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 -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 @gol
506 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
507 -mno-dwarf2-asm -mearly-stop-bits @gol
508 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
509 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64}
511 @emph{M32R/D Options}
512 @gccoptlist{-m32r2 -m32rx -m32r @gol
514 -malign-loops -mno-align-loops @gol
515 -missue-rate=@var{number} @gol
516 -mbranch-cost=@var{number} @gol
517 -mmodel=@var{code-size-model-type} @gol
518 -msdata=@var{sdata-type} @gol
519 -mno-flush-func -mflush-func=@var{name} @gol
520 -mno-flush-trap -mflush-trap=@var{number} @gol
523 @emph{M680x0 Options}
524 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
525 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
526 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
527 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
528 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
530 @emph{M68hc1x Options}
531 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
532 -mauto-incdec -minmax -mlong-calls -mshort @gol
533 -msoft-reg-count=@var{count}}
536 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
537 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
538 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
539 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
540 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
543 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
544 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
545 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
546 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
547 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
548 -mpaired-single -mips3d @gol
549 -mint64 -mlong64 -mlong32 @gol
550 -G@var{num} -membedded-data -mno-embedded-data @gol
551 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
552 -msplit-addresses -mno-split-addresses @gol
553 -mexplicit-relocs -mno-explicit-relocs @gol
554 -mcheck-zero-division -mno-check-zero-division @gol
555 -mdivide-traps -mdivide-breaks @gol
556 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
557 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
558 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
559 -mfix-vr4120 -mno-fix-vr4120 -mfix-sb1 -mno-fix-sb1 @gol
560 -mflush-func=@var{func} -mno-flush-func @gol
561 -mbranch-likely -mno-branch-likely @gol
562 -mfp-exceptions -mno-fp-exceptions @gol
563 -mvr4130-align -mno-vr4130-align}
566 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
567 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
568 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
569 -mno-base-addresses -msingle-exit -mno-single-exit}
571 @emph{MN10300 Options}
572 @gccoptlist{-mmult-bug -mno-mult-bug @gol
573 -mam33 -mno-am33 @gol
574 -mam33-2 -mno-am33-2 @gol
578 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
579 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
580 -mregparam -mnoregparam -msb -mnosb @gol
581 -mbitfield -mnobitfield -mhimem -mnohimem}
583 @emph{PDP-11 Options}
584 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
585 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
586 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
587 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
588 -mbranch-expensive -mbranch-cheap @gol
589 -msplit -mno-split -munix-asm -mdec-asm}
591 @emph{PowerPC Options}
592 See RS/6000 and PowerPC Options.
594 @emph{RS/6000 and PowerPC Options}
595 @gccoptlist{-mcpu=@var{cpu-type} @gol
596 -mtune=@var{cpu-type} @gol
597 -mpower -mno-power -mpower2 -mno-power2 @gol
598 -mpowerpc -mpowerpc64 -mno-powerpc @gol
599 -maltivec -mno-altivec @gol
600 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
601 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
602 -mnew-mnemonics -mold-mnemonics @gol
603 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
604 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
605 -malign-power -malign-natural @gol
606 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
607 -mstring -mno-string -mupdate -mno-update @gol
608 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
609 -mstrict-align -mno-strict-align -mrelocatable @gol
610 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
611 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
612 -mdynamic-no-pic @gol
613 -mprioritize-restricted-insns=@var{priority} @gol
614 -msched-costly-dep=@var{dependence_type} @gol
615 -minsert-sched-nops=@var{scheme} @gol
616 -mcall-sysv -mcall-netbsd @gol
617 -maix-struct-return -msvr4-struct-return @gol
618 -mabi=altivec -mabi=no-altivec @gol
619 -mabi=spe -mabi=no-spe @gol
620 -misel=yes -misel=no @gol
621 -mspe=yes -mspe=no @gol
622 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
623 -mprototype -mno-prototype @gol
624 -msim -mmvme -mads -myellowknife -memb -msdata @gol
625 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
627 @emph{S/390 and zSeries Options}
628 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
629 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
630 -mpacked-stack -mno-packed-stack @gol
631 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
632 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
633 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
634 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
637 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
638 -m4-nofpu -m4-single-only -m4-single -m4 @gol
639 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
640 -m5-64media -m5-64media-nofpu @gol
641 -m5-32media -m5-32media-nofpu @gol
642 -m5-compact -m5-compact-nofpu @gol
643 -mb -ml -mdalign -mrelax @gol
644 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
645 -mieee -misize -mpadstruct -mspace @gol
646 -mprefergot -musermode}
649 @gccoptlist{-mcpu=@var{cpu-type} @gol
650 -mtune=@var{cpu-type} @gol
651 -mcmodel=@var{code-model} @gol
652 -m32 -m64 -mapp-regs -mno-app-regs @gol
653 -mfaster-structs -mno-faster-structs @gol
654 -mfpu -mno-fpu -mhard-float -msoft-float @gol
655 -mhard-quad-float -msoft-quad-float @gol
656 -mimpure-text -mno-impure-text -mlittle-endian @gol
657 -mstack-bias -mno-stack-bias @gol
658 -munaligned-doubles -mno-unaligned-doubles @gol
659 -mv8plus -mno-v8plus -mvis -mno-vis
662 @emph{System V Options}
663 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
665 @emph{TMS320C3x/C4x Options}
666 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
667 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
668 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
669 -mparallel-insns -mparallel-mpy -mpreserve-float}
672 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
673 -mprolog-function -mno-prolog-function -mspace @gol
674 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
675 -mapp-regs -mno-app-regs @gol
676 -mdisable-callt -mno-disable-callt @gol
682 @gccoptlist{-mg -mgnu -munix}
684 @emph{x86-64 Options}
685 See i386 and x86-64 Options.
687 @emph{Xstormy16 Options}
690 @emph{Xtensa Options}
691 @gccoptlist{-mconst16 -mno-const16 @gol
692 -mfused-madd -mno-fused-madd @gol
693 -mtext-section-literals -mno-text-section-literals @gol
694 -mtarget-align -mno-target-align @gol
695 -mlongcalls -mno-longcalls}
697 @emph{zSeries Options}
698 See S/390 and zSeries Options.
700 @item Code Generation Options
701 @xref{Code Gen Options,,Options for Code Generation Conventions}.
702 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
703 -ffixed-@var{reg} -fexceptions @gol
704 -fnon-call-exceptions -funwind-tables @gol
705 -fasynchronous-unwind-tables @gol
706 -finhibit-size-directive -finstrument-functions @gol
707 -fno-common -fno-ident @gol
708 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
709 -freg-struct-return -fshared-data -fshort-enums @gol
710 -fshort-double -fshort-wchar @gol
711 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
712 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
713 -fargument-alias -fargument-noalias @gol
714 -fargument-noalias-global -fleading-underscore @gol
715 -ftls-model=@var{model} @gol
716 -ftrapv -fwrapv -fbounds-check @gol
721 * Overall Options:: Controlling the kind of output:
722 an executable, object files, assembler files,
723 or preprocessed source.
724 * C Dialect Options:: Controlling the variant of C language compiled.
725 * C++ Dialect Options:: Variations on C++.
726 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
728 * Language Independent Options:: Controlling how diagnostics should be
730 * Warning Options:: How picky should the compiler be?
731 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
732 * Optimize Options:: How much optimization?
733 * Preprocessor Options:: Controlling header files and macro definitions.
734 Also, getting dependency information for Make.
735 * Assembler Options:: Passing options to the assembler.
736 * Link Options:: Specifying libraries and so on.
737 * Directory Options:: Where to find header files and libraries.
738 Where to find the compiler executable files.
739 * Spec Files:: How to pass switches to sub-processes.
740 * Target Options:: Running a cross-compiler, or an old version of GCC.
743 @node Overall Options
744 @section Options Controlling the Kind of Output
746 Compilation can involve up to four stages: preprocessing, compilation
747 proper, assembly and linking, always in that order. GCC is capable of
748 preprocessing and compiling several files either into several
749 assembler input files, or into one assembler input file; then each
750 assembler input file produces an object file, and linking combines all
751 the object files (those newly compiled, and those specified as input)
752 into an executable file.
754 @cindex file name suffix
755 For any given input file, the file name suffix determines what kind of
760 C source code which must be preprocessed.
763 C source code which should not be preprocessed.
766 C++ source code which should not be preprocessed.
769 Objective-C source code. Note that you must link with the @file{libobjc}
770 library to make an Objective-C program work.
773 Objective-C source code which should not be preprocessed.
777 Objective-C++ source code. Note that you must link with the @file{libobjc}
778 library to make an Objective-C++ program work. Note that @samp{.M} refers
779 to a literal capital M@.
782 Objective-C++ source code which should not be preprocessed.
785 C, C++, Objective-C or Objective-C++ header file to be turned into a
790 @itemx @var{file}.cxx
791 @itemx @var{file}.cpp
792 @itemx @var{file}.CPP
793 @itemx @var{file}.c++
795 C++ source code which must be preprocessed. Note that in @samp{.cxx},
796 the last two letters must both be literally @samp{x}. Likewise,
797 @samp{.C} refers to a literal capital C@.
801 C++ header file to be turned into a precompiled header.
804 @itemx @var{file}.for
805 @itemx @var{file}.FOR
806 Fortran source code which should not be preprocessed.
809 @itemx @var{file}.fpp
810 @itemx @var{file}.FPP
811 Fortran source code which must be preprocessed (with the traditional
815 Fortran source code which must be preprocessed with a RATFOR
816 preprocessor (not included with GCC)@.
819 @itemx @var{file}.f95
820 Fortran 90/95 source code which should not be preprocessed.
822 @c FIXME: Descriptions of Java file types.
829 Ada source code file which contains a library unit declaration (a
830 declaration of a package, subprogram, or generic, or a generic
831 instantiation), or a library unit renaming declaration (a package,
832 generic, or subprogram renaming declaration). Such files are also
835 @itemx @var{file}.adb
836 Ada source code file containing a library unit body (a subprogram or
837 package body). Such files are also called @dfn{bodies}.
839 @c GCC also knows about some suffixes for languages not yet included:
848 Assembler code which must be preprocessed.
851 An object file to be fed straight into linking.
852 Any file name with no recognized suffix is treated this way.
856 You can specify the input language explicitly with the @option{-x} option:
859 @item -x @var{language}
860 Specify explicitly the @var{language} for the following input files
861 (rather than letting the compiler choose a default based on the file
862 name suffix). This option applies to all following input files until
863 the next @option{-x} option. Possible values for @var{language} are:
865 c c-header c-cpp-output
866 c++ c++-header c++-cpp-output
867 objective-c objective-c-header objective-c-cpp-output
868 objective-c++ objective-c++-header objective-c++-cpp-output
869 assembler assembler-with-cpp
871 f77 f77-cpp-input ratfor
878 Turn off any specification of a language, so that subsequent files are
879 handled according to their file name suffixes (as they are if @option{-x}
880 has not been used at all).
882 @item -pass-exit-codes
883 @opindex pass-exit-codes
884 Normally the @command{gcc} program will exit with the code of 1 if any
885 phase of the compiler returns a non-success return code. If you specify
886 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
887 numerically highest error produced by any phase that returned an error
891 If you only want some of the stages of compilation, you can use
892 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
893 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
894 @command{gcc} is to stop. Note that some combinations (for example,
895 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
900 Compile or assemble the source files, but do not link. The linking
901 stage simply is not done. The ultimate output is in the form of an
902 object file for each source file.
904 By default, the object file name for a source file is made by replacing
905 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
907 Unrecognized input files, not requiring compilation or assembly, are
912 Stop after the stage of compilation proper; do not assemble. The output
913 is in the form of an assembler code file for each non-assembler input
916 By default, the assembler file name for a source file is made by
917 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
919 Input files that don't require compilation are ignored.
923 Stop after the preprocessing stage; do not run the compiler proper. The
924 output is in the form of preprocessed source code, which is sent to the
927 Input files which don't require preprocessing are ignored.
929 @cindex output file option
932 Place output in file @var{file}. This applies regardless to whatever
933 sort of output is being produced, whether it be an executable file,
934 an object file, an assembler file or preprocessed C code.
936 If @option{-o} is not specified, the default is to put an executable
937 file in @file{a.out}, the object file for
938 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
939 assembler file in @file{@var{source}.s}, a precompiled header file in
940 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
945 Print (on standard error output) the commands executed to run the stages
946 of compilation. Also print the version number of the compiler driver
947 program and of the preprocessor and the compiler proper.
951 Like @option{-v} except the commands are not executed and all command
952 arguments are quoted. This is useful for shell scripts to capture the
953 driver-generated command lines.
957 Use pipes rather than temporary files for communication between the
958 various stages of compilation. This fails to work on some systems where
959 the assembler is unable to read from a pipe; but the GNU assembler has
964 If you are compiling multiple source files, this option tells the driver
965 to pass all the source files to the compiler at once (for those
966 languages for which the compiler can handle this). This will allow
967 intermodule analysis (IMA) to be performed by the compiler. Currently the only
968 language for which this is supported is C@. If you pass source files for
969 multiple languages to the driver, using this option, the driver will invoke
970 the compiler(s) that support IMA once each, passing each compiler all the
971 source files appropriate for it. For those languages that do not support
972 IMA this option will be ignored, and the compiler will be invoked once for
973 each source file in that language. If you use this option in conjunction
974 with @option{-save-temps}, the compiler will generate multiple
976 (one for each source file), but only one (combined) @file{.o} or
981 Print (on the standard output) a description of the command line options
982 understood by @command{gcc}. If the @option{-v} option is also specified
983 then @option{--help} will also be passed on to the various processes
984 invoked by @command{gcc}, so that they can display the command line options
985 they accept. If the @option{-Wextra} option is also specified then command
986 line options which have no documentation associated with them will also
991 Print (on the standard output) a description of target specific command
992 line options for each tool.
996 Display the version number and copyrights of the invoked GCC@.
1000 @section Compiling C++ Programs
1002 @cindex suffixes for C++ source
1003 @cindex C++ source file suffixes
1004 C++ source files conventionally use one of the suffixes @samp{.C},
1005 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1006 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1007 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1008 files with these names and compiles them as C++ programs even if you
1009 call the compiler the same way as for compiling C programs (usually
1010 with the name @command{gcc}).
1014 However, C++ programs often require class libraries as well as a
1015 compiler that understands the C++ language---and under some
1016 circumstances, you might want to compile programs or header files from
1017 standard input, or otherwise without a suffix that flags them as C++
1018 programs. You might also like to precompile a C header file with a
1019 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1020 program that calls GCC with the default language set to C++, and
1021 automatically specifies linking against the C++ library. On many
1022 systems, @command{g++} is also installed with the name @command{c++}.
1024 @cindex invoking @command{g++}
1025 When you compile C++ programs, you may specify many of the same
1026 command-line options that you use for compiling programs in any
1027 language; or command-line options meaningful for C and related
1028 languages; or options that are meaningful only for C++ programs.
1029 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1030 explanations of options for languages related to C@.
1031 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1032 explanations of options that are meaningful only for C++ programs.
1034 @node C Dialect Options
1035 @section Options Controlling C Dialect
1036 @cindex dialect options
1037 @cindex language dialect options
1038 @cindex options, dialect
1040 The following options control the dialect of C (or languages derived
1041 from C, such as C++, Objective-C and Objective-C++) that the compiler
1045 @cindex ANSI support
1049 In C mode, support all ISO C90 programs. In C++ mode,
1050 remove GNU extensions that conflict with ISO C++.
1052 This turns off certain features of GCC that are incompatible with ISO
1053 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1054 such as the @code{asm} and @code{typeof} keywords, and
1055 predefined macros such as @code{unix} and @code{vax} that identify the
1056 type of system you are using. It also enables the undesirable and
1057 rarely used ISO trigraph feature. For the C compiler,
1058 it disables recognition of C++ style @samp{//} comments as well as
1059 the @code{inline} keyword.
1061 The alternate keywords @code{__asm__}, @code{__extension__},
1062 @code{__inline__} and @code{__typeof__} continue to work despite
1063 @option{-ansi}. You would not want to use them in an ISO C program, of
1064 course, but it is useful to put them in header files that might be included
1065 in compilations done with @option{-ansi}. Alternate predefined macros
1066 such as @code{__unix__} and @code{__vax__} are also available, with or
1067 without @option{-ansi}.
1069 The @option{-ansi} option does not cause non-ISO programs to be
1070 rejected gratuitously. For that, @option{-pedantic} is required in
1071 addition to @option{-ansi}. @xref{Warning Options}.
1073 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1074 option is used. Some header files may notice this macro and refrain
1075 from declaring certain functions or defining certain macros that the
1076 ISO standard doesn't call for; this is to avoid interfering with any
1077 programs that might use these names for other things.
1079 Functions which would normally be built in but do not have semantics
1080 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1081 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1082 built-in functions provided by GCC}, for details of the functions
1087 Determine the language standard. This option is currently only
1088 supported when compiling C or C++. A value for this option must be
1089 provided; possible values are
1094 ISO C90 (same as @option{-ansi}).
1096 @item iso9899:199409
1097 ISO C90 as modified in amendment 1.
1103 ISO C99. Note that this standard is not yet fully supported; see
1104 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1105 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1108 Default, ISO C90 plus GNU extensions (including some C99 features).
1112 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1113 this will become the default. The name @samp{gnu9x} is deprecated.
1116 The 1998 ISO C++ standard plus amendments.
1119 The same as @option{-std=c++98} plus GNU extensions. This is the
1120 default for C++ code.
1123 Even when this option is not specified, you can still use some of the
1124 features of newer standards in so far as they do not conflict with
1125 previous C standards. For example, you may use @code{__restrict__} even
1126 when @option{-std=c99} is not specified.
1128 The @option{-std} options specifying some version of ISO C have the same
1129 effects as @option{-ansi}, except that features that were not in ISO C90
1130 but are in the specified version (for example, @samp{//} comments and
1131 the @code{inline} keyword in ISO C99) are not disabled.
1133 @xref{Standards,,Language Standards Supported by GCC}, for details of
1134 these standard versions.
1136 @item -aux-info @var{filename}
1138 Output to the given filename prototyped declarations for all functions
1139 declared and/or defined in a translation unit, including those in header
1140 files. This option is silently ignored in any language other than C@.
1142 Besides declarations, the file indicates, in comments, the origin of
1143 each declaration (source file and line), whether the declaration was
1144 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1145 @samp{O} for old, respectively, in the first character after the line
1146 number and the colon), and whether it came from a declaration or a
1147 definition (@samp{C} or @samp{F}, respectively, in the following
1148 character). In the case of function definitions, a K&R-style list of
1149 arguments followed by their declarations is also provided, inside
1150 comments, after the declaration.
1154 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1155 keyword, so that code can use these words as identifiers. You can use
1156 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1157 instead. @option{-ansi} implies @option{-fno-asm}.
1159 In C++, this switch only affects the @code{typeof} keyword, since
1160 @code{asm} and @code{inline} are standard keywords. You may want to
1161 use the @option{-fno-gnu-keywords} flag instead, which has the same
1162 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1163 switch only affects the @code{asm} and @code{typeof} keywords, since
1164 @code{inline} is a standard keyword in ISO C99.
1167 @itemx -fno-builtin-@var{function}
1168 @opindex fno-builtin
1169 @cindex built-in functions
1170 Don't recognize built-in functions that do not begin with
1171 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1172 functions provided by GCC}, for details of the functions affected,
1173 including those which are not built-in functions when @option{-ansi} or
1174 @option{-std} options for strict ISO C conformance are used because they
1175 do not have an ISO standard meaning.
1177 GCC normally generates special code to handle certain built-in functions
1178 more efficiently; for instance, calls to @code{alloca} may become single
1179 instructions that adjust the stack directly, and calls to @code{memcpy}
1180 may become inline copy loops. The resulting code is often both smaller
1181 and faster, but since the function calls no longer appear as such, you
1182 cannot set a breakpoint on those calls, nor can you change the behavior
1183 of the functions by linking with a different library. In addition,
1184 when a function is recognized as a built-in function, GCC may use
1185 information about that function to warn about problems with calls to
1186 that function, or to generate more efficient code, even if the
1187 resulting code still contains calls to that function. For example,
1188 warnings are given with @option{-Wformat} for bad calls to
1189 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1190 known not to modify global memory.
1192 With the @option{-fno-builtin-@var{function}} option
1193 only the built-in function @var{function} is
1194 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1195 function is named this is not built-in in this version of GCC, this
1196 option is ignored. There is no corresponding
1197 @option{-fbuiltin-@var{function}} option; if you wish to enable
1198 built-in functions selectively when using @option{-fno-builtin} or
1199 @option{-ffreestanding}, you may define macros such as:
1202 #define abs(n) __builtin_abs ((n))
1203 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1208 @cindex hosted environment
1210 Assert that compilation takes place in a hosted environment. This implies
1211 @option{-fbuiltin}. A hosted environment is one in which the
1212 entire standard library is available, and in which @code{main} has a return
1213 type of @code{int}. Examples are nearly everything except a kernel.
1214 This is equivalent to @option{-fno-freestanding}.
1216 @item -ffreestanding
1217 @opindex ffreestanding
1218 @cindex hosted environment
1220 Assert that compilation takes place in a freestanding environment. This
1221 implies @option{-fno-builtin}. A freestanding environment
1222 is one in which the standard library may not exist, and program startup may
1223 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1224 This is equivalent to @option{-fno-hosted}.
1226 @xref{Standards,,Language Standards Supported by GCC}, for details of
1227 freestanding and hosted environments.
1229 @item -fms-extensions
1230 @opindex fms-extensions
1231 Accept some non-standard constructs used in Microsoft header files.
1233 Some cases of unnamed fields in structures and unions are only
1234 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1235 fields within structs/unions}, for details.
1239 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1240 options for strict ISO C conformance) implies @option{-trigraphs}.
1242 @item -no-integrated-cpp
1243 @opindex no-integrated-cpp
1244 Performs a compilation in two passes: preprocessing and compiling. This
1245 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1246 @option{-B} option. The user supplied compilation step can then add in
1247 an additional preprocessing step after normal preprocessing but before
1248 compiling. The default is to use the integrated cpp (internal cpp)
1250 The semantics of this option will change if "cc1", "cc1plus", and
1251 "cc1obj" are merged.
1253 @cindex traditional C language
1254 @cindex C language, traditional
1256 @itemx -traditional-cpp
1257 @opindex traditional-cpp
1258 @opindex traditional
1259 Formerly, these options caused GCC to attempt to emulate a pre-standard
1260 C compiler. They are now only supported with the @option{-E} switch.
1261 The preprocessor continues to support a pre-standard mode. See the GNU
1262 CPP manual for details.
1264 @item -fcond-mismatch
1265 @opindex fcond-mismatch
1266 Allow conditional expressions with mismatched types in the second and
1267 third arguments. The value of such an expression is void. This option
1268 is not supported for C++.
1270 @item -funsigned-char
1271 @opindex funsigned-char
1272 Let the type @code{char} be unsigned, like @code{unsigned char}.
1274 Each kind of machine has a default for what @code{char} should
1275 be. It is either like @code{unsigned char} by default or like
1276 @code{signed char} by default.
1278 Ideally, a portable program should always use @code{signed char} or
1279 @code{unsigned char} when it depends on the signedness of an object.
1280 But many programs have been written to use plain @code{char} and
1281 expect it to be signed, or expect it to be unsigned, depending on the
1282 machines they were written for. This option, and its inverse, let you
1283 make such a program work with the opposite default.
1285 The type @code{char} is always a distinct type from each of
1286 @code{signed char} or @code{unsigned char}, even though its behavior
1287 is always just like one of those two.
1290 @opindex fsigned-char
1291 Let the type @code{char} be signed, like @code{signed char}.
1293 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1294 the negative form of @option{-funsigned-char}. Likewise, the option
1295 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1297 @item -fsigned-bitfields
1298 @itemx -funsigned-bitfields
1299 @itemx -fno-signed-bitfields
1300 @itemx -fno-unsigned-bitfields
1301 @opindex fsigned-bitfields
1302 @opindex funsigned-bitfields
1303 @opindex fno-signed-bitfields
1304 @opindex fno-unsigned-bitfields
1305 These options control whether a bit-field is signed or unsigned, when the
1306 declaration does not use either @code{signed} or @code{unsigned}. By
1307 default, such a bit-field is signed, because this is consistent: the
1308 basic integer types such as @code{int} are signed types.
1311 @node C++ Dialect Options
1312 @section Options Controlling C++ Dialect
1314 @cindex compiler options, C++
1315 @cindex C++ options, command line
1316 @cindex options, C++
1317 This section describes the command-line options that are only meaningful
1318 for C++ programs; but you can also use most of the GNU compiler options
1319 regardless of what language your program is in. For example, you
1320 might compile a file @code{firstClass.C} like this:
1323 g++ -g -frepo -O -c firstClass.C
1327 In this example, only @option{-frepo} is an option meant
1328 only for C++ programs; you can use the other options with any
1329 language supported by GCC@.
1331 Here is a list of options that are @emph{only} for compiling C++ programs:
1335 @item -fabi-version=@var{n}
1336 @opindex fabi-version
1337 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1338 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1339 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1340 the version that conforms most closely to the C++ ABI specification.
1341 Therefore, the ABI obtained using version 0 will change as ABI bugs
1344 The default is version 2.
1346 @item -fno-access-control
1347 @opindex fno-access-control
1348 Turn off all access checking. This switch is mainly useful for working
1349 around bugs in the access control code.
1353 Check that the pointer returned by @code{operator new} is non-null
1354 before attempting to modify the storage allocated. This check is
1355 normally unnecessary because the C++ standard specifies that
1356 @code{operator new} will only return @code{0} if it is declared
1357 @samp{throw()}, in which case the compiler will always check the
1358 return value even without this option. In all other cases, when
1359 @code{operator new} has a non-empty exception specification, memory
1360 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1361 @samp{new (nothrow)}.
1363 @item -fconserve-space
1364 @opindex fconserve-space
1365 Put uninitialized or runtime-initialized global variables into the
1366 common segment, as C does. This saves space in the executable at the
1367 cost of not diagnosing duplicate definitions. If you compile with this
1368 flag and your program mysteriously crashes after @code{main()} has
1369 completed, you may have an object that is being destroyed twice because
1370 two definitions were merged.
1372 This option is no longer useful on most targets, now that support has
1373 been added for putting variables into BSS without making them common.
1375 @item -fno-const-strings
1376 @opindex fno-const-strings
1377 Give string constants type @code{char *} instead of type @code{const
1378 char *}. By default, G++ uses type @code{const char *} as required by
1379 the standard. Even if you use @option{-fno-const-strings}, you cannot
1380 actually modify the value of a string constant.
1382 This option might be removed in a future release of G++. For maximum
1383 portability, you should structure your code so that it works with
1384 string constants that have type @code{const char *}.
1386 @item -fno-elide-constructors
1387 @opindex fno-elide-constructors
1388 The C++ standard allows an implementation to omit creating a temporary
1389 which is only used to initialize another object of the same type.
1390 Specifying this option disables that optimization, and forces G++ to
1391 call the copy constructor in all cases.
1393 @item -fno-enforce-eh-specs
1394 @opindex fno-enforce-eh-specs
1395 Don't check for violation of exception specifications at runtime. This
1396 option violates the C++ standard, but may be useful for reducing code
1397 size in production builds, much like defining @samp{NDEBUG}. The compiler
1398 will still optimize based on the exception specifications.
1401 @itemx -fno-for-scope
1403 @opindex fno-for-scope
1404 If @option{-ffor-scope} is specified, the scope of variables declared in
1405 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1406 as specified by the C++ standard.
1407 If @option{-fno-for-scope} is specified, the scope of variables declared in
1408 a @i{for-init-statement} extends to the end of the enclosing scope,
1409 as was the case in old versions of G++, and other (traditional)
1410 implementations of C++.
1412 The default if neither flag is given to follow the standard,
1413 but to allow and give a warning for old-style code that would
1414 otherwise be invalid, or have different behavior.
1416 @item -fno-gnu-keywords
1417 @opindex fno-gnu-keywords
1418 Do not recognize @code{typeof} as a keyword, so that code can use this
1419 word as an identifier. You can use the keyword @code{__typeof__} instead.
1420 @option{-ansi} implies @option{-fno-gnu-keywords}.
1422 @item -fno-implicit-templates
1423 @opindex fno-implicit-templates
1424 Never emit code for non-inline templates which are instantiated
1425 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1426 @xref{Template Instantiation}, for more information.
1428 @item -fno-implicit-inline-templates
1429 @opindex fno-implicit-inline-templates
1430 Don't emit code for implicit instantiations of inline templates, either.
1431 The default is to handle inlines differently so that compiles with and
1432 without optimization will need the same set of explicit instantiations.
1434 @item -fno-implement-inlines
1435 @opindex fno-implement-inlines
1436 To save space, do not emit out-of-line copies of inline functions
1437 controlled by @samp{#pragma implementation}. This will cause linker
1438 errors if these functions are not inlined everywhere they are called.
1440 @item -fms-extensions
1441 @opindex fms-extensions
1442 Disable pedantic warnings about constructs used in MFC, such as implicit
1443 int and getting a pointer to member function via non-standard syntax.
1445 @item -fno-nonansi-builtins
1446 @opindex fno-nonansi-builtins
1447 Disable built-in declarations of functions that are not mandated by
1448 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1449 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1451 @item -fno-operator-names
1452 @opindex fno-operator-names
1453 Do not treat the operator name keywords @code{and}, @code{bitand},
1454 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1455 synonyms as keywords.
1457 @item -fno-optional-diags
1458 @opindex fno-optional-diags
1459 Disable diagnostics that the standard says a compiler does not need to
1460 issue. Currently, the only such diagnostic issued by G++ is the one for
1461 a name having multiple meanings within a class.
1464 @opindex fpermissive
1465 Downgrade some diagnostics about nonconformant code from errors to
1466 warnings. Thus, using @option{-fpermissive} will allow some
1467 nonconforming code to compile.
1471 Enable automatic template instantiation at link time. This option also
1472 implies @option{-fno-implicit-templates}. @xref{Template
1473 Instantiation}, for more information.
1477 Disable generation of information about every class with virtual
1478 functions for use by the C++ runtime type identification features
1479 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1480 of the language, you can save some space by using this flag. Note that
1481 exception handling uses the same information, but it will generate it as
1486 Emit statistics about front-end processing at the end of the compilation.
1487 This information is generally only useful to the G++ development team.
1489 @item -ftemplate-depth-@var{n}
1490 @opindex ftemplate-depth
1491 Set the maximum instantiation depth for template classes to @var{n}.
1492 A limit on the template instantiation depth is needed to detect
1493 endless recursions during template class instantiation. ANSI/ISO C++
1494 conforming programs must not rely on a maximum depth greater than 17.
1496 @item -fno-threadsafe-statics
1497 @opindex fno-threadsafe-statics
1498 Do not emit the extra code to use the routines specified in the C++
1499 ABI for thread-safe initialization of local statics. You can use this
1500 option to reduce code size slightly in code that doesn't need to be
1503 @item -fuse-cxa-atexit
1504 @opindex fuse-cxa-atexit
1505 Register destructors for objects with static storage duration with the
1506 @code{__cxa_atexit} function rather than the @code{atexit} function.
1507 This option is required for fully standards-compliant handling of static
1508 destructors, but will only work if your C library supports
1509 @code{__cxa_atexit}.
1511 @item -fvisibility-inlines-hidden
1512 @opindex fvisibility-inlines-hidden
1513 Causes all inlined methods to be marked with
1514 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1515 appear in the export table of a DSO and do not require a PLT indirection
1516 when used within the DSO@. Enabling this option can have a dramatic effect
1517 on load and link times of a DSO as it massively reduces the size of the
1518 dynamic export table when the library makes heavy use of templates. While
1519 it can cause bloating through duplication of code within each DSO where
1520 it is used, often the wastage is less than the considerable space occupied
1521 by a long symbol name in the export table which is typical when using
1522 templates and namespaces. For even more savings, combine with the
1523 @option{-fvisibility=hidden} switch.
1527 Do not use weak symbol support, even if it is provided by the linker.
1528 By default, G++ will use weak symbols if they are available. This
1529 option exists only for testing, and should not be used by end-users;
1530 it will result in inferior code and has no benefits. This option may
1531 be removed in a future release of G++.
1535 Do not search for header files in the standard directories specific to
1536 C++, but do still search the other standard directories. (This option
1537 is used when building the C++ library.)
1540 In addition, these optimization, warning, and code generation options
1541 have meanings only for C++ programs:
1544 @item -fno-default-inline
1545 @opindex fno-default-inline
1546 Do not assume @samp{inline} for functions defined inside a class scope.
1547 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1548 functions will have linkage like inline functions; they just won't be
1551 @item -Wabi @r{(C++ only)}
1553 Warn when G++ generates code that is probably not compatible with the
1554 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1555 all such cases, there are probably some cases that are not warned about,
1556 even though G++ is generating incompatible code. There may also be
1557 cases where warnings are emitted even though the code that is generated
1560 You should rewrite your code to avoid these warnings if you are
1561 concerned about the fact that code generated by G++ may not be binary
1562 compatible with code generated by other compilers.
1564 The known incompatibilities at this point include:
1569 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1570 pack data into the same byte as a base class. For example:
1573 struct A @{ virtual void f(); int f1 : 1; @};
1574 struct B : public A @{ int f2 : 1; @};
1578 In this case, G++ will place @code{B::f2} into the same byte
1579 as@code{A::f1}; other compilers will not. You can avoid this problem
1580 by explicitly padding @code{A} so that its size is a multiple of the
1581 byte size on your platform; that will cause G++ and other compilers to
1582 layout @code{B} identically.
1585 Incorrect handling of tail-padding for virtual bases. G++ does not use
1586 tail padding when laying out virtual bases. For example:
1589 struct A @{ virtual void f(); char c1; @};
1590 struct B @{ B(); char c2; @};
1591 struct C : public A, public virtual B @{@};
1595 In this case, G++ will not place @code{B} into the tail-padding for
1596 @code{A}; other compilers will. You can avoid this problem by
1597 explicitly padding @code{A} so that its size is a multiple of its
1598 alignment (ignoring virtual base classes); that will cause G++ and other
1599 compilers to layout @code{C} identically.
1602 Incorrect handling of bit-fields with declared widths greater than that
1603 of their underlying types, when the bit-fields appear in a union. For
1607 union U @{ int i : 4096; @};
1611 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1612 union too small by the number of bits in an @code{int}.
1615 Empty classes can be placed at incorrect offsets. For example:
1625 struct C : public B, public A @{@};
1629 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1630 it should be placed at offset zero. G++ mistakenly believes that the
1631 @code{A} data member of @code{B} is already at offset zero.
1634 Names of template functions whose types involve @code{typename} or
1635 template template parameters can be mangled incorrectly.
1638 template <typename Q>
1639 void f(typename Q::X) @{@}
1641 template <template <typename> class Q>
1642 void f(typename Q<int>::X) @{@}
1646 Instantiations of these templates may be mangled incorrectly.
1650 @item -Wctor-dtor-privacy @r{(C++ only)}
1651 @opindex Wctor-dtor-privacy
1652 Warn when a class seems unusable because all the constructors or
1653 destructors in that class are private, and it has neither friends nor
1654 public static member functions.
1656 @item -Wnon-virtual-dtor @r{(C++ only)}
1657 @opindex Wnon-virtual-dtor
1658 Warn when a class appears to be polymorphic, thereby requiring a virtual
1659 destructor, yet it declares a non-virtual one.
1660 This warning is enabled by @option{-Wall}.
1662 @item -Wreorder @r{(C++ only)}
1664 @cindex reordering, warning
1665 @cindex warning for reordering of member initializers
1666 Warn when the order of member initializers given in the code does not
1667 match the order in which they must be executed. For instance:
1673 A(): j (0), i (1) @{ @}
1677 The compiler will rearrange the member initializers for @samp{i}
1678 and @samp{j} to match the declaration order of the members, emitting
1679 a warning to that effect. This warning is enabled by @option{-Wall}.
1682 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1685 @item -Weffc++ @r{(C++ only)}
1687 Warn about violations of the following style guidelines from Scott Meyers'
1688 @cite{Effective C++} book:
1692 Item 11: Define a copy constructor and an assignment operator for classes
1693 with dynamically allocated memory.
1696 Item 12: Prefer initialization to assignment in constructors.
1699 Item 14: Make destructors virtual in base classes.
1702 Item 15: Have @code{operator=} return a reference to @code{*this}.
1705 Item 23: Don't try to return a reference when you must return an object.
1709 Also warn about violations of the following style guidelines from
1710 Scott Meyers' @cite{More Effective C++} book:
1714 Item 6: Distinguish between prefix and postfix forms of increment and
1715 decrement operators.
1718 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1722 When selecting this option, be aware that the standard library
1723 headers do not obey all of these guidelines; use @samp{grep -v}
1724 to filter out those warnings.
1726 @item -Wno-deprecated @r{(C++ only)}
1727 @opindex Wno-deprecated
1728 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1730 @item -Wno-non-template-friend @r{(C++ only)}
1731 @opindex Wno-non-template-friend
1732 Disable warnings when non-templatized friend functions are declared
1733 within a template. Since the advent of explicit template specification
1734 support in G++, if the name of the friend is an unqualified-id (i.e.,
1735 @samp{friend foo(int)}), the C++ language specification demands that the
1736 friend declare or define an ordinary, nontemplate function. (Section
1737 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1738 could be interpreted as a particular specialization of a templatized
1739 function. Because this non-conforming behavior is no longer the default
1740 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1741 check existing code for potential trouble spots and is on by default.
1742 This new compiler behavior can be turned off with
1743 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1744 but disables the helpful warning.
1746 @item -Wold-style-cast @r{(C++ only)}
1747 @opindex Wold-style-cast
1748 Warn if an old-style (C-style) cast to a non-void type is used within
1749 a C++ program. The new-style casts (@samp{static_cast},
1750 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1751 unintended effects and much easier to search for.
1753 @item -Woverloaded-virtual @r{(C++ only)}
1754 @opindex Woverloaded-virtual
1755 @cindex overloaded virtual fn, warning
1756 @cindex warning for overloaded virtual fn
1757 Warn when a function declaration hides virtual functions from a
1758 base class. For example, in:
1765 struct B: public A @{
1770 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1778 will fail to compile.
1780 @item -Wno-pmf-conversions @r{(C++ only)}
1781 @opindex Wno-pmf-conversions
1782 Disable the diagnostic for converting a bound pointer to member function
1785 @item -Wsign-promo @r{(C++ only)}
1786 @opindex Wsign-promo
1787 Warn when overload resolution chooses a promotion from unsigned or
1788 enumerated type to a signed type, over a conversion to an unsigned type of
1789 the same size. Previous versions of G++ would try to preserve
1790 unsignedness, but the standard mandates the current behavior.
1792 @item -Wsynth @r{(C++ only)}
1794 @cindex warning for synthesized methods
1795 @cindex synthesized methods, warning
1796 Warn when G++'s synthesis behavior does not match that of cfront. For
1802 A& operator = (int);
1812 In this example, G++ will synthesize a default @samp{A& operator =
1813 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1816 @node Objective-C and Objective-C++ Dialect Options
1817 @section Options Controlling Objective-C and Objective-C++ Dialects
1819 @cindex compiler options, Objective-C and Objective-C++
1820 @cindex Objective-C and Objective-C++ options, command line
1821 @cindex options, Objective-C and Objective-C++
1822 (NOTE: This manual does not describe the Objective-C and Objective-C++
1823 languages themselves. See @xref{Standards,,Language Standards
1824 Supported by GCC}, for references.)
1826 This section describes the command-line options that are only meaningful
1827 for Objective-C and Objective-C++ programs, but you can also use most of
1828 the language-independent GNU compiler options.
1829 For example, you might compile a file @code{some_class.m} like this:
1832 gcc -g -fgnu-runtime -O -c some_class.m
1836 In this example, @option{-fgnu-runtime} is an option meant only for
1837 Objective-C and Objective-C++ programs; you can use the other options with
1838 any language supported by GCC@.
1840 Note that since Objective-C is an extension of the C language, Objective-C
1841 compilations may also use options specific to the C front-end (e.g.,
1842 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1843 C++-specific options (e.g., @option{-Wabi}).
1845 Here is a list of options that are @emph{only} for compiling Objective-C
1846 and Objective-C++ programs:
1849 @item -fconstant-string-class=@var{class-name}
1850 @opindex fconstant-string-class
1851 Use @var{class-name} as the name of the class to instantiate for each
1852 literal string specified with the syntax @code{@@"@dots{}"}. The default
1853 class name is @code{NXConstantString} if the GNU runtime is being used, and
1854 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1855 @option{-fconstant-cfstrings} option, if also present, will override the
1856 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1857 to be laid out as constant CoreFoundation strings.
1860 @opindex fgnu-runtime
1861 Generate object code compatible with the standard GNU Objective-C
1862 runtime. This is the default for most types of systems.
1864 @item -fnext-runtime
1865 @opindex fnext-runtime
1866 Generate output compatible with the NeXT runtime. This is the default
1867 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1868 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1871 @item -fno-nil-receivers
1872 @opindex fno-nil-receivers
1873 Assume that all Objective-C message dispatches (e.g.,
1874 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1875 is not @code{nil}. This allows for more efficient entry points in the runtime
1876 to be used. Currently, this option is only available in conjunction with
1877 the NeXT runtime on Mac OS X 10.3 and later.
1879 @item -fobjc-exceptions
1880 @opindex fobjc-exceptions
1881 Enable syntactic support for structured exception handling in Objective-C,
1882 similar to what is offered by C++ and Java. Currently, this option is only
1883 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1891 @@catch (AnObjCClass *exc) @{
1898 @@catch (AnotherClass *exc) @{
1901 @@catch (id allOthers) @{
1911 The @code{@@throw} statement may appear anywhere in an Objective-C or
1912 Objective-C++ program; when used inside of a @code{@@catch} block, the
1913 @code{@@throw} may appear without an argument (as shown above), in which case
1914 the object caught by the @code{@@catch} will be rethrown.
1916 Note that only (pointers to) Objective-C objects may be thrown and
1917 caught using this scheme. When an object is thrown, it will be caught
1918 by the nearest @code{@@catch} clause capable of handling objects of that type,
1919 analogously to how @code{catch} blocks work in C++ and Java. A
1920 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1921 any and all Objective-C exceptions not caught by previous @code{@@catch}
1924 The @code{@@finally} clause, if present, will be executed upon exit from the
1925 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1926 regardless of whether any exceptions are thrown, caught or rethrown
1927 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1928 of the @code{finally} clause in Java.
1930 There are several caveats to using the new exception mechanism:
1934 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1935 idioms provided by the @code{NSException} class, the new
1936 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1937 systems, due to additional functionality needed in the (NeXT) Objective-C
1941 As mentioned above, the new exceptions do not support handling
1942 types other than Objective-C objects. Furthermore, when used from
1943 Objective-C++, the Objective-C exception model does not interoperate with C++
1944 exceptions at this time. This means you cannot @code{@@throw} an exception
1945 from Objective-C and @code{catch} it in C++, or vice versa
1946 (i.e., @code{throw @dots{} @@catch}).
1949 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1950 blocks for thread-safe execution:
1953 @@synchronized (ObjCClass *guard) @{
1958 Upon entering the @code{@@synchronized} block, a thread of execution shall
1959 first check whether a lock has been placed on the corresponding @code{guard}
1960 object by another thread. If it has, the current thread shall wait until
1961 the other thread relinquishes its lock. Once @code{guard} becomes available,
1962 the current thread will place its own lock on it, execute the code contained in
1963 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1964 making @code{guard} available to other threads).
1966 Unlike Java, Objective-C does not allow for entire methods to be marked
1967 @code{@@synchronized}. Note that throwing exceptions out of
1968 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1969 to be unlocked properly.
1971 @item -freplace-objc-classes
1972 @opindex freplace-objc-classes
1973 Emit a special marker instructing @command{ld(1)} not to statically link in
1974 the resulting object file, and allow @command{dyld(1)} to load it in at
1975 run time instead. This is used in conjunction with the Fix-and-Continue
1976 debugging mode, where the object file in question may be recompiled and
1977 dynamically reloaded in the course of program execution, without the need
1978 to restart the program itself. Currently, Fix-and-Continue functionality
1979 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1984 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1985 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1986 compile time) with static class references that get initialized at load time,
1987 which improves run-time performance. Specifying the @option{-fzero-link} flag
1988 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1989 to be retained. This is useful in Zero-Link debugging mode, since it allows
1990 for individual class implementations to be modified during program execution.
1994 Dump interface declarations for all classes seen in the source file to a
1995 file named @file{@var{sourcename}.decl}.
1998 @opindex Wno-protocol
1999 If a class is declared to implement a protocol, a warning is issued for
2000 every method in the protocol that is not implemented by the class. The
2001 default behavior is to issue a warning for every method not explicitly
2002 implemented in the class, even if a method implementation is inherited
2003 from the superclass. If you use the @option{-Wno-protocol} option, then
2004 methods inherited from the superclass are considered to be implemented,
2005 and no warning is issued for them.
2009 Warn if multiple methods of different types for the same selector are
2010 found during compilation. The check is performed on the list of methods
2011 in the final stage of compilation. Additionally, a check is performed
2012 for each selector appearing in a @code{@@selector(@dots{})}
2013 expression, and a corresponding method for that selector has been found
2014 during compilation. Because these checks scan the method table only at
2015 the end of compilation, these warnings are not produced if the final
2016 stage of compilation is not reached, for example because an error is
2017 found during compilation, or because the @option{-fsyntax-only} option is
2020 @item -Wundeclared-selector
2021 @opindex Wundeclared-selector
2022 Warn if a @code{@@selector(@dots{})} expression referring to an
2023 undeclared selector is found. A selector is considered undeclared if no
2024 method with that name has been declared before the
2025 @code{@@selector(@dots{})} expression, either explicitly in an
2026 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2027 an @code{@@implementation} section. This option always performs its
2028 checks as soon as a @code{@@selector(@dots{})} expression is found,
2029 while @option{-Wselector} only performs its checks in the final stage of
2030 compilation. This also enforces the coding style convention
2031 that methods and selectors must be declared before being used.
