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 @gol
419 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
420 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
422 @emph{Blackfin Options}
423 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
424 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
425 -mlow-64k -mno-low64k -mid-shared-library @gol
426 -mno-id-shared-library -mshared-library-id=@var{n} @gol
427 -mlong-calls -mno-long-calls}
430 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
431 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
432 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
433 -mstack-align -mdata-align -mconst-align @gol
434 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
435 -melf -maout -melinux -mlinux -sim -sim2 @gol
436 -mmul-bug-workaround -mno-mul-bug-workaround}
439 @gccoptlist{-mmac -mpush-args}
441 @emph{Darwin Options}
442 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
443 -arch_only -bind_at_load -bundle -bundle_loader @gol
444 -client_name -compatibility_version -current_version @gol
446 -dependency-file -dylib_file -dylinker_install_name @gol
447 -dynamic -dynamiclib -exported_symbols_list @gol
448 -filelist -flat_namespace -force_cpusubtype_ALL @gol
449 -force_flat_namespace -headerpad_max_install_names @gol
450 -image_base -init -install_name -keep_private_externs @gol
451 -multi_module -multiply_defined -multiply_defined_unused @gol
452 -noall_load -no_dead_strip_inits_and_terms @gol
453 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
454 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
455 -private_bundle -read_only_relocs -sectalign @gol
456 -sectobjectsymbols -whyload -seg1addr @gol
457 -sectcreate -sectobjectsymbols -sectorder @gol
458 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
459 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
460 -segprot -segs_read_only_addr -segs_read_write_addr @gol
461 -single_module -static -sub_library -sub_umbrella @gol
462 -twolevel_namespace -umbrella -undefined @gol
463 -unexported_symbols_list -weak_reference_mismatches @gol
464 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
467 @emph{DEC Alpha Options}
468 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
469 -mieee -mieee-with-inexact -mieee-conformant @gol
470 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
471 -mtrap-precision=@var{mode} -mbuild-constants @gol
472 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
473 -mbwx -mmax -mfix -mcix @gol
474 -mfloat-vax -mfloat-ieee @gol
475 -mexplicit-relocs -msmall-data -mlarge-data @gol
476 -msmall-text -mlarge-text @gol
477 -mmemory-latency=@var{time}}
479 @emph{DEC Alpha/VMS Options}
480 @gccoptlist{-mvms-return-codes}
483 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
484 -mhard-float -msoft-float @gol
485 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
486 -mdouble -mno-double @gol
487 -mmedia -mno-media -mmuladd -mno-muladd @gol
488 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
489 -mlinked-fp -mlong-calls -malign-labels @gol
490 -mlibrary-pic -macc-4 -macc-8 @gol
491 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
492 -moptimize-membar -mno-optimize-membar @gol
493 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
494 -mvliw-branch -mno-vliw-branch @gol
495 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
496 -mno-nested-cond-exec -mtomcat-stats @gol
500 @emph{H8/300 Options}
501 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
504 @gccoptlist{-march=@var{architecture-type} @gol
505 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
506 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
507 -mfixed-range=@var{register-range} @gol
508 -mjump-in-delay -mlinker-opt -mlong-calls @gol
509 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
510 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
511 -mno-jump-in-delay -mno-long-load-store @gol
512 -mno-portable-runtime -mno-soft-float @gol
513 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
514 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
515 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
516 -munix=@var{unix-std} -nolibdld -static -threads}
518 @emph{i386 and x86-64 Options}
519 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
520 -mfpmath=@var{unit} @gol
521 -masm=@var{dialect} -mno-fancy-math-387 @gol
522 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
523 -mno-wide-multiply -mrtd -malign-double @gol
524 -mpreferred-stack-boundary=@var{num} @gol
525 -mmmx -msse -msse2 -msse3 -m3dnow @gol
526 -mthreads -mno-align-stringops -minline-all-stringops @gol
527 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
528 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
529 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
530 -mcmodel=@var{code-model} @gol
531 -m32 -m64 -mlarge-data-threshold=@var{num}}
534 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
535 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
536 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
537 -minline-float-divide-max-throughput @gol
538 -minline-int-divide-min-latency @gol
539 -minline-int-divide-max-throughput @gol
540 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
541 -mno-dwarf2-asm -mearly-stop-bits @gol
542 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
543 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64}
545 @emph{M32R/D Options}
546 @gccoptlist{-m32r2 -m32rx -m32r @gol
548 -malign-loops -mno-align-loops @gol
549 -missue-rate=@var{number} @gol
550 -mbranch-cost=@var{number} @gol
551 -mmodel=@var{code-size-model-type} @gol
552 -msdata=@var{sdata-type} @gol
553 -mno-flush-func -mflush-func=@var{name} @gol
554 -mno-flush-trap -mflush-trap=@var{number} @gol
558 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
560 @emph{M680x0 Options}
561 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
562 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
563 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
564 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
565 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
567 @emph{M68hc1x Options}
568 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
569 -mauto-incdec -minmax -mlong-calls -mshort @gol
570 -msoft-reg-count=@var{count}}
573 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
574 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
575 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
576 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
577 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
580 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
581 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
582 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
583 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
584 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
585 -mdsp -mpaired-single -mips3d @gol
586 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
587 -G@var{num} -membedded-data -mno-embedded-data @gol
588 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
589 -msplit-addresses -mno-split-addresses @gol
590 -mexplicit-relocs -mno-explicit-relocs @gol
591 -mcheck-zero-division -mno-check-zero-division @gol
592 -mdivide-traps -mdivide-breaks @gol
593 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
594 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
595 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
596 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
597 -mfix-sb1 -mno-fix-sb1 @gol
598 -mflush-func=@var{func} -mno-flush-func @gol
599 -mbranch-likely -mno-branch-likely @gol
600 -mfp-exceptions -mno-fp-exceptions @gol
601 -mvr4130-align -mno-vr4130-align}
604 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
605 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
606 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
607 -mno-base-addresses -msingle-exit -mno-single-exit}
609 @emph{MN10300 Options}
610 @gccoptlist{-mmult-bug -mno-mult-bug @gol
611 -mam33 -mno-am33 @gol
612 -mam33-2 -mno-am33-2 @gol
613 -mreturn-pointer-on-d0 @gol
617 @gccoptlist{-mno-crt0 -mmul -mbacc -msim @gol
618 -march=@var{cpu-type} }
620 @emph{PDP-11 Options}
621 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
622 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
623 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
624 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
625 -mbranch-expensive -mbranch-cheap @gol
626 -msplit -mno-split -munix-asm -mdec-asm}
628 @emph{PowerPC Options}
629 See RS/6000 and PowerPC Options.
631 @emph{RS/6000 and PowerPC Options}
632 @gccoptlist{-mcpu=@var{cpu-type} @gol
633 -mtune=@var{cpu-type} @gol
634 -mpower -mno-power -mpower2 -mno-power2 @gol
635 -mpowerpc -mpowerpc64 -mno-powerpc @gol
636 -maltivec -mno-altivec @gol
637 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
638 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
639 -mnew-mnemonics -mold-mnemonics @gol
640 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
641 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
642 -malign-power -malign-natural @gol
643 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
644 -mstring -mno-string -mupdate -mno-update @gol
645 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
646 -mstrict-align -mno-strict-align -mrelocatable @gol
647 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
648 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
649 -mdynamic-no-pic -maltivec -mswdiv @gol
650 -mprioritize-restricted-insns=@var{priority} @gol
651 -msched-costly-dep=@var{dependence_type} @gol
652 -minsert-sched-nops=@var{scheme} @gol
653 -mcall-sysv -mcall-netbsd @gol
654 -maix-struct-return -msvr4-struct-return @gol
655 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
656 -misel -mno-isel @gol
657 -misel=yes -misel=no @gol
659 -mspe=yes -mspe=no @gol
660 -mvrsave -mno-vrsave @gol
661 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
662 -mprototype -mno-prototype @gol
663 -msim -mmvme -mads -myellowknife -memb -msdata @gol
664 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
666 @emph{S/390 and zSeries Options}
667 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
668 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
669 -mpacked-stack -mno-packed-stack @gol
670 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
671 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
672 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
673 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
676 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
677 -m4-nofpu -m4-single-only -m4-single -m4 @gol
678 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
679 -m5-64media -m5-64media-nofpu @gol
680 -m5-32media -m5-32media-nofpu @gol
681 -m5-compact -m5-compact-nofpu @gol
682 -mb -ml -mdalign -mrelax @gol
683 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
684 -mieee -misize -mpadstruct -mspace @gol
685 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
686 -mdivsi3_libfunc=@var{name} @gol
687 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
691 @gccoptlist{-mcpu=@var{cpu-type} @gol
692 -mtune=@var{cpu-type} @gol
693 -mcmodel=@var{code-model} @gol
694 -m32 -m64 -mapp-regs -mno-app-regs @gol
695 -mfaster-structs -mno-faster-structs @gol
696 -mfpu -mno-fpu -mhard-float -msoft-float @gol
697 -mhard-quad-float -msoft-quad-float @gol
698 -mimpure-text -mno-impure-text -mlittle-endian @gol
699 -mstack-bias -mno-stack-bias @gol
700 -munaligned-doubles -mno-unaligned-doubles @gol
701 -mv8plus -mno-v8plus -mvis -mno-vis
704 @emph{System V Options}
705 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
707 @emph{TMS320C3x/C4x Options}
708 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
709 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
710 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
711 -mparallel-insns -mparallel-mpy -mpreserve-float}
714 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
715 -mprolog-function -mno-prolog-function -mspace @gol
716 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
717 -mapp-regs -mno-app-regs @gol
718 -mdisable-callt -mno-disable-callt @gol
724 @gccoptlist{-mg -mgnu -munix}
726 @emph{x86-64 Options}
727 See i386 and x86-64 Options.
729 @emph{Xstormy16 Options}
732 @emph{Xtensa Options}
733 @gccoptlist{-mconst16 -mno-const16 @gol
734 -mfused-madd -mno-fused-madd @gol
735 -mtext-section-literals -mno-text-section-literals @gol
736 -mtarget-align -mno-target-align @gol
737 -mlongcalls -mno-longcalls}
739 @emph{zSeries Options}
740 See S/390 and zSeries Options.
742 @item Code Generation Options
743 @xref{Code Gen Options,,Options for Code Generation Conventions}.
744 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
745 -ffixed-@var{reg} -fexceptions @gol
746 -fnon-call-exceptions -funwind-tables @gol
747 -fasynchronous-unwind-tables @gol
748 -finhibit-size-directive -finstrument-functions @gol
749 -fno-common -fno-ident @gol
750 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
751 -fno-jump-tables @gol
752 -freg-struct-return -fshared-data -fshort-enums @gol
753 -fshort-double -fshort-wchar @gol
754 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
755 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
756 -fargument-alias -fargument-noalias @gol
757 -fargument-noalias-global -fleading-underscore @gol
758 -ftls-model=@var{model} @gol
759 -ftrapv -fwrapv -fbounds-check @gol
764 * Overall Options:: Controlling the kind of output:
765 an executable, object files, assembler files,
766 or preprocessed source.
767 * C Dialect Options:: Controlling the variant of C language compiled.
768 * C++ Dialect Options:: Variations on C++.
769 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
771 * Language Independent Options:: Controlling how diagnostics should be
773 * Warning Options:: How picky should the compiler be?
774 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
775 * Optimize Options:: How much optimization?
776 * Preprocessor Options:: Controlling header files and macro definitions.
777 Also, getting dependency information for Make.
778 * Assembler Options:: Passing options to the assembler.
779 * Link Options:: Specifying libraries and so on.
780 * Directory Options:: Where to find header files and libraries.
781 Where to find the compiler executable files.
782 * Spec Files:: How to pass switches to sub-processes.
783 * Target Options:: Running a cross-compiler, or an old version of GCC.
786 @node Overall Options
787 @section Options Controlling the Kind of Output
789 Compilation can involve up to four stages: preprocessing, compilation
790 proper, assembly and linking, always in that order. GCC is capable of
791 preprocessing and compiling several files either into several
792 assembler input files, or into one assembler input file; then each
793 assembler input file produces an object file, and linking combines all
794 the object files (those newly compiled, and those specified as input)
795 into an executable file.
797 @cindex file name suffix
798 For any given input file, the file name suffix determines what kind of
803 C source code which must be preprocessed.
806 C source code which should not be preprocessed.
809 C++ source code which should not be preprocessed.
812 Objective-C source code. Note that you must link with the @file{libobjc}
813 library to make an Objective-C program work.
816 Objective-C source code which should not be preprocessed.
820 Objective-C++ source code. Note that you must link with the @file{libobjc}
821 library to make an Objective-C++ program work. Note that @samp{.M} refers
822 to a literal capital M@.
825 Objective-C++ source code which should not be preprocessed.
828 C, C++, Objective-C or Objective-C++ header file to be turned into a
833 @itemx @var{file}.cxx
834 @itemx @var{file}.cpp
835 @itemx @var{file}.CPP
836 @itemx @var{file}.c++
838 C++ source code which must be preprocessed. Note that in @samp{.cxx},
839 the last two letters must both be literally @samp{x}. Likewise,
840 @samp{.C} refers to a literal capital C@.
844 Objective-C++ source code which must be preprocessed.
847 Objective-C++ source code which should not be preprocessed.
851 C++ header file to be turned into a precompiled header.
854 @itemx @var{file}.for
855 @itemx @var{file}.FOR
856 Fortran source code which should not be preprocessed.
859 @itemx @var{file}.fpp
860 @itemx @var{file}.FPP
861 Fortran source code which must be preprocessed (with the traditional
865 Fortran source code which must be preprocessed with a RATFOR
866 preprocessor (not included with GCC)@.
869 @itemx @var{file}.f95
870 Fortran 90/95 source code which should not be preprocessed.
873 @itemx @var{file}.F95
874 Fortran 90/95 source code which must be preprocessed (with the
875 traditional preprocessor).
877 @c FIXME: Descriptions of Java file types.
884 Ada source code file which contains a library unit declaration (a
885 declaration of a package, subprogram, or generic, or a generic
886 instantiation), or a library unit renaming declaration (a package,
887 generic, or subprogram renaming declaration). Such files are also
890 @itemx @var{file}.adb
891 Ada source code file containing a library unit body (a subprogram or
892 package body). Such files are also called @dfn{bodies}.
894 @c GCC also knows about some suffixes for languages not yet included:
903 Assembler code which must be preprocessed.
906 An object file to be fed straight into linking.
907 Any file name with no recognized suffix is treated this way.
911 You can specify the input language explicitly with the @option{-x} option:
914 @item -x @var{language}
915 Specify explicitly the @var{language} for the following input files
916 (rather than letting the compiler choose a default based on the file
917 name suffix). This option applies to all following input files until
918 the next @option{-x} option. Possible values for @var{language} are:
920 c c-header c-cpp-output
921 c++ c++-header c++-cpp-output
922 objective-c objective-c-header objective-c-cpp-output
923 objective-c++ objective-c++-header objective-c++-cpp-output
924 assembler assembler-with-cpp
926 f77 f77-cpp-input ratfor
933 Turn off any specification of a language, so that subsequent files are
934 handled according to their file name suffixes (as they are if @option{-x}
935 has not been used at all).
937 @item -pass-exit-codes
938 @opindex pass-exit-codes
939 Normally the @command{gcc} program will exit with the code of 1 if any
940 phase of the compiler returns a non-success return code. If you specify
941 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
942 numerically highest error produced by any phase that returned an error
946 If you only want some of the stages of compilation, you can use
947 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
948 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
949 @command{gcc} is to stop. Note that some combinations (for example,
950 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
955 Compile or assemble the source files, but do not link. The linking
956 stage simply is not done. The ultimate output is in the form of an
957 object file for each source file.
959 By default, the object file name for a source file is made by replacing
960 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
962 Unrecognized input files, not requiring compilation or assembly, are
967 Stop after the stage of compilation proper; do not assemble. The output
968 is in the form of an assembler code file for each non-assembler input
971 By default, the assembler file name for a source file is made by
972 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
974 Input files that don't require compilation are ignored.
978 Stop after the preprocessing stage; do not run the compiler proper. The
979 output is in the form of preprocessed source code, which is sent to the
982 Input files which don't require preprocessing are ignored.
984 @cindex output file option
987 Place output in file @var{file}. This applies regardless to whatever
988 sort of output is being produced, whether it be an executable file,
989 an object file, an assembler file or preprocessed C code.
991 If @option{-o} is not specified, the default is to put an executable
992 file in @file{a.out}, the object file for
993 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
994 assembler file in @file{@var{source}.s}, a precompiled header file in
995 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1000 Print (on standard error output) the commands executed to run the stages
1001 of compilation. Also print the version number of the compiler driver
1002 program and of the preprocessor and the compiler proper.
1006 Like @option{-v} except the commands are not executed and all command
1007 arguments are quoted. This is useful for shell scripts to capture the
1008 driver-generated command lines.
1012 Use pipes rather than temporary files for communication between the
1013 various stages of compilation. This fails to work on some systems where
1014 the assembler is unable to read from a pipe; but the GNU assembler has
1019 If you are compiling multiple source files, this option tells the driver
1020 to pass all the source files to the compiler at once (for those
1021 languages for which the compiler can handle this). This will allow
1022 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1023 language for which this is supported is C@. If you pass source files for
1024 multiple languages to the driver, using this option, the driver will invoke
1025 the compiler(s) that support IMA once each, passing each compiler all the
1026 source files appropriate for it. For those languages that do not support
1027 IMA this option will be ignored, and the compiler will be invoked once for
1028 each source file in that language. If you use this option in conjunction
1029 with @option{-save-temps}, the compiler will generate multiple
1031 (one for each source file), but only one (combined) @file{.o} or
1036 Print (on the standard output) a description of the command line options
1037 understood by @command{gcc}. If the @option{-v} option is also specified
1038 then @option{--help} will also be passed on to the various processes
1039 invoked by @command{gcc}, so that they can display the command line options
1040 they accept. If the @option{-Wextra} option is also specified then command
1041 line options which have no documentation associated with them will also
1045 @opindex target-help
1046 Print (on the standard output) a description of target specific command
1047 line options for each tool.
1051 Display the version number and copyrights of the invoked GCC@.
1055 @section Compiling C++ Programs
1057 @cindex suffixes for C++ source
1058 @cindex C++ source file suffixes
1059 C++ source files conventionally use one of the suffixes @samp{.C},
1060 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1061 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1062 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1063 files with these names and compiles them as C++ programs even if you
1064 call the compiler the same way as for compiling C programs (usually
1065 with the name @command{gcc}).
1069 However, C++ programs often require class libraries as well as a
1070 compiler that understands the C++ language---and under some
1071 circumstances, you might want to compile programs or header files from
1072 standard input, or otherwise without a suffix that flags them as C++
1073 programs. You might also like to precompile a C header file with a
1074 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1075 program that calls GCC with the default language set to C++, and
1076 automatically specifies linking against the C++ library. On many
1077 systems, @command{g++} is also installed with the name @command{c++}.
1079 @cindex invoking @command{g++}
1080 When you compile C++ programs, you may specify many of the same
1081 command-line options that you use for compiling programs in any
1082 language; or command-line options meaningful for C and related
1083 languages; or options that are meaningful only for C++ programs.
1084 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1085 explanations of options for languages related to C@.
1086 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1087 explanations of options that are meaningful only for C++ programs.
1089 @node C Dialect Options
1090 @section Options Controlling C Dialect
1091 @cindex dialect options
1092 @cindex language dialect options
1093 @cindex options, dialect
1095 The following options control the dialect of C (or languages derived
1096 from C, such as C++, Objective-C and Objective-C++) that the compiler
1100 @cindex ANSI support
1104 In C mode, support all ISO C90 programs. In C++ mode,
1105 remove GNU extensions that conflict with ISO C++.
1107 This turns off certain features of GCC that are incompatible with ISO
1108 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1109 such as the @code{asm} and @code{typeof} keywords, and
1110 predefined macros such as @code{unix} and @code{vax} that identify the
1111 type of system you are using. It also enables the undesirable and
1112 rarely used ISO trigraph feature. For the C compiler,
1113 it disables recognition of C++ style @samp{//} comments as well as
1114 the @code{inline} keyword.
1116 The alternate keywords @code{__asm__}, @code{__extension__},
1117 @code{__inline__} and @code{__typeof__} continue to work despite
1118 @option{-ansi}. You would not want to use them in an ISO C program, of
1119 course, but it is useful to put them in header files that might be included
1120 in compilations done with @option{-ansi}. Alternate predefined macros
1121 such as @code{__unix__} and @code{__vax__} are also available, with or
1122 without @option{-ansi}.
1124 The @option{-ansi} option does not cause non-ISO programs to be
1125 rejected gratuitously. For that, @option{-pedantic} is required in
1126 addition to @option{-ansi}. @xref{Warning Options}.
1128 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1129 option is used. Some header files may notice this macro and refrain
1130 from declaring certain functions or defining certain macros that the
1131 ISO standard doesn't call for; this is to avoid interfering with any
1132 programs that might use these names for other things.
1134 Functions which would normally be built in but do not have semantics
1135 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1136 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1137 built-in functions provided by GCC}, for details of the functions
1142 Determine the language standard. This option is currently only
1143 supported when compiling C or C++. A value for this option must be
1144 provided; possible values are
1149 ISO C90 (same as @option{-ansi}).
1151 @item iso9899:199409
1152 ISO C90 as modified in amendment 1.
1158 ISO C99. Note that this standard is not yet fully supported; see
1159 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1160 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1163 Default, ISO C90 plus GNU extensions (including some C99 features).
1167 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1168 this will become the default. The name @samp{gnu9x} is deprecated.
1171 The 1998 ISO C++ standard plus amendments.
1174 The same as @option{-std=c++98} plus GNU extensions. This is the
1175 default for C++ code.
1178 Even when this option is not specified, you can still use some of the
1179 features of newer standards in so far as they do not conflict with
1180 previous C standards. For example, you may use @code{__restrict__} even
1181 when @option{-std=c99} is not specified.
1183 The @option{-std} options specifying some version of ISO C have the same
1184 effects as @option{-ansi}, except that features that were not in ISO C90
1185 but are in the specified version (for example, @samp{//} comments and
1186 the @code{inline} keyword in ISO C99) are not disabled.
1188 @xref{Standards,,Language Standards Supported by GCC}, for details of
1189 these standard versions.
1191 @item -aux-info @var{filename}
1193 Output to the given filename prototyped declarations for all functions
1194 declared and/or defined in a translation unit, including those in header
1195 files. This option is silently ignored in any language other than C@.
1197 Besides declarations, the file indicates, in comments, the origin of
1198 each declaration (source file and line), whether the declaration was
1199 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1200 @samp{O} for old, respectively, in the first character after the line
1201 number and the colon), and whether it came from a declaration or a
1202 definition (@samp{C} or @samp{F}, respectively, in the following
1203 character). In the case of function definitions, a K&R-style list of
1204 arguments followed by their declarations is also provided, inside
1205 comments, after the declaration.
1209 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1210 keyword, so that code can use these words as identifiers. You can use
1211 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1212 instead. @option{-ansi} implies @option{-fno-asm}.
1214 In C++, this switch only affects the @code{typeof} keyword, since
1215 @code{asm} and @code{inline} are standard keywords. You may want to
1216 use the @option{-fno-gnu-keywords} flag instead, which has the same
1217 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1218 switch only affects the @code{asm} and @code{typeof} keywords, since
1219 @code{inline} is a standard keyword in ISO C99.
1222 @itemx -fno-builtin-@var{function}
1223 @opindex fno-builtin
1224 @cindex built-in functions
1225 Don't recognize built-in functions that do not begin with
1226 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1227 functions provided by GCC}, for details of the functions affected,
1228 including those which are not built-in functions when @option{-ansi} or
1229 @option{-std} options for strict ISO C conformance are used because they
1230 do not have an ISO standard meaning.
1232 GCC normally generates special code to handle certain built-in functions
1233 more efficiently; for instance, calls to @code{alloca} may become single
1234 instructions that adjust the stack directly, and calls to @code{memcpy}
1235 may become inline copy loops. The resulting code is often both smaller
1236 and faster, but since the function calls no longer appear as such, you
1237 cannot set a breakpoint on those calls, nor can you change the behavior
1238 of the functions by linking with a different library. In addition,
1239 when a function is recognized as a built-in function, GCC may use
1240 information about that function to warn about problems with calls to
1241 that function, or to generate more efficient code, even if the
1242 resulting code still contains calls to that function. For example,
1243 warnings are given with @option{-Wformat} for bad calls to
1244 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1245 known not to modify global memory.
1247 With the @option{-fno-builtin-@var{function}} option
1248 only the built-in function @var{function} is
1249 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1250 function is named this is not built-in in this version of GCC, this
1251 option is ignored. There is no corresponding
1252 @option{-fbuiltin-@var{function}} option; if you wish to enable
1253 built-in functions selectively when using @option{-fno-builtin} or
1254 @option{-ffreestanding}, you may define macros such as:
1257 #define abs(n) __builtin_abs ((n))
1258 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1263 @cindex hosted environment
1265 Assert that compilation takes place in a hosted environment. This implies
1266 @option{-fbuiltin}. A hosted environment is one in which the
1267 entire standard library is available, and in which @code{main} has a return
1268 type of @code{int}. Examples are nearly everything except a kernel.
1269 This is equivalent to @option{-fno-freestanding}.
1271 @item -ffreestanding
1272 @opindex ffreestanding
1273 @cindex hosted environment
1275 Assert that compilation takes place in a freestanding environment. This
1276 implies @option{-fno-builtin}. A freestanding environment
1277 is one in which the standard library may not exist, and program startup may
1278 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1279 This is equivalent to @option{-fno-hosted}.
1281 @xref{Standards,,Language Standards Supported by GCC}, for details of
1282 freestanding and hosted environments.
1284 @item -fms-extensions
1285 @opindex fms-extensions
1286 Accept some non-standard constructs used in Microsoft header files.
1288 Some cases of unnamed fields in structures and unions are only
1289 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1290 fields within structs/unions}, for details.
1294 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1295 options for strict ISO C conformance) implies @option{-trigraphs}.
1297 @item -no-integrated-cpp
1298 @opindex no-integrated-cpp
1299 Performs a compilation in two passes: preprocessing and compiling. This
1300 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1301 @option{-B} option. The user supplied compilation step can then add in
1302 an additional preprocessing step after normal preprocessing but before
1303 compiling. The default is to use the integrated cpp (internal cpp)
1305 The semantics of this option will change if "cc1", "cc1plus", and
1306 "cc1obj" are merged.
1308 @cindex traditional C language
1309 @cindex C language, traditional
1311 @itemx -traditional-cpp
1312 @opindex traditional-cpp
1313 @opindex traditional
1314 Formerly, these options caused GCC to attempt to emulate a pre-standard
1315 C compiler. They are now only supported with the @option{-E} switch.
1316 The preprocessor continues to support a pre-standard mode. See the GNU
1317 CPP manual for details.
1319 @item -fcond-mismatch
1320 @opindex fcond-mismatch
1321 Allow conditional expressions with mismatched types in the second and
1322 third arguments. The value of such an expression is void. This option
1323 is not supported for C++.
1325 @item -funsigned-char
1326 @opindex funsigned-char
1327 Let the type @code{char} be unsigned, like @code{unsigned char}.
1329 Each kind of machine has a default for what @code{char} should
1330 be. It is either like @code{unsigned char} by default or like
1331 @code{signed char} by default.
1333 Ideally, a portable program should always use @code{signed char} or
1334 @code{unsigned char} when it depends on the signedness of an object.
1335 But many programs have been written to use plain @code{char} and
1336 expect it to be signed, or expect it to be unsigned, depending on the
1337 machines they were written for. This option, and its inverse, let you
1338 make such a program work with the opposite default.
1340 The type @code{char} is always a distinct type from each of
1341 @code{signed char} or @code{unsigned char}, even though its behavior
1342 is always just like one of those two.
1345 @opindex fsigned-char
1346 Let the type @code{char} be signed, like @code{signed char}.
1348 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1349 the negative form of @option{-funsigned-char}. Likewise, the option
1350 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1352 @item -fsigned-bitfields
1353 @itemx -funsigned-bitfields
1354 @itemx -fno-signed-bitfields
1355 @itemx -fno-unsigned-bitfields
1356 @opindex fsigned-bitfields
1357 @opindex funsigned-bitfields
1358 @opindex fno-signed-bitfields
1359 @opindex fno-unsigned-bitfields
1360 These options control whether a bit-field is signed or unsigned, when the
1361 declaration does not use either @code{signed} or @code{unsigned}. By
1362 default, such a bit-field is signed, because this is consistent: the
1363 basic integer types such as @code{int} are signed types.
1366 @node C++ Dialect Options
1367 @section Options Controlling C++ Dialect
1369 @cindex compiler options, C++
1370 @cindex C++ options, command line
1371 @cindex options, C++
1372 This section describes the command-line options that are only meaningful
1373 for C++ programs; but you can also use most of the GNU compiler options
1374 regardless of what language your program is in. For example, you
1375 might compile a file @code{firstClass.C} like this:
1378 g++ -g -frepo -O -c firstClass.C
1382 In this example, only @option{-frepo} is an option meant
1383 only for C++ programs; you can use the other options with any
1384 language supported by GCC@.
1386 Here is a list of options that are @emph{only} for compiling C++ programs:
1390 @item -fabi-version=@var{n}
1391 @opindex fabi-version
1392 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1393 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1394 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1395 the version that conforms most closely to the C++ ABI specification.
1396 Therefore, the ABI obtained using version 0 will change as ABI bugs
1399 The default is version 2.
1401 @item -fno-access-control
1402 @opindex fno-access-control
1403 Turn off all access checking. This switch is mainly useful for working
1404 around bugs in the access control code.
1408 Check that the pointer returned by @code{operator new} is non-null
1409 before attempting to modify the storage allocated. This check is
1410 normally unnecessary because the C++ standard specifies that
1411 @code{operator new} will only return @code{0} if it is declared
1412 @samp{throw()}, in which case the compiler will always check the
1413 return value even without this option. In all other cases, when
1414 @code{operator new} has a non-empty exception specification, memory
1415 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1416 @samp{new (nothrow)}.
1418 @item -fconserve-space
1419 @opindex fconserve-space
1420 Put uninitialized or runtime-initialized global variables into the
1421 common segment, as C does. This saves space in the executable at the
1422 cost of not diagnosing duplicate definitions. If you compile with this
1423 flag and your program mysteriously crashes after @code{main()} has
1424 completed, you may have an object that is being destroyed twice because
1425 two definitions were merged.
1427 This option is no longer useful on most targets, now that support has
1428 been added for putting variables into BSS without making them common.
1430 @item -ffriend-injection
1431 @opindex ffriend-injection
1432 Inject friend functions into the enclosing namespace, so that they are
1433 visible outside the scope of the class in which they are declared.
1434 Friend functions were documented to work this way in the old Annotated
1435 C++ Reference Manual, and versions of G++ before 4.1 always worked
1436 that way. However, in ISO C++ a friend function which is not declared
1437 in an enclosing scope can only be found using argument dependent
1438 lookup. This option causes friends to be injected as they were in
1441 This option is for compatibility, and may be removed in a future
1444 @item -fno-const-strings
1445 @opindex fno-const-strings
1446 Give string constants type @code{char *} instead of type @code{const
1447 char *}. By default, G++ uses type @code{const char *} as required by
1448 the standard. Even if you use @option{-fno-const-strings}, you cannot
1449 actually modify the value of a string constant.
1451 This option might be removed in a future release of G++. For maximum
1452 portability, you should structure your code so that it works with
1453 string constants that have type @code{const char *}.
1455 @item -fno-elide-constructors
1456 @opindex fno-elide-constructors
1457 The C++ standard allows an implementation to omit creating a temporary
1458 which is only used to initialize another object of the same type.
1459 Specifying this option disables that optimization, and forces G++ to
1460 call the copy constructor in all cases.
1462 @item -fno-enforce-eh-specs
1463 @opindex fno-enforce-eh-specs
1464 Don't generate code to check for violation of exception specifications
1465 at runtime. This option violates the C++ standard, but may be useful
1466 for reducing code size in production builds, much like defining
1467 @samp{NDEBUG}. This does not give user code permission to throw
1468 exceptions in violation of the exception specifications; the compiler
1469 will still optimize based on the specifications, so throwing an
1470 unexpected exception will result in undefined behavior.
1473 @itemx -fno-for-scope
1475 @opindex fno-for-scope
1476 If @option{-ffor-scope} is specified, the scope of variables declared in
1477 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1478 as specified by the C++ standard.
1479 If @option{-fno-for-scope} is specified, the scope of variables declared in
1480 a @i{for-init-statement} extends to the end of the enclosing scope,
1481 as was the case in old versions of G++, and other (traditional)
1482 implementations of C++.
1484 The default if neither flag is given to follow the standard,
1485 but to allow and give a warning for old-style code that would
1486 otherwise be invalid, or have different behavior.
1488 @item -fno-gnu-keywords
1489 @opindex fno-gnu-keywords
1490 Do not recognize @code{typeof} as a keyword, so that code can use this
1491 word as an identifier. You can use the keyword @code{__typeof__} instead.
1492 @option{-ansi} implies @option{-fno-gnu-keywords}.
1494 @item -fno-implicit-templates
1495 @opindex fno-implicit-templates
1496 Never emit code for non-inline templates which are instantiated
1497 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1498 @xref{Template Instantiation}, for more information.
1500 @item -fno-implicit-inline-templates
1501 @opindex fno-implicit-inline-templates
1502 Don't emit code for implicit instantiations of inline templates, either.
1503 The default is to handle inlines differently so that compiles with and
1504 without optimization will need the same set of explicit instantiations.
1506 @item -fno-implement-inlines
1507 @opindex fno-implement-inlines
1508 To save space, do not emit out-of-line copies of inline functions
1509 controlled by @samp{#pragma implementation}. This will cause linker
1510 errors if these functions are not inlined everywhere they are called.
1512 @item -fms-extensions
1513 @opindex fms-extensions
1514 Disable pedantic warnings about constructs used in MFC, such as implicit
1515 int and getting a pointer to member function via non-standard syntax.
1517 @item -fno-nonansi-builtins
1518 @opindex fno-nonansi-builtins
1519 Disable built-in declarations of functions that are not mandated by
1520 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1521 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1523 @item -fno-operator-names
1524 @opindex fno-operator-names
1525 Do not treat the operator name keywords @code{and}, @code{bitand},
1526 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1527 synonyms as keywords.
1529 @item -fno-optional-diags
1530 @opindex fno-optional-diags
1531 Disable diagnostics that the standard says a compiler does not need to
1532 issue. Currently, the only such diagnostic issued by G++ is the one for
1533 a name having multiple meanings within a class.
1536 @opindex fpermissive
1537 Downgrade some diagnostics about nonconformant code from errors to
1538 warnings. Thus, using @option{-fpermissive} will allow some
1539 nonconforming code to compile.
1543 Enable automatic template instantiation at link time. This option also
1544 implies @option{-fno-implicit-templates}. @xref{Template
1545 Instantiation}, for more information.
1549 Disable generation of information about every class with virtual
1550 functions for use by the C++ runtime type identification features
1551 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1552 of the language, you can save some space by using this flag. Note that
1553 exception handling uses the same information, but it will generate it as
1558 Emit statistics about front-end processing at the end of the compilation.
1559 This information is generally only useful to the G++ development team.
1561 @item -ftemplate-depth-@var{n}
1562 @opindex ftemplate-depth
1563 Set the maximum instantiation depth for template classes to @var{n}.
1564 A limit on the template instantiation depth is needed to detect
1565 endless recursions during template class instantiation. ANSI/ISO C++
1566 conforming programs must not rely on a maximum depth greater than 17.
1568 @item -fno-threadsafe-statics
1569 @opindex fno-threadsafe-statics
1570 Do not emit the extra code to use the routines specified in the C++
1571 ABI for thread-safe initialization of local statics. You can use this
1572 option to reduce code size slightly in code that doesn't need to be
1575 @item -fuse-cxa-atexit
1576 @opindex fuse-cxa-atexit
1577 Register destructors for objects with static storage duration with the
1578 @code{__cxa_atexit} function rather than the @code{atexit} function.
1579 This option is required for fully standards-compliant handling of static
1580 destructors, but will only work if your C library supports
1581 @code{__cxa_atexit}.
1583 @item -fvisibility-inlines-hidden
1584 @opindex fvisibility-inlines-hidden
1585 Causes all inlined methods to be marked with
1586 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1587 appear in the export table of a DSO and do not require a PLT indirection
1588 when used within the DSO@. Enabling this option can have a dramatic effect
1589 on load and link times of a DSO as it massively reduces the size of the
1590 dynamic export table when the library makes heavy use of templates. While
1591 it can cause bloating through duplication of code within each DSO where
1592 it is used, often the wastage is less than the considerable space occupied
1593 by a long symbol name in the export table which is typical when using
1594 templates and namespaces. For even more savings, combine with the
1595 @option{-fvisibility=hidden} switch.
1599 Do not use weak symbol support, even if it is provided by the linker.
1600 By default, G++ will use weak symbols if they are available. This
1601 option exists only for testing, and should not be used by end-users;
1602 it will result in inferior code and has no benefits. This option may
1603 be removed in a future release of G++.
1607 Do not search for header files in the standard directories specific to
1608 C++, but do still search the other standard directories. (This option
1609 is used when building the C++ library.)
1612 In addition, these optimization, warning, and code generation options
1613 have meanings only for C++ programs:
1616 @item -fno-default-inline
1617 @opindex fno-default-inline
1618 Do not assume @samp{inline} for functions defined inside a class scope.
1619 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1620 functions will have linkage like inline functions; they just won't be
1623 @item -Wabi @r{(C++ only)}
1625 Warn when G++ generates code that is probably not compatible with the
1626 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1627 all such cases, there are probably some cases that are not warned about,
1628 even though G++ is generating incompatible code. There may also be
1629 cases where warnings are emitted even though the code that is generated
1632 You should rewrite your code to avoid these warnings if you are
1633 concerned about the fact that code generated by G++ may not be binary
1634 compatible with code generated by other compilers.
1636 The known incompatibilities at this point include:
1641 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1642 pack data into the same byte as a base class. For example:
1645 struct A @{ virtual void f(); int f1 : 1; @};
1646 struct B : public A @{ int f2 : 1; @};
1650 In this case, G++ will place @code{B::f2} into the same byte
1651 as@code{A::f1}; other compilers will not. You can avoid this problem
1652 by explicitly padding @code{A} so that its size is a multiple of the
1653 byte size on your platform; that will cause G++ and other compilers to
1654 layout @code{B} identically.
1657 Incorrect handling of tail-padding for virtual bases. G++ does not use
1658 tail padding when laying out virtual bases. For example:
1661 struct A @{ virtual void f(); char c1; @};
1662 struct B @{ B(); char c2; @};
1663 struct C : public A, public virtual B @{@};
1667 In this case, G++ will not place @code{B} into the tail-padding for
1668 @code{A}; other compilers will. You can avoid this problem by
1669 explicitly padding @code{A} so that its size is a multiple of its
1670 alignment (ignoring virtual base classes); that will cause G++ and other
1671 compilers to layout @code{C} identically.
1674 Incorrect handling of bit-fields with declared widths greater than that
1675 of their underlying types, when the bit-fields appear in a union. For
1679 union U @{ int i : 4096; @};
1683 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1684 union too small by the number of bits in an @code{int}.
1687 Empty classes can be placed at incorrect offsets. For example:
1697 struct C : public B, public A @{@};
1701 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1702 it should be placed at offset zero. G++ mistakenly believes that the
1703 @code{A} data member of @code{B} is already at offset zero.
1706 Names of template functions whose types involve @code{typename} or
1707 template template parameters can be mangled incorrectly.
1710 template <typename Q>
1711 void f(typename Q::X) @{@}
1713 template <template <typename> class Q>
1714 void f(typename Q<int>::X) @{@}
1718 Instantiations of these templates may be mangled incorrectly.
1722 @item -Wctor-dtor-privacy @r{(C++ only)}
1723 @opindex Wctor-dtor-privacy
1724 Warn when a class seems unusable because all the constructors or
1725 destructors in that class are private, and it has neither friends nor
1726 public static member functions.
1728 @item -Wnon-virtual-dtor @r{(C++ only)}
1729 @opindex Wnon-virtual-dtor
1730 Warn when a class appears to be polymorphic, thereby requiring a virtual
1731 destructor, yet it declares a non-virtual one.
1732 This warning is enabled by @option{-Wall}.
1734 @item -Wreorder @r{(C++ only)}
1736 @cindex reordering, warning
1737 @cindex warning for reordering of member initializers
1738 Warn when the order of member initializers given in the code does not
1739 match the order in which they must be executed. For instance:
1745 A(): j (0), i (1) @{ @}
1749 The compiler will rearrange the member initializers for @samp{i}
1750 and @samp{j} to match the declaration order of the members, emitting
1751 a warning to that effect. This warning is enabled by @option{-Wall}.
1754 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1757 @item -Weffc++ @r{(C++ only)}
1759 Warn about violations of the following style guidelines from Scott Meyers'
1760 @cite{Effective C++} book:
1764 Item 11: Define a copy constructor and an assignment operator for classes
1765 with dynamically allocated memory.
1768 Item 12: Prefer initialization to assignment in constructors.
1771 Item 14: Make destructors virtual in base classes.
1774 Item 15: Have @code{operator=} return a reference to @code{*this}.
1777 Item 23: Don't try to return a reference when you must return an object.
1781 Also warn about violations of the following style guidelines from
1782 Scott Meyers' @cite{More Effective C++} book:
1786 Item 6: Distinguish between prefix and postfix forms of increment and
1787 decrement operators.
1790 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1794 When selecting this option, be aware that the standard library
1795 headers do not obey all of these guidelines; use @samp{grep -v}
1796 to filter out those warnings.
1798 @item -Wno-deprecated @r{(C++ only)}
1799 @opindex Wno-deprecated
1800 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1802 @item -Wstrict-null-sentinel @r{(C++ only)}
1803 @opindex Wstrict-null-sentinel
1804 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1805 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1806 to @code{__null}. Although it is a null pointer constant not a null pointer,
1807 it is guaranteed to of the same size as a pointer. But this use is
1808 not portable across different compilers.
1810 @item -Wno-non-template-friend @r{(C++ only)}
1811 @opindex Wno-non-template-friend
1812 Disable warnings when non-templatized friend functions are declared
1813 within a template. Since the advent of explicit template specification
1814 support in G++, if the name of the friend is an unqualified-id (i.e.,
1815 @samp{friend foo(int)}), the C++ language specification demands that the
1816 friend declare or define an ordinary, nontemplate function. (Section
1817 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1818 could be interpreted as a particular specialization of a templatized
1819 function. Because this non-conforming behavior is no longer the default
1820 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1821 check existing code for potential trouble spots and is on by default.
1822 This new compiler behavior can be turned off with
1823 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1824 but disables the helpful warning.
1826 @item -Wold-style-cast @r{(C++ only)}
1827 @opindex Wold-style-cast
1828 Warn if an old-style (C-style) cast to a non-void type is used within
1829 a C++ program. The new-style casts (@samp{dynamic_cast},
1830 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1831 less vulnerable to unintended effects and much easier to search for.
1833 @item -Woverloaded-virtual @r{(C++ only)}
1834 @opindex Woverloaded-virtual
1835 @cindex overloaded virtual fn, warning
1836 @cindex warning for overloaded virtual fn
1837 Warn when a function declaration hides virtual functions from a
1838 base class. For example, in:
1845 struct B: public A @{
1850 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1858 will fail to compile.
1860 @item -Wno-pmf-conversions @r{(C++ only)}
1861 @opindex Wno-pmf-conversions
1862 Disable the diagnostic for converting a bound pointer to member function
1865 @item -Wsign-promo @r{(C++ only)}
1866 @opindex Wsign-promo
1867 Warn when overload resolution chooses a promotion from unsigned or
1868 enumerated type to a signed type, over a conversion to an unsigned type of
1869 the same size. Previous versions of G++ would try to preserve
1870 unsignedness, but the standard mandates the current behavior.
1875 A& operator = (int);
1885 In this example, G++ will synthesize a default @samp{A& operator =
1886 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1889 @node Objective-C and Objective-C++ Dialect Options
1890 @section Options Controlling Objective-C and Objective-C++ Dialects
1892 @cindex compiler options, Objective-C and Objective-C++
1893 @cindex Objective-C and Objective-C++ options, command line
1894 @cindex options, Objective-C and Objective-C++
1895 (NOTE: This manual does not describe the Objective-C and Objective-C++
1896 languages themselves. See @xref{Standards,,Language Standards
1897 Supported by GCC}, for references.)
1899 This section describes the command-line options that are only meaningful
1900 for Objective-C and Objective-C++ programs, but you can also use most of
1901 the language-independent GNU compiler options.
1902 For example, you might compile a file @code{some_class.m} like this:
1905 gcc -g -fgnu-runtime -O -c some_class.m
1909 In this example, @option{-fgnu-runtime} is an option meant only for
1910 Objective-C and Objective-C++ programs; you can use the other options with
1911 any language supported by GCC@.
1913 Note that since Objective-C is an extension of the C language, Objective-C
1914 compilations may also use options specific to the C front-end (e.g.,
1915 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1916 C++-specific options (e.g., @option{-Wabi}).
1918 Here is a list of options that are @emph{only} for compiling Objective-C
1919 and Objective-C++ programs:
1922 @item -fconstant-string-class=@var{class-name}
1923 @opindex fconstant-string-class
1924 Use @var{class-name} as the name of the class to instantiate for each
1925 literal string specified with the syntax @code{@@"@dots{}"}. The default
1926 class name is @code{NXConstantString} if the GNU runtime is being used, and
1927 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1928 @option{-fconstant-cfstrings} option, if also present, will override the
1929 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1930 to be laid out as constant CoreFoundation strings.
1933 @opindex fgnu-runtime
1934 Generate object code compatible with the standard GNU Objective-C
1935 runtime. This is the default for most types of systems.
1937 @item -fnext-runtime
1938 @opindex fnext-runtime
1939 Generate output compatible with the NeXT runtime. This is the default
1940 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1941 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1944 @item -fno-nil-receivers
1945 @opindex fno-nil-receivers
1946 Assume that all Objective-C message dispatches (e.g.,
1947 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1948 is not @code{nil}. This allows for more efficient entry points in the runtime
1949 to be used. Currently, this option is only available in conjunction with
1950 the NeXT runtime on Mac OS X 10.3 and later.
1952 @item -fobjc-call-cxx-cdtors
1953 @opindex fobjc-call-cxx-cdtors
1954 For each Objective-C class, check if any of its instance variables is a
1955 C++ object with a non-trivial default constructor. If so, synthesize a
1956 special @code{- (id) .cxx_construct} instance method that will run
1957 non-trivial default constructors on any such instance variables, in order,
1958 and then return @code{self}. Similarly, check if any instance variable
1959 is a C++ object with a non-trivial destructor, and if so, synthesize a
1960 special @code{- (void) .cxx_destruct} method that will run
1961 all such default destructors, in reverse order.
