only when @option{-Winit-self} has been specified:
2286 @item -Wimplicit-int
2287 @opindex Wimplicit-int
2288 Warn when a declaration does not specify a type.
2290 @item -Wimplicit-function-declaration
2291 @itemx -Werror-implicit-function-declaration
2292 @opindex Wimplicit-function-declaration
2293 @opindex Werror-implicit-function-declaration
2294 Give a warning (or error) whenever a function is used before being
2295 declared. The form @option{-Wno-error-implicit-function-declaration}
2300 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2304 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2305 function with external linkage, returning int, taking either zero
2306 arguments, two, or three arguments of appropriate types.
2308 @item -Wmissing-braces
2309 @opindex Wmissing-braces
2310 Warn if an aggregate or union initializer is not fully bracketed. In
2311 the following example, the initializer for @samp{a} is not fully
2312 bracketed, but that for @samp{b} is fully bracketed.
2315 int a[2][2] = @{ 0, 1, 2, 3 @};
2316 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2319 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2320 @opindex Wmissing-include-dirs
2321 Warn if a user-supplied include directory does not exist.
2324 @opindex Wparentheses
2325 Warn if parentheses are omitted in certain contexts, such
2326 as when there is an assignment in a context where a truth value
2327 is expected, or when operators are nested whose precedence people
2328 often get confused about. Only the warning for an assignment used as
2329 a truth value is supported when compiling C++; the other warnings are
2330 only supported when compiling C@.
2332 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2333 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2334 interpretation from that of ordinary mathematical notation.
2336 Also warn about constructions where there may be confusion to which
2337 @code{if} statement an @code{else} branch belongs. Here is an example of
2352 In C, every @code{else} branch belongs to the innermost possible @code{if}
2353 statement, which in this example is @code{if (b)}. This is often not
2354 what the programmer expected, as illustrated in the above example by
2355 indentation the programmer chose. When there is the potential for this
2356 confusion, GCC will issue a warning when this flag is specified.
2357 To eliminate the warning, add explicit braces around the innermost
2358 @code{if} statement so there is no way the @code{else} could belong to
2359 the enclosing @code{if}. The resulting code would look like this:
2375 @item -Wsequence-point
2376 @opindex Wsequence-point
2377 Warn about code that may have undefined semantics because of violations
2378 of sequence point rules in the C standard.
2380 The C standard defines the order in which expressions in a C program are
2381 evaluated in terms of @dfn{sequence points}, which represent a partial
2382 ordering between the execution of parts of the program: those executed
2383 before the sequence point, and those executed after it. These occur
2384 after the evaluation of a full expression (one which is not part of a
2385 larger expression), after the evaluation of the first operand of a
2386 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2387 function is called (but after the evaluation of its arguments and the
2388 expression denoting the called function), and in certain other places.
2389 Other than as expressed by the sequence point rules, the order of
2390 evaluation of subexpressions of an expression is not specified. All
2391 these rules describe only a partial order rather than a total order,
2392 since, for example, if two functions are called within one expression
2393 with no sequence point between them, the order in which the functions
2394 are called is not specified. However, the standards committee have
2395 ruled that function calls do not overlap.
2397 It is not specified when between sequence points modifications to the
2398 values of objects take effect. Programs whose behavior depends on this
2399 have undefined behavior; the C standard specifies that ``Between the
2400 previous and next sequence point an object shall have its stored value
2401 modified at most once by the evaluation of an expression. Furthermore,
2402 the prior value shall be read only to determine the value to be
2403 stored.''. If a program breaks these rules, the results on any
2404 particular implementation are entirely unpredictable.
2406 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2407 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2408 diagnosed by this option, and it may give an occasional false positive
2409 result, but in general it has been found fairly effective at detecting
2410 this sort of problem in programs.
2412 The present implementation of this option only works for C programs. A
2413 future implementation may also work for C++ programs.
2415 The C standard is worded confusingly, therefore there is some debate
2416 over the precise meaning of the sequence point rules in subtle cases.
2417 Links to discussions of the problem, including proposed formal
2418 definitions, may be found on the GCC readings page, at
2419 @w{@uref{http://gcc.gnu.org/readings.html}}.
2422 @opindex Wreturn-type
2423 Warn whenever a function is defined with a return-type that defaults to
2424 @code{int}. Also warn about any @code{return} statement with no
2425 return-value in a function whose return-type is not @code{void}.
2427 For C, also warn if the return type of a function has a type qualifier
2428 such as @code{const}. Such a type qualifier has no effect, since the
2429 value returned by a function is not an lvalue. ISO C prohibits
2430 qualified @code{void} return types on function definitions, so such
2431 return types always receive a warning even without this option.
2433 For C++, a function without return type always produces a diagnostic
2434 message, even when @option{-Wno-return-type} is specified. The only
2435 exceptions are @samp{main} and functions defined in system headers.
2439 Warn whenever a @code{switch} statement has an index of enumerated type
2440 and lacks a @code{case} for one or more of the named codes of that
2441 enumeration. (The presence of a @code{default} label prevents this
2442 warning.) @code{case} labels outside the enumeration range also
2443 provoke warnings when this option is used.
2445 @item -Wswitch-default
2446 @opindex Wswitch-switch
2447 Warn whenever a @code{switch} statement does not have a @code{default}
2451 @opindex Wswitch-enum
2452 Warn whenever a @code{switch} statement has an index of enumerated type
2453 and lacks a @code{case} for one or more of the named codes of that
2454 enumeration. @code{case} labels outside the enumeration range also
2455 provoke warnings when this option is used.
2459 Warn if any trigraphs are encountered that might change the meaning of
2460 the program (trigraphs within comments are not warned about).
2462 @item -Wunused-function
2463 @opindex Wunused-function
2464 Warn whenever a static function is declared but not defined or a
2465 non\-inline static function is unused.
2467 @item -Wunused-label
2468 @opindex Wunused-label
2469 Warn whenever a label is declared but not used.
2471 To suppress this warning use the @samp{unused} attribute
2472 (@pxref{Variable Attributes}).
2474 @item -Wunused-parameter
2475 @opindex Wunused-parameter
2476 Warn whenever a function parameter is unused aside from its declaration.
2478 To suppress this warning use the @samp{unused} attribute
2479 (@pxref{Variable Attributes}).
2481 @item -Wunused-variable
2482 @opindex Wunused-variable
2483 Warn whenever a local variable or non-constant static variable is unused
2484 aside from its declaration
2486 To suppress this warning use the @samp{unused} attribute
2487 (@pxref{Variable Attributes}).
2489 @item -Wunused-value
2490 @opindex Wunused-value
2491 Warn whenever a statement computes a result that is explicitly not used.
2493 To suppress this warning cast the expression to @samp{void}.
2497 All the above @option{-Wunused} options combined.
2499 In order to get a warning about an unused function parameter, you must
2500 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2501 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2503 @item -Wuninitialized
2504 @opindex Wuninitialized
2505 Warn if an automatic variable is used without first being initialized or
2506 if a variable may be clobbered by a @code{setjmp} call.
2508 These warnings are possible only in optimizing compilation,
2509 because they require data flow information that is computed only
2510 when optimizing. If you don't specify @option{-O}, you simply won't
2513 If you want to warn about code which uses the uninitialized value of the
2514 variable in its own initializer, use the @option{-Winit-self} option.
2516 These warnings occur only for variables that are candidates for
2517 register allocation. Therefore, they do not occur for a variable that
2518 is declared @code{volatile}, or whose address is taken, or whose size
2519 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2520 structures, unions or arrays, even when they are in registers.
2522 Note that there may be no warning about a variable that is used only
2523 to compute a value that itself is never used, because such
2524 computations may be deleted by data flow analysis before the warnings
2527 These warnings are made optional because GCC is not smart
2528 enough to see all the reasons why the code might be correct
2529 despite appearing to have an error. Here is one example of how
2550 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2551 always initialized, but GCC doesn't know this. Here is
2552 another common case:
2557 if (change_y) save_y = y, y = new_y;
2559 if (change_y) y = save_y;
2564 This has no bug because @code{save_y} is used only if it is set.
2566 @cindex @code{longjmp} warnings
2567 This option also warns when a non-volatile automatic variable might be
2568 changed by a call to @code{longjmp}. These warnings as well are possible
2569 only in optimizing compilation.
2571 The compiler sees only the calls to @code{setjmp}. It cannot know
2572 where @code{longjmp} will be called; in fact, a signal handler could
2573 call it at any point in the code. As a result, you may get a warning
2574 even when there is in fact no problem because @code{longjmp} cannot
2575 in fact be called at the place which would cause a problem.
2577 Some spurious warnings can be avoided if you declare all the functions
2578 you use that never return as @code{noreturn}. @xref{Function
2581 @item -Wunknown-pragmas
2582 @opindex Wunknown-pragmas
2583 @cindex warning for unknown pragmas
2584 @cindex unknown pragmas, warning
2585 @cindex pragmas, warning of unknown
2586 Warn when a #pragma directive is encountered which is not understood by
2587 GCC@. If this command line option is used, warnings will even be issued
2588 for unknown pragmas in system header files. This is not the case if
2589 the warnings were only enabled by the @option{-Wall} command line option.
2591 @item -Wstrict-aliasing
2592 @opindex Wstrict-aliasing
2593 This option is only active when @option{-fstrict-aliasing} is active.
2594 It warns about code which might break the strict aliasing rules that the
2595 compiler is using for optimization. The warning does not catch all
2596 cases, but does attempt to catch the more common pitfalls. It is
2597 included in @option{-Wall}.
2599 @item -Wstrict-aliasing=2
2600 @opindex Wstrict-aliasing=2
2601 This option is only active when @option{-fstrict-aliasing} is active.
2602 It warns about all code which might break the strict aliasing rules that the
2603 compiler is using for optimization. This warning catches all cases, but
2604 it will also give a warning for some ambiguous cases that are safe.
2608 All of the above @samp{-W} options combined. This enables all the
2609 warnings about constructions that some users consider questionable, and
2610 that are easy to avoid (or modify to prevent the warning), even in
2611 conjunction with macros. This also enables some language-specific
2612 warnings described in @ref{C++ Dialect Options} and
2613 @ref{Objective-C and Objective-C++ Dialect Options}.
2616 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2617 Some of them warn about constructions that users generally do not
2618 consider questionable, but which occasionally you might wish to check
2619 for; others warn about constructions that are necessary or hard to avoid
2620 in some cases, and there is no simple way to modify the code to suppress
2627 (This option used to be called @option{-W}. The older name is still
2628 supported, but the newer name is more descriptive.) Print extra warning
2629 messages for these events:
2633 A function can return either with or without a value. (Falling
2634 off the end of the function body is considered returning without
2635 a value.) For example, this function would evoke such a
2649 An expression-statement or the left-hand side of a comma expression
2650 contains no side effects.
2651 To suppress the warning, cast the unused expression to void.
2652 For example, an expression such as @samp{x[i,j]} will cause a warning,
2653 but @samp{x[(void)i,j]} will not.
2656 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2659 Storage-class specifiers like @code{static} are not the first things in
2660 a declaration. According to the C Standard, this usage is obsolescent.
2663 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2667 A comparison between signed and unsigned values could produce an
2668 incorrect result when the signed value is converted to unsigned.
2669 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2672 An aggregate has an initializer which does not initialize all members.
2673 This warning can be independently controlled by
2674 @option{-Wmissing-field-initializers}.
2677 A function parameter is declared without a type specifier in K&R-style
2685 An empty body occurs in an @samp{if} or @samp{else} statement.
2688 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2689 @samp{>}, or @samp{>=}.
2692 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2695 Any of several floating-point events that often indicate errors, such as
2696 overflow, underflow, loss of precision, etc.
2698 @item @r{(C++ only)}
2699 An enumerator and a non-enumerator both appear in a conditional expression.
2701 @item @r{(C++ only)}
2702 A non-static reference or non-static @samp{const} member appears in a
2703 class without constructors.
2705 @item @r{(C++ only)}
2706 Ambiguous virtual bases.
2708 @item @r{(C++ only)}
2709 Subscripting an array which has been declared @samp{register}.
2711 @item @r{(C++ only)}
2712 Taking the address of a variable which has been declared @samp{register}.
2714 @item @r{(C++ only)}
2715 A base class is not initialized in a derived class' copy constructor.
2718 @item -Wno-div-by-zero
2719 @opindex Wno-div-by-zero
2720 @opindex Wdiv-by-zero
2721 Do not warn about compile-time integer division by zero. Floating point
2722 division by zero is not warned about, as it can be a legitimate way of
2723 obtaining infinities and NaNs.
2725 @item -Wsystem-headers
2726 @opindex Wsystem-headers
2727 @cindex warnings from system headers
2728 @cindex system headers, warnings from
2729 Print warning messages for constructs found in system header files.
2730 Warnings from system headers are normally suppressed, on the assumption
2731 that they usually do not indicate real problems and would only make the
2732 compiler output harder to read. Using this command line option tells
2733 GCC to emit warnings from system headers as if they occurred in user
2734 code. However, note that using @option{-Wall} in conjunction with this
2735 option will @emph{not} warn about unknown pragmas in system
2736 headers---for that, @option{-Wunknown-pragmas} must also be used.
2739 @opindex Wfloat-equal
2740 Warn if floating point values are used in equality comparisons.
2742 The idea behind this is that sometimes it is convenient (for the
2743 programmer) to consider floating-point values as approximations to
2744 infinitely precise real numbers. If you are doing this, then you need
2745 to compute (by analyzing the code, or in some other way) the maximum or
2746 likely maximum error that the computation introduces, and allow for it
2747 when performing comparisons (and when producing output, but that's a
2748 different problem). In particular, instead of testing for equality, you
2749 would check to see whether the two values have ranges that overlap; and
2750 this is done with the relational operators, so equality comparisons are
2753 @item -Wtraditional @r{(C only)}
2754 @opindex Wtraditional
2755 Warn about certain constructs that behave differently in traditional and
2756 ISO C@. Also warn about ISO C constructs that have no traditional C
2757 equivalent, and/or problematic constructs which should be avoided.
2761 Macro parameters that appear within string literals in the macro body.
2762 In traditional C macro replacement takes place within string literals,
2763 but does not in ISO C@.
2766 In traditional C, some preprocessor directives did not exist.
2767 Traditional preprocessors would only consider a line to be a directive
2768 if the @samp{#} appeared in column 1 on the line. Therefore
2769 @option{-Wtraditional} warns about directives that traditional C
2770 understands but would ignore because the @samp{#} does not appear as the
2771 first character on the line. It also suggests you hide directives like
2772 @samp{#pragma} not understood by traditional C by indenting them. Some
2773 traditional implementations would not recognize @samp{#elif}, so it
2774 suggests avoiding it altogether.
2777 A function-like macro that appears without arguments.
2780 The unary plus operator.
2783 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2784 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2785 constants.) Note, these suffixes appear in macros defined in the system
2786 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2787 Use of these macros in user code might normally lead to spurious
2788 warnings, however GCC's integrated preprocessor has enough context to
2789 avoid warning in these cases.
2792 A function declared external in one block and then used after the end of
2796 A @code{switch} statement has an operand of type @code{long}.
2799 A non-@code{static} function declaration follows a @code{static} one.
2800 This construct is not accepted by some traditional C compilers.
2803 The ISO type of an integer constant has a different width or
2804 signedness from its traditional type. This warning is only issued if
2805 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2806 typically represent bit patterns, are not warned about.
2809 Usage of ISO string concatenation is detected.
2812 Initialization of automatic aggregates.
2815 Identifier conflicts with labels. Traditional C lacks a separate
2816 namespace for labels.
2819 Initialization of unions. If the initializer is zero, the warning is
2820 omitted. This is done under the assumption that the zero initializer in
2821 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2822 initializer warnings and relies on default initialization to zero in the
2826 Conversions by prototypes between fixed/floating point values and vice
2827 versa. The absence of these prototypes when compiling with traditional
2828 C would cause serious problems. This is a subset of the possible
2829 conversion warnings, for the full set use @option{-Wconversion}.
2832 Use of ISO C style function definitions. This warning intentionally is
2833 @emph{not} issued for prototype declarations or variadic functions
2834 because these ISO C features will appear in your code when using
2835 libiberty's traditional C compatibility macros, @code{PARAMS} and
2836 @code{VPARAMS}. This warning is also bypassed for nested functions
2837 because that feature is already a GCC extension and thus not relevant to
2838 traditional C compatibility.
2841 @item -Wdeclaration-after-statement @r{(C only)}
2842 @opindex Wdeclaration-after-statement
2843 Warn when a declaration is found after a statement in a block. This
2844 construct, known from C++, was introduced with ISO C99 and is by default
2845 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2846 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2850 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2852 @item -Wendif-labels
2853 @opindex Wendif-labels
2854 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2858 Warn whenever a local variable shadows another local variable, parameter or
2859 global variable or whenever a built-in function is shadowed.
2861 @item -Wlarger-than-@var{len}
2862 @opindex Wlarger-than
2863 Warn whenever an object of larger than @var{len} bytes is defined.
2865 @item -Wpointer-arith
2866 @opindex Wpointer-arith
2867 Warn about anything that depends on the ``size of'' a function type or
2868 of @code{void}. GNU C assigns these types a size of 1, for
2869 convenience in calculations with @code{void *} pointers and pointers
2872 @item -Wbad-function-cast @r{(C only)}
2873 @opindex Wbad-function-cast
2874 Warn whenever a function call is cast to a non-matching type.
2875 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2879 Warn whenever a pointer is cast so as to remove a type qualifier from
2880 the target type. For example, warn if a @code{const char *} is cast
2881 to an ordinary @code{char *}.
2884 @opindex Wcast-align
2885 Warn whenever a pointer is cast such that the required alignment of the
2886 target is increased. For example, warn if a @code{char *} is cast to
2887 an @code{int *} on machines where integers can only be accessed at
2888 two- or four-byte boundaries.
2890 @item -Wwrite-strings
2891 @opindex Wwrite-strings
2892 When compiling C, give string constants the type @code{const
2893 char[@var{length}]} so that
2894 copying the address of one into a non-@code{const} @code{char *}
2895 pointer will get a warning; when compiling C++, warn about the
2896 deprecated conversion from string constants to @code{char *}.
2897 These warnings will help you find at
2898 compile time code that can try to write into a string constant, but
2899 only if you have been very careful about using @code{const} in
2900 declarations and prototypes. Otherwise, it will just be a nuisance;
2901 this is why we did not make @option{-Wall} request these warnings.
2904 @opindex Wconversion
2905 Warn if a prototype causes a type conversion that is different from what
2906 would happen to the same argument in the absence of a prototype. This
2907 includes conversions of fixed point to floating and vice versa, and
2908 conversions changing the width or signedness of a fixed point argument
2909 except when the same as the default promotion.
2911 Also, warn if a negative integer constant expression is implicitly
2912 converted to an unsigned type. For example, warn about the assignment
2913 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2914 casts like @code{(unsigned) -1}.
2916 @item -Wsign-compare
2917 @opindex Wsign-compare
2918 @cindex warning for comparison of signed and unsigned values
2919 @cindex comparison of signed and unsigned values, warning
2920 @cindex signed and unsigned values, comparison warning
2921 Warn when a comparison between signed and unsigned values could produce
2922 an incorrect result when the signed value is converted to unsigned.
2923 This warning is also enabled by @option{-Wextra}; to get the other warnings
2924 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2926 @item -Waggregate-return
2927 @opindex Waggregate-return
2928 Warn if any functions that return structures or unions are defined or
2929 called. (In languages where you can return an array, this also elicits
2932 @item -Wstrict-prototypes @r{(C only)}
2933 @opindex Wstrict-prototypes
2934 Warn if a function is declared or defined without specifying the
2935 argument types. (An old-style function definition is permitted without
2936 a warning if preceded by a declaration which specifies the argument
2939 @item -Wold-style-definition @r{(C only)}
2940 @opindex Wold-style-definition
2941 Warn if an old-style function definition is used. A warning is given
2942 even if there is a previous prototype.
2944 @item -Wmissing-prototypes @r{(C only)}
2945 @opindex Wmissing-prototypes
2946 Warn if a global function is defined without a previous prototype
2947 declaration. This warning is issued even if the definition itself
2948 provides a prototype. The aim is to detect global functions that fail
2949 to be declared in header files.
2951 @item -Wmissing-declarations @r{(C only)}
2952 @opindex Wmissing-declarations
2953 Warn if a global function is defined without a previous declaration.
2954 Do so even if the definition itself provides a prototype.
2955 Use this option to detect global functions that are not declared in
2958 @item -Wmissing-field-initializers
2959 @opindex Wmissing-field-initializers
2962 Warn if a structure's initializer has some fields missing. For
2963 example, the following code would cause such a warning, because
2964 @code{x.h} is implicitly zero:
2967 struct s @{ int f, g, h; @};
2968 struct s x = @{ 3, 4 @};
2971 This option does not warn about designated initializers, so the following
2972 modification would not trigger a warning:
2975 struct s @{ int f, g, h; @};
2976 struct s x = @{ .f = 3, .g = 4 @};
2979 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
2980 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
2982 @item -Wmissing-noreturn
2983 @opindex Wmissing-noreturn
2984 Warn about functions which might be candidates for attribute @code{noreturn}.
2985 Note these are only possible candidates, not absolute ones. Care should
2986 be taken to manually verify functions actually do not ever return before
2987 adding the @code{noreturn} attribute, otherwise subtle code generation
2988 bugs could be introduced. You will not get a warning for @code{main} in
2989 hosted C environments.
2991 @item -Wmissing-format-attribute
2992 @opindex Wmissing-format-attribute
2994 If @option{-Wformat} is enabled, also warn about functions which might be
2995 candidates for @code{format} attributes. Note these are only possible
2996 candidates, not absolute ones. GCC will guess that @code{format}
2997 attributes might be appropriate for any function that calls a function
2998 like @code{vprintf} or @code{vscanf}, but this might not always be the
2999 case, and some functions for which @code{format} attributes are
3000 appropriate may not be detected. This option has no effect unless
3001 @option{-Wformat} is enabled (possibly by @option{-Wall}).
3003 @item -Wno-multichar
3004 @opindex Wno-multichar
3006 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3007 Usually they indicate a typo in the user's code, as they have
3008 implementation-defined values, and should not be used in portable code.
3010 @item -Wno-deprecated-declarations
3011 @opindex Wno-deprecated-declarations
3012 Do not warn about uses of functions, variables, and types marked as
3013 deprecated by using the @code{deprecated} attribute.
3014 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3015 @pxref{Type Attributes}.)
3019 Warn if a structure is given the packed attribute, but the packed
3020 attribute has no effect on the layout or size of the structure.
3021 Such structures may be mis-aligned for little benefit. For
3022 instance, in this code, the variable @code{f.x} in @code{struct bar}
3023 will be misaligned even though @code{struct bar} does not itself
3024 have the packed attribute:
3031 @} __attribute__((packed));
3041 Warn if padding is included in a structure, either to align an element
3042 of the structure or to align the whole structure. Sometimes when this
3043 happens it is possible to rearrange the fields of the structure to
3044 reduce the padding and so make the structure smaller.
3046 @item -Wredundant-decls
3047 @opindex Wredundant-decls
3048 Warn if anything is declared more than once in the same scope, even in
3049 cases where multiple declaration is valid and changes nothing.
3051 @item -Wnested-externs @r{(C only)}
3052 @opindex Wnested-externs
3053 Warn if an @code{extern} declaration is encountered within a function.
3055 @item -Wunreachable-code
3056 @opindex Wunreachable-code
3057 Warn if the compiler detects that code will never be executed.
3059 This option is intended to warn when the compiler detects that at
3060 least a whole line of source code will never be executed, because
3061 some condition is never satisfied or because it is after a
3062 procedure that never returns.
3064 It is possible for this option to produce a warning even though there
3065 are circumstances under which part of the affected line can be executed,
3066 so care should be taken when removing apparently-unreachable code.
3068 For instance, when a function is inlined, a warning may mean that the
3069 line is unreachable in only one inlined copy of the function.
3071 This option is not made part of @option{-Wall} because in a debugging
3072 version of a program there is often substantial code which checks
3073 correct functioning of the program and is, hopefully, unreachable
3074 because the program does work. Another common use of unreachable
3075 code is to provide behavior which is selectable at compile-time.
3079 Warn if a function can not be inlined and it was declared as inline.
3080 Even with this option, the compiler will not warn about failures to
3081 inline functions declared in system headers.
3083 The compiler uses a variety of heuristics to determine whether or not
3084 to inline a function. For example, the compiler takes into account
3085 the size of the function being inlined and the the amount of inlining
3086 that has already been done in the current function. Therefore,
3087 seemingly insignificant changes in the source program can cause the
3088 warnings produced by @option{-Winline} to appear or disappear.
3090 @item -Wno-invalid-offsetof @r{(C++ only)}
3091 @opindex Wno-invalid-offsetof
3092 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3093 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3094 to a non-POD type is undefined. In existing C++ implementations,
3095 however, @samp{offsetof} typically gives meaningful results even when
3096 applied to certain kinds of non-POD types. (Such as a simple
3097 @samp{struct} that fails to be a POD type only by virtue of having a
3098 constructor.) This flag is for users who are aware that they are
3099 writing nonportable code and who have deliberately chosen to ignore the
3102 The restrictions on @samp{offsetof} may be relaxed in a future version
3103 of the C++ standard.
3106 @opindex Winvalid-pch
3107 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3108 the search path but can't be used.
3112 @opindex Wno-long-long
3113 Warn if @samp{long long} type is used. This is default. To inhibit
3114 the warning messages, use @option{-Wno-long-long}. Flags
3115 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3116 only when @option{-pedantic} flag is used.
3118 @item -Wvariadic-macros
3119 @opindex Wvariadic-macros
3120 @opindex Wno-variadic-macros
3121 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3122 alternate syntax when in pedantic ISO C99 mode. This is default.
3123 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3125 @item -Wdisabled-optimization
3126 @opindex Wdisabled-optimization
3127 Warn if a requested optimization pass is disabled. This warning does
3128 not generally indicate that there is anything wrong with your code; it
3129 merely indicates that GCC's optimizers were unable to handle the code
3130 effectively. Often, the problem is that your code is too big or too
3131 complex; GCC will refuse to optimize programs when the optimization
3132 itself is likely to take inordinate amounts of time.
3136 Make all warnings into errors.
3139 @node Debugging Options
3140 @section Options for Debugging Your Program or GCC
3141 @cindex options, debugging
3142 @cindex debugging information options
3144 GCC has various special options that are used for debugging
3145 either your program or GCC:
3150 Produce debugging information in the operating system's native format
3151 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3154 On most systems that use stabs format, @option{-g} enables use of extra
3155 debugging information that only GDB can use; this extra information
3156 makes debugging work better in GDB but will probably make other debuggers
3158 refuse to read the program. If you want to control for certain whether
3159 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3160 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3162 GCC allows you to use @option{-g} with
3163 @option{-O}. The shortcuts taken by optimized code may occasionally
3164 produce surprising results: some variables you declared may not exist
3165 at all; flow of control may briefly move where you did not expect it;
3166 some statements may not be executed because they compute constant
3167 results or their values were already at hand; some statements may
3168 execute in different places because they were moved out of loops.
3170 Nevertheless it proves possible to debug optimized output. This makes
3171 it reasonable to use the optimizer for programs that might have bugs.
3173 The following options are useful when GCC is generated with the
3174 capability for more than one debugging format.
3178 Produce debugging information for use by GDB@. This means to use the
3179 most expressive format available (DWARF 2, stabs, or the native format
3180 if neither of those are supported), including GDB extensions if at all
3185 Produce debugging information in stabs format (if that is supported),
3186 without GDB extensions. This is the format used by DBX on most BSD
3187 systems. On MIPS, Alpha and System V Release 4 systems this option
3188 produces stabs debugging output which is not understood by DBX or SDB@.
3189 On System V Release 4 systems this option requires the GNU assembler.
3191 @item -feliminate-unused-debug-symbols
3192 @opindex feliminate-unused-debug-symbols
3193 Produce debugging information in stabs format (if that is supported),
3194 for only symbols that are actually used.
3198 Produce debugging information in stabs format (if that is supported),
3199 using GNU extensions understood only by the GNU debugger (GDB)@. The
3200 use of these extensions is likely to make other debuggers crash or
3201 refuse to read the program.
3205 Produce debugging information in COFF format (if that is supported).
3206 This is the format used by SDB on most System V systems prior to
3211 Produce debugging information in XCOFF format (if that is supported).
3212 This is the format used by the DBX debugger on IBM RS/6000 systems.
3216 Produce debugging information in XCOFF format (if that is supported),
3217 using GNU extensions understood only by the GNU debugger (GDB)@. The
3218 use of these extensions is likely to make other debuggers crash or
3219 refuse to read the program, and may cause assemblers other than the GNU
3220 assembler (GAS) to fail with an error.
3224 Produce debugging information in DWARF version 2 format (if that is
3225 supported). This is the format used by DBX on IRIX 6. With this
3226 option, GCC uses features of DWARF version 3 when they are useful;
3227 version 3 is upward compatible with version 2, but may still cause
3228 problems for older debuggers.
3232 Produce debugging information in VMS debug format (if that is
3233 supported). This is the format used by DEBUG on VMS systems.
3236 @itemx -ggdb@var{level}
3237 @itemx -gstabs@var{level}
3238 @itemx -gcoff@var{level}
3239 @itemx -gxcoff@var{level}
3240 @itemx -gvms@var{level}
3241 Request debugging information and also use @var{level} to specify how
3242 much information. The default level is 2.
3244 Level 1 produces minimal information, enough for making backtraces in
3245 parts of the program that you don't plan to debug. This includes
3246 descriptions of functions and external variables, but no information
3247 about local variables and no line numbers.
3249 Level 3 includes extra information, such as all the macro definitions
3250 present in the program. Some debuggers support macro expansion when
3251 you use @option{-g3}.
3253 @option{-gdwarf-2} does not accept a concatenated debug level, because
3254 GCC used to support an option @option{-gdwarf} that meant to generate
3255 debug information in version 1 of the DWARF format (which is very
3256 different from version 2), and it would have been too confusing. That
3257 debug format is long obsolete, but the option cannot be changed now.
3258 Instead use an additional @option{-g@var{level}} option to change the
3259 debug level for DWARF2.
3261 @item -feliminate-dwarf2-dups
3262 @opindex feliminate-dwarf2-dups
3263 Compress DWARF2 debugging information by eliminating duplicated
3264 information about each symbol. This option only makes sense when
3265 generating DWARF2 debugging information with @option{-gdwarf-2}.
3267 @cindex @command{prof}
3270 Generate extra code to write profile information suitable for the
3271 analysis program @command{prof}. You must use this option when compiling
3272 the source files you want data about, and you must also use it when
3275 @cindex @command{gprof}
3278 Generate extra code to write profile information suitable for the
3279 analysis program @command{gprof}. You must use this option when compiling
3280 the source files you want data about, and you must also use it when
3285 Makes the compiler print out each function name as it is compiled, and
3286 print some statistics about each pass when it finishes.
3289 @opindex ftime-report
3290 Makes the compiler print some statistics about the time consumed by each
3291 pass when it finishes.
3294 @opindex fmem-report
3295 Makes the compiler print some statistics about permanent memory
3296 allocation when it finishes.
3298 @item -fprofile-arcs
3299 @opindex fprofile-arcs
3300 Add code so that program flow @dfn{arcs} are instrumented. During
3301 execution the program records how many times each branch and call is
3302 executed and how many times it is taken or returns. When the compiled
3303 program exits it saves this data to a file called
3304 @file{@var{auxname}.gcda} for each source file. The data may be used for
3305 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3306 test coverage analysis (@option{-ftest-coverage}). Each object file's
3307 @var{auxname} is generated from the name of the output file, if
3308 explicitly specified and it is not the final executable, otherwise it is
3309 the basename of the source file. In both cases any suffix is removed
3310 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3311 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3316 Compile the source files with @option{-fprofile-arcs} plus optimization
3317 and code generation options. For test coverage analysis, use the
3318 additional @option{-ftest-coverage} option. You do not need to profile
3319 every source file in a program.
3322 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3323 (the latter implies the former).
3326 Run the program on a representative workload to generate the arc profile
3327 information. This may be repeated any number of times. You can run
3328 concurrent instances of your program, and provided that the file system
3329 supports locking, the data files will be correctly updated. Also
3330 @code{fork} calls are detected and correctly handled (double counting
3334 For profile-directed optimizations, compile the source files again with
3335 the same optimization and code generation options plus
3336 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3337 Control Optimization}).
3340 For test coverage analysis, use @command{gcov} to produce human readable
3341 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3342 @command{gcov} documentation for further information.
3346 With @option{-fprofile-arcs}, for each function of your program GCC
3347 creates a program flow graph, then finds a spanning tree for the graph.
3348 Only arcs that are not on the spanning tree have to be instrumented: the
3349 compiler adds code to count the number of times that these arcs are
3350 executed. When an arc is the only exit or only entrance to a block, the
3351 instrumentation code can be added to the block; otherwise, a new basic
3352 block must be created to hold the instrumentation code.
3354 @item -ftree-based-profiling
3355 @opindex ftree-based-profiling
3356 This option is used in addition to @option{-fprofile-arcs} or
3357 @option{-fbranch-probabilities} to control whether those optimizations
3358 are performed on a tree-based or rtl-based internal representation.
3359 If you use this option when compiling with @option{-fprofile-arcs},
3360 you must also use it when compiling later with @option{-fbranch-probabilities}.
3361 Currently the tree-based optimization is in an early stage of
3362 development, and this option is recommended only for those people
3363 working on improving it.
3366 @item -ftest-coverage
3367 @opindex ftest-coverage
3368 Produce a notes file that the @command{gcov} code-coverage utility
3369 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3370 show program coverage. Each source file's note file is called
3371 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3372 above for a description of @var{auxname} and instructions on how to
3373 generate test coverage data. Coverage data will match the source files
3374 more closely, if you do not optimize.
3376 @item -d@var{letters}
3377 @item -fdump-rtl-@var{pass}
3379 Says to make debugging dumps during compilation at times specified by
3380 @var{letters}. This is used for debugging the RTL-based passes of the
3381 compiler. The file names for most of the dumps are made by appending a
3382 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3383 from the name of the output file, if explicitly specified and it is not
3384 an executable, otherwise it is the basename of the source file.
3386 Most debug dumps can be enabled either passing a letter to the @option{-d}
3387 option, or with a long @option{-fdump-rtl} switch; here are the possible
3388 letters for use in @var{letters} and @var{pass}, and their meanings:
3393 Annotate the assembler output with miscellaneous debugging information.
3396 @itemx -fdump-rtl-bp
3398 @opindex fdump-rtl-bp
3399 Dump after computing branch probabilities, to @file{@var{file}.09.bp}.
3402 @itemx -fdump-rtl-bbro
3404 @opindex fdump-rtl-bbro
3405 Dump after block reordering, to @file{@var{file}.30.bbro}.
3408 @itemx -fdump-rtl-combine
3410 @opindex fdump-rtl-combine
3411 Dump after instruction combination, to the file @file{@var{file}.17.combine}.
3414 @itemx -fdump-rtl-ce1
3415 @itemx -fdump-rtl-ce2
3417 @opindex fdump-rtl-ce1
3418 @opindex fdump-rtl-ce2
3419 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3420 first if conversion, to the file @file{@var{file}.11.ce1}. @option{-dC}
3421 and @option{-fdump-rtl-ce2} enable dumping after the second if
3422 conversion, to the file @file{@var{file}.18.ce2}.
3425 @itemx -fdump-rtl-btl
3426 @itemx -fdump-rtl-dbr
3428 @opindex fdump-rtl-btl
3429 @opindex fdump-rtl-dbr
3430 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3431 target load optimization, to to @file{@var{file}.31.btl}. @option{-dd}
3432 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3433 scheduling, to @file{@var{file}.36.dbr}.
3437 Dump all macro definitions, at the end of preprocessing, in addition to
3441 @itemx -fdump-rtl-ce3
3443 @opindex fdump-rtl-ce3
3444 Dump after the third if conversion, to @file{@var{file}.28.ce3}.
3447 @itemx -fdump-rtl-cfg
3448 @itemx -fdump-rtl-life
3450 @opindex fdump-rtl-cfg
3451 @opindex fdump-rtl-life
3452 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3453 and data flow analysis, to @file{@var{file}.08.cfg}. @option{-df}
3454 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3455 to @file{@var{file}.16.life}.
3458 @itemx -fdump-rtl-greg
3460 @opindex fdump-rtl-greg
3461 Dump after global register allocation, to @file{@var{file}.23.greg}.
3464 @itemx -fdump-rtl-gcse
3465 @itemx -fdump-rtl-bypass
3467 @opindex fdump-rtl-gcse
3468 @opindex fdump-rtl-bypass
3469 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3470 @file{@var{file}.05.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3471 enable dumping after jump bypassing and control flow optimizations, to
3472 @file{@var{file}.07.bypass}.
3475 @itemx -fdump-rtl-eh
3477 @opindex fdump-rtl-eh
3478 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3481 @itemx -fdump-rtl-sibling
3483 @opindex fdump-rtl-sibling
3484 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3487 @itemx -fdump-rtl-jump
3489 @opindex fdump-rtl-jump
3490 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3493 @itemx -fdump-rtl-stack
3495 @opindex fdump-rtl-stack
3496 Dump after conversion from registers to stack, to @file{@var{file}.33.stack}.
3499 @itemx -fdump-rtl-lreg
3501 @opindex fdump-rtl-lreg
3502 Dump after local register allocation, to @file{@var{file}.22.lreg}.
3505 @itemx -fdump-rtl-loop
3506 @itemx -fdump-rtl-loop2
3508 @opindex fdump-rtl-loop
3509 @opindex fdump-rtl-loop2
3510 @option{-dL} and @option{-fdump-rtl-loop} enable dumping after the first
3511 loop optimization pass, to @file{@var{file}.06.loop}. @option{-dL} and
3512 @option{-fdump-rtl-loop2} enable dumping after the second pass, to
3513 @file{@var{file}.13.loop2}.
3516 @itemx -fdump-rtl-sms
3518 @opindex fdump-rtl-sms
3519 Dump after modulo scheduling, to @file{@var{file}.20.sms}.
3522 @itemx -fdump-rtl-mach
3524 @opindex fdump-rtl-mach
3525 Dump after performing the machine dependent reorganization pass, to
3526 @file{@var{file}.35.mach}.
3529 @itemx -fdump-rtl-rnreg
3531 @opindex fdump-rtl-rnreg
3532 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3535 @itemx -fdump-rtl-regmove
3537 @opindex fdump-rtl-regmove
3538 Dump after the register move pass, to @file{@var{file}.19.regmove}.
3541 @itemx -fdump-rtl-postreload
3543 @opindex fdump-rtl-postreload
3544 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3547 @itemx -fdump-rtl-expand
3549 @opindex fdump-rtl-expand
3550 Dump after RTL generation, to @file{@var{file}.00.expand}.
3553 @itemx -fdump-rtl-sched2
3555 @opindex fdump-rtl-sched2
3556 Dump after the second scheduling pass, to @file{@var{file}.32.sched2}.
3559 @itemx -fdump-rtl-cse
3561 @opindex fdump-rtl-cse
3562 Dump after CSE (including the jump optimization that sometimes follows
3563 CSE), to @file{@var{file}.04.cse}.
3566 @itemx -fdump-rtl-sched
3568 @opindex fdump-rtl-sched
3569 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3572 @itemx -fdump-rtl-cse2
3574 @opindex fdump-rtl-cse2
3575 Dump after the second CSE pass (including the jump optimization that
3576 sometimes follows CSE), to @file{@var{file}.15.cse2}.