2033 @item -print-objc-runtime-info
2034 @opindex print-objc-runtime-info
2035 Generate C header describing the largest structure that is passed by
2040 @node Language Independent Options
2041 @section Options to Control Diagnostic Messages Formatting
2042 @cindex options to control diagnostics formatting
2043 @cindex diagnostic messages
2044 @cindex message formatting
2046 Traditionally, diagnostic messages have been formatted irrespective of
2047 the output device's aspect (e.g.@: its width, @dots{}). The options described
2048 below can be used to control the diagnostic messages formatting
2049 algorithm, e.g.@: how many characters per line, how often source location
2050 information should be reported. Right now, only the C++ front end can
2051 honor these options. However it is expected, in the near future, that
2052 the remaining front ends would be able to digest them correctly.
2055 @item -fmessage-length=@var{n}
2056 @opindex fmessage-length
2057 Try to format error messages so that they fit on lines of about @var{n}
2058 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2059 the front ends supported by GCC@. If @var{n} is zero, then no
2060 line-wrapping will be done; each error message will appear on a single
2063 @opindex fdiagnostics-show-location
2064 @item -fdiagnostics-show-location=once
2065 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2066 reporter to emit @emph{once} source location information; that is, in
2067 case the message is too long to fit on a single physical line and has to
2068 be wrapped, the source location won't be emitted (as prefix) again,
2069 over and over, in subsequent continuation lines. This is the default
2072 @item -fdiagnostics-show-location=every-line
2073 Only meaningful in line-wrapping mode. Instructs the diagnostic
2074 messages reporter to emit the same source location information (as
2075 prefix) for physical lines that result from the process of breaking
2076 a message which is too long to fit on a single line.
2080 @node Warning Options
2081 @section Options to Request or Suppress Warnings
2082 @cindex options to control warnings
2083 @cindex warning messages
2084 @cindex messages, warning
2085 @cindex suppressing warnings
2087 Warnings are diagnostic messages that report constructions which
2088 are not inherently erroneous but which are risky or suggest there
2089 may have been an error.
2091 You can request many specific warnings with options beginning @samp{-W},
2092 for example @option{-Wimplicit} to request warnings on implicit
2093 declarations. Each of these specific warning options also has a
2094 negative form beginning @samp{-Wno-} to turn off warnings;
2095 for example, @option{-Wno-implicit}. This manual lists only one of the
2096 two forms, whichever is not the default.
2098 The following options control the amount and kinds of warnings produced
2099 by GCC; for further, language-specific options also refer to
2100 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2104 @cindex syntax checking
2106 @opindex fsyntax-only
2107 Check the code for syntax errors, but don't do anything beyond that.
2111 Issue all the warnings demanded by strict ISO C and ISO C++;
2112 reject all programs that use forbidden extensions, and some other
2113 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2114 version of the ISO C standard specified by any @option{-std} option used.
2116 Valid ISO C and ISO C++ programs should compile properly with or without
2117 this option (though a rare few will require @option{-ansi} or a
2118 @option{-std} option specifying the required version of ISO C)@. However,
2119 without this option, certain GNU extensions and traditional C and C++
2120 features are supported as well. With this option, they are rejected.
2122 @option{-pedantic} does not cause warning messages for use of the
2123 alternate keywords whose names begin and end with @samp{__}. Pedantic
2124 warnings are also disabled in the expression that follows
2125 @code{__extension__}. However, only system header files should use
2126 these escape routes; application programs should avoid them.
2127 @xref{Alternate Keywords}.
2129 Some users try to use @option{-pedantic} to check programs for strict ISO
2130 C conformance. They soon find that it does not do quite what they want:
2131 it finds some non-ISO practices, but not all---only those for which
2132 ISO C @emph{requires} a diagnostic, and some others for which
2133 diagnostics have been added.
2135 A feature to report any failure to conform to ISO C might be useful in
2136 some instances, but would require considerable additional work and would
2137 be quite different from @option{-pedantic}. We don't have plans to
2138 support such a feature in the near future.
2140 Where the standard specified with @option{-std} represents a GNU
2141 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2142 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2143 extended dialect is based. Warnings from @option{-pedantic} are given
2144 where they are required by the base standard. (It would not make sense
2145 for such warnings to be given only for features not in the specified GNU
2146 C dialect, since by definition the GNU dialects of C include all
2147 features the compiler supports with the given option, and there would be
2148 nothing to warn about.)
2150 @item -pedantic-errors
2151 @opindex pedantic-errors
2152 Like @option{-pedantic}, except that errors are produced rather than
2157 Inhibit all warning messages.
2161 Inhibit warning messages about the use of @samp{#import}.
2163 @item -Wchar-subscripts
2164 @opindex Wchar-subscripts
2165 Warn if an array subscript has type @code{char}. This is a common cause
2166 of error, as programmers often forget that this type is signed on some
2171 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2172 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2174 @item -Wfatal-errors
2175 @opindex Wfatal-errors
2176 This option causes the compiler to abort compilation on the first error
2177 occurred rather than trying to keep going and printing further error
2182 @opindex ffreestanding
2183 @opindex fno-builtin
2184 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2185 the arguments supplied have types appropriate to the format string
2186 specified, and that the conversions specified in the format string make
2187 sense. This includes standard functions, and others specified by format
2188 attributes (@pxref{Function Attributes}), in the @code{printf},
2189 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2190 not in the C standard) families (or other target-specific families).
2191 Which functions are checked without format attributes having been
2192 specified depends on the standard version selected, and such checks of
2193 functions without the attribute specified are disabled by
2194 @option{-ffreestanding} or @option{-fno-builtin}.
2196 The formats are checked against the format features supported by GNU
2197 libc version 2.2. These include all ISO C90 and C99 features, as well
2198 as features from the Single Unix Specification and some BSD and GNU
2199 extensions. Other library implementations may not support all these
2200 features; GCC does not support warning about features that go beyond a
2201 particular library's limitations. However, if @option{-pedantic} is used
2202 with @option{-Wformat}, warnings will be given about format features not
2203 in the selected standard version (but not for @code{strfmon} formats,
2204 since those are not in any version of the C standard). @xref{C Dialect
2205 Options,,Options Controlling C Dialect}.
2207 Since @option{-Wformat} also checks for null format arguments for
2208 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2210 @option{-Wformat} is included in @option{-Wall}. For more control over some
2211 aspects of format checking, the options @option{-Wformat-y2k},
2212 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2213 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2214 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2217 @opindex Wformat-y2k
2218 If @option{-Wformat} is specified, also warn about @code{strftime}
2219 formats which may yield only a two-digit year.
2221 @item -Wno-format-extra-args
2222 @opindex Wno-format-extra-args
2223 If @option{-Wformat} is specified, do not warn about excess arguments to a
2224 @code{printf} or @code{scanf} format function. The C standard specifies
2225 that such arguments are ignored.
2227 Where the unused arguments lie between used arguments that are
2228 specified with @samp{$} operand number specifications, normally
2229 warnings are still given, since the implementation could not know what
2230 type to pass to @code{va_arg} to skip the unused arguments. However,
2231 in the case of @code{scanf} formats, this option will suppress the
2232 warning if the unused arguments are all pointers, since the Single
2233 Unix Specification says that such unused arguments are allowed.
2235 @item -Wno-format-zero-length
2236 @opindex Wno-format-zero-length
2237 If @option{-Wformat} is specified, do not warn about zero-length formats.
2238 The C standard specifies that zero-length formats are allowed.
2240 @item -Wformat-nonliteral
2241 @opindex Wformat-nonliteral
2242 If @option{-Wformat} is specified, also warn if the format string is not a
2243 string literal and so cannot be checked, unless the format function
2244 takes its format arguments as a @code{va_list}.
2246 @item -Wformat-security
2247 @opindex Wformat-security
2248 If @option{-Wformat} is specified, also warn about uses of format
2249 functions that represent possible security problems. At present, this
2250 warns about calls to @code{printf} and @code{scanf} functions where the
2251 format string is not a string literal and there are no format arguments,
2252 as in @code{printf (foo);}. This may be a security hole if the format
2253 string came from untrusted input and contains @samp{%n}. (This is
2254 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2255 in future warnings may be added to @option{-Wformat-security} that are not
2256 included in @option{-Wformat-nonliteral}.)
2260 Enable @option{-Wformat} plus format checks not included in
2261 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2262 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2266 Warn about passing a null pointer for arguments marked as
2267 requiring a non-null value by the @code{nonnull} function attribute.
2269 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2270 can be disabled with the @option{-Wno-nonnull} option.
2272 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2274 Warn about uninitialized variables which are initialized with themselves.
2275 Note this option can only be used with the @option{-Wuninitialized} option,
2276 which in turn only works with @option{-O1} and above.
2278 For example, GCC will warn about @code{i} being uninitialized in the
2279 following snippet only when @option{-Winit-self} has been specified:
2290 @item -Wimplicit-int
2291 @opindex Wimplicit-int
2292 Warn when a declaration does not specify a type.
2294 @item -Wimplicit-function-declaration
2295 @itemx -Werror-implicit-function-declaration
2296 @opindex Wimplicit-function-declaration
2297 @opindex Werror-implicit-function-declaration
2298 Give a warning (or error) whenever a function is used before being
2299 declared. The form @option{-Wno-error-implicit-function-declaration}
2304 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2308 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2309 function with external linkage, returning int, taking either zero
2310 arguments, two, or three arguments of appropriate types.
2312 @item -Wmissing-braces
2313 @opindex Wmissing-braces
2314 Warn if an aggregate or union initializer is not fully bracketed. In
2315 the following example, the initializer for @samp{a} is not fully
2316 bracketed, but that for @samp{b} is fully bracketed.
2319 int a[2][2] = @{ 0, 1, 2, 3 @};
2320 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2323 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2324 @opindex Wmissing-include-dirs
2325 Warn if a user-supplied include directory does not exist.
2328 @opindex Wparentheses
2329 Warn if parentheses are omitted in certain contexts, such
2330 as when there is an assignment in a context where a truth value
2331 is expected, or when operators are nested whose precedence people
2332 often get confused about. Only the warning for an assignment used as
2333 a truth value is supported when compiling C++; the other warnings are
2334 only supported when compiling C@.
2336 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2337 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2338 interpretation from that of ordinary mathematical notation.
2340 Also warn about constructions where there may be confusion to which
2341 @code{if} statement an @code{else} branch belongs. Here is an example of
2356 In C, every @code{else} branch belongs to the innermost possible @code{if}
2357 statement, which in this example is @code{if (b)}. This is often not
2358 what the programmer expected, as illustrated in the above example by
2359 indentation the programmer chose. When there is the potential for this
2360 confusion, GCC will issue a warning when this flag is specified.
2361 To eliminate the warning, add explicit braces around the innermost
2362 @code{if} statement so there is no way the @code{else} could belong to
2363 the enclosing @code{if}. The resulting code would look like this:
2379 @item -Wsequence-point
2380 @opindex Wsequence-point
2381 Warn about code that may have undefined semantics because of violations
2382 of sequence point rules in the C standard.
2384 The C standard defines the order in which expressions in a C program are
2385 evaluated in terms of @dfn{sequence points}, which represent a partial
2386 ordering between the execution of parts of the program: those executed
2387 before the sequence point, and those executed after it. These occur
2388 after the evaluation of a full expression (one which is not part of a
2389 larger expression), after the evaluation of the first operand of a
2390 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2391 function is called (but after the evaluation of its arguments and the
2392 expression denoting the called function), and in certain other places.
2393 Other than as expressed by the sequence point rules, the order of
2394 evaluation of subexpressions of an expression is not specified. All
2395 these rules describe only a partial order rather than a total order,
2396 since, for example, if two functions are called within one expression
2397 with no sequence point between them, the order in which the functions
2398 are called is not specified. However, the standards committee have
2399 ruled that function calls do not overlap.
2401 It is not specified when between sequence points modifications to the
2402 values of objects take effect. Programs whose behavior depends on this
2403 have undefined behavior; the C standard specifies that ``Between the
2404 previous and next sequence point an object shall have its stored value
2405 modified at most once by the evaluation of an expression. Furthermore,
2406 the prior value shall be read only to determine the value to be
2407 stored.''. If a program breaks these rules, the results on any
2408 particular implementation are entirely unpredictable.
2410 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2411 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2412 diagnosed by this option, and it may give an occasional false positive
2413 result, but in general it has been found fairly effective at detecting
2414 this sort of problem in programs.
2416 The present implementation of this option only works for C programs. A
2417 future implementation may also work for C++ programs.
2419 The C standard is worded confusingly, therefore there is some debate
2420 over the precise meaning of the sequence point rules in subtle cases.
2421 Links to discussions of the problem, including proposed formal
2422 definitions, may be found on the GCC readings page, at
2423 @w{@uref{http://gcc.gnu.org/readings.html}}.
2426 @opindex Wreturn-type
2427 Warn whenever a function is defined with a return-type that defaults to
2428 @code{int}. Also warn about any @code{return} statement with no
2429 return-value in a function whose return-type is not @code{void}.
2431 For C, also warn if the return type of a function has a type qualifier
2432 such as @code{const}. Such a type qualifier has no effect, since the
2433 value returned by a function is not an lvalue. ISO C prohibits
2434 qualified @code{void} return types on function definitions, so such
2435 return types always receive a warning even without this option.
2437 For C++, a function without return type always produces a diagnostic
2438 message, even when @option{-Wno-return-type} is specified. The only
2439 exceptions are @samp{main} and functions defined in system headers.
2443 Warn whenever a @code{switch} statement has an index of enumerated type
2444 and lacks a @code{case} for one or more of the named codes of that
2445 enumeration. (The presence of a @code{default} label prevents this
2446 warning.) @code{case} labels outside the enumeration range also
2447 provoke warnings when this option is used.
2449 @item -Wswitch-default
2450 @opindex Wswitch-switch
2451 Warn whenever a @code{switch} statement does not have a @code{default}
2455 @opindex Wswitch-enum
2456 Warn whenever a @code{switch} statement has an index of enumerated type
2457 and lacks a @code{case} for one or more of the named codes of that
2458 enumeration. @code{case} labels outside the enumeration range also
2459 provoke warnings when this option is used.
2463 Warn if any trigraphs are encountered that might change the meaning of
2464 the program (trigraphs within comments are not warned about).
2466 @item -Wunused-function
2467 @opindex Wunused-function
2468 Warn whenever a static function is declared but not defined or a
2469 non\-inline static function is unused.
2471 @item -Wunused-label
2472 @opindex Wunused-label
2473 Warn whenever a label is declared but not used.
2475 To suppress this warning use the @samp{unused} attribute
2476 (@pxref{Variable Attributes}).
2478 @item -Wunused-parameter
2479 @opindex Wunused-parameter
2480 Warn whenever a function parameter is unused aside from its declaration.
2482 To suppress this warning use the @samp{unused} attribute
2483 (@pxref{Variable Attributes}).
2485 @item -Wunused-variable
2486 @opindex Wunused-variable
2487 Warn whenever a local variable or non-constant static variable is unused
2488 aside from its declaration
2490 To suppress this warning use the @samp{unused} attribute
2491 (@pxref{Variable Attributes}).
2493 @item -Wunused-value
2494 @opindex Wunused-value
2495 Warn whenever a statement computes a result that is explicitly not used.
2497 To suppress this warning cast the expression to @samp{void}.
2501 All the above @option{-Wunused} options combined.
2503 In order to get a warning about an unused function parameter, you must
2504 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2505 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2507 @item -Wuninitialized
2508 @opindex Wuninitialized
2509 Warn if an automatic variable is used without first being initialized or
2510 if a variable may be clobbered by a @code{setjmp} call.
2512 These warnings are possible only in optimizing compilation,
2513 because they require data flow information that is computed only
2514 when optimizing. If you don't specify @option{-O}, you simply won't
2517 If you want to warn about code which uses the uninitialized value of the
2518 variable in its own initializer, use the @option{-Winit-self} option.
2520 These warnings occur only for variables that are candidates for
2521 register allocation. Therefore, they do not occur for a variable that
2522 is declared @code{volatile}, or whose address is taken, or whose size
2523 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2524 structures, unions or arrays, even when they are in registers.
2526 Note that there may be no warning about a variable that is used only
2527 to compute a value that itself is never used, because such
2528 computations may be deleted by data flow analysis before the warnings
2531 These warnings are made optional because GCC is not smart
2532 enough to see all the reasons why the code might be correct
2533 despite appearing to have an error. Here is one example of how
2554 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2555 always initialized, but GCC doesn't know this. Here is
2556 another common case:
2561 if (change_y) save_y = y, y = new_y;
2563 if (change_y) y = save_y;
2568 This has no bug because @code{save_y} is used only if it is set.
2570 @cindex @code{longjmp} warnings
2571 This option also warns when a non-volatile automatic variable might be
2572 changed by a call to @code{longjmp}. These warnings as well are possible
2573 only in optimizing compilation.
2575 The compiler sees only the calls to @code{setjmp}. It cannot know
2576 where @code{longjmp} will be called; in fact, a signal handler could
2577 call it at any point in the code. As a result, you may get a warning
2578 even when there is in fact no problem because @code{longjmp} cannot
2579 in fact be called at the place which would cause a problem.
2581 Some spurious warnings can be avoided if you declare all the functions
2582 you use that never return as @code{noreturn}. @xref{Function
2585 @item -Wunknown-pragmas
2586 @opindex Wunknown-pragmas
2587 @cindex warning for unknown pragmas
2588 @cindex unknown pragmas, warning
2589 @cindex pragmas, warning of unknown
2590 Warn when a #pragma directive is encountered which is not understood by
2591 GCC@. If this command line option is used, warnings will even be issued
2592 for unknown pragmas in system header files. This is not the case if
2593 the warnings were only enabled by the @option{-Wall} command line option.
2595 @item -Wstrict-aliasing
2596 @opindex Wstrict-aliasing
2597 This option is only active when @option{-fstrict-aliasing} is active.
2598 It warns about code which might break the strict aliasing rules that the
2599 compiler is using for optimization. The warning does not catch all
2600 cases, but does attempt to catch the more common pitfalls. It is
2601 included in @option{-Wall}.
2603 @item -Wstrict-aliasing=2
2604 @opindex Wstrict-aliasing=2
2605 This option is only active when @option{-fstrict-aliasing} is active.
2606 It warns about all code which might break the strict aliasing rules that the
2607 compiler is using for optimization. This warning catches all cases, but
2608 it will also give a warning for some ambiguous cases that are safe.
2612 All of the above @samp{-W} options combined. This enables all the
2613 warnings about constructions that some users consider questionable, and
2614 that are easy to avoid (or modify to prevent the warning), even in
2615 conjunction with macros. This also enables some language-specific
2616 warnings described in @ref{C++ Dialect Options} and
2617 @ref{Objective-C and Objective-C++ Dialect Options}.
2620 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2621 Some of them warn about constructions that users generally do not
2622 consider questionable, but which occasionally you might wish to check
2623 for; others warn about constructions that are necessary or hard to avoid
2624 in some cases, and there is no simple way to modify the code to suppress
2631 (This option used to be called @option{-W}. The older name is still
2632 supported, but the newer name is more descriptive.) Print extra warning
2633 messages for these events:
2637 A function can return either with or without a value. (Falling
2638 off the end of the function body is considered returning without
2639 a value.) For example, this function would evoke such a
2653 An expression-statement or the left-hand side of a comma expression
2654 contains no side effects.
2655 To suppress the warning, cast the unused expression to void.
2656 For example, an expression such as @samp{x[i,j]} will cause a warning,
2657 but @samp{x[(void)i,j]} will not.
2660 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2663 Storage-class specifiers like @code{static} are not the first things in
2664 a declaration. According to the C Standard, this usage is obsolescent.
2667 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2671 A comparison between signed and unsigned values could produce an
2672 incorrect result when the signed value is converted to unsigned.
2673 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2676 An aggregate has an initializer which does not initialize all members.
2677 This warning can be independently controlled by
2678 @option{-Wmissing-field-initializers}.
2681 A function parameter is declared without a type specifier in K&R-style
2689 An empty body occurs in an @samp{if} or @samp{else} statement.
2692 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2693 @samp{>}, or @samp{>=}.
2696 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2699 Any of several floating-point events that often indicate errors, such as
2700 overflow, underflow, loss of precision, etc.
2702 @item @r{(C++ only)}
2703 An enumerator and a non-enumerator both appear in a conditional expression.
2705 @item @r{(C++ only)}
2706 A non-static reference or non-static @samp{const} member appears in a
2707 class without constructors.
2709 @item @r{(C++ only)}
2710 Ambiguous virtual bases.
2712 @item @r{(C++ only)}
2713 Subscripting an array which has been declared @samp{register}.
2715 @item @r{(C++ only)}
2716 Taking the address of a variable which has been declared @samp{register}.
2718 @item @r{(C++ only)}
2719 A base class is not initialized in a derived class' copy constructor.
2722 @item -Wno-div-by-zero
2723 @opindex Wno-div-by-zero
2724 @opindex Wdiv-by-zero
2725 Do not warn about compile-time integer division by zero. Floating point
2726 division by zero is not warned about, as it can be a legitimate way of
2727 obtaining infinities and NaNs.
2729 @item -Wsystem-headers
2730 @opindex Wsystem-headers
2731 @cindex warnings from system headers
2732 @cindex system headers, warnings from
2733 Print warning messages for constructs found in system header files.
2734 Warnings from system headers are normally suppressed, on the assumption
2735 that they usually do not indicate real problems and would only make the
2736 compiler output harder to read. Using this command line option tells
2737 GCC to emit warnings from system headers as if they occurred in user
2738 code. However, note that using @option{-Wall} in conjunction with this
2739 option will @emph{not} warn about unknown pragmas in system
2740 headers---for that, @option{-Wunknown-pragmas} must also be used.
2743 @opindex Wfloat-equal
2744 Warn if floating point values are used in equality comparisons.
2746 The idea behind this is that sometimes it is convenient (for the
2747 programmer) to consider floating-point values as approximations to
2748 infinitely precise real numbers. If you are doing this, then you need
2749 to compute (by analyzing the code, or in some other way) the maximum or
2750 likely maximum error that the computation introduces, and allow for it
2751 when performing comparisons (and when producing output, but that's a
2752 different problem). In particular, instead of testing for equality, you
2753 would check to see whether the two values have ranges that overlap; and
2754 this is done with the relational operators, so equality comparisons are
2757 @item -Wtraditional @r{(C only)}
2758 @opindex Wtraditional
2759 Warn about certain constructs that behave differently in traditional and
2760 ISO C@. Also warn about ISO C constructs that have no traditional C
2761 equivalent, and/or problematic constructs which should be avoided.
2765 Macro parameters that appear within string literals in the macro body.
2766 In traditional C macro replacement takes place within string literals,
2767 but does not in ISO C@.
2770 In traditional C, some preprocessor directives did not exist.
2771 Traditional preprocessors would only consider a line to be a directive
2772 if the @samp{#} appeared in column 1 on the line. Therefore
2773 @option{-Wtraditional} warns about directives that traditional C
2774 understands but would ignore because the @samp{#} does not appear as the
2775 first character on the line. It also suggests you hide directives like
2776 @samp{#pragma} not understood by traditional C by indenting them. Some
2777 traditional implementations would not recognize @samp{#elif}, so it
2778 suggests avoiding it altogether.
2781 A function-like macro that appears without arguments.
2784 The unary plus operator.
2787 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2788 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2789 constants.) Note, these suffixes appear in macros defined in the system
2790 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2791 Use of these macros in user code might normally lead to spurious
2792 warnings, however GCC's integrated preprocessor has enough context to
2793 avoid warning in these cases.
2796 A function declared external in one block and then used after the end of
2800 A @code{switch} statement has an operand of type @code{long}.
2803 A non-@code{static} function declaration follows a @code{static} one.
2804 This construct is not accepted by some traditional C compilers.
2807 The ISO type of an integer constant has a different width or
2808 signedness from its traditional type. This warning is only issued if
2809 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2810 typically represent bit patterns, are not warned about.
2813 Usage of ISO string concatenation is detected.
2816 Initialization of automatic aggregates.
2819 Identifier conflicts with labels. Traditional C lacks a separate
2820 namespace for labels.
2823 Initialization of unions. If the initializer is zero, the warning is
2824 omitted. This is done under the assumption that the zero initializer in
2825 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2826 initializer warnings and relies on default initialization to zero in the
2830 Conversions by prototypes between fixed/floating point values and vice
2831 versa. The absence of these prototypes when compiling with traditional
2832 C would cause serious problems. This is a subset of the possible
2833 conversion warnings, for the full set use @option{-Wconversion}.
2836 Use of ISO C style function definitions. This warning intentionally is
2837 @emph{not} issued for prototype declarations or variadic functions
2838 because these ISO C features will appear in your code when using
2839 libiberty's traditional C compatibility macros, @code{PARAMS} and
2840 @code{VPARAMS}. This warning is also bypassed for nested functions
2841 because that feature is already a GCC extension and thus not relevant to
2842 traditional C compatibility.
2845 @item -Wdeclaration-after-statement @r{(C only)}
2846 @opindex Wdeclaration-after-statement
2847 Warn when a declaration is found after a statement in a block. This
2848 construct, known from C++, was introduced with ISO C99 and is by default
2849 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2850 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2854 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2856 @item -Wendif-labels
2857 @opindex Wendif-labels
2858 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2862 Warn whenever a local variable shadows another local variable, parameter or
2863 global variable or whenever a built-in function is shadowed.
2865 @item -Wlarger-than-@var{len}
2866 @opindex Wlarger-than
2867 Warn whenever an object of larger than @var{len} bytes is defined.
2869 @item -Wpointer-arith
2870 @opindex Wpointer-arith
2871 Warn about anything that depends on the ``size of'' a function type or
2872 of @code{void}. GNU C assigns these types a size of 1, for
2873 convenience in calculations with @code{void *} pointers and pointers
2876 @item -Wbad-function-cast @r{(C only)}
2877 @opindex Wbad-function-cast
2878 Warn whenever a function call is cast to a non-matching type.
2879 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2883 Warn whenever a pointer is cast so as to remove a type qualifier from
2884 the target type. For example, warn if a @code{const char *} is cast
2885 to an ordinary @code{char *}.
2888 @opindex Wcast-align
2889 Warn whenever a pointer is cast such that the required alignment of the
2890 target is increased. For example, warn if a @code{char *} is cast to
2891 an @code{int *} on machines where integers can only be accessed at
2892 two- or four-byte boundaries.
2894 @item -Wwrite-strings
2895 @opindex Wwrite-strings
2896 When compiling C, give string constants the type @code{const
2897 char[@var{length}]} so that
2898 copying the address of one into a non-@code{const} @code{char *}
2899 pointer will get a warning; when compiling C++, warn about the
2900 deprecated conversion from string constants to @code{char *}.
2901 These warnings will help you find at
2902 compile time code that can try to write into a string constant, but
2903 only if you have been very careful about using @code{const} in
2904 declarations and prototypes. Otherwise, it will just be a nuisance;
2905 this is why we did not make @option{-Wall} request these warnings.
2908 @opindex Wconversion
2909 Warn if a prototype causes a type conversion that is different from what
2910 would happen to the same argument in the absence of a prototype. This
2911 includes conversions of fixed point to floating and vice versa, and
2912 conversions changing the width or signedness of a fixed point argument
2913 except when the same as the default promotion.
2915 Also, warn if a negative integer constant expression is implicitly
2916 converted to an unsigned type. For example, warn about the assignment
2917 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2918 casts like @code{(unsigned) -1}.
2920 @item -Wsign-compare
2921 @opindex Wsign-compare
2922 @cindex warning for comparison of signed and unsigned values
2923 @cindex comparison of signed and unsigned values, warning
2924 @cindex signed and unsigned values, comparison warning
2925 Warn when a comparison between signed and unsigned values could produce
2926 an incorrect result when the signed value is converted to unsigned.
2927 This warning is also enabled by @option{-Wextra}; to get the other warnings
2928 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2930 @item -Waggregate-return
2931 @opindex Waggregate-return
2932 Warn if any functions that return structures or unions are defined or
2933 called. (In languages where you can return an array, this also elicits
2936 @item -Wstrict-prototypes @r{(C only)}
2937 @opindex Wstrict-prototypes
2938 Warn if a function is declared or defined without specifying the
2939 argument types. (An old-style function definition is permitted without
2940 a warning if preceded by a declaration which specifies the argument
2943 @item -Wold-style-definition @r{(C only)}
2944 @opindex Wold-style-definition
2945 Warn if an old-style function definition is used. A warning is given
2946 even if there is a previous prototype.
2948 @item -Wmissing-prototypes @r{(C only)}
2949 @opindex Wmissing-prototypes
2950 Warn if a global function is defined without a previous prototype
2951 declaration. This warning is issued even if the definition itself
2952 provides a prototype. The aim is to detect global functions that fail
2953 to be declared in header files.
2955 @item -Wmissing-declarations @r{(C only)}
2956 @opindex Wmissing-declarations
2957 Warn if a global function is defined without a previous declaration.
2958 Do so even if the definition itself provides a prototype.
2959 Use this option to detect global functions that are not declared in
2962 @item -Wmissing-field-initializers
2963 @opindex Wmissing-field-initializers
2966 Warn if a structure's initializer has some fields missing. For
2967 example, the following code would cause such a warning, because
2968 @code{x.h} is implicitly zero:
2971 struct s @{ int f, g, h; @};
2972 struct s x = @{ 3, 4 @};
2975 This option does not warn about designated initializers, so the following
2976 modification would not trigger a warning:
2979 struct s @{ int f, g, h; @};
2980 struct s x = @{ .f = 3, .g = 4 @};
2983 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
2984 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
2986 @item -Wmissing-noreturn
2987 @opindex Wmissing-noreturn
2988 Warn about functions which might be candidates for attribute @code{noreturn}.
2989 Note these are only possible candidates, not absolute ones. Care should
2990 be taken to manually verify functions actually do not ever return before
2991 adding the @code{noreturn} attribute, otherwise subtle code generation
2992 bugs could be introduced. You will not get a warning for @code{main} in
2993 hosted C environments.
2995 @item -Wmissing-format-attribute
2996 @opindex Wmissing-format-attribute
2998 If @option{-Wformat} is enabled, also warn about functions which might be
2999 candidates for @code{format} attributes. Note these are only possible
3000 candidates, not absolute ones. GCC will guess that @code{format}
3001 attributes might be appropriate for any function that calls a function
3002 like @code{vprintf} or @code{vscanf}, but this might not always be the
3003 case, and some functions for which @code{format} attributes are
3004 appropriate may not be detected. This option has no effect unless
3005 @option{-Wformat} is enabled (possibly by @option{-Wall}).
3007 @item -Wno-multichar
3008 @opindex Wno-multichar
3010 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3011 Usually they indicate a typo in the user's code, as they have
3012 implementation-defined values, and should not be used in portable code.
3014 @item -Wno-deprecated-declarations
3015 @opindex Wno-deprecated-declarations
3016 Do not warn about uses of functions, variables, and types marked as
3017 deprecated by using the @code{deprecated} attribute.
3018 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3019 @pxref{Type Attributes}.)
3023 Warn if a structure is given the packed attribute, but the packed
3024 attribute has no effect on the layout or size of the structure.
3025 Such structures may be mis-aligned for little benefit. For
3026 instance, in this code, the variable @code{f.x} in @code{struct bar}
3027 will be misaligned even though @code{struct bar} does not itself
3028 have the packed attribute:
3035 @} __attribute__((packed));
3045 Warn if padding is included in a structure, either to align an element
3046 of the structure or to align the whole structure. Sometimes when this
3047 happens it is possible to rearrange the fields of the structure to
3048 reduce the padding and so make the structure smaller.
3050 @item -Wredundant-decls
3051 @opindex Wredundant-decls
3052 Warn if anything is declared more than once in the same scope, even in
3053 cases where multiple declaration is valid and changes nothing.
3055 @item -Wnested-externs @r{(C only)}
3056 @opindex Wnested-externs
3057 Warn if an @code{extern} declaration is encountered within a function.
3059 @item -Wunreachable-code
3060 @opindex Wunreachable-code
3061 Warn if the compiler detects that code will never be executed.
3063 This option is intended to warn when the compiler detects that at
3064 least a whole line of source code will never be executed, because
3065 some condition is never satisfied or because it is after a
3066 procedure that never returns.
3068 It is possible for this option to produce a warning even though there
3069 are circumstances under which part of the affected line can be executed,
3070 so care should be taken when removing apparently-unreachable code.
3072 For instance, when a function is inlined, a warning may mean that the
3073 line is unreachable in only one inlined copy of the function.
3075 This option is not made part of @option{-Wall} because in a debugging
3076 version of a program there is often substantial code which checks
3077 correct functioning of the program and is, hopefully, unreachable
3078 because the program does work. Another common use of unreachable
3079 code is to provide behavior which is selectable at compile-time.
3083 Warn if a function can not be inlined and it was declared as inline.
3084 Even with this option, the compiler will not warn about failures to
3085 inline functions declared in system headers.
3087 The compiler uses a variety of heuristics to determine whether or not
3088 to inline a function. For example, the compiler takes into account
3089 the size of the function being inlined and the the amount of inlining
3090 that has already been done in the current function. Therefore,
3091 seemingly insignificant changes in the source program can cause the
3092 warnings produced by @option{-Winline} to appear or disappear.
3094 @item -Wno-invalid-offsetof @r{(C++ only)}
3095 @opindex Wno-invalid-offsetof
3096 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3097 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3098 to a non-POD type is undefined. In existing C++ implementations,
3099 however, @samp{offsetof} typically gives meaningful results even when
3100 applied to certain kinds of non-POD types. (Such as a simple
3101 @samp{struct} that fails to be a POD type only by virtue of having a
3102 constructor.) This flag is for users who are aware that they are
3103 writing nonportable code and who have deliberately chosen to ignore the
3106 The restrictions on @samp{offsetof} may be relaxed in a future version
3107 of the C++ standard.
3110 @opindex Winvalid-pch
3111 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3112 the search path but can't be used.
3116 @opindex Wno-long-long
3117 Warn if @samp{long long} type is used. This is default. To inhibit
3118 the warning messages, use @option{-Wno-long-long}. Flags
3119 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3120 only when @option{-pedantic} flag is used.
3122 @item -Wvariadic-macros
3123 @opindex Wvariadic-macros
3124 @opindex Wno-variadic-macros
3125 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3126 alternate syntax when in pedantic ISO C99 mode. This is default.
3127 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3129 @item -Wdisabled-optimization
3130 @opindex Wdisabled-optimization
3131 Warn if a requested optimization pass is disabled. This warning does
3132 not generally indicate that there is anything wrong with your code; it
3133 merely indicates that GCC's optimizers were unable to handle the code
3134 effectively. Often, the problem is that your code is too big or too
3135 complex; GCC will refuse to optimize programs when the optimization
3136 itself is likely to take inordinate amounts of time.
3140 Make all warnings into errors.
3143 @node Debugging Options
3144 @section Options for Debugging Your Program or GCC
3145 @cindex options, debugging
3146 @cindex debugging information options
3148 GCC has various special options that are used for debugging
3149 either your program or GCC:
3154 Produce debugging information in the operating system's native format
3155 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3158 On most systems that use stabs format, @option{-g} enables use of extra
3159 debugging information that only GDB can use; this extra information
3160 makes debugging work better in GDB but will probably make other debuggers
3162 refuse to read the program. If you want to control for certain whether
3163 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3164 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3166 GCC allows you to use @option{-g} with
3167 @option{-O}. The shortcuts taken by optimized code may occasionally
3168 produce surprising results: some variables you declared may not exist
3169 at all; flow of control may briefly move where you did not expect it;
3170 some statements may not be executed because they compute constant
3171 results or their values were already at hand; some statements may
3172 execute in different places because they were moved out of loops.
3174 Nevertheless it proves possible to debug optimized output. This makes
3175 it reasonable to use the optimizer for programs that might have bugs.
3177 The following options are useful when GCC is generated with the
3178 capability for more than one debugging format.
3182 Produce debugging information for use by GDB@. This means to use the
3183 most expressive format available (DWARF 2, stabs, or the native format
3184 if neither of those are supported), including GDB extensions if at all
3189 Produce debugging information in stabs format (if that is supported),
3190 without GDB extensions. This is the format used by DBX on most BSD
3191 systems. On MIPS, Alpha and System V Release 4 systems this option
3192 produces stabs debugging output which is not understood by DBX or SDB@.
3193 On System V Release 4 systems this option requires the GNU assembler.
3195 @item -feliminate-unused-debug-symbols
3196 @opindex feliminate-unused-debug-symbols
3197 Produce debugging information in stabs format (if that is supported),
3198 for only symbols that are actually used.
3202 Produce debugging information in stabs format (if that is supported),
3203 using GNU extensions understood only by the GNU debugger (GDB)@. The
3204 use of these extensions is likely to make other debuggers crash or
3205 refuse to read the program.
3209 Produce debugging information in COFF format (if that is supported).
3210 This is the format used by SDB on most System V systems prior to
3215 Produce debugging information in XCOFF format (if that is supported).
3216 This is the format used by the DBX debugger on IBM RS/6000 systems.
3220 Produce debugging information in XCOFF format (if that is supported),
3221 using GNU extensions understood only by the GNU debugger (GDB)@. The
3222 use of these extensions is likely to make other debuggers crash or
3223 refuse to read the program, and may cause assemblers other than the GNU
3224 assembler (GAS) to fail with an error.
3228 Produce debugging information in DWARF version 2 format (if that is
3229 supported). This is the format used by DBX on IRIX 6. With this
3230 option, GCC uses features of DWARF version 3 when they are useful;
3231 version 3 is upward compatible with version 2, but may still cause
3232 problems for older debuggers.
3236 Produce debugging information in VMS debug format (if that is
3237 supported). This is the format used by DEBUG on VMS systems.
3240 @itemx -ggdb@var{level}
3241 @itemx -gstabs@var{level}
3242 @itemx -gcoff@var{level}
3243 @itemx -gxcoff@var{level}
3244 @itemx -gvms@var{level}
3245 Request debugging information and also use @var{level} to specify how
3246 much information. The default level is 2.
3248 Level 1 produces minimal information, enough for making backtraces in
3249 parts of the program that you don't plan to debug. This includes
3250 descriptions of functions and external variables, but no information
3251 about local variables and no line numbers.
3253 Level 3 includes extra information, such as all the macro definitions
3254 present in the program. Some debuggers support macro expansion when
3255 you use @option{-g3}.
3257 @option{-gdwarf-2} does not accept a concatenated debug level, because
3258 GCC used to support an option @option{-gdwarf} that meant to generate
3259 debug information in version 1 of the DWARF format (which is very
3260 different from version 2), and it would have been too confusing. That
3261 debug format is long obsolete, but the option cannot be changed now.
3262 Instead use an additional @option{-g@var{level}} option to change the
3263 debug level for DWARF2.
3265 @item -feliminate-dwarf2-dups
3266 @opindex feliminate-dwarf2-dups
3267 Compress DWARF2 debugging information by eliminating duplicated
3268 information about each symbol. This option only makes sense when
3269 generating DWARF2 debugging information with @option{-gdwarf-2}.
3271 @cindex @command{prof}
3274 Generate extra code to write profile information suitable for the
3275 analysis program @command{prof}. You must use this option when compiling
3276 the source files you want data about, and you must also use it when
3279 @cindex @command{gprof}
3282 Generate extra code to write profile information suitable for the
3283 analysis program @command{gprof}. You must use this option when compiling
3284 the source files you want data about, and you must also use it when
3289 Makes the compiler print out each function name as it is compiled, and
3290 print some statistics about each pass when it finishes.
3293 @opindex ftime-report
3294 Makes the compiler print some statistics about the time consumed by each
3295 pass when it finishes.
3298 @opindex fmem-report
3299 Makes the compiler print some statistics about permanent memory
3300 allocation when it finishes.
3302 @item -fprofile-arcs
3303 @opindex fprofile-arcs
3304 Add code so that program flow @dfn{arcs} are instrumented. During
3305 execution the program records how many times each branch and call is
3306 executed and how many times it is taken or returns. When the compiled
3307 program exits it saves this data to a file called
3308 @file{@var{auxname}.gcda} for each source file. The data may be used for
3309 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3310 test coverage analysis (@option{-ftest-coverage}). Each object file's
3311 @var{auxname} is generated from the name of the output file, if
3312 explicitly specified and it is not the final executable, otherwise it is
3313 the basename of the source file. In both cases any suffix is removed
3314 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3315 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3320 Compile the source files with @option{-fprofile-arcs} plus optimization
3321 and code generation options. For test coverage analysis, use the
3322 additional @option{-ftest-coverage} option. You do not need to profile
3323 every source file in a program.
3326 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3327 (the latter implies the former).
3330 Run the program on a representative workload to generate the arc profile
3331 information. This may be repeated any number of times. You can run
3332 concurrent instances of your program, and provided that the file system
3333 supports locking, the data files will be correctly updated. Also
3334 @code{fork} calls are detected and correctly handled (double counting
3338 For profile-directed optimizations, compile the source files again with
3339 the same optimization and code generation options plus
3340 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3341 Control Optimization}).
3344 For test coverage analysis, use @command{gcov} to produce human readable
3345 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3346 @command{gcov} documentation for further information.
3350 With @option{-fprofile-arcs}, for each function of your program GCC
3351 creates a program flow graph, then finds a spanning tree for the graph.
3352 Only arcs that are not on the spanning tree have to be instrumented: the
3353 compiler adds code to count the number of times that these arcs are
3354 executed. When an arc is the only exit or only entrance to a block, the
3355 instrumentation code can be added to the block; otherwise, a new basic
3356 block must be created to hold the instrumentation code.
3358 @item -ftree-based-profiling
3359 @opindex ftree-based-profiling
3360 This option is used in addition to @option{-fprofile-arcs} or
3361 @option{-fbranch-probabilities} to control whether those optimizations
3362 are performed on a tree-based or rtl-based internal representation.
3363 If you use this option when compiling with @option{-fprofile-arcs},
3364 you must also use it when compiling later with @option{-fbranch-probabilities}.
3365 Currently the tree-based optimization is in an early stage of
3366 development, and this option is recommended only for those people
3367 working on improving it.
3370 @item -ftest-coverage
3371 @opindex ftest-coverage
3372 Produce a notes file that the @command{gcov} code-coverage utility
3373 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3374 show program coverage. Each source file's note file is called
3375 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3376 above for a description of @var{auxname} and instructions on how to
3377 generate test coverage data. Coverage data will match the source files
3378 more closely, if you do not optimize.
3380 @item -d@var{letters}
3381 @item -fdump-rtl-@var{pass}
3383 Says to make debugging dumps during compilation at times specified by
3384 @var{letters}. This is used for debugging the RTL-based passes of the
3385 compiler. The file names for most of the dumps are made by appending a
3386 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3387 from the name of the output file, if explicitly specified and it is not
3388 an executable, otherwise it is the basename of the source file.
3390 Most debug dumps can be enabled either passing a letter to the @option{-d}
3391 option, or with a long @option{-fdump-rtl} switch; here are the possible
3392 letters for use in @var{letters} and @var{pass}, and their meanings:
3397 Annotate the assembler output with miscellaneous debugging information.
3400 @itemx -fdump-rtl-bp
3402 @opindex fdump-rtl-bp
3403 Dump after computing branch probabilities, to @file{@var{file}.09.bp}.
3406 @itemx -fdump-rtl-bbro
3408 @opindex fdump-rtl-bbro
3409 Dump after block reordering, to @file{@var{file}.30.bbro}.
3412 @itemx -fdump-rtl-combine
3414 @opindex fdump-rtl-combine
3415 Dump after instruction combination, to the file @file{@var{file}.17.combine}.
3418 @itemx -fdump-rtl-ce1
3419 @itemx -fdump-rtl-ce2
3421 @opindex fdump-rtl-ce1
3422 @opindex fdump-rtl-ce2
3423 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3424 first if conversion, to the file @file{@var{file}.11.ce1}. @option{-dC}
3425 and @option{-fdump-rtl-ce2} enable dumping after the second if
3426 conversion, to the file @file{@var{file}.18.ce2}.
3429 @itemx -fdump-rtl-btl
3430 @itemx -fdump-rtl-dbr
3432 @opindex fdump-rtl-btl
3433 @opindex fdump-rtl-dbr
3434 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3435 target load optimization, to to @file{@var{file}.31.btl}. @option{-dd}
3436 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3437 scheduling, to @file{@var{file}.36.dbr}.
3441 Dump all macro definitions, at the end of preprocessing, in addition to
3445 @itemx -fdump-rtl-ce3
3447 @opindex fdump-rtl-ce3
3448 Dump after the third if conversion, to @file{@var{file}.28.ce3}.
3451 @itemx -fdump-rtl-cfg
3452 @itemx -fdump-rtl-life
3454 @opindex fdump-rtl-cfg
3455 @opindex fdump-rtl-life
3456 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3457 and data flow analysis, to @file{@var{file}.08.cfg}. @option{-df}
3458 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3459 to @file{@var{file}.16.life}.
3462 @itemx -fdump-rtl-greg
3464 @opindex fdump-rtl-greg
3465 Dump after global register allocation, to @file{@var{file}.23.greg}.
3468 @itemx -fdump-rtl-gcse
3469 @itemx -fdump-rtl-bypass
3471 @opindex fdump-rtl-gcse
3472 @opindex fdump-rtl-bypass
3473 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3474 @file{@var{file}.05.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3475 enable dumping after jump bypassing and control flow optimizations, to
3476 @file{@var{file}.07.bypass}.
3479 @itemx -fdump-rtl-eh
3481 @opindex fdump-rtl-eh
3482 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3485 @itemx -fdump-rtl-sibling
3487 @opindex fdump-rtl-sibling
3488 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3491 @itemx -fdump-rtl-jump
3493 @opindex fdump-rtl-jump
3494 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3497 @itemx -fdump-rtl-stack
3499 @opindex fdump-rtl-stack
3500 Dump after conversion from registers to stack, to @file{@var{file}.33.stack}.
3503 @itemx -fdump-rtl-lreg
3505 @opindex fdump-rtl-lreg
3506 Dump after local register allocation, to @file{@var{file}.22.lreg}.
3509 @itemx -fdump-rtl-loop
3510 @itemx -fdump-rtl-loop2
3512 @opindex fdump-rtl-loop
3513 @opindex fdump-rtl-loop2
3514 @option{-dL} and @option{-fdump-rtl-loop} enable dumping after the first
3515 loop optimization pass, to @file{@var{file}.06.loop}. @option{-dL} and
3516 @option{-fdump-rtl-loop2} enable dumping after the second pass, to
3517 @file{@var{file}.13.loop2}.