1963 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
1964 thusly generated will only operate on instance variables declared in the
1965 current Objective-C class, and not those inherited from superclasses. It
1966 is the responsibility of the Objective-C runtime to invoke all such methods
1967 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
1968 will be invoked by the runtime immediately after a new object
1969 instance is allocated; the @code{- (void) .cxx_destruct} methods will
1970 be invoked immediately before the runtime deallocates an object instance.
1972 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
1973 support for invoking the @code{- (id) .cxx_construct} and
1974 @code{- (void) .cxx_destruct} methods.
1976 @item -fobjc-direct-dispatch
1977 @opindex fobjc-direct-dispatch
1978 Allow fast jumps to the message dispatcher. On Darwin this is
1979 accomplished via the comm page.
1981 @item -fobjc-exceptions
1982 @opindex fobjc-exceptions
1983 Enable syntactic support for structured exception handling in Objective-C,
1984 similar to what is offered by C++ and Java. Currently, this option is only
1985 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1993 @@catch (AnObjCClass *exc) @{
2000 @@catch (AnotherClass *exc) @{
2003 @@catch (id allOthers) @{
2013 The @code{@@throw} statement may appear anywhere in an Objective-C or
2014 Objective-C++ program; when used inside of a @code{@@catch} block, the
2015 @code{@@throw} may appear without an argument (as shown above), in which case
2016 the object caught by the @code{@@catch} will be rethrown.
2018 Note that only (pointers to) Objective-C objects may be thrown and
2019 caught using this scheme. When an object is thrown, it will be caught
2020 by the nearest @code{@@catch} clause capable of handling objects of that type,
2021 analogously to how @code{catch} blocks work in C++ and Java. A
2022 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2023 any and all Objective-C exceptions not caught by previous @code{@@catch}
2026 The @code{@@finally} clause, if present, will be executed upon exit from the
2027 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2028 regardless of whether any exceptions are thrown, caught or rethrown
2029 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2030 of the @code{finally} clause in Java.
2032 There are several caveats to using the new exception mechanism:
2036 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2037 idioms provided by the @code{NSException} class, the new
2038 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2039 systems, due to additional functionality needed in the (NeXT) Objective-C
2043 As mentioned above, the new exceptions do not support handling
2044 types other than Objective-C objects. Furthermore, when used from
2045 Objective-C++, the Objective-C exception model does not interoperate with C++
2046 exceptions at this time. This means you cannot @code{@@throw} an exception
2047 from Objective-C and @code{catch} it in C++, or vice versa
2048 (i.e., @code{throw @dots{} @@catch}).
2051 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2052 blocks for thread-safe execution:
2055 @@synchronized (ObjCClass *guard) @{
2060 Upon entering the @code{@@synchronized} block, a thread of execution shall
2061 first check whether a lock has been placed on the corresponding @code{guard}
2062 object by another thread. If it has, the current thread shall wait until
2063 the other thread relinquishes its lock. Once @code{guard} becomes available,
2064 the current thread will place its own lock on it, execute the code contained in
2065 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2066 making @code{guard} available to other threads).
2068 Unlike Java, Objective-C does not allow for entire methods to be marked
2069 @code{@@synchronized}. Note that throwing exceptions out of
2070 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2071 to be unlocked properly.
2075 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2077 @item -freplace-objc-classes
2078 @opindex freplace-objc-classes
2079 Emit a special marker instructing @command{ld(1)} not to statically link in
2080 the resulting object file, and allow @command{dyld(1)} to load it in at
2081 run time instead. This is used in conjunction with the Fix-and-Continue
2082 debugging mode, where the object file in question may be recompiled and
2083 dynamically reloaded in the course of program execution, without the need
2084 to restart the program itself. Currently, Fix-and-Continue functionality
2085 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2090 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2091 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2092 compile time) with static class references that get initialized at load time,
2093 which improves run-time performance. Specifying the @option{-fzero-link} flag
2094 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2095 to be retained. This is useful in Zero-Link debugging mode, since it allows
2096 for individual class implementations to be modified during program execution.
2100 Dump interface declarations for all classes seen in the source file to a
2101 file named @file{@var{sourcename}.decl}.
2103 @item -Wassign-intercept
2104 @opindex Wassign-intercept
2105 Warn whenever an Objective-C assignment is being intercepted by the
2109 @opindex Wno-protocol
2110 If a class is declared to implement a protocol, a warning is issued for
2111 every method in the protocol that is not implemented by the class. The
2112 default behavior is to issue a warning for every method not explicitly
2113 implemented in the class, even if a method implementation is inherited
2114 from the superclass. If you use the @option{-Wno-protocol} option, then
2115 methods inherited from the superclass are considered to be implemented,
2116 and no warning is issued for them.
2120 Warn if multiple methods of different types for the same selector are
2121 found during compilation. The check is performed on the list of methods
2122 in the final stage of compilation. Additionally, a check is performed
2123 for each selector appearing in a @code{@@selector(@dots{})}
2124 expression, and a corresponding method for that selector has been found
2125 during compilation. Because these checks scan the method table only at
2126 the end of compilation, these warnings are not produced if the final
2127 stage of compilation is not reached, for example because an error is
2128 found during compilation, or because the @option{-fsyntax-only} option is
2131 @item -Wstrict-selector-match
2132 @opindex Wstrict-selector-match
2133 Warn if multiple methods with differing argument and/or return types are
2134 found for a given selector when attempting to send a message using this
2135 selector to a receiver of type @code{id} or @code{Class}. When this flag
2136 is off (which is the default behavior), the compiler will omit such warnings
2137 if any differences found are confined to types which share the same size
2140 @item -Wundeclared-selector
2141 @opindex Wundeclared-selector
2142 Warn if a @code{@@selector(@dots{})} expression referring to an
2143 undeclared selector is found. A selector is considered undeclared if no
2144 method with that name has been declared before the
2145 @code{@@selector(@dots{})} expression, either explicitly in an
2146 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2147 an @code{@@implementation} section. This option always performs its
2148 checks as soon as a @code{@@selector(@dots{})} expression is found,
2149 while @option{-Wselector} only performs its checks in the final stage of
2150 compilation. This also enforces the coding style convention
2151 that methods and selectors must be declared before being used.
2153 @item -print-objc-runtime-info
2154 @opindex print-objc-runtime-info
2155 Generate C header describing the largest structure that is passed by
2160 @node Language Independent Options
2161 @section Options to Control Diagnostic Messages Formatting
2162 @cindex options to control diagnostics formatting
2163 @cindex diagnostic messages
2164 @cindex message formatting
2166 Traditionally, diagnostic messages have been formatted irrespective of
2167 the output device's aspect (e.g.@: its width, @dots{}). The options described
2168 below can be used to control the diagnostic messages formatting
2169 algorithm, e.g.@: how many characters per line, how often source location
2170 information should be reported. Right now, only the C++ front end can
2171 honor these options. However it is expected, in the near future, that
2172 the remaining front ends would be able to digest them correctly.
2175 @item -fmessage-length=@var{n}
2176 @opindex fmessage-length
2177 Try to format error messages so that they fit on lines of about @var{n}
2178 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2179 the front ends supported by GCC@. If @var{n} is zero, then no
2180 line-wrapping will be done; each error message will appear on a single
2183 @opindex fdiagnostics-show-location
2184 @item -fdiagnostics-show-location=once
2185 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2186 reporter to emit @emph{once} source location information; that is, in
2187 case the message is too long to fit on a single physical line and has to
2188 be wrapped, the source location won't be emitted (as prefix) again,
2189 over and over, in subsequent continuation lines. This is the default
2192 @item -fdiagnostics-show-location=every-line
2193 Only meaningful in line-wrapping mode. Instructs the diagnostic
2194 messages reporter to emit the same source location information (as
2195 prefix) for physical lines that result from the process of breaking
2196 a message which is too long to fit on a single line.
2198 @item -fdiagnostics-show-options
2199 @opindex fdiagnostics-show-options
2200 This option instructs the diagnostic machinery to add text to each
2201 diagnostic emitted, which indicates which command line option directly
2202 controls that diagnostic, when such an option is known to the
2203 diagnostic machinery.
2207 @node Warning Options
2208 @section Options to Request or Suppress Warnings
2209 @cindex options to control warnings
2210 @cindex warning messages
2211 @cindex messages, warning
2212 @cindex suppressing warnings
2214 Warnings are diagnostic messages that report constructions which
2215 are not inherently erroneous but which are risky or suggest there
2216 may have been an error.
2218 You can request many specific warnings with options beginning @samp{-W},
2219 for example @option{-Wimplicit} to request warnings on implicit
2220 declarations. Each of these specific warning options also has a
2221 negative form beginning @samp{-Wno-} to turn off warnings;
2222 for example, @option{-Wno-implicit}. This manual lists only one of the
2223 two forms, whichever is not the default.
2225 The following options control the amount and kinds of warnings produced
2226 by GCC; for further, language-specific options also refer to
2227 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2231 @cindex syntax checking
2233 @opindex fsyntax-only
2234 Check the code for syntax errors, but don't do anything beyond that.
2238 Issue all the warnings demanded by strict ISO C and ISO C++;
2239 reject all programs that use forbidden extensions, and some other
2240 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2241 version of the ISO C standard specified by any @option{-std} option used.
2243 Valid ISO C and ISO C++ programs should compile properly with or without
2244 this option (though a rare few will require @option{-ansi} or a
2245 @option{-std} option specifying the required version of ISO C)@. However,
2246 without this option, certain GNU extensions and traditional C and C++
2247 features are supported as well. With this option, they are rejected.
2249 @option{-pedantic} does not cause warning messages for use of the
2250 alternate keywords whose names begin and end with @samp{__}. Pedantic
2251 warnings are also disabled in the expression that follows
2252 @code{__extension__}. However, only system header files should use
2253 these escape routes; application programs should avoid them.
2254 @xref{Alternate Keywords}.
2256 Some users try to use @option{-pedantic} to check programs for strict ISO
2257 C conformance. They soon find that it does not do quite what they want:
2258 it finds some non-ISO practices, but not all---only those for which
2259 ISO C @emph{requires} a diagnostic, and some others for which
2260 diagnostics have been added.
2262 A feature to report any failure to conform to ISO C might be useful in
2263 some instances, but would require considerable additional work and would
2264 be quite different from @option{-pedantic}. We don't have plans to
2265 support such a feature in the near future.
2267 Where the standard specified with @option{-std} represents a GNU
2268 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2269 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2270 extended dialect is based. Warnings from @option{-pedantic} are given
2271 where they are required by the base standard. (It would not make sense
2272 for such warnings to be given only for features not in the specified GNU
2273 C dialect, since by definition the GNU dialects of C include all
2274 features the compiler supports with the given option, and there would be
2275 nothing to warn about.)
2277 @item -pedantic-errors
2278 @opindex pedantic-errors
2279 Like @option{-pedantic}, except that errors are produced rather than
2284 Inhibit all warning messages.
2288 Inhibit warning messages about the use of @samp{#import}.
2290 @item -Wchar-subscripts
2291 @opindex Wchar-subscripts
2292 Warn if an array subscript has type @code{char}. This is a common cause
2293 of error, as programmers often forget that this type is signed on some
2295 This warning is enabled by @option{-Wall}.
2299 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2300 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2301 This warning is enabled by @option{-Wall}.
2303 @item -Wfatal-errors
2304 @opindex Wfatal-errors
2305 This option causes the compiler to abort compilation on the first error
2306 occurred rather than trying to keep going and printing further error
2311 @opindex ffreestanding
2312 @opindex fno-builtin
2313 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2314 the arguments supplied have types appropriate to the format string
2315 specified, and that the conversions specified in the format string make
2316 sense. This includes standard functions, and others specified by format
2317 attributes (@pxref{Function Attributes}), in the @code{printf},
2318 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2319 not in the C standard) families (or other target-specific families).
2320 Which functions are checked without format attributes having been
2321 specified depends on the standard version selected, and such checks of
2322 functions without the attribute specified are disabled by
2323 @option{-ffreestanding} or @option{-fno-builtin}.
2325 The formats are checked against the format features supported by GNU
2326 libc version 2.2. These include all ISO C90 and C99 features, as well
2327 as features from the Single Unix Specification and some BSD and GNU
2328 extensions. Other library implementations may not support all these
2329 features; GCC does not support warning about features that go beyond a
2330 particular library's limitations. However, if @option{-pedantic} is used
2331 with @option{-Wformat}, warnings will be given about format features not
2332 in the selected standard version (but not for @code{strfmon} formats,
2333 since those are not in any version of the C standard). @xref{C Dialect
2334 Options,,Options Controlling C Dialect}.
2336 Since @option{-Wformat} also checks for null format arguments for
2337 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2339 @option{-Wformat} is included in @option{-Wall}. For more control over some
2340 aspects of format checking, the options @option{-Wformat-y2k},
2341 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2342 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2343 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2346 @opindex Wformat-y2k
2347 If @option{-Wformat} is specified, also warn about @code{strftime}
2348 formats which may yield only a two-digit year.
2350 @item -Wno-format-extra-args
2351 @opindex Wno-format-extra-args
2352 If @option{-Wformat} is specified, do not warn about excess arguments to a
2353 @code{printf} or @code{scanf} format function. The C standard specifies
2354 that such arguments are ignored.
2356 Where the unused arguments lie between used arguments that are
2357 specified with @samp{$} operand number specifications, normally
2358 warnings are still given, since the implementation could not know what
2359 type to pass to @code{va_arg} to skip the unused arguments. However,
2360 in the case of @code{scanf} formats, this option will suppress the
2361 warning if the unused arguments are all pointers, since the Single
2362 Unix Specification says that such unused arguments are allowed.
2364 @item -Wno-format-zero-length
2365 @opindex Wno-format-zero-length
2366 If @option{-Wformat} is specified, do not warn about zero-length formats.
2367 The C standard specifies that zero-length formats are allowed.
2369 @item -Wformat-nonliteral
2370 @opindex Wformat-nonliteral
2371 If @option{-Wformat} is specified, also warn if the format string is not a
2372 string literal and so cannot be checked, unless the format function
2373 takes its format arguments as a @code{va_list}.
2375 @item -Wformat-security
2376 @opindex Wformat-security
2377 If @option{-Wformat} is specified, also warn about uses of format
2378 functions that represent possible security problems. At present, this
2379 warns about calls to @code{printf} and @code{scanf} functions where the
2380 format string is not a string literal and there are no format arguments,
2381 as in @code{printf (foo);}. This may be a security hole if the format
2382 string came from untrusted input and contains @samp{%n}. (This is
2383 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2384 in future warnings may be added to @option{-Wformat-security} that are not
2385 included in @option{-Wformat-nonliteral}.)
2389 Enable @option{-Wformat} plus format checks not included in
2390 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2391 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2395 Warn about passing a null pointer for arguments marked as
2396 requiring a non-null value by the @code{nonnull} function attribute.
2398 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2399 can be disabled with the @option{-Wno-nonnull} option.
2401 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2403 Warn about uninitialized variables which are initialized with themselves.
2404 Note this option can only be used with the @option{-Wuninitialized} option,
2405 which in turn only works with @option{-O1} and above.
2407 For example, GCC will warn about @code{i} being uninitialized in the
2408 following snippet only when @option{-Winit-self} has been specified:
2419 @item -Wimplicit-int
2420 @opindex Wimplicit-int
2421 Warn when a declaration does not specify a type.
2422 This warning is enabled by @option{-Wall}.
2424 @item -Wimplicit-function-declaration
2425 @itemx -Werror-implicit-function-declaration
2426 @opindex Wimplicit-function-declaration
2427 @opindex Werror-implicit-function-declaration
2428 Give a warning (or error) whenever a function is used before being
2429 declared. The form @option{-Wno-error-implicit-function-declaration}
2431 This warning is enabled by @option{-Wall} (as a warning, not an error).
2435 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2436 This warning is enabled by @option{-Wall}.
2440 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2441 function with external linkage, returning int, taking either zero
2442 arguments, two, or three arguments of appropriate types.
2443 This warning is enabled by @option{-Wall}.
2445 @item -Wmissing-braces
2446 @opindex Wmissing-braces
2447 Warn if an aggregate or union initializer is not fully bracketed. In
2448 the following example, the initializer for @samp{a} is not fully
2449 bracketed, but that for @samp{b} is fully bracketed.
2452 int a[2][2] = @{ 0, 1, 2, 3 @};
2453 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2456 This warning is enabled by @option{-Wall}.
2458 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2459 @opindex Wmissing-include-dirs
2460 Warn if a user-supplied include directory does not exist.
2463 @opindex Wparentheses
2464 Warn if parentheses are omitted in certain contexts, such
2465 as when there is an assignment in a context where a truth value
2466 is expected, or when operators are nested whose precedence people
2467 often get confused about. Only the warning for an assignment used as
2468 a truth value is supported when compiling C++; the other warnings are
2469 only supported when compiling C@.
2471 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2472 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2473 interpretation from that of ordinary mathematical notation.
2475 Also warn about constructions where there may be confusion to which
2476 @code{if} statement an @code{else} branch belongs. Here is an example of
2491 In C, every @code{else} branch belongs to the innermost possible @code{if}
2492 statement, which in this example is @code{if (b)}. This is often not
2493 what the programmer expected, as illustrated in the above example by
2494 indentation the programmer chose. When there is the potential for this
2495 confusion, GCC will issue a warning when this flag is specified.
2496 To eliminate the warning, add explicit braces around the innermost
2497 @code{if} statement so there is no way the @code{else} could belong to
2498 the enclosing @code{if}. The resulting code would look like this:
2514 This warning is enabled by @option{-Wall}.
2516 @item -Wsequence-point
2517 @opindex Wsequence-point
2518 Warn about code that may have undefined semantics because of violations
2519 of sequence point rules in the C standard.
2521 The C standard defines the order in which expressions in a C program are
2522 evaluated in terms of @dfn{sequence points}, which represent a partial
2523 ordering between the execution of parts of the program: those executed
2524 before the sequence point, and those executed after it. These occur
2525 after the evaluation of a full expression (one which is not part of a
2526 larger expression), after the evaluation of the first operand of a
2527 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2528 function is called (but after the evaluation of its arguments and the
2529 expression denoting the called function), and in certain other places.
2530 Other than as expressed by the sequence point rules, the order of
2531 evaluation of subexpressions of an expression is not specified. All
2532 these rules describe only a partial order rather than a total order,
2533 since, for example, if two functions are called within one expression
2534 with no sequence point between them, the order in which the functions
2535 are called is not specified. However, the standards committee have
2536 ruled that function calls do not overlap.
2538 It is not specified when between sequence points modifications to the
2539 values of objects take effect. Programs whose behavior depends on this
2540 have undefined behavior; the C standard specifies that ``Between the
2541 previous and next sequence point an object shall have its stored value
2542 modified at most once by the evaluation of an expression. Furthermore,
2543 the prior value shall be read only to determine the value to be
2544 stored.''. If a program breaks these rules, the results on any
2545 particular implementation are entirely unpredictable.
2547 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2548 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2549 diagnosed by this option, and it may give an occasional false positive
2550 result, but in general it has been found fairly effective at detecting
2551 this sort of problem in programs.
2553 The present implementation of this option only works for C programs. A
2554 future implementation may also work for C++ programs.
2556 The C standard is worded confusingly, therefore there is some debate
2557 over the precise meaning of the sequence point rules in subtle cases.
2558 Links to discussions of the problem, including proposed formal
2559 definitions, may be found on the GCC readings page, at
2560 @w{@uref{http://gcc.gnu.org/readings.html}}.
2562 This warning is enabled by @option{-Wall}.
2565 @opindex Wreturn-type
2566 Warn whenever a function is defined with a return-type that defaults to
2567 @code{int}. Also warn about any @code{return} statement with no
2568 return-value in a function whose return-type is not @code{void}.
2570 For C, also warn if the return type of a function has a type qualifier
2571 such as @code{const}. Such a type qualifier has no effect, since the
2572 value returned by a function is not an lvalue. ISO C prohibits
2573 qualified @code{void} return types on function definitions, so such
2574 return types always receive a warning even without this option.
2576 For C++, a function without return type always produces a diagnostic
2577 message, even when @option{-Wno-return-type} is specified. The only
2578 exceptions are @samp{main} and functions defined in system headers.
2580 This warning is enabled by @option{-Wall}.
2584 Warn whenever a @code{switch} statement has an index of enumerated type
2585 and lacks a @code{case} for one or more of the named codes of that
2586 enumeration. (The presence of a @code{default} label prevents this
2587 warning.) @code{case} labels outside the enumeration range also
2588 provoke warnings when this option is used.
2589 This warning is enabled by @option{-Wall}.
2591 @item -Wswitch-default
2592 @opindex Wswitch-switch
2593 Warn whenever a @code{switch} statement does not have a @code{default}
2597 @opindex Wswitch-enum
2598 Warn whenever a @code{switch} statement has an index of enumerated type
2599 and lacks a @code{case} for one or more of the named codes of that
2600 enumeration. @code{case} labels outside the enumeration range also
2601 provoke warnings when this option is used.
2605 Warn if any trigraphs are encountered that might change the meaning of
2606 the program (trigraphs within comments are not warned about).
2607 This warning is enabled by @option{-Wall}.
2609 @item -Wunused-function
2610 @opindex Wunused-function
2611 Warn whenever a static function is declared but not defined or a
2612 non-inline static function is unused.
2613 This warning is enabled by @option{-Wall}.
2615 @item -Wunused-label
2616 @opindex Wunused-label
2617 Warn whenever a label is declared but not used.
2618 This warning is enabled by @option{-Wall}.
2620 To suppress this warning use the @samp{unused} attribute
2621 (@pxref{Variable Attributes}).
2623 @item -Wunused-parameter
2624 @opindex Wunused-parameter
2625 Warn whenever a function parameter is unused aside from its declaration.
2627 To suppress this warning use the @samp{unused} attribute
2628 (@pxref{Variable Attributes}).
2630 @item -Wunused-variable
2631 @opindex Wunused-variable
2632 Warn whenever a local variable or non-constant static variable is unused
2633 aside from its declaration
2634 This warning is enabled by @option{-Wall}.
2636 To suppress this warning use the @samp{unused} attribute
2637 (@pxref{Variable Attributes}).
2639 @item -Wunused-value
2640 @opindex Wunused-value
2641 Warn whenever a statement computes a result that is explicitly not used.
2642 This warning is enabled by @option{-Wall}.
2644 To suppress this warning cast the expression to @samp{void}.
2648 All the above @option{-Wunused} options combined.
2650 In order to get a warning about an unused function parameter, you must
2651 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2652 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2654 @item -Wuninitialized
2655 @opindex Wuninitialized
2656 Warn if an automatic variable is used without first being initialized or
2657 if a variable may be clobbered by a @code{setjmp} call.
2659 These warnings are possible only in optimizing compilation,
2660 because they require data flow information that is computed only
2661 when optimizing. If you don't specify @option{-O}, you simply won't
2664 If you want to warn about code which uses the uninitialized value of the
2665 variable in its own initializer, use the @option{-Winit-self} option.
2667 These warnings occur for individual uninitialized or clobbered
2668 elements of structure, union or array variables as well as for
2669 variables which are uninitialized or clobbered as a whole. They do
2670 not occur for variables or elements declared @code{volatile}. Because
2671 these warnings depend on optimization, the exact variables or elements
2672 for which there are warnings will depend on the precise optimization
2673 options and version of GCC used.
2675 Note that there may be no warning about a variable that is used only
2676 to compute a value that itself is never used, because such
2677 computations may be deleted by data flow analysis before the warnings
2680 These warnings are made optional because GCC is not smart
2681 enough to see all the reasons why the code might be correct
2682 despite appearing to have an error. Here is one example of how
2703 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2704 always initialized, but GCC doesn't know this. Here is
2705 another common case:
2710 if (change_y) save_y = y, y = new_y;
2712 if (change_y) y = save_y;
2717 This has no bug because @code{save_y} is used only if it is set.
2719 @cindex @code{longjmp} warnings
2720 This option also warns when a non-volatile automatic variable might be
2721 changed by a call to @code{longjmp}. These warnings as well are possible
2722 only in optimizing compilation.
2724 The compiler sees only the calls to @code{setjmp}. It cannot know
2725 where @code{longjmp} will be called; in fact, a signal handler could
2726 call it at any point in the code. As a result, you may get a warning
2727 even when there is in fact no problem because @code{longjmp} cannot
2728 in fact be called at the place which would cause a problem.
2730 Some spurious warnings can be avoided if you declare all the functions
2731 you use that never return as @code{noreturn}. @xref{Function
2734 This warning is enabled by @option{-Wall}.
2736 @item -Wunknown-pragmas
2737 @opindex Wunknown-pragmas
2738 @cindex warning for unknown pragmas
2739 @cindex unknown pragmas, warning
2740 @cindex pragmas, warning of unknown
2741 Warn when a #pragma directive is encountered which is not understood by
2742 GCC@. If this command line option is used, warnings will even be issued
2743 for unknown pragmas in system header files. This is not the case if
2744 the warnings were only enabled by the @option{-Wall} command line option.
2747 @opindex Wno-pragmas
2749 Do not warn about misuses of pragmas, such as incorrect parameters,
2750 invalid syntax, or conflicts between pragmas. See also
2751 @samp{-Wunknown-pragmas}.
2753 @item -Wstrict-aliasing
2754 @opindex Wstrict-aliasing
2755 This option is only active when @option{-fstrict-aliasing} is active.
2756 It warns about code which might break the strict aliasing rules that the
2757 compiler is using for optimization. The warning does not catch all
2758 cases, but does attempt to catch the more common pitfalls. It is
2759 included in @option{-Wall}.
2761 @item -Wstrict-aliasing=2
2762 @opindex Wstrict-aliasing=2
2763 This option is only active when @option{-fstrict-aliasing} is active.
2764 It warns about code which might break the strict aliasing rules that the
2765 compiler is using for optimization. This warning catches more cases than
2766 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2767 cases that are safe.
2771 All of the above @samp{-W} options combined. This enables all the
2772 warnings about constructions that some users consider questionable, and
2773 that are easy to avoid (or modify to prevent the warning), even in
2774 conjunction with macros. This also enables some language-specific
2775 warnings described in @ref{C++ Dialect Options} and
2776 @ref{Objective-C and Objective-C++ Dialect Options}.
2779 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2780 Some of them warn about constructions that users generally do not
2781 consider questionable, but which occasionally you might wish to check
2782 for; others warn about constructions that are necessary or hard to avoid
2783 in some cases, and there is no simple way to modify the code to suppress
2790 (This option used to be called @option{-W}. The older name is still
2791 supported, but the newer name is more descriptive.) Print extra warning
2792 messages for these events:
2796 A function can return either with or without a value. (Falling
2797 off the end of the function body is considered returning without
2798 a value.) For example, this function would evoke such a
2812 An expression-statement or the left-hand side of a comma expression
2813 contains no side effects.
2814 To suppress the warning, cast the unused expression to void.
2815 For example, an expression such as @samp{x[i,j]} will cause a warning,
2816 but @samp{x[(void)i,j]} will not.
2819 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2822 Storage-class specifiers like @code{static} are not the first things in
2823 a declaration. According to the C Standard, this usage is obsolescent.
2826 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2830 A comparison between signed and unsigned values could produce an
2831 incorrect result when the signed value is converted to unsigned.
2832 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2835 An aggregate has an initializer which does not initialize all members.
2836 This warning can be independently controlled by
2837 @option{-Wmissing-field-initializers}.
2840 A function parameter is declared without a type specifier in K&R-style
2848 An empty body occurs in an @samp{if} or @samp{else} statement.
2851 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2852 @samp{>}, or @samp{>=}.
2855 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2858 Any of several floating-point events that often indicate errors, such as
2859 overflow, underflow, loss of precision, etc.
2861 @item @r{(C++ only)}
2862 An enumerator and a non-enumerator both appear in a conditional expression.
2864 @item @r{(C++ only)}
2865 A non-static reference or non-static @samp{const} member appears in a
2866 class without constructors.
2868 @item @r{(C++ only)}
2869 Ambiguous virtual bases.
2871 @item @r{(C++ only)}
2872 Subscripting an array which has been declared @samp{register}.
2874 @item @r{(C++ only)}
2875 Taking the address of a variable which has been declared @samp{register}.
2877 @item @r{(C++ only)}
2878 A base class is not initialized in a derived class' copy constructor.
2881 @item -Wno-div-by-zero
2882 @opindex Wno-div-by-zero
2883 @opindex Wdiv-by-zero
2884 Do not warn about compile-time integer division by zero. Floating point
2885 division by zero is not warned about, as it can be a legitimate way of
2886 obtaining infinities and NaNs.
2888 @item -Wsystem-headers
2889 @opindex Wsystem-headers
2890 @cindex warnings from system headers
2891 @cindex system headers, warnings from
2892 Print warning messages for constructs found in system header files.
2893 Warnings from system headers are normally suppressed, on the assumption
2894 that they usually do not indicate real problems and would only make the
2895 compiler output harder to read. Using this command line option tells
2896 GCC to emit warnings from system headers as if they occurred in user
2897 code. However, note that using @option{-Wall} in conjunction with this
2898 option will @emph{not} warn about unknown pragmas in system
2899 headers---for that, @option{-Wunknown-pragmas} must also be used.
2902 @opindex Wfloat-equal
2903 Warn if floating point values are used in equality comparisons.
2905 The idea behind this is that sometimes it is convenient (for the
2906 programmer) to consider floating-point values as approximations to
2907 infinitely precise real numbers. If you are doing this, then you need
2908 to compute (by analyzing the code, or in some other way) the maximum or
2909 likely maximum error that the computation introduces, and allow for it
2910 when performing comparisons (and when producing output, but that's a
2911 different problem). In particular, instead of testing for equality, you
2912 would check to see whether the two values have ranges that overlap; and
2913 this is done with the relational operators, so equality comparisons are
2916 @item -Wtraditional @r{(C only)}
2917 @opindex Wtraditional
2918 Warn about certain constructs that behave differently in traditional and
2919 ISO C@. Also warn about ISO C constructs that have no traditional C
2920 equivalent, and/or problematic constructs which should be avoided.
2924 Macro parameters that appear within string literals in the macro body.
2925 In traditional C macro replacement takes place within string literals,
2926 but does not in ISO C@.
2929 In traditional C, some preprocessor directives did not exist.
2930 Traditional preprocessors would only consider a line to be a directive
2931 if the @samp{#} appeared in column 1 on the line. Therefore
2932 @option{-Wtraditional} warns about directives that traditional C
2933 understands but would ignore because the @samp{#} does not appear as the
2934 first character on the line. It also suggests you hide directives like
2935 @samp{#pragma} not understood by traditional C by indenting them. Some
2936 traditional implementations would not recognize @samp{#elif}, so it
2937 suggests avoiding it altogether.
2940 A function-like macro that appears without arguments.
2943 The unary plus operator.
2946 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2947 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2948 constants.) Note, these suffixes appear in macros defined in the system
2949 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2950 Use of these macros in user code might normally lead to spurious
2951 warnings, however GCC's integrated preprocessor has enough context to
2952 avoid warning in these cases.
2955 A function declared external in one block and then used after the end of
2959 A @code{switch} statement has an operand of type @code{long}.
2962 A non-@code{static} function declaration follows a @code{static} one.
2963 This construct is not accepted by some traditional C compilers.
2966 The ISO type of an integer constant has a different width or
2967 signedness from its traditional type. This warning is only issued if
2968 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2969 typically represent bit patterns, are not warned about.
2972 Usage of ISO string concatenation is detected.
2975 Initialization of automatic aggregates.
2978 Identifier conflicts with labels. Traditional C lacks a separate
2979 namespace for labels.
2982 Initialization of unions. If the initializer is zero, the warning is
2983 omitted. This is done under the assumption that the zero initializer in
2984 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2985 initializer warnings and relies on default initialization to zero in the
2989 Conversions by prototypes between fixed/floating point values and vice
2990 versa. The absence of these prototypes when compiling with traditional
2991 C would cause serious problems. This is a subset of the possible
2992 conversion warnings, for the full set use @option{-Wconversion}.
2995 Use of ISO C style function definitions. This warning intentionally is
2996 @emph{not} issued for prototype declarations or variadic functions
2997 because these ISO C features will appear in your code when using
2998 libiberty's traditional C compatibility macros, @code{PARAMS} and
2999 @code{VPARAMS}. This warning is also bypassed for nested functions
3000 because that feature is already a GCC extension and thus not relevant to
3001 traditional C compatibility.
3004 @item -Wdeclaration-after-statement @r{(C only)}
3005 @opindex Wdeclaration-after-statement
3006 Warn when a declaration is found after a statement in a block. This
3007 construct, known from C++, was introduced with ISO C99 and is by default
3008 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3009 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3013 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3015 @item -Wno-endif-labels
3016 @opindex Wno-endif-labels
3017 @opindex Wendif-labels
3018 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3022 Warn whenever a local variable shadows another local variable, parameter or
3023 global variable or whenever a built-in function is shadowed.
3025 @item -Wlarger-than-@var{len}
3026 @opindex Wlarger-than
3027 Warn whenever an object of larger than @var{len} bytes is defined.
3029 @item -Wunsafe-loop-optimizations
3030 @opindex Wunsafe-loop-optimizations
3031 Warn if the loop cannot be optimized because the compiler could not
3032 assume anything on the bounds of the loop indices. With
3033 @option{-funsafe-loop-optimizations} warn if the compiler made
3036 @item -Wpointer-arith
3037 @opindex Wpointer-arith
3038 Warn about anything that depends on the ``size of'' a function type or
3039 of @code{void}. GNU C assigns these types a size of 1, for
3040 convenience in calculations with @code{void *} pointers and pointers
3043 @item -Wbad-function-cast @r{(C only)}
3044 @opindex Wbad-function-cast
3045 Warn whenever a function call is cast to a non-matching type.
3046 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3049 Warn about ISO C constructs that are outside of the common subset of
3050 ISO C and ISO C++, e.g.@: request for implicit conversion from
3051 @code{void *} to a pointer to non-@code{void} type.
3055 Warn whenever a pointer is cast so as to remove a type qualifier from
3056 the target type. For example, warn if a @code{const char *} is cast
3057 to an ordinary @code{char *}.
3060 @opindex Wcast-align
3061 Warn whenever a pointer is cast such that the required alignment of the
3062 target is increased. For example, warn if a @code{char *} is cast to
3063 an @code{int *} on machines where integers can only be accessed at
3064 two- or four-byte boundaries.
3066 @item -Wwrite-strings
3067 @opindex Wwrite-strings
3068 When compiling C, give string constants the type @code{const
3069 char[@var{length}]} so that
3070 copying the address of one into a non-@code{const} @code{char *}
3071 pointer will get a warning; when compiling C++, warn about the
3072 deprecated conversion from string constants to @code{char *}.
3073 These warnings will help you find at
3074 compile time code that can try to write into a string constant, but
3075 only if you have been very careful about using @code{const} in
3076 declarations and prototypes. Otherwise, it will just be a nuisance;
3077 this is why we did not make @option{-Wall} request these warnings.
3080 @opindex Wconversion
3081 Warn if a prototype causes a type conversion that is different from what
3082 would happen to the same argument in the absence of a prototype. This
3083 includes conversions of fixed point to floating and vice versa, and
3084 conversions changing the width or signedness of a fixed point argument
3085 except when the same as the default promotion.
3087 Also, warn if a negative integer constant expression is implicitly
3088 converted to an unsigned type. For example, warn about the assignment
3089 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3090 casts like @code{(unsigned) -1}.
3092 @item -Wsign-compare
3093 @opindex Wsign-compare
3094 @cindex warning for comparison of signed and unsigned values
3095 @cindex comparison of signed and unsigned values, warning
3096 @cindex signed and unsigned values, comparison warning
3097 Warn when a comparison between signed and unsigned values could produce
3098 an incorrect result when the signed value is converted to unsigned.
3099 This warning is also enabled by @option{-Wextra}; to get the other warnings
3100 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3102 @item -Waggregate-return
3103 @opindex Waggregate-return
3104 Warn if any functions that return structures or unions are defined or
3105 called. (In languages where you can return an array, this also elicits
3108 @item -Wno-attributes
3109 @opindex Wno-attributes
3110 @opindex Wattributes
3111 Do not warn if an unexpected @code{__attribute__} is used, such as
3112 unrecognized attributes, function attributes applied to variables,
3113 etc. This will not stop errors for incorrect use of supported
3116 @item -Wstrict-prototypes @r{(C only)}
3117 @opindex Wstrict-prototypes
3118 Warn if a function is declared or defined without specifying the
3119 argument types. (An old-style function definition is permitted without
3120 a warning if preceded by a declaration which specifies the argument
3123 @item -Wold-style-definition @r{(C only)}
3124 @opindex Wold-style-definition
3125 Warn if an old-style function definition is used. A warning is given
3126 even if there is a previous prototype.
3128 @item -Wmissing-prototypes @r{(C only)}
3129 @opindex Wmissing-prototypes
3130 Warn if a global function is defined without a previous prototype
3131 declaration. This warning is issued even if the definition itself
3132 provides a prototype. The aim is to detect global functions that fail
3133 to be declared in header files.
3135 @item -Wmissing-declarations @r{(C only)}
3136 @opindex Wmissing-declarations
3137 Warn if a global function is defined without a previous declaration.
3138 Do so even if the definition itself provides a prototype.
3139 Use this option to detect global functions that are not declared in
3142 @item -Wmissing-field-initializers
3143 @opindex Wmissing-field-initializers
3146 Warn if a structure's initializer has some fields missing. For
3147 example, the following code would cause such a warning, because
3148 @code{x.h} is implicitly zero:
3151 struct s @{ int f, g, h; @};
3152 struct s x = @{ 3, 4 @};
3155 This option does not warn about designated initializers, so the following
3156 modification would not trigger a warning:
3159 struct s @{ int f, g, h; @};
3160 struct s x = @{ .f = 3, .g = 4 @};
3163 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3164 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3166 @item -Wmissing-noreturn
3167 @opindex Wmissing-noreturn
3168 Warn about functions which might be candidates for attribute @code{noreturn}.
3169 Note these are only possible candidates, not absolute ones. Care should
3170 be taken to manually verify functions actually do not ever return before
3171 adding the @code{noreturn} attribute, otherwise subtle code generation
3172 bugs could be introduced. You will not get a warning for @code{main} in
3173 hosted C environments.
3175 @item -Wmissing-format-attribute
3176 @opindex Wmissing-format-attribute
3178 Warn about function pointers which might be candidates for @code{format}
3179 attributes. Note these are only possible candidates, not absolute ones.
3180 GCC will guess that function pointers with @code{format} attributes that
3181 are used in assignment, initialization, parameter passing or return
3182 statements should have a corresponding @code{format} attribute in the
3183 resulting type. I.e.@: the left-hand side of the assignment or
3184 initialization, the type of the parameter variable, or the return type
3185 of the containing function respectively should also have a @code{format}
3186 attribute to avoid the warning.
3188 GCC will also warn about function definitions which might be
3189 candidates for @code{format} attributes. Again, these are only
3190 possible candidates. GCC will guess that @code{format} attributes
3191 might be appropriate for any function that calls a function like
3192 @code{vprintf} or @code{vscanf}, but this might not always be the
3193 case, and some functions for which @code{format} attributes are
3194 appropriate may not be detected.
3196 @item -Wno-multichar
3197 @opindex Wno-multichar
3199 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3200 Usually they indicate a typo in the user's code, as they have
3201 implementation-defined values, and should not be used in portable code.
3203 @item -Wnormalized=<none|id|nfc|nfkc>
3204 @opindex Wnormalized
3207 @cindex character set, input normalization
3208 In ISO C and ISO C++, two identifiers are different if they are
3209 different sequences of characters. However, sometimes when characters
3210 outside the basic ASCII character set are used, you can have two
3211 different character sequences that look the same. To avoid confusion,
3212 the ISO 10646 standard sets out some @dfn{normalization rules} which
3213 when applied ensure that two sequences that look the same are turned into
3214 the same sequence. GCC can warn you if you are using identifiers which
3215 have not been normalized; this option controls that warning.
3217 There are four levels of warning that GCC supports. The default is
3218 @option{-Wnormalized=nfc}, which warns about any identifier which is
3219 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3220 recommended form for most uses.
3222 Unfortunately, there are some characters which ISO C and ISO C++ allow
3223 in identifiers that when turned into NFC aren't allowable as
3224 identifiers. That is, there's no way to use these symbols in portable
3225 ISO C or C++ and have all your identifiers in NFC.
3226 @option{-Wnormalized=id} suppresses the warning for these characters.
3227 It is hoped that future versions of the standards involved will correct
3228 this, which is why this option is not the default.
3230 You can switch the warning off for all characters by writing
3231 @option{-Wnormalized=none}. You would only want to do this if you
3232 were using some other normalization scheme (like ``D''), because
3233 otherwise you can easily create bugs that are literally impossible to see.
3235 Some characters in ISO 10646 have distinct meanings but look identical
3236 in some fonts or display methodologies, especially once formatting has
3237 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3238 LETTER N'', will display just like a regular @code{n} which has been
3239 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3240 normalisation scheme to convert all these into a standard form as
3241 well, and GCC will warn if your code is not in NFKC if you use
3242 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3243 about every identifier that contains the letter O because it might be
3244 confused with the digit 0, and so is not the default, but may be
3245 useful as a local coding convention if the programming environment is
3246 unable to be fixed to display these characters distinctly.
3248 @item -Wno-deprecated-declarations
3249 @opindex Wno-deprecated-declarations
3250 Do not warn about uses of functions, variables, and types marked as
3251 deprecated by using the @code{deprecated} attribute.
3252 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3253 @pxref{Type Attributes}.)
3257 Warn if a structure is given the packed attribute, but the packed
3258 attribute has no effect on the layout or size of the structure.
3259 Such structures may be mis-aligned for little benefit. For
3260 instance, in this code, the variable @code{f.x} in @code{struct bar}
3261 will be misaligned even though @code{struct bar} does not itself
3262 have the packed attribute:
3269 @} __attribute__((packed));
3279 Warn if padding is included in a structure, either to align an element
3280 of the structure or to align the whole structure. Sometimes when this
3281 happens it is possible to rearrange the fields of the structure to
3282 reduce the padding and so make the structure smaller.
3284 @item -Wredundant-decls
3285 @opindex Wredundant-decls
3286 Warn if anything is declared more than once in the same scope, even in
3287 cases where multiple declaration is valid and changes nothing.
3289 @item -Wnested-externs @r{(C only)}
3290 @opindex Wnested-externs
3291 Warn if an @code{extern} declaration is encountered within a function.
3293 @item -Wunreachable-code
3294 @opindex Wunreachable-code
3295 Warn if the compiler detects that code will never be executed.
3297 This option is intended to warn when the compiler detects that at
3298 least a whole line of source code will never be executed, because
3299 some condition is never satisfied or because it is after a
3300 procedure that never returns.
3302 It is possible for this option to produce a warning even though there
3303 are circumstances under which part of the affected line can be executed,
3304 so care should be taken when removing apparently-unreachable code.
3306 For instance, when a function is inlined, a warning may mean that the
3307 line is unreachable in only one inlined copy of the function.
3309 This option is not made part of @option{-Wall} because in a debugging
3310 version of a program there is often substantial code which checks
3311 correct functioning of the program and is, hopefully, unreachable
3312 because the program does work. Another common use of unreachable
3313 code is to provide behavior which is selectable at compile-time.
3317 Warn if a function can not be inlined and it was declared as inline.
3318 Even with this option, the compiler will not warn about failures to
3319 inline functions declared in system headers.
3321 The compiler uses a variety of heuristics to determine whether or not
3322 to inline a function. For example, the compiler takes into account
3323 the size of the function being inlined and the amount of inlining
3324 that has already been done in the current function. Therefore,
3325 seemingly insignificant changes in the source program can cause the
3326 warnings produced by @option{-Winline} to appear or disappear.
3328 @item -Wno-invalid-offsetof @r{(C++ only)}
3329 @opindex Wno-invalid-offsetof
3330 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3331 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3332 to a non-POD type is undefined. In existing C++ implementations,
3333 however, @samp{offsetof} typically gives meaningful results even when
3334 applied to certain kinds of non-POD types. (Such as a simple
3335 @samp{struct} that fails to be a POD type only by virtue of having a
3336 constructor.) This flag is for users who are aware that they are
3337 writing nonportable code and who have deliberately chosen to ignore the
3340 The restrictions on @samp{offsetof} may be relaxed in a future version
3341 of the C++ standard.
3343 @item -Wno-int-to-pointer-cast @r{(C only)}
3344 @opindex Wno-int-to-pointer-cast
3345 Suppress warnings from casts to pointer type of an integer of a
3348 @item -Wno-pointer-to-int-cast @r{(C only)}
3349 @opindex Wno-pointer-to-int-cast
3350 Suppress warnings from casts from a pointer to an integer type of a
3354 @opindex Winvalid-pch
3355 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3356 the search path but can't be used.
3360 @opindex Wno-long-long
3361 Warn if @samp{long long} type is used. This is default. To inhibit
3362 the warning messages, use @option{-Wno-long-long}. Flags
3363 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3364 only when @option{-pedantic} flag is used.
3366 @item -Wvariadic-macros
3367 @opindex Wvariadic-macros
3368 @opindex Wno-variadic-macros
3369 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3370 alternate syntax when in pedantic ISO C99 mode. This is default.
3371 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3373 @item -Wdisabled-optimization
3374 @opindex Wdisabled-optimization
3375 Warn if a requested optimization pass is disabled. This warning does
3376 not generally indicate that there is anything wrong with your code; it
3377 merely indicates that GCC's optimizers were unable to handle the code
3378 effectively. Often, the problem is that your code is too big or too
3379 complex; GCC will refuse to optimize programs when the optimization
3380 itself is likely to take inordinate amounts of time.
3382 @item -Wno-pointer-sign
3383 @opindex Wno-pointer-sign
3384 Don't warn for pointer argument passing or assignment with different signedness.
3385 Only useful in the negative form since this warning is enabled by default.
3386 This option is only supported for C and Objective-C@.
3390 Make all warnings into errors.
3392 @item -Wstack-protector
3393 This option is only active when @option{-fstack-protector} is active. It
3394 warns about functions that will not be protected against stack smashing.
3398 @node Debugging Options
3399 @section Options for Debugging Your Program or GCC
3400 @cindex options, debugging
3401 @cindex debugging information options
3403 GCC has various special options that are used for debugging
3404 either your program or GCC:
3409 Produce debugging information in the operating system's native format
3410 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3413 On most systems that use stabs format, @option{-g} enables use of extra
3414 debugging information that only GDB can use; this extra information
3415 makes debugging work better in GDB but will probably make other debuggers
3417 refuse to read the program. If you want to control for certain whether
3418 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3419 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3421 GCC allows you to use @option{-g} with
3422 @option{-O}. The shortcuts taken by optimized code may occasionally
3423 produce surprising results: some variables you declared may not exist
3424 at all; flow of control may briefly move where you did not expect it;
3425 some statements may not be executed because they compute constant
3426 results or their values were already at hand; some statements may
3427 execute in different places because they were moved out of loops.
3429 Nevertheless it proves possible to debug optimized output. This makes
3430 it reasonable to use the optimizer for programs that might have bugs.