3579 @itemx -fdump-rtl-tracer
3581 @opindex fdump-rtl-tracer
3582 Dump after running tracer, to @file{@var{file}.12.tracer}.
3585 @itemx -fdump-rtl-vpt
3586 @itemx -fdump-rtl-vartrack
3588 @opindex fdump-rtl-vpt
3589 @opindex fdump-rtl-vartrack
3590 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3591 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3592 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3593 to @file{@var{file}.34.vartrack}.
3596 @itemx -fdump-rtl-flow2
3598 @opindex fdump-rtl-flow2
3599 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3602 @itemx -fdump-rtl-peephole2
3604 @opindex fdump-rtl-peephole2
3605 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3608 @itemx -fdump-rtl-web
3610 @opindex fdump-rtl-web
3611 Dump after live range splitting, to @file{@var{file}.14.web}.
3614 @itemx -fdump-rtl-all
3616 @opindex fdump-rtl-all
3617 Produce all the dumps listed above.
3621 Produce a core dump whenever an error occurs.
3625 Print statistics on memory usage, at the end of the run, to
3630 Annotate the assembler output with a comment indicating which
3631 pattern and alternative was used. The length of each instruction is
3636 Dump the RTL in the assembler output as a comment before each instruction.
3637 Also turns on @option{-dp} annotation.
3641 For each of the other indicated dump files (either with @option{-d} or
3642 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3643 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3647 Just generate RTL for a function instead of compiling it. Usually used
3648 with @samp{r} (@option{-fdump-rtl-expand}).
3652 Dump debugging information during parsing, to standard error.
3655 @item -fdump-unnumbered
3656 @opindex fdump-unnumbered
3657 When doing debugging dumps (see @option{-d} option above), suppress instruction
3658 numbers and line number note output. This makes it more feasible to
3659 use diff on debugging dumps for compiler invocations with different
3660 options, in particular with and without @option{-g}.
3662 @item -fdump-translation-unit @r{(C and C++ only)}
3663 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3664 @opindex fdump-translation-unit
3665 Dump a representation of the tree structure for the entire translation
3666 unit to a file. The file name is made by appending @file{.tu} to the
3667 source file name. If the @samp{-@var{options}} form is used, @var{options}
3668 controls the details of the dump as described for the
3669 @option{-fdump-tree} options.
3671 @item -fdump-class-hierarchy @r{(C++ only)}
3672 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3673 @opindex fdump-class-hierarchy
3674 Dump a representation of each class's hierarchy and virtual function
3675 table layout to a file. The file name is made by appending @file{.class}
3676 to the source file name. If the @samp{-@var{options}} form is used,
3677 @var{options} controls the details of the dump as described for the
3678 @option{-fdump-tree} options.
3680 @item -fdump-ipa-@var{switch}
3682 Control the dumping at various stages of inter-procedural analysis
3683 language tree to a file. The file name is generated by appending a switch
3684 specific suffix to the source file name. The following dumps are possible:
3688 Enables all inter-procedural analysis dumps; currently the only produced
3689 dump is the @samp{cgraph} dump.
3692 Dumps information about call-graph optimization, unused function removal,
3693 and inlining decisions.
3696 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3697 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3699 Control the dumping at various stages of processing the intermediate
3700 language tree to a file. The file name is generated by appending a switch
3701 specific suffix to the source file name. If the @samp{-@var{options}}
3702 form is used, @var{options} is a list of @samp{-} separated options that
3703 control the details of the dump. Not all options are applicable to all
3704 dumps, those which are not meaningful will be ignored. The following
3705 options are available
3709 Print the address of each node. Usually this is not meaningful as it
3710 changes according to the environment and source file. Its primary use
3711 is for tying up a dump file with a debug environment.
3713 Inhibit dumping of members of a scope or body of a function merely
3714 because that scope has been reached. Only dump such items when they
3715 are directly reachable by some other path. When dumping pretty-printed
3716 trees, this option inhibits dumping the bodies of control structures.
3718 Print a raw representation of the tree. By default, trees are
3719 pretty-printed into a C-like representation.
3721 Enable more detailed dumps (not honored by every dump option).
3723 Enable dumping various statistics about the pass (not honored by every dump
3726 Enable showing basic block boundaries (disabled in raw dumps).
3728 Enable showing virtual operands for every statement.
3730 Enable showing line numbers for statements.
3732 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3734 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3737 The following tree dumps are possible:
3741 Dump before any tree based optimization, to @file{@var{file}.original}.
3744 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3747 Dump after function inlining, to @file{@var{file}.inlined}.
3750 @opindex fdump-tree-gimple
3751 Dump each function before and after the gimplification pass to a file. The
3752 file name is made by appending @file{.gimple} to the source file name.
3755 @opindex fdump-tree-cfg
3756 Dump the control flow graph of each function to a file. The file name is
3757 made by appending @file{.cfg} to the source file name.
3760 @opindex fdump-tree-vcg
3761 Dump the control flow graph of each function to a file in VCG format. The
3762 file name is made by appending @file{.vcg} to the source file name. Note
3763 that if the file contains more than one function, the generated file cannot
3764 be used directly by VCG. You will need to cut and paste each function's
3765 graph into its own separate file first.
3768 @opindex fdump-tree-ch
3769 Dump each function after copying loop headers. The file name is made by
3770 appending @file{.ch} to the source file name.
3773 @opindex fdump-tree-ssa
3774 Dump SSA related information to a file. The file name is made by appending
3775 @file{.ssa} to the source file name.
3778 @opindex fdump-tree-alias
3779 Dump aliasing information for each function. The file name is made by
3780 appending @file{.alias} to the source file name.
3783 @opindex fdump-tree-ccp
3784 Dump each function after CCP. The file name is made by appending
3785 @file{.ccp} to the source file name.
3788 @opindex fdump-tree-pre
3789 Dump trees after partial redundancy elimination. The file name is made
3790 by appending @file{.pre} to the source file name.
3793 @opindex fdump-tree-fre
3794 Dump trees after full redundancy elimination. The file name is made
3795 by appending @file{.fre} to the source file name.
3798 @opindex fdump-tree-dce
3799 Dump each function after dead code elimination. The file name is made by
3800 appending @file{.dce} to the source file name.
3803 @opindex fdump-tree-mudflap
3804 Dump each function after adding mudflap instrumentation. The file name is
3805 made by appending @file{.mudflap} to the source file name.
3808 @opindex fdump-tree-sra
3809 Dump each function after performing scalar replacement of aggregates. The
3810 file name is made by appending @file{.sra} to the source file name.
3813 @opindex fdump-tree-dom
3814 Dump each function after applying dominator tree optimizations. The file
3815 name is made by appending @file{.dom} to the source file name.
3818 @opindex fdump-tree-dse
3819 Dump each function after applying dead store elimination. The file
3820 name is made by appending @file{.dse} to the source file name.
3823 @opindex fdump-tree-phiopt
3824 Dump each function after optimizing PHI nodes into straightline code. The file
3825 name is made by appending @file{.phiopt} to the source file name.
3828 @opindex fdump-tree-forwprop
3829 Dump each function after forward propagating single use variables. The file
3830 name is made by appending @file{.forwprop} to the source file name.
3833 @opindex fdump-tree-copyrename
3834 Dump each function after applying the copy rename optimization. The file
3835 name is made by appending @file{.copyrename} to the source file name.
3838 @opindex fdump-tree-nrv
3839 Dump each function after applying the named return value optimization on
3840 generic trees. The file name is made by appending @file{.nrv} to the source
3844 @opindex fdump-tree-vect
3845 Dump each function after applying vectorization of loops. The file name is
3846 made by appending @file{.vect} to the source file name.
3849 @opindex fdump-tree-all
3850 Enable all the available tree dumps with the flags provided in this option.
3853 @item -frandom-seed=@var{string}
3854 @opindex frandom-string
3855 This option provides a seed that GCC uses when it would otherwise use
3856 random numbers. It is used to generate certain symbol names
3857 that have to be different in every compiled file. It is also used to
3858 place unique stamps in coverage data files and the object files that
3859 produce them. You can use the @option{-frandom-seed} option to produce
3860 reproducibly identical object files.
3862 The @var{string} should be different for every file you compile.
3864 @item -fsched-verbose=@var{n}
3865 @opindex fsched-verbose
3866 On targets that use instruction scheduling, this option controls the
3867 amount of debugging output the scheduler prints. This information is
3868 written to standard error, unless @option{-dS} or @option{-dR} is
3869 specified, in which case it is output to the usual dump
3870 listing file, @file{.sched} or @file{.sched2} respectively. However
3871 for @var{n} greater than nine, the output is always printed to standard
3874 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3875 same information as @option{-dRS}. For @var{n} greater than one, it
3876 also output basic block probabilities, detailed ready list information
3877 and unit/insn info. For @var{n} greater than two, it includes RTL
3878 at abort point, control-flow and regions info. And for @var{n} over
3879 four, @option{-fsched-verbose} also includes dependence info.
3883 Store the usual ``temporary'' intermediate files permanently; place them
3884 in the current directory and name them based on the source file. Thus,
3885 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3886 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3887 preprocessed @file{foo.i} output file even though the compiler now
3888 normally uses an integrated preprocessor.
3892 Report the CPU time taken by each subprocess in the compilation
3893 sequence. For C source files, this is the compiler proper and assembler
3894 (plus the linker if linking is done). The output looks like this:
3901 The first number on each line is the ``user time'', that is time spent
3902 executing the program itself. The second number is ``system time'',
3903 time spent executing operating system routines on behalf of the program.
3904 Both numbers are in seconds.
3906 @item -fvar-tracking
3907 @opindex fvar-tracking
3908 Run variable tracking pass. It computes where variables are stored at each
3909 position in code. Better debugging information is then generated
3910 (if the debugging information format supports this information).
3912 It is enabled by default when compiling with optimization (@option{-Os},
3913 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3914 the debug info format supports it.
3916 @item -print-file-name=@var{library}
3917 @opindex print-file-name
3918 Print the full absolute name of the library file @var{library} that
3919 would be used when linking---and don't do anything else. With this
3920 option, GCC does not compile or link anything; it just prints the
3923 @item -print-multi-directory
3924 @opindex print-multi-directory
3925 Print the directory name corresponding to the multilib selected by any
3926 other switches present in the command line. This directory is supposed
3927 to exist in @env{GCC_EXEC_PREFIX}.
3929 @item -print-multi-lib
3930 @opindex print-multi-lib
3931 Print the mapping from multilib directory names to compiler switches
3932 that enable them. The directory name is separated from the switches by
3933 @samp{;}, and each switch starts with an @samp{@@} instead of the
3934 @samp{-}, without spaces between multiple switches. This is supposed to
3935 ease shell-processing.
3937 @item -print-prog-name=@var{program}
3938 @opindex print-prog-name
3939 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3941 @item -print-libgcc-file-name
3942 @opindex print-libgcc-file-name
3943 Same as @option{-print-file-name=libgcc.a}.
3945 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3946 but you do want to link with @file{libgcc.a}. You can do
3949 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3952 @item -print-search-dirs
3953 @opindex print-search-dirs
3954 Print the name of the configured installation directory and a list of
3955 program and library directories @command{gcc} will search---and don't do anything else.
3957 This is useful when @command{gcc} prints the error message
3958 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3959 To resolve this you either need to put @file{cpp0} and the other compiler
3960 components where @command{gcc} expects to find them, or you can set the environment
3961 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3962 Don't forget the trailing @samp{/}.
3963 @xref{Environment Variables}.
3966 @opindex dumpmachine
3967 Print the compiler's target machine (for example,
3968 @samp{i686-pc-linux-gnu})---and don't do anything else.
3971 @opindex dumpversion
3972 Print the compiler version (for example, @samp{3.0})---and don't do
3977 Print the compiler's built-in specs---and don't do anything else. (This
3978 is used when GCC itself is being built.) @xref{Spec Files}.
3980 @item -feliminate-unused-debug-types
3981 @opindex feliminate-unused-debug-types
3982 Normally, when producing DWARF2 output, GCC will emit debugging
3983 information for all types declared in a compilation
3984 unit, regardless of whether or not they are actually used
3985 in that compilation unit. Sometimes this is useful, such as
3986 if, in the debugger, you want to cast a value to a type that is
3987 not actually used in your program (but is declared). More often,
3988 however, this results in a significant amount of wasted space.
3989 With this option, GCC will avoid producing debug symbol output
3990 for types that are nowhere used in the source file being compiled.
3993 @node Optimize Options
3994 @section Options That Control Optimization
3995 @cindex optimize options
3996 @cindex options, optimization
3998 These options control various sorts of optimizations.
4000 Without any optimization option, the compiler's goal is to reduce the
4001 cost of compilation and to make debugging produce the expected
4002 results. Statements are independent: if you stop the program with a
4003 breakpoint between statements, you can then assign a new value to any
4004 variable or change the program counter to any other statement in the
4005 function and get exactly the results you would expect from the source
4008 Turning on optimization flags makes the compiler attempt to improve
4009 the performance and/or code size at the expense of compilation time
4010 and possibly the ability to debug the program.
4012 The compiler performs optimization based on the knowledge it has of
4013 the program. Optimization levels @option{-O2} and above, in
4014 particular, enable @emph{unit-at-a-time} mode, which allows the
4015 compiler to consider information gained from later functions in
4016 the file when compiling a function. Compiling multiple files at
4017 once to a single output file in @emph{unit-at-a-time} mode allows
4018 the compiler to use information gained from all of the files when
4019 compiling each of them.
4021 Not all optimizations are controlled directly by a flag. Only
4022 optimizations that have a flag are listed.
4029 Optimize. Optimizing compilation takes somewhat more time, and a lot
4030 more memory for a large function.
4032 With @option{-O}, the compiler tries to reduce code size and execution
4033 time, without performing any optimizations that take a great deal of
4036 @option{-O} turns on the following optimization flags:
4037 @gccoptlist{-fdefer-pop @gol
4038 -fmerge-constants @gol
4040 -floop-optimize @gol
4041 -fif-conversion @gol
4042 -fif-conversion2 @gol
4043 -fdelayed-branch @gol
4044 -fguess-branch-probability @gol
4047 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4048 where doing so does not interfere with debugging.
4052 Optimize even more. GCC performs nearly all supported optimizations
4053 that do not involve a space-speed tradeoff. The compiler does not
4054 perform loop unrolling or function inlining when you specify @option{-O2}.
4055 As compared to @option{-O}, this option increases both compilation time
4056 and the performance of the generated code.
4058 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4059 also turns on the following optimization flags:
4060 @gccoptlist{-fforce-mem @gol
4061 -foptimize-sibling-calls @gol
4062 -fstrength-reduce @gol
4063 -fcse-follow-jumps -fcse-skip-blocks @gol
4064 -frerun-cse-after-loop -frerun-loop-opt @gol
4065 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
4066 -fdelete-null-pointer-checks @gol
4067 -fexpensive-optimizations @gol
4069 -fschedule-insns -fschedule-insns2 @gol
4070 -fsched-interblock -fsched-spec @gol
4073 -freorder-blocks -freorder-functions @gol
4074 -fstrict-aliasing @gol
4075 -funit-at-a-time @gol
4076 -falign-functions -falign-jumps @gol
4077 -falign-loops -falign-labels @gol
4080 Please note the warning under @option{-fgcse} about
4081 invoking @option{-O2} on programs that use computed gotos.
4085 Optimize yet more. @option{-O3} turns on all optimizations specified by
4086 @option{-O2} and also turns on the @option{-finline-functions},
4087 @option{-fweb} and @option{-fgcse-after-reload} options.
4091 Do not optimize. This is the default.
4095 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4096 do not typically increase code size. It also performs further
4097 optimizations designed to reduce code size.
4099 @option{-Os} disables the following optimization flags:
4100 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4101 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
4103 If you use multiple @option{-O} options, with or without level numbers,
4104 the last such option is the one that is effective.
4107 Options of the form @option{-f@var{flag}} specify machine-independent
4108 flags. Most flags have both positive and negative forms; the negative
4109 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4110 below, only one of the forms is listed---the one you typically will
4111 use. You can figure out the other form by either removing @samp{no-}
4114 The following options control specific optimizations. They are either
4115 activated by @option{-O} options or are related to ones that are. You
4116 can use the following flags in the rare cases when ``fine-tuning'' of
4117 optimizations to be performed is desired.
4120 @item -fno-default-inline
4121 @opindex fno-default-inline
4122 Do not make member functions inline by default merely because they are
4123 defined inside the class scope (C++ only). Otherwise, when you specify
4124 @w{@option{-O}}, member functions defined inside class scope are compiled
4125 inline by default; i.e., you don't need to add @samp{inline} in front of
4126 the member function name.
4128 @item -fno-defer-pop
4129 @opindex fno-defer-pop
4130 Always pop the arguments to each function call as soon as that function
4131 returns. For machines which must pop arguments after a function call,
4132 the compiler normally lets arguments accumulate on the stack for several
4133 function calls and pops them all at once.
4135 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4139 Force memory operands to be copied into registers before doing
4140 arithmetic on them. This produces better code by making all memory
4141 references potential common subexpressions. When they are not common
4142 subexpressions, instruction combination should eliminate the separate
4145 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4148 @opindex fforce-addr
4149 Force memory address constants to be copied into registers before
4150 doing arithmetic on them. This may produce better code just as
4151 @option{-fforce-mem} may.
4153 @item -fomit-frame-pointer
4154 @opindex fomit-frame-pointer
4155 Don't keep the frame pointer in a register for functions that
4156 don't need one. This avoids the instructions to save, set up and
4157 restore frame pointers; it also makes an extra register available
4158 in many functions. @strong{It also makes debugging impossible on
4161 On some machines, such as the VAX, this flag has no effect, because
4162 the standard calling sequence automatically handles the frame pointer
4163 and nothing is saved by pretending it doesn't exist. The
4164 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4165 whether a target machine supports this flag. @xref{Registers,,Register
4166 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4168 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4170 @item -foptimize-sibling-calls
4171 @opindex foptimize-sibling-calls
4172 Optimize sibling and tail recursive calls.
4174 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4178 Don't pay attention to the @code{inline} keyword. Normally this option
4179 is used to keep the compiler from expanding any functions inline.
4180 Note that if you are not optimizing, no functions can be expanded inline.
4182 @item -finline-functions
4183 @opindex finline-functions
4184 Integrate all simple functions into their callers. The compiler
4185 heuristically decides which functions are simple enough to be worth
4186 integrating in this way.
4188 If all calls to a given function are integrated, and the function is
4189 declared @code{static}, then the function is normally not output as
4190 assembler code in its own right.
4192 Enabled at level @option{-O3}.
4194 @item -finline-limit=@var{n}
4195 @opindex finline-limit
4196 By default, GCC limits the size of functions that can be inlined. This flag
4197 allows the control of this limit for functions that are explicitly marked as
4198 inline (i.e., marked with the inline keyword or defined within the class
4199 definition in c++). @var{n} is the size of functions that can be inlined in
4200 number of pseudo instructions (not counting parameter handling). The default
4201 value of @var{n} is 600.
4202 Increasing this value can result in more inlined code at
4203 the cost of compilation time and memory consumption. Decreasing usually makes
4204 the compilation faster and less code will be inlined (which presumably
4205 means slower programs). This option is particularly useful for programs that
4206 use inlining heavily such as those based on recursive templates with C++.
4208 Inlining is actually controlled by a number of parameters, which may be
4209 specified individually by using @option{--param @var{name}=@var{value}}.
4210 The @option{-finline-limit=@var{n}} option sets some of these parameters
4214 @item max-inline-insns-single
4215 is set to @var{n}/2.
4216 @item max-inline-insns-auto
4217 is set to @var{n}/2.
4218 @item min-inline-insns
4219 is set to 130 or @var{n}/4, whichever is smaller.
4220 @item max-inline-insns-rtl
4224 See below for a documentation of the individual
4225 parameters controlling inlining.
4227 @emph{Note:} pseudo instruction represents, in this particular context, an
4228 abstract measurement of function's size. In no way, it represents a count
4229 of assembly instructions and as such its exact meaning might change from one
4230 release to an another.
4232 @item -fkeep-inline-functions
4233 @opindex fkeep-inline-functions
4234 In C, emit @code{static} functions that are declared @code{inline}
4235 into the object file, even if the function has been inlined into all
4236 of its callers. This switch does not affect functions using the
4237 @code{extern inline} extension in GNU C. In C++, emit any and all
4238 inline functions into the object file.
4240 @item -fkeep-static-consts
4241 @opindex fkeep-static-consts
4242 Emit variables declared @code{static const} when optimization isn't turned
4243 on, even if the variables aren't referenced.
4245 GCC enables this option by default. If you want to force the compiler to
4246 check if the variable was referenced, regardless of whether or not
4247 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4249 @item -fmerge-constants
4250 Attempt to merge identical constants (string constants and floating point
4251 constants) across compilation units.
4253 This option is the default for optimized compilation if the assembler and
4254 linker support it. Use @option{-fno-merge-constants} to inhibit this
4257 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4259 @item -fmerge-all-constants
4260 Attempt to merge identical constants and identical variables.
4262 This option implies @option{-fmerge-constants}. In addition to
4263 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4264 arrays or initialized constant variables with integral or floating point
4265 types. Languages like C or C++ require each non-automatic variable to
4266 have distinct location, so using this option will result in non-conforming
4269 @item -fmodulo-sched
4270 @opindex fmodulo-sched
4271 Perform swing modulo scheduling immediately before the first scheduling
4272 pass. This pass looks at innermost loops and reorders their
4273 instructions by overlapping different iterations.
4277 Use a graph coloring register allocator. Currently this option is meant
4278 only for testing. Users should not specify this option, since it is not
4279 yet ready for production use.
4281 @item -fno-branch-count-reg
4282 @opindex fno-branch-count-reg
4283 Do not use ``decrement and branch'' instructions on a count register,
4284 but instead generate a sequence of instructions that decrement a
4285 register, compare it against zero, then branch based upon the result.
4286 This option is only meaningful on architectures that support such
4287 instructions, which include x86, PowerPC, IA-64 and S/390.
4289 The default is @option{-fbranch-count-reg}, enabled when
4290 @option{-fstrength-reduce} is enabled.
4292 @item -fno-function-cse
4293 @opindex fno-function-cse
4294 Do not put function addresses in registers; make each instruction that
4295 calls a constant function contain the function's address explicitly.
4297 This option results in less efficient code, but some strange hacks
4298 that alter the assembler output may be confused by the optimizations
4299 performed when this option is not used.
4301 The default is @option{-ffunction-cse}
4303 @item -fno-zero-initialized-in-bss
4304 @opindex fno-zero-initialized-in-bss
4305 If the target supports a BSS section, GCC by default puts variables that
4306 are initialized to zero into BSS@. This can save space in the resulting
4309 This option turns off this behavior because some programs explicitly
4310 rely on variables going to the data section. E.g., so that the
4311 resulting executable can find the beginning of that section and/or make
4312 assumptions based on that.
4314 The default is @option{-fzero-initialized-in-bss}.
4316 @item -fbounds-check
4317 @opindex fbounds-check
4318 For front-ends that support it, generate additional code to check that
4319 indices used to access arrays are within the declared range. This is
4320 currently only supported by the Java and Fortran front-ends, where
4321 this option defaults to true and false respectively.
4323 @item -fmudflap -fmudflapth -fmudflapir
4327 @cindex bounds checking
4329 For front-ends that support it (C and C++), instrument all risky
4330 pointer/array dereferencing operations, some standard library
4331 string/heap functions, and some other associated constructs with
4332 range/validity tests. Modules so instrumented should be immune to
4333 buffer overflows, invalid heap use, and some other classes of C/C++
4334 programming errors. The instrumentation relies on a separate runtime
4335 library (@file{libmudflap}), which will be linked into a program if
4336 @option{-fmudflap} is given at link time. Run-time behavior of the
4337 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4338 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4341 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4342 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4343 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4344 instrumentation should ignore pointer reads. This produces less
4345 instrumentation (and therefore faster execution) and still provides
4346 some protection against outright memory corrupting writes, but allows
4347 erroneously read data to propagate within a program.
4349 @item -fstrength-reduce
4350 @opindex fstrength-reduce
4351 Perform the optimizations of loop strength reduction and
4352 elimination of iteration variables.
4354 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4356 @item -fthread-jumps
4357 @opindex fthread-jumps
4358 Perform optimizations where we check to see if a jump branches to a
4359 location where another comparison subsumed by the first is found. If
4360 so, the first branch is redirected to either the destination of the
4361 second branch or a point immediately following it, depending on whether
4362 the condition is known to be true or false.
4364 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4366 @item -fcse-follow-jumps
4367 @opindex fcse-follow-jumps
4368 In common subexpression elimination, scan through jump instructions
4369 when the target of the jump is not reached by any other path. For
4370 example, when CSE encounters an @code{if} statement with an
4371 @code{else} clause, CSE will follow the jump when the condition
4374 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4376 @item -fcse-skip-blocks
4377 @opindex fcse-skip-blocks
4378 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4379 follow jumps which conditionally skip over blocks. When CSE
4380 encounters a simple @code{if} statement with no else clause,
4381 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4382 body of the @code{if}.
4384 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4386 @item -frerun-cse-after-loop
4387 @opindex frerun-cse-after-loop
4388 Re-run common subexpression elimination after loop optimizations has been
4391 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4393 @item -frerun-loop-opt
4394 @opindex frerun-loop-opt
4395 Run the loop optimizer twice.
4397 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4401 Perform a global common subexpression elimination pass.
4402 This pass also performs global constant and copy propagation.
4404 @emph{Note:} When compiling a program using computed gotos, a GCC
4405 extension, you may get better runtime performance if you disable
4406 the global common subexpression elimination pass by adding
4407 @option{-fno-gcse} to the command line.
4409 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4413 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4414 attempt to move loads which are only killed by stores into themselves. This
4415 allows a loop containing a load/store sequence to be changed to a load outside
4416 the loop, and a copy/store within the loop.
4418 Enabled by default when gcse is enabled.
4422 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4423 global common subexpression elimination. This pass will attempt to move
4424 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4425 loops containing a load/store sequence can be changed to a load before
4426 the loop and a store after the loop.
4428 Enabled by default when gcse is enabled.
4432 When @option{-fgcse-las} is enabled, the global common subexpression
4433 elimination pass eliminates redundant loads that come after stores to the
4434 same memory location (both partial and full redundancies).
4436 Enabled by default when gcse is enabled.
4438 @item -fgcse-after-reload
4439 @opindex fgcse-after-reload
4440 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4441 pass is performed after reload. The purpose of this pass is to cleanup
4444 @item -floop-optimize
4445 @opindex floop-optimize
4446 Perform loop optimizations: move constant expressions out of loops, simplify
4447 exit test conditions and optionally do strength-reduction as well.
4449 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4451 @item -floop-optimize2
4452 @opindex floop-optimize2
4453 Perform loop optimizations using the new loop optimizer. The optimizations
4454 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4457 @item -fcrossjumping
4458 @opindex crossjumping
4459 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4460 resulting code may or may not perform better than without cross-jumping.
4462 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4464 @item -fif-conversion
4465 @opindex if-conversion
4466 Attempt to transform conditional jumps into branch-less equivalents. This
4467 include use of conditional moves, min, max, set flags and abs instructions, and
4468 some tricks doable by standard arithmetics. The use of conditional execution
4469 on chips where it is available is controlled by @code{if-conversion2}.
4471 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4473 @item -fif-conversion2
4474 @opindex if-conversion2
4475 Use conditional execution (where available) to transform conditional jumps into
4476 branch-less equivalents.
4478 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4480 @item -fdelete-null-pointer-checks
4481 @opindex fdelete-null-pointer-checks
4482 Use global dataflow analysis to identify and eliminate useless checks
4483 for null pointers. The compiler assumes that dereferencing a null
4484 pointer would have halted the program. If a pointer is checked after
4485 it has already been dereferenced, it cannot be null.
4487 In some environments, this assumption is not true, and programs can
4488 safely dereference null pointers. Use
4489 @option{-fno-delete-null-pointer-checks} to disable this optimization
4490 for programs which depend on that behavior.
4492 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4494 @item -fexpensive-optimizations
4495 @opindex fexpensive-optimizations
4496 Perform a number of minor optimizations that are relatively expensive.
4498 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4500 @item -foptimize-register-move
4502 @opindex foptimize-register-move
4504 Attempt to reassign register numbers in move instructions and as
4505 operands of other simple instructions in order to maximize the amount of
4506 register tying. This is especially helpful on machines with two-operand
4509 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4512 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4514 @item -fdelayed-branch
4515 @opindex fdelayed-branch
4516 If supported for the target machine, attempt to reorder instructions
4517 to exploit instruction slots available after delayed branch
4520 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4522 @item -fschedule-insns
4523 @opindex fschedule-insns
4524 If supported for the target machine, attempt to reorder instructions to
4525 eliminate execution stalls due to required data being unavailable. This
4526 helps machines that have slow floating point or memory load instructions
4527 by allowing other instructions to be issued until the result of the load
4528 or floating point instruction is required.
4530 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4532 @item -fschedule-insns2
4533 @opindex fschedule-insns2
4534 Similar to @option{-fschedule-insns}, but requests an additional pass of
4535 instruction scheduling after register allocation has been done. This is
4536 especially useful on machines with a relatively small number of
4537 registers and where memory load instructions take more than one cycle.
4539 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4541 @item -fno-sched-interblock
4542 @opindex fno-sched-interblock
4543 Don't schedule instructions across basic blocks. This is normally
4544 enabled by default when scheduling before register allocation, i.e.@:
4545 with @option{-fschedule-insns} or at @option{-O2} or higher.
4547 @item -fno-sched-spec
4548 @opindex fno-sched-spec
4549 Don't allow speculative motion of non-load instructions. This is normally
4550 enabled by default when scheduling before register allocation, i.e.@:
4551 with @option{-fschedule-insns} or at @option{-O2} or higher.
4553 @item -fsched-spec-load
4554 @opindex fsched-spec-load
4555 Allow speculative motion of some load instructions. This only makes
4556 sense when scheduling before register allocation, i.e.@: with
4557 @option{-fschedule-insns} or at @option{-O2} or higher.
4559 @item -fsched-spec-load-dangerous
4560 @opindex fsched-spec-load-dangerous
4561 Allow speculative motion of more load instructions. This only makes
4562 sense when scheduling before register allocation, i.e.@: with
4563 @option{-fschedule-insns} or at @option{-O2} or higher.
4565 @item -fsched-stalled-insns=@var{n}
4566 @opindex fsched-stalled-insns
4567 Define how many insns (if any) can be moved prematurely from the queue
4568 of stalled insns into the ready list, during the second scheduling pass.
4570 @item -fsched-stalled-insns-dep=@var{n}
4571 @opindex fsched-stalled-insns-dep
4572 Define how many insn groups (cycles) will be examined for a dependency
4573 on a stalled insn that is candidate for premature removal from the queue
4574 of stalled insns. Has an effect only during the second scheduling pass,
4575 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4577 @item -fsched2-use-superblocks
4578 @opindex fsched2-use-superblocks
4579 When scheduling after register allocation, do use superblock scheduling
4580 algorithm. Superblock scheduling allows motion across basic block boundaries
4581 resulting on faster schedules. This option is experimental, as not all machine
4582 descriptions used by GCC model the CPU closely enough to avoid unreliable
4583 results from the algorithm.
4585 This only makes sense when scheduling after register allocation, i.e.@: with
4586 @option{-fschedule-insns2} or at @option{-O2} or higher.
4588 @item -fsched2-use-traces
4589 @opindex fsched2-use-traces
4590 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4591 allocation and additionally perform code duplication in order to increase the
4592 size of superblocks using tracer pass. See @option{-ftracer} for details on
4595 This mode should produce faster but significantly longer programs. Also
4596 without @option{-fbranch-probabilities} the traces constructed may not
4597 match the reality and hurt the performance. This only makes
4598 sense when scheduling after register allocation, i.e.@: with
4599 @option{-fschedule-insns2} or at @option{-O2} or higher.
4601 @item -freschedule-modulo-scheduled-loops
4602 @opindex fscheduling-in-modulo-scheduled-loops
4603 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4604 we may want to prevent the later scheduling passes from changing its schedule, we use this
4605 option to control that.
4607 @item -fcaller-saves
4608 @opindex fcaller-saves
4609 Enable values to be allocated in registers that will be clobbered by
4610 function calls, by emitting extra instructions to save and restore the
4611 registers around such calls. Such allocation is done only when it
4612 seems to result in better code than would otherwise be produced.
4614 This option is always enabled by default on certain machines, usually
4615 those which have no call-preserved registers to use instead.
4617 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4620 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4621 enabled by default at @option{-O} and higher.
4624 Perform Full Redundancy Elimination (FRE) on trees. The difference
4625 between FRE and PRE is that FRE only considers expressions
4626 that are computed on all paths leading to the redundant computation.
4627 This analysis faster than PRE, though it exposes fewer redundancies.
4628 This flag is enabled by default at @option{-O} and higher.
4631 Perform sparse conditional constant propagation (CCP) on trees. This flag
4632 is enabled by default at @option{-O} and higher.
4635 Perform dead code elimination (DCE) on trees. This flag is enabled by
4636 default at @option{-O} and higher.
4638 @item -ftree-dominator-opts
4639 Perform dead code elimination (DCE) on trees. This flag is enabled by
4640 default at @option{-O} and higher.
4643 Perform loop header copying on trees. This is beneficial since it increases
4644 effectiveness of code motion optimizations. It also saves one jump. This flag
4645 is enabled by default at @option{-O} and higher. It is not enabled
4646 for @option{-Os}, since it usually increases code size.
4648 @item -ftree-loop-optimize
4649 Perform loop optimizations on trees. This flag is enabled by default
4650 at @option{-O} and higher.
4652 @item -ftree-loop-linear
4653 Perform linear loop transformations on tree. This flag can improve cache
4654 performance and allow further loop optimizations to take place.
4657 Perform loop invariant motion on trees. This pass moves only invartiants that
4658 would be hard to handle on rtl level (function calls, operations that expand to
4659 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
4660 operands of conditions that are invariant out of the loop, so that we can use
4661 just trivial invariantness analysis in loop unswitching. The pass also includes
4665 Create a canonical counter for number of iterations in the loop for that
4666 determining number of iterations requires complicated analysis. Later
4667 optimizations then may determine the number easily. Useful especially
4668 in connection with unrolling.
4671 Perform induction variable optimizations (strength reduction, induction
4672 variable merging and induction variable elimination) on trees.
4675 Perform scalar replacement of aggregates. This pass replaces structure
4676 references with scalars to prevent committing structures to memory too
4677 early. This flag is enabled by default at @option{-O} and higher.
4679 @item -ftree-copyrename
4680 Perform copy renaming on trees. This pass attempts to rename compiler
4681 temporaries to other variables at copy locations, usually resulting in
4682 variable names which more closely resemble the original variables. This flag
4683 is enabled by default at @option{-O} and higher.
4686 Perform temporary expression replacement during the SSA->normal phase. Single
4687 use/single def temporaries are replaced at their use location with their
4688 defining expression. This results in non-GIMPLE code, but gives the expanders
4689 much more complex trees to work on resulting in better RTL generation. This is
4690 enabled by default at @option{-O} and higher.
4693 Perform live range splitting during the SSA->normal phase. Distinct live
4694 ranges of a variable are split into unique variables, allowing for better
4695 optimization later. This is enabled by default at @option{-O} and higher.
4697 @item -ftree-vectorize
4698 Perform loop vectorization on trees.
4702 Perform tail duplication to enlarge superblock size. This transformation
4703 simplifies the control flow of the function allowing other optimizations to do
4706 @item -funroll-loops
4707 @opindex funroll-loops
4708 Unroll loops whose number of iterations can be determined at compile
4709 time or upon entry to the loop. @option{-funroll-loops} implies both
4710 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4711 option makes code larger, and may or may not make it run faster.
4713 @item -funroll-all-loops
4714 @opindex funroll-all-loops
4715 Unroll all loops, even if their number of iterations is uncertain when
4716 the loop is entered. This usually makes programs run more slowly.
4717 @option{-funroll-all-loops} implies the same options as
4718 @option{-funroll-loops},
4720 @item -fsplit-ivs-in-unroller
4721 @opindex -fsplit-ivs-in-unroller
4722 Enables expressing of values of induction variables in later iterations
4723 of the unrolled loop using the value in the first iteration. This breaks
4724 long dependency chains, thus improving efficiency of the scheduling passes
4725 (for best results, @option{-fweb} should be used as well).
4727 Combination of @option{-fweb} and CSE is often sufficient to obtain the
4728 same effect. However in cases the loop body is more complicated than
4729 a single basic block, this is not reliable. It also does not work at all
4730 on some of the architectures due to restrictions in the CSE pass.
4732 This optimization is enabled by default.
4734 @item -fvariable-expansion-in-unroller
4735 @opindex -fvariable-expansion-in-unroller
4736 With this option, the compiler will create multiple copies of some
4737 local variables when unrolling a loop which can result in superior code.
4739 @item -fprefetch-loop-arrays
4740 @opindex fprefetch-loop-arrays
4741 If supported by the target machine, generate instructions to prefetch
4742 memory to improve the performance of loops that access large arrays.
4744 These options may generate better or worse code; results are highly
4745 dependent on the structure of loops within the source code.
4748 @itemx -fno-peephole2
4749 @opindex fno-peephole
4750 @opindex fno-peephole2
4751 Disable any machine-specific peephole optimizations. The difference
4752 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4753 are implemented in the compiler; some targets use one, some use the
4754 other, a few use both.
4756 @option{-fpeephole} is enabled by default.
4757 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4759 @item -fno-guess-branch-probability
4760 @opindex fno-guess-branch-probability
4761 Do not guess branch probabilities using a randomized model.
4763 Sometimes GCC will opt to use a randomized model to guess branch
4764 probabilities, when none are available from either profiling feedback
4765 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4766 different runs of the compiler on the same program may produce different
4769 In a hard real-time system, people don't want different runs of the
4770 compiler to produce code that has different behavior; minimizing
4771 non-determinism is of paramount import. This switch allows users to
4772 reduce non-determinism, possibly at the expense of inferior
4775 The default is @option{-fguess-branch-probability} at levels
4776 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4778 @item -freorder-blocks
4779 @opindex freorder-blocks
4780 Reorder basic blocks in the compiled function in order to reduce number of
4781 taken branches and improve code locality.
4783 Enabled at levels @option{-O2}, @option{-O3}.
4785 @item -freorder-blocks-and-partition
4786 @opindex freorder-blocks-and-partition
4787 In addition to reordering basic blocks in the compiled function, in order
4788 to reduce number of taken branches, partitions hot and cold basic blocks
4789 into separate sections of the assembly and .o files, to improve
4790 paging and cache locality performance.