3520 @itemx -fdump-rtl-sms
3522 @opindex fdump-rtl-sms
3523 Dump after modulo scheduling, to @file{@var{file}.20.sms}.
3526 @itemx -fdump-rtl-mach
3528 @opindex fdump-rtl-mach
3529 Dump after performing the machine dependent reorganization pass, to
3530 @file{@var{file}.35.mach}.
3533 @itemx -fdump-rtl-rnreg
3535 @opindex fdump-rtl-rnreg
3536 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3539 @itemx -fdump-rtl-regmove
3541 @opindex fdump-rtl-regmove
3542 Dump after the register move pass, to @file{@var{file}.19.regmove}.
3545 @itemx -fdump-rtl-postreload
3547 @opindex fdump-rtl-postreload
3548 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3551 @itemx -fdump-rtl-expand
3553 @opindex fdump-rtl-expand
3554 Dump after RTL generation, to @file{@var{file}.00.expand}.
3557 @itemx -fdump-rtl-sched2
3559 @opindex fdump-rtl-sched2
3560 Dump after the second scheduling pass, to @file{@var{file}.32.sched2}.
3563 @itemx -fdump-rtl-cse
3565 @opindex fdump-rtl-cse
3566 Dump after CSE (including the jump optimization that sometimes follows
3567 CSE), to @file{@var{file}.04.cse}.
3570 @itemx -fdump-rtl-sched
3572 @opindex fdump-rtl-sched
3573 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3576 @itemx -fdump-rtl-cse2
3578 @opindex fdump-rtl-cse2
3579 Dump after the second CSE pass (including the jump optimization that
3580 sometimes follows CSE), to @file{@var{file}.15.cse2}.
3583 @itemx -fdump-rtl-tracer
3585 @opindex fdump-rtl-tracer
3586 Dump after running tracer, to @file{@var{file}.12.tracer}.
3589 @itemx -fdump-rtl-vpt
3590 @itemx -fdump-rtl-vartrack
3592 @opindex fdump-rtl-vpt
3593 @opindex fdump-rtl-vartrack
3594 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3595 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3596 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3597 to @file{@var{file}.34.vartrack}.
3600 @itemx -fdump-rtl-flow2
3602 @opindex fdump-rtl-flow2
3603 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3606 @itemx -fdump-rtl-peephole2
3608 @opindex fdump-rtl-peephole2
3609 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3612 @itemx -fdump-rtl-web
3614 @opindex fdump-rtl-web
3615 Dump after live range splitting, to @file{@var{file}.14.web}.
3618 @itemx -fdump-rtl-all
3620 @opindex fdump-rtl-all
3621 Produce all the dumps listed above.
3625 Produce a core dump whenever an error occurs.
3629 Print statistics on memory usage, at the end of the run, to
3634 Annotate the assembler output with a comment indicating which
3635 pattern and alternative was used. The length of each instruction is
3640 Dump the RTL in the assembler output as a comment before each instruction.
3641 Also turns on @option{-dp} annotation.
3645 For each of the other indicated dump files (either with @option{-d} or
3646 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3647 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3651 Just generate RTL for a function instead of compiling it. Usually used
3652 with @samp{r} (@option{-fdump-rtl-expand}).
3656 Dump debugging information during parsing, to standard error.
3659 @item -fdump-unnumbered
3660 @opindex fdump-unnumbered
3661 When doing debugging dumps (see @option{-d} option above), suppress instruction
3662 numbers and line number note output. This makes it more feasible to
3663 use diff on debugging dumps for compiler invocations with different
3664 options, in particular with and without @option{-g}.
3666 @item -fdump-translation-unit @r{(C and C++ only)}
3667 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3668 @opindex fdump-translation-unit
3669 Dump a representation of the tree structure for the entire translation
3670 unit to a file. The file name is made by appending @file{.tu} to the
3671 source file name. If the @samp{-@var{options}} form is used, @var{options}
3672 controls the details of the dump as described for the
3673 @option{-fdump-tree} options.
3675 @item -fdump-class-hierarchy @r{(C++ only)}
3676 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3677 @opindex fdump-class-hierarchy
3678 Dump a representation of each class's hierarchy and virtual function
3679 table layout to a file. The file name is made by appending @file{.class}
3680 to the source file name. If the @samp{-@var{options}} form is used,
3681 @var{options} controls the details of the dump as described for the
3682 @option{-fdump-tree} options.
3684 @item -fdump-ipa-@var{switch}
3686 Control the dumping at various stages of inter-procedural analysis
3687 language tree to a file. The file name is generated by appending a switch
3688 specific suffix to the source file name. The following dumps are possible:
3692 Enables all inter-procedural analysis dumps; currently the only produced
3693 dump is the @samp{cgraph} dump.
3696 Dumps information about call-graph optimization, unused function removal,
3697 and inlining decisions.
3700 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3701 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3703 Control the dumping at various stages of processing the intermediate
3704 language tree to a file. The file name is generated by appending a switch
3705 specific suffix to the source file name. If the @samp{-@var{options}}
3706 form is used, @var{options} is a list of @samp{-} separated options that
3707 control the details of the dump. Not all options are applicable to all
3708 dumps, those which are not meaningful will be ignored. The following
3709 options are available
3713 Print the address of each node. Usually this is not meaningful as it
3714 changes according to the environment and source file. Its primary use
3715 is for tying up a dump file with a debug environment.
3717 Inhibit dumping of members of a scope or body of a function merely
3718 because that scope has been reached. Only dump such items when they
3719 are directly reachable by some other path. When dumping pretty-printed
3720 trees, this option inhibits dumping the bodies of control structures.
3722 Print a raw representation of the tree. By default, trees are
3723 pretty-printed into a C-like representation.
3725 Enable more detailed dumps (not honored by every dump option).
3727 Enable dumping various statistics about the pass (not honored by every dump
3730 Enable showing basic block boundaries (disabled in raw dumps).
3732 Enable showing virtual operands for every statement.
3734 Enable showing line numbers for statements.
3736 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3738 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3741 The following tree dumps are possible:
3745 Dump before any tree based optimization, to @file{@var{file}.original}.
3748 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3751 Dump after function inlining, to @file{@var{file}.inlined}.
3754 @opindex fdump-tree-gimple
3755 Dump each function before and after the gimplification pass to a file. The
3756 file name is made by appending @file{.gimple} to the source file name.
3759 @opindex fdump-tree-cfg
3760 Dump the control flow graph of each function to a file. The file name is
3761 made by appending @file{.cfg} to the source file name.
3764 @opindex fdump-tree-vcg
3765 Dump the control flow graph of each function to a file in VCG format. The
3766 file name is made by appending @file{.vcg} to the source file name. Note
3767 that if the file contains more than one function, the generated file cannot
3768 be used directly by VCG@. You will need to cut and paste each function's
3769 graph into its own separate file first.
3772 @opindex fdump-tree-ch
3773 Dump each function after copying loop headers. The file name is made by
3774 appending @file{.ch} to the source file name.
3777 @opindex fdump-tree-ssa
3778 Dump SSA related information to a file. The file name is made by appending
3779 @file{.ssa} to the source file name.
3782 @opindex fdump-tree-alias
3783 Dump aliasing information for each function. The file name is made by
3784 appending @file{.alias} to the source file name.
3787 @opindex fdump-tree-ccp
3788 Dump each function after CCP@. The file name is made by appending
3789 @file{.ccp} to the source file name.
3792 @opindex fdump-tree-pre
3793 Dump trees after partial redundancy elimination. The file name is made
3794 by appending @file{.pre} to the source file name.
3797 @opindex fdump-tree-fre
3798 Dump trees after full redundancy elimination. The file name is made
3799 by appending @file{.fre} to the source file name.
3802 @opindex fdump-tree-dce
3803 Dump each function after dead code elimination. The file name is made by
3804 appending @file{.dce} to the source file name.
3807 @opindex fdump-tree-mudflap
3808 Dump each function after adding mudflap instrumentation. The file name is
3809 made by appending @file{.mudflap} to the source file name.
3812 @opindex fdump-tree-sra
3813 Dump each function after performing scalar replacement of aggregates. The
3814 file name is made by appending @file{.sra} to the source file name.
3817 @opindex fdump-tree-dom
3818 Dump each function after applying dominator tree optimizations. The file
3819 name is made by appending @file{.dom} to the source file name.
3822 @opindex fdump-tree-dse
3823 Dump each function after applying dead store elimination. The file
3824 name is made by appending @file{.dse} to the source file name.
3827 @opindex fdump-tree-phiopt
3828 Dump each function after optimizing PHI nodes into straightline code. The file
3829 name is made by appending @file{.phiopt} to the source file name.
3832 @opindex fdump-tree-forwprop
3833 Dump each function after forward propagating single use variables. The file
3834 name is made by appending @file{.forwprop} to the source file name.
3837 @opindex fdump-tree-copyrename
3838 Dump each function after applying the copy rename optimization. The file
3839 name is made by appending @file{.copyrename} to the source file name.
3842 @opindex fdump-tree-nrv
3843 Dump each function after applying the named return value optimization on
3844 generic trees. The file name is made by appending @file{.nrv} to the source
3848 @opindex fdump-tree-vect
3849 Dump each function after applying vectorization of loops. The file name is
3850 made by appending @file{.vect} to the source file name.
3853 @opindex fdump-tree-all
3854 Enable all the available tree dumps with the flags provided in this option.
3857 @item -frandom-seed=@var{string}
3858 @opindex frandom-string
3859 This option provides a seed that GCC uses when it would otherwise use
3860 random numbers. It is used to generate certain symbol names
3861 that have to be different in every compiled file. It is also used to
3862 place unique stamps in coverage data files and the object files that
3863 produce them. You can use the @option{-frandom-seed} option to produce
3864 reproducibly identical object files.
3866 The @var{string} should be different for every file you compile.
3868 @item -fsched-verbose=@var{n}
3869 @opindex fsched-verbose
3870 On targets that use instruction scheduling, this option controls the
3871 amount of debugging output the scheduler prints. This information is
3872 written to standard error, unless @option{-dS} or @option{-dR} is
3873 specified, in which case it is output to the usual dump
3874 listing file, @file{.sched} or @file{.sched2} respectively. However
3875 for @var{n} greater than nine, the output is always printed to standard
3878 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3879 same information as @option{-dRS}. For @var{n} greater than one, it
3880 also output basic block probabilities, detailed ready list information
3881 and unit/insn info. For @var{n} greater than two, it includes RTL
3882 at abort point, control-flow and regions info. And for @var{n} over
3883 four, @option{-fsched-verbose} also includes dependence info.
3887 Store the usual ``temporary'' intermediate files permanently; place them
3888 in the current directory and name them based on the source file. Thus,
3889 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3890 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3891 preprocessed @file{foo.i} output file even though the compiler now
3892 normally uses an integrated preprocessor.
3894 When used in combination with the @option{-x} command line option,
3895 @option{-save-temps} is sensible enough to avoid over writing an
3896 input source file with the same extension as an intermediate file.
3897 The corresponding intermediate file may be obtained by renaming the
3898 source file before using @option{-save-temps}.
3902 Report the CPU time taken by each subprocess in the compilation
3903 sequence. For C source files, this is the compiler proper and assembler
3904 (plus the linker if linking is done). The output looks like this:
3911 The first number on each line is the ``user time'', that is time spent
3912 executing the program itself. The second number is ``system time'',
3913 time spent executing operating system routines on behalf of the program.
3914 Both numbers are in seconds.
3916 @item -fvar-tracking
3917 @opindex fvar-tracking
3918 Run variable tracking pass. It computes where variables are stored at each
3919 position in code. Better debugging information is then generated
3920 (if the debugging information format supports this information).
3922 It is enabled by default when compiling with optimization (@option{-Os},
3923 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3924 the debug info format supports it.
3926 @item -print-file-name=@var{library}
3927 @opindex print-file-name
3928 Print the full absolute name of the library file @var{library} that
3929 would be used when linking---and don't do anything else. With this
3930 option, GCC does not compile or link anything; it just prints the
3933 @item -print-multi-directory
3934 @opindex print-multi-directory
3935 Print the directory name corresponding to the multilib selected by any
3936 other switches present in the command line. This directory is supposed
3937 to exist in @env{GCC_EXEC_PREFIX}.
3939 @item -print-multi-lib
3940 @opindex print-multi-lib
3941 Print the mapping from multilib directory names to compiler switches
3942 that enable them. The directory name is separated from the switches by
3943 @samp{;}, and each switch starts with an @samp{@@} instead of the
3944 @samp{-}, without spaces between multiple switches. This is supposed to
3945 ease shell-processing.
3947 @item -print-prog-name=@var{program}
3948 @opindex print-prog-name
3949 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3951 @item -print-libgcc-file-name
3952 @opindex print-libgcc-file-name
3953 Same as @option{-print-file-name=libgcc.a}.
3955 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3956 but you do want to link with @file{libgcc.a}. You can do
3959 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3962 @item -print-search-dirs
3963 @opindex print-search-dirs
3964 Print the name of the configured installation directory and a list of
3965 program and library directories @command{gcc} will search---and don't do anything else.
3967 This is useful when @command{gcc} prints the error message
3968 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3969 To resolve this you either need to put @file{cpp0} and the other compiler
3970 components where @command{gcc} expects to find them, or you can set the environment
3971 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3972 Don't forget the trailing @samp{/}.
3973 @xref{Environment Variables}.
3976 @opindex dumpmachine
3977 Print the compiler's target machine (for example,
3978 @samp{i686-pc-linux-gnu})---and don't do anything else.
3981 @opindex dumpversion
3982 Print the compiler version (for example, @samp{3.0})---and don't do
3987 Print the compiler's built-in specs---and don't do anything else. (This
3988 is used when GCC itself is being built.) @xref{Spec Files}.
3990 @item -feliminate-unused-debug-types
3991 @opindex feliminate-unused-debug-types
3992 Normally, when producing DWARF2 output, GCC will emit debugging
3993 information for all types declared in a compilation
3994 unit, regardless of whether or not they are actually used
3995 in that compilation unit. Sometimes this is useful, such as
3996 if, in the debugger, you want to cast a value to a type that is
3997 not actually used in your program (but is declared). More often,
3998 however, this results in a significant amount of wasted space.
3999 With this option, GCC will avoid producing debug symbol output
4000 for types that are nowhere used in the source file being compiled.
4003 @node Optimize Options
4004 @section Options That Control Optimization
4005 @cindex optimize options
4006 @cindex options, optimization
4008 These options control various sorts of optimizations.
4010 Without any optimization option, the compiler's goal is to reduce the
4011 cost of compilation and to make debugging produce the expected
4012 results. Statements are independent: if you stop the program with a
4013 breakpoint between statements, you can then assign a new value to any
4014 variable or change the program counter to any other statement in the
4015 function and get exactly the results you would expect from the source
4018 Turning on optimization flags makes the compiler attempt to improve
4019 the performance and/or code size at the expense of compilation time
4020 and possibly the ability to debug the program.
4022 The compiler performs optimization based on the knowledge it has of
4023 the program. Optimization levels @option{-O2} and above, in
4024 particular, enable @emph{unit-at-a-time} mode, which allows the
4025 compiler to consider information gained from later functions in
4026 the file when compiling a function. Compiling multiple files at
4027 once to a single output file in @emph{unit-at-a-time} mode allows
4028 the compiler to use information gained from all of the files when
4029 compiling each of them.
4031 Not all optimizations are controlled directly by a flag. Only
4032 optimizations that have a flag are listed.
4039 Optimize. Optimizing compilation takes somewhat more time, and a lot
4040 more memory for a large function.
4042 With @option{-O}, the compiler tries to reduce code size and execution
4043 time, without performing any optimizations that take a great deal of
4046 @option{-O} turns on the following optimization flags:
4047 @gccoptlist{-fdefer-pop @gol
4048 -fmerge-constants @gol
4050 -floop-optimize @gol
4051 -fif-conversion @gol
4052 -fif-conversion2 @gol
4053 -fdelayed-branch @gol
4054 -fguess-branch-probability @gol
4057 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4058 where doing so does not interfere with debugging.
4062 Optimize even more. GCC performs nearly all supported optimizations
4063 that do not involve a space-speed tradeoff. The compiler does not
4064 perform loop unrolling or function inlining when you specify @option{-O2}.
4065 As compared to @option{-O}, this option increases both compilation time
4066 and the performance of the generated code.
4068 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4069 also turns on the following optimization flags:
4070 @gccoptlist{-fforce-mem @gol
4071 -foptimize-sibling-calls @gol
4072 -fstrength-reduce @gol
4073 -fcse-follow-jumps -fcse-skip-blocks @gol
4074 -frerun-cse-after-loop -frerun-loop-opt @gol
4075 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
4076 -fdelete-null-pointer-checks @gol
4077 -fexpensive-optimizations @gol
4079 -fschedule-insns -fschedule-insns2 @gol
4080 -fsched-interblock -fsched-spec @gol
4083 -freorder-blocks -freorder-functions @gol
4084 -fstrict-aliasing @gol
4085 -funit-at-a-time @gol
4086 -falign-functions -falign-jumps @gol
4087 -falign-loops -falign-labels @gol
4090 Please note the warning under @option{-fgcse} about
4091 invoking @option{-O2} on programs that use computed gotos.
4095 Optimize yet more. @option{-O3} turns on all optimizations specified by
4096 @option{-O2} and also turns on the @option{-finline-functions},
4097 @option{-fweb} and @option{-fgcse-after-reload} options.
4101 Do not optimize. This is the default.
4105 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4106 do not typically increase code size. It also performs further
4107 optimizations designed to reduce code size.
4109 @option{-Os} disables the following optimization flags:
4110 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4111 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
4113 If you use multiple @option{-O} options, with or without level numbers,
4114 the last such option is the one that is effective.
4117 Options of the form @option{-f@var{flag}} specify machine-independent
4118 flags. Most flags have both positive and negative forms; the negative
4119 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4120 below, only one of the forms is listed---the one you typically will
4121 use. You can figure out the other form by either removing @samp{no-}
4124 The following options control specific optimizations. They are either
4125 activated by @option{-O} options or are related to ones that are. You
4126 can use the following flags in the rare cases when ``fine-tuning'' of
4127 optimizations to be performed is desired.
4130 @item -fno-default-inline
4131 @opindex fno-default-inline
4132 Do not make member functions inline by default merely because they are
4133 defined inside the class scope (C++ only). Otherwise, when you specify
4134 @w{@option{-O}}, member functions defined inside class scope are compiled
4135 inline by default; i.e., you don't need to add @samp{inline} in front of
4136 the member function name.
4138 @item -fno-defer-pop
4139 @opindex fno-defer-pop
4140 Always pop the arguments to each function call as soon as that function
4141 returns. For machines which must pop arguments after a function call,
4142 the compiler normally lets arguments accumulate on the stack for several
4143 function calls and pops them all at once.
4145 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4149 Force memory operands to be copied into registers before doing
4150 arithmetic on them. This produces better code by making all memory
4151 references potential common subexpressions. When they are not common
4152 subexpressions, instruction combination should eliminate the separate
4155 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4158 @opindex fforce-addr
4159 Force memory address constants to be copied into registers before
4160 doing arithmetic on them. This may produce better code just as
4161 @option{-fforce-mem} may.
4163 @item -fomit-frame-pointer
4164 @opindex fomit-frame-pointer
4165 Don't keep the frame pointer in a register for functions that
4166 don't need one. This avoids the instructions to save, set up and
4167 restore frame pointers; it also makes an extra register available
4168 in many functions. @strong{It also makes debugging impossible on
4171 On some machines, such as the VAX, this flag has no effect, because
4172 the standard calling sequence automatically handles the frame pointer
4173 and nothing is saved by pretending it doesn't exist. The
4174 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4175 whether a target machine supports this flag. @xref{Registers,,Register
4176 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4178 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4180 @item -foptimize-sibling-calls
4181 @opindex foptimize-sibling-calls
4182 Optimize sibling and tail recursive calls.
4184 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4188 Don't pay attention to the @code{inline} keyword. Normally this option
4189 is used to keep the compiler from expanding any functions inline.
4190 Note that if you are not optimizing, no functions can be expanded inline.
4192 @item -finline-functions
4193 @opindex finline-functions
4194 Integrate all simple functions into their callers. The compiler
4195 heuristically decides which functions are simple enough to be worth
4196 integrating in this way.
4198 If all calls to a given function are integrated, and the function is
4199 declared @code{static}, then the function is normally not output as
4200 assembler code in its own right.
4202 Enabled at level @option{-O3}.
4204 @item -finline-limit=@var{n}
4205 @opindex finline-limit
4206 By default, GCC limits the size of functions that can be inlined. This flag
4207 allows the control of this limit for functions that are explicitly marked as
4208 inline (i.e., marked with the inline keyword or defined within the class
4209 definition in c++). @var{n} is the size of functions that can be inlined in
4210 number of pseudo instructions (not counting parameter handling). The default
4211 value of @var{n} is 600.
4212 Increasing this value can result in more inlined code at
4213 the cost of compilation time and memory consumption. Decreasing usually makes
4214 the compilation faster and less code will be inlined (which presumably
4215 means slower programs). This option is particularly useful for programs that
4216 use inlining heavily such as those based on recursive templates with C++.
4218 Inlining is actually controlled by a number of parameters, which may be
4219 specified individually by using @option{--param @var{name}=@var{value}}.
4220 The @option{-finline-limit=@var{n}} option sets some of these parameters
4224 @item max-inline-insns-single
4225 is set to @var{n}/2.
4226 @item max-inline-insns-auto
4227 is set to @var{n}/2.
4228 @item min-inline-insns
4229 is set to 130 or @var{n}/4, whichever is smaller.
4230 @item max-inline-insns-rtl
4234 See below for a documentation of the individual
4235 parameters controlling inlining.
4237 @emph{Note:} pseudo instruction represents, in this particular context, an
4238 abstract measurement of function's size. In no way, it represents a count
4239 of assembly instructions and as such its exact meaning might change from one
4240 release to an another.
4242 @item -fkeep-inline-functions
4243 @opindex fkeep-inline-functions
4244 In C, emit @code{static} functions that are declared @code{inline}
4245 into the object file, even if the function has been inlined into all
4246 of its callers. This switch does not affect functions using the
4247 @code{extern inline} extension in GNU C@. In C++, emit any and all
4248 inline functions into the object file.
4250 @item -fkeep-static-consts
4251 @opindex fkeep-static-consts
4252 Emit variables declared @code{static const} when optimization isn't turned
4253 on, even if the variables aren't referenced.
4255 GCC enables this option by default. If you want to force the compiler to
4256 check if the variable was referenced, regardless of whether or not
4257 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4259 @item -fmerge-constants
4260 Attempt to merge identical constants (string constants and floating point
4261 constants) across compilation units.
4263 This option is the default for optimized compilation if the assembler and
4264 linker support it. Use @option{-fno-merge-constants} to inhibit this
4267 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4269 @item -fmerge-all-constants
4270 Attempt to merge identical constants and identical variables.
4272 This option implies @option{-fmerge-constants}. In addition to
4273 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4274 arrays or initialized constant variables with integral or floating point
4275 types. Languages like C or C++ require each non-automatic variable to
4276 have distinct location, so using this option will result in non-conforming
4279 @item -fmodulo-sched
4280 @opindex fmodulo-sched
4281 Perform swing modulo scheduling immediately before the first scheduling
4282 pass. This pass looks at innermost loops and reorders their
4283 instructions by overlapping different iterations.
4285 @item -fno-branch-count-reg
4286 @opindex fno-branch-count-reg
4287 Do not use ``decrement and branch'' instructions on a count register,
4288 but instead generate a sequence of instructions that decrement a
4289 register, compare it against zero, then branch based upon the result.
4290 This option is only meaningful on architectures that support such
4291 instructions, which include x86, PowerPC, IA-64 and S/390.
4293 The default is @option{-fbranch-count-reg}, enabled when
4294 @option{-fstrength-reduce} is enabled.
4296 @item -fno-function-cse
4297 @opindex fno-function-cse
4298 Do not put function addresses in registers; make each instruction that
4299 calls a constant function contain the function's address explicitly.
4301 This option results in less efficient code, but some strange hacks
4302 that alter the assembler output may be confused by the optimizations
4303 performed when this option is not used.
4305 The default is @option{-ffunction-cse}
4307 @item -fno-zero-initialized-in-bss
4308 @opindex fno-zero-initialized-in-bss
4309 If the target supports a BSS section, GCC by default puts variables that
4310 are initialized to zero into BSS@. This can save space in the resulting
4313 This option turns off this behavior because some programs explicitly
4314 rely on variables going to the data section. E.g., so that the
4315 resulting executable can find the beginning of that section and/or make
4316 assumptions based on that.
4318 The default is @option{-fzero-initialized-in-bss}.
4320 @item -fbounds-check
4321 @opindex fbounds-check
4322 For front-ends that support it, generate additional code to check that
4323 indices used to access arrays are within the declared range. This is
4324 currently only supported by the Java and Fortran front-ends, where
4325 this option defaults to true and false respectively.
4327 @item -fmudflap -fmudflapth -fmudflapir
4331 @cindex bounds checking
4333 For front-ends that support it (C and C++), instrument all risky
4334 pointer/array dereferencing operations, some standard library
4335 string/heap functions, and some other associated constructs with
4336 range/validity tests. Modules so instrumented should be immune to
4337 buffer overflows, invalid heap use, and some other classes of C/C++
4338 programming errors. The instrumentation relies on a separate runtime
4339 library (@file{libmudflap}), which will be linked into a program if
4340 @option{-fmudflap} is given at link time. Run-time behavior of the
4341 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4342 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4345 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4346 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4347 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4348 instrumentation should ignore pointer reads. This produces less
4349 instrumentation (and therefore faster execution) and still provides
4350 some protection against outright memory corrupting writes, but allows
4351 erroneously read data to propagate within a program.
4353 @item -fstrength-reduce
4354 @opindex fstrength-reduce
4355 Perform the optimizations of loop strength reduction and
4356 elimination of iteration variables.
4358 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4360 @item -fthread-jumps
4361 @opindex fthread-jumps
4362 Perform optimizations where we check to see if a jump branches to a
4363 location where another comparison subsumed by the first is found. If
4364 so, the first branch is redirected to either the destination of the
4365 second branch or a point immediately following it, depending on whether
4366 the condition is known to be true or false.
4368 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4370 @item -fcse-follow-jumps
4371 @opindex fcse-follow-jumps
4372 In common subexpression elimination, scan through jump instructions
4373 when the target of the jump is not reached by any other path. For
4374 example, when CSE encounters an @code{if} statement with an
4375 @code{else} clause, CSE will follow the jump when the condition
4378 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4380 @item -fcse-skip-blocks
4381 @opindex fcse-skip-blocks
4382 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4383 follow jumps which conditionally skip over blocks. When CSE
4384 encounters a simple @code{if} statement with no else clause,
4385 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4386 body of the @code{if}.
4388 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4390 @item -frerun-cse-after-loop
4391 @opindex frerun-cse-after-loop
4392 Re-run common subexpression elimination after loop optimizations has been
4395 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4397 @item -frerun-loop-opt
4398 @opindex frerun-loop-opt
4399 Run the loop optimizer twice.
4401 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4405 Perform a global common subexpression elimination pass.
4406 This pass also performs global constant and copy propagation.
4408 @emph{Note:} When compiling a program using computed gotos, a GCC
4409 extension, you may get better runtime performance if you disable
4410 the global common subexpression elimination pass by adding
4411 @option{-fno-gcse} to the command line.
4413 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4417 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4418 attempt to move loads which are only killed by stores into themselves. This
4419 allows a loop containing a load/store sequence to be changed to a load outside
4420 the loop, and a copy/store within the loop.
4422 Enabled by default when gcse is enabled.
4426 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4427 global common subexpression elimination. This pass will attempt to move
4428 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4429 loops containing a load/store sequence can be changed to a load before
4430 the loop and a store after the loop.
4432 Enabled by default when gcse is enabled.
4436 When @option{-fgcse-las} is enabled, the global common subexpression
4437 elimination pass eliminates redundant loads that come after stores to the
4438 same memory location (both partial and full redundancies).
4440 Enabled by default when gcse is enabled.
4442 @item -fgcse-after-reload
4443 @opindex fgcse-after-reload
4444 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4445 pass is performed after reload. The purpose of this pass is to cleanup
4448 @item -floop-optimize
4449 @opindex floop-optimize
4450 Perform loop optimizations: move constant expressions out of loops, simplify
4451 exit test conditions and optionally do strength-reduction as well.
4453 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4455 @item -floop-optimize2
4456 @opindex floop-optimize2
4457 Perform loop optimizations using the new loop optimizer. The optimizations
4458 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4461 @item -fcrossjumping
4462 @opindex crossjumping
4463 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4464 resulting code may or may not perform better than without cross-jumping.
4466 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4468 @item -fif-conversion
4469 @opindex if-conversion
4470 Attempt to transform conditional jumps into branch-less equivalents. This
4471 include use of conditional moves, min, max, set flags and abs instructions, and
4472 some tricks doable by standard arithmetics. The use of conditional execution
4473 on chips where it is available is controlled by @code{if-conversion2}.
4475 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4477 @item -fif-conversion2
4478 @opindex if-conversion2
4479 Use conditional execution (where available) to transform conditional jumps into
4480 branch-less equivalents.
4482 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4484 @item -fdelete-null-pointer-checks
4485 @opindex fdelete-null-pointer-checks
4486 Use global dataflow analysis to identify and eliminate useless checks
4487 for null pointers. The compiler assumes that dereferencing a null
4488 pointer would have halted the program. If a pointer is checked after
4489 it has already been dereferenced, it cannot be null.
4491 In some environments, this assumption is not true, and programs can
4492 safely dereference null pointers. Use
4493 @option{-fno-delete-null-pointer-checks} to disable this optimization
4494 for programs which depend on that behavior.
4496 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4498 @item -fexpensive-optimizations
4499 @opindex fexpensive-optimizations
4500 Perform a number of minor optimizations that are relatively expensive.
4502 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4504 @item -foptimize-register-move
4506 @opindex foptimize-register-move
4508 Attempt to reassign register numbers in move instructions and as
4509 operands of other simple instructions in order to maximize the amount of
4510 register tying. This is especially helpful on machines with two-operand
4513 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4516 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4518 @item -fdelayed-branch
4519 @opindex fdelayed-branch
4520 If supported for the target machine, attempt to reorder instructions
4521 to exploit instruction slots available after delayed branch
4524 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4526 @item -fschedule-insns
4527 @opindex fschedule-insns
4528 If supported for the target machine, attempt to reorder instructions to
4529 eliminate execution stalls due to required data being unavailable. This
4530 helps machines that have slow floating point or memory load instructions
4531 by allowing other instructions to be issued until the result of the load
4532 or floating point instruction is required.
4534 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4536 @item -fschedule-insns2
4537 @opindex fschedule-insns2
4538 Similar to @option{-fschedule-insns}, but requests an additional pass of
4539 instruction scheduling after register allocation has been done. This is
4540 especially useful on machines with a relatively small number of
4541 registers and where memory load instructions take more than one cycle.
4543 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4545 @item -fno-sched-interblock
4546 @opindex fno-sched-interblock
4547 Don't schedule instructions across basic blocks. This is normally
4548 enabled by default when scheduling before register allocation, i.e.@:
4549 with @option{-fschedule-insns} or at @option{-O2} or higher.
4551 @item -fno-sched-spec
4552 @opindex fno-sched-spec
4553 Don't allow speculative motion of non-load instructions. This is normally
4554 enabled by default when scheduling before register allocation, i.e.@:
4555 with @option{-fschedule-insns} or at @option{-O2} or higher.
4557 @item -fsched-spec-load
4558 @opindex fsched-spec-load
4559 Allow speculative motion of some load instructions. This only makes
4560 sense when scheduling before register allocation, i.e.@: with
4561 @option{-fschedule-insns} or at @option{-O2} or higher.
4563 @item -fsched-spec-load-dangerous
4564 @opindex fsched-spec-load-dangerous
4565 Allow speculative motion of more load instructions. This only makes
4566 sense when scheduling before register allocation, i.e.@: with
4567 @option{-fschedule-insns} or at @option{-O2} or higher.
4569 @item -fsched-stalled-insns=@var{n}
4570 @opindex fsched-stalled-insns
4571 Define how many insns (if any) can be moved prematurely from the queue
4572 of stalled insns into the ready list, during the second scheduling pass.
4574 @item -fsched-stalled-insns-dep=@var{n}
4575 @opindex fsched-stalled-insns-dep
4576 Define how many insn groups (cycles) will be examined for a dependency
4577 on a stalled insn that is candidate for premature removal from the queue
4578 of stalled insns. Has an effect only during the second scheduling pass,
4579 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4581 @item -fsched2-use-superblocks
4582 @opindex fsched2-use-superblocks
4583 When scheduling after register allocation, do use superblock scheduling
4584 algorithm. Superblock scheduling allows motion across basic block boundaries
4585 resulting on faster schedules. This option is experimental, as not all machine
4586 descriptions used by GCC model the CPU closely enough to avoid unreliable
4587 results from the algorithm.
4589 This only makes sense when scheduling after register allocation, i.e.@: with
4590 @option{-fschedule-insns2} or at @option{-O2} or higher.
4592 @item -fsched2-use-traces
4593 @opindex fsched2-use-traces
4594 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4595 allocation and additionally perform code duplication in order to increase the
4596 size of superblocks using tracer pass. See @option{-ftracer} for details on
4599 This mode should produce faster but significantly longer programs. Also
4600 without @option{-fbranch-probabilities} the traces constructed may not
4601 match the reality and hurt the performance. This only makes
4602 sense when scheduling after register allocation, i.e.@: with
4603 @option{-fschedule-insns2} or at @option{-O2} or higher.
4605 @item -freschedule-modulo-scheduled-loops
4606 @opindex fscheduling-in-modulo-scheduled-loops
4607 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4608 we may want to prevent the later scheduling passes from changing its schedule, we use this
4609 option to control that.
4611 @item -fcaller-saves
4612 @opindex fcaller-saves
4613 Enable values to be allocated in registers that will be clobbered by
4614 function calls, by emitting extra instructions to save and restore the
4615 registers around such calls. Such allocation is done only when it
4616 seems to result in better code than would otherwise be produced.
4618 This option is always enabled by default on certain machines, usually
4619 those which have no call-preserved registers to use instead.
4621 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4624 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4625 enabled by default at @option{-O} and higher.
4628 Perform Full Redundancy Elimination (FRE) on trees. The difference
4629 between FRE and PRE is that FRE only considers expressions
4630 that are computed on all paths leading to the redundant computation.
4631 This analysis faster than PRE, though it exposes fewer redundancies.
4632 This flag is enabled by default at @option{-O} and higher.
4635 Perform sparse conditional constant propagation (CCP) on trees. This flag
4636 is enabled by default at @option{-O} and higher.
4639 Perform dead code elimination (DCE) on trees. This flag is enabled by
4640 default at @option{-O} and higher.
4642 @item -ftree-dominator-opts
4643 Perform dead code elimination (DCE) on trees. This flag is enabled by
4644 default at @option{-O} and higher.
4647 Perform loop header copying on trees. This is beneficial since it increases
4648 effectiveness of code motion optimizations. It also saves one jump. This flag
4649 is enabled by default at @option{-O} and higher. It is not enabled
4650 for @option{-Os}, since it usually increases code size.
4652 @item -ftree-loop-optimize
4653 Perform loop optimizations on trees. This flag is enabled by default
4654 at @option{-O} and higher.
4656 @item -ftree-loop-linear
4657 Perform linear loop transformations on tree. This flag can improve cache
4658 performance and allow further loop optimizations to take place.
4661 Perform loop invariant motion on trees. This pass moves only invartiants that
4662 would be hard to handle on rtl level (function calls, operations that expand to
4663 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
4664 operands of conditions that are invariant out of the loop, so that we can use
4665 just trivial invariantness analysis in loop unswitching. The pass also includes
4669 Create a canonical counter for number of iterations in the loop for that
4670 determining number of iterations requires complicated analysis. Later
4671 optimizations then may determine the number easily. Useful especially
4672 in connection with unrolling.
4675 Perform induction variable optimizations (strength reduction, induction
4676 variable merging and induction variable elimination) on trees.
4679 Perform scalar replacement of aggregates. This pass replaces structure
4680 references with scalars to prevent committing structures to memory too
4681 early. This flag is enabled by default at @option{-O} and higher.
4683 @item -ftree-copyrename
4684 Perform copy renaming on trees. This pass attempts to rename compiler
4685 temporaries to other variables at copy locations, usually resulting in
4686 variable names which more closely resemble the original variables. This flag
4687 is enabled by default at @option{-O} and higher.
4690 Perform temporary expression replacement during the SSA->normal phase. Single
4691 use/single def temporaries are replaced at their use location with their
4692 defining expression. This results in non-GIMPLE code, but gives the expanders
4693 much more complex trees to work on resulting in better RTL generation. This is
4694 enabled by default at @option{-O} and higher.
4697 Perform live range splitting during the SSA->normal phase. Distinct live
4698 ranges of a variable are split into unique variables, allowing for better
4699 optimization later. This is enabled by default at @option{-O} and higher.
4701 @item -ftree-vectorize
4702 Perform loop vectorization on trees.
4706 Perform tail duplication to enlarge superblock size. This transformation
4707 simplifies the control flow of the function allowing other optimizations to do
4710 @item -funroll-loops
4711 @opindex funroll-loops
4712 Unroll loops whose number of iterations can be determined at compile
4713 time or upon entry to the loop. @option{-funroll-loops} implies both
4714 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4715 option makes code larger, and may or may not make it run faster.
4717 @item -funroll-all-loops
4718 @opindex funroll-all-loops
4719 Unroll all loops, even if their number of iterations is uncertain when
4720 the loop is entered. This usually makes programs run more slowly.
4721 @option{-funroll-all-loops} implies the same options as
4722 @option{-funroll-loops},
4724 @item -fsplit-ivs-in-unroller
4725 @opindex -fsplit-ivs-in-unroller
4726 Enables expressing of values of induction variables in later iterations
4727 of the unrolled loop using the value in the first iteration. This breaks
4728 long dependency chains, thus improving efficiency of the scheduling passes
4729 (for best results, @option{-fweb} should be used as well).
4731 Combination of @option{-fweb} and CSE is often sufficient to obtain the
4732 same effect. However in cases the loop body is more complicated than
4733 a single basic block, this is not reliable. It also does not work at all
4734 on some of the architectures due to restrictions in the CSE pass.
4736 This optimization is enabled by default.
4738 @item -fvariable-expansion-in-unroller
4739 @opindex -fvariable-expansion-in-unroller
4740 With this option, the compiler will create multiple copies of some
4741 local variables when unrolling a loop which can result in superior code.
4743 @item -fprefetch-loop-arrays
4744 @opindex fprefetch-loop-arrays
4745 If supported by the target machine, generate instructions to prefetch
4746 memory to improve the performance of loops that access large arrays.
4748 These options may generate better or worse code; results are highly
4749 dependent on the structure of loops within the source code.
4752 @itemx -fno-peephole2
4753 @opindex fno-peephole
4754 @opindex fno-peephole2
4755 Disable any machine-specific peephole optimizations. The difference
4756 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4757 are implemented in the compiler; some targets use one, some use the
4758 other, a few use both.
4760 @option{-fpeephole} is enabled by default.
4761 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4763 @item -fno-guess-branch-probability
4764 @opindex fno-guess-branch-probability
4765 Do not guess branch probabilities using heuristics.
4767 GCC will use heuristics to guess branch probabilities if they are
4768 not provided by profiling feedback (@option{-fprofile-arcs}). These
4769 heuristics are based on the control flow graph. If some branch probabilities
4770 are specified by @samp{__builtin_expect}, then the heuristics will be
4771 used to guess branch probabilities for the rest of the control flow graph,
4772 taking the @samp{__builtin_expect} info into account. The interactions
4773 between the heuristics and @samp{__builtin_expect} can be complex, and in
4774 some cases, it may be useful to disable the heuristics so that the effects
4775 of @samp{__builtin_expect} are easier to understand.
4777 The default is @option{-fguess-branch-probability} at levels
4778 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4780 @item -freorder-blocks
4781 @opindex freorder-blocks
4782 Reorder basic blocks in the compiled function in order to reduce number of
4783 taken branches and improve code locality.
4785 Enabled at levels @option{-O2}, @option{-O3}.
4787 @item -freorder-blocks-and-partition
4788 @opindex freorder-blocks-and-partition
4789 In addition to reordering basic blocks in the compiled function, in order
4790 to reduce number of taken branches, partitions hot and cold basic blocks
4791 into separate sections of the assembly and .o files, to improve
4792 paging and cache locality performance.
4794 This optimization is automatically turned off in the presence of
4795 exception handling, for linkonce sections, for functions with a user-defined
4796 section attribute and on any architecture that does not support named
4799 @item -freorder-functions
4800 @opindex freorder-functions
4801 Reorder basic blocks in the compiled function in order to reduce number of
4802 taken branches and improve code locality. This is implemented by using special
4803 subsections @code{.text.hot} for most frequently executed functions and
4804 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4805 the linker so object file format must support named sections and linker must
4806 place them in a reasonable way.
4808 Also profile feedback must be available in to make this option effective. See
4809 @option{-fprofile-arcs} for details.
4811 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4813 @item -fstrict-aliasing
4814 @opindex fstrict-aliasing
4815 Allows the compiler to assume the strictest aliasing rules applicable to
4816 the language being compiled. For C (and C++), this activates
4817 optimizations based on the type of expressions. In particular, an
4818 object of one type is assumed never to reside at the same address as an
4819 object of a different type, unless the types are almost the same. For
4820 example, an @code{unsigned int} can alias an @code{int}, but not a
4821 @code{void*} or a @code{double}. A character type may alias any other
4824 Pay special attention to code like this:
4837 The practice of reading from a different union member than the one most
4838 recently written to (called ``type-punning'') is common. Even with
4839 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4840 is accessed through the union type. So, the code above will work as
4841 expected. However, this code might not:
4852 Every language that wishes to perform language-specific alias analysis
4853 should define a function that computes, given an @code{tree}
4854 node, an alias set for the node. Nodes in different alias sets are not
4855 allowed to alias. For an example, see the C front-end function
4856 @code{c_get_alias_set}.
4858 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4860 @item -falign-functions
4861 @itemx -falign-functions=@var{n}
4862 @opindex falign-functions
4863 Align the start of functions to the next power-of-two greater than
4864 @var{n}, skipping up to @var{n} bytes. For instance,
4865 @option{-falign-functions=32} aligns functions to the next 32-byte
4866 boundary, but @option{-falign-functions=24} would align to the next
4867 32-byte boundary only if this can be done by skipping 23 bytes or less.
4869 @option{-fno-align-functions} and @option{-falign-functions=1} are
4870 equivalent and mean that functions will not be aligned.
4872 Some assemblers only support this flag when @var{n} is a power of two;
4873 in that case, it is rounded up.
4875 If @var{n} is not specified or is zero, use a machine-dependent default.
4877 Enabled at levels @option{-O2}, @option{-O3}.
4879 @item -falign-labels
4880 @itemx -falign-labels=@var{n}
4881 @opindex falign-labels
4882 Align all branch targets to a power-of-two boundary, skipping up to
4883 @var{n} bytes like @option{-falign-functions}. This option can easily
4884 make code slower, because it must insert dummy operations for when the
4885 branch target is reached in the usual flow of the code.
4887 @option{-fno-align-labels} and @option{-falign-labels=1} are
4888 equivalent and mean that labels will not be aligned.
4890 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4891 are greater than this value, then their values are used instead.
4893 If @var{n} is not specified or is zero, use a machine-dependent default
4894 which is very likely to be @samp{1}, meaning no alignment.
4896 Enabled at levels @option{-O2}, @option{-O3}.
4899 @itemx -falign-loops=@var{n}
4900 @opindex falign-loops
4901 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4902 like @option{-falign-functions}. The hope is that the loop will be
4903 executed many times, which will make up for any execution of the dummy
4906 @option{-fno-align-loops} and @option{-falign-loops=1} are
4907 equivalent and mean that loops will not be aligned.
4909 If @var{n} is not specified or is zero, use a machine-dependent default.
4911 Enabled at levels @option{-O2}, @option{-O3}.
4914 @itemx -falign-jumps=@var{n}
4915 @opindex falign-jumps
4916 Align branch targets to a power-of-two boundary, for branch targets
4917 where the targets can only be reached by jumping, skipping up to @var{n}
4918 bytes like @option{-falign-functions}. In this case, no dummy operations
4921 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4922 equivalent and mean that loops will not be aligned.
4924 If @var{n} is not specified or is zero, use a machine-dependent default.
4926 Enabled at levels @option{-O2}, @option{-O3}.
4928 @item -funit-at-a-time
4929 @opindex funit-at-a-time
4930 Parse the whole compilation unit before starting to produce code.
4931 This allows some extra optimizations to take place but consumes
4932 more memory (in general). There are some compatibility issues
4933 with @emph{unit-at-at-time} mode:
4936 enabling @emph{unit-at-a-time} mode may change the order
4937 in which functions, variables, and top-level @code{asm} statements
4938 are emitted, and will likely break code relying on some particular
4939 ordering. The majority of such top-level @code{asm} statements,
4940 though, can be replaced by @code{section} attributes.
4943 @emph{unit-at-a-time} mode removes unreferenced static variables
4944 and functions are removed. This may result in undefined references
4945 when an @code{asm} statement refers directly to variables or functions
4946 that are otherwise unused. In that case either the variable/function
4947 shall be listed as an operand of the @code{asm} statement operand or,
4948 in the case of top-level @code{asm} statements the attribute @code{used}
4949 shall be used on the declaration.
4952 Static functions now can use non-standard passing conventions that
4953 may break @code{asm} statements calling functions directly. Again,
4954 attribute @code{used} will prevent this behavior.
4957 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
4958 but this scheme may not be supported by future releases of GCC@.
