1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005 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, 2005 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}}.
55 See the Info entry for @command{gcc}, or
56 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
57 for contributors to GCC@.
62 @chapter GCC Command Options
63 @cindex GCC command options
64 @cindex command options
65 @cindex options, GCC command
67 @c man begin DESCRIPTION
68 When you invoke GCC, it normally does preprocessing, compilation,
69 assembly and linking. The ``overall options'' allow you to stop this
70 process at an intermediate stage. For example, the @option{-c} option
71 says not to run the linker. Then the output consists of object files
72 output by the assembler.
74 Other options are passed on to one stage of processing. Some options
75 control the preprocessor and others the compiler itself. Yet other
76 options control the assembler and linker; most of these are not
77 documented here, since you rarely need to use any of them.
79 @cindex C compilation options
80 Most of the command line options that you can use with GCC are useful
81 for C programs; when an option is only useful with another language
82 (usually C++), the explanation says so explicitly. If the description
83 for a particular option does not mention a source language, you can use
84 that option with all supported languages.
86 @cindex C++ compilation options
87 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
88 options for compiling C++ programs.
90 @cindex grouping options
91 @cindex options, grouping
92 The @command{gcc} program accepts options and file names as operands. Many
93 options have multi-letter names; therefore multiple single-letter options
94 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
97 @cindex order of options
98 @cindex options, order
99 You can mix options and other arguments. For the most part, the order
100 you use doesn't matter. Order does matter when you use several options
101 of the same kind; for example, if you specify @option{-L} more than once,
102 the directories are searched in the order specified.
104 Many options have long names starting with @samp{-f} or with
105 @samp{-W}---for example,
106 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
107 these have both positive and negative forms; the negative form of
108 @option{-ffoo} would be @option{-fno-foo}. This manual documents
109 only one of these two forms, whichever one is not the default.
113 @xref{Option Index}, for an index to GCC's options.
116 * Option Summary:: Brief list of all options, without explanations.
117 * Overall Options:: Controlling the kind of output:
118 an executable, object files, assembler files,
119 or preprocessed source.
120 * Invoking G++:: Compiling C++ programs.
121 * C Dialect Options:: Controlling the variant of C language compiled.
122 * C++ Dialect Options:: Variations on C++.
123 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char}
171 @item C++ Language Options
172 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
173 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
174 -fconserve-space -ffriend-injection -fno-const-strings @gol
175 -fno-elide-constructors @gol
176 -fno-enforce-eh-specs @gol
177 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
178 -fno-implicit-templates @gol
179 -fno-implicit-inline-templates @gol
180 -fno-implement-inlines -fms-extensions @gol
181 -fno-nonansi-builtins -fno-operator-names @gol
182 -fno-optional-diags -fpermissive @gol
183 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
184 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
185 -fno-default-inline -fvisibility-inlines-hidden @gol
186 -Wabi -Wctor-dtor-privacy @gol
187 -Wnon-virtual-dtor -Wreorder @gol
188 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
189 -Wno-non-template-friend -Wold-style-cast @gol
190 -Woverloaded-virtual -Wno-pmf-conversions @gol
193 @item Objective-C and Objective-C++ Language Options
194 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
195 Objective-C and Objective-C++ Dialects}.
197 -fconstant-string-class=@var{class-name} @gol
198 -fgnu-runtime -fnext-runtime @gol
199 -fno-nil-receivers @gol
200 -fobjc-call-cxx-cdtors @gol
201 -fobjc-direct-dispatch @gol
202 -fobjc-exceptions @gol
204 -freplace-objc-classes @gol
207 -Wassign-intercept @gol
208 -Wno-protocol -Wselector @gol
209 -Wstrict-selector-match @gol
210 -Wundeclared-selector}
212 @item Language Independent Options
213 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
214 @gccoptlist{-fmessage-length=@var{n} @gol
215 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}} @gol
216 -fdiagnostics-show-options
218 @item Warning Options
219 @xref{Warning Options,,Options to Request or Suppress Warnings}.
220 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
221 -w -Wextra -Wall -Waggregate-return -Wno-attributes @gol
222 -Wc++-compat -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
223 -Wconversion -Wno-deprecated-declarations @gol
224 -Wdisabled-optimization -Wno-div-by-zero -Wno-endif-labels @gol
225 -Werror -Werror-implicit-function-declaration @gol
226 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
227 -Wno-format-extra-args -Wformat-nonliteral @gol
228 -Wformat-security -Wformat-y2k @gol
229 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
230 -Wimport -Wno-import -Winit-self -Winline @gol
231 -Wno-int-to-pointer-cast @gol
232 -Wno-invalid-offsetof -Winvalid-pch @gol
233 -Wlarger-than-@var{len} -Wunsafe-loop-optimizations -Wlong-long @gol
234 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
235 -Wmissing-format-attribute -Wmissing-include-dirs @gol
236 -Wmissing-noreturn @gol
237 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
238 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
239 -Wredundant-decls @gol
240 -Wreturn-type -Wsequence-point -Wshadow @gol
241 -Wsign-compare -Wstack-protector @gol
242 -Wstrict-aliasing -Wstrict-aliasing=2 @gol
243 -Wswitch -Wswitch-default -Wswitch-enum @gol
244 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
245 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
246 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
247 -Wunused-value -Wunused-variable -Wvariadic-macros @gol
250 @item C-only Warning Options
251 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
252 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
253 -Wstrict-prototypes -Wtraditional @gol
254 -Wdeclaration-after-statement -Wno-pointer-sign}
256 @item Debugging Options
257 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
258 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
259 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
260 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
261 -fdump-ipa-all -fdump-ipa-cgraph @gol
263 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
264 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
265 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
266 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
268 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
269 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
270 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
271 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
272 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
273 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
274 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
275 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
276 -fdump-tree-nrv -fdump-tree-vect @gol
277 -fdump-tree-sink @gol
278 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
279 -fdump-tree-salias @gol
280 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
282 -ftree-vectorizer-verbose=@var{n} @gol
283 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
284 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
285 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
286 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
287 -ftest-coverage -ftime-report -fvar-tracking @gol
288 -g -g@var{level} -gcoff -gdwarf-2 @gol
289 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
290 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
291 -print-multi-directory -print-multi-lib @gol
292 -print-prog-name=@var{program} -print-search-dirs -Q @gol
295 @item Optimization Options
296 @xref{Optimize Options,,Options that Control Optimization}.
297 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
298 -falign-labels=@var{n} -falign-loops=@var{n} @gol
299 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
300 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
301 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
302 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
303 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
304 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
305 -fexpensive-optimizations -ffast-math -ffloat-store @gol
306 -fforce-addr -ffunction-sections @gol
307 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
308 -floop-optimize -fcrossjumping -fif-conversion -fif-conversion2 @gol
309 -finline-functions -finline-functions-called-once @gol
310 -finline-limit=@var{n} -fkeep-inline-functions @gol
311 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
312 -fmodulo-sched -fno-branch-count-reg @gol
313 -fno-default-inline -fno-defer-pop -floop-optimize2 -fmove-loop-invariants @gol
314 -fno-function-cse -fno-guess-branch-probability @gol
315 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
316 -funsafe-math-optimizations -funsafe-loop-optimizations -ffinite-math-only @gol
317 -fno-trapping-math -fno-zero-initialized-in-bss @gol
318 -fomit-frame-pointer -foptimize-register-move @gol
319 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
320 -fprofile-generate -fprofile-use @gol
321 -fregmove -frename-registers @gol
322 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
323 -frerun-cse-after-loop -frerun-loop-opt @gol
324 -frounding-math -fschedule-insns -fschedule-insns2 @gol
325 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
326 -fsched-spec-load-dangerous @gol
327 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
328 -fsched2-use-superblocks @gol
329 -fsched2-use-traces -freschedule-modulo-scheduled-loops @gol
330 -fsignaling-nans -fsingle-precision-constant @gol
331 -fstack-protector -fstack-protector-all @gol
332 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
333 -funroll-all-loops -funroll-loops -fpeel-loops @gol
334 -fsplit-ivs-in-unroller -funswitch-loops @gol
335 -fvariable-expansion-in-unroller @gol
336 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
337 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
338 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
339 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
340 -ftree-vect-loop-version -ftree-salias -fweb @gol
341 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
342 --param @var{name}=@var{value}
343 -O -O0 -O1 -O2 -O3 -Os}
345 @item Preprocessor Options
346 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
347 @gccoptlist{-A@var{question}=@var{answer} @gol
348 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
349 -C -dD -dI -dM -dN @gol
350 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
351 -idirafter @var{dir} @gol
352 -include @var{file} -imacros @var{file} @gol
353 -iprefix @var{file} -iwithprefix @var{dir} @gol
354 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
355 -isysroot @var{dir} @gol
356 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
357 -P -fworking-directory -remap @gol
358 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
359 -Xpreprocessor @var{option}}
361 @item Assembler Option
362 @xref{Assembler Options,,Passing Options to the Assembler}.
363 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
366 @xref{Link Options,,Options for Linking}.
367 @gccoptlist{@var{object-file-name} -l@var{library} @gol
368 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
369 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
370 -Wl,@var{option} -Xlinker @var{option} @gol
373 @item Directory Options
374 @xref{Directory Options,,Options for Directory Search}.
375 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
376 -specs=@var{file} -I- --sysroot=@var{dir}}
379 @c I wrote this xref this way to avoid overfull hbox. -- rms
380 @xref{Target Options}.
381 @gccoptlist{-V @var{version} -b @var{machine}}
383 @item Machine Dependent Options
384 @xref{Submodel Options,,Hardware Models and Configurations}.
385 @c This list is ordered alphanumerically by subsection name.
386 @c Try and put the significant identifier (CPU or system) first,
387 @c so users have a clue at guessing where the ones they want will be.
390 @gccoptlist{-EB -EL @gol
391 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
392 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
395 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
396 -mabi=@var{name} @gol
397 -mapcs-stack-check -mno-apcs-stack-check @gol
398 -mapcs-float -mno-apcs-float @gol
399 -mapcs-reentrant -mno-apcs-reentrant @gol
400 -msched-prolog -mno-sched-prolog @gol
401 -mlittle-endian -mbig-endian -mwords-little-endian @gol
402 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
403 -mthumb-interwork -mno-thumb-interwork @gol
404 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
405 -mstructure-size-boundary=@var{n} @gol
406 -mabort-on-noreturn @gol
407 -mlong-calls -mno-long-calls @gol
408 -msingle-pic-base -mno-single-pic-base @gol
409 -mpic-register=@var{reg} @gol
410 -mnop-fun-dllimport @gol
411 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
412 -mpoke-function-name @gol
414 -mtpcs-frame -mtpcs-leaf-frame @gol
415 -mcaller-super-interworking -mcallee-super-interworking}
418 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
419 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
421 @emph{Blackfin Options}
422 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
423 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
424 -mlow-64k -mno-low64k -mid-shared-library @gol
425 -mno-id-shared-library -mshared-library-id=@var{n} @gol
426 -mlong-calls -mno-long-calls}
429 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
430 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
431 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
432 -mstack-align -mdata-align -mconst-align @gol
433 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
434 -melf -maout -melinux -mlinux -sim -sim2 @gol
435 -mmul-bug-workaround -mno-mul-bug-workaround}
438 @gccoptlist{-mmac -mpush-args}
440 @emph{Darwin Options}
441 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
442 -arch_only -bind_at_load -bundle -bundle_loader @gol
443 -client_name -compatibility_version -current_version @gol
445 -dependency-file -dylib_file -dylinker_install_name @gol
446 -dynamic -dynamiclib -exported_symbols_list @gol
447 -filelist -flat_namespace -force_cpusubtype_ALL @gol
448 -force_flat_namespace -headerpad_max_install_names @gol
449 -image_base -init -install_name -keep_private_externs @gol
450 -multi_module -multiply_defined -multiply_defined_unused @gol
451 -noall_load -no_dead_strip_inits_and_terms @gol
452 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
453 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
454 -private_bundle -read_only_relocs -sectalign @gol
455 -sectobjectsymbols -whyload -seg1addr @gol
456 -sectcreate -sectobjectsymbols -sectorder @gol
457 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
458 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
459 -segprot -segs_read_only_addr -segs_read_write_addr @gol
460 -single_module -static -sub_library -sub_umbrella @gol
461 -twolevel_namespace -umbrella -undefined @gol
462 -unexported_symbols_list -weak_reference_mismatches @gol
463 -whatsloaded -F -gused -gfull -mmacosx-min-version=@var{version} @gol
466 @emph{DEC Alpha Options}
467 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
468 -mieee -mieee-with-inexact -mieee-conformant @gol
469 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
470 -mtrap-precision=@var{mode} -mbuild-constants @gol
471 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
472 -mbwx -mmax -mfix -mcix @gol
473 -mfloat-vax -mfloat-ieee @gol
474 -mexplicit-relocs -msmall-data -mlarge-data @gol
475 -msmall-text -mlarge-text @gol
476 -mmemory-latency=@var{time}}
478 @emph{DEC Alpha/VMS Options}
479 @gccoptlist{-mvms-return-codes}
482 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
483 -mhard-float -msoft-float @gol
484 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
485 -mdouble -mno-double @gol
486 -mmedia -mno-media -mmuladd -mno-muladd @gol
487 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
488 -mlinked-fp -mlong-calls -malign-labels @gol
489 -mlibrary-pic -macc-4 -macc-8 @gol
490 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
491 -moptimize-membar -mno-optimize-membar @gol
492 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
493 -mvliw-branch -mno-vliw-branch @gol
494 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
495 -mno-nested-cond-exec -mtomcat-stats @gol
499 @emph{H8/300 Options}
500 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
503 @gccoptlist{-march=@var{architecture-type} @gol
504 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
505 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
506 -mfixed-range=@var{register-range} @gol
507 -mjump-in-delay -mlinker-opt -mlong-calls @gol
508 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
509 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
510 -mno-jump-in-delay -mno-long-load-store @gol
511 -mno-portable-runtime -mno-soft-float @gol
512 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
513 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
514 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
515 -munix=@var{unix-std} -nolibdld -static -threads}
517 @emph{i386 and x86-64 Options}
518 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
519 -mfpmath=@var{unit} @gol
520 -masm=@var{dialect} -mno-fancy-math-387 @gol
521 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
522 -mno-wide-multiply -mrtd -malign-double @gol
523 -mpreferred-stack-boundary=@var{num} @gol
524 -mmmx -msse -msse2 -msse3 -m3dnow @gol
525 -mthreads -mno-align-stringops -minline-all-stringops @gol
526 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
527 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
528 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
529 -mcmodel=@var{code-model} @gol
530 -m32 -m64 -mlarge-data-threshold=@var{num}}
533 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
534 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
535 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
536 -minline-float-divide-max-throughput @gol
537 -minline-int-divide-min-latency @gol
538 -minline-int-divide-max-throughput @gol
539 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
540 -mno-dwarf2-asm -mearly-stop-bits @gol
541 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
542 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64}
544 @emph{M32R/D Options}
545 @gccoptlist{-m32r2 -m32rx -m32r @gol
547 -malign-loops -mno-align-loops @gol
548 -missue-rate=@var{number} @gol
549 -mbranch-cost=@var{number} @gol
550 -mmodel=@var{code-size-model-type} @gol
551 -msdata=@var{sdata-type} @gol
552 -mno-flush-func -mflush-func=@var{name} @gol
553 -mno-flush-trap -mflush-trap=@var{number} @gol
557 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
559 @emph{M680x0 Options}
560 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
561 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
562 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
563 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
564 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
566 @emph{M68hc1x Options}
567 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
568 -mauto-incdec -minmax -mlong-calls -mshort @gol
569 -msoft-reg-count=@var{count}}
572 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
573 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
574 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
575 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
576 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
579 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
580 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
581 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
582 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
583 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
584 -mdsp -mpaired-single -mips3d @gol
585 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
586 -G@var{num} -membedded-data -mno-embedded-data @gol
587 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
588 -msplit-addresses -mno-split-addresses @gol
589 -mexplicit-relocs -mno-explicit-relocs @gol
590 -mcheck-zero-division -mno-check-zero-division @gol
591 -mdivide-traps -mdivide-breaks @gol
592 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
593 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
594 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
595 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
596 -mfix-sb1 -mno-fix-sb1 @gol
597 -mflush-func=@var{func} -mno-flush-func @gol
598 -mbranch-likely -mno-branch-likely @gol
599 -mfp-exceptions -mno-fp-exceptions @gol
600 -mvr4130-align -mno-vr4130-align}
603 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
604 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
605 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
606 -mno-base-addresses -msingle-exit -mno-single-exit}
608 @emph{MN10300 Options}
609 @gccoptlist{-mmult-bug -mno-mult-bug @gol
610 -mam33 -mno-am33 @gol
611 -mam33-2 -mno-am33-2 @gol
615 @gccoptlist{-mno-crt0 -mmul -mbacc -msim @gol
616 -march=@var{cpu-type} }
618 @emph{PDP-11 Options}
619 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
620 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
621 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
622 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
623 -mbranch-expensive -mbranch-cheap @gol
624 -msplit -mno-split -munix-asm -mdec-asm}
626 @emph{PowerPC Options}
627 See RS/6000 and PowerPC Options.
629 @emph{RS/6000 and PowerPC Options}
630 @gccoptlist{-mcpu=@var{cpu-type} @gol
631 -mtune=@var{cpu-type} @gol
632 -mpower -mno-power -mpower2 -mno-power2 @gol
633 -mpowerpc -mpowerpc64 -mno-powerpc @gol
634 -maltivec -mno-altivec @gol
635 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
636 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
637 -mnew-mnemonics -mold-mnemonics @gol
638 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
639 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
640 -malign-power -malign-natural @gol
641 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
642 -mstring -mno-string -mupdate -mno-update @gol
643 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
644 -mstrict-align -mno-strict-align -mrelocatable @gol
645 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
646 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
647 -mdynamic-no-pic -maltivec -mswdiv @gol
648 -mprioritize-restricted-insns=@var{priority} @gol
649 -msched-costly-dep=@var{dependence_type} @gol
650 -minsert-sched-nops=@var{scheme} @gol
651 -mcall-sysv -mcall-netbsd @gol
652 -maix-struct-return -msvr4-struct-return @gol
653 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
654 -misel -mno-isel @gol
655 -misel=yes -misel=no @gol
657 -mspe=yes -mspe=no @gol
658 -mvrsave -mno-vrsave @gol
659 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
660 -mprototype -mno-prototype @gol
661 -msim -mmvme -mads -myellowknife -memb -msdata @gol
662 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
664 @emph{S/390 and zSeries Options}
665 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
666 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
667 -mpacked-stack -mno-packed-stack @gol
668 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
669 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
670 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
671 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
674 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
675 -m4-nofpu -m4-single-only -m4-single -m4 @gol
676 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
677 -m5-64media -m5-64media-nofpu @gol
678 -m5-32media -m5-32media-nofpu @gol
679 -m5-compact -m5-compact-nofpu @gol
680 -mb -ml -mdalign -mrelax @gol
681 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
682 -mieee -misize -mpadstruct -mspace @gol
683 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
684 -mdivsi3_libfunc=@var{name} @gol
685 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
689 @gccoptlist{-mcpu=@var{cpu-type} @gol
690 -mtune=@var{cpu-type} @gol
691 -mcmodel=@var{code-model} @gol
692 -m32 -m64 -mapp-regs -mno-app-regs @gol
693 -mfaster-structs -mno-faster-structs @gol
694 -mfpu -mno-fpu -mhard-float -msoft-float @gol
695 -mhard-quad-float -msoft-quad-float @gol
696 -mimpure-text -mno-impure-text -mlittle-endian @gol
697 -mstack-bias -mno-stack-bias @gol
698 -munaligned-doubles -mno-unaligned-doubles @gol
699 -mv8plus -mno-v8plus -mvis -mno-vis
702 @emph{System V Options}
703 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
705 @emph{TMS320C3x/C4x Options}
706 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
707 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
708 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
709 -mparallel-insns -mparallel-mpy -mpreserve-float}
712 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
713 -mprolog-function -mno-prolog-function -mspace @gol
714 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
715 -mapp-regs -mno-app-regs @gol
716 -mdisable-callt -mno-disable-callt @gol
722 @gccoptlist{-mg -mgnu -munix}
724 @emph{x86-64 Options}
725 See i386 and x86-64 Options.
727 @emph{Xstormy16 Options}
730 @emph{Xtensa Options}
731 @gccoptlist{-mconst16 -mno-const16 @gol
732 -mfused-madd -mno-fused-madd @gol
733 -mtext-section-literals -mno-text-section-literals @gol
734 -mtarget-align -mno-target-align @gol
735 -mlongcalls -mno-longcalls}
737 @emph{zSeries Options}
738 See S/390 and zSeries Options.
740 @item Code Generation Options
741 @xref{Code Gen Options,,Options for Code Generation Conventions}.
742 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
743 -ffixed-@var{reg} -fexceptions @gol
744 -fnon-call-exceptions -funwind-tables @gol
745 -fasynchronous-unwind-tables @gol
746 -finhibit-size-directive -finstrument-functions @gol
747 -fno-common -fno-ident @gol
748 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
749 -fno-jump-tables @gol
750 -freg-struct-return -fshared-data -fshort-enums @gol
751 -fshort-double -fshort-wchar @gol
752 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
753 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
754 -fargument-alias -fargument-noalias @gol
755 -fargument-noalias-global -fleading-underscore @gol
756 -ftls-model=@var{model} @gol
757 -ftrapv -fwrapv -fbounds-check @gol
762 * Overall Options:: Controlling the kind of output:
763 an executable, object files, assembler files,
764 or preprocessed source.
765 * C Dialect Options:: Controlling the variant of C language compiled.
766 * C++ Dialect Options:: Variations on C++.
767 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
769 * Language Independent Options:: Controlling how diagnostics should be
771 * Warning Options:: How picky should the compiler be?
772 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
773 * Optimize Options:: How much optimization?
774 * Preprocessor Options:: Controlling header files and macro definitions.
775 Also, getting dependency information for Make.
776 * Assembler Options:: Passing options to the assembler.
777 * Link Options:: Specifying libraries and so on.
778 * Directory Options:: Where to find header files and libraries.
779 Where to find the compiler executable files.
780 * Spec Files:: How to pass switches to sub-processes.
781 * Target Options:: Running a cross-compiler, or an old version of GCC.
784 @node Overall Options
785 @section Options Controlling the Kind of Output
787 Compilation can involve up to four stages: preprocessing, compilation
788 proper, assembly and linking, always in that order. GCC is capable of
789 preprocessing and compiling several files either into several
790 assembler input files, or into one assembler input file; then each
791 assembler input file produces an object file, and linking combines all
792 the object files (those newly compiled, and those specified as input)
793 into an executable file.
795 @cindex file name suffix
796 For any given input file, the file name suffix determines what kind of
801 C source code which must be preprocessed.
804 C source code which should not be preprocessed.
807 C++ source code which should not be preprocessed.
810 Objective-C source code. Note that you must link with the @file{libobjc}
811 library to make an Objective-C program work.
814 Objective-C source code which should not be preprocessed.
818 Objective-C++ source code. Note that you must link with the @file{libobjc}
819 library to make an Objective-C++ program work. Note that @samp{.M} refers
820 to a literal capital M@.
823 Objective-C++ source code which should not be preprocessed.
826 C, C++, Objective-C or Objective-C++ header file to be turned into a
831 @itemx @var{file}.cxx
832 @itemx @var{file}.cpp
833 @itemx @var{file}.CPP
834 @itemx @var{file}.c++
836 C++ source code which must be preprocessed. Note that in @samp{.cxx},
837 the last two letters must both be literally @samp{x}. Likewise,
838 @samp{.C} refers to a literal capital C@.
842 Objective-C++ source code which must be preprocessed.
845 Objective-C++ source code which should not be preprocessed.
849 C++ header file to be turned into a precompiled header.
852 @itemx @var{file}.for
853 @itemx @var{file}.FOR
854 Fortran source code which should not be preprocessed.
857 @itemx @var{file}.fpp
858 @itemx @var{file}.FPP
859 Fortran source code which must be preprocessed (with the traditional
863 Fortran source code which must be preprocessed with a RATFOR
864 preprocessor (not included with GCC)@.
867 @itemx @var{file}.f95
868 Fortran 90/95 source code which should not be preprocessed.
871 @itemx @var{file}.F95
872 Fortran 90/95 source code which must be preprocessed (with the
873 traditional preprocessor).
875 @c FIXME: Descriptions of Java file types.
882 Ada source code file which contains a library unit declaration (a
883 declaration of a package, subprogram, or generic, or a generic
884 instantiation), or a library unit renaming declaration (a package,
885 generic, or subprogram renaming declaration). Such files are also
888 @itemx @var{file}.adb
889 Ada source code file containing a library unit body (a subprogram or
890 package body). Such files are also called @dfn{bodies}.
892 @c GCC also knows about some suffixes for languages not yet included:
901 Assembler code which must be preprocessed.
904 An object file to be fed straight into linking.
905 Any file name with no recognized suffix is treated this way.
909 You can specify the input language explicitly with the @option{-x} option:
912 @item -x @var{language}
913 Specify explicitly the @var{language} for the following input files
914 (rather than letting the compiler choose a default based on the file
915 name suffix). This option applies to all following input files until
916 the next @option{-x} option. Possible values for @var{language} are:
918 c c-header c-cpp-output
919 c++ c++-header c++-cpp-output
920 objective-c objective-c-header objective-c-cpp-output
921 objective-c++ objective-c++-header objective-c++-cpp-output
922 assembler assembler-with-cpp
924 f77 f77-cpp-input ratfor
931 Turn off any specification of a language, so that subsequent files are
932 handled according to their file name suffixes (as they are if @option{-x}
933 has not been used at all).
935 @item -pass-exit-codes
936 @opindex pass-exit-codes
937 Normally the @command{gcc} program will exit with the code of 1 if any
938 phase of the compiler returns a non-success return code. If you specify
939 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
940 numerically highest error produced by any phase that returned an error
944 If you only want some of the stages of compilation, you can use
945 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
946 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
947 @command{gcc} is to stop. Note that some combinations (for example,
948 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
953 Compile or assemble the source files, but do not link. The linking
954 stage simply is not done. The ultimate output is in the form of an
955 object file for each source file.
957 By default, the object file name for a source file is made by replacing
958 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
960 Unrecognized input files, not requiring compilation or assembly, are
965 Stop after the stage of compilation proper; do not assemble. The output
966 is in the form of an assembler code file for each non-assembler input
969 By default, the assembler file name for a source file is made by
970 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
972 Input files that don't require compilation are ignored.
976 Stop after the preprocessing stage; do not run the compiler proper. The
977 output is in the form of preprocessed source code, which is sent to the
980 Input files which don't require preprocessing are ignored.
982 @cindex output file option
985 Place output in file @var{file}. This applies regardless to whatever
986 sort of output is being produced, whether it be an executable file,
987 an object file, an assembler file or preprocessed C code.
989 If @option{-o} is not specified, the default is to put an executable
990 file in @file{a.out}, the object file for
991 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
992 assembler file in @file{@var{source}.s}, a precompiled header file in
993 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
998 Print (on standard error output) the commands executed to run the stages
999 of compilation. Also print the version number of the compiler driver
1000 program and of the preprocessor and the compiler proper.
1004 Like @option{-v} except the commands are not executed and all command
1005 arguments are quoted. This is useful for shell scripts to capture the
1006 driver-generated command lines.
1010 Use pipes rather than temporary files for communication between the
1011 various stages of compilation. This fails to work on some systems where
1012 the assembler is unable to read from a pipe; but the GNU assembler has
1017 If you are compiling multiple source files, this option tells the driver
1018 to pass all the source files to the compiler at once (for those
1019 languages for which the compiler can handle this). This will allow
1020 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1021 language for which this is supported is C@. If you pass source files for
1022 multiple languages to the driver, using this option, the driver will invoke
1023 the compiler(s) that support IMA once each, passing each compiler all the
1024 source files appropriate for it. For those languages that do not support
1025 IMA this option will be ignored, and the compiler will be invoked once for
1026 each source file in that language. If you use this option in conjunction
1027 with @option{-save-temps}, the compiler will generate multiple
1029 (one for each source file), but only one (combined) @file{.o} or
1034 Print (on the standard output) a description of the command line options
1035 understood by @command{gcc}. If the @option{-v} option is also specified
1036 then @option{--help} will also be passed on to the various processes
1037 invoked by @command{gcc}, so that they can display the command line options
1038 they accept. If the @option{-Wextra} option is also specified then command
1039 line options which have no documentation associated with them will also
1043 @opindex target-help
1044 Print (on the standard output) a description of target specific command
1045 line options for each tool.
1049 Display the version number and copyrights of the invoked GCC@.
1053 @section Compiling C++ Programs
1055 @cindex suffixes for C++ source
1056 @cindex C++ source file suffixes
1057 C++ source files conventionally use one of the suffixes @samp{.C},
1058 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1059 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1060 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1061 files with these names and compiles them as C++ programs even if you
1062 call the compiler the same way as for compiling C programs (usually
1063 with the name @command{gcc}).
1067 However, C++ programs often require class libraries as well as a
1068 compiler that understands the C++ language---and under some
1069 circumstances, you might want to compile programs or header files from
1070 standard input, or otherwise without a suffix that flags them as C++
1071 programs. You might also like to precompile a C header file with a
1072 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1073 program that calls GCC with the default language set to C++, and
1074 automatically specifies linking against the C++ library. On many
1075 systems, @command{g++} is also installed with the name @command{c++}.
1077 @cindex invoking @command{g++}
1078 When you compile C++ programs, you may specify many of the same
1079 command-line options that you use for compiling programs in any
1080 language; or command-line options meaningful for C and related
1081 languages; or options that are meaningful only for C++ programs.
1082 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1083 explanations of options for languages related to C@.
1084 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1085 explanations of options that are meaningful only for C++ programs.
1087 @node C Dialect Options
1088 @section Options Controlling C Dialect
1089 @cindex dialect options
1090 @cindex language dialect options
1091 @cindex options, dialect
1093 The following options control the dialect of C (or languages derived
1094 from C, such as C++, Objective-C and Objective-C++) that the compiler
1098 @cindex ANSI support
1102 In C mode, support all ISO C90 programs. In C++ mode,
1103 remove GNU extensions that conflict with ISO C++.
1105 This turns off certain features of GCC that are incompatible with ISO
1106 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1107 such as the @code{asm} and @code{typeof} keywords, and
1108 predefined macros such as @code{unix} and @code{vax} that identify the
1109 type of system you are using. It also enables the undesirable and
1110 rarely used ISO trigraph feature. For the C compiler,
1111 it disables recognition of C++ style @samp{//} comments as well as
1112 the @code{inline} keyword.
1114 The alternate keywords @code{__asm__}, @code{__extension__},
1115 @code{__inline__} and @code{__typeof__} continue to work despite
1116 @option{-ansi}. You would not want to use them in an ISO C program, of
1117 course, but it is useful to put them in header files that might be included
1118 in compilations done with @option{-ansi}. Alternate predefined macros
1119 such as @code{__unix__} and @code{__vax__} are also available, with or
1120 without @option{-ansi}.
1122 The @option{-ansi} option does not cause non-ISO programs to be
1123 rejected gratuitously. For that, @option{-pedantic} is required in
1124 addition to @option{-ansi}. @xref{Warning Options}.
1126 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1127 option is used. Some header files may notice this macro and refrain
1128 from declaring certain functions or defining certain macros that the
1129 ISO standard doesn't call for; this is to avoid interfering with any
1130 programs that might use these names for other things.
1132 Functions which would normally be built in but do not have semantics
1133 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1134 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1135 built-in functions provided by GCC}, for details of the functions
1140 Determine the language standard. This option is currently only
1141 supported when compiling C or C++. A value for this option must be
1142 provided; possible values are
1147 ISO C90 (same as @option{-ansi}).
1149 @item iso9899:199409
1150 ISO C90 as modified in amendment 1.
1156 ISO C99. Note that this standard is not yet fully supported; see
1157 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1158 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1161 Default, ISO C90 plus GNU extensions (including some C99 features).
1165 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1166 this will become the default. The name @samp{gnu9x} is deprecated.
1169 The 1998 ISO C++ standard plus amendments.
1172 The same as @option{-std=c++98} plus GNU extensions. This is the
1173 default for C++ code.
1176 Even when this option is not specified, you can still use some of the
1177 features of newer standards in so far as they do not conflict with
1178 previous C standards. For example, you may use @code{__restrict__} even
1179 when @option{-std=c99} is not specified.
1181 The @option{-std} options specifying some version of ISO C have the same
1182 effects as @option{-ansi}, except that features that were not in ISO C90
1183 but are in the specified version (for example, @samp{//} comments and
1184 the @code{inline} keyword in ISO C99) are not disabled.
1186 @xref{Standards,,Language Standards Supported by GCC}, for details of
1187 these standard versions.
1189 @item -aux-info @var{filename}
1191 Output to the given filename prototyped declarations for all functions
1192 declared and/or defined in a translation unit, including those in header
1193 files. This option is silently ignored in any language other than C@.
1195 Besides declarations, the file indicates, in comments, the origin of
1196 each declaration (source file and line), whether the declaration was
1197 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1198 @samp{O} for old, respectively, in the first character after the line
1199 number and the colon), and whether it came from a declaration or a
1200 definition (@samp{C} or @samp{F}, respectively, in the following
1201 character). In the case of function definitions, a K&R-style list of
1202 arguments followed by their declarations is also provided, inside
1203 comments, after the declaration.
1207 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1208 keyword, so that code can use these words as identifiers. You can use
1209 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1210 instead. @option{-ansi} implies @option{-fno-asm}.
1212 In C++, this switch only affects the @code{typeof} keyword, since
1213 @code{asm} and @code{inline} are standard keywords. You may want to
1214 use the @option{-fno-gnu-keywords} flag instead, which has the same
1215 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1216 switch only affects the @code{asm} and @code{typeof} keywords, since
1217 @code{inline} is a standard keyword in ISO C99.
1220 @itemx -fno-builtin-@var{function}
1221 @opindex fno-builtin
1222 @cindex built-in functions
1223 Don't recognize built-in functions that do not begin with
1224 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1225 functions provided by GCC}, for details of the functions affected,
1226 including those which are not built-in functions when @option{-ansi} or
1227 @option{-std} options for strict ISO C conformance are used because they
1228 do not have an ISO standard meaning.
1230 GCC normally generates special code to handle certain built-in functions
1231 more efficiently; for instance, calls to @code{alloca} may become single
1232 instructions that adjust the stack directly, and calls to @code{memcpy}
1233 may become inline copy loops. The resulting code is often both smaller
1234 and faster, but since the function calls no longer appear as such, you
1235 cannot set a breakpoint on those calls, nor can you change the behavior
1236 of the functions by linking with a different library. In addition,
1237 when a function is recognized as a built-in function, GCC may use
1238 information about that function to warn about problems with calls to
1239 that function, or to generate more efficient code, even if the
1240 resulting code still contains calls to that function. For example,
1241 warnings are given with @option{-Wformat} for bad calls to
1242 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1243 known not to modify global memory.
1245 With the @option{-fno-builtin-@var{function}} option
1246 only the built-in function @var{function} is
1247 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1248 function is named this is not built-in in this version of GCC, this
1249 option is ignored. There is no corresponding
1250 @option{-fbuiltin-@var{function}} option; if you wish to enable
1251 built-in functions selectively when using @option{-fno-builtin} or
1252 @option{-ffreestanding}, you may define macros such as:
1255 #define abs(n) __builtin_abs ((n))
1256 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1261 @cindex hosted environment
1263 Assert that compilation takes place in a hosted environment. This implies
1264 @option{-fbuiltin}. A hosted environment is one in which the
1265 entire standard library is available, and in which @code{main} has a return
1266 type of @code{int}. Examples are nearly everything except a kernel.
1267 This is equivalent to @option{-fno-freestanding}.
1269 @item -ffreestanding
1270 @opindex ffreestanding
1271 @cindex hosted environment
1273 Assert that compilation takes place in a freestanding environment. This
1274 implies @option{-fno-builtin}. A freestanding environment
1275 is one in which the standard library may not exist, and program startup may
1276 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1277 This is equivalent to @option{-fno-hosted}.
1279 @xref{Standards,,Language Standards Supported by GCC}, for details of
1280 freestanding and hosted environments.
1282 @item -fms-extensions
1283 @opindex fms-extensions
1284 Accept some non-standard constructs used in Microsoft header files.
1286 Some cases of unnamed fields in structures and unions are only
1287 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1288 fields within structs/unions}, for details.
1292 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1293 options for strict ISO C conformance) implies @option{-trigraphs}.
1295 @item -no-integrated-cpp
1296 @opindex no-integrated-cpp
1297 Performs a compilation in two passes: preprocessing and compiling. This
1298 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1299 @option{-B} option. The user supplied compilation step can then add in
1300 an additional preprocessing step after normal preprocessing but before
1301 compiling. The default is to use the integrated cpp (internal cpp)
1303 The semantics of this option will change if "cc1", "cc1plus", and
1304 "cc1obj" are merged.
1306 @cindex traditional C language
1307 @cindex C language, traditional
1309 @itemx -traditional-cpp
1310 @opindex traditional-cpp
1311 @opindex traditional
1312 Formerly, these options caused GCC to attempt to emulate a pre-standard
1313 C compiler. They are now only supported with the @option{-E} switch.
1314 The preprocessor continues to support a pre-standard mode. See the GNU
1315 CPP manual for details.
1317 @item -fcond-mismatch
1318 @opindex fcond-mismatch
1319 Allow conditional expressions with mismatched types in the second and
1320 third arguments. The value of such an expression is void. This option
1321 is not supported for C++.
1323 @item -funsigned-char
1324 @opindex funsigned-char
1325 Let the type @code{char} be unsigned, like @code{unsigned char}.
1327 Each kind of machine has a default for what @code{char} should
1328 be. It is either like @code{unsigned char} by default or like
1329 @code{signed char} by default.
1331 Ideally, a portable program should always use @code{signed char} or
1332 @code{unsigned char} when it depends on the signedness of an object.
1333 But many programs have been written to use plain @code{char} and
1334 expect it to be signed, or expect it to be unsigned, depending on the
1335 machines they were written for. This option, and its inverse, let you
1336 make such a program work with the opposite default.
1338 The type @code{char} is always a distinct type from each of
1339 @code{signed char} or @code{unsigned char}, even though its behavior
1340 is always just like one of those two.
1343 @opindex fsigned-char
1344 Let the type @code{char} be signed, like @code{signed char}.
1346 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1347 the negative form of @option{-funsigned-char}. Likewise, the option
1348 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1350 @item -fsigned-bitfields
1351 @itemx -funsigned-bitfields
1352 @itemx -fno-signed-bitfields
1353 @itemx -fno-unsigned-bitfields
1354 @opindex fsigned-bitfields
1355 @opindex funsigned-bitfields
1356 @opindex fno-signed-bitfields
1357 @opindex fno-unsigned-bitfields
1358 These options control whether a bit-field is signed or unsigned, when the
1359 declaration does not use either @code{signed} or @code{unsigned}. By
1360 default, such a bit-field is signed, because this is consistent: the
1361 basic integer types such as @code{int} are signed types.
1364 @node C++ Dialect Options
1365 @section Options Controlling C++ Dialect
1367 @cindex compiler options, C++
1368 @cindex C++ options, command line
1369 @cindex options, C++
1370 This section describes the command-line options that are only meaningful
1371 for C++ programs; but you can also use most of the GNU compiler options
1372 regardless of what language your program is in. For example, you
1373 might compile a file @code{firstClass.C} like this:
1376 g++ -g -frepo -O -c firstClass.C
1380 In this example, only @option{-frepo} is an option meant
1381 only for C++ programs; you can use the other options with any
1382 language supported by GCC@.
1384 Here is a list of options that are @emph{only} for compiling C++ programs:
1388 @item -fabi-version=@var{n}
1389 @opindex fabi-version
1390 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1391 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1392 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1393 the version that conforms most closely to the C++ ABI specification.
1394 Therefore, the ABI obtained using version 0 will change as ABI bugs
1397 The default is version 2.
1399 @item -fno-access-control
1400 @opindex fno-access-control
1401 Turn off all access checking. This switch is mainly useful for working
1402 around bugs in the access control code.
1406 Check that the pointer returned by @code{operator new} is non-null
1407 before attempting to modify the storage allocated. This check is
1408 normally unnecessary because the C++ standard specifies that
1409 @code{operator new} will only return @code{0} if it is declared
1410 @samp{throw()}, in which case the compiler will always check the
1411 return value even without this option. In all other cases, when
1412 @code{operator new} has a non-empty exception specification, memory
1413 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1414 @samp{new (nothrow)}.
1416 @item -fconserve-space
1417 @opindex fconserve-space
1418 Put uninitialized or runtime-initialized global variables into the
1419 common segment, as C does. This saves space in the executable at the
1420 cost of not diagnosing duplicate definitions. If you compile with this
1421 flag and your program mysteriously crashes after @code{main()} has
1422 completed, you may have an object that is being destroyed twice because
1423 two definitions were merged.
1425 This option is no longer useful on most targets, now that support has
1426 been added for putting variables into BSS without making them common.
1428 @item -ffriend-injection
1429 @opindex ffriend-injection
1430 Inject friend functions into the enclosing namespace, so that they are
1431 visible outside the scope of the class in which they are declared.
1432 Friend functions were documented to work this way in the old Annotated
1433 C++ Reference Manual, and versions of G++ before 4.1 always worked
1434 that way. However, in ISO C++ a friend function which is not declared
1435 in an enclosing scope can only be found using argument dependent
1436 lookup. This option causes friends to be injected as they were in
1439 This option is for compatibility, and may be removed in a future
1442 @item -fno-const-strings
1443 @opindex fno-const-strings
1444 Give string constants type @code{char *} instead of type @code{const
1445 char *}. By default, G++ uses type @code{const char *} as required by
1446 the standard. Even if you use @option{-fno-const-strings}, you cannot
1447 actually modify the value of a string constant.
1449 This option might be removed in a future release of G++. For maximum
1450 portability, you should structure your code so that it works with
1451 string constants that have type @code{const char *}.
1453 @item -fno-elide-constructors
1454 @opindex fno-elide-constructors
1455 The C++ standard allows an implementation to omit creating a temporary
1456 which is only used to initialize another object of the same type.
1457 Specifying this option disables that optimization, and forces G++ to
1458 call the copy constructor in all cases.
1460 @item -fno-enforce-eh-specs
1461 @opindex fno-enforce-eh-specs
1462 Don't check for violation of exception specifications at runtime. This
1463 option violates the C++ standard, but may be useful for reducing code
1464 size in production builds, much like defining @samp{NDEBUG}. The compiler
1465 will still optimize based on the exception specifications.
1468 @itemx -fno-for-scope
1470 @opindex fno-for-scope
1471 If @option{-ffor-scope} is specified, the scope of variables declared in
1472 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1473 as specified by the C++ standard.
1474 If @option{-fno-for-scope} is specified, the scope of variables declared in
1475 a @i{for-init-statement} extends to the end of the enclosing scope,
1476 as was the case in old versions of G++, and other (traditional)
1477 implementations of C++.
1479 The default if neither flag is given to follow the standard,
1480 but to allow and give a warning for old-style code that would
1481 otherwise be invalid, or have different behavior.
1483 @item -fno-gnu-keywords
1484 @opindex fno-gnu-keywords
1485 Do not recognize @code{typeof} as a keyword, so that code can use this
1486 word as an identifier. You can use the keyword @code{__typeof__} instead.
1487 @option{-ansi} implies @option{-fno-gnu-keywords}.
1489 @item -fno-implicit-templates
1490 @opindex fno-implicit-templates
1491 Never emit code for non-inline templates which are instantiated
1492 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1493 @xref{Template Instantiation}, for more information.
1495 @item -fno-implicit-inline-templates
1496 @opindex fno-implicit-inline-templates
1497 Don't emit code for implicit instantiations of inline templates, either.
1498 The default is to handle inlines differently so that compiles with and
1499 without optimization will need the same set of explicit instantiations.
1501 @item -fno-implement-inlines
1502 @opindex fno-implement-inlines
1503 To save space, do not emit out-of-line copies of inline functions
1504 controlled by @samp{#pragma implementation}. This will cause linker
1505 errors if these functions are not inlined everywhere they are called.
1507 @item -fms-extensions
1508 @opindex fms-extensions
1509 Disable pedantic warnings about constructs used in MFC, such as implicit
1510 int and getting a pointer to member function via non-standard syntax.
1512 @item -fno-nonansi-builtins
1513 @opindex fno-nonansi-builtins
1514 Disable built-in declarations of functions that are not mandated by
1515 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1516 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1518 @item -fno-operator-names
1519 @opindex fno-operator-names
1520 Do not treat the operator name keywords @code{and}, @code{bitand},
1521 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1522 synonyms as keywords.
1524 @item -fno-optional-diags
1525 @opindex fno-optional-diags
1526 Disable diagnostics that the standard says a compiler does not need to
1527 issue. Currently, the only such diagnostic issued by G++ is the one for
1528 a name having multiple meanings within a class.
1531 @opindex fpermissive
1532 Downgrade some diagnostics about nonconformant code from errors to
1533 warnings. Thus, using @option{-fpermissive} will allow some
1534 nonconforming code to compile.
1538 Enable automatic template instantiation at link time. This option also
1539 implies @option{-fno-implicit-templates}. @xref{Template
1540 Instantiation}, for more information.
1544 Disable generation of information about every class with virtual
1545 functions for use by the C++ runtime type identification features
1546 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1547 of the language, you can save some space by using this flag. Note that
1548 exception handling uses the same information, but it will generate it as
1553 Emit statistics about front-end processing at the end of the compilation.
1554 This information is generally only useful to the G++ development team.
1556 @item -ftemplate-depth-@var{n}
1557 @opindex ftemplate-depth
1558 Set the maximum instantiation depth for template classes to @var{n}.
1559 A limit on the template instantiation depth is needed to detect
1560 endless recursions during template class instantiation. ANSI/ISO C++
1561 conforming programs must not rely on a maximum depth greater than 17.
1563 @item -fno-threadsafe-statics
1564 @opindex fno-threadsafe-statics
1565 Do not emit the extra code to use the routines specified in the C++
1566 ABI for thread-safe initialization of local statics. You can use this
1567 option to reduce code size slightly in code that doesn't need to be
1570 @item -fuse-cxa-atexit
1571 @opindex fuse-cxa-atexit
1572 Register destructors for objects with static storage duration with the
1573 @code{__cxa_atexit} function rather than the @code{atexit} function.
1574 This option is required for fully standards-compliant handling of static
1575 destructors, but will only work if your C library supports
1576 @code{__cxa_atexit}.
1578 @item -fvisibility-inlines-hidden
1579 @opindex fvisibility-inlines-hidden
1580 Causes all inlined methods to be marked with
1581 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1582 appear in the export table of a DSO and do not require a PLT indirection
1583 when used within the DSO@. Enabling this option can have a dramatic effect
1584 on load and link times of a DSO as it massively reduces the size of the
1585 dynamic export table when the library makes heavy use of templates. While
1586 it can cause bloating through duplication of code within each DSO where
1587 it is used, often the wastage is less than the considerable space occupied
1588 by a long symbol name in the export table which is typical when using
1589 templates and namespaces. For even more savings, combine with the
1590 @option{-fvisibility=hidden} switch.
1594 Do not use weak symbol support, even if it is provided by the linker.
1595 By default, G++ will use weak symbols if they are available. This
1596 option exists only for testing, and should not be used by end-users;
1597 it will result in inferior code and has no benefits. This option may
1598 be removed in a future release of G++.
1602 Do not search for header files in the standard directories specific to
1603 C++, but do still search the other standard directories. (This option
1604 is used when building the C++ library.)
1607 In addition, these optimization, warning, and code generation options
1608 have meanings only for C++ programs:
1611 @item -fno-default-inline
1612 @opindex fno-default-inline
1613 Do not assume @samp{inline} for functions defined inside a class scope.
1614 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1615 functions will have linkage like inline functions; they just won't be
1618 @item -Wabi @r{(C++ only)}
1620 Warn when G++ generates code that is probably not compatible with the
1621 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1622 all such cases, there are probably some cases that are not warned about,
1623 even though G++ is generating incompatible code. There may also be
1624 cases where warnings are emitted even though the code that is generated
1627 You should rewrite your code to avoid these warnings if you are
1628 concerned about the fact that code generated by G++ may not be binary
1629 compatible with code generated by other compilers.
1631 The known incompatibilities at this point include:
1636 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1637 pack data into the same byte as a base class. For example:
1640 struct A @{ virtual void f(); int f1 : 1; @};
1641 struct B : public A @{ int f2 : 1; @};
1645 In this case, G++ will place @code{B::f2} into the same byte
1646 as@code{A::f1}; other compilers will not. You can avoid this problem
1647 by explicitly padding @code{A} so that its size is a multiple of the
1648 byte size on your platform; that will cause G++ and other compilers to
1649 layout @code{B} identically.
1652 Incorrect handling of tail-padding for virtual bases. G++ does not use
1653 tail padding when laying out virtual bases. For example:
1656 struct A @{ virtual void f(); char c1; @};
1657 struct B @{ B(); char c2; @};
1658 struct C : public A, public virtual B @{@};
1662 In this case, G++ will not place @code{B} into the tail-padding for
1663 @code{A}; other compilers will. You can avoid this problem by
1664 explicitly padding @code{A} so that its size is a multiple of its
1665 alignment (ignoring virtual base classes); that will cause G++ and other
1666 compilers to layout @code{C} identically.
1669 Incorrect handling of bit-fields with declared widths greater than that
1670 of their underlying types, when the bit-fields appear in a union. For
1674 union U @{ int i : 4096; @};
1678 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1679 union too small by the number of bits in an @code{int}.
1682 Empty classes can be placed at incorrect offsets. For example:
1692 struct C : public B, public A @{@};
1696 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1697 it should be placed at offset zero. G++ mistakenly believes that the
1698 @code{A} data member of @code{B} is already at offset zero.
1701 Names of template functions whose types involve @code{typename} or
1702 template template parameters can be mangled incorrectly.
1705 template <typename Q>
1706 void f(typename Q::X) @{@}
1708 template <template <typename> class Q>
1709 void f(typename Q<int>::X) @{@}
1713 Instantiations of these templates may be mangled incorrectly.
1717 @item -Wctor-dtor-privacy @r{(C++ only)}
1718 @opindex Wctor-dtor-privacy
1719 Warn when a class seems unusable because all the constructors or
1720 destructors in that class are private, and it has neither friends nor
1721 public static member functions.
1723 @item -Wnon-virtual-dtor @r{(C++ only)}
1724 @opindex Wnon-virtual-dtor
1725 Warn when a class appears to be polymorphic, thereby requiring a virtual
1726 destructor, yet it declares a non-virtual one.