3432 The following options are useful when GCC is generated with the
3433 capability for more than one debugging format.
3437 Produce debugging information for use by GDB@. This means to use the
3438 most expressive format available (DWARF 2, stabs, or the native format
3439 if neither of those are supported), including GDB extensions if at all
3444 Produce debugging information in stabs format (if that is supported),
3445 without GDB extensions. This is the format used by DBX on most BSD
3446 systems. On MIPS, Alpha and System V Release 4 systems this option
3447 produces stabs debugging output which is not understood by DBX or SDB@.
3448 On System V Release 4 systems this option requires the GNU assembler.
3450 @item -feliminate-unused-debug-symbols
3451 @opindex feliminate-unused-debug-symbols
3452 Produce debugging information in stabs format (if that is supported),
3453 for only symbols that are actually used.
3457 Produce debugging information in stabs format (if that is supported),
3458 using GNU extensions understood only by the GNU debugger (GDB)@. The
3459 use of these extensions is likely to make other debuggers crash or
3460 refuse to read the program.
3464 Produce debugging information in COFF format (if that is supported).
3465 This is the format used by SDB on most System V systems prior to
3470 Produce debugging information in XCOFF format (if that is supported).
3471 This is the format used by the DBX debugger on IBM RS/6000 systems.
3475 Produce debugging information in XCOFF format (if that is supported),
3476 using GNU extensions understood only by the GNU debugger (GDB)@. The
3477 use of these extensions is likely to make other debuggers crash or
3478 refuse to read the program, and may cause assemblers other than the GNU
3479 assembler (GAS) to fail with an error.
3483 Produce debugging information in DWARF version 2 format (if that is
3484 supported). This is the format used by DBX on IRIX 6. With this
3485 option, GCC uses features of DWARF version 3 when they are useful;
3486 version 3 is upward compatible with version 2, but may still cause
3487 problems for older debuggers.
3491 Produce debugging information in VMS debug format (if that is
3492 supported). This is the format used by DEBUG on VMS systems.
3495 @itemx -ggdb@var{level}
3496 @itemx -gstabs@var{level}
3497 @itemx -gcoff@var{level}
3498 @itemx -gxcoff@var{level}
3499 @itemx -gvms@var{level}
3500 Request debugging information and also use @var{level} to specify how
3501 much information. The default level is 2.
3503 Level 1 produces minimal information, enough for making backtraces in
3504 parts of the program that you don't plan to debug. This includes
3505 descriptions of functions and external variables, but no information
3506 about local variables and no line numbers.
3508 Level 3 includes extra information, such as all the macro definitions
3509 present in the program. Some debuggers support macro expansion when
3510 you use @option{-g3}.
3512 @option{-gdwarf-2} does not accept a concatenated debug level, because
3513 GCC used to support an option @option{-gdwarf} that meant to generate
3514 debug information in version 1 of the DWARF format (which is very
3515 different from version 2), and it would have been too confusing. That
3516 debug format is long obsolete, but the option cannot be changed now.
3517 Instead use an additional @option{-g@var{level}} option to change the
3518 debug level for DWARF2.
3520 @item -feliminate-dwarf2-dups
3521 @opindex feliminate-dwarf2-dups
3522 Compress DWARF2 debugging information by eliminating duplicated
3523 information about each symbol. This option only makes sense when
3524 generating DWARF2 debugging information with @option{-gdwarf-2}.
3526 @cindex @command{prof}
3529 Generate extra code to write profile information suitable for the
3530 analysis program @command{prof}. You must use this option when compiling
3531 the source files you want data about, and you must also use it when
3534 @cindex @command{gprof}
3537 Generate extra code to write profile information suitable for the
3538 analysis program @command{gprof}. You must use this option when compiling
3539 the source files you want data about, and you must also use it when
3544 Makes the compiler print out each function name as it is compiled, and
3545 print some statistics about each pass when it finishes.
3548 @opindex ftime-report
3549 Makes the compiler print some statistics about the time consumed by each
3550 pass when it finishes.
3553 @opindex fmem-report
3554 Makes the compiler print some statistics about permanent memory
3555 allocation when it finishes.
3557 @item -fprofile-arcs
3558 @opindex fprofile-arcs
3559 Add code so that program flow @dfn{arcs} are instrumented. During
3560 execution the program records how many times each branch and call is
3561 executed and how many times it is taken or returns. When the compiled
3562 program exits it saves this data to a file called
3563 @file{@var{auxname}.gcda} for each source file. The data may be used for
3564 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3565 test coverage analysis (@option{-ftest-coverage}). Each object file's
3566 @var{auxname} is generated from the name of the output file, if
3567 explicitly specified and it is not the final executable, otherwise it is
3568 the basename of the source file. In both cases any suffix is removed
3569 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3570 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3571 @xref{Cross-profiling}.
3573 @cindex @command{gcov}
3577 This option is used to compile and link code instrumented for coverage
3578 analysis. The option is a synonym for @option{-fprofile-arcs}
3579 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3580 linking). See the documentation for those options for more details.
3585 Compile the source files with @option{-fprofile-arcs} plus optimization
3586 and code generation options. For test coverage analysis, use the
3587 additional @option{-ftest-coverage} option. You do not need to profile
3588 every source file in a program.
3591 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3592 (the latter implies the former).
3595 Run the program on a representative workload to generate the arc profile
3596 information. This may be repeated any number of times. You can run
3597 concurrent instances of your program, and provided that the file system
3598 supports locking, the data files will be correctly updated. Also
3599 @code{fork} calls are detected and correctly handled (double counting
3603 For profile-directed optimizations, compile the source files again with
3604 the same optimization and code generation options plus
3605 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3606 Control Optimization}).
3609 For test coverage analysis, use @command{gcov} to produce human readable
3610 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3611 @command{gcov} documentation for further information.
3615 With @option{-fprofile-arcs}, for each function of your program GCC
3616 creates a program flow graph, then finds a spanning tree for the graph.
3617 Only arcs that are not on the spanning tree have to be instrumented: the
3618 compiler adds code to count the number of times that these arcs are
3619 executed. When an arc is the only exit or only entrance to a block, the
3620 instrumentation code can be added to the block; otherwise, a new basic
3621 block must be created to hold the instrumentation code.
3624 @item -ftest-coverage
3625 @opindex ftest-coverage
3626 Produce a notes file that the @command{gcov} code-coverage utility
3627 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3628 show program coverage. Each source file's note file is called
3629 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3630 above for a description of @var{auxname} and instructions on how to
3631 generate test coverage data. Coverage data will match the source files
3632 more closely, if you do not optimize.
3634 @item -d@var{letters}
3635 @item -fdump-rtl-@var{pass}
3637 Says to make debugging dumps during compilation at times specified by
3638 @var{letters}. This is used for debugging the RTL-based passes of the
3639 compiler. The file names for most of the dumps are made by appending a
3640 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3641 from the name of the output file, if explicitly specified and it is not
3642 an executable, otherwise it is the basename of the source file.
3644 Most debug dumps can be enabled either passing a letter to the @option{-d}
3645 option, or with a long @option{-fdump-rtl} switch; here are the possible
3646 letters for use in @var{letters} and @var{pass}, and their meanings:
3651 Annotate the assembler output with miscellaneous debugging information.
3654 @itemx -fdump-rtl-bp
3656 @opindex fdump-rtl-bp
3657 Dump after computing branch probabilities, to @file{@var{file}.09.bp}.
3660 @itemx -fdump-rtl-bbro
3662 @opindex fdump-rtl-bbro
3663 Dump after block reordering, to @file{@var{file}.30.bbro}.
3666 @itemx -fdump-rtl-combine
3668 @opindex fdump-rtl-combine
3669 Dump after instruction combination, to the file @file{@var{file}.17.combine}.
3672 @itemx -fdump-rtl-ce1
3673 @itemx -fdump-rtl-ce2
3675 @opindex fdump-rtl-ce1
3676 @opindex fdump-rtl-ce2
3677 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3678 first if conversion, to the file @file{@var{file}.11.ce1}. @option{-dC}
3679 and @option{-fdump-rtl-ce2} enable dumping after the second if
3680 conversion, to the file @file{@var{file}.18.ce2}.
3683 @itemx -fdump-rtl-btl
3684 @itemx -fdump-rtl-dbr
3686 @opindex fdump-rtl-btl
3687 @opindex fdump-rtl-dbr
3688 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3689 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3690 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3691 scheduling, to @file{@var{file}.36.dbr}.
3695 Dump all macro definitions, at the end of preprocessing, in addition to
3699 @itemx -fdump-rtl-ce3
3701 @opindex fdump-rtl-ce3
3702 Dump after the third if conversion, to @file{@var{file}.28.ce3}.
3705 @itemx -fdump-rtl-cfg
3706 @itemx -fdump-rtl-life
3708 @opindex fdump-rtl-cfg
3709 @opindex fdump-rtl-life
3710 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3711 and data flow analysis, to @file{@var{file}.08.cfg}. @option{-df}
3712 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3713 to @file{@var{file}.16.life}.
3716 @itemx -fdump-rtl-greg
3718 @opindex fdump-rtl-greg
3719 Dump after global register allocation, to @file{@var{file}.23.greg}.
3722 @itemx -fdump-rtl-gcse
3723 @itemx -fdump-rtl-bypass
3725 @opindex fdump-rtl-gcse
3726 @opindex fdump-rtl-bypass
3727 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3728 @file{@var{file}.05.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3729 enable dumping after jump bypassing and control flow optimizations, to
3730 @file{@var{file}.07.bypass}.
3733 @itemx -fdump-rtl-eh
3735 @opindex fdump-rtl-eh
3736 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3739 @itemx -fdump-rtl-sibling
3741 @opindex fdump-rtl-sibling
3742 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3745 @itemx -fdump-rtl-jump
3747 @opindex fdump-rtl-jump
3748 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3751 @itemx -fdump-rtl-stack
3753 @opindex fdump-rtl-stack
3754 Dump after conversion from registers to stack, to @file{@var{file}.33.stack}.
3757 @itemx -fdump-rtl-lreg
3759 @opindex fdump-rtl-lreg
3760 Dump after local register allocation, to @file{@var{file}.22.lreg}.
3763 @itemx -fdump-rtl-loop
3764 @itemx -fdump-rtl-loop2
3766 @opindex fdump-rtl-loop
3767 @opindex fdump-rtl-loop2
3768 @option{-dL} and @option{-fdump-rtl-loop} enable dumping after the first
3769 loop optimization pass, to @file{@var{file}.06.loop}. @option{-dL} and
3770 @option{-fdump-rtl-loop2} enable dumping after the second pass, to
3771 @file{@var{file}.13.loop2}.
3774 @itemx -fdump-rtl-sms
3776 @opindex fdump-rtl-sms
3777 Dump after modulo scheduling, to @file{@var{file}.20.sms}.
3780 @itemx -fdump-rtl-mach
3782 @opindex fdump-rtl-mach
3783 Dump after performing the machine dependent reorganization pass, to
3784 @file{@var{file}.35.mach}.
3787 @itemx -fdump-rtl-rnreg
3789 @opindex fdump-rtl-rnreg
3790 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3793 @itemx -fdump-rtl-regmove
3795 @opindex fdump-rtl-regmove
3796 Dump after the register move pass, to @file{@var{file}.19.regmove}.
3799 @itemx -fdump-rtl-postreload
3801 @opindex fdump-rtl-postreload
3802 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3805 @itemx -fdump-rtl-expand
3807 @opindex fdump-rtl-expand
3808 Dump after RTL generation, to @file{@var{file}.00.expand}.
3811 @itemx -fdump-rtl-sched2
3813 @opindex fdump-rtl-sched2
3814 Dump after the second scheduling pass, to @file{@var{file}.32.sched2}.
3817 @itemx -fdump-rtl-cse
3819 @opindex fdump-rtl-cse
3820 Dump after CSE (including the jump optimization that sometimes follows
3821 CSE), to @file{@var{file}.04.cse}.
3824 @itemx -fdump-rtl-sched
3826 @opindex fdump-rtl-sched
3827 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3830 @itemx -fdump-rtl-cse2
3832 @opindex fdump-rtl-cse2
3833 Dump after the second CSE pass (including the jump optimization that
3834 sometimes follows CSE), to @file{@var{file}.15.cse2}.
3837 @itemx -fdump-rtl-tracer
3839 @opindex fdump-rtl-tracer
3840 Dump after running tracer, to @file{@var{file}.12.tracer}.
3843 @itemx -fdump-rtl-vpt
3844 @itemx -fdump-rtl-vartrack
3846 @opindex fdump-rtl-vpt
3847 @opindex fdump-rtl-vartrack
3848 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3849 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3850 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3851 to @file{@var{file}.34.vartrack}.
3854 @itemx -fdump-rtl-flow2
3856 @opindex fdump-rtl-flow2
3857 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3860 @itemx -fdump-rtl-peephole2
3862 @opindex fdump-rtl-peephole2
3863 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3866 @itemx -fdump-rtl-web
3868 @opindex fdump-rtl-web
3869 Dump after live range splitting, to @file{@var{file}.14.web}.
3872 @itemx -fdump-rtl-all
3874 @opindex fdump-rtl-all
3875 Produce all the dumps listed above.
3879 Produce a core dump whenever an error occurs.
3883 Print statistics on memory usage, at the end of the run, to
3888 Annotate the assembler output with a comment indicating which
3889 pattern and alternative was used. The length of each instruction is
3894 Dump the RTL in the assembler output as a comment before each instruction.
3895 Also turns on @option{-dp} annotation.
3899 For each of the other indicated dump files (either with @option{-d} or
3900 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3901 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3905 Just generate RTL for a function instead of compiling it. Usually used
3906 with @samp{r} (@option{-fdump-rtl-expand}).
3910 Dump debugging information during parsing, to standard error.
3913 @item -fdump-unnumbered
3914 @opindex fdump-unnumbered
3915 When doing debugging dumps (see @option{-d} option above), suppress instruction
3916 numbers and line number note output. This makes it more feasible to
3917 use diff on debugging dumps for compiler invocations with different
3918 options, in particular with and without @option{-g}.
3920 @item -fdump-translation-unit @r{(C++ only)}
3921 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
3922 @opindex fdump-translation-unit
3923 Dump a representation of the tree structure for the entire translation
3924 unit to a file. The file name is made by appending @file{.tu} to the
3925 source file name. If the @samp{-@var{options}} form is used, @var{options}
3926 controls the details of the dump as described for the
3927 @option{-fdump-tree} options.
3929 @item -fdump-class-hierarchy @r{(C++ only)}
3930 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3931 @opindex fdump-class-hierarchy
3932 Dump a representation of each class's hierarchy and virtual function
3933 table layout to a file. The file name is made by appending @file{.class}
3934 to the source file name. If the @samp{-@var{options}} form is used,
3935 @var{options} controls the details of the dump as described for the
3936 @option{-fdump-tree} options.
3938 @item -fdump-ipa-@var{switch}
3940 Control the dumping at various stages of inter-procedural analysis
3941 language tree to a file. The file name is generated by appending a switch
3942 specific suffix to the source file name. The following dumps are possible:
3946 Enables all inter-procedural analysis dumps; currently the only produced
3947 dump is the @samp{cgraph} dump.
3950 Dumps information about call-graph optimization, unused function removal,
3951 and inlining decisions.
3954 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3955 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3957 Control the dumping at various stages of processing the intermediate
3958 language tree to a file. The file name is generated by appending a switch
3959 specific suffix to the source file name. If the @samp{-@var{options}}
3960 form is used, @var{options} is a list of @samp{-} separated options that
3961 control the details of the dump. Not all options are applicable to all
3962 dumps, those which are not meaningful will be ignored. The following
3963 options are available
3967 Print the address of each node. Usually this is not meaningful as it
3968 changes according to the environment and source file. Its primary use
3969 is for tying up a dump file with a debug environment.
3971 Inhibit dumping of members of a scope or body of a function merely
3972 because that scope has been reached. Only dump such items when they
3973 are directly reachable by some other path. When dumping pretty-printed
3974 trees, this option inhibits dumping the bodies of control structures.
3976 Print a raw representation of the tree. By default, trees are
3977 pretty-printed into a C-like representation.
3979 Enable more detailed dumps (not honored by every dump option).
3981 Enable dumping various statistics about the pass (not honored by every dump
3984 Enable showing basic block boundaries (disabled in raw dumps).
3986 Enable showing virtual operands for every statement.
3988 Enable showing line numbers for statements.
3990 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3992 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3995 The following tree dumps are possible:
3999 Dump before any tree based optimization, to @file{@var{file}.original}.
4002 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4005 Dump after function inlining, to @file{@var{file}.inlined}.
4008 @opindex fdump-tree-gimple
4009 Dump each function before and after the gimplification pass to a file. The
4010 file name is made by appending @file{.gimple} to the source file name.
4013 @opindex fdump-tree-cfg
4014 Dump the control flow graph of each function to a file. The file name is
4015 made by appending @file{.cfg} to the source file name.
4018 @opindex fdump-tree-vcg
4019 Dump the control flow graph of each function to a file in VCG format. The
4020 file name is made by appending @file{.vcg} to the source file name. Note
4021 that if the file contains more than one function, the generated file cannot
4022 be used directly by VCG@. You will need to cut and paste each function's
4023 graph into its own separate file first.
4026 @opindex fdump-tree-ch
4027 Dump each function after copying loop headers. The file name is made by
4028 appending @file{.ch} to the source file name.
4031 @opindex fdump-tree-ssa
4032 Dump SSA related information to a file. The file name is made by appending
4033 @file{.ssa} to the source file name.
4036 @opindex fdump-tree-salias
4037 Dump structure aliasing variable information to a file. This file name
4038 is made by appending @file{.salias} to the source file name.
4041 @opindex fdump-tree-alias
4042 Dump aliasing information for each function. The file name is made by
4043 appending @file{.alias} to the source file name.
4046 @opindex fdump-tree-ccp
4047 Dump each function after CCP@. The file name is made by appending
4048 @file{.ccp} to the source file name.
4051 @opindex fdump-tree-storeccp
4052 Dump each function after STORE-CCP. The file name is made by appending
4053 @file{.storeccp} to the source file name.
4056 @opindex fdump-tree-pre
4057 Dump trees after partial redundancy elimination. The file name is made
4058 by appending @file{.pre} to the source file name.
4061 @opindex fdump-tree-fre
4062 Dump trees after full redundancy elimination. The file name is made
4063 by appending @file{.fre} to the source file name.
4066 @opindex fdump-tree-copyprop
4067 Dump trees after copy propagation. The file name is made
4068 by appending @file{.copyprop} to the source file name.
4070 @item store_copyprop
4071 @opindex fdump-tree-store_copyprop
4072 Dump trees after store copy-propagation. The file name is made
4073 by appending @file{.store_copyprop} to the source file name.
4076 @opindex fdump-tree-dce
4077 Dump each function after dead code elimination. The file name is made by
4078 appending @file{.dce} to the source file name.
4081 @opindex fdump-tree-mudflap
4082 Dump each function after adding mudflap instrumentation. The file name is
4083 made by appending @file{.mudflap} to the source file name.
4086 @opindex fdump-tree-sra
4087 Dump each function after performing scalar replacement of aggregates. The
4088 file name is made by appending @file{.sra} to the source file name.
4091 @opindex fdump-tree-sink
4092 Dump each function after performing code sinking. The file name is made
4093 by appending @file{.sink} to the source file name.
4096 @opindex fdump-tree-dom
4097 Dump each function after applying dominator tree optimizations. The file
4098 name is made by appending @file{.dom} to the source file name.
4101 @opindex fdump-tree-dse
4102 Dump each function after applying dead store elimination. The file
4103 name is made by appending @file{.dse} to the source file name.
4106 @opindex fdump-tree-phiopt
4107 Dump each function after optimizing PHI nodes into straightline code. The file
4108 name is made by appending @file{.phiopt} to the source file name.
4111 @opindex fdump-tree-forwprop
4112 Dump each function after forward propagating single use variables. The file
4113 name is made by appending @file{.forwprop} to the source file name.
4116 @opindex fdump-tree-copyrename
4117 Dump each function after applying the copy rename optimization. The file
4118 name is made by appending @file{.copyrename} to the source file name.
4121 @opindex fdump-tree-nrv
4122 Dump each function after applying the named return value optimization on
4123 generic trees. The file name is made by appending @file{.nrv} to the source
4127 @opindex fdump-tree-vect
4128 Dump each function after applying vectorization of loops. The file name is
4129 made by appending @file{.vect} to the source file name.
4132 @opindex fdump-tree-vrp
4133 Dump each function after Value Range Propagation (VRP). The file name
4134 is made by appending @file{.vrp} to the source file name.
4137 @opindex fdump-tree-all
4138 Enable all the available tree dumps with the flags provided in this option.
4141 @item -ftree-vectorizer-verbose=@var{n}
4142 @opindex ftree-vectorizer-verbose
4143 This option controls the amount of debugging output the vectorizer prints.
4144 This information is written to standard error, unless @option{-fdump-tree-all}
4145 or @option{-fdump-tree-vect} is specified, in which case it is output to the
4146 usual dump listing file, @file{.vect}.
4148 @item -frandom-seed=@var{string}
4149 @opindex frandom-string
4150 This option provides a seed that GCC uses when it would otherwise use
4151 random numbers. It is used to generate certain symbol names
4152 that have to be different in every compiled file. It is also used to
4153 place unique stamps in coverage data files and the object files that
4154 produce them. You can use the @option{-frandom-seed} option to produce
4155 reproducibly identical object files.
4157 The @var{string} should be different for every file you compile.
4159 @item -fsched-verbose=@var{n}
4160 @opindex fsched-verbose
4161 On targets that use instruction scheduling, this option controls the
4162 amount of debugging output the scheduler prints. This information is
4163 written to standard error, unless @option{-dS} or @option{-dR} is
4164 specified, in which case it is output to the usual dump
4165 listing file, @file{.sched} or @file{.sched2} respectively. However
4166 for @var{n} greater than nine, the output is always printed to standard
4169 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4170 same information as @option{-dRS}. For @var{n} greater than one, it
4171 also output basic block probabilities, detailed ready list information
4172 and unit/insn info. For @var{n} greater than two, it includes RTL
4173 at abort point, control-flow and regions info. And for @var{n} over
4174 four, @option{-fsched-verbose} also includes dependence info.
4178 Store the usual ``temporary'' intermediate files permanently; place them
4179 in the current directory and name them based on the source file. Thus,
4180 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4181 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4182 preprocessed @file{foo.i} output file even though the compiler now
4183 normally uses an integrated preprocessor.
4185 When used in combination with the @option{-x} command line option,
4186 @option{-save-temps} is sensible enough to avoid over writing an
4187 input source file with the same extension as an intermediate file.
4188 The corresponding intermediate file may be obtained by renaming the
4189 source file before using @option{-save-temps}.
4193 Report the CPU time taken by each subprocess in the compilation
4194 sequence. For C source files, this is the compiler proper and assembler
4195 (plus the linker if linking is done). The output looks like this:
4202 The first number on each line is the ``user time'', that is time spent
4203 executing the program itself. The second number is ``system time'',
4204 time spent executing operating system routines on behalf of the program.
4205 Both numbers are in seconds.
4207 @item -fvar-tracking
4208 @opindex fvar-tracking
4209 Run variable tracking pass. It computes where variables are stored at each
4210 position in code. Better debugging information is then generated
4211 (if the debugging information format supports this information).
4213 It is enabled by default when compiling with optimization (@option{-Os},
4214 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4215 the debug info format supports it.
4217 @item -print-file-name=@var{library}
4218 @opindex print-file-name
4219 Print the full absolute name of the library file @var{library} that
4220 would be used when linking---and don't do anything else. With this
4221 option, GCC does not compile or link anything; it just prints the
4224 @item -print-multi-directory
4225 @opindex print-multi-directory
4226 Print the directory name corresponding to the multilib selected by any
4227 other switches present in the command line. This directory is supposed
4228 to exist in @env{GCC_EXEC_PREFIX}.
4230 @item -print-multi-lib
4231 @opindex print-multi-lib
4232 Print the mapping from multilib directory names to compiler switches
4233 that enable them. The directory name is separated from the switches by
4234 @samp{;}, and each switch starts with an @samp{@@} instead of the
4235 @samp{-}, without spaces between multiple switches. This is supposed to
4236 ease shell-processing.
4238 @item -print-prog-name=@var{program}
4239 @opindex print-prog-name
4240 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4242 @item -print-libgcc-file-name
4243 @opindex print-libgcc-file-name
4244 Same as @option{-print-file-name=libgcc.a}.
4246 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4247 but you do want to link with @file{libgcc.a}. You can do
4250 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4253 @item -print-search-dirs
4254 @opindex print-search-dirs
4255 Print the name of the configured installation directory and a list of
4256 program and library directories @command{gcc} will search---and don't do anything else.
4258 This is useful when @command{gcc} prints the error message
4259 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4260 To resolve this you either need to put @file{cpp0} and the other compiler
4261 components where @command{gcc} expects to find them, or you can set the environment
4262 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4263 Don't forget the trailing @samp{/}.
4264 @xref{Environment Variables}.
4267 @opindex dumpmachine
4268 Print the compiler's target machine (for example,
4269 @samp{i686-pc-linux-gnu})---and don't do anything else.
4272 @opindex dumpversion
4273 Print the compiler version (for example, @samp{3.0})---and don't do
4278 Print the compiler's built-in specs---and don't do anything else. (This
4279 is used when GCC itself is being built.) @xref{Spec Files}.
4281 @item -feliminate-unused-debug-types
4282 @opindex feliminate-unused-debug-types
4283 Normally, when producing DWARF2 output, GCC will emit debugging
4284 information for all types declared in a compilation
4285 unit, regardless of whether or not they are actually used
4286 in that compilation unit. Sometimes this is useful, such as
4287 if, in the debugger, you want to cast a value to a type that is
4288 not actually used in your program (but is declared). More often,
4289 however, this results in a significant amount of wasted space.
4290 With this option, GCC will avoid producing debug symbol output
4291 for types that are nowhere used in the source file being compiled.
4294 @node Optimize Options
4295 @section Options That Control Optimization
4296 @cindex optimize options
4297 @cindex options, optimization
4299 These options control various sorts of optimizations.
4301 Without any optimization option, the compiler's goal is to reduce the
4302 cost of compilation and to make debugging produce the expected
4303 results. Statements are independent: if you stop the program with a
4304 breakpoint between statements, you can then assign a new value to any
4305 variable or change the program counter to any other statement in the
4306 function and get exactly the results you would expect from the source
4309 Turning on optimization flags makes the compiler attempt to improve
4310 the performance and/or code size at the expense of compilation time
4311 and possibly the ability to debug the program.
4313 The compiler performs optimization based on the knowledge it has of
4314 the program. Optimization levels @option{-O2} and above, in
4315 particular, enable @emph{unit-at-a-time} mode, which allows the
4316 compiler to consider information gained from later functions in
4317 the file when compiling a function. Compiling multiple files at
4318 once to a single output file in @emph{unit-at-a-time} mode allows
4319 the compiler to use information gained from all of the files when
4320 compiling each of them.
4322 Not all optimizations are controlled directly by a flag. Only
4323 optimizations that have a flag are listed.
4330 Optimize. Optimizing compilation takes somewhat more time, and a lot
4331 more memory for a large function.
4333 With @option{-O}, the compiler tries to reduce code size and execution
4334 time, without performing any optimizations that take a great deal of
4337 @option{-O} turns on the following optimization flags:
4338 @gccoptlist{-fdefer-pop @gol
4339 -fdelayed-branch @gol
4340 -fguess-branch-probability @gol
4341 -fcprop-registers @gol
4342 -floop-optimize @gol
4343 -fif-conversion @gol
4344 -fif-conversion2 @gol
4347 -ftree-dominator-opts @gol
4352 -ftree-copyrename @gol
4357 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4358 where doing so does not interfere with debugging.
4362 Optimize even more. GCC performs nearly all supported optimizations
4363 that do not involve a space-speed tradeoff. The compiler does not
4364 perform loop unrolling or function inlining when you specify @option{-O2}.
4365 As compared to @option{-O}, this option increases both compilation time
4366 and the performance of the generated code.
4368 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4369 also turns on the following optimization flags:
4370 @gccoptlist{-fthread-jumps @gol
4372 -foptimize-sibling-calls @gol
4373 -fcse-follow-jumps -fcse-skip-blocks @gol
4374 -fgcse -fgcse-lm @gol
4375 -fexpensive-optimizations @gol
4376 -fstrength-reduce @gol
4377 -frerun-cse-after-loop -frerun-loop-opt @gol
4380 -fschedule-insns -fschedule-insns2 @gol
4381 -fsched-interblock -fsched-spec @gol
4383 -fstrict-aliasing @gol
4384 -fdelete-null-pointer-checks @gol
4385 -freorder-blocks -freorder-functions @gol
4386 -funit-at-a-time @gol
4387 -falign-functions -falign-jumps @gol
4388 -falign-loops -falign-labels @gol
4392 Please note the warning under @option{-fgcse} about
4393 invoking @option{-O2} on programs that use computed gotos.
4397 Optimize yet more. @option{-O3} turns on all optimizations specified by
4398 @option{-O2} and also turns on the @option{-finline-functions},
4399 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4403 Do not optimize. This is the default.
4407 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4408 do not typically increase code size. It also performs further
4409 optimizations designed to reduce code size.
4411 @option{-Os} disables the following optimization flags:
4412 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4413 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4414 -fprefetch-loop-arrays -ftree-vect-loop-version}
4416 If you use multiple @option{-O} options, with or without level numbers,
4417 the last such option is the one that is effective.
4420 Options of the form @option{-f@var{flag}} specify machine-independent
4421 flags. Most flags have both positive and negative forms; the negative
4422 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4423 below, only one of the forms is listed---the one you typically will
4424 use. You can figure out the other form by either removing @samp{no-}
4427 The following options control specific optimizations. They are either
4428 activated by @option{-O} options or are related to ones that are. You
4429 can use the following flags in the rare cases when ``fine-tuning'' of
4430 optimizations to be performed is desired.
4433 @item -fno-default-inline
4434 @opindex fno-default-inline
4435 Do not make member functions inline by default merely because they are
4436 defined inside the class scope (C++ only). Otherwise, when you specify
4437 @w{@option{-O}}, member functions defined inside class scope are compiled
4438 inline by default; i.e., you don't need to add @samp{inline} in front of
4439 the member function name.
4441 @item -fno-defer-pop
4442 @opindex fno-defer-pop
4443 Always pop the arguments to each function call as soon as that function
4444 returns. For machines which must pop arguments after a function call,
4445 the compiler normally lets arguments accumulate on the stack for several
4446 function calls and pops them all at once.
4448 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4452 Force memory operands to be copied into registers before doing
4453 arithmetic on them. This produces better code by making all memory
4454 references potential common subexpressions. When they are not common
4455 subexpressions, instruction combination should eliminate the separate
4456 register-load. This option is now a nop and will be removed in 4.2.
4459 @opindex fforce-addr
4460 Force memory address constants to be copied into registers before
4461 doing arithmetic on them.
4463 @item -fomit-frame-pointer
4464 @opindex fomit-frame-pointer
4465 Don't keep the frame pointer in a register for functions that
4466 don't need one. This avoids the instructions to save, set up and
4467 restore frame pointers; it also makes an extra register available
4468 in many functions. @strong{It also makes debugging impossible on
4471 On some machines, such as the VAX, this flag has no effect, because
4472 the standard calling sequence automatically handles the frame pointer
4473 and nothing is saved by pretending it doesn't exist. The
4474 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4475 whether a target machine supports this flag. @xref{Registers,,Register
4476 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4478 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4480 @item -foptimize-sibling-calls
4481 @opindex foptimize-sibling-calls
4482 Optimize sibling and tail recursive calls.
4484 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4488 Don't pay attention to the @code{inline} keyword. Normally this option
4489 is used to keep the compiler from expanding any functions inline.
4490 Note that if you are not optimizing, no functions can be expanded inline.
4492 @item -finline-functions
4493 @opindex finline-functions
4494 Integrate all simple functions into their callers. The compiler
4495 heuristically decides which functions are simple enough to be worth
4496 integrating in this way.
4498 If all calls to a given function are integrated, and the function is
4499 declared @code{static}, then the function is normally not output as
4500 assembler code in its own right.
4502 Enabled at level @option{-O3}.
4504 @item -finline-functions-called-once
4505 @opindex finline-functions-called-once
4506 Consider all @code{static} functions called once for inlining into their
4507 caller even if they are not marked @code{inline}. If a call to a given
4508 function is integrated, then the function is not output as assembler code
4511 Enabled if @option{-funit-at-a-time} is enabled.
4513 @item -fearly-inlining
4514 @opindex fearly-inlining
4515 Inline functions marked by @code{always_inline} and functions whose body seems
4516 smaller than the function call overhead early before doing
4517 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4518 makes profiling significantly cheaper and usually inlining faster on programs
4519 having large chains of nested wrapper functions.
4523 @item -finline-limit=@var{n}
4524 @opindex finline-limit
4525 By default, GCC limits the size of functions that can be inlined. This flag
4526 allows the control of this limit for functions that are explicitly marked as
4527 inline (i.e., marked with the inline keyword or defined within the class
4528 definition in c++). @var{n} is the size of functions that can be inlined in
4529 number of pseudo instructions (not counting parameter handling). The default
4530 value of @var{n} is 600.
4531 Increasing this value can result in more inlined code at
4532 the cost of compilation time and memory consumption. Decreasing usually makes
4533 the compilation faster and less code will be inlined (which presumably
4534 means slower programs). This option is particularly useful for programs that
4535 use inlining heavily such as those based on recursive templates with C++.
4537 Inlining is actually controlled by a number of parameters, which may be
4538 specified individually by using @option{--param @var{name}=@var{value}}.
4539 The @option{-finline-limit=@var{n}} option sets some of these parameters
4543 @item max-inline-insns-single
4544 is set to @var{n}/2.
4545 @item max-inline-insns-auto
4546 is set to @var{n}/2.
4547 @item min-inline-insns
4548 is set to 130 or @var{n}/4, whichever is smaller.
4549 @item max-inline-insns-rtl
4553 See below for a documentation of the individual
4554 parameters controlling inlining.
4556 @emph{Note:} pseudo instruction represents, in this particular context, an
4557 abstract measurement of function's size. In no way does it represent a count
4558 of assembly instructions and as such its exact meaning might change from one
4559 release to an another.
4561 @item -fkeep-inline-functions
4562 @opindex fkeep-inline-functions
4563 In C, emit @code{static} functions that are declared @code{inline}
4564 into the object file, even if the function has been inlined into all
4565 of its callers. This switch does not affect functions using the
4566 @code{extern inline} extension in GNU C@. In C++, emit any and all
4567 inline functions into the object file.
4569 @item -fkeep-static-consts
4570 @opindex fkeep-static-consts
4571 Emit variables declared @code{static const} when optimization isn't turned
4572 on, even if the variables aren't referenced.
4574 GCC enables this option by default. If you want to force the compiler to
4575 check if the variable was referenced, regardless of whether or not
4576 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4578 @item -fmerge-constants
4579 Attempt to merge identical constants (string constants and floating point
4580 constants) across compilation units.
4582 This option is the default for optimized compilation if the assembler and
4583 linker support it. Use @option{-fno-merge-constants} to inhibit this
4586 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4588 @item -fmerge-all-constants
4589 Attempt to merge identical constants and identical variables.
4591 This option implies @option{-fmerge-constants}. In addition to
4592 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4593 arrays or initialized constant variables with integral or floating point
4594 types. Languages like C or C++ require each non-automatic variable to
4595 have distinct location, so using this option will result in non-conforming
4598 @item -fmodulo-sched
4599 @opindex fmodulo-sched
4600 Perform swing modulo scheduling immediately before the first scheduling
4601 pass. This pass looks at innermost loops and reorders their
4602 instructions by overlapping different iterations.
4604 @item -fno-branch-count-reg
4605 @opindex fno-branch-count-reg
4606 Do not use ``decrement and branch'' instructions on a count register,
4607 but instead generate a sequence of instructions that decrement a
4608 register, compare it against zero, then branch based upon the result.
4609 This option is only meaningful on architectures that support such
4610 instructions, which include x86, PowerPC, IA-64 and S/390.
4612 The default is @option{-fbranch-count-reg}, enabled when
4613 @option{-fstrength-reduce} is enabled.
4615 @item -fno-function-cse
4616 @opindex fno-function-cse
4617 Do not put function addresses in registers; make each instruction that
4618 calls a constant function contain the function's address explicitly.
4620 This option results in less efficient code, but some strange hacks
4621 that alter the assembler output may be confused by the optimizations
4622 performed when this option is not used.
4624 The default is @option{-ffunction-cse}
4626 @item -fno-zero-initialized-in-bss
4627 @opindex fno-zero-initialized-in-bss
4628 If the target supports a BSS section, GCC by default puts variables that
4629 are initialized to zero into BSS@. This can save space in the resulting
4632 This option turns off this behavior because some programs explicitly
4633 rely on variables going to the data section. E.g., so that the
4634 resulting executable can find the beginning of that section and/or make
4635 assumptions based on that.
4637 The default is @option{-fzero-initialized-in-bss}.
4639 @item -fbounds-check
4640 @opindex fbounds-check
4641 For front-ends that support it, generate additional code to check that
4642 indices used to access arrays are within the declared range. This is
4643 currently only supported by the Java and Fortran front-ends, where
4644 this option defaults to true and false respectively.
4646 @item -fmudflap -fmudflapth -fmudflapir
4650 @cindex bounds checking
4652 For front-ends that support it (C and C++), instrument all risky
4653 pointer/array dereferencing operations, some standard library
4654 string/heap functions, and some other associated constructs with
4655 range/validity tests. Modules so instrumented should be immune to
4656 buffer overflows, invalid heap use, and some other classes of C/C++
4657 programming errors. The instrumentation relies on a separate runtime
4658 library (@file{libmudflap}), which will be linked into a program if
4659 @option{-fmudflap} is given at link time. Run-time behavior of the
4660 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4661 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4664 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4665 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4666 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4667 instrumentation should ignore pointer reads. This produces less
4668 instrumentation (and therefore faster execution) and still provides
4669 some protection against outright memory corrupting writes, but allows
4670 erroneously read data to propagate within a program.
4672 @item -fstrength-reduce
4673 @opindex fstrength-reduce
4674 Perform the optimizations of loop strength reduction and
4675 elimination of iteration variables.
4677 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4679 @item -fthread-jumps
4680 @opindex fthread-jumps
4681 Perform optimizations where we check to see if a jump branches to a
4682 location where another comparison subsumed by the first is found. If
4683 so, the first branch is redirected to either the destination of the
4684 second branch or a point immediately following it, depending on whether
4685 the condition is known to be true or false.
4687 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4689 @item -fcse-follow-jumps
4690 @opindex fcse-follow-jumps
4691 In common subexpression elimination, scan through jump instructions
4692 when the target of the jump is not reached by any other path. For
4693 example, when CSE encounters an @code{if} statement with an
4694 @code{else} clause, CSE will follow the jump when the condition
4697 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4699 @item -fcse-skip-blocks
4700 @opindex fcse-skip-blocks
4701 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4702 follow jumps which conditionally skip over blocks. When CSE
4703 encounters a simple @code{if} statement with no else clause,
4704 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4705 body of the @code{if}.
4707 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4709 @item -frerun-cse-after-loop
4710 @opindex frerun-cse-after-loop
4711 Re-run common subexpression elimination after loop optimizations has been
4714 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4716 @item -frerun-loop-opt
4717 @opindex frerun-loop-opt
4718 Run the loop optimizer twice.
4720 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4724 Perform a global common subexpression elimination pass.
4725 This pass also performs global constant and copy propagation.
4727 @emph{Note:} When compiling a program using computed gotos, a GCC
4728 extension, you may get better runtime performance if you disable
4729 the global common subexpression elimination pass by adding
4730 @option{-fno-gcse} to the command line.
4732 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4736 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4737 attempt to move loads which are only killed by stores into themselves. This
4738 allows a loop containing a load/store sequence to be changed to a load outside
4739 the loop, and a copy/store within the loop.
4741 Enabled by default when gcse is enabled.
4745 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4746 global common subexpression elimination. This pass will attempt to move
4747 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4748 loops containing a load/store sequence can be changed to a load before
4749 the loop and a store after the loop.
4751 Not enabled at any optimization level.
4755 When @option{-fgcse-las} is enabled, the global common subexpression
4756 elimination pass eliminates redundant loads that come after stores to the
4757 same memory location (both partial and full redundancies).
4759 Not enabled at any optimization level.
4761 @item -fgcse-after-reload
4762 @opindex fgcse-after-reload
4763 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4764 pass is performed after reload. The purpose of this pass is to cleanup
4767 @item -floop-optimize
4768 @opindex floop-optimize
4769 Perform loop optimizations: move constant expressions out of loops, simplify
4770 exit test conditions and optionally do strength-reduction as well.
4772 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4774 @item -floop-optimize2
4775 @opindex floop-optimize2
4776 Perform loop optimizations using the new loop optimizer. The optimizations
4777 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4780 @item -funsafe-loop-optimizations
4781 @opindex funsafe-loop-optimizations
4782 If given, the loop optimizer will assume that loop indices do not
4783 overflow, and that the loops with nontrivial exit condition are not
4784 infinite. This enables a wider range of loop optimizations even if
4785 the loop optimizer itself cannot prove that these assumptions are valid.
4786 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4787 if it finds this kind of loop.
4789 @item -fcrossjumping
4790 @opindex crossjumping
4791 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4792 resulting code may or may not perform better than without cross-jumping.
4794 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4796 @item -fif-conversion
4797 @opindex if-conversion
4798 Attempt to transform conditional jumps into branch-less equivalents. This
4799 include use of conditional moves, min, max, set flags and abs instructions, and
4800 some tricks doable by standard arithmetics. The use of conditional execution
4801 on chips where it is available is controlled by @code{if-conversion2}.
4803 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4805 @item -fif-conversion2
4806 @opindex if-conversion2
4807 Use conditional execution (where available) to transform conditional jumps into
4808 branch-less equivalents.
4810 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4812 @item -fdelete-null-pointer-checks
4813 @opindex fdelete-null-pointer-checks
4814 Use global dataflow analysis to identify and eliminate useless checks
4815 for null pointers. The compiler assumes that dereferencing a null
4816 pointer would have halted the program. If a pointer is checked after
4817 it has already been dereferenced, it cannot be null.
4819 In some environments, this assumption is not true, and programs can
4820 safely dereference null pointers. Use
4821 @option{-fno-delete-null-pointer-checks} to disable this optimization
4822 for programs which depend on that behavior.
4824 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4826 @item -fexpensive-optimizations
4827 @opindex fexpensive-optimizations
4828 Perform a number of minor optimizations that are relatively expensive.
4830 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4832 @item -foptimize-register-move
4834 @opindex foptimize-register-move
4836 Attempt to reassign register numbers in move instructions and as
4837 operands of other simple instructions in order to maximize the amount of
4838 register tying. This is especially helpful on machines with two-operand
4841 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4844 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4846 @item -fdelayed-branch
4847 @opindex fdelayed-branch
4848 If supported for the target machine, attempt to reorder instructions
4849 to exploit instruction slots available after delayed branch
4852 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4854 @item -fschedule-insns
4855 @opindex fschedule-insns
4856 If supported for the target machine, attempt to reorder instructions to
4857 eliminate execution stalls due to required data being unavailable. This
4858 helps machines that have slow floating point or memory load instructions
4859 by allowing other instructions to be issued until the result of the load
4860 or floating point instruction is required.
4862 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4864 @item -fschedule-insns2
4865 @opindex fschedule-insns2
4866 Similar to @option{-fschedule-insns}, but requests an additional pass of
4867 instruction scheduling after register allocation has been done. This is
4868 especially useful on machines with a relatively small number of
4869 registers and where memory load instructions take more than one cycle.
4871 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4873 @item -fno-sched-interblock
4874 @opindex fno-sched-interblock
4875 Don't schedule instructions across basic blocks. This is normally
4876 enabled by default when scheduling before register allocation, i.e.@:
4877 with @option{-fschedule-insns} or at @option{-O2} or higher.
4879 @item -fno-sched-spec
4880 @opindex fno-sched-spec
4881 Don't allow speculative motion of non-load instructions. This is normally
4882 enabled by default when scheduling before register allocation, i.e.@:
4883 with @option{-fschedule-insns} or at @option{-O2} or higher.
4885 @item -fsched-spec-load
4886 @opindex fsched-spec-load
4887 Allow speculative motion of some load instructions. This only makes
4888 sense when scheduling before register allocation, i.e.@: with
4889 @option{-fschedule-insns} or at @option{-O2} or higher.
4891 @item -fsched-spec-load-dangerous
4892 @opindex fsched-spec-load-dangerous
4893 Allow speculative motion of more load instructions. This only makes
4894 sense when scheduling before register allocation, i.e.@: with
4895 @option{-fschedule-insns} or at @option{-O2} or higher.
4897 @item -fsched-stalled-insns=@var{n}
4898 @opindex fsched-stalled-insns
4899 Define how many insns (if any) can be moved prematurely from the queue
4900 of stalled insns into the ready list, during the second scheduling pass.
4902 @item -fsched-stalled-insns-dep=@var{n}
4903 @opindex fsched-stalled-insns-dep
4904 Define how many insn groups (cycles) will be examined for a dependency
4905 on a stalled insn that is candidate for premature removal from the queue
4906 of stalled insns. Has an effect only during the second scheduling pass,
4907 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4909 @item -fsched2-use-superblocks
4910 @opindex fsched2-use-superblocks
4911 When scheduling after register allocation, do use superblock scheduling
4912 algorithm. Superblock scheduling allows motion across basic block boundaries
4913 resulting on faster schedules. This option is experimental, as not all machine
4914 descriptions used by GCC model the CPU closely enough to avoid unreliable
4915 results from the algorithm.
4917 This only makes sense when scheduling after register allocation, i.e.@: with
4918 @option{-fschedule-insns2} or at @option{-O2} or higher.
4920 @item -fsched2-use-traces
4921 @opindex fsched2-use-traces
4922 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4923 allocation and additionally perform code duplication in order to increase the
4924 size of superblocks using tracer pass. See @option{-ftracer} for details on
4927 This mode should produce faster but significantly longer programs. Also
4928 without @option{-fbranch-probabilities} the traces constructed may not
4929 match the reality and hurt the performance. This only makes
4930 sense when scheduling after register allocation, i.e.@: with
4931 @option{-fschedule-insns2} or at @option{-O2} or higher.
4933 @item -freschedule-modulo-scheduled-loops
4934 @opindex fscheduling-in-modulo-scheduled-loops
4935 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4936 we may want to prevent the later scheduling passes from changing its schedule, we use this
4937 option to control that.
4939 @item -fcaller-saves
4940 @opindex fcaller-saves
4941 Enable values to be allocated in registers that will be clobbered by
4942 function calls, by emitting extra instructions to save and restore the
4943 registers around such calls. Such allocation is done only when it
4944 seems to result in better code than would otherwise be produced.