4960 Enabled at levels @option{-O2}, @option{-O3}.
4964 Constructs webs as commonly used for register allocation purposes and assign
4965 each web individual pseudo register. This allows the register allocation pass
4966 to operate on pseudos directly, but also strengthens several other optimization
4967 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4968 however, make debugging impossible, since variables will no longer stay in a
4971 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
4972 on targets where the default format for debugging information supports
4975 @item -fno-cprop-registers
4976 @opindex fno-cprop-registers
4977 After register allocation and post-register allocation instruction splitting,
4978 we perform a copy-propagation pass to try to reduce scheduling dependencies
4979 and occasionally eliminate the copy.
4981 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4983 @item -fprofile-generate
4984 @opindex fprofile-generate
4986 Enable options usually used for instrumenting application to produce
4987 profile useful for later recompilation with profile feedback based
4988 optimization. You must use @option{-fprofile-generate} both when
4989 compiling and when linking your program.
4991 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
4994 @opindex fprofile-use
4995 Enable profile feedback directed optimizations, and optimizations
4996 generally profitable only with profile feedback available.
4998 The following options are enabled: @code{-fbranch-probabilities},
4999 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
5003 The following options control compiler behavior regarding floating
5004 point arithmetic. These options trade off between speed and
5005 correctness. All must be specifically enabled.
5009 @opindex ffloat-store
5010 Do not store floating point variables in registers, and inhibit other
5011 options that might change whether a floating point value is taken from a
5014 @cindex floating point precision
5015 This option prevents undesirable excess precision on machines such as
5016 the 68000 where the floating registers (of the 68881) keep more
5017 precision than a @code{double} is supposed to have. Similarly for the
5018 x86 architecture. For most programs, the excess precision does only
5019 good, but a few programs rely on the precise definition of IEEE floating
5020 point. Use @option{-ffloat-store} for such programs, after modifying
5021 them to store all pertinent intermediate computations into variables.
5025 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5026 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5027 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
5029 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5031 This option should never be turned on by any @option{-O} option since
5032 it can result in incorrect output for programs which depend on
5033 an exact implementation of IEEE or ISO rules/specifications for
5036 @item -fno-math-errno
5037 @opindex fno-math-errno
5038 Do not set ERRNO after calling math functions that are executed
5039 with a single instruction, e.g., sqrt. A program that relies on
5040 IEEE exceptions for math error handling may want to use this flag
5041 for speed while maintaining IEEE arithmetic compatibility.
5043 This option should never be turned on by any @option{-O} option since
5044 it can result in incorrect output for programs which depend on
5045 an exact implementation of IEEE or ISO rules/specifications for
5048 The default is @option{-fmath-errno}.
5050 @item -funsafe-math-optimizations
5051 @opindex funsafe-math-optimizations
5052 Allow optimizations for floating-point arithmetic that (a) assume
5053 that arguments and results are valid and (b) may violate IEEE or
5054 ANSI standards. When used at link-time, it may include libraries
5055 or startup files that change the default FPU control word or other
5056 similar optimizations.
5058 This option should never be turned on by any @option{-O} option since
5059 it can result in incorrect output for programs which depend on
5060 an exact implementation of IEEE or ISO rules/specifications for
5063 The default is @option{-fno-unsafe-math-optimizations}.
5065 @item -ffinite-math-only
5066 @opindex ffinite-math-only
5067 Allow optimizations for floating-point arithmetic that assume
5068 that arguments and results are not NaNs or +-Infs.
5070 This option should never be turned on by any @option{-O} option since
5071 it can result in incorrect output for programs which depend on
5072 an exact implementation of IEEE or ISO rules/specifications.
5074 The default is @option{-fno-finite-math-only}.
5076 @item -fno-trapping-math
5077 @opindex fno-trapping-math
5078 Compile code assuming that floating-point operations cannot generate
5079 user-visible traps. These traps include division by zero, overflow,
5080 underflow, inexact result and invalid operation. This option implies
5081 @option{-fno-signaling-nans}. Setting this option may allow faster
5082 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5084 This option should never be turned on by any @option{-O} option since
5085 it can result in incorrect output for programs which depend on
5086 an exact implementation of IEEE or ISO rules/specifications for
5089 The default is @option{-ftrapping-math}.
5091 @item -frounding-math
5092 @opindex frounding-math
5093 Disable transformations and optimizations that assume default floating
5094 point rounding behavior. This is round-to-zero for all floating point
5095 to integer conversions, and round-to-nearest for all other arithmetic
5096 truncations. This option should be specified for programs that change
5097 the FP rounding mode dynamically, or that may be executed with a
5098 non-default rounding mode. This option disables constant folding of
5099 floating point expressions at compile-time (which may be affected by
5100 rounding mode) and arithmetic transformations that are unsafe in the
5101 presence of sign-dependent rounding modes.
5103 The default is @option{-fno-rounding-math}.
5105 This option is experimental and does not currently guarantee to
5106 disable all GCC optimizations that are affected by rounding mode.
5107 Future versions of GCC may provide finer control of this setting
5108 using C99's @code{FENV_ACCESS} pragma. This command line option
5109 will be used to specify the default state for @code{FENV_ACCESS}.
5111 @item -fsignaling-nans
5112 @opindex fsignaling-nans
5113 Compile code assuming that IEEE signaling NaNs may generate user-visible
5114 traps during floating-point operations. Setting this option disables
5115 optimizations that may change the number of exceptions visible with
5116 signaling NaNs. This option implies @option{-ftrapping-math}.
5118 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5121 The default is @option{-fno-signaling-nans}.
5123 This option is experimental and does not currently guarantee to
5124 disable all GCC optimizations that affect signaling NaN behavior.
5126 @item -fsingle-precision-constant
5127 @opindex fsingle-precision-constant
5128 Treat floating point constant as single precision constant instead of
5129 implicitly converting it to double precision constant.
5134 The following options control optimizations that may improve
5135 performance, but are not enabled by any @option{-O} options. This
5136 section includes experimental options that may produce broken code.
5139 @item -fbranch-probabilities
5140 @opindex fbranch-probabilities
5141 After running a program compiled with @option{-fprofile-arcs}
5142 (@pxref{Debugging Options,, Options for Debugging Your Program or
5143 @command{gcc}}), you can compile it a second time using
5144 @option{-fbranch-probabilities}, to improve optimizations based on
5145 the number of times each branch was taken. When the program
5146 compiled with @option{-fprofile-arcs} exits it saves arc execution
5147 counts to a file called @file{@var{sourcename}.gcda} for each source
5148 file The information in this data file is very dependent on the
5149 structure of the generated code, so you must use the same source code
5150 and the same optimization options for both compilations.
5152 With @option{-fbranch-probabilities}, GCC puts a
5153 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5154 These can be used to improve optimization. Currently, they are only
5155 used in one place: in @file{reorg.c}, instead of guessing which path a
5156 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5157 exactly determine which path is taken more often.
5159 @item -fprofile-values
5160 @opindex fprofile-values
5161 If combined with @option{-fprofile-arcs}, it adds code so that some
5162 data about values of expressions in the program is gathered.
5164 With @option{-fbranch-probabilities}, it reads back the data gathered
5165 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5166 notes to instructions for their later usage in optimizations.
5168 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5172 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5173 a code to gather information about values of expressions.
5175 With @option{-fbranch-probabilities}, it reads back the data gathered
5176 and actually performs the optimizations based on them.
5177 Currently the optimizations include specialization of division operation
5178 using the knowledge about the value of the denominator.
5180 @item -fspeculative-prefetching
5181 @opindex fspeculative-prefetching
5182 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5183 a code to gather information about addresses of memory references in the
5186 With @option{-fbranch-probabilities}, it reads back the data gathered
5187 and issues prefetch instructions according to them. In addition to the opportunities
5188 noticed by @option{-fprefetch-loop-arrays}, it also notices more complicated
5189 memory access patterns---for example accesses to the data stored in linked
5190 list whose elements are usually allocated sequentially.
5192 In order to prevent issuing double prefetches, usage of
5193 @option{-fspeculative-prefetching} implies @option{-fno-prefetch-loop-arrays}.
5195 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5197 @item -frename-registers
5198 @opindex frename-registers
5199 Attempt to avoid false dependencies in scheduled code by making use
5200 of registers left over after register allocation. This optimization
5201 will most benefit processors with lots of registers. Depending on the
5202 debug information format adopted by the target, however, it can
5203 make debugging impossible, since variables will no longer stay in
5204 a ``home register''.
5206 Not enabled by default at any level because it has known bugs.
5210 Perform tail duplication to enlarge superblock size. This transformation
5211 simplifies the control flow of the function allowing other optimizations to do
5214 Enabled with @option{-fprofile-use}.
5216 @item -funroll-loops
5217 @opindex funroll-loops
5218 Unroll loops whose number of iterations can be determined at compile time or
5219 upon entry to the loop. @option{-funroll-loops} implies
5220 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
5221 (i.e.@: complete removal of loops with small constant number of iterations).
5222 This option makes code larger, and may or may not make it run faster.
5224 Enabled with @option{-fprofile-use}.
5226 @item -funroll-all-loops
5227 @opindex funroll-all-loops
5228 Unroll all loops, even if their number of iterations is uncertain when
5229 the loop is entered. This usually makes programs run more slowly.
5230 @option{-funroll-all-loops} implies the same options as
5231 @option{-funroll-loops}.
5234 @opindex fpeel-loops
5235 Peels the loops for that there is enough information that they do not
5236 roll much (from profile feedback). It also turns on complete loop peeling
5237 (i.e.@: complete removal of loops with small constant number of iterations).
5239 Enabled with @option{-fprofile-use}.
5241 @item -fmove-loop-invariants
5242 @opindex fmove-loop-invariants
5243 Enables the loop invariant motion pass in the new loop optimizer. Enabled
5244 at level @option{-O1}
5246 @item -funswitch-loops
5247 @opindex funswitch-loops
5248 Move branches with loop invariant conditions out of the loop, with duplicates
5249 of the loop on both branches (modified according to result of the condition).
5251 @item -fprefetch-loop-arrays
5252 @opindex fprefetch-loop-arrays
5253 If supported by the target machine, generate instructions to prefetch
5254 memory to improve the performance of loops that access large arrays.
5256 Disabled at level @option{-Os}.
5258 @item -ffunction-sections
5259 @itemx -fdata-sections
5260 @opindex ffunction-sections
5261 @opindex fdata-sections
5262 Place each function or data item into its own section in the output
5263 file if the target supports arbitrary sections. The name of the
5264 function or the name of the data item determines the section's name
5267 Use these options on systems where the linker can perform optimizations
5268 to improve locality of reference in the instruction space. Most systems
5269 using the ELF object format and SPARC processors running Solaris 2 have
5270 linkers with such optimizations. AIX may have these optimizations in
5273 Only use these options when there are significant benefits from doing
5274 so. When you specify these options, the assembler and linker will
5275 create larger object and executable files and will also be slower.
5276 You will not be able to use @code{gprof} on all systems if you
5277 specify this option and you may have problems with debugging if
5278 you specify both this option and @option{-g}.
5280 @item -fbranch-target-load-optimize
5281 @opindex fbranch-target-load-optimize
5282 Perform branch target register load optimization before prologue / epilogue
5284 The use of target registers can typically be exposed only during reload,
5285 thus hoisting loads out of loops and doing inter-block scheduling needs
5286 a separate optimization pass.
5288 @item -fbranch-target-load-optimize2
5289 @opindex fbranch-target-load-optimize2
5290 Perform branch target register load optimization after prologue / epilogue
5293 @item -fbtr-bb-exclusive
5294 @opindex fbtr-bb-exclusive
5295 When performing branch target register load optimization, don't reuse
5296 branch target registers in within any basic block.
5298 @item --param @var{name}=@var{value}
5300 In some places, GCC uses various constants to control the amount of
5301 optimization that is done. For example, GCC will not inline functions
5302 that contain more that a certain number of instructions. You can
5303 control some of these constants on the command-line using the
5304 @option{--param} option.
5306 The names of specific parameters, and the meaning of the values, are
5307 tied to the internals of the compiler, and are subject to change
5308 without notice in future releases.
5310 In each case, the @var{value} is an integer. The allowable choices for
5311 @var{name} are given in the following table:
5314 @item sra-max-structure-size
5315 The maximum structure size, in bytes, at which the scalar replacement
5316 of aggregates (SRA) optimization will perform block copies. The
5317 default value, 0, implies that GCC will select the most appropriate
5320 @item sra-field-structure-ratio
5321 The threshold ratio (as a percentage) between instantiated fields and
5322 the complete structure size. We say that if the ratio of the number
5323 of bytes in instantiated fields to the number of bytes in the complete
5324 structure exceeds this parameter, then block copies are not used. The
5327 @item max-crossjump-edges
5328 The maximum number of incoming edges to consider for crossjumping.
5329 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5330 the number of edges incoming to each block. Increasing values mean
5331 more aggressive optimization, making the compile time increase with
5332 probably small improvement in executable size.
5334 @item min-crossjump-insns
5335 The minimum number of instructions which must be matched at the end
5336 of two blocks before crossjumping will be performed on them. This
5337 value is ignored in the case where all instructions in the block being
5338 crossjumped from are matched. The default value is 5.
5340 @item max-delay-slot-insn-search
5341 The maximum number of instructions to consider when looking for an
5342 instruction to fill a delay slot. If more than this arbitrary number of
5343 instructions is searched, the time savings from filling the delay slot
5344 will be minimal so stop searching. Increasing values mean more
5345 aggressive optimization, making the compile time increase with probably
5346 small improvement in executable run time.
5348 @item max-delay-slot-live-search
5349 When trying to fill delay slots, the maximum number of instructions to
5350 consider when searching for a block with valid live register
5351 information. Increasing this arbitrarily chosen value means more
5352 aggressive optimization, increasing the compile time. This parameter
5353 should be removed when the delay slot code is rewritten to maintain the
5356 @item max-gcse-memory
5357 The approximate maximum amount of memory that will be allocated in
5358 order to perform the global common subexpression elimination
5359 optimization. If more memory than specified is required, the
5360 optimization will not be done.
5362 @item max-gcse-passes
5363 The maximum number of passes of GCSE to run. The default is 1.
5365 @item max-pending-list-length
5366 The maximum number of pending dependencies scheduling will allow
5367 before flushing the current state and starting over. Large functions
5368 with few branches or calls can create excessively large lists which
5369 needlessly consume memory and resources.
5371 @item max-inline-insns-single
5372 Several parameters control the tree inliner used in gcc.
5373 This number sets the maximum number of instructions (counted in GCC's
5374 internal representation) in a single function that the tree inliner
5375 will consider for inlining. This only affects functions declared
5376 inline and methods implemented in a class declaration (C++).
5377 The default value is 500.
5379 @item max-inline-insns-auto
5380 When you use @option{-finline-functions} (included in @option{-O3}),
5381 a lot of functions that would otherwise not be considered for inlining
5382 by the compiler will be investigated. To those functions, a different
5383 (more restrictive) limit compared to functions declared inline can
5385 The default value is 120.
5387 @item large-function-insns
5388 The limit specifying really large functions. For functions larger than this
5389 limit after inlining inlining is constrained by
5390 @option{--param large-function-growth}. This parameter is useful primarily
5391 to avoid extreme compilation time caused by non-linear algorithms used by the
5393 This parameter is ignored when @option{-funit-at-a-time} is not used.
5394 The default value is 3000.
5396 @item large-function-growth
5397 Specifies maximal growth of large function caused by inlining in percents.
5398 This parameter is ignored when @option{-funit-at-a-time} is not used.
5399 The default value is 100 which limits large function growth to 2.0 times
5402 @item inline-unit-growth
5403 Specifies maximal overall growth of the compilation unit caused by inlining.
5404 This parameter is ignored when @option{-funit-at-a-time} is not used.
5405 The default value is 50 which limits unit growth to 1.5 times the original
5408 @item max-inline-insns-recursive
5409 @itemx max-inline-insns-recursive-auto
5410 Specifies maximum number of instructions out-of-line copy of self recursive inline
5411 function can grow into by performing recursive inlining.
5413 For functions declared inline @option{--param max-inline-insns-recursive} is
5414 taken into acount. For function not declared inline, recursive inlining
5415 happens only when @option{-finline-functions} (included in @option{-O3}) is
5416 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5417 default value is 500.
5419 @item max-inline-recursive-depth
5420 @itemx max-inline-recursive-depth-auto
5421 Specifies maximum recursion depth used by the recursive inlining.
5423 For functions declared inline @option{--param max-inline-recursive-depth} is
5424 taken into acount. For function not declared inline, recursive inlining
5425 happens only when @option{-finline-functions} (included in @option{-O3}) is
5426 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5427 default value is 500.
5429 @item max-unrolled-insns
5430 The maximum number of instructions that a loop should have if that loop
5431 is unrolled, and if the loop is unrolled, it determines how many times
5432 the loop code is unrolled.
5434 @item max-average-unrolled-insns
5435 The maximum number of instructions biased by probabilities of their execution
5436 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5437 it determines how many times the loop code is unrolled.
5439 @item max-unroll-times
5440 The maximum number of unrollings of a single loop.
5442 @item max-peeled-insns
5443 The maximum number of instructions that a loop should have if that loop
5444 is peeled, and if the loop is peeled, it determines how many times
5445 the loop code is peeled.
5447 @item max-peel-times
5448 The maximum number of peelings of a single loop.
5450 @item max-completely-peeled-insns
5451 The maximum number of insns of a completely peeled loop.
5453 @item max-completely-peel-times
5454 The maximum number of iterations of a loop to be suitable for complete peeling.
5456 @item max-unswitch-insns
5457 The maximum number of insns of an unswitched loop.
5459 @item max-unswitch-level
5460 The maximum number of branches unswitched in a single loop.
5463 The minimum cost of an expensive expression in the loop invariant motion.
5465 @item iv-consider-all-candidates-bound
5466 Bound on number of candidates for induction variables below that
5467 all candidates are considered for each use in induction variable
5468 optimizations. Only the most relevant candidates are considered
5469 if there are more candidates, to avoid quadratic time complexity.
5471 @item iv-max-considered-uses
5472 The induction variable optimizations give up on loops that contain more
5473 induction variable uses.
5475 @item iv-always-prune-cand-set-bound
5476 If number of candidates in the set is smaller than this value,
5477 we always try to remove unnecessary ivs from the set during its
5478 optimization when a new iv is added to the set.
5480 @item max-iterations-to-track
5482 The maximum number of iterations of a loop the brute force algorithm
5483 for analysis of # of iterations of the loop tries to evaluate.
5485 @item hot-bb-count-fraction
5486 Select fraction of the maximal count of repetitions of basic block in program
5487 given basic block needs to have to be considered hot.
5489 @item hot-bb-frequency-fraction
5490 Select fraction of the maximal frequency of executions of basic block in
5491 function given basic block needs to have to be considered hot
5493 @item tracer-dynamic-coverage
5494 @itemx tracer-dynamic-coverage-feedback
5496 This value is used to limit superblock formation once the given percentage of
5497 executed instructions is covered. This limits unnecessary code size
5500 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5501 feedback is available. The real profiles (as opposed to statically estimated
5502 ones) are much less balanced allowing the threshold to be larger value.
5504 @item tracer-max-code-growth
5505 Stop tail duplication once code growth has reached given percentage. This is
5506 rather hokey argument, as most of the duplicates will be eliminated later in
5507 cross jumping, so it may be set to much higher values than is the desired code
5510 @item tracer-min-branch-ratio
5512 Stop reverse growth when the reverse probability of best edge is less than this
5513 threshold (in percent).
5515 @item tracer-min-branch-ratio
5516 @itemx tracer-min-branch-ratio-feedback
5518 Stop forward growth if the best edge do have probability lower than this
5521 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5522 compilation for profile feedback and one for compilation without. The value
5523 for compilation with profile feedback needs to be more conservative (higher) in
5524 order to make tracer effective.
5526 @item max-cse-path-length
5528 Maximum number of basic blocks on path that cse considers. The default is 10.
5530 @item global-var-threshold
5532 Counts the number of function calls (@var{n}) and the number of
5533 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
5534 single artificial variable will be created to represent all the
5535 call-clobbered variables at function call sites. This artificial
5536 variable will then be made to alias every call-clobbered variable.
5537 (done as @code{int * size_t} on the host machine; beware overflow).
5539 @item max-aliased-vops
5541 Maximum number of virtual operands allowed to represent aliases
5542 before triggering the alias grouping heuristic. Alias grouping
5543 reduces compile times and memory consumption needed for aliasing at
5544 the expense of precision loss in alias information.
5546 @item ggc-min-expand
5548 GCC uses a garbage collector to manage its own memory allocation. This
5549 parameter specifies the minimum percentage by which the garbage
5550 collector's heap should be allowed to expand between collections.
5551 Tuning this may improve compilation speed; it has no effect on code
5554 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
5555 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
5556 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
5557 GCC is not able to calculate RAM on a particular platform, the lower
5558 bound of 30% is used. Setting this parameter and
5559 @option{ggc-min-heapsize} to zero causes a full collection to occur at
5560 every opportunity. This is extremely slow, but can be useful for
5563 @item ggc-min-heapsize
5565 Minimum size of the garbage collector's heap before it begins bothering
5566 to collect garbage. The first collection occurs after the heap expands
5567 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
5568 tuning this may improve compilation speed, and has no effect on code
5571 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
5572 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
5573 with a lower bound of 4096 (four megabytes) and an upper bound of
5574 131072 (128 megabytes). If GCC is not able to calculate RAM on a
5575 particular platform, the lower bound is used. Setting this parameter
5576 very large effectively disables garbage collection. Setting this
5577 parameter and @option{ggc-min-expand} to zero causes a full collection
5578 to occur at every opportunity.
5580 @item max-reload-search-insns
5581 The maximum number of instruction reload should look backward for equivalent
5582 register. Increasing values mean more aggressive optimization, making the
5583 compile time increase with probably slightly better performance. The default
5586 @item max-cselib-memory-location
5587 The maximum number of memory locations cselib should take into acount.
5588 Increasing values mean more aggressive optimization, making the compile time
5589 increase with probably slightly better performance. The default value is 500.
5591 @item reorder-blocks-duplicate
5592 @itemx reorder-blocks-duplicate-feedback
5594 Used by basic block reordering pass to decide whether to use unconditional
5595 branch or duplicate the code on its destination. Code is duplicated when its
5596 estimated size is smaller than this value multiplied by the estimated size of
5597 unconditional jump in the hot spots of the program.
5599 The @option{reorder-block-duplicate-feedback} is used only when profile
5600 feedback is available and may be set to higher values than
5601 @option{reorder-block-duplicate} since information about the hot spots is more
5604 @item max-sched-region-blocks
5605 The maximum number of blocks in a region to be considered for
5606 interblock scheduling. The default value is 10.
5608 @item max-sched-region-insns
5609 The maximum number of insns in a region to be considered for
5610 interblock scheduling. The default value is 100.
5612 @item integer-share-limit
5613 Small integer constants can use a shared data structure, reducing the
5614 compiler's memory usage and increasing its speed. This sets the maximum
5615 value of a shared integer constant's. The default value is 256.
5620 @node Preprocessor Options
5621 @section Options Controlling the Preprocessor
5622 @cindex preprocessor options
5623 @cindex options, preprocessor
5625 These options control the C preprocessor, which is run on each C source
5626 file before actual compilation.
5628 If you use the @option{-E} option, nothing is done except preprocessing.
5629 Some of these options make sense only together with @option{-E} because
5630 they cause the preprocessor output to be unsuitable for actual
5635 You can use @option{-Wp,@var{option}} to bypass the compiler driver
5636 and pass @var{option} directly through to the preprocessor. If
5637 @var{option} contains commas, it is split into multiple options at the
5638 commas. However, many options are modified, translated or interpreted
5639 by the compiler driver before being passed to the preprocessor, and
5640 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
5641 interface is undocumented and subject to change, so whenever possible
5642 you should avoid using @option{-Wp} and let the driver handle the
5645 @item -Xpreprocessor @var{option}
5646 @opindex preprocessor
5647 Pass @var{option} as an option to the preprocessor. You can use this to
5648 supply system-specific preprocessor options which GCC does not know how to
5651 If you want to pass an option that takes an argument, you must use
5652 @option{-Xpreprocessor} twice, once for the option and once for the argument.
5655 @include cppopts.texi
5657 @node Assembler Options
5658 @section Passing Options to the Assembler
5660 @c prevent bad page break with this line
5661 You can pass options to the assembler.
5664 @item -Wa,@var{option}
5666 Pass @var{option} as an option to the assembler. If @var{option}
5667 contains commas, it is split into multiple options at the commas.
5669 @item -Xassembler @var{option}
5671 Pass @var{option} as an option to the assembler. You can use this to
5672 supply system-specific assembler options which GCC does not know how to
5675 If you want to pass an option that takes an argument, you must use
5676 @option{-Xassembler} twice, once for the option and once for the argument.
5681 @section Options for Linking
5682 @cindex link options
5683 @cindex options, linking
5685 These options come into play when the compiler links object files into
5686 an executable output file. They are meaningless if the compiler is
5687 not doing a link step.
5691 @item @var{object-file-name}
5692 A file name that does not end in a special recognized suffix is
5693 considered to name an object file or library. (Object files are
5694 distinguished from libraries by the linker according to the file
5695 contents.) If linking is done, these object files are used as input
5704 If any of these options is used, then the linker is not run, and
5705 object file names should not be used as arguments. @xref{Overall
5709 @item -l@var{library}
5710 @itemx -l @var{library}
5712 Search the library named @var{library} when linking. (The second
5713 alternative with the library as a separate argument is only for
5714 POSIX compliance and is not recommended.)
5716 It makes a difference where in the command you write this option; the
5717 linker searches and processes libraries and object files in the order they
5718 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5719 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5720 to functions in @samp{z}, those functions may not be loaded.
5722 The linker searches a standard list of directories for the library,
5723 which is actually a file named @file{lib@var{library}.a}. The linker
5724 then uses this file as if it had been specified precisely by name.
5726 The directories searched include several standard system directories
5727 plus any that you specify with @option{-L}.
5729 Normally the files found this way are library files---archive files
5730 whose members are object files. The linker handles an archive file by
5731 scanning through it for members which define symbols that have so far
5732 been referenced but not defined. But if the file that is found is an
5733 ordinary object file, it is linked in the usual fashion. The only
5734 difference between using an @option{-l} option and specifying a file name
5735 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5736 and searches several directories.
5740 You need this special case of the @option{-l} option in order to
5741 link an Objective-C or Objective-C++ program.
5744 @opindex nostartfiles
5745 Do not use the standard system startup files when linking.
5746 The standard system libraries are used normally, unless @option{-nostdlib}
5747 or @option{-nodefaultlibs} is used.
5749 @item -nodefaultlibs
5750 @opindex nodefaultlibs
5751 Do not use the standard system libraries when linking.
5752 Only the libraries you specify will be passed to the linker.
5753 The standard startup files are used normally, unless @option{-nostartfiles}
5754 is used. The compiler may generate calls to @code{memcmp},
5755 @code{memset}, @code{memcpy} and @code{memmove}.
5756 These entries are usually resolved by entries in
5757 libc. These entry points should be supplied through some other
5758 mechanism when this option is specified.
5762 Do not use the standard system startup files or libraries when linking.
5763 No startup files and only the libraries you specify will be passed to
5764 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
5765 @code{memcpy} and @code{memmove}.
5766 These entries are usually resolved by entries in
5767 libc. These entry points should be supplied through some other
5768 mechanism when this option is specified.
5770 @cindex @option{-lgcc}, use with @option{-nostdlib}
5771 @cindex @option{-nostdlib} and unresolved references
5772 @cindex unresolved references and @option{-nostdlib}
5773 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5774 @cindex @option{-nodefaultlibs} and unresolved references
5775 @cindex unresolved references and @option{-nodefaultlibs}
5776 One of the standard libraries bypassed by @option{-nostdlib} and
5777 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5778 that GCC uses to overcome shortcomings of particular machines, or special
5779 needs for some languages.
5780 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5781 Collection (GCC) Internals},
5782 for more discussion of @file{libgcc.a}.)
5783 In most cases, you need @file{libgcc.a} even when you want to avoid
5784 other standard libraries. In other words, when you specify @option{-nostdlib}
5785 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5786 This ensures that you have no unresolved references to internal GCC
5787 library subroutines. (For example, @samp{__main}, used to ensure C++
5788 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5789 GNU Compiler Collection (GCC) Internals}.)
5793 Produce a position independent executable on targets which support it.
5794 For predictable results, you must also specify the same set of options
5795 that were used to generate code (@option{-fpie}, @option{-fPIE},
5796 or model suboptions) when you specify this option.
5800 Remove all symbol table and relocation information from the executable.
5804 On systems that support dynamic linking, this prevents linking with the shared
5805 libraries. On other systems, this option has no effect.
5809 Produce a shared object which can then be linked with other objects to
5810 form an executable. Not all systems support this option. For predictable
5811 results, you must also specify the same set of options that were used to
5812 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5813 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5814 needs to build supplementary stub code for constructors to work. On
5815 multi-libbed systems, @samp{gcc -shared} must select the correct support
5816 libraries to link against. Failing to supply the correct flags may lead
5817 to subtle defects. Supplying them in cases where they are not necessary
5820 @item -shared-libgcc
5821 @itemx -static-libgcc
5822 @opindex shared-libgcc
5823 @opindex static-libgcc
5824 On systems that provide @file{libgcc} as a shared library, these options
5825 force the use of either the shared or static version respectively.
5826 If no shared version of @file{libgcc} was built when the compiler was
5827 configured, these options have no effect.
5829 There are several situations in which an application should use the
5830 shared @file{libgcc} instead of the static version. The most common
5831 of these is when the application wishes to throw and catch exceptions
5832 across different shared libraries. In that case, each of the libraries
5833 as well as the application itself should use the shared @file{libgcc}.
5835 Therefore, the G++ and GCJ drivers automatically add
5836 @option{-shared-libgcc} whenever you build a shared library or a main
5837 executable, because C++ and Java programs typically use exceptions, so
5838 this is the right thing to do.
5840 If, instead, you use the GCC driver to create shared libraries, you may
5841 find that they will not always be linked with the shared @file{libgcc}.
5842 If GCC finds, at its configuration time, that you have a non-GNU linker
5843 or a GNU linker that does not support option @option{--eh-frame-hdr},
5844 it will link the shared version of @file{libgcc} into shared libraries
5845 by default. Otherwise, it will take advantage of the linker and optimize
5846 away the linking with the shared version of @file{libgcc}, linking with
5847 the static version of libgcc by default. This allows exceptions to
5848 propagate through such shared libraries, without incurring relocation
5849 costs at library load time.
5851 However, if a library or main executable is supposed to throw or catch
5852 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5853 for the languages used in the program, or using the option
5854 @option{-shared-libgcc}, such that it is linked with the shared
5859 Bind references to global symbols when building a shared object. Warn
5860 about any unresolved references (unless overridden by the link editor
5861 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5864 @item -Xlinker @var{option}
5866 Pass @var{option} as an option to the linker. You can use this to
5867 supply system-specific linker options which GCC does not know how to
5870 If you want to pass an option that takes an argument, you must use
5871 @option{-Xlinker} twice, once for the option and once for the argument.
5872 For example, to pass @option{-assert definitions}, you must write
5873 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5874 @option{-Xlinker "-assert definitions"}, because this passes the entire
5875 string as a single argument, which is not what the linker expects.
5877 @item -Wl,@var{option}
5879 Pass @var{option} as an option to the linker. If @var{option} contains
5880 commas, it is split into multiple options at the commas.
5882 @item -u @var{symbol}
5884 Pretend the symbol @var{symbol} is undefined, to force linking of
5885 library modules to define it. You can use @option{-u} multiple times with
5886 different symbols to force loading of additional library modules.
5889 @node Directory Options
5890 @section Options for Directory Search
5891 @cindex directory options
5892 @cindex options, directory search
5895 These options specify directories to search for header files, for
5896 libraries and for parts of the compiler:
5901 Add the directory @var{dir} to the head of the list of directories to be
5902 searched for header files. This can be used to override a system header
5903 file, substituting your own version, since these directories are
5904 searched before the system header file directories. However, you should
5905 not use this option to add directories that contain vendor-supplied
5906 system header files (use @option{-isystem} for that). If you use more than
5907 one @option{-I} option, the directories are scanned in left-to-right
5908 order; the standard system directories come after.
5910 If a standard system include directory, or a directory specified with
5911 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5912 option will be ignored. The directory will still be searched but as a
5913 system directory at its normal position in the system include chain.
5914 This is to ensure that GCC's procedure to fix buggy system headers and
5915 the ordering for the include_next directive are not inadvertently changed.
5916 If you really need to change the search order for system directories,
5917 use the @option{-nostdinc} and/or @option{-isystem} options.
5919 @item -iquote@var{dir}
5921 Add the directory @var{dir} to the head of the list of directories to
5922 be searched for header files only for the case of @samp{#include
5923 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
5924 otherwise just like @option{-I}.
5928 Add directory @var{dir} to the list of directories to be searched
5931 @item -B@var{prefix}
5933 This option specifies where to find the executables, libraries,
5934 include files, and data files of the compiler itself.
5936 The compiler driver program runs one or more of the subprograms
5937 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5938 @var{prefix} as a prefix for each program it tries to run, both with and
5939 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5941 For each subprogram to be run, the compiler driver first tries the
5942 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5943 was not specified, the driver tries two standard prefixes, which are
5944 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5945 those results in a file name that is found, the unmodified program
5946 name is searched for using the directories specified in your
5947 @env{PATH} environment variable.
5949 The compiler will check to see if the path provided by the @option{-B}
5950 refers to a directory, and if necessary it will add a directory
5951 separator character at the end of the path.
5953 @option{-B} prefixes that effectively specify directory names also apply
5954 to libraries in the linker, because the compiler translates these
5955 options into @option{-L} options for the linker. They also apply to
5956 includes files in the preprocessor, because the compiler translates these
5957 options into @option{-isystem} options for the preprocessor. In this case,
5958 the compiler appends @samp{include} to the prefix.
5960 The run-time support file @file{libgcc.a} can also be searched for using
5961 the @option{-B} prefix, if needed. If it is not found there, the two
5962 standard prefixes above are tried, and that is all. The file is left
5963 out of the link if it is not found by those means.
5965 Another way to specify a prefix much like the @option{-B} prefix is to use
5966 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5969 As a special kludge, if the path provided by @option{-B} is
5970 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5971 9, then it will be replaced by @file{[dir/]include}. This is to help
5972 with boot-strapping the compiler.
5974 @item -specs=@var{file}
5976 Process @var{file} after the compiler reads in the standard @file{specs}
5977 file, in order to override the defaults that the @file{gcc} driver
5978 program uses when determining what switches to pass to @file{cc1},
5979 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5980 @option{-specs=@var{file}} can be specified on the command line, and they
5981 are processed in order, from left to right.
5985 This option has been deprecated. Please use @option{-iquote} instead for
5986 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
5987 Any directories you specify with @option{-I} options before the @option{-I-}
5988 option are searched only for the case of @samp{#include "@var{file}"};
5989 they are not searched for @samp{#include <@var{file}>}.
5991 If additional directories are specified with @option{-I} options after
5992 the @option{-I-}, these directories are searched for all @samp{#include}
5993 directives. (Ordinarily @emph{all} @option{-I} directories are used
5996 In addition, the @option{-I-} option inhibits the use of the current
5997 directory (where the current input file came from) as the first search
5998 directory for @samp{#include "@var{file}"}. There is no way to
5999 override this effect of @option{-I-}. With @option{-I.} you can specify
6000 searching the directory which was current when the compiler was
6001 invoked. That is not exactly the same as what the preprocessor does
6002 by default, but it is often satisfactory.
6004 @option{-I-} does not inhibit the use of the standard system directories
6005 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6012 @section Specifying subprocesses and the switches to pass to them
6015 @command{gcc} is a driver program. It performs its job by invoking a
6016 sequence of other programs to do the work of compiling, assembling and
6017 linking. GCC interprets its command-line parameters and uses these to
6018 deduce which programs it should invoke, and which command-line options
6019 it ought to place on their command lines. This behavior is controlled
6020 by @dfn{spec strings}. In most cases there is one spec string for each
6021 program that GCC can invoke, but a few programs have multiple spec
6022 strings to control their behavior. The spec strings built into GCC can
6023 be overridden by using the @option{-specs=} command-line switch to specify
6026 @dfn{Spec files} are plaintext files that are used to construct spec
6027 strings. They consist of a sequence of directives separated by blank
6028 lines. The type of directive is determined by the first non-whitespace
6029 character on the line and it can be one of the following:
6032 @item %@var{command}
6033 Issues a @var{command} to the spec file processor. The commands that can
6037 @item %include <@var{file}>
6039 Search for @var{file} and insert its text at the current point in the
6042 @item %include_noerr <@var{file}>
6043 @cindex %include_noerr
6044 Just like @samp{%include}, but do not generate an error message if the include
6045 file cannot be found.
6047 @item %rename @var{old_name} @var{new_name}
6049 Rename the spec string @var{old_name} to @var{new_name}.
6053 @item *[@var{spec_name}]:
6054 This tells the compiler to create, override or delete the named spec
6055 string. All lines after this directive up to the next directive or
6056 blank line are considered to be the text for the spec string. If this
6057 results in an empty string then the spec will be deleted. (Or, if the
6058 spec did not exist, then nothing will happened.) Otherwise, if the spec
6059 does not currently exist a new spec will be created. If the spec does
6060 exist then its contents will be overridden by the text of this
6061 directive, unless the first character of that text is the @samp{+}
6062 character, in which case the text will be appended to the spec.
6064 @item [@var{suffix}]:
6065 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6066 and up to the next directive or blank line are considered to make up the
6067 spec string for the indicated suffix. When the compiler encounters an
6068 input file with the named suffix, it will processes the spec string in
6069 order to work out how to compile that file. For example:
6076 This says that any input file whose name ends in @samp{.ZZ} should be
6077 passed to the program @samp{z-compile}, which should be invoked with the
6078 command-line switch @option{-input} and with the result of performing the
6079 @samp{%i} substitution. (See below.)
6081 As an alternative to providing a spec string, the text that follows a
6082 suffix directive can be one of the following:
6085 @item @@@var{language}
6086 This says that the suffix is an alias for a known @var{language}. This is
6087 similar to using the @option{-x} command-line switch to GCC to specify a
6088 language explicitly. For example:
6095 Says that .ZZ files are, in fact, C++ source files.
6098 This causes an error messages saying:
6101 @var{name} compiler not installed on this system.
6105 GCC already has an extensive list of suffixes built into it.
6106 This directive will add an entry to the end of the list of suffixes, but
6107 since the list is searched from the end backwards, it is effectively
6108 possible to override earlier entries using this technique.
6112 GCC has the following spec strings built into it. Spec files can
6113 override these strings or create their own. Note that individual
6114 targets can also add their own spec strings to this list.
6117 asm Options to pass to the assembler
6118 asm_final Options to pass to the assembler post-processor
6119 cpp Options to pass to the C preprocessor
6120 cc1 Options to pass to the C compiler
6121 cc1plus Options to pass to the C++ compiler
6122 endfile Object files to include at the end of the link
6123 link Options to pass to the linker
6124 lib Libraries to include on the command line to the linker
6125 libgcc Decides which GCC support library to pass to the linker
6126 linker Sets the name of the linker
6127 predefines Defines to be passed to the C preprocessor
6128 signed_char Defines to pass to CPP to say whether @code{char} is signed
6130 startfile Object files to include at the start of the link
6133 Here is a small example of a spec file:
6139 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6142 This example renames the spec called @samp{lib} to @samp{old_lib} and
6143 then overrides the previous definition of @samp{lib} with a new one.
6144 The new definition adds in some extra command-line options before
6145 including the text of the old definition.
6147 @dfn{Spec strings} are a list of command-line options to be passed to their
6148 corresponding program. In addition, the spec strings can contain
6149 @samp{%}-prefixed sequences to substitute variable text or to
6150 conditionally insert text into the command line. Using these constructs
6151 it is possible to generate quite complex command lines.
6153 Here is a table of all defined @samp{%}-sequences for spec
6154 strings. Note that spaces are not generated automatically around the
6155 results of expanding these sequences. Therefore you can concatenate them
6156 together or combine them with constant text in a single argument.
6160 Substitute one @samp{%} into the program name or argument.
6163 Substitute the name of the input file being processed.
6166 Substitute the basename of the input file being processed.
6167 This is the substring up to (and not including) the last period
6168 and not including the directory.
6171 This is the same as @samp{%b}, but include the file suffix (text after
6175 Marks the argument containing or following the @samp{%d} as a
6176 temporary file name, so that that file will be deleted if GCC exits
6177 successfully. Unlike @samp{%g}, this contributes no text to the
6180 @item %g@var{suffix}
6181 Substitute a file name that has suffix @var{suffix} and is chosen
6182 once per compilation, and mark the argument in the same way as
6183 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6184 name is now chosen in a way that is hard to predict even when previously
6185 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6186 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6187 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6188 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6189 was simply substituted with a file name chosen once per compilation,
6190 without regard to any appended suffix (which was therefore treated
6191 just like ordinary text), making such attacks more likely to succeed.
6193 @item %u@var{suffix}
6194 Like @samp{%g}, but generates a new temporary file name even if
6195 @samp{%u@var{suffix}} was already seen.
6197 @item %U@var{suffix}
6198 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6199 new one if there is no such last file name. In the absence of any
6200 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6201 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6202 would involve the generation of two distinct file names, one
6203 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6204 simply substituted with a file name chosen for the previous @samp{%u},
6205 without regard to any appended suffix.
6207 @item %j@var{suffix}
6208 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6209 writable, and if save-temps is off; otherwise, substitute the name
6210 of a temporary file, just like @samp{%u}. This temporary file is not
6211 meant for communication between processes, but rather as a junk
6214 @item %|@var{suffix}
6215 @itemx %m@var{suffix}
6216 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6217 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6218 all. These are the two most common ways to instruct a program that it
6219 should read from standard input or write to standard output. If you
6220 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6221 construct: see for example @file{f/lang-specs.h}.
6223 @item %.@var{SUFFIX}
6224 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6225 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6226 terminated by the next space or %.
6229 Marks the argument containing or following the @samp{%w} as the
6230 designated output file of this compilation. This puts the argument
6231 into the sequence of arguments that @samp{%o} will substitute later.
6234 Substitutes the names of all the output files, with spaces
6235 automatically placed around them. You should write spaces
6236 around the @samp{%o} as well or the results are undefined.
6237 @samp{%o} is for use in the specs for running the linker.
6238 Input files whose names have no recognized suffix are not compiled
6239 at all, but they are included among the output files, so they will
6243 Substitutes the suffix for object files. Note that this is
6244 handled specially when it immediately follows @samp{%g, %u, or %U},
6245 because of the need for those to form complete file names. The
6246 handling is such that @samp{%O} is treated exactly as if it had already
6247 been substituted, except that @samp{%g, %u, and %U} do not currently
6248 support additional @var{suffix} characters following @samp{%O} as they would
6249 following, for example, @samp{.o}.
6252 Substitutes the standard macro predefinitions for the
6253 current target machine. Use this when running @code{cpp}.
6256 Like @samp{%p}, but puts @samp{__} before and after the name of each
6257 predefined macro, except for macros that start with @samp{__} or with
6258 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6262 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6263 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
6264 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6268 Current argument is the name of a library or startup file of some sort.
6269 Search for that file in a standard list of directories and substitute
6270 the full name found.
6273 Print @var{str} as an error message. @var{str} is terminated by a newline.
6274 Use this when inconsistent options are detected.
6277 Substitute the contents of spec string @var{name} at this point.
6280 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6282 @item %x@{@var{option}@}
6283 Accumulate an option for @samp{%X}.
6286 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6290 Output the accumulated assembler options specified by @option{-Wa}.
6293 Output the accumulated preprocessor options specified by @option{-Wp}.
6296 Process the @code{asm} spec. This is used to compute the
6297 switches to be passed to the assembler.
6300 Process the @code{asm_final} spec. This is a spec string for
6301 passing switches to an assembler post-processor, if such a program is
6305 Process the @code{link} spec. This is the spec for computing the
6306 command line passed to the linker. Typically it will make use of the
6307 @samp{%L %G %S %D and %E} sequences.
6310 Dump out a @option{-L} option for each directory that GCC believes might
6311 contain startup files. If the target supports multilibs then the
6312 current multilib directory will be prepended to each of these paths.
6315 Output the multilib directory with directory separators replaced with
6316 @samp{_}. If multilib directories are not set, or the multilib directory is
6317 @file{.} then this option emits nothing.
6320 Process the @code{lib} spec. This is a spec string for deciding which
6321 libraries should be included on the command line to the linker.
6324 Process the @code{libgcc} spec. This is a spec string for deciding
6325 which GCC support library should be included on the command line to the linker.
6328 Process the @code{startfile} spec. This is a spec for deciding which
6329 object files should be the first ones passed to the linker. Typically
6330 this might be a file named @file{crt0.o}.
6333 Process the @code{endfile} spec. This is a spec string that specifies
6334 the last object files that will be passed to the linker.
6337 Process the @code{cpp} spec. This is used to construct the arguments
6338 to be passed to the C preprocessor.
6341 Process the @code{cc1} spec. This is used to construct the options to be
6342 passed to the actual C compiler (@samp{cc1}).
6345 Process the @code{cc1plus} spec. This is used to construct the options to be
6346 passed to the actual C++ compiler (@samp{cc1plus}).
6349 Substitute the variable part of a matched option. See below.
6350 Note that each comma in the substituted string is replaced by
6354 Remove all occurrences of @code{-S} from the command line. Note---this
6355 command is position dependent. @samp{%} commands in the spec string
6356 before this one will see @code{-S}, @samp{%} commands in the spec string
6357 after this one will not.
6359 @item %:@var{function}(@var{args})
6360 Call the named function @var{function}, passing it @var{args}.
6361 @var{args} is first processed as a nested spec string, then split
6362 into an argument vector in the usual fashion. The function returns
6363 a string which is processed as if it had appeared literally as part
6364 of the current spec.
6366 The following built-in spec functions are provided:
6369 @item @code{if-exists}
6370 The @code{if-exists} spec function takes one argument, an absolute
6371 pathname to a file. If the file exists, @code{if-exists} returns the
6372 pathname. Here is a small example of its usage:
6376 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6379 @item @code{if-exists-else}
6380 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6381 spec function, except that it takes two arguments. The first argument is
6382 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6383 returns the pathname. If it does not exist, it returns the second argument.