1727 This warning is enabled by @option{-Wall}.
1729 @item -Wreorder @r{(C++ only)}
1731 @cindex reordering, warning
1732 @cindex warning for reordering of member initializers
1733 Warn when the order of member initializers given in the code does not
1734 match the order in which they must be executed. For instance:
1740 A(): j (0), i (1) @{ @}
1744 The compiler will rearrange the member initializers for @samp{i}
1745 and @samp{j} to match the declaration order of the members, emitting
1746 a warning to that effect. This warning is enabled by @option{-Wall}.
1749 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1752 @item -Weffc++ @r{(C++ only)}
1754 Warn about violations of the following style guidelines from Scott Meyers'
1755 @cite{Effective C++} book:
1759 Item 11: Define a copy constructor and an assignment operator for classes
1760 with dynamically allocated memory.
1763 Item 12: Prefer initialization to assignment in constructors.
1766 Item 14: Make destructors virtual in base classes.
1769 Item 15: Have @code{operator=} return a reference to @code{*this}.
1772 Item 23: Don't try to return a reference when you must return an object.
1776 Also warn about violations of the following style guidelines from
1777 Scott Meyers' @cite{More Effective C++} book:
1781 Item 6: Distinguish between prefix and postfix forms of increment and
1782 decrement operators.
1785 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1789 When selecting this option, be aware that the standard library
1790 headers do not obey all of these guidelines; use @samp{grep -v}
1791 to filter out those warnings.
1793 @item -Wno-deprecated @r{(C++ only)}
1794 @opindex Wno-deprecated
1795 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1797 @item -Wstrict-null-sentinel @r{(C++ only)}
1798 @opindex Wstrict-null-sentinel
1799 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1800 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1801 to @code{__null}. Although it is a null pointer constant not a null pointer,
1802 it is guaranteed to of the same size as a pointer. But this use is
1803 not portable across different compilers.
1805 @item -Wno-non-template-friend @r{(C++ only)}
1806 @opindex Wno-non-template-friend
1807 Disable warnings when non-templatized friend functions are declared
1808 within a template. Since the advent of explicit template specification
1809 support in G++, if the name of the friend is an unqualified-id (i.e.,
1810 @samp{friend foo(int)}), the C++ language specification demands that the
1811 friend declare or define an ordinary, nontemplate function. (Section
1812 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1813 could be interpreted as a particular specialization of a templatized
1814 function. Because this non-conforming behavior is no longer the default
1815 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1816 check existing code for potential trouble spots and is on by default.
1817 This new compiler behavior can be turned off with
1818 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1819 but disables the helpful warning.
1821 @item -Wold-style-cast @r{(C++ only)}
1822 @opindex Wold-style-cast
1823 Warn if an old-style (C-style) cast to a non-void type is used within
1824 a C++ program. The new-style casts (@samp{dynamic_cast},
1825 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1826 less vulnerable to unintended effects and much easier to search for.
1828 @item -Woverloaded-virtual @r{(C++ only)}
1829 @opindex Woverloaded-virtual
1830 @cindex overloaded virtual fn, warning
1831 @cindex warning for overloaded virtual fn
1832 Warn when a function declaration hides virtual functions from a
1833 base class. For example, in:
1840 struct B: public A @{
1845 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1853 will fail to compile.
1855 @item -Wno-pmf-conversions @r{(C++ only)}
1856 @opindex Wno-pmf-conversions
1857 Disable the diagnostic for converting a bound pointer to member function
1860 @item -Wsign-promo @r{(C++ only)}
1861 @opindex Wsign-promo
1862 Warn when overload resolution chooses a promotion from unsigned or
1863 enumerated type to a signed type, over a conversion to an unsigned type of
1864 the same size. Previous versions of G++ would try to preserve
1865 unsignedness, but the standard mandates the current behavior.
1870 A& operator = (int);
1880 In this example, G++ will synthesize a default @samp{A& operator =
1881 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1884 @node Objective-C and Objective-C++ Dialect Options
1885 @section Options Controlling Objective-C and Objective-C++ Dialects
1887 @cindex compiler options, Objective-C and Objective-C++
1888 @cindex Objective-C and Objective-C++ options, command line
1889 @cindex options, Objective-C and Objective-C++
1890 (NOTE: This manual does not describe the Objective-C and Objective-C++
1891 languages themselves. See @xref{Standards,,Language Standards
1892 Supported by GCC}, for references.)
1894 This section describes the command-line options that are only meaningful
1895 for Objective-C and Objective-C++ programs, but you can also use most of
1896 the language-independent GNU compiler options.
1897 For example, you might compile a file @code{some_class.m} like this:
1900 gcc -g -fgnu-runtime -O -c some_class.m
1904 In this example, @option{-fgnu-runtime} is an option meant only for
1905 Objective-C and Objective-C++ programs; you can use the other options with
1906 any language supported by GCC@.
1908 Note that since Objective-C is an extension of the C language, Objective-C
1909 compilations may also use options specific to the C front-end (e.g.,
1910 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1911 C++-specific options (e.g., @option{-Wabi}).
1913 Here is a list of options that are @emph{only} for compiling Objective-C
1914 and Objective-C++ programs:
1917 @item -fconstant-string-class=@var{class-name}
1918 @opindex fconstant-string-class
1919 Use @var{class-name} as the name of the class to instantiate for each
1920 literal string specified with the syntax @code{@@"@dots{}"}. The default
1921 class name is @code{NXConstantString} if the GNU runtime is being used, and
1922 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1923 @option{-fconstant-cfstrings} option, if also present, will override the
1924 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1925 to be laid out as constant CoreFoundation strings.
1928 @opindex fgnu-runtime
1929 Generate object code compatible with the standard GNU Objective-C
1930 runtime. This is the default for most types of systems.
1932 @item -fnext-runtime
1933 @opindex fnext-runtime
1934 Generate output compatible with the NeXT runtime. This is the default
1935 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1936 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1939 @item -fno-nil-receivers
1940 @opindex fno-nil-receivers
1941 Assume that all Objective-C message dispatches (e.g.,
1942 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1943 is not @code{nil}. This allows for more efficient entry points in the runtime
1944 to be used. Currently, this option is only available in conjunction with
1945 the NeXT runtime on Mac OS X 10.3 and later.
1947 @item -fobjc-call-cxx-cdtors
1948 @opindex fobjc-call-cxx-cdtors
1949 For each Objective-C class, check if any of its instance variables is a
1950 C++ object with a non-trivial default constructor. If so, synthesize a
1951 special @code{- (id) .cxx_construct} instance method that will run
1952 non-trivial default constructors on any such instance variables, in order,
1953 and then return @code{self}. Similarly, check if any instance variable
1954 is a C++ object with a non-trivial destructor, and if so, synthesize a
1955 special @code{- (void) .cxx_destruct} method that will run
1956 all such default destructors, in reverse order.
1958 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
1959 thusly generated will only operate on instance variables declared in the
1960 current Objective-C class, and not those inherited from superclasses. It
1961 is the responsibility of the Objective-C runtime to invoke all such methods
1962 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
1963 will be invoked by the runtime immediately after a new object
1964 instance is allocated; the @code{- (void) .cxx_destruct} methods will
1965 be invoked immediately before the runtime deallocates an object instance.
1967 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
1968 support for invoking the @code{- (id) .cxx_construct} and
1969 @code{- (void) .cxx_destruct} methods.
1971 @item -fobjc-direct-dispatch
1972 @opindex fobjc-direct-dispatch
1973 Allow fast jumps to the message dispatcher. On Darwin this is
1974 accomplished via the comm page.
1976 @item -fobjc-exceptions
1977 @opindex fobjc-exceptions
1978 Enable syntactic support for structured exception handling in Objective-C,
1979 similar to what is offered by C++ and Java. Currently, this option is only
1980 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1988 @@catch (AnObjCClass *exc) @{
1995 @@catch (AnotherClass *exc) @{
1998 @@catch (id allOthers) @{
2008 The @code{@@throw} statement may appear anywhere in an Objective-C or
2009 Objective-C++ program; when used inside of a @code{@@catch} block, the
2010 @code{@@throw} may appear without an argument (as shown above), in which case
2011 the object caught by the @code{@@catch} will be rethrown.
2013 Note that only (pointers to) Objective-C objects may be thrown and
2014 caught using this scheme. When an object is thrown, it will be caught
2015 by the nearest @code{@@catch} clause capable of handling objects of that type,
2016 analogously to how @code{catch} blocks work in C++ and Java. A
2017 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2018 any and all Objective-C exceptions not caught by previous @code{@@catch}
2021 The @code{@@finally} clause, if present, will be executed upon exit from the
2022 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2023 regardless of whether any exceptions are thrown, caught or rethrown
2024 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2025 of the @code{finally} clause in Java.
2027 There are several caveats to using the new exception mechanism:
2031 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2032 idioms provided by the @code{NSException} class, the new
2033 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2034 systems, due to additional functionality needed in the (NeXT) Objective-C
2038 As mentioned above, the new exceptions do not support handling
2039 types other than Objective-C objects. Furthermore, when used from
2040 Objective-C++, the Objective-C exception model does not interoperate with C++
2041 exceptions at this time. This means you cannot @code{@@throw} an exception
2042 from Objective-C and @code{catch} it in C++, or vice versa
2043 (i.e., @code{throw @dots{} @@catch}).
2046 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2047 blocks for thread-safe execution:
2050 @@synchronized (ObjCClass *guard) @{
2055 Upon entering the @code{@@synchronized} block, a thread of execution shall
2056 first check whether a lock has been placed on the corresponding @code{guard}
2057 object by another thread. If it has, the current thread shall wait until
2058 the other thread relinquishes its lock. Once @code{guard} becomes available,
2059 the current thread will place its own lock on it, execute the code contained in
2060 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2061 making @code{guard} available to other threads).
2063 Unlike Java, Objective-C does not allow for entire methods to be marked
2064 @code{@@synchronized}. Note that throwing exceptions out of
2065 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2066 to be unlocked properly.
2070 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2072 @item -freplace-objc-classes
2073 @opindex freplace-objc-classes
2074 Emit a special marker instructing @command{ld(1)} not to statically link in
2075 the resulting object file, and allow @command{dyld(1)} to load it in at
2076 run time instead. This is used in conjunction with the Fix-and-Continue
2077 debugging mode, where the object file in question may be recompiled and
2078 dynamically reloaded in the course of program execution, without the need
2079 to restart the program itself. Currently, Fix-and-Continue functionality
2080 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2085 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2086 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2087 compile time) with static class references that get initialized at load time,
2088 which improves run-time performance. Specifying the @option{-fzero-link} flag
2089 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2090 to be retained. This is useful in Zero-Link debugging mode, since it allows
2091 for individual class implementations to be modified during program execution.
2095 Dump interface declarations for all classes seen in the source file to a
2096 file named @file{@var{sourcename}.decl}.
2098 @item -Wassign-intercept
2099 @opindex Wassign-intercept
2100 Warn whenever an Objective-C assignment is being intercepted by the
2104 @opindex Wno-protocol
2105 If a class is declared to implement a protocol, a warning is issued for
2106 every method in the protocol that is not implemented by the class. The
2107 default behavior is to issue a warning for every method not explicitly
2108 implemented in the class, even if a method implementation is inherited
2109 from the superclass. If you use the @option{-Wno-protocol} option, then
2110 methods inherited from the superclass are considered to be implemented,
2111 and no warning is issued for them.
2115 Warn if multiple methods of different types for the same selector are
2116 found during compilation. The check is performed on the list of methods
2117 in the final stage of compilation. Additionally, a check is performed
2118 for each selector appearing in a @code{@@selector(@dots{})}
2119 expression, and a corresponding method for that selector has been found
2120 during compilation. Because these checks scan the method table only at
2121 the end of compilation, these warnings are not produced if the final
2122 stage of compilation is not reached, for example because an error is
2123 found during compilation, or because the @option{-fsyntax-only} option is
2126 @item -Wstrict-selector-match
2127 @opindex Wstrict-selector-match
2128 Warn if multiple methods with differing argument and/or return types are
2129 found for a given selector when attempting to send a message using this
2130 selector to a receiver of type @code{id} or @code{Class}. When this flag
2131 is off (which is the default behavior), the compiler will omit such warnings
2132 if any differences found are confined to types which share the same size
2135 @item -Wundeclared-selector
2136 @opindex Wundeclared-selector
2137 Warn if a @code{@@selector(@dots{})} expression referring to an
2138 undeclared selector is found. A selector is considered undeclared if no
2139 method with that name has been declared before the
2140 @code{@@selector(@dots{})} expression, either explicitly in an
2141 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2142 an @code{@@implementation} section. This option always performs its
2143 checks as soon as a @code{@@selector(@dots{})} expression is found,
2144 while @option{-Wselector} only performs its checks in the final stage of
2145 compilation. This also enforces the coding style convention
2146 that methods and selectors must be declared before being used.
2148 @item -print-objc-runtime-info
2149 @opindex print-objc-runtime-info
2150 Generate C header describing the largest structure that is passed by
2155 @node Language Independent Options
2156 @section Options to Control Diagnostic Messages Formatting
2157 @cindex options to control diagnostics formatting
2158 @cindex diagnostic messages
2159 @cindex message formatting
2161 Traditionally, diagnostic messages have been formatted irrespective of
2162 the output device's aspect (e.g.@: its width, @dots{}). The options described
2163 below can be used to control the diagnostic messages formatting
2164 algorithm, e.g.@: how many characters per line, how often source location
2165 information should be reported. Right now, only the C++ front end can
2166 honor these options. However it is expected, in the near future, that
2167 the remaining front ends would be able to digest them correctly.
2170 @item -fmessage-length=@var{n}
2171 @opindex fmessage-length
2172 Try to format error messages so that they fit on lines of about @var{n}
2173 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2174 the front ends supported by GCC@. If @var{n} is zero, then no
2175 line-wrapping will be done; each error message will appear on a single
2178 @opindex fdiagnostics-show-location
2179 @item -fdiagnostics-show-location=once
2180 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2181 reporter to emit @emph{once} source location information; that is, in
2182 case the message is too long to fit on a single physical line and has to
2183 be wrapped, the source location won't be emitted (as prefix) again,
2184 over and over, in subsequent continuation lines. This is the default
2187 @item -fdiagnostics-show-location=every-line
2188 Only meaningful in line-wrapping mode. Instructs the diagnostic
2189 messages reporter to emit the same source location information (as
2190 prefix) for physical lines that result from the process of breaking
2191 a message which is too long to fit on a single line.
2193 @item -fdiagnostics-show-options
2194 @opindex fdiagnostics-show-options
2195 This option instructs the diagnostic machinery to add text to each
2196 diagnostic emitted, which indicates which command line option directly
2197 controls that diagnostic, when such an option is known to the
2198 diagnostic machinery.
2202 @node Warning Options
2203 @section Options to Request or Suppress Warnings
2204 @cindex options to control warnings
2205 @cindex warning messages
2206 @cindex messages, warning
2207 @cindex suppressing warnings
2209 Warnings are diagnostic messages that report constructions which
2210 are not inherently erroneous but which are risky or suggest there
2211 may have been an error.
2213 You can request many specific warnings with options beginning @samp{-W},
2214 for example @option{-Wimplicit} to request warnings on implicit
2215 declarations. Each of these specific warning options also has a
2216 negative form beginning @samp{-Wno-} to turn off warnings;
2217 for example, @option{-Wno-implicit}. This manual lists only one of the
2218 two forms, whichever is not the default.
2220 The following options control the amount and kinds of warnings produced
2221 by GCC; for further, language-specific options also refer to
2222 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2226 @cindex syntax checking
2228 @opindex fsyntax-only
2229 Check the code for syntax errors, but don't do anything beyond that.
2233 Issue all the warnings demanded by strict ISO C and ISO C++;
2234 reject all programs that use forbidden extensions, and some other
2235 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2236 version of the ISO C standard specified by any @option{-std} option used.
2238 Valid ISO C and ISO C++ programs should compile properly with or without
2239 this option (though a rare few will require @option{-ansi} or a
2240 @option{-std} option specifying the required version of ISO C)@. However,
2241 without this option, certain GNU extensions and traditional C and C++
2242 features are supported as well. With this option, they are rejected.
2244 @option{-pedantic} does not cause warning messages for use of the
2245 alternate keywords whose names begin and end with @samp{__}. Pedantic
2246 warnings are also disabled in the expression that follows
2247 @code{__extension__}. However, only system header files should use
2248 these escape routes; application programs should avoid them.
2249 @xref{Alternate Keywords}.
2251 Some users try to use @option{-pedantic} to check programs for strict ISO
2252 C conformance. They soon find that it does not do quite what they want:
2253 it finds some non-ISO practices, but not all---only those for which
2254 ISO C @emph{requires} a diagnostic, and some others for which
2255 diagnostics have been added.
2257 A feature to report any failure to conform to ISO C might be useful in
2258 some instances, but would require considerable additional work and would
2259 be quite different from @option{-pedantic}. We don't have plans to
2260 support such a feature in the near future.
2262 Where the standard specified with @option{-std} represents a GNU
2263 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2264 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2265 extended dialect is based. Warnings from @option{-pedantic} are given
2266 where they are required by the base standard. (It would not make sense
2267 for such warnings to be given only for features not in the specified GNU
2268 C dialect, since by definition the GNU dialects of C include all
2269 features the compiler supports with the given option, and there would be
2270 nothing to warn about.)
2272 @item -pedantic-errors
2273 @opindex pedantic-errors
2274 Like @option{-pedantic}, except that errors are produced rather than
2279 Inhibit all warning messages.
2283 Inhibit warning messages about the use of @samp{#import}.
2285 @item -Wchar-subscripts
2286 @opindex Wchar-subscripts
2287 Warn if an array subscript has type @code{char}. This is a common cause
2288 of error, as programmers often forget that this type is signed on some
2290 This warning is enabled by @option{-Wall}.
2294 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2295 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2296 This warning is enabled by @option{-Wall}.
2298 @item -Wfatal-errors
2299 @opindex Wfatal-errors
2300 This option causes the compiler to abort compilation on the first error
2301 occurred rather than trying to keep going and printing further error
2306 @opindex ffreestanding
2307 @opindex fno-builtin
2308 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2309 the arguments supplied have types appropriate to the format string
2310 specified, and that the conversions specified in the format string make
2311 sense. This includes standard functions, and others specified by format
2312 attributes (@pxref{Function Attributes}), in the @code{printf},
2313 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2314 not in the C standard) families (or other target-specific families).
2315 Which functions are checked without format attributes having been
2316 specified depends on the standard version selected, and such checks of
2317 functions without the attribute specified are disabled by
2318 @option{-ffreestanding} or @option{-fno-builtin}.
2320 The formats are checked against the format features supported by GNU
2321 libc version 2.2. These include all ISO C90 and C99 features, as well
2322 as features from the Single Unix Specification and some BSD and GNU
2323 extensions. Other library implementations may not support all these
2324 features; GCC does not support warning about features that go beyond a
2325 particular library's limitations. However, if @option{-pedantic} is used
2326 with @option{-Wformat}, warnings will be given about format features not
2327 in the selected standard version (but not for @code{strfmon} formats,
2328 since those are not in any version of the C standard). @xref{C Dialect
2329 Options,,Options Controlling C Dialect}.
2331 Since @option{-Wformat} also checks for null format arguments for
2332 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2334 @option{-Wformat} is included in @option{-Wall}. For more control over some
2335 aspects of format checking, the options @option{-Wformat-y2k},
2336 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2337 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2338 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2341 @opindex Wformat-y2k
2342 If @option{-Wformat} is specified, also warn about @code{strftime}
2343 formats which may yield only a two-digit year.
2345 @item -Wno-format-extra-args
2346 @opindex Wno-format-extra-args
2347 If @option{-Wformat} is specified, do not warn about excess arguments to a
2348 @code{printf} or @code{scanf} format function. The C standard specifies
2349 that such arguments are ignored.
2351 Where the unused arguments lie between used arguments that are
2352 specified with @samp{$} operand number specifications, normally
2353 warnings are still given, since the implementation could not know what
2354 type to pass to @code{va_arg} to skip the unused arguments. However,
2355 in the case of @code{scanf} formats, this option will suppress the
2356 warning if the unused arguments are all pointers, since the Single
2357 Unix Specification says that such unused arguments are allowed.
2359 @item -Wno-format-zero-length
2360 @opindex Wno-format-zero-length
2361 If @option{-Wformat} is specified, do not warn about zero-length formats.
2362 The C standard specifies that zero-length formats are allowed.
2364 @item -Wformat-nonliteral
2365 @opindex Wformat-nonliteral
2366 If @option{-Wformat} is specified, also warn if the format string is not a
2367 string literal and so cannot be checked, unless the format function
2368 takes its format arguments as a @code{va_list}.
2370 @item -Wformat-security
2371 @opindex Wformat-security
2372 If @option{-Wformat} is specified, also warn about uses of format
2373 functions that represent possible security problems. At present, this
2374 warns about calls to @code{printf} and @code{scanf} functions where the
2375 format string is not a string literal and there are no format arguments,
2376 as in @code{printf (foo);}. This may be a security hole if the format
2377 string came from untrusted input and contains @samp{%n}. (This is
2378 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2379 in future warnings may be added to @option{-Wformat-security} that are not
2380 included in @option{-Wformat-nonliteral}.)
2384 Enable @option{-Wformat} plus format checks not included in
2385 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2386 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2390 Warn about passing a null pointer for arguments marked as
2391 requiring a non-null value by the @code{nonnull} function attribute.
2393 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2394 can be disabled with the @option{-Wno-nonnull} option.
2396 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2398 Warn about uninitialized variables which are initialized with themselves.
2399 Note this option can only be used with the @option{-Wuninitialized} option,
2400 which in turn only works with @option{-O1} and above.
2402 For example, GCC will warn about @code{i} being uninitialized in the
2403 following snippet only when @option{-Winit-self} has been specified:
2414 @item -Wimplicit-int
2415 @opindex Wimplicit-int
2416 Warn when a declaration does not specify a type.
2417 This warning is enabled by @option{-Wall}.
2419 @item -Wimplicit-function-declaration
2420 @itemx -Werror-implicit-function-declaration
2421 @opindex Wimplicit-function-declaration
2422 @opindex Werror-implicit-function-declaration
2423 Give a warning (or error) whenever a function is used before being
2424 declared. The form @option{-Wno-error-implicit-function-declaration}
2426 This warning is enabled by @option{-Wall} (as a warning, not an error).
2430 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2431 This warning is enabled by @option{-Wall}.
2435 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2436 function with external linkage, returning int, taking either zero
2437 arguments, two, or three arguments of appropriate types.
2438 This warning is enabled by @option{-Wall}.
2440 @item -Wmissing-braces
2441 @opindex Wmissing-braces
2442 Warn if an aggregate or union initializer is not fully bracketed. In
2443 the following example, the initializer for @samp{a} is not fully
2444 bracketed, but that for @samp{b} is fully bracketed.
2447 int a[2][2] = @{ 0, 1, 2, 3 @};
2448 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2451 This warning is enabled by @option{-Wall}.
2453 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2454 @opindex Wmissing-include-dirs
2455 Warn if a user-supplied include directory does not exist.
2458 @opindex Wparentheses
2459 Warn if parentheses are omitted in certain contexts, such
2460 as when there is an assignment in a context where a truth value
2461 is expected, or when operators are nested whose precedence people
2462 often get confused about. Only the warning for an assignment used as
2463 a truth value is supported when compiling C++; the other warnings are
2464 only supported when compiling C@.
2466 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2467 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2468 interpretation from that of ordinary mathematical notation.
2470 Also warn about constructions where there may be confusion to which
2471 @code{if} statement an @code{else} branch belongs. Here is an example of
2486 In C, every @code{else} branch belongs to the innermost possible @code{if}
2487 statement, which in this example is @code{if (b)}. This is often not
2488 what the programmer expected, as illustrated in the above example by
2489 indentation the programmer chose. When there is the potential for this
2490 confusion, GCC will issue a warning when this flag is specified.
2491 To eliminate the warning, add explicit braces around the innermost
2492 @code{if} statement so there is no way the @code{else} could belong to
2493 the enclosing @code{if}. The resulting code would look like this:
2509 This warning is enabled by @option{-Wall}.
2511 @item -Wsequence-point
2512 @opindex Wsequence-point
2513 Warn about code that may have undefined semantics because of violations
2514 of sequence point rules in the C standard.
2516 The C standard defines the order in which expressions in a C program are
2517 evaluated in terms of @dfn{sequence points}, which represent a partial
2518 ordering between the execution of parts of the program: those executed
2519 before the sequence point, and those executed after it. These occur
2520 after the evaluation of a full expression (one which is not part of a
2521 larger expression), after the evaluation of the first operand of a
2522 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2523 function is called (but after the evaluation of its arguments and the
2524 expression denoting the called function), and in certain other places.
2525 Other than as expressed by the sequence point rules, the order of
2526 evaluation of subexpressions of an expression is not specified. All
2527 these rules describe only a partial order rather than a total order,
2528 since, for example, if two functions are called within one expression
2529 with no sequence point between them, the order in which the functions
2530 are called is not specified. However, the standards committee have
2531 ruled that function calls do not overlap.
2533 It is not specified when between sequence points modifications to the
2534 values of objects take effect. Programs whose behavior depends on this
2535 have undefined behavior; the C standard specifies that ``Between the
2536 previous and next sequence point an object shall have its stored value
2537 modified at most once by the evaluation of an expression. Furthermore,
2538 the prior value shall be read only to determine the value to be
2539 stored.''. If a program breaks these rules, the results on any
2540 particular implementation are entirely unpredictable.
2542 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2543 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2544 diagnosed by this option, and it may give an occasional false positive
2545 result, but in general it has been found fairly effective at detecting
2546 this sort of problem in programs.
2548 The present implementation of this option only works for C programs. A
2549 future implementation may also work for C++ programs.
2551 The C standard is worded confusingly, therefore there is some debate
2552 over the precise meaning of the sequence point rules in subtle cases.
2553 Links to discussions of the problem, including proposed formal
2554 definitions, may be found on the GCC readings page, at
2555 @w{@uref{http://gcc.gnu.org/readings.html}}.
2557 This warning is enabled by @option{-Wall}.
2560 @opindex Wreturn-type
2561 Warn whenever a function is defined with a return-type that defaults to
2562 @code{int}. Also warn about any @code{return} statement with no
2563 return-value in a function whose return-type is not @code{void}.
2565 For C, also warn if the return type of a function has a type qualifier
2566 such as @code{const}. Such a type qualifier has no effect, since the
2567 value returned by a function is not an lvalue. ISO C prohibits
2568 qualified @code{void} return types on function definitions, so such
2569 return types always receive a warning even without this option.
2571 For C++, a function without return type always produces a diagnostic
2572 message, even when @option{-Wno-return-type} is specified. The only
2573 exceptions are @samp{main} and functions defined in system headers.
2575 This warning is enabled by @option{-Wall}.
2579 Warn whenever a @code{switch} statement has an index of enumerated type
2580 and lacks a @code{case} for one or more of the named codes of that
2581 enumeration. (The presence of a @code{default} label prevents this
2582 warning.) @code{case} labels outside the enumeration range also
2583 provoke warnings when this option is used.
2584 This warning is enabled by @option{-Wall}.
2586 @item -Wswitch-default
2587 @opindex Wswitch-switch
2588 Warn whenever a @code{switch} statement does not have a @code{default}
2592 @opindex Wswitch-enum
2593 Warn whenever a @code{switch} statement has an index of enumerated type
2594 and lacks a @code{case} for one or more of the named codes of that
2595 enumeration. @code{case} labels outside the enumeration range also
2596 provoke warnings when this option is used.
2600 Warn if any trigraphs are encountered that might change the meaning of
2601 the program (trigraphs within comments are not warned about).
2602 This warning is enabled by @option{-Wall}.
2604 @item -Wunused-function
2605 @opindex Wunused-function
2606 Warn whenever a static function is declared but not defined or a
2607 non\-inline static function is unused.
2608 This warning is enabled by @option{-Wall}.
2610 @item -Wunused-label
2611 @opindex Wunused-label
2612 Warn whenever a label is declared but not used.
2613 This warning is enabled by @option{-Wall}.
2615 To suppress this warning use the @samp{unused} attribute
2616 (@pxref{Variable Attributes}).
2618 @item -Wunused-parameter
2619 @opindex Wunused-parameter
2620 Warn whenever a function parameter is unused aside from its declaration.
2622 To suppress this warning use the @samp{unused} attribute
2623 (@pxref{Variable Attributes}).
2625 @item -Wunused-variable
2626 @opindex Wunused-variable
2627 Warn whenever a local variable or non-constant static variable is unused
2628 aside from its declaration
2629 This warning is enabled by @option{-Wall}.
2631 To suppress this warning use the @samp{unused} attribute
2632 (@pxref{Variable Attributes}).
2634 @item -Wunused-value
2635 @opindex Wunused-value
2636 Warn whenever a statement computes a result that is explicitly not used.
2637 This warning is enabled by @option{-Wall}.
2639 To suppress this warning cast the expression to @samp{void}.
2643 All the above @option{-Wunused} options combined.
2645 In order to get a warning about an unused function parameter, you must
2646 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2647 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2649 @item -Wuninitialized
2650 @opindex Wuninitialized
2651 Warn if an automatic variable is used without first being initialized or
2652 if a variable may be clobbered by a @code{setjmp} call.
2654 These warnings are possible only in optimizing compilation,
2655 because they require data flow information that is computed only
2656 when optimizing. If you don't specify @option{-O}, you simply won't
2659 If you want to warn about code which uses the uninitialized value of the
2660 variable in its own initializer, use the @option{-Winit-self} option.
2662 These warnings occur for individual uninitialized or clobbered
2663 elements of structure, union or array variables as well as for
2664 variables which are uninitialized or clobbered as a whole. They do
2665 not occur for variables or elements declared @code{volatile}. Because
2666 these warnings depend on optimization, the exact variables or elements
2667 for which there are warnings will depend on the precise optimization
2668 options and version of GCC used.
2670 Note that there may be no warning about a variable that is used only
2671 to compute a value that itself is never used, because such
2672 computations may be deleted by data flow analysis before the warnings
2675 These warnings are made optional because GCC is not smart
2676 enough to see all the reasons why the code might be correct
2677 despite appearing to have an error. Here is one example of how
2698 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2699 always initialized, but GCC doesn't know this. Here is
2700 another common case:
2705 if (change_y) save_y = y, y = new_y;
2707 if (change_y) y = save_y;
2712 This has no bug because @code{save_y} is used only if it is set.
2714 @cindex @code{longjmp} warnings
2715 This option also warns when a non-volatile automatic variable might be
2716 changed by a call to @code{longjmp}. These warnings as well are possible
2717 only in optimizing compilation.
2719 The compiler sees only the calls to @code{setjmp}. It cannot know
2720 where @code{longjmp} will be called; in fact, a signal handler could
2721 call it at any point in the code. As a result, you may get a warning
2722 even when there is in fact no problem because @code{longjmp} cannot
2723 in fact be called at the place which would cause a problem.
2725 Some spurious warnings can be avoided if you declare all the functions
2726 you use that never return as @code{noreturn}. @xref{Function
2729 This warning is enabled by @option{-Wall}.
2731 @item -Wunknown-pragmas
2732 @opindex Wunknown-pragmas
2733 @cindex warning for unknown pragmas
2734 @cindex unknown pragmas, warning
2735 @cindex pragmas, warning of unknown
2736 Warn when a #pragma directive is encountered which is not understood by
2737 GCC@. If this command line option is used, warnings will even be issued
2738 for unknown pragmas in system header files. This is not the case if
2739 the warnings were only enabled by the @option{-Wall} command line option.
2742 @opindex Wno-pragmas
2744 Do not warn about misuses of pragmas, such as incorrect parameters,
2745 invalid syntax, or conflicts between pragmas. See also
2746 @samp{-Wunknown-pragmas}.
2748 @item -Wstrict-aliasing
2749 @opindex Wstrict-aliasing
2750 This option is only active when @option{-fstrict-aliasing} is active.
2751 It warns about code which might break the strict aliasing rules that the
2752 compiler is using for optimization. The warning does not catch all
2753 cases, but does attempt to catch the more common pitfalls. It is
2754 included in @option{-Wall}.
2756 @item -Wstrict-aliasing=2
2757 @opindex Wstrict-aliasing=2
2758 This option is only active when @option{-fstrict-aliasing} is active.
2759 It warns about code which might break the strict aliasing rules that the
2760 compiler is using for optimization. This warning catches more cases than
2761 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2762 cases that are safe.
2766 All of the above @samp{-W} options combined. This enables all the
2767 warnings about constructions that some users consider questionable, and
2768 that are easy to avoid (or modify to prevent the warning), even in
2769 conjunction with macros. This also enables some language-specific
2770 warnings described in @ref{C++ Dialect Options} and
2771 @ref{Objective-C and Objective-C++ Dialect Options}.
2774 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2775 Some of them warn about constructions that users generally do not
2776 consider questionable, but which occasionally you might wish to check
2777 for; others warn about constructions that are necessary or hard to avoid
2778 in some cases, and there is no simple way to modify the code to suppress
2785 (This option used to be called @option{-W}. The older name is still
2786 supported, but the newer name is more descriptive.) Print extra warning
2787 messages for these events:
2791 A function can return either with or without a value. (Falling
2792 off the end of the function body is considered returning without
2793 a value.) For example, this function would evoke such a
2807 An expression-statement or the left-hand side of a comma expression
2808 contains no side effects.
2809 To suppress the warning, cast the unused expression to void.
2810 For example, an expression such as @samp{x[i,j]} will cause a warning,
2811 but @samp{x[(void)i,j]} will not.
2814 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2817 Storage-class specifiers like @code{static} are not the first things in
2818 a declaration. According to the C Standard, this usage is obsolescent.
2821 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2825 A comparison between signed and unsigned values could produce an
2826 incorrect result when the signed value is converted to unsigned.
2827 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2830 An aggregate has an initializer which does not initialize all members.
2831 This warning can be independently controlled by
2832 @option{-Wmissing-field-initializers}.
2835 A function parameter is declared without a type specifier in K&R-style
2843 An empty body occurs in an @samp{if} or @samp{else} statement.
2846 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2847 @samp{>}, or @samp{>=}.
2850 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2853 Any of several floating-point events that often indicate errors, such as
2854 overflow, underflow, loss of precision, etc.
2856 @item @r{(C++ only)}
2857 An enumerator and a non-enumerator both appear in a conditional expression.
2859 @item @r{(C++ only)}
2860 A non-static reference or non-static @samp{const} member appears in a
2861 class without constructors.
2863 @item @r{(C++ only)}
2864 Ambiguous virtual bases.
2866 @item @r{(C++ only)}
2867 Subscripting an array which has been declared @samp{register}.
2869 @item @r{(C++ only)}
2870 Taking the address of a variable which has been declared @samp{register}.
2872 @item @r{(C++ only)}
2873 A base class is not initialized in a derived class' copy constructor.
2876 @item -Wno-div-by-zero
2877 @opindex Wno-div-by-zero
2878 @opindex Wdiv-by-zero
2879 Do not warn about compile-time integer division by zero. Floating point
2880 division by zero is not warned about, as it can be a legitimate way of
2881 obtaining infinities and NaNs.
2883 @item -Wsystem-headers
2884 @opindex Wsystem-headers
2885 @cindex warnings from system headers
2886 @cindex system headers, warnings from
2887 Print warning messages for constructs found in system header files.
2888 Warnings from system headers are normally suppressed, on the assumption
2889 that they usually do not indicate real problems and would only make the
2890 compiler output harder to read. Using this command line option tells
2891 GCC to emit warnings from system headers as if they occurred in user
2892 code. However, note that using @option{-Wall} in conjunction with this
2893 option will @emph{not} warn about unknown pragmas in system
2894 headers---for that, @option{-Wunknown-pragmas} must also be used.
2897 @opindex Wfloat-equal
2898 Warn if floating point values are used in equality comparisons.
2900 The idea behind this is that sometimes it is convenient (for the
2901 programmer) to consider floating-point values as approximations to
2902 infinitely precise real numbers. If you are doing this, then you need
2903 to compute (by analyzing the code, or in some other way) the maximum or
2904 likely maximum error that the computation introduces, and allow for it
2905 when performing comparisons (and when producing output, but that's a
2906 different problem). In particular, instead of testing for equality, you
2907 would check to see whether the two values have ranges that overlap; and
2908 this is done with the relational operators, so equality comparisons are
2911 @item -Wtraditional @r{(C only)}
2912 @opindex Wtraditional
2913 Warn about certain constructs that behave differently in traditional and
2914 ISO C@. Also warn about ISO C constructs that have no traditional C
2915 equivalent, and/or problematic constructs which should be avoided.
2919 Macro parameters that appear within string literals in the macro body.
2920 In traditional C macro replacement takes place within string literals,
2921 but does not in ISO C@.
2924 In traditional C, some preprocessor directives did not exist.
2925 Traditional preprocessors would only consider a line to be a directive
2926 if the @samp{#} appeared in column 1 on the line. Therefore
2927 @option{-Wtraditional} warns about directives that traditional C
2928 understands but would ignore because the @samp{#} does not appear as the
2929 first character on the line. It also suggests you hide directives like
2930 @samp{#pragma} not understood by traditional C by indenting them. Some
2931 traditional implementations would not recognize @samp{#elif}, so it
2932 suggests avoiding it altogether.
2935 A function-like macro that appears without arguments.
2938 The unary plus operator.
2941 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2942 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2943 constants.) Note, these suffixes appear in macros defined in the system
2944 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2945 Use of these macros in user code might normally lead to spurious
2946 warnings, however GCC's integrated preprocessor has enough context to
2947 avoid warning in these cases.
2950 A function declared external in one block and then used after the end of
2954 A @code{switch} statement has an operand of type @code{long}.
2957 A non-@code{static} function declaration follows a @code{static} one.
2958 This construct is not accepted by some traditional C compilers.
2961 The ISO type of an integer constant has a different width or
2962 signedness from its traditional type. This warning is only issued if
2963 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2964 typically represent bit patterns, are not warned about.
2967 Usage of ISO string concatenation is detected.
2970 Initialization of automatic aggregates.
2973 Identifier conflicts with labels. Traditional C lacks a separate
2974 namespace for labels.
2977 Initialization of unions. If the initializer is zero, the warning is
2978 omitted. This is done under the assumption that the zero initializer in
2979 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2980 initializer warnings and relies on default initialization to zero in the
2984 Conversions by prototypes between fixed/floating point values and vice
2985 versa. The absence of these prototypes when compiling with traditional
2986 C would cause serious problems. This is a subset of the possible
2987 conversion warnings, for the full set use @option{-Wconversion}.
2990 Use of ISO C style function definitions. This warning intentionally is
2991 @emph{not} issued for prototype declarations or variadic functions
2992 because these ISO C features will appear in your code when using
2993 libiberty's traditional C compatibility macros, @code{PARAMS} and
2994 @code{VPARAMS}. This warning is also bypassed for nested functions
2995 because that feature is already a GCC extension and thus not relevant to
2996 traditional C compatibility.
2999 @item -Wdeclaration-after-statement @r{(C only)}
3000 @opindex Wdeclaration-after-statement
3001 Warn when a declaration is found after a statement in a block. This
3002 construct, known from C++, was introduced with ISO C99 and is by default
3003 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3004 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3008 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3010 @item -Wno-endif-labels
3011 @opindex Wno-endif-labels
3012 @opindex Wendif-labels
3013 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3017 Warn whenever a local variable shadows another local variable, parameter or
3018 global variable or whenever a built-in function is shadowed.
3020 @item -Wlarger-than-@var{len}
3021 @opindex Wlarger-than
3022 Warn whenever an object of larger than @var{len} bytes is defined.
3024 @item -Wunsafe-loop-optimizations
3025 @opindex Wunsafe-loop-optimizations
3026 Warn if the loop cannot be optimized because the compiler could not
3027 assume anything on the bounds of the loop indices. With
3028 @option{-funsafe-loop-optimizations} warn if the compiler made
3031 @item -Wpointer-arith
3032 @opindex Wpointer-arith
3033 Warn about anything that depends on the ``size of'' a function type or
3034 of @code{void}. GNU C assigns these types a size of 1, for
3035 convenience in calculations with @code{void *} pointers and pointers
3038 @item -Wbad-function-cast @r{(C only)}
3039 @opindex Wbad-function-cast
3040 Warn whenever a function call is cast to a non-matching type.
3041 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3044 Warn about ISO C constructs that are outside of the common subset of
3045 ISO C and ISO C++, e.g.@: request for implicit conversion from
3046 @code{void *} to a pointer to non-@code{void} type.
3050 Warn whenever a pointer is cast so as to remove a type qualifier from
3051 the target type. For example, warn if a @code{const char *} is cast
3052 to an ordinary @code{char *}.
3055 @opindex Wcast-align
3056 Warn whenever a pointer is cast such that the required alignment of the
3057 target is increased. For example, warn if a @code{char *} is cast to
3058 an @code{int *} on machines where integers can only be accessed at
3059 two- or four-byte boundaries.
3061 @item -Wwrite-strings
3062 @opindex Wwrite-strings
3063 When compiling C, give string constants the type @code{const
3064 char[@var{length}]} so that
3065 copying the address of one into a non-@code{const} @code{char *}
3066 pointer will get a warning; when compiling C++, warn about the
3067 deprecated conversion from string constants to @code{char *}.
3068 These warnings will help you find at
3069 compile time code that can try to write into a string constant, but
3070 only if you have been very careful about using @code{const} in
3071 declarations and prototypes. Otherwise, it will just be a nuisance;
3072 this is why we did not make @option{-Wall} request these warnings.
3075 @opindex Wconversion
3076 Warn if a prototype causes a type conversion that is different from what
3077 would happen to the same argument in the absence of a prototype. This
3078 includes conversions of fixed point to floating and vice versa, and
3079 conversions changing the width or signedness of a fixed point argument
3080 except when the same as the default promotion.
3082 Also, warn if a negative integer constant expression is implicitly
3083 converted to an unsigned type. For example, warn about the assignment
3084 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3085 casts like @code{(unsigned) -1}.
3087 @item -Wsign-compare
3088 @opindex Wsign-compare
3089 @cindex warning for comparison of signed and unsigned values
3090 @cindex comparison of signed and unsigned values, warning
3091 @cindex signed and unsigned values, comparison warning
3092 Warn when a comparison between signed and unsigned values could produce
3093 an incorrect result when the signed value is converted to unsigned.
3094 This warning is also enabled by @option{-Wextra}; to get the other warnings
3095 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3097 @item -Waggregate-return
3098 @opindex Waggregate-return
3099 Warn if any functions that return structures or unions are defined or
3100 called. (In languages where you can return an array, this also elicits
3103 @item -Wno-attributes
3104 @opindex Wno-attributes
3105 @opindex Wattributes
3106 Do not warn if an unexpected @code{__attribute__} is used, such as
3107 unrecognized attributes, function attributes applied to variables,
3108 etc. This will not stop errors for incorrect use of supported
3111 @item -Wstrict-prototypes @r{(C only)}
3112 @opindex Wstrict-prototypes
3113 Warn if a function is declared or defined without specifying the
3114 argument types. (An old-style function definition is permitted without
3115 a warning if preceded by a declaration which specifies the argument
3118 @item -Wold-style-definition @r{(C only)}
3119 @opindex Wold-style-definition
3120 Warn if an old-style function definition is used. A warning is given
3121 even if there is a previous prototype.
3123 @item -Wmissing-prototypes @r{(C only)}
3124 @opindex Wmissing-prototypes
3125 Warn if a global function is defined without a previous prototype
3126 declaration. This warning is issued even if the definition itself
3127 provides a prototype. The aim is to detect global functions that fail
3128 to be declared in header files.
3130 @item -Wmissing-declarations @r{(C only)}
3131 @opindex Wmissing-declarations
3132 Warn if a global function is defined without a previous declaration.
3133 Do so even if the definition itself provides a prototype.
3134 Use this option to detect global functions that are not declared in
3137 @item -Wmissing-field-initializers
3138 @opindex Wmissing-field-initializers
3141 Warn if a structure's initializer has some fields missing. For
3142 example, the following code would cause such a warning, because
3143 @code{x.h} is implicitly zero:
3146 struct s @{ int f, g, h; @};
3147 struct s x = @{ 3, 4 @};
3150 This option does not warn about designated initializers, so the following
3151 modification would not trigger a warning:
3154 struct s @{ int f, g, h; @};
3155 struct s x = @{ .f = 3, .g = 4 @};
3158 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3159 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3161 @item -Wmissing-noreturn
3162 @opindex Wmissing-noreturn
3163 Warn about functions which might be candidates for attribute @code{noreturn}.
3164 Note these are only possible candidates, not absolute ones. Care should
3165 be taken to manually verify functions actually do not ever return before
3166 adding the @code{noreturn} attribute, otherwise subtle code generation
3167 bugs could be introduced. You will not get a warning for @code{main} in
3168 hosted C environments.
3170 @item -Wmissing-format-attribute
3171 @opindex Wmissing-format-attribute
3173 Warn about function pointers which might be candidates for @code{format}
3174 attributes. Note these are only possible candidates, not absolute ones.
3175 GCC will guess that function pointers with @code{format} attributes that
3176 are used in assignment, initialization, parameter passing or return
3177 statements should have a corresponding @code{format} attribute in the
3178 resulting type. I.e.@: the left-hand side of the assignment or
3179 initialization, the type of the parameter variable, or the return type
3180 of the containing function respectively should also have a @code{format}
3181 attribute to avoid the warning.
3183 GCC will also warn about function definitions which might be
3184 candidates for @code{format} attributes. Again, these are only
3185 possible candidates. GCC will guess that @code{format} attributes
3186 might be appropriate for any function that calls a function like
3187 @code{vprintf} or @code{vscanf}, but this might not always be the
3188 case, and some functions for which @code{format} attributes are
3189 appropriate may not be detected.
3191 @item -Wno-multichar
3192 @opindex Wno-multichar
3194 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3195 Usually they indicate a typo in the user's code, as they have
3196 implementation-defined values, and should not be used in portable code.
3198 @item -Wnormalized=<none|id|nfc|nfkc>
3199 @opindex Wnormalized
3202 @cindex character set, input normalization
3203 In ISO C and ISO C++, two identifiers are different if they are
3204 different sequences of characters. However, sometimes when characters
3205 outside the basic ASCII character set are used, you can have two
3206 different character sequences that look the same. To avoid confusion,
3207 the ISO 10646 standard sets out some @dfn{normalization rules} which
3208 when applied ensure that two sequences that look the same are turned into
3209 the same sequence. GCC can warn you if you are using identifiers which
3210 have not been normalized; this option controls that warning.
3212 There are four levels of warning that GCC supports. The default is
3213 @option{-Wnormalized=nfc}, which warns about any identifier which is
3214 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3215 recommended form for most uses.
3217 Unfortunately, there are some characters which ISO C and ISO C++ allow
3218 in identifiers that when turned into NFC aren't allowable as
3219 identifiers. That is, there's no way to use these symbols in portable
3220 ISO C or C++ and have all your identifiers in NFC.
3221 @option{-Wnormalized=id} suppresses the warning for these characters.
3222 It is hoped that future versions of the standards involved will correct
3223 this, which is why this option is not the default.
3225 You can switch the warning off for all characters by writing
3226 @option{-Wnormalized=none}. You would only want to do this if you
3227 were using some other normalization scheme (like ``D''), because
3228 otherwise you can easily create bugs that are literally impossible to see.
3230 Some characters in ISO 10646 have distinct meanings but look identical
3231 in some fonts or display methodologies, especially once formatting has
3232 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3233 LETTER N'', will display just like a regular @code{n} which has been
3234 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3235 normalisation scheme to convert all these into a standard form as
3236 well, and GCC will warn if your code is not in NFKC if you use
3237 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3238 about every identifier that contains the letter O because it might be
3239 confused with the digit 0, and so is not the default, but may be
3240 useful as a local coding convention if the programming environment is
3241 unable to be fixed to display these characters distinctly.
3243 @item -Wno-deprecated-declarations
3244 @opindex Wno-deprecated-declarations
3245 Do not warn about uses of functions, variables, and types marked as
3246 deprecated by using the @code{deprecated} attribute.
3247 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3248 @pxref{Type Attributes}.)
3252 Warn if a structure is given the packed attribute, but the packed
3253 attribute has no effect on the layout or size of the structure.
3254 Such structures may be mis-aligned for little benefit. For
3255 instance, in this code, the variable @code{f.x} in @code{struct bar}
3256 will be misaligned even though @code{struct bar} does not itself
3257 have the packed attribute:
3264 @} __attribute__((packed));
3274 Warn if padding is included in a structure, either to align an element
3275 of the structure or to align the whole structure. Sometimes when this
3276 happens it is possible to rearrange the fields of the structure to
3277 reduce the padding and so make the structure smaller.
3279 @item -Wredundant-decls
3280 @opindex Wredundant-decls
3281 Warn if anything is declared more than once in the same scope, even in
3282 cases where multiple declaration is valid and changes nothing.
3284 @item -Wnested-externs @r{(C only)}
3285 @opindex Wnested-externs
3286 Warn if an @code{extern} declaration is encountered within a function.
3288 @item -Wunreachable-code
3289 @opindex Wunreachable-code
3290 Warn if the compiler detects that code will never be executed.
3292 This option is intended to warn when the compiler detects that at
3293 least a whole line of source code will never be executed, because
3294 some condition is never satisfied or because it is after a
3295 procedure that never returns.
3297 It is possible for this option to produce a warning even though there
3298 are circumstances under which part of the affected line can be executed,
3299 so care should be taken when removing apparently-unreachable code.
3301 For instance, when a function is inlined, a warning may mean that the
3302 line is unreachable in only one inlined copy of the function.
3304 This option is not made part of @option{-Wall} because in a debugging
3305 version of a program there is often substantial code which checks
3306 correct functioning of the program and is, hopefully, unreachable
3307 because the program does work. Another common use of unreachable
3308 code is to provide behavior which is selectable at compile-time.
3312 Warn if a function can not be inlined and it was declared as inline.
3313 Even with this option, the compiler will not warn about failures to
3314 inline functions declared in system headers.
3316 The compiler uses a variety of heuristics to determine whether or not
3317 to inline a function. For example, the compiler takes into account
3318 the size of the function being inlined and the amount of inlining
3319 that has already been done in the current function. Therefore,
3320 seemingly insignificant changes in the source program can cause the
3321 warnings produced by @option{-Winline} to appear or disappear.
3323 @item -Wno-invalid-offsetof @r{(C++ only)}
3324 @opindex Wno-invalid-offsetof
3325 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3326 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3327 to a non-POD type is undefined. In existing C++ implementations,
3328 however, @samp{offsetof} typically gives meaningful results even when
3329 applied to certain kinds of non-POD types. (Such as a simple
3330 @samp{struct} that fails to be a POD type only by virtue of having a
3331 constructor.) This flag is for users who are aware that they are
3332 writing nonportable code and who have deliberately chosen to ignore the
3335 The restrictions on @samp{offsetof} may be relaxed in a future version
3336 of the C++ standard.