4946 This option is always enabled by default on certain machines, usually
4947 those which have no call-preserved registers to use instead.
4949 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4952 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4953 enabled by default at @option{-O2} and @option{-O3}.
4956 Perform Full Redundancy Elimination (FRE) on trees. The difference
4957 between FRE and PRE is that FRE only considers expressions
4958 that are computed on all paths leading to the redundant computation.
4959 This analysis faster than PRE, though it exposes fewer redundancies.
4960 This flag is enabled by default at @option{-O} and higher.
4962 @item -ftree-copy-prop
4963 Perform copy propagation on trees. This pass eliminates unnecessary
4964 copy operations. This flag is enabled by default at @option{-O} and
4967 @item -ftree-store-copy-prop
4968 Perform copy propagation of memory loads and stores. This pass
4969 eliminates unnecessary copy operations in memory references
4970 (structures, global variables, arrays, etc). This flag is enabled by
4971 default at @option{-O2} and higher.
4974 Perform structural alias analysis on trees. This flag
4975 is enabled by default at @option{-O} and higher.
4978 Perform forward store motion on trees. This flag is
4979 enabled by default at @option{-O} and higher.
4982 Perform sparse conditional constant propagation (CCP) on trees. This
4983 pass only operates on local scalar variables and is enabled by default
4984 at @option{-O} and higher.
4986 @item -ftree-store-ccp
4987 Perform sparse conditional constant propagation (CCP) on trees. This
4988 pass operates on both local scalar variables and memory stores and
4989 loads (global variables, structures, arrays, etc). This flag is
4990 enabled by default at @option{-O2} and higher.
4993 Perform dead code elimination (DCE) on trees. This flag is enabled by
4994 default at @option{-O} and higher.
4996 @item -ftree-dominator-opts
4997 Perform a variety of simple scalar cleanups (constant/copy
4998 propagation, redundancy elimination, range propagation and expression
4999 simplification) based on a dominator tree traversal. This also
5000 performs jump threading (to reduce jumps to jumps). This flag is
5001 enabled by default at @option{-O} and higher.
5004 Perform loop header copying on trees. This is beneficial since it increases
5005 effectiveness of code motion optimizations. It also saves one jump. This flag
5006 is enabled by default at @option{-O} and higher. It is not enabled
5007 for @option{-Os}, since it usually increases code size.
5009 @item -ftree-loop-optimize
5010 Perform loop optimizations on trees. This flag is enabled by default
5011 at @option{-O} and higher.
5013 @item -ftree-loop-linear
5014 Perform linear loop transformations on tree. This flag can improve cache
5015 performance and allow further loop optimizations to take place.
5017 @item -ftree-loop-im
5018 Perform loop invariant motion on trees. This pass moves only invariants that
5019 would be hard to handle at RTL level (function calls, operations that expand to
5020 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5021 operands of conditions that are invariant out of the loop, so that we can use
5022 just trivial invariantness analysis in loop unswitching. The pass also includes
5025 @item -ftree-loop-ivcanon
5026 Create a canonical counter for number of iterations in the loop for that
5027 determining number of iterations requires complicated analysis. Later
5028 optimizations then may determine the number easily. Useful especially
5029 in connection with unrolling.
5032 Perform induction variable optimizations (strength reduction, induction
5033 variable merging and induction variable elimination) on trees.
5036 Perform scalar replacement of aggregates. This pass replaces structure
5037 references with scalars to prevent committing structures to memory too
5038 early. This flag is enabled by default at @option{-O} and higher.
5040 @item -ftree-copyrename
5041 Perform copy renaming on trees. This pass attempts to rename compiler
5042 temporaries to other variables at copy locations, usually resulting in
5043 variable names which more closely resemble the original variables. This flag
5044 is enabled by default at @option{-O} and higher.
5047 Perform temporary expression replacement during the SSA->normal phase. Single
5048 use/single def temporaries are replaced at their use location with their
5049 defining expression. This results in non-GIMPLE code, but gives the expanders
5050 much more complex trees to work on resulting in better RTL generation. This is
5051 enabled by default at @option{-O} and higher.
5054 Perform live range splitting during the SSA->normal phase. Distinct live
5055 ranges of a variable are split into unique variables, allowing for better
5056 optimization later. This is enabled by default at @option{-O} and higher.
5058 @item -ftree-vectorize
5059 Perform loop vectorization on trees.
5061 @item -ftree-vect-loop-version
5062 @opindex ftree-vect-loop-version
5063 Perform loop versioning when doing loop vectorization on trees. When a loop
5064 appears to be vectorizable except that data alignment or data dependence cannot
5065 be determined at compile time then vectorized and non-vectorized versions of
5066 the loop are generated along with runtime checks for alignment or dependence
5067 to control which version is executed. This option is enabled by default
5068 except at level @option{-Os} where it is disabled.
5071 Perform Value Range Propagation on trees. This is similar to the
5072 constant propagation pass, but instead of values, ranges of values are
5073 propagated. This allows the optimizers to remove unnecessary range
5074 checks like array bound checks and null pointer checks. This is
5075 enabled by default at @option{-O2} and higher. Null pointer check
5076 elimination is only done if @option{-fdelete-null-pointer-checks} is
5081 Perform tail duplication to enlarge superblock size. This transformation
5082 simplifies the control flow of the function allowing other optimizations to do
5085 @item -funroll-loops
5086 @opindex funroll-loops
5087 Unroll loops whose number of iterations can be determined at compile
5088 time or upon entry to the loop. @option{-funroll-loops} implies both
5089 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
5090 option makes code larger, and may or may not make it run faster.
5092 @item -funroll-all-loops
5093 @opindex funroll-all-loops
5094 Unroll all loops, even if their number of iterations is uncertain when
5095 the loop is entered. This usually makes programs run more slowly.
5096 @option{-funroll-all-loops} implies the same options as
5097 @option{-funroll-loops},
5099 @item -fsplit-ivs-in-unroller
5100 @opindex -fsplit-ivs-in-unroller
5101 Enables expressing of values of induction variables in later iterations
5102 of the unrolled loop using the value in the first iteration. This breaks
5103 long dependency chains, thus improving efficiency of the scheduling passes.
5105 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5106 same effect. However in cases the loop body is more complicated than
5107 a single basic block, this is not reliable. It also does not work at all
5108 on some of the architectures due to restrictions in the CSE pass.
5110 This optimization is enabled by default.
5112 @item -fvariable-expansion-in-unroller
5113 @opindex -fvariable-expansion-in-unroller
5114 With this option, the compiler will create multiple copies of some
5115 local variables when unrolling a loop which can result in superior code.
5117 @item -fprefetch-loop-arrays
5118 @opindex fprefetch-loop-arrays
5119 If supported by the target machine, generate instructions to prefetch
5120 memory to improve the performance of loops that access large arrays.
5122 These options may generate better or worse code; results are highly
5123 dependent on the structure of loops within the source code.
5126 @itemx -fno-peephole2
5127 @opindex fno-peephole
5128 @opindex fno-peephole2
5129 Disable any machine-specific peephole optimizations. The difference
5130 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5131 are implemented in the compiler; some targets use one, some use the
5132 other, a few use both.
5134 @option{-fpeephole} is enabled by default.
5135 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5137 @item -fno-guess-branch-probability
5138 @opindex fno-guess-branch-probability
5139 Do not guess branch probabilities using heuristics.
5141 GCC will use heuristics to guess branch probabilities if they are
5142 not provided by profiling feedback (@option{-fprofile-arcs}). These
5143 heuristics are based on the control flow graph. If some branch probabilities
5144 are specified by @samp{__builtin_expect}, then the heuristics will be
5145 used to guess branch probabilities for the rest of the control flow graph,
5146 taking the @samp{__builtin_expect} info into account. The interactions
5147 between the heuristics and @samp{__builtin_expect} can be complex, and in
5148 some cases, it may be useful to disable the heuristics so that the effects
5149 of @samp{__builtin_expect} are easier to understand.
5151 The default is @option{-fguess-branch-probability} at levels
5152 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5154 @item -freorder-blocks
5155 @opindex freorder-blocks
5156 Reorder basic blocks in the compiled function in order to reduce number of
5157 taken branches and improve code locality.
5159 Enabled at levels @option{-O2}, @option{-O3}.
5161 @item -freorder-blocks-and-partition
5162 @opindex freorder-blocks-and-partition
5163 In addition to reordering basic blocks in the compiled function, in order
5164 to reduce number of taken branches, partitions hot and cold basic blocks
5165 into separate sections of the assembly and .o files, to improve
5166 paging and cache locality performance.
5168 This optimization is automatically turned off in the presence of
5169 exception handling, for linkonce sections, for functions with a user-defined
5170 section attribute and on any architecture that does not support named
5173 @item -freorder-functions
5174 @opindex freorder-functions
5175 Reorder functions in the object file in order to
5176 improve code locality. This is implemented by using special
5177 subsections @code{.text.hot} for most frequently executed functions and
5178 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5179 the linker so object file format must support named sections and linker must
5180 place them in a reasonable way.
5182 Also profile feedback must be available in to make this option effective. See
5183 @option{-fprofile-arcs} for details.
5185 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5187 @item -fstrict-aliasing
5188 @opindex fstrict-aliasing
5189 Allows the compiler to assume the strictest aliasing rules applicable to
5190 the language being compiled. For C (and C++), this activates
5191 optimizations based on the type of expressions. In particular, an
5192 object of one type is assumed never to reside at the same address as an
5193 object of a different type, unless the types are almost the same. For
5194 example, an @code{unsigned int} can alias an @code{int}, but not a
5195 @code{void*} or a @code{double}. A character type may alias any other
5198 Pay special attention to code like this:
5211 The practice of reading from a different union member than the one most
5212 recently written to (called ``type-punning'') is common. Even with
5213 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5214 is accessed through the union type. So, the code above will work as
5215 expected. However, this code might not:
5226 Every language that wishes to perform language-specific alias analysis
5227 should define a function that computes, given an @code{tree}
5228 node, an alias set for the node. Nodes in different alias sets are not
5229 allowed to alias. For an example, see the C front-end function
5230 @code{c_get_alias_set}.
5232 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5234 @item -falign-functions
5235 @itemx -falign-functions=@var{n}
5236 @opindex falign-functions
5237 Align the start of functions to the next power-of-two greater than
5238 @var{n}, skipping up to @var{n} bytes. For instance,
5239 @option{-falign-functions=32} aligns functions to the next 32-byte
5240 boundary, but @option{-falign-functions=24} would align to the next
5241 32-byte boundary only if this can be done by skipping 23 bytes or less.
5243 @option{-fno-align-functions} and @option{-falign-functions=1} are
5244 equivalent and mean that functions will not be aligned.
5246 Some assemblers only support this flag when @var{n} is a power of two;
5247 in that case, it is rounded up.
5249 If @var{n} is not specified or is zero, use a machine-dependent default.
5251 Enabled at levels @option{-O2}, @option{-O3}.
5253 @item -falign-labels
5254 @itemx -falign-labels=@var{n}
5255 @opindex falign-labels
5256 Align all branch targets to a power-of-two boundary, skipping up to
5257 @var{n} bytes like @option{-falign-functions}. This option can easily
5258 make code slower, because it must insert dummy operations for when the
5259 branch target is reached in the usual flow of the code.
5261 @option{-fno-align-labels} and @option{-falign-labels=1} are
5262 equivalent and mean that labels will not be aligned.
5264 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5265 are greater than this value, then their values are used instead.
5267 If @var{n} is not specified or is zero, use a machine-dependent default
5268 which is very likely to be @samp{1}, meaning no alignment.
5270 Enabled at levels @option{-O2}, @option{-O3}.
5273 @itemx -falign-loops=@var{n}
5274 @opindex falign-loops
5275 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5276 like @option{-falign-functions}. The hope is that the loop will be
5277 executed many times, which will make up for any execution of the dummy
5280 @option{-fno-align-loops} and @option{-falign-loops=1} are
5281 equivalent and mean that loops will not be aligned.
5283 If @var{n} is not specified or is zero, use a machine-dependent default.
5285 Enabled at levels @option{-O2}, @option{-O3}.
5288 @itemx -falign-jumps=@var{n}
5289 @opindex falign-jumps
5290 Align branch targets to a power-of-two boundary, for branch targets
5291 where the targets can only be reached by jumping, skipping up to @var{n}
5292 bytes like @option{-falign-functions}. In this case, no dummy operations
5295 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5296 equivalent and mean that loops will not be aligned.
5298 If @var{n} is not specified or is zero, use a machine-dependent default.
5300 Enabled at levels @option{-O2}, @option{-O3}.
5302 @item -funit-at-a-time
5303 @opindex funit-at-a-time
5304 Parse the whole compilation unit before starting to produce code.
5305 This allows some extra optimizations to take place but consumes
5306 more memory (in general). There are some compatibility issues
5307 with @emph{unit-at-at-time} mode:
5310 enabling @emph{unit-at-a-time} mode may change the order
5311 in which functions, variables, and top-level @code{asm} statements
5312 are emitted, and will likely break code relying on some particular
5313 ordering. The majority of such top-level @code{asm} statements,
5314 though, can be replaced by @code{section} attributes.
5317 @emph{unit-at-a-time} mode removes unreferenced static variables
5318 and functions. This may result in undefined references
5319 when an @code{asm} statement refers directly to variables or functions
5320 that are otherwise unused. In that case either the variable/function
5321 shall be listed as an operand of the @code{asm} statement operand or,
5322 in the case of top-level @code{asm} statements the attribute @code{used}
5323 shall be used on the declaration.
5326 Static functions now can use non-standard passing conventions that
5327 may break @code{asm} statements calling functions directly. Again,
5328 attribute @code{used} will prevent this behavior.
5331 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5332 but this scheme may not be supported by future releases of GCC@.
5334 Enabled at levels @option{-O2}, @option{-O3}.
5338 Constructs webs as commonly used for register allocation purposes and assign
5339 each web individual pseudo register. This allows the register allocation pass
5340 to operate on pseudos directly, but also strengthens several other optimization
5341 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5342 however, make debugging impossible, since variables will no longer stay in a
5345 Enabled by default with @option{-funroll-loops}.
5347 @item -fwhole-program
5348 @opindex fwhole-program
5349 Assume that the current compilation unit represents whole program being
5350 compiled. All public functions and variables with the exception of @code{main}
5351 and those merged by attribute @code{externally_visible} become static functions
5352 and in a affect gets more aggressively optimized by interprocedural optimizers.
5353 While this option is equivalent to proper use of @code{static} keyword for
5354 programs consisting of single file, in combination with option
5355 @option{--combine} this flag can be used to compile most of smaller scale C
5356 programs since the functions and variables become local for the whole combined
5357 compilation unit, not for the single source file itself.
5360 @item -fno-cprop-registers
5361 @opindex fno-cprop-registers
5362 After register allocation and post-register allocation instruction splitting,
5363 we perform a copy-propagation pass to try to reduce scheduling dependencies
5364 and occasionally eliminate the copy.
5366 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5368 @item -fprofile-generate
5369 @opindex fprofile-generate
5371 Enable options usually used for instrumenting application to produce
5372 profile useful for later recompilation with profile feedback based
5373 optimization. You must use @option{-fprofile-generate} both when
5374 compiling and when linking your program.
5376 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5379 @opindex fprofile-use
5380 Enable profile feedback directed optimizations, and optimizations
5381 generally profitable only with profile feedback available.
5383 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5384 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer},
5385 @code{-fno-loop-optimize}.
5389 The following options control compiler behavior regarding floating
5390 point arithmetic. These options trade off between speed and
5391 correctness. All must be specifically enabled.
5395 @opindex ffloat-store
5396 Do not store floating point variables in registers, and inhibit other
5397 options that might change whether a floating point value is taken from a
5400 @cindex floating point precision
5401 This option prevents undesirable excess precision on machines such as
5402 the 68000 where the floating registers (of the 68881) keep more
5403 precision than a @code{double} is supposed to have. Similarly for the
5404 x86 architecture. For most programs, the excess precision does only
5405 good, but a few programs rely on the precise definition of IEEE floating
5406 point. Use @option{-ffloat-store} for such programs, after modifying
5407 them to store all pertinent intermediate computations into variables.
5411 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5412 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5413 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5414 and @option{fcx-limited-range}.
5416 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5418 This option should never be turned on by any @option{-O} option since
5419 it can result in incorrect output for programs which depend on
5420 an exact implementation of IEEE or ISO rules/specifications for
5423 @item -fno-math-errno
5424 @opindex fno-math-errno
5425 Do not set ERRNO after calling math functions that are executed
5426 with a single instruction, e.g., sqrt. A program that relies on
5427 IEEE exceptions for math error handling may want to use this flag
5428 for speed while maintaining IEEE arithmetic compatibility.
5430 This option should never be turned on by any @option{-O} option since
5431 it can result in incorrect output for programs which depend on
5432 an exact implementation of IEEE or ISO rules/specifications for
5435 The default is @option{-fmath-errno}.
5437 On Darwin systems, the math library never sets @code{errno}. There is therefore
5438 no reason for the compiler to consider the possibility that it might,
5439 and @option{-fno-math-errno} is the default.
5441 @item -funsafe-math-optimizations
5442 @opindex funsafe-math-optimizations
5443 Allow optimizations for floating-point arithmetic that (a) assume
5444 that arguments and results are valid and (b) may violate IEEE or
5445 ANSI standards. When used at link-time, it may include libraries
5446 or startup files that change the default FPU control word or other
5447 similar optimizations.
5449 This option should never be turned on by any @option{-O} option since
5450 it can result in incorrect output for programs which depend on
5451 an exact implementation of IEEE or ISO rules/specifications for
5454 The default is @option{-fno-unsafe-math-optimizations}.
5456 @item -ffinite-math-only
5457 @opindex ffinite-math-only
5458 Allow optimizations for floating-point arithmetic that assume
5459 that arguments and results are not NaNs or +-Infs.
5461 This option should never be turned on by any @option{-O} option since
5462 it can result in incorrect output for programs which depend on
5463 an exact implementation of IEEE or ISO rules/specifications.
5465 The default is @option{-fno-finite-math-only}.
5467 @item -fno-trapping-math
5468 @opindex fno-trapping-math
5469 Compile code assuming that floating-point operations cannot generate
5470 user-visible traps. These traps include division by zero, overflow,
5471 underflow, inexact result and invalid operation. This option implies
5472 @option{-fno-signaling-nans}. Setting this option may allow faster
5473 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5475 This option should never be turned on by any @option{-O} option since
5476 it can result in incorrect output for programs which depend on
5477 an exact implementation of IEEE or ISO rules/specifications for
5480 The default is @option{-ftrapping-math}.
5482 @item -frounding-math
5483 @opindex frounding-math
5484 Disable transformations and optimizations that assume default floating
5485 point rounding behavior. This is round-to-zero for all floating point
5486 to integer conversions, and round-to-nearest for all other arithmetic
5487 truncations. This option should be specified for programs that change
5488 the FP rounding mode dynamically, or that may be executed with a
5489 non-default rounding mode. This option disables constant folding of
5490 floating point expressions at compile-time (which may be affected by
5491 rounding mode) and arithmetic transformations that are unsafe in the
5492 presence of sign-dependent rounding modes.
5494 The default is @option{-fno-rounding-math}.
5496 This option is experimental and does not currently guarantee to
5497 disable all GCC optimizations that are affected by rounding mode.
5498 Future versions of GCC may provide finer control of this setting
5499 using C99's @code{FENV_ACCESS} pragma. This command line option
5500 will be used to specify the default state for @code{FENV_ACCESS}.
5502 @item -fsignaling-nans
5503 @opindex fsignaling-nans
5504 Compile code assuming that IEEE signaling NaNs may generate user-visible
5505 traps during floating-point operations. Setting this option disables
5506 optimizations that may change the number of exceptions visible with
5507 signaling NaNs. This option implies @option{-ftrapping-math}.
5509 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5512 The default is @option{-fno-signaling-nans}.
5514 This option is experimental and does not currently guarantee to
5515 disable all GCC optimizations that affect signaling NaN behavior.
5517 @item -fsingle-precision-constant
5518 @opindex fsingle-precision-constant
5519 Treat floating point constant as single precision constant instead of
5520 implicitly converting it to double precision constant.
5522 @item -fcx-limited-range
5523 @itemx -fno-cx-limited-range
5524 @opindex fcx-limited-range
5525 @opindex fno-cx-limited-range
5526 When enabled, this option states that a range reduction step is not
5527 needed when performing complex division. The default is
5528 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5530 This option controls the default setting of the ISO C99
5531 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5536 The following options control optimizations that may improve
5537 performance, but are not enabled by any @option{-O} options. This
5538 section includes experimental options that may produce broken code.
5541 @item -fbranch-probabilities
5542 @opindex fbranch-probabilities
5543 After running a program compiled with @option{-fprofile-arcs}
5544 (@pxref{Debugging Options,, Options for Debugging Your Program or
5545 @command{gcc}}), you can compile it a second time using
5546 @option{-fbranch-probabilities}, to improve optimizations based on
5547 the number of times each branch was taken. When the program
5548 compiled with @option{-fprofile-arcs} exits it saves arc execution
5549 counts to a file called @file{@var{sourcename}.gcda} for each source
5550 file The information in this data file is very dependent on the
5551 structure of the generated code, so you must use the same source code
5552 and the same optimization options for both compilations.
5554 With @option{-fbranch-probabilities}, GCC puts a
5555 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5556 These can be used to improve optimization. Currently, they are only
5557 used in one place: in @file{reorg.c}, instead of guessing which path a
5558 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5559 exactly determine which path is taken more often.
5561 @item -fprofile-values
5562 @opindex fprofile-values
5563 If combined with @option{-fprofile-arcs}, it adds code so that some
5564 data about values of expressions in the program is gathered.
5566 With @option{-fbranch-probabilities}, it reads back the data gathered
5567 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5568 notes to instructions for their later usage in optimizations.
5570 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5574 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5575 a code to gather information about values of expressions.
5577 With @option{-fbranch-probabilities}, it reads back the data gathered
5578 and actually performs the optimizations based on them.
5579 Currently the optimizations include specialization of division operation
5580 using the knowledge about the value of the denominator.
5582 @item -frename-registers
5583 @opindex frename-registers
5584 Attempt to avoid false dependencies in scheduled code by making use
5585 of registers left over after register allocation. This optimization
5586 will most benefit processors with lots of registers. Depending on the
5587 debug information format adopted by the target, however, it can
5588 make debugging impossible, since variables will no longer stay in
5589 a ``home register''.
5591 Enabled by default with @option{-funroll-loops}.
5595 Perform tail duplication to enlarge superblock size. This transformation
5596 simplifies the control flow of the function allowing other optimizations to do
5599 Enabled with @option{-fprofile-use}.
5601 @item -funroll-loops
5602 @opindex funroll-loops
5603 Unroll loops whose number of iterations can be determined at compile time or
5604 upon entry to the loop. @option{-funroll-loops} implies
5605 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5606 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5607 small constant number of iterations). This option makes code larger, and may
5608 or may not make it run faster.
5610 Enabled with @option{-fprofile-use}.
5612 @item -funroll-all-loops
5613 @opindex funroll-all-loops
5614 Unroll all loops, even if their number of iterations is uncertain when
5615 the loop is entered. This usually makes programs run more slowly.
5616 @option{-funroll-all-loops} implies the same options as
5617 @option{-funroll-loops}.
5620 @opindex fpeel-loops
5621 Peels the loops for that there is enough information that they do not
5622 roll much (from profile feedback). It also turns on complete loop peeling
5623 (i.e.@: complete removal of loops with small constant number of iterations).
5625 Enabled with @option{-fprofile-use}.
5627 @item -fmove-loop-invariants
5628 @opindex fmove-loop-invariants
5629 Enables the loop invariant motion pass in the new loop optimizer. Enabled
5630 at level @option{-O1}
5632 @item -funswitch-loops
5633 @opindex funswitch-loops
5634 Move branches with loop invariant conditions out of the loop, with duplicates
5635 of the loop on both branches (modified according to result of the condition).
5637 @item -fprefetch-loop-arrays
5638 @opindex fprefetch-loop-arrays
5639 If supported by the target machine, generate instructions to prefetch
5640 memory to improve the performance of loops that access large arrays.
5642 Disabled at level @option{-Os}.
5644 @item -ffunction-sections
5645 @itemx -fdata-sections
5646 @opindex ffunction-sections
5647 @opindex fdata-sections
5648 Place each function or data item into its own section in the output
5649 file if the target supports arbitrary sections. The name of the
5650 function or the name of the data item determines the section's name
5653 Use these options on systems where the linker can perform optimizations
5654 to improve locality of reference in the instruction space. Most systems
5655 using the ELF object format and SPARC processors running Solaris 2 have
5656 linkers with such optimizations. AIX may have these optimizations in
5659 Only use these options when there are significant benefits from doing
5660 so. When you specify these options, the assembler and linker will
5661 create larger object and executable files and will also be slower.
5662 You will not be able to use @code{gprof} on all systems if you
5663 specify this option and you may have problems with debugging if
5664 you specify both this option and @option{-g}.
5666 @item -fbranch-target-load-optimize
5667 @opindex fbranch-target-load-optimize
5668 Perform branch target register load optimization before prologue / epilogue
5670 The use of target registers can typically be exposed only during reload,
5671 thus hoisting loads out of loops and doing inter-block scheduling needs
5672 a separate optimization pass.
5674 @item -fbranch-target-load-optimize2
5675 @opindex fbranch-target-load-optimize2
5676 Perform branch target register load optimization after prologue / epilogue
5679 @item -fbtr-bb-exclusive
5680 @opindex fbtr-bb-exclusive
5681 When performing branch target register load optimization, don't reuse
5682 branch target registers in within any basic block.
5684 @item -fstack-protector
5685 Emit extra code to check for buffer overflows, such as stack smashing
5686 attacks. This is done by adding a guard variable to functions with
5687 vulnerable objects. This includes functions that call alloca, and
5688 functions with buffers larger than 8 bytes. The guards are initialized
5689 when a function is entered and then checked when the function exits.
5690 If a guard check fails, an error message is printed and the program exits.
5692 @item -fstack-protector-all
5693 Like @option{-fstack-protector} except that all functions are protected.
5695 @item --param @var{name}=@var{value}
5697 In some places, GCC uses various constants to control the amount of
5698 optimization that is done. For example, GCC will not inline functions
5699 that contain more that a certain number of instructions. You can
5700 control some of these constants on the command-line using the
5701 @option{--param} option.
5703 The names of specific parameters, and the meaning of the values, are
5704 tied to the internals of the compiler, and are subject to change
5705 without notice in future releases.
5707 In each case, the @var{value} is an integer. The allowable choices for
5708 @var{name} are given in the following table:
5711 @item salias-max-implicit-fields
5712 The maximum number of fields in a variable without direct
5713 structure accesses for which structure aliasing will consider trying
5714 to track each field. The default is 5
5716 @item sra-max-structure-size
5717 The maximum structure size, in bytes, at which the scalar replacement
5718 of aggregates (SRA) optimization will perform block copies. The
5719 default value, 0, implies that GCC will select the most appropriate
5722 @item sra-field-structure-ratio
5723 The threshold ratio (as a percentage) between instantiated fields and
5724 the complete structure size. We say that if the ratio of the number
5725 of bytes in instantiated fields to the number of bytes in the complete
5726 structure exceeds this parameter, then block copies are not used. The
5729 @item max-crossjump-edges
5730 The maximum number of incoming edges to consider for crossjumping.
5731 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5732 the number of edges incoming to each block. Increasing values mean
5733 more aggressive optimization, making the compile time increase with
5734 probably small improvement in executable size.
5736 @item min-crossjump-insns
5737 The minimum number of instructions which must be matched at the end
5738 of two blocks before crossjumping will be performed on them. This
5739 value is ignored in the case where all instructions in the block being
5740 crossjumped from are matched. The default value is 5.
5742 @item max-grow-copy-bb-insns
5743 The maximum code size expansion factor when copying basic blocks
5744 instead of jumping. The expansion is relative to a jump instruction.
5745 The default value is 8.
5747 @item max-goto-duplication-insns
5748 The maximum number of instructions to duplicate to a block that jumps
5749 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5750 passes, GCC factors computed gotos early in the compilation process,
5751 and unfactors them as late as possible. Only computed jumps at the
5752 end of a basic blocks with no more than max-goto-duplication-insns are
5753 unfactored. The default value is 8.
5755 @item max-delay-slot-insn-search
5756 The maximum number of instructions to consider when looking for an
5757 instruction to fill a delay slot. If more than this arbitrary number of
5758 instructions is searched, the time savings from filling the delay slot
5759 will be minimal so stop searching. Increasing values mean more
5760 aggressive optimization, making the compile time increase with probably
5761 small improvement in executable run time.
5763 @item max-delay-slot-live-search
5764 When trying to fill delay slots, the maximum number of instructions to
5765 consider when searching for a block with valid live register
5766 information. Increasing this arbitrarily chosen value means more
5767 aggressive optimization, increasing the compile time. This parameter
5768 should be removed when the delay slot code is rewritten to maintain the
5771 @item max-gcse-memory
5772 The approximate maximum amount of memory that will be allocated in
5773 order to perform the global common subexpression elimination
5774 optimization. If more memory than specified is required, the
5775 optimization will not be done.
5777 @item max-gcse-passes
5778 The maximum number of passes of GCSE to run. The default is 1.
5780 @item max-pending-list-length
5781 The maximum number of pending dependencies scheduling will allow
5782 before flushing the current state and starting over. Large functions
5783 with few branches or calls can create excessively large lists which
5784 needlessly consume memory and resources.
5786 @item max-inline-insns-single
5787 Several parameters control the tree inliner used in gcc.
5788 This number sets the maximum number of instructions (counted in GCC's
5789 internal representation) in a single function that the tree inliner
5790 will consider for inlining. This only affects functions declared
5791 inline and methods implemented in a class declaration (C++).
5792 The default value is 450.
5794 @item max-inline-insns-auto
5795 When you use @option{-finline-functions} (included in @option{-O3}),
5796 a lot of functions that would otherwise not be considered for inlining
5797 by the compiler will be investigated. To those functions, a different
5798 (more restrictive) limit compared to functions declared inline can
5800 The default value is 90.
5802 @item large-function-insns
5803 The limit specifying really large functions. For functions larger than this
5804 limit after inlining inlining is constrained by
5805 @option{--param large-function-growth}. This parameter is useful primarily
5806 to avoid extreme compilation time caused by non-linear algorithms used by the
5808 This parameter is ignored when @option{-funit-at-a-time} is not used.
5809 The default value is 2700.
5811 @item large-function-growth
5812 Specifies maximal growth of large function caused by inlining in percents.
5813 This parameter is ignored when @option{-funit-at-a-time} is not used.
5814 The default value is 100 which limits large function growth to 2.0 times
5817 @item inline-unit-growth
5818 Specifies maximal overall growth of the compilation unit caused by inlining.
5819 This parameter is ignored when @option{-funit-at-a-time} is not used.
5820 The default value is 50 which limits unit growth to 1.5 times the original
5823 @item max-inline-insns-recursive
5824 @itemx max-inline-insns-recursive-auto
5825 Specifies maximum number of instructions out-of-line copy of self recursive inline
5826 function can grow into by performing recursive inlining.
5828 For functions declared inline @option{--param max-inline-insns-recursive} 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-insns-recursive-auto} is used. The
5832 default value is 450.
5834 @item max-inline-recursive-depth
5835 @itemx max-inline-recursive-depth-auto
5836 Specifies maximum recursion depth used by the recursive inlining.
5838 For functions declared inline @option{--param max-inline-recursive-depth} is
5839 taken into acount. For function not declared inline, recursive inlining
5840 happens only when @option{-finline-functions} (included in @option{-O3}) is
5841 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5842 default value is 450.
5844 @item min-inline-recursive-probability
5845 Recursive inlining is profitable only for function having deep recursion
5846 in average and can hurt for function having little recursion depth by
5847 increasing the prologue size or complexity of function body to other
5850 When profile feedback is available (see @option{-fprofile-generate}) the actual
5851 recursion depth can be guessed from probability that function will recurse via
5852 given call expression. This parameter limits inlining only to call expression
5853 whose probability exceeds given threshold (in percents). The default value is
5856 @item inline-call-cost
5857 Specify cost of call instruction relative to simple arithmetics operations
5858 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
5859 functions and at the same time increases size of leaf function that is believed to
5860 reduce function size by being inlined. In effect it increases amount of
5861 inlining for code having large abstraction penalty (many functions that just
5862 pass the arguments to other functions) and decrease inlining for code with low
5863 abstraction penalty. The default value is 16.
5865 @item max-unrolled-insns
5866 The maximum number of instructions that a loop should have if that loop
5867 is unrolled, and if the loop is unrolled, it determines how many times
5868 the loop code is unrolled.
5870 @item max-average-unrolled-insns
5871 The maximum number of instructions biased by probabilities of their execution
5872 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5873 it determines how many times the loop code is unrolled.
5875 @item max-unroll-times
5876 The maximum number of unrollings of a single loop.
5878 @item max-peeled-insns
5879 The maximum number of instructions that a loop should have if that loop
5880 is peeled, and if the loop is peeled, it determines how many times
5881 the loop code is peeled.
5883 @item max-peel-times
5884 The maximum number of peelings of a single loop.
5886 @item max-completely-peeled-insns
5887 The maximum number of insns of a completely peeled loop.
5889 @item max-completely-peel-times
5890 The maximum number of iterations of a loop to be suitable for complete peeling.
5892 @item max-unswitch-insns
5893 The maximum number of insns of an unswitched loop.
5895 @item max-unswitch-level
5896 The maximum number of branches unswitched in a single loop.
5899 The minimum cost of an expensive expression in the loop invariant motion.
5901 @item iv-consider-all-candidates-bound
5902 Bound on number of candidates for induction variables below that
5903 all candidates are considered for each use in induction variable
5904 optimizations. Only the most relevant candidates are considered
5905 if there are more candidates, to avoid quadratic time complexity.
5907 @item iv-max-considered-uses
5908 The induction variable optimizations give up on loops that contain more
5909 induction variable uses.
5911 @item iv-always-prune-cand-set-bound
5912 If number of candidates in the set is smaller than this value,
5913 we always try to remove unnecessary ivs from the set during its
5914 optimization when a new iv is added to the set.
5916 @item scev-max-expr-size
5917 Bound on size of expressions used in the scalar evolutions analyzer.
5918 Large expressions slow the analyzer.
5920 @item vect-max-version-checks
5921 The maximum number of runtime checks that can be performed when doing
5922 loop versioning in the vectorizer. See option ftree-vect-loop-version
5923 for more information.
5925 @item max-iterations-to-track
5927 The maximum number of iterations of a loop the brute force algorithm
5928 for analysis of # of iterations of the loop tries to evaluate.
5930 @item hot-bb-count-fraction
5931 Select fraction of the maximal count of repetitions of basic block in program
5932 given basic block needs to have to be considered hot.
5934 @item hot-bb-frequency-fraction
5935 Select fraction of the maximal frequency of executions of basic block in
5936 function given basic block needs to have to be considered hot
5938 @item max-predicted-iterations
5939 The maximum number of loop iterations we predict statically. This is useful
5940 in cases where function contain single loop with known bound and other loop
5941 with unknown. We predict the known number of iterations correctly, while
5942 the unknown nummber of iterations average to roughly 10. This means that the
5943 loop without bounds would appear artifically cold relative to the other one.
5945 @item tracer-dynamic-coverage
5946 @itemx tracer-dynamic-coverage-feedback
5948 This value is used to limit superblock formation once the given percentage of
5949 executed instructions is covered. This limits unnecessary code size
5952 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5953 feedback is available. The real profiles (as opposed to statically estimated
5954 ones) are much less balanced allowing the threshold to be larger value.
5956 @item tracer-max-code-growth
5957 Stop tail duplication once code growth has reached given percentage. This is
5958 rather hokey argument, as most of the duplicates will be eliminated later in
5959 cross jumping, so it may be set to much higher values than is the desired code
5962 @item tracer-min-branch-ratio
5964 Stop reverse growth when the reverse probability of best edge is less than this
5965 threshold (in percent).
5967 @item tracer-min-branch-ratio
5968 @itemx tracer-min-branch-ratio-feedback
5970 Stop forward growth if the best edge do have probability lower than this
5973 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5974 compilation for profile feedback and one for compilation without. The value
5975 for compilation with profile feedback needs to be more conservative (higher) in
5976 order to make tracer effective.
5978 @item max-cse-path-length
5980 Maximum number of basic blocks on path that cse considers. The default is 10.
5983 The maximum instructions CSE process before flushing. The default is 1000.
5985 @item global-var-threshold
5987 Counts the number of function calls (@var{n}) and the number of
5988 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
5989 single artificial variable will be created to represent all the
5990 call-clobbered variables at function call sites. This artificial
5991 variable will then be made to alias every call-clobbered variable.
5992 (done as @code{int * size_t} on the host machine; beware overflow).
5994 @item max-aliased-vops
5996 Maximum number of virtual operands allowed to represent aliases
5997 before triggering the alias grouping heuristic. Alias grouping
5998 reduces compile times and memory consumption needed for aliasing at
5999 the expense of precision loss in alias information.
6001 @item ggc-min-expand
6003 GCC uses a garbage collector to manage its own memory allocation. This
6004 parameter specifies the minimum percentage by which the garbage
6005 collector's heap should be allowed to expand between collections.
6006 Tuning this may improve compilation speed; it has no effect on code
6009 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6010 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6011 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6012 GCC is not able to calculate RAM on a particular platform, the lower
6013 bound of 30% is used. Setting this parameter and
6014 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6015 every opportunity. This is extremely slow, but can be useful for
6018 @item ggc-min-heapsize
6020 Minimum size of the garbage collector's heap before it begins bothering
6021 to collect garbage. The first collection occurs after the heap expands
6022 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6023 tuning this may improve compilation speed, and has no effect on code
6026 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6027 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6028 with a lower bound of 4096 (four megabytes) and an upper bound of
6029 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6030 particular platform, the lower bound is used. Setting this parameter
6031 very large effectively disables garbage collection. Setting this
6032 parameter and @option{ggc-min-expand} to zero causes a full collection
6033 to occur at every opportunity.
6035 @item max-reload-search-insns
6036 The maximum number of instruction reload should look backward for equivalent
6037 register. Increasing values mean more aggressive optimization, making the
6038 compile time increase with probably slightly better performance. The default
6041 @item max-cselib-memory-location
6042 The maximum number of memory locations cselib should take into acount.
6043 Increasing values mean more aggressive optimization, making the compile time
6044 increase with probably slightly better performance. The default value is 500.
6046 @item max-flow-memory-location
6047 Similar as @option{max-cselib-memory-location} but for dataflow liveness.
6048 The default value is 100.
6050 @item reorder-blocks-duplicate
6051 @itemx reorder-blocks-duplicate-feedback
6053 Used by basic block reordering pass to decide whether to use unconditional
6054 branch or duplicate the code on its destination. Code is duplicated when its
6055 estimated size is smaller than this value multiplied by the estimated size of
6056 unconditional jump in the hot spots of the program.
6058 The @option{reorder-block-duplicate-feedback} is used only when profile
6059 feedback is available and may be set to higher values than
6060 @option{reorder-block-duplicate} since information about the hot spots is more
6063 @item max-sched-region-blocks
6064 The maximum number of blocks in a region to be considered for
6065 interblock scheduling. The default value is 10.
6067 @item max-sched-region-insns
6068 The maximum number of insns in a region to be considered for
6069 interblock scheduling. The default value is 100.
6071 @item min-sched-prob
6072 The minimum probability of reaching a source block for interblock
6073 speculative scheduling. The default value is 40.
6075 @item max-last-value-rtl
6077 The maximum size measured as number of RTLs that can be recorded in an expression
6078 in combiner for a pseudo register as last known value of that register. The default
6081 @item integer-share-limit
6082 Small integer constants can use a shared data structure, reducing the
6083 compiler's memory usage and increasing its speed. This sets the maximum
6084 value of a shared integer constant's. The default value is 256.
6086 @item min-virtual-mappings
6087 Specifies the minimum number of virtual mappings in the incremental
6088 SSA updater that should be registered to trigger the virtual mappings
6089 heuristic defined by virtual-mappings-ratio. The default value is
6092 @item virtual-mappings-ratio
6093 If the number of virtual mappings is virtual-mappings-ratio bigger
6094 than the number of virtual symbols to be updated, then the incremental
6095 SSA updater switches to a full update for those symbols. The default
6098 @item ssp-buffer-size
6099 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6100 protection when @option{-fstack-protection} is used.
6102 @item max-jump-thread-duplication-stmts
6103 Maximum number of statements allowed in a block that needs to be
6104 duplicated when threading jumps.
6108 @node Preprocessor Options
6109 @section Options Controlling the Preprocessor
6110 @cindex preprocessor options
6111 @cindex options, preprocessor
6113 These options control the C preprocessor, which is run on each C source
6114 file before actual compilation.
6116 If you use the @option{-E} option, nothing is done except preprocessing.
6117 Some of these options make sense only together with @option{-E} because
6118 they cause the preprocessor output to be unsuitable for actual
6123 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6124 and pass @var{option} directly through to the preprocessor. If
6125 @var{option} contains commas, it is split into multiple options at the
6126 commas. However, many options are modified, translated or interpreted
6127 by the compiler driver before being passed to the preprocessor, and
6128 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6129 interface is undocumented and subject to change, so whenever possible
6130 you should avoid using @option{-Wp} and let the driver handle the
6133 @item -Xpreprocessor @var{option}
6134 @opindex preprocessor
6135 Pass @var{option} as an option to the preprocessor. You can use this to
6136 supply system-specific preprocessor options which GCC does not know how to
6139 If you want to pass an option that takes an argument, you must use
6140 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6143 @include cppopts.texi
6145 @node Assembler Options
6146 @section Passing Options to the Assembler
6148 @c prevent bad page break with this line
6149 You can pass options to the assembler.
6152 @item -Wa,@var{option}
6154 Pass @var{option} as an option to the assembler. If @var{option}
6155 contains commas, it is split into multiple options at the commas.
6157 @item -Xassembler @var{option}
6159 Pass @var{option} as an option to the assembler. You can use this to
6160 supply system-specific assembler options which GCC does not know how to
6163 If you want to pass an option that takes an argument, you must use
6164 @option{-Xassembler} twice, once for the option and once for the argument.
6169 @section Options for Linking
6170 @cindex link options
6171 @cindex options, linking
6173 These options come into play when the compiler links object files into
6174 an executable output file. They are meaningless if the compiler is
6175 not doing a link step.
6179 @item @var{object-file-name}
6180 A file name that does not end in a special recognized suffix is
6181 considered to name an object file or library. (Object files are
6182 distinguished from libraries by the linker according to the file
6183 contents.) If linking is done, these object files are used as input
6192 If any of these options is used, then the linker is not run, and
6193 object file names should not be used as arguments. @xref{Overall
6197 @item -l@var{library}
6198 @itemx -l @var{library}
6200 Search the library named @var{library} when linking. (The second
6201 alternative with the library as a separate argument is only for
6202 POSIX compliance and is not recommended.)
6204 It makes a difference where in the command you write this option; the
6205 linker searches and processes libraries and object files in the order they
6206 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6207 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6208 to functions in @samp{z}, those functions may not be loaded.
6210 The linker searches a standard list of directories for the library,
6211 which is actually a file named @file{lib@var{library}.a}. The linker
6212 then uses this file as if it had been specified precisely by name.
6214 The directories searched include several standard system directories
6215 plus any that you specify with @option{-L}.
6217 Normally the files found this way are library files---archive files
6218 whose members are object files. The linker handles an archive file by
6219 scanning through it for members which define symbols that have so far
6220 been referenced but not defined. But if the file that is found is an
6221 ordinary object file, it is linked in the usual fashion. The only
6222 difference between using an @option{-l} option and specifying a file name
6223 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6224 and searches several directories.
6228 You need this special case of the @option{-l} option in order to
6229 link an Objective-C or Objective-C++ program.
6232 @opindex nostartfiles
6233 Do not use the standard system startup files when linking.
6234 The standard system libraries are used normally, unless @option{-nostdlib}
6235 or @option{-nodefaultlibs} is used.
6237 @item -nodefaultlibs
6238 @opindex nodefaultlibs
6239 Do not use the standard system libraries when linking.
6240 Only the libraries you specify will be passed to the linker.
6241 The standard startup files are used normally, unless @option{-nostartfiles}
6242 is used. The compiler may generate calls to @code{memcmp},
6243 @code{memset}, @code{memcpy} and @code{memmove}.
6244 These entries are usually resolved by entries in
6245 libc. These entry points should be supplied through some other
6246 mechanism when this option is specified.
6250 Do not use the standard system startup files or libraries when linking.
6251 No startup files and only the libraries you specify will be passed to
6252 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6253 @code{memcpy} and @code{memmove}.
6254 These entries are usually resolved by entries in
6255 libc. These entry points should be supplied through some other
6256 mechanism when this option is specified.
6258 @cindex @option{-lgcc}, use with @option{-nostdlib}
6259 @cindex @option{-nostdlib} and unresolved references
6260 @cindex unresolved references and @option{-nostdlib}
6261 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6262 @cindex @option{-nodefaultlibs} and unresolved references
6263 @cindex unresolved references and @option{-nodefaultlibs}
6264 One of the standard libraries bypassed by @option{-nostdlib} and
6265 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6266 that GCC uses to overcome shortcomings of particular machines, or special
6267 needs for some languages.
6268 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6269 Collection (GCC) Internals},
6270 for more discussion of @file{libgcc.a}.)
6271 In most cases, you need @file{libgcc.a} even when you want to avoid
6272 other standard libraries. In other words, when you specify @option{-nostdlib}
6273 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6274 This ensures that you have no unresolved references to internal GCC
6275 library subroutines. (For example, @samp{__main}, used to ensure C++
6276 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6277 GNU Compiler Collection (GCC) Internals}.)
6281 Produce a position independent executable on targets which support it.
6282 For predictable results, you must also specify the same set of options
6283 that were used to generate code (@option{-fpie}, @option{-fPIE},
6284 or model suboptions) when you specify this option.
6288 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6289 that support it. This instructs the linker to add all symbols, not
6290 only used ones, to the dynamic symbol table. This option is needed
6291 for some uses of @code{dlopen} or to allow obtaining backtraces
6292 from within a program.
6296 Remove all symbol table and relocation information from the executable.
6300 On systems that support dynamic linking, this prevents linking with the shared
6301 libraries. On other systems, this option has no effect.
6305 Produce a shared object which can then be linked with other objects to
6306 form an executable. Not all systems support this option. For predictable
6307 results, you must also specify the same set of options that were used to
6308 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6309 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6310 needs to build supplementary stub code for constructors to work. On
6311 multi-libbed systems, @samp{gcc -shared} must select the correct support
6312 libraries to link against. Failing to supply the correct flags may lead
6313 to subtle defects. Supplying them in cases where they are not necessary
6316 @item -shared-libgcc
6317 @itemx -static-libgcc
6318 @opindex shared-libgcc
6319 @opindex static-libgcc
6320 On systems that provide @file{libgcc} as a shared library, these options
6321 force the use of either the shared or static version respectively.