6384 This way, @code{if-exists-else} can be used to select one file or another,
6385 based on the existence of the first. Here is a small example of its usage:
6389 crt0%O%s %:if-exists(crti%O%s) \
6390 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6393 @item @code{replace-outfile}
6394 The @code{replace-outfile} spec function takes two arguments. It looks for the
6395 first argument in the outfiles array and replaces it with the second argument. Here
6396 is a small example of its usage:
6399 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
6405 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6406 If that switch was not specified, this substitutes nothing. Note that
6407 the leading dash is omitted when specifying this option, and it is
6408 automatically inserted if the substitution is performed. Thus the spec
6409 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6410 and would output the command line option @option{-foo}.
6412 @item %W@{@code{S}@}
6413 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6416 @item %@{@code{S}*@}
6417 Substitutes all the switches specified to GCC whose names start
6418 with @code{-S}, but which also take an argument. This is used for
6419 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6420 GCC considers @option{-o foo} as being
6421 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6422 text, including the space. Thus two arguments would be generated.
6424 @item %@{@code{S}*&@code{T}*@}
6425 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6426 (the order of @code{S} and @code{T} in the spec is not significant).
6427 There can be any number of ampersand-separated variables; for each the
6428 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6430 @item %@{@code{S}:@code{X}@}
6431 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6433 @item %@{!@code{S}:@code{X}@}
6434 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6436 @item %@{@code{S}*:@code{X}@}
6437 Substitutes @code{X} if one or more switches whose names start with
6438 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6439 once, no matter how many such switches appeared. However, if @code{%*}
6440 appears somewhere in @code{X}, then @code{X} will be substituted once
6441 for each matching switch, with the @code{%*} replaced by the part of
6442 that switch that matched the @code{*}.
6444 @item %@{.@code{S}:@code{X}@}
6445 Substitutes @code{X}, if processing a file with suffix @code{S}.
6447 @item %@{!.@code{S}:@code{X}@}
6448 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6450 @item %@{@code{S}|@code{P}:@code{X}@}
6451 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6452 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6453 although they have a stronger binding than the @samp{|}. If @code{%*}
6454 appears in @code{X}, all of the alternatives must be starred, and only
6455 the first matching alternative is substituted.
6457 For example, a spec string like this:
6460 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6463 will output the following command-line options from the following input
6464 command-line options:
6469 -d fred.c -foo -baz -boggle
6470 -d jim.d -bar -baz -boggle
6473 @item %@{S:X; T:Y; :D@}
6475 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6476 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6477 be as many clauses as you need. This may be combined with @code{.},
6478 @code{!}, @code{|}, and @code{*} as needed.
6483 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6484 construct may contain other nested @samp{%} constructs or spaces, or
6485 even newlines. They are processed as usual, as described above.
6486 Trailing white space in @code{X} is ignored. White space may also
6487 appear anywhere on the left side of the colon in these constructs,
6488 except between @code{.} or @code{*} and the corresponding word.
6490 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6491 handled specifically in these constructs. If another value of
6492 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
6493 @option{-W} switch is found later in the command line, the earlier
6494 switch value is ignored, except with @{@code{S}*@} where @code{S} is
6495 just one letter, which passes all matching options.
6497 The character @samp{|} at the beginning of the predicate text is used to
6498 indicate that a command should be piped to the following command, but
6499 only if @option{-pipe} is specified.
6501 It is built into GCC which switches take arguments and which do not.
6502 (You might think it would be useful to generalize this to allow each
6503 compiler's spec to say which switches take arguments. But this cannot
6504 be done in a consistent fashion. GCC cannot even decide which input
6505 files have been specified without knowing which switches take arguments,
6506 and it must know which input files to compile in order to tell which
6509 GCC also knows implicitly that arguments starting in @option{-l} are to be
6510 treated as compiler output files, and passed to the linker in their
6511 proper position among the other output files.
6513 @c man begin OPTIONS
6515 @node Target Options
6516 @section Specifying Target Machine and Compiler Version
6517 @cindex target options
6518 @cindex cross compiling
6519 @cindex specifying machine version
6520 @cindex specifying compiler version and target machine
6521 @cindex compiler version, specifying
6522 @cindex target machine, specifying
6524 The usual way to run GCC is to run the executable called @file{gcc}, or
6525 @file{<machine>-gcc} when cross-compiling, or
6526 @file{<machine>-gcc-<version>} to run a version other than the one that
6527 was installed last. Sometimes this is inconvenient, so GCC provides
6528 options that will switch to another cross-compiler or version.
6531 @item -b @var{machine}
6533 The argument @var{machine} specifies the target machine for compilation.
6535 The value to use for @var{machine} is the same as was specified as the
6536 machine type when configuring GCC as a cross-compiler. For
6537 example, if a cross-compiler was configured with @samp{configure
6538 i386v}, meaning to compile for an 80386 running System V, then you
6539 would specify @option{-b i386v} to run that cross compiler.
6541 @item -V @var{version}
6543 The argument @var{version} specifies which version of GCC to run.
6544 This is useful when multiple versions are installed. For example,
6545 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
6548 The @option{-V} and @option{-b} options work by running the
6549 @file{<machine>-gcc-<version>} executable, so there's no real reason to
6550 use them if you can just run that directly.
6552 @node Submodel Options
6553 @section Hardware Models and Configurations
6554 @cindex submodel options
6555 @cindex specifying hardware config
6556 @cindex hardware models and configurations, specifying
6557 @cindex machine dependent options
6559 Earlier we discussed the standard option @option{-b} which chooses among
6560 different installed compilers for completely different target
6561 machines, such as VAX vs.@: 68000 vs.@: 80386.
6563 In addition, each of these target machine types can have its own
6564 special options, starting with @samp{-m}, to choose among various
6565 hardware models or configurations---for example, 68010 vs 68020,
6566 floating coprocessor or none. A single installed version of the
6567 compiler can compile for any model or configuration, according to the
6570 Some configurations of the compiler also support additional special
6571 options, usually for compatibility with other compilers on the same
6574 These options are defined by the macro @code{TARGET_SWITCHES} in the
6575 machine description. The default for the options is also defined by
6576 that macro, which enables you to change the defaults.
6578 @c This list is ordered alphanumerically by subsection name.
6579 @c It should be the same order and spelling as these options are listed
6580 @c in Machine Dependent Options
6588 * DEC Alpha Options::
6589 * DEC Alpha/VMS Options::
6593 * i386 and x86-64 Options::
6605 * RS/6000 and PowerPC Options::
6606 * S/390 and zSeries Options::
6609 * System V Options::
6610 * TMS320C3x/C4x Options::
6614 * Xstormy16 Options::
6620 @subsection ARC Options
6623 These options are defined for ARC implementations:
6628 Compile code for little endian mode. This is the default.
6632 Compile code for big endian mode.
6635 @opindex mmangle-cpu
6636 Prepend the name of the cpu to all public symbol names.
6637 In multiple-processor systems, there are many ARC variants with different
6638 instruction and register set characteristics. This flag prevents code
6639 compiled for one cpu to be linked with code compiled for another.
6640 No facility exists for handling variants that are ``almost identical''.
6641 This is an all or nothing option.
6643 @item -mcpu=@var{cpu}
6645 Compile code for ARC variant @var{cpu}.
6646 Which variants are supported depend on the configuration.
6647 All variants support @option{-mcpu=base}, this is the default.
6649 @item -mtext=@var{text-section}
6650 @itemx -mdata=@var{data-section}
6651 @itemx -mrodata=@var{readonly-data-section}
6655 Put functions, data, and readonly data in @var{text-section},
6656 @var{data-section}, and @var{readonly-data-section} respectively
6657 by default. This can be overridden with the @code{section} attribute.
6658 @xref{Variable Attributes}.
6663 @subsection ARM Options
6666 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6670 @item -mabi=@var{name}
6672 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
6673 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6676 @opindex mapcs-frame
6677 Generate a stack frame that is compliant with the ARM Procedure Call
6678 Standard for all functions, even if this is not strictly necessary for
6679 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6680 with this option will cause the stack frames not to be generated for
6681 leaf functions. The default is @option{-mno-apcs-frame}.
6685 This is a synonym for @option{-mapcs-frame}.
6688 @c not currently implemented
6689 @item -mapcs-stack-check
6690 @opindex mapcs-stack-check
6691 Generate code to check the amount of stack space available upon entry to
6692 every function (that actually uses some stack space). If there is
6693 insufficient space available then either the function
6694 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6695 called, depending upon the amount of stack space required. The run time
6696 system is required to provide these functions. The default is
6697 @option{-mno-apcs-stack-check}, since this produces smaller code.
6699 @c not currently implemented
6701 @opindex mapcs-float
6702 Pass floating point arguments using the float point registers. This is
6703 one of the variants of the APCS@. This option is recommended if the
6704 target hardware has a floating point unit or if a lot of floating point
6705 arithmetic is going to be performed by the code. The default is
6706 @option{-mno-apcs-float}, since integer only code is slightly increased in
6707 size if @option{-mapcs-float} is used.
6709 @c not currently implemented
6710 @item -mapcs-reentrant
6711 @opindex mapcs-reentrant
6712 Generate reentrant, position independent code. The default is
6713 @option{-mno-apcs-reentrant}.
6716 @item -mthumb-interwork
6717 @opindex mthumb-interwork
6718 Generate code which supports calling between the ARM and Thumb
6719 instruction sets. Without this option the two instruction sets cannot
6720 be reliably used inside one program. The default is
6721 @option{-mno-thumb-interwork}, since slightly larger code is generated
6722 when @option{-mthumb-interwork} is specified.
6724 @item -mno-sched-prolog
6725 @opindex mno-sched-prolog
6726 Prevent the reordering of instructions in the function prolog, or the
6727 merging of those instruction with the instructions in the function's
6728 body. This means that all functions will start with a recognizable set
6729 of instructions (or in fact one of a choice from a small set of
6730 different function prologues), and this information can be used to
6731 locate the start if functions inside an executable piece of code. The
6732 default is @option{-msched-prolog}.
6735 @opindex mhard-float
6736 Generate output containing floating point instructions. This is the
6740 @opindex msoft-float
6741 Generate output containing library calls for floating point.
6742 @strong{Warning:} the requisite libraries are not available for all ARM
6743 targets. Normally the facilities of the machine's usual C compiler are
6744 used, but this cannot be done directly in cross-compilation. You must make
6745 your own arrangements to provide suitable library functions for
6748 @option{-msoft-float} changes the calling convention in the output file;
6749 therefore, it is only useful if you compile @emph{all} of a program with
6750 this option. In particular, you need to compile @file{libgcc.a}, the
6751 library that comes with GCC, with @option{-msoft-float} in order for
6754 @item -mfloat-abi=@var{name}
6756 Specifies which ABI to use for floating point values. Permissible values
6757 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6759 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6760 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6761 of floating point instructions, but still uses the soft-float calling
6764 @item -mlittle-endian
6765 @opindex mlittle-endian
6766 Generate code for a processor running in little-endian mode. This is
6767 the default for all standard configurations.
6770 @opindex mbig-endian
6771 Generate code for a processor running in big-endian mode; the default is
6772 to compile code for a little-endian processor.
6774 @item -mwords-little-endian
6775 @opindex mwords-little-endian
6776 This option only applies when generating code for big-endian processors.
6777 Generate code for a little-endian word order but a big-endian byte
6778 order. That is, a byte order of the form @samp{32107654}. Note: this
6779 option should only be used if you require compatibility with code for
6780 big-endian ARM processors generated by versions of the compiler prior to
6783 @item -mcpu=@var{name}
6785 This specifies the name of the target ARM processor. GCC uses this name
6786 to determine what kind of instructions it can emit when generating
6787 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6788 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6789 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6790 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6791 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6792 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
6793 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6794 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6795 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
6796 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
6797 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
6798 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
6799 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
6800 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
6803 @itemx -mtune=@var{name}
6805 This option is very similar to the @option{-mcpu=} option, except that
6806 instead of specifying the actual target processor type, and hence
6807 restricting which instructions can be used, it specifies that GCC should
6808 tune the performance of the code as if the target were of the type
6809 specified in this option, but still choosing the instructions that it
6810 will generate based on the cpu specified by a @option{-mcpu=} option.
6811 For some ARM implementations better performance can be obtained by using
6814 @item -march=@var{name}
6816 This specifies the name of the target ARM architecture. GCC uses this
6817 name to determine what kind of instructions it can emit when generating
6818 assembly code. This option can be used in conjunction with or instead
6819 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6820 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6821 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6822 @samp{iwmmxt}, @samp{ep9312}.
6824 @item -mfpu=@var{name}
6825 @itemx -mfpe=@var{number}
6826 @itemx -mfp=@var{number}
6830 This specifies what floating point hardware (or hardware emulation) is
6831 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6832 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6833 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6834 with older versions of GCC@.
6836 If @option{-msoft-float} is specified this specifies the format of
6837 floating point values.
6839 @item -mstructure-size-boundary=@var{n}
6840 @opindex mstructure-size-boundary
6841 The size of all structures and unions will be rounded up to a multiple
6842 of the number of bits set by this option. Permissible values are 8, 32
6843 and 64. The default value varies for different toolchains. For the COFF
6844 targeted toolchain the default value is 8. A value of 64 is only allowed
6845 if the underlying ABI supports it.
6847 Specifying the larger number can produce faster, more efficient code, but
6848 can also increase the size of the program. Different values are potentially
6849 incompatible. Code compiled with one value cannot necessarily expect to
6850 work with code or libraries compiled with another value, if they exchange
6851 information using structures or unions.
6853 @item -mabort-on-noreturn
6854 @opindex mabort-on-noreturn
6855 Generate a call to the function @code{abort} at the end of a
6856 @code{noreturn} function. It will be executed if the function tries to
6860 @itemx -mno-long-calls
6861 @opindex mlong-calls
6862 @opindex mno-long-calls
6863 Tells the compiler to perform function calls by first loading the
6864 address of the function into a register and then performing a subroutine
6865 call on this register. This switch is needed if the target function
6866 will lie outside of the 64 megabyte addressing range of the offset based
6867 version of subroutine call instruction.
6869 Even if this switch is enabled, not all function calls will be turned
6870 into long calls. The heuristic is that static functions, functions
6871 which have the @samp{short-call} attribute, functions that are inside
6872 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6873 definitions have already been compiled within the current compilation
6874 unit, will not be turned into long calls. The exception to this rule is
6875 that weak function definitions, functions with the @samp{long-call}
6876 attribute or the @samp{section} attribute, and functions that are within
6877 the scope of a @samp{#pragma long_calls} directive, will always be
6878 turned into long calls.
6880 This feature is not enabled by default. Specifying
6881 @option{-mno-long-calls} will restore the default behavior, as will
6882 placing the function calls within the scope of a @samp{#pragma
6883 long_calls_off} directive. Note these switches have no effect on how
6884 the compiler generates code to handle function calls via function
6887 @item -mnop-fun-dllimport
6888 @opindex mnop-fun-dllimport
6889 Disable support for the @code{dllimport} attribute.
6891 @item -msingle-pic-base
6892 @opindex msingle-pic-base
6893 Treat the register used for PIC addressing as read-only, rather than
6894 loading it in the prologue for each function. The run-time system is
6895 responsible for initializing this register with an appropriate value
6896 before execution begins.
6898 @item -mpic-register=@var{reg}
6899 @opindex mpic-register
6900 Specify the register to be used for PIC addressing. The default is R10
6901 unless stack-checking is enabled, when R9 is used.
6903 @item -mcirrus-fix-invalid-insns
6904 @opindex mcirrus-fix-invalid-insns
6905 @opindex mno-cirrus-fix-invalid-insns
6906 Insert NOPs into the instruction stream to in order to work around
6907 problems with invalid Maverick instruction combinations. This option
6908 is only valid if the @option{-mcpu=ep9312} option has been used to
6909 enable generation of instructions for the Cirrus Maverick floating
6910 point co-processor. This option is not enabled by default, since the
6911 problem is only present in older Maverick implementations. The default
6912 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6915 @item -mpoke-function-name
6916 @opindex mpoke-function-name
6917 Write the name of each function into the text section, directly
6918 preceding the function prologue. The generated code is similar to this:
6922 .ascii "arm_poke_function_name", 0
6925 .word 0xff000000 + (t1 - t0)
6926 arm_poke_function_name
6928 stmfd sp!, @{fp, ip, lr, pc@}
6932 When performing a stack backtrace, code can inspect the value of
6933 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6934 location @code{pc - 12} and the top 8 bits are set, then we know that
6935 there is a function name embedded immediately preceding this location
6936 and has length @code{((pc[-3]) & 0xff000000)}.
6940 Generate code for the 16-bit Thumb instruction set. The default is to
6941 use the 32-bit ARM instruction set.
6944 @opindex mtpcs-frame
6945 Generate a stack frame that is compliant with the Thumb Procedure Call
6946 Standard for all non-leaf functions. (A leaf function is one that does
6947 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6949 @item -mtpcs-leaf-frame
6950 @opindex mtpcs-leaf-frame
6951 Generate a stack frame that is compliant with the Thumb Procedure Call
6952 Standard for all leaf functions. (A leaf function is one that does
6953 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6955 @item -mcallee-super-interworking
6956 @opindex mcallee-super-interworking
6957 Gives all externally visible functions in the file being compiled an ARM
6958 instruction set header which switches to Thumb mode before executing the
6959 rest of the function. This allows these functions to be called from
6960 non-interworking code.
6962 @item -mcaller-super-interworking
6963 @opindex mcaller-super-interworking
6964 Allows calls via function pointers (including virtual functions) to
6965 execute correctly regardless of whether the target code has been
6966 compiled for interworking or not. There is a small overhead in the cost
6967 of executing a function pointer if this option is enabled.
6972 @subsection AVR Options
6975 These options are defined for AVR implementations:
6978 @item -mmcu=@var{mcu}
6980 Specify ATMEL AVR instruction set or MCU type.
6982 Instruction set avr1 is for the minimal AVR core, not supported by the C
6983 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
6984 attiny11, attiny12, attiny15, attiny28).
6986 Instruction set avr2 (default) is for the classic AVR core with up to
6987 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
6988 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
6989 at90c8534, at90s8535).
6991 Instruction set avr3 is for the classic AVR core with up to 128K program
6992 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
6994 Instruction set avr4 is for the enhanced AVR core with up to 8K program
6995 memory space (MCU types: atmega8, atmega83, atmega85).
6997 Instruction set avr5 is for the enhanced AVR core with up to 128K program
6998 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
6999 atmega64, atmega128, at43usb355, at94k).
7003 Output instruction sizes to the asm file.
7005 @item -minit-stack=@var{N}
7006 @opindex minit-stack
7007 Specify the initial stack address, which may be a symbol or numeric value,
7008 @samp{__stack} is the default.
7010 @item -mno-interrupts
7011 @opindex mno-interrupts
7012 Generated code is not compatible with hardware interrupts.
7013 Code size will be smaller.
7015 @item -mcall-prologues
7016 @opindex mcall-prologues
7017 Functions prologues/epilogues expanded as call to appropriate
7018 subroutines. Code size will be smaller.
7020 @item -mno-tablejump
7021 @opindex mno-tablejump
7022 Do not generate tablejump insns which sometimes increase code size.
7025 @opindex mtiny-stack
7026 Change only the low 8 bits of the stack pointer.
7030 Assume int to be 8 bit integer. This affects the sizes of all types: A
7031 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7032 and long long will be 4 bytes. Please note that this option does not
7033 comply to the C standards, but it will provide you with smaller code
7038 @subsection CRIS Options
7039 @cindex CRIS Options
7041 These options are defined specifically for the CRIS ports.
7044 @item -march=@var{architecture-type}
7045 @itemx -mcpu=@var{architecture-type}
7048 Generate code for the specified architecture. The choices for
7049 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7050 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7051 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7054 @item -mtune=@var{architecture-type}
7056 Tune to @var{architecture-type} everything applicable about the generated
7057 code, except for the ABI and the set of available instructions. The
7058 choices for @var{architecture-type} are the same as for
7059 @option{-march=@var{architecture-type}}.
7061 @item -mmax-stack-frame=@var{n}
7062 @opindex mmax-stack-frame
7063 Warn when the stack frame of a function exceeds @var{n} bytes.
7065 @item -melinux-stacksize=@var{n}
7066 @opindex melinux-stacksize
7067 Only available with the @samp{cris-axis-aout} target. Arranges for
7068 indications in the program to the kernel loader that the stack of the
7069 program should be set to @var{n} bytes.
7075 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7076 @option{-march=v3} and @option{-march=v8} respectively.
7078 @item -mmul-bug-workaround
7079 @itemx -mno-mul-bug-workaround
7080 @opindex mmul-bug-workaround
7081 @opindex mno-mul-bug-workaround
7082 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7083 models where it applies. This option is active by default.
7087 Enable CRIS-specific verbose debug-related information in the assembly
7088 code. This option also has the effect to turn off the @samp{#NO_APP}
7089 formatted-code indicator to the assembler at the beginning of the
7094 Do not use condition-code results from previous instruction; always emit
7095 compare and test instructions before use of condition codes.
7097 @item -mno-side-effects
7098 @opindex mno-side-effects
7099 Do not emit instructions with side-effects in addressing modes other than
7103 @itemx -mno-stack-align
7105 @itemx -mno-data-align
7106 @itemx -mconst-align
7107 @itemx -mno-const-align
7108 @opindex mstack-align
7109 @opindex mno-stack-align
7110 @opindex mdata-align
7111 @opindex mno-data-align
7112 @opindex mconst-align
7113 @opindex mno-const-align
7114 These options (no-options) arranges (eliminate arrangements) for the
7115 stack-frame, individual data and constants to be aligned for the maximum
7116 single data access size for the chosen CPU model. The default is to
7117 arrange for 32-bit alignment. ABI details such as structure layout are
7118 not affected by these options.
7126 Similar to the stack- data- and const-align options above, these options
7127 arrange for stack-frame, writable data and constants to all be 32-bit,
7128 16-bit or 8-bit aligned. The default is 32-bit alignment.
7130 @item -mno-prologue-epilogue
7131 @itemx -mprologue-epilogue
7132 @opindex mno-prologue-epilogue
7133 @opindex mprologue-epilogue
7134 With @option{-mno-prologue-epilogue}, the normal function prologue and
7135 epilogue that sets up the stack-frame are omitted and no return
7136 instructions or return sequences are generated in the code. Use this
7137 option only together with visual inspection of the compiled code: no
7138 warnings or errors are generated when call-saved registers must be saved,
7139 or storage for local variable needs to be allocated.
7145 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7146 instruction sequences that load addresses for functions from the PLT part
7147 of the GOT rather than (traditional on other architectures) calls to the
7148 PLT@. The default is @option{-mgotplt}.
7152 Legacy no-op option only recognized with the cris-axis-aout target.
7156 Legacy no-op option only recognized with the cris-axis-elf and
7157 cris-axis-linux-gnu targets.
7161 Only recognized with the cris-axis-aout target, where it selects a
7162 GNU/linux-like multilib, include files and instruction set for
7167 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7171 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7172 to link with input-output functions from a simulator library. Code,
7173 initialized data and zero-initialized data are allocated consecutively.
7177 Like @option{-sim}, but pass linker options to locate initialized data at
7178 0x40000000 and zero-initialized data at 0x80000000.
7181 @node Darwin Options
7182 @subsection Darwin Options
7183 @cindex Darwin options
7185 These options are defined for all architectures running the Darwin operating
7188 FSF GCC on Darwin does not create ``fat'' object files; it will create
7189 an object file for the single architecture that it was built to
7190 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7191 @option{-arch} options are used; it does so by running the compiler or
7192 linker multiple times and joining the results together with
7195 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7196 @samp{i686}) is determined by the flags that specify the ISA
7197 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7198 @option{-force_cpusubtype_ALL} option can be used to override this.
7200 The Darwin tools vary in their behaviour when presented with an ISA
7201 mismatch. The assembler, @file{as}, will only permit instructions to
7202 be used that are valid for the subtype of the file it is generating,
7203 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7204 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7205 and print an error if asked to create a shared library with a less
7206 restrictive subtype than its input files (for instance, trying to put
7207 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7208 for executables, @file{ld}, will quietly give the executable the most
7209 restrictive subtype of any of its input files.
7214 Add the framework directory @var{dir} to the head of the list of
7215 directories to be searched for header files. These directories are
7216 interleaved with those specified by @option{-I} options and are
7217 scanned in a left-to-right order.
7219 A framework directory is a directory with frameworks in it. A
7220 framework is a directory with a @samp{"Headers"} and/or
7221 @samp{"PrivateHeaders"} directory contained directly in it that ends
7222 in @samp{".framework"}. The name of a framework is the name of this
7223 directory excluding the @samp{".framework"}. Headers associated with
7224 the framework are found in one of those two directories, with
7225 @samp{"Headers"} being searched first. A subframework is a framework
7226 directory that is in a framework's @samp{"Frameworks"} directory.
7227 Includes of subframework headers can only appear in a header of a
7228 framework that contains the subframework, or in a sibling subframework
7229 header. Two subframeworks are siblings if they occur in the same
7230 framework. A subframework should not have the same name as a
7231 framework, a warning will be issued if this is violated. Currently a
7232 subframework cannot have subframeworks, in the future, the mechanism
7233 may be extended to support this. The standard frameworks can be found
7234 in @samp{"/System/Library/Frameworks"} and
7235 @samp{"/Library/Frameworks"}. An example include looks like
7236 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7237 the name of the framework and header.h is found in the
7238 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7242 Emit debugging information for symbols that are used. For STABS
7243 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7244 This is by default ON@.
7248 Emit debugging information for all symbols and types.
7250 @item -mone-byte-bool
7251 @opindex -mone-byte-bool
7252 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
7253 By default @samp{sizeof(bool)} is @samp{4} when compiling for
7254 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
7255 option has no effect on x86.
7257 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
7258 to generate code that is not binary compatible with code generated
7259 without that switch. Using this switch may require recompiling all
7260 other modules in a program, including system libraries. Use this
7261 switch to conform to a non-default data model.
7263 @item -mfix-and-continue
7264 @itemx -ffix-and-continue
7265 @itemx -findirect-data
7266 @opindex mfix-and-continue
7267 @opindex ffix-and-continue
7268 @opindex findirect-data
7269 Generate code suitable for fast turn around development. Needed to
7270 enable gdb to dynamically load @code{.o} files into already running
7271 programs. @option{-findirect-data} and @option{-ffix-and-continue}
7272 are provided for backwards compatibility.
7276 Loads all members of static archive libraries.
7277 See man ld(1) for more information.
7279 @item -arch_errors_fatal
7280 @opindex arch_errors_fatal
7281 Cause the errors having to do with files that have the wrong architecture
7285 @opindex bind_at_load
7286 Causes the output file to be marked such that the dynamic linker will
7287 bind all undefined references when the file is loaded or launched.
7291 Produce a Mach-o bundle format file.
7292 See man ld(1) for more information.
7294 @item -bundle_loader @var{executable}
7295 @opindex bundle_loader
7296 This option specifies the @var{executable} that will be loading the build
7297 output file being linked. See man ld(1) for more information.
7300 @opindex -dynamiclib
7301 When passed this option, GCC will produce a dynamic library instead of
7302 an executable when linking, using the Darwin @file{libtool} command.
7304 @item -force_cpusubtype_ALL
7305 @opindex -force_cpusubtype_ALL
7306 This causes GCC's output file to have the @var{ALL} subtype, instead of
7307 one controlled by the @option{-mcpu} or @option{-march} option.
7309 @item -allowable_client @var{client_name}
7311 @itemx -compatibility_version
7312 @itemx -current_version
7314 @itemx -dependency-file
7316 @itemx -dylinker_install_name
7318 @itemx -exported_symbols_list
7320 @itemx -flat_namespace
7321 @itemx -force_flat_namespace
7322 @itemx -headerpad_max_install_names
7325 @itemx -install_name
7326 @itemx -keep_private_externs
7327 @itemx -multi_module
7328 @itemx -multiply_defined
7329 @itemx -multiply_defined_unused
7331 @itemx -no_dead_strip_inits_and_terms
7332 @itemx -nofixprebinding
7335 @itemx -noseglinkedit
7336 @itemx -pagezero_size
7338 @itemx -prebind_all_twolevel_modules
7339 @itemx -private_bundle
7340 @itemx -read_only_relocs
7342 @itemx -sectobjectsymbols
7346 @itemx -sectobjectsymbols
7349 @itemx -segs_read_only_addr
7350 @itemx -segs_read_write_addr
7351 @itemx -seg_addr_table
7352 @itemx -seg_addr_table_filename
7355 @itemx -segs_read_only_addr
7356 @itemx -segs_read_write_addr
7357 @itemx -single_module
7360 @itemx -sub_umbrella
7361 @itemx -twolevel_namespace
7364 @itemx -unexported_symbols_list
7365 @itemx -weak_reference_mismatches
7368 @opindex allowable_client
7369 @opindex client_name
7370 @opindex compatibility_version
7371 @opindex current_version
7373 @opindex dependency-file
7375 @opindex dylinker_install_name
7377 @opindex exported_symbols_list
7379 @opindex flat_namespace
7380 @opindex force_flat_namespace
7381 @opindex headerpad_max_install_names
7384 @opindex install_name
7385 @opindex keep_private_externs
7386 @opindex multi_module
7387 @opindex multiply_defined
7388 @opindex multiply_defined_unused
7390 @opindex no_dead_strip_inits_and_terms
7391 @opindex nofixprebinding
7392 @opindex nomultidefs
7394 @opindex noseglinkedit
7395 @opindex pagezero_size
7397 @opindex prebind_all_twolevel_modules
7398 @opindex private_bundle
7399 @opindex read_only_relocs
7401 @opindex sectobjectsymbols
7405 @opindex sectobjectsymbols
7408 @opindex segs_read_only_addr
7409 @opindex segs_read_write_addr
7410 @opindex seg_addr_table
7411 @opindex seg_addr_table_filename
7412 @opindex seglinkedit
7414 @opindex segs_read_only_addr
7415 @opindex segs_read_write_addr
7416 @opindex single_module
7418 @opindex sub_library
7419 @opindex sub_umbrella
7420 @opindex twolevel_namespace
7423 @opindex unexported_symbols_list
7424 @opindex weak_reference_mismatches
7425 @opindex whatsloaded
7427 These options are passed to the Darwin linker. The Darwin linker man page
7428 describes them in detail.
7431 @node DEC Alpha Options
7432 @subsection DEC Alpha Options
7434 These @samp{-m} options are defined for the DEC Alpha implementations:
7437 @item -mno-soft-float
7439 @opindex mno-soft-float
7440 @opindex msoft-float
7441 Use (do not use) the hardware floating-point instructions for
7442 floating-point operations. When @option{-msoft-float} is specified,
7443 functions in @file{libgcc.a} will be used to perform floating-point
7444 operations. Unless they are replaced by routines that emulate the
7445 floating-point operations, or compiled in such a way as to call such
7446 emulations routines, these routines will issue floating-point
7447 operations. If you are compiling for an Alpha without floating-point
7448 operations, you must ensure that the library is built so as not to call
7451 Note that Alpha implementations without floating-point operations are
7452 required to have floating-point registers.
7457 @opindex mno-fp-regs
7458 Generate code that uses (does not use) the floating-point register set.
7459 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7460 register set is not used, floating point operands are passed in integer
7461 registers as if they were integers and floating-point results are passed
7462 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
7463 so any function with a floating-point argument or return value called by code
7464 compiled with @option{-mno-fp-regs} must also be compiled with that
7467 A typical use of this option is building a kernel that does not use,
7468 and hence need not save and restore, any floating-point registers.
7472 The Alpha architecture implements floating-point hardware optimized for
7473 maximum performance. It is mostly compliant with the IEEE floating
7474 point standard. However, for full compliance, software assistance is
7475 required. This option generates code fully IEEE compliant code
7476 @emph{except} that the @var{inexact-flag} is not maintained (see below).
7477 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
7478 defined during compilation. The resulting code is less efficient but is
7479 able to correctly support denormalized numbers and exceptional IEEE
7480 values such as not-a-number and plus/minus infinity. Other Alpha
7481 compilers call this option @option{-ieee_with_no_inexact}.
7483 @item -mieee-with-inexact
7484 @opindex mieee-with-inexact
7485 This is like @option{-mieee} except the generated code also maintains
7486 the IEEE @var{inexact-flag}. Turning on this option causes the
7487 generated code to implement fully-compliant IEEE math. In addition to
7488 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
7489 macro. On some Alpha implementations the resulting code may execute
7490 significantly slower than the code generated by default. Since there is
7491 very little code that depends on the @var{inexact-flag}, you should
7492 normally not specify this option. Other Alpha compilers call this
7493 option @option{-ieee_with_inexact}.
7495 @item -mfp-trap-mode=@var{trap-mode}
7496 @opindex mfp-trap-mode
7497 This option controls what floating-point related traps are enabled.
7498 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
7499 The trap mode can be set to one of four values:
7503 This is the default (normal) setting. The only traps that are enabled
7504 are the ones that cannot be disabled in software (e.g., division by zero
7508 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7512 Like @samp{su}, but the instructions are marked to be safe for software
7513 completion (see Alpha architecture manual for details).
7516 Like @samp{su}, but inexact traps are enabled as well.
7519 @item -mfp-rounding-mode=@var{rounding-mode}
7520 @opindex mfp-rounding-mode
7521 Selects the IEEE rounding mode. Other Alpha compilers call this option
7522 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
7527 Normal IEEE rounding mode. Floating point numbers are rounded towards
7528 the nearest machine number or towards the even machine number in case
7532 Round towards minus infinity.
7535 Chopped rounding mode. Floating point numbers are rounded towards zero.
7538 Dynamic rounding mode. A field in the floating point control register
7539 (@var{fpcr}, see Alpha architecture reference manual) controls the
7540 rounding mode in effect. The C library initializes this register for
7541 rounding towards plus infinity. Thus, unless your program modifies the
7542 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7545 @item -mtrap-precision=@var{trap-precision}
7546 @opindex mtrap-precision
7547 In the Alpha architecture, floating point traps are imprecise. This
7548 means without software assistance it is impossible to recover from a
7549 floating trap and program execution normally needs to be terminated.
7550 GCC can generate code that can assist operating system trap handlers
7551 in determining the exact location that caused a floating point trap.
7552 Depending on the requirements of an application, different levels of
7553 precisions can be selected:
7557 Program precision. This option is the default and means a trap handler
7558 can only identify which program caused a floating point exception.
7561 Function precision. The trap handler can determine the function that
7562 caused a floating point exception.
7565 Instruction precision. The trap handler can determine the exact
7566 instruction that caused a floating point exception.
7569 Other Alpha compilers provide the equivalent options called
7570 @option{-scope_safe} and @option{-resumption_safe}.
7572 @item -mieee-conformant
7573 @opindex mieee-conformant
7574 This option marks the generated code as IEEE conformant. You must not
7575 use this option unless you also specify @option{-mtrap-precision=i} and either
7576 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
7577 is to emit the line @samp{.eflag 48} in the function prologue of the
7578 generated assembly file. Under DEC Unix, this has the effect that
7579 IEEE-conformant math library routines will be linked in.
7581 @item -mbuild-constants
7582 @opindex mbuild-constants
7583 Normally GCC examines a 32- or 64-bit integer constant to
7584 see if it can construct it from smaller constants in two or three
7585 instructions. If it cannot, it will output the constant as a literal and
7586 generate code to load it from the data segment at runtime.
7588 Use this option to require GCC to construct @emph{all} integer constants
7589 using code, even if it takes more instructions (the maximum is six).
7591 You would typically use this option to build a shared library dynamic
7592 loader. Itself a shared library, it must relocate itself in memory
7593 before it can find the variables and constants in its own data segment.
7599 Select whether to generate code to be assembled by the vendor-supplied
7600 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
7618 Indicate whether GCC should generate code to use the optional BWX,
7619 CIX, FIX and MAX instruction sets. The default is to use the instruction
7620 sets supported by the CPU type specified via @option{-mcpu=} option or that
7621 of the CPU on which GCC was built if none was specified.
7626 @opindex mfloat-ieee
7627 Generate code that uses (does not use) VAX F and G floating point
7628 arithmetic instead of IEEE single and double precision.
7630 @item -mexplicit-relocs
7631 @itemx -mno-explicit-relocs
7632 @opindex mexplicit-relocs
7633 @opindex mno-explicit-relocs
7634 Older Alpha assemblers provided no way to generate symbol relocations
7635 except via assembler macros. Use of these macros does not allow
7636 optimal instruction scheduling. GNU binutils as of version 2.12
7637 supports a new syntax that allows the compiler to explicitly mark
7638 which relocations should apply to which instructions. This option
7639 is mostly useful for debugging, as GCC detects the capabilities of
7640 the assembler when it is built and sets the default accordingly.
7644 @opindex msmall-data
7645 @opindex mlarge-data
7646 When @option{-mexplicit-relocs} is in effect, static data is
7647 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
7648 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
7649 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
7650 16-bit relocations off of the @code{$gp} register. This limits the
7651 size of the small data area to 64KB, but allows the variables to be
7652 directly accessed via a single instruction.
7654 The default is @option{-mlarge-data}. With this option the data area
7655 is limited to just below 2GB@. Programs that require more than 2GB of
7656 data must use @code{malloc} or @code{mmap} to allocate the data in the
7657 heap instead of in the program's data segment.
7659 When generating code for shared libraries, @option{-fpic} implies
7660 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
7664 @opindex msmall-text
7665 @opindex mlarge-text
7666 When @option{-msmall-text} is used, the compiler assumes that the
7667 code of the entire program (or shared library) fits in 4MB, and is
7668 thus reachable with a branch instruction. When @option{-msmall-data}
7669 is used, the compiler can assume that all local symbols share the
7670 same @code{$gp} value, and thus reduce the number of instructions
7671 required for a function call from 4 to 1.
7673 The default is @option{-mlarge-text}.
7675 @item -mcpu=@var{cpu_type}
7677 Set the instruction set and instruction scheduling parameters for
7678 machine type @var{cpu_type}. You can specify either the @samp{EV}
7679 style name or the corresponding chip number. GCC supports scheduling
7680 parameters for the EV4, EV5 and EV6 family of processors and will
7681 choose the default values for the instruction set from the processor
7682 you specify. If you do not specify a processor type, GCC will default
7683 to the processor on which the compiler was built.
7685 Supported values for @var{cpu_type} are
7691 Schedules as an EV4 and has no instruction set extensions.
7695 Schedules as an EV5 and has no instruction set extensions.
7699 Schedules as an EV5 and supports the BWX extension.
7704 Schedules as an EV5 and supports the BWX and MAX extensions.
7708 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
7712 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
7715 @item -mtune=@var{cpu_type}
7717 Set only the instruction scheduling parameters for machine type
7718 @var{cpu_type}. The instruction set is not changed.
7720 @item -mmemory-latency=@var{time}
7721 @opindex mmemory-latency
7722 Sets the latency the scheduler should assume for typical memory
7723 references as seen by the application. This number is highly
7724 dependent on the memory access patterns used by the application
7725 and the size of the external cache on the machine.
7727 Valid options for @var{time} are
7731 A decimal number representing clock cycles.
7737 The compiler contains estimates of the number of clock cycles for
7738 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7739 (also called Dcache, Scache, and Bcache), as well as to main memory.
7740 Note that L3 is only valid for EV5.
7745 @node DEC Alpha/VMS Options
7746 @subsection DEC Alpha/VMS Options
7748 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
7751 @item -mvms-return-codes
7752 @opindex mvms-return-codes
7753 Return VMS condition codes from main. The default is to return POSIX
7754 style condition (e.g.@ error) codes.
7758 @subsection FRV Options
7765 Only use the first 32 general purpose registers.
7770 Use all 64 general purpose registers.
7775 Use only the first 32 floating point registers.
7780 Use all 64 floating point registers
7783 @opindex mhard-float
7785 Use hardware instructions for floating point operations.
7788 @opindex msoft-float
7790 Use library routines for floating point operations.
7795 Dynamically allocate condition code registers.
7800 Do not try to dynamically allocate condition code registers, only
7801 use @code{icc0} and @code{fcc0}.
7806 Change ABI to use double word insns.
7811 Do not use double word instructions.
7816 Use floating point double instructions.
7821 Do not use floating point double instructions.
7826 Use media instructions.
7831 Do not use media instructions.
7836 Use multiply and add/subtract instructions.
7841 Do not use multiply and add/subtract instructions.
7846 Select the FDPIC ABI, that uses function descriptors to represent
7847 pointers to functions. Without any PIC/PIE-related options, it
7848 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
7849 assumes GOT entries and small data are within a 12-bit range from the
7850 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
7851 are computed with 32 bits.
7854 @opindex minline-plt
7856 Enable inlining of PLT entries in function calls to functions that are
7857 not known to bind locally. It has no effect without @option{-mfdpic}.
7858 It's enabled by default if optimizing for speed and compiling for
7859 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
7860 optimization option such as @option{-O3} or above is present in the
7866 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
7867 that is known to be in read-only sections. It's enabled by default,
7868 except for @option{-fpic} or @option{-fpie}: even though it may help
7869 make the global offset table smaller, it trades 1 instruction for 4.
7870 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
7871 one of which may be shared by multiple symbols, and it avoids the need
7872 for a GOT entry for the referenced symbol, so it's more likely to be a
7873 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
7875 @item -multilib-library-pic
7876 @opindex multilib-library-pic
7878 Link with the (library, not FD) pic libraries. It's implied by
7879 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
7880 @option{-fpic} without @option{-mfdpic}. You should never have to use
7886 Follow the EABI requirement of always creating a frame pointer whenever
7887 a stack frame is allocated. This option is enabled by default and can
7888 be disabled with @option{-mno-linked-fp}.
7891 @opindex mlong-calls
7893 Use indirect addressing to call functions outside the current
7894 compilation unit. This allows the functions to be placed anywhere
7895 within the 32-bit address space.
7897 @item -malign-labels
7898 @opindex malign-labels
7900 Try to align labels to an 8-byte boundary by inserting nops into the
7901 previous packet. This option only has an effect when VLIW packing
7902 is enabled. It doesn't create new packets; it merely adds nops to
7906 @opindex mlibrary-pic
7908 Generate position-independent EABI code.
7913 Use only the first four media accumulator registers.
7918 Use all eight media accumulator registers.
7923 Pack VLIW instructions.
7928 Do not pack VLIW instructions.
7933 Do not mark ABI switches in e_flags.
7938 Enable the use of conditional-move instructions (default).
7940 This switch is mainly for debugging the compiler and will likely be removed
7941 in a future version.
7943 @item -mno-cond-move
7944 @opindex mno-cond-move
7946 Disable the use of conditional-move instructions.
7948 This switch is mainly for debugging the compiler and will likely be removed
7949 in a future version.
7954 Enable the use of conditional set instructions (default).
7956 This switch is mainly for debugging the compiler and will likely be removed
7957 in a future version.
7962 Disable the use of conditional set instructions.
7964 This switch is mainly for debugging the compiler and will likely be removed
7965 in a future version.
7970 Enable the use of conditional execution (default).
7972 This switch is mainly for debugging the compiler and will likely be removed
7973 in a future version.
7975 @item -mno-cond-exec
7976 @opindex mno-cond-exec
7978 Disable the use of conditional execution.
7980 This switch is mainly for debugging the compiler and will likely be removed
7981 in a future version.
7984 @opindex mvliw-branch
7986 Run a pass to pack branches into VLIW instructions (default).
7988 This switch is mainly for debugging the compiler and will likely be removed
7989 in a future version.
7991 @item -mno-vliw-branch
7992 @opindex mno-vliw-branch
7994 Do not run a pass to pack branches into VLIW instructions.
7996 This switch is mainly for debugging the compiler and will likely be removed
7997 in a future version.
7999 @item -mmulti-cond-exec
8000 @opindex mmulti-cond-exec
8002 Enable optimization of @code{&&} and @code{||} in conditional execution
8005 This switch is mainly for debugging the compiler and will likely be removed
8006 in a future version.
8008 @item -mno-multi-cond-exec
8009 @opindex mno-multi-cond-exec
8011 Disable optimization of @code{&&} and @code{||} in conditional execution.
8013 This switch is mainly for debugging the compiler and will likely be removed
8014 in a future version.
8016 @item -mnested-cond-exec
8017 @opindex mnested-cond-exec
8019 Enable nested conditional execution optimizations (default).
8021 This switch is mainly for debugging the compiler and will likely be removed
8022 in a future version.
8024 @item -mno-nested-cond-exec
8025 @opindex mno-nested-cond-exec
8027 Disable nested conditional execution optimizations.
8029 This switch is mainly for debugging the compiler and will likely be removed
8030 in a future version.
8032 @item -mtomcat-stats
8033 @opindex mtomcat-stats
8035 Cause gas to print out tomcat statistics.
8037 @item -mcpu=@var{cpu}
8040 Select the processor type for which to generate code. Possible values are
8041 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8042 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8046 @node H8/300 Options
8047 @subsection H8/300 Options
8049 These @samp{-m} options are defined for the H8/300 implementations:
8054 Shorten some address references at link time, when possible; uses the
8055 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8056 ld, Using ld}, for a fuller description.
8060 Generate code for the H8/300H@.
8064 Generate code for the H8S@.
8068 Generate code for the H8S and H8/300H in the normal mode. This switch
8069 must be used either with @option{-mh} or @option{-ms}.
8073 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8077 Make @code{int} data 32 bits by default.
8081 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8082 The default for the H8/300H and H8S is to align longs and floats on 4
8084 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8085 This option has no effect on the H8/300.
8089 @subsection HPPA Options
8090 @cindex HPPA Options
8092 These @samp{-m} options are defined for the HPPA family of computers:
8095 @item -march=@var{architecture-type}
8097 Generate code for the specified architecture. The choices for
8098 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8099 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8100 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8101 architecture option for your machine. Code compiled for lower numbered
8102 architectures will run on higher numbered architectures, but not the
8106 @itemx -mpa-risc-1-1
8107 @itemx -mpa-risc-2-0
8108 @opindex mpa-risc-1-0
8109 @opindex mpa-risc-1-1
8110 @opindex mpa-risc-2-0
8111 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8114 @opindex mbig-switch
8115 Generate code suitable for big switch tables. Use this option only if
8116 the assembler/linker complain about out of range branches within a switch
8119 @item -mjump-in-delay
8120 @opindex mjump-in-delay
8121 Fill delay slots of function calls with unconditional jump instructions
8122 by modifying the return pointer for the function call to be the target
8123 of the conditional jump.