3338 @item -Wno-int-to-pointer-cast @r{(C only)}
3339 @opindex Wno-int-to-pointer-cast
3340 Suppress warnings from casts to pointer type of an integer of a
3343 @item -Wno-pointer-to-int-cast @r{(C only)}
3344 @opindex Wno-pointer-to-int-cast
3345 Suppress warnings from casts from a pointer to an integer type of a
3349 @opindex Winvalid-pch
3350 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3351 the search path but can't be used.
3355 @opindex Wno-long-long
3356 Warn if @samp{long long} type is used. This is default. To inhibit
3357 the warning messages, use @option{-Wno-long-long}. Flags
3358 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3359 only when @option{-pedantic} flag is used.
3361 @item -Wvariadic-macros
3362 @opindex Wvariadic-macros
3363 @opindex Wno-variadic-macros
3364 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3365 alternate syntax when in pedantic ISO C99 mode. This is default.
3366 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3368 @item -Wdisabled-optimization
3369 @opindex Wdisabled-optimization
3370 Warn if a requested optimization pass is disabled. This warning does
3371 not generally indicate that there is anything wrong with your code; it
3372 merely indicates that GCC's optimizers were unable to handle the code
3373 effectively. Often, the problem is that your code is too big or too
3374 complex; GCC will refuse to optimize programs when the optimization
3375 itself is likely to take inordinate amounts of time.
3377 @item -Wno-pointer-sign
3378 @opindex Wno-pointer-sign
3379 Don't warn for pointer argument passing or assignment with different signedness.
3380 Only useful in the negative form since this warning is enabled by default.
3381 This option is only supported for C and Objective-C@.
3385 Make all warnings into errors.
3387 @item -Wstack-protector
3388 This option is only active when @option{-fstack-protector} is active. It
3389 warns about functions that will not be protected against stack smashing.
3393 @node Debugging Options
3394 @section Options for Debugging Your Program or GCC
3395 @cindex options, debugging
3396 @cindex debugging information options
3398 GCC has various special options that are used for debugging
3399 either your program or GCC:
3404 Produce debugging information in the operating system's native format
3405 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3408 On most systems that use stabs format, @option{-g} enables use of extra
3409 debugging information that only GDB can use; this extra information
3410 makes debugging work better in GDB but will probably make other debuggers
3412 refuse to read the program. If you want to control for certain whether
3413 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3414 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3416 GCC allows you to use @option{-g} with
3417 @option{-O}. The shortcuts taken by optimized code may occasionally
3418 produce surprising results: some variables you declared may not exist
3419 at all; flow of control may briefly move where you did not expect it;
3420 some statements may not be executed because they compute constant
3421 results or their values were already at hand; some statements may
3422 execute in different places because they were moved out of loops.
3424 Nevertheless it proves possible to debug optimized output. This makes
3425 it reasonable to use the optimizer for programs that might have bugs.
3427 The following options are useful when GCC is generated with the
3428 capability for more than one debugging format.
3432 Produce debugging information for use by GDB@. This means to use the
3433 most expressive format available (DWARF 2, stabs, or the native format
3434 if neither of those are supported), including GDB extensions if at all
3439 Produce debugging information in stabs format (if that is supported),
3440 without GDB extensions. This is the format used by DBX on most BSD
3441 systems. On MIPS, Alpha and System V Release 4 systems this option
3442 produces stabs debugging output which is not understood by DBX or SDB@.
3443 On System V Release 4 systems this option requires the GNU assembler.
3445 @item -feliminate-unused-debug-symbols
3446 @opindex feliminate-unused-debug-symbols
3447 Produce debugging information in stabs format (if that is supported),
3448 for only symbols that are actually used.
3452 Produce debugging information in stabs format (if that is supported),
3453 using GNU extensions understood only by the GNU debugger (GDB)@. The
3454 use of these extensions is likely to make other debuggers crash or
3455 refuse to read the program.
3459 Produce debugging information in COFF format (if that is supported).
3460 This is the format used by SDB on most System V systems prior to
3465 Produce debugging information in XCOFF format (if that is supported).
3466 This is the format used by the DBX debugger on IBM RS/6000 systems.
3470 Produce debugging information in XCOFF format (if that is supported),
3471 using GNU extensions understood only by the GNU debugger (GDB)@. The
3472 use of these extensions is likely to make other debuggers crash or
3473 refuse to read the program, and may cause assemblers other than the GNU
3474 assembler (GAS) to fail with an error.
3478 Produce debugging information in DWARF version 2 format (if that is
3479 supported). This is the format used by DBX on IRIX 6. With this
3480 option, GCC uses features of DWARF version 3 when they are useful;
3481 version 3 is upward compatible with version 2, but may still cause
3482 problems for older debuggers.
3486 Produce debugging information in VMS debug format (if that is
3487 supported). This is the format used by DEBUG on VMS systems.
3490 @itemx -ggdb@var{level}
3491 @itemx -gstabs@var{level}
3492 @itemx -gcoff@var{level}
3493 @itemx -gxcoff@var{level}
3494 @itemx -gvms@var{level}
3495 Request debugging information and also use @var{level} to specify how
3496 much information. The default level is 2.
3498 Level 1 produces minimal information, enough for making backtraces in
3499 parts of the program that you don't plan to debug. This includes
3500 descriptions of functions and external variables, but no information
3501 about local variables and no line numbers.
3503 Level 3 includes extra information, such as all the macro definitions
3504 present in the program. Some debuggers support macro expansion when
3505 you use @option{-g3}.
3507 @option{-gdwarf-2} does not accept a concatenated debug level, because
3508 GCC used to support an option @option{-gdwarf} that meant to generate
3509 debug information in version 1 of the DWARF format (which is very
3510 different from version 2), and it would have been too confusing. That
3511 debug format is long obsolete, but the option cannot be changed now.
3512 Instead use an additional @option{-g@var{level}} option to change the
3513 debug level for DWARF2.
3515 @item -feliminate-dwarf2-dups
3516 @opindex feliminate-dwarf2-dups
3517 Compress DWARF2 debugging information by eliminating duplicated
3518 information about each symbol. This option only makes sense when
3519 generating DWARF2 debugging information with @option{-gdwarf-2}.
3521 @cindex @command{prof}
3524 Generate extra code to write profile information suitable for the
3525 analysis program @command{prof}. You must use this option when compiling
3526 the source files you want data about, and you must also use it when
3529 @cindex @command{gprof}
3532 Generate extra code to write profile information suitable for the
3533 analysis program @command{gprof}. You must use this option when compiling
3534 the source files you want data about, and you must also use it when
3539 Makes the compiler print out each function name as it is compiled, and
3540 print some statistics about each pass when it finishes.
3543 @opindex ftime-report
3544 Makes the compiler print some statistics about the time consumed by each
3545 pass when it finishes.
3548 @opindex fmem-report
3549 Makes the compiler print some statistics about permanent memory
3550 allocation when it finishes.
3552 @item -fprofile-arcs
3553 @opindex fprofile-arcs
3554 Add code so that program flow @dfn{arcs} are instrumented. During
3555 execution the program records how many times each branch and call is
3556 executed and how many times it is taken or returns. When the compiled
3557 program exits it saves this data to a file called
3558 @file{@var{auxname}.gcda} for each source file. The data may be used for
3559 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3560 test coverage analysis (@option{-ftest-coverage}). Each object file's
3561 @var{auxname} is generated from the name of the output file, if
3562 explicitly specified and it is not the final executable, otherwise it is
3563 the basename of the source file. In both cases any suffix is removed
3564 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3565 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3566 @xref{Cross-profiling}.
3568 @cindex @command{gcov}
3572 This option is used to compile and link code instrumented for coverage
3573 analysis. The option is a synonym for @option{-fprofile-arcs}
3574 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3575 linking). See the documentation for those options for more details.
3580 Compile the source files with @option{-fprofile-arcs} plus optimization
3581 and code generation options. For test coverage analysis, use the
3582 additional @option{-ftest-coverage} option. You do not need to profile
3583 every source file in a program.
3586 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3587 (the latter implies the former).
3590 Run the program on a representative workload to generate the arc profile
3591 information. This may be repeated any number of times. You can run
3592 concurrent instances of your program, and provided that the file system
3593 supports locking, the data files will be correctly updated. Also
3594 @code{fork} calls are detected and correctly handled (double counting
3598 For profile-directed optimizations, compile the source files again with
3599 the same optimization and code generation options plus
3600 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3601 Control Optimization}).
3604 For test coverage analysis, use @command{gcov} to produce human readable
3605 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3606 @command{gcov} documentation for further information.
3610 With @option{-fprofile-arcs}, for each function of your program GCC
3611 creates a program flow graph, then finds a spanning tree for the graph.
3612 Only arcs that are not on the spanning tree have to be instrumented: the
3613 compiler adds code to count the number of times that these arcs are
3614 executed. When an arc is the only exit or only entrance to a block, the
3615 instrumentation code can be added to the block; otherwise, a new basic
3616 block must be created to hold the instrumentation code.
3619 @item -ftest-coverage
3620 @opindex ftest-coverage
3621 Produce a notes file that the @command{gcov} code-coverage utility
3622 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3623 show program coverage. Each source file's note file is called
3624 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3625 above for a description of @var{auxname} and instructions on how to
3626 generate test coverage data. Coverage data will match the source files
3627 more closely, if you do not optimize.
3629 @item -d@var{letters}
3630 @item -fdump-rtl-@var{pass}
3632 Says to make debugging dumps during compilation at times specified by
3633 @var{letters}. This is used for debugging the RTL-based passes of the
3634 compiler. The file names for most of the dumps are made by appending a
3635 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3636 from the name of the output file, if explicitly specified and it is not
3637 an executable, otherwise it is the basename of the source file.
3639 Most debug dumps can be enabled either passing a letter to the @option{-d}
3640 option, or with a long @option{-fdump-rtl} switch; here are the possible
3641 letters for use in @var{letters} and @var{pass}, and their meanings:
3646 Annotate the assembler output with miscellaneous debugging information.
3649 @itemx -fdump-rtl-bp
3651 @opindex fdump-rtl-bp
3652 Dump after computing branch probabilities, to @file{@var{file}.09.bp}.
3655 @itemx -fdump-rtl-bbro
3657 @opindex fdump-rtl-bbro
3658 Dump after block reordering, to @file{@var{file}.30.bbro}.
3661 @itemx -fdump-rtl-combine
3663 @opindex fdump-rtl-combine
3664 Dump after instruction combination, to the file @file{@var{file}.17.combine}.
3667 @itemx -fdump-rtl-ce1
3668 @itemx -fdump-rtl-ce2
3670 @opindex fdump-rtl-ce1
3671 @opindex fdump-rtl-ce2
3672 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3673 first if conversion, to the file @file{@var{file}.11.ce1}. @option{-dC}
3674 and @option{-fdump-rtl-ce2} enable dumping after the second if
3675 conversion, to the file @file{@var{file}.18.ce2}.
3678 @itemx -fdump-rtl-btl
3679 @itemx -fdump-rtl-dbr
3681 @opindex fdump-rtl-btl
3682 @opindex fdump-rtl-dbr
3683 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3684 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3685 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3686 scheduling, to @file{@var{file}.36.dbr}.
3690 Dump all macro definitions, at the end of preprocessing, in addition to
3694 @itemx -fdump-rtl-ce3
3696 @opindex fdump-rtl-ce3
3697 Dump after the third if conversion, to @file{@var{file}.28.ce3}.
3700 @itemx -fdump-rtl-cfg
3701 @itemx -fdump-rtl-life
3703 @opindex fdump-rtl-cfg
3704 @opindex fdump-rtl-life
3705 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3706 and data flow analysis, to @file{@var{file}.08.cfg}. @option{-df}
3707 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3708 to @file{@var{file}.16.life}.
3711 @itemx -fdump-rtl-greg
3713 @opindex fdump-rtl-greg
3714 Dump after global register allocation, to @file{@var{file}.23.greg}.
3717 @itemx -fdump-rtl-gcse
3718 @itemx -fdump-rtl-bypass
3720 @opindex fdump-rtl-gcse
3721 @opindex fdump-rtl-bypass
3722 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3723 @file{@var{file}.05.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3724 enable dumping after jump bypassing and control flow optimizations, to
3725 @file{@var{file}.07.bypass}.
3728 @itemx -fdump-rtl-eh
3730 @opindex fdump-rtl-eh
3731 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3734 @itemx -fdump-rtl-sibling
3736 @opindex fdump-rtl-sibling
3737 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3740 @itemx -fdump-rtl-jump
3742 @opindex fdump-rtl-jump
3743 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3746 @itemx -fdump-rtl-stack
3748 @opindex fdump-rtl-stack
3749 Dump after conversion from registers to stack, to @file{@var{file}.33.stack}.
3752 @itemx -fdump-rtl-lreg
3754 @opindex fdump-rtl-lreg
3755 Dump after local register allocation, to @file{@var{file}.22.lreg}.
3758 @itemx -fdump-rtl-loop
3759 @itemx -fdump-rtl-loop2
3761 @opindex fdump-rtl-loop
3762 @opindex fdump-rtl-loop2
3763 @option{-dL} and @option{-fdump-rtl-loop} enable dumping after the first
3764 loop optimization pass, to @file{@var{file}.06.loop}. @option{-dL} and
3765 @option{-fdump-rtl-loop2} enable dumping after the second pass, to
3766 @file{@var{file}.13.loop2}.
3769 @itemx -fdump-rtl-sms
3771 @opindex fdump-rtl-sms
3772 Dump after modulo scheduling, to @file{@var{file}.20.sms}.
3775 @itemx -fdump-rtl-mach
3777 @opindex fdump-rtl-mach
3778 Dump after performing the machine dependent reorganization pass, to
3779 @file{@var{file}.35.mach}.
3782 @itemx -fdump-rtl-rnreg
3784 @opindex fdump-rtl-rnreg
3785 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3788 @itemx -fdump-rtl-regmove
3790 @opindex fdump-rtl-regmove
3791 Dump after the register move pass, to @file{@var{file}.19.regmove}.
3794 @itemx -fdump-rtl-postreload
3796 @opindex fdump-rtl-postreload
3797 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3800 @itemx -fdump-rtl-expand
3802 @opindex fdump-rtl-expand
3803 Dump after RTL generation, to @file{@var{file}.00.expand}.
3806 @itemx -fdump-rtl-sched2
3808 @opindex fdump-rtl-sched2
3809 Dump after the second scheduling pass, to @file{@var{file}.32.sched2}.
3812 @itemx -fdump-rtl-cse
3814 @opindex fdump-rtl-cse
3815 Dump after CSE (including the jump optimization that sometimes follows
3816 CSE), to @file{@var{file}.04.cse}.
3819 @itemx -fdump-rtl-sched
3821 @opindex fdump-rtl-sched
3822 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3825 @itemx -fdump-rtl-cse2
3827 @opindex fdump-rtl-cse2
3828 Dump after the second CSE pass (including the jump optimization that
3829 sometimes follows CSE), to @file{@var{file}.15.cse2}.
3832 @itemx -fdump-rtl-tracer
3834 @opindex fdump-rtl-tracer
3835 Dump after running tracer, to @file{@var{file}.12.tracer}.
3838 @itemx -fdump-rtl-vpt
3839 @itemx -fdump-rtl-vartrack
3841 @opindex fdump-rtl-vpt
3842 @opindex fdump-rtl-vartrack
3843 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3844 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3845 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3846 to @file{@var{file}.34.vartrack}.
3849 @itemx -fdump-rtl-flow2
3851 @opindex fdump-rtl-flow2
3852 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3855 @itemx -fdump-rtl-peephole2
3857 @opindex fdump-rtl-peephole2
3858 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3861 @itemx -fdump-rtl-web
3863 @opindex fdump-rtl-web
3864 Dump after live range splitting, to @file{@var{file}.14.web}.
3867 @itemx -fdump-rtl-all
3869 @opindex fdump-rtl-all
3870 Produce all the dumps listed above.
3874 Produce a core dump whenever an error occurs.
3878 Print statistics on memory usage, at the end of the run, to
3883 Annotate the assembler output with a comment indicating which
3884 pattern and alternative was used. The length of each instruction is
3889 Dump the RTL in the assembler output as a comment before each instruction.
3890 Also turns on @option{-dp} annotation.
3894 For each of the other indicated dump files (either with @option{-d} or
3895 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3896 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3900 Just generate RTL for a function instead of compiling it. Usually used
3901 with @samp{r} (@option{-fdump-rtl-expand}).
3905 Dump debugging information during parsing, to standard error.
3908 @item -fdump-unnumbered
3909 @opindex fdump-unnumbered
3910 When doing debugging dumps (see @option{-d} option above), suppress instruction
3911 numbers and line number note output. This makes it more feasible to
3912 use diff on debugging dumps for compiler invocations with different
3913 options, in particular with and without @option{-g}.
3915 @item -fdump-translation-unit @r{(C++ only)}
3916 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
3917 @opindex fdump-translation-unit
3918 Dump a representation of the tree structure for the entire translation
3919 unit to a file. The file name is made by appending @file{.tu} to the
3920 source file name. If the @samp{-@var{options}} form is used, @var{options}
3921 controls the details of the dump as described for the
3922 @option{-fdump-tree} options.
3924 @item -fdump-class-hierarchy @r{(C++ only)}
3925 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3926 @opindex fdump-class-hierarchy
3927 Dump a representation of each class's hierarchy and virtual function
3928 table layout to a file. The file name is made by appending @file{.class}
3929 to the source file name. If the @samp{-@var{options}} form is used,
3930 @var{options} controls the details of the dump as described for the
3931 @option{-fdump-tree} options.
3933 @item -fdump-ipa-@var{switch}
3935 Control the dumping at various stages of inter-procedural analysis
3936 language tree to a file. The file name is generated by appending a switch
3937 specific suffix to the source file name. The following dumps are possible:
3941 Enables all inter-procedural analysis dumps; currently the only produced
3942 dump is the @samp{cgraph} dump.
3945 Dumps information about call-graph optimization, unused function removal,
3946 and inlining decisions.
3949 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3950 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3952 Control the dumping at various stages of processing the intermediate
3953 language tree to a file. The file name is generated by appending a switch
3954 specific suffix to the source file name. If the @samp{-@var{options}}
3955 form is used, @var{options} is a list of @samp{-} separated options that
3956 control the details of the dump. Not all options are applicable to all
3957 dumps, those which are not meaningful will be ignored. The following
3958 options are available
3962 Print the address of each node. Usually this is not meaningful as it
3963 changes according to the environment and source file. Its primary use
3964 is for tying up a dump file with a debug environment.
3966 Inhibit dumping of members of a scope or body of a function merely
3967 because that scope has been reached. Only dump such items when they
3968 are directly reachable by some other path. When dumping pretty-printed
3969 trees, this option inhibits dumping the bodies of control structures.
3971 Print a raw representation of the tree. By default, trees are
3972 pretty-printed into a C-like representation.
3974 Enable more detailed dumps (not honored by every dump option).
3976 Enable dumping various statistics about the pass (not honored by every dump
3979 Enable showing basic block boundaries (disabled in raw dumps).
3981 Enable showing virtual operands for every statement.
3983 Enable showing line numbers for statements.
3985 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3987 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3990 The following tree dumps are possible:
3994 Dump before any tree based optimization, to @file{@var{file}.original}.
3997 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4000 Dump after function inlining, to @file{@var{file}.inlined}.
4003 @opindex fdump-tree-gimple
4004 Dump each function before and after the gimplification pass to a file. The
4005 file name is made by appending @file{.gimple} to the source file name.
4008 @opindex fdump-tree-cfg
4009 Dump the control flow graph of each function to a file. The file name is
4010 made by appending @file{.cfg} to the source file name.
4013 @opindex fdump-tree-vcg
4014 Dump the control flow graph of each function to a file in VCG format. The
4015 file name is made by appending @file{.vcg} to the source file name. Note
4016 that if the file contains more than one function, the generated file cannot
4017 be used directly by VCG@. You will need to cut and paste each function's
4018 graph into its own separate file first.
4021 @opindex fdump-tree-ch
4022 Dump each function after copying loop headers. The file name is made by
4023 appending @file{.ch} to the source file name.
4026 @opindex fdump-tree-ssa
4027 Dump SSA related information to a file. The file name is made by appending
4028 @file{.ssa} to the source file name.
4031 @opindex fdump-tree-salias
4032 Dump structure aliasing variable information to a file. This file name
4033 is made by appending @file{.salias} to the source file name.
4036 @opindex fdump-tree-alias
4037 Dump aliasing information for each function. The file name is made by
4038 appending @file{.alias} to the source file name.
4041 @opindex fdump-tree-ccp
4042 Dump each function after CCP@. The file name is made by appending
4043 @file{.ccp} to the source file name.
4046 @opindex fdump-tree-storeccp
4047 Dump each function after STORE-CCP. The file name is made by appending
4048 @file{.storeccp} to the source file name.
4051 @opindex fdump-tree-pre
4052 Dump trees after partial redundancy elimination. The file name is made
4053 by appending @file{.pre} to the source file name.
4056 @opindex fdump-tree-fre
4057 Dump trees after full redundancy elimination. The file name is made
4058 by appending @file{.fre} to the source file name.
4061 @opindex fdump-tree-copyprop
4062 Dump trees after copy propagation. The file name is made
4063 by appending @file{.copyprop} to the source file name.
4065 @item store_copyprop
4066 @opindex fdump-tree-store_copyprop
4067 Dump trees after store copy-propagation. The file name is made
4068 by appending @file{.store_copyprop} to the source file name.
4071 @opindex fdump-tree-dce
4072 Dump each function after dead code elimination. The file name is made by
4073 appending @file{.dce} to the source file name.
4076 @opindex fdump-tree-mudflap
4077 Dump each function after adding mudflap instrumentation. The file name is
4078 made by appending @file{.mudflap} to the source file name.
4081 @opindex fdump-tree-sra
4082 Dump each function after performing scalar replacement of aggregates. The
4083 file name is made by appending @file{.sra} to the source file name.
4086 @opindex fdump-tree-sink
4087 Dump each function after performing code sinking. The file name is made
4088 by appending @file{.sink} to the source file name.
4091 @opindex fdump-tree-dom
4092 Dump each function after applying dominator tree optimizations. The file
4093 name is made by appending @file{.dom} to the source file name.
4096 @opindex fdump-tree-dse
4097 Dump each function after applying dead store elimination. The file
4098 name is made by appending @file{.dse} to the source file name.
4101 @opindex fdump-tree-phiopt
4102 Dump each function after optimizing PHI nodes into straightline code. The file
4103 name is made by appending @file{.phiopt} to the source file name.
4106 @opindex fdump-tree-forwprop
4107 Dump each function after forward propagating single use variables. The file
4108 name is made by appending @file{.forwprop} to the source file name.
4111 @opindex fdump-tree-copyrename
4112 Dump each function after applying the copy rename optimization. The file
4113 name is made by appending @file{.copyrename} to the source file name.
4116 @opindex fdump-tree-nrv
4117 Dump each function after applying the named return value optimization on
4118 generic trees. The file name is made by appending @file{.nrv} to the source
4122 @opindex fdump-tree-vect
4123 Dump each function after applying vectorization of loops. The file name is
4124 made by appending @file{.vect} to the source file name.
4127 @opindex fdump-tree-vrp
4128 Dump each function after Value Range Propagation (VRP). The file name
4129 is made by appending @file{.vrp} to the source file name.
4132 @opindex fdump-tree-all
4133 Enable all the available tree dumps with the flags provided in this option.
4136 @item -ftree-vectorizer-verbose=@var{n}
4137 @opindex ftree-vectorizer-verbose
4138 This option controls the amount of debugging output the vectorizer prints.
4139 This information is written to standard error, unless @option{-fdump-tree-all}
4140 or @option{-fdump-tree-vect} is specified, in which case it is output to the
4141 usual dump listing file, @file{.vect}.
4143 @item -frandom-seed=@var{string}
4144 @opindex frandom-string
4145 This option provides a seed that GCC uses when it would otherwise use
4146 random numbers. It is used to generate certain symbol names
4147 that have to be different in every compiled file. It is also used to
4148 place unique stamps in coverage data files and the object files that
4149 produce them. You can use the @option{-frandom-seed} option to produce
4150 reproducibly identical object files.
4152 The @var{string} should be different for every file you compile.
4154 @item -fsched-verbose=@var{n}
4155 @opindex fsched-verbose
4156 On targets that use instruction scheduling, this option controls the
4157 amount of debugging output the scheduler prints. This information is
4158 written to standard error, unless @option{-dS} or @option{-dR} is
4159 specified, in which case it is output to the usual dump
4160 listing file, @file{.sched} or @file{.sched2} respectively. However
4161 for @var{n} greater than nine, the output is always printed to standard
4164 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4165 same information as @option{-dRS}. For @var{n} greater than one, it
4166 also output basic block probabilities, detailed ready list information
4167 and unit/insn info. For @var{n} greater than two, it includes RTL
4168 at abort point, control-flow and regions info. And for @var{n} over
4169 four, @option{-fsched-verbose} also includes dependence info.
4173 Store the usual ``temporary'' intermediate files permanently; place them
4174 in the current directory and name them based on the source file. Thus,
4175 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4176 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4177 preprocessed @file{foo.i} output file even though the compiler now
4178 normally uses an integrated preprocessor.
4180 When used in combination with the @option{-x} command line option,
4181 @option{-save-temps} is sensible enough to avoid over writing an
4182 input source file with the same extension as an intermediate file.
4183 The corresponding intermediate file may be obtained by renaming the
4184 source file before using @option{-save-temps}.
4188 Report the CPU time taken by each subprocess in the compilation
4189 sequence. For C source files, this is the compiler proper and assembler
4190 (plus the linker if linking is done). The output looks like this:
4197 The first number on each line is the ``user time'', that is time spent
4198 executing the program itself. The second number is ``system time'',
4199 time spent executing operating system routines on behalf of the program.
4200 Both numbers are in seconds.
4202 @item -fvar-tracking
4203 @opindex fvar-tracking
4204 Run variable tracking pass. It computes where variables are stored at each
4205 position in code. Better debugging information is then generated
4206 (if the debugging information format supports this information).
4208 It is enabled by default when compiling with optimization (@option{-Os},
4209 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4210 the debug info format supports it.
4212 @item -print-file-name=@var{library}
4213 @opindex print-file-name
4214 Print the full absolute name of the library file @var{library} that
4215 would be used when linking---and don't do anything else. With this
4216 option, GCC does not compile or link anything; it just prints the
4219 @item -print-multi-directory
4220 @opindex print-multi-directory
4221 Print the directory name corresponding to the multilib selected by any
4222 other switches present in the command line. This directory is supposed
4223 to exist in @env{GCC_EXEC_PREFIX}.
4225 @item -print-multi-lib
4226 @opindex print-multi-lib
4227 Print the mapping from multilib directory names to compiler switches
4228 that enable them. The directory name is separated from the switches by
4229 @samp{;}, and each switch starts with an @samp{@@} instead of the
4230 @samp{-}, without spaces between multiple switches. This is supposed to
4231 ease shell-processing.
4233 @item -print-prog-name=@var{program}
4234 @opindex print-prog-name
4235 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4237 @item -print-libgcc-file-name
4238 @opindex print-libgcc-file-name
4239 Same as @option{-print-file-name=libgcc.a}.
4241 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4242 but you do want to link with @file{libgcc.a}. You can do
4245 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4248 @item -print-search-dirs
4249 @opindex print-search-dirs
4250 Print the name of the configured installation directory and a list of
4251 program and library directories @command{gcc} will search---and don't do anything else.
4253 This is useful when @command{gcc} prints the error message
4254 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4255 To resolve this you either need to put @file{cpp0} and the other compiler
4256 components where @command{gcc} expects to find them, or you can set the environment
4257 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4258 Don't forget the trailing @samp{/}.
4259 @xref{Environment Variables}.
4262 @opindex dumpmachine
4263 Print the compiler's target machine (for example,
4264 @samp{i686-pc-linux-gnu})---and don't do anything else.
4267 @opindex dumpversion
4268 Print the compiler version (for example, @samp{3.0})---and don't do
4273 Print the compiler's built-in specs---and don't do anything else. (This
4274 is used when GCC itself is being built.) @xref{Spec Files}.
4276 @item -feliminate-unused-debug-types
4277 @opindex feliminate-unused-debug-types
4278 Normally, when producing DWARF2 output, GCC will emit debugging
4279 information for all types declared in a compilation
4280 unit, regardless of whether or not they are actually used
4281 in that compilation unit. Sometimes this is useful, such as
4282 if, in the debugger, you want to cast a value to a type that is
4283 not actually used in your program (but is declared). More often,
4284 however, this results in a significant amount of wasted space.
4285 With this option, GCC will avoid producing debug symbol output
4286 for types that are nowhere used in the source file being compiled.
4289 @node Optimize Options
4290 @section Options That Control Optimization
4291 @cindex optimize options
4292 @cindex options, optimization
4294 These options control various sorts of optimizations.
4296 Without any optimization option, the compiler's goal is to reduce the
4297 cost of compilation and to make debugging produce the expected
4298 results. Statements are independent: if you stop the program with a
4299 breakpoint between statements, you can then assign a new value to any
4300 variable or change the program counter to any other statement in the
4301 function and get exactly the results you would expect from the source
4304 Turning on optimization flags makes the compiler attempt to improve
4305 the performance and/or code size at the expense of compilation time
4306 and possibly the ability to debug the program.
4308 The compiler performs optimization based on the knowledge it has of
4309 the program. Optimization levels @option{-O2} and above, in
4310 particular, enable @emph{unit-at-a-time} mode, which allows the
4311 compiler to consider information gained from later functions in
4312 the file when compiling a function. Compiling multiple files at
4313 once to a single output file in @emph{unit-at-a-time} mode allows
4314 the compiler to use information gained from all of the files when
4315 compiling each of them.
4317 Not all optimizations are controlled directly by a flag. Only
4318 optimizations that have a flag are listed.
4325 Optimize. Optimizing compilation takes somewhat more time, and a lot
4326 more memory for a large function.
4328 With @option{-O}, the compiler tries to reduce code size and execution
4329 time, without performing any optimizations that take a great deal of
4332 @option{-O} turns on the following optimization flags:
4333 @gccoptlist{-fdefer-pop @gol
4334 -fdelayed-branch @gol
4335 -fguess-branch-probability @gol
4336 -fcprop-registers @gol
4337 -floop-optimize @gol
4338 -fif-conversion @gol
4339 -fif-conversion2 @gol
4342 -ftree-dominator-opts @gol
4347 -ftree-copyrename @gol
4352 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4353 where doing so does not interfere with debugging.
4357 Optimize even more. GCC performs nearly all supported optimizations
4358 that do not involve a space-speed tradeoff. The compiler does not
4359 perform loop unrolling or function inlining when you specify @option{-O2}.
4360 As compared to @option{-O}, this option increases both compilation time
4361 and the performance of the generated code.
4363 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4364 also turns on the following optimization flags:
4365 @gccoptlist{-fthread-jumps @gol
4367 -foptimize-sibling-calls @gol
4368 -fcse-follow-jumps -fcse-skip-blocks @gol
4369 -fgcse -fgcse-lm @gol
4370 -fexpensive-optimizations @gol
4371 -fstrength-reduce @gol
4372 -frerun-cse-after-loop -frerun-loop-opt @gol
4375 -fschedule-insns -fschedule-insns2 @gol
4376 -fsched-interblock -fsched-spec @gol
4378 -fstrict-aliasing @gol
4379 -fdelete-null-pointer-checks @gol
4380 -freorder-blocks -freorder-functions @gol
4381 -funit-at-a-time @gol
4382 -falign-functions -falign-jumps @gol
4383 -falign-loops -falign-labels @gol
4387 Please note the warning under @option{-fgcse} about
4388 invoking @option{-O2} on programs that use computed gotos.
4392 Optimize yet more. @option{-O3} turns on all optimizations specified by
4393 @option{-O2} and also turns on the @option{-finline-functions},
4394 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4398 Do not optimize. This is the default.
4402 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4403 do not typically increase code size. It also performs further
4404 optimizations designed to reduce code size.
4406 @option{-Os} disables the following optimization flags:
4407 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4408 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4409 -fprefetch-loop-arrays -ftree-vect-loop-version}
4411 If you use multiple @option{-O} options, with or without level numbers,
4412 the last such option is the one that is effective.
4415 Options of the form @option{-f@var{flag}} specify machine-independent
4416 flags. Most flags have both positive and negative forms; the negative
4417 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4418 below, only one of the forms is listed---the one you typically will
4419 use. You can figure out the other form by either removing @samp{no-}
4422 The following options control specific optimizations. They are either
4423 activated by @option{-O} options or are related to ones that are. You
4424 can use the following flags in the rare cases when ``fine-tuning'' of
4425 optimizations to be performed is desired.
4428 @item -fno-default-inline
4429 @opindex fno-default-inline
4430 Do not make member functions inline by default merely because they are
4431 defined inside the class scope (C++ only). Otherwise, when you specify
4432 @w{@option{-O}}, member functions defined inside class scope are compiled
4433 inline by default; i.e., you don't need to add @samp{inline} in front of
4434 the member function name.
4436 @item -fno-defer-pop
4437 @opindex fno-defer-pop
4438 Always pop the arguments to each function call as soon as that function
4439 returns. For machines which must pop arguments after a function call,
4440 the compiler normally lets arguments accumulate on the stack for several
4441 function calls and pops them all at once.
4443 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4447 Force memory operands to be copied into registers before doing
4448 arithmetic on them. This produces better code by making all memory
4449 references potential common subexpressions. When they are not common
4450 subexpressions, instruction combination should eliminate the separate
4451 register-load. This option is now a nop and will be removed in 4.2.
4454 @opindex fforce-addr
4455 Force memory address constants to be copied into registers before
4456 doing arithmetic on them.
4458 @item -fomit-frame-pointer
4459 @opindex fomit-frame-pointer
4460 Don't keep the frame pointer in a register for functions that
4461 don't need one. This avoids the instructions to save, set up and
4462 restore frame pointers; it also makes an extra register available
4463 in many functions. @strong{It also makes debugging impossible on
4466 On some machines, such as the VAX, this flag has no effect, because
4467 the standard calling sequence automatically handles the frame pointer
4468 and nothing is saved by pretending it doesn't exist. The
4469 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4470 whether a target machine supports this flag. @xref{Registers,,Register
4471 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4473 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4475 @item -foptimize-sibling-calls
4476 @opindex foptimize-sibling-calls
4477 Optimize sibling and tail recursive calls.
4479 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4483 Don't pay attention to the @code{inline} keyword. Normally this option
4484 is used to keep the compiler from expanding any functions inline.
4485 Note that if you are not optimizing, no functions can be expanded inline.
4487 @item -finline-functions
4488 @opindex finline-functions
4489 Integrate all simple functions into their callers. The compiler
4490 heuristically decides which functions are simple enough to be worth
4491 integrating in this way.
4493 If all calls to a given function are integrated, and the function is
4494 declared @code{static}, then the function is normally not output as
4495 assembler code in its own right.
4497 Enabled at level @option{-O3}.
4499 @item -finline-functions-called-once
4500 @opindex finline-functions-called-once
4501 Consider all @code{static} functions called once for inlining into their
4502 caller even if they are not marked @code{inline}. If a call to a given
4503 function is integrated, then the function is not output as assembler code
4506 Enabled if @option{-funit-at-a-time} is enabled.
4508 @item -fearly-inlining
4509 @opindex fearly-inlining
4510 Inline functions marked by @code{always_inline} and functions whose body seems
4511 smaller than the function call overhead early before doing
4512 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4513 makes profiling significantly cheaper and usually inlining faster on programs
4514 having large chains of nested wrapper functions.
4518 @item -finline-limit=@var{n}
4519 @opindex finline-limit
4520 By default, GCC limits the size of functions that can be inlined. This flag
4521 allows the control of this limit for functions that are explicitly marked as
4522 inline (i.e., marked with the inline keyword or defined within the class
4523 definition in c++). @var{n} is the size of functions that can be inlined in
4524 number of pseudo instructions (not counting parameter handling). The default
4525 value of @var{n} is 600.
4526 Increasing this value can result in more inlined code at
4527 the cost of compilation time and memory consumption. Decreasing usually makes
4528 the compilation faster and less code will be inlined (which presumably
4529 means slower programs). This option is particularly useful for programs that
4530 use inlining heavily such as those based on recursive templates with C++.
4532 Inlining is actually controlled by a number of parameters, which may be
4533 specified individually by using @option{--param @var{name}=@var{value}}.
4534 The @option{-finline-limit=@var{n}} option sets some of these parameters
4538 @item max-inline-insns-single
4539 is set to @var{n}/2.
4540 @item max-inline-insns-auto
4541 is set to @var{n}/2.
4542 @item min-inline-insns
4543 is set to 130 or @var{n}/4, whichever is smaller.
4544 @item max-inline-insns-rtl
4548 See below for a documentation of the individual
4549 parameters controlling inlining.
4551 @emph{Note:} pseudo instruction represents, in this particular context, an
4552 abstract measurement of function's size. In no way does it represent a count
4553 of assembly instructions and as such its exact meaning might change from one
4554 release to an another.
4556 @item -fkeep-inline-functions
4557 @opindex fkeep-inline-functions
4558 In C, emit @code{static} functions that are declared @code{inline}
4559 into the object file, even if the function has been inlined into all
4560 of its callers. This switch does not affect functions using the
4561 @code{extern inline} extension in GNU C@. In C++, emit any and all
4562 inline functions into the object file.
4564 @item -fkeep-static-consts
4565 @opindex fkeep-static-consts
4566 Emit variables declared @code{static const} when optimization isn't turned
4567 on, even if the variables aren't referenced.
4569 GCC enables this option by default. If you want to force the compiler to
4570 check if the variable was referenced, regardless of whether or not
4571 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4573 @item -fmerge-constants
4574 Attempt to merge identical constants (string constants and floating point
4575 constants) across compilation units.
4577 This option is the default for optimized compilation if the assembler and
4578 linker support it. Use @option{-fno-merge-constants} to inhibit this
4581 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4583 @item -fmerge-all-constants
4584 Attempt to merge identical constants and identical variables.
4586 This option implies @option{-fmerge-constants}. In addition to
4587 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4588 arrays or initialized constant variables with integral or floating point
4589 types. Languages like C or C++ require each non-automatic variable to
4590 have distinct location, so using this option will result in non-conforming
4593 @item -fmodulo-sched
4594 @opindex fmodulo-sched
4595 Perform swing modulo scheduling immediately before the first scheduling
4596 pass. This pass looks at innermost loops and reorders their
4597 instructions by overlapping different iterations.
4599 @item -fno-branch-count-reg
4600 @opindex fno-branch-count-reg
4601 Do not use ``decrement and branch'' instructions on a count register,
4602 but instead generate a sequence of instructions that decrement a
4603 register, compare it against zero, then branch based upon the result.
4604 This option is only meaningful on architectures that support such
4605 instructions, which include x86, PowerPC, IA-64 and S/390.
4607 The default is @option{-fbranch-count-reg}, enabled when
4608 @option{-fstrength-reduce} is enabled.
4610 @item -fno-function-cse
4611 @opindex fno-function-cse
4612 Do not put function addresses in registers; make each instruction that
4613 calls a constant function contain the function's address explicitly.
4615 This option results in less efficient code, but some strange hacks
4616 that alter the assembler output may be confused by the optimizations
4617 performed when this option is not used.
4619 The default is @option{-ffunction-cse}
4621 @item -fno-zero-initialized-in-bss
4622 @opindex fno-zero-initialized-in-bss
4623 If the target supports a BSS section, GCC by default puts variables that
4624 are initialized to zero into BSS@. This can save space in the resulting
4627 This option turns off this behavior because some programs explicitly
4628 rely on variables going to the data section. E.g., so that the
4629 resulting executable can find the beginning of that section and/or make
4630 assumptions based on that.
4632 The default is @option{-fzero-initialized-in-bss}.
4634 @item -fbounds-check
4635 @opindex fbounds-check
4636 For front-ends that support it, generate additional code to check that
4637 indices used to access arrays are within the declared range. This is
4638 currently only supported by the Java and Fortran front-ends, where
4639 this option defaults to true and false respectively.
4641 @item -fmudflap -fmudflapth -fmudflapir
4645 @cindex bounds checking
4647 For front-ends that support it (C and C++), instrument all risky
4648 pointer/array dereferencing operations, some standard library
4649 string/heap functions, and some other associated constructs with
4650 range/validity tests. Modules so instrumented should be immune to
4651 buffer overflows, invalid heap use, and some other classes of C/C++
4652 programming errors. The instrumentation relies on a separate runtime
4653 library (@file{libmudflap}), which will be linked into a program if
4654 @option{-fmudflap} is given at link time. Run-time behavior of the
4655 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4656 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4659 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4660 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4661 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4662 instrumentation should ignore pointer reads. This produces less
4663 instrumentation (and therefore faster execution) and still provides
4664 some protection against outright memory corrupting writes, but allows
4665 erroneously read data to propagate within a program.
4667 @item -fstrength-reduce
4668 @opindex fstrength-reduce
4669 Perform the optimizations of loop strength reduction and
4670 elimination of iteration variables.
4672 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4674 @item -fthread-jumps
4675 @opindex fthread-jumps
4676 Perform optimizations where we check to see if a jump branches to a
4677 location where another comparison subsumed by the first is found. If
4678 so, the first branch is redirected to either the destination of the
4679 second branch or a point immediately following it, depending on whether
4680 the condition is known to be true or false.
4682 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4684 @item -fcse-follow-jumps
4685 @opindex fcse-follow-jumps
4686 In common subexpression elimination, scan through jump instructions
4687 when the target of the jump is not reached by any other path. For
4688 example, when CSE encounters an @code{if} statement with an
4689 @code{else} clause, CSE will follow the jump when the condition
4692 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4694 @item -fcse-skip-blocks
4695 @opindex fcse-skip-blocks
4696 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4697 follow jumps which conditionally skip over blocks. When CSE
4698 encounters a simple @code{if} statement with no else clause,
4699 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4700 body of the @code{if}.
4702 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4704 @item -frerun-cse-after-loop
4705 @opindex frerun-cse-after-loop
4706 Re-run common subexpression elimination after loop optimizations has been
4709 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4711 @item -frerun-loop-opt
4712 @opindex frerun-loop-opt
4713 Run the loop optimizer twice.
4715 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4719 Perform a global common subexpression elimination pass.
4720 This pass also performs global constant and copy propagation.
4722 @emph{Note:} When compiling a program using computed gotos, a GCC
4723 extension, you may get better runtime performance if you disable
4724 the global common subexpression elimination pass by adding
4725 @option{-fno-gcse} to the command line.
4727 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4731 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4732 attempt to move loads which are only killed by stores into themselves. This
4733 allows a loop containing a load/store sequence to be changed to a load outside
4734 the loop, and a copy/store within the loop.
4736 Enabled by default when gcse is enabled.
4740 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4741 global common subexpression elimination. This pass will attempt to move
4742 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4743 loops containing a load/store sequence can be changed to a load before
4744 the loop and a store after the loop.
4746 Not enabled at any optimization level.
4750 When @option{-fgcse-las} is enabled, the global common subexpression
4751 elimination pass eliminates redundant loads that come after stores to the
4752 same memory location (both partial and full redundancies).
4754 Not enabled at any optimization level.
4756 @item -fgcse-after-reload
4757 @opindex fgcse-after-reload
4758 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4759 pass is performed after reload. The purpose of this pass is to cleanup
4762 @item -floop-optimize
4763 @opindex floop-optimize
4764 Perform loop optimizations: move constant expressions out of loops, simplify
4765 exit test conditions and optionally do strength-reduction as well.
4767 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4769 @item -floop-optimize2
4770 @opindex floop-optimize2
4771 Perform loop optimizations using the new loop optimizer. The optimizations
4772 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4775 @item -funsafe-loop-optimizations
4776 @opindex funsafe-loop-optimizations
4777 If given, the loop optimizer will assume that loop indices do not
4778 overflow, and that the loops with nontrivial exit condition are not
4779 infinite. This enables a wider range of loop optimizations even if
4780 the loop optimizer itself cannot prove that these assumptions are valid.
4781 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4782 if it finds this kind of loop.
4784 @item -fcrossjumping
4785 @opindex crossjumping
4786 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4787 resulting code may or may not perform better than without cross-jumping.
4789 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4791 @item -fif-conversion
4792 @opindex if-conversion
4793 Attempt to transform conditional jumps into branch-less equivalents. This
4794 include use of conditional moves, min, max, set flags and abs instructions, and
4795 some tricks doable by standard arithmetics. The use of conditional execution
4796 on chips where it is available is controlled by @code{if-conversion2}.
4798 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4800 @item -fif-conversion2
4801 @opindex if-conversion2
4802 Use conditional execution (where available) to transform conditional jumps into
4803 branch-less equivalents.
4805 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4807 @item -fdelete-null-pointer-checks
4808 @opindex fdelete-null-pointer-checks
4809 Use global dataflow analysis to identify and eliminate useless checks
4810 for null pointers. The compiler assumes that dereferencing a null
4811 pointer would have halted the program. If a pointer is checked after
4812 it has already been dereferenced, it cannot be null.
4814 In some environments, this assumption is not true, and programs can
4815 safely dereference null pointers. Use
4816 @option{-fno-delete-null-pointer-checks} to disable this optimization
4817 for programs which depend on that behavior.
4819 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4821 @item -fexpensive-optimizations
4822 @opindex fexpensive-optimizations
4823 Perform a number of minor optimizations that are relatively expensive.
4825 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4827 @item -foptimize-register-move
4829 @opindex foptimize-register-move
4831 Attempt to reassign register numbers in move instructions and as
4832 operands of other simple instructions in order to maximize the amount of
4833 register tying. This is especially helpful on machines with two-operand
4836 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4839 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4841 @item -fdelayed-branch
4842 @opindex fdelayed-branch
4843 If supported for the target machine, attempt to reorder instructions
4844 to exploit instruction slots available after delayed branch
4847 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4849 @item -fschedule-insns
4850 @opindex fschedule-insns
4851 If supported for the target machine, attempt to reorder instructions to
4852 eliminate execution stalls due to required data being unavailable. This
4853 helps machines that have slow floating point or memory load instructions
4854 by allowing other instructions to be issued until the result of the load
4855 or floating point instruction is required.
4857 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4859 @item -fschedule-insns2
4860 @opindex fschedule-insns2
4861 Similar to @option{-fschedule-insns}, but requests an additional pass of
4862 instruction scheduling after register allocation has been done. This is
4863 especially useful on machines with a relatively small number of
4864 registers and where memory load instructions take more than one cycle.
4866 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4868 @item -fno-sched-interblock
4869 @opindex fno-sched-interblock
4870 Don't schedule instructions across basic blocks. This is normally
4871 enabled by default when scheduling before register allocation, i.e.@:
4872 with @option{-fschedule-insns} or at @option{-O2} or higher.
4874 @item -fno-sched-spec
4875 @opindex fno-sched-spec
4876 Don't allow speculative motion of non-load instructions. This is normally
4877 enabled by default when scheduling before register allocation, i.e.@:
4878 with @option{-fschedule-insns} or at @option{-O2} or higher.
4880 @item -fsched-spec-load
4881 @opindex fsched-spec-load
4882 Allow speculative motion of some load instructions. This only makes
4883 sense when scheduling before register allocation, i.e.@: with
4884 @option{-fschedule-insns} or at @option{-O2} or higher.
4886 @item -fsched-spec-load-dangerous
4887 @opindex fsched-spec-load-dangerous
4888 Allow speculative motion of more load instructions. This only makes
4889 sense when scheduling before register allocation, i.e.@: with
4890 @option{-fschedule-insns} or at @option{-O2} or higher.
4892 @item -fsched-stalled-insns=@var{n}
4893 @opindex fsched-stalled-insns
4894 Define how many insns (if any) can be moved prematurely from the queue
4895 of stalled insns into the ready list, during the second scheduling pass.
4897 @item -fsched-stalled-insns-dep=@var{n}
4898 @opindex fsched-stalled-insns-dep
4899 Define how many insn groups (cycles) will be examined for a dependency
4900 on a stalled insn that is candidate for premature removal from the queue
4901 of stalled insns. Has an effect only during the second scheduling pass,
4902 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4904 @item -fsched2-use-superblocks
4905 @opindex fsched2-use-superblocks
4906 When scheduling after register allocation, do use superblock scheduling
4907 algorithm. Superblock scheduling allows motion across basic block boundaries
4908 resulting on faster schedules. This option is experimental, as not all machine
4909 descriptions used by GCC model the CPU closely enough to avoid unreliable
4910 results from the algorithm.
4912 This only makes sense when scheduling after register allocation, i.e.@: with
4913 @option{-fschedule-insns2} or at @option{-O2} or higher.
4915 @item -fsched2-use-traces
4916 @opindex fsched2-use-traces
4917 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4918 allocation and additionally perform code duplication in order to increase the
4919 size of superblocks using tracer pass. See @option{-ftracer} for details on
4922 This mode should produce faster but significantly longer programs. Also
4923 without @option{-fbranch-probabilities} the traces constructed may not
4924 match the reality and hurt the performance. This only makes
4925 sense when scheduling after register allocation, i.e.@: with
4926 @option{-fschedule-insns2} or at @option{-O2} or higher.
4928 @item -freschedule-modulo-scheduled-loops
4929 @opindex fscheduling-in-modulo-scheduled-loops
4930 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4931 we may want to prevent the later scheduling passes from changing its schedule, we use this
4932 option to control that.
4934 @item -fcaller-saves
4935 @opindex fcaller-saves
4936 Enable values to be allocated in registers that will be clobbered by
4937 function calls, by emitting extra instructions to save and restore the
4938 registers around such calls. Such allocation is done only when it
4939 seems to result in better code than would otherwise be produced.
4941 This option is always enabled by default on certain machines, usually
4942 those which have no call-preserved registers to use instead.
4944 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4947 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4948 enabled by default at @option{-O2} and @option{-O3}.
4951 Perform Full Redundancy Elimination (FRE) on trees. The difference
4952 between FRE and PRE is that FRE only considers expressions
4953 that are computed on all paths leading to the redundant computation.
4954 This analysis faster than PRE, though it exposes fewer redundancies.
4955 This flag is enabled by default at @option{-O} and higher.
4957 @item -ftree-copy-prop
4958 Perform copy propagation on trees. This pass eliminates unnecessary
4959 copy operations. This flag is enabled by default at @option{-O} and
4962 @item -ftree-store-copy-prop
4963 Perform copy propagation of memory loads and stores. This pass
4964 eliminates unnecessary copy operations in memory references
4965 (structures, global variables, arrays, etc). This flag is enabled by
4966 default at @option{-O2} and higher.
4969 Perform structural alias analysis on trees. This flag
4970 is enabled by default at @option{-O} and higher.
4973 Perform forward store motion on trees. This flag is
4974 enabled by default at @option{-O} and higher.
4977 Perform sparse conditional constant propagation (CCP) on trees. This
4978 pass only operates on local scalar variables and is enabled by default
4979 at @option{-O} and higher.