6322 If no shared version of @file{libgcc} was built when the compiler was
6323 configured, these options have no effect.
6325 There are several situations in which an application should use the
6326 shared @file{libgcc} instead of the static version. The most common
6327 of these is when the application wishes to throw and catch exceptions
6328 across different shared libraries. In that case, each of the libraries
6329 as well as the application itself should use the shared @file{libgcc}.
6331 Therefore, the G++ and GCJ drivers automatically add
6332 @option{-shared-libgcc} whenever you build a shared library or a main
6333 executable, because C++ and Java programs typically use exceptions, so
6334 this is the right thing to do.
6336 If, instead, you use the GCC driver to create shared libraries, you may
6337 find that they will not always be linked with the shared @file{libgcc}.
6338 If GCC finds, at its configuration time, that you have a non-GNU linker
6339 or a GNU linker that does not support option @option{--eh-frame-hdr},
6340 it will link the shared version of @file{libgcc} into shared libraries
6341 by default. Otherwise, it will take advantage of the linker and optimize
6342 away the linking with the shared version of @file{libgcc}, linking with
6343 the static version of libgcc by default. This allows exceptions to
6344 propagate through such shared libraries, without incurring relocation
6345 costs at library load time.
6347 However, if a library or main executable is supposed to throw or catch
6348 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6349 for the languages used in the program, or using the option
6350 @option{-shared-libgcc}, such that it is linked with the shared
6355 Bind references to global symbols when building a shared object. Warn
6356 about any unresolved references (unless overridden by the link editor
6357 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6360 @item -Xlinker @var{option}
6362 Pass @var{option} as an option to the linker. You can use this to
6363 supply system-specific linker options which GCC does not know how to
6366 If you want to pass an option that takes an argument, you must use
6367 @option{-Xlinker} twice, once for the option and once for the argument.
6368 For example, to pass @option{-assert definitions}, you must write
6369 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6370 @option{-Xlinker "-assert definitions"}, because this passes the entire
6371 string as a single argument, which is not what the linker expects.
6373 @item -Wl,@var{option}
6375 Pass @var{option} as an option to the linker. If @var{option} contains
6376 commas, it is split into multiple options at the commas.
6378 @item -u @var{symbol}
6380 Pretend the symbol @var{symbol} is undefined, to force linking of
6381 library modules to define it. You can use @option{-u} multiple times with
6382 different symbols to force loading of additional library modules.
6385 @node Directory Options
6386 @section Options for Directory Search
6387 @cindex directory options
6388 @cindex options, directory search
6391 These options specify directories to search for header files, for
6392 libraries and for parts of the compiler:
6397 Add the directory @var{dir} to the head of the list of directories to be
6398 searched for header files. This can be used to override a system header
6399 file, substituting your own version, since these directories are
6400 searched before the system header file directories. However, you should
6401 not use this option to add directories that contain vendor-supplied
6402 system header files (use @option{-isystem} for that). If you use more than
6403 one @option{-I} option, the directories are scanned in left-to-right
6404 order; the standard system directories come after.
6406 If a standard system include directory, or a directory specified with
6407 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6408 option will be ignored. The directory will still be searched but as a
6409 system directory at its normal position in the system include chain.
6410 This is to ensure that GCC's procedure to fix buggy system headers and
6411 the ordering for the include_next directive are not inadvertently changed.
6412 If you really need to change the search order for system directories,
6413 use the @option{-nostdinc} and/or @option{-isystem} options.
6415 @item -iquote@var{dir}
6417 Add the directory @var{dir} to the head of the list of directories to
6418 be searched for header files only for the case of @samp{#include
6419 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6420 otherwise just like @option{-I}.
6424 Add directory @var{dir} to the list of directories to be searched
6427 @item -B@var{prefix}
6429 This option specifies where to find the executables, libraries,
6430 include files, and data files of the compiler itself.
6432 The compiler driver program runs one or more of the subprograms
6433 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6434 @var{prefix} as a prefix for each program it tries to run, both with and
6435 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6437 For each subprogram to be run, the compiler driver first tries the
6438 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6439 was not specified, the driver tries two standard prefixes, which are
6440 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6441 those results in a file name that is found, the unmodified program
6442 name is searched for using the directories specified in your
6443 @env{PATH} environment variable.
6445 The compiler will check to see if the path provided by the @option{-B}
6446 refers to a directory, and if necessary it will add a directory
6447 separator character at the end of the path.
6449 @option{-B} prefixes that effectively specify directory names also apply
6450 to libraries in the linker, because the compiler translates these
6451 options into @option{-L} options for the linker. They also apply to
6452 includes files in the preprocessor, because the compiler translates these
6453 options into @option{-isystem} options for the preprocessor. In this case,
6454 the compiler appends @samp{include} to the prefix.
6456 The run-time support file @file{libgcc.a} can also be searched for using
6457 the @option{-B} prefix, if needed. If it is not found there, the two
6458 standard prefixes above are tried, and that is all. The file is left
6459 out of the link if it is not found by those means.
6461 Another way to specify a prefix much like the @option{-B} prefix is to use
6462 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6465 As a special kludge, if the path provided by @option{-B} is
6466 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6467 9, then it will be replaced by @file{[dir/]include}. This is to help
6468 with boot-strapping the compiler.
6470 @item -specs=@var{file}
6472 Process @var{file} after the compiler reads in the standard @file{specs}
6473 file, in order to override the defaults that the @file{gcc} driver
6474 program uses when determining what switches to pass to @file{cc1},
6475 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6476 @option{-specs=@var{file}} can be specified on the command line, and they
6477 are processed in order, from left to right.
6479 @item --sysroot=@var{dir}
6481 Use @var{dir} as the logical root directory for headers and libraries.
6482 For example, if the compiler would normally search for headers in
6483 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6484 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6486 If you use both this option and the @option{-isysroot} option, then
6487 the @option{--sysroot} option will apply to libraries, but the
6488 @option{-isysroot} option will apply to header files.
6490 The GNU linker (beginning with version 2.16) has the necessary support
6491 for this option. If your linker does not support this option, the
6492 header file aspect of @option{--sysroot} will still work, but the
6493 library aspect will not.
6497 This option has been deprecated. Please use @option{-iquote} instead for
6498 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6499 Any directories you specify with @option{-I} options before the @option{-I-}
6500 option are searched only for the case of @samp{#include "@var{file}"};
6501 they are not searched for @samp{#include <@var{file}>}.
6503 If additional directories are specified with @option{-I} options after
6504 the @option{-I-}, these directories are searched for all @samp{#include}
6505 directives. (Ordinarily @emph{all} @option{-I} directories are used
6508 In addition, the @option{-I-} option inhibits the use of the current
6509 directory (where the current input file came from) as the first search
6510 directory for @samp{#include "@var{file}"}. There is no way to
6511 override this effect of @option{-I-}. With @option{-I.} you can specify
6512 searching the directory which was current when the compiler was
6513 invoked. That is not exactly the same as what the preprocessor does
6514 by default, but it is often satisfactory.
6516 @option{-I-} does not inhibit the use of the standard system directories
6517 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6524 @section Specifying subprocesses and the switches to pass to them
6527 @command{gcc} is a driver program. It performs its job by invoking a
6528 sequence of other programs to do the work of compiling, assembling and
6529 linking. GCC interprets its command-line parameters and uses these to
6530 deduce which programs it should invoke, and which command-line options
6531 it ought to place on their command lines. This behavior is controlled
6532 by @dfn{spec strings}. In most cases there is one spec string for each
6533 program that GCC can invoke, but a few programs have multiple spec
6534 strings to control their behavior. The spec strings built into GCC can
6535 be overridden by using the @option{-specs=} command-line switch to specify
6538 @dfn{Spec files} are plaintext files that are used to construct spec
6539 strings. They consist of a sequence of directives separated by blank
6540 lines. The type of directive is determined by the first non-whitespace
6541 character on the line and it can be one of the following:
6544 @item %@var{command}
6545 Issues a @var{command} to the spec file processor. The commands that can
6549 @item %include <@var{file}>
6551 Search for @var{file} and insert its text at the current point in the
6554 @item %include_noerr <@var{file}>
6555 @cindex %include_noerr
6556 Just like @samp{%include}, but do not generate an error message if the include
6557 file cannot be found.
6559 @item %rename @var{old_name} @var{new_name}
6561 Rename the spec string @var{old_name} to @var{new_name}.
6565 @item *[@var{spec_name}]:
6566 This tells the compiler to create, override or delete the named spec
6567 string. All lines after this directive up to the next directive or
6568 blank line are considered to be the text for the spec string. If this
6569 results in an empty string then the spec will be deleted. (Or, if the
6570 spec did not exist, then nothing will happened.) Otherwise, if the spec
6571 does not currently exist a new spec will be created. If the spec does
6572 exist then its contents will be overridden by the text of this
6573 directive, unless the first character of that text is the @samp{+}
6574 character, in which case the text will be appended to the spec.
6576 @item [@var{suffix}]:
6577 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6578 and up to the next directive or blank line are considered to make up the
6579 spec string for the indicated suffix. When the compiler encounters an
6580 input file with the named suffix, it will processes the spec string in
6581 order to work out how to compile that file. For example:
6588 This says that any input file whose name ends in @samp{.ZZ} should be
6589 passed to the program @samp{z-compile}, which should be invoked with the
6590 command-line switch @option{-input} and with the result of performing the
6591 @samp{%i} substitution. (See below.)
6593 As an alternative to providing a spec string, the text that follows a
6594 suffix directive can be one of the following:
6597 @item @@@var{language}
6598 This says that the suffix is an alias for a known @var{language}. This is
6599 similar to using the @option{-x} command-line switch to GCC to specify a
6600 language explicitly. For example:
6607 Says that .ZZ files are, in fact, C++ source files.
6610 This causes an error messages saying:
6613 @var{name} compiler not installed on this system.
6617 GCC already has an extensive list of suffixes built into it.
6618 This directive will add an entry to the end of the list of suffixes, but
6619 since the list is searched from the end backwards, it is effectively
6620 possible to override earlier entries using this technique.
6624 GCC has the following spec strings built into it. Spec files can
6625 override these strings or create their own. Note that individual
6626 targets can also add their own spec strings to this list.
6629 asm Options to pass to the assembler
6630 asm_final Options to pass to the assembler post-processor
6631 cpp Options to pass to the C preprocessor
6632 cc1 Options to pass to the C compiler
6633 cc1plus Options to pass to the C++ compiler
6634 endfile Object files to include at the end of the link
6635 link Options to pass to the linker
6636 lib Libraries to include on the command line to the linker
6637 libgcc Decides which GCC support library to pass to the linker
6638 linker Sets the name of the linker
6639 predefines Defines to be passed to the C preprocessor
6640 signed_char Defines to pass to CPP to say whether @code{char} is signed
6642 startfile Object files to include at the start of the link
6645 Here is a small example of a spec file:
6651 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6654 This example renames the spec called @samp{lib} to @samp{old_lib} and
6655 then overrides the previous definition of @samp{lib} with a new one.
6656 The new definition adds in some extra command-line options before
6657 including the text of the old definition.
6659 @dfn{Spec strings} are a list of command-line options to be passed to their
6660 corresponding program. In addition, the spec strings can contain
6661 @samp{%}-prefixed sequences to substitute variable text or to
6662 conditionally insert text into the command line. Using these constructs
6663 it is possible to generate quite complex command lines.
6665 Here is a table of all defined @samp{%}-sequences for spec
6666 strings. Note that spaces are not generated automatically around the
6667 results of expanding these sequences. Therefore you can concatenate them
6668 together or combine them with constant text in a single argument.
6672 Substitute one @samp{%} into the program name or argument.
6675 Substitute the name of the input file being processed.
6678 Substitute the basename of the input file being processed.
6679 This is the substring up to (and not including) the last period
6680 and not including the directory.
6683 This is the same as @samp{%b}, but include the file suffix (text after
6687 Marks the argument containing or following the @samp{%d} as a
6688 temporary file name, so that that file will be deleted if GCC exits
6689 successfully. Unlike @samp{%g}, this contributes no text to the
6692 @item %g@var{suffix}
6693 Substitute a file name that has suffix @var{suffix} and is chosen
6694 once per compilation, and mark the argument in the same way as
6695 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6696 name is now chosen in a way that is hard to predict even when previously
6697 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6698 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6699 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6700 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6701 was simply substituted with a file name chosen once per compilation,
6702 without regard to any appended suffix (which was therefore treated
6703 just like ordinary text), making such attacks more likely to succeed.
6705 @item %u@var{suffix}
6706 Like @samp{%g}, but generates a new temporary file name even if
6707 @samp{%u@var{suffix}} was already seen.
6709 @item %U@var{suffix}
6710 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6711 new one if there is no such last file name. In the absence of any
6712 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6713 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6714 would involve the generation of two distinct file names, one
6715 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6716 simply substituted with a file name chosen for the previous @samp{%u},
6717 without regard to any appended suffix.
6719 @item %j@var{suffix}
6720 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6721 writable, and if save-temps is off; otherwise, substitute the name
6722 of a temporary file, just like @samp{%u}. This temporary file is not
6723 meant for communication between processes, but rather as a junk
6726 @item %|@var{suffix}
6727 @itemx %m@var{suffix}
6728 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6729 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6730 all. These are the two most common ways to instruct a program that it
6731 should read from standard input or write to standard output. If you
6732 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6733 construct: see for example @file{f/lang-specs.h}.
6735 @item %.@var{SUFFIX}
6736 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6737 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6738 terminated by the next space or %.
6741 Marks the argument containing or following the @samp{%w} as the
6742 designated output file of this compilation. This puts the argument
6743 into the sequence of arguments that @samp{%o} will substitute later.
6746 Substitutes the names of all the output files, with spaces
6747 automatically placed around them. You should write spaces
6748 around the @samp{%o} as well or the results are undefined.
6749 @samp{%o} is for use in the specs for running the linker.
6750 Input files whose names have no recognized suffix are not compiled
6751 at all, but they are included among the output files, so they will
6755 Substitutes the suffix for object files. Note that this is
6756 handled specially when it immediately follows @samp{%g, %u, or %U},
6757 because of the need for those to form complete file names. The
6758 handling is such that @samp{%O} is treated exactly as if it had already
6759 been substituted, except that @samp{%g, %u, and %U} do not currently
6760 support additional @var{suffix} characters following @samp{%O} as they would
6761 following, for example, @samp{.o}.
6764 Substitutes the standard macro predefinitions for the
6765 current target machine. Use this when running @code{cpp}.
6768 Like @samp{%p}, but puts @samp{__} before and after the name of each
6769 predefined macro, except for macros that start with @samp{__} or with
6770 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6774 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6775 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
6776 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6780 Current argument is the name of a library or startup file of some sort.
6781 Search for that file in a standard list of directories and substitute
6782 the full name found.
6785 Print @var{str} as an error message. @var{str} is terminated by a newline.
6786 Use this when inconsistent options are detected.
6789 Substitute the contents of spec string @var{name} at this point.
6792 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6794 @item %x@{@var{option}@}
6795 Accumulate an option for @samp{%X}.
6798 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6802 Output the accumulated assembler options specified by @option{-Wa}.
6805 Output the accumulated preprocessor options specified by @option{-Wp}.
6808 Process the @code{asm} spec. This is used to compute the
6809 switches to be passed to the assembler.
6812 Process the @code{asm_final} spec. This is a spec string for
6813 passing switches to an assembler post-processor, if such a program is
6817 Process the @code{link} spec. This is the spec for computing the
6818 command line passed to the linker. Typically it will make use of the
6819 @samp{%L %G %S %D and %E} sequences.
6822 Dump out a @option{-L} option for each directory that GCC believes might
6823 contain startup files. If the target supports multilibs then the
6824 current multilib directory will be prepended to each of these paths.
6827 Process the @code{lib} spec. This is a spec string for deciding which
6828 libraries should be included on the command line to the linker.
6831 Process the @code{libgcc} spec. This is a spec string for deciding
6832 which GCC support library should be included on the command line to the linker.
6835 Process the @code{startfile} spec. This is a spec for deciding which
6836 object files should be the first ones passed to the linker. Typically
6837 this might be a file named @file{crt0.o}.
6840 Process the @code{endfile} spec. This is a spec string that specifies
6841 the last object files that will be passed to the linker.
6844 Process the @code{cpp} spec. This is used to construct the arguments
6845 to be passed to the C preprocessor.
6848 Process the @code{cc1} spec. This is used to construct the options to be
6849 passed to the actual C compiler (@samp{cc1}).
6852 Process the @code{cc1plus} spec. This is used to construct the options to be
6853 passed to the actual C++ compiler (@samp{cc1plus}).
6856 Substitute the variable part of a matched option. See below.
6857 Note that each comma in the substituted string is replaced by
6861 Remove all occurrences of @code{-S} from the command line. Note---this
6862 command is position dependent. @samp{%} commands in the spec string
6863 before this one will see @code{-S}, @samp{%} commands in the spec string
6864 after this one will not.
6866 @item %:@var{function}(@var{args})
6867 Call the named function @var{function}, passing it @var{args}.
6868 @var{args} is first processed as a nested spec string, then split
6869 into an argument vector in the usual fashion. The function returns
6870 a string which is processed as if it had appeared literally as part
6871 of the current spec.
6873 The following built-in spec functions are provided:
6876 @item @code{if-exists}
6877 The @code{if-exists} spec function takes one argument, an absolute
6878 pathname to a file. If the file exists, @code{if-exists} returns the
6879 pathname. Here is a small example of its usage:
6883 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6886 @item @code{if-exists-else}
6887 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6888 spec function, except that it takes two arguments. The first argument is
6889 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6890 returns the pathname. If it does not exist, it returns the second argument.
6891 This way, @code{if-exists-else} can be used to select one file or another,
6892 based on the existence of the first. Here is a small example of its usage:
6896 crt0%O%s %:if-exists(crti%O%s) \
6897 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6900 @item @code{replace-outfile}
6901 The @code{replace-outfile} spec function takes two arguments. It looks for the
6902 first argument in the outfiles array and replaces it with the second argument. Here
6903 is a small example of its usage:
6906 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
6912 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6913 If that switch was not specified, this substitutes nothing. Note that
6914 the leading dash is omitted when specifying this option, and it is
6915 automatically inserted if the substitution is performed. Thus the spec
6916 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6917 and would output the command line option @option{-foo}.
6919 @item %W@{@code{S}@}
6920 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6923 @item %@{@code{S}*@}
6924 Substitutes all the switches specified to GCC whose names start
6925 with @code{-S}, but which also take an argument. This is used for
6926 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6927 GCC considers @option{-o foo} as being
6928 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6929 text, including the space. Thus two arguments would be generated.
6931 @item %@{@code{S}*&@code{T}*@}
6932 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6933 (the order of @code{S} and @code{T} in the spec is not significant).
6934 There can be any number of ampersand-separated variables; for each the
6935 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6937 @item %@{@code{S}:@code{X}@}
6938 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6940 @item %@{!@code{S}:@code{X}@}
6941 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6943 @item %@{@code{S}*:@code{X}@}
6944 Substitutes @code{X} if one or more switches whose names start with
6945 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6946 once, no matter how many such switches appeared. However, if @code{%*}
6947 appears somewhere in @code{X}, then @code{X} will be substituted once
6948 for each matching switch, with the @code{%*} replaced by the part of
6949 that switch that matched the @code{*}.
6951 @item %@{.@code{S}:@code{X}@}
6952 Substitutes @code{X}, if processing a file with suffix @code{S}.
6954 @item %@{!.@code{S}:@code{X}@}
6955 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6957 @item %@{@code{S}|@code{P}:@code{X}@}
6958 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6959 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6960 although they have a stronger binding than the @samp{|}. If @code{%*}
6961 appears in @code{X}, all of the alternatives must be starred, and only
6962 the first matching alternative is substituted.
6964 For example, a spec string like this:
6967 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6970 will output the following command-line options from the following input
6971 command-line options:
6976 -d fred.c -foo -baz -boggle
6977 -d jim.d -bar -baz -boggle
6980 @item %@{S:X; T:Y; :D@}
6982 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6983 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6984 be as many clauses as you need. This may be combined with @code{.},
6985 @code{!}, @code{|}, and @code{*} as needed.
6990 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6991 construct may contain other nested @samp{%} constructs or spaces, or
6992 even newlines. They are processed as usual, as described above.
6993 Trailing white space in @code{X} is ignored. White space may also
6994 appear anywhere on the left side of the colon in these constructs,
6995 except between @code{.} or @code{*} and the corresponding word.
6997 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6998 handled specifically in these constructs. If another value of
6999 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7000 @option{-W} switch is found later in the command line, the earlier
7001 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7002 just one letter, which passes all matching options.
7004 The character @samp{|} at the beginning of the predicate text is used to
7005 indicate that a command should be piped to the following command, but
7006 only if @option{-pipe} is specified.
7008 It is built into GCC which switches take arguments and which do not.
7009 (You might think it would be useful to generalize this to allow each
7010 compiler's spec to say which switches take arguments. But this cannot
7011 be done in a consistent fashion. GCC cannot even decide which input
7012 files have been specified without knowing which switches take arguments,
7013 and it must know which input files to compile in order to tell which
7016 GCC also knows implicitly that arguments starting in @option{-l} are to be
7017 treated as compiler output files, and passed to the linker in their
7018 proper position among the other output files.
7020 @c man begin OPTIONS
7022 @node Target Options
7023 @section Specifying Target Machine and Compiler Version
7024 @cindex target options
7025 @cindex cross compiling
7026 @cindex specifying machine version
7027 @cindex specifying compiler version and target machine
7028 @cindex compiler version, specifying
7029 @cindex target machine, specifying
7031 The usual way to run GCC is to run the executable called @file{gcc}, or
7032 @file{<machine>-gcc} when cross-compiling, or
7033 @file{<machine>-gcc-<version>} to run a version other than the one that
7034 was installed last. Sometimes this is inconvenient, so GCC provides
7035 options that will switch to another cross-compiler or version.
7038 @item -b @var{machine}
7040 The argument @var{machine} specifies the target machine for compilation.
7042 The value to use for @var{machine} is the same as was specified as the
7043 machine type when configuring GCC as a cross-compiler. For
7044 example, if a cross-compiler was configured with @samp{configure
7045 arm-elf}, meaning to compile for an arm processor with elf binaries,
7046 then you would specify @option{-b arm-elf} to run that cross compiler.
7047 Because there are other options beginning with @option{-b}, the
7048 configuration must contain a hyphen.
7050 @item -V @var{version}
7052 The argument @var{version} specifies which version of GCC to run.
7053 This is useful when multiple versions are installed. For example,
7054 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7057 The @option{-V} and @option{-b} options work by running the
7058 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7059 use them if you can just run that directly.
7061 @node Submodel Options
7062 @section Hardware Models and Configurations
7063 @cindex submodel options
7064 @cindex specifying hardware config
7065 @cindex hardware models and configurations, specifying
7066 @cindex machine dependent options
7068 Earlier we discussed the standard option @option{-b} which chooses among
7069 different installed compilers for completely different target
7070 machines, such as VAX vs.@: 68000 vs.@: 80386.
7072 In addition, each of these target machine types can have its own
7073 special options, starting with @samp{-m}, to choose among various
7074 hardware models or configurations---for example, 68010 vs 68020,
7075 floating coprocessor or none. A single installed version of the
7076 compiler can compile for any model or configuration, according to the
7079 Some configurations of the compiler also support additional special
7080 options, usually for compatibility with other compilers on the same
7083 @c This list is ordered alphanumerically by subsection name.
7084 @c It should be the same order and spelling as these options are listed
7085 @c in Machine Dependent Options
7091 * Blackfin Options::
7095 * DEC Alpha Options::
7096 * DEC Alpha/VMS Options::
7100 * i386 and x86-64 Options::
7113 * RS/6000 and PowerPC Options::
7114 * S/390 and zSeries Options::
7117 * System V Options::
7118 * TMS320C3x/C4x Options::
7122 * Xstormy16 Options::
7128 @subsection ARC Options
7131 These options are defined for ARC implementations:
7136 Compile code for little endian mode. This is the default.
7140 Compile code for big endian mode.
7143 @opindex mmangle-cpu
7144 Prepend the name of the cpu to all public symbol names.
7145 In multiple-processor systems, there are many ARC variants with different
7146 instruction and register set characteristics. This flag prevents code
7147 compiled for one cpu to be linked with code compiled for another.
7148 No facility exists for handling variants that are ``almost identical''.
7149 This is an all or nothing option.
7151 @item -mcpu=@var{cpu}
7153 Compile code for ARC variant @var{cpu}.
7154 Which variants are supported depend on the configuration.
7155 All variants support @option{-mcpu=base}, this is the default.
7157 @item -mtext=@var{text-section}
7158 @itemx -mdata=@var{data-section}
7159 @itemx -mrodata=@var{readonly-data-section}
7163 Put functions, data, and readonly data in @var{text-section},
7164 @var{data-section}, and @var{readonly-data-section} respectively
7165 by default. This can be overridden with the @code{section} attribute.
7166 @xref{Variable Attributes}.
7171 @subsection ARM Options
7174 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7178 @item -mabi=@var{name}
7180 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7181 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7184 @opindex mapcs-frame
7185 Generate a stack frame that is compliant with the ARM Procedure Call
7186 Standard for all functions, even if this is not strictly necessary for
7187 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7188 with this option will cause the stack frames not to be generated for
7189 leaf functions. The default is @option{-mno-apcs-frame}.
7193 This is a synonym for @option{-mapcs-frame}.
7196 @c not currently implemented
7197 @item -mapcs-stack-check
7198 @opindex mapcs-stack-check
7199 Generate code to check the amount of stack space available upon entry to
7200 every function (that actually uses some stack space). If there is
7201 insufficient space available then either the function
7202 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7203 called, depending upon the amount of stack space required. The run time
7204 system is required to provide these functions. The default is
7205 @option{-mno-apcs-stack-check}, since this produces smaller code.
7207 @c not currently implemented
7209 @opindex mapcs-float
7210 Pass floating point arguments using the float point registers. This is
7211 one of the variants of the APCS@. This option is recommended if the
7212 target hardware has a floating point unit or if a lot of floating point
7213 arithmetic is going to be performed by the code. The default is
7214 @option{-mno-apcs-float}, since integer only code is slightly increased in
7215 size if @option{-mapcs-float} is used.
7217 @c not currently implemented
7218 @item -mapcs-reentrant
7219 @opindex mapcs-reentrant
7220 Generate reentrant, position independent code. The default is
7221 @option{-mno-apcs-reentrant}.
7224 @item -mthumb-interwork
7225 @opindex mthumb-interwork
7226 Generate code which supports calling between the ARM and Thumb
7227 instruction sets. Without this option the two instruction sets cannot
7228 be reliably used inside one program. The default is
7229 @option{-mno-thumb-interwork}, since slightly larger code is generated
7230 when @option{-mthumb-interwork} is specified.
7232 @item -mno-sched-prolog
7233 @opindex mno-sched-prolog
7234 Prevent the reordering of instructions in the function prolog, or the
7235 merging of those instruction with the instructions in the function's
7236 body. This means that all functions will start with a recognizable set
7237 of instructions (or in fact one of a choice from a small set of
7238 different function prologues), and this information can be used to
7239 locate the start if functions inside an executable piece of code. The
7240 default is @option{-msched-prolog}.
7243 @opindex mhard-float
7244 Generate output containing floating point instructions. This is the
7248 @opindex msoft-float
7249 Generate output containing library calls for floating point.
7250 @strong{Warning:} the requisite libraries are not available for all ARM
7251 targets. Normally the facilities of the machine's usual C compiler are
7252 used, but this cannot be done directly in cross-compilation. You must make
7253 your own arrangements to provide suitable library functions for
7256 @option{-msoft-float} changes the calling convention in the output file;
7257 therefore, it is only useful if you compile @emph{all} of a program with
7258 this option. In particular, you need to compile @file{libgcc.a}, the
7259 library that comes with GCC, with @option{-msoft-float} in order for
7262 @item -mfloat-abi=@var{name}
7264 Specifies which ABI to use for floating point values. Permissible values
7265 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7267 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7268 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7269 of floating point instructions, but still uses the soft-float calling
7272 @item -mlittle-endian
7273 @opindex mlittle-endian
7274 Generate code for a processor running in little-endian mode. This is
7275 the default for all standard configurations.
7278 @opindex mbig-endian
7279 Generate code for a processor running in big-endian mode; the default is
7280 to compile code for a little-endian processor.
7282 @item -mwords-little-endian
7283 @opindex mwords-little-endian
7284 This option only applies when generating code for big-endian processors.
7285 Generate code for a little-endian word order but a big-endian byte
7286 order. That is, a byte order of the form @samp{32107654}. Note: this
7287 option should only be used if you require compatibility with code for
7288 big-endian ARM processors generated by versions of the compiler prior to
7291 @item -mcpu=@var{name}
7293 This specifies the name of the target ARM processor. GCC uses this name
7294 to determine what kind of instructions it can emit when generating
7295 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7296 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7297 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7298 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7299 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7300 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7301 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7302 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7303 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7304 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7305 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7306 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7307 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7308 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7311 @itemx -mtune=@var{name}
7313 This option is very similar to the @option{-mcpu=} option, except that
7314 instead of specifying the actual target processor type, and hence
7315 restricting which instructions can be used, it specifies that GCC should
7316 tune the performance of the code as if the target were of the type
7317 specified in this option, but still choosing the instructions that it
7318 will generate based on the cpu specified by a @option{-mcpu=} option.
7319 For some ARM implementations better performance can be obtained by using
7322 @item -march=@var{name}
7324 This specifies the name of the target ARM architecture. GCC uses this
7325 name to determine what kind of instructions it can emit when generating
7326 assembly code. This option can be used in conjunction with or instead
7327 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7328 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7329 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7330 @samp{iwmmxt}, @samp{ep9312}.
7332 @item -mfpu=@var{name}
7333 @itemx -mfpe=@var{number}
7334 @itemx -mfp=@var{number}
7338 This specifies what floating point hardware (or hardware emulation) is
7339 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7340 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7341 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7342 with older versions of GCC@.
7344 If @option{-msoft-float} is specified this specifies the format of
7345 floating point values.
7347 @item -mstructure-size-boundary=@var{n}
7348 @opindex mstructure-size-boundary
7349 The size of all structures and unions will be rounded up to a multiple
7350 of the number of bits set by this option. Permissible values are 8, 32
7351 and 64. The default value varies for different toolchains. For the COFF
7352 targeted toolchain the default value is 8. A value of 64 is only allowed
7353 if the underlying ABI supports it.
7355 Specifying the larger number can produce faster, more efficient code, but
7356 can also increase the size of the program. Different values are potentially
7357 incompatible. Code compiled with one value cannot necessarily expect to
7358 work with code or libraries compiled with another value, if they exchange
7359 information using structures or unions.
7361 @item -mabort-on-noreturn
7362 @opindex mabort-on-noreturn
7363 Generate a call to the function @code{abort} at the end of a
7364 @code{noreturn} function. It will be executed if the function tries to
7368 @itemx -mno-long-calls
7369 @opindex mlong-calls
7370 @opindex mno-long-calls
7371 Tells the compiler to perform function calls by first loading the
7372 address of the function into a register and then performing a subroutine
7373 call on this register. This switch is needed if the target function
7374 will lie outside of the 64 megabyte addressing range of the offset based
7375 version of subroutine call instruction.
7377 Even if this switch is enabled, not all function calls will be turned
7378 into long calls. The heuristic is that static functions, functions
7379 which have the @samp{short-call} attribute, functions that are inside
7380 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7381 definitions have already been compiled within the current compilation
7382 unit, will not be turned into long calls. The exception to this rule is
7383 that weak function definitions, functions with the @samp{long-call}
7384 attribute or the @samp{section} attribute, and functions that are within
7385 the scope of a @samp{#pragma long_calls} directive, will always be
7386 turned into long calls.
7388 This feature is not enabled by default. Specifying
7389 @option{-mno-long-calls} will restore the default behavior, as will
7390 placing the function calls within the scope of a @samp{#pragma
7391 long_calls_off} directive. Note these switches have no effect on how
7392 the compiler generates code to handle function calls via function
7395 @item -mnop-fun-dllimport
7396 @opindex mnop-fun-dllimport
7397 Disable support for the @code{dllimport} attribute.
7399 @item -msingle-pic-base
7400 @opindex msingle-pic-base
7401 Treat the register used for PIC addressing as read-only, rather than
7402 loading it in the prologue for each function. The run-time system is
7403 responsible for initializing this register with an appropriate value
7404 before execution begins.
7406 @item -mpic-register=@var{reg}
7407 @opindex mpic-register
7408 Specify the register to be used for PIC addressing. The default is R10
7409 unless stack-checking is enabled, when R9 is used.
7411 @item -mcirrus-fix-invalid-insns
7412 @opindex mcirrus-fix-invalid-insns
7413 @opindex mno-cirrus-fix-invalid-insns
7414 Insert NOPs into the instruction stream to in order to work around
7415 problems with invalid Maverick instruction combinations. This option
7416 is only valid if the @option{-mcpu=ep9312} option has been used to
7417 enable generation of instructions for the Cirrus Maverick floating
7418 point co-processor. This option is not enabled by default, since the
7419 problem is only present in older Maverick implementations. The default
7420 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7423 @item -mpoke-function-name
7424 @opindex mpoke-function-name
7425 Write the name of each function into the text section, directly
7426 preceding the function prologue. The generated code is similar to this:
7430 .ascii "arm_poke_function_name", 0
7433 .word 0xff000000 + (t1 - t0)
7434 arm_poke_function_name
7436 stmfd sp!, @{fp, ip, lr, pc@}
7440 When performing a stack backtrace, code can inspect the value of
7441 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7442 location @code{pc - 12} and the top 8 bits are set, then we know that
7443 there is a function name embedded immediately preceding this location
7444 and has length @code{((pc[-3]) & 0xff000000)}.
7448 Generate code for the 16-bit Thumb instruction set. The default is to
7449 use the 32-bit ARM instruction set.
7452 @opindex mtpcs-frame
7453 Generate a stack frame that is compliant with the Thumb Procedure Call
7454 Standard for all non-leaf functions. (A leaf function is one that does
7455 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7457 @item -mtpcs-leaf-frame
7458 @opindex mtpcs-leaf-frame
7459 Generate a stack frame that is compliant with the Thumb Procedure Call
7460 Standard for all leaf functions. (A leaf function is one that does
7461 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7463 @item -mcallee-super-interworking
7464 @opindex mcallee-super-interworking
7465 Gives all externally visible functions in the file being compiled an ARM
7466 instruction set header which switches to Thumb mode before executing the
7467 rest of the function. This allows these functions to be called from
7468 non-interworking code.
7470 @item -mcaller-super-interworking
7471 @opindex mcaller-super-interworking
7472 Allows calls via function pointers (including virtual functions) to
7473 execute correctly regardless of whether the target code has been
7474 compiled for interworking or not. There is a small overhead in the cost
7475 of executing a function pointer if this option is enabled.
7477 @item -mtp=@var{name}
7479 Specify the access model for the thread local storage pointer. The valid
7480 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7481 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7482 (supported in the arm6k architecture), and @option{auto}, which uses the
7483 best available method for the selected processor. The default setting is
7489 @subsection AVR Options
7492 These options are defined for AVR implementations:
7495 @item -mmcu=@var{mcu}
7497 Specify ATMEL AVR instruction set or MCU type.
7499 Instruction set avr1 is for the minimal AVR core, not supported by the C
7500 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7501 attiny11, attiny12, attiny15, attiny28).
7503 Instruction set avr2 (default) is for the classic AVR core with up to
7504 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7505 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7506 at90c8534, at90s8535).
7508 Instruction set avr3 is for the classic AVR core with up to 128K program
7509 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7511 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7512 memory space (MCU types: atmega8, atmega83, atmega85).
7514 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7515 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7516 atmega64, atmega128, at43usb355, at94k).
7520 Output instruction sizes to the asm file.
7522 @item -minit-stack=@var{N}
7523 @opindex minit-stack
7524 Specify the initial stack address, which may be a symbol or numeric value,
7525 @samp{__stack} is the default.
7527 @item -mno-interrupts
7528 @opindex mno-interrupts
7529 Generated code is not compatible with hardware interrupts.
7530 Code size will be smaller.
7532 @item -mcall-prologues
7533 @opindex mcall-prologues
7534 Functions prologues/epilogues expanded as call to appropriate
7535 subroutines. Code size will be smaller.
7537 @item -mno-tablejump
7538 @opindex mno-tablejump
7539 Do not generate tablejump insns which sometimes increase code size.
7542 @opindex mtiny-stack
7543 Change only the low 8 bits of the stack pointer.
7547 Assume int to be 8 bit integer. This affects the sizes of all types: A
7548 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7549 and long long will be 4 bytes. Please note that this option does not
7550 comply to the C standards, but it will provide you with smaller code
7554 @node Blackfin Options
7555 @subsection Blackfin Options
7556 @cindex Blackfin Options
7559 @item -momit-leaf-frame-pointer
7560 @opindex momit-leaf-frame-pointer
7561 Don't keep the frame pointer in a register for leaf functions. This
7562 avoids the instructions to save, set up and restore frame pointers and
7563 makes an extra register available in leaf functions. The option
7564 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7565 which might make debugging harder.
7567 @item -mspecld-anomaly
7568 @opindex mspecld-anomaly
7569 When enabled, the compiler will ensure that the generated code does not
7570 contain speculative loads after jump instructions. This option is enabled
7573 @item -mno-specld-anomaly
7574 @opindex mno-specld-anomaly
7575 Don't generate extra code to prevent speculative loads from occurring.
7577 @item -mcsync-anomaly
7578 @opindex mcsync-anomaly
7579 When enabled, the compiler will ensure that the generated code does not
7580 contain CSYNC or SSYNC instructions too soon after conditional branches.
7581 This option is enabled by default.
7583 @item -mno-csync-anomaly
7584 @opindex mno-csync-anomaly
7585 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7586 occurring too soon after a conditional branch.
7590 When enabled, the compiler is free to take advantage of the knowledge that
7591 the entire program fits into the low 64k of memory.
7594 @opindex mno-low-64k
7595 Assume that the program is arbitrarily large. This is the default.
7597 @item -mid-shared-library
7598 @opindex mid-shared-library
7599 Generate code that supports shared libraries via the library ID method.
7600 This allows for execute in place and shared libraries in an environment
7601 without virtual memory management. This option implies @option{-fPIC}.
7603 @item -mno-id-shared-library
7604 @opindex mno-id-shared-library
7605 Generate code that doesn't assume ID based shared libraries are being used.
7606 This is the default.
7608 @item -mshared-library-id=n
7609 @opindex mshared-library-id
7610 Specified the identification number of the ID based shared library being
7611 compiled. Specifying a value of 0 will generate more compact code, specifying
7612 other values will force the allocation of that number to the current
7613 library but is no more space or time efficient than omitting this option.
7616 @itemx -mno-long-calls
7617 @opindex mlong-calls
7618 @opindex mno-long-calls
7619 Tells the compiler to perform function calls by first loading the
7620 address of the function into a register and then performing a subroutine
7621 call on this register. This switch is needed if the target function
7622 will lie outside of the 24 bit addressing range of the offset based
7623 version of subroutine call instruction.
7625 This feature is not enabled by default. Specifying
7626 @option{-mno-long-calls} will restore the default behavior. Note these
7627 switches have no effect on how the compiler generates code to handle
7628 function calls via function pointers.
7632 @subsection CRIS Options
7633 @cindex CRIS Options
7635 These options are defined specifically for the CRIS ports.
7638 @item -march=@var{architecture-type}
7639 @itemx -mcpu=@var{architecture-type}
7642 Generate code for the specified architecture. The choices for
7643 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7644 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7645 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7648 @item -mtune=@var{architecture-type}
7650 Tune to @var{architecture-type} everything applicable about the generated
7651 code, except for the ABI and the set of available instructions. The
7652 choices for @var{architecture-type} are the same as for
7653 @option{-march=@var{architecture-type}}.
7655 @item -mmax-stack-frame=@var{n}
7656 @opindex mmax-stack-frame
7657 Warn when the stack frame of a function exceeds @var{n} bytes.
7659 @item -melinux-stacksize=@var{n}
7660 @opindex melinux-stacksize
7661 Only available with the @samp{cris-axis-aout} target. Arranges for
7662 indications in the program to the kernel loader that the stack of the
7663 program should be set to @var{n} bytes.
7669 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7670 @option{-march=v3} and @option{-march=v8} respectively.
7672 @item -mmul-bug-workaround
7673 @itemx -mno-mul-bug-workaround
7674 @opindex mmul-bug-workaround
7675 @opindex mno-mul-bug-workaround
7676 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7677 models where it applies. This option is active by default.
7681 Enable CRIS-specific verbose debug-related information in the assembly
7682 code. This option also has the effect to turn off the @samp{#NO_APP}
7683 formatted-code indicator to the assembler at the beginning of the
7688 Do not use condition-code results from previous instruction; always emit
7689 compare and test instructions before use of condition codes.
7691 @item -mno-side-effects
7692 @opindex mno-side-effects
7693 Do not emit instructions with side-effects in addressing modes other than
7697 @itemx -mno-stack-align
7699 @itemx -mno-data-align
7700 @itemx -mconst-align
7701 @itemx -mno-const-align
7702 @opindex mstack-align
7703 @opindex mno-stack-align
7704 @opindex mdata-align
7705 @opindex mno-data-align
7706 @opindex mconst-align
7707 @opindex mno-const-align
7708 These options (no-options) arranges (eliminate arrangements) for the
7709 stack-frame, individual data and constants to be aligned for the maximum
7710 single data access size for the chosen CPU model. The default is to
7711 arrange for 32-bit alignment. ABI details such as structure layout are
7712 not affected by these options.
7720 Similar to the stack- data- and const-align options above, these options
7721 arrange for stack-frame, writable data and constants to all be 32-bit,
7722 16-bit or 8-bit aligned. The default is 32-bit alignment.
7724 @item -mno-prologue-epilogue
7725 @itemx -mprologue-epilogue
7726 @opindex mno-prologue-epilogue
7727 @opindex mprologue-epilogue
7728 With @option{-mno-prologue-epilogue}, the normal function prologue and
7729 epilogue that sets up the stack-frame are omitted and no return
7730 instructions or return sequences are generated in the code. Use this
7731 option only together with visual inspection of the compiled code: no
7732 warnings or errors are generated when call-saved registers must be saved,
7733 or storage for local variable needs to be allocated.
7739 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7740 instruction sequences that load addresses for functions from the PLT part
7741 of the GOT rather than (traditional on other architectures) calls to the
7742 PLT@. The default is @option{-mgotplt}.
7746 Legacy no-op option only recognized with the cris-axis-aout target.
7750 Legacy no-op option only recognized with the cris-axis-elf and
7751 cris-axis-linux-gnu targets.
7755 Only recognized with the cris-axis-aout target, where it selects a
7756 GNU/linux-like multilib, include files and instruction set for
7761 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7765 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7766 to link with input-output functions from a simulator library. Code,
7767 initialized data and zero-initialized data are allocated consecutively.
7771 Like @option{-sim}, but pass linker options to locate initialized data at
7772 0x40000000 and zero-initialized data at 0x80000000.
7776 @subsection CRX Options
7779 These options are defined specifically for the CRX ports.
7785 Enable the use of multiply-accumulate instructions. Disabled by default.
7789 Push instructions will be used to pass outgoing arguments when functions
7790 are called. Enabled by default.
7793 @node Darwin Options
7794 @subsection Darwin Options
7795 @cindex Darwin options
7797 These options are defined for all architectures running the Darwin operating
7800 FSF GCC on Darwin does not create ``fat'' object files; it will create
7801 an object file for the single architecture that it was built to
7802 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7803 @option{-arch} options are used; it does so by running the compiler or
7804 linker multiple times and joining the results together with
7807 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7808 @samp{i686}) is determined by the flags that specify the ISA
7809 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7810 @option{-force_cpusubtype_ALL} option can be used to override this.
7812 The Darwin tools vary in their behavior when presented with an ISA
7813 mismatch. The assembler, @file{as}, will only permit instructions to
7814 be used that are valid for the subtype of the file it is generating,
7815 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7816 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7817 and print an error if asked to create a shared library with a less
7818 restrictive subtype than its input files (for instance, trying to put
7819 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7820 for executables, @file{ld}, will quietly give the executable the most
7821 restrictive subtype of any of its input files.
7826 Add the framework directory @var{dir} to the head of the list of
7827 directories to be searched for header files. These directories are
7828 interleaved with those specified by @option{-I} options and are
7829 scanned in a left-to-right order.
7831 A framework directory is a directory with frameworks in it. A
7832 framework is a directory with a @samp{"Headers"} and/or
7833 @samp{"PrivateHeaders"} directory contained directly in it that ends
7834 in @samp{".framework"}. The name of a framework is the name of this
7835 directory excluding the @samp{".framework"}. Headers associated with
7836 the framework are found in one of those two directories, with
7837 @samp{"Headers"} being searched first. A subframework is a framework
7838 directory that is in a framework's @samp{"Frameworks"} directory.
7839 Includes of subframework headers can only appear in a header of a
7840 framework that contains the subframework, or in a sibling subframework
7841 header. Two subframeworks are siblings if they occur in the same
7842 framework. A subframework should not have the same name as a
7843 framework, a warning will be issued if this is violated. Currently a
7844 subframework cannot have subframeworks, in the future, the mechanism
7845 may be extended to support this. The standard frameworks can be found
7846 in @samp{"/System/Library/Frameworks"} and
7847 @samp{"/Library/Frameworks"}. An example include looks like
7848 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7849 the name of the framework and header.h is found in the
7850 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7854 Emit debugging information for symbols that are used. For STABS
7855 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7856 This is by default ON@.
7860 Emit debugging information for all symbols and types.
7862 @item -mmacosx-version-min=@var{version}
7863 The earliest version of MacOS X that this executable will run on
7864 is @var{version}. Typical values of @var{version} include @code{10.1},
7865 @code{10.2}, and @code{10.3.9}.
7867 The default for this option is to make choices that seem to be most
7870 @item -mone-byte-bool
7871 @opindex -mone-byte-bool
7872 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
7873 By default @samp{sizeof(bool)} is @samp{4} when compiling for
7874 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
7875 option has no effect on x86.
7877 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
7878 to generate code that is not binary compatible with code generated
7879 without that switch. Using this switch may require recompiling all
7880 other modules in a program, including system libraries. Use this
7881 switch to conform to a non-default data model.
7883 @item -mfix-and-continue
7884 @itemx -ffix-and-continue
7885 @itemx -findirect-data
7886 @opindex mfix-and-continue
7887 @opindex ffix-and-continue
7888 @opindex findirect-data
7889 Generate code suitable for fast turn around development. Needed to
7890 enable gdb to dynamically load @code{.o} files into already running
7891 programs. @option{-findirect-data} and @option{-ffix-and-continue}
7892 are provided for backwards compatibility.
7896 Loads all members of static archive libraries.
7897 See man ld(1) for more information.
7899 @item -arch_errors_fatal
7900 @opindex arch_errors_fatal
7901 Cause the errors having to do with files that have the wrong architecture
7905 @opindex bind_at_load
7906 Causes the output file to be marked such that the dynamic linker will
7907 bind all undefined references when the file is loaded or launched.