8125 @item -mdisable-fpregs
8126 @opindex mdisable-fpregs
8127 Prevent floating point registers from being used in any manner. This is
8128 necessary for compiling kernels which perform lazy context switching of
8129 floating point registers. If you use this option and attempt to perform
8130 floating point operations, the compiler will abort.
8132 @item -mdisable-indexing
8133 @opindex mdisable-indexing
8134 Prevent the compiler from using indexing address modes. This avoids some
8135 rather obscure problems when compiling MIG generated code under MACH@.
8137 @item -mno-space-regs
8138 @opindex mno-space-regs
8139 Generate code that assumes the target has no space registers. This allows
8140 GCC to generate faster indirect calls and use unscaled index address modes.
8142 Such code is suitable for level 0 PA systems and kernels.
8144 @item -mfast-indirect-calls
8145 @opindex mfast-indirect-calls
8146 Generate code that assumes calls never cross space boundaries. This
8147 allows GCC to emit code which performs faster indirect calls.
8149 This option will not work in the presence of shared libraries or nested
8152 @item -mfixed-range=@var{register-range}
8153 @opindex mfixed-range
8154 Generate code treating the given register range as fixed registers.
8155 A fixed register is one that the register allocator can not use. This is
8156 useful when compiling kernel code. A register range is specified as
8157 two registers separated by a dash. Multiple register ranges can be
8158 specified separated by a comma.
8160 @item -mlong-load-store
8161 @opindex mlong-load-store
8162 Generate 3-instruction load and store sequences as sometimes required by
8163 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8166 @item -mportable-runtime
8167 @opindex mportable-runtime
8168 Use the portable calling conventions proposed by HP for ELF systems.
8172 Enable the use of assembler directives only GAS understands.
8174 @item -mschedule=@var{cpu-type}
8176 Schedule code according to the constraints for the machine type
8177 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8178 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8179 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8180 proper scheduling option for your machine. The default scheduling is
8184 @opindex mlinker-opt
8185 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8186 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8187 linkers in which they give bogus error messages when linking some programs.
8190 @opindex msoft-float
8191 Generate output containing library calls for floating point.
8192 @strong{Warning:} the requisite libraries are not available for all HPPA
8193 targets. Normally the facilities of the machine's usual C compiler are
8194 used, but this cannot be done directly in cross-compilation. You must make
8195 your own arrangements to provide suitable library functions for
8196 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8197 does provide software floating point support.
8199 @option{-msoft-float} changes the calling convention in the output file;
8200 therefore, it is only useful if you compile @emph{all} of a program with
8201 this option. In particular, you need to compile @file{libgcc.a}, the
8202 library that comes with GCC, with @option{-msoft-float} in order for
8207 Generate the predefine, @code{_SIO}, for server IO@. The default is
8208 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8209 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
8210 options are available under HP-UX and HI-UX@.
8214 Use GNU ld specific options. This passes @option{-shared} to ld when
8215 building a shared library. It is the default when GCC is configured,
8216 explicitly or implicitly, with the GNU linker. This option does not
8217 have any affect on which ld is called, it only changes what parameters
8218 are passed to that ld. The ld that is called is determined by the
8219 @option{--with-ld} configure option, GCC's program search path, and
8220 finally by the user's @env{PATH}. The linker used by GCC can be printed
8221 using @samp{which `gcc -print-prog-name=ld`}.
8225 Use HP ld specific options. This passes @option{-b} to ld when building
8226 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8227 links. It is the default when GCC is configured, explicitly or
8228 implicitly, with the HP linker. This option does not have any affect on
8229 which ld is called, it only changes what parameters are passed to that
8230 ld. The ld that is called is determined by the @option{--with-ld}
8231 configure option, GCC's program search path, and finally by the user's
8232 @env{PATH}. The linker used by GCC can be printed using @samp{which
8233 `gcc -print-prog-name=ld`}.
8236 @opindex mno-long-calls
8237 Generate code that uses long call sequences. This ensures that a call
8238 is always able to reach linker generated stubs. The default is to generate
8239 long calls only when the distance from the call site to the beginning
8240 of the function or translation unit, as the case may be, exceeds a
8241 predefined limit set by the branch type being used. The limits for
8242 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8243 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8246 Distances are measured from the beginning of functions when using the
8247 @option{-ffunction-sections} option, or when using the @option{-mgas}
8248 and @option{-mno-portable-runtime} options together under HP-UX with
8251 It is normally not desirable to use this option as it will degrade
8252 performance. However, it may be useful in large applications,
8253 particularly when partial linking is used to build the application.
8255 The types of long calls used depends on the capabilities of the
8256 assembler and linker, and the type of code being generated. The
8257 impact on systems that support long absolute calls, and long pic
8258 symbol-difference or pc-relative calls should be relatively small.
8259 However, an indirect call is used on 32-bit ELF systems in pic code
8260 and it is quite long.
8262 @item -munix=@var{unix-std}
8264 Generate compiler predefines and select a startfile for the specified
8265 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
8266 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
8267 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
8268 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
8269 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
8272 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
8273 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
8274 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
8275 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
8276 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
8277 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
8279 It is @emph{important} to note that this option changes the interfaces
8280 for various library routines. It also affects the operational behavior
8281 of the C library. Thus, @emph{extreme} care is needed in using this
8284 Library code that is intended to operate with more than one UNIX
8285 standard must test, set and restore the variable @var{__xpg4_extended_mask}
8286 as appropriate. Most GNU software doesn't provide this capability.
8290 Suppress the generation of link options to search libdld.sl when the
8291 @option{-static} option is specified on HP-UX 10 and later.
8295 The HP-UX implementation of setlocale in libc has a dependency on
8296 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8297 when the @option{-static} option is specified, special link options
8298 are needed to resolve this dependency.
8300 On HP-UX 10 and later, the GCC driver adds the necessary options to
8301 link with libdld.sl when the @option{-static} option is specified.
8302 This causes the resulting binary to be dynamic. On the 64-bit port,
8303 the linkers generate dynamic binaries by default in any case. The
8304 @option{-nolibdld} option can be used to prevent the GCC driver from
8305 adding these link options.
8309 Add support for multithreading with the @dfn{dce thread} library
8310 under HP-UX@. This option sets flags for both the preprocessor and
8314 @node i386 and x86-64 Options
8315 @subsection Intel 386 and AMD x86-64 Options
8316 @cindex i386 Options
8317 @cindex x86-64 Options
8318 @cindex Intel 386 Options
8319 @cindex AMD x86-64 Options
8321 These @samp{-m} options are defined for the i386 and x86-64 family of
8325 @item -mtune=@var{cpu-type}
8327 Tune to @var{cpu-type} everything applicable about the generated code, except
8328 for the ABI and the set of available instructions. The choices for
8332 Original Intel's i386 CPU@.
8334 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
8336 Intel Pentium CPU with no MMX support.
8338 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8339 @item i686, pentiumpro
8340 Intel PentiumPro CPU@.
8342 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8343 @item pentium3, pentium3m
8344 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8347 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8348 support. Used by Centrino notebooks.
8349 @item pentium4, pentium4m
8350 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8352 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8355 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8356 SSE2 and SSE3 instruction set support.
8358 AMD K6 CPU with MMX instruction set support.
8360 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8361 @item athlon, athlon-tbird
8362 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8364 @item athlon-4, athlon-xp, athlon-mp
8365 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8366 instruction set support.
8367 @item k8, opteron, athlon64, athlon-fx
8368 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8369 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8371 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8374 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8375 instruction set support.
8377 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8378 implemented for this chip.)
8380 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8381 implemented for this chip.)
8384 While picking a specific @var{cpu-type} will schedule things appropriately
8385 for that particular chip, the compiler will not generate any code that
8386 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8389 @item -march=@var{cpu-type}
8391 Generate instructions for the machine type @var{cpu-type}. The choices
8392 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8393 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8395 @item -mcpu=@var{cpu-type}
8397 A deprecated synonym for @option{-mtune}.
8406 @opindex mpentiumpro
8407 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8408 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8409 These synonyms are deprecated.
8411 @item -mfpmath=@var{unit}
8413 Generate floating point arithmetics for selected unit @var{unit}. The choices
8418 Use the standard 387 floating point coprocessor present majority of chips and
8419 emulated otherwise. Code compiled with this option will run almost everywhere.
8420 The temporary results are computed in 80bit precision instead of precision
8421 specified by the type resulting in slightly different results compared to most
8422 of other chips. See @option{-ffloat-store} for more detailed description.
8424 This is the default choice for i386 compiler.
8427 Use scalar floating point instructions present in the SSE instruction set.
8428 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8429 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8430 instruction set supports only single precision arithmetics, thus the double and
8431 extended precision arithmetics is still done using 387. Later version, present
8432 only in Pentium4 and the future AMD x86-64 chips supports double precision
8435 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
8436 or @option{-msse2} switches to enable SSE extensions and make this option
8437 effective. For the x86-64 compiler, these extensions are enabled by default.
8439 The resulting code should be considerably faster in the majority of cases and avoid
8440 the numerical instability problems of 387 code, but may break some existing
8441 code that expects temporaries to be 80bit.
8443 This is the default choice for the x86-64 compiler.
8446 Attempt to utilize both instruction sets at once. This effectively double the
8447 amount of available registers and on chips with separate execution units for
8448 387 and SSE the execution resources too. Use this option with care, as it is
8449 still experimental, because the GCC register allocator does not model separate
8450 functional units well resulting in instable performance.
8453 @item -masm=@var{dialect}
8454 @opindex masm=@var{dialect}
8455 Output asm instructions using selected @var{dialect}. Supported choices are
8456 @samp{intel} or @samp{att} (the default one).
8461 @opindex mno-ieee-fp
8462 Control whether or not the compiler uses IEEE floating point
8463 comparisons. These handle correctly the case where the result of a
8464 comparison is unordered.
8467 @opindex msoft-float
8468 Generate output containing library calls for floating point.
8469 @strong{Warning:} the requisite libraries are not part of GCC@.
8470 Normally the facilities of the machine's usual C compiler are used, but
8471 this can't be done directly in cross-compilation. You must make your
8472 own arrangements to provide suitable library functions for
8475 On machines where a function returns floating point results in the 80387
8476 register stack, some floating point opcodes may be emitted even if
8477 @option{-msoft-float} is used.
8479 @item -mno-fp-ret-in-387
8480 @opindex mno-fp-ret-in-387
8481 Do not use the FPU registers for return values of functions.
8483 The usual calling convention has functions return values of types
8484 @code{float} and @code{double} in an FPU register, even if there
8485 is no FPU@. The idea is that the operating system should emulate
8488 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8489 in ordinary CPU registers instead.
8491 @item -mno-fancy-math-387
8492 @opindex mno-fancy-math-387
8493 Some 387 emulators do not support the @code{sin}, @code{cos} and
8494 @code{sqrt} instructions for the 387. Specify this option to avoid
8495 generating those instructions. This option is the default on FreeBSD,
8496 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8497 indicates that the target cpu will always have an FPU and so the
8498 instruction will not need emulation. As of revision 2.6.1, these
8499 instructions are not generated unless you also use the
8500 @option{-funsafe-math-optimizations} switch.
8502 @item -malign-double
8503 @itemx -mno-align-double
8504 @opindex malign-double
8505 @opindex mno-align-double
8506 Control whether GCC aligns @code{double}, @code{long double}, and
8507 @code{long long} variables on a two word boundary or a one word
8508 boundary. Aligning @code{double} variables on a two word boundary will
8509 produce code that runs somewhat faster on a @samp{Pentium} at the
8510 expense of more memory.
8512 @strong{Warning:} if you use the @option{-malign-double} switch,
8513 structures containing the above types will be aligned differently than
8514 the published application binary interface specifications for the 386
8515 and will not be binary compatible with structures in code compiled
8516 without that switch.
8518 @item -m96bit-long-double
8519 @itemx -m128bit-long-double
8520 @opindex m96bit-long-double
8521 @opindex m128bit-long-double
8522 These switches control the size of @code{long double} type. The i386
8523 application binary interface specifies the size to be 96 bits,
8524 so @option{-m96bit-long-double} is the default in 32 bit mode.
8526 Modern architectures (Pentium and newer) would prefer @code{long double}
8527 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8528 conforming to the ABI, this would not be possible. So specifying a
8529 @option{-m128bit-long-double} will align @code{long double}
8530 to a 16 byte boundary by padding the @code{long double} with an additional
8533 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8534 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8536 Notice that neither of these options enable any extra precision over the x87
8537 standard of 80 bits for a @code{long double}.
8539 @strong{Warning:} if you override the default value for your target ABI, the
8540 structures and arrays containing @code{long double} variables will change
8541 their size as well as function calling convention for function taking
8542 @code{long double} will be modified. Hence they will not be binary
8543 compatible with arrays or structures in code compiled without that switch.
8547 @itemx -mno-svr3-shlib
8548 @opindex msvr3-shlib
8549 @opindex mno-svr3-shlib
8550 Control whether GCC places uninitialized local variables into the
8551 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8552 into @code{bss}. These options are meaningful only on System V Release 3.
8556 Use a different function-calling convention, in which functions that
8557 take a fixed number of arguments return with the @code{ret} @var{num}
8558 instruction, which pops their arguments while returning. This saves one
8559 instruction in the caller since there is no need to pop the arguments
8562 You can specify that an individual function is called with this calling
8563 sequence with the function attribute @samp{stdcall}. You can also
8564 override the @option{-mrtd} option by using the function attribute
8565 @samp{cdecl}. @xref{Function Attributes}.
8567 @strong{Warning:} this calling convention is incompatible with the one
8568 normally used on Unix, so you cannot use it if you need to call
8569 libraries compiled with the Unix compiler.
8571 Also, you must provide function prototypes for all functions that
8572 take variable numbers of arguments (including @code{printf});
8573 otherwise incorrect code will be generated for calls to those
8576 In addition, seriously incorrect code will result if you call a
8577 function with too many arguments. (Normally, extra arguments are
8578 harmlessly ignored.)
8580 @item -mregparm=@var{num}
8582 Control how many registers are used to pass integer arguments. By
8583 default, no registers are used to pass arguments, and at most 3
8584 registers can be used. You can control this behavior for a specific
8585 function by using the function attribute @samp{regparm}.
8586 @xref{Function Attributes}.
8588 @strong{Warning:} if you use this switch, and
8589 @var{num} is nonzero, then you must build all modules with the same
8590 value, including any libraries. This includes the system libraries and
8593 @item -mpreferred-stack-boundary=@var{num}
8594 @opindex mpreferred-stack-boundary
8595 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8596 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8597 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8598 size (@option{-Os}), in which case the default is the minimum correct
8599 alignment (4 bytes for x86, and 8 bytes for x86-64).
8601 On Pentium and PentiumPro, @code{double} and @code{long double} values
8602 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8603 suffer significant run time performance penalties. On Pentium III, the
8604 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8605 penalties if it is not 16 byte aligned.
8607 To ensure proper alignment of this values on the stack, the stack boundary
8608 must be as aligned as that required by any value stored on the stack.
8609 Further, every function must be generated such that it keeps the stack
8610 aligned. Thus calling a function compiled with a higher preferred
8611 stack boundary from a function compiled with a lower preferred stack
8612 boundary will most likely misalign the stack. It is recommended that
8613 libraries that use callbacks always use the default setting.
8615 This extra alignment does consume extra stack space, and generally
8616 increases code size. Code that is sensitive to stack space usage, such
8617 as embedded systems and operating system kernels, may want to reduce the
8618 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8636 These switches enable or disable the use of built-in functions that allow
8637 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8640 @xref{X86 Built-in Functions}, for details of the functions enabled
8641 and disabled by these switches.
8643 To have SSE/SSE2 instructions generated automatically from floating-point
8644 code, see @option{-mfpmath=sse}.
8647 @itemx -mno-push-args
8649 @opindex mno-push-args
8650 Use PUSH operations to store outgoing parameters. This method is shorter
8651 and usually equally fast as method using SUB/MOV operations and is enabled
8652 by default. In some cases disabling it may improve performance because of
8653 improved scheduling and reduced dependencies.
8655 @item -maccumulate-outgoing-args
8656 @opindex maccumulate-outgoing-args
8657 If enabled, the maximum amount of space required for outgoing arguments will be
8658 computed in the function prologue. This is faster on most modern CPUs
8659 because of reduced dependencies, improved scheduling and reduced stack usage
8660 when preferred stack boundary is not equal to 2. The drawback is a notable
8661 increase in code size. This switch implies @option{-mno-push-args}.
8665 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8666 on thread-safe exception handling must compile and link all code with the
8667 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8668 @option{-D_MT}; when linking, it links in a special thread helper library
8669 @option{-lmingwthrd} which cleans up per thread exception handling data.
8671 @item -mno-align-stringops
8672 @opindex mno-align-stringops
8673 Do not align destination of inlined string operations. This switch reduces
8674 code size and improves performance in case the destination is already aligned,
8675 but GCC doesn't know about it.
8677 @item -minline-all-stringops
8678 @opindex minline-all-stringops
8679 By default GCC inlines string operations only when destination is known to be
8680 aligned at least to 4 byte boundary. This enables more inlining, increase code
8681 size, but may improve performance of code that depends on fast memcpy, strlen
8682 and memset for short lengths.
8684 @item -momit-leaf-frame-pointer
8685 @opindex momit-leaf-frame-pointer
8686 Don't keep the frame pointer in a register for leaf functions. This
8687 avoids the instructions to save, set up and restore frame pointers and
8688 makes an extra register available in leaf functions. The option
8689 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8690 which might make debugging harder.
8692 @item -mtls-direct-seg-refs
8693 @itemx -mno-tls-direct-seg-refs
8694 @opindex mtls-direct-seg-refs
8695 Controls whether TLS variables may be accessed with offsets from the
8696 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8697 or whether the thread base pointer must be added. Whether or not this
8698 is legal depends on the operating system, and whether it maps the
8699 segment to cover the entire TLS area.
8701 For systems that use GNU libc, the default is on.
8704 These @samp{-m} switches are supported in addition to the above
8705 on AMD x86-64 processors in 64-bit environments.
8712 Generate code for a 32-bit or 64-bit environment.
8713 The 32-bit environment sets int, long and pointer to 32 bits and
8714 generates code that runs on any i386 system.
8715 The 64-bit environment sets int to 32 bits and long and pointer
8716 to 64 bits and generates code for AMD's x86-64 architecture.
8719 @opindex no-red-zone
8720 Do not use a so called red zone for x86-64 code. The red zone is mandated
8721 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8722 stack pointer that will not be modified by signal or interrupt handlers
8723 and therefore can be used for temporary data without adjusting the stack
8724 pointer. The flag @option{-mno-red-zone} disables this red zone.
8726 @item -mcmodel=small
8727 @opindex mcmodel=small
8728 Generate code for the small code model: the program and its symbols must
8729 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8730 Programs can be statically or dynamically linked. This is the default
8733 @item -mcmodel=kernel
8734 @opindex mcmodel=kernel
8735 Generate code for the kernel code model. The kernel runs in the
8736 negative 2 GB of the address space.
8737 This model has to be used for Linux kernel code.
8739 @item -mcmodel=medium
8740 @opindex mcmodel=medium
8741 Generate code for the medium model: The program is linked in the lower 2
8742 GB of the address space but symbols can be located anywhere in the
8743 address space. Programs can be statically or dynamically linked, but
8744 building of shared libraries are not supported with the medium model.
8746 @item -mcmodel=large
8747 @opindex mcmodel=large
8748 Generate code for the large model: This model makes no assumptions
8749 about addresses and sizes of sections. Currently GCC does not implement
8754 @subsection IA-64 Options
8755 @cindex IA-64 Options
8757 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8761 @opindex mbig-endian
8762 Generate code for a big endian target. This is the default for HP-UX@.
8764 @item -mlittle-endian
8765 @opindex mlittle-endian
8766 Generate code for a little endian target. This is the default for AIX5
8773 Generate (or don't) code for the GNU assembler. This is the default.
8774 @c Also, this is the default if the configure option @option{--with-gnu-as}
8781 Generate (or don't) code for the GNU linker. This is the default.
8782 @c Also, this is the default if the configure option @option{--with-gnu-ld}
8787 Generate code that does not use a global pointer register. The result
8788 is not position independent code, and violates the IA-64 ABI@.
8790 @item -mvolatile-asm-stop
8791 @itemx -mno-volatile-asm-stop
8792 @opindex mvolatile-asm-stop
8793 @opindex mno-volatile-asm-stop
8794 Generate (or don't) a stop bit immediately before and after volatile asm
8799 Generate code that works around Itanium B step errata.
8801 @item -mregister-names
8802 @itemx -mno-register-names
8803 @opindex mregister-names
8804 @opindex mno-register-names
8805 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8806 the stacked registers. This may make assembler output more readable.
8812 Disable (or enable) optimizations that use the small data section. This may
8813 be useful for working around optimizer bugs.
8816 @opindex mconstant-gp
8817 Generate code that uses a single constant global pointer value. This is
8818 useful when compiling kernel code.
8822 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
8823 This is useful when compiling firmware code.
8825 @item -minline-float-divide-min-latency
8826 @opindex minline-float-divide-min-latency
8827 Generate code for inline divides of floating point values
8828 using the minimum latency algorithm.
8830 @item -minline-float-divide-max-throughput
8831 @opindex minline-float-divide-max-throughput
8832 Generate code for inline divides of floating point values
8833 using the maximum throughput algorithm.
8835 @item -minline-int-divide-min-latency
8836 @opindex minline-int-divide-min-latency
8837 Generate code for inline divides of integer values
8838 using the minimum latency algorithm.
8840 @item -minline-int-divide-max-throughput
8841 @opindex minline-int-divide-max-throughput
8842 Generate code for inline divides of integer values
8843 using the maximum throughput algorithm.
8845 @item -minline-sqrt-min-latency
8846 @opindex minline-sqrt-min-latency
8847 Generate code for inline square roots
8848 using the minimum latency algorithm.
8850 @item -minline-sqrt-max-throughput
8851 @opindex minline-sqrt-max-throughput
8852 Generate code for inline square roots
8853 using the maximum throughput algorithm.
8855 @item -mno-dwarf2-asm
8857 @opindex mno-dwarf2-asm
8858 @opindex mdwarf2-asm
8859 Don't (or do) generate assembler code for the DWARF2 line number debugging
8860 info. This may be useful when not using the GNU assembler.
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.
8870 @item -mfixed-range=@var{register-range}
8871 @opindex mfixed-range
8872 Generate code treating the given register range as fixed registers.
8873 A fixed register is one that the register allocator can not use. This is
8874 useful when compiling kernel code. A register range is specified as
8875 two registers separated by a dash. Multiple register ranges can be
8876 specified separated by a comma.
8878 @item -mtls-size=@var{tls-size}
8880 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
8883 @item -mtune-arch=@var{cpu-type}
8885 Tune the instruction scheduling for a particular CPU, Valid values are
8886 itanium, itanium1, merced, itanium2, and mckinley.
8892 Add support for multithreading using the POSIX threads library. This
8893 option sets flags for both the preprocessor and linker. It does
8894 not affect the thread safety of object code produced by the compiler or
8895 that of libraries supplied with it. These are HP-UX specific flags.
8901 Generate code for a 32-bit or 64-bit environment.
8902 The 32-bit environment sets int, long and pointer to 32 bits.
8903 The 64-bit environment sets int to 32 bits and long and pointer
8904 to 64 bits. These are HP-UX specific flags.
8908 @node M32R/D Options
8909 @subsection M32R/D Options
8910 @cindex M32R/D options
8912 These @option{-m} options are defined for Renesas M32R/D architectures:
8917 Generate code for the M32R/2@.
8921 Generate code for the M32R/X@.
8925 Generate code for the M32R@. This is the default.
8928 @opindex mmodel=small
8929 Assume all objects live in the lower 16MB of memory (so that their addresses
8930 can be loaded with the @code{ld24} instruction), and assume all subroutines
8931 are reachable with the @code{bl} instruction.
8932 This is the default.
8934 The addressability of a particular object can be set with the
8935 @code{model} attribute.
8937 @item -mmodel=medium
8938 @opindex mmodel=medium
8939 Assume objects may be anywhere in the 32-bit address space (the compiler
8940 will generate @code{seth/add3} instructions to load their addresses), and
8941 assume all subroutines are reachable with the @code{bl} instruction.
8944 @opindex mmodel=large
8945 Assume objects may be anywhere in the 32-bit address space (the compiler
8946 will generate @code{seth/add3} instructions to load their addresses), and
8947 assume subroutines may not be reachable with the @code{bl} instruction
8948 (the compiler will generate the much slower @code{seth/add3/jl}
8949 instruction sequence).
8952 @opindex msdata=none
8953 Disable use of the small data area. Variables will be put into
8954 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
8955 @code{section} attribute has been specified).
8956 This is the default.
8958 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
8959 Objects may be explicitly put in the small data area with the
8960 @code{section} attribute using one of these sections.
8963 @opindex msdata=sdata
8964 Put small global and static data in the small data area, but do not
8965 generate special code to reference them.
8969 Put small global and static data in the small data area, and generate
8970 special instructions to reference them.
8974 @cindex smaller data references
8975 Put global and static objects less than or equal to @var{num} bytes
8976 into the small data or bss sections instead of the normal data or bss
8977 sections. The default value of @var{num} is 8.
8978 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
8979 for this option to have any effect.
8981 All modules should be compiled with the same @option{-G @var{num}} value.
8982 Compiling with different values of @var{num} may or may not work; if it
8983 doesn't the linker will give an error message---incorrect code will not be
8988 Makes the M32R specific code in the compiler display some statistics
8989 that might help in debugging programs.
8992 @opindex malign-loops
8993 Align all loops to a 32-byte boundary.
8995 @item -mno-align-loops
8996 @opindex mno-align-loops
8997 Do not enforce a 32-byte alignment for loops. This is the default.
8999 @item -missue-rate=@var{number}
9000 @opindex missue-rate=@var{number}
9001 Issue @var{number} instructions per cycle. @var{number} can only be 1
9004 @item -mbranch-cost=@var{number}
9005 @opindex mbranch-cost=@var{number}
9006 @var{number} can only be 1 or 2. If it is 1 then branches will be
9007 preferred over conditional code, if it is 2, then the opposite will
9010 @item -mflush-trap=@var{number}
9011 @opindex mflush-trap=@var{number}
9012 Specifies the trap number to use to flush the cache. The default is
9013 12. Valid numbers are between 0 and 15 inclusive.
9015 @item -mno-flush-trap
9016 @opindex mno-flush-trap
9017 Specifies that the cache cannot be flushed by using a trap.
9019 @item -mflush-func=@var{name}
9020 @opindex mflush-func=@var{name}
9021 Specifies the name of the operating system function to call to flush
9022 the cache. The default is @emph{_flush_cache}, but a function call
9023 will only be used if a trap is not available.
9025 @item -mno-flush-func
9026 @opindex mno-flush-func
9027 Indicates that there is no OS function for flushing the cache.
9031 @node M680x0 Options
9032 @subsection M680x0 Options
9033 @cindex M680x0 options
9035 These are the @samp{-m} options defined for the 68000 series. The default
9036 values for these options depends on which style of 68000 was selected when
9037 the compiler was configured; the defaults for the most common choices are
9045 Generate output for a 68000. This is the default
9046 when the compiler is configured for 68000-based systems.
9048 Use this option for microcontrollers with a 68000 or EC000 core,
9049 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
9055 Generate output for a 68020. This is the default
9056 when the compiler is configured for 68020-based systems.
9060 Generate output containing 68881 instructions for floating point.
9061 This is the default for most 68020 systems unless @option{--nfp} was
9062 specified when the compiler was configured.
9066 Generate output for a 68030. This is the default when the compiler is
9067 configured for 68030-based systems.
9071 Generate output for a 68040. This is the default when the compiler is
9072 configured for 68040-based systems.
9074 This option inhibits the use of 68881/68882 instructions that have to be
9075 emulated by software on the 68040. Use this option if your 68040 does not
9076 have code to emulate those instructions.
9080 Generate output for a 68060. This is the default when the compiler is
9081 configured for 68060-based systems.
9083 This option inhibits the use of 68020 and 68881/68882 instructions that
9084 have to be emulated by software on the 68060. Use this option if your 68060
9085 does not have code to emulate those instructions.
9089 Generate output for a CPU32. This is the default
9090 when the compiler is configured for CPU32-based systems.
9092 Use this option for microcontrollers with a
9093 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
9094 68336, 68340, 68341, 68349 and 68360.
9098 Generate output for a 520X ``coldfire'' family cpu. This is the default
9099 when the compiler is configured for 520X-based systems.
9101 Use this option for microcontroller with a 5200 core, including
9102 the MCF5202, MCF5203, MCF5204 and MCF5202.
9107 Generate output for a 68040, without using any of the new instructions.
9108 This results in code which can run relatively efficiently on either a
9109 68020/68881 or a 68030 or a 68040. The generated code does use the
9110 68881 instructions that are emulated on the 68040.
9114 Generate output for a 68060, without using any of the new instructions.
9115 This results in code which can run relatively efficiently on either a
9116 68020/68881 or a 68030 or a 68040. The generated code does use the
9117 68881 instructions that are emulated on the 68060.
9120 @opindex msoft-float
9121 Generate output containing library calls for floating point.
9122 @strong{Warning:} the requisite libraries are not available for all m68k
9123 targets. Normally the facilities of the machine's usual C compiler are
9124 used, but this can't be done directly in cross-compilation. You must
9125 make your own arrangements to provide suitable library functions for
9126 cross-compilation. The embedded targets @samp{m68k-*-aout} and
9127 @samp{m68k-*-coff} do provide software floating point support.
9131 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9132 Additionally, parameters passed on the stack are also aligned to a
9133 16-bit boundary even on targets whose API mandates promotion to 32-bit.
9136 @opindex mnobitfield
9137 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
9138 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
9142 Do use the bit-field instructions. The @option{-m68020} option implies
9143 @option{-mbitfield}. This is the default if you use a configuration
9144 designed for a 68020.
9148 Use a different function-calling convention, in which functions
9149 that take a fixed number of arguments return with the @code{rtd}
9150 instruction, which pops their arguments while returning. This
9151 saves one instruction in the caller since there is no need to pop
9152 the arguments there.
9154 This calling convention is incompatible with the one normally
9155 used on Unix, so you cannot use it if you need to call libraries
9156 compiled with the Unix compiler.
9158 Also, you must provide function prototypes for all functions that
9159 take variable numbers of arguments (including @code{printf});
9160 otherwise incorrect code will be generated for calls to those
9163 In addition, seriously incorrect code will result if you call a
9164 function with too many arguments. (Normally, extra arguments are
9165 harmlessly ignored.)
9167 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
9168 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
9171 @itemx -mno-align-int
9173 @opindex mno-align-int
9174 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
9175 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
9176 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
9177 Aligning variables on 32-bit boundaries produces code that runs somewhat
9178 faster on processors with 32-bit busses at the expense of more memory.
9180 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
9181 align structures containing the above types differently than
9182 most published application binary interface specifications for the m68k.
9186 Use the pc-relative addressing mode of the 68000 directly, instead of
9187 using a global offset table. At present, this option implies @option{-fpic},
9188 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
9189 not presently supported with @option{-mpcrel}, though this could be supported for
9190 68020 and higher processors.
9192 @item -mno-strict-align
9193 @itemx -mstrict-align
9194 @opindex mno-strict-align
9195 @opindex mstrict-align
9196 Do not (do) assume that unaligned memory references will be handled by
9200 Generate code that allows the data segment to be located in a different
9201 area of memory from the text segment. This allows for execute in place in
9202 an environment without virtual memory management. This option implies
9206 Generate code that assumes that the data segment follows the text segment.
9207 This is the default.
9209 @item -mid-shared-library
9210 Generate code that supports shared libraries via the library ID method.
9211 This allows for execute in place and shared libraries in an environment
9212 without virtual memory management. This option implies @option{-fPIC}.
9214 @item -mno-id-shared-library
9215 Generate code that doesn't assume ID based shared libraries are being used.
9216 This is the default.
9218 @item -mshared-library-id=n
9219 Specified the identification number of the ID based shared library being
9220 compiled. Specifying a value of 0 will generate more compact code, specifying
9221 other values will force the allocation of that number to the current
9222 library but is no more space or time efficient than omitting this option.
9226 @node M68hc1x Options
9227 @subsection M68hc1x Options
9228 @cindex M68hc1x options
9230 These are the @samp{-m} options defined for the 68hc11 and 68hc12
9231 microcontrollers. The default values for these options depends on
9232 which style of microcontroller was selected when the compiler was configured;
9233 the defaults for the most common choices are given below.
9240 Generate output for a 68HC11. This is the default
9241 when the compiler is configured for 68HC11-based systems.
9247 Generate output for a 68HC12. This is the default
9248 when the compiler is configured for 68HC12-based systems.
9254 Generate output for a 68HCS12.
9257 @opindex mauto-incdec
9258 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
9265 Enable the use of 68HC12 min and max instructions.
9268 @itemx -mno-long-calls
9269 @opindex mlong-calls
9270 @opindex mno-long-calls
9271 Treat all calls as being far away (near). If calls are assumed to be
9272 far away, the compiler will use the @code{call} instruction to
9273 call a function and the @code{rtc} instruction for returning.
9277 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9279 @item -msoft-reg-count=@var{count}
9280 @opindex msoft-reg-count
9281 Specify the number of pseudo-soft registers which are used for the
9282 code generation. The maximum number is 32. Using more pseudo-soft
9283 register may or may not result in better code depending on the program.
9284 The default is 4 for 68HC11 and 2 for 68HC12.
9289 @subsection MCore Options
9290 @cindex MCore options
9292 These are the @samp{-m} options defined for the Motorola M*Core
9300 @opindex mno-hardlit
9301 Inline constants into the code stream if it can be done in two
9302 instructions or less.
9308 Use the divide instruction. (Enabled by default).
9310 @item -mrelax-immediate
9311 @itemx -mno-relax-immediate
9312 @opindex mrelax-immediate
9313 @opindex mno-relax-immediate
9314 Allow arbitrary sized immediates in bit operations.
9316 @item -mwide-bitfields
9317 @itemx -mno-wide-bitfields
9318 @opindex mwide-bitfields
9319 @opindex mno-wide-bitfields
9320 Always treat bit-fields as int-sized.
9322 @item -m4byte-functions
9323 @itemx -mno-4byte-functions
9324 @opindex m4byte-functions
9325 @opindex mno-4byte-functions
9326 Force all functions to be aligned to a four byte boundary.
9328 @item -mcallgraph-data
9329 @itemx -mno-callgraph-data
9330 @opindex mcallgraph-data
9331 @opindex mno-callgraph-data
9332 Emit callgraph information.
9335 @itemx -mno-slow-bytes
9336 @opindex mslow-bytes
9337 @opindex mno-slow-bytes
9338 Prefer word access when reading byte quantities.
9340 @item -mlittle-endian
9342 @opindex mlittle-endian
9343 @opindex mbig-endian
9344 Generate code for a little endian target.
9350 Generate code for the 210 processor.
9354 @subsection MIPS Options
9355 @cindex MIPS options
9361 Generate big-endian code.
9365 Generate little-endian code. This is the default for @samp{mips*el-*-*}
9368 @item -march=@var{arch}
9370 Generate code that will run on @var{arch}, which can be the name of a
9371 generic MIPS ISA, or the name of a particular processor.
9373 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
9374 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
9375 The processor names are:
9376 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
9378 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
9379 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
9383 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
9384 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
9385 The special value @samp{from-abi} selects the
9386 most compatible architecture for the selected ABI (that is,
9387 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
9389 In processor names, a final @samp{000} can be abbreviated as @samp{k}
9390 (for example, @samp{-march=r2k}). Prefixes are optional, and
9391 @samp{vr} may be written @samp{r}.
9393 GCC defines two macros based on the value of this option. The first
9394 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
9395 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
9396 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
9397 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
9398 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
9400 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
9401 above. In other words, it will have the full prefix and will not
9402 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
9403 the macro names the resolved architecture (either @samp{"mips1"} or
9404 @samp{"mips3"}). It names the default architecture when no
9405 @option{-march} option is given.
9407 @item -mtune=@var{arch}
9409 Optimize for @var{arch}. Among other things, this option controls
9410 the way instructions are scheduled, and the perceived cost of arithmetic
9411 operations. The list of @var{arch} values is the same as for
9414 When this option is not used, GCC will optimize for the processor
9415 specified by @option{-march}. By using @option{-march} and
9416 @option{-mtune} together, it is possible to generate code that will
9417 run on a family of processors, but optimize the code for one
9418 particular member of that family.
9420 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
9421 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
9422 @samp{-march} ones described above.
9426 Equivalent to @samp{-march=mips1}.
9430 Equivalent to @samp{-march=mips2}.
9434 Equivalent to @samp{-march=mips3}.
9438 Equivalent to @samp{-march=mips4}.
9442 Equivalent to @samp{-march=mips32}.
9446 Equivalent to @samp{-march=mips32r2}.
9450 Equivalent to @samp{-march=mips64}.
9456 Use (do not use) the MIPS16 ISA@.
9468 Generate code for the given ABI@.
9470 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
9471 generates 64-bit code when you select a 64-bit architecture, but you
9472 can use @option{-mgp32} to get 32-bit code instead.
9474 For information about the O64 ABI, see
9475 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
9478 @itemx -mno-abicalls
9480 @opindex mno-abicalls
9481 Generate (do not generate) SVR4-style position-independent code.
9482 @option{-mabicalls} is the default for SVR4-based systems.
9488 Lift (do not lift) the usual restrictions on the size of the global
9491 GCC normally uses a single instruction to load values from the GOT@.
9492 While this is relatively efficient, it will only work if the GOT
9493 is smaller than about 64k. Anything larger will cause the linker
9494 to report an error such as:
9496 @cindex relocation truncated to fit (MIPS)
9498 relocation truncated to fit: R_MIPS_GOT16 foobar
9501 If this happens, you should recompile your code with @option{-mxgot}.
9502 It should then work with very large GOTs, although it will also be
9503 less efficient, since it will take three instructions to fetch the
9504 value of a global symbol.
9506 Note that some linkers can create multiple GOTs. If you have such a
9507 linker, you should only need to use @option{-mxgot} when a single object
9508 file accesses more than 64k's worth of GOT entries. Very few do.
9510 These options have no effect unless GCC is generating position
9515 Assume that general-purpose registers are 32 bits wide.
9519 Assume that general-purpose registers are 64 bits wide.
9523 Assume that floating-point registers are 32 bits wide.
9527 Assume that floating-point registers are 64 bits wide.
9530 @opindex mhard-float
9531 Use floating-point coprocessor instructions.
9534 @opindex msoft-float
9535 Do not use floating-point coprocessor instructions. Implement
9536 floating-point calculations using library calls instead.
9538 @item -msingle-float
9539 @opindex msingle-float
9540 Assume that the floating-point coprocessor only supports single-precision
9543 @itemx -mdouble-float
9544 @opindex mdouble-float
9545 Assume that the floating-point coprocessor supports double-precision
9546 operations. This is the default.
9548 @itemx -mpaired-single
9549 @itemx -mno-paired-single
9550 @opindex mpaired-single
9551 @opindex mno-paired-single
9552 Use (do not use) paired-single floating-point instructions.
9553 @xref{MIPS Paired-Single Support}. This option can only be used
9554 when generating 64-bit code and requires hardware floating-point
9555 support to be enabled.
9561 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
9562 The option @option{-mips3d} implies @option{-mpaired-single}.
9566 Force @code{int} and @code{long} types to be 64 bits wide. See
9567 @option{-mlong32} for an explanation of the default and the way
9568 that the pointer size is determined.
9572 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
9573 an explanation of the default and the way that the pointer size is
9578 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
9580 The default size of @code{int}s, @code{long}s and pointers depends on
9581 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
9582 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
9583 32-bit @code{long}s. Pointers are the same size as @code{long}s,
9584 or the same size as integer registers, whichever is smaller.
9588 @cindex smaller data references (MIPS)
9589 @cindex gp-relative references (MIPS)
9590 Put global and static items less than or equal to @var{num} bytes into
9591 the small data or bss section instead of the normal data or bss section.
9592 This allows the data to be accessed using a single instruction.
9594 All modules should be compiled with the same @option{-G @var{num}}
9597 @item -membedded-data
9598 @itemx -mno-embedded-data
9599 @opindex membedded-data
9600 @opindex mno-embedded-data
9601 Allocate variables to the read-only data section first if possible, then
9602 next in the small data section if possible, otherwise in data. This gives
9603 slightly slower code than the default, but reduces the amount of RAM required
9604 when executing, and thus may be preferred for some embedded systems.
9606 @item -muninit-const-in-rodata
9607 @itemx -mno-uninit-const-in-rodata
9608 @opindex muninit-const-in-rodata
9609 @opindex mno-uninit-const-in-rodata
9610 Put uninitialized @code{const} variables in the read-only data section.
9611 This option is only meaningful in conjunction with @option{-membedded-data}.
9613 @item -msplit-addresses
9614 @itemx -mno-split-addresses
9615 @opindex msplit-addresses
9616 @opindex mno-split-addresses
9617 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
9618 relocation operators. This option has been superceded by
9619 @option{-mexplicit-relocs} but is retained for backwards compatibility.
9621 @item -mexplicit-relocs
9622 @itemx -mno-explicit-relocs
9623 @opindex mexplicit-relocs
9624 @opindex mno-explicit-relocs
9625 Use (do not use) assembler relocation operators when dealing with symbolic
9626 addresses. The alternative, selected by @option{-mno-explicit-relocs},
9627 is to use assembler macros instead.
9629 @option{-mexplicit-relocs} is the default if GCC was configured
9630 to use an assembler that supports relocation operators.
9632 @item -mcheck-zero-division
9633 @itemx -mno-check-zero-division
9634 @opindex mcheck-zero-division
9635 @opindex mno-check-zero-division
9636 Trap (do not trap) on integer division by zero. The default is
9637 @option{-mcheck-zero-division}.
9639 @item -mdivide-traps
9640 @itemx -mdivide-breaks
9641 @opindex mdivide-traps
9642 @opindex mdivide-breaks
9643 MIPS systems check for division by zero by generating either a
9644 conditional trap or a break instruction. Using traps results in
9645 smaller code, but is only supported on MIPS II and later. Also, some
9646 versions of the Linux kernel have a bug that prevents trap from
9647 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
9648 allow conditional traps on architectures that support them and
9649 @option{-mdivide-breaks} to force the use of breaks.
9651 The default is usually @option{-mdivide-traps}, but this can be
9652 overridden at configure time using @option{--with-divide=breaks}.
9653 Divide-by-zero checks can be completely disabled using
9654 @option{-mno-check-zero-division}.
9660 Force (do not force) the use of @code{memcpy()} for non-trivial block
9661 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
9662 most constant-sized copies.
9665 @itemx -mno-long-calls
9666 @opindex mlong-calls
9667 @opindex mno-long-calls
9668 Disable (do not disable) use of the @code{jal} instruction. Calling
9669 functions using @code{jal} is more efficient but requires the caller
9670 and callee to be in the same 256 megabyte segment.
9672 This option has no effect on abicalls code. The default is
9673 @option{-mno-long-calls}.
9679 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
9680 instructions, as provided by the R4650 ISA@.
9683 @itemx -mno-fused-madd
9684 @opindex mfused-madd
9685 @opindex mno-fused-madd
9686 Enable (disable) use of the floating point multiply-accumulate
9687 instructions, when they are available. The default is
9688 @option{-mfused-madd}.
9690 When multiply-accumulate instructions are used, the intermediate
9691 product is calculated to infinite precision and is not subject to
9692 the FCSR Flush to Zero bit. This may be undesirable in some
9697 Tell the MIPS assembler to not run its preprocessor over user
9698 assembler files (with a @samp{.s} suffix) when assembling them.
9701 @itemx -mno-fix-r4000
9703 @opindex mno-fix-r4000
9704 Work around certain R4000 CPU errata:
9707 A double-word or a variable shift may give an incorrect result if executed
9708 immediately after starting an integer division.
9710 A double-word or a variable shift may give an incorrect result if executed
9711 while an integer multiplication is in progress.
9713 An integer division may give an incorrect result if started in a delay slot
9714 of a taken branch or a jump.
9718 @itemx -mno-fix-r4400
9720 @opindex mno-fix-r4400
9721 Work around certain R4400 CPU errata:
9724 A double-word or a variable shift may give an incorrect result if executed
9725 immediately after starting an integer division.
9729 @itemx -mno-fix-vr4120
9730 @opindex mfix-vr4120
9731 Work around certain VR4120 errata:
9734 @code{dmultu} does not always produce the correct result.
9736 @code{div} and @code{ddiv} do not always produce the correct result if one
9737 of the operands is negative.
9739 The workarounds for the division errata rely on special functions in
9740 @file{libgcc.a}. At present, these functions are only provided by
9741 the @code{mips64vr*-elf} configurations.
9743 Other VR4120 errata require a nop to be inserted between certain pairs of
9744 instructions. These errata are handled by the assembler, not by GCC itself.
9749 Work around certain SB-1 CPU core errata.
9750 (This flag currently works around the SB-1 revision 2
9751 ``F1'' and ``F2'' floating point errata.)
9753 @item -mflush-func=@var{func}
9754 @itemx -mno-flush-func
9755 @opindex mflush-func
9756 Specifies the function to call to flush the I and D caches, or to not
9757 call any such function. If called, the function must take the same
9758 arguments as the common @code{_flush_func()}, that is, the address of the
9759 memory range for which the cache is being flushed, the size of the
9760 memory range, and the number 3 (to flush both caches). The default
9761 depends on the target GCC was configured for, but commonly is either
9762 @samp{_flush_func} or @samp{__cpu_flush}.
9764 @item -mbranch-likely
9765 @itemx -mno-branch-likely
9766 @opindex mbranch-likely
9767 @opindex mno-branch-likely
9768 Enable or disable use of Branch Likely instructions, regardless of the
9769 default for the selected architecture. By default, Branch Likely
9770 instructions may be generated if they are supported by the selected
9771 architecture. An exception is for the MIPS32 and MIPS64 architectures
9772 and processors which implement those architectures; for those, Branch
9773 Likely instructions will not be generated by default because the MIPS32
9774 and MIPS64 architectures specifically deprecate their use.