4981 @item -ftree-store-ccp
4982 Perform sparse conditional constant propagation (CCP) on trees. This
4983 pass operates on both local scalar variables and memory stores and
4984 loads (global variables, structures, arrays, etc). This flag is
4985 enabled by default at @option{-O2} and higher.
4988 Perform dead code elimination (DCE) on trees. This flag is enabled by
4989 default at @option{-O} and higher.
4991 @item -ftree-dominator-opts
4992 Perform a variety of simple scalar cleanups (constant/copy
4993 propagation, redundancy elimination, range propagation and expression
4994 simplification) based on a dominator tree traversal. This also
4995 performs jump threading (to reduce jumps to jumps). This flag is
4996 enabled by default at @option{-O} and higher.
4999 Perform loop header copying on trees. This is beneficial since it increases
5000 effectiveness of code motion optimizations. It also saves one jump. This flag
5001 is enabled by default at @option{-O} and higher. It is not enabled
5002 for @option{-Os}, since it usually increases code size.
5004 @item -ftree-loop-optimize
5005 Perform loop optimizations on trees. This flag is enabled by default
5006 at @option{-O} and higher.
5008 @item -ftree-loop-linear
5009 Perform linear loop transformations on tree. This flag can improve cache
5010 performance and allow further loop optimizations to take place.
5012 @item -ftree-loop-im
5013 Perform loop invariant motion on trees. This pass moves only invariants that
5014 would be hard to handle at RTL level (function calls, operations that expand to
5015 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5016 operands of conditions that are invariant out of the loop, so that we can use
5017 just trivial invariantness analysis in loop unswitching. The pass also includes
5020 @item -ftree-loop-ivcanon
5021 Create a canonical counter for number of iterations in the loop for that
5022 determining number of iterations requires complicated analysis. Later
5023 optimizations then may determine the number easily. Useful especially
5024 in connection with unrolling.
5027 Perform induction variable optimizations (strength reduction, induction
5028 variable merging and induction variable elimination) on trees.
5031 Perform scalar replacement of aggregates. This pass replaces structure
5032 references with scalars to prevent committing structures to memory too
5033 early. This flag is enabled by default at @option{-O} and higher.
5035 @item -ftree-copyrename
5036 Perform copy renaming on trees. This pass attempts to rename compiler
5037 temporaries to other variables at copy locations, usually resulting in
5038 variable names which more closely resemble the original variables. This flag
5039 is enabled by default at @option{-O} and higher.
5042 Perform temporary expression replacement during the SSA->normal phase. Single
5043 use/single def temporaries are replaced at their use location with their
5044 defining expression. This results in non-GIMPLE code, but gives the expanders
5045 much more complex trees to work on resulting in better RTL generation. This is
5046 enabled by default at @option{-O} and higher.
5049 Perform live range splitting during the SSA->normal phase. Distinct live
5050 ranges of a variable are split into unique variables, allowing for better
5051 optimization later. This is enabled by default at @option{-O} and higher.
5053 @item -ftree-vectorize
5054 Perform loop vectorization on trees.
5056 @item -ftree-vect-loop-version
5057 @opindex ftree-vect-loop-version
5058 Perform loop versioning when doing loop vectorization on trees. When a loop
5059 appears to be vectorizable except that data alignment or data dependence cannot
5060 be determined at compile time then vectorized and non-vectorized versions of
5061 the loop are generated along with runtime checks for alignment or dependence
5062 to control which version is executed. This option is enabled by default
5063 except at level @option{-Os} where it is disabled.
5066 Perform Value Range Propagation on trees. This is similar to the
5067 constant propagation pass, but instead of values, ranges of values are
5068 propagated. This allows the optimizers to remove unnecessary range
5069 checks like array bound checks and null pointer checks. This is
5070 enabled by default at @option{-O2} and higher. Null pointer check
5071 elimination is only done if @option{-fdelete-null-pointer-checks} is
5076 Perform tail duplication to enlarge superblock size. This transformation
5077 simplifies the control flow of the function allowing other optimizations to do
5080 @item -funroll-loops
5081 @opindex funroll-loops
5082 Unroll loops whose number of iterations can be determined at compile
5083 time or upon entry to the loop. @option{-funroll-loops} implies both
5084 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
5085 option makes code larger, and may or may not make it run faster.
5087 @item -funroll-all-loops
5088 @opindex funroll-all-loops
5089 Unroll all loops, even if their number of iterations is uncertain when
5090 the loop is entered. This usually makes programs run more slowly.
5091 @option{-funroll-all-loops} implies the same options as
5092 @option{-funroll-loops},
5094 @item -fsplit-ivs-in-unroller
5095 @opindex -fsplit-ivs-in-unroller
5096 Enables expressing of values of induction variables in later iterations
5097 of the unrolled loop using the value in the first iteration. This breaks
5098 long dependency chains, thus improving efficiency of the scheduling passes.
5100 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5101 same effect. However in cases the loop body is more complicated than
5102 a single basic block, this is not reliable. It also does not work at all
5103 on some of the architectures due to restrictions in the CSE pass.
5105 This optimization is enabled by default.
5107 @item -fvariable-expansion-in-unroller
5108 @opindex -fvariable-expansion-in-unroller
5109 With this option, the compiler will create multiple copies of some
5110 local variables when unrolling a loop which can result in superior code.
5112 @item -fprefetch-loop-arrays
5113 @opindex fprefetch-loop-arrays
5114 If supported by the target machine, generate instructions to prefetch
5115 memory to improve the performance of loops that access large arrays.
5117 These options may generate better or worse code; results are highly
5118 dependent on the structure of loops within the source code.
5121 @itemx -fno-peephole2
5122 @opindex fno-peephole
5123 @opindex fno-peephole2
5124 Disable any machine-specific peephole optimizations. The difference
5125 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5126 are implemented in the compiler; some targets use one, some use the
5127 other, a few use both.
5129 @option{-fpeephole} is enabled by default.
5130 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5132 @item -fno-guess-branch-probability
5133 @opindex fno-guess-branch-probability
5134 Do not guess branch probabilities using heuristics.
5136 GCC will use heuristics to guess branch probabilities if they are
5137 not provided by profiling feedback (@option{-fprofile-arcs}). These
5138 heuristics are based on the control flow graph. If some branch probabilities
5139 are specified by @samp{__builtin_expect}, then the heuristics will be
5140 used to guess branch probabilities for the rest of the control flow graph,
5141 taking the @samp{__builtin_expect} info into account. The interactions
5142 between the heuristics and @samp{__builtin_expect} can be complex, and in
5143 some cases, it may be useful to disable the heuristics so that the effects
5144 of @samp{__builtin_expect} are easier to understand.
5146 The default is @option{-fguess-branch-probability} at levels
5147 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5149 @item -freorder-blocks
5150 @opindex freorder-blocks
5151 Reorder basic blocks in the compiled function in order to reduce number of
5152 taken branches and improve code locality.
5154 Enabled at levels @option{-O2}, @option{-O3}.
5156 @item -freorder-blocks-and-partition
5157 @opindex freorder-blocks-and-partition
5158 In addition to reordering basic blocks in the compiled function, in order
5159 to reduce number of taken branches, partitions hot and cold basic blocks
5160 into separate sections of the assembly and .o files, to improve
5161 paging and cache locality performance.
5163 This optimization is automatically turned off in the presence of
5164 exception handling, for linkonce sections, for functions with a user-defined
5165 section attribute and on any architecture that does not support named
5168 @item -freorder-functions
5169 @opindex freorder-functions
5170 Reorder functions in the object file in order to
5171 improve code locality. This is implemented by using special
5172 subsections @code{.text.hot} for most frequently executed functions and
5173 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5174 the linker so object file format must support named sections and linker must
5175 place them in a reasonable way.
5177 Also profile feedback must be available in to make this option effective. See
5178 @option{-fprofile-arcs} for details.
5180 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5182 @item -fstrict-aliasing
5183 @opindex fstrict-aliasing
5184 Allows the compiler to assume the strictest aliasing rules applicable to
5185 the language being compiled. For C (and C++), this activates
5186 optimizations based on the type of expressions. In particular, an
5187 object of one type is assumed never to reside at the same address as an
5188 object of a different type, unless the types are almost the same. For
5189 example, an @code{unsigned int} can alias an @code{int}, but not a
5190 @code{void*} or a @code{double}. A character type may alias any other
5193 Pay special attention to code like this:
5206 The practice of reading from a different union member than the one most
5207 recently written to (called ``type-punning'') is common. Even with
5208 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5209 is accessed through the union type. So, the code above will work as
5210 expected. However, this code might not:
5221 Every language that wishes to perform language-specific alias analysis
5222 should define a function that computes, given an @code{tree}
5223 node, an alias set for the node. Nodes in different alias sets are not
5224 allowed to alias. For an example, see the C front-end function
5225 @code{c_get_alias_set}.
5227 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5229 @item -falign-functions
5230 @itemx -falign-functions=@var{n}
5231 @opindex falign-functions
5232 Align the start of functions to the next power-of-two greater than
5233 @var{n}, skipping up to @var{n} bytes. For instance,
5234 @option{-falign-functions=32} aligns functions to the next 32-byte
5235 boundary, but @option{-falign-functions=24} would align to the next
5236 32-byte boundary only if this can be done by skipping 23 bytes or less.
5238 @option{-fno-align-functions} and @option{-falign-functions=1} are
5239 equivalent and mean that functions will not be aligned.
5241 Some assemblers only support this flag when @var{n} is a power of two;
5242 in that case, it is rounded up.
5244 If @var{n} is not specified or is zero, use a machine-dependent default.
5246 Enabled at levels @option{-O2}, @option{-O3}.
5248 @item -falign-labels
5249 @itemx -falign-labels=@var{n}
5250 @opindex falign-labels
5251 Align all branch targets to a power-of-two boundary, skipping up to
5252 @var{n} bytes like @option{-falign-functions}. This option can easily
5253 make code slower, because it must insert dummy operations for when the
5254 branch target is reached in the usual flow of the code.
5256 @option{-fno-align-labels} and @option{-falign-labels=1} are
5257 equivalent and mean that labels will not be aligned.
5259 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5260 are greater than this value, then their values are used instead.
5262 If @var{n} is not specified or is zero, use a machine-dependent default
5263 which is very likely to be @samp{1}, meaning no alignment.
5265 Enabled at levels @option{-O2}, @option{-O3}.
5268 @itemx -falign-loops=@var{n}
5269 @opindex falign-loops
5270 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5271 like @option{-falign-functions}. The hope is that the loop will be
5272 executed many times, which will make up for any execution of the dummy
5275 @option{-fno-align-loops} and @option{-falign-loops=1} are
5276 equivalent and mean that loops will not be aligned.
5278 If @var{n} is not specified or is zero, use a machine-dependent default.
5280 Enabled at levels @option{-O2}, @option{-O3}.
5283 @itemx -falign-jumps=@var{n}
5284 @opindex falign-jumps
5285 Align branch targets to a power-of-two boundary, for branch targets
5286 where the targets can only be reached by jumping, skipping up to @var{n}
5287 bytes like @option{-falign-functions}. In this case, no dummy operations
5290 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5291 equivalent and mean that loops will not be aligned.
5293 If @var{n} is not specified or is zero, use a machine-dependent default.
5295 Enabled at levels @option{-O2}, @option{-O3}.
5297 @item -funit-at-a-time
5298 @opindex funit-at-a-time
5299 Parse the whole compilation unit before starting to produce code.
5300 This allows some extra optimizations to take place but consumes
5301 more memory (in general). There are some compatibility issues
5302 with @emph{unit-at-at-time} mode:
5305 enabling @emph{unit-at-a-time} mode may change the order
5306 in which functions, variables, and top-level @code{asm} statements
5307 are emitted, and will likely break code relying on some particular
5308 ordering. The majority of such top-level @code{asm} statements,
5309 though, can be replaced by @code{section} attributes.
5312 @emph{unit-at-a-time} mode removes unreferenced static variables
5313 and functions. This may result in undefined references
5314 when an @code{asm} statement refers directly to variables or functions
5315 that are otherwise unused. In that case either the variable/function
5316 shall be listed as an operand of the @code{asm} statement operand or,
5317 in the case of top-level @code{asm} statements the attribute @code{used}
5318 shall be used on the declaration.
5321 Static functions now can use non-standard passing conventions that
5322 may break @code{asm} statements calling functions directly. Again,
5323 attribute @code{used} will prevent this behavior.
5326 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5327 but this scheme may not be supported by future releases of GCC@.
5329 Enabled at levels @option{-O2}, @option{-O3}.
5333 Constructs webs as commonly used for register allocation purposes and assign
5334 each web individual pseudo register. This allows the register allocation pass
5335 to operate on pseudos directly, but also strengthens several other optimization
5336 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5337 however, make debugging impossible, since variables will no longer stay in a
5340 Enabled by default with @option{-funroll-loops}.
5342 @item -fwhole-program
5343 @opindex fwhole-program
5344 Assume that the current compilation unit represents whole program being
5345 compiled. All public functions and variables with the exception of @code{main}
5346 and those merged by attribute @code{externally_visible} become static functions
5347 and in a affect gets more aggressively optimized by interprocedural optimizers.
5348 While this option is equivalent to proper use of @code{static} keyword for
5349 programs consisting of single file, in combination with option
5350 @option{--combine} this flag can be used to compile most of smaller scale C
5351 programs since the functions and variables become local for the whole combined
5352 compilation unit, not for the single source file itself.
5355 @item -fno-cprop-registers
5356 @opindex fno-cprop-registers
5357 After register allocation and post-register allocation instruction splitting,
5358 we perform a copy-propagation pass to try to reduce scheduling dependencies
5359 and occasionally eliminate the copy.
5361 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5363 @item -fprofile-generate
5364 @opindex fprofile-generate
5366 Enable options usually used for instrumenting application to produce
5367 profile useful for later recompilation with profile feedback based
5368 optimization. You must use @option{-fprofile-generate} both when
5369 compiling and when linking your program.
5371 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5374 @opindex fprofile-use
5375 Enable profile feedback directed optimizations, and optimizations
5376 generally profitable only with profile feedback available.
5378 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5379 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer},
5380 @code{-fno-loop-optimize}.
5384 The following options control compiler behavior regarding floating
5385 point arithmetic. These options trade off between speed and
5386 correctness. All must be specifically enabled.
5390 @opindex ffloat-store
5391 Do not store floating point variables in registers, and inhibit other
5392 options that might change whether a floating point value is taken from a
5395 @cindex floating point precision
5396 This option prevents undesirable excess precision on machines such as
5397 the 68000 where the floating registers (of the 68881) keep more
5398 precision than a @code{double} is supposed to have. Similarly for the
5399 x86 architecture. For most programs, the excess precision does only
5400 good, but a few programs rely on the precise definition of IEEE floating
5401 point. Use @option{-ffloat-store} for such programs, after modifying
5402 them to store all pertinent intermediate computations into variables.
5406 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5407 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5408 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5409 and @option{fcx-limited-range}.
5411 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5413 This option should never be turned on by any @option{-O} option since
5414 it can result in incorrect output for programs which depend on
5415 an exact implementation of IEEE or ISO rules/specifications for
5418 @item -fno-math-errno
5419 @opindex fno-math-errno
5420 Do not set ERRNO after calling math functions that are executed
5421 with a single instruction, e.g., sqrt. A program that relies on
5422 IEEE exceptions for math error handling may want to use this flag
5423 for speed while maintaining IEEE arithmetic compatibility.
5425 This option should never be turned on by any @option{-O} option since
5426 it can result in incorrect output for programs which depend on
5427 an exact implementation of IEEE or ISO rules/specifications for
5430 The default is @option{-fmath-errno}.
5432 On Darwin systems, the math library never sets @code{errno}. There is therefore
5433 no reason for the compiler to consider the possibility that it might,
5434 and @option{-fno-math-errno} is the default.
5436 @item -funsafe-math-optimizations
5437 @opindex funsafe-math-optimizations
5438 Allow optimizations for floating-point arithmetic that (a) assume
5439 that arguments and results are valid and (b) may violate IEEE or
5440 ANSI standards. When used at link-time, it may include libraries
5441 or startup files that change the default FPU control word or other
5442 similar optimizations.
5444 This option should never be turned on by any @option{-O} option since
5445 it can result in incorrect output for programs which depend on
5446 an exact implementation of IEEE or ISO rules/specifications for
5449 The default is @option{-fno-unsafe-math-optimizations}.
5451 @item -ffinite-math-only
5452 @opindex ffinite-math-only
5453 Allow optimizations for floating-point arithmetic that assume
5454 that arguments and results are not NaNs or +-Infs.
5456 This option should never be turned on by any @option{-O} option since
5457 it can result in incorrect output for programs which depend on
5458 an exact implementation of IEEE or ISO rules/specifications.
5460 The default is @option{-fno-finite-math-only}.
5462 @item -fno-trapping-math
5463 @opindex fno-trapping-math
5464 Compile code assuming that floating-point operations cannot generate
5465 user-visible traps. These traps include division by zero, overflow,
5466 underflow, inexact result and invalid operation. This option implies
5467 @option{-fno-signaling-nans}. Setting this option may allow faster
5468 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5470 This option should never be turned on by any @option{-O} option since
5471 it can result in incorrect output for programs which depend on
5472 an exact implementation of IEEE or ISO rules/specifications for
5475 The default is @option{-ftrapping-math}.
5477 @item -frounding-math
5478 @opindex frounding-math
5479 Disable transformations and optimizations that assume default floating
5480 point rounding behavior. This is round-to-zero for all floating point
5481 to integer conversions, and round-to-nearest for all other arithmetic
5482 truncations. This option should be specified for programs that change
5483 the FP rounding mode dynamically, or that may be executed with a
5484 non-default rounding mode. This option disables constant folding of
5485 floating point expressions at compile-time (which may be affected by
5486 rounding mode) and arithmetic transformations that are unsafe in the
5487 presence of sign-dependent rounding modes.
5489 The default is @option{-fno-rounding-math}.
5491 This option is experimental and does not currently guarantee to
5492 disable all GCC optimizations that are affected by rounding mode.
5493 Future versions of GCC may provide finer control of this setting
5494 using C99's @code{FENV_ACCESS} pragma. This command line option
5495 will be used to specify the default state for @code{FENV_ACCESS}.
5497 @item -fsignaling-nans
5498 @opindex fsignaling-nans
5499 Compile code assuming that IEEE signaling NaNs may generate user-visible
5500 traps during floating-point operations. Setting this option disables
5501 optimizations that may change the number of exceptions visible with
5502 signaling NaNs. This option implies @option{-ftrapping-math}.
5504 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5507 The default is @option{-fno-signaling-nans}.
5509 This option is experimental and does not currently guarantee to
5510 disable all GCC optimizations that affect signaling NaN behavior.
5512 @item -fsingle-precision-constant
5513 @opindex fsingle-precision-constant
5514 Treat floating point constant as single precision constant instead of
5515 implicitly converting it to double precision constant.
5517 @item -fcx-limited-range
5518 @itemx -fno-cx-limited-range
5519 @opindex fcx-limited-range
5520 @opindex fno-cx-limited-range
5521 When enabled, this option states that a range reduction step is not
5522 needed when performing complex division. The default is
5523 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5525 This option controls the default setting of the ISO C99
5526 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5531 The following options control optimizations that may improve
5532 performance, but are not enabled by any @option{-O} options. This
5533 section includes experimental options that may produce broken code.
5536 @item -fbranch-probabilities
5537 @opindex fbranch-probabilities
5538 After running a program compiled with @option{-fprofile-arcs}
5539 (@pxref{Debugging Options,, Options for Debugging Your Program or
5540 @command{gcc}}), you can compile it a second time using
5541 @option{-fbranch-probabilities}, to improve optimizations based on
5542 the number of times each branch was taken. When the program
5543 compiled with @option{-fprofile-arcs} exits it saves arc execution
5544 counts to a file called @file{@var{sourcename}.gcda} for each source
5545 file The information in this data file is very dependent on the
5546 structure of the generated code, so you must use the same source code
5547 and the same optimization options for both compilations.
5549 With @option{-fbranch-probabilities}, GCC puts a
5550 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5551 These can be used to improve optimization. Currently, they are only
5552 used in one place: in @file{reorg.c}, instead of guessing which path a
5553 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5554 exactly determine which path is taken more often.
5556 @item -fprofile-values
5557 @opindex fprofile-values
5558 If combined with @option{-fprofile-arcs}, it adds code so that some
5559 data about values of expressions in the program is gathered.
5561 With @option{-fbranch-probabilities}, it reads back the data gathered
5562 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5563 notes to instructions for their later usage in optimizations.
5565 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5569 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5570 a code to gather information about values of expressions.
5572 With @option{-fbranch-probabilities}, it reads back the data gathered
5573 and actually performs the optimizations based on them.
5574 Currently the optimizations include specialization of division operation
5575 using the knowledge about the value of the denominator.
5577 @item -frename-registers
5578 @opindex frename-registers
5579 Attempt to avoid false dependencies in scheduled code by making use
5580 of registers left over after register allocation. This optimization
5581 will most benefit processors with lots of registers. Depending on the
5582 debug information format adopted by the target, however, it can
5583 make debugging impossible, since variables will no longer stay in
5584 a ``home register''.
5586 Enabled by default with @option{-funroll-loops}.
5590 Perform tail duplication to enlarge superblock size. This transformation
5591 simplifies the control flow of the function allowing other optimizations to do
5594 Enabled with @option{-fprofile-use}.
5596 @item -funroll-loops
5597 @opindex funroll-loops
5598 Unroll loops whose number of iterations can be determined at compile time or
5599 upon entry to the loop. @option{-funroll-loops} implies
5600 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5601 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5602 small constant number of iterations). This option makes code larger, and may
5603 or may not make it run faster.
5605 Enabled with @option{-fprofile-use}.
5607 @item -funroll-all-loops
5608 @opindex funroll-all-loops
5609 Unroll all loops, even if their number of iterations is uncertain when
5610 the loop is entered. This usually makes programs run more slowly.
5611 @option{-funroll-all-loops} implies the same options as
5612 @option{-funroll-loops}.
5615 @opindex fpeel-loops
5616 Peels the loops for that there is enough information that they do not
5617 roll much (from profile feedback). It also turns on complete loop peeling
5618 (i.e.@: complete removal of loops with small constant number of iterations).
5620 Enabled with @option{-fprofile-use}.
5622 @item -fmove-loop-invariants
5623 @opindex fmove-loop-invariants
5624 Enables the loop invariant motion pass in the new loop optimizer. Enabled
5625 at level @option{-O1}
5627 @item -funswitch-loops
5628 @opindex funswitch-loops
5629 Move branches with loop invariant conditions out of the loop, with duplicates
5630 of the loop on both branches (modified according to result of the condition).
5632 @item -fprefetch-loop-arrays
5633 @opindex fprefetch-loop-arrays
5634 If supported by the target machine, generate instructions to prefetch
5635 memory to improve the performance of loops that access large arrays.
5637 Disabled at level @option{-Os}.
5639 @item -ffunction-sections
5640 @itemx -fdata-sections
5641 @opindex ffunction-sections
5642 @opindex fdata-sections
5643 Place each function or data item into its own section in the output
5644 file if the target supports arbitrary sections. The name of the
5645 function or the name of the data item determines the section's name
5648 Use these options on systems where the linker can perform optimizations
5649 to improve locality of reference in the instruction space. Most systems
5650 using the ELF object format and SPARC processors running Solaris 2 have
5651 linkers with such optimizations. AIX may have these optimizations in
5654 Only use these options when there are significant benefits from doing
5655 so. When you specify these options, the assembler and linker will
5656 create larger object and executable files and will also be slower.
5657 You will not be able to use @code{gprof} on all systems if you
5658 specify this option and you may have problems with debugging if
5659 you specify both this option and @option{-g}.
5661 @item -fbranch-target-load-optimize
5662 @opindex fbranch-target-load-optimize
5663 Perform branch target register load optimization before prologue / epilogue
5665 The use of target registers can typically be exposed only during reload,
5666 thus hoisting loads out of loops and doing inter-block scheduling needs
5667 a separate optimization pass.
5669 @item -fbranch-target-load-optimize2
5670 @opindex fbranch-target-load-optimize2
5671 Perform branch target register load optimization after prologue / epilogue
5674 @item -fbtr-bb-exclusive
5675 @opindex fbtr-bb-exclusive
5676 When performing branch target register load optimization, don't reuse
5677 branch target registers in within any basic block.
5679 @item -fstack-protector
5680 Emit extra code to check for buffer overflows, such as stack smashing
5681 attacks. This is done by adding a guard variable to functions with
5682 vulnerable objects. This includes functions that call alloca, and
5683 functions with buffers larger than 8 bytes. The guards are initialized
5684 when a function is entered and then checked when the function exits.
5685 If a guard check fails, an error message is printed and the program exits.
5687 @item -fstack-protector-all
5688 Like @option{-fstack-protector} except that all functions are protected.
5690 @item --param @var{name}=@var{value}
5692 In some places, GCC uses various constants to control the amount of
5693 optimization that is done. For example, GCC will not inline functions
5694 that contain more that a certain number of instructions. You can
5695 control some of these constants on the command-line using the
5696 @option{--param} option.
5698 The names of specific parameters, and the meaning of the values, are
5699 tied to the internals of the compiler, and are subject to change
5700 without notice in future releases.
5702 In each case, the @var{value} is an integer. The allowable choices for
5703 @var{name} are given in the following table:
5706 @item salias-max-implicit-fields
5707 The maximum number of fields in a variable without direct
5708 structure accesses for which structure aliasing will consider trying
5709 to track each field. The default is 5
5711 @item sra-max-structure-size
5712 The maximum structure size, in bytes, at which the scalar replacement
5713 of aggregates (SRA) optimization will perform block copies. The
5714 default value, 0, implies that GCC will select the most appropriate
5717 @item sra-field-structure-ratio
5718 The threshold ratio (as a percentage) between instantiated fields and
5719 the complete structure size. We say that if the ratio of the number
5720 of bytes in instantiated fields to the number of bytes in the complete
5721 structure exceeds this parameter, then block copies are not used. The
5724 @item max-crossjump-edges
5725 The maximum number of incoming edges to consider for crossjumping.
5726 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5727 the number of edges incoming to each block. Increasing values mean
5728 more aggressive optimization, making the compile time increase with
5729 probably small improvement in executable size.
5731 @item min-crossjump-insns
5732 The minimum number of instructions which must be matched at the end
5733 of two blocks before crossjumping will be performed on them. This
5734 value is ignored in the case where all instructions in the block being
5735 crossjumped from are matched. The default value is 5.
5737 @item max-goto-duplication-insns
5738 The maximum number of instructions to duplicate to a block that jumps
5739 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5740 passes, GCC factors computed gotos early in the compilation process,
5741 and unfactors them as late as possible. Only computed jumps at the
5742 end of a basic blocks with no more than max-goto-duplication-insns are
5743 unfactored. The default value is 8.
5745 @item max-delay-slot-insn-search
5746 The maximum number of instructions to consider when looking for an
5747 instruction to fill a delay slot. If more than this arbitrary number of
5748 instructions is searched, the time savings from filling the delay slot
5749 will be minimal so stop searching. Increasing values mean more
5750 aggressive optimization, making the compile time increase with probably
5751 small improvement in executable run time.
5753 @item max-delay-slot-live-search
5754 When trying to fill delay slots, the maximum number of instructions to
5755 consider when searching for a block with valid live register
5756 information. Increasing this arbitrarily chosen value means more
5757 aggressive optimization, increasing the compile time. This parameter
5758 should be removed when the delay slot code is rewritten to maintain the
5761 @item max-gcse-memory
5762 The approximate maximum amount of memory that will be allocated in
5763 order to perform the global common subexpression elimination
5764 optimization. If more memory than specified is required, the
5765 optimization will not be done.
5767 @item max-gcse-passes
5768 The maximum number of passes of GCSE to run. The default is 1.
5770 @item max-pending-list-length
5771 The maximum number of pending dependencies scheduling will allow
5772 before flushing the current state and starting over. Large functions
5773 with few branches or calls can create excessively large lists which
5774 needlessly consume memory and resources.
5776 @item max-inline-insns-single
5777 Several parameters control the tree inliner used in gcc.
5778 This number sets the maximum number of instructions (counted in GCC's
5779 internal representation) in a single function that the tree inliner
5780 will consider for inlining. This only affects functions declared
5781 inline and methods implemented in a class declaration (C++).
5782 The default value is 450.
5784 @item max-inline-insns-auto
5785 When you use @option{-finline-functions} (included in @option{-O3}),
5786 a lot of functions that would otherwise not be considered for inlining
5787 by the compiler will be investigated. To those functions, a different
5788 (more restrictive) limit compared to functions declared inline can
5790 The default value is 90.
5792 @item large-function-insns
5793 The limit specifying really large functions. For functions larger than this
5794 limit after inlining inlining is constrained by
5795 @option{--param large-function-growth}. This parameter is useful primarily
5796 to avoid extreme compilation time caused by non-linear algorithms used by the
5798 This parameter is ignored when @option{-funit-at-a-time} is not used.
5799 The default value is 2700.
5801 @item large-function-growth
5802 Specifies maximal growth of large function caused by inlining in percents.
5803 This parameter is ignored when @option{-funit-at-a-time} is not used.
5804 The default value is 100 which limits large function growth to 2.0 times
5807 @item inline-unit-growth
5808 Specifies maximal overall growth of the compilation unit caused by inlining.
5809 This parameter is ignored when @option{-funit-at-a-time} is not used.
5810 The default value is 50 which limits unit growth to 1.5 times the original
5813 @item max-inline-insns-recursive
5814 @itemx max-inline-insns-recursive-auto
5815 Specifies maximum number of instructions out-of-line copy of self recursive inline
5816 function can grow into by performing recursive inlining.
5818 For functions declared inline @option{--param max-inline-insns-recursive} is
5819 taken into acount. For function not declared inline, recursive inlining
5820 happens only when @option{-finline-functions} (included in @option{-O3}) is
5821 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5822 default value is 450.
5824 @item max-inline-recursive-depth
5825 @itemx max-inline-recursive-depth-auto
5826 Specifies maximum recursion depth used by the recursive inlining.
5828 For functions declared inline @option{--param max-inline-recursive-depth} is
5829 taken into acount. For function not declared inline, recursive inlining
5830 happens only when @option{-finline-functions} (included in @option{-O3}) is
5831 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5832 default value is 450.
5834 @item min-inline-recursive-probability
5835 Recursive inlining is profitable only for function having deep recursion
5836 in average and can hurt for function having little recursion depth by
5837 increasing the prologue size or complexity of function body to other
5840 When profile feedback is available (see @option{-fprofile-generate}) the actual
5841 recursion depth can be guessed from probability that function will recurse via
5842 given call expression. This parameter limits inlining only to call expression
5843 whose probability exceeds given threshold (in percents). The default value is
5846 @item inline-call-cost
5847 Specify cost of call instruction relative to simple arithmetics operations
5848 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
5849 functions and at the same time increases size of leaf function that is believed to
5850 reduce function size by being inlined. In effect it increases amount of
5851 inlining for code having large abstraction penalty (many functions that just
5852 pass the arguments to other functions) and decrease inlining for code with low
5853 abstraction penalty. The default value is 16.
5855 @item max-unrolled-insns
5856 The maximum number of instructions that a loop should have if that loop
5857 is unrolled, and if the loop is unrolled, it determines how many times
5858 the loop code is unrolled.
5860 @item max-average-unrolled-insns
5861 The maximum number of instructions biased by probabilities of their execution
5862 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5863 it determines how many times the loop code is unrolled.
5865 @item max-unroll-times
5866 The maximum number of unrollings of a single loop.
5868 @item max-peeled-insns
5869 The maximum number of instructions that a loop should have if that loop
5870 is peeled, and if the loop is peeled, it determines how many times
5871 the loop code is peeled.
5873 @item max-peel-times
5874 The maximum number of peelings of a single loop.
5876 @item max-completely-peeled-insns
5877 The maximum number of insns of a completely peeled loop.
5879 @item max-completely-peel-times
5880 The maximum number of iterations of a loop to be suitable for complete peeling.
5882 @item max-unswitch-insns
5883 The maximum number of insns of an unswitched loop.
5885 @item max-unswitch-level
5886 The maximum number of branches unswitched in a single loop.
5889 The minimum cost of an expensive expression in the loop invariant motion.
5891 @item iv-consider-all-candidates-bound
5892 Bound on number of candidates for induction variables below that
5893 all candidates are considered for each use in induction variable
5894 optimizations. Only the most relevant candidates are considered
5895 if there are more candidates, to avoid quadratic time complexity.
5897 @item iv-max-considered-uses
5898 The induction variable optimizations give up on loops that contain more
5899 induction variable uses.
5901 @item iv-always-prune-cand-set-bound
5902 If number of candidates in the set is smaller than this value,
5903 we always try to remove unnecessary ivs from the set during its
5904 optimization when a new iv is added to the set.
5906 @item scev-max-expr-size
5907 Bound on size of expressions used in the scalar evolutions analyzer.
5908 Large expressions slow the analyzer.
5910 @item vect-max-version-checks
5911 The maxinum number of runtime checks that can be performed when doing
5912 loop versioning in the vectorizer. See option ftree-vect-loop-version
5913 for more information.
5915 @item max-iterations-to-track
5917 The maximum number of iterations of a loop the brute force algorithm
5918 for analysis of # of iterations of the loop tries to evaluate.
5920 @item hot-bb-count-fraction
5921 Select fraction of the maximal count of repetitions of basic block in program
5922 given basic block needs to have to be considered hot.
5924 @item hot-bb-frequency-fraction
5925 Select fraction of the maximal frequency of executions of basic block in
5926 function given basic block needs to have to be considered hot
5928 @item tracer-dynamic-coverage
5929 @itemx tracer-dynamic-coverage-feedback
5931 This value is used to limit superblock formation once the given percentage of
5932 executed instructions is covered. This limits unnecessary code size
5935 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5936 feedback is available. The real profiles (as opposed to statically estimated
5937 ones) are much less balanced allowing the threshold to be larger value.
5939 @item tracer-max-code-growth
5940 Stop tail duplication once code growth has reached given percentage. This is
5941 rather hokey argument, as most of the duplicates will be eliminated later in
5942 cross jumping, so it may be set to much higher values than is the desired code
5945 @item tracer-min-branch-ratio
5947 Stop reverse growth when the reverse probability of best edge is less than this
5948 threshold (in percent).
5950 @item tracer-min-branch-ratio
5951 @itemx tracer-min-branch-ratio-feedback
5953 Stop forward growth if the best edge do have probability lower than this
5956 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5957 compilation for profile feedback and one for compilation without. The value
5958 for compilation with profile feedback needs to be more conservative (higher) in
5959 order to make tracer effective.
5961 @item max-cse-path-length
5963 Maximum number of basic blocks on path that cse considers. The default is 10.
5965 @item global-var-threshold
5967 Counts the number of function calls (@var{n}) and the number of
5968 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
5969 single artificial variable will be created to represent all the
5970 call-clobbered variables at function call sites. This artificial
5971 variable will then be made to alias every call-clobbered variable.
5972 (done as @code{int * size_t} on the host machine; beware overflow).
5974 @item max-aliased-vops
5976 Maximum number of virtual operands allowed to represent aliases
5977 before triggering the alias grouping heuristic. Alias grouping
5978 reduces compile times and memory consumption needed for aliasing at
5979 the expense of precision loss in alias information.
5981 @item ggc-min-expand
5983 GCC uses a garbage collector to manage its own memory allocation. This
5984 parameter specifies the minimum percentage by which the garbage
5985 collector's heap should be allowed to expand between collections.
5986 Tuning this may improve compilation speed; it has no effect on code
5989 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
5990 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
5991 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
5992 GCC is not able to calculate RAM on a particular platform, the lower
5993 bound of 30% is used. Setting this parameter and
5994 @option{ggc-min-heapsize} to zero causes a full collection to occur at
5995 every opportunity. This is extremely slow, but can be useful for
5998 @item ggc-min-heapsize
6000 Minimum size of the garbage collector's heap before it begins bothering
6001 to collect garbage. The first collection occurs after the heap expands
6002 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6003 tuning this may improve compilation speed, and has no effect on code
6006 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6007 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6008 with a lower bound of 4096 (four megabytes) and an upper bound of
6009 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6010 particular platform, the lower bound is used. Setting this parameter
6011 very large effectively disables garbage collection. Setting this
6012 parameter and @option{ggc-min-expand} to zero causes a full collection
6013 to occur at every opportunity.
6015 @item max-reload-search-insns
6016 The maximum number of instruction reload should look backward for equivalent
6017 register. Increasing values mean more aggressive optimization, making the
6018 compile time increase with probably slightly better performance. The default
6021 @item max-cselib-memory-location
6022 The maximum number of memory locations cselib should take into acount.
6023 Increasing values mean more aggressive optimization, making the compile time
6024 increase with probably slightly better performance. The default value is 500.
6026 @item reorder-blocks-duplicate
6027 @itemx reorder-blocks-duplicate-feedback
6029 Used by basic block reordering pass to decide whether to use unconditional
6030 branch or duplicate the code on its destination. Code is duplicated when its
6031 estimated size is smaller than this value multiplied by the estimated size of
6032 unconditional jump in the hot spots of the program.
6034 The @option{reorder-block-duplicate-feedback} is used only when profile
6035 feedback is available and may be set to higher values than
6036 @option{reorder-block-duplicate} since information about the hot spots is more
6039 @item max-sched-region-blocks
6040 The maximum number of blocks in a region to be considered for
6041 interblock scheduling. The default value is 10.
6043 @item max-sched-region-insns
6044 The maximum number of insns in a region to be considered for
6045 interblock scheduling. The default value is 100.
6047 @item min-sched-prob
6048 The minimum probability of reaching a source block for interblock
6049 speculative scheduling. The default value is 40.
6051 @item max-last-value-rtl
6053 The maximum size measured as number of RTLs that can be recorded in an expression
6054 in combiner for a pseudo register as last known value of that register. The default
6057 @item integer-share-limit
6058 Small integer constants can use a shared data structure, reducing the
6059 compiler's memory usage and increasing its speed. This sets the maximum
6060 value of a shared integer constant's. The default value is 256.
6062 @item min-virtual-mappings
6063 Specifies the minimum number of virtual mappings in the incremental
6064 SSA updater that should be registered to trigger the virtual mappings
6065 heuristic defined by virtual-mappings-ratio. The default value is
6068 @item virtual-mappings-ratio
6069 If the number of virtual mappings is virtual-mappings-ratio bigger
6070 than the number of virtual symbols to be updated, then the incremental
6071 SSA updater switches to a full update for those symbols. The default
6074 @item ssp-buffer-size
6075 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6076 protection when @option{-fstack-protection} is used.
6081 @node Preprocessor Options
6082 @section Options Controlling the Preprocessor
6083 @cindex preprocessor options
6084 @cindex options, preprocessor
6086 These options control the C preprocessor, which is run on each C source
6087 file before actual compilation.
6089 If you use the @option{-E} option, nothing is done except preprocessing.
6090 Some of these options make sense only together with @option{-E} because
6091 they cause the preprocessor output to be unsuitable for actual
6096 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6097 and pass @var{option} directly through to the preprocessor. If
6098 @var{option} contains commas, it is split into multiple options at the
6099 commas. However, many options are modified, translated or interpreted
6100 by the compiler driver before being passed to the preprocessor, and
6101 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6102 interface is undocumented and subject to change, so whenever possible
6103 you should avoid using @option{-Wp} and let the driver handle the
6106 @item -Xpreprocessor @var{option}
6107 @opindex preprocessor
6108 Pass @var{option} as an option to the preprocessor. You can use this to
6109 supply system-specific preprocessor options which GCC does not know how to
6112 If you want to pass an option that takes an argument, you must use
6113 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6116 @include cppopts.texi
6118 @node Assembler Options
6119 @section Passing Options to the Assembler
6121 @c prevent bad page break with this line
6122 You can pass options to the assembler.
6125 @item -Wa,@var{option}
6127 Pass @var{option} as an option to the assembler. If @var{option}
6128 contains commas, it is split into multiple options at the commas.
6130 @item -Xassembler @var{option}
6132 Pass @var{option} as an option to the assembler. You can use this to
6133 supply system-specific assembler options which GCC does not know how to
6136 If you want to pass an option that takes an argument, you must use
6137 @option{-Xassembler} twice, once for the option and once for the argument.
6142 @section Options for Linking
6143 @cindex link options
6144 @cindex options, linking
6146 These options come into play when the compiler links object files into
6147 an executable output file. They are meaningless if the compiler is
6148 not doing a link step.
6152 @item @var{object-file-name}
6153 A file name that does not end in a special recognized suffix is
6154 considered to name an object file or library. (Object files are
6155 distinguished from libraries by the linker according to the file
6156 contents.) If linking is done, these object files are used as input
6165 If any of these options is used, then the linker is not run, and
6166 object file names should not be used as arguments. @xref{Overall
6170 @item -l@var{library}
6171 @itemx -l @var{library}
6173 Search the library named @var{library} when linking. (The second
6174 alternative with the library as a separate argument is only for
6175 POSIX compliance and is not recommended.)
6177 It makes a difference where in the command you write this option; the
6178 linker searches and processes libraries and object files in the order they
6179 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6180 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6181 to functions in @samp{z}, those functions may not be loaded.
6183 The linker searches a standard list of directories for the library,
6184 which is actually a file named @file{lib@var{library}.a}. The linker
6185 then uses this file as if it had been specified precisely by name.
6187 The directories searched include several standard system directories
6188 plus any that you specify with @option{-L}.
6190 Normally the files found this way are library files---archive files
6191 whose members are object files. The linker handles an archive file by
6192 scanning through it for members which define symbols that have so far
6193 been referenced but not defined. But if the file that is found is an
6194 ordinary object file, it is linked in the usual fashion. The only
6195 difference between using an @option{-l} option and specifying a file name
6196 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6197 and searches several directories.
6201 You need this special case of the @option{-l} option in order to
6202 link an Objective-C or Objective-C++ program.
6205 @opindex nostartfiles
6206 Do not use the standard system startup files when linking.
6207 The standard system libraries are used normally, unless @option{-nostdlib}
6208 or @option{-nodefaultlibs} is used.
6210 @item -nodefaultlibs
6211 @opindex nodefaultlibs
6212 Do not use the standard system libraries when linking.
6213 Only the libraries you specify will be passed to the linker.
6214 The standard startup files are used normally, unless @option{-nostartfiles}
6215 is used. The compiler may generate calls to @code{memcmp},
6216 @code{memset}, @code{memcpy} and @code{memmove}.
6217 These entries are usually resolved by entries in
6218 libc. These entry points should be supplied through some other
6219 mechanism when this option is specified.
6223 Do not use the standard system startup files or libraries when linking.
6224 No startup files and only the libraries you specify will be passed to
6225 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6226 @code{memcpy} and @code{memmove}.
6227 These entries are usually resolved by entries in
6228 libc. These entry points should be supplied through some other
6229 mechanism when this option is specified.
6231 @cindex @option{-lgcc}, use with @option{-nostdlib}
6232 @cindex @option{-nostdlib} and unresolved references
6233 @cindex unresolved references and @option{-nostdlib}
6234 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6235 @cindex @option{-nodefaultlibs} and unresolved references
6236 @cindex unresolved references and @option{-nodefaultlibs}
6237 One of the standard libraries bypassed by @option{-nostdlib} and
6238 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6239 that GCC uses to overcome shortcomings of particular machines, or special
6240 needs for some languages.
6241 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6242 Collection (GCC) Internals},
6243 for more discussion of @file{libgcc.a}.)
6244 In most cases, you need @file{libgcc.a} even when you want to avoid
6245 other standard libraries. In other words, when you specify @option{-nostdlib}
6246 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6247 This ensures that you have no unresolved references to internal GCC
6248 library subroutines. (For example, @samp{__main}, used to ensure C++
6249 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6250 GNU Compiler Collection (GCC) Internals}.)
6254 Produce a position independent executable on targets which support it.
6255 For predictable results, you must also specify the same set of options
6256 that were used to generate code (@option{-fpie}, @option{-fPIE},
6257 or model suboptions) when you specify this option.
6261 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6262 that support it. This instructs the linker to add all symbols, not
6263 only used ones, to the dynamic symbol table. This option is needed
6264 for some uses of @code{dlopen} or to allow obtaining backtraces
6265 from within a program.
6269 Remove all symbol table and relocation information from the executable.
6273 On systems that support dynamic linking, this prevents linking with the shared
6274 libraries. On other systems, this option has no effect.
6278 Produce a shared object which can then be linked with other objects to
6279 form an executable. Not all systems support this option. For predictable
6280 results, you must also specify the same set of options that were used to
6281 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6282 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6283 needs to build supplementary stub code for constructors to work. On
6284 multi-libbed systems, @samp{gcc -shared} must select the correct support
6285 libraries to link against. Failing to supply the correct flags may lead
6286 to subtle defects. Supplying them in cases where they are not necessary
6289 @item -shared-libgcc
6290 @itemx -static-libgcc
6291 @opindex shared-libgcc
6292 @opindex static-libgcc
6293 On systems that provide @file{libgcc} as a shared library, these options
6294 force the use of either the shared or static version respectively.
6295 If no shared version of @file{libgcc} was built when the compiler was
6296 configured, these options have no effect.
6298 There are several situations in which an application should use the
6299 shared @file{libgcc} instead of the static version. The most common
6300 of these is when the application wishes to throw and catch exceptions
6301 across different shared libraries. In that case, each of the libraries
6302 as well as the application itself should use the shared @file{libgcc}.
6304 Therefore, the G++ and GCJ drivers automatically add
6305 @option{-shared-libgcc} whenever you build a shared library or a main
6306 executable, because C++ and Java programs typically use exceptions, so
6307 this is the right thing to do.
6309 If, instead, you use the GCC driver to create shared libraries, you may
6310 find that they will not always be linked with the shared @file{libgcc}.
6311 If GCC finds, at its configuration time, that you have a non-GNU linker
6312 or a GNU linker that does not support option @option{--eh-frame-hdr},
6313 it will link the shared version of @file{libgcc} into shared libraries
6314 by default. Otherwise, it will take advantage of the linker and optimize
6315 away the linking with the shared version of @file{libgcc}, linking with
6316 the static version of libgcc by default. This allows exceptions to
6317 propagate through such shared libraries, without incurring relocation
6318 costs at library load time.
6320 However, if a library or main executable is supposed to throw or catch
6321 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6322 for the languages used in the program, or using the option
6323 @option{-shared-libgcc}, such that it is linked with the shared
6328 Bind references to global symbols when building a shared object. Warn
6329 about any unresolved references (unless overridden by the link editor
6330 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6333 @item -Xlinker @var{option}
6335 Pass @var{option} as an option to the linker. You can use this to
6336 supply system-specific linker options which GCC does not know how to
6339 If you want to pass an option that takes an argument, you must use
6340 @option{-Xlinker} twice, once for the option and once for the argument.
6341 For example, to pass @option{-assert definitions}, you must write
6342 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6343 @option{-Xlinker "-assert definitions"}, because this passes the entire
6344 string as a single argument, which is not what the linker expects.
6346 @item -Wl,@var{option}
6348 Pass @var{option} as an option to the linker. If @var{option} contains
6349 commas, it is split into multiple options at the commas.
6351 @item -u @var{symbol}
6353 Pretend the symbol @var{symbol} is undefined, to force linking of
6354 library modules to define it. You can use @option{-u} multiple times with
6355 different symbols to force loading of additional library modules.
6358 @node Directory Options
6359 @section Options for Directory Search
6360 @cindex directory options
6361 @cindex options, directory search
6364 These options specify directories to search for header files, for
6365 libraries and for parts of the compiler:
6370 Add the directory @var{dir} to the head of the list of directories to be
6371 searched for header files. This can be used to override a system header
6372 file, substituting your own version, since these directories are
6373 searched before the system header file directories. However, you should
6374 not use this option to add directories that contain vendor-supplied
6375 system header files (use @option{-isystem} for that). If you use more than
6376 one @option{-I} option, the directories are scanned in left-to-right
6377 order; the standard system directories come after.
6379 If a standard system include directory, or a directory specified with
6380 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6381 option will be ignored. The directory will still be searched but as a
6382 system directory at its normal position in the system include chain.
6383 This is to ensure that GCC's procedure to fix buggy system headers and
6384 the ordering for the include_next directive are not inadvertently changed.
6385 If you really need to change the search order for system directories,
6386 use the @option{-nostdinc} and/or @option{-isystem} options.
6388 @item -iquote@var{dir}
6390 Add the directory @var{dir} to the head of the list of directories to
6391 be searched for header files only for the case of @samp{#include
6392 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6393 otherwise just like @option{-I}.
6397 Add directory @var{dir} to the list of directories to be searched
6400 @item -B@var{prefix}
6402 This option specifies where to find the executables, libraries,
6403 include files, and data files of the compiler itself.
6405 The compiler driver program runs one or more of the subprograms
6406 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6407 @var{prefix} as a prefix for each program it tries to run, both with and
6408 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6410 For each subprogram to be run, the compiler driver first tries the
6411 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6412 was not specified, the driver tries two standard prefixes, which are
6413 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6414 those results in a file name that is found, the unmodified program
6415 name is searched for using the directories specified in your
6416 @env{PATH} environment variable.
6418 The compiler will check to see if the path provided by the @option{-B}
6419 refers to a directory, and if necessary it will add a directory
6420 separator character at the end of the path.
6422 @option{-B} prefixes that effectively specify directory names also apply
6423 to libraries in the linker, because the compiler translates these
6424 options into @option{-L} options for the linker. They also apply to
6425 includes files in the preprocessor, because the compiler translates these
6426 options into @option{-isystem} options for the preprocessor. In this case,
6427 the compiler appends @samp{include} to the prefix.
6429 The run-time support file @file{libgcc.a} can also be searched for using
6430 the @option{-B} prefix, if needed. If it is not found there, the two
6431 standard prefixes above are tried, and that is all. The file is left
6432 out of the link if it is not found by those means.
6434 Another way to specify a prefix much like the @option{-B} prefix is to use
6435 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6438 As a special kludge, if the path provided by @option{-B} is
6439 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6440 9, then it will be replaced by @file{[dir/]include}. This is to help
6441 with boot-strapping the compiler.
6443 @item -specs=@var{file}
6445 Process @var{file} after the compiler reads in the standard @file{specs}
6446 file, in order to override the defaults that the @file{gcc} driver
6447 program uses when determining what switches to pass to @file{cc1},
6448 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6449 @option{-specs=@var{file}} can be specified on the command line, and they
6450 are processed in order, from left to right.
6452 @item --sysroot=@var{dir}
6454 Use @var{dir} as the logical root directory for headers and libraries.
6455 For example, if the compiler would normally search for headers in
6456 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6457 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6459 If you use both this option and the @option{-isysroot} option, then
6460 the @option{--sysroot} option will apply to libraries, but the
6461 @option{-isysroot} option will apply to header files.
6463 The GNU linker (beginning with version 2.16) has the necessary support
6464 for this option. If your linker does not support this option, the
6465 header file aspect of @option{--sysroot} will still work, but the
6466 library aspect will not.