4792 This optimization is automatically turned off in the presence of
4793 exception handling, for linkonce sections, for functions with a user-defined
4794 section attribute and on any architecture that does not support named
4797 @item -freorder-functions
4798 @opindex freorder-functions
4799 Reorder basic blocks in the compiled function in order to reduce number of
4800 taken branches and improve code locality. This is implemented by using special
4801 subsections @code{.text.hot} for most frequently executed functions and
4802 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4803 the linker so object file format must support named sections and linker must
4804 place them in a reasonable way.
4806 Also profile feedback must be available in to make this option effective. See
4807 @option{-fprofile-arcs} for details.
4809 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4811 @item -fstrict-aliasing
4812 @opindex fstrict-aliasing
4813 Allows the compiler to assume the strictest aliasing rules applicable to
4814 the language being compiled. For C (and C++), this activates
4815 optimizations based on the type of expressions. In particular, an
4816 object of one type is assumed never to reside at the same address as an
4817 object of a different type, unless the types are almost the same. For
4818 example, an @code{unsigned int} can alias an @code{int}, but not a
4819 @code{void*} or a @code{double}. A character type may alias any other
4822 Pay special attention to code like this:
4835 The practice of reading from a different union member than the one most
4836 recently written to (called ``type-punning'') is common. Even with
4837 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4838 is accessed through the union type. So, the code above will work as
4839 expected. However, this code might not:
4850 Every language that wishes to perform language-specific alias analysis
4851 should define a function that computes, given an @code{tree}
4852 node, an alias set for the node. Nodes in different alias sets are not
4853 allowed to alias. For an example, see the C front-end function
4854 @code{c_get_alias_set}.
4856 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4858 @item -falign-functions
4859 @itemx -falign-functions=@var{n}
4860 @opindex falign-functions
4861 Align the start of functions to the next power-of-two greater than
4862 @var{n}, skipping up to @var{n} bytes. For instance,
4863 @option{-falign-functions=32} aligns functions to the next 32-byte
4864 boundary, but @option{-falign-functions=24} would align to the next
4865 32-byte boundary only if this can be done by skipping 23 bytes or less.
4867 @option{-fno-align-functions} and @option{-falign-functions=1} are
4868 equivalent and mean that functions will not be aligned.
4870 Some assemblers only support this flag when @var{n} is a power of two;
4871 in that case, it is rounded up.
4873 If @var{n} is not specified or is zero, use a machine-dependent default.
4875 Enabled at levels @option{-O2}, @option{-O3}.
4877 @item -falign-labels
4878 @itemx -falign-labels=@var{n}
4879 @opindex falign-labels
4880 Align all branch targets to a power-of-two boundary, skipping up to
4881 @var{n} bytes like @option{-falign-functions}. This option can easily
4882 make code slower, because it must insert dummy operations for when the
4883 branch target is reached in the usual flow of the code.
4885 @option{-fno-align-labels} and @option{-falign-labels=1} are
4886 equivalent and mean that labels will not be aligned.
4888 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4889 are greater than this value, then their values are used instead.
4891 If @var{n} is not specified or is zero, use a machine-dependent default
4892 which is very likely to be @samp{1}, meaning no alignment.
4894 Enabled at levels @option{-O2}, @option{-O3}.
4897 @itemx -falign-loops=@var{n}
4898 @opindex falign-loops
4899 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4900 like @option{-falign-functions}. The hope is that the loop will be
4901 executed many times, which will make up for any execution of the dummy
4904 @option{-fno-align-loops} and @option{-falign-loops=1} are
4905 equivalent and mean that loops will not be aligned.
4907 If @var{n} is not specified or is zero, use a machine-dependent default.
4909 Enabled at levels @option{-O2}, @option{-O3}.
4912 @itemx -falign-jumps=@var{n}
4913 @opindex falign-jumps
4914 Align branch targets to a power-of-two boundary, for branch targets
4915 where the targets can only be reached by jumping, skipping up to @var{n}
4916 bytes like @option{-falign-functions}. In this case, no dummy operations
4919 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4920 equivalent and mean that loops will not be aligned.
4922 If @var{n} is not specified or is zero, use a machine-dependent default.
4924 Enabled at levels @option{-O2}, @option{-O3}.
4926 @item -funit-at-a-time
4927 @opindex funit-at-a-time
4928 Parse the whole compilation unit before starting to produce code.
4929 This allows some extra optimizations to take place but consumes
4930 more memory (in general). There are some compatibility issues
4931 with @emph{unit-at-at-time} mode:
4934 enabling @emph{unit-at-a-time} mode may change the order
4935 in which functions, variables, and top-level @code{asm} statements
4936 are emitted, and will likely break code relying on some particular
4937 ordering. The majority of such top-level @code{asm} statements,
4938 though, can be replaced by @code{section} attributes.
4941 @emph{unit-at-a-time} mode removes unreferenced static variables
4942 and functions are removed. This may result in undefined references
4943 when an @code{asm} statement refers directly to variables or functions
4944 that are otherwise unused. In that case either the variable/function
4945 shall be listed as an operand of the @code{asm} statement operand or,
4946 in the case of top-level @code{asm} statements the attribute @code{used}
4947 shall be used on the declaration.
4950 Static functions now can use non-standard passing conventions that
4951 may break @code{asm} statements calling functions directly. Again,
4952 attribute @code{used} will prevent this behavior.
4955 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
4956 but this scheme may not be supported by future releases of GCC.
4958 Enabled at levels @option{-O2}, @option{-O3}.
4962 Constructs webs as commonly used for register allocation purposes and assign
4963 each web individual pseudo register. This allows the register allocation pass
4964 to operate on pseudos directly, but also strengthens several other optimization
4965 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4966 however, make debugging impossible, since variables will no longer stay in a
4969 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
4970 on targets where the default format for debugging information supports
4973 @item -fno-cprop-registers
4974 @opindex fno-cprop-registers
4975 After register allocation and post-register allocation instruction splitting,
4976 we perform a copy-propagation pass to try to reduce scheduling dependencies
4977 and occasionally eliminate the copy.
4979 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4981 @item -fprofile-generate
4982 @opindex fprofile-generate
4984 Enable options usually used for instrumenting application to produce
4985 profile useful for later recompilation with profile feedback based
4986 optimization. You must use @option{-fprofile-generate} both when
4987 compiling and when linking your program.
4989 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
4992 @opindex fprofile-use
4993 Enable profile feedback directed optimizations, and optimizations
4994 generally profitable only with profile feedback available.
4996 The following options are enabled: @code{-fbranch-probabilities},
4997 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
5001 The following options control compiler behavior regarding floating
5002 point arithmetic. These options trade off between speed and
5003 correctness. All must be specifically enabled.
5007 @opindex ffloat-store
5008 Do not store floating point variables in registers, and inhibit other
5009 options that might change whether a floating point value is taken from a
5012 @cindex floating point precision
5013 This option prevents undesirable excess precision on machines such as
5014 the 68000 where the floating registers (of the 68881) keep more
5015 precision than a @code{double} is supposed to have. Similarly for the
5016 x86 architecture. For most programs, the excess precision does only
5017 good, but a few programs rely on the precise definition of IEEE floating
5018 point. Use @option{-ffloat-store} for such programs, after modifying
5019 them to store all pertinent intermediate computations into variables.
5023 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5024 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5025 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
5027 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
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 for
5034 @item -fno-math-errno
5035 @opindex fno-math-errno
5036 Do not set ERRNO after calling math functions that are executed
5037 with a single instruction, e.g., sqrt. A program that relies on
5038 IEEE exceptions for math error handling may want to use this flag
5039 for speed while maintaining IEEE arithmetic compatibility.
5041 This option should never be turned on by any @option{-O} option since
5042 it can result in incorrect output for programs which depend on
5043 an exact implementation of IEEE or ISO rules/specifications for
5046 The default is @option{-fmath-errno}.
5048 @item -funsafe-math-optimizations
5049 @opindex funsafe-math-optimizations
5050 Allow optimizations for floating-point arithmetic that (a) assume
5051 that arguments and results are valid and (b) may violate IEEE or
5052 ANSI standards. When used at link-time, it may include libraries
5053 or startup files that change the default FPU control word or other
5054 similar optimizations.
5056 This option should never be turned on by any @option{-O} option since
5057 it can result in incorrect output for programs which depend on
5058 an exact implementation of IEEE or ISO rules/specifications for
5061 The default is @option{-fno-unsafe-math-optimizations}.
5063 @item -ffinite-math-only
5064 @opindex ffinite-math-only
5065 Allow optimizations for floating-point arithmetic that assume
5066 that arguments and results are not NaNs or +-Infs.
5068 This option should never be turned on by any @option{-O} option since
5069 it can result in incorrect output for programs which depend on
5070 an exact implementation of IEEE or ISO rules/specifications.
5072 The default is @option{-fno-finite-math-only}.
5074 @item -fno-trapping-math
5075 @opindex fno-trapping-math
5076 Compile code assuming that floating-point operations cannot generate
5077 user-visible traps. These traps include division by zero, overflow,
5078 underflow, inexact result and invalid operation. This option implies
5079 @option{-fno-signaling-nans}. Setting this option may allow faster
5080 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5082 This option should never be turned on by any @option{-O} option since
5083 it can result in incorrect output for programs which depend on
5084 an exact implementation of IEEE or ISO rules/specifications for
5087 The default is @option{-ftrapping-math}.
5089 @item -frounding-math
5090 @opindex frounding-math
5091 Disable transformations and optimizations that assume default floating
5092 point rounding behavior. This is round-to-zero for all floating point
5093 to integer conversions, and round-to-nearest for all other arithmetic
5094 truncations. This option should be specified for programs that change
5095 the FP rounding mode dynamically, or that may be executed with a
5096 non-default rounding mode. This option disables constant folding of
5097 floating point expressions at compile-time (which may be affected by
5098 rounding mode) and arithmetic transformations that are unsafe in the
5099 presence of sign-dependent rounding modes.
5101 The default is @option{-fno-rounding-math}.
5103 This option is experimental and does not currently guarantee to
5104 disable all GCC optimizations that are affected by rounding mode.
5105 Future versions of GCC may provide finer control of this setting
5106 using C99's @code{FENV_ACCESS} pragma. This command line option
5107 will be used to specify the default state for @code{FENV_ACCESS}.
5109 @item -fsignaling-nans
5110 @opindex fsignaling-nans
5111 Compile code assuming that IEEE signaling NaNs may generate user-visible
5112 traps during floating-point operations. Setting this option disables
5113 optimizations that may change the number of exceptions visible with
5114 signaling NaNs. This option implies @option{-ftrapping-math}.
5116 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5119 The default is @option{-fno-signaling-nans}.
5121 This option is experimental and does not currently guarantee to
5122 disable all GCC optimizations that affect signaling NaN behavior.
5124 @item -fsingle-precision-constant
5125 @opindex fsingle-precision-constant
5126 Treat floating point constant as single precision constant instead of
5127 implicitly converting it to double precision constant.
5132 The following options control optimizations that may improve
5133 performance, but are not enabled by any @option{-O} options. This
5134 section includes experimental options that may produce broken code.
5137 @item -fbranch-probabilities
5138 @opindex fbranch-probabilities
5139 After running a program compiled with @option{-fprofile-arcs}
5140 (@pxref{Debugging Options,, Options for Debugging Your Program or
5141 @command{gcc}}), you can compile it a second time using
5142 @option{-fbranch-probabilities}, to improve optimizations based on
5143 the number of times each branch was taken. When the program
5144 compiled with @option{-fprofile-arcs} exits it saves arc execution
5145 counts to a file called @file{@var{sourcename}.gcda} for each source
5146 file The information in this data file is very dependent on the
5147 structure of the generated code, so you must use the same source code
5148 and the same optimization options for both compilations.
5150 With @option{-fbranch-probabilities}, GCC puts a
5151 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5152 These can be used to improve optimization. Currently, they are only
5153 used in one place: in @file{reorg.c}, instead of guessing which path a
5154 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5155 exactly determine which path is taken more often.
5157 @item -fprofile-values
5158 @opindex fprofile-values
5159 If combined with @option{-fprofile-arcs}, it adds code so that some
5160 data about values of expressions in the program is gathered.
5162 With @option{-fbranch-probabilities}, it reads back the data gathered
5163 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5164 notes to instructions for their later usage in optimizations.
5166 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5170 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5171 a code to gather information about values of expressions.
5173 With @option{-fbranch-probabilities}, it reads back the data gathered
5174 and actually performs the optimizations based on them.
5175 Currently the optimizations include specialization of division operation
5176 using the knowledge about the value of the denominator.
5178 @item -fspeculative-prefetching
5179 @opindex fspeculative-prefetching
5180 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5181 a code to gather information about addresses of memory references in the
5184 With @option{-fbranch-probabilities}, it reads back the data gathered
5185 and issues prefetch instructions according to them. In addition to the opportunities
5186 noticed by @option{-fprefetch-loop-arrays}, it also notices more complicated
5187 memory access patterns---for example accesses to the data stored in linked
5188 list whose elements are usually allocated sequentially.
5190 In order to prevent issuing double prefetches, usage of
5191 @option{-fspeculative-prefetching} implies @option{-fno-prefetch-loop-arrays}.
5193 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5195 @item -frename-registers
5196 @opindex frename-registers
5197 Attempt to avoid false dependencies in scheduled code by making use
5198 of registers left over after register allocation. This optimization
5199 will most benefit processors with lots of registers. Depending on the
5200 debug information format adopted by the target, however, it can
5201 make debugging impossible, since variables will no longer stay in
5202 a ``home register''.
5204 Not enabled by default at any level because it has known bugs.
5208 Use a graph coloring register allocator. Currently this option is meant
5209 for testing, so we are interested to hear about miscompilations with
5214 Perform tail duplication to enlarge superblock size. This transformation
5215 simplifies the control flow of the function allowing other optimizations to do
5218 Enabled with @option{-fprofile-use}.
5220 @item -funroll-loops
5221 @opindex funroll-loops
5222 Unroll loops whose number of iterations can be determined at compile time or
5223 upon entry to the loop. @option{-funroll-loops} implies
5224 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
5225 (i.e.@: complete removal of loops with small constant number of iterations).
5226 This option makes code larger, and may or may not make it run faster.
5228 Enabled with @option{-fprofile-use}.
5230 @item -funroll-all-loops
5231 @opindex funroll-all-loops
5232 Unroll all loops, even if their number of iterations is uncertain when
5233 the loop is entered. This usually makes programs run more slowly.
5234 @option{-funroll-all-loops} implies the same options as
5235 @option{-funroll-loops}.
5238 @opindex fpeel-loops
5239 Peels the loops for that there is enough information that they do not
5240 roll much (from profile feedback). It also turns on complete loop peeling
5241 (i.e.@: complete removal of loops with small constant number of iterations).
5243 Enabled with @option{-fprofile-use}.
5245 @item -fmove-loop-invariants
5246 @opindex fmove-loop-invariants
5247 Enables the loop invariant motion pass in the new loop optimizer. Enabled
5248 at level @option{-O1}
5250 @item -funswitch-loops
5251 @opindex funswitch-loops
5252 Move branches with loop invariant conditions out of the loop, with duplicates
5253 of the loop on both branches (modified according to result of the condition).
5255 @item -fprefetch-loop-arrays
5256 @opindex fprefetch-loop-arrays
5257 If supported by the target machine, generate instructions to prefetch
5258 memory to improve the performance of loops that access large arrays.
5260 Disabled at level @option{-Os}.
5262 @item -ffunction-sections
5263 @itemx -fdata-sections
5264 @opindex ffunction-sections
5265 @opindex fdata-sections
5266 Place each function or data item into its own section in the output
5267 file if the target supports arbitrary sections. The name of the
5268 function or the name of the data item determines the section's name
5271 Use these options on systems where the linker can perform optimizations
5272 to improve locality of reference in the instruction space. Most systems
5273 using the ELF object format and SPARC processors running Solaris 2 have
5274 linkers with such optimizations. AIX may have these optimizations in
5277 Only use these options when there are significant benefits from doing
5278 so. When you specify these options, the assembler and linker will
5279 create larger object and executable files and will also be slower.
5280 You will not be able to use @code{gprof} on all systems if you
5281 specify this option and you may have problems with debugging if
5282 you specify both this option and @option{-g}.
5284 @item -fbranch-target-load-optimize
5285 @opindex fbranch-target-load-optimize
5286 Perform branch target register load optimization before prologue / epilogue
5288 The use of target registers can typically be exposed only during reload,
5289 thus hoisting loads out of loops and doing inter-block scheduling needs
5290 a separate optimization pass.
5292 @item -fbranch-target-load-optimize2
5293 @opindex fbranch-target-load-optimize2
5294 Perform branch target register load optimization after prologue / epilogue
5297 @item -fbtr-bb-exclusive
5298 @opindex fbtr-bb-exclusive
5299 When performing branch target register load optimization, don't reuse
5300 branch target registers in within any basic block.
5302 @item --param @var{name}=@var{value}
5304 In some places, GCC uses various constants to control the amount of
5305 optimization that is done. For example, GCC will not inline functions
5306 that contain more that a certain number of instructions. You can
5307 control some of these constants on the command-line using the
5308 @option{--param} option.
5310 The names of specific parameters, and the meaning of the values, are
5311 tied to the internals of the compiler, and are subject to change
5312 without notice in future releases.
5314 In each case, the @var{value} is an integer. The allowable choices for
5315 @var{name} are given in the following table:
5318 @item sra-max-structure-size
5319 The maximum structure size, in bytes, at which the scalar replacement
5320 of aggregates (SRA) optimization will perform block copies. The
5321 default value, 0, implies that GCC will select the most appropriate
5324 @item sra-field-structure-ratio
5325 The threshold ratio (as a percentage) between instantiated fields and
5326 the complete structure size. We say that if the ratio of the number
5327 of bytes in instantiated fields to the number of bytes in the complete
5328 structure exceeds this parameter, then block copies are not used. The
5331 @item max-crossjump-edges
5332 The maximum number of incoming edges to consider for crossjumping.
5333 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5334 the number of edges incoming to each block. Increasing values mean
5335 more aggressive optimization, making the compile time increase with
5336 probably small improvement in executable size.
5338 @item min-crossjump-insns
5339 The minimum number of instructions which must be matched at the end
5340 of two blocks before crossjumping will be performed on them. This
5341 value is ignored in the case where all instructions in the block being
5342 crossjumped from are matched. The default value is 5.
5344 @item max-delay-slot-insn-search
5345 The maximum number of instructions to consider when looking for an
5346 instruction to fill a delay slot. If more than this arbitrary number of
5347 instructions is searched, the time savings from filling the delay slot
5348 will be minimal so stop searching. Increasing values mean more
5349 aggressive optimization, making the compile time increase with probably
5350 small improvement in executable run time.
5352 @item max-delay-slot-live-search
5353 When trying to fill delay slots, the maximum number of instructions to
5354 consider when searching for a block with valid live register
5355 information. Increasing this arbitrarily chosen value means more
5356 aggressive optimization, increasing the compile time. This parameter
5357 should be removed when the delay slot code is rewritten to maintain the
5360 @item max-gcse-memory
5361 The approximate maximum amount of memory that will be allocated in
5362 order to perform the global common subexpression elimination
5363 optimization. If more memory than specified is required, the
5364 optimization will not be done.
5366 @item max-gcse-passes
5367 The maximum number of passes of GCSE to run. The default is 1.
5369 @item max-pending-list-length
5370 The maximum number of pending dependencies scheduling will allow
5371 before flushing the current state and starting over. Large functions
5372 with few branches or calls can create excessively large lists which
5373 needlessly consume memory and resources.
5375 @item max-inline-insns-single
5376 Several parameters control the tree inliner used in gcc.
5377 This number sets the maximum number of instructions (counted in GCC's
5378 internal representation) in a single function that the tree inliner
5379 will consider for inlining. This only affects functions declared
5380 inline and methods implemented in a class declaration (C++).
5381 The default value is 500.
5383 @item max-inline-insns-auto
5384 When you use @option{-finline-functions} (included in @option{-O3}),
5385 a lot of functions that would otherwise not be considered for inlining
5386 by the compiler will be investigated. To those functions, a different
5387 (more restrictive) limit compared to functions declared inline can
5389 The default value is 120.
5391 @item large-function-insns
5392 The limit specifying really large functions. For functions greater than this
5393 limit inlining is constrained by @option{--param large-function-growth}.
5394 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
5395 algorithms used by the backend.
5396 This parameter is ignored when @option{-funit-at-a-time} is not used.
5397 The default value is 3000.
5399 @item large-function-growth
5400 Specifies maximal growth of large function caused by inlining in percents.
5401 This parameter is ignored when @option{-funit-at-a-time} is not used.
5402 The default value is 200.
5404 @item inline-unit-growth
5405 Specifies maximal overall growth of the compilation unit caused by inlining.
5406 This parameter is ignored when @option{-funit-at-a-time} is not used.
5407 The default value is 150.
5409 @item max-inline-insns-recursive
5410 @itemx max-inline-insns-recursive-auto
5411 Specifies maximum number of instructions out-of-line copy of self recursive inline
5412 function can grow into by performing recursive inlining.
5414 For functions declared inline @option{--param max-inline-insns-recursive} is
5415 taken into acount. For function not declared inline, recursive inlining
5416 happens only when @option{-finline-functions} (included in @option{-O3}) is
5417 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5418 default value is 500.
5420 @item max-inline-recursive-depth
5421 @itemx max-inline-recursive-depth-auto
5422 Specifies maximum recursion depth used by the recursive inlining.
5424 For functions declared inline @option{--param max-inline-recursive-depth} is
5425 taken into acount. For function not declared inline, recursive inlining
5426 happens only when @option{-finline-functions} (included in @option{-O3}) is
5427 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5428 default value is 500.
5430 @item max-inline-insns-rtl
5431 For languages that use the RTL inliner (this happens at a later stage
5432 than tree inlining), you can set the maximum allowable size (counted
5433 in RTL instructions) for the RTL inliner with this parameter.
5434 The default value is 600.
5436 @item max-unrolled-insns
5437 The maximum number of instructions that a loop should have if that loop
5438 is unrolled, and if the loop is unrolled, it determines how many times
5439 the loop code is unrolled.
5441 @item max-average-unrolled-insns
5442 The maximum number of instructions biased by probabilities of their execution
5443 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5444 it determines how many times the loop code is unrolled.
5446 @item max-unroll-times
5447 The maximum number of unrollings of a single loop.
5449 @item max-peeled-insns
5450 The maximum number of instructions that a loop should have if that loop
5451 is peeled, and if the loop is peeled, it determines how many times
5452 the loop code is peeled.
5454 @item max-peel-times
5455 The maximum number of peelings of a single loop.
5457 @item max-completely-peeled-insns
5458 The maximum number of insns of a completely peeled loop.
5460 @item max-completely-peel-times
5461 The maximum number of iterations of a loop to be suitable for complete peeling.
5463 @item max-unswitch-insns
5464 The maximum number of insns of an unswitched loop.
5466 @item max-unswitch-level
5467 The maximum number of branches unswitched in a single loop.
5470 The minimum cost of an expensive expression in the loop invariant motion.
5472 @item iv-consider-all-candidates-bound
5473 Bound on number of candidates for induction variables below that
5474 all candidates are considered for each use in induction variable
5475 optimizations. Only the most relevant candidates are considered
5476 if there are more candidates, to avoid quadratic time complexity.
5478 @item iv-max-considered-uses
5479 The induction variable optimizations give up on loops that contain more
5480 induction variable uses.
5482 @item max-iterations-to-track
5484 The maximum number of iterations of a loop the brute force algorithm
5485 for analysis of # of iterations of the loop tries to evaluate.
5487 @item hot-bb-count-fraction
5488 Select fraction of the maximal count of repetitions of basic block in program
5489 given basic block needs to have to be considered hot.
5491 @item hot-bb-frequency-fraction
5492 Select fraction of the maximal frequency of executions of basic block in
5493 function given basic block needs to have to be considered hot
5495 @item tracer-dynamic-coverage
5496 @itemx tracer-dynamic-coverage-feedback
5498 This value is used to limit superblock formation once the given percentage of
5499 executed instructions is covered. This limits unnecessary code size
5502 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5503 feedback is available. The real profiles (as opposed to statically estimated
5504 ones) are much less balanced allowing the threshold to be larger value.
5506 @item tracer-max-code-growth
5507 Stop tail duplication once code growth has reached given percentage. This is
5508 rather hokey argument, as most of the duplicates will be eliminated later in
5509 cross jumping, so it may be set to much higher values than is the desired code
5512 @item tracer-min-branch-ratio
5514 Stop reverse growth when the reverse probability of best edge is less than this
5515 threshold (in percent).
5517 @item tracer-min-branch-ratio
5518 @itemx tracer-min-branch-ratio-feedback
5520 Stop forward growth if the best edge do have probability lower than this
5523 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5524 compilation for profile feedback and one for compilation without. The value
5525 for compilation with profile feedback needs to be more conservative (higher) in
5526 order to make tracer effective.
5528 @item max-cse-path-length
5530 Maximum number of basic blocks on path that cse considers. The default is 10.
5532 @item global-var-threshold
5534 Counts the number of function calls (N) and the number of
5535 call-clobbered variables (V). If NxV is larger than this limit, a
5536 single artificial variable will be created to represent all the
5537 call-clobbered variables at function call sites. This artificial
5538 variable will then be made to alias every call-clobbered variable.
5539 (done as int * size_t on the host machine; beware overflow).
5541 @item max-aliased-vops
5543 Maximum number of virtual operands allowed to represent aliases
5544 before triggering the alias grouping heuristic. Alias grouping
5545 reduces compile times and memory consumption needed for aliasing at
5546 the expense of precision loss in alias information.
5548 @item ggc-min-expand
5550 GCC uses a garbage collector to manage its own memory allocation. This
5551 parameter specifies the minimum percentage by which the garbage
5552 collector's heap should be allowed to expand between collections.
5553 Tuning this may improve compilation speed; it has no effect on code
5556 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
5557 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
5558 the smallest of actual RAM and RLIMIT_DATA or RLIMIT_AS. If
5559 GCC is not able to calculate RAM on a particular platform, the lower
5560 bound of 30% is used. Setting this parameter and
5561 @option{ggc-min-heapsize} to zero causes a full collection to occur at
5562 every opportunity. This is extremely slow, but can be useful for
5565 @item ggc-min-heapsize
5567 Minimum size of the garbage collector's heap before it begins bothering
5568 to collect garbage. The first collection occurs after the heap expands
5569 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
5570 tuning this may improve compilation speed, and has no effect on code
5573 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
5574 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
5575 with a lower bound of 4096 (four megabytes) and an upper bound of
5576 131072 (128 megabytes). If GCC is not able to calculate RAM on a
5577 particular platform, the lower bound is used. Setting this parameter
5578 very large effectively disables garbage collection. Setting this
5579 parameter and @option{ggc-min-expand} to zero causes a full collection
5580 to occur at every opportunity.
5582 @item max-reload-search-insns
5583 The maximum number of instruction reload should look backward for equivalent
5584 register. Increasing values mean more aggressive optimization, making the
5585 compile time increase with probably slightly better performance. The default
5588 @item max-cselib-memory-location
5589 The maximum number of memory locations cselib should take into acount.
5590 Increasing values mean more aggressive optimization, making the compile time
5591 increase with probably slightly better performance. The default value is 500.
5593 @item reorder-blocks-duplicate
5594 @itemx reorder-blocks-duplicate-feedback
5596 Used by basic block reordering pass to decide whether to use unconditional
5597 branch or duplicate the code on its destination. Code is duplicated when its
5598 estimated size is smaller than this value multiplied by the estimated size of
5599 unconditional jump in the hot spots of the program.
5601 The @option{reorder-block-duplicate-feedback} is used only when profile
5602 feedback is available and may be set to higher values than
5603 @option{reorder-block-duplicate} since information about the hot spots is more
5606 @item max-sched-region-blocks
5607 The maximum number of blocks in a region to be considered for
5608 interblock scheduling. The default value is 10.
5610 @item max-sched-region-insns
5611 The maximum number of insns in a region to be considered for
5612 interblock scheduling. The default value is 100.
5614 @item integer-share-limit
5615 Small integer constants can use a shared data structure, reducing the
5616 compiler's memory usage and increasing its speed. This sets the maximum
5617 value of a shared integer constant's. The default value is 256.
5622 @node Preprocessor Options
5623 @section Options Controlling the Preprocessor
5624 @cindex preprocessor options
5625 @cindex options, preprocessor
5627 These options control the C preprocessor, which is run on each C source
5628 file before actual compilation.
5630 If you use the @option{-E} option, nothing is done except preprocessing.
5631 Some of these options make sense only together with @option{-E} because
5632 they cause the preprocessor output to be unsuitable for actual
5637 You can use @option{-Wp,@var{option}} to bypass the compiler driver
5638 and pass @var{option} directly through to the preprocessor. If
5639 @var{option} contains commas, it is split into multiple options at the
5640 commas. However, many options are modified, translated or interpreted
5641 by the compiler driver before being passed to the preprocessor, and
5642 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
5643 interface is undocumented and subject to change, so whenever possible
5644 you should avoid using @option{-Wp} and let the driver handle the
5647 @item -Xpreprocessor @var{option}
5648 @opindex preprocessor
5649 Pass @var{option} as an option to the preprocessor. You can use this to
5650 supply system-specific preprocessor options which GCC does not know how to
5653 If you want to pass an option that takes an argument, you must use
5654 @option{-Xpreprocessor} twice, once for the option and once for the argument.
5657 @include cppopts.texi
5659 @node Assembler Options
5660 @section Passing Options to the Assembler
5662 @c prevent bad page break with this line
5663 You can pass options to the assembler.
5666 @item -Wa,@var{option}
5668 Pass @var{option} as an option to the assembler. If @var{option}
5669 contains commas, it is split into multiple options at the commas.
5671 @item -Xassembler @var{option}
5673 Pass @var{option} as an option to the assembler. You can use this to
5674 supply system-specific assembler options which GCC does not know how to
5677 If you want to pass an option that takes an argument, you must use
5678 @option{-Xassembler} twice, once for the option and once for the argument.
5683 @section Options for Linking
5684 @cindex link options
5685 @cindex options, linking
5687 These options come into play when the compiler links object files into
5688 an executable output file. They are meaningless if the compiler is
5689 not doing a link step.
5693 @item @var{object-file-name}
5694 A file name that does not end in a special recognized suffix is
5695 considered to name an object file or library. (Object files are
5696 distinguished from libraries by the linker according to the file
5697 contents.) If linking is done, these object files are used as input
5706 If any of these options is used, then the linker is not run, and
5707 object file names should not be used as arguments. @xref{Overall
5711 @item -l@var{library}
5712 @itemx -l @var{library}
5714 Search the library named @var{library} when linking. (The second
5715 alternative with the library as a separate argument is only for
5716 POSIX compliance and is not recommended.)
5718 It makes a difference where in the command you write this option; the
5719 linker searches and processes libraries and object files in the order they
5720 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5721 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5722 to functions in @samp{z}, those functions may not be loaded.
5724 The linker searches a standard list of directories for the library,
5725 which is actually a file named @file{lib@var{library}.a}. The linker
5726 then uses this file as if it had been specified precisely by name.
5728 The directories searched include several standard system directories
5729 plus any that you specify with @option{-L}.
5731 Normally the files found this way are library files---archive files
5732 whose members are object files. The linker handles an archive file by
5733 scanning through it for members which define symbols that have so far
5734 been referenced but not defined. But if the file that is found is an
5735 ordinary object file, it is linked in the usual fashion. The only
5736 difference between using an @option{-l} option and specifying a file name
5737 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5738 and searches several directories.
5742 You need this special case of the @option{-l} option in order to
5743 link an Objective-C or Objective-C++ program.
5746 @opindex nostartfiles
5747 Do not use the standard system startup files when linking.
5748 The standard system libraries are used normally, unless @option{-nostdlib}
5749 or @option{-nodefaultlibs} is used.
5751 @item -nodefaultlibs
5752 @opindex nodefaultlibs
5753 Do not use the standard system libraries when linking.
5754 Only the libraries you specify will be passed to the linker.
5755 The standard startup files are used normally, unless @option{-nostartfiles}
5756 is used. The compiler may generate calls to @code{memcmp},
5757 @code{memset}, @code{memcpy} and @code{memmove}.
5758 These entries are usually resolved by entries in
5759 libc. These entry points should be supplied through some other
5760 mechanism when this option is specified.
5764 Do not use the standard system startup files or libraries when linking.
5765 No startup files and only the libraries you specify will be passed to
5766 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
5767 @code{memcpy} and @code{memmove}.
5768 These entries are usually resolved by entries in
5769 libc. These entry points should be supplied through some other
5770 mechanism when this option is specified.
5772 @cindex @option{-lgcc}, use with @option{-nostdlib}
5773 @cindex @option{-nostdlib} and unresolved references
5774 @cindex unresolved references and @option{-nostdlib}
5775 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5776 @cindex @option{-nodefaultlibs} and unresolved references
5777 @cindex unresolved references and @option{-nodefaultlibs}
5778 One of the standard libraries bypassed by @option{-nostdlib} and
5779 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5780 that GCC uses to overcome shortcomings of particular machines, or special
5781 needs for some languages.
5782 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5783 Collection (GCC) Internals},
5784 for more discussion of @file{libgcc.a}.)
5785 In most cases, you need @file{libgcc.a} even when you want to avoid
5786 other standard libraries. In other words, when you specify @option{-nostdlib}
5787 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5788 This ensures that you have no unresolved references to internal GCC
5789 library subroutines. (For example, @samp{__main}, used to ensure C++
5790 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5791 GNU Compiler Collection (GCC) Internals}.)
5795 Produce a position independent executable on targets which support it.
5796 For predictable results, you must also specify the same set of options
5797 that were used to generate code (@option{-fpie}, @option{-fPIE},
5798 or model suboptions) when you specify this option.
5802 Remove all symbol table and relocation information from the executable.
5806 On systems that support dynamic linking, this prevents linking with the shared
5807 libraries. On other systems, this option has no effect.
5811 Produce a shared object which can then be linked with other objects to
5812 form an executable. Not all systems support this option. For predictable
5813 results, you must also specify the same set of options that were used to
5814 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5815 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5816 needs to build supplementary stub code for constructors to work. On
5817 multi-libbed systems, @samp{gcc -shared} must select the correct support
5818 libraries to link against. Failing to supply the correct flags may lead
5819 to subtle defects. Supplying them in cases where they are not necessary
5822 @item -shared-libgcc
5823 @itemx -static-libgcc
5824 @opindex shared-libgcc
5825 @opindex static-libgcc
5826 On systems that provide @file{libgcc} as a shared library, these options
5827 force the use of either the shared or static version respectively.
5828 If no shared version of @file{libgcc} was built when the compiler was
5829 configured, these options have no effect.
5831 There are several situations in which an application should use the
5832 shared @file{libgcc} instead of the static version. The most common
5833 of these is when the application wishes to throw and catch exceptions
5834 across different shared libraries. In that case, each of the libraries
5835 as well as the application itself should use the shared @file{libgcc}.
5837 Therefore, the G++ and GCJ drivers automatically add
5838 @option{-shared-libgcc} whenever you build a shared library or a main
5839 executable, because C++ and Java programs typically use exceptions, so
5840 this is the right thing to do.
5842 If, instead, you use the GCC driver to create shared libraries, you may
5843 find that they will not always be linked with the shared @file{libgcc}.
5844 If GCC finds, at its configuration time, that you have a non-GNU linker
5845 or a GNU linker that does not support option @option{--eh-frame-hdr},
5846 it will link the shared version of @file{libgcc} into shared libraries
5847 by default. Otherwise, it will take advantage of the linker and optimize
5848 away the linking with the shared version of @file{libgcc}, linking with
5849 the static version of libgcc by default. This allows exceptions to
5850 propagate through such shared libraries, without incurring relocation
5851 costs at library load time.
5853 However, if a library or main executable is supposed to throw or catch
5854 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5855 for the languages used in the program, or using the option
5856 @option{-shared-libgcc}, such that it is linked with the shared
5861 Bind references to global symbols when building a shared object. Warn
5862 about any unresolved references (unless overridden by the link editor
5863 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5866 @item -Xlinker @var{option}
5868 Pass @var{option} as an option to the linker. You can use this to
5869 supply system-specific linker options which GCC does not know how to
5872 If you want to pass an option that takes an argument, you must use
5873 @option{-Xlinker} twice, once for the option and once for the argument.
5874 For example, to pass @option{-assert definitions}, you must write
5875 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5876 @option{-Xlinker "-assert definitions"}, because this passes the entire
5877 string as a single argument, which is not what the linker expects.
5879 @item -Wl,@var{option}
5881 Pass @var{option} as an option to the linker. If @var{option} contains
5882 commas, it is split into multiple options at the commas.
5884 @item -u @var{symbol}
5886 Pretend the symbol @var{symbol} is undefined, to force linking of
5887 library modules to define it. You can use @option{-u} multiple times with
5888 different symbols to force loading of additional library modules.
5891 @node Directory Options
5892 @section Options for Directory Search
5893 @cindex directory options
5894 @cindex options, directory search
5897 These options specify directories to search for header files, for
5898 libraries and for parts of the compiler:
5903 Add the directory @var{dir} to the head of the list of directories to be
5904 searched for header files. This can be used to override a system header
5905 file, substituting your own version, since these directories are
5906 searched before the system header file directories. However, you should
5907 not use this option to add directories that contain vendor-supplied
5908 system header files (use @option{-isystem} for that). If you use more than
5909 one @option{-I} option, the directories are scanned in left-to-right
5910 order; the standard system directories come after.
5912 If a standard system include directory, or a directory specified with
5913 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5914 option will be ignored. The directory will still be searched but as a
5915 system directory at its normal position in the system include chain.
5916 This is to ensure that GCC's procedure to fix buggy system headers and
5917 the ordering for the include_next directive are not inadvertently changed.
5918 If you really need to change the search order for system directories,
5919 use the @option{-nostdinc} and/or @option{-isystem} options.
5921 @item -iquote@var{dir}
5923 Add the directory @var{dir} to the head of the list of directories to
5924 be searched for header files only for the case of @samp{#include
5925 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
5926 otherwise just like @option{-I}.
5930 Add directory @var{dir} to the list of directories to be searched
5933 @item -B@var{prefix}
5935 This option specifies where to find the executables, libraries,
5936 include files, and data files of the compiler itself.
5938 The compiler driver program runs one or more of the subprograms
5939 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5940 @var{prefix} as a prefix for each program it tries to run, both with and
5941 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5943 For each subprogram to be run, the compiler driver first tries the
5944 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5945 was not specified, the driver tries two standard prefixes, which are
5946 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5947 those results in a file name that is found, the unmodified program
5948 name is searched for using the directories specified in your
5949 @env{PATH} environment variable.
5951 The compiler will check to see if the path provided by the @option{-B}
5952 refers to a directory, and if necessary it will add a directory
5953 separator character at the end of the path.
5955 @option{-B} prefixes that effectively specify directory names also apply
5956 to libraries in the linker, because the compiler translates these
5957 options into @option{-L} options for the linker. They also apply to
5958 includes files in the preprocessor, because the compiler translates these
5959 options into @option{-isystem} options for the preprocessor. In this case,
5960 the compiler appends @samp{include} to the prefix.
5962 The run-time support file @file{libgcc.a} can also be searched for using
5963 the @option{-B} prefix, if needed. If it is not found there, the two
5964 standard prefixes above are tried, and that is all. The file is left
5965 out of the link if it is not found by those means.