9776 @item -mfp-exceptions
9777 @itemx -mno-fp-exceptions
9778 @opindex mfp-exceptions
9779 Specifies whether FP exceptions are enabled. This affects how we schedule
9780 FP instructions for some processors. The default is that FP exceptions are
9783 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
9784 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
9787 @item -mvr4130-align
9788 @itemx -mno-vr4130-align
9789 @opindex mvr4130-align
9790 The VR4130 pipeline is two-way superscalar, but can only issue two
9791 instructions together if the first one is 8-byte aligned. When this
9792 option is enabled, GCC will align pairs of instructions that it
9793 thinks should execute in parallel.
9795 This option only has an effect when optimizing for the VR4130.
9796 It normally makes code faster, but at the expense of making it bigger.
9797 It is enabled by default at optimization level @option{-O3}.
9801 @subsection MMIX Options
9802 @cindex MMIX Options
9804 These options are defined for the MMIX:
9808 @itemx -mno-libfuncs
9810 @opindex mno-libfuncs
9811 Specify that intrinsic library functions are being compiled, passing all
9812 values in registers, no matter the size.
9817 @opindex mno-epsilon
9818 Generate floating-point comparison instructions that compare with respect
9819 to the @code{rE} epsilon register.
9821 @item -mabi=mmixware
9823 @opindex mabi-mmixware
9825 Generate code that passes function parameters and return values that (in
9826 the called function) are seen as registers @code{$0} and up, as opposed to
9827 the GNU ABI which uses global registers @code{$231} and up.
9830 @itemx -mno-zero-extend
9831 @opindex mzero-extend
9832 @opindex mno-zero-extend
9833 When reading data from memory in sizes shorter than 64 bits, use (do not
9834 use) zero-extending load instructions by default, rather than
9835 sign-extending ones.
9838 @itemx -mno-knuthdiv
9840 @opindex mno-knuthdiv
9841 Make the result of a division yielding a remainder have the same sign as
9842 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9843 remainder follows the sign of the dividend. Both methods are
9844 arithmetically valid, the latter being almost exclusively used.
9846 @item -mtoplevel-symbols
9847 @itemx -mno-toplevel-symbols
9848 @opindex mtoplevel-symbols
9849 @opindex mno-toplevel-symbols
9850 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9851 code can be used with the @code{PREFIX} assembly directive.
9855 Generate an executable in the ELF format, rather than the default
9856 @samp{mmo} format used by the @command{mmix} simulator.
9858 @item -mbranch-predict
9859 @itemx -mno-branch-predict
9860 @opindex mbranch-predict
9861 @opindex mno-branch-predict
9862 Use (do not use) the probable-branch instructions, when static branch
9863 prediction indicates a probable branch.
9865 @item -mbase-addresses
9866 @itemx -mno-base-addresses
9867 @opindex mbase-addresses
9868 @opindex mno-base-addresses
9869 Generate (do not generate) code that uses @emph{base addresses}. Using a
9870 base address automatically generates a request (handled by the assembler
9871 and the linker) for a constant to be set up in a global register. The
9872 register is used for one or more base address requests within the range 0
9873 to 255 from the value held in the register. The generally leads to short
9874 and fast code, but the number of different data items that can be
9875 addressed is limited. This means that a program that uses lots of static
9876 data may require @option{-mno-base-addresses}.
9879 @itemx -mno-single-exit
9880 @opindex msingle-exit
9881 @opindex mno-single-exit
9882 Force (do not force) generated code to have a single exit point in each
9886 @node MN10300 Options
9887 @subsection MN10300 Options
9888 @cindex MN10300 options
9890 These @option{-m} options are defined for Matsushita MN10300 architectures:
9895 Generate code to avoid bugs in the multiply instructions for the MN10300
9896 processors. This is the default.
9899 @opindex mno-mult-bug
9900 Do not generate code to avoid bugs in the multiply instructions for the
9905 Generate code which uses features specific to the AM33 processor.
9909 Do not generate code which uses features specific to the AM33 processor. This
9914 Do not link in the C run-time initialization object file.
9918 Indicate to the linker that it should perform a relaxation optimization pass
9919 to shorten branches, calls and absolute memory addresses. This option only
9920 has an effect when used on the command line for the final link step.
9922 This option makes symbolic debugging impossible.
9926 @subsection NS32K Options
9927 @cindex NS32K options
9929 These are the @samp{-m} options defined for the 32000 series. The default
9930 values for these options depends on which style of 32000 was selected when
9931 the compiler was configured; the defaults for the most common choices are
9939 Generate output for a 32032. This is the default
9940 when the compiler is configured for 32032 and 32016 based systems.
9946 Generate output for a 32332. This is the default
9947 when the compiler is configured for 32332-based systems.
9953 Generate output for a 32532. This is the default
9954 when the compiler is configured for 32532-based systems.
9958 Generate output containing 32081 instructions for floating point.
9959 This is the default for all systems.
9963 Generate output containing 32381 instructions for floating point. This
9964 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9965 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9969 Try and generate multiply-add floating point instructions @code{polyF}
9970 and @code{dotF}. This option is only available if the @option{-m32381}
9971 option is in effect. Using these instructions requires changes to
9972 register allocation which generally has a negative impact on
9973 performance. This option should only be enabled when compiling code
9974 particularly likely to make heavy use of multiply-add instructions.
9977 @opindex mnomulti-add
9978 Do not try and generate multiply-add floating point instructions
9979 @code{polyF} and @code{dotF}. This is the default on all platforms.
9982 @opindex msoft-float
9983 Generate output containing library calls for floating point.
9984 @strong{Warning:} the requisite libraries may not be available.
9986 @item -mieee-compare
9987 @itemx -mno-ieee-compare
9988 @opindex mieee-compare
9989 @opindex mno-ieee-compare
9990 Control whether or not the compiler uses IEEE floating point
9991 comparisons. These handle correctly the case where the result of a
9992 comparison is unordered.
9993 @strong{Warning:} the requisite kernel support may not be available.
9996 @opindex mnobitfield
9997 Do not use the bit-field instructions. On some machines it is faster to
9998 use shifting and masking operations. This is the default for the pc532.
10002 Do use the bit-field instructions. This is the default for all platforms
10007 Use a different function-calling convention, in which functions
10008 that take a fixed number of arguments return pop their
10009 arguments on return with the @code{ret} instruction.
10011 This calling convention is incompatible with the one normally
10012 used on Unix, so you cannot use it if you need to call libraries
10013 compiled with the Unix compiler.
10015 Also, you must provide function prototypes for all functions that
10016 take variable numbers of arguments (including @code{printf});
10017 otherwise incorrect code will be generated for calls to those
10020 In addition, seriously incorrect code will result if you call a
10021 function with too many arguments. (Normally, extra arguments are
10022 harmlessly ignored.)
10024 This option takes its name from the 680x0 @code{rtd} instruction.
10029 Use a different function-calling convention where the first two arguments
10030 are passed in registers.
10032 This calling convention is incompatible with the one normally
10033 used on Unix, so you cannot use it if you need to call libraries
10034 compiled with the Unix compiler.
10037 @opindex mnoregparam
10038 Do not pass any arguments in registers. This is the default for all
10043 It is OK to use the sb as an index register which is always loaded with
10044 zero. This is the default for the pc532-netbsd target.
10048 The sb register is not available for use or has not been initialized to
10049 zero by the run time system. This is the default for all targets except
10050 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
10051 @option{-fpic} is set.
10055 Many ns32000 series addressing modes use displacements of up to 512MB@.
10056 If an address is above 512MB then displacements from zero can not be used.
10057 This option causes code to be generated which can be loaded above 512MB@.
10058 This may be useful for operating systems or ROM code.
10062 Assume code will be loaded in the first 512MB of virtual address space.
10063 This is the default for all platforms.
10067 @node PDP-11 Options
10068 @subsection PDP-11 Options
10069 @cindex PDP-11 Options
10071 These options are defined for the PDP-11:
10076 Use hardware FPP floating point. This is the default. (FIS floating
10077 point on the PDP-11/40 is not supported.)
10080 @opindex msoft-float
10081 Do not use hardware floating point.
10085 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10089 Return floating-point results in memory. This is the default.
10093 Generate code for a PDP-11/40.
10097 Generate code for a PDP-11/45. This is the default.
10101 Generate code for a PDP-11/10.
10103 @item -mbcopy-builtin
10104 @opindex bcopy-builtin
10105 Use inline @code{movmemhi} patterns for copying memory. This is the
10110 Do not use inline @code{movmemhi} patterns for copying memory.
10116 Use 16-bit @code{int}. This is the default.
10122 Use 32-bit @code{int}.
10125 @itemx -mno-float32
10127 @opindex mno-float32
10128 Use 64-bit @code{float}. This is the default.
10131 @itemx -mno-float64
10133 @opindex mno-float64
10134 Use 32-bit @code{float}.
10138 Use @code{abshi2} pattern. This is the default.
10142 Do not use @code{abshi2} pattern.
10144 @item -mbranch-expensive
10145 @opindex mbranch-expensive
10146 Pretend that branches are expensive. This is for experimenting with
10147 code generation only.
10149 @item -mbranch-cheap
10150 @opindex mbranch-cheap
10151 Do not pretend that branches are expensive. This is the default.
10155 Generate code for a system with split I&D@.
10159 Generate code for a system without split I&D@. This is the default.
10163 Use Unix assembler syntax. This is the default when configured for
10164 @samp{pdp11-*-bsd}.
10168 Use DEC assembler syntax. This is the default when configured for any
10169 PDP-11 target other than @samp{pdp11-*-bsd}.
10172 @node PowerPC Options
10173 @subsection PowerPC Options
10174 @cindex PowerPC options
10176 These are listed under @xref{RS/6000 and PowerPC Options}.
10178 @node RS/6000 and PowerPC Options
10179 @subsection IBM RS/6000 and PowerPC Options
10180 @cindex RS/6000 and PowerPC Options
10181 @cindex IBM RS/6000 and PowerPC Options
10183 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
10190 @itemx -mno-powerpc
10191 @itemx -mpowerpc-gpopt
10192 @itemx -mno-powerpc-gpopt
10193 @itemx -mpowerpc-gfxopt
10194 @itemx -mno-powerpc-gfxopt
10196 @itemx -mno-powerpc64
10200 @opindex mno-power2
10202 @opindex mno-powerpc
10203 @opindex mpowerpc-gpopt
10204 @opindex mno-powerpc-gpopt
10205 @opindex mpowerpc-gfxopt
10206 @opindex mno-powerpc-gfxopt
10207 @opindex mpowerpc64
10208 @opindex mno-powerpc64
10209 GCC supports two related instruction set architectures for the
10210 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
10211 instructions supported by the @samp{rios} chip set used in the original
10212 RS/6000 systems and the @dfn{PowerPC} instruction set is the
10213 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
10214 the IBM 4xx microprocessors.
10216 Neither architecture is a subset of the other. However there is a
10217 large common subset of instructions supported by both. An MQ
10218 register is included in processors supporting the POWER architecture.
10220 You use these options to specify which instructions are available on the
10221 processor you are using. The default value of these options is
10222 determined when configuring GCC@. Specifying the
10223 @option{-mcpu=@var{cpu_type}} overrides the specification of these
10224 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
10225 rather than the options listed above.
10227 The @option{-mpower} option allows GCC to generate instructions that
10228 are found only in the POWER architecture and to use the MQ register.
10229 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
10230 to generate instructions that are present in the POWER2 architecture but
10231 not the original POWER architecture.
10233 The @option{-mpowerpc} option allows GCC to generate instructions that
10234 are found only in the 32-bit subset of the PowerPC architecture.
10235 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
10236 GCC to use the optional PowerPC architecture instructions in the
10237 General Purpose group, including floating-point square root. Specifying
10238 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
10239 use the optional PowerPC architecture instructions in the Graphics
10240 group, including floating-point select.
10242 The @option{-mpowerpc64} option allows GCC to generate the additional
10243 64-bit instructions that are found in the full PowerPC64 architecture
10244 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
10245 @option{-mno-powerpc64}.
10247 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
10248 will use only the instructions in the common subset of both
10249 architectures plus some special AIX common-mode calls, and will not use
10250 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
10251 permits GCC to use any instruction from either architecture and to
10252 allow use of the MQ register; specify this for the Motorola MPC601.
10254 @item -mnew-mnemonics
10255 @itemx -mold-mnemonics
10256 @opindex mnew-mnemonics
10257 @opindex mold-mnemonics
10258 Select which mnemonics to use in the generated assembler code. With
10259 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
10260 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
10261 assembler mnemonics defined for the POWER architecture. Instructions
10262 defined in only one architecture have only one mnemonic; GCC uses that
10263 mnemonic irrespective of which of these options is specified.
10265 GCC defaults to the mnemonics appropriate for the architecture in
10266 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
10267 value of these option. Unless you are building a cross-compiler, you
10268 should normally not specify either @option{-mnew-mnemonics} or
10269 @option{-mold-mnemonics}, but should instead accept the default.
10271 @item -mcpu=@var{cpu_type}
10273 Set architecture type, register usage, choice of mnemonics, and
10274 instruction scheduling parameters for machine type @var{cpu_type}.
10275 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
10276 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
10277 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
10278 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
10279 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
10280 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
10281 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
10282 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
10283 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
10285 @option{-mcpu=common} selects a completely generic processor. Code
10286 generated under this option will run on any POWER or PowerPC processor.
10287 GCC will use only the instructions in the common subset of both
10288 architectures, and will not use the MQ register. GCC assumes a generic
10289 processor model for scheduling purposes.
10291 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
10292 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
10293 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
10294 types, with an appropriate, generic processor model assumed for
10295 scheduling purposes.
10297 The other options specify a specific processor. Code generated under
10298 those options will run best on that processor, and may not run at all on
10301 The @option{-mcpu} options automatically enable or disable the
10302 following options: @option{-maltivec}, @option{-mhard-float},
10303 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
10304 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
10305 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
10306 @option{-mstring}. The particular options set for any particular CPU
10307 will vary between compiler versions, depending on what setting seems
10308 to produce optimal code for that CPU; it doesn't necessarily reflect
10309 the actual hardware's capabilities. If you wish to set an individual
10310 option to a particular value, you may specify it after the
10311 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
10313 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
10314 not enabled or disabled by the @option{-mcpu} option at present, since
10315 AIX does not have full support for these options. You may still
10316 enable or disable them individually if you're sure it'll work in your
10319 @item -mtune=@var{cpu_type}
10321 Set the instruction scheduling parameters for machine type
10322 @var{cpu_type}, but do not set the architecture type, register usage, or
10323 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
10324 values for @var{cpu_type} are used for @option{-mtune} as for
10325 @option{-mcpu}. If both are specified, the code generated will use the
10326 architecture, registers, and mnemonics set by @option{-mcpu}, but the
10327 scheduling parameters set by @option{-mtune}.
10330 @itemx -mno-altivec
10332 @opindex mno-altivec
10333 Generate code that uses (does not use) AltiVec instructions, and also
10334 enable the use of built-in functions that allow more direct access to
10335 the AltiVec instruction set. You may also need to set
10336 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
10341 Extend the current ABI with SPE ABI extensions. This does not change
10342 the default ABI, instead it adds the SPE ABI extensions to the current
10346 @opindex mabi=no-spe
10347 Disable Booke SPE ABI extensions for the current ABI@.
10349 @item -misel=@var{yes/no}
10352 This switch enables or disables the generation of ISEL instructions.
10354 @item -mspe=@var{yes/no}
10357 This switch enables or disables the generation of SPE simd
10360 @item -mfloat-gprs=@var{yes/single/double/no}
10361 @itemx -mfloat-gprs
10362 @opindex mfloat-gprs
10363 This switch enables or disables the generation of floating point
10364 operations on the general purpose registers for architectures that
10367 The argument @var{yes} or @var{single} enables the use of
10368 single-precision floating point operations.
10370 The argument @var{double} enables the use of single and
10371 double-precision floating point operations.
10373 The argument @var{no} disables floating point operations on the
10374 general purpose registers.
10376 This option is currently only available on the MPC854x.
10382 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
10383 targets (including GNU/Linux). The 32-bit environment sets int, long
10384 and pointer to 32 bits and generates code that runs on any PowerPC
10385 variant. The 64-bit environment sets int to 32 bits and long and
10386 pointer to 64 bits, and generates code for PowerPC64, as for
10387 @option{-mpowerpc64}.
10390 @itemx -mno-fp-in-toc
10391 @itemx -mno-sum-in-toc
10392 @itemx -mminimal-toc
10394 @opindex mno-fp-in-toc
10395 @opindex mno-sum-in-toc
10396 @opindex mminimal-toc
10397 Modify generation of the TOC (Table Of Contents), which is created for
10398 every executable file. The @option{-mfull-toc} option is selected by
10399 default. In that case, GCC will allocate at least one TOC entry for
10400 each unique non-automatic variable reference in your program. GCC
10401 will also place floating-point constants in the TOC@. However, only
10402 16,384 entries are available in the TOC@.
10404 If you receive a linker error message that saying you have overflowed
10405 the available TOC space, you can reduce the amount of TOC space used
10406 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
10407 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
10408 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
10409 generate code to calculate the sum of an address and a constant at
10410 run-time instead of putting that sum into the TOC@. You may specify one
10411 or both of these options. Each causes GCC to produce very slightly
10412 slower and larger code at the expense of conserving TOC space.
10414 If you still run out of space in the TOC even when you specify both of
10415 these options, specify @option{-mminimal-toc} instead. This option causes
10416 GCC to make only one TOC entry for every file. When you specify this
10417 option, GCC will produce code that is slower and larger but which
10418 uses extremely little TOC space. You may wish to use this option
10419 only on files that contain less frequently executed code.
10425 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
10426 @code{long} type, and the infrastructure needed to support them.
10427 Specifying @option{-maix64} implies @option{-mpowerpc64} and
10428 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
10429 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
10432 @itemx -mno-xl-call
10434 @opindex mno-xl-call
10435 On AIX, pass floating-point arguments to prototyped functions beyond the
10436 register save area (RSA) on the stack in addition to argument FPRs. The
10437 AIX calling convention was extended but not initially documented to
10438 handle an obscure K&R C case of calling a function that takes the
10439 address of its arguments with fewer arguments than declared. AIX XL
10440 compilers access floating point arguments which do not fit in the
10441 RSA from the stack when a subroutine is compiled without
10442 optimization. Because always storing floating-point arguments on the
10443 stack is inefficient and rarely needed, this option is not enabled by
10444 default and only is necessary when calling subroutines compiled by AIX
10445 XL compilers without optimization.
10449 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
10450 application written to use message passing with special startup code to
10451 enable the application to run. The system must have PE installed in the
10452 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
10453 must be overridden with the @option{-specs=} option to specify the
10454 appropriate directory location. The Parallel Environment does not
10455 support threads, so the @option{-mpe} option and the @option{-pthread}
10456 option are incompatible.
10458 @item -malign-natural
10459 @itemx -malign-power
10460 @opindex malign-natural
10461 @opindex malign-power
10462 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
10463 @option{-malign-natural} overrides the ABI-defined alignment of larger
10464 types, such as floating-point doubles, on their natural size-based boundary.
10465 The option @option{-malign-power} instructs GCC to follow the ABI-specified
10466 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
10469 @itemx -mhard-float
10470 @opindex msoft-float
10471 @opindex mhard-float
10472 Generate code that does not use (uses) the floating-point register set.
10473 Software floating point emulation is provided if you use the
10474 @option{-msoft-float} option, and pass the option to GCC when linking.
10477 @itemx -mno-multiple
10479 @opindex mno-multiple
10480 Generate code that uses (does not use) the load multiple word
10481 instructions and the store multiple word instructions. These
10482 instructions are generated by default on POWER systems, and not
10483 generated on PowerPC systems. Do not use @option{-mmultiple} on little
10484 endian PowerPC systems, since those instructions do not work when the
10485 processor is in little endian mode. The exceptions are PPC740 and
10486 PPC750 which permit the instructions usage in little endian mode.
10491 @opindex mno-string
10492 Generate code that uses (does not use) the load string instructions
10493 and the store string word instructions to save multiple registers and
10494 do small block moves. These instructions are generated by default on
10495 POWER systems, and not generated on PowerPC systems. Do not use
10496 @option{-mstring} on little endian PowerPC systems, since those
10497 instructions do not work when the processor is in little endian mode.
10498 The exceptions are PPC740 and PPC750 which permit the instructions
10499 usage in little endian mode.
10504 @opindex mno-update
10505 Generate code that uses (does not use) the load or store instructions
10506 that update the base register to the address of the calculated memory
10507 location. These instructions are generated by default. If you use
10508 @option{-mno-update}, there is a small window between the time that the
10509 stack pointer is updated and the address of the previous frame is
10510 stored, which means code that walks the stack frame across interrupts or
10511 signals may get corrupted data.
10514 @itemx -mno-fused-madd
10515 @opindex mfused-madd
10516 @opindex mno-fused-madd
10517 Generate code that uses (does not use) the floating point multiply and
10518 accumulate instructions. These instructions are generated by default if
10519 hardware floating is used.
10521 @item -mno-bit-align
10523 @opindex mno-bit-align
10524 @opindex mbit-align
10525 On System V.4 and embedded PowerPC systems do not (do) force structures
10526 and unions that contain bit-fields to be aligned to the base type of the
10529 For example, by default a structure containing nothing but 8
10530 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
10531 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
10532 the structure would be aligned to a 1 byte boundary and be one byte in
10535 @item -mno-strict-align
10536 @itemx -mstrict-align
10537 @opindex mno-strict-align
10538 @opindex mstrict-align
10539 On System V.4 and embedded PowerPC systems do not (do) assume that
10540 unaligned memory references will be handled by the system.
10542 @item -mrelocatable
10543 @itemx -mno-relocatable
10544 @opindex mrelocatable
10545 @opindex mno-relocatable
10546 On embedded PowerPC systems generate code that allows (does not allow)
10547 the program to be relocated to a different address at runtime. If you
10548 use @option{-mrelocatable} on any module, all objects linked together must
10549 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
10551 @item -mrelocatable-lib
10552 @itemx -mno-relocatable-lib
10553 @opindex mrelocatable-lib
10554 @opindex mno-relocatable-lib
10555 On embedded PowerPC systems generate code that allows (does not allow)
10556 the program to be relocated to a different address at runtime. Modules
10557 compiled with @option{-mrelocatable-lib} can be linked with either modules
10558 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
10559 with modules compiled with the @option{-mrelocatable} options.
10565 On System V.4 and embedded PowerPC systems do not (do) assume that
10566 register 2 contains a pointer to a global area pointing to the addresses
10567 used in the program.
10570 @itemx -mlittle-endian
10572 @opindex mlittle-endian
10573 On System V.4 and embedded PowerPC systems compile code for the
10574 processor in little endian mode. The @option{-mlittle-endian} option is
10575 the same as @option{-mlittle}.
10578 @itemx -mbig-endian
10580 @opindex mbig-endian
10581 On System V.4 and embedded PowerPC systems compile code for the
10582 processor in big endian mode. The @option{-mbig-endian} option is
10583 the same as @option{-mbig}.
10585 @item -mdynamic-no-pic
10586 @opindex mdynamic-no-pic
10587 On Darwin and Mac OS X systems, compile code so that it is not
10588 relocatable, but that its external references are relocatable. The
10589 resulting code is suitable for applications, but not shared
10592 @item -mprioritize-restricted-insns=@var{priority}
10593 @opindex mprioritize-restricted-insns
10594 This option controls the priority that is assigned to
10595 dispatch-slot restricted instructions during the second scheduling
10596 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
10597 @var{no/highest/second-highest} priority to dispatch slot restricted
10600 @item -msched-costly-dep=@var{dependence_type}
10601 @opindex msched-costly-dep
10602 This option controls which dependences are considered costly
10603 by the target during instruction scheduling. The argument
10604 @var{dependence_type} takes one of the following values:
10605 @var{no}: no dependence is costly,
10606 @var{all}: all dependences are costly,
10607 @var{true_store_to_load}: a true dependence from store to load is costly,
10608 @var{store_to_load}: any dependence from store to load is costly,
10609 @var{number}: any dependence which latency >= @var{number} is costly.
10611 @item -minsert-sched-nops=@var{scheme}
10612 @opindex minsert-sched-nops
10613 This option controls which nop insertion scheme will be used during
10614 the second scheduling pass. The argument @var{scheme} takes one of the
10616 @var{no}: Don't insert nops.
10617 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
10618 according to the scheduler's grouping.
10619 @var{regroup_exact}: Insert nops to force costly dependent insns into
10620 separate groups. Insert exactly as many nops as needed to force an insn
10621 to a new group, according to the estimated processor grouping.
10622 @var{number}: Insert nops to force costly dependent insns into
10623 separate groups. Insert @var{number} nops to force an insn to a new group.
10626 @opindex mcall-sysv
10627 On System V.4 and embedded PowerPC systems compile code using calling
10628 conventions that adheres to the March 1995 draft of the System V
10629 Application Binary Interface, PowerPC processor supplement. This is the
10630 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
10632 @item -mcall-sysv-eabi
10633 @opindex mcall-sysv-eabi
10634 Specify both @option{-mcall-sysv} and @option{-meabi} options.
10636 @item -mcall-sysv-noeabi
10637 @opindex mcall-sysv-noeabi
10638 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
10640 @item -mcall-solaris
10641 @opindex mcall-solaris
10642 On System V.4 and embedded PowerPC systems compile code for the Solaris
10646 @opindex mcall-linux
10647 On System V.4 and embedded PowerPC systems compile code for the
10648 Linux-based GNU system.
10652 On System V.4 and embedded PowerPC systems compile code for the
10653 Hurd-based GNU system.
10655 @item -mcall-netbsd
10656 @opindex mcall-netbsd
10657 On System V.4 and embedded PowerPC systems compile code for the
10658 NetBSD operating system.
10660 @item -maix-struct-return
10661 @opindex maix-struct-return
10662 Return all structures in memory (as specified by the AIX ABI)@.
10664 @item -msvr4-struct-return
10665 @opindex msvr4-struct-return
10666 Return structures smaller than 8 bytes in registers (as specified by the
10669 @item -mabi=altivec
10670 @opindex mabi=altivec
10671 Extend the current ABI with AltiVec ABI extensions. This does not
10672 change the default ABI, instead it adds the AltiVec ABI extensions to
10675 @item -mabi=no-altivec
10676 @opindex mabi=no-altivec
10677 Disable AltiVec ABI extensions for the current ABI@.
10680 @itemx -mno-prototype
10681 @opindex mprototype
10682 @opindex mno-prototype
10683 On System V.4 and embedded PowerPC systems assume that all calls to
10684 variable argument functions are properly prototyped. Otherwise, the
10685 compiler must insert an instruction before every non prototyped call to
10686 set or clear bit 6 of the condition code register (@var{CR}) to
10687 indicate whether floating point values were passed in the floating point
10688 registers in case the function takes a variable arguments. With
10689 @option{-mprototype}, only calls to prototyped variable argument functions
10690 will set or clear the bit.
10694 On embedded PowerPC systems, assume that the startup module is called
10695 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
10696 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
10701 On embedded PowerPC systems, assume that the startup module is called
10702 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
10707 On embedded PowerPC systems, assume that the startup module is called
10708 @file{crt0.o} and the standard C libraries are @file{libads.a} and
10711 @item -myellowknife
10712 @opindex myellowknife
10713 On embedded PowerPC systems, assume that the startup module is called
10714 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
10719 On System V.4 and embedded PowerPC systems, specify that you are
10720 compiling for a VxWorks system.
10724 Specify that you are compiling for the WindISS simulation environment.
10728 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
10729 header to indicate that @samp{eabi} extended relocations are used.
10735 On System V.4 and embedded PowerPC systems do (do not) adhere to the
10736 Embedded Applications Binary Interface (eabi) which is a set of
10737 modifications to the System V.4 specifications. Selecting @option{-meabi}
10738 means that the stack is aligned to an 8 byte boundary, a function
10739 @code{__eabi} is called to from @code{main} to set up the eabi
10740 environment, and the @option{-msdata} option can use both @code{r2} and
10741 @code{r13} to point to two separate small data areas. Selecting
10742 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
10743 do not call an initialization function from @code{main}, and the
10744 @option{-msdata} option will only use @code{r13} to point to a single
10745 small data area. The @option{-meabi} option is on by default if you
10746 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
10749 @opindex msdata=eabi
10750 On System V.4 and embedded PowerPC systems, put small initialized
10751 @code{const} global and static data in the @samp{.sdata2} section, which
10752 is pointed to by register @code{r2}. Put small initialized
10753 non-@code{const} global and static data in the @samp{.sdata} section,
10754 which is pointed to by register @code{r13}. Put small uninitialized
10755 global and static data in the @samp{.sbss} section, which is adjacent to
10756 the @samp{.sdata} section. The @option{-msdata=eabi} option is
10757 incompatible with the @option{-mrelocatable} option. The
10758 @option{-msdata=eabi} option also sets the @option{-memb} option.
10761 @opindex msdata=sysv
10762 On System V.4 and embedded PowerPC systems, put small global and static
10763 data in the @samp{.sdata} section, which is pointed to by register
10764 @code{r13}. Put small uninitialized global and static data in the
10765 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
10766 The @option{-msdata=sysv} option is incompatible with the
10767 @option{-mrelocatable} option.
10769 @item -msdata=default
10771 @opindex msdata=default
10773 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
10774 compile code the same as @option{-msdata=eabi}, otherwise compile code the
10775 same as @option{-msdata=sysv}.
10778 @opindex msdata-data
10779 On System V.4 and embedded PowerPC systems, put small global and static
10780 data in the @samp{.sdata} section. Put small uninitialized global and
10781 static data in the @samp{.sbss} section. Do not use register @code{r13}
10782 to address small data however. This is the default behavior unless
10783 other @option{-msdata} options are used.
10787 @opindex msdata=none
10789 On embedded PowerPC systems, put all initialized global and static data
10790 in the @samp{.data} section, and all uninitialized data in the
10791 @samp{.bss} section.
10795 @cindex smaller data references (PowerPC)
10796 @cindex .sdata/.sdata2 references (PowerPC)
10797 On embedded PowerPC systems, put global and static items less than or
10798 equal to @var{num} bytes into the small data or bss sections instead of
10799 the normal data or bss section. By default, @var{num} is 8. The
10800 @option{-G @var{num}} switch is also passed to the linker.
10801 All modules should be compiled with the same @option{-G @var{num}} value.
10804 @itemx -mno-regnames
10806 @opindex mno-regnames
10807 On System V.4 and embedded PowerPC systems do (do not) emit register
10808 names in the assembly language output using symbolic forms.
10811 @itemx -mno-longcall
10813 @opindex mno-longcall
10814 Default to making all function calls indirectly, using a register, so
10815 that functions which reside further than 32 megabytes (33,554,432
10816 bytes) from the current location can be called. This setting can be
10817 overridden by the @code{shortcall} function attribute, or by
10818 @code{#pragma longcall(0)}.
10820 Some linkers are capable of detecting out-of-range calls and generating
10821 glue code on the fly. On these systems, long calls are unnecessary and
10822 generate slower code. As of this writing, the AIX linker can do this,
10823 as can the GNU linker for PowerPC/64. It is planned to add this feature
10824 to the GNU linker for 32-bit PowerPC systems as well.
10826 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
10827 callee, L42'', plus a ``branch island'' (glue code). The two target
10828 addresses represent the callee and the ``branch island''. The
10829 Darwin/PPC linker will prefer the first address and generate a ``bl
10830 callee'' if the PPC ``bl'' instruction will reach the callee directly;
10831 otherwise, the linker will generate ``bl L42'' to call the ``branch
10832 island''. The ``branch island'' is appended to the body of the
10833 calling function; it computes the full 32-bit address of the callee
10836 On Mach-O (Darwin) systems, this option directs the compiler emit to
10837 the glue for every direct call, and the Darwin linker decides whether
10838 to use or discard it.
10840 In the future, we may cause GCC to ignore all longcall specifications
10841 when the linker is known to generate glue.
10845 Adds support for multithreading with the @dfn{pthreads} library.
10846 This option sets flags for both the preprocessor and linker.
10850 @node S/390 and zSeries Options
10851 @subsection S/390 and zSeries Options
10852 @cindex S/390 and zSeries Options
10854 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10858 @itemx -msoft-float
10859 @opindex mhard-float
10860 @opindex msoft-float
10861 Use (do not use) the hardware floating-point instructions and registers
10862 for floating-point operations. When @option{-msoft-float} is specified,
10863 functions in @file{libgcc.a} will be used to perform floating-point
10864 operations. When @option{-mhard-float} is specified, the compiler
10865 generates IEEE floating-point instructions. This is the default.
10868 @itemx -mno-backchain
10869 @opindex mbackchain
10870 @opindex mno-backchain
10871 Store (do not store) the address of the caller's frame as backchain pointer
10872 into the callee's stack frame.
10873 A backchain may be needed to allow debugging using tools that do not understand
10874 DWARF-2 call frame information.
10875 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
10876 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
10877 the backchain is placed into the topmost word of the 96/160 byte register
10880 In general, code compiled with @option{-mbackchain} is call-compatible with
10881 code compiled with @option{-mmo-backchain}; however, use of the backchain
10882 for debugging purposes usually requires that the whole binary is built with
10883 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
10884 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
10885 to build a linux kernel use @option{-msoft-float}.
10887 The default is to not maintain the backchain.
10889 @item -mpacked-stack
10890 @item -mno-packed-stack
10891 @opindex mpacked-stack
10892 @opindex mno-packed-stack
10893 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
10894 specified, the compiler uses the all fields of the 96/160 byte register save
10895 area only for their default purpose; unused fields still take up stack space.
10896 When @option{-mpacked-stack} is specified, register save slots are densely
10897 packed at the top of the register save area; unused space is reused for other
10898 purposes, allowing for more efficient use of the available stack space.
10899 However, when @option{-mbackchain} is also in effect, the topmost word of
10900 the save area is always used to store the backchain, and the return address
10901 register is always saved two words below the backchain.
10903 As long as the stack frame backchain is not used, code generated with
10904 @option{-mpacked-stack} is call-compatible with code generated with
10905 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
10906 S/390 or zSeries generated code that uses the stack frame backchain at run
10907 time, not just for debugging purposes. Such code is not call-compatible
10908 with code compiled with @option{-mpacked-stack}. Also, note that the
10909 combination of @option{-mbackchain},
10910 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
10911 to build a linux kernel use @option{-msoft-float}.
10913 The default is to not use the packed stack layout.
10916 @itemx -mno-small-exec
10917 @opindex msmall-exec
10918 @opindex mno-small-exec
10919 Generate (or do not generate) code using the @code{bras} instruction
10920 to do subroutine calls.
10921 This only works reliably if the total executable size does not
10922 exceed 64k. The default is to use the @code{basr} instruction instead,
10923 which does not have this limitation.
10929 When @option{-m31} is specified, generate code compliant to the
10930 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10931 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10932 particular to generate 64-bit instructions. For the @samp{s390}
10933 targets, the default is @option{-m31}, while the @samp{s390x}
10934 targets default to @option{-m64}.
10940 When @option{-mzarch} is specified, generate code using the
10941 instructions available on z/Architecture.
10942 When @option{-mesa} is specified, generate code using the
10943 instructions available on ESA/390. Note that @option{-mesa} is
10944 not possible with @option{-m64}.
10945 When generating code compliant to the GNU/Linux for S/390 ABI,
10946 the default is @option{-mesa}. When generating code compliant
10947 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10953 Generate (or do not generate) code using the @code{mvcle} instruction
10954 to perform block moves. When @option{-mno-mvcle} is specified,
10955 use a @code{mvc} loop instead. This is the default.
10961 Print (or do not print) additional debug information when compiling.
10962 The default is to not print debug information.
10964 @item -march=@var{cpu-type}
10966 Generate code that will run on @var{cpu-type}, which is the name of a system
10967 representing a certain processor type. Possible values for
10968 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10969 When generating code using the instructions available on z/Architecture,
10970 the default is @option{-march=z900}. Otherwise, the default is
10971 @option{-march=g5}.
10973 @item -mtune=@var{cpu-type}
10975 Tune to @var{cpu-type} everything applicable about the generated code,
10976 except for the ABI and the set of available instructions.
10977 The list of @var{cpu-type} values is the same as for @option{-march}.
10978 The default is the value used for @option{-march}.
10981 @itemx -mno-tpf-trace
10982 @opindex mtpf-trace
10983 @opindex mno-tpf-trace
10984 Generate code that adds (does not add) in TPF OS specific branches to trace
10985 routines in the operating system. This option is off by default, even
10986 when compiling for the TPF OS@.
10989 @itemx -mno-fused-madd
10990 @opindex mfused-madd
10991 @opindex mno-fused-madd
10992 Generate code that uses (does not use) the floating point multiply and
10993 accumulate instructions. These instructions are generated by default if
10994 hardware floating point is used.
10996 @item -mwarn-framesize=@var{framesize}
10997 @opindex mwarn-framesize
10998 Emit a warning if the current function exceeds the given frame size. Because
10999 this is a compile time check it doesn't need to be a real problem when the program
11000 runs. It is intended to identify functions which most probably cause
11001 a stack overflow. It is useful to be used in an environment with limited stack
11002 size e.g.@: the linux kernel.
11004 @item -mwarn-dynamicstack
11005 @opindex mwarn-dynamicstack
11006 Emit a warning if the function calls alloca or uses dynamically
11007 sized arrays. This is generally a bad idea with a limited stack size.
11009 @item -mstack-guard=@var{stack-guard}
11010 @item -mstack-size=@var{stack-size}
11011 @opindex mstack-guard
11012 @opindex mstack-size
11013 These arguments always have to be used in conjunction. If they are present the s390
11014 back end emits additional instructions in the function prologue which trigger a trap
11015 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
11016 (remember that the stack on s390 grows downward). These options are intended to
11017 be used to help debugging stack overflow problems. The additionally emitted code
11018 cause only little overhead and hence can also be used in production like systems
11019 without greater performance degradation. The given values have to be exact
11020 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard}.
11021 In order to be efficient the extra code makes the assumption that the stack starts
11022 at an address aligned to the value given by @var{stack-size}.
11026 @subsection SH Options
11028 These @samp{-m} options are defined for the SH implementations:
11033 Generate code for the SH1.
11037 Generate code for the SH2.
11040 Generate code for the SH2e.
11044 Generate code for the SH3.
11048 Generate code for the SH3e.
11052 Generate code for the SH4 without a floating-point unit.
11054 @item -m4-single-only
11055 @opindex m4-single-only
11056 Generate code for the SH4 with a floating-point unit that only
11057 supports single-precision arithmetic.
11061 Generate code for the SH4 assuming the floating-point unit is in
11062 single-precision mode by default.
11066 Generate code for the SH4.
11070 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
11071 floating-point unit is not used.
11073 @item -m4a-single-only
11074 @opindex m4a-single-only
11075 Generate code for the SH4a, in such a way that no double-precision
11076 floating point operations are used.
11079 @opindex m4a-single
11080 Generate code for the SH4a assuming the floating-point unit is in
11081 single-precision mode by default.
11085 Generate code for the SH4a.
11089 Same as @option{-m4a-nofpu}, except that it implicitly passes
11090 @option{-dsp} to the assembler. GCC doesn't generate any DSP
11091 instructions at the moment.
11095 Compile code for the processor in big endian mode.
11099 Compile code for the processor in little endian mode.
11103 Align doubles at 64-bit boundaries. Note that this changes the calling
11104 conventions, and thus some functions from the standard C library will
11105 not work unless you recompile it first with @option{-mdalign}.
11109 Shorten some address references at link time, when possible; uses the
11110 linker option @option{-relax}.
11114 Use 32-bit offsets in @code{switch} tables. The default is to use
11119 Enable the use of the instruction @code{fmovd}.
11123 Comply with the calling conventions defined by Renesas.
11127 Comply with the calling conventions defined by Renesas.
11131 Comply with the calling conventions defined for GCC before the Renesas
11132 conventions were available. This option is the default for all
11133 targets of the SH toolchain except for @samp{sh-symbianelf}.
11136 @opindex mnomacsave
11137 Mark the @code{MAC} register as call-clobbered, even if
11138 @option{-mhitachi} is given.
11142 Increase IEEE-compliance of floating-point code.
11146 Dump instruction size and location in the assembly code.
11149 @opindex mpadstruct
11150 This option is deprecated. It pads structures to multiple of 4 bytes,
11151 which is incompatible with the SH ABI@.
11155 Optimize for space instead of speed. Implied by @option{-Os}.
11158 @opindex mprefergot
11159 When generating position-independent code, emit function calls using
11160 the Global Offset Table instead of the Procedure Linkage Table.
11164 Generate a library function call to invalidate instruction cache
11165 entries, after fixing up a trampoline. This library function call
11166 doesn't assume it can write to the whole memory address space. This
11167 is the default when the target is @code{sh-*-linux*}.
11170 @node SPARC Options
11171 @subsection SPARC Options
11172 @cindex SPARC options
11174 These @samp{-m} options are supported on the SPARC:
11177 @item -mno-app-regs
11179 @opindex mno-app-regs
11181 Specify @option{-mapp-regs} to generate output using the global registers
11182 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
11185 To be fully SVR4 ABI compliant at the cost of some performance loss,
11186 specify @option{-mno-app-regs}. You should compile libraries and system
11187 software with this option.
11190 @itemx -mhard-float
11192 @opindex mhard-float
11193 Generate output containing floating point instructions. This is the
11197 @itemx -msoft-float
11199 @opindex msoft-float
11200 Generate output containing library calls for floating point.
11201 @strong{Warning:} the requisite libraries are not available for all SPARC
11202 targets. Normally the facilities of the machine's usual C compiler are
11203 used, but this cannot be done directly in cross-compilation. You must make
11204 your own arrangements to provide suitable library functions for
11205 cross-compilation. The embedded targets @samp{sparc-*-aout} and
11206 @samp{sparclite-*-*} do provide software floating point support.
11208 @option{-msoft-float} changes the calling convention in the output file;
11209 therefore, it is only useful if you compile @emph{all} of a program with
11210 this option. In particular, you need to compile @file{libgcc.a}, the
11211 library that comes with GCC, with @option{-msoft-float} in order for
11214 @item -mhard-quad-float
11215 @opindex mhard-quad-float
11216 Generate output containing quad-word (long double) floating point
11219 @item -msoft-quad-float
11220 @opindex msoft-quad-float
11221 Generate output containing library calls for quad-word (long double)
11222 floating point instructions. The functions called are those specified
11223 in the SPARC ABI@. This is the default.
11225 As of this writing, there are no SPARC implementations that have hardware
11226 support for the quad-word floating point instructions. They all invoke
11227 a trap handler for one of these instructions, and then the trap handler
11228 emulates the effect of the instruction. Because of the trap handler overhead,
11229 this is much slower than calling the ABI library routines. Thus the
11230 @option{-msoft-quad-float} option is the default.
11232 @item -mno-unaligned-doubles
11233 @itemx -munaligned-doubles
11234 @opindex mno-unaligned-doubles
11235 @opindex munaligned-doubles
11236 Assume that doubles have 8 byte alignment. This is the default.
11238 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
11239 alignment only if they are contained in another type, or if they have an
11240 absolute address. Otherwise, it assumes they have 4 byte alignment.
11241 Specifying this option avoids some rare compatibility problems with code
11242 generated by other compilers. It is not the default because it results
11243 in a performance loss, especially for floating point code.
11245 @item -mno-faster-structs
11246 @itemx -mfaster-structs
11247 @opindex mno-faster-structs
11248 @opindex mfaster-structs
11249 With @option{-mfaster-structs}, the compiler assumes that structures
11250 should have 8 byte alignment. This enables the use of pairs of
11251 @code{ldd} and @code{std} instructions for copies in structure
11252 assignment, in place of twice as many @code{ld} and @code{st} pairs.
11253 However, the use of this changed alignment directly violates the SPARC
11254 ABI@. Thus, it's intended only for use on targets where the developer
11255 acknowledges that their resulting code will not be directly in line with
11256 the rules of the ABI@.
11258 @item -mimpure-text
11259 @opindex mimpure-text
11260 @option{-mimpure-text}, used in addition to @option{-shared}, tells
11261 the compiler to not pass @option{-z text} to the linker when linking a
11262 shared object. Using this option, you can link position-dependent
11263 code into a shared object.
11265 @option{-mimpure-text} suppresses the ``relocations remain against
11266 allocatable but non-writable sections'' linker error message.
11267 However, the necessary relocations will trigger copy-on-write, and the
11268 shared object is not actually shared across processes. Instead of
11269 using @option{-mimpure-text}, you should compile all source code with
11270 @option{-fpic} or @option{-fPIC}.
11272 This option is only available on SunOS and Solaris.
11274 @item -mcpu=@var{cpu_type}
11276 Set the instruction set, register set, and instruction scheduling parameters
11277 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
11278 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
11279 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
11280 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
11281 @samp{ultrasparc3}.
11283 Default instruction scheduling parameters are used for values that select
11284 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
11285 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
11287 Here is a list of each supported architecture and their supported
11292 v8: supersparc, hypersparc
11293 sparclite: f930, f934, sparclite86x
11295 v9: ultrasparc, ultrasparc3
11298 By default (unless configured otherwise), GCC generates code for the V7
11299 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
11300 additionally optimizes it for the Cypress CY7C602 chip, as used in the
11301 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
11302 SPARCStation 1, 2, IPX etc.
11304 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
11305 architecture. The only difference from V7 code is that the compiler emits
11306 the integer multiply and integer divide instructions which exist in SPARC-V8
11307 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
11308 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
11311 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
11312 the SPARC architecture. This adds the integer multiply, integer divide step
11313 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
11314 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
11315 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
11316 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
11317 MB86934 chip, which is the more recent SPARClite with FPU@.
11319 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
11320 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
11321 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
11322 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
11323 optimizes it for the TEMIC SPARClet chip.
11325 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
11326 architecture. This adds 64-bit integer and floating-point move instructions,
11327 3 additional floating-point condition code registers and conditional move
11328 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
11329 optimizes it for the Sun UltraSPARC I/II chips. With
11330 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
11331 Sun UltraSPARC III chip.
11333 @item -mtune=@var{cpu_type}
11335 Set the instruction scheduling parameters for machine type
11336 @var{cpu_type}, but do not set the instruction set or register set that the
11337 option @option{-mcpu=@var{cpu_type}} would.