6470 This option has been deprecated. Please use @option{-iquote} instead for
6471 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6472 Any directories you specify with @option{-I} options before the @option{-I-}
6473 option are searched only for the case of @samp{#include "@var{file}"};
6474 they are not searched for @samp{#include <@var{file}>}.
6476 If additional directories are specified with @option{-I} options after
6477 the @option{-I-}, these directories are searched for all @samp{#include}
6478 directives. (Ordinarily @emph{all} @option{-I} directories are used
6481 In addition, the @option{-I-} option inhibits the use of the current
6482 directory (where the current input file came from) as the first search
6483 directory for @samp{#include "@var{file}"}. There is no way to
6484 override this effect of @option{-I-}. With @option{-I.} you can specify
6485 searching the directory which was current when the compiler was
6486 invoked. That is not exactly the same as what the preprocessor does
6487 by default, but it is often satisfactory.
6489 @option{-I-} does not inhibit the use of the standard system directories
6490 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6497 @section Specifying subprocesses and the switches to pass to them
6500 @command{gcc} is a driver program. It performs its job by invoking a
6501 sequence of other programs to do the work of compiling, assembling and
6502 linking. GCC interprets its command-line parameters and uses these to
6503 deduce which programs it should invoke, and which command-line options
6504 it ought to place on their command lines. This behavior is controlled
6505 by @dfn{spec strings}. In most cases there is one spec string for each
6506 program that GCC can invoke, but a few programs have multiple spec
6507 strings to control their behavior. The spec strings built into GCC can
6508 be overridden by using the @option{-specs=} command-line switch to specify
6511 @dfn{Spec files} are plaintext files that are used to construct spec
6512 strings. They consist of a sequence of directives separated by blank
6513 lines. The type of directive is determined by the first non-whitespace
6514 character on the line and it can be one of the following:
6517 @item %@var{command}
6518 Issues a @var{command} to the spec file processor. The commands that can
6522 @item %include <@var{file}>
6524 Search for @var{file} and insert its text at the current point in the
6527 @item %include_noerr <@var{file}>
6528 @cindex %include_noerr
6529 Just like @samp{%include}, but do not generate an error message if the include
6530 file cannot be found.
6532 @item %rename @var{old_name} @var{new_name}
6534 Rename the spec string @var{old_name} to @var{new_name}.
6538 @item *[@var{spec_name}]:
6539 This tells the compiler to create, override or delete the named spec
6540 string. All lines after this directive up to the next directive or
6541 blank line are considered to be the text for the spec string. If this
6542 results in an empty string then the spec will be deleted. (Or, if the
6543 spec did not exist, then nothing will happened.) Otherwise, if the spec
6544 does not currently exist a new spec will be created. If the spec does
6545 exist then its contents will be overridden by the text of this
6546 directive, unless the first character of that text is the @samp{+}
6547 character, in which case the text will be appended to the spec.
6549 @item [@var{suffix}]:
6550 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6551 and up to the next directive or blank line are considered to make up the
6552 spec string for the indicated suffix. When the compiler encounters an
6553 input file with the named suffix, it will processes the spec string in
6554 order to work out how to compile that file. For example:
6561 This says that any input file whose name ends in @samp{.ZZ} should be
6562 passed to the program @samp{z-compile}, which should be invoked with the
6563 command-line switch @option{-input} and with the result of performing the
6564 @samp{%i} substitution. (See below.)
6566 As an alternative to providing a spec string, the text that follows a
6567 suffix directive can be one of the following:
6570 @item @@@var{language}
6571 This says that the suffix is an alias for a known @var{language}. This is
6572 similar to using the @option{-x} command-line switch to GCC to specify a
6573 language explicitly. For example:
6580 Says that .ZZ files are, in fact, C++ source files.
6583 This causes an error messages saying:
6586 @var{name} compiler not installed on this system.
6590 GCC already has an extensive list of suffixes built into it.
6591 This directive will add an entry to the end of the list of suffixes, but
6592 since the list is searched from the end backwards, it is effectively
6593 possible to override earlier entries using this technique.
6597 GCC has the following spec strings built into it. Spec files can
6598 override these strings or create their own. Note that individual
6599 targets can also add their own spec strings to this list.
6602 asm Options to pass to the assembler
6603 asm_final Options to pass to the assembler post-processor
6604 cpp Options to pass to the C preprocessor
6605 cc1 Options to pass to the C compiler
6606 cc1plus Options to pass to the C++ compiler
6607 endfile Object files to include at the end of the link
6608 link Options to pass to the linker
6609 lib Libraries to include on the command line to the linker
6610 libgcc Decides which GCC support library to pass to the linker
6611 linker Sets the name of the linker
6612 predefines Defines to be passed to the C preprocessor
6613 signed_char Defines to pass to CPP to say whether @code{char} is signed
6615 startfile Object files to include at the start of the link
6618 Here is a small example of a spec file:
6624 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6627 This example renames the spec called @samp{lib} to @samp{old_lib} and
6628 then overrides the previous definition of @samp{lib} with a new one.
6629 The new definition adds in some extra command-line options before
6630 including the text of the old definition.
6632 @dfn{Spec strings} are a list of command-line options to be passed to their
6633 corresponding program. In addition, the spec strings can contain
6634 @samp{%}-prefixed sequences to substitute variable text or to
6635 conditionally insert text into the command line. Using these constructs
6636 it is possible to generate quite complex command lines.
6638 Here is a table of all defined @samp{%}-sequences for spec
6639 strings. Note that spaces are not generated automatically around the
6640 results of expanding these sequences. Therefore you can concatenate them
6641 together or combine them with constant text in a single argument.
6645 Substitute one @samp{%} into the program name or argument.
6648 Substitute the name of the input file being processed.
6651 Substitute the basename of the input file being processed.
6652 This is the substring up to (and not including) the last period
6653 and not including the directory.
6656 This is the same as @samp{%b}, but include the file suffix (text after
6660 Marks the argument containing or following the @samp{%d} as a
6661 temporary file name, so that that file will be deleted if GCC exits
6662 successfully. Unlike @samp{%g}, this contributes no text to the
6665 @item %g@var{suffix}
6666 Substitute a file name that has suffix @var{suffix} and is chosen
6667 once per compilation, and mark the argument in the same way as
6668 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6669 name is now chosen in a way that is hard to predict even when previously
6670 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6671 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6672 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6673 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6674 was simply substituted with a file name chosen once per compilation,
6675 without regard to any appended suffix (which was therefore treated
6676 just like ordinary text), making such attacks more likely to succeed.
6678 @item %u@var{suffix}
6679 Like @samp{%g}, but generates a new temporary file name even if
6680 @samp{%u@var{suffix}} was already seen.
6682 @item %U@var{suffix}
6683 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6684 new one if there is no such last file name. In the absence of any
6685 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6686 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6687 would involve the generation of two distinct file names, one
6688 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6689 simply substituted with a file name chosen for the previous @samp{%u},
6690 without regard to any appended suffix.
6692 @item %j@var{suffix}
6693 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6694 writable, and if save-temps is off; otherwise, substitute the name
6695 of a temporary file, just like @samp{%u}. This temporary file is not
6696 meant for communication between processes, but rather as a junk
6699 @item %|@var{suffix}
6700 @itemx %m@var{suffix}
6701 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6702 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6703 all. These are the two most common ways to instruct a program that it
6704 should read from standard input or write to standard output. If you
6705 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6706 construct: see for example @file{f/lang-specs.h}.
6708 @item %.@var{SUFFIX}
6709 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6710 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6711 terminated by the next space or %.
6714 Marks the argument containing or following the @samp{%w} as the
6715 designated output file of this compilation. This puts the argument
6716 into the sequence of arguments that @samp{%o} will substitute later.
6719 Substitutes the names of all the output files, with spaces
6720 automatically placed around them. You should write spaces
6721 around the @samp{%o} as well or the results are undefined.
6722 @samp{%o} is for use in the specs for running the linker.
6723 Input files whose names have no recognized suffix are not compiled
6724 at all, but they are included among the output files, so they will
6728 Substitutes the suffix for object files. Note that this is
6729 handled specially when it immediately follows @samp{%g, %u, or %U},
6730 because of the need for those to form complete file names. The
6731 handling is such that @samp{%O} is treated exactly as if it had already
6732 been substituted, except that @samp{%g, %u, and %U} do not currently
6733 support additional @var{suffix} characters following @samp{%O} as they would
6734 following, for example, @samp{.o}.
6737 Substitutes the standard macro predefinitions for the
6738 current target machine. Use this when running @code{cpp}.
6741 Like @samp{%p}, but puts @samp{__} before and after the name of each
6742 predefined macro, except for macros that start with @samp{__} or with
6743 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6747 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6748 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
6749 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6753 Current argument is the name of a library or startup file of some sort.
6754 Search for that file in a standard list of directories and substitute
6755 the full name found.
6758 Print @var{str} as an error message. @var{str} is terminated by a newline.
6759 Use this when inconsistent options are detected.
6762 Substitute the contents of spec string @var{name} at this point.
6765 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6767 @item %x@{@var{option}@}
6768 Accumulate an option for @samp{%X}.
6771 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6775 Output the accumulated assembler options specified by @option{-Wa}.
6778 Output the accumulated preprocessor options specified by @option{-Wp}.
6781 Process the @code{asm} spec. This is used to compute the
6782 switches to be passed to the assembler.
6785 Process the @code{asm_final} spec. This is a spec string for
6786 passing switches to an assembler post-processor, if such a program is
6790 Process the @code{link} spec. This is the spec for computing the
6791 command line passed to the linker. Typically it will make use of the
6792 @samp{%L %G %S %D and %E} sequences.
6795 Dump out a @option{-L} option for each directory that GCC believes might
6796 contain startup files. If the target supports multilibs then the
6797 current multilib directory will be prepended to each of these paths.
6800 Process the @code{lib} spec. This is a spec string for deciding which
6801 libraries should be included on the command line to the linker.
6804 Process the @code{libgcc} spec. This is a spec string for deciding
6805 which GCC support library should be included on the command line to the linker.
6808 Process the @code{startfile} spec. This is a spec for deciding which
6809 object files should be the first ones passed to the linker. Typically
6810 this might be a file named @file{crt0.o}.
6813 Process the @code{endfile} spec. This is a spec string that specifies
6814 the last object files that will be passed to the linker.
6817 Process the @code{cpp} spec. This is used to construct the arguments
6818 to be passed to the C preprocessor.
6821 Process the @code{cc1} spec. This is used to construct the options to be
6822 passed to the actual C compiler (@samp{cc1}).
6825 Process the @code{cc1plus} spec. This is used to construct the options to be
6826 passed to the actual C++ compiler (@samp{cc1plus}).
6829 Substitute the variable part of a matched option. See below.
6830 Note that each comma in the substituted string is replaced by
6834 Remove all occurrences of @code{-S} from the command line. Note---this
6835 command is position dependent. @samp{%} commands in the spec string
6836 before this one will see @code{-S}, @samp{%} commands in the spec string
6837 after this one will not.
6839 @item %:@var{function}(@var{args})
6840 Call the named function @var{function}, passing it @var{args}.
6841 @var{args} is first processed as a nested spec string, then split
6842 into an argument vector in the usual fashion. The function returns
6843 a string which is processed as if it had appeared literally as part
6844 of the current spec.
6846 The following built-in spec functions are provided:
6849 @item @code{if-exists}
6850 The @code{if-exists} spec function takes one argument, an absolute
6851 pathname to a file. If the file exists, @code{if-exists} returns the
6852 pathname. Here is a small example of its usage:
6856 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6859 @item @code{if-exists-else}
6860 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6861 spec function, except that it takes two arguments. The first argument is
6862 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6863 returns the pathname. If it does not exist, it returns the second argument.
6864 This way, @code{if-exists-else} can be used to select one file or another,
6865 based on the existence of the first. Here is a small example of its usage:
6869 crt0%O%s %:if-exists(crti%O%s) \
6870 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6873 @item @code{replace-outfile}
6874 The @code{replace-outfile} spec function takes two arguments. It looks for the
6875 first argument in the outfiles array and replaces it with the second argument. Here
6876 is a small example of its usage:
6879 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
6885 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6886 If that switch was not specified, this substitutes nothing. Note that
6887 the leading dash is omitted when specifying this option, and it is
6888 automatically inserted if the substitution is performed. Thus the spec
6889 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6890 and would output the command line option @option{-foo}.
6892 @item %W@{@code{S}@}
6893 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6896 @item %@{@code{S}*@}
6897 Substitutes all the switches specified to GCC whose names start
6898 with @code{-S}, but which also take an argument. This is used for
6899 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6900 GCC considers @option{-o foo} as being
6901 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6902 text, including the space. Thus two arguments would be generated.
6904 @item %@{@code{S}*&@code{T}*@}
6905 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6906 (the order of @code{S} and @code{T} in the spec is not significant).
6907 There can be any number of ampersand-separated variables; for each the
6908 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6910 @item %@{@code{S}:@code{X}@}
6911 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6913 @item %@{!@code{S}:@code{X}@}
6914 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6916 @item %@{@code{S}*:@code{X}@}
6917 Substitutes @code{X} if one or more switches whose names start with
6918 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6919 once, no matter how many such switches appeared. However, if @code{%*}
6920 appears somewhere in @code{X}, then @code{X} will be substituted once
6921 for each matching switch, with the @code{%*} replaced by the part of
6922 that switch that matched the @code{*}.
6924 @item %@{.@code{S}:@code{X}@}
6925 Substitutes @code{X}, if processing a file with suffix @code{S}.
6927 @item %@{!.@code{S}:@code{X}@}
6928 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6930 @item %@{@code{S}|@code{P}:@code{X}@}
6931 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6932 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6933 although they have a stronger binding than the @samp{|}. If @code{%*}
6934 appears in @code{X}, all of the alternatives must be starred, and only
6935 the first matching alternative is substituted.
6937 For example, a spec string like this:
6940 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6943 will output the following command-line options from the following input
6944 command-line options:
6949 -d fred.c -foo -baz -boggle
6950 -d jim.d -bar -baz -boggle
6953 @item %@{S:X; T:Y; :D@}
6955 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6956 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6957 be as many clauses as you need. This may be combined with @code{.},
6958 @code{!}, @code{|}, and @code{*} as needed.
6963 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6964 construct may contain other nested @samp{%} constructs or spaces, or
6965 even newlines. They are processed as usual, as described above.
6966 Trailing white space in @code{X} is ignored. White space may also
6967 appear anywhere on the left side of the colon in these constructs,
6968 except between @code{.} or @code{*} and the corresponding word.
6970 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6971 handled specifically in these constructs. If another value of
6972 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
6973 @option{-W} switch is found later in the command line, the earlier
6974 switch value is ignored, except with @{@code{S}*@} where @code{S} is
6975 just one letter, which passes all matching options.
6977 The character @samp{|} at the beginning of the predicate text is used to
6978 indicate that a command should be piped to the following command, but
6979 only if @option{-pipe} is specified.
6981 It is built into GCC which switches take arguments and which do not.
6982 (You might think it would be useful to generalize this to allow each
6983 compiler's spec to say which switches take arguments. But this cannot
6984 be done in a consistent fashion. GCC cannot even decide which input
6985 files have been specified without knowing which switches take arguments,
6986 and it must know which input files to compile in order to tell which
6989 GCC also knows implicitly that arguments starting in @option{-l} are to be
6990 treated as compiler output files, and passed to the linker in their
6991 proper position among the other output files.
6993 @c man begin OPTIONS
6995 @node Target Options
6996 @section Specifying Target Machine and Compiler Version
6997 @cindex target options
6998 @cindex cross compiling
6999 @cindex specifying machine version
7000 @cindex specifying compiler version and target machine
7001 @cindex compiler version, specifying
7002 @cindex target machine, specifying
7004 The usual way to run GCC is to run the executable called @file{gcc}, or
7005 @file{<machine>-gcc} when cross-compiling, or
7006 @file{<machine>-gcc-<version>} to run a version other than the one that
7007 was installed last. Sometimes this is inconvenient, so GCC provides
7008 options that will switch to another cross-compiler or version.
7011 @item -b @var{machine}
7013 The argument @var{machine} specifies the target machine for compilation.
7015 The value to use for @var{machine} is the same as was specified as the
7016 machine type when configuring GCC as a cross-compiler. For
7017 example, if a cross-compiler was configured with @samp{configure
7018 arm-elf}, meaning to compile for an arm processor with elf binaries,
7019 then you would specify @option{-b arm-elf} to run that cross compiler.
7020 Because there are other options beginning with @option{-b}, the
7021 configuration must contain a hyphen.
7023 @item -V @var{version}
7025 The argument @var{version} specifies which version of GCC to run.
7026 This is useful when multiple versions are installed. For example,
7027 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7030 The @option{-V} and @option{-b} options work by running the
7031 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7032 use them if you can just run that directly.
7034 @node Submodel Options
7035 @section Hardware Models and Configurations
7036 @cindex submodel options
7037 @cindex specifying hardware config
7038 @cindex hardware models and configurations, specifying
7039 @cindex machine dependent options
7041 Earlier we discussed the standard option @option{-b} which chooses among
7042 different installed compilers for completely different target
7043 machines, such as VAX vs.@: 68000 vs.@: 80386.
7045 In addition, each of these target machine types can have its own
7046 special options, starting with @samp{-m}, to choose among various
7047 hardware models or configurations---for example, 68010 vs 68020,
7048 floating coprocessor or none. A single installed version of the
7049 compiler can compile for any model or configuration, according to the
7052 Some configurations of the compiler also support additional special
7053 options, usually for compatibility with other compilers on the same
7056 @c This list is ordered alphanumerically by subsection name.
7057 @c It should be the same order and spelling as these options are listed
7058 @c in Machine Dependent Options
7064 * Blackfin Options::
7068 * DEC Alpha Options::
7069 * DEC Alpha/VMS Options::
7073 * i386 and x86-64 Options::
7086 * RS/6000 and PowerPC Options::
7087 * S/390 and zSeries Options::
7090 * System V Options::
7091 * TMS320C3x/C4x Options::
7095 * Xstormy16 Options::
7101 @subsection ARC Options
7104 These options are defined for ARC implementations:
7109 Compile code for little endian mode. This is the default.
7113 Compile code for big endian mode.
7116 @opindex mmangle-cpu
7117 Prepend the name of the cpu to all public symbol names.
7118 In multiple-processor systems, there are many ARC variants with different
7119 instruction and register set characteristics. This flag prevents code
7120 compiled for one cpu to be linked with code compiled for another.
7121 No facility exists for handling variants that are ``almost identical''.
7122 This is an all or nothing option.
7124 @item -mcpu=@var{cpu}
7126 Compile code for ARC variant @var{cpu}.
7127 Which variants are supported depend on the configuration.
7128 All variants support @option{-mcpu=base}, this is the default.
7130 @item -mtext=@var{text-section}
7131 @itemx -mdata=@var{data-section}
7132 @itemx -mrodata=@var{readonly-data-section}
7136 Put functions, data, and readonly data in @var{text-section},
7137 @var{data-section}, and @var{readonly-data-section} respectively
7138 by default. This can be overridden with the @code{section} attribute.
7139 @xref{Variable Attributes}.
7144 @subsection ARM Options
7147 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7151 @item -mabi=@var{name}
7153 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7154 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
7157 @opindex mapcs-frame
7158 Generate a stack frame that is compliant with the ARM Procedure Call
7159 Standard for all functions, even if this is not strictly necessary for
7160 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7161 with this option will cause the stack frames not to be generated for
7162 leaf functions. The default is @option{-mno-apcs-frame}.
7166 This is a synonym for @option{-mapcs-frame}.
7169 @c not currently implemented
7170 @item -mapcs-stack-check
7171 @opindex mapcs-stack-check
7172 Generate code to check the amount of stack space available upon entry to
7173 every function (that actually uses some stack space). If there is
7174 insufficient space available then either the function
7175 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7176 called, depending upon the amount of stack space required. The run time
7177 system is required to provide these functions. The default is
7178 @option{-mno-apcs-stack-check}, since this produces smaller code.
7180 @c not currently implemented
7182 @opindex mapcs-float
7183 Pass floating point arguments using the float point registers. This is
7184 one of the variants of the APCS@. This option is recommended if the
7185 target hardware has a floating point unit or if a lot of floating point
7186 arithmetic is going to be performed by the code. The default is
7187 @option{-mno-apcs-float}, since integer only code is slightly increased in
7188 size if @option{-mapcs-float} is used.
7190 @c not currently implemented
7191 @item -mapcs-reentrant
7192 @opindex mapcs-reentrant
7193 Generate reentrant, position independent code. The default is
7194 @option{-mno-apcs-reentrant}.
7197 @item -mthumb-interwork
7198 @opindex mthumb-interwork
7199 Generate code which supports calling between the ARM and Thumb
7200 instruction sets. Without this option the two instruction sets cannot
7201 be reliably used inside one program. The default is
7202 @option{-mno-thumb-interwork}, since slightly larger code is generated
7203 when @option{-mthumb-interwork} is specified.
7205 @item -mno-sched-prolog
7206 @opindex mno-sched-prolog
7207 Prevent the reordering of instructions in the function prolog, or the
7208 merging of those instruction with the instructions in the function's
7209 body. This means that all functions will start with a recognizable set
7210 of instructions (or in fact one of a choice from a small set of
7211 different function prologues), and this information can be used to
7212 locate the start if functions inside an executable piece of code. The
7213 default is @option{-msched-prolog}.
7216 @opindex mhard-float
7217 Generate output containing floating point instructions. This is the
7221 @opindex msoft-float
7222 Generate output containing library calls for floating point.
7223 @strong{Warning:} the requisite libraries are not available for all ARM
7224 targets. Normally the facilities of the machine's usual C compiler are
7225 used, but this cannot be done directly in cross-compilation. You must make
7226 your own arrangements to provide suitable library functions for
7229 @option{-msoft-float} changes the calling convention in the output file;
7230 therefore, it is only useful if you compile @emph{all} of a program with
7231 this option. In particular, you need to compile @file{libgcc.a}, the
7232 library that comes with GCC, with @option{-msoft-float} in order for
7235 @item -mfloat-abi=@var{name}
7237 Specifies which ABI to use for floating point values. Permissible values
7238 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7240 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7241 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7242 of floating point instructions, but still uses the soft-float calling
7245 @item -mlittle-endian
7246 @opindex mlittle-endian
7247 Generate code for a processor running in little-endian mode. This is
7248 the default for all standard configurations.
7251 @opindex mbig-endian
7252 Generate code for a processor running in big-endian mode; the default is
7253 to compile code for a little-endian processor.
7255 @item -mwords-little-endian
7256 @opindex mwords-little-endian
7257 This option only applies when generating code for big-endian processors.
7258 Generate code for a little-endian word order but a big-endian byte
7259 order. That is, a byte order of the form @samp{32107654}. Note: this
7260 option should only be used if you require compatibility with code for
7261 big-endian ARM processors generated by versions of the compiler prior to
7264 @item -mcpu=@var{name}
7266 This specifies the name of the target ARM processor. GCC uses this name
7267 to determine what kind of instructions it can emit when generating
7268 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7269 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7270 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7271 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7272 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7273 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7274 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7275 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7276 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7277 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7278 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7279 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7280 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7281 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7284 @itemx -mtune=@var{name}
7286 This option is very similar to the @option{-mcpu=} option, except that
7287 instead of specifying the actual target processor type, and hence
7288 restricting which instructions can be used, it specifies that GCC should
7289 tune the performance of the code as if the target were of the type
7290 specified in this option, but still choosing the instructions that it
7291 will generate based on the cpu specified by a @option{-mcpu=} option.
7292 For some ARM implementations better performance can be obtained by using
7295 @item -march=@var{name}
7297 This specifies the name of the target ARM architecture. GCC uses this
7298 name to determine what kind of instructions it can emit when generating
7299 assembly code. This option can be used in conjunction with or instead
7300 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7301 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7302 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7303 @samp{iwmmxt}, @samp{ep9312}.
7305 @item -mfpu=@var{name}
7306 @itemx -mfpe=@var{number}
7307 @itemx -mfp=@var{number}
7311 This specifies what floating point hardware (or hardware emulation) is
7312 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7313 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7314 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7315 with older versions of GCC@.
7317 If @option{-msoft-float} is specified this specifies the format of
7318 floating point values.
7320 @item -mstructure-size-boundary=@var{n}
7321 @opindex mstructure-size-boundary
7322 The size of all structures and unions will be rounded up to a multiple
7323 of the number of bits set by this option. Permissible values are 8, 32
7324 and 64. The default value varies for different toolchains. For the COFF
7325 targeted toolchain the default value is 8. A value of 64 is only allowed
7326 if the underlying ABI supports it.
7328 Specifying the larger number can produce faster, more efficient code, but
7329 can also increase the size of the program. Different values are potentially
7330 incompatible. Code compiled with one value cannot necessarily expect to
7331 work with code or libraries compiled with another value, if they exchange
7332 information using structures or unions.
7334 @item -mabort-on-noreturn
7335 @opindex mabort-on-noreturn
7336 Generate a call to the function @code{abort} at the end of a
7337 @code{noreturn} function. It will be executed if the function tries to
7341 @itemx -mno-long-calls
7342 @opindex mlong-calls
7343 @opindex mno-long-calls
7344 Tells the compiler to perform function calls by first loading the
7345 address of the function into a register and then performing a subroutine
7346 call on this register. This switch is needed if the target function
7347 will lie outside of the 64 megabyte addressing range of the offset based
7348 version of subroutine call instruction.
7350 Even if this switch is enabled, not all function calls will be turned
7351 into long calls. The heuristic is that static functions, functions
7352 which have the @samp{short-call} attribute, functions that are inside
7353 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7354 definitions have already been compiled within the current compilation
7355 unit, will not be turned into long calls. The exception to this rule is
7356 that weak function definitions, functions with the @samp{long-call}
7357 attribute or the @samp{section} attribute, and functions that are within
7358 the scope of a @samp{#pragma long_calls} directive, will always be
7359 turned into long calls.
7361 This feature is not enabled by default. Specifying
7362 @option{-mno-long-calls} will restore the default behavior, as will
7363 placing the function calls within the scope of a @samp{#pragma
7364 long_calls_off} directive. Note these switches have no effect on how
7365 the compiler generates code to handle function calls via function
7368 @item -mnop-fun-dllimport
7369 @opindex mnop-fun-dllimport
7370 Disable support for the @code{dllimport} attribute.
7372 @item -msingle-pic-base
7373 @opindex msingle-pic-base
7374 Treat the register used for PIC addressing as read-only, rather than
7375 loading it in the prologue for each function. The run-time system is
7376 responsible for initializing this register with an appropriate value
7377 before execution begins.
7379 @item -mpic-register=@var{reg}
7380 @opindex mpic-register
7381 Specify the register to be used for PIC addressing. The default is R10
7382 unless stack-checking is enabled, when R9 is used.
7384 @item -mcirrus-fix-invalid-insns
7385 @opindex mcirrus-fix-invalid-insns
7386 @opindex mno-cirrus-fix-invalid-insns
7387 Insert NOPs into the instruction stream to in order to work around
7388 problems with invalid Maverick instruction combinations. This option
7389 is only valid if the @option{-mcpu=ep9312} option has been used to
7390 enable generation of instructions for the Cirrus Maverick floating
7391 point co-processor. This option is not enabled by default, since the
7392 problem is only present in older Maverick implementations. The default
7393 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7396 @item -mpoke-function-name
7397 @opindex mpoke-function-name
7398 Write the name of each function into the text section, directly
7399 preceding the function prologue. The generated code is similar to this:
7403 .ascii "arm_poke_function_name", 0
7406 .word 0xff000000 + (t1 - t0)
7407 arm_poke_function_name
7409 stmfd sp!, @{fp, ip, lr, pc@}
7413 When performing a stack backtrace, code can inspect the value of
7414 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7415 location @code{pc - 12} and the top 8 bits are set, then we know that
7416 there is a function name embedded immediately preceding this location
7417 and has length @code{((pc[-3]) & 0xff000000)}.
7421 Generate code for the 16-bit Thumb instruction set. The default is to
7422 use the 32-bit ARM instruction set.
7425 @opindex mtpcs-frame
7426 Generate a stack frame that is compliant with the Thumb Procedure Call
7427 Standard for all non-leaf functions. (A leaf function is one that does
7428 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7430 @item -mtpcs-leaf-frame
7431 @opindex mtpcs-leaf-frame
7432 Generate a stack frame that is compliant with the Thumb Procedure Call
7433 Standard for all leaf functions. (A leaf function is one that does
7434 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7436 @item -mcallee-super-interworking
7437 @opindex mcallee-super-interworking
7438 Gives all externally visible functions in the file being compiled an ARM
7439 instruction set header which switches to Thumb mode before executing the
7440 rest of the function. This allows these functions to be called from
7441 non-interworking code.
7443 @item -mcaller-super-interworking
7444 @opindex mcaller-super-interworking
7445 Allows calls via function pointers (including virtual functions) to
7446 execute correctly regardless of whether the target code has been
7447 compiled for interworking or not. There is a small overhead in the cost
7448 of executing a function pointer if this option is enabled.
7453 @subsection AVR Options
7456 These options are defined for AVR implementations:
7459 @item -mmcu=@var{mcu}
7461 Specify ATMEL AVR instruction set or MCU type.
7463 Instruction set avr1 is for the minimal AVR core, not supported by the C
7464 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7465 attiny11, attiny12, attiny15, attiny28).
7467 Instruction set avr2 (default) is for the classic AVR core with up to
7468 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7469 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7470 at90c8534, at90s8535).
7472 Instruction set avr3 is for the classic AVR core with up to 128K program
7473 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7475 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7476 memory space (MCU types: atmega8, atmega83, atmega85).
7478 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7479 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7480 atmega64, atmega128, at43usb355, at94k).
7484 Output instruction sizes to the asm file.
7486 @item -minit-stack=@var{N}
7487 @opindex minit-stack
7488 Specify the initial stack address, which may be a symbol or numeric value,
7489 @samp{__stack} is the default.
7491 @item -mno-interrupts
7492 @opindex mno-interrupts
7493 Generated code is not compatible with hardware interrupts.
7494 Code size will be smaller.
7496 @item -mcall-prologues
7497 @opindex mcall-prologues
7498 Functions prologues/epilogues expanded as call to appropriate
7499 subroutines. Code size will be smaller.
7501 @item -mno-tablejump
7502 @opindex mno-tablejump
7503 Do not generate tablejump insns which sometimes increase code size.
7506 @opindex mtiny-stack
7507 Change only the low 8 bits of the stack pointer.
7511 Assume int to be 8 bit integer. This affects the sizes of all types: A
7512 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7513 and long long will be 4 bytes. Please note that this option does not
7514 comply to the C standards, but it will provide you with smaller code
7518 @node Blackfin Options
7519 @subsection Blackfin Options
7520 @cindex Blackfin Options
7523 @item -momit-leaf-frame-pointer
7524 @opindex momit-leaf-frame-pointer
7525 Don't keep the frame pointer in a register for leaf functions. This
7526 avoids the instructions to save, set up and restore frame pointers and
7527 makes an extra register available in leaf functions. The option
7528 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7529 which might make debugging harder.
7531 @item -mspecld-anomaly
7532 @opindex mspecld-anomaly
7533 When enabled, the compiler will ensure that the generated code does not
7534 contain speculative loads after jump instructions. This option is enabled
7537 @item -mno-specld-anomaly
7538 @opindex mno-specld-anomaly
7539 Don't generate extra code to prevent speculative loads from occurring.
7541 @item -mcsync-anomaly
7542 @opindex mcsync-anomaly
7543 When enabled, the compiler will ensure that the generated code does not
7544 contain CSYNC or SSYNC instructions too soon after conditional branches.
7545 This option is enabled by default.
7547 @item -mno-csync-anomaly
7548 @opindex mno-csync-anomaly
7549 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7550 occurring too soon after a conditional branch.
7554 When enabled, the compiler is free to take advantage of the knowledge that
7555 the entire program fits into the low 64k of memory.
7558 @opindex mno-low-64k
7559 Assume that the program is arbitrarily large. This is the default.
7561 @item -mid-shared-library
7562 @opindex mid-shared-library
7563 Generate code that supports shared libraries via the library ID method.
7564 This allows for execute in place and shared libraries in an environment
7565 without virtual memory management. This option implies @option{-fPIC}.
7567 @item -mno-id-shared-library
7568 @opindex mno-id-shared-library
7569 Generate code that doesn't assume ID based shared libraries are being used.
7570 This is the default.
7572 @item -mshared-library-id=n
7573 @opindex mshared-library-id
7574 Specified the identification number of the ID based shared library being
7575 compiled. Specifying a value of 0 will generate more compact code, specifying
7576 other values will force the allocation of that number to the current
7577 library but is no more space or time efficient than omitting this option.
7580 @itemx -mno-long-calls
7581 @opindex mlong-calls
7582 @opindex mno-long-calls
7583 Tells the compiler to perform function calls by first loading the
7584 address of the function into a register and then performing a subroutine
7585 call on this register. This switch is needed if the target function
7586 will lie outside of the 24 bit addressing range of the offset based
7587 version of subroutine call instruction.
7589 This feature is not enabled by default. Specifying
7590 @option{-mno-long-calls} will restore the default behavior. Note these
7591 switches have no effect on how the compiler generates code to handle
7592 function calls via function pointers.
7596 @subsection CRIS Options
7597 @cindex CRIS Options
7599 These options are defined specifically for the CRIS ports.
7602 @item -march=@var{architecture-type}
7603 @itemx -mcpu=@var{architecture-type}
7606 Generate code for the specified architecture. The choices for
7607 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7608 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7609 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7612 @item -mtune=@var{architecture-type}
7614 Tune to @var{architecture-type} everything applicable about the generated
7615 code, except for the ABI and the set of available instructions. The
7616 choices for @var{architecture-type} are the same as for
7617 @option{-march=@var{architecture-type}}.
7619 @item -mmax-stack-frame=@var{n}
7620 @opindex mmax-stack-frame
7621 Warn when the stack frame of a function exceeds @var{n} bytes.
7623 @item -melinux-stacksize=@var{n}
7624 @opindex melinux-stacksize
7625 Only available with the @samp{cris-axis-aout} target. Arranges for
7626 indications in the program to the kernel loader that the stack of the
7627 program should be set to @var{n} bytes.
7633 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7634 @option{-march=v3} and @option{-march=v8} respectively.
7636 @item -mmul-bug-workaround
7637 @itemx -mno-mul-bug-workaround
7638 @opindex mmul-bug-workaround
7639 @opindex mno-mul-bug-workaround
7640 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7641 models where it applies. This option is active by default.
7645 Enable CRIS-specific verbose debug-related information in the assembly
7646 code. This option also has the effect to turn off the @samp{#NO_APP}
7647 formatted-code indicator to the assembler at the beginning of the
7652 Do not use condition-code results from previous instruction; always emit
7653 compare and test instructions before use of condition codes.
7655 @item -mno-side-effects
7656 @opindex mno-side-effects
7657 Do not emit instructions with side-effects in addressing modes other than
7661 @itemx -mno-stack-align
7663 @itemx -mno-data-align
7664 @itemx -mconst-align
7665 @itemx -mno-const-align
7666 @opindex mstack-align
7667 @opindex mno-stack-align
7668 @opindex mdata-align
7669 @opindex mno-data-align
7670 @opindex mconst-align
7671 @opindex mno-const-align
7672 These options (no-options) arranges (eliminate arrangements) for the
7673 stack-frame, individual data and constants to be aligned for the maximum
7674 single data access size for the chosen CPU model. The default is to
7675 arrange for 32-bit alignment. ABI details such as structure layout are
7676 not affected by these options.
7684 Similar to the stack- data- and const-align options above, these options
7685 arrange for stack-frame, writable data and constants to all be 32-bit,
7686 16-bit or 8-bit aligned. The default is 32-bit alignment.
7688 @item -mno-prologue-epilogue
7689 @itemx -mprologue-epilogue
7690 @opindex mno-prologue-epilogue
7691 @opindex mprologue-epilogue
7692 With @option{-mno-prologue-epilogue}, the normal function prologue and
7693 epilogue that sets up the stack-frame are omitted and no return
7694 instructions or return sequences are generated in the code. Use this
7695 option only together with visual inspection of the compiled code: no
7696 warnings or errors are generated when call-saved registers must be saved,
7697 or storage for local variable needs to be allocated.
7703 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7704 instruction sequences that load addresses for functions from the PLT part
7705 of the GOT rather than (traditional on other architectures) calls to the
7706 PLT@. The default is @option{-mgotplt}.
7710 Legacy no-op option only recognized with the cris-axis-aout target.
7714 Legacy no-op option only recognized with the cris-axis-elf and
7715 cris-axis-linux-gnu targets.
7719 Only recognized with the cris-axis-aout target, where it selects a
7720 GNU/linux-like multilib, include files and instruction set for
7725 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7729 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7730 to link with input-output functions from a simulator library. Code,
7731 initialized data and zero-initialized data are allocated consecutively.
7735 Like @option{-sim}, but pass linker options to locate initialized data at
7736 0x40000000 and zero-initialized data at 0x80000000.
7740 @subsection CRX Options
7743 These options are defined specifically for the CRX ports.
7749 Enable the use of multiply-accumulate instructions. Disabled by default.
7753 Push instructions will be used to pass outgoing arguments when functions
7754 are called. Enabled by default.
7757 @node Darwin Options
7758 @subsection Darwin Options
7759 @cindex Darwin options
7761 These options are defined for all architectures running the Darwin operating
7764 FSF GCC on Darwin does not create ``fat'' object files; it will create
7765 an object file for the single architecture that it was built to
7766 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7767 @option{-arch} options are used; it does so by running the compiler or
7768 linker multiple times and joining the results together with
7771 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7772 @samp{i686}) is determined by the flags that specify the ISA
7773 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7774 @option{-force_cpusubtype_ALL} option can be used to override this.
7776 The Darwin tools vary in their behavior when presented with an ISA
7777 mismatch. The assembler, @file{as}, will only permit instructions to
7778 be used that are valid for the subtype of the file it is generating,
7779 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7780 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7781 and print an error if asked to create a shared library with a less
7782 restrictive subtype than its input files (for instance, trying to put
7783 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7784 for executables, @file{ld}, will quietly give the executable the most
7785 restrictive subtype of any of its input files.
7790 Add the framework directory @var{dir} to the head of the list of
7791 directories to be searched for header files. These directories are
7792 interleaved with those specified by @option{-I} options and are
7793 scanned in a left-to-right order.
7795 A framework directory is a directory with frameworks in it. A
7796 framework is a directory with a @samp{"Headers"} and/or
7797 @samp{"PrivateHeaders"} directory contained directly in it that ends
7798 in @samp{".framework"}. The name of a framework is the name of this
7799 directory excluding the @samp{".framework"}. Headers associated with
7800 the framework are found in one of those two directories, with
7801 @samp{"Headers"} being searched first. A subframework is a framework
7802 directory that is in a framework's @samp{"Frameworks"} directory.
7803 Includes of subframework headers can only appear in a header of a
7804 framework that contains the subframework, or in a sibling subframework
7805 header. Two subframeworks are siblings if they occur in the same
7806 framework. A subframework should not have the same name as a
7807 framework, a warning will be issued if this is violated. Currently a
7808 subframework cannot have subframeworks, in the future, the mechanism
7809 may be extended to support this. The standard frameworks can be found
7810 in @samp{"/System/Library/Frameworks"} and
7811 @samp{"/Library/Frameworks"}. An example include looks like
7812 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7813 the name of the framework and header.h is found in the
7814 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7818 Emit debugging information for symbols that are used. For STABS
7819 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7820 This is by default ON@.
7824 Emit debugging information for all symbols and types.
7826 @item -mmacosx-version-min=@var{version}
7827 The earliest version of MacOS X that this executable will run on
7828 is @var{version}. Typical values of @var{version} include @code{10.1},
7829 @code{10.2}, and @code{10.3.9}.
7831 The default for this option is to make choices that seem to be most
7834 @item -mone-byte-bool
7835 @opindex -mone-byte-bool
7836 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
7837 By default @samp{sizeof(bool)} is @samp{4} when compiling for
7838 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
7839 option has no effect on x86.
7841 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
7842 to generate code that is not binary compatible with code generated
7843 without that switch. Using this switch may require recompiling all
7844 other modules in a program, including system libraries. Use this
7845 switch to conform to a non-default data model.
7847 @item -mfix-and-continue
7848 @itemx -ffix-and-continue
7849 @itemx -findirect-data
7850 @opindex mfix-and-continue
7851 @opindex ffix-and-continue
7852 @opindex findirect-data
7853 Generate code suitable for fast turn around development. Needed to
7854 enable gdb to dynamically load @code{.o} files into already running
7855 programs. @option{-findirect-data} and @option{-ffix-and-continue}
7856 are provided for backwards compatibility.
7860 Loads all members of static archive libraries.
7861 See man ld(1) for more information.
7863 @item -arch_errors_fatal
7864 @opindex arch_errors_fatal
7865 Cause the errors having to do with files that have the wrong architecture
7869 @opindex bind_at_load
7870 Causes the output file to be marked such that the dynamic linker will
7871 bind all undefined references when the file is loaded or launched.
7875 Produce a Mach-o bundle format file.
7876 See man ld(1) for more information.
7878 @item -bundle_loader @var{executable}
7879 @opindex bundle_loader
7880 This option specifies the @var{executable} that will be loading the build
7881 output file being linked. See man ld(1) for more information.
7884 @opindex -dynamiclib
7885 When passed this option, GCC will produce a dynamic library instead of
7886 an executable when linking, using the Darwin @file{libtool} command.
7888 @item -force_cpusubtype_ALL
7889 @opindex -force_cpusubtype_ALL
7890 This causes GCC's output file to have the @var{ALL} subtype, instead of
7891 one controlled by the @option{-mcpu} or @option{-march} option.
7893 @item -allowable_client @var{client_name}
7895 @itemx -compatibility_version
7896 @itemx -current_version
7898 @itemx -dependency-file
7900 @itemx -dylinker_install_name
7902 @itemx -exported_symbols_list
7904 @itemx -flat_namespace
7905 @itemx -force_flat_namespace
7906 @itemx -headerpad_max_install_names
7909 @itemx -install_name
7910 @itemx -keep_private_externs
7911 @itemx -multi_module
7912 @itemx -multiply_defined
7913 @itemx -multiply_defined_unused
7915 @itemx -no_dead_strip_inits_and_terms
7916 @itemx -nofixprebinding
7919 @itemx -noseglinkedit
7920 @itemx -pagezero_size
7922 @itemx -prebind_all_twolevel_modules
7923 @itemx -private_bundle
7924 @itemx -read_only_relocs
7926 @itemx -sectobjectsymbols
7930 @itemx -sectobjectsymbols
7933 @itemx -segs_read_only_addr
7934 @itemx -segs_read_write_addr
7935 @itemx -seg_addr_table
7936 @itemx -seg_addr_table_filename
7939 @itemx -segs_read_only_addr
7940 @itemx -segs_read_write_addr
7941 @itemx -single_module
7944 @itemx -sub_umbrella
7945 @itemx -twolevel_namespace
7948 @itemx -unexported_symbols_list
7949 @itemx -weak_reference_mismatches
7952 @opindex allowable_client
7953 @opindex client_name
7954 @opindex compatibility_version
7955 @opindex current_version
7957 @opindex dependency-file
7959 @opindex dylinker_install_name
7961 @opindex exported_symbols_list
7963 @opindex flat_namespace
7964 @opindex force_flat_namespace
7965 @opindex headerpad_max_install_names
7968 @opindex install_name
7969 @opindex keep_private_externs
7970 @opindex multi_module
7971 @opindex multiply_defined
7972 @opindex multiply_defined_unused
7974 @opindex no_dead_strip_inits_and_terms
7975 @opindex nofixprebinding
7976 @opindex nomultidefs
7978 @opindex noseglinkedit
7979 @opindex pagezero_size
7981 @opindex prebind_all_twolevel_modules
7982 @opindex private_bundle
7983 @opindex read_only_relocs
7985 @opindex sectobjectsymbols
7989 @opindex sectobjectsymbols
7992 @opindex segs_read_only_addr
7993 @opindex segs_read_write_addr
7994 @opindex seg_addr_table
7995 @opindex seg_addr_table_filename
7996 @opindex seglinkedit
7998 @opindex segs_read_only_addr
7999 @opindex segs_read_write_addr
8000 @opindex single_module
8002 @opindex sub_library
8003 @opindex sub_umbrella
8004 @opindex twolevel_namespace
8007 @opindex unexported_symbols_list
8008 @opindex weak_reference_mismatches
8009 @opindex whatsloaded
8011 These options are passed to the Darwin linker. The Darwin linker man page
8012 describes them in detail.
8015 @node DEC Alpha Options
8016 @subsection DEC Alpha Options
8018 These @samp{-m} options are defined for the DEC Alpha implementations:
8021 @item -mno-soft-float
8023 @opindex mno-soft-float
8024 @opindex msoft-float
8025 Use (do not use) the hardware floating-point instructions for
8026 floating-point operations. When @option{-msoft-float} is specified,
8027 functions in @file{libgcc.a} will be used to perform floating-point
8028 operations. Unless they are replaced by routines that emulate the
8029 floating-point operations, or compiled in such a way as to call such
8030 emulations routines, these routines will issue floating-point
8031 operations. If you are compiling for an Alpha without floating-point
8032 operations, you must ensure that the library is built so as not to call
8035 Note that Alpha implementations without floating-point operations are
8036 required to have floating-point registers.
8041 @opindex mno-fp-regs
8042 Generate code that uses (does not use) the floating-point register set.
8043 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8044 register set is not used, floating point operands are passed in integer
8045 registers as if they were integers and floating-point results are passed
8046 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8047 so any function with a floating-point argument or return value called by code
8048 compiled with @option{-mno-fp-regs} must also be compiled with that
8051 A typical use of this option is building a kernel that does not use,
8052 and hence need not save and restore, any floating-point registers.
8056 The Alpha architecture implements floating-point hardware optimized for
8057 maximum performance. It is mostly compliant with the IEEE floating
8058 point standard. However, for full compliance, software assistance is
8059 required. This option generates code fully IEEE compliant code
8060 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8061 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8062 defined during compilation. The resulting code is less efficient but is
8063 able to correctly support denormalized numbers and exceptional IEEE
8064 values such as not-a-number and plus/minus infinity. Other Alpha
8065 compilers call this option @option{-ieee_with_no_inexact}.
8067 @item -mieee-with-inexact
8068 @opindex mieee-with-inexact
8069 This is like @option{-mieee} except the generated code also maintains
8070 the IEEE @var{inexact-flag}. Turning on this option causes the
8071 generated code to implement fully-compliant IEEE math. In addition to
8072 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8073 macro. On some Alpha implementations the resulting code may execute
8074 significantly slower than the code generated by default. Since there is
8075 very little code that depends on the @var{inexact-flag}, you should
8076 normally not specify this option. Other Alpha compilers call this
8077 option @option{-ieee_with_inexact}.
8079 @item -mfp-trap-mode=@var{trap-mode}
8080 @opindex mfp-trap-mode
8081 This option controls what floating-point related traps are enabled.
8082 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8083 The trap mode can be set to one of four values:
8087 This is the default (normal) setting. The only traps that are enabled
8088 are the ones that cannot be disabled in software (e.g., division by zero
8092 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8096 Like @samp{su}, but the instructions are marked to be safe for software
8097 completion (see Alpha architecture manual for details).
8100 Like @samp{su}, but inexact traps are enabled as well.
8103 @item -mfp-rounding-mode=@var{rounding-mode}
8104 @opindex mfp-rounding-mode
8105 Selects the IEEE rounding mode. Other Alpha compilers call this option
8106 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8111 Normal IEEE rounding mode. Floating point numbers are rounded towards
8112 the nearest machine number or towards the even machine number in case
8116 Round towards minus infinity.
8119 Chopped rounding mode. Floating point numbers are rounded towards zero.
8122 Dynamic rounding mode. A field in the floating point control register
8123 (@var{fpcr}, see Alpha architecture reference manual) controls the
8124 rounding mode in effect. The C library initializes this register for
8125 rounding towards plus infinity. Thus, unless your program modifies the
8126 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8129 @item -mtrap-precision=@var{trap-precision}
8130 @opindex mtrap-precision
8131 In the Alpha architecture, floating point traps are imprecise. This
8132 means without software assistance it is impossible to recover from a
8133 floating trap and program execution normally needs to be terminated.
8134 GCC can generate code that can assist operating system trap handlers
8135 in determining the exact location that caused a floating point trap.
8136 Depending on the requirements of an application, different levels of
8137 precisions can be selected:
8141 Program precision. This option is the default and means a trap handler
8142 can only identify which program caused a floating point exception.
8145 Function precision. The trap handler can determine the function that
8146 caused a floating point exception.
8149 Instruction precision. The trap handler can determine the exact
8150 instruction that caused a floating point exception.
8153 Other Alpha compilers provide the equivalent options called
8154 @option{-scope_safe} and @option{-resumption_safe}.
8156 @item -mieee-conformant
8157 @opindex mieee-conformant
8158 This option marks the generated code as IEEE conformant. You must not
8159 use this option unless you also specify @option{-mtrap-precision=i} and either
8160 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8161 is to emit the line @samp{.eflag 48} in the function prologue of the
8162 generated assembly file. Under DEC Unix, this has the effect that
8163 IEEE-conformant math library routines will be linked in.
8165 @item -mbuild-constants
8166 @opindex mbuild-constants
8167 Normally GCC examines a 32- or 64-bit integer constant to
8168 see if it can construct it from smaller constants in two or three
8169 instructions. If it cannot, it will output the constant as a literal and
8170 generate code to load it from the data segment at runtime.
8172 Use this option to require GCC to construct @emph{all} integer constants
8173 using code, even if it takes more instructions (the maximum is six).
8175 You would typically use this option to build a shared library dynamic
8176 loader. Itself a shared library, it must relocate itself in memory
8177 before it can find the variables and constants in its own data segment.
8183 Select whether to generate code to be assembled by the vendor-supplied
8184 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8202 Indicate whether GCC should generate code to use the optional BWX,
8203 CIX, FIX and MAX instruction sets. The default is to use the instruction
8204 sets supported by the CPU type specified via @option{-mcpu=} option or that
8205 of the CPU on which GCC was built if none was specified.
8210 @opindex mfloat-ieee
8211 Generate code that uses (does not use) VAX F and G floating point
8212 arithmetic instead of IEEE single and double precision.
8214 @item -mexplicit-relocs
8215 @itemx -mno-explicit-relocs
8216 @opindex mexplicit-relocs
8217 @opindex mno-explicit-relocs
8218 Older Alpha assemblers provided no way to generate symbol relocations
8219 except via assembler macros. Use of these macros does not allow
8220 optimal instruction scheduling. GNU binutils as of version 2.12
8221 supports a new syntax that allows the compiler to explicitly mark
8222 which relocations should apply to which instructions. This option
8223 is mostly useful for debugging, as GCC detects the capabilities of
8224 the assembler when it is built and sets the default accordingly.