5967 Another way to specify a prefix much like the @option{-B} prefix is to use
5968 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5971 As a special kludge, if the path provided by @option{-B} is
5972 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5973 9, then it will be replaced by @file{[dir/]include}. This is to help
5974 with boot-strapping the compiler.
5976 @item -specs=@var{file}
5978 Process @var{file} after the compiler reads in the standard @file{specs}
5979 file, in order to override the defaults that the @file{gcc} driver
5980 program uses when determining what switches to pass to @file{cc1},
5981 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5982 @option{-specs=@var{file}} can be specified on the command line, and they
5983 are processed in order, from left to right.
5987 This option has been deprecated. Please use @option{-iquote} instead for
5988 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
5989 Any directories you specify with @option{-I} options before the @option{-I-}
5990 option are searched only for the case of @samp{#include "@var{file}"};
5991 they are not searched for @samp{#include <@var{file}>}.
5993 If additional directories are specified with @option{-I} options after
5994 the @option{-I-}, these directories are searched for all @samp{#include}
5995 directives. (Ordinarily @emph{all} @option{-I} directories are used
5998 In addition, the @option{-I-} option inhibits the use of the current
5999 directory (where the current input file came from) as the first search
6000 directory for @samp{#include "@var{file}"}. There is no way to
6001 override this effect of @option{-I-}. With @option{-I.} you can specify
6002 searching the directory which was current when the compiler was
6003 invoked. That is not exactly the same as what the preprocessor does
6004 by default, but it is often satisfactory.
6006 @option{-I-} does not inhibit the use of the standard system directories
6007 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6014 @section Specifying subprocesses and the switches to pass to them
6017 @command{gcc} is a driver program. It performs its job by invoking a
6018 sequence of other programs to do the work of compiling, assembling and
6019 linking. GCC interprets its command-line parameters and uses these to
6020 deduce which programs it should invoke, and which command-line options
6021 it ought to place on their command lines. This behavior is controlled
6022 by @dfn{spec strings}. In most cases there is one spec string for each
6023 program that GCC can invoke, but a few programs have multiple spec
6024 strings to control their behavior. The spec strings built into GCC can
6025 be overridden by using the @option{-specs=} command-line switch to specify
6028 @dfn{Spec files} are plaintext files that are used to construct spec
6029 strings. They consist of a sequence of directives separated by blank
6030 lines. The type of directive is determined by the first non-whitespace
6031 character on the line and it can be one of the following:
6034 @item %@var{command}
6035 Issues a @var{command} to the spec file processor. The commands that can
6039 @item %include <@var{file}>
6041 Search for @var{file} and insert its text at the current point in the
6044 @item %include_noerr <@var{file}>
6045 @cindex %include_noerr
6046 Just like @samp{%include}, but do not generate an error message if the include
6047 file cannot be found.
6049 @item %rename @var{old_name} @var{new_name}
6051 Rename the spec string @var{old_name} to @var{new_name}.
6055 @item *[@var{spec_name}]:
6056 This tells the compiler to create, override or delete the named spec
6057 string. All lines after this directive up to the next directive or
6058 blank line are considered to be the text for the spec string. If this
6059 results in an empty string then the spec will be deleted. (Or, if the
6060 spec did not exist, then nothing will happened.) Otherwise, if the spec
6061 does not currently exist a new spec will be created. If the spec does
6062 exist then its contents will be overridden by the text of this
6063 directive, unless the first character of that text is the @samp{+}
6064 character, in which case the text will be appended to the spec.
6066 @item [@var{suffix}]:
6067 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6068 and up to the next directive or blank line are considered to make up the
6069 spec string for the indicated suffix. When the compiler encounters an
6070 input file with the named suffix, it will processes the spec string in
6071 order to work out how to compile that file. For example:
6078 This says that any input file whose name ends in @samp{.ZZ} should be
6079 passed to the program @samp{z-compile}, which should be invoked with the
6080 command-line switch @option{-input} and with the result of performing the
6081 @samp{%i} substitution. (See below.)
6083 As an alternative to providing a spec string, the text that follows a
6084 suffix directive can be one of the following:
6087 @item @@@var{language}
6088 This says that the suffix is an alias for a known @var{language}. This is
6089 similar to using the @option{-x} command-line switch to GCC to specify a
6090 language explicitly. For example:
6097 Says that .ZZ files are, in fact, C++ source files.
6100 This causes an error messages saying:
6103 @var{name} compiler not installed on this system.
6107 GCC already has an extensive list of suffixes built into it.
6108 This directive will add an entry to the end of the list of suffixes, but
6109 since the list is searched from the end backwards, it is effectively
6110 possible to override earlier entries using this technique.
6114 GCC has the following spec strings built into it. Spec files can
6115 override these strings or create their own. Note that individual
6116 targets can also add their own spec strings to this list.
6119 asm Options to pass to the assembler
6120 asm_final Options to pass to the assembler post-processor
6121 cpp Options to pass to the C preprocessor
6122 cc1 Options to pass to the C compiler
6123 cc1plus Options to pass to the C++ compiler
6124 endfile Object files to include at the end of the link
6125 link Options to pass to the linker
6126 lib Libraries to include on the command line to the linker
6127 libgcc Decides which GCC support library to pass to the linker
6128 linker Sets the name of the linker
6129 predefines Defines to be passed to the C preprocessor
6130 signed_char Defines to pass to CPP to say whether @code{char} is signed
6132 startfile Object files to include at the start of the link
6135 Here is a small example of a spec file:
6141 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6144 This example renames the spec called @samp{lib} to @samp{old_lib} and
6145 then overrides the previous definition of @samp{lib} with a new one.
6146 The new definition adds in some extra command-line options before
6147 including the text of the old definition.
6149 @dfn{Spec strings} are a list of command-line options to be passed to their
6150 corresponding program. In addition, the spec strings can contain
6151 @samp{%}-prefixed sequences to substitute variable text or to
6152 conditionally insert text into the command line. Using these constructs
6153 it is possible to generate quite complex command lines.
6155 Here is a table of all defined @samp{%}-sequences for spec
6156 strings. Note that spaces are not generated automatically around the
6157 results of expanding these sequences. Therefore you can concatenate them
6158 together or combine them with constant text in a single argument.
6162 Substitute one @samp{%} into the program name or argument.
6165 Substitute the name of the input file being processed.
6168 Substitute the basename of the input file being processed.
6169 This is the substring up to (and not including) the last period
6170 and not including the directory.
6173 This is the same as @samp{%b}, but include the file suffix (text after
6177 Marks the argument containing or following the @samp{%d} as a
6178 temporary file name, so that that file will be deleted if GCC exits
6179 successfully. Unlike @samp{%g}, this contributes no text to the
6182 @item %g@var{suffix}
6183 Substitute a file name that has suffix @var{suffix} and is chosen
6184 once per compilation, and mark the argument in the same way as
6185 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6186 name is now chosen in a way that is hard to predict even when previously
6187 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6188 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6189 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6190 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6191 was simply substituted with a file name chosen once per compilation,
6192 without regard to any appended suffix (which was therefore treated
6193 just like ordinary text), making such attacks more likely to succeed.
6195 @item %u@var{suffix}
6196 Like @samp{%g}, but generates a new temporary file name even if
6197 @samp{%u@var{suffix}} was already seen.
6199 @item %U@var{suffix}
6200 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6201 new one if there is no such last file name. In the absence of any
6202 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6203 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6204 would involve the generation of two distinct file names, one
6205 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6206 simply substituted with a file name chosen for the previous @samp{%u},
6207 without regard to any appended suffix.
6209 @item %j@var{suffix}
6210 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6211 writable, and if save-temps is off; otherwise, substitute the name
6212 of a temporary file, just like @samp{%u}. This temporary file is not
6213 meant for communication between processes, but rather as a junk
6216 @item %|@var{suffix}
6217 @itemx %m@var{suffix}
6218 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6219 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6220 all. These are the two most common ways to instruct a program that it
6221 should read from standard input or write to standard output. If you
6222 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6223 construct: see for example @file{f/lang-specs.h}.
6225 @item %.@var{SUFFIX}
6226 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6227 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6228 terminated by the next space or %.
6231 Marks the argument containing or following the @samp{%w} as the
6232 designated output file of this compilation. This puts the argument
6233 into the sequence of arguments that @samp{%o} will substitute later.
6236 Substitutes the names of all the output files, with spaces
6237 automatically placed around them. You should write spaces
6238 around the @samp{%o} as well or the results are undefined.
6239 @samp{%o} is for use in the specs for running the linker.
6240 Input files whose names have no recognized suffix are not compiled
6241 at all, but they are included among the output files, so they will
6245 Substitutes the suffix for object files. Note that this is
6246 handled specially when it immediately follows @samp{%g, %u, or %U},
6247 because of the need for those to form complete file names. The
6248 handling is such that @samp{%O} is treated exactly as if it had already
6249 been substituted, except that @samp{%g, %u, and %U} do not currently
6250 support additional @var{suffix} characters following @samp{%O} as they would
6251 following, for example, @samp{.o}.
6254 Substitutes the standard macro predefinitions for the
6255 current target machine. Use this when running @code{cpp}.
6258 Like @samp{%p}, but puts @samp{__} before and after the name of each
6259 predefined macro, except for macros that start with @samp{__} or with
6260 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6264 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6265 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
6266 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6270 Current argument is the name of a library or startup file of some sort.
6271 Search for that file in a standard list of directories and substitute
6272 the full name found.
6275 Print @var{str} as an error message. @var{str} is terminated by a newline.
6276 Use this when inconsistent options are detected.
6279 Substitute the contents of spec string @var{name} at this point.
6282 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6284 @item %x@{@var{option}@}
6285 Accumulate an option for @samp{%X}.
6288 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6292 Output the accumulated assembler options specified by @option{-Wa}.
6295 Output the accumulated preprocessor options specified by @option{-Wp}.
6298 Process the @code{asm} spec. This is used to compute the
6299 switches to be passed to the assembler.
6302 Process the @code{asm_final} spec. This is a spec string for
6303 passing switches to an assembler post-processor, if such a program is
6307 Process the @code{link} spec. This is the spec for computing the
6308 command line passed to the linker. Typically it will make use of the
6309 @samp{%L %G %S %D and %E} sequences.
6312 Dump out a @option{-L} option for each directory that GCC believes might
6313 contain startup files. If the target supports multilibs then the
6314 current multilib directory will be prepended to each of these paths.
6317 Output the multilib directory with directory separators replaced with
6318 @samp{_}. If multilib directories are not set, or the multilib directory is
6319 @file{.} then this option emits nothing.
6322 Process the @code{lib} spec. This is a spec string for deciding which
6323 libraries should be included on the command line to the linker.
6326 Process the @code{libgcc} spec. This is a spec string for deciding
6327 which GCC support library should be included on the command line to the linker.
6330 Process the @code{startfile} spec. This is a spec for deciding which
6331 object files should be the first ones passed to the linker. Typically
6332 this might be a file named @file{crt0.o}.
6335 Process the @code{endfile} spec. This is a spec string that specifies
6336 the last object files that will be passed to the linker.
6339 Process the @code{cpp} spec. This is used to construct the arguments
6340 to be passed to the C preprocessor.
6343 Process the @code{cc1} spec. This is used to construct the options to be
6344 passed to the actual C compiler (@samp{cc1}).
6347 Process the @code{cc1plus} spec. This is used to construct the options to be
6348 passed to the actual C++ compiler (@samp{cc1plus}).
6351 Substitute the variable part of a matched option. See below.
6352 Note that each comma in the substituted string is replaced by
6356 Remove all occurrences of @code{-S} from the command line. Note---this
6357 command is position dependent. @samp{%} commands in the spec string
6358 before this one will see @code{-S}, @samp{%} commands in the spec string
6359 after this one will not.
6361 @item %:@var{function}(@var{args})
6362 Call the named function @var{function}, passing it @var{args}.
6363 @var{args} is first processed as a nested spec string, then split
6364 into an argument vector in the usual fashion. The function returns
6365 a string which is processed as if it had appeared literally as part
6366 of the current spec.
6368 The following built-in spec functions are provided:
6371 @item @code{if-exists}
6372 The @code{if-exists} spec function takes one argument, an absolute
6373 pathname to a file. If the file exists, @code{if-exists} returns the
6374 pathname. Here is a small example of its usage:
6378 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6381 @item @code{if-exists-else}
6382 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6383 spec function, except that it takes two arguments. The first argument is
6384 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6385 returns the pathname. If it does not exist, it returns the second argument.
6386 This way, @code{if-exists-else} can be used to select one file or another,
6387 based on the existence of the first. Here is a small example of its usage:
6391 crt0%O%s %:if-exists(crti%O%s) \
6392 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6395 @item @code{replace-outfile}
6396 The @code{replace-outfile} spec function takes two arguments. It looks for the
6397 first argument in the outfiles array and replaces it with the second argument. Here
6398 is a small example of its usage:
6401 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
6407 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6408 If that switch was not specified, this substitutes nothing. Note that
6409 the leading dash is omitted when specifying this option, and it is
6410 automatically inserted if the substitution is performed. Thus the spec
6411 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6412 and would output the command line option @option{-foo}.
6414 @item %W@{@code{S}@}
6415 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6418 @item %@{@code{S}*@}
6419 Substitutes all the switches specified to GCC whose names start
6420 with @code{-S}, but which also take an argument. This is used for
6421 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6422 GCC considers @option{-o foo} as being
6423 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6424 text, including the space. Thus two arguments would be generated.
6426 @item %@{@code{S}*&@code{T}*@}
6427 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6428 (the order of @code{S} and @code{T} in the spec is not significant).
6429 There can be any number of ampersand-separated variables; for each the
6430 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6432 @item %@{@code{S}:@code{X}@}
6433 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6435 @item %@{!@code{S}:@code{X}@}
6436 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6438 @item %@{@code{S}*:@code{X}@}
6439 Substitutes @code{X} if one or more switches whose names start with
6440 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6441 once, no matter how many such switches appeared. However, if @code{%*}
6442 appears somewhere in @code{X}, then @code{X} will be substituted once
6443 for each matching switch, with the @code{%*} replaced by the part of
6444 that switch that matched the @code{*}.
6446 @item %@{.@code{S}:@code{X}@}
6447 Substitutes @code{X}, if processing a file with suffix @code{S}.
6449 @item %@{!.@code{S}:@code{X}@}
6450 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6452 @item %@{@code{S}|@code{P}:@code{X}@}
6453 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6454 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6455 although they have a stronger binding than the @samp{|}. If @code{%*}
6456 appears in @code{X}, all of the alternatives must be starred, and only
6457 the first matching alternative is substituted.
6459 For example, a spec string like this:
6462 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6465 will output the following command-line options from the following input
6466 command-line options:
6471 -d fred.c -foo -baz -boggle
6472 -d jim.d -bar -baz -boggle
6475 @item %@{S:X; T:Y; :D@}
6477 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6478 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6479 be as many clauses as you need. This may be combined with @code{.},
6480 @code{!}, @code{|}, and @code{*} as needed.
6485 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6486 construct may contain other nested @samp{%} constructs or spaces, or
6487 even newlines. They are processed as usual, as described above.
6488 Trailing white space in @code{X} is ignored. White space may also
6489 appear anywhere on the left side of the colon in these constructs,
6490 except between @code{.} or @code{*} and the corresponding word.
6492 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6493 handled specifically in these constructs. If another value of
6494 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
6495 @option{-W} switch is found later in the command line, the earlier
6496 switch value is ignored, except with @{@code{S}*@} where @code{S} is
6497 just one letter, which passes all matching options.
6499 The character @samp{|} at the beginning of the predicate text is used to
6500 indicate that a command should be piped to the following command, but
6501 only if @option{-pipe} is specified.
6503 It is built into GCC which switches take arguments and which do not.
6504 (You might think it would be useful to generalize this to allow each
6505 compiler's spec to say which switches take arguments. But this cannot
6506 be done in a consistent fashion. GCC cannot even decide which input
6507 files have been specified without knowing which switches take arguments,
6508 and it must know which input files to compile in order to tell which
6511 GCC also knows implicitly that arguments starting in @option{-l} are to be
6512 treated as compiler output files, and passed to the linker in their
6513 proper position among the other output files.
6515 @c man begin OPTIONS
6517 @node Target Options
6518 @section Specifying Target Machine and Compiler Version
6519 @cindex target options
6520 @cindex cross compiling
6521 @cindex specifying machine version
6522 @cindex specifying compiler version and target machine
6523 @cindex compiler version, specifying
6524 @cindex target machine, specifying
6526 The usual way to run GCC is to run the executable called @file{gcc}, or
6527 @file{<machine>-gcc} when cross-compiling, or
6528 @file{<machine>-gcc-<version>} to run a version other than the one that
6529 was installed last. Sometimes this is inconvenient, so GCC provides
6530 options that will switch to another cross-compiler or version.
6533 @item -b @var{machine}
6535 The argument @var{machine} specifies the target machine for compilation.
6537 The value to use for @var{machine} is the same as was specified as the
6538 machine type when configuring GCC as a cross-compiler. For
6539 example, if a cross-compiler was configured with @samp{configure
6540 i386v}, meaning to compile for an 80386 running System V, then you
6541 would specify @option{-b i386v} to run that cross compiler.
6543 @item -V @var{version}
6545 The argument @var{version} specifies which version of GCC to run.
6546 This is useful when multiple versions are installed. For example,
6547 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
6550 The @option{-V} and @option{-b} options work by running the
6551 @file{<machine>-gcc-<version>} executable, so there's no real reason to
6552 use them if you can just run that directly.
6554 @node Submodel Options
6555 @section Hardware Models and Configurations
6556 @cindex submodel options
6557 @cindex specifying hardware config
6558 @cindex hardware models and configurations, specifying
6559 @cindex machine dependent options
6561 Earlier we discussed the standard option @option{-b} which chooses among
6562 different installed compilers for completely different target
6563 machines, such as VAX vs.@: 68000 vs.@: 80386.
6565 In addition, each of these target machine types can have its own
6566 special options, starting with @samp{-m}, to choose among various
6567 hardware models or configurations---for example, 68010 vs 68020,
6568 floating coprocessor or none. A single installed version of the
6569 compiler can compile for any model or configuration, according to the
6572 Some configurations of the compiler also support additional special
6573 options, usually for compatibility with other compilers on the same
6576 These options are defined by the macro @code{TARGET_SWITCHES} in the
6577 machine description. The default for the options is also defined by
6578 that macro, which enables you to change the defaults.
6580 @c This list is ordered alphanumerically by subsection name.
6581 @c It should be the same order and spelling as these options are listed
6582 @c in Machine Dependent Options
6590 * DEC Alpha Options::
6591 * DEC Alpha/VMS Options::
6595 * i386 and x86-64 Options::
6607 * RS/6000 and PowerPC Options::
6608 * S/390 and zSeries Options::
6611 * System V Options::
6612 * TMS320C3x/C4x Options::
6616 * Xstormy16 Options::
6622 @subsection ARC Options
6625 These options are defined for ARC implementations:
6630 Compile code for little endian mode. This is the default.
6634 Compile code for big endian mode.
6637 @opindex mmangle-cpu
6638 Prepend the name of the cpu to all public symbol names.
6639 In multiple-processor systems, there are many ARC variants with different
6640 instruction and register set characteristics. This flag prevents code
6641 compiled for one cpu to be linked with code compiled for another.
6642 No facility exists for handling variants that are ``almost identical''.
6643 This is an all or nothing option.
6645 @item -mcpu=@var{cpu}
6647 Compile code for ARC variant @var{cpu}.
6648 Which variants are supported depend on the configuration.
6649 All variants support @option{-mcpu=base}, this is the default.
6651 @item -mtext=@var{text-section}
6652 @itemx -mdata=@var{data-section}
6653 @itemx -mrodata=@var{readonly-data-section}
6657 Put functions, data, and readonly data in @var{text-section},
6658 @var{data-section}, and @var{readonly-data-section} respectively
6659 by default. This can be overridden with the @code{section} attribute.
6660 @xref{Variable Attributes}.
6665 @subsection ARM Options
6668 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6672 @item -mabi=@var{name}
6674 Generate code for the specified ABI. Permissible values are: @samp{apcs-gnu},
6675 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6678 @opindex mapcs-frame
6679 Generate a stack frame that is compliant with the ARM Procedure Call
6680 Standard for all functions, even if this is not strictly necessary for
6681 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6682 with this option will cause the stack frames not to be generated for
6683 leaf functions. The default is @option{-mno-apcs-frame}.
6687 This is a synonym for @option{-mapcs-frame}.
6690 @c not currently implemented
6691 @item -mapcs-stack-check
6692 @opindex mapcs-stack-check
6693 Generate code to check the amount of stack space available upon entry to
6694 every function (that actually uses some stack space). If there is
6695 insufficient space available then either the function
6696 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6697 called, depending upon the amount of stack space required. The run time
6698 system is required to provide these functions. The default is
6699 @option{-mno-apcs-stack-check}, since this produces smaller code.
6701 @c not currently implemented
6703 @opindex mapcs-float
6704 Pass floating point arguments using the float point registers. This is
6705 one of the variants of the APCS@. This option is recommended if the
6706 target hardware has a floating point unit or if a lot of floating point
6707 arithmetic is going to be performed by the code. The default is
6708 @option{-mno-apcs-float}, since integer only code is slightly increased in
6709 size if @option{-mapcs-float} is used.
6711 @c not currently implemented
6712 @item -mapcs-reentrant
6713 @opindex mapcs-reentrant
6714 Generate reentrant, position independent code. The default is
6715 @option{-mno-apcs-reentrant}.
6718 @item -mthumb-interwork
6719 @opindex mthumb-interwork
6720 Generate code which supports calling between the ARM and Thumb
6721 instruction sets. Without this option the two instruction sets cannot
6722 be reliably used inside one program. The default is
6723 @option{-mno-thumb-interwork}, since slightly larger code is generated
6724 when @option{-mthumb-interwork} is specified.
6726 @item -mno-sched-prolog
6727 @opindex mno-sched-prolog
6728 Prevent the reordering of instructions in the function prolog, or the
6729 merging of those instruction with the instructions in the function's
6730 body. This means that all functions will start with a recognizable set
6731 of instructions (or in fact one of a choice from a small set of
6732 different function prologues), and this information can be used to
6733 locate the start if functions inside an executable piece of code. The
6734 default is @option{-msched-prolog}.
6737 @opindex mhard-float
6738 Generate output containing floating point instructions. This is the
6742 @opindex msoft-float
6743 Generate output containing library calls for floating point.
6744 @strong{Warning:} the requisite libraries are not available for all ARM
6745 targets. Normally the facilities of the machine's usual C compiler are
6746 used, but this cannot be done directly in cross-compilation. You must make
6747 your own arrangements to provide suitable library functions for
6750 @option{-msoft-float} changes the calling convention in the output file;
6751 therefore, it is only useful if you compile @emph{all} of a program with
6752 this option. In particular, you need to compile @file{libgcc.a}, the
6753 library that comes with GCC, with @option{-msoft-float} in order for
6756 @item -mfloat-abi=@var{name}
6758 Specifies which ABI to use for floating point values. Permissible values
6759 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6761 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6762 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6763 of floating point instructions, but still uses the soft-float calling
6766 @item -mlittle-endian
6767 @opindex mlittle-endian
6768 Generate code for a processor running in little-endian mode. This is
6769 the default for all standard configurations.
6772 @opindex mbig-endian
6773 Generate code for a processor running in big-endian mode; the default is
6774 to compile code for a little-endian processor.
6776 @item -mwords-little-endian
6777 @opindex mwords-little-endian
6778 This option only applies when generating code for big-endian processors.
6779 Generate code for a little-endian word order but a big-endian byte
6780 order. That is, a byte order of the form @samp{32107654}. Note: this
6781 option should only be used if you require compatibility with code for
6782 big-endian ARM processors generated by versions of the compiler prior to
6785 @item -mcpu=@var{name}
6787 This specifies the name of the target ARM processor. GCC uses this name
6788 to determine what kind of instructions it can emit when generating
6789 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6790 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6791 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6792 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6793 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6794 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
6795 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6796 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6797 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
6798 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
6799 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
6800 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
6801 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
6802 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
6805 @itemx -mtune=@var{name}
6807 This option is very similar to the @option{-mcpu=} option, except that
6808 instead of specifying the actual target processor type, and hence
6809 restricting which instructions can be used, it specifies that GCC should
6810 tune the performance of the code as if the target were of the type
6811 specified in this option, but still choosing the instructions that it
6812 will generate based on the cpu specified by a @option{-mcpu=} option.
6813 For some ARM implementations better performance can be obtained by using
6816 @item -march=@var{name}
6818 This specifies the name of the target ARM architecture. GCC uses this
6819 name to determine what kind of instructions it can emit when generating
6820 assembly code. This option can be used in conjunction with or instead
6821 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6822 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6823 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6824 @samp{iwmmxt}, @samp{ep9312}.
6826 @item -mfpu=@var{name}
6827 @itemx -mfpe=@var{number}
6828 @itemx -mfp=@var{number}
6832 This specifies what floating point hardware (or hardware emulation) is
6833 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6834 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6835 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6836 with older versions of GCC@.
6838 If @option{-msoft-float} is specified this specifies the format of
6839 floating point values.
6841 @item -mstructure-size-boundary=@var{n}
6842 @opindex mstructure-size-boundary
6843 The size of all structures and unions will be rounded up to a multiple
6844 of the number of bits set by this option. Permissible values are 8, 32
6845 and 64. The default value varies for different toolchains. For the COFF
6846 targeted toolchain the default value is 8. A value of 64 is only allowed
6847 if the underlying ABI supports it.
6849 Specifying the larger number can produce faster, more efficient code, but
6850 can also increase the size of the program. Different values are potentially
6851 incompatible. Code compiled with one value cannot necessarily expect to
6852 work with code or libraries compiled with another value, if they exchange
6853 information using structures or unions.
6855 @item -mabort-on-noreturn
6856 @opindex mabort-on-noreturn
6857 Generate a call to the function @code{abort} at the end of a
6858 @code{noreturn} function. It will be executed if the function tries to
6862 @itemx -mno-long-calls
6863 @opindex mlong-calls
6864 @opindex mno-long-calls
6865 Tells the compiler to perform function calls by first loading the
6866 address of the function into a register and then performing a subroutine
6867 call on this register. This switch is needed if the target function
6868 will lie outside of the 64 megabyte addressing range of the offset based
6869 version of subroutine call instruction.
6871 Even if this switch is enabled, not all function calls will be turned
6872 into long calls. The heuristic is that static functions, functions
6873 which have the @samp{short-call} attribute, functions that are inside
6874 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6875 definitions have already been compiled within the current compilation
6876 unit, will not be turned into long calls. The exception to this rule is
6877 that weak function definitions, functions with the @samp{long-call}
6878 attribute or the @samp{section} attribute, and functions that are within
6879 the scope of a @samp{#pragma long_calls} directive, will always be
6880 turned into long calls.
6882 This feature is not enabled by default. Specifying
6883 @option{-mno-long-calls} will restore the default behavior, as will
6884 placing the function calls within the scope of a @samp{#pragma
6885 long_calls_off} directive. Note these switches have no effect on how
6886 the compiler generates code to handle function calls via function
6889 @item -mnop-fun-dllimport
6890 @opindex mnop-fun-dllimport
6891 Disable support for the @code{dllimport} attribute.
6893 @item -msingle-pic-base
6894 @opindex msingle-pic-base
6895 Treat the register used for PIC addressing as read-only, rather than
6896 loading it in the prologue for each function. The run-time system is
6897 responsible for initializing this register with an appropriate value
6898 before execution begins.
6900 @item -mpic-register=@var{reg}
6901 @opindex mpic-register
6902 Specify the register to be used for PIC addressing. The default is R10
6903 unless stack-checking is enabled, when R9 is used.
6905 @item -mcirrus-fix-invalid-insns
6906 @opindex mcirrus-fix-invalid-insns
6907 @opindex mno-cirrus-fix-invalid-insns
6908 Insert NOPs into the instruction stream to in order to work around
6909 problems with invalid Maverick instruction combinations. This option
6910 is only valid if the @option{-mcpu=ep9312} option has been used to
6911 enable generation of instructions for the Cirrus Maverick floating
6912 point co-processor. This option is not enabled by default, since the
6913 problem is only present in older Maverick implementations. The default
6914 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6917 @item -mpoke-function-name
6918 @opindex mpoke-function-name
6919 Write the name of each function into the text section, directly
6920 preceding the function prologue. The generated code is similar to this:
6924 .ascii "arm_poke_function_name", 0
6927 .word 0xff000000 + (t1 - t0)
6928 arm_poke_function_name
6930 stmfd sp!, @{fp, ip, lr, pc@}
6934 When performing a stack backtrace, code can inspect the value of
6935 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6936 location @code{pc - 12} and the top 8 bits are set, then we know that
6937 there is a function name embedded immediately preceding this location
6938 and has length @code{((pc[-3]) & 0xff000000)}.
6942 Generate code for the 16-bit Thumb instruction set. The default is to
6943 use the 32-bit ARM instruction set.
6946 @opindex mtpcs-frame
6947 Generate a stack frame that is compliant with the Thumb Procedure Call
6948 Standard for all non-leaf functions. (A leaf function is one that does
6949 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6951 @item -mtpcs-leaf-frame
6952 @opindex mtpcs-leaf-frame
6953 Generate a stack frame that is compliant with the Thumb Procedure Call
6954 Standard for all leaf functions. (A leaf function is one that does
6955 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6957 @item -mcallee-super-interworking
6958 @opindex mcallee-super-interworking
6959 Gives all externally visible functions in the file being compiled an ARM
6960 instruction set header which switches to Thumb mode before executing the
6961 rest of the function. This allows these functions to be called from
6962 non-interworking code.
6964 @item -mcaller-super-interworking
6965 @opindex mcaller-super-interworking
6966 Allows calls via function pointers (including virtual functions) to
6967 execute correctly regardless of whether the target code has been
6968 compiled for interworking or not. There is a small overhead in the cost
6969 of executing a function pointer if this option is enabled.
6974 @subsection AVR Options
6977 These options are defined for AVR implementations:
6980 @item -mmcu=@var{mcu}
6982 Specify ATMEL AVR instruction set or MCU type.
6984 Instruction set avr1 is for the minimal AVR core, not supported by the C
6985 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
6986 attiny11, attiny12, attiny15, attiny28).
6988 Instruction set avr2 (default) is for the classic AVR core with up to
6989 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
6990 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
6991 at90c8534, at90s8535).
6993 Instruction set avr3 is for the classic AVR core with up to 128K program
6994 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
6996 Instruction set avr4 is for the enhanced AVR core with up to 8K program
6997 memory space (MCU types: atmega8, atmega83, atmega85).
6999 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7000 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7001 atmega64, atmega128, at43usb355, at94k).
7005 Output instruction sizes to the asm file.
7007 @item -minit-stack=@var{N}
7008 @opindex minit-stack
7009 Specify the initial stack address, which may be a symbol or numeric value,
7010 @samp{__stack} is the default.
7012 @item -mno-interrupts
7013 @opindex mno-interrupts
7014 Generated code is not compatible with hardware interrupts.
7015 Code size will be smaller.
7017 @item -mcall-prologues
7018 @opindex mcall-prologues
7019 Functions prologues/epilogues expanded as call to appropriate
7020 subroutines. Code size will be smaller.
7022 @item -mno-tablejump
7023 @opindex mno-tablejump
7024 Do not generate tablejump insns which sometimes increase code size.
7027 @opindex mtiny-stack
7028 Change only the low 8 bits of the stack pointer.
7032 Assume int to be 8 bit integer. This affects the sizes of all types: A
7033 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7034 and long long will be 4 bytes. Please note that this option does not
7035 comply to the C standards, but it will provide you with smaller code
7040 @subsection CRIS Options
7041 @cindex CRIS Options
7043 These options are defined specifically for the CRIS ports.
7046 @item -march=@var{architecture-type}
7047 @itemx -mcpu=@var{architecture-type}
7050 Generate code for the specified architecture. The choices for
7051 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7052 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
7053 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7056 @item -mtune=@var{architecture-type}
7058 Tune to @var{architecture-type} everything applicable about the generated
7059 code, except for the ABI and the set of available instructions. The
7060 choices for @var{architecture-type} are the same as for
7061 @option{-march=@var{architecture-type}}.
7063 @item -mmax-stack-frame=@var{n}
7064 @opindex mmax-stack-frame
7065 Warn when the stack frame of a function exceeds @var{n} bytes.
7067 @item -melinux-stacksize=@var{n}
7068 @opindex melinux-stacksize
7069 Only available with the @samp{cris-axis-aout} target. Arranges for
7070 indications in the program to the kernel loader that the stack of the
7071 program should be set to @var{n} bytes.
7077 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7078 @option{-march=v3} and @option{-march=v8} respectively.
7080 @item -mmul-bug-workaround
7081 @itemx -mno-mul-bug-workaround
7082 @opindex mmul-bug-workaround
7083 @opindex mno-mul-bug-workaround
7084 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7085 models where it applies. This option is active by default.
7089 Enable CRIS-specific verbose debug-related information in the assembly
7090 code. This option also has the effect to turn off the @samp{#NO_APP}
7091 formatted-code indicator to the assembler at the beginning of the
7096 Do not use condition-code results from previous instruction; always emit
7097 compare and test instructions before use of condition codes.
7099 @item -mno-side-effects
7100 @opindex mno-side-effects
7101 Do not emit instructions with side-effects in addressing modes other than
7105 @itemx -mno-stack-align
7107 @itemx -mno-data-align
7108 @itemx -mconst-align
7109 @itemx -mno-const-align
7110 @opindex mstack-align
7111 @opindex mno-stack-align
7112 @opindex mdata-align
7113 @opindex mno-data-align
7114 @opindex mconst-align
7115 @opindex mno-const-align
7116 These options (no-options) arranges (eliminate arrangements) for the
7117 stack-frame, individual data and constants to be aligned for the maximum
7118 single data access size for the chosen CPU model. The default is to
7119 arrange for 32-bit alignment. ABI details such as structure layout are
7120 not affected by these options.
7128 Similar to the stack- data- and const-align options above, these options
7129 arrange for stack-frame, writable data and constants to all be 32-bit,
7130 16-bit or 8-bit aligned. The default is 32-bit alignment.
7132 @item -mno-prologue-epilogue
7133 @itemx -mprologue-epilogue
7134 @opindex mno-prologue-epilogue
7135 @opindex mprologue-epilogue
7136 With @option{-mno-prologue-epilogue}, the normal function prologue and
7137 epilogue that sets up the stack-frame are omitted and no return
7138 instructions or return sequences are generated in the code. Use this
7139 option only together with visual inspection of the compiled code: no
7140 warnings or errors are generated when call-saved registers must be saved,
7141 or storage for local variable needs to be allocated.
7147 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7148 instruction sequences that load addresses for functions from the PLT part
7149 of the GOT rather than (traditional on other architectures) calls to the
7150 PLT. The default is @option{-mgotplt}.
7154 Legacy no-op option only recognized with the cris-axis-aout target.
7158 Legacy no-op option only recognized with the cris-axis-elf and
7159 cris-axis-linux-gnu targets.
7163 Only recognized with the cris-axis-aout target, where it selects a
7164 GNU/linux-like multilib, include files and instruction set for
7169 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7173 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7174 to link with input-output functions from a simulator library. Code,
7175 initialized data and zero-initialized data are allocated consecutively.
7179 Like @option{-sim}, but pass linker options to locate initialized data at
7180 0x40000000 and zero-initialized data at 0x80000000.
7183 @node Darwin Options
7184 @subsection Darwin Options
7185 @cindex Darwin options
7187 These options are defined for all architectures running the Darwin operating
7190 FSF GCC on Darwin does not create ``fat'' object files; it will create
7191 an object file for the single architecture that it was built to
7192 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7193 @option{-arch} options are used; it does so by running the compiler or
7194 linker multiple times and joining the results together with
7197 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7198 @samp{i686}) is determined by the flags that specify the ISA
7199 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7200 @option{-force_cpusubtype_ALL} option can be used to override this.
7202 The Darwin tools vary in their behaviour when presented with an ISA
7203 mismatch. The assembler, @file{as}, will only permit instructions to
7204 be used that are valid for the subtype of the file it is generating,
7205 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7206 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7207 and print an error if asked to create a shared library with a less
7208 restrictive subtype than its input files (for instance, trying to put
7209 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7210 for executables, @file{ld}, will quietly give the executable the most
7211 restrictive subtype of any of its input files.
7216 Add the framework directory @var{dir} to the head of the list of
7217 directories to be searched for header files. These directories are
7218 interleaved with those specified by @option{-I} options and are
7219 scanned in a left-to-right order.
7221 A framework directory is a directory with frameworks in it. A
7222 framework is a directory with a @samp{"Headers"} and/or
7223 @samp{"PrivateHeaders"} directory contained directly in it that ends
7224 in @samp{".framework"}. The name of a framework is the name of this
7225 directory excluding the @samp{".framework"}. Headers associated with
7226 the framework are found in one of those two directories, with
7227 @samp{"Headers"} being searched first. A subframework is a framework
7228 directory that is in a framework's @samp{"Frameworks"} directory.
7229 Includes of subframework headers can only appear in a header of a
7230 framework that contains the subframework, or in a sibling subframework
7231 header. Two subframeworks are siblings if they occur in the same
7232 framework. A subframework should not have the same name as a
7233 framework, a warning will be issued if this is violated. Currently a
7234 subframework cannot have subframeworks, in the future, the mechanism
7235 may be extended to support this. The standard frameworks can be found
7236 in @samp{"/System/Library/Frameworks"} and
7237 @samp{"/Library/Frameworks"}. An example include looks like
7238 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7239 the name of the framework and header.h is found in the
7240 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7244 Emit debugging information for symbols that are used. For STABS
7245 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7246 This is by default ON.
7250 Emit debugging information for all symbols and types.
7252 @item -mone-byte-bool
7253 @opindex -mone-byte-bool
7254 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
7255 By default @samp{sizeof(bool)} is @samp{4} when compiling for
7256 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
7257 option has no effect on x86.
7259 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
7260 to generate code that is not binary compatible with code generated
7261 without that switch. Using this switch may require recompiling all
7262 other modules in a program, including system libraries. Use this
7263 switch to conform to a non-default data model.
7265 @item -mfix-and-continue
7266 @itemx -ffix-and-continue
7267 @itemx -findirect-data
7268 @opindex mfix-and-continue
7269 @opindex ffix-and-continue
7270 @opindex findirect-data
7271 Generate code suitable for fast turn around development. Needed to
7272 enable gdb to dynamically load @code{.o} files into already running
7273 programs. @option{-findirect-data} and @option{-ffix-and-continue}
7274 are provided for backwards compatibility.
7278 Loads all members of static archive libraries.
7279 See man ld(1) for more information.
7281 @item -arch_errors_fatal
7282 @opindex arch_errors_fatal
7283 Cause the errors having to do with files that have the wrong architecture
7287 @opindex bind_at_load
7288 Causes the output file to be marked such that the dynamic linker will
7289 bind all undefined references when the file is loaded or launched.
7293 Produce a Mach-o bundle format file.
7294 See man ld(1) for more information.
7296 @item -bundle_loader @var{executable}
7297 @opindex bundle_loader
7298 This option specifies the @var{executable} that will be loading the build
7299 output file being linked. See man ld(1) for more information.