11339 The same values for @option{-mcpu=@var{cpu_type}} can be used for
11340 @option{-mtune=@var{cpu_type}}, but the only useful values are those
11341 that select a particular cpu implementation. Those are @samp{cypress},
11342 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
11343 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
11344 @samp{ultrasparc3}.
11349 @opindex mno-v8plus
11350 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
11351 difference from the V8 ABI is that the global and out registers are
11352 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
11353 mode for all SPARC-V9 processors.
11359 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
11360 Visual Instruction Set extensions. The default is @option{-mno-vis}.
11363 These @samp{-m} options are supported in addition to the above
11364 on SPARC-V9 processors in 64-bit environments:
11367 @item -mlittle-endian
11368 @opindex mlittle-endian
11369 Generate code for a processor running in little-endian mode. It is only
11370 available for a few configurations and most notably not on Solaris.
11376 Generate code for a 32-bit or 64-bit environment.
11377 The 32-bit environment sets int, long and pointer to 32 bits.
11378 The 64-bit environment sets int to 32 bits and long and pointer
11381 @item -mcmodel=medlow
11382 @opindex mcmodel=medlow
11383 Generate code for the Medium/Low code model: 64-bit addresses, programs
11384 must be linked in the low 32 bits of memory. Programs can be statically
11385 or dynamically linked.
11387 @item -mcmodel=medmid
11388 @opindex mcmodel=medmid
11389 Generate code for the Medium/Middle code model: 64-bit addresses, programs
11390 must be linked in the low 44 bits of memory, the text and data segments must
11391 be less than 2GB in size and the data segment must be located within 2GB of
11394 @item -mcmodel=medany
11395 @opindex mcmodel=medany
11396 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
11397 may be linked anywhere in memory, the text and data segments must be less
11398 than 2GB in size and the data segment must be located within 2GB of the
11401 @item -mcmodel=embmedany
11402 @opindex mcmodel=embmedany
11403 Generate code for the Medium/Anywhere code model for embedded systems:
11404 64-bit addresses, the text and data segments must be less than 2GB in
11405 size, both starting anywhere in memory (determined at link time). The
11406 global register %g4 points to the base of the data segment. Programs
11407 are statically linked and PIC is not supported.
11410 @itemx -mno-stack-bias
11411 @opindex mstack-bias
11412 @opindex mno-stack-bias
11413 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
11414 frame pointer if present, are offset by @minus{}2047 which must be added back
11415 when making stack frame references. This is the default in 64-bit mode.
11416 Otherwise, assume no such offset is present.
11419 These switches are supported in addition to the above on Solaris:
11424 Add support for multithreading using the Solaris threads library. This
11425 option sets flags for both the preprocessor and linker. This option does
11426 not affect the thread safety of object code produced by the compiler or
11427 that of libraries supplied with it.
11431 Add support for multithreading using the POSIX threads library. This
11432 option sets flags for both the preprocessor and linker. This option does
11433 not affect the thread safety of object code produced by the compiler or
11434 that of libraries supplied with it.
11437 @node System V Options
11438 @subsection Options for System V
11440 These additional options are available on System V Release 4 for
11441 compatibility with other compilers on those systems:
11446 Create a shared object.
11447 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
11451 Identify the versions of each tool used by the compiler, in a
11452 @code{.ident} assembler directive in the output.
11456 Refrain from adding @code{.ident} directives to the output file (this is
11459 @item -YP,@var{dirs}
11461 Search the directories @var{dirs}, and no others, for libraries
11462 specified with @option{-l}.
11464 @item -Ym,@var{dir}
11466 Look in the directory @var{dir} to find the M4 preprocessor.
11467 The assembler uses this option.
11468 @c This is supposed to go with a -Yd for predefined M4 macro files, but
11469 @c the generic assembler that comes with Solaris takes just -Ym.
11472 @node TMS320C3x/C4x Options
11473 @subsection TMS320C3x/C4x Options
11474 @cindex TMS320C3x/C4x Options
11476 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
11480 @item -mcpu=@var{cpu_type}
11482 Set the instruction set, register set, and instruction scheduling
11483 parameters for machine type @var{cpu_type}. Supported values for
11484 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
11485 @samp{c44}. The default is @samp{c40} to generate code for the
11490 @itemx -msmall-memory
11492 @opindex mbig-memory
11494 @opindex msmall-memory
11496 Generates code for the big or small memory model. The small memory
11497 model assumed that all data fits into one 64K word page. At run-time
11498 the data page (DP) register must be set to point to the 64K page
11499 containing the .bss and .data program sections. The big memory model is
11500 the default and requires reloading of the DP register for every direct
11507 Allow (disallow) allocation of general integer operands into the block
11508 count register BK@.
11514 Enable (disable) generation of code using decrement and branch,
11515 DBcond(D), instructions. This is enabled by default for the C4x. To be
11516 on the safe side, this is disabled for the C3x, since the maximum
11517 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
11518 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
11519 that it can utilize the decrement and branch instruction, but will give
11520 up if there is more than one memory reference in the loop. Thus a loop
11521 where the loop counter is decremented can generate slightly more
11522 efficient code, in cases where the RPTB instruction cannot be utilized.
11524 @item -mdp-isr-reload
11526 @opindex mdp-isr-reload
11528 Force the DP register to be saved on entry to an interrupt service
11529 routine (ISR), reloaded to point to the data section, and restored on
11530 exit from the ISR@. This should not be required unless someone has
11531 violated the small memory model by modifying the DP register, say within
11538 For the C3x use the 24-bit MPYI instruction for integer multiplies
11539 instead of a library call to guarantee 32-bit results. Note that if one
11540 of the operands is a constant, then the multiplication will be performed
11541 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
11542 then squaring operations are performed inline instead of a library call.
11545 @itemx -mno-fast-fix
11547 @opindex mno-fast-fix
11548 The C3x/C4x FIX instruction to convert a floating point value to an
11549 integer value chooses the nearest integer less than or equal to the
11550 floating point value rather than to the nearest integer. Thus if the
11551 floating point number is negative, the result will be incorrectly
11552 truncated an additional code is necessary to detect and correct this
11553 case. This option can be used to disable generation of the additional
11554 code required to correct the result.
11560 Enable (disable) generation of repeat block sequences using the RPTB
11561 instruction for zero overhead looping. The RPTB construct is only used
11562 for innermost loops that do not call functions or jump across the loop
11563 boundaries. There is no advantage having nested RPTB loops due to the
11564 overhead required to save and restore the RC, RS, and RE registers.
11565 This is enabled by default with @option{-O2}.
11567 @item -mrpts=@var{count}
11571 Enable (disable) the use of the single instruction repeat instruction
11572 RPTS@. If a repeat block contains a single instruction, and the loop
11573 count can be guaranteed to be less than the value @var{count}, GCC will
11574 emit a RPTS instruction instead of a RPTB@. If no value is specified,
11575 then a RPTS will be emitted even if the loop count cannot be determined
11576 at compile time. Note that the repeated instruction following RPTS does
11577 not have to be reloaded from memory each iteration, thus freeing up the
11578 CPU buses for operands. However, since interrupts are blocked by this
11579 instruction, it is disabled by default.
11581 @item -mloop-unsigned
11582 @itemx -mno-loop-unsigned
11583 @opindex mloop-unsigned
11584 @opindex mno-loop-unsigned
11585 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
11586 is @math{2^{31} + 1} since these instructions test if the iteration count is
11587 negative to terminate the loop. If the iteration count is unsigned
11588 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
11589 exceeded. This switch allows an unsigned iteration count.
11593 Try to emit an assembler syntax that the TI assembler (asm30) is happy
11594 with. This also enforces compatibility with the API employed by the TI
11595 C3x C compiler. For example, long doubles are passed as structures
11596 rather than in floating point registers.
11602 Generate code that uses registers (stack) for passing arguments to functions.
11603 By default, arguments are passed in registers where possible rather
11604 than by pushing arguments on to the stack.
11606 @item -mparallel-insns
11607 @itemx -mno-parallel-insns
11608 @opindex mparallel-insns
11609 @opindex mno-parallel-insns
11610 Allow the generation of parallel instructions. This is enabled by
11611 default with @option{-O2}.
11613 @item -mparallel-mpy
11614 @itemx -mno-parallel-mpy
11615 @opindex mparallel-mpy
11616 @opindex mno-parallel-mpy
11617 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
11618 provided @option{-mparallel-insns} is also specified. These instructions have
11619 tight register constraints which can pessimize the code generation
11620 of large functions.
11625 @subsection V850 Options
11626 @cindex V850 Options
11628 These @samp{-m} options are defined for V850 implementations:
11632 @itemx -mno-long-calls
11633 @opindex mlong-calls
11634 @opindex mno-long-calls
11635 Treat all calls as being far away (near). If calls are assumed to be
11636 far away, the compiler will always load the functions address up into a
11637 register, and call indirect through the pointer.
11643 Do not optimize (do optimize) basic blocks that use the same index
11644 pointer 4 or more times to copy pointer into the @code{ep} register, and
11645 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
11646 option is on by default if you optimize.
11648 @item -mno-prolog-function
11649 @itemx -mprolog-function
11650 @opindex mno-prolog-function
11651 @opindex mprolog-function
11652 Do not use (do use) external functions to save and restore registers
11653 at the prologue and epilogue of a function. The external functions
11654 are slower, but use less code space if more than one function saves
11655 the same number of registers. The @option{-mprolog-function} option
11656 is on by default if you optimize.
11660 Try to make the code as small as possible. At present, this just turns
11661 on the @option{-mep} and @option{-mprolog-function} options.
11663 @item -mtda=@var{n}
11665 Put static or global variables whose size is @var{n} bytes or less into
11666 the tiny data area that register @code{ep} points to. The tiny data
11667 area can hold up to 256 bytes in total (128 bytes for byte references).
11669 @item -msda=@var{n}
11671 Put static or global variables whose size is @var{n} bytes or less into
11672 the small data area that register @code{gp} points to. The small data
11673 area can hold up to 64 kilobytes.
11675 @item -mzda=@var{n}
11677 Put static or global variables whose size is @var{n} bytes or less into
11678 the first 32 kilobytes of memory.
11682 Specify that the target processor is the V850.
11685 @opindex mbig-switch
11686 Generate code suitable for big switch tables. Use this option only if
11687 the assembler/linker complain about out of range branches within a switch
11692 This option will cause r2 and r5 to be used in the code generated by
11693 the compiler. This setting is the default.
11695 @item -mno-app-regs
11696 @opindex mno-app-regs
11697 This option will cause r2 and r5 to be treated as fixed registers.
11701 Specify that the target processor is the V850E1. The preprocessor
11702 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
11703 this option is used.
11707 Specify that the target processor is the V850E@. The preprocessor
11708 constant @samp{__v850e__} will be defined if this option is used.
11710 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
11711 are defined then a default target processor will be chosen and the
11712 relevant @samp{__v850*__} preprocessor constant will be defined.
11714 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
11715 defined, regardless of which processor variant is the target.
11717 @item -mdisable-callt
11718 @opindex mdisable-callt
11719 This option will suppress generation of the CALLT instruction for the
11720 v850e and v850e1 flavors of the v850 architecture. The default is
11721 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
11726 @subsection VAX Options
11727 @cindex VAX options
11729 These @samp{-m} options are defined for the VAX:
11734 Do not output certain jump instructions (@code{aobleq} and so on)
11735 that the Unix assembler for the VAX cannot handle across long
11740 Do output those jump instructions, on the assumption that you
11741 will assemble with the GNU assembler.
11745 Output code for g-format floating point numbers instead of d-format.
11748 @node x86-64 Options
11749 @subsection x86-64 Options
11750 @cindex x86-64 options
11752 These are listed under @xref{i386 and x86-64 Options}.
11754 @node Xstormy16 Options
11755 @subsection Xstormy16 Options
11756 @cindex Xstormy16 Options
11758 These options are defined for Xstormy16:
11763 Choose startup files and linker script suitable for the simulator.
11766 @node Xtensa Options
11767 @subsection Xtensa Options
11768 @cindex Xtensa Options
11770 These options are supported for Xtensa targets:
11774 @itemx -mno-const16
11776 @opindex mno-const16
11777 Enable or disable use of @code{CONST16} instructions for loading
11778 constant values. The @code{CONST16} instruction is currently not a
11779 standard option from Tensilica. When enabled, @code{CONST16}
11780 instructions are always used in place of the standard @code{L32R}
11781 instructions. The use of @code{CONST16} is enabled by default only if
11782 the @code{L32R} instruction is not available.
11785 @itemx -mno-fused-madd
11786 @opindex mfused-madd
11787 @opindex mno-fused-madd
11788 Enable or disable use of fused multiply/add and multiply/subtract
11789 instructions in the floating-point option. This has no effect if the
11790 floating-point option is not also enabled. Disabling fused multiply/add
11791 and multiply/subtract instructions forces the compiler to use separate
11792 instructions for the multiply and add/subtract operations. This may be
11793 desirable in some cases where strict IEEE 754-compliant results are
11794 required: the fused multiply add/subtract instructions do not round the
11795 intermediate result, thereby producing results with @emph{more} bits of
11796 precision than specified by the IEEE standard. Disabling fused multiply
11797 add/subtract instructions also ensures that the program output is not
11798 sensitive to the compiler's ability to combine multiply and add/subtract
11801 @item -mtext-section-literals
11802 @itemx -mno-text-section-literals
11803 @opindex mtext-section-literals
11804 @opindex mno-text-section-literals
11805 Control the treatment of literal pools. The default is
11806 @option{-mno-text-section-literals}, which places literals in a separate
11807 section in the output file. This allows the literal pool to be placed
11808 in a data RAM/ROM, and it also allows the linker to combine literal
11809 pools from separate object files to remove redundant literals and
11810 improve code size. With @option{-mtext-section-literals}, the literals
11811 are interspersed in the text section in order to keep them as close as
11812 possible to their references. This may be necessary for large assembly
11815 @item -mtarget-align
11816 @itemx -mno-target-align
11817 @opindex mtarget-align
11818 @opindex mno-target-align
11819 When this option is enabled, GCC instructs the assembler to
11820 automatically align instructions to reduce branch penalties at the
11821 expense of some code density. The assembler attempts to widen density
11822 instructions to align branch targets and the instructions following call
11823 instructions. If there are not enough preceding safe density
11824 instructions to align a target, no widening will be performed. The
11825 default is @option{-mtarget-align}. These options do not affect the
11826 treatment of auto-aligned instructions like @code{LOOP}, which the
11827 assembler will always align, either by widening density instructions or
11828 by inserting no-op instructions.
11831 @itemx -mno-longcalls
11832 @opindex mlongcalls
11833 @opindex mno-longcalls
11834 When this option is enabled, GCC instructs the assembler to translate
11835 direct calls to indirect calls unless it can determine that the target
11836 of a direct call is in the range allowed by the call instruction. This
11837 translation typically occurs for calls to functions in other source
11838 files. Specifically, the assembler translates a direct @code{CALL}
11839 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11840 The default is @option{-mno-longcalls}. This option should be used in
11841 programs where the call target can potentially be out of range. This
11842 option is implemented in the assembler, not the compiler, so the
11843 assembly code generated by GCC will still show direct call
11844 instructions---look at the disassembled object code to see the actual
11845 instructions. Note that the assembler will use an indirect call for
11846 every cross-file call, not just those that really will be out of range.
11849 @node zSeries Options
11850 @subsection zSeries Options
11851 @cindex zSeries options
11853 These are listed under @xref{S/390 and zSeries Options}.
11855 @node Code Gen Options
11856 @section Options for Code Generation Conventions
11857 @cindex code generation conventions
11858 @cindex options, code generation
11859 @cindex run-time options
11861 These machine-independent options control the interface conventions
11862 used in code generation.
11864 Most of them have both positive and negative forms; the negative form
11865 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11866 one of the forms is listed---the one which is not the default. You
11867 can figure out the other form by either removing @samp{no-} or adding
11871 @item -fbounds-check
11872 @opindex fbounds-check
11873 For front-ends that support it, generate additional code to check that
11874 indices used to access arrays are within the declared range. This is
11875 currently only supported by the Java and Fortran 77 front-ends, where
11876 this option defaults to true and false respectively.
11880 This option generates traps for signed overflow on addition, subtraction,
11881 multiplication operations.
11885 This option instructs the compiler to assume that signed arithmetic
11886 overflow of addition, subtraction and multiplication wraps around
11887 using twos-complement representation. This flag enables some optimizations
11888 and disables other. This option is enabled by default for the Java
11889 front-end, as required by the Java language specification.
11892 @opindex fexceptions
11893 Enable exception handling. Generates extra code needed to propagate
11894 exceptions. For some targets, this implies GCC will generate frame
11895 unwind information for all functions, which can produce significant data
11896 size overhead, although it does not affect execution. If you do not
11897 specify this option, GCC will enable it by default for languages like
11898 C++ which normally require exception handling, and disable it for
11899 languages like C that do not normally require it. However, you may need
11900 to enable this option when compiling C code that needs to interoperate
11901 properly with exception handlers written in C++. You may also wish to
11902 disable this option if you are compiling older C++ programs that don't
11903 use exception handling.
11905 @item -fnon-call-exceptions
11906 @opindex fnon-call-exceptions
11907 Generate code that allows trapping instructions to throw exceptions.
11908 Note that this requires platform-specific runtime support that does
11909 not exist everywhere. Moreover, it only allows @emph{trapping}
11910 instructions to throw exceptions, i.e.@: memory references or floating
11911 point instructions. It does not allow exceptions to be thrown from
11912 arbitrary signal handlers such as @code{SIGALRM}.
11914 @item -funwind-tables
11915 @opindex funwind-tables
11916 Similar to @option{-fexceptions}, except that it will just generate any needed
11917 static data, but will not affect the generated code in any other way.
11918 You will normally not enable this option; instead, a language processor
11919 that needs this handling would enable it on your behalf.
11921 @item -fasynchronous-unwind-tables
11922 @opindex fasynchronous-unwind-tables
11923 Generate unwind table in dwarf2 format, if supported by target machine. The
11924 table is exact at each instruction boundary, so it can be used for stack
11925 unwinding from asynchronous events (such as debugger or garbage collector).
11927 @item -fpcc-struct-return
11928 @opindex fpcc-struct-return
11929 Return ``short'' @code{struct} and @code{union} values in memory like
11930 longer ones, rather than in registers. This convention is less
11931 efficient, but it has the advantage of allowing intercallability between
11932 GCC-compiled files and files compiled with other compilers, particularly
11933 the Portable C Compiler (pcc).
11935 The precise convention for returning structures in memory depends
11936 on the target configuration macros.
11938 Short structures and unions are those whose size and alignment match
11939 that of some integer type.
11941 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11942 switch is not binary compatible with code compiled with the
11943 @option{-freg-struct-return} switch.
11944 Use it to conform to a non-default application binary interface.
11946 @item -freg-struct-return
11947 @opindex freg-struct-return
11948 Return @code{struct} and @code{union} values in registers when possible.
11949 This is more efficient for small structures than
11950 @option{-fpcc-struct-return}.
11952 If you specify neither @option{-fpcc-struct-return} nor
11953 @option{-freg-struct-return}, GCC defaults to whichever convention is
11954 standard for the target. If there is no standard convention, GCC
11955 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11956 the principal compiler. In those cases, we can choose the standard, and
11957 we chose the more efficient register return alternative.
11959 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11960 switch is not binary compatible with code compiled with the
11961 @option{-fpcc-struct-return} switch.
11962 Use it to conform to a non-default application binary interface.
11964 @item -fshort-enums
11965 @opindex fshort-enums
11966 Allocate to an @code{enum} type only as many bytes as it needs for the
11967 declared range of possible values. Specifically, the @code{enum} type
11968 will be equivalent to the smallest integer type which has enough room.
11970 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11971 code that is not binary compatible with code generated without that switch.
11972 Use it to conform to a non-default application binary interface.
11974 @item -fshort-double
11975 @opindex fshort-double
11976 Use the same size for @code{double} as for @code{float}.
11978 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11979 code that is not binary compatible with code generated without that switch.
11980 Use it to conform to a non-default application binary interface.
11982 @item -fshort-wchar
11983 @opindex fshort-wchar
11984 Override the underlying type for @samp{wchar_t} to be @samp{short
11985 unsigned int} instead of the default for the target. This option is
11986 useful for building programs to run under WINE@.
11988 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11989 code that is not binary compatible with code generated without that switch.
11990 Use it to conform to a non-default application binary interface.
11992 @item -fshared-data
11993 @opindex fshared-data
11994 Requests that the data and non-@code{const} variables of this
11995 compilation be shared data rather than private data. The distinction
11996 makes sense only on certain operating systems, where shared data is
11997 shared between processes running the same program, while private data
11998 exists in one copy per process.
12001 @opindex fno-common
12002 In C, allocate even uninitialized global variables in the data section of the
12003 object file, rather than generating them as common blocks. This has the
12004 effect that if the same variable is declared (without @code{extern}) in
12005 two different compilations, you will get an error when you link them.
12006 The only reason this might be useful is if you wish to verify that the
12007 program will work on other systems which always work this way.
12011 Ignore the @samp{#ident} directive.
12013 @item -finhibit-size-directive
12014 @opindex finhibit-size-directive
12015 Don't output a @code{.size} assembler directive, or anything else that
12016 would cause trouble if the function is split in the middle, and the
12017 two halves are placed at locations far apart in memory. This option is
12018 used when compiling @file{crtstuff.c}; you should not need to use it
12021 @item -fverbose-asm
12022 @opindex fverbose-asm
12023 Put extra commentary information in the generated assembly code to
12024 make it more readable. This option is generally only of use to those
12025 who actually need to read the generated assembly code (perhaps while
12026 debugging the compiler itself).
12028 @option{-fno-verbose-asm}, the default, causes the
12029 extra information to be omitted and is useful when comparing two assembler
12034 @cindex global offset table
12036 Generate position-independent code (PIC) suitable for use in a shared
12037 library, if supported for the target machine. Such code accesses all
12038 constant addresses through a global offset table (GOT)@. The dynamic
12039 loader resolves the GOT entries when the program starts (the dynamic
12040 loader is not part of GCC; it is part of the operating system). If
12041 the GOT size for the linked executable exceeds a machine-specific
12042 maximum size, you get an error message from the linker indicating that
12043 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
12044 instead. (These maximums are 8k on the SPARC and 32k
12045 on the m68k and RS/6000. The 386 has no such limit.)
12047 Position-independent code requires special support, and therefore works
12048 only on certain machines. For the 386, GCC supports PIC for System V
12049 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
12050 position-independent.
12054 If supported for the target machine, emit position-independent code,
12055 suitable for dynamic linking and avoiding any limit on the size of the
12056 global offset table. This option makes a difference on the m68k,
12057 PowerPC and SPARC@.
12059 Position-independent code requires special support, and therefore works
12060 only on certain machines.
12066 These options are similar to @option{-fpic} and @option{-fPIC}, but
12067 generated position independent code can be only linked into executables.
12068 Usually these options are used when @option{-pie} GCC option will be
12069 used during linking.
12071 @item -ffixed-@var{reg}
12073 Treat the register named @var{reg} as a fixed register; generated code
12074 should never refer to it (except perhaps as a stack pointer, frame
12075 pointer or in some other fixed role).
12077 @var{reg} must be the name of a register. The register names accepted
12078 are machine-specific and are defined in the @code{REGISTER_NAMES}
12079 macro in the machine description macro file.
12081 This flag does not have a negative form, because it specifies a
12084 @item -fcall-used-@var{reg}
12085 @opindex fcall-used
12086 Treat the register named @var{reg} as an allocable register that is
12087 clobbered by function calls. It may be allocated for temporaries or
12088 variables that do not live across a call. Functions compiled this way
12089 will not save and restore the register @var{reg}.
12091 It is an error to used this flag with the frame pointer or stack pointer.
12092 Use of this flag for other registers that have fixed pervasive roles in
12093 the machine's execution model will produce disastrous results.
12095 This flag does not have a negative form, because it specifies a
12098 @item -fcall-saved-@var{reg}
12099 @opindex fcall-saved
12100 Treat the register named @var{reg} as an allocable register saved by
12101 functions. It may be allocated even for temporaries or variables that
12102 live across a call. Functions compiled this way will save and restore
12103 the register @var{reg} if they use it.
12105 It is an error to used this flag with the frame pointer or stack pointer.
12106 Use of this flag for other registers that have fixed pervasive roles in
12107 the machine's execution model will produce disastrous results.
12109 A different sort of disaster will result from the use of this flag for
12110 a register in which function values may be returned.
12112 This flag does not have a negative form, because it specifies a
12115 @item -fpack-struct[=@var{n}]
12116 @opindex fpack-struct
12117 Without a value specified, pack all structure members together without
12118 holes. When a value is specified (which must be a small power of two), pack
12119 structure members according to this value, representing the maximum
12120 alignment (that is, objects with default alignment requirements larger than
12121 this will be output potentially unaligned at the next fitting location.
12123 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
12124 code that is not binary compatible with code generated without that switch.
12125 Additionally, it makes the code suboptimal.
12126 Use it to conform to a non-default application binary interface.
12128 @item -finstrument-functions
12129 @opindex finstrument-functions
12130 Generate instrumentation calls for entry and exit to functions. Just
12131 after function entry and just before function exit, the following
12132 profiling functions will be called with the address of the current
12133 function and its call site. (On some platforms,
12134 @code{__builtin_return_address} does not work beyond the current
12135 function, so the call site information may not be available to the
12136 profiling functions otherwise.)
12139 void __cyg_profile_func_enter (void *this_fn,
12141 void __cyg_profile_func_exit (void *this_fn,
12145 The first argument is the address of the start of the current function,
12146 which may be looked up exactly in the symbol table.
12148 This instrumentation is also done for functions expanded inline in other
12149 functions. The profiling calls will indicate where, conceptually, the
12150 inline function is entered and exited. This means that addressable
12151 versions of such functions must be available. If all your uses of a
12152 function are expanded inline, this may mean an additional expansion of
12153 code size. If you use @samp{extern inline} in your C code, an
12154 addressable version of such functions must be provided. (This is
12155 normally the case anyways, but if you get lucky and the optimizer always
12156 expands the functions inline, you might have gotten away without
12157 providing static copies.)
12159 A function may be given the attribute @code{no_instrument_function}, in
12160 which case this instrumentation will not be done. This can be used, for
12161 example, for the profiling functions listed above, high-priority
12162 interrupt routines, and any functions from which the profiling functions
12163 cannot safely be called (perhaps signal handlers, if the profiling
12164 routines generate output or allocate memory).
12166 @item -fstack-check
12167 @opindex fstack-check
12168 Generate code to verify that you do not go beyond the boundary of the
12169 stack. You should specify this flag if you are running in an
12170 environment with multiple threads, but only rarely need to specify it in
12171 a single-threaded environment since stack overflow is automatically
12172 detected on nearly all systems if there is only one stack.
12174 Note that this switch does not actually cause checking to be done; the
12175 operating system must do that. The switch causes generation of code
12176 to ensure that the operating system sees the stack being extended.
12178 @item -fstack-limit-register=@var{reg}
12179 @itemx -fstack-limit-symbol=@var{sym}
12180 @itemx -fno-stack-limit
12181 @opindex fstack-limit-register
12182 @opindex fstack-limit-symbol
12183 @opindex fno-stack-limit
12184 Generate code to ensure that the stack does not grow beyond a certain value,
12185 either the value of a register or the address of a symbol. If the stack
12186 would grow beyond the value, a signal is raised. For most targets,
12187 the signal is raised before the stack overruns the boundary, so
12188 it is possible to catch the signal without taking special precautions.
12190 For instance, if the stack starts at absolute address @samp{0x80000000}
12191 and grows downwards, you can use the flags
12192 @option{-fstack-limit-symbol=__stack_limit} and
12193 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
12194 of 128KB@. Note that this may only work with the GNU linker.
12196 @cindex aliasing of parameters
12197 @cindex parameters, aliased
12198 @item -fargument-alias
12199 @itemx -fargument-noalias
12200 @itemx -fargument-noalias-global
12201 @opindex fargument-alias
12202 @opindex fargument-noalias
12203 @opindex fargument-noalias-global
12204 Specify the possible relationships among parameters and between
12205 parameters and global data.
12207 @option{-fargument-alias} specifies that arguments (parameters) may
12208 alias each other and may alias global storage.@*
12209 @option{-fargument-noalias} specifies that arguments do not alias
12210 each other, but may alias global storage.@*
12211 @option{-fargument-noalias-global} specifies that arguments do not
12212 alias each other and do not alias global storage.
12214 Each language will automatically use whatever option is required by
12215 the language standard. You should not need to use these options yourself.
12217 @item -fleading-underscore
12218 @opindex fleading-underscore
12219 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
12220 change the way C symbols are represented in the object file. One use
12221 is to help link with legacy assembly code.
12223 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
12224 generate code that is not binary compatible with code generated without that
12225 switch. Use it to conform to a non-default application binary interface.
12226 Not all targets provide complete support for this switch.
12228 @item -ftls-model=@var{model}
12229 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
12230 The @var{model} argument should be one of @code{global-dynamic},
12231 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
12233 The default without @option{-fpic} is @code{initial-exec}; with
12234 @option{-fpic} the default is @code{global-dynamic}.
12236 @item -fvisibility=@var{default|internal|hidden|protected}
12237 @opindex fvisibility
12238 Set the default ELF image symbol visibility to the specified option---all
12239 symbols will be marked with this unless overridden within the code.
12240 Using this feature can very substantially improve linking and
12241 load times of shared object libraries, produce more optimized
12242 code, provide near-perfect API export and prevent symbol clashes.
12243 It is @strong{strongly} recommended that you use this in any shared objects
12246 Despite the nomenclature, @code{default} always means public ie;
12247 available to be linked against from outside the shared object.
12248 @code{protected} and @code{internal} are pretty useless in real-world
12249 usage so the only other commonly used option will be @code{hidden}.
12250 The default if @option{-fvisibility} isn't specified is
12251 @code{default}, i.e., make every
12252 symbol public---this causes the same behavior as previous versions of
12255 A good explanation of the benefits offered by ensuring ELF
12256 symbols have the correct visibility is given by ``How To Write
12257 Shared Libraries'' by Ulrich Drepper (which can be found at
12258 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
12259 solution made possible by this option to marking things hidden when
12260 the default is public is to make the default hidden and mark things
12261 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
12262 and @code{__attribute__ ((visibility("default")))} instead of
12263 @code{__declspec(dllexport)} you get almost identical semantics with
12264 identical syntax. This is a great boon to those working with
12265 cross-platform projects.
12267 For those adding visibility support to existing code, you may find
12268 @samp{#pragma GCC visibility} of use. This works by you enclosing
12269 the declarations you wish to set visibility for with (for example)
12270 @samp{#pragma GCC visibility push(hidden)} and
12271 @samp{#pragma GCC visibility pop}. These can be nested up to sixteen
12272 times. Bear in mind that symbol visibility should be viewed @strong{as
12273 part of the API interface contract} and thus all new code should
12274 always specify visibility when it is not the default ie; declarations
12275 only for use within the local DSO should @strong{always} be marked explicitly
12276 as hidden as so to avoid PLT indirection overheads---making this
12277 abundantly clear also aids readability and self-documentation of the code.
12278 Note that due to ISO C++ specification requirements, operator new and
12279 operator delete must always be of default visibility.
12281 An overview of these techniques, their benefits and how to use them
12282 is at @w{@uref{http://www.nedprod.com/programs/gccvisibility.html}}.
12288 @node Environment Variables
12289 @section Environment Variables Affecting GCC
12290 @cindex environment variables
12292 @c man begin ENVIRONMENT
12293 This section describes several environment variables that affect how GCC
12294 operates. Some of them work by specifying directories or prefixes to use
12295 when searching for various kinds of files. Some are used to specify other
12296 aspects of the compilation environment.
12298 Note that you can also specify places to search using options such as
12299 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
12300 take precedence over places specified using environment variables, which
12301 in turn take precedence over those specified by the configuration of GCC@.
12302 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
12303 GNU Compiler Collection (GCC) Internals}.
12308 @c @itemx LC_COLLATE
12310 @c @itemx LC_MONETARY
12311 @c @itemx LC_NUMERIC
12316 @c @findex LC_COLLATE
12317 @findex LC_MESSAGES
12318 @c @findex LC_MONETARY
12319 @c @findex LC_NUMERIC
12323 These environment variables control the way that GCC uses
12324 localization information that allow GCC to work with different
12325 national conventions. GCC inspects the locale categories
12326 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
12327 so. These locale categories can be set to any value supported by your
12328 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
12329 Kingdom encoded in UTF-8.
12331 The @env{LC_CTYPE} environment variable specifies character
12332 classification. GCC uses it to determine the character boundaries in
12333 a string; this is needed for some multibyte encodings that contain quote
12334 and escape characters that would otherwise be interpreted as a string
12337 The @env{LC_MESSAGES} environment variable specifies the language to
12338 use in diagnostic messages.
12340 If the @env{LC_ALL} environment variable is set, it overrides the value
12341 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
12342 and @env{LC_MESSAGES} default to the value of the @env{LANG}
12343 environment variable. If none of these variables are set, GCC
12344 defaults to traditional C English behavior.
12348 If @env{TMPDIR} is set, it specifies the directory to use for temporary
12349 files. GCC uses temporary files to hold the output of one stage of
12350 compilation which is to be used as input to the next stage: for example,
12351 the output of the preprocessor, which is the input to the compiler
12354 @item GCC_EXEC_PREFIX
12355 @findex GCC_EXEC_PREFIX
12356 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
12357 names of the subprograms executed by the compiler. No slash is added
12358 when this prefix is combined with the name of a subprogram, but you can
12359 specify a prefix that ends with a slash if you wish.
12361 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
12362 an appropriate prefix to use based on the pathname it was invoked with.
12364 If GCC cannot find the subprogram using the specified prefix, it
12365 tries looking in the usual places for the subprogram.
12367 The default value of @env{GCC_EXEC_PREFIX} is
12368 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
12369 of @code{prefix} when you ran the @file{configure} script.
12371 Other prefixes specified with @option{-B} take precedence over this prefix.
12373 This prefix is also used for finding files such as @file{crt0.o} that are
12376 In addition, the prefix is used in an unusual way in finding the
12377 directories to search for header files. For each of the standard
12378 directories whose name normally begins with @samp{/usr/local/lib/gcc}
12379 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
12380 replacing that beginning with the specified prefix to produce an
12381 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
12382 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
12383 These alternate directories are searched first; the standard directories
12386 @item COMPILER_PATH
12387 @findex COMPILER_PATH
12388 The value of @env{COMPILER_PATH} is a colon-separated list of
12389 directories, much like @env{PATH}. GCC tries the directories thus
12390 specified when searching for subprograms, if it can't find the
12391 subprograms using @env{GCC_EXEC_PREFIX}.
12394 @findex LIBRARY_PATH
12395 The value of @env{LIBRARY_PATH} is a colon-separated list of
12396 directories, much like @env{PATH}. When configured as a native compiler,
12397 GCC tries the directories thus specified when searching for special
12398 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
12399 using GCC also uses these directories when searching for ordinary
12400 libraries for the @option{-l} option (but directories specified with
12401 @option{-L} come first).
12405 @cindex locale definition
12406 This variable is used to pass locale information to the compiler. One way in
12407 which this information is used is to determine the character set to be used
12408 when character literals, string literals and comments are parsed in C and C++.
12409 When the compiler is configured to allow multibyte characters,
12410 the following values for @env{LANG} are recognized:
12414 Recognize JIS characters.
12416 Recognize SJIS characters.
12418 Recognize EUCJP characters.
12421 If @env{LANG} is not defined, or if it has some other value, then the
12422 compiler will use mblen and mbtowc as defined by the default locale to
12423 recognize and translate multibyte characters.
12427 Some additional environments variables affect the behavior of the
12430 @include cppenv.texi
12434 @node Precompiled Headers
12435 @section Using Precompiled Headers
12436 @cindex precompiled headers
12437 @cindex speed of compilation
12439 Often large projects have many header files that are included in every
12440 source file. The time the compiler takes to process these header files
12441 over and over again can account for nearly all of the time required to
12442 build the project. To make builds faster, GCC allows users to
12443 `precompile' a header file; then, if builds can use the precompiled
12444 header file they will be much faster.
12446 @strong{Caution:} There are a few known situations where GCC will
12447 crash when trying to use a precompiled header. If you have trouble
12448 with a precompiled header, you should remove the precompiled header
12449 and compile without it. In addition, please use GCC's on-line
12450 defect-tracking system to report any problems you encounter with
12451 precompiled headers. @xref{Bugs}.
12453 To create a precompiled header file, simply compile it as you would any
12454 other file, if necessary using the @option{-x} option to make the driver
12455 treat it as a C or C++ header file. You will probably want to use a
12456 tool like @command{make} to keep the precompiled header up-to-date when
12457 the headers it contains change.
12459 A precompiled header file will be searched for when @code{#include} is
12460 seen in the compilation. As it searches for the included file
12461 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
12462 compiler looks for a precompiled header in each directory just before it
12463 looks for the include file in that directory. The name searched for is
12464 the name specified in the @code{#include} with @samp{.gch} appended. If
12465 the precompiled header file can't be used, it is ignored.
12467 For instance, if you have @code{#include "all.h"}, and you have
12468 @file{all.h.gch} in the same directory as @file{all.h}, then the
12469 precompiled header file will be used if possible, and the original
12470 header will be used otherwise.
12472 Alternatively, you might decide to put the precompiled header file in a
12473 directory and use @option{-I} to ensure that directory is searched
12474 before (or instead of) the directory containing the original header.
12475 Then, if you want to check that the precompiled header file is always
12476 used, you can put a file of the same name as the original header in this
12477 directory containing an @code{#error} command.
12479 This also works with @option{-include}. So yet another way to use
12480 precompiled headers, good for projects not designed with precompiled
12481 header files in mind, is to simply take most of the header files used by
12482 a project, include them from another header file, precompile that header
12483 file, and @option{-include} the precompiled header. If the header files
12484 have guards against multiple inclusion, they will be skipped because
12485 they've already been included (in the precompiled header).
12487 If you need to precompile the same header file for different
12488 languages, targets, or compiler options, you can instead make a
12489 @emph{directory} named like @file{all.h.gch}, and put each precompiled
12490 header in the directory, perhaps using @option{-o}. It doesn't matter
12491 what you call the files in the directory, every precompiled header in
12492 the directory will be considered. The first precompiled header
12493 encountered in the directory that is valid for this compilation will
12494 be used; they're searched in no particular order.
12496 There are many other possibilities, limited only by your imagination,
12497 good sense, and the constraints of your build system.
12499 A precompiled header file can be used only when these conditions apply:
12503 Only one precompiled header can be used in a particular compilation.
12506 A precompiled header can't be used once the first C token is seen. You
12507 can have preprocessor directives before a precompiled header; you can
12508 even include a precompiled header from inside another header, so long as
12509 there are no C tokens before the @code{#include}.
12512 The precompiled header file must be produced for the same language as
12513 the current compilation. You can't use a C precompiled header for a C++
12517 The precompiled header file must be produced by the same compiler
12518 version and configuration as the current compilation is using.
12519 The easiest way to guarantee this is to use the same compiler binary
12520 for creating and using precompiled headers.
12523 Any macros defined before the precompiled header is included must
12524 either be defined in the same way as when the precompiled header was
12525 generated, or must not affect the precompiled header, which usually
12526 means that the they don't appear in the precompiled header at all.
12528 The @option{-D} option is one way to define a macro before a
12529 precompiled header is included; using a @code{#define} can also do it.
12530 There are also some options that define macros implicitly, like
12531 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
12534 @item If debugging information is output when using the precompiled
12535 header, using @option{-g} or similar, the same kind of debugging information
12536 must have been output when building the precompiled header. However,
12537 a precompiled header built using @option{-g} can be used in a compilation
12538 when no debugging information is being output.
12540 @item The same @option{-m} options must generally be used when building
12541 and using the precompiled header. @xref{Submodel Options},
12542 for any cases where this rule is relaxed.
12544 @item Each of the following options must be the same when building and using
12545 the precompiled header:
12547 @gccoptlist{-fexceptions -funit-at-a-time}
12550 Some other command-line options starting with @option{-f},
12551 @option{-p}, or @option{-O} must be defined in the same way as when
12552 the precompiled header was generated. At present, it's not clear
12553 which options are safe to change and which are not; the safest choice
12554 is to use exactly the same options when generating and using the
12555 precompiled header. The following are known to be safe:
12557 @gccoptlist{-fpreprocessed -pedantic-errors}
12561 For all of these except the last, the compiler will automatically
12562 ignore the precompiled header if the conditions aren't met. If you
12563 find an option combination that doesn't work and doesn't cause the
12564 precompiled header to be ignored, please consider filing a bug report,
12567 If you do use differing options when generating and using the
12568 precompiled header, the actual behavior will be a mixture of the
12569 behavior for the options. For instance, if you use @option{-g} to
12570 generate the precompiled header but not when using it, you may or may
12571 not get debugging information for routines in the precompiled header.
12573 @node Running Protoize
12574 @section Running Protoize
12576 The program @code{protoize} is an optional part of GCC@. You can use
12577 it to add prototypes to a program, thus converting the program to ISO
12578 C in one respect. The companion program @code{unprotoize} does the
12579 reverse: it removes argument types from any prototypes that are found.
12581 When you run these programs, you must specify a set of source files as
12582 command line arguments. The conversion programs start out by compiling
12583 these files to see what functions they define. The information gathered
12584 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
12586 After scanning comes actual conversion. The specified files are all
12587 eligible to be converted; any files they include (whether sources or
12588 just headers) are eligible as well.
12590 But not all the eligible files are converted. By default,
12591 @code{protoize} and @code{unprotoize} convert only source and header
12592 files in the current directory. You can specify additional directories
12593 whose files should be converted with the @option{-d @var{directory}}
12594 option. You can also specify particular files to exclude with the
12595 @option{-x @var{file}} option. A file is converted if it is eligible, its
12596 directory name matches one of the specified directory names, and its
12597 name within the directory has not been excluded.
12599 Basic conversion with @code{protoize} consists of rewriting most
12600 function definitions and function declarations to specify the types of
12601 the arguments. The only ones not rewritten are those for varargs
12604 @code{protoize} optionally inserts prototype declarations at the
12605 beginning of the source file, to make them available for any calls that
12606 precede the function's definition. Or it can insert prototype
12607 declarations with block scope in the blocks where undeclared functions
12610 Basic conversion with @code{unprotoize} consists of rewriting most
12611 function declarations to remove any argument types, and rewriting
12612 function definitions to the old-style pre-ISO form.
12614 Both conversion programs print a warning for any function declaration or
12615 definition that they can't convert. You can suppress these warnings
12618 The output from @code{protoize} or @code{unprotoize} replaces the
12619 original source file. The original file is renamed to a name ending
12620 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
12621 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
12622 for DOS) file already exists, then the source file is simply discarded.
12624 @code{protoize} and @code{unprotoize} both depend on GCC itself to
12625 scan the program and collect information about the functions it uses.
12626 So neither of these programs will work until GCC is installed.
12628 Here is a table of the options you can use with @code{protoize} and
12629 @code{unprotoize}. Each option works with both programs unless
12633 @item -B @var{directory}
12634 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
12635 usual directory (normally @file{/usr/local/lib}). This file contains
12636 prototype information about standard system functions. This option
12637 applies only to @code{protoize}.
12639 @item -c @var{compilation-options}
12640 Use @var{compilation-options} as the options when running @command{gcc} to
12641 produce the @samp{.X} files. The special option @option{-aux-info} is
12642 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
12644 Note that the compilation options must be given as a single argument to
12645 @code{protoize} or @code{unprotoize}. If you want to specify several
12646 @command{gcc} options, you must quote the entire set of compilation options
12647 to make them a single word in the shell.
12649 There are certain @command{gcc} arguments that you cannot use, because they
12650 would produce the wrong kind of output. These include @option{-g},
12651 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
12652 the @var{compilation-options}, they are ignored.
12655 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
12656 systems) instead of @samp{.c}. This is convenient if you are converting
12657 a C program to C++. This option applies only to @code{protoize}.
12660 Add explicit global declarations. This means inserting explicit
12661 declarations at the beginning of each source file for each function
12662 that is called in the file and was not declared. These declarations
12663 precede the first function definition that contains a call to an
12664 undeclared function. This option applies only to @code{protoize}.
12666 @item -i @var{string}
12667 Indent old-style parameter declarations with the string @var{string}.
12668 This option applies only to @code{protoize}.
12670 @code{unprotoize} converts prototyped function definitions to old-style
12671 function definitions, where the arguments are declared between the
12672 argument list and the initial @samp{@{}. By default, @code{unprotoize}
12673 uses five spaces as the indentation. If you want to indent with just
12674 one space instead, use @option{-i " "}.
12677 Keep the @samp{.X} files. Normally, they are deleted after conversion
12681 Add explicit local declarations. @code{protoize} with @option{-l} inserts
12682 a prototype declaration for each function in each block which calls the
12683 function without any declaration. This option applies only to
12687 Make no real changes. This mode just prints information about the conversions
12688 that would have been done without @option{-n}.
12691 Make no @samp{.save} files. The original files are simply deleted.
12692 Use this option with caution.
12694 @item -p @var{program}
12695 Use the program @var{program} as the compiler. Normally, the name
12696 @file{gcc} is used.
12699 Work quietly. Most warnings are suppressed.
12702 Print the version number, just like @option{-v} for @command{gcc}.
12705 If you need special compiler options to compile one of your program's
12706 source files, then you should generate that file's @samp{.X} file
12707 specially, by running @command{gcc} on that source file with the
12708 appropriate options and the option @option{-aux-info}. Then run
12709 @code{protoize} on the entire set of files. @code{protoize} will use
12710 the existing @samp{.X} file because it is newer than the source file.
12714 gcc -Dfoo=bar file1.c -aux-info file1.X
12719 You need to include the special files along with the rest in the
12720 @code{protoize} command, even though their @samp{.X} files already
12721 exist, because otherwise they won't get converted.
12723 @xref{Protoize Caveats}, for more information on how to use
12724 @code{protoize} successfully.