8228 @opindex msmall-data
8229 @opindex mlarge-data
8230 When @option{-mexplicit-relocs} is in effect, static data is
8231 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8232 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8233 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8234 16-bit relocations off of the @code{$gp} register. This limits the
8235 size of the small data area to 64KB, but allows the variables to be
8236 directly accessed via a single instruction.
8238 The default is @option{-mlarge-data}. With this option the data area
8239 is limited to just below 2GB@. Programs that require more than 2GB of
8240 data must use @code{malloc} or @code{mmap} to allocate the data in the
8241 heap instead of in the program's data segment.
8243 When generating code for shared libraries, @option{-fpic} implies
8244 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8248 @opindex msmall-text
8249 @opindex mlarge-text
8250 When @option{-msmall-text} is used, the compiler assumes that the
8251 code of the entire program (or shared library) fits in 4MB, and is
8252 thus reachable with a branch instruction. When @option{-msmall-data}
8253 is used, the compiler can assume that all local symbols share the
8254 same @code{$gp} value, and thus reduce the number of instructions
8255 required for a function call from 4 to 1.
8257 The default is @option{-mlarge-text}.
8259 @item -mcpu=@var{cpu_type}
8261 Set the instruction set and instruction scheduling parameters for
8262 machine type @var{cpu_type}. You can specify either the @samp{EV}
8263 style name or the corresponding chip number. GCC supports scheduling
8264 parameters for the EV4, EV5 and EV6 family of processors and will
8265 choose the default values for the instruction set from the processor
8266 you specify. If you do not specify a processor type, GCC will default
8267 to the processor on which the compiler was built.
8269 Supported values for @var{cpu_type} are
8275 Schedules as an EV4 and has no instruction set extensions.
8279 Schedules as an EV5 and has no instruction set extensions.
8283 Schedules as an EV5 and supports the BWX extension.
8288 Schedules as an EV5 and supports the BWX and MAX extensions.
8292 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8296 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8299 @item -mtune=@var{cpu_type}
8301 Set only the instruction scheduling parameters for machine type
8302 @var{cpu_type}. The instruction set is not changed.
8304 @item -mmemory-latency=@var{time}
8305 @opindex mmemory-latency
8306 Sets the latency the scheduler should assume for typical memory
8307 references as seen by the application. This number is highly
8308 dependent on the memory access patterns used by the application
8309 and the size of the external cache on the machine.
8311 Valid options for @var{time} are
8315 A decimal number representing clock cycles.
8321 The compiler contains estimates of the number of clock cycles for
8322 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8323 (also called Dcache, Scache, and Bcache), as well as to main memory.
8324 Note that L3 is only valid for EV5.
8329 @node DEC Alpha/VMS Options
8330 @subsection DEC Alpha/VMS Options
8332 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8335 @item -mvms-return-codes
8336 @opindex mvms-return-codes
8337 Return VMS condition codes from main. The default is to return POSIX
8338 style condition (e.g.@ error) codes.
8342 @subsection FRV Options
8349 Only use the first 32 general purpose registers.
8354 Use all 64 general purpose registers.
8359 Use only the first 32 floating point registers.
8364 Use all 64 floating point registers
8367 @opindex mhard-float
8369 Use hardware instructions for floating point operations.
8372 @opindex msoft-float
8374 Use library routines for floating point operations.
8379 Dynamically allocate condition code registers.
8384 Do not try to dynamically allocate condition code registers, only
8385 use @code{icc0} and @code{fcc0}.
8390 Change ABI to use double word insns.
8395 Do not use double word instructions.
8400 Use floating point double instructions.
8405 Do not use floating point double instructions.
8410 Use media instructions.
8415 Do not use media instructions.
8420 Use multiply and add/subtract instructions.
8425 Do not use multiply and add/subtract instructions.
8430 Select the FDPIC ABI, that uses function descriptors to represent
8431 pointers to functions. Without any PIC/PIE-related options, it
8432 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8433 assumes GOT entries and small data are within a 12-bit range from the
8434 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8435 are computed with 32 bits.
8438 @opindex minline-plt
8440 Enable inlining of PLT entries in function calls to functions that are
8441 not known to bind locally. It has no effect without @option{-mfdpic}.
8442 It's enabled by default if optimizing for speed and compiling for
8443 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8444 optimization option such as @option{-O3} or above is present in the
8450 Assume a large TLS segment when generating thread-local code.
8455 Do not assume a large TLS segment when generating thread-local code.
8460 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8461 that is known to be in read-only sections. It's enabled by default,
8462 except for @option{-fpic} or @option{-fpie}: even though it may help
8463 make the global offset table smaller, it trades 1 instruction for 4.
8464 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8465 one of which may be shared by multiple symbols, and it avoids the need
8466 for a GOT entry for the referenced symbol, so it's more likely to be a
8467 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8469 @item -multilib-library-pic
8470 @opindex multilib-library-pic
8472 Link with the (library, not FD) pic libraries. It's implied by
8473 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8474 @option{-fpic} without @option{-mfdpic}. You should never have to use
8480 Follow the EABI requirement of always creating a frame pointer whenever
8481 a stack frame is allocated. This option is enabled by default and can
8482 be disabled with @option{-mno-linked-fp}.
8485 @opindex mlong-calls
8487 Use indirect addressing to call functions outside the current
8488 compilation unit. This allows the functions to be placed anywhere
8489 within the 32-bit address space.
8491 @item -malign-labels
8492 @opindex malign-labels
8494 Try to align labels to an 8-byte boundary by inserting nops into the
8495 previous packet. This option only has an effect when VLIW packing
8496 is enabled. It doesn't create new packets; it merely adds nops to
8500 @opindex mlibrary-pic
8502 Generate position-independent EABI code.
8507 Use only the first four media accumulator registers.
8512 Use all eight media accumulator registers.
8517 Pack VLIW instructions.
8522 Do not pack VLIW instructions.
8527 Do not mark ABI switches in e_flags.
8532 Enable the use of conditional-move instructions (default).
8534 This switch is mainly for debugging the compiler and will likely be removed
8535 in a future version.
8537 @item -mno-cond-move
8538 @opindex mno-cond-move
8540 Disable the use of conditional-move instructions.
8542 This switch is mainly for debugging the compiler and will likely be removed
8543 in a future version.
8548 Enable the use of conditional set instructions (default).
8550 This switch is mainly for debugging the compiler and will likely be removed
8551 in a future version.
8556 Disable the use of conditional set instructions.
8558 This switch is mainly for debugging the compiler and will likely be removed
8559 in a future version.
8564 Enable the use of conditional execution (default).
8566 This switch is mainly for debugging the compiler and will likely be removed
8567 in a future version.
8569 @item -mno-cond-exec
8570 @opindex mno-cond-exec
8572 Disable the use of conditional execution.
8574 This switch is mainly for debugging the compiler and will likely be removed
8575 in a future version.
8578 @opindex mvliw-branch
8580 Run a pass to pack branches into VLIW instructions (default).
8582 This switch is mainly for debugging the compiler and will likely be removed
8583 in a future version.
8585 @item -mno-vliw-branch
8586 @opindex mno-vliw-branch
8588 Do not run a pass to pack branches into VLIW instructions.
8590 This switch is mainly for debugging the compiler and will likely be removed
8591 in a future version.
8593 @item -mmulti-cond-exec
8594 @opindex mmulti-cond-exec
8596 Enable optimization of @code{&&} and @code{||} in conditional execution
8599 This switch is mainly for debugging the compiler and will likely be removed
8600 in a future version.
8602 @item -mno-multi-cond-exec
8603 @opindex mno-multi-cond-exec
8605 Disable optimization of @code{&&} and @code{||} in conditional execution.
8607 This switch is mainly for debugging the compiler and will likely be removed
8608 in a future version.
8610 @item -mnested-cond-exec
8611 @opindex mnested-cond-exec
8613 Enable nested conditional execution optimizations (default).
8615 This switch is mainly for debugging the compiler and will likely be removed
8616 in a future version.
8618 @item -mno-nested-cond-exec
8619 @opindex mno-nested-cond-exec
8621 Disable nested conditional execution optimizations.
8623 This switch is mainly for debugging the compiler and will likely be removed
8624 in a future version.
8626 @item -moptimize-membar
8627 @opindex moptimize-membar
8629 This switch removes redundant @code{membar} instructions from the
8630 compiler generated code. It is enabled by default.
8632 @item -mno-optimize-membar
8633 @opindex mno-optimize-membar
8635 This switch disables the automatic removal of redundant @code{membar}
8636 instructions from the generated code.
8638 @item -mtomcat-stats
8639 @opindex mtomcat-stats
8641 Cause gas to print out tomcat statistics.
8643 @item -mcpu=@var{cpu}
8646 Select the processor type for which to generate code. Possible values are
8647 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8648 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8652 @node H8/300 Options
8653 @subsection H8/300 Options
8655 These @samp{-m} options are defined for the H8/300 implementations:
8660 Shorten some address references at link time, when possible; uses the
8661 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8662 ld, Using ld}, for a fuller description.
8666 Generate code for the H8/300H@.
8670 Generate code for the H8S@.
8674 Generate code for the H8S and H8/300H in the normal mode. This switch
8675 must be used either with @option{-mh} or @option{-ms}.
8679 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8683 Make @code{int} data 32 bits by default.
8687 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8688 The default for the H8/300H and H8S is to align longs and floats on 4
8690 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8691 This option has no effect on the H8/300.
8695 @subsection HPPA Options
8696 @cindex HPPA Options
8698 These @samp{-m} options are defined for the HPPA family of computers:
8701 @item -march=@var{architecture-type}
8703 Generate code for the specified architecture. The choices for
8704 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8705 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8706 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8707 architecture option for your machine. Code compiled for lower numbered
8708 architectures will run on higher numbered architectures, but not the
8712 @itemx -mpa-risc-1-1
8713 @itemx -mpa-risc-2-0
8714 @opindex mpa-risc-1-0
8715 @opindex mpa-risc-1-1
8716 @opindex mpa-risc-2-0
8717 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8720 @opindex mbig-switch
8721 Generate code suitable for big switch tables. Use this option only if
8722 the assembler/linker complain about out of range branches within a switch
8725 @item -mjump-in-delay
8726 @opindex mjump-in-delay
8727 Fill delay slots of function calls with unconditional jump instructions
8728 by modifying the return pointer for the function call to be the target
8729 of the conditional jump.
8731 @item -mdisable-fpregs
8732 @opindex mdisable-fpregs
8733 Prevent floating point registers from being used in any manner. This is
8734 necessary for compiling kernels which perform lazy context switching of
8735 floating point registers. If you use this option and attempt to perform
8736 floating point operations, the compiler will abort.
8738 @item -mdisable-indexing
8739 @opindex mdisable-indexing
8740 Prevent the compiler from using indexing address modes. This avoids some
8741 rather obscure problems when compiling MIG generated code under MACH@.
8743 @item -mno-space-regs
8744 @opindex mno-space-regs
8745 Generate code that assumes the target has no space registers. This allows
8746 GCC to generate faster indirect calls and use unscaled index address modes.
8748 Such code is suitable for level 0 PA systems and kernels.
8750 @item -mfast-indirect-calls
8751 @opindex mfast-indirect-calls
8752 Generate code that assumes calls never cross space boundaries. This
8753 allows GCC to emit code which performs faster indirect calls.
8755 This option will not work in the presence of shared libraries or nested
8758 @item -mfixed-range=@var{register-range}
8759 @opindex mfixed-range
8760 Generate code treating the given register range as fixed registers.
8761 A fixed register is one that the register allocator can not use. This is
8762 useful when compiling kernel code. A register range is specified as
8763 two registers separated by a dash. Multiple register ranges can be
8764 specified separated by a comma.
8766 @item -mlong-load-store
8767 @opindex mlong-load-store
8768 Generate 3-instruction load and store sequences as sometimes required by
8769 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8772 @item -mportable-runtime
8773 @opindex mportable-runtime
8774 Use the portable calling conventions proposed by HP for ELF systems.
8778 Enable the use of assembler directives only GAS understands.
8780 @item -mschedule=@var{cpu-type}
8782 Schedule code according to the constraints for the machine type
8783 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8784 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8785 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8786 proper scheduling option for your machine. The default scheduling is
8790 @opindex mlinker-opt
8791 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8792 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8793 linkers in which they give bogus error messages when linking some programs.
8796 @opindex msoft-float
8797 Generate output containing library calls for floating point.
8798 @strong{Warning:} the requisite libraries are not available for all HPPA
8799 targets. Normally the facilities of the machine's usual C compiler are
8800 used, but this cannot be done directly in cross-compilation. You must make
8801 your own arrangements to provide suitable library functions for
8802 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8803 does provide software floating point support.
8805 @option{-msoft-float} changes the calling convention in the output file;
8806 therefore, it is only useful if you compile @emph{all} of a program with
8807 this option. In particular, you need to compile @file{libgcc.a}, the
8808 library that comes with GCC, with @option{-msoft-float} in order for
8813 Generate the predefine, @code{_SIO}, for server IO@. The default is
8814 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8815 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
8816 options are available under HP-UX and HI-UX@.
8820 Use GNU ld specific options. This passes @option{-shared} to ld when
8821 building a shared library. It is the default when GCC is configured,
8822 explicitly or implicitly, with the GNU linker. This option does not
8823 have any affect on which ld is called, it only changes what parameters
8824 are passed to that ld. The ld that is called is determined by the
8825 @option{--with-ld} configure option, GCC's program search path, and
8826 finally by the user's @env{PATH}. The linker used by GCC can be printed
8827 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
8828 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
8832 Use HP ld specific options. This passes @option{-b} to ld when building
8833 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8834 links. It is the default when GCC is configured, explicitly or
8835 implicitly, with the HP linker. This option does not have any affect on
8836 which ld is called, it only changes what parameters are passed to that
8837 ld. The ld that is called is determined by the @option{--with-ld}
8838 configure option, GCC's program search path, and finally by the user's
8839 @env{PATH}. The linker used by GCC can be printed using @samp{which
8840 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
8841 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
8844 @opindex mno-long-calls
8845 Generate code that uses long call sequences. This ensures that a call
8846 is always able to reach linker generated stubs. The default is to generate
8847 long calls only when the distance from the call site to the beginning
8848 of the function or translation unit, as the case may be, exceeds a
8849 predefined limit set by the branch type being used. The limits for
8850 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8851 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8854 Distances are measured from the beginning of functions when using the
8855 @option{-ffunction-sections} option, or when using the @option{-mgas}
8856 and @option{-mno-portable-runtime} options together under HP-UX with
8859 It is normally not desirable to use this option as it will degrade
8860 performance. However, it may be useful in large applications,
8861 particularly when partial linking is used to build the application.
8863 The types of long calls used depends on the capabilities of the
8864 assembler and linker, and the type of code being generated. The
8865 impact on systems that support long absolute calls, and long pic
8866 symbol-difference or pc-relative calls should be relatively small.
8867 However, an indirect call is used on 32-bit ELF systems in pic code
8868 and it is quite long.
8870 @item -munix=@var{unix-std}
8872 Generate compiler predefines and select a startfile for the specified
8873 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
8874 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
8875 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
8876 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
8877 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
8880 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
8881 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
8882 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
8883 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
8884 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
8885 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
8887 It is @emph{important} to note that this option changes the interfaces
8888 for various library routines. It also affects the operational behavior
8889 of the C library. Thus, @emph{extreme} care is needed in using this
8892 Library code that is intended to operate with more than one UNIX
8893 standard must test, set and restore the variable @var{__xpg4_extended_mask}
8894 as appropriate. Most GNU software doesn't provide this capability.
8898 Suppress the generation of link options to search libdld.sl when the
8899 @option{-static} option is specified on HP-UX 10 and later.
8903 The HP-UX implementation of setlocale in libc has a dependency on
8904 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8905 when the @option{-static} option is specified, special link options
8906 are needed to resolve this dependency.
8908 On HP-UX 10 and later, the GCC driver adds the necessary options to
8909 link with libdld.sl when the @option{-static} option is specified.
8910 This causes the resulting binary to be dynamic. On the 64-bit port,
8911 the linkers generate dynamic binaries by default in any case. The
8912 @option{-nolibdld} option can be used to prevent the GCC driver from
8913 adding these link options.
8917 Add support for multithreading with the @dfn{dce thread} library
8918 under HP-UX@. This option sets flags for both the preprocessor and
8922 @node i386 and x86-64 Options
8923 @subsection Intel 386 and AMD x86-64 Options
8924 @cindex i386 Options
8925 @cindex x86-64 Options
8926 @cindex Intel 386 Options
8927 @cindex AMD x86-64 Options
8929 These @samp{-m} options are defined for the i386 and x86-64 family of
8933 @item -mtune=@var{cpu-type}
8935 Tune to @var{cpu-type} everything applicable about the generated code, except
8936 for the ABI and the set of available instructions. The choices for
8940 Original Intel's i386 CPU@.
8942 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
8944 Intel Pentium CPU with no MMX support.
8946 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8947 @item i686, pentiumpro
8948 Intel PentiumPro CPU@.
8950 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8951 @item pentium3, pentium3m
8952 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8955 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8956 support. Used by Centrino notebooks.
8957 @item pentium4, pentium4m
8958 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8960 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8963 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8964 SSE2 and SSE3 instruction set support.
8966 AMD K6 CPU with MMX instruction set support.
8968 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8969 @item athlon, athlon-tbird
8970 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8972 @item athlon-4, athlon-xp, athlon-mp
8973 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8974 instruction set support.
8975 @item k8, opteron, athlon64, athlon-fx
8976 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8977 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8979 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8982 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8983 instruction set support.
8985 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8986 implemented for this chip.)
8988 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8989 implemented for this chip.)
8992 While picking a specific @var{cpu-type} will schedule things appropriately
8993 for that particular chip, the compiler will not generate any code that
8994 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8997 @item -march=@var{cpu-type}
8999 Generate instructions for the machine type @var{cpu-type}. The choices
9000 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9001 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9003 @item -mcpu=@var{cpu-type}
9005 A deprecated synonym for @option{-mtune}.
9014 @opindex mpentiumpro
9015 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9016 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9017 These synonyms are deprecated.
9019 @item -mfpmath=@var{unit}
9021 Generate floating point arithmetics for selected unit @var{unit}. The choices
9026 Use the standard 387 floating point coprocessor present majority of chips and
9027 emulated otherwise. Code compiled with this option will run almost everywhere.
9028 The temporary results are computed in 80bit precision instead of precision
9029 specified by the type resulting in slightly different results compared to most
9030 of other chips. See @option{-ffloat-store} for more detailed description.
9032 This is the default choice for i386 compiler.
9035 Use scalar floating point instructions present in the SSE instruction set.
9036 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9037 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9038 instruction set supports only single precision arithmetics, thus the double and
9039 extended precision arithmetics is still done using 387. Later version, present
9040 only in Pentium4 and the future AMD x86-64 chips supports double precision
9043 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9044 or @option{-msse2} switches to enable SSE extensions and make this option
9045 effective. For the x86-64 compiler, these extensions are enabled by default.
9047 The resulting code should be considerably faster in the majority of cases and avoid
9048 the numerical instability problems of 387 code, but may break some existing
9049 code that expects temporaries to be 80bit.
9051 This is the default choice for the x86-64 compiler.
9054 Attempt to utilize both instruction sets at once. This effectively double the
9055 amount of available registers and on chips with separate execution units for
9056 387 and SSE the execution resources too. Use this option with care, as it is
9057 still experimental, because the GCC register allocator does not model separate
9058 functional units well resulting in instable performance.
9061 @item -masm=@var{dialect}
9062 @opindex masm=@var{dialect}
9063 Output asm instructions using selected @var{dialect}. Supported choices are
9064 @samp{intel} or @samp{att} (the default one).
9069 @opindex mno-ieee-fp
9070 Control whether or not the compiler uses IEEE floating point
9071 comparisons. These handle correctly the case where the result of a
9072 comparison is unordered.
9075 @opindex msoft-float
9076 Generate output containing library calls for floating point.
9077 @strong{Warning:} the requisite libraries are not part of GCC@.
9078 Normally the facilities of the machine's usual C compiler are used, but
9079 this can't be done directly in cross-compilation. You must make your
9080 own arrangements to provide suitable library functions for
9083 On machines where a function returns floating point results in the 80387
9084 register stack, some floating point opcodes may be emitted even if
9085 @option{-msoft-float} is used.
9087 @item -mno-fp-ret-in-387
9088 @opindex mno-fp-ret-in-387
9089 Do not use the FPU registers for return values of functions.
9091 The usual calling convention has functions return values of types
9092 @code{float} and @code{double} in an FPU register, even if there
9093 is no FPU@. The idea is that the operating system should emulate
9096 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9097 in ordinary CPU registers instead.
9099 @item -mno-fancy-math-387
9100 @opindex mno-fancy-math-387
9101 Some 387 emulators do not support the @code{sin}, @code{cos} and
9102 @code{sqrt} instructions for the 387. Specify this option to avoid
9103 generating those instructions. This option is the default on FreeBSD,
9104 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9105 indicates that the target cpu will always have an FPU and so the
9106 instruction will not need emulation. As of revision 2.6.1, these
9107 instructions are not generated unless you also use the
9108 @option{-funsafe-math-optimizations} switch.
9110 @item -malign-double
9111 @itemx -mno-align-double
9112 @opindex malign-double
9113 @opindex mno-align-double
9114 Control whether GCC aligns @code{double}, @code{long double}, and
9115 @code{long long} variables on a two word boundary or a one word
9116 boundary. Aligning @code{double} variables on a two word boundary will
9117 produce code that runs somewhat faster on a @samp{Pentium} at the
9118 expense of more memory.
9120 @strong{Warning:} if you use the @option{-malign-double} switch,
9121 structures containing the above types will be aligned differently than
9122 the published application binary interface specifications for the 386
9123 and will not be binary compatible with structures in code compiled
9124 without that switch.
9126 @item -m96bit-long-double
9127 @itemx -m128bit-long-double
9128 @opindex m96bit-long-double
9129 @opindex m128bit-long-double
9130 These switches control the size of @code{long double} type. The i386
9131 application binary interface specifies the size to be 96 bits,
9132 so @option{-m96bit-long-double} is the default in 32 bit mode.
9134 Modern architectures (Pentium and newer) would prefer @code{long double}
9135 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9136 conforming to the ABI, this would not be possible. So specifying a
9137 @option{-m128bit-long-double} will align @code{long double}
9138 to a 16 byte boundary by padding the @code{long double} with an additional
9141 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9142 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9144 Notice that neither of these options enable any extra precision over the x87
9145 standard of 80 bits for a @code{long double}.
9147 @strong{Warning:} if you override the default value for your target ABI, the
9148 structures and arrays containing @code{long double} variables will change
9149 their size as well as function calling convention for function taking
9150 @code{long double} will be modified. Hence they will not be binary
9151 compatible with arrays or structures in code compiled without that switch.
9153 @item -mmlarge-data-threshold=@var{number}
9154 @opindex mlarge-data-threshold=@var{number}
9155 When @option{-mcmodel=medium} is specified, the data greater than
9156 @var{threshold} are placed in large data section. This value must be the
9157 same across all object linked into the binary and defaults to 65535.
9160 @itemx -mno-svr3-shlib
9161 @opindex msvr3-shlib
9162 @opindex mno-svr3-shlib
9163 Control whether GCC places uninitialized local variables into the
9164 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9165 into @code{bss}. These options are meaningful only on System V Release 3.
9169 Use a different function-calling convention, in which functions that
9170 take a fixed number of arguments return with the @code{ret} @var{num}
9171 instruction, which pops their arguments while returning. This saves one
9172 instruction in the caller since there is no need to pop the arguments
9175 You can specify that an individual function is called with this calling
9176 sequence with the function attribute @samp{stdcall}. You can also
9177 override the @option{-mrtd} option by using the function attribute
9178 @samp{cdecl}. @xref{Function Attributes}.
9180 @strong{Warning:} this calling convention is incompatible with the one
9181 normally used on Unix, so you cannot use it if you need to call
9182 libraries compiled with the Unix compiler.
9184 Also, you must provide function prototypes for all functions that
9185 take variable numbers of arguments (including @code{printf});
9186 otherwise incorrect code will be generated for calls to those
9189 In addition, seriously incorrect code will result if you call a
9190 function with too many arguments. (Normally, extra arguments are
9191 harmlessly ignored.)
9193 @item -mregparm=@var{num}
9195 Control how many registers are used to pass integer arguments. By
9196 default, no registers are used to pass arguments, and at most 3
9197 registers can be used. You can control this behavior for a specific
9198 function by using the function attribute @samp{regparm}.
9199 @xref{Function Attributes}.
9201 @strong{Warning:} if you use this switch, and
9202 @var{num} is nonzero, then you must build all modules with the same
9203 value, including any libraries. This includes the system libraries and
9207 @opindex msseregparm
9208 Use SSE register passing conventions for float and double arguments
9209 and return values. You can control this behavior for a specific
9210 function by using the function attribute @samp{sseregparm}.
9211 @xref{Function Attributes}.
9213 @strong{Warning:} if you use this switch then you must build all
9214 modules with the same value, including any libraries. This includes
9215 the system libraries and startup modules.
9217 @item -mpreferred-stack-boundary=@var{num}
9218 @opindex mpreferred-stack-boundary
9219 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9220 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9221 the default is 4 (16 bytes or 128 bits), except when optimizing for code
9222 size (@option{-Os}), in which case the default is the minimum correct
9223 alignment (4 bytes for x86, and 8 bytes for x86-64).
9225 On Pentium and PentiumPro, @code{double} and @code{long double} values
9226 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9227 suffer significant run time performance penalties. On Pentium III, the
9228 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
9229 penalties if it is not 16 byte aligned.
9231 To ensure proper alignment of this values on the stack, the stack boundary
9232 must be as aligned as that required by any value stored on the stack.
9233 Further, every function must be generated such that it keeps the stack
9234 aligned. Thus calling a function compiled with a higher preferred
9235 stack boundary from a function compiled with a lower preferred stack
9236 boundary will most likely misalign the stack. It is recommended that
9237 libraries that use callbacks always use the default setting.
9239 This extra alignment does consume extra stack space, and generally
9240 increases code size. Code that is sensitive to stack space usage, such
9241 as embedded systems and operating system kernels, may want to reduce the
9242 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9260 These switches enable or disable the use of built-in functions that allow
9261 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
9264 @xref{X86 Built-in Functions}, for details of the functions enabled
9265 and disabled by these switches.
9267 To have SSE/SSE2 instructions generated automatically from floating-point
9268 code, see @option{-mfpmath=sse}.
9271 @itemx -mno-push-args
9273 @opindex mno-push-args
9274 Use PUSH operations to store outgoing parameters. This method is shorter
9275 and usually equally fast as method using SUB/MOV operations and is enabled
9276 by default. In some cases disabling it may improve performance because of
9277 improved scheduling and reduced dependencies.
9279 @item -maccumulate-outgoing-args
9280 @opindex maccumulate-outgoing-args
9281 If enabled, the maximum amount of space required for outgoing arguments will be
9282 computed in the function prologue. This is faster on most modern CPUs
9283 because of reduced dependencies, improved scheduling and reduced stack usage
9284 when preferred stack boundary is not equal to 2. The drawback is a notable
9285 increase in code size. This switch implies @option{-mno-push-args}.
9289 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9290 on thread-safe exception handling must compile and link all code with the
9291 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9292 @option{-D_MT}; when linking, it links in a special thread helper library
9293 @option{-lmingwthrd} which cleans up per thread exception handling data.
9295 @item -mno-align-stringops
9296 @opindex mno-align-stringops
9297 Do not align destination of inlined string operations. This switch reduces
9298 code size and improves performance in case the destination is already aligned,
9299 but GCC doesn't know about it.
9301 @item -minline-all-stringops
9302 @opindex minline-all-stringops
9303 By default GCC inlines string operations only when destination is known to be
9304 aligned at least to 4 byte boundary. This enables more inlining, increase code
9305 size, but may improve performance of code that depends on fast memcpy, strlen
9306 and memset for short lengths.
9308 @item -momit-leaf-frame-pointer
9309 @opindex momit-leaf-frame-pointer
9310 Don't keep the frame pointer in a register for leaf functions. This
9311 avoids the instructions to save, set up and restore frame pointers and
9312 makes an extra register available in leaf functions. The option
9313 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9314 which might make debugging harder.
9316 @item -mtls-direct-seg-refs
9317 @itemx -mno-tls-direct-seg-refs
9318 @opindex mtls-direct-seg-refs
9319 Controls whether TLS variables may be accessed with offsets from the
9320 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9321 or whether the thread base pointer must be added. Whether or not this
9322 is legal depends on the operating system, and whether it maps the
9323 segment to cover the entire TLS area.
9325 For systems that use GNU libc, the default is on.
9328 These @samp{-m} switches are supported in addition to the above
9329 on AMD x86-64 processors in 64-bit environments.
9336 Generate code for a 32-bit or 64-bit environment.
9337 The 32-bit environment sets int, long and pointer to 32 bits and
9338 generates code that runs on any i386 system.
9339 The 64-bit environment sets int to 32 bits and long and pointer
9340 to 64 bits and generates code for AMD's x86-64 architecture.
9343 @opindex no-red-zone
9344 Do not use a so called red zone for x86-64 code. The red zone is mandated
9345 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9346 stack pointer that will not be modified by signal or interrupt handlers
9347 and therefore can be used for temporary data without adjusting the stack
9348 pointer. The flag @option{-mno-red-zone} disables this red zone.
9350 @item -mcmodel=small
9351 @opindex mcmodel=small
9352 Generate code for the small code model: the program and its symbols must
9353 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9354 Programs can be statically or dynamically linked. This is the default
9357 @item -mcmodel=kernel
9358 @opindex mcmodel=kernel
9359 Generate code for the kernel code model. The kernel runs in the
9360 negative 2 GB of the address space.
9361 This model has to be used for Linux kernel code.
9363 @item -mcmodel=medium
9364 @opindex mcmodel=medium
9365 Generate code for the medium model: The program is linked in the lower 2
9366 GB of the address space but symbols can be located anywhere in the
9367 address space. Programs can be statically or dynamically linked, but
9368 building of shared libraries are not supported with the medium model.
9370 @item -mcmodel=large
9371 @opindex mcmodel=large
9372 Generate code for the large model: This model makes no assumptions
9373 about addresses and sizes of sections. Currently GCC does not implement
9378 @subsection IA-64 Options
9379 @cindex IA-64 Options
9381 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9385 @opindex mbig-endian
9386 Generate code for a big endian target. This is the default for HP-UX@.
9388 @item -mlittle-endian
9389 @opindex mlittle-endian
9390 Generate code for a little endian target. This is the default for AIX5
9397 Generate (or don't) code for the GNU assembler. This is the default.
9398 @c Also, this is the default if the configure option @option{--with-gnu-as}
9405 Generate (or don't) code for the GNU linker. This is the default.
9406 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9411 Generate code that does not use a global pointer register. The result
9412 is not position independent code, and violates the IA-64 ABI@.
9414 @item -mvolatile-asm-stop
9415 @itemx -mno-volatile-asm-stop
9416 @opindex mvolatile-asm-stop
9417 @opindex mno-volatile-asm-stop
9418 Generate (or don't) a stop bit immediately before and after volatile asm
9421 @item -mregister-names
9422 @itemx -mno-register-names
9423 @opindex mregister-names
9424 @opindex mno-register-names
9425 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9426 the stacked registers. This may make assembler output more readable.
9432 Disable (or enable) optimizations that use the small data section. This may
9433 be useful for working around optimizer bugs.
9436 @opindex mconstant-gp
9437 Generate code that uses a single constant global pointer value. This is
9438 useful when compiling kernel code.
9442 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9443 This is useful when compiling firmware code.
9445 @item -minline-float-divide-min-latency
9446 @opindex minline-float-divide-min-latency
9447 Generate code for inline divides of floating point values
9448 using the minimum latency algorithm.
9450 @item -minline-float-divide-max-throughput
9451 @opindex minline-float-divide-max-throughput
9452 Generate code for inline divides of floating point values
9453 using the maximum throughput algorithm.
9455 @item -minline-int-divide-min-latency
9456 @opindex minline-int-divide-min-latency
9457 Generate code for inline divides of integer values
9458 using the minimum latency algorithm.
9460 @item -minline-int-divide-max-throughput
9461 @opindex minline-int-divide-max-throughput
9462 Generate code for inline divides of integer values
9463 using the maximum throughput algorithm.
9465 @item -minline-sqrt-min-latency
9466 @opindex minline-sqrt-min-latency
9467 Generate code for inline square roots
9468 using the minimum latency algorithm.
9470 @item -minline-sqrt-max-throughput
9471 @opindex minline-sqrt-max-throughput
9472 Generate code for inline square roots
9473 using the maximum throughput algorithm.
9475 @item -mno-dwarf2-asm
9477 @opindex mno-dwarf2-asm
9478 @opindex mdwarf2-asm
9479 Don't (or do) generate assembler code for the DWARF2 line number debugging
9480 info. This may be useful when not using the GNU assembler.
9482 @item -mearly-stop-bits
9483 @itemx -mno-early-stop-bits
9484 @opindex mearly-stop-bits
9485 @opindex mno-early-stop-bits
9486 Allow stop bits to be placed earlier than immediately preceding the
9487 instruction that triggered the stop bit. This can improve instruction
9488 scheduling, but does not always do so.
9490 @item -mfixed-range=@var{register-range}
9491 @opindex mfixed-range
9492 Generate code treating the given register range as fixed registers.
9493 A fixed register is one that the register allocator can not use. This is
9494 useful when compiling kernel code. A register range is specified as
9495 two registers separated by a dash. Multiple register ranges can be
9496 specified separated by a comma.
9498 @item -mtls-size=@var{tls-size}
9500 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9503 @item -mtune-arch=@var{cpu-type}
9505 Tune the instruction scheduling for a particular CPU, Valid values are
9506 itanium, itanium1, merced, itanium2, and mckinley.
9512 Add support for multithreading using the POSIX threads library. This
9513 option sets flags for both the preprocessor and linker. It does
9514 not affect the thread safety of object code produced by the compiler or
9515 that of libraries supplied with it. These are HP-UX specific flags.
9521 Generate code for a 32-bit or 64-bit environment.
9522 The 32-bit environment sets int, long and pointer to 32 bits.
9523 The 64-bit environment sets int to 32 bits and long and pointer
9524 to 64 bits. These are HP-UX specific flags.
9529 @subsection M32C Options
9530 @cindex M32C options
9533 @item -mcpu=@var{name}
9535 Select the CPU for which code is generated. @var{name} may be one of
9536 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
9537 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
9542 Specifies that the program will be run on the simulator. This causes
9543 an alternate runtime library to be linked in which supports, for
9544 example, file I/O. You must not use this option when generating
9545 programs that will run on real hardware; you must provide your own
9546 runtime library for whatever I/O functions are needed.
9548 @item -memregs=@var{number}
9550 Specifies the number of memory-based pseudo-registers GCC will use
9551 during code generation. These pseudo-registers will be used like real
9552 registers, so there is a tradeoff between GCC's ability to fit the
9553 code into available registers, and the performance penalty of using
9554 memory instead of registers. Note that all modules in a program must
9555 be compiled with the same value for this option. Because of that, you
9556 must not use this option with the default runtime libraries gcc
9561 @node M32R/D Options
9562 @subsection M32R/D Options
9563 @cindex M32R/D options
9565 These @option{-m} options are defined for Renesas M32R/D architectures:
9570 Generate code for the M32R/2@.
9574 Generate code for the M32R/X@.
9578 Generate code for the M32R@. This is the default.
9581 @opindex mmodel=small
9582 Assume all objects live in the lower 16MB of memory (so that their addresses
9583 can be loaded with the @code{ld24} instruction), and assume all subroutines
9584 are reachable with the @code{bl} instruction.
9585 This is the default.
9587 The addressability of a particular object can be set with the
9588 @code{model} attribute.
9590 @item -mmodel=medium
9591 @opindex mmodel=medium
9592 Assume objects may be anywhere in the 32-bit address space (the compiler
9593 will generate @code{seth/add3} instructions to load their addresses), and
9594 assume all subroutines are reachable with the @code{bl} instruction.
9597 @opindex mmodel=large
9598 Assume objects may be anywhere in the 32-bit address space (the compiler
9599 will generate @code{seth/add3} instructions to load their addresses), and
9600 assume subroutines may not be reachable with the @code{bl} instruction
9601 (the compiler will generate the much slower @code{seth/add3/jl}
9602 instruction sequence).
9605 @opindex msdata=none
9606 Disable use of the small data area. Variables will be put into
9607 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
9608 @code{section} attribute has been specified).
9609 This is the default.
9611 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
9612 Objects may be explicitly put in the small data area with the
9613 @code{section} attribute using one of these sections.
9616 @opindex msdata=sdata
9617 Put small global and static data in the small data area, but do not
9618 generate special code to reference them.
9622 Put small global and static data in the small data area, and generate
9623 special instructions to reference them.
9627 @cindex smaller data references
9628 Put global and static objects less than or equal to @var{num} bytes
9629 into the small data or bss sections instead of the normal data or bss
9630 sections. The default value of @var{num} is 8.
9631 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
9632 for this option to have any effect.
9634 All modules should be compiled with the same @option{-G @var{num}} value.
9635 Compiling with different values of @var{num} may or may not work; if it
9636 doesn't the linker will give an error message---incorrect code will not be
9641 Makes the M32R specific code in the compiler display some statistics
9642 that might help in debugging programs.
9645 @opindex malign-loops
9646 Align all loops to a 32-byte boundary.
9648 @item -mno-align-loops
9649 @opindex mno-align-loops
9650 Do not enforce a 32-byte alignment for loops. This is the default.
9652 @item -missue-rate=@var{number}
9653 @opindex missue-rate=@var{number}
9654 Issue @var{number} instructions per cycle. @var{number} can only be 1
9657 @item -mbranch-cost=@var{number}
9658 @opindex mbranch-cost=@var{number}
9659 @var{number} can only be 1 or 2. If it is 1 then branches will be
9660 preferred over conditional code, if it is 2, then the opposite will
9663 @item -mflush-trap=@var{number}
9664 @opindex mflush-trap=@var{number}
9665 Specifies the trap number to use to flush the cache. The default is
9666 12. Valid numbers are between 0 and 15 inclusive.
9668 @item -mno-flush-trap
9669 @opindex mno-flush-trap
9670 Specifies that the cache cannot be flushed by using a trap.
9672 @item -mflush-func=@var{name}
9673 @opindex mflush-func=@var{name}
9674 Specifies the name of the operating system function to call to flush
9675 the cache. The default is @emph{_flush_cache}, but a function call
9676 will only be used if a trap is not available.
9678 @item -mno-flush-func
9679 @opindex mno-flush-func
9680 Indicates that there is no OS function for flushing the cache.
9684 @node M680x0 Options
9685 @subsection M680x0 Options
9686 @cindex M680x0 options
9688 These are the @samp{-m} options defined for the 68000 series. The default
9689 values for these options depends on which style of 68000 was selected when
9690 the compiler was configured; the defaults for the most common choices are
9698 Generate output for a 68000. This is the default
9699 when the compiler is configured for 68000-based systems.
9701 Use this option for microcontrollers with a 68000 or EC000 core,
9702 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
9708 Generate output for a 68020. This is the default
9709 when the compiler is configured for 68020-based systems.
9713 Generate output containing 68881 instructions for floating point.
9714 This is the default for most 68020 systems unless @option{--nfp} was
9715 specified when the compiler was configured.
9719 Generate output for a 68030. This is the default when the compiler is
9720 configured for 68030-based systems.
9724 Generate output for a 68040. This is the default when the compiler is
9725 configured for 68040-based systems.
9727 This option inhibits the use of 68881/68882 instructions that have to be
9728 emulated by software on the 68040. Use this option if your 68040 does not
9729 have code to emulate those instructions.
9733 Generate output for a 68060. This is the default when the compiler is
9734 configured for 68060-based systems.
9736 This option inhibits the use of 68020 and 68881/68882 instructions that
9737 have to be emulated by software on the 68060. Use this option if your 68060
9738 does not have code to emulate those instructions.
9742 Generate output for a CPU32. This is the default
9743 when the compiler is configured for CPU32-based systems.
9745 Use this option for microcontrollers with a
9746 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
9747 68336, 68340, 68341, 68349 and 68360.
9751 Generate output for a 520X ``coldfire'' family cpu. This is the default
9752 when the compiler is configured for 520X-based systems.
9754 Use this option for microcontroller with a 5200 core, including
9755 the MCF5202, MCF5203, MCF5204 and MCF5202.
9760 Generate output for a 68040, without using any of the new instructions.
9761 This results in code which can run relatively efficiently on either a
9762 68020/68881 or a 68030 or a 68040. The generated code does use the
9763 68881 instructions that are emulated on the 68040.
9767 Generate output for a 68060, without using any of the new instructions.
9768 This results in code which can run relatively efficiently on either a
9769 68020/68881 or a 68030 or a 68040. The generated code does use the
9770 68881 instructions that are emulated on the 68060.
9773 @opindex msoft-float
9774 Generate output containing library calls for floating point.
9775 @strong{Warning:} the requisite libraries are not available for all m68k
9776 targets. Normally the facilities of the machine's usual C compiler are
9777 used, but this can't be done directly in cross-compilation. You must
9778 make your own arrangements to provide suitable library functions for
9779 cross-compilation. The embedded targets @samp{m68k-*-aout} and
9780 @samp{m68k-*-coff} do provide software floating point support.
9784 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9785 Additionally, parameters passed on the stack are also aligned to a
9786 16-bit boundary even on targets whose API mandates promotion to 32-bit.
9789 @opindex mnobitfield
9790 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
9791 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
9795 Do use the bit-field instructions. The @option{-m68020} option implies
9796 @option{-mbitfield}. This is the default if you use a configuration
9797 designed for a 68020.
9801 Use a different function-calling convention, in which functions
9802 that take a fixed number of arguments return with the @code{rtd}
9803 instruction, which pops their arguments while returning. This
9804 saves one instruction in the caller since there is no need to pop
9805 the arguments there.
9807 This calling convention is incompatible with the one normally
9808 used on Unix, so you cannot use it if you need to call libraries
9809 compiled with the Unix compiler.
9811 Also, you must provide function prototypes for all functions that
9812 take variable numbers of arguments (including @code{printf});
9813 otherwise incorrect code will be generated for calls to those
9816 In addition, seriously incorrect code will result if you call a
9817 function with too many arguments. (Normally, extra arguments are
9818 harmlessly ignored.)
9820 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
9821 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
9824 @itemx -mno-align-int
9826 @opindex mno-align-int
9827 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
9828 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
9829 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
9830 Aligning variables on 32-bit boundaries produces code that runs somewhat
9831 faster on processors with 32-bit busses at the expense of more memory.
9833 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
9834 align structures containing the above types differently than
9835 most published application binary interface specifications for the m68k.
9839 Use the pc-relative addressing mode of the 68000 directly, instead of
9840 using a global offset table. At present, this option implies @option{-fpic},
9841 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
9842 not presently supported with @option{-mpcrel}, though this could be supported for
9843 68020 and higher processors.
9845 @item -mno-strict-align
9846 @itemx -mstrict-align
9847 @opindex mno-strict-align
9848 @opindex mstrict-align
9849 Do not (do) assume that unaligned memory references will be handled by
9853 Generate code that allows the data segment to be located in a different
9854 area of memory from the text segment. This allows for execute in place in
9855 an environment without virtual memory management. This option implies
9859 Generate code that assumes that the data segment follows the text segment.
9860 This is the default.
9862 @item -mid-shared-library
9863 Generate code that supports shared libraries via the library ID method.
9864 This allows for execute in place and shared libraries in an environment
9865 without virtual memory management. This option implies @option{-fPIC}.
9867 @item -mno-id-shared-library
9868 Generate code that doesn't assume ID based shared libraries are being used.
9869 This is the default.
9871 @item -mshared-library-id=n
9872 Specified the identification number of the ID based shared library being
9873 compiled. Specifying a value of 0 will generate more compact code, specifying
9874 other values will force the allocation of that number to the current
9875 library but is no more space or time efficient than omitting this option.
9879 @node M68hc1x Options
9880 @subsection M68hc1x Options
9881 @cindex M68hc1x options
9883 These are the @samp{-m} options defined for the 68hc11 and 68hc12
9884 microcontrollers. The default values for these options depends on
9885 which style of microcontroller was selected when the compiler was configured;
9886 the defaults for the most common choices are given below.
9893 Generate output for a 68HC11. This is the default
9894 when the compiler is configured for 68HC11-based systems.
9900 Generate output for a 68HC12. This is the default
9901 when the compiler is configured for 68HC12-based systems.
9907 Generate output for a 68HCS12.
9910 @opindex mauto-incdec
9911 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
9918 Enable the use of 68HC12 min and max instructions.
9921 @itemx -mno-long-calls
9922 @opindex mlong-calls
9923 @opindex mno-long-calls
9924 Treat all calls as being far away (near). If calls are assumed to be
9925 far away, the compiler will use the @code{call} instruction to
9926 call a function and the @code{rtc} instruction for returning.
9930 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9932 @item -msoft-reg-count=@var{count}
9933 @opindex msoft-reg-count
9934 Specify the number of pseudo-soft registers which are used for the
9935 code generation. The maximum number is 32. Using more pseudo-soft
9936 register may or may not result in better code depending on the program.
9937 The default is 4 for 68HC11 and 2 for 68HC12.
9942 @subsection MCore Options
9943 @cindex MCore options
9945 These are the @samp{-m} options defined for the Motorola M*Core
9953 @opindex mno-hardlit
9954 Inline constants into the code stream if it can be done in two
9955 instructions or less.
9961 Use the divide instruction. (Enabled by default).
9963 @item -mrelax-immediate
9964 @itemx -mno-relax-immediate
9965 @opindex mrelax-immediate
9966 @opindex mno-relax-immediate
9967 Allow arbitrary sized immediates in bit operations.
9969 @item -mwide-bitfields
9970 @itemx -mno-wide-bitfields
9971 @opindex mwide-bitfields
9972 @opindex mno-wide-bitfields
9973 Always treat bit-fields as int-sized.
9975 @item -m4byte-functions
9976 @itemx -mno-4byte-functions
9977 @opindex m4byte-functions
9978 @opindex mno-4byte-functions
9979 Force all functions to be aligned to a four byte boundary.
9981 @item -mcallgraph-data
9982 @itemx -mno-callgraph-data
9983 @opindex mcallgraph-data
9984 @opindex mno-callgraph-data
9985 Emit callgraph information.
9988 @itemx -mno-slow-bytes
9989 @opindex mslow-bytes
9990 @opindex mno-slow-bytes
9991 Prefer word access when reading byte quantities.
9993 @item -mlittle-endian
9995 @opindex mlittle-endian
9996 @opindex mbig-endian
9997 Generate code for a little endian target.
10003 Generate code for the 210 processor.
10007 @subsection MIPS Options
10008 @cindex MIPS options
10014 Generate big-endian code.
10018 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10021 @item -march=@var{arch}
10023 Generate code that will run on @var{arch}, which can be the name of a
10024 generic MIPS ISA, or the name of a particular processor.
10026 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10027 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10028 The processor names are:
10029 @samp{4kc}, @samp{4km}, @samp{4kp},
10030 @samp{5kc}, @samp{5kf},
10032 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10035 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10036 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10037 @samp{rm7000}, @samp{rm9000},
10040 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10041 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10042 The special value @samp{from-abi} selects the
10043 most compatible architecture for the selected ABI (that is,
10044 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10046 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10047 (for example, @samp{-march=r2k}). Prefixes are optional, and
10048 @samp{vr} may be written @samp{r}.
10050 GCC defines two macros based on the value of this option. The first
10051 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10052 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10053 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10054 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10055 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10057 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10058 above. In other words, it will have the full prefix and will not
10059 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10060 the macro names the resolved architecture (either @samp{"mips1"} or
10061 @samp{"mips3"}). It names the default architecture when no
10062 @option{-march} option is given.
10064 @item -mtune=@var{arch}
10066 Optimize for @var{arch}. Among other things, this option controls
10067 the way instructions are scheduled, and the perceived cost of arithmetic
10068 operations. The list of @var{arch} values is the same as for
10071 When this option is not used, GCC will optimize for the processor
10072 specified by @option{-march}. By using @option{-march} and
10073 @option{-mtune} together, it is possible to generate code that will
10074 run on a family of processors, but optimize the code for one
10075 particular member of that family.
10077 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10078 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10079 @samp{-march} ones described above.
10083 Equivalent to @samp{-march=mips1}.
10087 Equivalent to @samp{-march=mips2}.
10091 Equivalent to @samp{-march=mips3}.
10095 Equivalent to @samp{-march=mips4}.
10099 Equivalent to @samp{-march=mips32}.
10103 Equivalent to @samp{-march=mips32r2}.
10107 Equivalent to @samp{-march=mips64}.
10112 @opindex mno-mips16
10113 Generate (do not generate) MIPS16 code. If GCC is targetting a
10114 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10126 Generate code for the given ABI@.
10128 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10129 generates 64-bit code when you select a 64-bit architecture, but you
10130 can use @option{-mgp32} to get 32-bit code instead.
10132 For information about the O64 ABI, see
10133 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10136 @itemx -mno-abicalls
10138 @opindex mno-abicalls
10139 Generate (do not generate) SVR4-style position-independent code.
10140 @option{-mabicalls} is the default for SVR4-based systems.
10146 Lift (do not lift) the usual restrictions on the size of the global
10149 GCC normally uses a single instruction to load values from the GOT@.
10150 While this is relatively efficient, it will only work if the GOT
10151 is smaller than about 64k. Anything larger will cause the linker
10152 to report an error such as:
10154 @cindex relocation truncated to fit (MIPS)
10156 relocation truncated to fit: R_MIPS_GOT16 foobar
10159 If this happens, you should recompile your code with @option{-mxgot}.
10160 It should then work with very large GOTs, although it will also be
10161 less efficient, since it will take three instructions to fetch the
10162 value of a global symbol.
10164 Note that some linkers can create multiple GOTs. If you have such a
10165 linker, you should only need to use @option{-mxgot} when a single object
10166 file accesses more than 64k's worth of GOT entries. Very few do.
10168 These options have no effect unless GCC is generating position
10173 Assume that general-purpose registers are 32 bits wide.
10177 Assume that general-purpose registers are 64 bits wide.
10181 Assume that floating-point registers are 32 bits wide.
10185 Assume that floating-point registers are 64 bits wide.
10188 @opindex mhard-float
10189 Use floating-point coprocessor instructions.
10192 @opindex msoft-float
10193 Do not use floating-point coprocessor instructions. Implement
10194 floating-point calculations using library calls instead.
10196 @item -msingle-float
10197 @opindex msingle-float
10198 Assume that the floating-point coprocessor only supports single-precision
10201 @itemx -mdouble-float
10202 @opindex mdouble-float
10203 Assume that the floating-point coprocessor supports double-precision
10204 operations. This is the default.