7911 Produce a Mach-o bundle format file.
7912 See man ld(1) for more information.
7914 @item -bundle_loader @var{executable}
7915 @opindex bundle_loader
7916 This option specifies the @var{executable} that will be loading the build
7917 output file being linked. See man ld(1) for more information.
7920 @opindex -dynamiclib
7921 When passed this option, GCC will produce a dynamic library instead of
7922 an executable when linking, using the Darwin @file{libtool} command.
7924 @item -force_cpusubtype_ALL
7925 @opindex -force_cpusubtype_ALL
7926 This causes GCC's output file to have the @var{ALL} subtype, instead of
7927 one controlled by the @option{-mcpu} or @option{-march} option.
7929 @item -allowable_client @var{client_name}
7931 @itemx -compatibility_version
7932 @itemx -current_version
7934 @itemx -dependency-file
7936 @itemx -dylinker_install_name
7938 @itemx -exported_symbols_list
7940 @itemx -flat_namespace
7941 @itemx -force_flat_namespace
7942 @itemx -headerpad_max_install_names
7945 @itemx -install_name
7946 @itemx -keep_private_externs
7947 @itemx -multi_module
7948 @itemx -multiply_defined
7949 @itemx -multiply_defined_unused
7951 @itemx -no_dead_strip_inits_and_terms
7952 @itemx -nofixprebinding
7955 @itemx -noseglinkedit
7956 @itemx -pagezero_size
7958 @itemx -prebind_all_twolevel_modules
7959 @itemx -private_bundle
7960 @itemx -read_only_relocs
7962 @itemx -sectobjectsymbols
7966 @itemx -sectobjectsymbols
7969 @itemx -segs_read_only_addr
7970 @itemx -segs_read_write_addr
7971 @itemx -seg_addr_table
7972 @itemx -seg_addr_table_filename
7975 @itemx -segs_read_only_addr
7976 @itemx -segs_read_write_addr
7977 @itemx -single_module
7980 @itemx -sub_umbrella
7981 @itemx -twolevel_namespace
7984 @itemx -unexported_symbols_list
7985 @itemx -weak_reference_mismatches
7988 @opindex allowable_client
7989 @opindex client_name
7990 @opindex compatibility_version
7991 @opindex current_version
7993 @opindex dependency-file
7995 @opindex dylinker_install_name
7997 @opindex exported_symbols_list
7999 @opindex flat_namespace
8000 @opindex force_flat_namespace
8001 @opindex headerpad_max_install_names
8004 @opindex install_name
8005 @opindex keep_private_externs
8006 @opindex multi_module
8007 @opindex multiply_defined
8008 @opindex multiply_defined_unused
8010 @opindex no_dead_strip_inits_and_terms
8011 @opindex nofixprebinding
8012 @opindex nomultidefs
8014 @opindex noseglinkedit
8015 @opindex pagezero_size
8017 @opindex prebind_all_twolevel_modules
8018 @opindex private_bundle
8019 @opindex read_only_relocs
8021 @opindex sectobjectsymbols
8025 @opindex sectobjectsymbols
8028 @opindex segs_read_only_addr
8029 @opindex segs_read_write_addr
8030 @opindex seg_addr_table
8031 @opindex seg_addr_table_filename
8032 @opindex seglinkedit
8034 @opindex segs_read_only_addr
8035 @opindex segs_read_write_addr
8036 @opindex single_module
8038 @opindex sub_library
8039 @opindex sub_umbrella
8040 @opindex twolevel_namespace
8043 @opindex unexported_symbols_list
8044 @opindex weak_reference_mismatches
8045 @opindex whatsloaded
8047 These options are passed to the Darwin linker. The Darwin linker man page
8048 describes them in detail.
8051 @node DEC Alpha Options
8052 @subsection DEC Alpha Options
8054 These @samp{-m} options are defined for the DEC Alpha implementations:
8057 @item -mno-soft-float
8059 @opindex mno-soft-float
8060 @opindex msoft-float
8061 Use (do not use) the hardware floating-point instructions for
8062 floating-point operations. When @option{-msoft-float} is specified,
8063 functions in @file{libgcc.a} will be used to perform floating-point
8064 operations. Unless they are replaced by routines that emulate the
8065 floating-point operations, or compiled in such a way as to call such
8066 emulations routines, these routines will issue floating-point
8067 operations. If you are compiling for an Alpha without floating-point
8068 operations, you must ensure that the library is built so as not to call
8071 Note that Alpha implementations without floating-point operations are
8072 required to have floating-point registers.
8077 @opindex mno-fp-regs
8078 Generate code that uses (does not use) the floating-point register set.
8079 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8080 register set is not used, floating point operands are passed in integer
8081 registers as if they were integers and floating-point results are passed
8082 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8083 so any function with a floating-point argument or return value called by code
8084 compiled with @option{-mno-fp-regs} must also be compiled with that
8087 A typical use of this option is building a kernel that does not use,
8088 and hence need not save and restore, any floating-point registers.
8092 The Alpha architecture implements floating-point hardware optimized for
8093 maximum performance. It is mostly compliant with the IEEE floating
8094 point standard. However, for full compliance, software assistance is
8095 required. This option generates code fully IEEE compliant code
8096 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8097 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8098 defined during compilation. The resulting code is less efficient but is
8099 able to correctly support denormalized numbers and exceptional IEEE
8100 values such as not-a-number and plus/minus infinity. Other Alpha
8101 compilers call this option @option{-ieee_with_no_inexact}.
8103 @item -mieee-with-inexact
8104 @opindex mieee-with-inexact
8105 This is like @option{-mieee} except the generated code also maintains
8106 the IEEE @var{inexact-flag}. Turning on this option causes the
8107 generated code to implement fully-compliant IEEE math. In addition to
8108 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8109 macro. On some Alpha implementations the resulting code may execute
8110 significantly slower than the code generated by default. Since there is
8111 very little code that depends on the @var{inexact-flag}, you should
8112 normally not specify this option. Other Alpha compilers call this
8113 option @option{-ieee_with_inexact}.
8115 @item -mfp-trap-mode=@var{trap-mode}
8116 @opindex mfp-trap-mode
8117 This option controls what floating-point related traps are enabled.
8118 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8119 The trap mode can be set to one of four values:
8123 This is the default (normal) setting. The only traps that are enabled
8124 are the ones that cannot be disabled in software (e.g., division by zero
8128 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8132 Like @samp{su}, but the instructions are marked to be safe for software
8133 completion (see Alpha architecture manual for details).
8136 Like @samp{su}, but inexact traps are enabled as well.
8139 @item -mfp-rounding-mode=@var{rounding-mode}
8140 @opindex mfp-rounding-mode
8141 Selects the IEEE rounding mode. Other Alpha compilers call this option
8142 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8147 Normal IEEE rounding mode. Floating point numbers are rounded towards
8148 the nearest machine number or towards the even machine number in case
8152 Round towards minus infinity.
8155 Chopped rounding mode. Floating point numbers are rounded towards zero.
8158 Dynamic rounding mode. A field in the floating point control register
8159 (@var{fpcr}, see Alpha architecture reference manual) controls the
8160 rounding mode in effect. The C library initializes this register for
8161 rounding towards plus infinity. Thus, unless your program modifies the
8162 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8165 @item -mtrap-precision=@var{trap-precision}
8166 @opindex mtrap-precision
8167 In the Alpha architecture, floating point traps are imprecise. This
8168 means without software assistance it is impossible to recover from a
8169 floating trap and program execution normally needs to be terminated.
8170 GCC can generate code that can assist operating system trap handlers
8171 in determining the exact location that caused a floating point trap.
8172 Depending on the requirements of an application, different levels of
8173 precisions can be selected:
8177 Program precision. This option is the default and means a trap handler
8178 can only identify which program caused a floating point exception.
8181 Function precision. The trap handler can determine the function that
8182 caused a floating point exception.
8185 Instruction precision. The trap handler can determine the exact
8186 instruction that caused a floating point exception.
8189 Other Alpha compilers provide the equivalent options called
8190 @option{-scope_safe} and @option{-resumption_safe}.
8192 @item -mieee-conformant
8193 @opindex mieee-conformant
8194 This option marks the generated code as IEEE conformant. You must not
8195 use this option unless you also specify @option{-mtrap-precision=i} and either
8196 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8197 is to emit the line @samp{.eflag 48} in the function prologue of the
8198 generated assembly file. Under DEC Unix, this has the effect that
8199 IEEE-conformant math library routines will be linked in.
8201 @item -mbuild-constants
8202 @opindex mbuild-constants
8203 Normally GCC examines a 32- or 64-bit integer constant to
8204 see if it can construct it from smaller constants in two or three
8205 instructions. If it cannot, it will output the constant as a literal and
8206 generate code to load it from the data segment at runtime.
8208 Use this option to require GCC to construct @emph{all} integer constants
8209 using code, even if it takes more instructions (the maximum is six).
8211 You would typically use this option to build a shared library dynamic
8212 loader. Itself a shared library, it must relocate itself in memory
8213 before it can find the variables and constants in its own data segment.
8219 Select whether to generate code to be assembled by the vendor-supplied
8220 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8238 Indicate whether GCC should generate code to use the optional BWX,
8239 CIX, FIX and MAX instruction sets. The default is to use the instruction
8240 sets supported by the CPU type specified via @option{-mcpu=} option or that
8241 of the CPU on which GCC was built if none was specified.
8246 @opindex mfloat-ieee
8247 Generate code that uses (does not use) VAX F and G floating point
8248 arithmetic instead of IEEE single and double precision.
8250 @item -mexplicit-relocs
8251 @itemx -mno-explicit-relocs
8252 @opindex mexplicit-relocs
8253 @opindex mno-explicit-relocs
8254 Older Alpha assemblers provided no way to generate symbol relocations
8255 except via assembler macros. Use of these macros does not allow
8256 optimal instruction scheduling. GNU binutils as of version 2.12
8257 supports a new syntax that allows the compiler to explicitly mark
8258 which relocations should apply to which instructions. This option
8259 is mostly useful for debugging, as GCC detects the capabilities of
8260 the assembler when it is built and sets the default accordingly.
8264 @opindex msmall-data
8265 @opindex mlarge-data
8266 When @option{-mexplicit-relocs} is in effect, static data is
8267 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8268 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8269 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8270 16-bit relocations off of the @code{$gp} register. This limits the
8271 size of the small data area to 64KB, but allows the variables to be
8272 directly accessed via a single instruction.
8274 The default is @option{-mlarge-data}. With this option the data area
8275 is limited to just below 2GB@. Programs that require more than 2GB of
8276 data must use @code{malloc} or @code{mmap} to allocate the data in the
8277 heap instead of in the program's data segment.
8279 When generating code for shared libraries, @option{-fpic} implies
8280 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8284 @opindex msmall-text
8285 @opindex mlarge-text
8286 When @option{-msmall-text} is used, the compiler assumes that the
8287 code of the entire program (or shared library) fits in 4MB, and is
8288 thus reachable with a branch instruction. When @option{-msmall-data}
8289 is used, the compiler can assume that all local symbols share the
8290 same @code{$gp} value, and thus reduce the number of instructions
8291 required for a function call from 4 to 1.
8293 The default is @option{-mlarge-text}.
8295 @item -mcpu=@var{cpu_type}
8297 Set the instruction set and instruction scheduling parameters for
8298 machine type @var{cpu_type}. You can specify either the @samp{EV}
8299 style name or the corresponding chip number. GCC supports scheduling
8300 parameters for the EV4, EV5 and EV6 family of processors and will
8301 choose the default values for the instruction set from the processor
8302 you specify. If you do not specify a processor type, GCC will default
8303 to the processor on which the compiler was built.
8305 Supported values for @var{cpu_type} are
8311 Schedules as an EV4 and has no instruction set extensions.
8315 Schedules as an EV5 and has no instruction set extensions.
8319 Schedules as an EV5 and supports the BWX extension.
8324 Schedules as an EV5 and supports the BWX and MAX extensions.
8328 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8332 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8335 @item -mtune=@var{cpu_type}
8337 Set only the instruction scheduling parameters for machine type
8338 @var{cpu_type}. The instruction set is not changed.
8340 @item -mmemory-latency=@var{time}
8341 @opindex mmemory-latency
8342 Sets the latency the scheduler should assume for typical memory
8343 references as seen by the application. This number is highly
8344 dependent on the memory access patterns used by the application
8345 and the size of the external cache on the machine.
8347 Valid options for @var{time} are
8351 A decimal number representing clock cycles.
8357 The compiler contains estimates of the number of clock cycles for
8358 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8359 (also called Dcache, Scache, and Bcache), as well as to main memory.
8360 Note that L3 is only valid for EV5.
8365 @node DEC Alpha/VMS Options
8366 @subsection DEC Alpha/VMS Options
8368 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8371 @item -mvms-return-codes
8372 @opindex mvms-return-codes
8373 Return VMS condition codes from main. The default is to return POSIX
8374 style condition (e.g.@ error) codes.
8378 @subsection FRV Options
8385 Only use the first 32 general purpose registers.
8390 Use all 64 general purpose registers.
8395 Use only the first 32 floating point registers.
8400 Use all 64 floating point registers
8403 @opindex mhard-float
8405 Use hardware instructions for floating point operations.
8408 @opindex msoft-float
8410 Use library routines for floating point operations.
8415 Dynamically allocate condition code registers.
8420 Do not try to dynamically allocate condition code registers, only
8421 use @code{icc0} and @code{fcc0}.
8426 Change ABI to use double word insns.
8431 Do not use double word instructions.
8436 Use floating point double instructions.
8441 Do not use floating point double instructions.
8446 Use media instructions.
8451 Do not use media instructions.
8456 Use multiply and add/subtract instructions.
8461 Do not use multiply and add/subtract instructions.
8466 Select the FDPIC ABI, that uses function descriptors to represent
8467 pointers to functions. Without any PIC/PIE-related options, it
8468 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8469 assumes GOT entries and small data are within a 12-bit range from the
8470 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8471 are computed with 32 bits.
8474 @opindex minline-plt
8476 Enable inlining of PLT entries in function calls to functions that are
8477 not known to bind locally. It has no effect without @option{-mfdpic}.
8478 It's enabled by default if optimizing for speed and compiling for
8479 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8480 optimization option such as @option{-O3} or above is present in the
8486 Assume a large TLS segment when generating thread-local code.
8491 Do not assume a large TLS segment when generating thread-local code.
8496 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8497 that is known to be in read-only sections. It's enabled by default,
8498 except for @option{-fpic} or @option{-fpie}: even though it may help
8499 make the global offset table smaller, it trades 1 instruction for 4.
8500 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8501 one of which may be shared by multiple symbols, and it avoids the need
8502 for a GOT entry for the referenced symbol, so it's more likely to be a
8503 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8505 @item -multilib-library-pic
8506 @opindex multilib-library-pic
8508 Link with the (library, not FD) pic libraries. It's implied by
8509 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8510 @option{-fpic} without @option{-mfdpic}. You should never have to use
8516 Follow the EABI requirement of always creating a frame pointer whenever
8517 a stack frame is allocated. This option is enabled by default and can
8518 be disabled with @option{-mno-linked-fp}.
8521 @opindex mlong-calls
8523 Use indirect addressing to call functions outside the current
8524 compilation unit. This allows the functions to be placed anywhere
8525 within the 32-bit address space.
8527 @item -malign-labels
8528 @opindex malign-labels
8530 Try to align labels to an 8-byte boundary by inserting nops into the
8531 previous packet. This option only has an effect when VLIW packing
8532 is enabled. It doesn't create new packets; it merely adds nops to
8536 @opindex mlibrary-pic
8538 Generate position-independent EABI code.
8543 Use only the first four media accumulator registers.
8548 Use all eight media accumulator registers.
8553 Pack VLIW instructions.
8558 Do not pack VLIW instructions.
8563 Do not mark ABI switches in e_flags.
8568 Enable the use of conditional-move instructions (default).
8570 This switch is mainly for debugging the compiler and will likely be removed
8571 in a future version.
8573 @item -mno-cond-move
8574 @opindex mno-cond-move
8576 Disable the use of conditional-move instructions.
8578 This switch is mainly for debugging the compiler and will likely be removed
8579 in a future version.
8584 Enable the use of conditional set instructions (default).
8586 This switch is mainly for debugging the compiler and will likely be removed
8587 in a future version.
8592 Disable the use of conditional set instructions.
8594 This switch is mainly for debugging the compiler and will likely be removed
8595 in a future version.
8600 Enable the use of conditional execution (default).
8602 This switch is mainly for debugging the compiler and will likely be removed
8603 in a future version.
8605 @item -mno-cond-exec
8606 @opindex mno-cond-exec
8608 Disable the use of conditional execution.
8610 This switch is mainly for debugging the compiler and will likely be removed
8611 in a future version.
8614 @opindex mvliw-branch
8616 Run a pass to pack branches into VLIW instructions (default).
8618 This switch is mainly for debugging the compiler and will likely be removed
8619 in a future version.
8621 @item -mno-vliw-branch
8622 @opindex mno-vliw-branch
8624 Do not run a pass to pack branches into VLIW instructions.
8626 This switch is mainly for debugging the compiler and will likely be removed
8627 in a future version.
8629 @item -mmulti-cond-exec
8630 @opindex mmulti-cond-exec
8632 Enable optimization of @code{&&} and @code{||} in conditional execution
8635 This switch is mainly for debugging the compiler and will likely be removed
8636 in a future version.
8638 @item -mno-multi-cond-exec
8639 @opindex mno-multi-cond-exec
8641 Disable optimization of @code{&&} and @code{||} in conditional execution.
8643 This switch is mainly for debugging the compiler and will likely be removed
8644 in a future version.
8646 @item -mnested-cond-exec
8647 @opindex mnested-cond-exec
8649 Enable nested conditional execution optimizations (default).
8651 This switch is mainly for debugging the compiler and will likely be removed
8652 in a future version.
8654 @item -mno-nested-cond-exec
8655 @opindex mno-nested-cond-exec
8657 Disable nested conditional execution optimizations.
8659 This switch is mainly for debugging the compiler and will likely be removed
8660 in a future version.
8662 @item -moptimize-membar
8663 @opindex moptimize-membar
8665 This switch removes redundant @code{membar} instructions from the
8666 compiler generated code. It is enabled by default.
8668 @item -mno-optimize-membar
8669 @opindex mno-optimize-membar
8671 This switch disables the automatic removal of redundant @code{membar}
8672 instructions from the generated code.
8674 @item -mtomcat-stats
8675 @opindex mtomcat-stats
8677 Cause gas to print out tomcat statistics.
8679 @item -mcpu=@var{cpu}
8682 Select the processor type for which to generate code. Possible values are
8683 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8684 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8688 @node H8/300 Options
8689 @subsection H8/300 Options
8691 These @samp{-m} options are defined for the H8/300 implementations:
8696 Shorten some address references at link time, when possible; uses the
8697 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8698 ld, Using ld}, for a fuller description.
8702 Generate code for the H8/300H@.
8706 Generate code for the H8S@.
8710 Generate code for the H8S and H8/300H in the normal mode. This switch
8711 must be used either with @option{-mh} or @option{-ms}.
8715 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8719 Make @code{int} data 32 bits by default.
8723 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8724 The default for the H8/300H and H8S is to align longs and floats on 4
8726 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8727 This option has no effect on the H8/300.
8731 @subsection HPPA Options
8732 @cindex HPPA Options
8734 These @samp{-m} options are defined for the HPPA family of computers:
8737 @item -march=@var{architecture-type}
8739 Generate code for the specified architecture. The choices for
8740 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8741 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8742 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8743 architecture option for your machine. Code compiled for lower numbered
8744 architectures will run on higher numbered architectures, but not the
8748 @itemx -mpa-risc-1-1
8749 @itemx -mpa-risc-2-0
8750 @opindex mpa-risc-1-0
8751 @opindex mpa-risc-1-1
8752 @opindex mpa-risc-2-0
8753 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8756 @opindex mbig-switch
8757 Generate code suitable for big switch tables. Use this option only if
8758 the assembler/linker complain about out of range branches within a switch
8761 @item -mjump-in-delay
8762 @opindex mjump-in-delay
8763 Fill delay slots of function calls with unconditional jump instructions
8764 by modifying the return pointer for the function call to be the target
8765 of the conditional jump.
8767 @item -mdisable-fpregs
8768 @opindex mdisable-fpregs
8769 Prevent floating point registers from being used in any manner. This is
8770 necessary for compiling kernels which perform lazy context switching of
8771 floating point registers. If you use this option and attempt to perform
8772 floating point operations, the compiler will abort.
8774 @item -mdisable-indexing
8775 @opindex mdisable-indexing
8776 Prevent the compiler from using indexing address modes. This avoids some
8777 rather obscure problems when compiling MIG generated code under MACH@.
8779 @item -mno-space-regs
8780 @opindex mno-space-regs
8781 Generate code that assumes the target has no space registers. This allows
8782 GCC to generate faster indirect calls and use unscaled index address modes.
8784 Such code is suitable for level 0 PA systems and kernels.
8786 @item -mfast-indirect-calls
8787 @opindex mfast-indirect-calls
8788 Generate code that assumes calls never cross space boundaries. This
8789 allows GCC to emit code which performs faster indirect calls.
8791 This option will not work in the presence of shared libraries or nested
8794 @item -mfixed-range=@var{register-range}
8795 @opindex mfixed-range
8796 Generate code treating the given register range as fixed registers.
8797 A fixed register is one that the register allocator can not use. This is
8798 useful when compiling kernel code. A register range is specified as
8799 two registers separated by a dash. Multiple register ranges can be
8800 specified separated by a comma.
8802 @item -mlong-load-store
8803 @opindex mlong-load-store
8804 Generate 3-instruction load and store sequences as sometimes required by
8805 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8808 @item -mportable-runtime
8809 @opindex mportable-runtime
8810 Use the portable calling conventions proposed by HP for ELF systems.
8814 Enable the use of assembler directives only GAS understands.
8816 @item -mschedule=@var{cpu-type}
8818 Schedule code according to the constraints for the machine type
8819 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8820 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8821 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8822 proper scheduling option for your machine. The default scheduling is
8826 @opindex mlinker-opt
8827 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8828 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8829 linkers in which they give bogus error messages when linking some programs.
8832 @opindex msoft-float
8833 Generate output containing library calls for floating point.
8834 @strong{Warning:} the requisite libraries are not available for all HPPA
8835 targets. Normally the facilities of the machine's usual C compiler are
8836 used, but this cannot be done directly in cross-compilation. You must make
8837 your own arrangements to provide suitable library functions for
8838 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8839 does provide software floating point support.
8841 @option{-msoft-float} changes the calling convention in the output file;
8842 therefore, it is only useful if you compile @emph{all} of a program with
8843 this option. In particular, you need to compile @file{libgcc.a}, the
8844 library that comes with GCC, with @option{-msoft-float} in order for
8849 Generate the predefine, @code{_SIO}, for server IO@. The default is
8850 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8851 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
8852 options are available under HP-UX and HI-UX@.
8856 Use GNU ld specific options. This passes @option{-shared} to ld when
8857 building a shared library. It is the default when GCC is configured,
8858 explicitly or implicitly, with the GNU linker. This option does not
8859 have any affect on which ld is called, it only changes what parameters
8860 are passed to that ld. The ld that is called is determined by the
8861 @option{--with-ld} configure option, GCC's program search path, and
8862 finally by the user's @env{PATH}. The linker used by GCC can be printed
8863 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
8864 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
8868 Use HP ld specific options. This passes @option{-b} to ld when building
8869 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8870 links. It is the default when GCC is configured, explicitly or
8871 implicitly, with the HP linker. This option does not have any affect on
8872 which ld is called, it only changes what parameters are passed to that
8873 ld. The ld that is called is determined by the @option{--with-ld}
8874 configure option, GCC's program search path, and finally by the user's
8875 @env{PATH}. The linker used by GCC can be printed using @samp{which
8876 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
8877 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
8880 @opindex mno-long-calls
8881 Generate code that uses long call sequences. This ensures that a call
8882 is always able to reach linker generated stubs. The default is to generate
8883 long calls only when the distance from the call site to the beginning
8884 of the function or translation unit, as the case may be, exceeds a
8885 predefined limit set by the branch type being used. The limits for
8886 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8887 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8890 Distances are measured from the beginning of functions when using the
8891 @option{-ffunction-sections} option, or when using the @option{-mgas}
8892 and @option{-mno-portable-runtime} options together under HP-UX with
8895 It is normally not desirable to use this option as it will degrade
8896 performance. However, it may be useful in large applications,
8897 particularly when partial linking is used to build the application.
8899 The types of long calls used depends on the capabilities of the
8900 assembler and linker, and the type of code being generated. The
8901 impact on systems that support long absolute calls, and long pic
8902 symbol-difference or pc-relative calls should be relatively small.
8903 However, an indirect call is used on 32-bit ELF systems in pic code
8904 and it is quite long.
8906 @item -munix=@var{unix-std}
8908 Generate compiler predefines and select a startfile for the specified
8909 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
8910 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
8911 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
8912 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
8913 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
8916 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
8917 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
8918 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
8919 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
8920 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
8921 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
8923 It is @emph{important} to note that this option changes the interfaces
8924 for various library routines. It also affects the operational behavior
8925 of the C library. Thus, @emph{extreme} care is needed in using this
8928 Library code that is intended to operate with more than one UNIX
8929 standard must test, set and restore the variable @var{__xpg4_extended_mask}
8930 as appropriate. Most GNU software doesn't provide this capability.
8934 Suppress the generation of link options to search libdld.sl when the
8935 @option{-static} option is specified on HP-UX 10 and later.
8939 The HP-UX implementation of setlocale in libc has a dependency on
8940 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8941 when the @option{-static} option is specified, special link options
8942 are needed to resolve this dependency.
8944 On HP-UX 10 and later, the GCC driver adds the necessary options to
8945 link with libdld.sl when the @option{-static} option is specified.
8946 This causes the resulting binary to be dynamic. On the 64-bit port,
8947 the linkers generate dynamic binaries by default in any case. The
8948 @option{-nolibdld} option can be used to prevent the GCC driver from
8949 adding these link options.
8953 Add support for multithreading with the @dfn{dce thread} library
8954 under HP-UX@. This option sets flags for both the preprocessor and
8958 @node i386 and x86-64 Options
8959 @subsection Intel 386 and AMD x86-64 Options
8960 @cindex i386 Options
8961 @cindex x86-64 Options
8962 @cindex Intel 386 Options
8963 @cindex AMD x86-64 Options
8965 These @samp{-m} options are defined for the i386 and x86-64 family of
8969 @item -mtune=@var{cpu-type}
8971 Tune to @var{cpu-type} everything applicable about the generated code, except
8972 for the ABI and the set of available instructions. The choices for
8976 Original Intel's i386 CPU@.
8978 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
8980 Intel Pentium CPU with no MMX support.
8982 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8983 @item i686, pentiumpro
8984 Intel PentiumPro CPU@.
8986 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8987 @item pentium3, pentium3m
8988 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8991 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8992 support. Used by Centrino notebooks.
8993 @item pentium4, pentium4m
8994 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8996 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8999 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9000 SSE2 and SSE3 instruction set support.
9002 AMD K6 CPU with MMX instruction set support.
9004 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9005 @item athlon, athlon-tbird
9006 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9008 @item athlon-4, athlon-xp, athlon-mp
9009 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9010 instruction set support.
9011 @item k8, opteron, athlon64, athlon-fx
9012 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9013 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9015 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9018 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9019 instruction set support.
9021 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9022 implemented for this chip.)
9024 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9025 implemented for this chip.)
9028 While picking a specific @var{cpu-type} will schedule things appropriately
9029 for that particular chip, the compiler will not generate any code that
9030 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9033 @item -march=@var{cpu-type}
9035 Generate instructions for the machine type @var{cpu-type}. The choices
9036 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9037 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9039 @item -mcpu=@var{cpu-type}
9041 A deprecated synonym for @option{-mtune}.
9050 @opindex mpentiumpro
9051 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9052 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9053 These synonyms are deprecated.
9055 @item -mfpmath=@var{unit}
9057 Generate floating point arithmetics for selected unit @var{unit}. The choices
9062 Use the standard 387 floating point coprocessor present majority of chips and
9063 emulated otherwise. Code compiled with this option will run almost everywhere.
9064 The temporary results are computed in 80bit precision instead of precision
9065 specified by the type resulting in slightly different results compared to most
9066 of other chips. See @option{-ffloat-store} for more detailed description.
9068 This is the default choice for i386 compiler.
9071 Use scalar floating point instructions present in the SSE instruction set.
9072 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9073 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9074 instruction set supports only single precision arithmetics, thus the double and
9075 extended precision arithmetics is still done using 387. Later version, present
9076 only in Pentium4 and the future AMD x86-64 chips supports double precision
9079 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9080 or @option{-msse2} switches to enable SSE extensions and make this option
9081 effective. For the x86-64 compiler, these extensions are enabled by default.
9083 The resulting code should be considerably faster in the majority of cases and avoid
9084 the numerical instability problems of 387 code, but may break some existing
9085 code that expects temporaries to be 80bit.
9087 This is the default choice for the x86-64 compiler.
9090 Attempt to utilize both instruction sets at once. This effectively double the
9091 amount of available registers and on chips with separate execution units for
9092 387 and SSE the execution resources too. Use this option with care, as it is
9093 still experimental, because the GCC register allocator does not model separate
9094 functional units well resulting in instable performance.
9097 @item -masm=@var{dialect}
9098 @opindex masm=@var{dialect}
9099 Output asm instructions using selected @var{dialect}. Supported choices are
9100 @samp{intel} or @samp{att} (the default one).
9105 @opindex mno-ieee-fp
9106 Control whether or not the compiler uses IEEE floating point
9107 comparisons. These handle correctly the case where the result of a
9108 comparison is unordered.
9111 @opindex msoft-float
9112 Generate output containing library calls for floating point.
9113 @strong{Warning:} the requisite libraries are not part of GCC@.
9114 Normally the facilities of the machine's usual C compiler are used, but
9115 this can't be done directly in cross-compilation. You must make your
9116 own arrangements to provide suitable library functions for
9119 On machines where a function returns floating point results in the 80387
9120 register stack, some floating point opcodes may be emitted even if
9121 @option{-msoft-float} is used.
9123 @item -mno-fp-ret-in-387
9124 @opindex mno-fp-ret-in-387
9125 Do not use the FPU registers for return values of functions.
9127 The usual calling convention has functions return values of types
9128 @code{float} and @code{double} in an FPU register, even if there
9129 is no FPU@. The idea is that the operating system should emulate
9132 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9133 in ordinary CPU registers instead.
9135 @item -mno-fancy-math-387
9136 @opindex mno-fancy-math-387
9137 Some 387 emulators do not support the @code{sin}, @code{cos} and
9138 @code{sqrt} instructions for the 387. Specify this option to avoid
9139 generating those instructions. This option is the default on FreeBSD,
9140 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9141 indicates that the target cpu will always have an FPU and so the
9142 instruction will not need emulation. As of revision 2.6.1, these
9143 instructions are not generated unless you also use the
9144 @option{-funsafe-math-optimizations} switch.
9146 @item -malign-double
9147 @itemx -mno-align-double
9148 @opindex malign-double
9149 @opindex mno-align-double
9150 Control whether GCC aligns @code{double}, @code{long double}, and
9151 @code{long long} variables on a two word boundary or a one word
9152 boundary. Aligning @code{double} variables on a two word boundary will
9153 produce code that runs somewhat faster on a @samp{Pentium} at the
9154 expense of more memory.
9156 @strong{Warning:} if you use the @option{-malign-double} switch,
9157 structures containing the above types will be aligned differently than
9158 the published application binary interface specifications for the 386
9159 and will not be binary compatible with structures in code compiled
9160 without that switch.
9162 @item -m96bit-long-double
9163 @itemx -m128bit-long-double
9164 @opindex m96bit-long-double
9165 @opindex m128bit-long-double
9166 These switches control the size of @code{long double} type. The i386
9167 application binary interface specifies the size to be 96 bits,
9168 so @option{-m96bit-long-double} is the default in 32 bit mode.
9170 Modern architectures (Pentium and newer) would prefer @code{long double}
9171 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9172 conforming to the ABI, this would not be possible. So specifying a
9173 @option{-m128bit-long-double} will align @code{long double}
9174 to a 16 byte boundary by padding the @code{long double} with an additional
9177 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9178 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9180 Notice that neither of these options enable any extra precision over the x87
9181 standard of 80 bits for a @code{long double}.
9183 @strong{Warning:} if you override the default value for your target ABI, the
9184 structures and arrays containing @code{long double} variables will change
9185 their size as well as function calling convention for function taking
9186 @code{long double} will be modified. Hence they will not be binary
9187 compatible with arrays or structures in code compiled without that switch.
9189 @item -mmlarge-data-threshold=@var{number}
9190 @opindex mlarge-data-threshold=@var{number}
9191 When @option{-mcmodel=medium} is specified, the data greater than
9192 @var{threshold} are placed in large data section. This value must be the
9193 same across all object linked into the binary and defaults to 65535.
9196 @itemx -mno-svr3-shlib
9197 @opindex msvr3-shlib
9198 @opindex mno-svr3-shlib
9199 Control whether GCC places uninitialized local variables into the
9200 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9201 into @code{bss}. These options are meaningful only on System V Release 3.
9205 Use a different function-calling convention, in which functions that
9206 take a fixed number of arguments return with the @code{ret} @var{num}
9207 instruction, which pops their arguments while returning. This saves one
9208 instruction in the caller since there is no need to pop the arguments
9211 You can specify that an individual function is called with this calling
9212 sequence with the function attribute @samp{stdcall}. You can also
9213 override the @option{-mrtd} option by using the function attribute
9214 @samp{cdecl}. @xref{Function Attributes}.
9216 @strong{Warning:} this calling convention is incompatible with the one
9217 normally used on Unix, so you cannot use it if you need to call
9218 libraries compiled with the Unix compiler.
9220 Also, you must provide function prototypes for all functions that
9221 take variable numbers of arguments (including @code{printf});
9222 otherwise incorrect code will be generated for calls to those
9225 In addition, seriously incorrect code will result if you call a
9226 function with too many arguments. (Normally, extra arguments are
9227 harmlessly ignored.)
9229 @item -mregparm=@var{num}
9231 Control how many registers are used to pass integer arguments. By
9232 default, no registers are used to pass arguments, and at most 3
9233 registers can be used. You can control this behavior for a specific
9234 function by using the function attribute @samp{regparm}.
9235 @xref{Function Attributes}.
9237 @strong{Warning:} if you use this switch, and
9238 @var{num} is nonzero, then you must build all modules with the same
9239 value, including any libraries. This includes the system libraries and
9243 @opindex msseregparm
9244 Use SSE register passing conventions for float and double arguments
9245 and return values. You can control this behavior for a specific
9246 function by using the function attribute @samp{sseregparm}.
9247 @xref{Function Attributes}.
9249 @strong{Warning:} if you use this switch then you must build all
9250 modules with the same value, including any libraries. This includes
9251 the system libraries and startup modules.
9253 @item -mpreferred-stack-boundary=@var{num}
9254 @opindex mpreferred-stack-boundary
9255 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9256 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9257 the default is 4 (16 bytes or 128 bits), except when optimizing for code
9258 size (@option{-Os}), in which case the default is the minimum correct
9259 alignment (4 bytes for x86, and 8 bytes for x86-64).
9261 On Pentium and PentiumPro, @code{double} and @code{long double} values
9262 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9263 suffer significant run time performance penalties. On Pentium III, the
9264 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
9265 penalties if it is not 16 byte aligned.
9267 To ensure proper alignment of this values on the stack, the stack boundary
9268 must be as aligned as that required by any value stored on the stack.
9269 Further, every function must be generated such that it keeps the stack
9270 aligned. Thus calling a function compiled with a higher preferred
9271 stack boundary from a function compiled with a lower preferred stack
9272 boundary will most likely misalign the stack. It is recommended that
9273 libraries that use callbacks always use the default setting.
9275 This extra alignment does consume extra stack space, and generally
9276 increases code size. Code that is sensitive to stack space usage, such
9277 as embedded systems and operating system kernels, may want to reduce the
9278 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9296 These switches enable or disable the use of instructions in the MMX,
9297 SSE, SSE2 or 3DNow! extended instruction sets. These extensions are
9298 also available as built-in functions: see @ref{X86 Built-in Functions},
9299 for details of the functions enabled and disabled by these switches.
9301 To have SSE/SSE2 instructions generated automatically from floating-point
9302 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9304 These options will enable GCC to use these extended instructions in
9305 generated code, even without @option{-mfpmath=sse}. Applications which
9306 perform runtime CPU detection must compile separate files for each
9307 supported architecture, using the appropriate flags. In particular,
9308 the file containing the CPU detection code should be compiled without
9312 @itemx -mno-push-args
9314 @opindex mno-push-args
9315 Use PUSH operations to store outgoing parameters. This method is shorter
9316 and usually equally fast as method using SUB/MOV operations and is enabled
9317 by default. In some cases disabling it may improve performance because of
9318 improved scheduling and reduced dependencies.
9320 @item -maccumulate-outgoing-args
9321 @opindex maccumulate-outgoing-args
9322 If enabled, the maximum amount of space required for outgoing arguments will be
9323 computed in the function prologue. This is faster on most modern CPUs
9324 because of reduced dependencies, improved scheduling and reduced stack usage
9325 when preferred stack boundary is not equal to 2. The drawback is a notable
9326 increase in code size. This switch implies @option{-mno-push-args}.
9330 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9331 on thread-safe exception handling must compile and link all code with the
9332 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9333 @option{-D_MT}; when linking, it links in a special thread helper library
9334 @option{-lmingwthrd} which cleans up per thread exception handling data.
9336 @item -mno-align-stringops
9337 @opindex mno-align-stringops
9338 Do not align destination of inlined string operations. This switch reduces
9339 code size and improves performance in case the destination is already aligned,
9340 but GCC doesn't know about it.
9342 @item -minline-all-stringops
9343 @opindex minline-all-stringops
9344 By default GCC inlines string operations only when destination is known to be
9345 aligned at least to 4 byte boundary. This enables more inlining, increase code
9346 size, but may improve performance of code that depends on fast memcpy, strlen
9347 and memset for short lengths.
9349 @item -momit-leaf-frame-pointer
9350 @opindex momit-leaf-frame-pointer
9351 Don't keep the frame pointer in a register for leaf functions. This
9352 avoids the instructions to save, set up and restore frame pointers and
9353 makes an extra register available in leaf functions. The option
9354 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9355 which might make debugging harder.
9357 @item -mtls-direct-seg-refs
9358 @itemx -mno-tls-direct-seg-refs
9359 @opindex mtls-direct-seg-refs
9360 Controls whether TLS variables may be accessed with offsets from the
9361 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9362 or whether the thread base pointer must be added. Whether or not this
9363 is legal depends on the operating system, and whether it maps the
9364 segment to cover the entire TLS area.
9366 For systems that use GNU libc, the default is on.
9369 These @samp{-m} switches are supported in addition to the above
9370 on AMD x86-64 processors in 64-bit environments.
9377 Generate code for a 32-bit or 64-bit environment.
9378 The 32-bit environment sets int, long and pointer to 32 bits and
9379 generates code that runs on any i386 system.
9380 The 64-bit environment sets int to 32 bits and long and pointer
9381 to 64 bits and generates code for AMD's x86-64 architecture.
9384 @opindex no-red-zone
9385 Do not use a so called red zone for x86-64 code. The red zone is mandated
9386 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9387 stack pointer that will not be modified by signal or interrupt handlers
9388 and therefore can be used for temporary data without adjusting the stack
9389 pointer. The flag @option{-mno-red-zone} disables this red zone.
9391 @item -mcmodel=small
9392 @opindex mcmodel=small
9393 Generate code for the small code model: the program and its symbols must
9394 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9395 Programs can be statically or dynamically linked. This is the default
9398 @item -mcmodel=kernel
9399 @opindex mcmodel=kernel
9400 Generate code for the kernel code model. The kernel runs in the
9401 negative 2 GB of the address space.
9402 This model has to be used for Linux kernel code.
9404 @item -mcmodel=medium
9405 @opindex mcmodel=medium
9406 Generate code for the medium model: The program is linked in the lower 2
9407 GB of the address space but symbols can be located anywhere in the
9408 address space. Programs can be statically or dynamically linked, but
9409 building of shared libraries are not supported with the medium model.
9411 @item -mcmodel=large
9412 @opindex mcmodel=large
9413 Generate code for the large model: This model makes no assumptions
9414 about addresses and sizes of sections. Currently GCC does not implement
9419 @subsection IA-64 Options
9420 @cindex IA-64 Options
9422 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9426 @opindex mbig-endian
9427 Generate code for a big endian target. This is the default for HP-UX@.
9429 @item -mlittle-endian
9430 @opindex mlittle-endian
9431 Generate code for a little endian target. This is the default for AIX5
9438 Generate (or don't) code for the GNU assembler. This is the default.
9439 @c Also, this is the default if the configure option @option{--with-gnu-as}
9446 Generate (or don't) code for the GNU linker. This is the default.
9447 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9452 Generate code that does not use a global pointer register. The result
9453 is not position independent code, and violates the IA-64 ABI@.
9455 @item -mvolatile-asm-stop
9456 @itemx -mno-volatile-asm-stop
9457 @opindex mvolatile-asm-stop
9458 @opindex mno-volatile-asm-stop
9459 Generate (or don't) a stop bit immediately before and after volatile asm
9462 @item -mregister-names
9463 @itemx -mno-register-names
9464 @opindex mregister-names
9465 @opindex mno-register-names
9466 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9467 the stacked registers. This may make assembler output more readable.
9473 Disable (or enable) optimizations that use the small data section. This may
9474 be useful for working around optimizer bugs.
9477 @opindex mconstant-gp
9478 Generate code that uses a single constant global pointer value. This is
9479 useful when compiling kernel code.
9483 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9484 This is useful when compiling firmware code.
9486 @item -minline-float-divide-min-latency
9487 @opindex minline-float-divide-min-latency
9488 Generate code for inline divides of floating point values
9489 using the minimum latency algorithm.
9491 @item -minline-float-divide-max-throughput
9492 @opindex minline-float-divide-max-throughput
9493 Generate code for inline divides of floating point values
9494 using the maximum throughput algorithm.
9496 @item -minline-int-divide-min-latency
9497 @opindex minline-int-divide-min-latency
9498 Generate code for inline divides of integer values
9499 using the minimum latency algorithm.
9501 @item -minline-int-divide-max-throughput
9502 @opindex minline-int-divide-max-throughput
9503 Generate code for inline divides of integer values
9504 using the maximum throughput algorithm.
9506 @item -minline-sqrt-min-latency
9507 @opindex minline-sqrt-min-latency
9508 Generate code for inline square roots
9509 using the minimum latency algorithm.
9511 @item -minline-sqrt-max-throughput
9512 @opindex minline-sqrt-max-throughput
9513 Generate code for inline square roots
9514 using the maximum throughput algorithm.
9516 @item -mno-dwarf2-asm
9518 @opindex mno-dwarf2-asm
9519 @opindex mdwarf2-asm
9520 Don't (or do) generate assembler code for the DWARF2 line number debugging
9521 info. This may be useful when not using the GNU assembler.
9523 @item -mearly-stop-bits
9524 @itemx -mno-early-stop-bits
9525 @opindex mearly-stop-bits
9526 @opindex mno-early-stop-bits
9527 Allow stop bits to be placed earlier than immediately preceding the
9528 instruction that triggered the stop bit. This can improve instruction
9529 scheduling, but does not always do so.
9531 @item -mfixed-range=@var{register-range}
9532 @opindex mfixed-range
9533 Generate code treating the given register range as fixed registers.
9534 A fixed register is one that the register allocator can not use. This is
9535 useful when compiling kernel code. A register range is specified as
9536 two registers separated by a dash. Multiple register ranges can be
9537 specified separated by a comma.
9539 @item -mtls-size=@var{tls-size}
9541 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9544 @item -mtune=@var{cpu-type}
9546 Tune the instruction scheduling for a particular CPU, Valid values are
9547 itanium, itanium1, merced, itanium2, and mckinley.
9553 Add support for multithreading using the POSIX threads library. This
9554 option sets flags for both the preprocessor and linker. It does
9555 not affect the thread safety of object code produced by the compiler or
9556 that of libraries supplied with it. These are HP-UX specific flags.
9562 Generate code for a 32-bit or 64-bit environment.
9563 The 32-bit environment sets int, long and pointer to 32 bits.
9564 The 64-bit environment sets int to 32 bits and long and pointer
9565 to 64 bits. These are HP-UX specific flags.
9570 @subsection M32C Options
9571 @cindex M32C options
9574 @item -mcpu=@var{name}
9576 Select the CPU for which code is generated. @var{name} may be one of
9577 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
9578 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
9583 Specifies that the program will be run on the simulator. This causes
9584 an alternate runtime library to be linked in which supports, for
9585 example, file I/O. You must not use this option when generating
9586 programs that will run on real hardware; you must provide your own
9587 runtime library for whatever I/O functions are needed.
9589 @item -memregs=@var{number}
9591 Specifies the number of memory-based pseudo-registers GCC will use
9592 during code generation. These pseudo-registers will be used like real
9593 registers, so there is a tradeoff between GCC's ability to fit the
9594 code into available registers, and the performance penalty of using
9595 memory instead of registers. Note that all modules in a program must
9596 be compiled with the same value for this option. Because of that, you
9597 must not use this option with the default runtime libraries gcc
9602 @node M32R/D Options
9603 @subsection M32R/D Options
9604 @cindex M32R/D options
9606 These @option{-m} options are defined for Renesas M32R/D architectures:
9611 Generate code for the M32R/2@.
9615 Generate code for the M32R/X@.
9619 Generate code for the M32R@. This is the default.
9622 @opindex mmodel=small
9623 Assume all objects live in the lower 16MB of memory (so that their addresses
9624 can be loaded with the @code{ld24} instruction), and assume all subroutines
9625 are reachable with the @code{bl} instruction.
9626 This is the default.
9628 The addressability of a particular object can be set with the
9629 @code{model} attribute.
9631 @item -mmodel=medium
9632 @opindex mmodel=medium
9633 Assume objects may be anywhere in the 32-bit address space (the compiler
9634 will generate @code{seth/add3} instructions to load their addresses), and
9635 assume all subroutines are reachable with the @code{bl} instruction.
9638 @opindex mmodel=large
9639 Assume objects may be anywhere in the 32-bit address space (the compiler
9640 will generate @code{seth/add3} instructions to load their addresses), and
9641 assume subroutines may not be reachable with the @code{bl} instruction
9642 (the compiler will generate the much slower @code{seth/add3/jl}
9643 instruction sequence).
9646 @opindex msdata=none
9647 Disable use of the small data area. Variables will be put into
9648 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
9649 @code{section} attribute has been specified).
9650 This is the default.
9652 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
9653 Objects may be explicitly put in the small data area with the
9654 @code{section} attribute using one of these sections.