7302 @opindex -dynamiclib
7303 When passed this option, GCC will produce a dynamic library instead of
7304 an executable when linking, using the Darwin @file{libtool} command.
7306 @item -force_cpusubtype_ALL
7307 @opindex -force_cpusubtype_ALL
7308 This causes GCC's output file to have the @var{ALL} subtype, instead of
7309 one controlled by the @option{-mcpu} or @option{-march} option.
7311 @item -allowable_client @var{client_name}
7313 @itemx -compatibility_version
7314 @itemx -current_version
7316 @itemx -dependency-file
7318 @itemx -dylinker_install_name
7320 @itemx -exported_symbols_list
7322 @itemx -flat_namespace
7323 @itemx -force_flat_namespace
7324 @itemx -headerpad_max_install_names
7327 @itemx -install_name
7328 @itemx -keep_private_externs
7329 @itemx -multi_module
7330 @itemx -multiply_defined
7331 @itemx -multiply_defined_unused
7333 @itemx -no_dead_strip_inits_and_terms
7334 @itemx -nofixprebinding
7337 @itemx -noseglinkedit
7338 @itemx -pagezero_size
7340 @itemx -prebind_all_twolevel_modules
7341 @itemx -private_bundle
7342 @itemx -read_only_relocs
7344 @itemx -sectobjectsymbols
7348 @itemx -sectobjectsymbols
7351 @itemx -segs_read_only_addr
7352 @itemx -segs_read_write_addr
7353 @itemx -seg_addr_table
7354 @itemx -seg_addr_table_filename
7357 @itemx -segs_read_only_addr
7358 @itemx -segs_read_write_addr
7359 @itemx -single_module
7362 @itemx -sub_umbrella
7363 @itemx -twolevel_namespace
7366 @itemx -unexported_symbols_list
7367 @itemx -weak_reference_mismatches
7370 @opindex allowable_client
7371 @opindex client_name
7372 @opindex compatibility_version
7373 @opindex current_version
7375 @opindex dependency-file
7377 @opindex dylinker_install_name
7379 @opindex exported_symbols_list
7381 @opindex flat_namespace
7382 @opindex force_flat_namespace
7383 @opindex headerpad_max_install_names
7386 @opindex install_name
7387 @opindex keep_private_externs
7388 @opindex multi_module
7389 @opindex multiply_defined
7390 @opindex multiply_defined_unused
7392 @opindex no_dead_strip_inits_and_terms
7393 @opindex nofixprebinding
7394 @opindex nomultidefs
7396 @opindex noseglinkedit
7397 @opindex pagezero_size
7399 @opindex prebind_all_twolevel_modules
7400 @opindex private_bundle
7401 @opindex read_only_relocs
7403 @opindex sectobjectsymbols
7407 @opindex sectobjectsymbols
7410 @opindex segs_read_only_addr
7411 @opindex segs_read_write_addr
7412 @opindex seg_addr_table
7413 @opindex seg_addr_table_filename
7414 @opindex seglinkedit
7416 @opindex segs_read_only_addr
7417 @opindex segs_read_write_addr
7418 @opindex single_module
7420 @opindex sub_library
7421 @opindex sub_umbrella
7422 @opindex twolevel_namespace
7425 @opindex unexported_symbols_list
7426 @opindex weak_reference_mismatches
7427 @opindex whatsloaded
7429 These options are passed to the Darwin linker. The Darwin linker man page
7430 describes them in detail.
7433 @node DEC Alpha Options
7434 @subsection DEC Alpha Options
7436 These @samp{-m} options are defined for the DEC Alpha implementations:
7439 @item -mno-soft-float
7441 @opindex mno-soft-float
7442 @opindex msoft-float
7443 Use (do not use) the hardware floating-point instructions for
7444 floating-point operations. When @option{-msoft-float} is specified,
7445 functions in @file{libgcc.a} will be used to perform floating-point
7446 operations. Unless they are replaced by routines that emulate the
7447 floating-point operations, or compiled in such a way as to call such
7448 emulations routines, these routines will issue floating-point
7449 operations. If you are compiling for an Alpha without floating-point
7450 operations, you must ensure that the library is built so as not to call
7453 Note that Alpha implementations without floating-point operations are
7454 required to have floating-point registers.
7459 @opindex mno-fp-regs
7460 Generate code that uses (does not use) the floating-point register set.
7461 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7462 register set is not used, floating point operands are passed in integer
7463 registers as if they were integers and floating-point results are passed
7464 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
7465 so any function with a floating-point argument or return value called by code
7466 compiled with @option{-mno-fp-regs} must also be compiled with that
7469 A typical use of this option is building a kernel that does not use,
7470 and hence need not save and restore, any floating-point registers.
7474 The Alpha architecture implements floating-point hardware optimized for
7475 maximum performance. It is mostly compliant with the IEEE floating
7476 point standard. However, for full compliance, software assistance is
7477 required. This option generates code fully IEEE compliant code
7478 @emph{except} that the @var{inexact-flag} is not maintained (see below).
7479 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
7480 defined during compilation. The resulting code is less efficient but is
7481 able to correctly support denormalized numbers and exceptional IEEE
7482 values such as not-a-number and plus/minus infinity. Other Alpha
7483 compilers call this option @option{-ieee_with_no_inexact}.
7485 @item -mieee-with-inexact
7486 @opindex mieee-with-inexact
7487 This is like @option{-mieee} except the generated code also maintains
7488 the IEEE @var{inexact-flag}. Turning on this option causes the
7489 generated code to implement fully-compliant IEEE math. In addition to
7490 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
7491 macro. On some Alpha implementations the resulting code may execute
7492 significantly slower than the code generated by default. Since there is
7493 very little code that depends on the @var{inexact-flag}, you should
7494 normally not specify this option. Other Alpha compilers call this
7495 option @option{-ieee_with_inexact}.
7497 @item -mfp-trap-mode=@var{trap-mode}
7498 @opindex mfp-trap-mode
7499 This option controls what floating-point related traps are enabled.
7500 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
7501 The trap mode can be set to one of four values:
7505 This is the default (normal) setting. The only traps that are enabled
7506 are the ones that cannot be disabled in software (e.g., division by zero
7510 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7514 Like @samp{su}, but the instructions are marked to be safe for software
7515 completion (see Alpha architecture manual for details).
7518 Like @samp{su}, but inexact traps are enabled as well.
7521 @item -mfp-rounding-mode=@var{rounding-mode}
7522 @opindex mfp-rounding-mode
7523 Selects the IEEE rounding mode. Other Alpha compilers call this option
7524 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
7529 Normal IEEE rounding mode. Floating point numbers are rounded towards
7530 the nearest machine number or towards the even machine number in case
7534 Round towards minus infinity.
7537 Chopped rounding mode. Floating point numbers are rounded towards zero.
7540 Dynamic rounding mode. A field in the floating point control register
7541 (@var{fpcr}, see Alpha architecture reference manual) controls the
7542 rounding mode in effect. The C library initializes this register for
7543 rounding towards plus infinity. Thus, unless your program modifies the
7544 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7547 @item -mtrap-precision=@var{trap-precision}
7548 @opindex mtrap-precision
7549 In the Alpha architecture, floating point traps are imprecise. This
7550 means without software assistance it is impossible to recover from a
7551 floating trap and program execution normally needs to be terminated.
7552 GCC can generate code that can assist operating system trap handlers
7553 in determining the exact location that caused a floating point trap.
7554 Depending on the requirements of an application, different levels of
7555 precisions can be selected:
7559 Program precision. This option is the default and means a trap handler
7560 can only identify which program caused a floating point exception.
7563 Function precision. The trap handler can determine the function that
7564 caused a floating point exception.
7567 Instruction precision. The trap handler can determine the exact
7568 instruction that caused a floating point exception.
7571 Other Alpha compilers provide the equivalent options called
7572 @option{-scope_safe} and @option{-resumption_safe}.
7574 @item -mieee-conformant
7575 @opindex mieee-conformant
7576 This option marks the generated code as IEEE conformant. You must not
7577 use this option unless you also specify @option{-mtrap-precision=i} and either
7578 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
7579 is to emit the line @samp{.eflag 48} in the function prologue of the
7580 generated assembly file. Under DEC Unix, this has the effect that
7581 IEEE-conformant math library routines will be linked in.
7583 @item -mbuild-constants
7584 @opindex mbuild-constants
7585 Normally GCC examines a 32- or 64-bit integer constant to
7586 see if it can construct it from smaller constants in two or three
7587 instructions. If it cannot, it will output the constant as a literal and
7588 generate code to load it from the data segment at runtime.
7590 Use this option to require GCC to construct @emph{all} integer constants
7591 using code, even if it takes more instructions (the maximum is six).
7593 You would typically use this option to build a shared library dynamic
7594 loader. Itself a shared library, it must relocate itself in memory
7595 before it can find the variables and constants in its own data segment.
7601 Select whether to generate code to be assembled by the vendor-supplied
7602 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
7620 Indicate whether GCC should generate code to use the optional BWX,
7621 CIX, FIX and MAX instruction sets. The default is to use the instruction
7622 sets supported by the CPU type specified via @option{-mcpu=} option or that
7623 of the CPU on which GCC was built if none was specified.
7628 @opindex mfloat-ieee
7629 Generate code that uses (does not use) VAX F and G floating point
7630 arithmetic instead of IEEE single and double precision.
7632 @item -mexplicit-relocs
7633 @itemx -mno-explicit-relocs
7634 @opindex mexplicit-relocs
7635 @opindex mno-explicit-relocs
7636 Older Alpha assemblers provided no way to generate symbol relocations
7637 except via assembler macros. Use of these macros does not allow
7638 optimal instruction scheduling. GNU binutils as of version 2.12
7639 supports a new syntax that allows the compiler to explicitly mark
7640 which relocations should apply to which instructions. This option
7641 is mostly useful for debugging, as GCC detects the capabilities of
7642 the assembler when it is built and sets the default accordingly.
7646 @opindex msmall-data
7647 @opindex mlarge-data
7648 When @option{-mexplicit-relocs} is in effect, static data is
7649 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
7650 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
7651 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
7652 16-bit relocations off of the @code{$gp} register. This limits the
7653 size of the small data area to 64KB, but allows the variables to be
7654 directly accessed via a single instruction.
7656 The default is @option{-mlarge-data}. With this option the data area
7657 is limited to just below 2GB. Programs that require more than 2GB of
7658 data must use @code{malloc} or @code{mmap} to allocate the data in the
7659 heap instead of in the program's data segment.
7661 When generating code for shared libraries, @option{-fpic} implies
7662 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
7666 @opindex msmall-text
7667 @opindex mlarge-text
7668 When @option{-msmall-text} is used, the compiler assumes that the
7669 code of the entire program (or shared library) fits in 4MB, and is
7670 thus reachable with a branch instruction. When @option{-msmall-data}
7671 is used, the compiler can assume that all local symbols share the
7672 same @code{$gp} value, and thus reduce the number of instructions
7673 required for a function call from 4 to 1.
7675 The default is @option{-mlarge-text}.
7677 @item -mcpu=@var{cpu_type}
7679 Set the instruction set and instruction scheduling parameters for
7680 machine type @var{cpu_type}. You can specify either the @samp{EV}
7681 style name or the corresponding chip number. GCC supports scheduling
7682 parameters for the EV4, EV5 and EV6 family of processors and will
7683 choose the default values for the instruction set from the processor
7684 you specify. If you do not specify a processor type, GCC will default
7685 to the processor on which the compiler was built.
7687 Supported values for @var{cpu_type} are
7693 Schedules as an EV4 and has no instruction set extensions.
7697 Schedules as an EV5 and has no instruction set extensions.
7701 Schedules as an EV5 and supports the BWX extension.
7706 Schedules as an EV5 and supports the BWX and MAX extensions.
7710 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
7714 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
7717 @item -mtune=@var{cpu_type}
7719 Set only the instruction scheduling parameters for machine type
7720 @var{cpu_type}. The instruction set is not changed.
7722 @item -mmemory-latency=@var{time}
7723 @opindex mmemory-latency
7724 Sets the latency the scheduler should assume for typical memory
7725 references as seen by the application. This number is highly
7726 dependent on the memory access patterns used by the application
7727 and the size of the external cache on the machine.
7729 Valid options for @var{time} are
7733 A decimal number representing clock cycles.
7739 The compiler contains estimates of the number of clock cycles for
7740 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7741 (also called Dcache, Scache, and Bcache), as well as to main memory.
7742 Note that L3 is only valid for EV5.
7747 @node DEC Alpha/VMS Options
7748 @subsection DEC Alpha/VMS Options
7750 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
7753 @item -mvms-return-codes
7754 @opindex mvms-return-codes
7755 Return VMS condition codes from main. The default is to return POSIX
7756 style condition (e.g.@ error) codes.
7760 @subsection FRV Options
7767 Only use the first 32 general purpose registers.
7772 Use all 64 general purpose registers.
7777 Use only the first 32 floating point registers.
7782 Use all 64 floating point registers
7785 @opindex mhard-float
7787 Use hardware instructions for floating point operations.
7790 @opindex msoft-float
7792 Use library routines for floating point operations.
7797 Dynamically allocate condition code registers.
7802 Do not try to dynamically allocate condition code registers, only
7803 use @code{icc0} and @code{fcc0}.
7808 Change ABI to use double word insns.
7813 Do not use double word instructions.
7818 Use floating point double instructions.
7823 Do not use floating point double instructions.
7828 Use media instructions.
7833 Do not use media instructions.
7838 Use multiply and add/subtract instructions.
7843 Do not use multiply and add/subtract instructions.
7848 Select the FDPIC ABI, that uses function descriptors to represent
7849 pointers to functions. Without any PIC/PIE-related options, it
7850 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
7851 assumes GOT entries and small data are within a 12-bit range from the
7852 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
7853 are computed with 32 bits.
7856 @opindex minline-plt
7858 Enable inlining of PLT entries in function calls to functions that are
7859 not known to bind locally. It has no effect without @option{-mfdpic}.
7860 It's enabled by default if optimizing for speed and compiling for
7861 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
7862 optimization option such as @option{-O3} or above is present in the
7868 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
7869 that is known to be in read-only sections. It's enabled by default,
7870 except for @option{-fpic} or @option{-fpie}: even though it may help
7871 make the global offset table smaller, it trades 1 instruction for 4.
7872 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
7873 one of which may be shared by multiple symbols, and it avoids the need
7874 for a GOT entry for the referenced symbol, so it's more likely to be a
7875 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
7877 @item -multilib-library-pic
7878 @opindex multilib-library-pic
7880 Link with the (library, not FD) pic libraries. It's implied by
7881 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
7882 @option{-fpic} without @option{-mfdpic}. You should never have to use
7888 Follow the EABI requirement of always creating a frame pointer whenever
7889 a stack frame is allocated. This option is enabled by default and can
7890 be disabled with @option{-mno-linked-fp}.
7893 @opindex mlong-calls
7895 Use indirect addressing to call functions outside the current
7896 compilation unit. This allows the functions to be placed anywhere
7897 within the 32-bit address space.
7899 @item -malign-labels
7900 @opindex malign-labels
7902 Try to align labels to an 8-byte boundary by inserting nops into the
7903 previous packet. This option only has an effect when VLIW packing
7904 is enabled. It doesn't create new packets; it merely adds nops to
7908 @opindex mlibrary-pic
7910 Generate position-independent EABI code.
7915 Use only the first four media accumulator registers.
7920 Use all eight media accumulator registers.
7925 Pack VLIW instructions.
7930 Do not pack VLIW instructions.
7935 Do not mark ABI switches in e_flags.
7940 Enable the use of conditional-move instructions (default).
7942 This switch is mainly for debugging the compiler and will likely be removed
7943 in a future version.
7945 @item -mno-cond-move
7946 @opindex mno-cond-move
7948 Disable the use of conditional-move instructions.
7950 This switch is mainly for debugging the compiler and will likely be removed
7951 in a future version.
7956 Enable the use of conditional set instructions (default).
7958 This switch is mainly for debugging the compiler and will likely be removed
7959 in a future version.
7964 Disable the use of conditional set instructions.
7966 This switch is mainly for debugging the compiler and will likely be removed
7967 in a future version.
7972 Enable the use of conditional execution (default).
7974 This switch is mainly for debugging the compiler and will likely be removed
7975 in a future version.
7977 @item -mno-cond-exec
7978 @opindex mno-cond-exec
7980 Disable the use of conditional execution.
7982 This switch is mainly for debugging the compiler and will likely be removed
7983 in a future version.
7986 @opindex mvliw-branch
7988 Run a pass to pack branches into VLIW instructions (default).
7990 This switch is mainly for debugging the compiler and will likely be removed
7991 in a future version.
7993 @item -mno-vliw-branch
7994 @opindex mno-vliw-branch
7996 Do not run a pass to pack branches into VLIW instructions.
7998 This switch is mainly for debugging the compiler and will likely be removed
7999 in a future version.
8001 @item -mmulti-cond-exec
8002 @opindex mmulti-cond-exec
8004 Enable optimization of @code{&&} and @code{||} in conditional execution
8007 This switch is mainly for debugging the compiler and will likely be removed
8008 in a future version.
8010 @item -mno-multi-cond-exec
8011 @opindex mno-multi-cond-exec
8013 Disable optimization of @code{&&} and @code{||} in conditional execution.
8015 This switch is mainly for debugging the compiler and will likely be removed
8016 in a future version.
8018 @item -mnested-cond-exec
8019 @opindex mnested-cond-exec
8021 Enable nested conditional execution optimizations (default).
8023 This switch is mainly for debugging the compiler and will likely be removed
8024 in a future version.
8026 @item -mno-nested-cond-exec
8027 @opindex mno-nested-cond-exec
8029 Disable nested conditional execution optimizations.
8031 This switch is mainly for debugging the compiler and will likely be removed
8032 in a future version.
8034 @item -mtomcat-stats
8035 @opindex mtomcat-stats
8037 Cause gas to print out tomcat statistics.
8039 @item -mcpu=@var{cpu}
8042 Select the processor type for which to generate code. Possible values are
8043 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8044 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8048 @node H8/300 Options
8049 @subsection H8/300 Options
8051 These @samp{-m} options are defined for the H8/300 implementations:
8056 Shorten some address references at link time, when possible; uses the
8057 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8058 ld, Using ld}, for a fuller description.
8062 Generate code for the H8/300H@.
8066 Generate code for the H8S@.
8070 Generate code for the H8S and H8/300H in the normal mode. This switch
8071 must be used either with @option{-mh} or @option{-ms}.
8075 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8079 Make @code{int} data 32 bits by default.
8083 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8084 The default for the H8/300H and H8S is to align longs and floats on 4
8086 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8087 This option has no effect on the H8/300.
8091 @subsection HPPA Options
8092 @cindex HPPA Options
8094 These @samp{-m} options are defined for the HPPA family of computers:
8097 @item -march=@var{architecture-type}
8099 Generate code for the specified architecture. The choices for
8100 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8101 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8102 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8103 architecture option for your machine. Code compiled for lower numbered
8104 architectures will run on higher numbered architectures, but not the
8108 @itemx -mpa-risc-1-1
8109 @itemx -mpa-risc-2-0
8110 @opindex mpa-risc-1-0
8111 @opindex mpa-risc-1-1
8112 @opindex mpa-risc-2-0
8113 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8116 @opindex mbig-switch
8117 Generate code suitable for big switch tables. Use this option only if
8118 the assembler/linker complain about out of range branches within a switch
8121 @item -mjump-in-delay
8122 @opindex mjump-in-delay
8123 Fill delay slots of function calls with unconditional jump instructions
8124 by modifying the return pointer for the function call to be the target
8125 of the conditional jump.
8127 @item -mdisable-fpregs
8128 @opindex mdisable-fpregs
8129 Prevent floating point registers from being used in any manner. This is
8130 necessary for compiling kernels which perform lazy context switching of
8131 floating point registers. If you use this option and attempt to perform
8132 floating point operations, the compiler will abort.
8134 @item -mdisable-indexing
8135 @opindex mdisable-indexing
8136 Prevent the compiler from using indexing address modes. This avoids some
8137 rather obscure problems when compiling MIG generated code under MACH@.
8139 @item -mno-space-regs
8140 @opindex mno-space-regs
8141 Generate code that assumes the target has no space registers. This allows
8142 GCC to generate faster indirect calls and use unscaled index address modes.
8144 Such code is suitable for level 0 PA systems and kernels.
8146 @item -mfast-indirect-calls
8147 @opindex mfast-indirect-calls
8148 Generate code that assumes calls never cross space boundaries. This
8149 allows GCC to emit code which performs faster indirect calls.
8151 This option will not work in the presence of shared libraries or nested
8154 @item -mfixed-range=@var{register-range}
8155 @opindex mfixed-range
8156 Generate code treating the given register range as fixed registers.
8157 A fixed register is one that the register allocator can not use. This is
8158 useful when compiling kernel code. A register range is specified as
8159 two registers separated by a dash. Multiple register ranges can be
8160 specified separated by a comma.
8162 @item -mlong-load-store
8163 @opindex mlong-load-store
8164 Generate 3-instruction load and store sequences as sometimes required by
8165 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8168 @item -mportable-runtime
8169 @opindex mportable-runtime
8170 Use the portable calling conventions proposed by HP for ELF systems.
8174 Enable the use of assembler directives only GAS understands.
8176 @item -mschedule=@var{cpu-type}
8178 Schedule code according to the constraints for the machine type
8179 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8180 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8181 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8182 proper scheduling option for your machine. The default scheduling is
8186 @opindex mlinker-opt
8187 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8188 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8189 linkers in which they give bogus error messages when linking some programs.
8192 @opindex msoft-float
8193 Generate output containing library calls for floating point.
8194 @strong{Warning:} the requisite libraries are not available for all HPPA
8195 targets. Normally the facilities of the machine's usual C compiler are
8196 used, but this cannot be done directly in cross-compilation. You must make
8197 your own arrangements to provide suitable library functions for
8198 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8199 does provide software floating point support.
8201 @option{-msoft-float} changes the calling convention in the output file;
8202 therefore, it is only useful if you compile @emph{all} of a program with
8203 this option. In particular, you need to compile @file{libgcc.a}, the
8204 library that comes with GCC, with @option{-msoft-float} in order for
8209 Generate the predefine, @code{_SIO}, for server IO. The default is
8210 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8211 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8212 options are available under HP-UX and HI-UX.
8216 Use GNU ld specific options. This passes @option{-shared} to ld when
8217 building a shared library. It is the default when GCC is configured,
8218 explicitly or implicitly, with the GNU linker. This option does not
8219 have any affect on which ld is called, it only changes what parameters
8220 are passed to that ld. The ld that is called is determined by the
8221 @option{--with-ld} configure option, GCC's program search path, and
8222 finally by the user's @env{PATH}. The linker used by GCC can be printed
8223 using @samp{which `gcc -print-prog-name=ld`}.
8227 Use HP ld specific options. This passes @option{-b} to ld when building
8228 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8229 links. It is the default when GCC is configured, explicitly or
8230 implicitly, with the HP linker. This option does not have any affect on
8231 which ld is called, it only changes what parameters are passed to that
8232 ld. The ld that is called is determined by the @option{--with-ld}
8233 configure option, GCC's program search path, and finally by the user's
8234 @env{PATH}. The linker used by GCC can be printed using @samp{which
8235 `gcc -print-prog-name=ld`}.
8238 @opindex mno-long-calls
8239 Generate code that uses long call sequences. This ensures that a call
8240 is always able to reach linker generated stubs. The default is to generate
8241 long calls only when the distance from the call site to the beginning
8242 of the function or translation unit, as the case may be, exceeds a
8243 predefined limit set by the branch type being used. The limits for
8244 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8245 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8248 Distances are measured from the beginning of functions when using the
8249 @option{-ffunction-sections} option, or when using the @option{-mgas}
8250 and @option{-mno-portable-runtime} options together under HP-UX with
8253 It is normally not desirable to use this option as it will degrade
8254 performance. However, it may be useful in large applications,
8255 particularly when partial linking is used to build the application.
8257 The types of long calls used depends on the capabilities of the
8258 assembler and linker, and the type of code being generated. The
8259 impact on systems that support long absolute calls, and long pic
8260 symbol-difference or pc-relative calls should be relatively small.
8261 However, an indirect call is used on 32-bit ELF systems in pic code
8262 and it is quite long.
8264 @item -munix=@var{unix-std}
8266 Generate compiler predefines and select a startfile for the specified
8267 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
8268 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
8269 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
8270 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
8271 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
8274 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
8275 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
8276 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
8277 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
8278 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
8279 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
8281 It is @emph{important} to note that this option changes the interfaces
8282 for various library routines. It also affects the operational behavior
8283 of the C library. Thus, @emph{extreme} care is needed in using this
8286 Library code that is intended to operate with more than one UNIX
8287 standard must test, set and restore the variable @var{__xpg4_extended_mask}
8288 as appropriate. Most GNU software doesn't provide this capability.
8292 Suppress the generation of link options to search libdld.sl when the
8293 @option{-static} option is specified on HP-UX 10 and later.
8297 The HP-UX implementation of setlocale in libc has a dependency on
8298 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8299 when the @option{-static} option is specified, special link options
8300 are needed to resolve this dependency.
8302 On HP-UX 10 and later, the GCC driver adds the necessary options to
8303 link with libdld.sl when the @option{-static} option is specified.
8304 This causes the resulting binary to be dynamic. On the 64-bit port,
8305 the linkers generate dynamic binaries by default in any case. The
8306 @option{-nolibdld} option can be used to prevent the GCC driver from
8307 adding these link options.
8311 Add support for multithreading with the @dfn{dce thread} library
8312 under HP-UX. This option sets flags for both the preprocessor and
8316 @node i386 and x86-64 Options
8317 @subsection Intel 386 and AMD x86-64 Options
8318 @cindex i386 Options
8319 @cindex x86-64 Options
8320 @cindex Intel 386 Options
8321 @cindex AMD x86-64 Options
8323 These @samp{-m} options are defined for the i386 and x86-64 family of
8327 @item -mtune=@var{cpu-type}
8329 Tune to @var{cpu-type} everything applicable about the generated code, except
8330 for the ABI and the set of available instructions. The choices for
8334 Original Intel's i386 CPU.
8336 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8338 Intel Pentium CPU with no MMX support.
8340 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8341 @item i686, pentiumpro
8342 Intel PentiumPro CPU.
8344 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8345 @item pentium3, pentium3m
8346 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8349 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8350 support. Used by Centrino notebooks.
8351 @item pentium4, pentium4m
8352 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8354 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8357 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8358 SSE2 and SSE3 instruction set support.
8360 AMD K6 CPU with MMX instruction set support.
8362 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8363 @item athlon, athlon-tbird
8364 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8366 @item athlon-4, athlon-xp, athlon-mp
8367 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8368 instruction set support.
8369 @item k8, opteron, athlon64, athlon-fx
8370 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8371 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8373 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8376 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8377 instruction set support.
8379 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8380 implemented for this chip.)
8382 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8383 implemented for this chip.)
8386 While picking a specific @var{cpu-type} will schedule things appropriately
8387 for that particular chip, the compiler will not generate any code that
8388 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8391 @item -march=@var{cpu-type}
8393 Generate instructions for the machine type @var{cpu-type}. The choices
8394 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8395 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8397 @item -mcpu=@var{cpu-type}
8399 A deprecated synonym for @option{-mtune}.
8408 @opindex mpentiumpro
8409 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8410 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8411 These synonyms are deprecated.
8413 @item -mfpmath=@var{unit}
8415 Generate floating point arithmetics for selected unit @var{unit}. The choices
8420 Use the standard 387 floating point coprocessor present majority of chips and
8421 emulated otherwise. Code compiled with this option will run almost everywhere.
8422 The temporary results are computed in 80bit precision instead of precision
8423 specified by the type resulting in slightly different results compared to most
8424 of other chips. See @option{-ffloat-store} for more detailed description.
8426 This is the default choice for i386 compiler.
8429 Use scalar floating point instructions present in the SSE instruction set.
8430 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8431 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8432 instruction set supports only single precision arithmetics, thus the double and
8433 extended precision arithmetics is still done using 387. Later version, present
8434 only in Pentium4 and the future AMD x86-64 chips supports double precision
8437 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8438 @option{-msse2} switches to enable SSE extensions and make this option
8439 effective. For x86-64 compiler, these extensions are enabled by default.
8441 The resulting code should be considerably faster in the majority of cases and avoid
8442 the numerical instability problems of 387 code, but may break some existing
8443 code that expects temporaries to be 80bit.
8445 This is the default choice for the x86-64 compiler.
8448 Attempt to utilize both instruction sets at once. This effectively double the
8449 amount of available registers and on chips with separate execution units for
8450 387 and SSE the execution resources too. Use this option with care, as it is
8451 still experimental, because the GCC register allocator does not model separate
8452 functional units well resulting in instable performance.
8455 @item -masm=@var{dialect}
8456 @opindex masm=@var{dialect}
8457 Output asm instructions using selected @var{dialect}. Supported choices are
8458 @samp{intel} or @samp{att} (the default one).
8463 @opindex mno-ieee-fp
8464 Control whether or not the compiler uses IEEE floating point
8465 comparisons. These handle correctly the case where the result of a
8466 comparison is unordered.
8469 @opindex msoft-float
8470 Generate output containing library calls for floating point.
8471 @strong{Warning:} the requisite libraries are not part of GCC@.
8472 Normally the facilities of the machine's usual C compiler are used, but
8473 this can't be done directly in cross-compilation. You must make your
8474 own arrangements to provide suitable library functions for
8477 On machines where a function returns floating point results in the 80387
8478 register stack, some floating point opcodes may be emitted even if
8479 @option{-msoft-float} is used.
8481 @item -mno-fp-ret-in-387
8482 @opindex mno-fp-ret-in-387
8483 Do not use the FPU registers for return values of functions.
8485 The usual calling convention has functions return values of types
8486 @code{float} and @code{double} in an FPU register, even if there
8487 is no FPU@. The idea is that the operating system should emulate
8490 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8491 in ordinary CPU registers instead.
8493 @item -mno-fancy-math-387
8494 @opindex mno-fancy-math-387
8495 Some 387 emulators do not support the @code{sin}, @code{cos} and
8496 @code{sqrt} instructions for the 387. Specify this option to avoid
8497 generating those instructions. This option is the default on FreeBSD,
8498 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8499 indicates that the target cpu will always have an FPU and so the
8500 instruction will not need emulation. As of revision 2.6.1, these
8501 instructions are not generated unless you also use the
8502 @option{-funsafe-math-optimizations} switch.
8504 @item -malign-double
8505 @itemx -mno-align-double
8506 @opindex malign-double
8507 @opindex mno-align-double
8508 Control whether GCC aligns @code{double}, @code{long double}, and
8509 @code{long long} variables on a two word boundary or a one word
8510 boundary. Aligning @code{double} variables on a two word boundary will
8511 produce code that runs somewhat faster on a @samp{Pentium} at the
8512 expense of more memory.
8514 @strong{Warning:} if you use the @option{-malign-double} switch,
8515 structures containing the above types will be aligned differently than
8516 the published application binary interface specifications for the 386
8517 and will not be binary compatible with structures in code compiled
8518 without that switch.
8520 @item -m96bit-long-double
8521 @itemx -m128bit-long-double
8522 @opindex m96bit-long-double
8523 @opindex m128bit-long-double
8524 These switches control the size of @code{long double} type. The i386
8525 application binary interface specifies the size to be 96 bits,
8526 so @option{-m96bit-long-double} is the default in 32 bit mode.
8528 Modern architectures (Pentium and newer) would prefer @code{long double}
8529 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8530 conforming to the ABI, this would not be possible. So specifying a
8531 @option{-m128bit-long-double} will align @code{long double}
8532 to a 16 byte boundary by padding the @code{long double} with an additional
8535 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8536 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8538 Notice that neither of these options enable any extra precision over the x87
8539 standard of 80 bits for a @code{long double}.
8541 @strong{Warning:} if you override the default value for your target ABI, the
8542 structures and arrays containing @code{long double} variables will change
8543 their size as well as function calling convention for function taking
8544 @code{long double} will be modified. Hence they will not be binary
8545 compatible with arrays or structures in code compiled without that switch.
8549 @itemx -mno-svr3-shlib
8550 @opindex msvr3-shlib
8551 @opindex mno-svr3-shlib
8552 Control whether GCC places uninitialized local variables into the
8553 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8554 into @code{bss}. These options are meaningful only on System V Release 3.
8558 Use a different function-calling convention, in which functions that
8559 take a fixed number of arguments return with the @code{ret} @var{num}
8560 instruction, which pops their arguments while returning. This saves one
8561 instruction in the caller since there is no need to pop the arguments
8564 You can specify that an individual function is called with this calling
8565 sequence with the function attribute @samp{stdcall}. You can also
8566 override the @option{-mrtd} option by using the function attribute
8567 @samp{cdecl}. @xref{Function Attributes}.
8569 @strong{Warning:} this calling convention is incompatible with the one
8570 normally used on Unix, so you cannot use it if you need to call
8571 libraries compiled with the Unix compiler.
8573 Also, you must provide function prototypes for all functions that
8574 take variable numbers of arguments (including @code{printf});
8575 otherwise incorrect code will be generated for calls to those
8578 In addition, seriously incorrect code will result if you call a
8579 function with too many arguments. (Normally, extra arguments are
8580 harmlessly ignored.)
8582 @item -mregparm=@var{num}
8584 Control how many registers are used to pass integer arguments. By
8585 default, no registers are used to pass arguments, and at most 3
8586 registers can be used. You can control this behavior for a specific
8587 function by using the function attribute @samp{regparm}.
8588 @xref{Function Attributes}.
8590 @strong{Warning:} if you use this switch, and
8591 @var{num} is nonzero, then you must build all modules with the same
8592 value, including any libraries. This includes the system libraries and
8595 @item -mpreferred-stack-boundary=@var{num}
8596 @opindex mpreferred-stack-boundary
8597 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8598 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8599 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8600 size (@option{-Os}), in which case the default is the minimum correct
8601 alignment (4 bytes for x86, and 8 bytes for x86-64).
8603 On Pentium and PentiumPro, @code{double} and @code{long double} values
8604 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8605 suffer significant run time performance penalties. On Pentium III, the
8606 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8607 penalties if it is not 16 byte aligned.
8609 To ensure proper alignment of this values on the stack, the stack boundary
8610 must be as aligned as that required by any value stored on the stack.
8611 Further, every function must be generated such that it keeps the stack
8612 aligned. Thus calling a function compiled with a higher preferred
8613 stack boundary from a function compiled with a lower preferred stack
8614 boundary will most likely misalign the stack. It is recommended that
8615 libraries that use callbacks always use the default setting.
8617 This extra alignment does consume extra stack space, and generally
8618 increases code size. Code that is sensitive to stack space usage, such
8619 as embedded systems and operating system kernels, may want to reduce the
8620 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8638 These switches enable or disable the use of built-in functions that allow
8639 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8642 @xref{X86 Built-in Functions}, for details of the functions enabled
8643 and disabled by these switches.
8645 To have SSE/SSE2 instructions generated automatically from floating-point
8646 code, see @option{-mfpmath=sse}.
8649 @itemx -mno-push-args
8651 @opindex mno-push-args
8652 Use PUSH operations to store outgoing parameters. This method is shorter
8653 and usually equally fast as method using SUB/MOV operations and is enabled
8654 by default. In some cases disabling it may improve performance because of
8655 improved scheduling and reduced dependencies.
8657 @item -maccumulate-outgoing-args
8658 @opindex maccumulate-outgoing-args
8659 If enabled, the maximum amount of space required for outgoing arguments will be
8660 computed in the function prologue. This is faster on most modern CPUs
8661 because of reduced dependencies, improved scheduling and reduced stack usage
8662 when preferred stack boundary is not equal to 2. The drawback is a notable
8663 increase in code size. This switch implies @option{-mno-push-args}.
8667 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8668 on thread-safe exception handling must compile and link all code with the
8669 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8670 @option{-D_MT}; when linking, it links in a special thread helper library
8671 @option{-lmingwthrd} which cleans up per thread exception handling data.
8673 @item -mno-align-stringops
8674 @opindex mno-align-stringops
8675 Do not align destination of inlined string operations. This switch reduces
8676 code size and improves performance in case the destination is already aligned,
8677 but GCC doesn't know about it.
8679 @item -minline-all-stringops
8680 @opindex minline-all-stringops
8681 By default GCC inlines string operations only when destination is known to be
8682 aligned at least to 4 byte boundary. This enables more inlining, increase code
8683 size, but may improve performance of code that depends on fast memcpy, strlen
8684 and memset for short lengths.
8686 @item -momit-leaf-frame-pointer
8687 @opindex momit-leaf-frame-pointer
8688 Don't keep the frame pointer in a register for leaf functions. This
8689 avoids the instructions to save, set up and restore frame pointers and
8690 makes an extra register available in leaf functions. The option
8691 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8692 which might make debugging harder.
8694 @item -mtls-direct-seg-refs
8695 @itemx -mno-tls-direct-seg-refs
8696 @opindex mtls-direct-seg-refs
8697 Controls whether TLS variables may be accessed with offsets from the
8698 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8699 or whether the thread base pointer must be added. Whether or not this
8700 is legal depends on the operating system, and whether it maps the
8701 segment to cover the entire TLS area.
8703 For systems that use GNU libc, the default is on.
8706 These @samp{-m} switches are supported in addition to the above
8707 on AMD x86-64 processors in 64-bit environments.
8714 Generate code for a 32-bit or 64-bit environment.
8715 The 32-bit environment sets int, long and pointer to 32 bits and
8716 generates code that runs on any i386 system.
8717 The 64-bit environment sets int to 32 bits and long and pointer
8718 to 64 bits and generates code for AMD's x86-64 architecture.
8721 @opindex no-red-zone
8722 Do not use a so called red zone for x86-64 code. The red zone is mandated
8723 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8724 stack pointer that will not be modified by signal or interrupt handlers
8725 and therefore can be used for temporary data without adjusting the stack
8726 pointer. The flag @option{-mno-red-zone} disables this red zone.
8728 @item -mcmodel=small
8729 @opindex mcmodel=small
8730 Generate code for the small code model: the program and its symbols must
8731 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8732 Programs can be statically or dynamically linked. This is the default
8735 @item -mcmodel=kernel
8736 @opindex mcmodel=kernel
8737 Generate code for the kernel code model. The kernel runs in the
8738 negative 2 GB of the address space.
8739 This model has to be used for Linux kernel code.
8741 @item -mcmodel=medium
8742 @opindex mcmodel=medium
8743 Generate code for the medium model: The program is linked in the lower 2
8744 GB of the address space but symbols can be located anywhere in the
8745 address space. Programs can be statically or dynamically linked, but
8746 building of shared libraries are not supported with the medium model.