10210 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10212 @itemx -mpaired-single
10213 @itemx -mno-paired-single
10214 @opindex mpaired-single
10215 @opindex mno-paired-single
10216 Use (do not use) paired-single floating-point instructions.
10217 @xref{MIPS Paired-Single Support}. This option can only be used
10218 when generating 64-bit code and requires hardware floating-point
10219 support to be enabled.
10224 @opindex mno-mips3d
10225 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10226 The option @option{-mips3d} implies @option{-mpaired-single}.
10230 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10231 an explanation of the default and the way that the pointer size is
10236 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10238 The default size of @code{int}s, @code{long}s and pointers depends on
10239 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10240 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10241 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10242 or the same size as integer registers, whichever is smaller.
10248 Assume (do not assume) that all symbols have 32-bit values, regardless
10249 of the selected ABI@. This option is useful in combination with
10250 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10251 to generate shorter and faster references to symbolic addresses.
10255 @cindex smaller data references (MIPS)
10256 @cindex gp-relative references (MIPS)
10257 Put global and static items less than or equal to @var{num} bytes into
10258 the small data or bss section instead of the normal data or bss section.
10259 This allows the data to be accessed using a single instruction.
10261 All modules should be compiled with the same @option{-G @var{num}}
10264 @item -membedded-data
10265 @itemx -mno-embedded-data
10266 @opindex membedded-data
10267 @opindex mno-embedded-data
10268 Allocate variables to the read-only data section first if possible, then
10269 next in the small data section if possible, otherwise in data. This gives
10270 slightly slower code than the default, but reduces the amount of RAM required
10271 when executing, and thus may be preferred for some embedded systems.
10273 @item -muninit-const-in-rodata
10274 @itemx -mno-uninit-const-in-rodata
10275 @opindex muninit-const-in-rodata
10276 @opindex mno-uninit-const-in-rodata
10277 Put uninitialized @code{const} variables in the read-only data section.
10278 This option is only meaningful in conjunction with @option{-membedded-data}.
10280 @item -msplit-addresses
10281 @itemx -mno-split-addresses
10282 @opindex msplit-addresses
10283 @opindex mno-split-addresses
10284 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10285 relocation operators. This option has been superseded by
10286 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10288 @item -mexplicit-relocs
10289 @itemx -mno-explicit-relocs
10290 @opindex mexplicit-relocs
10291 @opindex mno-explicit-relocs
10292 Use (do not use) assembler relocation operators when dealing with symbolic
10293 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10294 is to use assembler macros instead.
10296 @option{-mexplicit-relocs} is the default if GCC was configured
10297 to use an assembler that supports relocation operators.
10299 @item -mcheck-zero-division
10300 @itemx -mno-check-zero-division
10301 @opindex mcheck-zero-division
10302 @opindex mno-check-zero-division
10303 Trap (do not trap) on integer division by zero. The default is
10304 @option{-mcheck-zero-division}.
10306 @item -mdivide-traps
10307 @itemx -mdivide-breaks
10308 @opindex mdivide-traps
10309 @opindex mdivide-breaks
10310 MIPS systems check for division by zero by generating either a
10311 conditional trap or a break instruction. Using traps results in
10312 smaller code, but is only supported on MIPS II and later. Also, some
10313 versions of the Linux kernel have a bug that prevents trap from
10314 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10315 allow conditional traps on architectures that support them and
10316 @option{-mdivide-breaks} to force the use of breaks.
10318 The default is usually @option{-mdivide-traps}, but this can be
10319 overridden at configure time using @option{--with-divide=breaks}.
10320 Divide-by-zero checks can be completely disabled using
10321 @option{-mno-check-zero-division}.
10326 @opindex mno-memcpy
10327 Force (do not force) the use of @code{memcpy()} for non-trivial block
10328 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10329 most constant-sized copies.
10332 @itemx -mno-long-calls
10333 @opindex mlong-calls
10334 @opindex mno-long-calls
10335 Disable (do not disable) use of the @code{jal} instruction. Calling
10336 functions using @code{jal} is more efficient but requires the caller
10337 and callee to be in the same 256 megabyte segment.
10339 This option has no effect on abicalls code. The default is
10340 @option{-mno-long-calls}.
10346 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10347 instructions, as provided by the R4650 ISA@.
10350 @itemx -mno-fused-madd
10351 @opindex mfused-madd
10352 @opindex mno-fused-madd
10353 Enable (disable) use of the floating point multiply-accumulate
10354 instructions, when they are available. The default is
10355 @option{-mfused-madd}.
10357 When multiply-accumulate instructions are used, the intermediate
10358 product is calculated to infinite precision and is not subject to
10359 the FCSR Flush to Zero bit. This may be undesirable in some
10364 Tell the MIPS assembler to not run its preprocessor over user
10365 assembler files (with a @samp{.s} suffix) when assembling them.
10368 @itemx -mno-fix-r4000
10369 @opindex mfix-r4000
10370 @opindex mno-fix-r4000
10371 Work around certain R4000 CPU errata:
10374 A double-word or a variable shift may give an incorrect result if executed
10375 immediately after starting an integer division.
10377 A double-word or a variable shift may give an incorrect result if executed
10378 while an integer multiplication is in progress.
10380 An integer division may give an incorrect result if started in a delay slot
10381 of a taken branch or a jump.
10385 @itemx -mno-fix-r4400
10386 @opindex mfix-r4400
10387 @opindex mno-fix-r4400
10388 Work around certain R4400 CPU errata:
10391 A double-word or a variable shift may give an incorrect result if executed
10392 immediately after starting an integer division.
10396 @itemx -mno-fix-vr4120
10397 @opindex mfix-vr4120
10398 Work around certain VR4120 errata:
10401 @code{dmultu} does not always produce the correct result.
10403 @code{div} and @code{ddiv} do not always produce the correct result if one
10404 of the operands is negative.
10406 The workarounds for the division errata rely on special functions in
10407 @file{libgcc.a}. At present, these functions are only provided by
10408 the @code{mips64vr*-elf} configurations.
10410 Other VR4120 errata require a nop to be inserted between certain pairs of
10411 instructions. These errata are handled by the assembler, not by GCC itself.
10414 @opindex mfix-vr4130
10415 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10416 workarounds are implemented by the assembler rather than by GCC,
10417 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10418 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10419 instructions are available instead.
10422 @itemx -mno-fix-sb1
10424 Work around certain SB-1 CPU core errata.
10425 (This flag currently works around the SB-1 revision 2
10426 ``F1'' and ``F2'' floating point errata.)
10428 @item -mflush-func=@var{func}
10429 @itemx -mno-flush-func
10430 @opindex mflush-func
10431 Specifies the function to call to flush the I and D caches, or to not
10432 call any such function. If called, the function must take the same
10433 arguments as the common @code{_flush_func()}, that is, the address of the
10434 memory range for which the cache is being flushed, the size of the
10435 memory range, and the number 3 (to flush both caches). The default
10436 depends on the target GCC was configured for, but commonly is either
10437 @samp{_flush_func} or @samp{__cpu_flush}.
10439 @item -mbranch-likely
10440 @itemx -mno-branch-likely
10441 @opindex mbranch-likely
10442 @opindex mno-branch-likely
10443 Enable or disable use of Branch Likely instructions, regardless of the
10444 default for the selected architecture. By default, Branch Likely
10445 instructions may be generated if they are supported by the selected
10446 architecture. An exception is for the MIPS32 and MIPS64 architectures
10447 and processors which implement those architectures; for those, Branch
10448 Likely instructions will not be generated by default because the MIPS32
10449 and MIPS64 architectures specifically deprecate their use.
10451 @item -mfp-exceptions
10452 @itemx -mno-fp-exceptions
10453 @opindex mfp-exceptions
10454 Specifies whether FP exceptions are enabled. This affects how we schedule
10455 FP instructions for some processors. The default is that FP exceptions are
10458 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
10459 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
10462 @item -mvr4130-align
10463 @itemx -mno-vr4130-align
10464 @opindex mvr4130-align
10465 The VR4130 pipeline is two-way superscalar, but can only issue two
10466 instructions together if the first one is 8-byte aligned. When this
10467 option is enabled, GCC will align pairs of instructions that it
10468 thinks should execute in parallel.
10470 This option only has an effect when optimizing for the VR4130.
10471 It normally makes code faster, but at the expense of making it bigger.
10472 It is enabled by default at optimization level @option{-O3}.
10476 @subsection MMIX Options
10477 @cindex MMIX Options
10479 These options are defined for the MMIX:
10483 @itemx -mno-libfuncs
10485 @opindex mno-libfuncs
10486 Specify that intrinsic library functions are being compiled, passing all
10487 values in registers, no matter the size.
10490 @itemx -mno-epsilon
10492 @opindex mno-epsilon
10493 Generate floating-point comparison instructions that compare with respect
10494 to the @code{rE} epsilon register.
10496 @item -mabi=mmixware
10498 @opindex mabi-mmixware
10500 Generate code that passes function parameters and return values that (in
10501 the called function) are seen as registers @code{$0} and up, as opposed to
10502 the GNU ABI which uses global registers @code{$231} and up.
10504 @item -mzero-extend
10505 @itemx -mno-zero-extend
10506 @opindex mzero-extend
10507 @opindex mno-zero-extend
10508 When reading data from memory in sizes shorter than 64 bits, use (do not
10509 use) zero-extending load instructions by default, rather than
10510 sign-extending ones.
10513 @itemx -mno-knuthdiv
10515 @opindex mno-knuthdiv
10516 Make the result of a division yielding a remainder have the same sign as
10517 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10518 remainder follows the sign of the dividend. Both methods are
10519 arithmetically valid, the latter being almost exclusively used.
10521 @item -mtoplevel-symbols
10522 @itemx -mno-toplevel-symbols
10523 @opindex mtoplevel-symbols
10524 @opindex mno-toplevel-symbols
10525 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10526 code can be used with the @code{PREFIX} assembly directive.
10530 Generate an executable in the ELF format, rather than the default
10531 @samp{mmo} format used by the @command{mmix} simulator.
10533 @item -mbranch-predict
10534 @itemx -mno-branch-predict
10535 @opindex mbranch-predict
10536 @opindex mno-branch-predict
10537 Use (do not use) the probable-branch instructions, when static branch
10538 prediction indicates a probable branch.
10540 @item -mbase-addresses
10541 @itemx -mno-base-addresses
10542 @opindex mbase-addresses
10543 @opindex mno-base-addresses
10544 Generate (do not generate) code that uses @emph{base addresses}. Using a
10545 base address automatically generates a request (handled by the assembler
10546 and the linker) for a constant to be set up in a global register. The
10547 register is used for one or more base address requests within the range 0
10548 to 255 from the value held in the register. The generally leads to short
10549 and fast code, but the number of different data items that can be
10550 addressed is limited. This means that a program that uses lots of static
10551 data may require @option{-mno-base-addresses}.
10553 @item -msingle-exit
10554 @itemx -mno-single-exit
10555 @opindex msingle-exit
10556 @opindex mno-single-exit
10557 Force (do not force) generated code to have a single exit point in each
10561 @node MN10300 Options
10562 @subsection MN10300 Options
10563 @cindex MN10300 options
10565 These @option{-m} options are defined for Matsushita MN10300 architectures:
10570 Generate code to avoid bugs in the multiply instructions for the MN10300
10571 processors. This is the default.
10573 @item -mno-mult-bug
10574 @opindex mno-mult-bug
10575 Do not generate code to avoid bugs in the multiply instructions for the
10576 MN10300 processors.
10580 Generate code which uses features specific to the AM33 processor.
10584 Do not generate code which uses features specific to the AM33 processor. This
10589 Do not link in the C run-time initialization object file.
10593 Indicate to the linker that it should perform a relaxation optimization pass
10594 to shorten branches, calls and absolute memory addresses. This option only
10595 has an effect when used on the command line for the final link step.
10597 This option makes symbolic debugging impossible.
10601 @subsection MS1 Options
10602 @cindex MS1 options
10604 These @option{-m} options are defined for Morpho MS1 architectures:
10608 @item -march=@var{cpu-type}
10610 Generate code that will run on @var{cpu-type}, which is the name of a system
10611 representing a certain processor type. Possible values for
10612 @var{cpu-type} are @samp{MS1-64-001}, @samp{MS1-16-002}, and
10615 When this option is not used, the default is @option{-march=MS1-16-003}.
10619 Generate multiply instructions.
10623 Do not generate multiply instructions.
10627 Use byte loads and stores when generating code.
10631 Do not use byte loads and stores when generating code.
10635 Use simulator runtime
10639 Do not link in the C run-time initialization object file
10640 @file{crti.o}. Other run-time initialization and termination files
10641 such as @file{startup.o} and @file{exit.o} are still included on the
10642 linker command line.
10646 @node PDP-11 Options
10647 @subsection PDP-11 Options
10648 @cindex PDP-11 Options
10650 These options are defined for the PDP-11:
10655 Use hardware FPP floating point. This is the default. (FIS floating
10656 point on the PDP-11/40 is not supported.)
10659 @opindex msoft-float
10660 Do not use hardware floating point.
10664 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10668 Return floating-point results in memory. This is the default.
10672 Generate code for a PDP-11/40.
10676 Generate code for a PDP-11/45. This is the default.
10680 Generate code for a PDP-11/10.
10682 @item -mbcopy-builtin
10683 @opindex bcopy-builtin
10684 Use inline @code{movmemhi} patterns for copying memory. This is the
10689 Do not use inline @code{movmemhi} patterns for copying memory.
10695 Use 16-bit @code{int}. This is the default.
10701 Use 32-bit @code{int}.
10704 @itemx -mno-float32
10706 @opindex mno-float32
10707 Use 64-bit @code{float}. This is the default.
10710 @itemx -mno-float64
10712 @opindex mno-float64
10713 Use 32-bit @code{float}.
10717 Use @code{abshi2} pattern. This is the default.
10721 Do not use @code{abshi2} pattern.
10723 @item -mbranch-expensive
10724 @opindex mbranch-expensive
10725 Pretend that branches are expensive. This is for experimenting with
10726 code generation only.
10728 @item -mbranch-cheap
10729 @opindex mbranch-cheap
10730 Do not pretend that branches are expensive. This is the default.
10734 Generate code for a system with split I&D@.
10738 Generate code for a system without split I&D@. This is the default.
10742 Use Unix assembler syntax. This is the default when configured for
10743 @samp{pdp11-*-bsd}.
10747 Use DEC assembler syntax. This is the default when configured for any
10748 PDP-11 target other than @samp{pdp11-*-bsd}.
10751 @node PowerPC Options
10752 @subsection PowerPC Options
10753 @cindex PowerPC options
10755 These are listed under @xref{RS/6000 and PowerPC Options}.
10757 @node RS/6000 and PowerPC Options
10758 @subsection IBM RS/6000 and PowerPC Options
10759 @cindex RS/6000 and PowerPC Options
10760 @cindex IBM RS/6000 and PowerPC Options
10762 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
10769 @itemx -mno-powerpc
10770 @itemx -mpowerpc-gpopt
10771 @itemx -mno-powerpc-gpopt
10772 @itemx -mpowerpc-gfxopt
10773 @itemx -mno-powerpc-gfxopt
10775 @itemx -mno-powerpc64
10779 @opindex mno-power2
10781 @opindex mno-powerpc
10782 @opindex mpowerpc-gpopt
10783 @opindex mno-powerpc-gpopt
10784 @opindex mpowerpc-gfxopt
10785 @opindex mno-powerpc-gfxopt
10786 @opindex mpowerpc64
10787 @opindex mno-powerpc64
10788 GCC supports two related instruction set architectures for the
10789 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
10790 instructions supported by the @samp{rios} chip set used in the original
10791 RS/6000 systems and the @dfn{PowerPC} instruction set is the
10792 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
10793 the IBM 4xx microprocessors.
10795 Neither architecture is a subset of the other. However there is a
10796 large common subset of instructions supported by both. An MQ
10797 register is included in processors supporting the POWER architecture.
10799 You use these options to specify which instructions are available on the
10800 processor you are using. The default value of these options is
10801 determined when configuring GCC@. Specifying the
10802 @option{-mcpu=@var{cpu_type}} overrides the specification of these
10803 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
10804 rather than the options listed above.
10806 The @option{-mpower} option allows GCC to generate instructions that
10807 are found only in the POWER architecture and to use the MQ register.
10808 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
10809 to generate instructions that are present in the POWER2 architecture but
10810 not the original POWER architecture.
10812 The @option{-mpowerpc} option allows GCC to generate instructions that
10813 are found only in the 32-bit subset of the PowerPC architecture.
10814 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
10815 GCC to use the optional PowerPC architecture instructions in the
10816 General Purpose group, including floating-point square root. Specifying
10817 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
10818 use the optional PowerPC architecture instructions in the Graphics
10819 group, including floating-point select.
10821 The @option{-mpowerpc64} option allows GCC to generate the additional
10822 64-bit instructions that are found in the full PowerPC64 architecture
10823 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
10824 @option{-mno-powerpc64}.
10826 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
10827 will use only the instructions in the common subset of both
10828 architectures plus some special AIX common-mode calls, and will not use
10829 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
10830 permits GCC to use any instruction from either architecture and to
10831 allow use of the MQ register; specify this for the Motorola MPC601.
10833 @item -mnew-mnemonics
10834 @itemx -mold-mnemonics
10835 @opindex mnew-mnemonics
10836 @opindex mold-mnemonics
10837 Select which mnemonics to use in the generated assembler code. With
10838 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
10839 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
10840 assembler mnemonics defined for the POWER architecture. Instructions
10841 defined in only one architecture have only one mnemonic; GCC uses that
10842 mnemonic irrespective of which of these options is specified.
10844 GCC defaults to the mnemonics appropriate for the architecture in
10845 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
10846 value of these option. Unless you are building a cross-compiler, you
10847 should normally not specify either @option{-mnew-mnemonics} or
10848 @option{-mold-mnemonics}, but should instead accept the default.
10850 @item -mcpu=@var{cpu_type}
10852 Set architecture type, register usage, choice of mnemonics, and
10853 instruction scheduling parameters for machine type @var{cpu_type}.
10854 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
10855 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
10856 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
10857 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
10858 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
10859 @samp{860}, @samp{970}, @samp{8540}, @samp{common}, @samp{ec603e}, @samp{G3},
10860 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
10861 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
10862 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
10864 @option{-mcpu=common} selects a completely generic processor. Code
10865 generated under this option will run on any POWER or PowerPC processor.
10866 GCC will use only the instructions in the common subset of both
10867 architectures, and will not use the MQ register. GCC assumes a generic
10868 processor model for scheduling purposes.
10870 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
10871 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
10872 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
10873 types, with an appropriate, generic processor model assumed for
10874 scheduling purposes.
10876 The other options specify a specific processor. Code generated under
10877 those options will run best on that processor, and may not run at all on
10880 The @option{-mcpu} options automatically enable or disable the
10881 following options: @option{-maltivec}, @option{-mhard-float},
10882 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
10883 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
10884 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
10885 @option{-mstring}. The particular options set for any particular CPU
10886 will vary between compiler versions, depending on what setting seems
10887 to produce optimal code for that CPU; it doesn't necessarily reflect
10888 the actual hardware's capabilities. If you wish to set an individual
10889 option to a particular value, you may specify it after the
10890 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
10892 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
10893 not enabled or disabled by the @option{-mcpu} option at present, since
10894 AIX does not have full support for these options. You may still
10895 enable or disable them individually if you're sure it'll work in your
10898 @item -mtune=@var{cpu_type}
10900 Set the instruction scheduling parameters for machine type
10901 @var{cpu_type}, but do not set the architecture type, register usage, or
10902 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
10903 values for @var{cpu_type} are used for @option{-mtune} as for
10904 @option{-mcpu}. If both are specified, the code generated will use the
10905 architecture, registers, and mnemonics set by @option{-mcpu}, but the
10906 scheduling parameters set by @option{-mtune}.
10912 Generate code to compute division as reciprocal estimate and iterative
10913 refinement, creating opportunities for increased throughput. This
10914 feature requires: optional PowerPC Graphics instruction set for single
10915 precision and FRE instruction for double precision, assuming divides
10916 cannot generate user-visible traps, and the domain values not include
10917 Infinities, denormals or zero denominator.
10920 @itemx -mno-altivec
10922 @opindex mno-altivec
10923 Generate code that uses (does not use) AltiVec instructions, and also
10924 enable the use of built-in functions that allow more direct access to
10925 the AltiVec instruction set. You may also need to set
10926 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
10932 @opindex mno-vrsave
10933 Generate VRSAVE instructions when generating AltiVec code.
10937 Extend the current ABI with SPE ABI extensions. This does not change
10938 the default ABI, instead it adds the SPE ABI extensions to the current
10942 @opindex mabi=no-spe
10943 Disable Booke SPE ABI extensions for the current ABI@.
10946 @opindex msecure-plt
10947 Generate code that allows ld and ld.so to build executables and shared
10948 libraries with non-exec .plt and .got sections. This is a PowerPC
10949 32-bit SYSV ABI option.
10953 Generate code that uses a BSS .plt section that ld.so fills in, and
10954 requires .plt and .got sections that are both writable and executable.
10955 This is a PowerPC 32-bit SYSV ABI option.
10961 This switch enables or disables the generation of ISEL instructions.
10963 @item -misel=@var{yes/no}
10964 This switch has been deprecated. Use @option{-misel} and
10965 @option{-mno-isel} instead.
10971 This switch enables or disables the generation of SPE simd
10974 @item -mspe=@var{yes/no}
10975 This option has been deprecated. Use @option{-mspe} and
10976 @option{-mno-spe} instead.
10978 @item -mfloat-gprs=@var{yes/single/double/no}
10979 @itemx -mfloat-gprs
10980 @opindex mfloat-gprs
10981 This switch enables or disables the generation of floating point
10982 operations on the general purpose registers for architectures that
10985 The argument @var{yes} or @var{single} enables the use of
10986 single-precision floating point operations.
10988 The argument @var{double} enables the use of single and
10989 double-precision floating point operations.
10991 The argument @var{no} disables floating point operations on the
10992 general purpose registers.
10994 This option is currently only available on the MPC854x.
11000 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11001 targets (including GNU/Linux). The 32-bit environment sets int, long
11002 and pointer to 32 bits and generates code that runs on any PowerPC
11003 variant. The 64-bit environment sets int to 32 bits and long and
11004 pointer to 64 bits, and generates code for PowerPC64, as for
11005 @option{-mpowerpc64}.
11008 @itemx -mno-fp-in-toc
11009 @itemx -mno-sum-in-toc
11010 @itemx -mminimal-toc
11012 @opindex mno-fp-in-toc
11013 @opindex mno-sum-in-toc
11014 @opindex mminimal-toc
11015 Modify generation of the TOC (Table Of Contents), which is created for
11016 every executable file. The @option{-mfull-toc} option is selected by
11017 default. In that case, GCC will allocate at least one TOC entry for
11018 each unique non-automatic variable reference in your program. GCC
11019 will also place floating-point constants in the TOC@. However, only
11020 16,384 entries are available in the TOC@.
11022 If you receive a linker error message that saying you have overflowed
11023 the available TOC space, you can reduce the amount of TOC space used
11024 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11025 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11026 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11027 generate code to calculate the sum of an address and a constant at
11028 run-time instead of putting that sum into the TOC@. You may specify one
11029 or both of these options. Each causes GCC to produce very slightly
11030 slower and larger code at the expense of conserving TOC space.
11032 If you still run out of space in the TOC even when you specify both of
11033 these options, specify @option{-mminimal-toc} instead. This option causes
11034 GCC to make only one TOC entry for every file. When you specify this
11035 option, GCC will produce code that is slower and larger but which
11036 uses extremely little TOC space. You may wish to use this option
11037 only on files that contain less frequently executed code.
11043 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11044 @code{long} type, and the infrastructure needed to support them.
11045 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11046 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11047 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11050 @itemx -mno-xl-compat
11051 @opindex mxl-compat
11052 @opindex mno-xl-compat
11053 Produce code that conforms more closely to IBM XLC semantics when using
11054 AIX-compatible ABI. Pass floating-point arguments to prototyped
11055 functions beyond the register save area (RSA) on the stack in addition
11056 to argument FPRs. Do not assume that most significant double in 128
11057 bit long double value is properly rounded when comparing values.
11059 The AIX calling convention was extended but not initially documented to
11060 handle an obscure K&R C case of calling a function that takes the
11061 address of its arguments with fewer arguments than declared. AIX XL
11062 compilers access floating point arguments which do not fit in the
11063 RSA from the stack when a subroutine is compiled without
11064 optimization. Because always storing floating-point arguments on the
11065 stack is inefficient and rarely needed, this option is not enabled by
11066 default and only is necessary when calling subroutines compiled by AIX
11067 XL compilers without optimization.
11071 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11072 application written to use message passing with special startup code to
11073 enable the application to run. The system must have PE installed in the
11074 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11075 must be overridden with the @option{-specs=} option to specify the
11076 appropriate directory location. The Parallel Environment does not
11077 support threads, so the @option{-mpe} option and the @option{-pthread}
11078 option are incompatible.
11080 @item -malign-natural
11081 @itemx -malign-power
11082 @opindex malign-natural
11083 @opindex malign-power
11084 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11085 @option{-malign-natural} overrides the ABI-defined alignment of larger
11086 types, such as floating-point doubles, on their natural size-based boundary.
11087 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11088 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11090 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11094 @itemx -mhard-float
11095 @opindex msoft-float
11096 @opindex mhard-float
11097 Generate code that does not use (uses) the floating-point register set.
11098 Software floating point emulation is provided if you use the
11099 @option{-msoft-float} option, and pass the option to GCC when linking.
11102 @itemx -mno-multiple
11104 @opindex mno-multiple
11105 Generate code that uses (does not use) the load multiple word
11106 instructions and the store multiple word instructions. These
11107 instructions are generated by default on POWER systems, and not
11108 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11109 endian PowerPC systems, since those instructions do not work when the
11110 processor is in little endian mode. The exceptions are PPC740 and
11111 PPC750 which permit the instructions usage in little endian mode.
11116 @opindex mno-string
11117 Generate code that uses (does not use) the load string instructions
11118 and the store string word instructions to save multiple registers and
11119 do small block moves. These instructions are generated by default on
11120 POWER systems, and not generated on PowerPC systems. Do not use
11121 @option{-mstring} on little endian PowerPC systems, since those
11122 instructions do not work when the processor is in little endian mode.
11123 The exceptions are PPC740 and PPC750 which permit the instructions
11124 usage in little endian mode.
11129 @opindex mno-update
11130 Generate code that uses (does not use) the load or store instructions
11131 that update the base register to the address of the calculated memory
11132 location. These instructions are generated by default. If you use
11133 @option{-mno-update}, there is a small window between the time that the
11134 stack pointer is updated and the address of the previous frame is
11135 stored, which means code that walks the stack frame across interrupts or
11136 signals may get corrupted data.
11139 @itemx -mno-fused-madd
11140 @opindex mfused-madd
11141 @opindex mno-fused-madd
11142 Generate code that uses (does not use) the floating point multiply and
11143 accumulate instructions. These instructions are generated by default if
11144 hardware floating is used.
11146 @item -mno-bit-align
11148 @opindex mno-bit-align
11149 @opindex mbit-align
11150 On System V.4 and embedded PowerPC systems do not (do) force structures
11151 and unions that contain bit-fields to be aligned to the base type of the
11154 For example, by default a structure containing nothing but 8
11155 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11156 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11157 the structure would be aligned to a 1 byte boundary and be one byte in
11160 @item -mno-strict-align
11161 @itemx -mstrict-align
11162 @opindex mno-strict-align
11163 @opindex mstrict-align
11164 On System V.4 and embedded PowerPC systems do not (do) assume that
11165 unaligned memory references will be handled by the system.
11167 @item -mrelocatable
11168 @itemx -mno-relocatable
11169 @opindex mrelocatable
11170 @opindex mno-relocatable
11171 On embedded PowerPC systems generate code that allows (does not allow)
11172 the program to be relocated to a different address at runtime. If you
11173 use @option{-mrelocatable} on any module, all objects linked together must
11174 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11176 @item -mrelocatable-lib
11177 @itemx -mno-relocatable-lib
11178 @opindex mrelocatable-lib
11179 @opindex mno-relocatable-lib
11180 On embedded PowerPC systems generate code that allows (does not allow)
11181 the program to be relocated to a different address at runtime. Modules
11182 compiled with @option{-mrelocatable-lib} can be linked with either modules
11183 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11184 with modules compiled with the @option{-mrelocatable} options.
11190 On System V.4 and embedded PowerPC systems do not (do) assume that
11191 register 2 contains a pointer to a global area pointing to the addresses
11192 used in the program.
11195 @itemx -mlittle-endian
11197 @opindex mlittle-endian
11198 On System V.4 and embedded PowerPC systems compile code for the
11199 processor in little endian mode. The @option{-mlittle-endian} option is
11200 the same as @option{-mlittle}.
11203 @itemx -mbig-endian
11205 @opindex mbig-endian
11206 On System V.4 and embedded PowerPC systems compile code for the
11207 processor in big endian mode. The @option{-mbig-endian} option is
11208 the same as @option{-mbig}.
11210 @item -mdynamic-no-pic
11211 @opindex mdynamic-no-pic
11212 On Darwin and Mac OS X systems, compile code so that it is not
11213 relocatable, but that its external references are relocatable. The
11214 resulting code is suitable for applications, but not shared
11217 @item -mprioritize-restricted-insns=@var{priority}
11218 @opindex mprioritize-restricted-insns
11219 This option controls the priority that is assigned to
11220 dispatch-slot restricted instructions during the second scheduling
11221 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11222 @var{no/highest/second-highest} priority to dispatch slot restricted
11225 @item -msched-costly-dep=@var{dependence_type}
11226 @opindex msched-costly-dep
11227 This option controls which dependences are considered costly
11228 by the target during instruction scheduling. The argument
11229 @var{dependence_type} takes one of the following values:
11230 @var{no}: no dependence is costly,
11231 @var{all}: all dependences are costly,
11232 @var{true_store_to_load}: a true dependence from store to load is costly,
11233 @var{store_to_load}: any dependence from store to load is costly,
11234 @var{number}: any dependence which latency >= @var{number} is costly.
11236 @item -minsert-sched-nops=@var{scheme}
11237 @opindex minsert-sched-nops
11238 This option controls which nop insertion scheme will be used during
11239 the second scheduling pass. The argument @var{scheme} takes one of the
11241 @var{no}: Don't insert nops.
11242 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11243 according to the scheduler's grouping.
11244 @var{regroup_exact}: Insert nops to force costly dependent insns into
11245 separate groups. Insert exactly as many nops as needed to force an insn
11246 to a new group, according to the estimated processor grouping.
11247 @var{number}: Insert nops to force costly dependent insns into
11248 separate groups. Insert @var{number} nops to force an insn to a new group.
11251 @opindex mcall-sysv
11252 On System V.4 and embedded PowerPC systems compile code using calling
11253 conventions that adheres to the March 1995 draft of the System V
11254 Application Binary Interface, PowerPC processor supplement. This is the
11255 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11257 @item -mcall-sysv-eabi
11258 @opindex mcall-sysv-eabi
11259 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11261 @item -mcall-sysv-noeabi
11262 @opindex mcall-sysv-noeabi
11263 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11265 @item -mcall-solaris
11266 @opindex mcall-solaris
11267 On System V.4 and embedded PowerPC systems compile code for the Solaris
11271 @opindex mcall-linux
11272 On System V.4 and embedded PowerPC systems compile code for the
11273 Linux-based GNU system.
11277 On System V.4 and embedded PowerPC systems compile code for the
11278 Hurd-based GNU system.
11280 @item -mcall-netbsd
11281 @opindex mcall-netbsd
11282 On System V.4 and embedded PowerPC systems compile code for the
11283 NetBSD operating system.
11285 @item -maix-struct-return
11286 @opindex maix-struct-return
11287 Return all structures in memory (as specified by the AIX ABI)@.
11289 @item -msvr4-struct-return
11290 @opindex msvr4-struct-return
11291 Return structures smaller than 8 bytes in registers (as specified by the
11294 @item -mabi=@var{abi-type}
11296 Extend the current ABI with a particular extension, or remove such extension.
11297 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11301 @itemx -mno-prototype
11302 @opindex mprototype
11303 @opindex mno-prototype
11304 On System V.4 and embedded PowerPC systems assume that all calls to
11305 variable argument functions are properly prototyped. Otherwise, the
11306 compiler must insert an instruction before every non prototyped call to
11307 set or clear bit 6 of the condition code register (@var{CR}) to
11308 indicate whether floating point values were passed in the floating point
11309 registers in case the function takes a variable arguments. With
11310 @option{-mprototype}, only calls to prototyped variable argument functions
11311 will set or clear the bit.
11315 On embedded PowerPC systems, assume that the startup module is called
11316 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11317 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11322 On embedded PowerPC systems, assume that the startup module is called
11323 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11328 On embedded PowerPC systems, assume that the startup module is called
11329 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11332 @item -myellowknife
11333 @opindex myellowknife
11334 On embedded PowerPC systems, assume that the startup module is called
11335 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11340 On System V.4 and embedded PowerPC systems, specify that you are
11341 compiling for a VxWorks system.
11345 Specify that you are compiling for the WindISS simulation environment.
11349 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11350 header to indicate that @samp{eabi} extended relocations are used.
11356 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11357 Embedded Applications Binary Interface (eabi) which is a set of
11358 modifications to the System V.4 specifications. Selecting @option{-meabi}
11359 means that the stack is aligned to an 8 byte boundary, a function
11360 @code{__eabi} is called to from @code{main} to set up the eabi
11361 environment, and the @option{-msdata} option can use both @code{r2} and
11362 @code{r13} to point to two separate small data areas. Selecting
11363 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11364 do not call an initialization function from @code{main}, and the
11365 @option{-msdata} option will only use @code{r13} to point to a single
11366 small data area. The @option{-meabi} option is on by default if you
11367 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11370 @opindex msdata=eabi
11371 On System V.4 and embedded PowerPC systems, put small initialized
11372 @code{const} global and static data in the @samp{.sdata2} section, which
11373 is pointed to by register @code{r2}. Put small initialized
11374 non-@code{const} global and static data in the @samp{.sdata} section,
11375 which is pointed to by register @code{r13}. Put small uninitialized
11376 global and static data in the @samp{.sbss} section, which is adjacent to
11377 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11378 incompatible with the @option{-mrelocatable} option. The
11379 @option{-msdata=eabi} option also sets the @option{-memb} option.
11382 @opindex msdata=sysv
11383 On System V.4 and embedded PowerPC systems, put small global and static
11384 data in the @samp{.sdata} section, which is pointed to by register
11385 @code{r13}. Put small uninitialized global and static data in the
11386 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
11387 The @option{-msdata=sysv} option is incompatible with the
11388 @option{-mrelocatable} option.
11390 @item -msdata=default
11392 @opindex msdata=default
11394 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
11395 compile code the same as @option{-msdata=eabi}, otherwise compile code the
11396 same as @option{-msdata=sysv}.
11399 @opindex msdata-data
11400 On System V.4 and embedded PowerPC systems, put small global and static
11401 data in the @samp{.sdata} section. Put small uninitialized global and
11402 static data in the @samp{.sbss} section. Do not use register @code{r13}
11403 to address small data however. This is the default behavior unless
11404 other @option{-msdata} options are used.
11408 @opindex msdata=none
11410 On embedded PowerPC systems, put all initialized global and static data
11411 in the @samp{.data} section, and all uninitialized data in the
11412 @samp{.bss} section.
11416 @cindex smaller data references (PowerPC)
11417 @cindex .sdata/.sdata2 references (PowerPC)
11418 On embedded PowerPC systems, put global and static items less than or
11419 equal to @var{num} bytes into the small data or bss sections instead of
11420 the normal data or bss section. By default, @var{num} is 8. The
11421 @option{-G @var{num}} switch is also passed to the linker.
11422 All modules should be compiled with the same @option{-G @var{num}} value.
11425 @itemx -mno-regnames
11427 @opindex mno-regnames
11428 On System V.4 and embedded PowerPC systems do (do not) emit register
11429 names in the assembly language output using symbolic forms.
11432 @itemx -mno-longcall
11434 @opindex mno-longcall
11435 Default to making all function calls indirectly, using a register, so
11436 that functions which reside further than 32 megabytes (33,554,432
11437 bytes) from the current location can be called. This setting can be
11438 overridden by the @code{shortcall} function attribute, or by
11439 @code{#pragma longcall(0)}.
11441 Some linkers are capable of detecting out-of-range calls and generating
11442 glue code on the fly. On these systems, long calls are unnecessary and
11443 generate slower code. As of this writing, the AIX linker can do this,
11444 as can the GNU linker for PowerPC/64. It is planned to add this feature
11445 to the GNU linker for 32-bit PowerPC systems as well.
11447 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
11448 callee, L42'', plus a ``branch island'' (glue code). The two target
11449 addresses represent the callee and the ``branch island''. The
11450 Darwin/PPC linker will prefer the first address and generate a ``bl
11451 callee'' if the PPC ``bl'' instruction will reach the callee directly;
11452 otherwise, the linker will generate ``bl L42'' to call the ``branch
11453 island''. The ``branch island'' is appended to the body of the
11454 calling function; it computes the full 32-bit address of the callee
11457 On Mach-O (Darwin) systems, this option directs the compiler emit to
11458 the glue for every direct call, and the Darwin linker decides whether
11459 to use or discard it.
11461 In the future, we may cause GCC to ignore all longcall specifications
11462 when the linker is known to generate glue.
11466 Adds support for multithreading with the @dfn{pthreads} library.
11467 This option sets flags for both the preprocessor and linker.
11471 @node S/390 and zSeries Options
11472 @subsection S/390 and zSeries Options
11473 @cindex S/390 and zSeries Options
11475 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
11479 @itemx -msoft-float
11480 @opindex mhard-float
11481 @opindex msoft-float
11482 Use (do not use) the hardware floating-point instructions and registers
11483 for floating-point operations. When @option{-msoft-float} is specified,
11484 functions in @file{libgcc.a} will be used to perform floating-point
11485 operations. When @option{-mhard-float} is specified, the compiler
11486 generates IEEE floating-point instructions. This is the default.
11489 @itemx -mno-backchain
11490 @opindex mbackchain
11491 @opindex mno-backchain
11492 Store (do not store) the address of the caller's frame as backchain pointer
11493 into the callee's stack frame.
11494 A backchain may be needed to allow debugging using tools that do not understand
11495 DWARF-2 call frame information.
11496 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
11497 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
11498 the backchain is placed into the topmost word of the 96/160 byte register
11501 In general, code compiled with @option{-mbackchain} is call-compatible with
11502 code compiled with @option{-mmo-backchain}; however, use of the backchain
11503 for debugging purposes usually requires that the whole binary is built with
11504 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
11505 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11506 to build a linux kernel use @option{-msoft-float}.
11508 The default is to not maintain the backchain.
11510 @item -mpacked-stack
11511 @item -mno-packed-stack
11512 @opindex mpacked-stack
11513 @opindex mno-packed-stack
11514 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
11515 specified, the compiler uses the all fields of the 96/160 byte register save
11516 area only for their default purpose; unused fields still take up stack space.
11517 When @option{-mpacked-stack} is specified, register save slots are densely
11518 packed at the top of the register save area; unused space is reused for other
11519 purposes, allowing for more efficient use of the available stack space.
11520 However, when @option{-mbackchain} is also in effect, the topmost word of
11521 the save area is always used to store the backchain, and the return address
11522 register is always saved two words below the backchain.
11524 As long as the stack frame backchain is not used, code generated with
11525 @option{-mpacked-stack} is call-compatible with code generated with
11526 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
11527 S/390 or zSeries generated code that uses the stack frame backchain at run
11528 time, not just for debugging purposes. Such code is not call-compatible
11529 with code compiled with @option{-mpacked-stack}. Also, note that the
11530 combination of @option{-mbackchain},
11531 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11532 to build a linux kernel use @option{-msoft-float}.
11534 The default is to not use the packed stack layout.
11537 @itemx -mno-small-exec
11538 @opindex msmall-exec
11539 @opindex mno-small-exec
11540 Generate (or do not generate) code using the @code{bras} instruction
11541 to do subroutine calls.
11542 This only works reliably if the total executable size does not
11543 exceed 64k. The default is to use the @code{basr} instruction instead,
11544 which does not have this limitation.
11550 When @option{-m31} is specified, generate code compliant to the
11551 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
11552 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
11553 particular to generate 64-bit instructions. For the @samp{s390}
11554 targets, the default is @option{-m31}, while the @samp{s390x}
11555 targets default to @option{-m64}.
11561 When @option{-mzarch} is specified, generate code using the
11562 instructions available on z/Architecture.
11563 When @option{-mesa} is specified, generate code using the
11564 instructions available on ESA/390. Note that @option{-mesa} is
11565 not possible with @option{-m64}.
11566 When generating code compliant to the GNU/Linux for S/390 ABI,
11567 the default is @option{-mesa}. When generating code compliant
11568 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
11574 Generate (or do not generate) code using the @code{mvcle} instruction
11575 to perform block moves. When @option{-mno-mvcle} is specified,
11576 use a @code{mvc} loop instead. This is the default unless optimizing for
11583 Print (or do not print) additional debug information when compiling.
11584 The default is to not print debug information.
11586 @item -march=@var{cpu-type}
11588 Generate code that will run on @var{cpu-type}, which is the name of a system
11589 representing a certain processor type. Possible values for
11590 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
11591 When generating code using the instructions available on z/Architecture,
11592 the default is @option{-march=z900}. Otherwise, the default is
11593 @option{-march=g5}.
11595 @item -mtune=@var{cpu-type}
11597 Tune to @var{cpu-type} everything applicable about the generated code,
11598 except for the ABI and the set of available instructions.
11599 The list of @var{cpu-type} values is the same as for @option{-march}.
11600 The default is the value used for @option{-march}.
11603 @itemx -mno-tpf-trace
11604 @opindex mtpf-trace
11605 @opindex mno-tpf-trace
11606 Generate code that adds (does not add) in TPF OS specific branches to trace
11607 routines in the operating system. This option is off by default, even
11608 when compiling for the TPF OS@.
11611 @itemx -mno-fused-madd
11612 @opindex mfused-madd
11613 @opindex mno-fused-madd
11614 Generate code that uses (does not use) the floating point multiply and
11615 accumulate instructions. These instructions are generated by default if
11616 hardware floating point is used.
11618 @item -mwarn-framesize=@var{framesize}
11619 @opindex mwarn-framesize
11620 Emit a warning if the current function exceeds the given frame size. Because
11621 this is a compile time check it doesn't need to be a real problem when the program
11622 runs. It is intended to identify functions which most probably cause
11623 a stack overflow. It is useful to be used in an environment with limited stack
11624 size e.g.@: the linux kernel.
11626 @item -mwarn-dynamicstack
11627 @opindex mwarn-dynamicstack
11628 Emit a warning if the function calls alloca or uses dynamically
11629 sized arrays. This is generally a bad idea with a limited stack size.
11631 @item -mstack-guard=@var{stack-guard}
11632 @item -mstack-size=@var{stack-size}
11633 @opindex mstack-guard
11634 @opindex mstack-size
11635 These arguments always have to be used in conjunction. If they are present the s390
11636 back end emits additional instructions in the function prologue which trigger a trap
11637 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
11638 (remember that the stack on s390 grows downward). These options are intended to
11639 be used to help debugging stack overflow problems. The additionally emitted code
11640 cause only little overhead and hence can also be used in production like systems
11641 without greater performance degradation. The given values have to be exact
11642 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard}.
11643 In order to be efficient the extra code makes the assumption that the stack starts
11644 at an address aligned to the value given by @var{stack-size}.
11648 @subsection SH Options
11650 These @samp{-m} options are defined for the SH implementations:
11655 Generate code for the SH1.
11659 Generate code for the SH2.
11662 Generate code for the SH2e.
11666 Generate code for the SH3.
11670 Generate code for the SH3e.
11674 Generate code for the SH4 without a floating-point unit.
11676 @item -m4-single-only
11677 @opindex m4-single-only
11678 Generate code for the SH4 with a floating-point unit that only
11679 supports single-precision arithmetic.
11683 Generate code for the SH4 assuming the floating-point unit is in
11684 single-precision mode by default.
11688 Generate code for the SH4.
11692 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
11693 floating-point unit is not used.
11695 @item -m4a-single-only
11696 @opindex m4a-single-only
11697 Generate code for the SH4a, in such a way that no double-precision
11698 floating point operations are used.
11701 @opindex m4a-single
11702 Generate code for the SH4a assuming the floating-point unit is in
11703 single-precision mode by default.
11707 Generate code for the SH4a.
11711 Same as @option{-m4a-nofpu}, except that it implicitly passes
11712 @option{-dsp} to the assembler. GCC doesn't generate any DSP
11713 instructions at the moment.
11717 Compile code for the processor in big endian mode.
11721 Compile code for the processor in little endian mode.
11725 Align doubles at 64-bit boundaries. Note that this changes the calling
11726 conventions, and thus some functions from the standard C library will
11727 not work unless you recompile it first with @option{-mdalign}.
11731 Shorten some address references at link time, when possible; uses the
11732 linker option @option{-relax}.
11736 Use 32-bit offsets in @code{switch} tables. The default is to use
11741 Enable the use of the instruction @code{fmovd}.
11745 Comply with the calling conventions defined by Renesas.
11749 Comply with the calling conventions defined by Renesas.
11753 Comply with the calling conventions defined for GCC before the Renesas
11754 conventions were available. This option is the default for all
11755 targets of the SH toolchain except for @samp{sh-symbianelf}.
11758 @opindex mnomacsave
11759 Mark the @code{MAC} register as call-clobbered, even if
11760 @option{-mhitachi} is given.
11764 Increase IEEE-compliance of floating-point code.
11765 At the moment, this is equivalent to @option{-fno-finite-math-only}.
11766 When generating 16 bit SH opcodes, getting IEEE-conforming results for
11767 comparisons of NANs / infinities incurs extra overhead in every
11768 floating point comparison, therefore the default is set to
11769 @option{-ffinite-math-only}.
11773 Dump instruction size and location in the assembly code.
11776 @opindex mpadstruct
11777 This option is deprecated. It pads structures to multiple of 4 bytes,
11778 which is incompatible with the SH ABI@.
11782 Optimize for space instead of speed. Implied by @option{-Os}.
11785 @opindex mprefergot
11786 When generating position-independent code, emit function calls using
11787 the Global Offset Table instead of the Procedure Linkage Table.
11791 Generate a library function call to invalidate instruction cache
11792 entries, after fixing up a trampoline. This library function call
11793 doesn't assume it can write to the whole memory address space. This
11794 is the default when the target is @code{sh-*-linux*}.
11796 @item -multcost=@var{number}
11797 @opindex multcost=@var{number}
11798 Set the cost to assume for a multiply insn.
11800 @item -mdiv=@var{strategy}
11801 @opindex mdiv=@var{strategy}
11802 Set the division strategy to use for SHmedia code. @var{strategy} must be
11803 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
11804 inv:call2, inv:fp .
11805 "fp" performs the operation in floating point. This has a very high latency,
11806 but needs only a few instructions, so it might be a good choice if
11807 your code has enough easily exploitable ILP to allow the compiler to
11808 schedule the floating point instructions together with other instructions.
11809 Division by zero causes a floating point exception.
11810 "inv" uses integer operations to calculate the inverse of the divisor,
11811 and then multiplies the dividend with the inverse. This strategy allows
11812 cse and hoisting of the inverse calculation. Division by zero calculates
11813 an unspecified result, but does not trap.
11814 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
11815 have been found, or if the entire operation has been hoisted to the same
11816 place, the last stages of the inverse calculation are intertwined with the
11817 final multiply to reduce the overall latency, at the expense of using a few
11818 more instructions, and thus offering fewer scheduling opportunities with
11820 "call" calls a library function that usually implements the inv:minlat
11822 This gives high code density for m5-*media-nofpu compilations.
11823 "call2" uses a different entry point of the same library function, where it
11824 assumes that a pointer to a lookup table has already been set up, which
11825 exposes the pointer load to cse / code hoisting optimizations.
11826 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
11827 code generation, but if the code stays unoptimized, revert to the "call",
11828 "call2", or "fp" strategies, respectively. Note that the
11829 potentially-trapping side effect of division by zero is carried by a
11830 separate instruction, so it is possible that all the integer instructions
11831 are hoisted out, but the marker for the side effect stays where it is.
11832 A recombination to fp operations or a call is not possible in that case.
11833 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
11834 that the inverse calculation was nor separated from the multiply, they speed
11835 up division where the dividend fits into 20 bits (plus sign where applicable),
11836 by inserting a test to skip a number of operations in this case; this test
11837 slows down the case of larger dividends. inv20u assumes the case of a such
11838 a small dividend to be unlikely, and inv20l assumes it to be likely.
11840 @item -mdivsi3_libfunc=@var{name}
11841 @opindex mdivsi3_libfunc=@var{name}
11842 Set the name of the library function used for 32 bit signed division to
11843 @var{name}. This only affect the name used in the call and inv:call
11844 division strategies, and the compiler will still expect the same
11845 sets of input/output/clobbered registers as if this option was not present.
11847 @item -madjust-unroll
11848 @opindex madjust-unroll
11849 Throttle unrolling to avoid thrashing target registers.
11850 This option only has an effect if the gcc code base supports the
11851 TARGET_ADJUST_UNROLL_MAX target hook.
11853 @item -mindexed-addressing
11854 @opindex mindexed-addressing
11855 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
11856 This is only safe if the hardware and/or OS implement 32 bit wrap-around
11857 semantics for the indexed addressing mode. The architecture allows the
11858 implementation of processors with 64 bit MMU, which the OS could use to
11859 get 32 bit addressing, but since no current hardware implementation supports
11860 this or any other way to make the indexed addressing mode safe to use in
11861 the 32 bit ABI, the default is -mno-indexed-addressing.
11863 @item -mgettrcost=@var{number}
11864 @opindex mgettrcost=@var{number}
11865 Set the cost assumed for the gettr instruction to @var{number}.
11866 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
11870 Assume pt* instructions won't trap. This will generally generate better
11871 scheduled code, but is unsafe on current hardware. The current architecture
11872 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
11873 This has the unintentional effect of making it unsafe to schedule ptabs /
11874 ptrel before a branch, or hoist it out of a loop. For example,
11875 __do_global_ctors, a part of libgcc that runs constructors at program
11876 startup, calls functions in a list which is delimited by -1. With the
11877 -mpt-fixed option, the ptabs will be done before testing against -1.
11878 That means that all the constructors will be run a bit quicker, but when
11879 the loop comes to the end of the list, the program crashes because ptabs
11880 loads -1 into a target register. Since this option is unsafe for any
11881 hardware implementing the current architecture specification, the default
11882 is -mno-pt-fixed. Unless the user specifies a specific cost with
11883 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
11884 this deters register allocation using target registers for storing
11887 @item -minvalid-symbols
11888 @opindex minvalid-symbols
11889 Assume symbols might be invalid. Ordinary function symbols generated by
11890 the compiler will always be valid to load with movi/shori/ptabs or
11891 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
11892 to generate symbols that will cause ptabs / ptrel to trap.