9657 @opindex msdata=sdata
9658 Put small global and static data in the small data area, but do not
9659 generate special code to reference them.
9663 Put small global and static data in the small data area, and generate
9664 special instructions to reference them.
9668 @cindex smaller data references
9669 Put global and static objects less than or equal to @var{num} bytes
9670 into the small data or bss sections instead of the normal data or bss
9671 sections. The default value of @var{num} is 8.
9672 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
9673 for this option to have any effect.
9675 All modules should be compiled with the same @option{-G @var{num}} value.
9676 Compiling with different values of @var{num} may or may not work; if it
9677 doesn't the linker will give an error message---incorrect code will not be
9682 Makes the M32R specific code in the compiler display some statistics
9683 that might help in debugging programs.
9686 @opindex malign-loops
9687 Align all loops to a 32-byte boundary.
9689 @item -mno-align-loops
9690 @opindex mno-align-loops
9691 Do not enforce a 32-byte alignment for loops. This is the default.
9693 @item -missue-rate=@var{number}
9694 @opindex missue-rate=@var{number}
9695 Issue @var{number} instructions per cycle. @var{number} can only be 1
9698 @item -mbranch-cost=@var{number}
9699 @opindex mbranch-cost=@var{number}
9700 @var{number} can only be 1 or 2. If it is 1 then branches will be
9701 preferred over conditional code, if it is 2, then the opposite will
9704 @item -mflush-trap=@var{number}
9705 @opindex mflush-trap=@var{number}
9706 Specifies the trap number to use to flush the cache. The default is
9707 12. Valid numbers are between 0 and 15 inclusive.
9709 @item -mno-flush-trap
9710 @opindex mno-flush-trap
9711 Specifies that the cache cannot be flushed by using a trap.
9713 @item -mflush-func=@var{name}
9714 @opindex mflush-func=@var{name}
9715 Specifies the name of the operating system function to call to flush
9716 the cache. The default is @emph{_flush_cache}, but a function call
9717 will only be used if a trap is not available.
9719 @item -mno-flush-func
9720 @opindex mno-flush-func
9721 Indicates that there is no OS function for flushing the cache.
9725 @node M680x0 Options
9726 @subsection M680x0 Options
9727 @cindex M680x0 options
9729 These are the @samp{-m} options defined for the 68000 series. The default
9730 values for these options depends on which style of 68000 was selected when
9731 the compiler was configured; the defaults for the most common choices are
9739 Generate output for a 68000. This is the default
9740 when the compiler is configured for 68000-based systems.
9742 Use this option for microcontrollers with a 68000 or EC000 core,
9743 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
9749 Generate output for a 68020. This is the default
9750 when the compiler is configured for 68020-based systems.
9754 Generate output containing 68881 instructions for floating point.
9755 This is the default for most 68020 systems unless @option{--nfp} was
9756 specified when the compiler was configured.
9760 Generate output for a 68030. This is the default when the compiler is
9761 configured for 68030-based systems.
9765 Generate output for a 68040. This is the default when the compiler is
9766 configured for 68040-based systems.
9768 This option inhibits the use of 68881/68882 instructions that have to be
9769 emulated by software on the 68040. Use this option if your 68040 does not
9770 have code to emulate those instructions.
9774 Generate output for a 68060. This is the default when the compiler is
9775 configured for 68060-based systems.
9777 This option inhibits the use of 68020 and 68881/68882 instructions that
9778 have to be emulated by software on the 68060. Use this option if your 68060
9779 does not have code to emulate those instructions.
9783 Generate output for a CPU32. This is the default
9784 when the compiler is configured for CPU32-based systems.
9786 Use this option for microcontrollers with a
9787 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
9788 68336, 68340, 68341, 68349 and 68360.
9792 Generate output for a 520X ``coldfire'' family cpu. This is the default
9793 when the compiler is configured for 520X-based systems.
9795 Use this option for microcontroller with a 5200 core, including
9796 the MCF5202, MCF5203, MCF5204 and MCF5202.
9801 Generate output for a 68040, without using any of the new instructions.
9802 This results in code which can run relatively efficiently on either a
9803 68020/68881 or a 68030 or a 68040. The generated code does use the
9804 68881 instructions that are emulated on the 68040.
9808 Generate output for a 68060, without using any of the new instructions.
9809 This results in code which can run relatively efficiently on either a
9810 68020/68881 or a 68030 or a 68040. The generated code does use the
9811 68881 instructions that are emulated on the 68060.
9814 @opindex msoft-float
9815 Generate output containing library calls for floating point.
9816 @strong{Warning:} the requisite libraries are not available for all m68k
9817 targets. Normally the facilities of the machine's usual C compiler are
9818 used, but this can't be done directly in cross-compilation. You must
9819 make your own arrangements to provide suitable library functions for
9820 cross-compilation. The embedded targets @samp{m68k-*-aout} and
9821 @samp{m68k-*-coff} do provide software floating point support.
9825 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9826 Additionally, parameters passed on the stack are also aligned to a
9827 16-bit boundary even on targets whose API mandates promotion to 32-bit.
9830 @opindex mnobitfield
9831 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
9832 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
9836 Do use the bit-field instructions. The @option{-m68020} option implies
9837 @option{-mbitfield}. This is the default if you use a configuration
9838 designed for a 68020.
9842 Use a different function-calling convention, in which functions
9843 that take a fixed number of arguments return with the @code{rtd}
9844 instruction, which pops their arguments while returning. This
9845 saves one instruction in the caller since there is no need to pop
9846 the arguments there.
9848 This calling convention is incompatible with the one normally
9849 used on Unix, so you cannot use it if you need to call libraries
9850 compiled with the Unix compiler.
9852 Also, you must provide function prototypes for all functions that
9853 take variable numbers of arguments (including @code{printf});
9854 otherwise incorrect code will be generated for calls to those
9857 In addition, seriously incorrect code will result if you call a
9858 function with too many arguments. (Normally, extra arguments are
9859 harmlessly ignored.)
9861 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
9862 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
9865 @itemx -mno-align-int
9867 @opindex mno-align-int
9868 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
9869 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
9870 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
9871 Aligning variables on 32-bit boundaries produces code that runs somewhat
9872 faster on processors with 32-bit busses at the expense of more memory.
9874 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
9875 align structures containing the above types differently than
9876 most published application binary interface specifications for the m68k.
9880 Use the pc-relative addressing mode of the 68000 directly, instead of
9881 using a global offset table. At present, this option implies @option{-fpic},
9882 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
9883 not presently supported with @option{-mpcrel}, though this could be supported for
9884 68020 and higher processors.
9886 @item -mno-strict-align
9887 @itemx -mstrict-align
9888 @opindex mno-strict-align
9889 @opindex mstrict-align
9890 Do not (do) assume that unaligned memory references will be handled by
9894 Generate code that allows the data segment to be located in a different
9895 area of memory from the text segment. This allows for execute in place in
9896 an environment without virtual memory management. This option implies
9900 Generate code that assumes that the data segment follows the text segment.
9901 This is the default.
9903 @item -mid-shared-library
9904 Generate code that supports shared libraries via the library ID method.
9905 This allows for execute in place and shared libraries in an environment
9906 without virtual memory management. This option implies @option{-fPIC}.
9908 @item -mno-id-shared-library
9909 Generate code that doesn't assume ID based shared libraries are being used.
9910 This is the default.
9912 @item -mshared-library-id=n
9913 Specified the identification number of the ID based shared library being
9914 compiled. Specifying a value of 0 will generate more compact code, specifying
9915 other values will force the allocation of that number to the current
9916 library but is no more space or time efficient than omitting this option.
9920 @node M68hc1x Options
9921 @subsection M68hc1x Options
9922 @cindex M68hc1x options
9924 These are the @samp{-m} options defined for the 68hc11 and 68hc12
9925 microcontrollers. The default values for these options depends on
9926 which style of microcontroller was selected when the compiler was configured;
9927 the defaults for the most common choices are given below.
9934 Generate output for a 68HC11. This is the default
9935 when the compiler is configured for 68HC11-based systems.
9941 Generate output for a 68HC12. This is the default
9942 when the compiler is configured for 68HC12-based systems.
9948 Generate output for a 68HCS12.
9951 @opindex mauto-incdec
9952 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
9959 Enable the use of 68HC12 min and max instructions.
9962 @itemx -mno-long-calls
9963 @opindex mlong-calls
9964 @opindex mno-long-calls
9965 Treat all calls as being far away (near). If calls are assumed to be
9966 far away, the compiler will use the @code{call} instruction to
9967 call a function and the @code{rtc} instruction for returning.
9971 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9973 @item -msoft-reg-count=@var{count}
9974 @opindex msoft-reg-count
9975 Specify the number of pseudo-soft registers which are used for the
9976 code generation. The maximum number is 32. Using more pseudo-soft
9977 register may or may not result in better code depending on the program.
9978 The default is 4 for 68HC11 and 2 for 68HC12.
9983 @subsection MCore Options
9984 @cindex MCore options
9986 These are the @samp{-m} options defined for the Motorola M*Core
9994 @opindex mno-hardlit
9995 Inline constants into the code stream if it can be done in two
9996 instructions or less.
10002 Use the divide instruction. (Enabled by default).
10004 @item -mrelax-immediate
10005 @itemx -mno-relax-immediate
10006 @opindex mrelax-immediate
10007 @opindex mno-relax-immediate
10008 Allow arbitrary sized immediates in bit operations.
10010 @item -mwide-bitfields
10011 @itemx -mno-wide-bitfields
10012 @opindex mwide-bitfields
10013 @opindex mno-wide-bitfields
10014 Always treat bit-fields as int-sized.
10016 @item -m4byte-functions
10017 @itemx -mno-4byte-functions
10018 @opindex m4byte-functions
10019 @opindex mno-4byte-functions
10020 Force all functions to be aligned to a four byte boundary.
10022 @item -mcallgraph-data
10023 @itemx -mno-callgraph-data
10024 @opindex mcallgraph-data
10025 @opindex mno-callgraph-data
10026 Emit callgraph information.
10029 @itemx -mno-slow-bytes
10030 @opindex mslow-bytes
10031 @opindex mno-slow-bytes
10032 Prefer word access when reading byte quantities.
10034 @item -mlittle-endian
10035 @itemx -mbig-endian
10036 @opindex mlittle-endian
10037 @opindex mbig-endian
10038 Generate code for a little endian target.
10044 Generate code for the 210 processor.
10048 @subsection MIPS Options
10049 @cindex MIPS options
10055 Generate big-endian code.
10059 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10062 @item -march=@var{arch}
10064 Generate code that will run on @var{arch}, which can be the name of a
10065 generic MIPS ISA, or the name of a particular processor.
10067 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10068 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10069 The processor names are:
10070 @samp{4kc}, @samp{4km}, @samp{4kp},
10071 @samp{5kc}, @samp{5kf},
10073 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10076 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10077 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10078 @samp{rm7000}, @samp{rm9000},
10081 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10082 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10083 The special value @samp{from-abi} selects the
10084 most compatible architecture for the selected ABI (that is,
10085 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10087 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10088 (for example, @samp{-march=r2k}). Prefixes are optional, and
10089 @samp{vr} may be written @samp{r}.
10091 GCC defines two macros based on the value of this option. The first
10092 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10093 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10094 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10095 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10096 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10098 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10099 above. In other words, it will have the full prefix and will not
10100 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10101 the macro names the resolved architecture (either @samp{"mips1"} or
10102 @samp{"mips3"}). It names the default architecture when no
10103 @option{-march} option is given.
10105 @item -mtune=@var{arch}
10107 Optimize for @var{arch}. Among other things, this option controls
10108 the way instructions are scheduled, and the perceived cost of arithmetic
10109 operations. The list of @var{arch} values is the same as for
10112 When this option is not used, GCC will optimize for the processor
10113 specified by @option{-march}. By using @option{-march} and
10114 @option{-mtune} together, it is possible to generate code that will
10115 run on a family of processors, but optimize the code for one
10116 particular member of that family.
10118 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10119 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10120 @samp{-march} ones described above.
10124 Equivalent to @samp{-march=mips1}.
10128 Equivalent to @samp{-march=mips2}.
10132 Equivalent to @samp{-march=mips3}.
10136 Equivalent to @samp{-march=mips4}.
10140 Equivalent to @samp{-march=mips32}.
10144 Equivalent to @samp{-march=mips32r2}.
10148 Equivalent to @samp{-march=mips64}.
10153 @opindex mno-mips16
10154 Generate (do not generate) MIPS16 code. If GCC is targetting a
10155 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10167 Generate code for the given ABI@.
10169 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10170 generates 64-bit code when you select a 64-bit architecture, but you
10171 can use @option{-mgp32} to get 32-bit code instead.
10173 For information about the O64 ABI, see
10174 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10177 @itemx -mno-abicalls
10179 @opindex mno-abicalls
10180 Generate (do not generate) SVR4-style position-independent code.
10181 @option{-mabicalls} is the default for SVR4-based systems.
10187 Lift (do not lift) the usual restrictions on the size of the global
10190 GCC normally uses a single instruction to load values from the GOT@.
10191 While this is relatively efficient, it will only work if the GOT
10192 is smaller than about 64k. Anything larger will cause the linker
10193 to report an error such as:
10195 @cindex relocation truncated to fit (MIPS)
10197 relocation truncated to fit: R_MIPS_GOT16 foobar
10200 If this happens, you should recompile your code with @option{-mxgot}.
10201 It should then work with very large GOTs, although it will also be
10202 less efficient, since it will take three instructions to fetch the
10203 value of a global symbol.
10205 Note that some linkers can create multiple GOTs. If you have such a
10206 linker, you should only need to use @option{-mxgot} when a single object
10207 file accesses more than 64k's worth of GOT entries. Very few do.
10209 These options have no effect unless GCC is generating position
10214 Assume that general-purpose registers are 32 bits wide.
10218 Assume that general-purpose registers are 64 bits wide.
10222 Assume that floating-point registers are 32 bits wide.
10226 Assume that floating-point registers are 64 bits wide.
10229 @opindex mhard-float
10230 Use floating-point coprocessor instructions.
10233 @opindex msoft-float
10234 Do not use floating-point coprocessor instructions. Implement
10235 floating-point calculations using library calls instead.
10237 @item -msingle-float
10238 @opindex msingle-float
10239 Assume that the floating-point coprocessor only supports single-precision
10242 @itemx -mdouble-float
10243 @opindex mdouble-float
10244 Assume that the floating-point coprocessor supports double-precision
10245 operations. This is the default.
10251 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10253 @itemx -mpaired-single
10254 @itemx -mno-paired-single
10255 @opindex mpaired-single
10256 @opindex mno-paired-single
10257 Use (do not use) paired-single floating-point instructions.
10258 @xref{MIPS Paired-Single Support}. This option can only be used
10259 when generating 64-bit code and requires hardware floating-point
10260 support to be enabled.
10265 @opindex mno-mips3d
10266 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10267 The option @option{-mips3d} implies @option{-mpaired-single}.
10271 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10272 an explanation of the default and the way that the pointer size is
10277 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10279 The default size of @code{int}s, @code{long}s and pointers depends on
10280 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10281 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10282 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10283 or the same size as integer registers, whichever is smaller.
10289 Assume (do not assume) that all symbols have 32-bit values, regardless
10290 of the selected ABI@. This option is useful in combination with
10291 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10292 to generate shorter and faster references to symbolic addresses.
10296 @cindex smaller data references (MIPS)
10297 @cindex gp-relative references (MIPS)
10298 Put global and static items less than or equal to @var{num} bytes into
10299 the small data or bss section instead of the normal data or bss section.
10300 This allows the data to be accessed using a single instruction.
10302 All modules should be compiled with the same @option{-G @var{num}}
10305 @item -membedded-data
10306 @itemx -mno-embedded-data
10307 @opindex membedded-data
10308 @opindex mno-embedded-data
10309 Allocate variables to the read-only data section first if possible, then
10310 next in the small data section if possible, otherwise in data. This gives
10311 slightly slower code than the default, but reduces the amount of RAM required
10312 when executing, and thus may be preferred for some embedded systems.
10314 @item -muninit-const-in-rodata
10315 @itemx -mno-uninit-const-in-rodata
10316 @opindex muninit-const-in-rodata
10317 @opindex mno-uninit-const-in-rodata
10318 Put uninitialized @code{const} variables in the read-only data section.
10319 This option is only meaningful in conjunction with @option{-membedded-data}.
10321 @item -msplit-addresses
10322 @itemx -mno-split-addresses
10323 @opindex msplit-addresses
10324 @opindex mno-split-addresses
10325 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10326 relocation operators. This option has been superseded by
10327 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10329 @item -mexplicit-relocs
10330 @itemx -mno-explicit-relocs
10331 @opindex mexplicit-relocs
10332 @opindex mno-explicit-relocs
10333 Use (do not use) assembler relocation operators when dealing with symbolic
10334 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10335 is to use assembler macros instead.
10337 @option{-mexplicit-relocs} is the default if GCC was configured
10338 to use an assembler that supports relocation operators.
10340 @item -mcheck-zero-division
10341 @itemx -mno-check-zero-division
10342 @opindex mcheck-zero-division
10343 @opindex mno-check-zero-division
10344 Trap (do not trap) on integer division by zero. The default is
10345 @option{-mcheck-zero-division}.
10347 @item -mdivide-traps
10348 @itemx -mdivide-breaks
10349 @opindex mdivide-traps
10350 @opindex mdivide-breaks
10351 MIPS systems check for division by zero by generating either a
10352 conditional trap or a break instruction. Using traps results in
10353 smaller code, but is only supported on MIPS II and later. Also, some
10354 versions of the Linux kernel have a bug that prevents trap from
10355 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10356 allow conditional traps on architectures that support them and
10357 @option{-mdivide-breaks} to force the use of breaks.
10359 The default is usually @option{-mdivide-traps}, but this can be
10360 overridden at configure time using @option{--with-divide=breaks}.
10361 Divide-by-zero checks can be completely disabled using
10362 @option{-mno-check-zero-division}.
10367 @opindex mno-memcpy
10368 Force (do not force) the use of @code{memcpy()} for non-trivial block
10369 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10370 most constant-sized copies.
10373 @itemx -mno-long-calls
10374 @opindex mlong-calls
10375 @opindex mno-long-calls
10376 Disable (do not disable) use of the @code{jal} instruction. Calling
10377 functions using @code{jal} is more efficient but requires the caller
10378 and callee to be in the same 256 megabyte segment.
10380 This option has no effect on abicalls code. The default is
10381 @option{-mno-long-calls}.
10387 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10388 instructions, as provided by the R4650 ISA@.
10391 @itemx -mno-fused-madd
10392 @opindex mfused-madd
10393 @opindex mno-fused-madd
10394 Enable (disable) use of the floating point multiply-accumulate
10395 instructions, when they are available. The default is
10396 @option{-mfused-madd}.
10398 When multiply-accumulate instructions are used, the intermediate
10399 product is calculated to infinite precision and is not subject to
10400 the FCSR Flush to Zero bit. This may be undesirable in some
10405 Tell the MIPS assembler to not run its preprocessor over user
10406 assembler files (with a @samp{.s} suffix) when assembling them.
10409 @itemx -mno-fix-r4000
10410 @opindex mfix-r4000
10411 @opindex mno-fix-r4000
10412 Work around certain R4000 CPU errata:
10415 A double-word or a variable shift may give an incorrect result if executed
10416 immediately after starting an integer division.
10418 A double-word or a variable shift may give an incorrect result if executed
10419 while an integer multiplication is in progress.
10421 An integer division may give an incorrect result if started in a delay slot
10422 of a taken branch or a jump.
10426 @itemx -mno-fix-r4400
10427 @opindex mfix-r4400
10428 @opindex mno-fix-r4400
10429 Work around certain R4400 CPU errata:
10432 A double-word or a variable shift may give an incorrect result if executed
10433 immediately after starting an integer division.
10437 @itemx -mno-fix-vr4120
10438 @opindex mfix-vr4120
10439 Work around certain VR4120 errata:
10442 @code{dmultu} does not always produce the correct result.
10444 @code{div} and @code{ddiv} do not always produce the correct result if one
10445 of the operands is negative.
10447 The workarounds for the division errata rely on special functions in
10448 @file{libgcc.a}. At present, these functions are only provided by
10449 the @code{mips64vr*-elf} configurations.
10451 Other VR4120 errata require a nop to be inserted between certain pairs of
10452 instructions. These errata are handled by the assembler, not by GCC itself.
10455 @opindex mfix-vr4130
10456 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10457 workarounds are implemented by the assembler rather than by GCC,
10458 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10459 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10460 instructions are available instead.
10463 @itemx -mno-fix-sb1
10465 Work around certain SB-1 CPU core errata.
10466 (This flag currently works around the SB-1 revision 2
10467 ``F1'' and ``F2'' floating point errata.)
10469 @item -mflush-func=@var{func}
10470 @itemx -mno-flush-func
10471 @opindex mflush-func
10472 Specifies the function to call to flush the I and D caches, or to not
10473 call any such function. If called, the function must take the same
10474 arguments as the common @code{_flush_func()}, that is, the address of the
10475 memory range for which the cache is being flushed, the size of the
10476 memory range, and the number 3 (to flush both caches). The default
10477 depends on the target GCC was configured for, but commonly is either
10478 @samp{_flush_func} or @samp{__cpu_flush}.
10480 @item -mbranch-likely
10481 @itemx -mno-branch-likely
10482 @opindex mbranch-likely
10483 @opindex mno-branch-likely
10484 Enable or disable use of Branch Likely instructions, regardless of the
10485 default for the selected architecture. By default, Branch Likely
10486 instructions may be generated if they are supported by the selected
10487 architecture. An exception is for the MIPS32 and MIPS64 architectures
10488 and processors which implement those architectures; for those, Branch
10489 Likely instructions will not be generated by default because the MIPS32
10490 and MIPS64 architectures specifically deprecate their use.
10492 @item -mfp-exceptions
10493 @itemx -mno-fp-exceptions
10494 @opindex mfp-exceptions
10495 Specifies whether FP exceptions are enabled. This affects how we schedule
10496 FP instructions for some processors. The default is that FP exceptions are
10499 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
10500 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
10503 @item -mvr4130-align
10504 @itemx -mno-vr4130-align
10505 @opindex mvr4130-align
10506 The VR4130 pipeline is two-way superscalar, but can only issue two
10507 instructions together if the first one is 8-byte aligned. When this
10508 option is enabled, GCC will align pairs of instructions that it
10509 thinks should execute in parallel.
10511 This option only has an effect when optimizing for the VR4130.
10512 It normally makes code faster, but at the expense of making it bigger.
10513 It is enabled by default at optimization level @option{-O3}.
10517 @subsection MMIX Options
10518 @cindex MMIX Options
10520 These options are defined for the MMIX:
10524 @itemx -mno-libfuncs
10526 @opindex mno-libfuncs
10527 Specify that intrinsic library functions are being compiled, passing all
10528 values in registers, no matter the size.
10531 @itemx -mno-epsilon
10533 @opindex mno-epsilon
10534 Generate floating-point comparison instructions that compare with respect
10535 to the @code{rE} epsilon register.
10537 @item -mabi=mmixware
10539 @opindex mabi-mmixware
10541 Generate code that passes function parameters and return values that (in
10542 the called function) are seen as registers @code{$0} and up, as opposed to
10543 the GNU ABI which uses global registers @code{$231} and up.
10545 @item -mzero-extend
10546 @itemx -mno-zero-extend
10547 @opindex mzero-extend
10548 @opindex mno-zero-extend
10549 When reading data from memory in sizes shorter than 64 bits, use (do not
10550 use) zero-extending load instructions by default, rather than
10551 sign-extending ones.
10554 @itemx -mno-knuthdiv
10556 @opindex mno-knuthdiv
10557 Make the result of a division yielding a remainder have the same sign as
10558 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10559 remainder follows the sign of the dividend. Both methods are
10560 arithmetically valid, the latter being almost exclusively used.
10562 @item -mtoplevel-symbols
10563 @itemx -mno-toplevel-symbols
10564 @opindex mtoplevel-symbols
10565 @opindex mno-toplevel-symbols
10566 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10567 code can be used with the @code{PREFIX} assembly directive.
10571 Generate an executable in the ELF format, rather than the default
10572 @samp{mmo} format used by the @command{mmix} simulator.
10574 @item -mbranch-predict
10575 @itemx -mno-branch-predict
10576 @opindex mbranch-predict
10577 @opindex mno-branch-predict
10578 Use (do not use) the probable-branch instructions, when static branch
10579 prediction indicates a probable branch.
10581 @item -mbase-addresses
10582 @itemx -mno-base-addresses
10583 @opindex mbase-addresses
10584 @opindex mno-base-addresses
10585 Generate (do not generate) code that uses @emph{base addresses}. Using a
10586 base address automatically generates a request (handled by the assembler
10587 and the linker) for a constant to be set up in a global register. The
10588 register is used for one or more base address requests within the range 0
10589 to 255 from the value held in the register. The generally leads to short
10590 and fast code, but the number of different data items that can be
10591 addressed is limited. This means that a program that uses lots of static
10592 data may require @option{-mno-base-addresses}.
10594 @item -msingle-exit
10595 @itemx -mno-single-exit
10596 @opindex msingle-exit
10597 @opindex mno-single-exit
10598 Force (do not force) generated code to have a single exit point in each
10602 @node MN10300 Options
10603 @subsection MN10300 Options
10604 @cindex MN10300 options
10606 These @option{-m} options are defined for Matsushita MN10300 architectures:
10611 Generate code to avoid bugs in the multiply instructions for the MN10300
10612 processors. This is the default.
10614 @item -mno-mult-bug
10615 @opindex mno-mult-bug
10616 Do not generate code to avoid bugs in the multiply instructions for the
10617 MN10300 processors.
10621 Generate code which uses features specific to the AM33 processor.
10625 Do not generate code which uses features specific to the AM33 processor. This
10628 @item -mreturn-pointer-on-d0
10629 @opindex mreturn-pointer-on-d0
10630 When generating a function which returns a pointer, return the pointer
10631 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
10632 only in a0, and attempts to call such functions without a prototype
10633 would result in errors. Note that this option is on by default; use
10634 @option{-mno-return-pointer-on-d0} to disable it.
10638 Do not link in the C run-time initialization object file.
10642 Indicate to the linker that it should perform a relaxation optimization pass
10643 to shorten branches, calls and absolute memory addresses. This option only
10644 has an effect when used on the command line for the final link step.
10646 This option makes symbolic debugging impossible.
10650 @subsection MS1 Options
10651 @cindex MS1 options
10653 These @option{-m} options are defined for Morpho MS1 architectures:
10657 @item -march=@var{cpu-type}
10659 Generate code that will run on @var{cpu-type}, which is the name of a system
10660 representing a certain processor type. Possible values for
10661 @var{cpu-type} are @samp{MS1-64-001}, @samp{MS1-16-002}, and
10664 When this option is not used, the default is @option{-march=MS1-16-003}.
10668 Generate multiply instructions.
10672 Do not generate multiply instructions.
10676 Use byte loads and stores when generating code.
10680 Do not use byte loads and stores when generating code.
10684 Use simulator runtime
10688 Do not link in the C run-time initialization object file
10689 @file{crti.o}. Other run-time initialization and termination files
10690 such as @file{startup.o} and @file{exit.o} are still included on the
10691 linker command line.
10695 @node PDP-11 Options
10696 @subsection PDP-11 Options
10697 @cindex PDP-11 Options
10699 These options are defined for the PDP-11:
10704 Use hardware FPP floating point. This is the default. (FIS floating
10705 point on the PDP-11/40 is not supported.)
10708 @opindex msoft-float
10709 Do not use hardware floating point.
10713 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10717 Return floating-point results in memory. This is the default.
10721 Generate code for a PDP-11/40.
10725 Generate code for a PDP-11/45. This is the default.
10729 Generate code for a PDP-11/10.
10731 @item -mbcopy-builtin
10732 @opindex bcopy-builtin
10733 Use inline @code{movmemhi} patterns for copying memory. This is the
10738 Do not use inline @code{movmemhi} patterns for copying memory.
10744 Use 16-bit @code{int}. This is the default.
10750 Use 32-bit @code{int}.
10753 @itemx -mno-float32
10755 @opindex mno-float32
10756 Use 64-bit @code{float}. This is the default.
10759 @itemx -mno-float64
10761 @opindex mno-float64
10762 Use 32-bit @code{float}.
10766 Use @code{abshi2} pattern. This is the default.
10770 Do not use @code{abshi2} pattern.
10772 @item -mbranch-expensive
10773 @opindex mbranch-expensive
10774 Pretend that branches are expensive. This is for experimenting with
10775 code generation only.
10777 @item -mbranch-cheap
10778 @opindex mbranch-cheap
10779 Do not pretend that branches are expensive. This is the default.
10783 Generate code for a system with split I&D@.
10787 Generate code for a system without split I&D@. This is the default.
10791 Use Unix assembler syntax. This is the default when configured for
10792 @samp{pdp11-*-bsd}.
10796 Use DEC assembler syntax. This is the default when configured for any
10797 PDP-11 target other than @samp{pdp11-*-bsd}.
10800 @node PowerPC Options
10801 @subsection PowerPC Options
10802 @cindex PowerPC options
10804 These are listed under @xref{RS/6000 and PowerPC Options}.
10806 @node RS/6000 and PowerPC Options
10807 @subsection IBM RS/6000 and PowerPC Options
10808 @cindex RS/6000 and PowerPC Options
10809 @cindex IBM RS/6000 and PowerPC Options
10811 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
10818 @itemx -mno-powerpc
10819 @itemx -mpowerpc-gpopt
10820 @itemx -mno-powerpc-gpopt
10821 @itemx -mpowerpc-gfxopt
10822 @itemx -mno-powerpc-gfxopt
10824 @itemx -mno-powerpc64
10828 @opindex mno-power2
10830 @opindex mno-powerpc
10831 @opindex mpowerpc-gpopt
10832 @opindex mno-powerpc-gpopt
10833 @opindex mpowerpc-gfxopt
10834 @opindex mno-powerpc-gfxopt
10835 @opindex mpowerpc64
10836 @opindex mno-powerpc64
10837 GCC supports two related instruction set architectures for the
10838 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
10839 instructions supported by the @samp{rios} chip set used in the original
10840 RS/6000 systems and the @dfn{PowerPC} instruction set is the
10841 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
10842 the IBM 4xx microprocessors.
10844 Neither architecture is a subset of the other. However there is a
10845 large common subset of instructions supported by both. An MQ
10846 register is included in processors supporting the POWER architecture.
10848 You use these options to specify which instructions are available on the
10849 processor you are using. The default value of these options is
10850 determined when configuring GCC@. Specifying the
10851 @option{-mcpu=@var{cpu_type}} overrides the specification of these
10852 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
10853 rather than the options listed above.
10855 The @option{-mpower} option allows GCC to generate instructions that
10856 are found only in the POWER architecture and to use the MQ register.
10857 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
10858 to generate instructions that are present in the POWER2 architecture but
10859 not the original POWER architecture.
10861 The @option{-mpowerpc} option allows GCC to generate instructions that
10862 are found only in the 32-bit subset of the PowerPC architecture.
10863 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
10864 GCC to use the optional PowerPC architecture instructions in the
10865 General Purpose group, including floating-point square root. Specifying
10866 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
10867 use the optional PowerPC architecture instructions in the Graphics
10868 group, including floating-point select.
10870 The @option{-mpowerpc64} option allows GCC to generate the additional
10871 64-bit instructions that are found in the full PowerPC64 architecture
10872 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
10873 @option{-mno-powerpc64}.
10875 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
10876 will use only the instructions in the common subset of both
10877 architectures plus some special AIX common-mode calls, and will not use
10878 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
10879 permits GCC to use any instruction from either architecture and to
10880 allow use of the MQ register; specify this for the Motorola MPC601.
10882 @item -mnew-mnemonics
10883 @itemx -mold-mnemonics
10884 @opindex mnew-mnemonics
10885 @opindex mold-mnemonics
10886 Select which mnemonics to use in the generated assembler code. With
10887 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
10888 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
10889 assembler mnemonics defined for the POWER architecture. Instructions
10890 defined in only one architecture have only one mnemonic; GCC uses that
10891 mnemonic irrespective of which of these options is specified.
10893 GCC defaults to the mnemonics appropriate for the architecture in
10894 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
10895 value of these option. Unless you are building a cross-compiler, you
10896 should normally not specify either @option{-mnew-mnemonics} or
10897 @option{-mold-mnemonics}, but should instead accept the default.
10899 @item -mcpu=@var{cpu_type}
10901 Set architecture type, register usage, choice of mnemonics, and
10902 instruction scheduling parameters for machine type @var{cpu_type}.
10903 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
10904 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
10905 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
10906 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
10907 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
10908 @samp{860}, @samp{970}, @samp{8540}, @samp{common}, @samp{ec603e}, @samp{G3},
10909 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
10910 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
10911 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
10913 @option{-mcpu=common} selects a completely generic processor. Code
10914 generated under this option will run on any POWER or PowerPC processor.
10915 GCC will use only the instructions in the common subset of both
10916 architectures, and will not use the MQ register. GCC assumes a generic
10917 processor model for scheduling purposes.
10919 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
10920 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
10921 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
10922 types, with an appropriate, generic processor model assumed for
10923 scheduling purposes.
10925 The other options specify a specific processor. Code generated under
10926 those options will run best on that processor, and may not run at all on
10929 The @option{-mcpu} options automatically enable or disable the
10930 following options: @option{-maltivec}, @option{-mhard-float},
10931 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
10932 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
10933 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
10934 @option{-mstring}. The particular options set for any particular CPU
10935 will vary between compiler versions, depending on what setting seems
10936 to produce optimal code for that CPU; it doesn't necessarily reflect
10937 the actual hardware's capabilities. If you wish to set an individual
10938 option to a particular value, you may specify it after the
10939 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
10941 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
10942 not enabled or disabled by the @option{-mcpu} option at present, since
10943 AIX does not have full support for these options. You may still
10944 enable or disable them individually if you're sure it'll work in your
10947 @item -mtune=@var{cpu_type}
10949 Set the instruction scheduling parameters for machine type
10950 @var{cpu_type}, but do not set the architecture type, register usage, or
10951 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
10952 values for @var{cpu_type} are used for @option{-mtune} as for
10953 @option{-mcpu}. If both are specified, the code generated will use the
10954 architecture, registers, and mnemonics set by @option{-mcpu}, but the
10955 scheduling parameters set by @option{-mtune}.
10961 Generate code to compute division as reciprocal estimate and iterative
10962 refinement, creating opportunities for increased throughput. This
10963 feature requires: optional PowerPC Graphics instruction set for single
10964 precision and FRE instruction for double precision, assuming divides
10965 cannot generate user-visible traps, and the domain values not include
10966 Infinities, denormals or zero denominator.
10969 @itemx -mno-altivec
10971 @opindex mno-altivec
10972 Generate code that uses (does not use) AltiVec instructions, and also
10973 enable the use of built-in functions that allow more direct access to
10974 the AltiVec instruction set. You may also need to set
10975 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
10981 @opindex mno-vrsave
10982 Generate VRSAVE instructions when generating AltiVec code.
10986 Extend the current ABI with SPE ABI extensions. This does not change
10987 the default ABI, instead it adds the SPE ABI extensions to the current
10991 @opindex mabi=no-spe
10992 Disable Booke SPE ABI extensions for the current ABI@.
10995 @opindex msecure-plt
10996 Generate code that allows ld and ld.so to build executables and shared
10997 libraries with non-exec .plt and .got sections. This is a PowerPC
10998 32-bit SYSV ABI option.
11002 Generate code that uses a BSS .plt section that ld.so fills in, and
11003 requires .plt and .got sections that are both writable and executable.
11004 This is a PowerPC 32-bit SYSV ABI option.
11010 This switch enables or disables the generation of ISEL instructions.
11012 @item -misel=@var{yes/no}
11013 This switch has been deprecated. Use @option{-misel} and
11014 @option{-mno-isel} instead.
11020 This switch enables or disables the generation of SPE simd
11023 @item -mspe=@var{yes/no}
11024 This option has been deprecated. Use @option{-mspe} and
11025 @option{-mno-spe} instead.
11027 @item -mfloat-gprs=@var{yes/single/double/no}
11028 @itemx -mfloat-gprs
11029 @opindex mfloat-gprs
11030 This switch enables or disables the generation of floating point
11031 operations on the general purpose registers for architectures that
11034 The argument @var{yes} or @var{single} enables the use of
11035 single-precision floating point operations.
11037 The argument @var{double} enables the use of single and
11038 double-precision floating point operations.
11040 The argument @var{no} disables floating point operations on the
11041 general purpose registers.
11043 This option is currently only available on the MPC854x.
11049 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11050 targets (including GNU/Linux). The 32-bit environment sets int, long
11051 and pointer to 32 bits and generates code that runs on any PowerPC
11052 variant. The 64-bit environment sets int to 32 bits and long and
11053 pointer to 64 bits, and generates code for PowerPC64, as for
11054 @option{-mpowerpc64}.
11057 @itemx -mno-fp-in-toc
11058 @itemx -mno-sum-in-toc
11059 @itemx -mminimal-toc
11061 @opindex mno-fp-in-toc
11062 @opindex mno-sum-in-toc
11063 @opindex mminimal-toc
11064 Modify generation of the TOC (Table Of Contents), which is created for
11065 every executable file. The @option{-mfull-toc} option is selected by
11066 default. In that case, GCC will allocate at least one TOC entry for
11067 each unique non-automatic variable reference in your program. GCC
11068 will also place floating-point constants in the TOC@. However, only
11069 16,384 entries are available in the TOC@.
11071 If you receive a linker error message that saying you have overflowed
11072 the available TOC space, you can reduce the amount of TOC space used
11073 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11074 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11075 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11076 generate code to calculate the sum of an address and a constant at
11077 run-time instead of putting that sum into the TOC@. You may specify one
11078 or both of these options. Each causes GCC to produce very slightly
11079 slower and larger code at the expense of conserving TOC space.
11081 If you still run out of space in the TOC even when you specify both of
11082 these options, specify @option{-mminimal-toc} instead. This option causes
11083 GCC to make only one TOC entry for every file. When you specify this
11084 option, GCC will produce code that is slower and larger but which
11085 uses extremely little TOC space. You may wish to use this option
11086 only on files that contain less frequently executed code.
11092 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11093 @code{long} type, and the infrastructure needed to support them.
11094 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11095 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11096 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11099 @itemx -mno-xl-compat
11100 @opindex mxl-compat
11101 @opindex mno-xl-compat
11102 Produce code that conforms more closely to IBM XLC semantics when using
11103 AIX-compatible ABI. Pass floating-point arguments to prototyped
11104 functions beyond the register save area (RSA) on the stack in addition
11105 to argument FPRs. Do not assume that most significant double in 128
11106 bit long double value is properly rounded when comparing values.
11108 The AIX calling convention was extended but not initially documented to
11109 handle an obscure K&R C case of calling a function that takes the
11110 address of its arguments with fewer arguments than declared. AIX XL
11111 compilers access floating point arguments which do not fit in the
11112 RSA from the stack when a subroutine is compiled without
11113 optimization. Because always storing floating-point arguments on the
11114 stack is inefficient and rarely needed, this option is not enabled by
11115 default and only is necessary when calling subroutines compiled by AIX
11116 XL compilers without optimization.
11120 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11121 application written to use message passing with special startup code to
11122 enable the application to run. The system must have PE installed in the
11123 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11124 must be overridden with the @option{-specs=} option to specify the
11125 appropriate directory location. The Parallel Environment does not
11126 support threads, so the @option{-mpe} option and the @option{-pthread}
11127 option are incompatible.
11129 @item -malign-natural
11130 @itemx -malign-power
11131 @opindex malign-natural
11132 @opindex malign-power
11133 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11134 @option{-malign-natural} overrides the ABI-defined alignment of larger
11135 types, such as floating-point doubles, on their natural size-based boundary.
11136 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11137 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11139 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11143 @itemx -mhard-float
11144 @opindex msoft-float
11145 @opindex mhard-float
11146 Generate code that does not use (uses) the floating-point register set.
11147 Software floating point emulation is provided if you use the
11148 @option{-msoft-float} option, and pass the option to GCC when linking.
11151 @itemx -mno-multiple
11153 @opindex mno-multiple
11154 Generate code that uses (does not use) the load multiple word
11155 instructions and the store multiple word instructions. These
11156 instructions are generated by default on POWER systems, and not
11157 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11158 endian PowerPC systems, since those instructions do not work when the
11159 processor is in little endian mode. The exceptions are PPC740 and
11160 PPC750 which permit the instructions usage in little endian mode.
11165 @opindex mno-string
11166 Generate code that uses (does not use) the load string instructions
11167 and the store string word instructions to save multiple registers and
11168 do small block moves. These instructions are generated by default on
11169 POWER systems, and not generated on PowerPC systems. Do not use
11170 @option{-mstring} on little endian PowerPC systems, since those
11171 instructions do not work when the processor is in little endian mode.
11172 The exceptions are PPC740 and PPC750 which permit the instructions
11173 usage in little endian mode.
11178 @opindex mno-update
11179 Generate code that uses (does not use) the load or store instructions
11180 that update the base register to the address of the calculated memory
11181 location. These instructions are generated by default. If you use
11182 @option{-mno-update}, there is a small window between the time that the
11183 stack pointer is updated and the address of the previous frame is
11184 stored, which means code that walks the stack frame across interrupts or
11185 signals may get corrupted data.
11188 @itemx -mno-fused-madd
11189 @opindex mfused-madd
11190 @opindex mno-fused-madd
11191 Generate code that uses (does not use) the floating point multiply and
11192 accumulate instructions. These instructions are generated by default if
11193 hardware floating is used.
11195 @item -mno-bit-align
11197 @opindex mno-bit-align
11198 @opindex mbit-align
11199 On System V.4 and embedded PowerPC systems do not (do) force structures
11200 and unions that contain bit-fields to be aligned to the base type of the
11203 For example, by default a structure containing nothing but 8
11204 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11205 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11206 the structure would be aligned to a 1 byte boundary and be one byte in
11209 @item -mno-strict-align
11210 @itemx -mstrict-align
11211 @opindex mno-strict-align
11212 @opindex mstrict-align
11213 On System V.4 and embedded PowerPC systems do not (do) assume that
11214 unaligned memory references will be handled by the system.
11216 @item -mrelocatable
11217 @itemx -mno-relocatable
11218 @opindex mrelocatable
11219 @opindex mno-relocatable
11220 On embedded PowerPC systems generate code that allows (does not allow)
11221 the program to be relocated to a different address at runtime. If you
11222 use @option{-mrelocatable} on any module, all objects linked together must
11223 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11225 @item -mrelocatable-lib
11226 @itemx -mno-relocatable-lib
11227 @opindex mrelocatable-lib
11228 @opindex mno-relocatable-lib
11229 On embedded PowerPC systems generate code that allows (does not allow)
11230 the program to be relocated to a different address at runtime. Modules
11231 compiled with @option{-mrelocatable-lib} can be linked with either modules
11232 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11233 with modules compiled with the @option{-mrelocatable} options.
11239 On System V.4 and embedded PowerPC systems do not (do) assume that
11240 register 2 contains a pointer to a global area pointing to the addresses
11241 used in the program.
11244 @itemx -mlittle-endian
11246 @opindex mlittle-endian
11247 On System V.4 and embedded PowerPC systems compile code for the
11248 processor in little endian mode. The @option{-mlittle-endian} option is
11249 the same as @option{-mlittle}.
11252 @itemx -mbig-endian
11254 @opindex mbig-endian
11255 On System V.4 and embedded PowerPC systems compile code for the
11256 processor in big endian mode. The @option{-mbig-endian} option is
11257 the same as @option{-mbig}.
11259 @item -mdynamic-no-pic
11260 @opindex mdynamic-no-pic
11261 On Darwin and Mac OS X systems, compile code so that it is not
11262 relocatable, but that its external references are relocatable. The
11263 resulting code is suitable for applications, but not shared
11266 @item -mprioritize-restricted-insns=@var{priority}
11267 @opindex mprioritize-restricted-insns
11268 This option controls the priority that is assigned to
11269 dispatch-slot restricted instructions during the second scheduling
11270 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11271 @var{no/highest/second-highest} priority to dispatch slot restricted
11274 @item -msched-costly-dep=@var{dependence_type}
11275 @opindex msched-costly-dep
11276 This option controls which dependences are considered costly
11277 by the target during instruction scheduling. The argument
11278 @var{dependence_type} takes one of the following values:
11279 @var{no}: no dependence is costly,
11280 @var{all}: all dependences are costly,
11281 @var{true_store_to_load}: a true dependence from store to load is costly,
11282 @var{store_to_load}: any dependence from store to load is costly,
11283 @var{number}: any dependence which latency >= @var{number} is costly.
11285 @item -minsert-sched-nops=@var{scheme}
11286 @opindex minsert-sched-nops
11287 This option controls which nop insertion scheme will be used during
11288 the second scheduling pass. The argument @var{scheme} takes one of the
11290 @var{no}: Don't insert nops.
11291 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11292 according to the scheduler's grouping.
11293 @var{regroup_exact}: Insert nops to force costly dependent insns into
11294 separate groups. Insert exactly as many nops as needed to force an insn
11295 to a new group, according to the estimated processor grouping.
11296 @var{number}: Insert nops to force costly dependent insns into
11297 separate groups. Insert @var{number} nops to force an insn to a new group.
11300 @opindex mcall-sysv
11301 On System V.4 and embedded PowerPC systems compile code using calling
11302 conventions that adheres to the March 1995 draft of the System V
11303 Application Binary Interface, PowerPC processor supplement. This is the
11304 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11306 @item -mcall-sysv-eabi
11307 @opindex mcall-sysv-eabi
11308 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11310 @item -mcall-sysv-noeabi
11311 @opindex mcall-sysv-noeabi
11312 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11314 @item -mcall-solaris
11315 @opindex mcall-solaris
11316 On System V.4 and embedded PowerPC systems compile code for the Solaris
11320 @opindex mcall-linux
11321 On System V.4 and embedded PowerPC systems compile code for the
11322 Linux-based GNU system.
11326 On System V.4 and embedded PowerPC systems compile code for the
11327 Hurd-based GNU system.
11329 @item -mcall-netbsd
11330 @opindex mcall-netbsd
11331 On System V.4 and embedded PowerPC systems compile code for the
11332 NetBSD operating system.
11334 @item -maix-struct-return
11335 @opindex maix-struct-return
11336 Return all structures in memory (as specified by the AIX ABI)@.
11338 @item -msvr4-struct-return
11339 @opindex msvr4-struct-return
11340 Return structures smaller than 8 bytes in registers (as specified by the
11343 @item -mabi=@var{abi-type}
11345 Extend the current ABI with a particular extension, or remove such extension.
11346 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11350 @itemx -mno-prototype
11351 @opindex mprototype
11352 @opindex mno-prototype
11353 On System V.4 and embedded PowerPC systems assume that all calls to
11354 variable argument functions are properly prototyped. Otherwise, the
11355 compiler must insert an instruction before every non prototyped call to
11356 set or clear bit 6 of the condition code register (@var{CR}) to
11357 indicate whether floating point values were passed in the floating point
11358 registers in case the function takes a variable arguments. With
11359 @option{-mprototype}, only calls to prototyped variable argument functions
11360 will set or clear the bit.