8748 @item -mcmodel=large
8749 @opindex mcmodel=large
8750 Generate code for the large model: This model makes no assumptions
8751 about addresses and sizes of sections. Currently GCC does not implement
8756 @subsection IA-64 Options
8757 @cindex IA-64 Options
8759 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8763 @opindex mbig-endian
8764 Generate code for a big endian target. This is the default for HP-UX@.
8766 @item -mlittle-endian
8767 @opindex mlittle-endian
8768 Generate code for a little endian target. This is the default for AIX5
8775 Generate (or don't) code for the GNU assembler. This is the default.
8776 @c Also, this is the default if the configure option @option{--with-gnu-as}
8783 Generate (or don't) code for the GNU linker. This is the default.
8784 @c Also, this is the default if the configure option @option{--with-gnu-ld}
8789 Generate code that does not use a global pointer register. The result
8790 is not position independent code, and violates the IA-64 ABI@.
8792 @item -mvolatile-asm-stop
8793 @itemx -mno-volatile-asm-stop
8794 @opindex mvolatile-asm-stop
8795 @opindex mno-volatile-asm-stop
8796 Generate (or don't) a stop bit immediately before and after volatile asm
8801 Generate code that works around Itanium B step errata.
8803 @item -mregister-names
8804 @itemx -mno-register-names
8805 @opindex mregister-names
8806 @opindex mno-register-names
8807 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8808 the stacked registers. This may make assembler output more readable.
8814 Disable (or enable) optimizations that use the small data section. This may
8815 be useful for working around optimizer bugs.
8818 @opindex mconstant-gp
8819 Generate code that uses a single constant global pointer value. This is
8820 useful when compiling kernel code.
8824 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
8825 This is useful when compiling firmware code.
8827 @item -minline-float-divide-min-latency
8828 @opindex minline-float-divide-min-latency
8829 Generate code for inline divides of floating point values
8830 using the minimum latency algorithm.
8832 @item -minline-float-divide-max-throughput
8833 @opindex minline-float-divide-max-throughput
8834 Generate code for inline divides of floating point values
8835 using the maximum throughput algorithm.
8837 @item -minline-int-divide-min-latency
8838 @opindex minline-int-divide-min-latency
8839 Generate code for inline divides of integer values
8840 using the minimum latency algorithm.
8842 @item -minline-int-divide-max-throughput
8843 @opindex minline-int-divide-max-throughput
8844 Generate code for inline divides of integer values
8845 using the maximum throughput algorithm.
8847 @item -mno-dwarf2-asm
8849 @opindex mno-dwarf2-asm
8850 @opindex mdwarf2-asm
8851 Don't (or do) generate assembler code for the DWARF2 line number debugging
8852 info. This may be useful when not using the GNU assembler.
8854 @item -mfixed-range=@var{register-range}
8855 @opindex mfixed-range
8856 Generate code treating the given register range as fixed registers.
8857 A fixed register is one that the register allocator can not use. This is
8858 useful when compiling kernel code. A register range is specified as
8859 two registers separated by a dash. Multiple register ranges can be
8860 specified separated by a comma.
8862 @item -mearly-stop-bits
8863 @itemx -mno-early-stop-bits
8864 @opindex mearly-stop-bits
8865 @opindex mno-early-stop-bits
8866 Allow stop bits to be placed earlier than immediately preceding the
8867 instruction that triggered the stop bit. This can improve instruction
8868 scheduling, but does not always do so.
8871 @node M32R/D Options
8872 @subsection M32R/D Options
8873 @cindex M32R/D options
8875 These @option{-m} options are defined for Renesas M32R/D architectures:
8880 Generate code for the M32R/2@.
8884 Generate code for the M32R/X@.
8888 Generate code for the M32R@. This is the default.
8891 @opindex mmodel=small
8892 Assume all objects live in the lower 16MB of memory (so that their addresses
8893 can be loaded with the @code{ld24} instruction), and assume all subroutines
8894 are reachable with the @code{bl} instruction.
8895 This is the default.
8897 The addressability of a particular object can be set with the
8898 @code{model} attribute.
8900 @item -mmodel=medium
8901 @opindex mmodel=medium
8902 Assume objects may be anywhere in the 32-bit address space (the compiler
8903 will generate @code{seth/add3} instructions to load their addresses), and
8904 assume all subroutines are reachable with the @code{bl} instruction.
8907 @opindex mmodel=large
8908 Assume objects may be anywhere in the 32-bit address space (the compiler
8909 will generate @code{seth/add3} instructions to load their addresses), and
8910 assume subroutines may not be reachable with the @code{bl} instruction
8911 (the compiler will generate the much slower @code{seth/add3/jl}
8912 instruction sequence).
8915 @opindex msdata=none
8916 Disable use of the small data area. Variables will be put into
8917 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
8918 @code{section} attribute has been specified).
8919 This is the default.
8921 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
8922 Objects may be explicitly put in the small data area with the
8923 @code{section} attribute using one of these sections.
8926 @opindex msdata=sdata
8927 Put small global and static data in the small data area, but do not
8928 generate special code to reference them.
8932 Put small global and static data in the small data area, and generate
8933 special instructions to reference them.
8937 @cindex smaller data references
8938 Put global and static objects less than or equal to @var{num} bytes
8939 into the small data or bss sections instead of the normal data or bss
8940 sections. The default value of @var{num} is 8.
8941 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
8942 for this option to have any effect.
8944 All modules should be compiled with the same @option{-G @var{num}} value.
8945 Compiling with different values of @var{num} may or may not work; if it
8946 doesn't the linker will give an error message---incorrect code will not be
8951 Makes the M32R specific code in the compiler display some statistics
8952 that might help in debugging programs.
8955 @opindex malign-loops
8956 Align all loops to a 32-byte boundary.
8958 @item -mno-align-loops
8959 @opindex mno-align-loops
8960 Do not enforce a 32-byte alignment for loops. This is the default.
8962 @item -missue-rate=@var{number}
8963 @opindex missue-rate=@var{number}
8964 Issue @var{number} instructions per cycle. @var{number} can only be 1
8967 @item -mbranch-cost=@var{number}
8968 @opindex mbranch-cost=@var{number}
8969 @var{number} can only be 1 or 2. If it is 1 then branches will be
8970 preferred over conditional code, if it is 2, then the opposite will
8973 @item -mflush-trap=@var{number}
8974 @opindex mflush-trap=@var{number}
8975 Specifies the trap number to use to flush the cache. The default is
8976 12. Valid numbers are between 0 and 15 inclusive.
8978 @item -mno-flush-trap
8979 @opindex mno-flush-trap
8980 Specifies that the cache cannot be flushed by using a trap.
8982 @item -mflush-func=@var{name}
8983 @opindex mflush-func=@var{name}
8984 Specifies the name of the operating system function to call to flush
8985 the cache. The default is @emph{_flush_cache}, but a function call
8986 will only be used if a trap is not available.
8988 @item -mno-flush-func
8989 @opindex mno-flush-func
8990 Indicates that there is no OS function for flushing the cache.
8994 @node M680x0 Options
8995 @subsection M680x0 Options
8996 @cindex M680x0 options
8998 These are the @samp{-m} options defined for the 68000 series. The default
8999 values for these options depends on which style of 68000 was selected when
9000 the compiler was configured; the defaults for the most common choices are
9008 Generate output for a 68000. This is the default
9009 when the compiler is configured for 68000-based systems.
9011 Use this option for microcontrollers with a 68000 or EC000 core,
9012 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
9018 Generate output for a 68020. This is the default
9019 when the compiler is configured for 68020-based systems.
9023 Generate output containing 68881 instructions for floating point.
9024 This is the default for most 68020 systems unless @option{--nfp} was
9025 specified when the compiler was configured.
9029 Generate output for a 68030. This is the default when the compiler is
9030 configured for 68030-based systems.
9034 Generate output for a 68040. This is the default when the compiler is
9035 configured for 68040-based systems.
9037 This option inhibits the use of 68881/68882 instructions that have to be
9038 emulated by software on the 68040. Use this option if your 68040 does not
9039 have code to emulate those instructions.
9043 Generate output for a 68060. This is the default when the compiler is
9044 configured for 68060-based systems.
9046 This option inhibits the use of 68020 and 68881/68882 instructions that
9047 have to be emulated by software on the 68060. Use this option if your 68060
9048 does not have code to emulate those instructions.
9052 Generate output for a CPU32. This is the default
9053 when the compiler is configured for CPU32-based systems.
9055 Use this option for microcontrollers with a
9056 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
9057 68336, 68340, 68341, 68349 and 68360.
9061 Generate output for a 520X ``coldfire'' family cpu. This is the default
9062 when the compiler is configured for 520X-based systems.
9064 Use this option for microcontroller with a 5200 core, including
9065 the MCF5202, MCF5203, MCF5204 and MCF5202.
9070 Generate output for a 68040, without using any of the new instructions.
9071 This results in code which can run relatively efficiently on either a
9072 68020/68881 or a 68030 or a 68040. The generated code does use the
9073 68881 instructions that are emulated on the 68040.
9077 Generate output for a 68060, without using any of the new instructions.
9078 This results in code which can run relatively efficiently on either a
9079 68020/68881 or a 68030 or a 68040. The generated code does use the
9080 68881 instructions that are emulated on the 68060.
9083 @opindex msoft-float
9084 Generate output containing library calls for floating point.
9085 @strong{Warning:} the requisite libraries are not available for all m68k
9086 targets. Normally the facilities of the machine's usual C compiler are
9087 used, but this can't be done directly in cross-compilation. You must
9088 make your own arrangements to provide suitable library functions for
9089 cross-compilation. The embedded targets @samp{m68k-*-aout} and
9090 @samp{m68k-*-coff} do provide software floating point support.
9094 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9095 Additionally, parameters passed on the stack are also aligned to a
9096 16-bit boundary even on targets whose API mandates promotion to 32-bit.
9099 @opindex mnobitfield
9100 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
9101 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
9105 Do use the bit-field instructions. The @option{-m68020} option implies
9106 @option{-mbitfield}. This is the default if you use a configuration
9107 designed for a 68020.
9111 Use a different function-calling convention, in which functions
9112 that take a fixed number of arguments return with the @code{rtd}
9113 instruction, which pops their arguments while returning. This
9114 saves one instruction in the caller since there is no need to pop
9115 the arguments there.
9117 This calling convention is incompatible with the one normally
9118 used on Unix, so you cannot use it if you need to call libraries
9119 compiled with the Unix compiler.
9121 Also, you must provide function prototypes for all functions that
9122 take variable numbers of arguments (including @code{printf});
9123 otherwise incorrect code will be generated for calls to those
9126 In addition, seriously incorrect code will result if you call a
9127 function with too many arguments. (Normally, extra arguments are
9128 harmlessly ignored.)
9130 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
9131 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
9134 @itemx -mno-align-int
9136 @opindex mno-align-int
9137 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
9138 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
9139 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
9140 Aligning variables on 32-bit boundaries produces code that runs somewhat
9141 faster on processors with 32-bit busses at the expense of more memory.
9143 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
9144 align structures containing the above types differently than
9145 most published application binary interface specifications for the m68k.
9149 Use the pc-relative addressing mode of the 68000 directly, instead of
9150 using a global offset table. At present, this option implies @option{-fpic},
9151 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
9152 not presently supported with @option{-mpcrel}, though this could be supported for
9153 68020 and higher processors.
9155 @item -mno-strict-align
9156 @itemx -mstrict-align
9157 @opindex mno-strict-align
9158 @opindex mstrict-align
9159 Do not (do) assume that unaligned memory references will be handled by
9163 Generate code that allows the data segment to be located in a different
9164 area of memory from the text segment. This allows for execute in place in
9165 an environment without virtual memory management. This option implies
9169 Generate code that assumes that the data segment follows the text segment.
9170 This is the default.
9172 @item -mid-shared-library
9173 Generate code that supports shared libraries via the library ID method.
9174 This allows for execute in place and shared libraries in an environment
9175 without virtual memory management. This option implies @option{-fPIC}.
9177 @item -mno-id-shared-library
9178 Generate code that doesn't assume ID based shared libraries are being used.
9179 This is the default.
9181 @item -mshared-library-id=n
9182 Specified the identification number of the ID based shared library being
9183 compiled. Specifying a value of 0 will generate more compact code, specifying
9184 other values will force the allocation of that number to the current
9185 library but is no more space or time efficient than omitting this option.
9189 @node M68hc1x Options
9190 @subsection M68hc1x Options
9191 @cindex M68hc1x options
9193 These are the @samp{-m} options defined for the 68hc11 and 68hc12
9194 microcontrollers. The default values for these options depends on
9195 which style of microcontroller was selected when the compiler was configured;
9196 the defaults for the most common choices are given below.
9203 Generate output for a 68HC11. This is the default
9204 when the compiler is configured for 68HC11-based systems.
9210 Generate output for a 68HC12. This is the default
9211 when the compiler is configured for 68HC12-based systems.
9217 Generate output for a 68HCS12.
9220 @opindex mauto-incdec
9221 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
9228 Enable the use of 68HC12 min and max instructions.
9231 @itemx -mno-long-calls
9232 @opindex mlong-calls
9233 @opindex mno-long-calls
9234 Treat all calls as being far away (near). If calls are assumed to be
9235 far away, the compiler will use the @code{call} instruction to
9236 call a function and the @code{rtc} instruction for returning.
9240 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9242 @item -msoft-reg-count=@var{count}
9243 @opindex msoft-reg-count
9244 Specify the number of pseudo-soft registers which are used for the
9245 code generation. The maximum number is 32. Using more pseudo-soft
9246 register may or may not result in better code depending on the program.
9247 The default is 4 for 68HC11 and 2 for 68HC12.
9252 @subsection MCore Options
9253 @cindex MCore options
9255 These are the @samp{-m} options defined for the Motorola M*Core
9263 @opindex mno-hardlit
9264 Inline constants into the code stream if it can be done in two
9265 instructions or less.
9271 Use the divide instruction. (Enabled by default).
9273 @item -mrelax-immediate
9274 @itemx -mno-relax-immediate
9275 @opindex mrelax-immediate
9276 @opindex mno-relax-immediate
9277 Allow arbitrary sized immediates in bit operations.
9279 @item -mwide-bitfields
9280 @itemx -mno-wide-bitfields
9281 @opindex mwide-bitfields
9282 @opindex mno-wide-bitfields
9283 Always treat bit-fields as int-sized.
9285 @item -m4byte-functions
9286 @itemx -mno-4byte-functions
9287 @opindex m4byte-functions
9288 @opindex mno-4byte-functions
9289 Force all functions to be aligned to a four byte boundary.
9291 @item -mcallgraph-data
9292 @itemx -mno-callgraph-data
9293 @opindex mcallgraph-data
9294 @opindex mno-callgraph-data
9295 Emit callgraph information.
9298 @itemx -mno-slow-bytes
9299 @opindex mslow-bytes
9300 @opindex mno-slow-bytes
9301 Prefer word access when reading byte quantities.
9303 @item -mlittle-endian
9305 @opindex mlittle-endian
9306 @opindex mbig-endian
9307 Generate code for a little endian target.
9313 Generate code for the 210 processor.
9317 @subsection MIPS Options
9318 @cindex MIPS options
9324 Generate big-endian code.
9328 Generate little-endian code. This is the default for @samp{mips*el-*-*}
9331 @item -march=@var{arch}
9333 Generate code that will run on @var{arch}, which can be the name of a
9334 generic MIPS ISA, or the name of a particular processor.
9336 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
9337 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
9338 The processor names are:
9339 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
9341 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
9342 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
9346 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
9347 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
9348 The special value @samp{from-abi} selects the
9349 most compatible architecture for the selected ABI (that is,
9350 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
9352 In processor names, a final @samp{000} can be abbreviated as @samp{k}
9353 (for example, @samp{-march=r2k}). Prefixes are optional, and
9354 @samp{vr} may be written @samp{r}.
9356 GCC defines two macros based on the value of this option. The first
9357 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
9358 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
9359 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
9360 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
9361 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
9363 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
9364 above. In other words, it will have the full prefix and will not
9365 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
9366 the macro names the resolved architecture (either @samp{"mips1"} or
9367 @samp{"mips3"}). It names the default architecture when no
9368 @option{-march} option is given.
9370 @item -mtune=@var{arch}
9372 Optimize for @var{arch}. Among other things, this option controls
9373 the way instructions are scheduled, and the perceived cost of arithmetic
9374 operations. The list of @var{arch} values is the same as for
9377 When this option is not used, GCC will optimize for the processor
9378 specified by @option{-march}. By using @option{-march} and
9379 @option{-mtune} together, it is possible to generate code that will
9380 run on a family of processors, but optimize the code for one
9381 particular member of that family.
9383 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
9384 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
9385 @samp{-march} ones described above.
9389 Equivalent to @samp{-march=mips1}.
9393 Equivalent to @samp{-march=mips2}.
9397 Equivalent to @samp{-march=mips3}.
9401 Equivalent to @samp{-march=mips4}.
9405 Equivalent to @samp{-march=mips32}.
9409 Equivalent to @samp{-march=mips32r2}.
9413 Equivalent to @samp{-march=mips64}.
9419 Use (do not use) the MIPS16 ISA.
9431 Generate code for the given ABI@.
9433 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
9434 generates 64-bit code when you select a 64-bit architecture, but you
9435 can use @option{-mgp32} to get 32-bit code instead.
9437 For information about the O64 ABI, see
9438 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
9441 @itemx -mno-abicalls
9443 @opindex mno-abicalls
9444 Generate (do not generate) SVR4-style position-independent code.
9445 @option{-mabicalls} is the default for SVR4-based systems.
9451 Lift (do not lift) the usual restrictions on the size of the global
9454 GCC normally uses a single instruction to load values from the GOT.
9455 While this is relatively efficient, it will only work if the GOT
9456 is smaller than about 64k. Anything larger will cause the linker
9457 to report an error such as:
9459 @cindex relocation truncated to fit (MIPS)
9461 relocation truncated to fit: R_MIPS_GOT16 foobar
9464 If this happens, you should recompile your code with @option{-mxgot}.
9465 It should then work with very large GOTs, although it will also be
9466 less efficient, since it will take three instructions to fetch the
9467 value of a global symbol.
9469 Note that some linkers can create multiple GOTs. If you have such a
9470 linker, you should only need to use @option{-mxgot} when a single object
9471 file accesses more than 64k's worth of GOT entries. Very few do.
9473 These options have no effect unless GCC is generating position
9478 Assume that general-purpose registers are 32 bits wide.
9482 Assume that general-purpose registers are 64 bits wide.
9486 Assume that floating-point registers are 32 bits wide.
9490 Assume that floating-point registers are 64 bits wide.
9493 @opindex mhard-float
9494 Use floating-point coprocessor instructions.
9497 @opindex msoft-float
9498 Do not use floating-point coprocessor instructions. Implement
9499 floating-point calculations using library calls instead.
9501 @item -msingle-float
9502 @opindex msingle-float
9503 Assume that the floating-point coprocessor only supports single-precision
9506 @itemx -mdouble-float
9507 @opindex mdouble-float
9508 Assume that the floating-point coprocessor supports double-precision
9509 operations. This is the default.
9511 @itemx -mpaired-single
9512 @itemx -mno-paired-single
9513 @opindex mpaired-single
9514 @opindex mno-paired-single
9515 Use (do not use) paired-single floating-point instructions.
9516 @xref{MIPS Paired-Single Support}. This option can only be used
9517 when generating 64-bit code and requires hardware floating-point
9518 support to be enabled.
9524 Use (do not use) the MIPS-3D ASE. @xref{MIPS-3D Built-in Functions}.
9525 The option @option{-mips3d} implies @option{-mpaired-single}.
9529 Force @code{int} and @code{long} types to be 64 bits wide. See
9530 @option{-mlong32} for an explanation of the default and the way
9531 that the pointer size is determined.
9535 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
9536 an explanation of the default and the way that the pointer size is
9541 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
9543 The default size of @code{int}s, @code{long}s and pointers depends on
9544 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
9545 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
9546 32-bit @code{long}s. Pointers are the same size as @code{long}s,
9547 or the same size as integer registers, whichever is smaller.
9551 @cindex smaller data references (MIPS)
9552 @cindex gp-relative references (MIPS)
9553 Put global and static items less than or equal to @var{num} bytes into
9554 the small data or bss section instead of the normal data or bss section.
9555 This allows the data to be accessed using a single instruction.
9557 All modules should be compiled with the same @option{-G @var{num}}
9560 @item -membedded-data
9561 @itemx -mno-embedded-data
9562 @opindex membedded-data
9563 @opindex mno-embedded-data
9564 Allocate variables to the read-only data section first if possible, then
9565 next in the small data section if possible, otherwise in data. This gives
9566 slightly slower code than the default, but reduces the amount of RAM required
9567 when executing, and thus may be preferred for some embedded systems.
9569 @item -muninit-const-in-rodata
9570 @itemx -mno-uninit-const-in-rodata
9571 @opindex muninit-const-in-rodata
9572 @opindex mno-uninit-const-in-rodata
9573 Put uninitialized @code{const} variables in the read-only data section.
9574 This option is only meaningful in conjunction with @option{-membedded-data}.
9576 @item -msplit-addresses
9577 @itemx -mno-split-addresses
9578 @opindex msplit-addresses
9579 @opindex mno-split-addresses
9580 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
9581 relocation operators. This option has been superceded by
9582 @option{-mexplicit-relocs} but is retained for backwards compatibility.
9584 @item -mexplicit-relocs
9585 @itemx -mno-explicit-relocs
9586 @opindex mexplicit-relocs
9587 @opindex mno-explicit-relocs
9588 Use (do not use) assembler relocation operators when dealing with symbolic
9589 addresses. The alternative, selected by @option{-mno-explicit-relocs},
9590 is to use assembler macros instead.
9592 @option{-mexplicit-relocs} is the default if GCC was configured
9593 to use an assembler that supports relocation operators.
9595 @item -mcheck-zero-division
9596 @itemx -mno-check-zero-division
9597 @opindex mcheck-zero-division
9598 @opindex mno-check-zero-division
9599 Trap (do not trap) on integer division by zero. The default is
9600 @option{-mcheck-zero-division}.
9602 @item -mdivide-traps
9603 @itemx -mdivide-breaks
9604 @opindex mdivide-traps
9605 @opindex mdivide-breaks
9606 MIPS systems check for division by zero by generating either a
9607 conditional trap or a break instruction. Using traps results in
9608 smaller code, but is only supported on MIPS II and later. Also, some
9609 versions of the Linux kernel have a bug that prevents trap from
9610 generating the proper signal (SIGFPE). Use @option{-mdivide-traps} to
9611 allow conditional traps on architectures that support them and
9612 @option{-mdivide-breaks} to force the use of breaks.
9614 The default is usually @option{-mdivide-traps}, but this can be
9615 overridden at configure time using @option{--with-divide=breaks}.
9616 Divide-by-zero checks can be completely disabled using
9617 @option{-mno-check-zero-division}.
9623 Force (do not force) the use of @code{memcpy()} for non-trivial block
9624 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
9625 most constant-sized copies.
9628 @itemx -mno-long-calls
9629 @opindex mlong-calls
9630 @opindex mno-long-calls
9631 Disable (do not disable) use of the @code{jal} instruction. Calling
9632 functions using @code{jal} is more efficient but requires the caller
9633 and callee to be in the same 256 megabyte segment.
9635 This option has no effect on abicalls code. The default is
9636 @option{-mno-long-calls}.
9642 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
9643 instructions, as provided by the R4650 ISA.
9646 @itemx -mno-fused-madd
9647 @opindex mfused-madd
9648 @opindex mno-fused-madd
9649 Enable (disable) use of the floating point multiply-accumulate
9650 instructions, when they are available. The default is
9651 @option{-mfused-madd}.
9653 When multiply-accumulate instructions are used, the intermediate
9654 product is calculated to infinite precision and is not subject to
9655 the FCSR Flush to Zero bit. This may be undesirable in some
9660 Tell the MIPS assembler to not run its preprocessor over user
9661 assembler files (with a @samp{.s} suffix) when assembling them.
9664 @itemx -mno-fix-r4000
9666 @opindex mno-fix-r4000
9667 Work around certain R4000 CPU errata:
9670 A double-word or a variable shift may give an incorrect result if executed
9671 immediately after starting an integer division.
9673 A double-word or a variable shift may give an incorrect result if executed
9674 while an integer multiplication is in progress.
9676 An integer division may give an incorrect result if started in a delay slot
9677 of a taken branch or a jump.
9681 @itemx -mno-fix-r4400
9683 @opindex mno-fix-r4400
9684 Work around certain R4400 CPU errata:
9687 A double-word or a variable shift may give an incorrect result if executed
9688 immediately after starting an integer division.
9692 @itemx -mno-fix-vr4120
9693 @opindex mfix-vr4120
9694 Work around certain VR4120 errata:
9697 @code{dmultu} does not always produce the correct result.
9699 @code{div} and @code{ddiv} do not always produce the correct result if one
9700 of the operands is negative.
9702 The workarounds for the division errata rely on special functions in
9703 @file{libgcc.a}. At present, these functions are only provided by
9704 the @code{mips64vr*-elf} configurations.
9706 Other VR4120 errata require a nop to be inserted between certain pairs of
9707 instructions. These errata are handled by the assembler, not by GCC itself.
9712 Work around certain SB-1 CPU core errata.
9713 (This flag currently works around the SB-1 revision 2
9714 ``F1'' and ``F2'' floating point errata.)
9716 @item -mflush-func=@var{func}
9717 @itemx -mno-flush-func
9718 @opindex mflush-func
9719 Specifies the function to call to flush the I and D caches, or to not
9720 call any such function. If called, the function must take the same
9721 arguments as the common @code{_flush_func()}, that is, the address of the
9722 memory range for which the cache is being flushed, the size of the
9723 memory range, and the number 3 (to flush both caches). The default
9724 depends on the target GCC was configured for, but commonly is either
9725 @samp{_flush_func} or @samp{__cpu_flush}.
9727 @item -mbranch-likely
9728 @itemx -mno-branch-likely
9729 @opindex mbranch-likely
9730 @opindex mno-branch-likely
9731 Enable or disable use of Branch Likely instructions, regardless of the
9732 default for the selected architecture. By default, Branch Likely
9733 instructions may be generated if they are supported by the selected
9734 architecture. An exception is for the MIPS32 and MIPS64 architectures
9735 and processors which implement those architectures; for those, Branch
9736 Likely instructions will not be generated by default because the MIPS32
9737 and MIPS64 architectures specifically deprecate their use.
9739 @item -mfp-exceptions
9740 @itemx -mno-fp-exceptions
9741 @opindex mfp-exceptions
9742 Specifies whether FP exceptions are enabled. This affects how we schedule
9743 FP instructions for some processors. The default is that FP exceptions are
9746 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
9747 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
9750 @item -mvr4130-align
9751 @itemx -mno-vr4130-align
9752 @opindex mvr4130-align
9753 The VR4130 pipeline is two-way superscalar, but can only issue two
9754 instructions together if the first one is 8-byte aligned. When this
9755 option is enabled, GCC will align pairs of instructions that it
9756 thinks should execute in parallel.
9758 This option only has an effect when optimizing for the VR4130.
9759 It normally makes code faster, but at the expense of making it bigger.
9760 It is enabled by default at optimization level @option{-O3}.
9764 @subsection MMIX Options
9765 @cindex MMIX Options
9767 These options are defined for the MMIX:
9771 @itemx -mno-libfuncs
9773 @opindex mno-libfuncs
9774 Specify that intrinsic library functions are being compiled, passing all
9775 values in registers, no matter the size.
9780 @opindex mno-epsilon
9781 Generate floating-point comparison instructions that compare with respect
9782 to the @code{rE} epsilon register.
9784 @item -mabi=mmixware
9786 @opindex mabi-mmixware
9788 Generate code that passes function parameters and return values that (in
9789 the called function) are seen as registers @code{$0} and up, as opposed to
9790 the GNU ABI which uses global registers @code{$231} and up.
9793 @itemx -mno-zero-extend
9794 @opindex mzero-extend
9795 @opindex mno-zero-extend
9796 When reading data from memory in sizes shorter than 64 bits, use (do not
9797 use) zero-extending load instructions by default, rather than
9798 sign-extending ones.
9801 @itemx -mno-knuthdiv
9803 @opindex mno-knuthdiv
9804 Make the result of a division yielding a remainder have the same sign as
9805 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9806 remainder follows the sign of the dividend. Both methods are
9807 arithmetically valid, the latter being almost exclusively used.
9809 @item -mtoplevel-symbols
9810 @itemx -mno-toplevel-symbols
9811 @opindex mtoplevel-symbols
9812 @opindex mno-toplevel-symbols
9813 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9814 code can be used with the @code{PREFIX} assembly directive.
9818 Generate an executable in the ELF format, rather than the default
9819 @samp{mmo} format used by the @command{mmix} simulator.
9821 @item -mbranch-predict
9822 @itemx -mno-branch-predict
9823 @opindex mbranch-predict
9824 @opindex mno-branch-predict
9825 Use (do not use) the probable-branch instructions, when static branch
9826 prediction indicates a probable branch.
9828 @item -mbase-addresses
9829 @itemx -mno-base-addresses
9830 @opindex mbase-addresses
9831 @opindex mno-base-addresses
9832 Generate (do not generate) code that uses @emph{base addresses}. Using a
9833 base address automatically generates a request (handled by the assembler
9834 and the linker) for a constant to be set up in a global register. The
9835 register is used for one or more base address requests within the range 0
9836 to 255 from the value held in the register. The generally leads to short
9837 and fast code, but the number of different data items that can be
9838 addressed is limited. This means that a program that uses lots of static
9839 data may require @option{-mno-base-addresses}.
9842 @itemx -mno-single-exit
9843 @opindex msingle-exit
9844 @opindex mno-single-exit
9845 Force (do not force) generated code to have a single exit point in each
9849 @node MN10300 Options
9850 @subsection MN10300 Options
9851 @cindex MN10300 options
9853 These @option{-m} options are defined for Matsushita MN10300 architectures:
9858 Generate code to avoid bugs in the multiply instructions for the MN10300
9859 processors. This is the default.
9862 @opindex mno-mult-bug
9863 Do not generate code to avoid bugs in the multiply instructions for the
9868 Generate code which uses features specific to the AM33 processor.
9872 Do not generate code which uses features specific to the AM33 processor. This
9877 Do not link in the C run-time initialization object file.
9881 Indicate to the linker that it should perform a relaxation optimization pass
9882 to shorten branches, calls and absolute memory addresses. This option only
9883 has an effect when used on the command line for the final link step.
9885 This option makes symbolic debugging impossible.
9889 @subsection NS32K Options
9890 @cindex NS32K options
9892 These are the @samp{-m} options defined for the 32000 series. The default
9893 values for these options depends on which style of 32000 was selected when
9894 the compiler was configured; the defaults for the most common choices are
9902 Generate output for a 32032. This is the default
9903 when the compiler is configured for 32032 and 32016 based systems.
9909 Generate output for a 32332. This is the default
9910 when the compiler is configured for 32332-based systems.
9916 Generate output for a 32532. This is the default
9917 when the compiler is configured for 32532-based systems.
9921 Generate output containing 32081 instructions for floating point.
9922 This is the default for all systems.
9926 Generate output containing 32381 instructions for floating point. This
9927 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9928 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9932 Try and generate multiply-add floating point instructions @code{polyF}
9933 and @code{dotF}. This option is only available if the @option{-m32381}
9934 option is in effect. Using these instructions requires changes to
9935 register allocation which generally has a negative impact on
9936 performance. This option should only be enabled when compiling code
9937 particularly likely to make heavy use of multiply-add instructions.
9940 @opindex mnomulti-add
9941 Do not try and generate multiply-add floating point instructions
9942 @code{polyF} and @code{dotF}. This is the default on all platforms.
9945 @opindex msoft-float
9946 Generate output containing library calls for floating point.
9947 @strong{Warning:} the requisite libraries may not be available.
9949 @item -mieee-compare
9950 @itemx -mno-ieee-compare
9951 @opindex mieee-compare
9952 @opindex mno-ieee-compare
9953 Control whether or not the compiler uses IEEE floating point
9954 comparisons. These handle correctly the case where the result of a
9955 comparison is unordered.
9956 @strong{Warning:} the requisite kernel support may not be available.
9959 @opindex mnobitfield
9960 Do not use the bit-field instructions. On some machines it is faster to
9961 use shifting and masking operations. This is the default for the pc532.
9965 Do use the bit-field instructions. This is the default for all platforms
9970 Use a different function-calling convention, in which functions
9971 that take a fixed number of arguments return pop their
9972 arguments on return with the @code{ret} instruction.
9974 This calling convention is incompatible with the one normally
9975 used on Unix, so you cannot use it if you need to call libraries
9976 compiled with the Unix compiler.
9978 Also, you must provide function prototypes for all functions that
9979 take variable numbers of arguments (including @code{printf});
9980 otherwise incorrect code will be generated for calls to those
9983 In addition, seriously incorrect code will result if you call a
9984 function with too many arguments. (Normally, extra arguments are
9985 harmlessly ignored.)
9987 This option takes its name from the 680x0 @code{rtd} instruction.
9992 Use a different function-calling convention where the first two arguments
9993 are passed in registers.
9995 This calling convention is incompatible with the one normally
9996 used on Unix, so you cannot use it if you need to call libraries
9997 compiled with the Unix compiler.
10000 @opindex mnoregparam
10001 Do not pass any arguments in registers. This is the default for all
10006 It is OK to use the sb as an index register which is always loaded with
10007 zero. This is the default for the pc532-netbsd target.
10011 The sb register is not available for use or has not been initialized to
10012 zero by the run time system. This is the default for all targets except
10013 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
10014 @option{-fpic} is set.
10018 Many ns32000 series addressing modes use displacements of up to 512MB@.
10019 If an address is above 512MB then displacements from zero can not be used.
10020 This option causes code to be generated which can be loaded above 512MB@.
10021 This may be useful for operating systems or ROM code.
10025 Assume code will be loaded in the first 512MB of virtual address space.
10026 This is the default for all platforms.
10030 @node PDP-11 Options
10031 @subsection PDP-11 Options
10032 @cindex PDP-11 Options
10034 These options are defined for the PDP-11:
10039 Use hardware FPP floating point. This is the default. (FIS floating
10040 point on the PDP-11/40 is not supported.)
10043 @opindex msoft-float
10044 Do not use hardware floating point.
10048 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10052 Return floating-point results in memory. This is the default.
10056 Generate code for a PDP-11/40.
10060 Generate code for a PDP-11/45. This is the default.
10064 Generate code for a PDP-11/10.
10066 @item -mbcopy-builtin
10067 @opindex bcopy-builtin
10068 Use inline @code{movmemhi} patterns for copying memory. This is the
10073 Do not use inline @code{movmemhi} patterns for copying memory.
10079 Use 16-bit @code{int}. This is the default.
10085 Use 32-bit @code{int}.
10088 @itemx -mno-float32
10090 @opindex mno-float32
10091 Use 64-bit @code{float}. This is the default.
10094 @itemx -mno-float64
10096 @opindex mno-float64
10097 Use 32-bit @code{float}.
10101 Use @code{abshi2} pattern. This is the default.
10105 Do not use @code{abshi2} pattern.
10107 @item -mbranch-expensive
10108 @opindex mbranch-expensive
10109 Pretend that branches are expensive. This is for experimenting with
10110 code generation only.
10112 @item -mbranch-cheap
10113 @opindex mbranch-cheap
10114 Do not pretend that branches are expensive. This is the default.
10118 Generate code for a system with split I&D.
10122 Generate code for a system without split I&D. This is the default.
10126 Use Unix assembler syntax. This is the default when configured for
10127 @samp{pdp11-*-bsd}.
10131 Use DEC assembler syntax. This is the default when configured for any
10132 PDP-11 target other than @samp{pdp11-*-bsd}.
10135 @node PowerPC Options
10136 @subsection PowerPC Options
10137 @cindex PowerPC options
10139 These are listed under @xref{RS/6000 and PowerPC Options}.
10141 @node RS/6000 and PowerPC Options
10142 @subsection IBM RS/6000 and PowerPC Options
10143 @cindex RS/6000 and PowerPC Options
10144 @cindex IBM RS/6000 and PowerPC Options
10146 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
10153 @itemx -mno-powerpc
10154 @itemx -mpowerpc-gpopt
10155 @itemx -mno-powerpc-gpopt
10156 @itemx -mpowerpc-gfxopt
10157 @itemx -mno-powerpc-gfxopt
10159 @itemx -mno-powerpc64
10163 @opindex mno-power2
10165 @opindex mno-powerpc
10166 @opindex mpowerpc-gpopt
10167 @opindex mno-powerpc-gpopt
10168 @opindex mpowerpc-gfxopt
10169 @opindex mno-powerpc-gfxopt
10170 @opindex mpowerpc64
10171 @opindex mno-powerpc64
10172 GCC supports two related instruction set architectures for the
10173 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
10174 instructions supported by the @samp{rios} chip set used in the original
10175 RS/6000 systems and the @dfn{PowerPC} instruction set is the
10176 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
10177 the IBM 4xx microprocessors.
10179 Neither architecture is a subset of the other. However there is a
10180 large common subset of instructions supported by both. An MQ
10181 register is included in processors supporting the POWER architecture.
10183 You use these options to specify which instructions are available on the
10184 processor you are using. The default value of these options is
10185 determined when configuring GCC@. Specifying the
10186 @option{-mcpu=@var{cpu_type}} overrides the specification of these
10187 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
10188 rather than the options listed above.
10190 The @option{-mpower} option allows GCC to generate instructions that
10191 are found only in the POWER architecture and to use the MQ register.
10192 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
10193 to generate instructions that are present in the POWER2 architecture but
10194 not the original POWER architecture.
10196 The @option{-mpowerpc} option allows GCC to generate instructions that
10197 are found only in the 32-bit subset of the PowerPC architecture.
10198 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
10199 GCC to use the optional PowerPC architecture instructions in the
10200 General Purpose group, including floating-point square root. Specifying
10201 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
10202 use the optional PowerPC architecture instructions in the Graphics
10203 group, including floating-point select.
10205 The @option{-mpowerpc64} option allows GCC to generate the additional
10206 64-bit instructions that are found in the full PowerPC64 architecture
10207 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
10208 @option{-mno-powerpc64}.
10210 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
10211 will use only the instructions in the common subset of both
10212 architectures plus some special AIX common-mode calls, and will not use
10213 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
10214 permits GCC to use any instruction from either architecture and to
10215 allow use of the MQ register; specify this for the Motorola MPC601.
10217 @item -mnew-mnemonics
10218 @itemx -mold-mnemonics
10219 @opindex mnew-mnemonics
10220 @opindex mold-mnemonics
10221 Select which mnemonics to use in the generated assembler code. With
10222 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
10223 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
10224 assembler mnemonics defined for the POWER architecture. Instructions
10225 defined in only one architecture have only one mnemonic; GCC uses that
10226 mnemonic irrespective of which of these options is specified.
10228 GCC defaults to the mnemonics appropriate for the architecture in
10229 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
10230 value of these option. Unless you are building a cross-compiler, you
10231 should normally not specify either @option{-mnew-mnemonics} or
10232 @option{-mold-mnemonics}, but should instead accept the default.
10234 @item -mcpu=@var{cpu_type}
10236 Set architecture type, register usage, choice of mnemonics, and
10237 instruction scheduling parameters for machine type @var{cpu_type}.
10238 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
10239 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
10240 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
10241 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
10242 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
10243 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
10244 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
10245 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
10246 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
10248 @option{-mcpu=common} selects a completely generic processor. Code
10249 generated under this option will run on any POWER or PowerPC processor.