11893 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
11894 It will then prevent cross-basic-block cse, hoisting and most scheduling
11895 of symbol loads. The default is @option{-mno-invalid-symbols}.
11898 @node SPARC Options
11899 @subsection SPARC Options
11900 @cindex SPARC options
11902 These @samp{-m} options are supported on the SPARC:
11905 @item -mno-app-regs
11907 @opindex mno-app-regs
11909 Specify @option{-mapp-regs} to generate output using the global registers
11910 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
11913 To be fully SVR4 ABI compliant at the cost of some performance loss,
11914 specify @option{-mno-app-regs}. You should compile libraries and system
11915 software with this option.
11918 @itemx -mhard-float
11920 @opindex mhard-float
11921 Generate output containing floating point instructions. This is the
11925 @itemx -msoft-float
11927 @opindex msoft-float
11928 Generate output containing library calls for floating point.
11929 @strong{Warning:} the requisite libraries are not available for all SPARC
11930 targets. Normally the facilities of the machine's usual C compiler are
11931 used, but this cannot be done directly in cross-compilation. You must make
11932 your own arrangements to provide suitable library functions for
11933 cross-compilation. The embedded targets @samp{sparc-*-aout} and
11934 @samp{sparclite-*-*} do provide software floating point support.
11936 @option{-msoft-float} changes the calling convention in the output file;
11937 therefore, it is only useful if you compile @emph{all} of a program with
11938 this option. In particular, you need to compile @file{libgcc.a}, the
11939 library that comes with GCC, with @option{-msoft-float} in order for
11942 @item -mhard-quad-float
11943 @opindex mhard-quad-float
11944 Generate output containing quad-word (long double) floating point
11947 @item -msoft-quad-float
11948 @opindex msoft-quad-float
11949 Generate output containing library calls for quad-word (long double)
11950 floating point instructions. The functions called are those specified
11951 in the SPARC ABI@. This is the default.
11953 As of this writing, there are no SPARC implementations that have hardware
11954 support for the quad-word floating point instructions. They all invoke
11955 a trap handler for one of these instructions, and then the trap handler
11956 emulates the effect of the instruction. Because of the trap handler overhead,
11957 this is much slower than calling the ABI library routines. Thus the
11958 @option{-msoft-quad-float} option is the default.
11960 @item -mno-unaligned-doubles
11961 @itemx -munaligned-doubles
11962 @opindex mno-unaligned-doubles
11963 @opindex munaligned-doubles
11964 Assume that doubles have 8 byte alignment. This is the default.
11966 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
11967 alignment only if they are contained in another type, or if they have an
11968 absolute address. Otherwise, it assumes they have 4 byte alignment.
11969 Specifying this option avoids some rare compatibility problems with code
11970 generated by other compilers. It is not the default because it results
11971 in a performance loss, especially for floating point code.
11973 @item -mno-faster-structs
11974 @itemx -mfaster-structs
11975 @opindex mno-faster-structs
11976 @opindex mfaster-structs
11977 With @option{-mfaster-structs}, the compiler assumes that structures
11978 should have 8 byte alignment. This enables the use of pairs of
11979 @code{ldd} and @code{std} instructions for copies in structure
11980 assignment, in place of twice as many @code{ld} and @code{st} pairs.
11981 However, the use of this changed alignment directly violates the SPARC
11982 ABI@. Thus, it's intended only for use on targets where the developer
11983 acknowledges that their resulting code will not be directly in line with
11984 the rules of the ABI@.
11986 @item -mimpure-text
11987 @opindex mimpure-text
11988 @option{-mimpure-text}, used in addition to @option{-shared}, tells
11989 the compiler to not pass @option{-z text} to the linker when linking a
11990 shared object. Using this option, you can link position-dependent
11991 code into a shared object.
11993 @option{-mimpure-text} suppresses the ``relocations remain against
11994 allocatable but non-writable sections'' linker error message.
11995 However, the necessary relocations will trigger copy-on-write, and the
11996 shared object is not actually shared across processes. Instead of
11997 using @option{-mimpure-text}, you should compile all source code with
11998 @option{-fpic} or @option{-fPIC}.
12000 This option is only available on SunOS and Solaris.
12002 @item -mcpu=@var{cpu_type}
12004 Set the instruction set, register set, and instruction scheduling parameters
12005 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12006 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12007 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12008 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
12009 @samp{ultrasparc3}.
12011 Default instruction scheduling parameters are used for values that select
12012 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12013 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12015 Here is a list of each supported architecture and their supported
12020 v8: supersparc, hypersparc
12021 sparclite: f930, f934, sparclite86x
12023 v9: ultrasparc, ultrasparc3
12026 By default (unless configured otherwise), GCC generates code for the V7
12027 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12028 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12029 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12030 SPARCStation 1, 2, IPX etc.
12032 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12033 architecture. The only difference from V7 code is that the compiler emits
12034 the integer multiply and integer divide instructions which exist in SPARC-V8
12035 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12036 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12039 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12040 the SPARC architecture. This adds the integer multiply, integer divide step
12041 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12042 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12043 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12044 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12045 MB86934 chip, which is the more recent SPARClite with FPU@.
12047 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12048 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12049 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12050 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12051 optimizes it for the TEMIC SPARClet chip.
12053 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12054 architecture. This adds 64-bit integer and floating-point move instructions,
12055 3 additional floating-point condition code registers and conditional move
12056 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12057 optimizes it for the Sun UltraSPARC I/II chips. With
12058 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12059 Sun UltraSPARC III chip.
12061 @item -mtune=@var{cpu_type}
12063 Set the instruction scheduling parameters for machine type
12064 @var{cpu_type}, but do not set the instruction set or register set that the
12065 option @option{-mcpu=@var{cpu_type}} would.
12067 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12068 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12069 that select a particular cpu implementation. Those are @samp{cypress},
12070 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12071 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
12072 @samp{ultrasparc3}.
12077 @opindex mno-v8plus
12078 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12079 difference from the V8 ABI is that the global and out registers are
12080 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12081 mode for all SPARC-V9 processors.
12087 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12088 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12091 These @samp{-m} options are supported in addition to the above
12092 on SPARC-V9 processors in 64-bit environments:
12095 @item -mlittle-endian
12096 @opindex mlittle-endian
12097 Generate code for a processor running in little-endian mode. It is only
12098 available for a few configurations and most notably not on Solaris and Linux.
12104 Generate code for a 32-bit or 64-bit environment.
12105 The 32-bit environment sets int, long and pointer to 32 bits.
12106 The 64-bit environment sets int to 32 bits and long and pointer
12109 @item -mcmodel=medlow
12110 @opindex mcmodel=medlow
12111 Generate code for the Medium/Low code model: 64-bit addresses, programs
12112 must be linked in the low 32 bits of memory. Programs can be statically
12113 or dynamically linked.
12115 @item -mcmodel=medmid
12116 @opindex mcmodel=medmid
12117 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12118 must be linked in the low 44 bits of memory, the text and data segments must
12119 be less than 2GB in size and the data segment must be located within 2GB of
12122 @item -mcmodel=medany
12123 @opindex mcmodel=medany
12124 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12125 may be linked anywhere in memory, the text and data segments must be less
12126 than 2GB in size and the data segment must be located within 2GB of the
12129 @item -mcmodel=embmedany
12130 @opindex mcmodel=embmedany
12131 Generate code for the Medium/Anywhere code model for embedded systems:
12132 64-bit addresses, the text and data segments must be less than 2GB in
12133 size, both starting anywhere in memory (determined at link time). The
12134 global register %g4 points to the base of the data segment. Programs
12135 are statically linked and PIC is not supported.
12138 @itemx -mno-stack-bias
12139 @opindex mstack-bias
12140 @opindex mno-stack-bias
12141 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12142 frame pointer if present, are offset by @minus{}2047 which must be added back
12143 when making stack frame references. This is the default in 64-bit mode.
12144 Otherwise, assume no such offset is present.
12147 These switches are supported in addition to the above on Solaris:
12152 Add support for multithreading using the Solaris threads library. This
12153 option sets flags for both the preprocessor and linker. This option does
12154 not affect the thread safety of object code produced by the compiler or
12155 that of libraries supplied with it.
12159 Add support for multithreading using the POSIX threads library. This
12160 option sets flags for both the preprocessor and linker. This option does
12161 not affect the thread safety of object code produced by the compiler or
12162 that of libraries supplied with it.
12165 @node System V Options
12166 @subsection Options for System V
12168 These additional options are available on System V Release 4 for
12169 compatibility with other compilers on those systems:
12174 Create a shared object.
12175 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12179 Identify the versions of each tool used by the compiler, in a
12180 @code{.ident} assembler directive in the output.
12184 Refrain from adding @code{.ident} directives to the output file (this is
12187 @item -YP,@var{dirs}
12189 Search the directories @var{dirs}, and no others, for libraries
12190 specified with @option{-l}.
12192 @item -Ym,@var{dir}
12194 Look in the directory @var{dir} to find the M4 preprocessor.
12195 The assembler uses this option.
12196 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12197 @c the generic assembler that comes with Solaris takes just -Ym.
12200 @node TMS320C3x/C4x Options
12201 @subsection TMS320C3x/C4x Options
12202 @cindex TMS320C3x/C4x Options
12204 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12208 @item -mcpu=@var{cpu_type}
12210 Set the instruction set, register set, and instruction scheduling
12211 parameters for machine type @var{cpu_type}. Supported values for
12212 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12213 @samp{c44}. The default is @samp{c40} to generate code for the
12218 @itemx -msmall-memory
12220 @opindex mbig-memory
12222 @opindex msmall-memory
12224 Generates code for the big or small memory model. The small memory
12225 model assumed that all data fits into one 64K word page. At run-time
12226 the data page (DP) register must be set to point to the 64K page
12227 containing the .bss and .data program sections. The big memory model is
12228 the default and requires reloading of the DP register for every direct
12235 Allow (disallow) allocation of general integer operands into the block
12236 count register BK@.
12242 Enable (disable) generation of code using decrement and branch,
12243 DBcond(D), instructions. This is enabled by default for the C4x. To be
12244 on the safe side, this is disabled for the C3x, since the maximum
12245 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12246 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12247 that it can utilize the decrement and branch instruction, but will give
12248 up if there is more than one memory reference in the loop. Thus a loop
12249 where the loop counter is decremented can generate slightly more
12250 efficient code, in cases where the RPTB instruction cannot be utilized.
12252 @item -mdp-isr-reload
12254 @opindex mdp-isr-reload
12256 Force the DP register to be saved on entry to an interrupt service
12257 routine (ISR), reloaded to point to the data section, and restored on
12258 exit from the ISR@. This should not be required unless someone has
12259 violated the small memory model by modifying the DP register, say within
12266 For the C3x use the 24-bit MPYI instruction for integer multiplies
12267 instead of a library call to guarantee 32-bit results. Note that if one
12268 of the operands is a constant, then the multiplication will be performed
12269 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12270 then squaring operations are performed inline instead of a library call.
12273 @itemx -mno-fast-fix
12275 @opindex mno-fast-fix
12276 The C3x/C4x FIX instruction to convert a floating point value to an
12277 integer value chooses the nearest integer less than or equal to the
12278 floating point value rather than to the nearest integer. Thus if the
12279 floating point number is negative, the result will be incorrectly
12280 truncated an additional code is necessary to detect and correct this
12281 case. This option can be used to disable generation of the additional
12282 code required to correct the result.
12288 Enable (disable) generation of repeat block sequences using the RPTB
12289 instruction for zero overhead looping. The RPTB construct is only used
12290 for innermost loops that do not call functions or jump across the loop
12291 boundaries. There is no advantage having nested RPTB loops due to the
12292 overhead required to save and restore the RC, RS, and RE registers.
12293 This is enabled by default with @option{-O2}.
12295 @item -mrpts=@var{count}
12299 Enable (disable) the use of the single instruction repeat instruction
12300 RPTS@. If a repeat block contains a single instruction, and the loop
12301 count can be guaranteed to be less than the value @var{count}, GCC will
12302 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12303 then a RPTS will be emitted even if the loop count cannot be determined
12304 at compile time. Note that the repeated instruction following RPTS does
12305 not have to be reloaded from memory each iteration, thus freeing up the
12306 CPU buses for operands. However, since interrupts are blocked by this
12307 instruction, it is disabled by default.
12309 @item -mloop-unsigned
12310 @itemx -mno-loop-unsigned
12311 @opindex mloop-unsigned
12312 @opindex mno-loop-unsigned
12313 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12314 is @math{2^{31} + 1} since these instructions test if the iteration count is
12315 negative to terminate the loop. If the iteration count is unsigned
12316 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12317 exceeded. This switch allows an unsigned iteration count.
12321 Try to emit an assembler syntax that the TI assembler (asm30) is happy
12322 with. This also enforces compatibility with the API employed by the TI
12323 C3x C compiler. For example, long doubles are passed as structures
12324 rather than in floating point registers.
12330 Generate code that uses registers (stack) for passing arguments to functions.
12331 By default, arguments are passed in registers where possible rather
12332 than by pushing arguments on to the stack.
12334 @item -mparallel-insns
12335 @itemx -mno-parallel-insns
12336 @opindex mparallel-insns
12337 @opindex mno-parallel-insns
12338 Allow the generation of parallel instructions. This is enabled by
12339 default with @option{-O2}.
12341 @item -mparallel-mpy
12342 @itemx -mno-parallel-mpy
12343 @opindex mparallel-mpy
12344 @opindex mno-parallel-mpy
12345 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
12346 provided @option{-mparallel-insns} is also specified. These instructions have
12347 tight register constraints which can pessimize the code generation
12348 of large functions.
12353 @subsection V850 Options
12354 @cindex V850 Options
12356 These @samp{-m} options are defined for V850 implementations:
12360 @itemx -mno-long-calls
12361 @opindex mlong-calls
12362 @opindex mno-long-calls
12363 Treat all calls as being far away (near). If calls are assumed to be
12364 far away, the compiler will always load the functions address up into a
12365 register, and call indirect through the pointer.
12371 Do not optimize (do optimize) basic blocks that use the same index
12372 pointer 4 or more times to copy pointer into the @code{ep} register, and
12373 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
12374 option is on by default if you optimize.
12376 @item -mno-prolog-function
12377 @itemx -mprolog-function
12378 @opindex mno-prolog-function
12379 @opindex mprolog-function
12380 Do not use (do use) external functions to save and restore registers
12381 at the prologue and epilogue of a function. The external functions
12382 are slower, but use less code space if more than one function saves
12383 the same number of registers. The @option{-mprolog-function} option
12384 is on by default if you optimize.
12388 Try to make the code as small as possible. At present, this just turns
12389 on the @option{-mep} and @option{-mprolog-function} options.
12391 @item -mtda=@var{n}
12393 Put static or global variables whose size is @var{n} bytes or less into
12394 the tiny data area that register @code{ep} points to. The tiny data
12395 area can hold up to 256 bytes in total (128 bytes for byte references).
12397 @item -msda=@var{n}
12399 Put static or global variables whose size is @var{n} bytes or less into
12400 the small data area that register @code{gp} points to. The small data
12401 area can hold up to 64 kilobytes.
12403 @item -mzda=@var{n}
12405 Put static or global variables whose size is @var{n} bytes or less into
12406 the first 32 kilobytes of memory.
12410 Specify that the target processor is the V850.
12413 @opindex mbig-switch
12414 Generate code suitable for big switch tables. Use this option only if
12415 the assembler/linker complain about out of range branches within a switch
12420 This option will cause r2 and r5 to be used in the code generated by
12421 the compiler. This setting is the default.
12423 @item -mno-app-regs
12424 @opindex mno-app-regs
12425 This option will cause r2 and r5 to be treated as fixed registers.
12429 Specify that the target processor is the V850E1. The preprocessor
12430 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
12431 this option is used.
12435 Specify that the target processor is the V850E@. The preprocessor
12436 constant @samp{__v850e__} will be defined if this option is used.
12438 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
12439 are defined then a default target processor will be chosen and the
12440 relevant @samp{__v850*__} preprocessor constant will be defined.
12442 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
12443 defined, regardless of which processor variant is the target.
12445 @item -mdisable-callt
12446 @opindex mdisable-callt
12447 This option will suppress generation of the CALLT instruction for the
12448 v850e and v850e1 flavors of the v850 architecture. The default is
12449 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
12454 @subsection VAX Options
12455 @cindex VAX options
12457 These @samp{-m} options are defined for the VAX:
12462 Do not output certain jump instructions (@code{aobleq} and so on)
12463 that the Unix assembler for the VAX cannot handle across long
12468 Do output those jump instructions, on the assumption that you
12469 will assemble with the GNU assembler.
12473 Output code for g-format floating point numbers instead of d-format.
12476 @node x86-64 Options
12477 @subsection x86-64 Options
12478 @cindex x86-64 options
12480 These are listed under @xref{i386 and x86-64 Options}.
12482 @node Xstormy16 Options
12483 @subsection Xstormy16 Options
12484 @cindex Xstormy16 Options
12486 These options are defined for Xstormy16:
12491 Choose startup files and linker script suitable for the simulator.
12494 @node Xtensa Options
12495 @subsection Xtensa Options
12496 @cindex Xtensa Options
12498 These options are supported for Xtensa targets:
12502 @itemx -mno-const16
12504 @opindex mno-const16
12505 Enable or disable use of @code{CONST16} instructions for loading
12506 constant values. The @code{CONST16} instruction is currently not a
12507 standard option from Tensilica. When enabled, @code{CONST16}
12508 instructions are always used in place of the standard @code{L32R}
12509 instructions. The use of @code{CONST16} is enabled by default only if
12510 the @code{L32R} instruction is not available.
12513 @itemx -mno-fused-madd
12514 @opindex mfused-madd
12515 @opindex mno-fused-madd
12516 Enable or disable use of fused multiply/add and multiply/subtract
12517 instructions in the floating-point option. This has no effect if the
12518 floating-point option is not also enabled. Disabling fused multiply/add
12519 and multiply/subtract instructions forces the compiler to use separate
12520 instructions for the multiply and add/subtract operations. This may be
12521 desirable in some cases where strict IEEE 754-compliant results are
12522 required: the fused multiply add/subtract instructions do not round the
12523 intermediate result, thereby producing results with @emph{more} bits of
12524 precision than specified by the IEEE standard. Disabling fused multiply
12525 add/subtract instructions also ensures that the program output is not
12526 sensitive to the compiler's ability to combine multiply and add/subtract
12529 @item -mtext-section-literals
12530 @itemx -mno-text-section-literals
12531 @opindex mtext-section-literals
12532 @opindex mno-text-section-literals
12533 Control the treatment of literal pools. The default is
12534 @option{-mno-text-section-literals}, which places literals in a separate
12535 section in the output file. This allows the literal pool to be placed
12536 in a data RAM/ROM, and it also allows the linker to combine literal
12537 pools from separate object files to remove redundant literals and
12538 improve code size. With @option{-mtext-section-literals}, the literals
12539 are interspersed in the text section in order to keep them as close as
12540 possible to their references. This may be necessary for large assembly
12543 @item -mtarget-align
12544 @itemx -mno-target-align
12545 @opindex mtarget-align
12546 @opindex mno-target-align
12547 When this option is enabled, GCC instructs the assembler to
12548 automatically align instructions to reduce branch penalties at the
12549 expense of some code density. The assembler attempts to widen density
12550 instructions to align branch targets and the instructions following call
12551 instructions. If there are not enough preceding safe density
12552 instructions to align a target, no widening will be performed. The
12553 default is @option{-mtarget-align}. These options do not affect the
12554 treatment of auto-aligned instructions like @code{LOOP}, which the
12555 assembler will always align, either by widening density instructions or
12556 by inserting no-op instructions.
12559 @itemx -mno-longcalls
12560 @opindex mlongcalls
12561 @opindex mno-longcalls
12562 When this option is enabled, GCC instructs the assembler to translate
12563 direct calls to indirect calls unless it can determine that the target
12564 of a direct call is in the range allowed by the call instruction. This
12565 translation typically occurs for calls to functions in other source
12566 files. Specifically, the assembler translates a direct @code{CALL}
12567 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
12568 The default is @option{-mno-longcalls}. This option should be used in
12569 programs where the call target can potentially be out of range. This
12570 option is implemented in the assembler, not the compiler, so the
12571 assembly code generated by GCC will still show direct call
12572 instructions---look at the disassembled object code to see the actual
12573 instructions. Note that the assembler will use an indirect call for
12574 every cross-file call, not just those that really will be out of range.
12577 @node zSeries Options
12578 @subsection zSeries Options
12579 @cindex zSeries options
12581 These are listed under @xref{S/390 and zSeries Options}.
12583 @node Code Gen Options
12584 @section Options for Code Generation Conventions
12585 @cindex code generation conventions
12586 @cindex options, code generation
12587 @cindex run-time options
12589 These machine-independent options control the interface conventions
12590 used in code generation.
12592 Most of them have both positive and negative forms; the negative form
12593 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
12594 one of the forms is listed---the one which is not the default. You
12595 can figure out the other form by either removing @samp{no-} or adding
12599 @item -fbounds-check
12600 @opindex fbounds-check
12601 For front-ends that support it, generate additional code to check that
12602 indices used to access arrays are within the declared range. This is
12603 currently only supported by the Java and Fortran 77 front-ends, where
12604 this option defaults to true and false respectively.
12608 This option generates traps for signed overflow on addition, subtraction,
12609 multiplication operations.
12613 This option instructs the compiler to assume that signed arithmetic
12614 overflow of addition, subtraction and multiplication wraps around
12615 using twos-complement representation. This flag enables some optimizations
12616 and disables other. This option is enabled by default for the Java
12617 front-end, as required by the Java language specification.
12620 @opindex fexceptions
12621 Enable exception handling. Generates extra code needed to propagate
12622 exceptions. For some targets, this implies GCC will generate frame
12623 unwind information for all functions, which can produce significant data
12624 size overhead, although it does not affect execution. If you do not
12625 specify this option, GCC will enable it by default for languages like
12626 C++ which normally require exception handling, and disable it for
12627 languages like C that do not normally require it. However, you may need
12628 to enable this option when compiling C code that needs to interoperate
12629 properly with exception handlers written in C++. You may also wish to
12630 disable this option if you are compiling older C++ programs that don't
12631 use exception handling.
12633 @item -fnon-call-exceptions
12634 @opindex fnon-call-exceptions
12635 Generate code that allows trapping instructions to throw exceptions.
12636 Note that this requires platform-specific runtime support that does
12637 not exist everywhere. Moreover, it only allows @emph{trapping}
12638 instructions to throw exceptions, i.e.@: memory references or floating
12639 point instructions. It does not allow exceptions to be thrown from
12640 arbitrary signal handlers such as @code{SIGALRM}.
12642 @item -funwind-tables
12643 @opindex funwind-tables
12644 Similar to @option{-fexceptions}, except that it will just generate any needed
12645 static data, but will not affect the generated code in any other way.
12646 You will normally not enable this option; instead, a language processor
12647 that needs this handling would enable it on your behalf.
12649 @item -fasynchronous-unwind-tables
12650 @opindex fasynchronous-unwind-tables
12651 Generate unwind table in dwarf2 format, if supported by target machine. The
12652 table is exact at each instruction boundary, so it can be used for stack
12653 unwinding from asynchronous events (such as debugger or garbage collector).
12655 @item -fpcc-struct-return
12656 @opindex fpcc-struct-return
12657 Return ``short'' @code{struct} and @code{union} values in memory like
12658 longer ones, rather than in registers. This convention is less
12659 efficient, but it has the advantage of allowing intercallability between
12660 GCC-compiled files and files compiled with other compilers, particularly
12661 the Portable C Compiler (pcc).
12663 The precise convention for returning structures in memory depends
12664 on the target configuration macros.
12666 Short structures and unions are those whose size and alignment match
12667 that of some integer type.
12669 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
12670 switch is not binary compatible with code compiled with the
12671 @option{-freg-struct-return} switch.
12672 Use it to conform to a non-default application binary interface.
12674 @item -freg-struct-return
12675 @opindex freg-struct-return
12676 Return @code{struct} and @code{union} values in registers when possible.
12677 This is more efficient for small structures than
12678 @option{-fpcc-struct-return}.
12680 If you specify neither @option{-fpcc-struct-return} nor
12681 @option{-freg-struct-return}, GCC defaults to whichever convention is
12682 standard for the target. If there is no standard convention, GCC
12683 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
12684 the principal compiler. In those cases, we can choose the standard, and
12685 we chose the more efficient register return alternative.
12687 @strong{Warning:} code compiled with the @option{-freg-struct-return}
12688 switch is not binary compatible with code compiled with the
12689 @option{-fpcc-struct-return} switch.
12690 Use it to conform to a non-default application binary interface.
12692 @item -fshort-enums
12693 @opindex fshort-enums
12694 Allocate to an @code{enum} type only as many bytes as it needs for the
12695 declared range of possible values. Specifically, the @code{enum} type
12696 will be equivalent to the smallest integer type which has enough room.
12698 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
12699 code that is not binary compatible with code generated without that switch.
12700 Use it to conform to a non-default application binary interface.
12702 @item -fshort-double
12703 @opindex fshort-double
12704 Use the same size for @code{double} as for @code{float}.
12706 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
12707 code that is not binary compatible with code generated without that switch.
12708 Use it to conform to a non-default application binary interface.
12710 @item -fshort-wchar
12711 @opindex fshort-wchar
12712 Override the underlying type for @samp{wchar_t} to be @samp{short
12713 unsigned int} instead of the default for the target. This option is
12714 useful for building programs to run under WINE@.
12716 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
12717 code that is not binary compatible with code generated without that switch.
12718 Use it to conform to a non-default application binary interface.
12720 @item -fshared-data
12721 @opindex fshared-data
12722 Requests that the data and non-@code{const} variables of this
12723 compilation be shared data rather than private data. The distinction
12724 makes sense only on certain operating systems, where shared data is
12725 shared between processes running the same program, while private data
12726 exists in one copy per process.
12729 @opindex fno-common
12730 In C, allocate even uninitialized global variables in the data section of the
12731 object file, rather than generating them as common blocks. This has the
12732 effect that if the same variable is declared (without @code{extern}) in
12733 two different compilations, you will get an error when you link them.
12734 The only reason this might be useful is if you wish to verify that the
12735 program will work on other systems which always work this way.
12739 Ignore the @samp{#ident} directive.
12741 @item -finhibit-size-directive
12742 @opindex finhibit-size-directive
12743 Don't output a @code{.size} assembler directive, or anything else that
12744 would cause trouble if the function is split in the middle, and the
12745 two halves are placed at locations far apart in memory. This option is
12746 used when compiling @file{crtstuff.c}; you should not need to use it
12749 @item -fverbose-asm
12750 @opindex fverbose-asm
12751 Put extra commentary information in the generated assembly code to
12752 make it more readable. This option is generally only of use to those
12753 who actually need to read the generated assembly code (perhaps while
12754 debugging the compiler itself).
12756 @option{-fno-verbose-asm}, the default, causes the
12757 extra information to be omitted and is useful when comparing two assembler
12762 @cindex global offset table
12764 Generate position-independent code (PIC) suitable for use in a shared
12765 library, if supported for the target machine. Such code accesses all
12766 constant addresses through a global offset table (GOT)@. The dynamic
12767 loader resolves the GOT entries when the program starts (the dynamic
12768 loader is not part of GCC; it is part of the operating system). If
12769 the GOT size for the linked executable exceeds a machine-specific
12770 maximum size, you get an error message from the linker indicating that
12771 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
12772 instead. (These maximums are 8k on the SPARC and 32k
12773 on the m68k and RS/6000. The 386 has no such limit.)
12775 Position-independent code requires special support, and therefore works
12776 only on certain machines. For the 386, GCC supports PIC for System V
12777 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
12778 position-independent.
12782 If supported for the target machine, emit position-independent code,
12783 suitable for dynamic linking and avoiding any limit on the size of the
12784 global offset table. This option makes a difference on the m68k,
12785 PowerPC and SPARC@.
12787 Position-independent code requires special support, and therefore works
12788 only on certain machines.
12794 These options are similar to @option{-fpic} and @option{-fPIC}, but
12795 generated position independent code can be only linked into executables.
12796 Usually these options are used when @option{-pie} GCC option will be
12797 used during linking.
12799 @item -fno-jump-tables
12800 @opindex fno-jump-tables
12801 Do not use jump tables for switch statements even where it would be
12802 more efficient than other code generation strategies. This option is
12803 of use in conjunction with @option{-fpic} or @option{-fPIC} for
12804 building code which forms part of a dynamic linker and cannot
12805 reference the address of a jump table. On some targets, jump tables
12806 do not require a GOT and this option is not needed.
12808 @item -ffixed-@var{reg}
12810 Treat the register named @var{reg} as a fixed register; generated code
12811 should never refer to it (except perhaps as a stack pointer, frame
12812 pointer or in some other fixed role).
12814 @var{reg} must be the name of a register. The register names accepted
12815 are machine-specific and are defined in the @code{REGISTER_NAMES}
12816 macro in the machine description macro file.
12818 This flag does not have a negative form, because it specifies a
12821 @item -fcall-used-@var{reg}
12822 @opindex fcall-used
12823 Treat the register named @var{reg} as an allocable register that is
12824 clobbered by function calls. It may be allocated for temporaries or
12825 variables that do not live across a call. Functions compiled this way
12826 will not save and restore the register @var{reg}.
12828 It is an error to used this flag with the frame pointer or stack pointer.
12829 Use of this flag for other registers that have fixed pervasive roles in
12830 the machine's execution model will produce disastrous results.
12832 This flag does not have a negative form, because it specifies a
12835 @item -fcall-saved-@var{reg}
12836 @opindex fcall-saved
12837 Treat the register named @var{reg} as an allocable register saved by
12838 functions. It may be allocated even for temporaries or variables that
12839 live across a call. Functions compiled this way will save and restore
12840 the register @var{reg} if they use it.
12842 It is an error to used this flag with the frame pointer or stack pointer.
12843 Use of this flag for other registers that have fixed pervasive roles in
12844 the machine's execution model will produce disastrous results.
12846 A different sort of disaster will result from the use of this flag for
12847 a register in which function values may be returned.
12849 This flag does not have a negative form, because it specifies a
12852 @item -fpack-struct[=@var{n}]
12853 @opindex fpack-struct
12854 Without a value specified, pack all structure members together without
12855 holes. When a value is specified (which must be a small power of two), pack
12856 structure members according to this value, representing the maximum
12857 alignment (that is, objects with default alignment requirements larger than
12858 this will be output potentially unaligned at the next fitting location.
12860 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
12861 code that is not binary compatible with code generated without that switch.
12862 Additionally, it makes the code suboptimal.
12863 Use it to conform to a non-default application binary interface.
12865 @item -finstrument-functions
12866 @opindex finstrument-functions
12867 Generate instrumentation calls for entry and exit to functions. Just
12868 after function entry and just before function exit, the following
12869 profiling functions will be called with the address of the current
12870 function and its call site. (On some platforms,
12871 @code{__builtin_return_address} does not work beyond the current
12872 function, so the call site information may not be available to the
12873 profiling functions otherwise.)
12876 void __cyg_profile_func_enter (void *this_fn,
12878 void __cyg_profile_func_exit (void *this_fn,
12882 The first argument is the address of the start of the current function,
12883 which may be looked up exactly in the symbol table.
12885 This instrumentation is also done for functions expanded inline in other
12886 functions. The profiling calls will indicate where, conceptually, the
12887 inline function is entered and exited. This means that addressable
12888 versions of such functions must be available. If all your uses of a
12889 function are expanded inline, this may mean an additional expansion of
12890 code size. If you use @samp{extern inline} in your C code, an
12891 addressable version of such functions must be provided. (This is
12892 normally the case anyways, but if you get lucky and the optimizer always
12893 expands the functions inline, you might have gotten away without
12894 providing static copies.)
12896 A function may be given the attribute @code{no_instrument_function}, in
12897 which case this instrumentation will not be done. This can be used, for
12898 example, for the profiling functions listed above, high-priority
12899 interrupt routines, and any functions from which the profiling functions
12900 cannot safely be called (perhaps signal handlers, if the profiling
12901 routines generate output or allocate memory).
12903 @item -fstack-check
12904 @opindex fstack-check
12905 Generate code to verify that you do not go beyond the boundary of the
12906 stack. You should specify this flag if you are running in an
12907 environment with multiple threads, but only rarely need to specify it in
12908 a single-threaded environment since stack overflow is automatically
12909 detected on nearly all systems if there is only one stack.
12911 Note that this switch does not actually cause checking to be done; the
12912 operating system must do that. The switch causes generation of code
12913 to ensure that the operating system sees the stack being extended.
12915 @item -fstack-limit-register=@var{reg}
12916 @itemx -fstack-limit-symbol=@var{sym}
12917 @itemx -fno-stack-limit
12918 @opindex fstack-limit-register
12919 @opindex fstack-limit-symbol
12920 @opindex fno-stack-limit
12921 Generate code to ensure that the stack does not grow beyond a certain value,
12922 either the value of a register or the address of a symbol. If the stack
12923 would grow beyond the value, a signal is raised. For most targets,
12924 the signal is raised before the stack overruns the boundary, so
12925 it is possible to catch the signal without taking special precautions.
12927 For instance, if the stack starts at absolute address @samp{0x80000000}
12928 and grows downwards, you can use the flags
12929 @option{-fstack-limit-symbol=__stack_limit} and
12930 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
12931 of 128KB@. Note that this may only work with the GNU linker.
12933 @cindex aliasing of parameters
12934 @cindex parameters, aliased
12935 @item -fargument-alias
12936 @itemx -fargument-noalias
12937 @itemx -fargument-noalias-global
12938 @opindex fargument-alias
12939 @opindex fargument-noalias
12940 @opindex fargument-noalias-global
12941 Specify the possible relationships among parameters and between
12942 parameters and global data.
12944 @option{-fargument-alias} specifies that arguments (parameters) may
12945 alias each other and may alias global storage.@*
12946 @option{-fargument-noalias} specifies that arguments do not alias
12947 each other, but may alias global storage.@*
12948 @option{-fargument-noalias-global} specifies that arguments do not
12949 alias each other and do not alias global storage.
12951 Each language will automatically use whatever option is required by
12952 the language standard. You should not need to use these options yourself.
12954 @item -fleading-underscore
12955 @opindex fleading-underscore
12956 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
12957 change the way C symbols are represented in the object file. One use
12958 is to help link with legacy assembly code.
12960 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
12961 generate code that is not binary compatible with code generated without that
12962 switch. Use it to conform to a non-default application binary interface.
12963 Not all targets provide complete support for this switch.
12965 @item -ftls-model=@var{model}
12966 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
12967 The @var{model} argument should be one of @code{global-dynamic},
12968 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
12970 The default without @option{-fpic} is @code{initial-exec}; with
12971 @option{-fpic} the default is @code{global-dynamic}.
12973 @item -fvisibility=@var{default|internal|hidden|protected}
12974 @opindex fvisibility
12975 Set the default ELF image symbol visibility to the specified option---all
12976 symbols will be marked with this unless overridden within the code.
12977 Using this feature can very substantially improve linking and
12978 load times of shared object libraries, produce more optimized
12979 code, provide near-perfect API export and prevent symbol clashes.
12980 It is @strong{strongly} recommended that you use this in any shared objects
12983 Despite the nomenclature, @code{default} always means public ie;
12984 available to be linked against from outside the shared object.
12985 @code{protected} and @code{internal} are pretty useless in real-world
12986 usage so the only other commonly used option will be @code{hidden}.
12987 The default if @option{-fvisibility} isn't specified is
12988 @code{default}, i.e., make every
12989 symbol public---this causes the same behavior as previous versions of
12992 A good explanation of the benefits offered by ensuring ELF
12993 symbols have the correct visibility is given by ``How To Write
12994 Shared Libraries'' by Ulrich Drepper (which can be found at
12995 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
12996 solution made possible by this option to marking things hidden when
12997 the default is public is to make the default hidden and mark things
12998 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
12999 and @code{__attribute__ ((visibility("default")))} instead of
13000 @code{__declspec(dllexport)} you get almost identical semantics with
13001 identical syntax. This is a great boon to those working with
13002 cross-platform projects.
13004 For those adding visibility support to existing code, you may find
13005 @samp{#pragma GCC visibility} of use. This works by you enclosing
13006 the declarations you wish to set visibility for with (for example)
13007 @samp{#pragma GCC visibility push(hidden)} and
13008 @samp{#pragma GCC visibility pop}.
13009 Bear in mind that symbol visibility should be viewed @strong{as
13010 part of the API interface contract} and thus all new code should
13011 always specify visibility when it is not the default ie; declarations
13012 only for use within the local DSO should @strong{always} be marked explicitly
13013 as hidden as so to avoid PLT indirection overheads---making this
13014 abundantly clear also aids readability and self-documentation of the code.
13015 Note that due to ISO C++ specification requirements, operator new and
13016 operator delete must always be of default visibility.
13018 An overview of these techniques, their benefits and how to use them
13019 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13025 @node Environment Variables
13026 @section Environment Variables Affecting GCC
13027 @cindex environment variables
13029 @c man begin ENVIRONMENT
13030 This section describes several environment variables that affect how GCC
13031 operates. Some of them work by specifying directories or prefixes to use
13032 when searching for various kinds of files. Some are used to specify other
13033 aspects of the compilation environment.
13035 Note that you can also specify places to search using options such as
13036 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13037 take precedence over places specified using environment variables, which
13038 in turn take precedence over those specified by the configuration of GCC@.
13039 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13040 GNU Compiler Collection (GCC) Internals}.
13045 @c @itemx LC_COLLATE
13047 @c @itemx LC_MONETARY
13048 @c @itemx LC_NUMERIC
13053 @c @findex LC_COLLATE
13054 @findex LC_MESSAGES
13055 @c @findex LC_MONETARY
13056 @c @findex LC_NUMERIC
13060 These environment variables control the way that GCC uses
13061 localization information that allow GCC to work with different
13062 national conventions. GCC inspects the locale categories
13063 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13064 so. These locale categories can be set to any value supported by your
13065 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13066 Kingdom encoded in UTF-8.
13068 The @env{LC_CTYPE} environment variable specifies character
13069 classification. GCC uses it to determine the character boundaries in
13070 a string; this is needed for some multibyte encodings that contain quote
13071 and escape characters that would otherwise be interpreted as a string
13074 The @env{LC_MESSAGES} environment variable specifies the language to
13075 use in diagnostic messages.
13077 If the @env{LC_ALL} environment variable is set, it overrides the value
13078 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13079 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13080 environment variable. If none of these variables are set, GCC
13081 defaults to traditional C English behavior.
13085 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13086 files. GCC uses temporary files to hold the output of one stage of
13087 compilation which is to be used as input to the next stage: for example,
13088 the output of the preprocessor, which is the input to the compiler
13091 @item GCC_EXEC_PREFIX
13092 @findex GCC_EXEC_PREFIX
13093 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13094 names of the subprograms executed by the compiler. No slash is added
13095 when this prefix is combined with the name of a subprogram, but you can
13096 specify a prefix that ends with a slash if you wish.
13098 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13099 an appropriate prefix to use based on the pathname it was invoked with.
13101 If GCC cannot find the subprogram using the specified prefix, it
13102 tries looking in the usual places for the subprogram.
13104 The default value of @env{GCC_EXEC_PREFIX} is
13105 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13106 of @code{prefix} when you ran the @file{configure} script.
13108 Other prefixes specified with @option{-B} take precedence over this prefix.
13110 This prefix is also used for finding files such as @file{crt0.o} that are
13113 In addition, the prefix is used in an unusual way in finding the
13114 directories to search for header files. For each of the standard
13115 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13116 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13117 replacing that beginning with the specified prefix to produce an
13118 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13119 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13120 These alternate directories are searched first; the standard directories
13123 @item COMPILER_PATH
13124 @findex COMPILER_PATH
13125 The value of @env{COMPILER_PATH} is a colon-separated list of
13126 directories, much like @env{PATH}. GCC tries the directories thus
13127 specified when searching for subprograms, if it can't find the
13128 subprograms using @env{GCC_EXEC_PREFIX}.
13131 @findex LIBRARY_PATH
13132 The value of @env{LIBRARY_PATH} is a colon-separated list of
13133 directories, much like @env{PATH}. When configured as a native compiler,
13134 GCC tries the directories thus specified when searching for special
13135 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13136 using GCC also uses these directories when searching for ordinary
13137 libraries for the @option{-l} option (but directories specified with
13138 @option{-L} come first).
13142 @cindex locale definition
13143 This variable is used to pass locale information to the compiler. One way in
13144 which this information is used is to determine the character set to be used
13145 when character literals, string literals and comments are parsed in C and C++.
13146 When the compiler is configured to allow multibyte characters,
13147 the following values for @env{LANG} are recognized:
13151 Recognize JIS characters.
13153 Recognize SJIS characters.
13155 Recognize EUCJP characters.
13158 If @env{LANG} is not defined, or if it has some other value, then the
13159 compiler will use mblen and mbtowc as defined by the default locale to
13160 recognize and translate multibyte characters.
13164 Some additional environments variables affect the behavior of the
13167 @include cppenv.texi
13171 @node Precompiled Headers
13172 @section Using Precompiled Headers
13173 @cindex precompiled headers
13174 @cindex speed of compilation
13176 Often large projects have many header files that are included in every
13177 source file. The time the compiler takes to process these header files
13178 over and over again can account for nearly all of the time required to
13179 build the project. To make builds faster, GCC allows users to
13180 `precompile' a header file; then, if builds can use the precompiled
13181 header file they will be much faster.
13183 To create a precompiled header file, simply compile it as you would any
13184 other file, if necessary using the @option{-x} option to make the driver
13185 treat it as a C or C++ header file. You will probably want to use a
13186 tool like @command{make} to keep the precompiled header up-to-date when
13187 the headers it contains change.
13189 A precompiled header file will be searched for when @code{#include} is
13190 seen in the compilation. As it searches for the included file
13191 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13192 compiler looks for a precompiled header in each directory just before it
13193 looks for the include file in that directory. The name searched for is
13194 the name specified in the @code{#include} with @samp{.gch} appended. If
13195 the precompiled header file can't be used, it is ignored.
13197 For instance, if you have @code{#include "all.h"}, and you have
13198 @file{all.h.gch} in the same directory as @file{all.h}, then the
13199 precompiled header file will be used if possible, and the original
13200 header will be used otherwise.
13202 Alternatively, you might decide to put the precompiled header file in a
13203 directory and use @option{-I} to ensure that directory is searched
13204 before (or instead of) the directory containing the original header.
13205 Then, if you want to check that the precompiled header file is always
13206 used, you can put a file of the same name as the original header in this
13207 directory containing an @code{#error} command.
13209 This also works with @option{-include}. So yet another way to use
13210 precompiled headers, good for projects not designed with precompiled
13211 header files in mind, is to simply take most of the header files used by
13212 a project, include them from another header file, precompile that header
13213 file, and @option{-include} the precompiled header. If the header files
13214 have guards against multiple inclusion, they will be skipped because
13215 they've already been included (in the precompiled header).
13217 If you need to precompile the same header file for different
13218 languages, targets, or compiler options, you can instead make a
13219 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13220 header in the directory, perhaps using @option{-o}. It doesn't matter
13221 what you call the files in the directory, every precompiled header in
13222 the directory will be considered. The first precompiled header
13223 encountered in the directory that is valid for this compilation will
13224 be used; they're searched in no particular order.
13226 There are many other possibilities, limited only by your imagination,
13227 good sense, and the constraints of your build system.
13229 A precompiled header file can be used only when these conditions apply:
13233 Only one precompiled header can be used in a particular compilation.
13236 A precompiled header can't be used once the first C token is seen. You
13237 can have preprocessor directives before a precompiled header; you can
13238 even include a precompiled header from inside another header, so long as
13239 there are no C tokens before the @code{#include}.
13242 The precompiled header file must be produced for the same language as
13243 the current compilation. You can't use a C precompiled header for a C++
13247 The precompiled header file must have been produced by the same compiler
13248 binary as the current compilation is using.
13251 Any macros defined before the precompiled header is included must
13252 either be defined in the same way as when the precompiled header was
13253 generated, or must not affect the precompiled header, which usually
13254 means that they don't appear in the precompiled header at all.
13256 The @option{-D} option is one way to define a macro before a
13257 precompiled header is included; using a @code{#define} can also do it.
13258 There are also some options that define macros implicitly, like
13259 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13262 @item If debugging information is output when using the precompiled
13263 header, using @option{-g} or similar, the same kind of debugging information
13264 must have been output when building the precompiled header. However,
13265 a precompiled header built using @option{-g} can be used in a compilation
13266 when no debugging information is being output.
13268 @item The same @option{-m} options must generally be used when building
13269 and using the precompiled header. @xref{Submodel Options},
13270 for any cases where this rule is relaxed.
13272 @item Each of the following options must be the same when building and using
13273 the precompiled header:
13275 @gccoptlist{-fexceptions -funit-at-a-time}
13278 Some other command-line options starting with @option{-f},
13279 @option{-p}, or @option{-O} must be defined in the same way as when
13280 the precompiled header was generated. At present, it's not clear
13281 which options are safe to change and which are not; the safest choice
13282 is to use exactly the same options when generating and using the
13283 precompiled header. The following are known to be safe:
13285 @gccoptlist{-fmessage-length= -fpreprocessed
13286 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13287 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13292 For all of these except the last, the compiler will automatically
13293 ignore the precompiled header if the conditions aren't met. If you
13294 find an option combination that doesn't work and doesn't cause the
13295 precompiled header to be ignored, please consider filing a bug report,
13298 If you do use differing options when generating and using the
13299 precompiled header, the actual behavior will be a mixture of the
13300 behavior for the options. For instance, if you use @option{-g} to
13301 generate the precompiled header but not when using it, you may or may
13302 not get debugging information for routines in the precompiled header.
13304 @node Running Protoize
13305 @section Running Protoize
13307 The program @code{protoize} is an optional part of GCC@. You can use
13308 it to add prototypes to a program, thus converting the program to ISO
13309 C in one respect. The companion program @code{unprotoize} does the
13310 reverse: it removes argument types from any prototypes that are found.
13312 When you run these programs, you must specify a set of source files as
13313 command line arguments. The conversion programs start out by compiling
13314 these files to see what functions they define. The information gathered
13315 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
13317 After scanning comes actual conversion. The specified files are all
13318 eligible to be converted; any files they include (whether sources or
13319 just headers) are eligible as well.
13321 But not all the eligible files are converted. By default,
13322 @code{protoize} and @code{unprotoize} convert only source and header
13323 files in the current directory. You can specify additional directories
13324 whose files should be converted with the @option{-d @var{directory}}
13325 option. You can also specify particular files to exclude with the
13326 @option{-x @var{file}} option. A file is converted if it is eligible, its
13327 directory name matches one of the specified directory names, and its
13328 name within the directory has not been excluded.
13330 Basic conversion with @code{protoize} consists of rewriting most
13331 function definitions and function declarations to specify the types of
13332 the arguments. The only ones not rewritten are those for varargs
13335 @code{protoize} optionally inserts prototype declarations at the
13336 beginning of the source file, to make them available for any calls that
13337 precede the function's definition. Or it can insert prototype
13338 declarations with block scope in the blocks where undeclared functions
13341 Basic conversion with @code{unprotoize} consists of rewriting most
13342 function declarations to remove any argument types, and rewriting
13343 function definitions to the old-style pre-ISO form.
13345 Both conversion programs print a warning for any function declaration or
13346 definition that they can't convert. You can suppress these warnings
13349 The output from @code{protoize} or @code{unprotoize} replaces the
13350 original source file. The original file is renamed to a name ending
13351 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
13352 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
13353 for DOS) file already exists, then the source file is simply discarded.
13355 @code{protoize} and @code{unprotoize} both depend on GCC itself to
13356 scan the program and collect information about the functions it uses.
13357 So neither of these programs will work until GCC is installed.
13359 Here is a table of the options you can use with @code{protoize} and
13360 @code{unprotoize}. Each option works with both programs unless
13364 @item -B @var{directory}
13365 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
13366 usual directory (normally @file{/usr/local/lib}). This file contains
13367 prototype information about standard system functions. This option
13368 applies only to @code{protoize}.
13370 @item -c @var{compilation-options}
13371 Use @var{compilation-options} as the options when running @command{gcc} to
13372 produce the @samp{.X} files. The special option @option{-aux-info} is
13373 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
13375 Note that the compilation options must be given as a single argument to
13376 @code{protoize} or @code{unprotoize}. If you want to specify several
13377 @command{gcc} options, you must quote the entire set of compilation options
13378 to make them a single word in the shell.
13380 There are certain @command{gcc} arguments that you cannot use, because they
13381 would produce the wrong kind of output. These include @option{-g},
13382 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
13383 the @var{compilation-options}, they are ignored.
13386 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
13387 systems) instead of @samp{.c}. This is convenient if you are converting
13388 a C program to C++. This option applies only to @code{protoize}.
13391 Add explicit global declarations. This means inserting explicit
13392 declarations at the beginning of each source file for each function
13393 that is called in the file and was not declared. These declarations
13394 precede the first function definition that contains a call to an
13395 undeclared function. This option applies only to @code{protoize}.
13397 @item -i @var{string}
13398 Indent old-style parameter declarations with the string @var{string}.
13399 This option applies only to @code{protoize}.
13401 @code{unprotoize} converts prototyped function definitions to old-style
13402 function definitions, where the arguments are declared between the
13403 argument list and the initial @samp{@{}. By default, @code{unprotoize}
13404 uses five spaces as the indentation. If you want to indent with just
13405 one space instead, use @option{-i " "}.
13408 Keep the @samp{.X} files. Normally, they are deleted after conversion
13412 Add explicit local declarations. @code{protoize} with @option{-l} inserts
13413 a prototype declaration for each function in each block which calls the
13414 function without any declaration. This option applies only to
13418 Make no real changes. This mode just prints information about the conversions
13419 that would have been done without @option{-n}.
13422 Make no @samp{.save} files. The original files are simply deleted.
13423 Use this option with caution.
13425 @item -p @var{program}
13426 Use the program @var{program} as the compiler. Normally, the name
13427 @file{gcc} is used.
13430 Work quietly. Most warnings are suppressed.
13433 Print the version number, just like @option{-v} for @command{gcc}.
13436 If you need special compiler options to compile one of your program's
13437 source files, then you should generate that file's @samp{.X} file
13438 specially, by running @command{gcc} on that source file with the
13439 appropriate options and the option @option{-aux-info}. Then run
13440 @code{protoize} on the entire set of files. @code{protoize} will use
13441 the existing @samp{.X} file because it is newer than the source file.
13445 gcc -Dfoo=bar file1.c -aux-info file1.X
13450 You need to include the special files along with the rest in the
13451 @code{protoize} command, even though their @samp{.X} files already
13452 exist, because otherwise they won't get converted.
13454 @xref{Protoize Caveats}, for more information on how to use
13455 @code{protoize} successfully.