10250 GCC will use only the instructions in the common subset of both
10251 architectures, and will not use the MQ register. GCC assumes a generic
10252 processor model for scheduling purposes.
10254 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
10255 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
10256 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
10257 types, with an appropriate, generic processor model assumed for
10258 scheduling purposes.
10260 The other options specify a specific processor. Code generated under
10261 those options will run best on that processor, and may not run at all on
10264 The @option{-mcpu} options automatically enable or disable the
10265 following options: @option{-maltivec}, @option{-mhard-float},
10266 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
10267 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
10268 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
10269 @option{-mstring}. The particular options set for any particular CPU
10270 will vary between compiler versions, depending on what setting seems
10271 to produce optimal code for that CPU; it doesn't necessarily reflect
10272 the actual hardware's capabilities. If you wish to set an individual
10273 option to a particular value, you may specify it after the
10274 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
10276 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
10277 not enabled or disabled by the @option{-mcpu} option at present, since
10278 AIX does not have full support for these options. You may still
10279 enable or disable them individually if you're sure it'll work in your
10282 @item -mtune=@var{cpu_type}
10284 Set the instruction scheduling parameters for machine type
10285 @var{cpu_type}, but do not set the architecture type, register usage, or
10286 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
10287 values for @var{cpu_type} are used for @option{-mtune} as for
10288 @option{-mcpu}. If both are specified, the code generated will use the
10289 architecture, registers, and mnemonics set by @option{-mcpu}, but the
10290 scheduling parameters set by @option{-mtune}.
10293 @itemx -mno-altivec
10295 @opindex mno-altivec
10296 Generate code that uses (does not use) AltiVec instructions, and also
10297 enable the use of built-in functions that allow more direct access to
10298 the AltiVec instruction set. You may also need to set
10299 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
10304 Extend the current ABI with SPE ABI extensions. This does not change
10305 the default ABI, instead it adds the SPE ABI extensions to the current
10309 @opindex mabi=no-spe
10310 Disable Booke SPE ABI extensions for the current ABI.
10312 @item -misel=@var{yes/no}
10315 This switch enables or disables the generation of ISEL instructions.
10317 @item -mspe=@var{yes/no}
10320 This switch enables or disables the generation of SPE simd
10323 @item -mfloat-gprs=@var{yes/single/double/no}
10324 @itemx -mfloat-gprs
10325 @opindex mfloat-gprs
10326 This switch enables or disables the generation of floating point
10327 operations on the general purpose registers for architectures that
10330 The argument @var{yes} or @var{single} enables the use of
10331 single-precision floating point operations.
10333 The argument @var{double} enables the use of single and
10334 double-precision floating point operations.
10336 The argument @var{no} disables floating point operations on the
10337 general purpose registers.
10339 This option is currently only available on the MPC854x.
10345 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
10346 targets (including GNU/Linux). The 32-bit environment sets int, long
10347 and pointer to 32 bits and generates code that runs on any PowerPC
10348 variant. The 64-bit environment sets int to 32 bits and long and
10349 pointer to 64 bits, and generates code for PowerPC64, as for
10350 @option{-mpowerpc64}.
10353 @itemx -mno-fp-in-toc
10354 @itemx -mno-sum-in-toc
10355 @itemx -mminimal-toc
10357 @opindex mno-fp-in-toc
10358 @opindex mno-sum-in-toc
10359 @opindex mminimal-toc
10360 Modify generation of the TOC (Table Of Contents), which is created for
10361 every executable file. The @option{-mfull-toc} option is selected by
10362 default. In that case, GCC will allocate at least one TOC entry for
10363 each unique non-automatic variable reference in your program. GCC
10364 will also place floating-point constants in the TOC@. However, only
10365 16,384 entries are available in the TOC@.
10367 If you receive a linker error message that saying you have overflowed
10368 the available TOC space, you can reduce the amount of TOC space used
10369 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
10370 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
10371 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
10372 generate code to calculate the sum of an address and a constant at
10373 run-time instead of putting that sum into the TOC@. You may specify one
10374 or both of these options. Each causes GCC to produce very slightly
10375 slower and larger code at the expense of conserving TOC space.
10377 If you still run out of space in the TOC even when you specify both of
10378 these options, specify @option{-mminimal-toc} instead. This option causes
10379 GCC to make only one TOC entry for every file. When you specify this
10380 option, GCC will produce code that is slower and larger but which
10381 uses extremely little TOC space. You may wish to use this option
10382 only on files that contain less frequently executed code.
10388 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
10389 @code{long} type, and the infrastructure needed to support them.
10390 Specifying @option{-maix64} implies @option{-mpowerpc64} and
10391 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
10392 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
10395 @itemx -mno-xl-call
10397 @opindex mno-xl-call
10398 On AIX, pass floating-point arguments to prototyped functions beyond the
10399 register save area (RSA) on the stack in addition to argument FPRs. The
10400 AIX calling convention was extended but not initially documented to
10401 handle an obscure K&R C case of calling a function that takes the
10402 address of its arguments with fewer arguments than declared. AIX XL
10403 compilers access floating point arguments which do not fit in the
10404 RSA from the stack when a subroutine is compiled without
10405 optimization. Because always storing floating-point arguments on the
10406 stack is inefficient and rarely needed, this option is not enabled by
10407 default and only is necessary when calling subroutines compiled by AIX
10408 XL compilers without optimization.
10412 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
10413 application written to use message passing with special startup code to
10414 enable the application to run. The system must have PE installed in the
10415 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
10416 must be overridden with the @option{-specs=} option to specify the
10417 appropriate directory location. The Parallel Environment does not
10418 support threads, so the @option{-mpe} option and the @option{-pthread}
10419 option are incompatible.
10421 @item -malign-natural
10422 @itemx -malign-power
10423 @opindex malign-natural
10424 @opindex malign-power
10425 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
10426 @option{-malign-natural} overrides the ABI-defined alignment of larger
10427 types, such as floating-point doubles, on their natural size-based boundary.
10428 The option @option{-malign-power} instructs GCC to follow the ABI-specified
10429 alignment rules. GCC defaults to the standard alignment defined in the ABI.
10432 @itemx -mhard-float
10433 @opindex msoft-float
10434 @opindex mhard-float
10435 Generate code that does not use (uses) the floating-point register set.
10436 Software floating point emulation is provided if you use the
10437 @option{-msoft-float} option, and pass the option to GCC when linking.
10440 @itemx -mno-multiple
10442 @opindex mno-multiple
10443 Generate code that uses (does not use) the load multiple word
10444 instructions and the store multiple word instructions. These
10445 instructions are generated by default on POWER systems, and not
10446 generated on PowerPC systems. Do not use @option{-mmultiple} on little
10447 endian PowerPC systems, since those instructions do not work when the
10448 processor is in little endian mode. The exceptions are PPC740 and
10449 PPC750 which permit the instructions usage in little endian mode.
10454 @opindex mno-string
10455 Generate code that uses (does not use) the load string instructions
10456 and the store string word instructions to save multiple registers and
10457 do small block moves. These instructions are generated by default on
10458 POWER systems, and not generated on PowerPC systems. Do not use
10459 @option{-mstring} on little endian PowerPC systems, since those
10460 instructions do not work when the processor is in little endian mode.
10461 The exceptions are PPC740 and PPC750 which permit the instructions
10462 usage in little endian mode.
10467 @opindex mno-update
10468 Generate code that uses (does not use) the load or store instructions
10469 that update the base register to the address of the calculated memory
10470 location. These instructions are generated by default. If you use
10471 @option{-mno-update}, there is a small window between the time that the
10472 stack pointer is updated and the address of the previous frame is
10473 stored, which means code that walks the stack frame across interrupts or
10474 signals may get corrupted data.
10477 @itemx -mno-fused-madd
10478 @opindex mfused-madd
10479 @opindex mno-fused-madd
10480 Generate code that uses (does not use) the floating point multiply and
10481 accumulate instructions. These instructions are generated by default if
10482 hardware floating is used.
10484 @item -mno-bit-align
10486 @opindex mno-bit-align
10487 @opindex mbit-align
10488 On System V.4 and embedded PowerPC systems do not (do) force structures
10489 and unions that contain bit-fields to be aligned to the base type of the
10492 For example, by default a structure containing nothing but 8
10493 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
10494 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
10495 the structure would be aligned to a 1 byte boundary and be one byte in
10498 @item -mno-strict-align
10499 @itemx -mstrict-align
10500 @opindex mno-strict-align
10501 @opindex mstrict-align
10502 On System V.4 and embedded PowerPC systems do not (do) assume that
10503 unaligned memory references will be handled by the system.
10505 @item -mrelocatable
10506 @itemx -mno-relocatable
10507 @opindex mrelocatable
10508 @opindex mno-relocatable
10509 On embedded PowerPC systems generate code that allows (does not allow)
10510 the program to be relocated to a different address at runtime. If you
10511 use @option{-mrelocatable} on any module, all objects linked together must
10512 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
10514 @item -mrelocatable-lib
10515 @itemx -mno-relocatable-lib
10516 @opindex mrelocatable-lib
10517 @opindex mno-relocatable-lib
10518 On embedded PowerPC systems generate code that allows (does not allow)
10519 the program to be relocated to a different address at runtime. Modules
10520 compiled with @option{-mrelocatable-lib} can be linked with either modules
10521 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
10522 with modules compiled with the @option{-mrelocatable} options.
10528 On System V.4 and embedded PowerPC systems do not (do) assume that
10529 register 2 contains a pointer to a global area pointing to the addresses
10530 used in the program.
10533 @itemx -mlittle-endian
10535 @opindex mlittle-endian
10536 On System V.4 and embedded PowerPC systems compile code for the
10537 processor in little endian mode. The @option{-mlittle-endian} option is
10538 the same as @option{-mlittle}.
10541 @itemx -mbig-endian
10543 @opindex mbig-endian
10544 On System V.4 and embedded PowerPC systems compile code for the
10545 processor in big endian mode. The @option{-mbig-endian} option is
10546 the same as @option{-mbig}.
10548 @item -mdynamic-no-pic
10549 @opindex mdynamic-no-pic
10550 On Darwin and Mac OS X systems, compile code so that it is not
10551 relocatable, but that its external references are relocatable. The
10552 resulting code is suitable for applications, but not shared
10555 @item -mprioritize-restricted-insns=@var{priority}
10556 @opindex mprioritize-restricted-insns
10557 This option controls the priority that is assigned to
10558 dispatch-slot restricted instructions during the second scheduling
10559 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
10560 @var{no/highest/second-highest} priority to dispatch slot restricted
10563 @item -msched-costly-dep=@var{dependence_type}
10564 @opindex msched-costly-dep
10565 This option controls which dependences are considered costly
10566 by the target during instruction scheduling. The argument
10567 @var{dependence_type} takes one of the following values:
10568 @var{no}: no dependence is costly,
10569 @var{all}: all dependences are costly,
10570 @var{true_store_to_load}: a true dependence from store to load is costly,
10571 @var{store_to_load}: any dependence from store to load is costly,
10572 @var{number}: any dependence which latency >= @var{number} is costly.
10574 @item -minsert-sched-nops=@var{scheme}
10575 @opindex minsert-sched-nops
10576 This option controls which nop insertion scheme will be used during
10577 the second scheduling pass. The argument @var{scheme} takes one of the
10579 @var{no}: Don't insert nops.
10580 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
10581 according to the scheduler's grouping.
10582 @var{regroup_exact}: Insert nops to force costly dependent insns into
10583 separate groups. Insert exactly as many nops as needed to force an insn
10584 to a new group, according to the estimated processor grouping.
10585 @var{number}: Insert nops to force costly dependent insns into
10586 separate groups. Insert @var{number} nops to force an insn to a new group.
10589 @opindex mcall-sysv
10590 On System V.4 and embedded PowerPC systems compile code using calling
10591 conventions that adheres to the March 1995 draft of the System V
10592 Application Binary Interface, PowerPC processor supplement. This is the
10593 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
10595 @item -mcall-sysv-eabi
10596 @opindex mcall-sysv-eabi
10597 Specify both @option{-mcall-sysv} and @option{-meabi} options.
10599 @item -mcall-sysv-noeabi
10600 @opindex mcall-sysv-noeabi
10601 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
10603 @item -mcall-solaris
10604 @opindex mcall-solaris
10605 On System V.4 and embedded PowerPC systems compile code for the Solaris
10609 @opindex mcall-linux
10610 On System V.4 and embedded PowerPC systems compile code for the
10611 Linux-based GNU system.
10615 On System V.4 and embedded PowerPC systems compile code for the
10616 Hurd-based GNU system.
10618 @item -mcall-netbsd
10619 @opindex mcall-netbsd
10620 On System V.4 and embedded PowerPC systems compile code for the
10621 NetBSD operating system.
10623 @item -maix-struct-return
10624 @opindex maix-struct-return
10625 Return all structures in memory (as specified by the AIX ABI)@.
10627 @item -msvr4-struct-return
10628 @opindex msvr4-struct-return
10629 Return structures smaller than 8 bytes in registers (as specified by the
10632 @item -mabi=altivec
10633 @opindex mabi=altivec
10634 Extend the current ABI with AltiVec ABI extensions. This does not
10635 change the default ABI, instead it adds the AltiVec ABI extensions to
10638 @item -mabi=no-altivec
10639 @opindex mabi=no-altivec
10640 Disable AltiVec ABI extensions for the current ABI.
10643 @itemx -mno-prototype
10644 @opindex mprototype
10645 @opindex mno-prototype
10646 On System V.4 and embedded PowerPC systems assume that all calls to
10647 variable argument functions are properly prototyped. Otherwise, the
10648 compiler must insert an instruction before every non prototyped call to
10649 set or clear bit 6 of the condition code register (@var{CR}) to
10650 indicate whether floating point values were passed in the floating point
10651 registers in case the function takes a variable arguments. With
10652 @option{-mprototype}, only calls to prototyped variable argument functions
10653 will set or clear the bit.
10657 On embedded PowerPC systems, assume that the startup module is called
10658 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
10659 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
10664 On embedded PowerPC systems, assume that the startup module is called
10665 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
10670 On embedded PowerPC systems, assume that the startup module is called
10671 @file{crt0.o} and the standard C libraries are @file{libads.a} and
10674 @item -myellowknife
10675 @opindex myellowknife
10676 On embedded PowerPC systems, assume that the startup module is called
10677 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
10682 On System V.4 and embedded PowerPC systems, specify that you are
10683 compiling for a VxWorks system.
10687 Specify that you are compiling for the WindISS simulation environment.
10691 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
10692 header to indicate that @samp{eabi} extended relocations are used.
10698 On System V.4 and embedded PowerPC systems do (do not) adhere to the
10699 Embedded Applications Binary Interface (eabi) which is a set of
10700 modifications to the System V.4 specifications. Selecting @option{-meabi}
10701 means that the stack is aligned to an 8 byte boundary, a function
10702 @code{__eabi} is called to from @code{main} to set up the eabi
10703 environment, and the @option{-msdata} option can use both @code{r2} and
10704 @code{r13} to point to two separate small data areas. Selecting
10705 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
10706 do not call an initialization function from @code{main}, and the
10707 @option{-msdata} option will only use @code{r13} to point to a single
10708 small data area. The @option{-meabi} option is on by default if you
10709 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
10712 @opindex msdata=eabi
10713 On System V.4 and embedded PowerPC systems, put small initialized
10714 @code{const} global and static data in the @samp{.sdata2} section, which
10715 is pointed to by register @code{r2}. Put small initialized
10716 non-@code{const} global and static data in the @samp{.sdata} section,
10717 which is pointed to by register @code{r13}. Put small uninitialized
10718 global and static data in the @samp{.sbss} section, which is adjacent to
10719 the @samp{.sdata} section. The @option{-msdata=eabi} option is
10720 incompatible with the @option{-mrelocatable} option. The
10721 @option{-msdata=eabi} option also sets the @option{-memb} option.
10724 @opindex msdata=sysv
10725 On System V.4 and embedded PowerPC systems, put small global and static
10726 data in the @samp{.sdata} section, which is pointed to by register
10727 @code{r13}. Put small uninitialized global and static data in the
10728 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
10729 The @option{-msdata=sysv} option is incompatible with the
10730 @option{-mrelocatable} option.
10732 @item -msdata=default
10734 @opindex msdata=default
10736 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
10737 compile code the same as @option{-msdata=eabi}, otherwise compile code the
10738 same as @option{-msdata=sysv}.
10741 @opindex msdata-data
10742 On System V.4 and embedded PowerPC systems, put small global and static
10743 data in the @samp{.sdata} section. Put small uninitialized global and
10744 static data in the @samp{.sbss} section. Do not use register @code{r13}
10745 to address small data however. This is the default behavior unless
10746 other @option{-msdata} options are used.
10750 @opindex msdata=none
10752 On embedded PowerPC systems, put all initialized global and static data
10753 in the @samp{.data} section, and all uninitialized data in the
10754 @samp{.bss} section.
10758 @cindex smaller data references (PowerPC)
10759 @cindex .sdata/.sdata2 references (PowerPC)
10760 On embedded PowerPC systems, put global and static items less than or
10761 equal to @var{num} bytes into the small data or bss sections instead of
10762 the normal data or bss section. By default, @var{num} is 8. The
10763 @option{-G @var{num}} switch is also passed to the linker.
10764 All modules should be compiled with the same @option{-G @var{num}} value.
10767 @itemx -mno-regnames
10769 @opindex mno-regnames
10770 On System V.4 and embedded PowerPC systems do (do not) emit register
10771 names in the assembly language output using symbolic forms.
10774 @itemx -mno-longcall
10776 @opindex mno-longcall
10777 Default to making all function calls indirectly, using a register, so
10778 that functions which reside further than 32 megabytes (33,554,432
10779 bytes) from the current location can be called. This setting can be
10780 overridden by the @code{shortcall} function attribute, or by
10781 @code{#pragma longcall(0)}.
10783 Some linkers are capable of detecting out-of-range calls and generating
10784 glue code on the fly. On these systems, long calls are unnecessary and
10785 generate slower code. As of this writing, the AIX linker can do this,
10786 as can the GNU linker for PowerPC/64. It is planned to add this feature
10787 to the GNU linker for 32-bit PowerPC systems as well.
10789 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
10790 callee, L42'', plus a ``branch island'' (glue code). The two target
10791 addresses represent the callee and the ``branch island''. The
10792 Darwin/PPC linker will prefer the first address and generate a ``bl
10793 callee'' if the PPC ``bl'' instruction will reach the callee directly;
10794 otherwise, the linker will generate ``bl L42'' to call the ``branch
10795 island''. The ``branch island'' is appended to the body of the
10796 calling function; it computes the full 32-bit address of the callee
10799 On Mach-O (Darwin) systems, this option directs the compiler emit to
10800 the glue for every direct call, and the Darwin linker decides whether
10801 to use or discard it.
10803 In the future, we may cause GCC to ignore all longcall specifications
10804 when the linker is known to generate glue.
10808 Adds support for multithreading with the @dfn{pthreads} library.
10809 This option sets flags for both the preprocessor and linker.
10813 @node S/390 and zSeries Options
10814 @subsection S/390 and zSeries Options
10815 @cindex S/390 and zSeries Options
10817 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10821 @itemx -msoft-float
10822 @opindex mhard-float
10823 @opindex msoft-float
10824 Use (do not use) the hardware floating-point instructions and registers
10825 for floating-point operations. When @option{-msoft-float} is specified,
10826 functions in @file{libgcc.a} will be used to perform floating-point
10827 operations. When @option{-mhard-float} is specified, the compiler
10828 generates IEEE floating-point instructions. This is the default.
10831 @itemx -mno-backchain
10832 @itemx -mkernel-backchain
10833 @opindex mbackchain
10834 @opindex mno-backchain
10835 @opindex mkernel-backchain
10836 In order to provide a backchain the address of the caller's frame
10837 is stored within the callee's stack frame.
10838 A backchain may be needed to allow debugging using tools that do not understand
10839 DWARF-2 call frame information.
10840 For @option{-mno-backchain} no backchain is maintained at all which is the
10842 If one of the other options is present the backchain pointer is placed either
10843 on top of the stack frame (@option{-mkernel-backchain}) or on
10844 the bottom (@option{-mbackchain}).
10845 Beside the different backchain location @option{-mkernel-backchain}
10846 also changes stack frame layout breaking the ABI. This option
10847 is intended to be used for code which internally needs a backchain but has
10848 to get by with a limited stack size e.g.@: the linux kernel.
10849 Internal unwinding code not using DWARF-2 info has to be able to locate the
10850 return address of a function. That will be eased be the fact that
10851 the return address of a function is placed two words below the backchain
10855 @itemx -mno-small-exec
10856 @opindex msmall-exec
10857 @opindex mno-small-exec
10858 Generate (or do not generate) code using the @code{bras} instruction
10859 to do subroutine calls.
10860 This only works reliably if the total executable size does not
10861 exceed 64k. The default is to use the @code{basr} instruction instead,
10862 which does not have this limitation.
10868 When @option{-m31} is specified, generate code compliant to the
10869 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10870 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10871 particular to generate 64-bit instructions. For the @samp{s390}
10872 targets, the default is @option{-m31}, while the @samp{s390x}
10873 targets default to @option{-m64}.
10879 When @option{-mzarch} is specified, generate code using the
10880 instructions available on z/Architecture.
10881 When @option{-mesa} is specified, generate code using the
10882 instructions available on ESA/390. Note that @option{-mesa} is
10883 not possible with @option{-m64}.
10884 When generating code compliant to the GNU/Linux for S/390 ABI,
10885 the default is @option{-mesa}. When generating code compliant
10886 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10892 Generate (or do not generate) code using the @code{mvcle} instruction
10893 to perform block moves. When @option{-mno-mvcle} is specified,
10894 use a @code{mvc} loop instead. This is the default.
10900 Print (or do not print) additional debug information when compiling.
10901 The default is to not print debug information.
10903 @item -march=@var{cpu-type}
10905 Generate code that will run on @var{cpu-type}, which is the name of a system
10906 representing a certain processor type. Possible values for
10907 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10908 When generating code using the instructions available on z/Architecture,
10909 the default is @option{-march=z900}. Otherwise, the default is
10910 @option{-march=g5}.
10912 @item -mtune=@var{cpu-type}
10914 Tune to @var{cpu-type} everything applicable about the generated code,
10915 except for the ABI and the set of available instructions.
10916 The list of @var{cpu-type} values is the same as for @option{-march}.
10917 The default is the value used for @option{-march}.
10920 @itemx -mno-tpf-trace
10921 @opindex mtpf-trace
10922 @opindex mno-tpf-trace
10923 Generate code that adds (does not add) in TPF OS specific branches to trace
10924 routines in the operating system. This option is off by default, even
10925 when compiling for the TPF OS.
10928 @itemx -mno-fused-madd
10929 @opindex mfused-madd
10930 @opindex mno-fused-madd
10931 Generate code that uses (does not use) the floating point multiply and
10932 accumulate instructions. These instructions are generated by default if
10933 hardware floating point is used.
10935 @item -mwarn-framesize=@var{framesize}
10936 @opindex mwarn-framesize
10937 Emit a warning if the current function exceeds the given frame size. Because
10938 this is a compile time check it doesn't need to be a real problem when the program
10939 runs. It is intended to identify functions which most probably cause
10940 a stack overflow. It is useful to be used in an environment with limited stack
10941 size e.g.@: the linux kernel.
10943 @item -mwarn-dynamicstack
10944 @opindex mwarn-dynamicstack
10945 Emit a warning if the function calls alloca or uses dynamically
10946 sized arrays. This is generally a bad idea with a limited stack size.
10948 @item -mstack-guard=@var{stack-guard}
10949 @item -mstack-size=@var{stack-size}
10950 @opindex mstack-guard
10951 @opindex mstack-size
10952 These arguments always have to be used in conjunction. If they are present the s390
10953 back end emits additional instructions in the function prologue which trigger a trap
10954 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
10955 (remember that the stack on s390 grows downward). These options are intended to
10956 be used to help debugging stack overflow problems. The additionally emitted code
10957 cause only little overhead and hence can also be used in production like systems
10958 without greater performance degradation. The given values have to be exact
10959 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard}.
10960 In order to be efficient the extra code makes the assumption that the stack starts
10961 at an address aligned to the value given by @var{stack-size}.
10965 @subsection SH Options
10967 These @samp{-m} options are defined for the SH implementations:
10972 Generate code for the SH1.
10976 Generate code for the SH2.
10979 Generate code for the SH2e.
10983 Generate code for the SH3.
10987 Generate code for the SH3e.
10991 Generate code for the SH4 without a floating-point unit.
10993 @item -m4-single-only
10994 @opindex m4-single-only
10995 Generate code for the SH4 with a floating-point unit that only
10996 supports single-precision arithmetic.
11000 Generate code for the SH4 assuming the floating-point unit is in
11001 single-precision mode by default.
11005 Generate code for the SH4.
11009 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
11010 floating-point unit is not used.
11012 @item -m4a-single-only
11013 @opindex m4a-single-only
11014 Generate code for the SH4a, in such a way that no double-precision
11015 floating point operations are used.
11018 @opindex m4a-single
11019 Generate code for the SH4a assuming the floating-point unit is in
11020 single-precision mode by default.
11024 Generate code for the SH4a.
11028 Same as @option{-m4a-nofpu}, except that it implicitly passes
11029 @option{-dsp} to the assembler. GCC doesn't generate any DSP
11030 instructions at the moment.
11034 Compile code for the processor in big endian mode.
11038 Compile code for the processor in little endian mode.
11042 Align doubles at 64-bit boundaries. Note that this changes the calling
11043 conventions, and thus some functions from the standard C library will
11044 not work unless you recompile it first with @option{-mdalign}.
11048 Shorten some address references at link time, when possible; uses the
11049 linker option @option{-relax}.
11053 Use 32-bit offsets in @code{switch} tables. The default is to use
11058 Enable the use of the instruction @code{fmovd}.
11062 Comply with the calling conventions defined by Renesas.
11066 Comply with the calling conventions defined by Renesas.
11070 Comply with the calling conventions defined for GCC before the Renesas
11071 conventions were available. This option is the default for all
11072 targets of the SH toolchain except for @samp{sh-symbianelf}.
11075 @opindex mnomacsave
11076 Mark the @code{MAC} register as call-clobbered, even if
11077 @option{-mhitachi} is given.
11081 Increase IEEE-compliance of floating-point code.
11085 Dump instruction size and location in the assembly code.
11088 @opindex mpadstruct
11089 This option is deprecated. It pads structures to multiple of 4 bytes,
11090 which is incompatible with the SH ABI@.
11094 Optimize for space instead of speed. Implied by @option{-Os}.
11097 @opindex mprefergot
11098 When generating position-independent code, emit function calls using
11099 the Global Offset Table instead of the Procedure Linkage Table.
11103 Generate a library function call to invalidate instruction cache
11104 entries, after fixing up a trampoline. This library function call
11105 doesn't assume it can write to the whole memory address space. This
11106 is the default when the target is @code{sh-*-linux*}.
11109 @node SPARC Options
11110 @subsection SPARC Options
11111 @cindex SPARC options
11113 These @samp{-m} options are supported on the SPARC:
11116 @item -mno-app-regs
11118 @opindex mno-app-regs
11120 Specify @option{-mapp-regs} to generate output using the global registers
11121 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
11124 To be fully SVR4 ABI compliant at the cost of some performance loss,
11125 specify @option{-mno-app-regs}. You should compile libraries and system
11126 software with this option.
11129 @itemx -mhard-float
11131 @opindex mhard-float
11132 Generate output containing floating point instructions. This is the
11136 @itemx -msoft-float
11138 @opindex msoft-float
11139 Generate output containing library calls for floating point.
11140 @strong{Warning:} the requisite libraries are not available for all SPARC
11141 targets. Normally the facilities of the machine's usual C compiler are
11142 used, but this cannot be done directly in cross-compilation. You must make
11143 your own arrangements to provide suitable library functions for
11144 cross-compilation. The embedded targets @samp{sparc-*-aout} and
11145 @samp{sparclite-*-*} do provide software floating point support.
11147 @option{-msoft-float} changes the calling convention in the output file;
11148 therefore, it is only useful if you compile @emph{all} of a program with
11149 this option. In particular, you need to compile @file{libgcc.a}, the
11150 library that comes with GCC, with @option{-msoft-float} in order for
11153 @item -mhard-quad-float
11154 @opindex mhard-quad-float
11155 Generate output containing quad-word (long double) floating point
11158 @item -msoft-quad-float
11159 @opindex msoft-quad-float
11160 Generate output containing library calls for quad-word (long double)
11161 floating point instructions. The functions called are those specified
11162 in the SPARC ABI@. This is the default.
11164 As of this writing, there are no SPARC implementations that have hardware
11165 support for the quad-word floating point instructions. They all invoke
11166 a trap handler for one of these instructions, and then the trap handler
11167 emulates the effect of the instruction. Because of the trap handler overhead,
11168 this is much slower than calling the ABI library routines. Thus the
11169 @option{-msoft-quad-float} option is the default.
11171 @item -mno-unaligned-doubles
11172 @itemx -munaligned-doubles
11173 @opindex mno-unaligned-doubles
11174 @opindex munaligned-doubles
11175 Assume that doubles have 8 byte alignment. This is the default.
11177 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
11178 alignment only if they are contained in another type, or if they have an
11179 absolute address. Otherwise, it assumes they have 4 byte alignment.
11180 Specifying this option avoids some rare compatibility problems with code
11181 generated by other compilers. It is not the default because it results
11182 in a performance loss, especially for floating point code.
11184 @item -mno-faster-structs
11185 @itemx -mfaster-structs
11186 @opindex mno-faster-structs
11187 @opindex mfaster-structs
11188 With @option{-mfaster-structs}, the compiler assumes that structures
11189 should have 8 byte alignment. This enables the use of pairs of
11190 @code{ldd} and @code{std} instructions for copies in structure
11191 assignment, in place of twice as many @code{ld} and @code{st} pairs.
11192 However, the use of this changed alignment directly violates the SPARC
11193 ABI@. Thus, it's intended only for use on targets where the developer
11194 acknowledges that their resulting code will not be directly in line with
11195 the rules of the ABI@.
11197 @item -mimpure-text
11198 @opindex mimpure-text
11199 @option{-mimpure-text}, used in addition to @option{-shared}, tells
11200 the compiler to not pass @option{-z text} to the linker when linking a
11201 shared object. Using this option, you can link position-dependent
11202 code into a shared object.
11204 @option{-mimpure-text} suppresses the ``relocations remain against
11205 allocatable but non-writable sections'' linker error message.
11206 However, the necessary relocations will trigger copy-on-write, and the
11207 shared object is not actually shared across processes. Instead of
11208 using @option{-mimpure-text}, you should compile all source code with
11209 @option{-fpic} or @option{-fPIC}.
11211 This option is only available on SunOS and Solaris.
11213 @item -mcpu=@var{cpu_type}
11215 Set the instruction set, register set, and instruction scheduling parameters
11216 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
11217 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
11218 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
11219 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
11220 @samp{ultrasparc3}.
11222 Default instruction scheduling parameters are used for values that select
11223 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
11224 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
11226 Here is a list of each supported architecture and their supported
11231 v8: supersparc, hypersparc
11232 sparclite: f930, f934, sparclite86x
11234 v9: ultrasparc, ultrasparc3
11237 By default (unless configured otherwise), GCC generates code for the V7
11238 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
11239 additionally optimizes it for the Cypress CY7C602 chip, as used in the
11240 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
11241 SPARCStation 1, 2, IPX etc.
11243 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
11244 architecture. The only difference from V7 code is that the compiler emits
11245 the integer multiply and integer divide instructions which exist in SPARC-V8
11246 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
11247 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
11250 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
11251 the SPARC architecture. This adds the integer multiply, integer divide step
11252 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
11253 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
11254 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
11255 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
11256 MB86934 chip, which is the more recent SPARClite with FPU.
11258 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
11259 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
11260 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
11261 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
11262 optimizes it for the TEMIC SPARClet chip.
11264 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
11265 architecture. This adds 64-bit integer and floating-point move instructions,
11266 3 additional floating-point condition code registers and conditional move
11267 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
11268 optimizes it for the Sun UltraSPARC I/II chips. With
11269 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
11270 Sun UltraSPARC III chip.
11272 @item -mtune=@var{cpu_type}
11274 Set the instruction scheduling parameters for machine type
11275 @var{cpu_type}, but do not set the instruction set or register set that the
11276 option @option{-mcpu=@var{cpu_type}} would.
11278 The same values for @option{-mcpu=@var{cpu_type}} can be used for
11279 @option{-mtune=@var{cpu_type}}, but the only useful values are those
11280 that select a particular cpu implementation. Those are @samp{cypress},
11281 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
11282 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
11283 @samp{ultrasparc3}.
11288 @opindex mno-v8plus
11289 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
11290 difference from the V8 ABI is that the global and out registers are
11291 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
11292 mode for all SPARC-V9 processors.
11298 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
11299 Visual Instruction Set extensions. The default is @option{-mno-vis}.
11302 These @samp{-m} options are supported in addition to the above
11303 on SPARC-V9 processors in 64-bit environments:
11306 @item -mlittle-endian
11307 @opindex mlittle-endian
11308 Generate code for a processor running in little-endian mode. It is only
11309 available for a few configurations and most notably not on Solaris.
11315 Generate code for a 32-bit or 64-bit environment.
11316 The 32-bit environment sets int, long and pointer to 32 bits.
11317 The 64-bit environment sets int to 32 bits and long and pointer
11320 @item -mcmodel=medlow
11321 @opindex mcmodel=medlow
11322 Generate code for the Medium/Low code model: 64-bit addresses, programs
11323 must be linked in the low 32 bits of memory. Programs can be statically
11324 or dynamically linked.
11326 @item -mcmodel=medmid
11327 @opindex mcmodel=medmid
11328 Generate code for the Medium/Middle code model: 64-bit addresses, programs
11329 must be linked in the low 44 bits of memory, the text and data segments must
11330 be less than 2GB in size and the data segment must be located within 2GB of
11333 @item -mcmodel=medany
11334 @opindex mcmodel=medany
11335 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
11336 may be linked anywhere in memory, the text and data segments must be less
11337 than 2GB in size and the data segment must be located within 2GB of the
11340 @item -mcmodel=embmedany
11341 @opindex mcmodel=embmedany
11342 Generate code for the Medium/Anywhere code model for embedded systems:
11343 64-bit addresses, the text and data segments must be less than 2GB in
11344 size, both starting anywhere in memory (determined at link time). The
11345 global register %g4 points to the base of the data segment. Programs
11346 are statically linked and PIC is not supported.
11349 @itemx -mno-stack-bias
11350 @opindex mstack-bias
11351 @opindex mno-stack-bias
11352 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
11353 frame pointer if present, are offset by @minus{}2047 which must be added back
11354 when making stack frame references. This is the default in 64-bit mode.
11355 Otherwise, assume no such offset is present.
11358 These switches are supported in addition to the above on Solaris:
11363 Add support for multithreading using the Solaris threads library. This
11364 option sets flags for both the preprocessor and linker. This option does
11365 not affect the thread safety of object code produced by the compiler or
11366 that of libraries supplied with it.
11370 Add support for multithreading using the POSIX threads library. This
11371 option sets flags for both the preprocessor and linker. This option does
11372 not affect the thread safety of object code produced by the compiler or
11373 that of libraries supplied with it.
11376 @node System V Options
11377 @subsection Options for System V
11379 These additional options are available on System V Release 4 for
11380 compatibility with other compilers on those systems:
11385 Create a shared object.
11386 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
11390 Identify the versions of each tool used by the compiler, in a
11391 @code{.ident} assembler directive in the output.
11395 Refrain from adding @code{.ident} directives to the output file (this is
11398 @item -YP,@var{dirs}
11400 Search the directories @var{dirs}, and no others, for libraries
11401 specified with @option{-l}.
11403 @item -Ym,@var{dir}
11405 Look in the directory @var{dir} to find the M4 preprocessor.
11406 The assembler uses this option.
11407 @c This is supposed to go with a -Yd for predefined M4 macro files, but
11408 @c the generic assembler that comes with Solaris takes just -Ym.
11411 @node TMS320C3x/C4x Options
11412 @subsection TMS320C3x/C4x Options
11413 @cindex TMS320C3x/C4x Options
11415 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
11419 @item -mcpu=@var{cpu_type}
11421 Set the instruction set, register set, and instruction scheduling
11422 parameters for machine type @var{cpu_type}. Supported values for
11423 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
11424 @samp{c44}. The default is @samp{c40} to generate code for the
11429 @itemx -msmall-memory
11431 @opindex mbig-memory
11433 @opindex msmall-memory
11435 Generates code for the big or small memory model. The small memory
11436 model assumed that all data fits into one 64K word page. At run-time
11437 the data page (DP) register must be set to point to the 64K page
11438 containing the .bss and .data program sections. The big memory model is
11439 the default and requires reloading of the DP register for every direct
11446 Allow (disallow) allocation of general integer operands into the block
11447 count register BK@.
11453 Enable (disable) generation of code using decrement and branch,
11454 DBcond(D), instructions. This is enabled by default for the C4x. To be
11455 on the safe side, this is disabled for the C3x, since the maximum
11456 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
11457 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
11458 that it can utilize the decrement and branch instruction, but will give
11459 up if there is more than one memory reference in the loop. Thus a loop
11460 where the loop counter is decremented can generate slightly more
11461 efficient code, in cases where the RPTB instruction cannot be utilized.
11463 @item -mdp-isr-reload
11465 @opindex mdp-isr-reload
11467 Force the DP register to be saved on entry to an interrupt service
11468 routine (ISR), reloaded to point to the data section, and restored on
11469 exit from the ISR@. This should not be required unless someone has
11470 violated the small memory model by modifying the DP register, say within
11477 For the C3x use the 24-bit MPYI instruction for integer multiplies
11478 instead of a library call to guarantee 32-bit results. Note that if one
11479 of the operands is a constant, then the multiplication will be performed
11480 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
11481 then squaring operations are performed inline instead of a library call.
11484 @itemx -mno-fast-fix
11486 @opindex mno-fast-fix
11487 The C3x/C4x FIX instruction to convert a floating point value to an
11488 integer value chooses the nearest integer less than or equal to the
11489 floating point value rather than to the nearest integer. Thus if the
11490 floating point number is negative, the result will be incorrectly
11491 truncated an additional code is necessary to detect and correct this
11492 case. This option can be used to disable generation of the additional
11493 code required to correct the result.
11499 Enable (disable) generation of repeat block sequences using the RPTB
11500 instruction for zero overhead looping. The RPTB construct is only used
11501 for innermost loops that do not call functions or jump across the loop
11502 boundaries. There is no advantage having nested RPTB loops due to the
11503 overhead required to save and restore the RC, RS, and RE registers.
11504 This is enabled by default with @option{-O2}.
11506 @item -mrpts=@var{count}
11510 Enable (disable) the use of the single instruction repeat instruction
11511 RPTS@. If a repeat block contains a single instruction, and the loop
11512 count can be guaranteed to be less than the value @var{count}, GCC will
11513 emit a RPTS instruction instead of a RPTB@. If no value is specified,
11514 then a RPTS will be emitted even if the loop count cannot be determined
11515 at compile time. Note that the repeated instruction following RPTS does
11516 not have to be reloaded from memory each iteration, thus freeing up the
11517 CPU buses for operands. However, since interrupts are blocked by this
11518 instruction, it is disabled by default.
11520 @item -mloop-unsigned
11521 @itemx -mno-loop-unsigned
11522 @opindex mloop-unsigned
11523 @opindex mno-loop-unsigned
11524 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
11525 is @math{2^{31} + 1} since these instructions test if the iteration count is
11526 negative to terminate the loop. If the iteration count is unsigned
11527 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
11528 exceeded. This switch allows an unsigned iteration count.
11532 Try to emit an assembler syntax that the TI assembler (asm30) is happy
11533 with. This also enforces compatibility with the API employed by the TI
11534 C3x C compiler. For example, long doubles are passed as structures
11535 rather than in floating point registers.
11541 Generate code that uses registers (stack) for passing arguments to functions.
11542 By default, arguments are passed in registers where possible rather
11543 than by pushing arguments on to the stack.
11545 @item -mparallel-insns
11546 @itemx -mno-parallel-insns
11547 @opindex mparallel-insns
11548 @opindex mno-parallel-insns
11549 Allow the generation of parallel instructions. This is enabled by
11550 default with @option{-O2}.
11552 @item -mparallel-mpy
11553 @itemx -mno-parallel-mpy
11554 @opindex mparallel-mpy
11555 @opindex mno-parallel-mpy
11556 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
11557 provided @option{-mparallel-insns} is also specified. These instructions have
11558 tight register constraints which can pessimize the code generation
11559 of large functions.
11564 @subsection V850 Options
11565 @cindex V850 Options
11567 These @samp{-m} options are defined for V850 implementations:
11571 @itemx -mno-long-calls
11572 @opindex mlong-calls
11573 @opindex mno-long-calls
11574 Treat all calls as being far away (near). If calls are assumed to be
11575 far away, the compiler will always load the functions address up into a
11576 register, and call indirect through the pointer.
11582 Do not optimize (do optimize) basic blocks that use the same index
11583 pointer 4 or more times to copy pointer into the @code{ep} register, and
11584 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
11585 option is on by default if you optimize.
11587 @item -mno-prolog-function
11588 @itemx -mprolog-function
11589 @opindex mno-prolog-function
11590 @opindex mprolog-function
11591 Do not use (do use) external functions to save and restore registers
11592 at the prologue and epilogue of a function. The external functions
11593 are slower, but use less code space if more than one function saves
11594 the same number of registers. The @option{-mprolog-function} option
11595 is on by default if you optimize.
11599 Try to make the code as small as possible. At present, this just turns
11600 on the @option{-mep} and @option{-mprolog-function} options.
11602 @item -mtda=@var{n}
11604 Put static or global variables whose size is @var{n} bytes or less into
11605 the tiny data area that register @code{ep} points to. The tiny data
11606 area can hold up to 256 bytes in total (128 bytes for byte references).
11608 @item -msda=@var{n}
11610 Put static or global variables whose size is @var{n} bytes or less into
11611 the small data area that register @code{gp} points to. The small data
11612 area can hold up to 64 kilobytes.
11614 @item -mzda=@var{n}
11616 Put static or global variables whose size is @var{n} bytes or less into
11617 the first 32 kilobytes of memory.
11621 Specify that the target processor is the V850.
11624 @opindex mbig-switch
11625 Generate code suitable for big switch tables. Use this option only if
11626 the assembler/linker complain about out of range branches within a switch
11631 This option will cause r2 and r5 to be used in the code generated by
11632 the compiler. This setting is the default.
11634 @item -mno-app-regs
11635 @opindex mno-app-regs
11636 This option will cause r2 and r5 to be treated as fixed registers.
11640 Specify that the target processor is the V850E1. The preprocessor
11641 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
11642 this option is used.
11646 Specify that the target processor is the V850E. The preprocessor
11647 constant @samp{__v850e__} will be defined if this option is used.
11649 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
11650 are defined then a default target processor will be chosen and the
11651 relevant @samp{__v850*__} preprocessor constant will be defined.
11653 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
11654 defined, regardless of which processor variant is the target.
11656 @item -mdisable-callt
11657 @opindex mdisable-callt
11658 This option will suppress generation of the CALLT instruction for the
11659 v850e and v850e1 flavors of the v850 architecture. The default is
11660 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
11665 @subsection VAX Options
11666 @cindex VAX options
11668 These @samp{-m} options are defined for the VAX:
11673 Do not output certain jump instructions (@code{aobleq} and so on)
11674 that the Unix assembler for the VAX cannot handle across long
11679 Do output those jump instructions, on the assumption that you
11680 will assemble with the GNU assembler.
11684 Output code for g-format floating point numbers instead of d-format.
11687 @node x86-64 Options
11688 @subsection x86-64 Options
11689 @cindex x86-64 options
11691 These are listed under @xref{i386 and x86-64 Options}.
11693 @node Xstormy16 Options
11694 @subsection Xstormy16 Options
11695 @cindex Xstormy16 Options
11697 These options are defined for Xstormy16:
11702 Choose startup files and linker script suitable for the simulator.
11705 @node Xtensa Options
11706 @subsection Xtensa Options
11707 @cindex Xtensa Options
11709 These options are supported for Xtensa targets:
11713 @itemx -mno-const16
11715 @opindex mno-const16
11716 Enable or disable use of @code{CONST16} instructions for loading
11717 constant values. The @code{CONST16} instruction is currently not a
11718 standard option from Tensilica. When enabled, @code{CONST16}
11719 instructions are always used in place of the standard @code{L32R}
11720 instructions. The use of @code{CONST16} is enabled by default only if
11721 the @code{L32R} instruction is not available.
11724 @itemx -mno-fused-madd
11725 @opindex mfused-madd
11726 @opindex mno-fused-madd
11727 Enable or disable use of fused multiply/add and multiply/subtract
11728 instructions in the floating-point option. This has no effect if the
11729 floating-point option is not also enabled. Disabling fused multiply/add
11730 and multiply/subtract instructions forces the compiler to use separate
11731 instructions for the multiply and add/subtract operations. This may be
11732 desirable in some cases where strict IEEE 754-compliant results are
11733 required: the fused multiply add/subtract instructions do not round the
11734 intermediate result, thereby producing results with @emph{more} bits of
11735 precision than specified by the IEEE standard. Disabling fused multiply
11736 add/subtract instructions also ensures that the program output is not
11737 sensitive to the compiler's ability to combine multiply and add/subtract
11740 @item -mtext-section-literals
11741 @itemx -mno-text-section-literals
11742 @opindex mtext-section-literals
11743 @opindex mno-text-section-literals
11744 Control the treatment of literal pools. The default is
11745 @option{-mno-text-section-literals}, which places literals in a separate
11746 section in the output file. This allows the literal pool to be placed
11747 in a data RAM/ROM, and it also allows the linker to combine literal
11748 pools from separate object files to remove redundant literals and
11749 improve code size. With @option{-mtext-section-literals}, the literals
11750 are interspersed in the text section in order to keep them as close as
11751 possible to their references. This may be necessary for large assembly
11754 @item -mtarget-align
11755 @itemx -mno-target-align
11756 @opindex mtarget-align
11757 @opindex mno-target-align
11758 When this option is enabled, GCC instructs the assembler to
11759 automatically align instructions to reduce branch penalties at the
11760 expense of some code density. The assembler attempts to widen density
11761 instructions to align branch targets and the instructions following call
11762 instructions. If there are not enough preceding safe density
11763 instructions to align a target, no widening will be performed. The
11764 default is @option{-mtarget-align}. These options do not affect the
11765 treatment of auto-aligned instructions like @code{LOOP}, which the
11766 assembler will always align, either by widening density instructions or
11767 by inserting no-op instructions.
11770 @itemx -mno-longcalls
11771 @opindex mlongcalls
11772 @opindex mno-longcalls
11773 When this option is enabled, GCC instructs the assembler to translate
11774 direct calls to indirect calls unless it can determine that the target
11775 of a direct call is in the range allowed by the call instruction. This
11776 translation typically occurs for calls to functions in other source
11777 files. Specifically, the assembler translates a direct @code{CALL}
11778 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11779 The default is @option{-mno-longcalls}. This option should be used in
11780 programs where the call target can potentially be out of range. This
11781 option is implemented in the assembler, not the compiler, so the
11782 assembly code generated by GCC will still show direct call
11783 instructions---look at the disassembled object code to see the actual
11784 instructions. Note that the assembler will use an indirect call for
11785 every cross-file call, not just those that really will be out of range.
11788 @node zSeries Options
11789 @subsection zSeries Options
11790 @cindex zSeries options
11792 These are listed under @xref{S/390 and zSeries Options}.
11794 @node Code Gen Options
11795 @section Options for Code Generation Conventions
11796 @cindex code generation conventions
11797 @cindex options, code generation
11798 @cindex run-time options
11800 These machine-independent options control the interface conventions
11801 used in code generation.
11803 Most of them have both positive and negative forms; the negative form
11804 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11805 one of the forms is listed---the one which is not the default. You
11806 can figure out the other form by either removing @samp{no-} or adding
11810 @item -fbounds-check
11811 @opindex fbounds-check
11812 For front-ends that support it, generate additional code to check that
11813 indices used to access arrays are within the declared range. This is
11814 currently only supported by the Java and Fortran 77 front-ends, where
11815 this option defaults to true and false respectively.
11819 This option generates traps for signed overflow on addition, subtraction,
11820 multiplication operations.
11824 This option instructs the compiler to assume that signed arithmetic
11825 overflow of addition, subtraction and multiplication wraps around
11826 using twos-complement representation. This flag enables some optimizations
11827 and disables other. This option is enabled by default for the Java
11828 front-end, as required by the Java language specification.
11831 @opindex fexceptions
11832 Enable exception handling. Generates extra code needed to propagate
11833 exceptions. For some targets, this implies GCC will generate frame
11834 unwind information for all functions, which can produce significant data
11835 size overhead, although it does not affect execution. If you do not
11836 specify this option, GCC will enable it by default for languages like
11837 C++ which normally require exception handling, and disable it for
11838 languages like C that do not normally require it. However, you may need
11839 to enable this option when compiling C code that needs to interoperate
11840 properly with exception handlers written in C++. You may also wish to
11841 disable this option if you are compiling older C++ programs that don't
11842 use exception handling.
11844 @item -fnon-call-exceptions
11845 @opindex fnon-call-exceptions
11846 Generate code that allows trapping instructions to throw exceptions.
11847 Note that this requires platform-specific runtime support that does
11848 not exist everywhere. Moreover, it only allows @emph{trapping}
11849 instructions to throw exceptions, i.e.@: memory references or floating
11850 point instructions. It does not allow exceptions to be thrown from
11851 arbitrary signal handlers such as @code{SIGALRM}.
11853 @item -funwind-tables
11854 @opindex funwind-tables
11855 Similar to @option{-fexceptions}, except that it will just generate any needed
11856 static data, but will not affect the generated code in any other way.
11857 You will normally not enable this option; instead, a language processor
11858 that needs this handling would enable it on your behalf.
11860 @item -fasynchronous-unwind-tables
11861 @opindex fasynchronous-unwind-tables
11862 Generate unwind table in dwarf2 format, if supported by target machine. The
11863 table is exact at each instruction boundary, so it can be used for stack
11864 unwinding from asynchronous events (such as debugger or garbage collector).
11866 @item -fpcc-struct-return
11867 @opindex fpcc-struct-return
11868 Return ``short'' @code{struct} and @code{union} values in memory like
11869 longer ones, rather than in registers. This convention is less
11870 efficient, but it has the advantage of allowing intercallability between
11871 GCC-compiled files and files compiled with other compilers, particularly
11872 the Portable C Compiler (pcc).
11874 The precise convention for returning structures in memory depends
11875 on the target configuration macros.
11877 Short structures and unions are those whose size and alignment match
11878 that of some integer type.
11880 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11881 switch is not binary compatible with code compiled with the
11882 @option{-freg-struct-return} switch.
11883 Use it to conform to a non-default application binary interface.
11885 @item -freg-struct-return
11886 @opindex freg-struct-return
11887 Return @code{struct} and @code{union} values in registers when possible.
11888 This is more efficient for small structures than
11889 @option{-fpcc-struct-return}.
11891 If you specify neither @option{-fpcc-struct-return} nor
11892 @option{-freg-struct-return}, GCC defaults to whichever convention is
11893 standard for the target. If there is no standard convention, GCC
11894 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11895 the principal compiler. In those cases, we can choose the standard, and
11896 we chose the more efficient register return alternative.
11898 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11899 switch is not binary compatible with code compiled with the
11900 @option{-fpcc-struct-return} switch.
11901 Use it to conform to a non-default application binary interface.
11903 @item -fshort-enums
11904 @opindex fshort-enums
11905 Allocate to an @code{enum} type only as many bytes as it needs for the
11906 declared range of possible values. Specifically, the @code{enum} type
11907 will be equivalent to the smallest integer type which has enough room.
11909 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11910 code that is not binary compatible with code generated without that switch.
11911 Use it to conform to a non-default application binary interface.
11913 @item -fshort-double
11914 @opindex fshort-double
11915 Use the same size for @code{double} as for @code{float}.
11917 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11918 code that is not binary compatible with code generated without that switch.
11919 Use it to conform to a non-default application binary interface.
11921 @item -fshort-wchar
11922 @opindex fshort-wchar
11923 Override the underlying type for @samp{wchar_t} to be @samp{short
11924 unsigned int} instead of the default for the target. This option is
11925 useful for building programs to run under WINE@.
11927 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11928 code that is not binary compatible with code generated without that switch.
11929 Use it to conform to a non-default application binary interface.
11931 @item -fshared-data
11932 @opindex fshared-data
11933 Requests that the data and non-@code{const} variables of this
11934 compilation be shared data rather than private data. The distinction
11935 makes sense only on certain operating systems, where shared data is
11936 shared between processes running the same program, while private data
11937 exists in one copy per process.
11940 @opindex fno-common
11941 In C, allocate even uninitialized global variables in the data section of the
11942 object file, rather than generating them as common blocks. This has the
11943 effect that if the same variable is declared (without @code{extern}) in
11944 two different compilations, you will get an error when you link them.
11945 The only reason this might be useful is if you wish to verify that the
11946 program will work on other systems which always work this way.
11950 Ignore the @samp{#ident} directive.
11952 @item -finhibit-size-directive
11953 @opindex finhibit-size-directive
11954 Don't output a @code{.size} assembler directive, or anything else that
11955 would cause trouble if the function is split in the middle, and the
11956 two halves are placed at locations far apart in memory. This option is
11957 used when compiling @file{crtstuff.c}; you should not need to use it
11960 @item -fverbose-asm
11961 @opindex fverbose-asm
11962 Put extra commentary information in the generated assembly code to
11963 make it more readable. This option is generally only of use to those
11964 who actually need to read the generated assembly code (perhaps while
11965 debugging the compiler itself).
11967 @option{-fno-verbose-asm}, the default, causes the
11968 extra information to be omitted and is useful when comparing two assembler
11973 @cindex global offset table
11975 Generate position-independent code (PIC) suitable for use in a shared
11976 library, if supported for the target machine. Such code accesses all
11977 constant addresses through a global offset table (GOT)@. The dynamic
11978 loader resolves the GOT entries when the program starts (the dynamic
11979 loader is not part of GCC; it is part of the operating system). If
11980 the GOT size for the linked executable exceeds a machine-specific
11981 maximum size, you get an error message from the linker indicating that
11982 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11983 instead. (These maximums are 8k on the SPARC and 32k
11984 on the m68k and RS/6000. The 386 has no such limit.)
11986 Position-independent code requires special support, and therefore works
11987 only on certain machines. For the 386, GCC supports PIC for System V
11988 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11989 position-independent.
11993 If supported for the target machine, emit position-independent code,
11994 suitable for dynamic linking and avoiding any limit on the size of the
11995 global offset table. This option makes a difference on the m68k
11998 Position-independent code requires special support, and therefore works
11999 only on certain machines.
12005 These options are similar to @option{-fpic} and @option{-fPIC}, but
12006 generated position independent code can be only linked into executables.
12007 Usually these options are used when @option{-pie} GCC option will be
12008 used during linking.
12010 @item -ffixed-@var{reg}
12012 Treat the register named @var{reg} as a fixed register; generated code
12013 should never refer to it (except perhaps as a stack pointer, frame
12014 pointer or in some other fixed role).
12016 @var{reg} must be the name of a register. The register names accepted
12017 are machine-specific and are defined in the @code{REGISTER_NAMES}
12018 macro in the machine description macro file.
12020 This flag does not have a negative form, because it specifies a
12023 @item -fcall-used-@var{reg}
12024 @opindex fcall-used
12025 Treat the register named @var{reg} as an allocable register that is
12026 clobbered by function calls. It may be allocated for temporaries or
12027 variables that do not live across a call. Functions compiled this way
12028 will not save and restore the register @var{reg}.
12030 It is an error to used this flag with the frame pointer or stack pointer.
12031 Use of this flag for other registers that have fixed pervasive roles in
12032 the machine's execution model will produce disastrous results.
12034 This flag does not have a negative form, because it specifies a
12037 @item -fcall-saved-@var{reg}
12038 @opindex fcall-saved
12039 Treat the register named @var{reg} as an allocable register saved by
12040 functions. It may be allocated even for temporaries or variables that
12041 live across a call. Functions compiled this way will save and restore
12042 the register @var{reg} if they use it.
12044 It is an error to used this flag with the frame pointer or stack pointer.
12045 Use of this flag for other registers that have fixed pervasive roles in
12046 the machine's execution model will produce disastrous results.
12048 A different sort of disaster will result from the use of this flag for
12049 a register in which function values may be returned.
12051 This flag does not have a negative form, because it specifies a
12054 @item -fpack-struct[=@var{n}]
12055 @opindex fpack-struct
12056 Without a value specified, pack all structure members together without
12057 holes. When a value is specified (which must be a small power of two), pack
12058 structure members according to this value, representing the maximum
12059 alignment (that is, objects with default alignment requirements larger than
12060 this will be output potentially unaligned at the next fitting location.
12062 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
12063 code that is not binary compatible with code generated without that switch.
12064 Additionally, it makes the code suboptimal.
12065 Use it to conform to a non-default application binary interface.
12067 @item -finstrument-functions
12068 @opindex finstrument-functions
12069 Generate instrumentation calls for entry and exit to functions. Just
12070 after function entry and just before function exit, the following
12071 profiling functions will be called with the address of the current
12072 function and its call site. (On some platforms,
12073 @code{__builtin_return_address} does not work beyond the current
12074 function, so the call site information may not be available to the
12075 profiling functions otherwise.)
12078 void __cyg_profile_func_enter (void *this_fn,
12080 void __cyg_profile_func_exit (void *this_fn,
12084 The first argument is the address of the start of the current function,
12085 which may be looked up exactly in the symbol table.
12087 This instrumentation is also done for functions expanded inline in other
12088 functions. The profiling calls will indicate where, conceptually, the
12089 inline function is entered and exited. This means that addressable
12090 versions of such functions must be available. If all your uses of a
12091 function are expanded inline, this may mean an additional expansion of
12092 code size. If you use @samp{extern inline} in your C code, an
12093 addressable version of such functions must be provided. (This is
12094 normally the case anyways, but if you get lucky and the optimizer always
12095 expands the functions inline, you might have gotten away without
12096 providing static copies.)
12098 A function may be given the attribute @code{no_instrument_function}, in
12099 which case this instrumentation will not be done. This can be used, for
12100 example, for the profiling functions listed above, high-priority
12101 interrupt routines, and any functions from which the profiling functions
12102 cannot safely be called (perhaps signal handlers, if the profiling
12103 routines generate output or allocate memory).
12105 @item -fstack-check
12106 @opindex fstack-check
12107 Generate code to verify that you do not go beyond the boundary of the
12108 stack. You should specify this flag if you are running in an
12109 environment with multiple threads, but only rarely need to specify it in
12110 a single-threaded environment since stack overflow is automatically
12111 detected on nearly all systems if there is only one stack.
12113 Note that this switch does not actually cause checking to be done; the
12114 operating system must do that. The switch causes generation of code
12115 to ensure that the operating system sees the stack being extended.
12117 @item -fstack-limit-register=@var{reg}
12118 @itemx -fstack-limit-symbol=@var{sym}
12119 @itemx -fno-stack-limit
12120 @opindex fstack-limit-register
12121 @opindex fstack-limit-symbol
12122 @opindex fno-stack-limit
12123 Generate code to ensure that the stack does not grow beyond a certain value,
12124 either the value of a register or the address of a symbol. If the stack
12125 would grow beyond the value, a signal is raised. For most targets,
12126 the signal is raised before the stack overruns the boundary, so
12127 it is possible to catch the signal without taking special precautions.
12129 For instance, if the stack starts at absolute address @samp{0x80000000}
12130 and grows downwards, you can use the flags
12131 @option{-fstack-limit-symbol=__stack_limit} and
12132 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
12133 of 128KB@. Note that this may only work with the GNU linker.
12135 @cindex aliasing of parameters
12136 @cindex parameters, aliased
12137 @item -fargument-alias
12138 @itemx -fargument-noalias
12139 @itemx -fargument-noalias-global
12140 @opindex fargument-alias
12141 @opindex fargument-noalias
12142 @opindex fargument-noalias-global
12143 Specify the possible relationships among parameters and between
12144 parameters and global data.
12146 @option{-fargument-alias} specifies that arguments (parameters) may
12147 alias each other and may alias global storage.@*
12148 @option{-fargument-noalias} specifies that arguments do not alias
12149 each other, but may alias global storage.@*
12150 @option{-fargument-noalias-global} specifies that arguments do not
12151 alias each other and do not alias global storage.
12153 Each language will automatically use whatever option is required by
12154 the language standard. You should not need to use these options yourself.
12156 @item -fleading-underscore
12157 @opindex fleading-underscore
12158 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
12159 change the way C symbols are represented in the object file. One use
12160 is to help link with legacy assembly code.
12162 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
12163 generate code that is not binary compatible with code generated without that
12164 switch. Use it to conform to a non-default application binary interface.
12165 Not all targets provide complete support for this switch.
12167 @item -ftls-model=@var{model}
12168 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
12169 The @var{model} argument should be one of @code{global-dynamic},
12170 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
12172 The default without @option{-fpic} is @code{initial-exec}; with
12173 @option{-fpic} the default is @code{global-dynamic}.
12175 @item -fvisibility=@var{default|internal|hidden|protected}
12176 @opindex fvisibility
12177 Set the default ELF image symbol visibility to the specified option---all
12178 symbols will be marked with this unless overridden within the code.
12179 Using this feature can very substantially improve linking and
12180 load times of shared object libraries, produce more optimized
12181 code, provide near-perfect API export and prevent symbol clashes.
12182 It is @strong{strongly} recommended that you use this in any shared objects
12185 Despite the nomenclature, @code{default} always means public ie;
12186 available to be linked against from outside the shared object.
12187 @code{protected} and @code{internal} are pretty useless in real-world
12188 usage so the only other commonly used option will be @code{hidden}.
12189 The default if @option{-fvisibility} isn't specified is
12190 @code{default}, i.e., make every
12191 symbol public---this causes the same behavior as previous versions of
12194 A good explanation of the benefits offered by ensuring ELF
12195 symbols have the correct visibility is given by ``How To Write
12196 Shared Libraries'' by Ulrich Drepper (which can be found at
12197 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
12198 solution made possible by this option to marking things hidden when
12199 the default is public is to make the default hidden and mark things
12200 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
12201 and @code{__attribute__ ((visibility("default")))} instead of
12202 @code{__declspec(dllexport)} you get almost identical semantics with
12203 identical syntax. This is a great boon to those working with
12204 cross-platform projects.
12206 For those adding visibility support to existing code, you may find
12207 @samp{#pragma GCC visibility} of use. This works by you enclosing
12208 the declarations you wish to set visibility for with (for example)
12209 @samp{#pragma GCC visibility push(hidden)} and
12210 @samp{#pragma GCC visibility pop}. These can be nested up to sixteen
12211 times. Bear in mind that symbol visibility should be viewed @strong{as
12212 part of the API interface contract} and thus all new code should
12213 always specify visibility when it is not the default ie; declarations
12214 only for use within the local DSO should @strong{always} be marked explicitly
12215 as hidden as so to avoid PLT indirection overheads---making this
12216 abundantly clear also aids readability and self-documentation of the code.
12217 Note that due to ISO C++ specification requirements, operator new and
12218 operator delete must always be of default visibility.
12220 An overview of these techniques, their benefits and how to use them
12221 is at @w{@uref{http://www.nedprod.com/programs/gccvisibility.html}}.
12227 @node Environment Variables
12228 @section Environment Variables Affecting GCC
12229 @cindex environment variables
12231 @c man begin ENVIRONMENT
12232 This section describes several environment variables that affect how GCC
12233 operates. Some of them work by specifying directories or prefixes to use
12234 when searching for various kinds of files. Some are used to specify other
12235 aspects of the compilation environment.
12237 Note that you can also specify places to search using options such as
12238 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
12239 take precedence over places specified using environment variables, which
12240 in turn take precedence over those specified by the configuration of GCC@.
12241 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
12242 GNU Compiler Collection (GCC) Internals}.
12247 @c @itemx LC_COLLATE
12249 @c @itemx LC_MONETARY
12250 @c @itemx LC_NUMERIC
12255 @c @findex LC_COLLATE
12256 @findex LC_MESSAGES
12257 @c @findex LC_MONETARY
12258 @c @findex LC_NUMERIC
12262 These environment variables control the way that GCC uses
12263 localization information that allow GCC to work with different
12264 national conventions. GCC inspects the locale categories
12265 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
12266 so. These locale categories can be set to any value supported by your
12267 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
12268 Kingdom encoded in UTF-8.
12270 The @env{LC_CTYPE} environment variable specifies character
12271 classification. GCC uses it to determine the character boundaries in
12272 a string; this is needed for some multibyte encodings that contain quote
12273 and escape characters that would otherwise be interpreted as a string
12276 The @env{LC_MESSAGES} environment variable specifies the language to
12277 use in diagnostic messages.
12279 If the @env{LC_ALL} environment variable is set, it overrides the value
12280 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
12281 and @env{LC_MESSAGES} default to the value of the @env{LANG}
12282 environment variable. If none of these variables are set, GCC
12283 defaults to traditional C English behavior.
12287 If @env{TMPDIR} is set, it specifies the directory to use for temporary
12288 files. GCC uses temporary files to hold the output of one stage of
12289 compilation which is to be used as input to the next stage: for example,
12290 the output of the preprocessor, which is the input to the compiler
12293 @item GCC_EXEC_PREFIX
12294 @findex GCC_EXEC_PREFIX
12295 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
12296 names of the subprograms executed by the compiler. No slash is added
12297 when this prefix is combined with the name of a subprogram, but you can
12298 specify a prefix that ends with a slash if you wish.
12300 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
12301 an appropriate prefix to use based on the pathname it was invoked with.
12303 If GCC cannot find the subprogram using the specified prefix, it
12304 tries looking in the usual places for the subprogram.
12306 The default value of @env{GCC_EXEC_PREFIX} is
12307 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
12308 of @code{prefix} when you ran the @file{configure} script.
12310 Other prefixes specified with @option{-B} take precedence over this prefix.
12312 This prefix is also used for finding files such as @file{crt0.o} that are
12315 In addition, the prefix is used in an unusual way in finding the
12316 directories to search for header files. For each of the standard
12317 directories whose name normally begins with @samp{/usr/local/lib/gcc}
12318 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
12319 replacing that beginning with the specified prefix to produce an
12320 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
12321 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
12322 These alternate directories are searched first; the standard directories
12325 @item COMPILER_PATH
12326 @findex COMPILER_PATH
12327 The value of @env{COMPILER_PATH} is a colon-separated list of
12328 directories, much like @env{PATH}. GCC tries the directories thus
12329 specified when searching for subprograms, if it can't find the
12330 subprograms using @env{GCC_EXEC_PREFIX}.
12333 @findex LIBRARY_PATH
12334 The value of @env{LIBRARY_PATH} is a colon-separated list of
12335 directories, much like @env{PATH}. When configured as a native compiler,
12336 GCC tries the directories thus specified when searching for special
12337 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
12338 using GCC also uses these directories when searching for ordinary
12339 libraries for the @option{-l} option (but directories specified with
12340 @option{-L} come first).
12344 @cindex locale definition
12345 This variable is used to pass locale information to the compiler. One way in
12346 which this information is used is to determine the character set to be used
12347 when character literals, string literals and comments are parsed in C and C++.
12348 When the compiler is configured to allow multibyte characters,
12349 the following values for @env{LANG} are recognized:
12353 Recognize JIS characters.
12355 Recognize SJIS characters.
12357 Recognize EUCJP characters.
12360 If @env{LANG} is not defined, or if it has some other value, then the
12361 compiler will use mblen and mbtowc as defined by the default locale to
12362 recognize and translate multibyte characters.
12366 Some additional environments variables affect the behavior of the
12369 @include cppenv.texi
12373 @node Precompiled Headers
12374 @section Using Precompiled Headers
12375 @cindex precompiled headers
12376 @cindex speed of compilation
12378 Often large projects have many header files that are included in every
12379 source file. The time the compiler takes to process these header files
12380 over and over again can account for nearly all of the time required to
12381 build the project. To make builds faster, GCC allows users to
12382 `precompile' a header file; then, if builds can use the precompiled
12383 header file they will be much faster.
12385 @strong{Caution:} There are a few known situations where GCC will
12386 crash when trying to use a precompiled header. If you have trouble
12387 with a precompiled header, you should remove the precompiled header
12388 and compile without it. In addition, please use GCC's on-line
12389 defect-tracking system to report any problems you encounter with
12390 precompiled headers. @xref{Bugs}.
12392 To create a precompiled header file, simply compile it as you would any
12393 other file, if necessary using the @option{-x} option to make the driver
12394 treat it as a C or C++ header file. You will probably want to use a
12395 tool like @command{make} to keep the precompiled header up-to-date when
12396 the headers it contains change.
12398 A precompiled header file will be searched for when @code{#include} is
12399 seen in the compilation. As it searches for the included file
12400 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
12401 compiler looks for a precompiled header in each directory just before it
12402 looks for the include file in that directory. The name searched for is
12403 the name specified in the @code{#include} with @samp{.gch} appended. If
12404 the precompiled header file can't be used, it is ignored.
12406 For instance, if you have @code{#include "all.h"}, and you have
12407 @file{all.h.gch} in the same directory as @file{all.h}, then the
12408 precompiled header file will be used if possible, and the original
12409 header will be used otherwise.
12411 Alternatively, you might decide to put the precompiled header file in a
12412 directory and use @option{-I} to ensure that directory is searched
12413 before (or instead of) the directory containing the original header.
12414 Then, if you want to check that the precompiled header file is always
12415 used, you can put a file of the same name as the original header in this
12416 directory containing an @code{#error} command.
12418 This also works with @option{-include}. So yet another way to use
12419 precompiled headers, good for projects not designed with precompiled
12420 header files in mind, is to simply take most of the header files used by
12421 a project, include them from another header file, precompile that header
12422 file, and @option{-include} the precompiled header. If the header files
12423 have guards against multiple inclusion, they will be skipped because
12424 they've already been included (in the precompiled header).
12426 If you need to precompile the same header file for different
12427 languages, targets, or compiler options, you can instead make a
12428 @emph{directory} named like @file{all.h.gch}, and put each precompiled
12429 header in the directory, perhaps using @option{-o}. It doesn't matter
12430 what you call the files in the directory, every precompiled header in
12431 the directory will be considered. The first precompiled header
12432 encountered in the directory that is valid for this compilation will
12433 be used; they're searched in no particular order.
12435 There are many other possibilities, limited only by your imagination,
12436 good sense, and the constraints of your build system.
12438 A precompiled header file can be used only when these conditions apply:
12442 Only one precompiled header can be used in a particular compilation.
12445 A precompiled header can't be used once the first C token is seen. You
12446 can have preprocessor directives before a precompiled header; you can
12447 even include a precompiled header from inside another header, so long as
12448 there are no C tokens before the @code{#include}.
12451 The precompiled header file must be produced for the same language as
12452 the current compilation. You can't use a C precompiled header for a C++
12456 The precompiled header file must be produced by the same compiler
12457 version and configuration as the current compilation is using.
12458 The easiest way to guarantee this is to use the same compiler binary
12459 for creating and using precompiled headers.
12462 Any macros defined before the precompiled header is included must
12463 either be defined in the same way as when the precompiled header was
12464 generated, or must not affect the precompiled header, which usually
12465 means that the they don't appear in the precompiled header at all.
12467 The @option{-D} option is one way to define a macro before a
12468 precompiled header is included; using a @code{#define} can also do it.
12469 There are also some options that define macros implicitly, like
12470 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
12473 @item If debugging information is output when using the precompiled
12474 header, using @option{-g} or similar, the same kind of debugging information
12475 must have been output when building the precompiled header. However,
12476 a precompiled header built using @option{-g} can be used in a compilation
12477 when no debugging information is being output.
12479 @item The same @option{-m} options must generally be used when building
12480 and using the precompiled header. @xref{Submodel Options},
12481 for any cases where this rule is relaxed.
12483 @item Each of the following options must be the same when building and using
12484 the precompiled header:
12486 @gccoptlist{-fexceptions -funit-at-a-time}
12489 Some other command-line options starting with @option{-f},
12490 @option{-p}, or @option{-O} must be defined in the same way as when
12491 the precompiled header was generated. At present, it's not clear
12492 which options are safe to change and which are not; the safest choice
12493 is to use exactly the same options when generating and using the
12494 precompiled header. The following are known to be safe:
12496 @gccoptlist{-fpreprocessed -pedantic-errors}
12500 For all of these except the last, the compiler will automatically
12501 ignore the precompiled header if the conditions aren't met. If you
12502 find an option combination that doesn't work and doesn't cause the
12503 precompiled header to be ignored, please consider filing a bug report,
12506 If you do use differing options when generating and using the
12507 precompiled header, the actual behavior will be a mixture of the
12508 behavior for the options. For instance, if you use @option{-g} to
12509 generate the precompiled header but not when using it, you may or may
12510 not get debugging information for routines in the precompiled header.
12512 @node Running Protoize
12513 @section Running Protoize
12515 The program @code{protoize} is an optional part of GCC@. You can use
12516 it to add prototypes to a program, thus converting the program to ISO
12517 C in one respect. The companion program @code{unprotoize} does the
12518 reverse: it removes argument types from any prototypes that are found.
12520 When you run these programs, you must specify a set of source files as
12521 command line arguments. The conversion programs start out by compiling
12522 these files to see what functions they define. The information gathered
12523 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
12525 After scanning comes actual conversion. The specified files are all
12526 eligible to be converted; any files they include (whether sources or
12527 just headers) are eligible as well.
12529 But not all the eligible files are converted. By default,
12530 @code{protoize} and @code{unprotoize} convert only source and header
12531 files in the current directory. You can specify additional directories
12532 whose files should be converted with the @option{-d @var{directory}}
12533 option. You can also specify particular files to exclude with the
12534 @option{-x @var{file}} option. A file is converted if it is eligible, its
12535 directory name matches one of the specified directory names, and its
12536 name within the directory has not been excluded.
12538 Basic conversion with @code{protoize} consists of rewriting most
12539 function definitions and function declarations to specify the types of
12540 the arguments. The only ones not rewritten are those for varargs
12543 @code{protoize} optionally inserts prototype declarations at the
12544 beginning of the source file, to make them available for any calls that
12545 precede the function's definition. Or it can insert prototype
12546 declarations with block scope in the blocks where undeclared functions
12549 Basic conversion with @code{unprotoize} consists of rewriting most
12550 function declarations to remove any argument types, and rewriting
12551 function definitions to the old-style pre-ISO form.
12553 Both conversion programs print a warning for any function declaration or
12554 definition that they can't convert. You can suppress these warnings
12557 The output from @code{protoize} or @code{unprotoize} replaces the
12558 original source file. The original file is renamed to a name ending
12559 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
12560 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
12561 for DOS) file already exists, then the source file is simply discarded.
12563 @code{protoize} and @code{unprotoize} both depend on GCC itself to
12564 scan the program and collect information about the functions it uses.
12565 So neither of these programs will work until GCC is installed.
12567 Here is a table of the options you can use with @code{protoize} and
12568 @code{unprotoize}. Each option works with both programs unless
12572 @item -B @var{directory}
12573 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
12574 usual directory (normally @file{/usr/local/lib}). This file contains
12575 prototype information about standard system functions. This option
12576 applies only to @code{protoize}.
12578 @item -c @var{compilation-options}
12579 Use @var{compilation-options} as the options when running @command{gcc} to
12580 produce the @samp{.X} files. The special option @option{-aux-info} is
12581 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
12583 Note that the compilation options must be given as a single argument to
12584 @code{protoize} or @code{unprotoize}. If you want to specify several
12585 @command{gcc} options, you must quote the entire set of compilation options
12586 to make them a single word in the shell.
12588 There are certain @command{gcc} arguments that you cannot use, because they
12589 would produce the wrong kind of output. These include @option{-g},
12590 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
12591 the @var{compilation-options}, they are ignored.
12594 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
12595 systems) instead of @samp{.c}. This is convenient if you are converting
12596 a C program to C++. This option applies only to @code{protoize}.
12599 Add explicit global declarations. This means inserting explicit
12600 declarations at the beginning of each source file for each function
12601 that is called in the file and was not declared. These declarations
12602 precede the first function definition that contains a call to an
12603 undeclared function. This option applies only to @code{protoize}.
12605 @item -i @var{string}
12606 Indent old-style parameter declarations with the string @var{string}.
12607 This option applies only to @code{protoize}.
12609 @code{unprotoize} converts prototyped function definitions to old-style
12610 function definitions, where the arguments are declared between the
12611 argument list and the initial @samp{@{}. By default, @code{unprotoize}
12612 uses five spaces as the indentation. If you want to indent with just
12613 one space instead, use @option{-i " "}.
12616 Keep the @samp{.X} files. Normally, they are deleted after conversion
12620 Add explicit local declarations. @code{protoize} with @option{-l} inserts
12621 a prototype declaration for each function in each block which calls the
12622 function without any declaration. This option applies only to
12626 Make no real changes. This mode just prints information about the conversions
12627 that would have been done without @option{-n}.
12630 Make no @samp{.save} files. The original files are simply deleted.
12631 Use this option with caution.
12633 @item -p @var{program}
12634 Use the program @var{program} as the compiler. Normally, the name
12635 @file{gcc} is used.
12638 Work quietly. Most warnings are suppressed.
12641 Print the version number, just like @option{-v} for @command{gcc}.
12644 If you need special compiler options to compile one of your program's
12645 source files, then you should generate that file's @samp{.X} file
12646 specially, by running @command{gcc} on that source file with the
12647 appropriate options and the option @option{-aux-info}. Then run
12648 @code{protoize} on the entire set of files. @code{protoize} will use
12649 the existing @samp{.X} file because it is newer than the source file.
12653 gcc -Dfoo=bar file1.c -aux-info file1.X
12658 You need to include the special files along with the rest in the
12659 @code{protoize} command, even though their @samp{.X} files already
12660 exist, because otherwise they won't get converted.
12662 @xref{Protoize Caveats}, for more information on how to use
12663 @code{protoize} successfully.