11364 On embedded PowerPC systems, assume that the startup module is called
11365 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11366 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11371 On embedded PowerPC systems, assume that the startup module is called
11372 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11377 On embedded PowerPC systems, assume that the startup module is called
11378 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11381 @item -myellowknife
11382 @opindex myellowknife
11383 On embedded PowerPC systems, assume that the startup module is called
11384 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11389 On System V.4 and embedded PowerPC systems, specify that you are
11390 compiling for a VxWorks system.
11394 Specify that you are compiling for the WindISS simulation environment.
11398 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11399 header to indicate that @samp{eabi} extended relocations are used.
11405 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11406 Embedded Applications Binary Interface (eabi) which is a set of
11407 modifications to the System V.4 specifications. Selecting @option{-meabi}
11408 means that the stack is aligned to an 8 byte boundary, a function
11409 @code{__eabi} is called to from @code{main} to set up the eabi
11410 environment, and the @option{-msdata} option can use both @code{r2} and
11411 @code{r13} to point to two separate small data areas. Selecting
11412 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11413 do not call an initialization function from @code{main}, and the
11414 @option{-msdata} option will only use @code{r13} to point to a single
11415 small data area. The @option{-meabi} option is on by default if you
11416 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11419 @opindex msdata=eabi
11420 On System V.4 and embedded PowerPC systems, put small initialized
11421 @code{const} global and static data in the @samp{.sdata2} section, which
11422 is pointed to by register @code{r2}. Put small initialized
11423 non-@code{const} global and static data in the @samp{.sdata} section,
11424 which is pointed to by register @code{r13}. Put small uninitialized
11425 global and static data in the @samp{.sbss} section, which is adjacent to
11426 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11427 incompatible with the @option{-mrelocatable} option. The
11428 @option{-msdata=eabi} option also sets the @option{-memb} option.
11431 @opindex msdata=sysv
11432 On System V.4 and embedded PowerPC systems, put small global and static
11433 data in the @samp{.sdata} section, which is pointed to by register
11434 @code{r13}. Put small uninitialized global and static data in the
11435 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
11436 The @option{-msdata=sysv} option is incompatible with the
11437 @option{-mrelocatable} option.
11439 @item -msdata=default
11441 @opindex msdata=default
11443 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
11444 compile code the same as @option{-msdata=eabi}, otherwise compile code the
11445 same as @option{-msdata=sysv}.
11448 @opindex msdata-data
11449 On System V.4 and embedded PowerPC systems, put small global and static
11450 data in the @samp{.sdata} section. Put small uninitialized global and
11451 static data in the @samp{.sbss} section. Do not use register @code{r13}
11452 to address small data however. This is the default behavior unless
11453 other @option{-msdata} options are used.
11457 @opindex msdata=none
11459 On embedded PowerPC systems, put all initialized global and static data
11460 in the @samp{.data} section, and all uninitialized data in the
11461 @samp{.bss} section.
11465 @cindex smaller data references (PowerPC)
11466 @cindex .sdata/.sdata2 references (PowerPC)
11467 On embedded PowerPC systems, put global and static items less than or
11468 equal to @var{num} bytes into the small data or bss sections instead of
11469 the normal data or bss section. By default, @var{num} is 8. The
11470 @option{-G @var{num}} switch is also passed to the linker.
11471 All modules should be compiled with the same @option{-G @var{num}} value.
11474 @itemx -mno-regnames
11476 @opindex mno-regnames
11477 On System V.4 and embedded PowerPC systems do (do not) emit register
11478 names in the assembly language output using symbolic forms.
11481 @itemx -mno-longcall
11483 @opindex mno-longcall
11484 Default to making all function calls indirectly, using a register, so
11485 that functions which reside further than 32 megabytes (33,554,432
11486 bytes) from the current location can be called. This setting can be
11487 overridden by the @code{shortcall} function attribute, or by
11488 @code{#pragma longcall(0)}.
11490 Some linkers are capable of detecting out-of-range calls and generating
11491 glue code on the fly. On these systems, long calls are unnecessary and
11492 generate slower code. As of this writing, the AIX linker can do this,
11493 as can the GNU linker for PowerPC/64. It is planned to add this feature
11494 to the GNU linker for 32-bit PowerPC systems as well.
11496 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
11497 callee, L42'', plus a ``branch island'' (glue code). The two target
11498 addresses represent the callee and the ``branch island''. The
11499 Darwin/PPC linker will prefer the first address and generate a ``bl
11500 callee'' if the PPC ``bl'' instruction will reach the callee directly;
11501 otherwise, the linker will generate ``bl L42'' to call the ``branch
11502 island''. The ``branch island'' is appended to the body of the
11503 calling function; it computes the full 32-bit address of the callee
11506 On Mach-O (Darwin) systems, this option directs the compiler emit to
11507 the glue for every direct call, and the Darwin linker decides whether
11508 to use or discard it.
11510 In the future, we may cause GCC to ignore all longcall specifications
11511 when the linker is known to generate glue.
11515 Adds support for multithreading with the @dfn{pthreads} library.
11516 This option sets flags for both the preprocessor and linker.
11520 @node S/390 and zSeries Options
11521 @subsection S/390 and zSeries Options
11522 @cindex S/390 and zSeries Options
11524 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
11528 @itemx -msoft-float
11529 @opindex mhard-float
11530 @opindex msoft-float
11531 Use (do not use) the hardware floating-point instructions and registers
11532 for floating-point operations. When @option{-msoft-float} is specified,
11533 functions in @file{libgcc.a} will be used to perform floating-point
11534 operations. When @option{-mhard-float} is specified, the compiler
11535 generates IEEE floating-point instructions. This is the default.
11538 @itemx -mno-backchain
11539 @opindex mbackchain
11540 @opindex mno-backchain
11541 Store (do not store) the address of the caller's frame as backchain pointer
11542 into the callee's stack frame.
11543 A backchain may be needed to allow debugging using tools that do not understand
11544 DWARF-2 call frame information.
11545 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
11546 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
11547 the backchain is placed into the topmost word of the 96/160 byte register
11550 In general, code compiled with @option{-mbackchain} is call-compatible with
11551 code compiled with @option{-mmo-backchain}; however, use of the backchain
11552 for debugging purposes usually requires that the whole binary is built with
11553 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
11554 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11555 to build a linux kernel use @option{-msoft-float}.
11557 The default is to not maintain the backchain.
11559 @item -mpacked-stack
11560 @item -mno-packed-stack
11561 @opindex mpacked-stack
11562 @opindex mno-packed-stack
11563 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
11564 specified, the compiler uses the all fields of the 96/160 byte register save
11565 area only for their default purpose; unused fields still take up stack space.
11566 When @option{-mpacked-stack} is specified, register save slots are densely
11567 packed at the top of the register save area; unused space is reused for other
11568 purposes, allowing for more efficient use of the available stack space.
11569 However, when @option{-mbackchain} is also in effect, the topmost word of
11570 the save area is always used to store the backchain, and the return address
11571 register is always saved two words below the backchain.
11573 As long as the stack frame backchain is not used, code generated with
11574 @option{-mpacked-stack} is call-compatible with code generated with
11575 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
11576 S/390 or zSeries generated code that uses the stack frame backchain at run
11577 time, not just for debugging purposes. Such code is not call-compatible
11578 with code compiled with @option{-mpacked-stack}. Also, note that the
11579 combination of @option{-mbackchain},
11580 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11581 to build a linux kernel use @option{-msoft-float}.
11583 The default is to not use the packed stack layout.
11586 @itemx -mno-small-exec
11587 @opindex msmall-exec
11588 @opindex mno-small-exec
11589 Generate (or do not generate) code using the @code{bras} instruction
11590 to do subroutine calls.
11591 This only works reliably if the total executable size does not
11592 exceed 64k. The default is to use the @code{basr} instruction instead,
11593 which does not have this limitation.
11599 When @option{-m31} is specified, generate code compliant to the
11600 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
11601 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
11602 particular to generate 64-bit instructions. For the @samp{s390}
11603 targets, the default is @option{-m31}, while the @samp{s390x}
11604 targets default to @option{-m64}.
11610 When @option{-mzarch} is specified, generate code using the
11611 instructions available on z/Architecture.
11612 When @option{-mesa} is specified, generate code using the
11613 instructions available on ESA/390. Note that @option{-mesa} is
11614 not possible with @option{-m64}.
11615 When generating code compliant to the GNU/Linux for S/390 ABI,
11616 the default is @option{-mesa}. When generating code compliant
11617 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
11623 Generate (or do not generate) code using the @code{mvcle} instruction
11624 to perform block moves. When @option{-mno-mvcle} is specified,
11625 use a @code{mvc} loop instead. This is the default unless optimizing for
11632 Print (or do not print) additional debug information when compiling.
11633 The default is to not print debug information.
11635 @item -march=@var{cpu-type}
11637 Generate code that will run on @var{cpu-type}, which is the name of a system
11638 representing a certain processor type. Possible values for
11639 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
11640 When generating code using the instructions available on z/Architecture,
11641 the default is @option{-march=z900}. Otherwise, the default is
11642 @option{-march=g5}.
11644 @item -mtune=@var{cpu-type}
11646 Tune to @var{cpu-type} everything applicable about the generated code,
11647 except for the ABI and the set of available instructions.
11648 The list of @var{cpu-type} values is the same as for @option{-march}.
11649 The default is the value used for @option{-march}.
11652 @itemx -mno-tpf-trace
11653 @opindex mtpf-trace
11654 @opindex mno-tpf-trace
11655 Generate code that adds (does not add) in TPF OS specific branches to trace
11656 routines in the operating system. This option is off by default, even
11657 when compiling for the TPF OS@.
11660 @itemx -mno-fused-madd
11661 @opindex mfused-madd
11662 @opindex mno-fused-madd
11663 Generate code that uses (does not use) the floating point multiply and
11664 accumulate instructions. These instructions are generated by default if
11665 hardware floating point is used.
11667 @item -mwarn-framesize=@var{framesize}
11668 @opindex mwarn-framesize
11669 Emit a warning if the current function exceeds the given frame size. Because
11670 this is a compile time check it doesn't need to be a real problem when the program
11671 runs. It is intended to identify functions which most probably cause
11672 a stack overflow. It is useful to be used in an environment with limited stack
11673 size e.g.@: the linux kernel.
11675 @item -mwarn-dynamicstack
11676 @opindex mwarn-dynamicstack
11677 Emit a warning if the function calls alloca or uses dynamically
11678 sized arrays. This is generally a bad idea with a limited stack size.
11680 @item -mstack-guard=@var{stack-guard}
11681 @item -mstack-size=@var{stack-size}
11682 @opindex mstack-guard
11683 @opindex mstack-size
11684 These arguments always have to be used in conjunction. If they are present the s390
11685 back end emits additional instructions in the function prologue which trigger a trap
11686 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
11687 (remember that the stack on s390 grows downward). These options are intended to
11688 be used to help debugging stack overflow problems. The additionally emitted code
11689 causes only little overhead and hence can also be used in production like systems
11690 without greater performance degradation. The given values have to be exact
11691 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
11693 In order to be efficient the extra code makes the assumption that the stack starts
11694 at an address aligned to the value given by @var{stack-size}.
11698 @subsection SH Options
11700 These @samp{-m} options are defined for the SH implementations:
11705 Generate code for the SH1.
11709 Generate code for the SH2.
11712 Generate code for the SH2e.
11716 Generate code for the SH3.
11720 Generate code for the SH3e.
11724 Generate code for the SH4 without a floating-point unit.
11726 @item -m4-single-only
11727 @opindex m4-single-only
11728 Generate code for the SH4 with a floating-point unit that only
11729 supports single-precision arithmetic.
11733 Generate code for the SH4 assuming the floating-point unit is in
11734 single-precision mode by default.
11738 Generate code for the SH4.
11742 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
11743 floating-point unit is not used.
11745 @item -m4a-single-only
11746 @opindex m4a-single-only
11747 Generate code for the SH4a, in such a way that no double-precision
11748 floating point operations are used.
11751 @opindex m4a-single
11752 Generate code for the SH4a assuming the floating-point unit is in
11753 single-precision mode by default.
11757 Generate code for the SH4a.
11761 Same as @option{-m4a-nofpu}, except that it implicitly passes
11762 @option{-dsp} to the assembler. GCC doesn't generate any DSP
11763 instructions at the moment.
11767 Compile code for the processor in big endian mode.
11771 Compile code for the processor in little endian mode.
11775 Align doubles at 64-bit boundaries. Note that this changes the calling
11776 conventions, and thus some functions from the standard C library will
11777 not work unless you recompile it first with @option{-mdalign}.
11781 Shorten some address references at link time, when possible; uses the
11782 linker option @option{-relax}.
11786 Use 32-bit offsets in @code{switch} tables. The default is to use
11791 Enable the use of the instruction @code{fmovd}.
11795 Comply with the calling conventions defined by Renesas.
11799 Comply with the calling conventions defined by Renesas.
11803 Comply with the calling conventions defined for GCC before the Renesas
11804 conventions were available. This option is the default for all
11805 targets of the SH toolchain except for @samp{sh-symbianelf}.
11808 @opindex mnomacsave
11809 Mark the @code{MAC} register as call-clobbered, even if
11810 @option{-mhitachi} is given.
11814 Increase IEEE-compliance of floating-point code.
11815 At the moment, this is equivalent to @option{-fno-finite-math-only}.
11816 When generating 16 bit SH opcodes, getting IEEE-conforming results for
11817 comparisons of NANs / infinities incurs extra overhead in every
11818 floating point comparison, therefore the default is set to
11819 @option{-ffinite-math-only}.
11823 Dump instruction size and location in the assembly code.
11826 @opindex mpadstruct
11827 This option is deprecated. It pads structures to multiple of 4 bytes,
11828 which is incompatible with the SH ABI@.
11832 Optimize for space instead of speed. Implied by @option{-Os}.
11835 @opindex mprefergot
11836 When generating position-independent code, emit function calls using
11837 the Global Offset Table instead of the Procedure Linkage Table.
11841 Generate a library function call to invalidate instruction cache
11842 entries, after fixing up a trampoline. This library function call
11843 doesn't assume it can write to the whole memory address space. This
11844 is the default when the target is @code{sh-*-linux*}.
11846 @item -multcost=@var{number}
11847 @opindex multcost=@var{number}
11848 Set the cost to assume for a multiply insn.
11850 @item -mdiv=@var{strategy}
11851 @opindex mdiv=@var{strategy}
11852 Set the division strategy to use for SHmedia code. @var{strategy} must be
11853 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
11854 inv:call2, inv:fp .
11855 "fp" performs the operation in floating point. This has a very high latency,
11856 but needs only a few instructions, so it might be a good choice if
11857 your code has enough easily exploitable ILP to allow the compiler to
11858 schedule the floating point instructions together with other instructions.
11859 Division by zero causes a floating point exception.
11860 "inv" uses integer operations to calculate the inverse of the divisor,
11861 and then multiplies the dividend with the inverse. This strategy allows
11862 cse and hoisting of the inverse calculation. Division by zero calculates
11863 an unspecified result, but does not trap.
11864 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
11865 have been found, or if the entire operation has been hoisted to the same
11866 place, the last stages of the inverse calculation are intertwined with the
11867 final multiply to reduce the overall latency, at the expense of using a few
11868 more instructions, and thus offering fewer scheduling opportunities with
11870 "call" calls a library function that usually implements the inv:minlat
11872 This gives high code density for m5-*media-nofpu compilations.
11873 "call2" uses a different entry point of the same library function, where it
11874 assumes that a pointer to a lookup table has already been set up, which
11875 exposes the pointer load to cse / code hoisting optimizations.
11876 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
11877 code generation, but if the code stays unoptimized, revert to the "call",
11878 "call2", or "fp" strategies, respectively. Note that the
11879 potentially-trapping side effect of division by zero is carried by a
11880 separate instruction, so it is possible that all the integer instructions
11881 are hoisted out, but the marker for the side effect stays where it is.
11882 A recombination to fp operations or a call is not possible in that case.
11883 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
11884 that the inverse calculation was nor separated from the multiply, they speed
11885 up division where the dividend fits into 20 bits (plus sign where applicable),
11886 by inserting a test to skip a number of operations in this case; this test
11887 slows down the case of larger dividends. inv20u assumes the case of a such
11888 a small dividend to be unlikely, and inv20l assumes it to be likely.
11890 @item -mdivsi3_libfunc=@var{name}
11891 @opindex mdivsi3_libfunc=@var{name}
11892 Set the name of the library function used for 32 bit signed division to
11893 @var{name}. This only affect the name used in the call and inv:call
11894 division strategies, and the compiler will still expect the same
11895 sets of input/output/clobbered registers as if this option was not present.
11897 @item -madjust-unroll
11898 @opindex madjust-unroll
11899 Throttle unrolling to avoid thrashing target registers.
11900 This option only has an effect if the gcc code base supports the
11901 TARGET_ADJUST_UNROLL_MAX target hook.
11903 @item -mindexed-addressing
11904 @opindex mindexed-addressing
11905 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
11906 This is only safe if the hardware and/or OS implement 32 bit wrap-around
11907 semantics for the indexed addressing mode. The architecture allows the
11908 implementation of processors with 64 bit MMU, which the OS could use to
11909 get 32 bit addressing, but since no current hardware implementation supports
11910 this or any other way to make the indexed addressing mode safe to use in
11911 the 32 bit ABI, the default is -mno-indexed-addressing.
11913 @item -mgettrcost=@var{number}
11914 @opindex mgettrcost=@var{number}
11915 Set the cost assumed for the gettr instruction to @var{number}.
11916 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
11920 Assume pt* instructions won't trap. This will generally generate better
11921 scheduled code, but is unsafe on current hardware. The current architecture
11922 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
11923 This has the unintentional effect of making it unsafe to schedule ptabs /
11924 ptrel before a branch, or hoist it out of a loop. For example,
11925 __do_global_ctors, a part of libgcc that runs constructors at program
11926 startup, calls functions in a list which is delimited by -1. With the
11927 -mpt-fixed option, the ptabs will be done before testing against -1.
11928 That means that all the constructors will be run a bit quicker, but when
11929 the loop comes to the end of the list, the program crashes because ptabs
11930 loads -1 into a target register. Since this option is unsafe for any
11931 hardware implementing the current architecture specification, the default
11932 is -mno-pt-fixed. Unless the user specifies a specific cost with
11933 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
11934 this deters register allocation using target registers for storing
11937 @item -minvalid-symbols
11938 @opindex minvalid-symbols
11939 Assume symbols might be invalid. Ordinary function symbols generated by
11940 the compiler will always be valid to load with movi/shori/ptabs or
11941 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
11942 to generate symbols that will cause ptabs / ptrel to trap.
11943 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
11944 It will then prevent cross-basic-block cse, hoisting and most scheduling
11945 of symbol loads. The default is @option{-mno-invalid-symbols}.
11948 @node SPARC Options
11949 @subsection SPARC Options
11950 @cindex SPARC options
11952 These @samp{-m} options are supported on the SPARC:
11955 @item -mno-app-regs
11957 @opindex mno-app-regs
11959 Specify @option{-mapp-regs} to generate output using the global registers
11960 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
11963 To be fully SVR4 ABI compliant at the cost of some performance loss,
11964 specify @option{-mno-app-regs}. You should compile libraries and system
11965 software with this option.
11968 @itemx -mhard-float
11970 @opindex mhard-float
11971 Generate output containing floating point instructions. This is the
11975 @itemx -msoft-float
11977 @opindex msoft-float
11978 Generate output containing library calls for floating point.
11979 @strong{Warning:} the requisite libraries are not available for all SPARC
11980 targets. Normally the facilities of the machine's usual C compiler are
11981 used, but this cannot be done directly in cross-compilation. You must make
11982 your own arrangements to provide suitable library functions for
11983 cross-compilation. The embedded targets @samp{sparc-*-aout} and
11984 @samp{sparclite-*-*} do provide software floating point support.
11986 @option{-msoft-float} changes the calling convention in the output file;
11987 therefore, it is only useful if you compile @emph{all} of a program with
11988 this option. In particular, you need to compile @file{libgcc.a}, the
11989 library that comes with GCC, with @option{-msoft-float} in order for
11992 @item -mhard-quad-float
11993 @opindex mhard-quad-float
11994 Generate output containing quad-word (long double) floating point
11997 @item -msoft-quad-float
11998 @opindex msoft-quad-float
11999 Generate output containing library calls for quad-word (long double)
12000 floating point instructions. The functions called are those specified
12001 in the SPARC ABI@. This is the default.
12003 As of this writing, there are no SPARC implementations that have hardware
12004 support for the quad-word floating point instructions. They all invoke
12005 a trap handler for one of these instructions, and then the trap handler
12006 emulates the effect of the instruction. Because of the trap handler overhead,
12007 this is much slower than calling the ABI library routines. Thus the
12008 @option{-msoft-quad-float} option is the default.
12010 @item -mno-unaligned-doubles
12011 @itemx -munaligned-doubles
12012 @opindex mno-unaligned-doubles
12013 @opindex munaligned-doubles
12014 Assume that doubles have 8 byte alignment. This is the default.
12016 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12017 alignment only if they are contained in another type, or if they have an
12018 absolute address. Otherwise, it assumes they have 4 byte alignment.
12019 Specifying this option avoids some rare compatibility problems with code
12020 generated by other compilers. It is not the default because it results
12021 in a performance loss, especially for floating point code.
12023 @item -mno-faster-structs
12024 @itemx -mfaster-structs
12025 @opindex mno-faster-structs
12026 @opindex mfaster-structs
12027 With @option{-mfaster-structs}, the compiler assumes that structures
12028 should have 8 byte alignment. This enables the use of pairs of
12029 @code{ldd} and @code{std} instructions for copies in structure
12030 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12031 However, the use of this changed alignment directly violates the SPARC
12032 ABI@. Thus, it's intended only for use on targets where the developer
12033 acknowledges that their resulting code will not be directly in line with
12034 the rules of the ABI@.
12036 @item -mimpure-text
12037 @opindex mimpure-text
12038 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12039 the compiler to not pass @option{-z text} to the linker when linking a
12040 shared object. Using this option, you can link position-dependent
12041 code into a shared object.
12043 @option{-mimpure-text} suppresses the ``relocations remain against
12044 allocatable but non-writable sections'' linker error message.
12045 However, the necessary relocations will trigger copy-on-write, and the
12046 shared object is not actually shared across processes. Instead of
12047 using @option{-mimpure-text}, you should compile all source code with
12048 @option{-fpic} or @option{-fPIC}.
12050 This option is only available on SunOS and Solaris.
12052 @item -mcpu=@var{cpu_type}
12054 Set the instruction set, register set, and instruction scheduling parameters
12055 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12056 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12057 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12058 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
12059 @samp{ultrasparc3}.
12061 Default instruction scheduling parameters are used for values that select
12062 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12063 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12065 Here is a list of each supported architecture and their supported
12070 v8: supersparc, hypersparc
12071 sparclite: f930, f934, sparclite86x
12073 v9: ultrasparc, ultrasparc3
12076 By default (unless configured otherwise), GCC generates code for the V7
12077 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12078 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12079 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12080 SPARCStation 1, 2, IPX etc.
12082 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12083 architecture. The only difference from V7 code is that the compiler emits
12084 the integer multiply and integer divide instructions which exist in SPARC-V8
12085 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12086 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12089 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12090 the SPARC architecture. This adds the integer multiply, integer divide step
12091 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12092 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12093 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12094 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12095 MB86934 chip, which is the more recent SPARClite with FPU@.
12097 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12098 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12099 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12100 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12101 optimizes it for the TEMIC SPARClet chip.
12103 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12104 architecture. This adds 64-bit integer and floating-point move instructions,
12105 3 additional floating-point condition code registers and conditional move
12106 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12107 optimizes it for the Sun UltraSPARC I/II chips. With
12108 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12109 Sun UltraSPARC III chip.
12111 @item -mtune=@var{cpu_type}
12113 Set the instruction scheduling parameters for machine type
12114 @var{cpu_type}, but do not set the instruction set or register set that the
12115 option @option{-mcpu=@var{cpu_type}} would.
12117 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12118 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12119 that select a particular cpu implementation. Those are @samp{cypress},
12120 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12121 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
12122 @samp{ultrasparc3}.
12127 @opindex mno-v8plus
12128 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12129 difference from the V8 ABI is that the global and out registers are
12130 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12131 mode for all SPARC-V9 processors.
12137 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12138 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12141 These @samp{-m} options are supported in addition to the above
12142 on SPARC-V9 processors in 64-bit environments:
12145 @item -mlittle-endian
12146 @opindex mlittle-endian
12147 Generate code for a processor running in little-endian mode. It is only
12148 available for a few configurations and most notably not on Solaris and Linux.
12154 Generate code for a 32-bit or 64-bit environment.
12155 The 32-bit environment sets int, long and pointer to 32 bits.
12156 The 64-bit environment sets int to 32 bits and long and pointer
12159 @item -mcmodel=medlow
12160 @opindex mcmodel=medlow
12161 Generate code for the Medium/Low code model: 64-bit addresses, programs
12162 must be linked in the low 32 bits of memory. Programs can be statically
12163 or dynamically linked.
12165 @item -mcmodel=medmid
12166 @opindex mcmodel=medmid
12167 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12168 must be linked in the low 44 bits of memory, the text and data segments must
12169 be less than 2GB in size and the data segment must be located within 2GB of
12172 @item -mcmodel=medany
12173 @opindex mcmodel=medany
12174 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12175 may be linked anywhere in memory, the text and data segments must be less
12176 than 2GB in size and the data segment must be located within 2GB of the
12179 @item -mcmodel=embmedany
12180 @opindex mcmodel=embmedany
12181 Generate code for the Medium/Anywhere code model for embedded systems:
12182 64-bit addresses, the text and data segments must be less than 2GB in
12183 size, both starting anywhere in memory (determined at link time). The
12184 global register %g4 points to the base of the data segment. Programs
12185 are statically linked and PIC is not supported.
12188 @itemx -mno-stack-bias
12189 @opindex mstack-bias
12190 @opindex mno-stack-bias
12191 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12192 frame pointer if present, are offset by @minus{}2047 which must be added back
12193 when making stack frame references. This is the default in 64-bit mode.
12194 Otherwise, assume no such offset is present.
12197 These switches are supported in addition to the above on Solaris:
12202 Add support for multithreading using the Solaris threads library. This
12203 option sets flags for both the preprocessor and linker. This option does
12204 not affect the thread safety of object code produced by the compiler or
12205 that of libraries supplied with it.
12209 Add support for multithreading using the POSIX threads library. This
12210 option sets flags for both the preprocessor and linker. This option does
12211 not affect the thread safety of object code produced by the compiler or
12212 that of libraries supplied with it.
12215 @node System V Options
12216 @subsection Options for System V
12218 These additional options are available on System V Release 4 for
12219 compatibility with other compilers on those systems:
12224 Create a shared object.
12225 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12229 Identify the versions of each tool used by the compiler, in a
12230 @code{.ident} assembler directive in the output.
12234 Refrain from adding @code{.ident} directives to the output file (this is
12237 @item -YP,@var{dirs}
12239 Search the directories @var{dirs}, and no others, for libraries
12240 specified with @option{-l}.
12242 @item -Ym,@var{dir}
12244 Look in the directory @var{dir} to find the M4 preprocessor.
12245 The assembler uses this option.
12246 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12247 @c the generic assembler that comes with Solaris takes just -Ym.
12250 @node TMS320C3x/C4x Options
12251 @subsection TMS320C3x/C4x Options
12252 @cindex TMS320C3x/C4x Options
12254 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12258 @item -mcpu=@var{cpu_type}
12260 Set the instruction set, register set, and instruction scheduling
12261 parameters for machine type @var{cpu_type}. Supported values for
12262 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12263 @samp{c44}. The default is @samp{c40} to generate code for the
12268 @itemx -msmall-memory
12270 @opindex mbig-memory
12272 @opindex msmall-memory
12274 Generates code for the big or small memory model. The small memory
12275 model assumed that all data fits into one 64K word page. At run-time
12276 the data page (DP) register must be set to point to the 64K page
12277 containing the .bss and .data program sections. The big memory model is
12278 the default and requires reloading of the DP register for every direct
12285 Allow (disallow) allocation of general integer operands into the block
12286 count register BK@.
12292 Enable (disable) generation of code using decrement and branch,
12293 DBcond(D), instructions. This is enabled by default for the C4x. To be
12294 on the safe side, this is disabled for the C3x, since the maximum
12295 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12296 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12297 that it can utilize the decrement and branch instruction, but will give
12298 up if there is more than one memory reference in the loop. Thus a loop
12299 where the loop counter is decremented can generate slightly more
12300 efficient code, in cases where the RPTB instruction cannot be utilized.
12302 @item -mdp-isr-reload
12304 @opindex mdp-isr-reload
12306 Force the DP register to be saved on entry to an interrupt service
12307 routine (ISR), reloaded to point to the data section, and restored on
12308 exit from the ISR@. This should not be required unless someone has
12309 violated the small memory model by modifying the DP register, say within
12316 For the C3x use the 24-bit MPYI instruction for integer multiplies
12317 instead of a library call to guarantee 32-bit results. Note that if one
12318 of the operands is a constant, then the multiplication will be performed
12319 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12320 then squaring operations are performed inline instead of a library call.
12323 @itemx -mno-fast-fix
12325 @opindex mno-fast-fix
12326 The C3x/C4x FIX instruction to convert a floating point value to an
12327 integer value chooses the nearest integer less than or equal to the
12328 floating point value rather than to the nearest integer. Thus if the
12329 floating point number is negative, the result will be incorrectly
12330 truncated an additional code is necessary to detect and correct this
12331 case. This option can be used to disable generation of the additional
12332 code required to correct the result.
12338 Enable (disable) generation of repeat block sequences using the RPTB
12339 instruction for zero overhead looping. The RPTB construct is only used
12340 for innermost loops that do not call functions or jump across the loop
12341 boundaries. There is no advantage having nested RPTB loops due to the
12342 overhead required to save and restore the RC, RS, and RE registers.
12343 This is enabled by default with @option{-O2}.
12345 @item -mrpts=@var{count}
12349 Enable (disable) the use of the single instruction repeat instruction
12350 RPTS@. If a repeat block contains a single instruction, and the loop
12351 count can be guaranteed to be less than the value @var{count}, GCC will
12352 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12353 then a RPTS will be emitted even if the loop count cannot be determined
12354 at compile time. Note that the repeated instruction following RPTS does
12355 not have to be reloaded from memory each iteration, thus freeing up the
12356 CPU buses for operands. However, since interrupts are blocked by this
12357 instruction, it is disabled by default.
12359 @item -mloop-unsigned
12360 @itemx -mno-loop-unsigned
12361 @opindex mloop-unsigned
12362 @opindex mno-loop-unsigned
12363 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12364 is @math{2^{31} + 1} since these instructions test if the iteration count is
12365 negative to terminate the loop. If the iteration count is unsigned
12366 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12367 exceeded. This switch allows an unsigned iteration count.
12371 Try to emit an assembler syntax that the TI assembler (asm30) is happy
12372 with. This also enforces compatibility with the API employed by the TI
12373 C3x C compiler. For example, long doubles are passed as structures
12374 rather than in floating point registers.
12380 Generate code that uses registers (stack) for passing arguments to functions.
12381 By default, arguments are passed in registers where possible rather
12382 than by pushing arguments on to the stack.
12384 @item -mparallel-insns
12385 @itemx -mno-parallel-insns
12386 @opindex mparallel-insns
12387 @opindex mno-parallel-insns
12388 Allow the generation of parallel instructions. This is enabled by
12389 default with @option{-O2}.
12391 @item -mparallel-mpy
12392 @itemx -mno-parallel-mpy
12393 @opindex mparallel-mpy
12394 @opindex mno-parallel-mpy
12395 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
12396 provided @option{-mparallel-insns} is also specified. These instructions have
12397 tight register constraints which can pessimize the code generation
12398 of large functions.
12403 @subsection V850 Options
12404 @cindex V850 Options
12406 These @samp{-m} options are defined for V850 implementations:
12410 @itemx -mno-long-calls
12411 @opindex mlong-calls
12412 @opindex mno-long-calls
12413 Treat all calls as being far away (near). If calls are assumed to be
12414 far away, the compiler will always load the functions address up into a
12415 register, and call indirect through the pointer.
12421 Do not optimize (do optimize) basic blocks that use the same index
12422 pointer 4 or more times to copy pointer into the @code{ep} register, and
12423 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
12424 option is on by default if you optimize.
12426 @item -mno-prolog-function
12427 @itemx -mprolog-function
12428 @opindex mno-prolog-function
12429 @opindex mprolog-function
12430 Do not use (do use) external functions to save and restore registers
12431 at the prologue and epilogue of a function. The external functions
12432 are slower, but use less code space if more than one function saves
12433 the same number of registers. The @option{-mprolog-function} option
12434 is on by default if you optimize.
12438 Try to make the code as small as possible. At present, this just turns
12439 on the @option{-mep} and @option{-mprolog-function} options.
12441 @item -mtda=@var{n}
12443 Put static or global variables whose size is @var{n} bytes or less into
12444 the tiny data area that register @code{ep} points to. The tiny data
12445 area can hold up to 256 bytes in total (128 bytes for byte references).
12447 @item -msda=@var{n}
12449 Put static or global variables whose size is @var{n} bytes or less into
12450 the small data area that register @code{gp} points to. The small data
12451 area can hold up to 64 kilobytes.
12453 @item -mzda=@var{n}
12455 Put static or global variables whose size is @var{n} bytes or less into
12456 the first 32 kilobytes of memory.
12460 Specify that the target processor is the V850.
12463 @opindex mbig-switch
12464 Generate code suitable for big switch tables. Use this option only if
12465 the assembler/linker complain about out of range branches within a switch
12470 This option will cause r2 and r5 to be used in the code generated by
12471 the compiler. This setting is the default.
12473 @item -mno-app-regs
12474 @opindex mno-app-regs
12475 This option will cause r2 and r5 to be treated as fixed registers.
12479 Specify that the target processor is the V850E1. The preprocessor
12480 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
12481 this option is used.
12485 Specify that the target processor is the V850E@. The preprocessor
12486 constant @samp{__v850e__} will be defined if this option is used.
12488 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
12489 are defined then a default target processor will be chosen and the
12490 relevant @samp{__v850*__} preprocessor constant will be defined.
12492 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
12493 defined, regardless of which processor variant is the target.
12495 @item -mdisable-callt
12496 @opindex mdisable-callt
12497 This option will suppress generation of the CALLT instruction for the
12498 v850e and v850e1 flavors of the v850 architecture. The default is
12499 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
12504 @subsection VAX Options
12505 @cindex VAX options
12507 These @samp{-m} options are defined for the VAX:
12512 Do not output certain jump instructions (@code{aobleq} and so on)
12513 that the Unix assembler for the VAX cannot handle across long
12518 Do output those jump instructions, on the assumption that you
12519 will assemble with the GNU assembler.
12523 Output code for g-format floating point numbers instead of d-format.
12526 @node x86-64 Options
12527 @subsection x86-64 Options
12528 @cindex x86-64 options
12530 These are listed under @xref{i386 and x86-64 Options}.
12532 @node Xstormy16 Options
12533 @subsection Xstormy16 Options
12534 @cindex Xstormy16 Options
12536 These options are defined for Xstormy16:
12541 Choose startup files and linker script suitable for the simulator.
12544 @node Xtensa Options
12545 @subsection Xtensa Options
12546 @cindex Xtensa Options
12548 These options are supported for Xtensa targets:
12552 @itemx -mno-const16
12554 @opindex mno-const16
12555 Enable or disable use of @code{CONST16} instructions for loading
12556 constant values. The @code{CONST16} instruction is currently not a
12557 standard option from Tensilica. When enabled, @code{CONST16}
12558 instructions are always used in place of the standard @code{L32R}
12559 instructions. The use of @code{CONST16} is enabled by default only if
12560 the @code{L32R} instruction is not available.
12563 @itemx -mno-fused-madd
12564 @opindex mfused-madd
12565 @opindex mno-fused-madd
12566 Enable or disable use of fused multiply/add and multiply/subtract
12567 instructions in the floating-point option. This has no effect if the
12568 floating-point option is not also enabled. Disabling fused multiply/add
12569 and multiply/subtract instructions forces the compiler to use separate
12570 instructions for the multiply and add/subtract operations. This may be
12571 desirable in some cases where strict IEEE 754-compliant results are
12572 required: the fused multiply add/subtract instructions do not round the
12573 intermediate result, thereby producing results with @emph{more} bits of
12574 precision than specified by the IEEE standard. Disabling fused multiply
12575 add/subtract instructions also ensures that the program output is not
12576 sensitive to the compiler's ability to combine multiply and add/subtract
12579 @item -mtext-section-literals
12580 @itemx -mno-text-section-literals
12581 @opindex mtext-section-literals
12582 @opindex mno-text-section-literals
12583 Control the treatment of literal pools. The default is
12584 @option{-mno-text-section-literals}, which places literals in a separate
12585 section in the output file. This allows the literal pool to be placed
12586 in a data RAM/ROM, and it also allows the linker to combine literal
12587 pools from separate object files to remove redundant literals and
12588 improve code size. With @option{-mtext-section-literals}, the literals
12589 are interspersed in the text section in order to keep them as close as
12590 possible to their references. This may be necessary for large assembly
12593 @item -mtarget-align
12594 @itemx -mno-target-align
12595 @opindex mtarget-align
12596 @opindex mno-target-align
12597 When this option is enabled, GCC instructs the assembler to
12598 automatically align instructions to reduce branch penalties at the
12599 expense of some code density. The assembler attempts to widen density
12600 instructions to align branch targets and the instructions following call
12601 instructions. If there are not enough preceding safe density
12602 instructions to align a target, no widening will be performed. The
12603 default is @option{-mtarget-align}. These options do not affect the
12604 treatment of auto-aligned instructions like @code{LOOP}, which the
12605 assembler will always align, either by widening density instructions or
12606 by inserting no-op instructions.
12609 @itemx -mno-longcalls
12610 @opindex mlongcalls
12611 @opindex mno-longcalls
12612 When this option is enabled, GCC instructs the assembler to translate
12613 direct calls to indirect calls unless it can determine that the target
12614 of a direct call is in the range allowed by the call instruction. This
12615 translation typically occurs for calls to functions in other source
12616 files. Specifically, the assembler translates a direct @code{CALL}
12617 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
12618 The default is @option{-mno-longcalls}. This option should be used in
12619 programs where the call target can potentially be out of range. This
12620 option is implemented in the assembler, not the compiler, so the
12621 assembly code generated by GCC will still show direct call
12622 instructions---look at the disassembled object code to see the actual
12623 instructions. Note that the assembler will use an indirect call for
12624 every cross-file call, not just those that really will be out of range.
12627 @node zSeries Options
12628 @subsection zSeries Options
12629 @cindex zSeries options
12631 These are listed under @xref{S/390 and zSeries Options}.
12633 @node Code Gen Options
12634 @section Options for Code Generation Conventions
12635 @cindex code generation conventions
12636 @cindex options, code generation
12637 @cindex run-time options
12639 These machine-independent options control the interface conventions
12640 used in code generation.
12642 Most of them have both positive and negative forms; the negative form
12643 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
12644 one of the forms is listed---the one which is not the default. You
12645 can figure out the other form by either removing @samp{no-} or adding
12649 @item -fbounds-check
12650 @opindex fbounds-check
12651 For front-ends that support it, generate additional code to check that
12652 indices used to access arrays are within the declared range. This is
12653 currently only supported by the Java and Fortran 77 front-ends, where
12654 this option defaults to true and false respectively.
12658 This option generates traps for signed overflow on addition, subtraction,
12659 multiplication operations.
12663 This option instructs the compiler to assume that signed arithmetic
12664 overflow of addition, subtraction and multiplication wraps around
12665 using twos-complement representation. This flag enables some optimizations
12666 and disables others. This option is enabled by default for the Java
12667 front-end, as required by the Java language specification.
12670 @opindex fexceptions
12671 Enable exception handling. Generates extra code needed to propagate
12672 exceptions. For some targets, this implies GCC will generate frame
12673 unwind information for all functions, which can produce significant data
12674 size overhead, although it does not affect execution. If you do not
12675 specify this option, GCC will enable it by default for languages like
12676 C++ which normally require exception handling, and disable it for
12677 languages like C that do not normally require it. However, you may need
12678 to enable this option when compiling C code that needs to interoperate
12679 properly with exception handlers written in C++. You may also wish to
12680 disable this option if you are compiling older C++ programs that don't
12681 use exception handling.
12683 @item -fnon-call-exceptions
12684 @opindex fnon-call-exceptions
12685 Generate code that allows trapping instructions to throw exceptions.
12686 Note that this requires platform-specific runtime support that does
12687 not exist everywhere. Moreover, it only allows @emph{trapping}
12688 instructions to throw exceptions, i.e.@: memory references or floating
12689 point instructions. It does not allow exceptions to be thrown from
12690 arbitrary signal handlers such as @code{SIGALRM}.
12692 @item -funwind-tables
12693 @opindex funwind-tables
12694 Similar to @option{-fexceptions}, except that it will just generate any needed
12695 static data, but will not affect the generated code in any other way.
12696 You will normally not enable this option; instead, a language processor
12697 that needs this handling would enable it on your behalf.
12699 @item -fasynchronous-unwind-tables
12700 @opindex fasynchronous-unwind-tables
12701 Generate unwind table in dwarf2 format, if supported by target machine. The
12702 table is exact at each instruction boundary, so it can be used for stack
12703 unwinding from asynchronous events (such as debugger or garbage collector).
12705 @item -fpcc-struct-return
12706 @opindex fpcc-struct-return
12707 Return ``short'' @code{struct} and @code{union} values in memory like
12708 longer ones, rather than in registers. This convention is less
12709 efficient, but it has the advantage of allowing intercallability between
12710 GCC-compiled files and files compiled with other compilers, particularly
12711 the Portable C Compiler (pcc).
12713 The precise convention for returning structures in memory depends
12714 on the target configuration macros.
12716 Short structures and unions are those whose size and alignment match
12717 that of some integer type.
12719 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
12720 switch is not binary compatible with code compiled with the
12721 @option{-freg-struct-return} switch.
12722 Use it to conform to a non-default application binary interface.
12724 @item -freg-struct-return
12725 @opindex freg-struct-return
12726 Return @code{struct} and @code{union} values in registers when possible.
12727 This is more efficient for small structures than
12728 @option{-fpcc-struct-return}.
12730 If you specify neither @option{-fpcc-struct-return} nor
12731 @option{-freg-struct-return}, GCC defaults to whichever convention is
12732 standard for the target. If there is no standard convention, GCC
12733 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
12734 the principal compiler. In those cases, we can choose the standard, and
12735 we chose the more efficient register return alternative.
12737 @strong{Warning:} code compiled with the @option{-freg-struct-return}
12738 switch is not binary compatible with code compiled with the
12739 @option{-fpcc-struct-return} switch.
12740 Use it to conform to a non-default application binary interface.
12742 @item -fshort-enums
12743 @opindex fshort-enums
12744 Allocate to an @code{enum} type only as many bytes as it needs for the
12745 declared range of possible values. Specifically, the @code{enum} type
12746 will be equivalent to the smallest integer type which has enough room.
12748 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
12749 code that is not binary compatible with code generated without that switch.
12750 Use it to conform to a non-default application binary interface.
12752 @item -fshort-double
12753 @opindex fshort-double
12754 Use the same size for @code{double} as for @code{float}.
12756 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
12757 code that is not binary compatible with code generated without that switch.
12758 Use it to conform to a non-default application binary interface.
12760 @item -fshort-wchar
12761 @opindex fshort-wchar
12762 Override the underlying type for @samp{wchar_t} to be @samp{short
12763 unsigned int} instead of the default for the target. This option is
12764 useful for building programs to run under WINE@.
12766 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
12767 code that is not binary compatible with code generated without that switch.
12768 Use it to conform to a non-default application binary interface.
12770 @item -fshared-data
12771 @opindex fshared-data
12772 Requests that the data and non-@code{const} variables of this
12773 compilation be shared data rather than private data. The distinction
12774 makes sense only on certain operating systems, where shared data is
12775 shared between processes running the same program, while private data
12776 exists in one copy per process.
12779 @opindex fno-common
12780 In C, allocate even uninitialized global variables in the data section of the
12781 object file, rather than generating them as common blocks. This has the
12782 effect that if the same variable is declared (without @code{extern}) in
12783 two different compilations, you will get an error when you link them.
12784 The only reason this might be useful is if you wish to verify that the
12785 program will work on other systems which always work this way.
12789 Ignore the @samp{#ident} directive.
12791 @item -finhibit-size-directive
12792 @opindex finhibit-size-directive
12793 Don't output a @code{.size} assembler directive, or anything else that
12794 would cause trouble if the function is split in the middle, and the
12795 two halves are placed at locations far apart in memory. This option is
12796 used when compiling @file{crtstuff.c}; you should not need to use it
12799 @item -fverbose-asm
12800 @opindex fverbose-asm
12801 Put extra commentary information in the generated assembly code to
12802 make it more readable. This option is generally only of use to those
12803 who actually need to read the generated assembly code (perhaps while
12804 debugging the compiler itself).
12806 @option{-fno-verbose-asm}, the default, causes the
12807 extra information to be omitted and is useful when comparing two assembler
12812 @cindex global offset table
12814 Generate position-independent code (PIC) suitable for use in a shared
12815 library, if supported for the target machine. Such code accesses all
12816 constant addresses through a global offset table (GOT)@. The dynamic
12817 loader resolves the GOT entries when the program starts (the dynamic
12818 loader is not part of GCC; it is part of the operating system). If
12819 the GOT size for the linked executable exceeds a machine-specific
12820 maximum size, you get an error message from the linker indicating that
12821 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
12822 instead. (These maximums are 8k on the SPARC and 32k
12823 on the m68k and RS/6000. The 386 has no such limit.)
12825 Position-independent code requires special support, and therefore works
12826 only on certain machines. For the 386, GCC supports PIC for System V
12827 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
12828 position-independent.
12832 If supported for the target machine, emit position-independent code,
12833 suitable for dynamic linking and avoiding any limit on the size of the
12834 global offset table. This option makes a difference on the m68k,
12835 PowerPC and SPARC@.
12837 Position-independent code requires special support, and therefore works
12838 only on certain machines.
12844 These options are similar to @option{-fpic} and @option{-fPIC}, but
12845 generated position independent code can be only linked into executables.
12846 Usually these options are used when @option{-pie} GCC option will be
12847 used during linking.
12849 @item -fno-jump-tables
12850 @opindex fno-jump-tables
12851 Do not use jump tables for switch statements even where it would be
12852 more efficient than other code generation strategies. This option is
12853 of use in conjunction with @option{-fpic} or @option{-fPIC} for
12854 building code which forms part of a dynamic linker and cannot
12855 reference the address of a jump table. On some targets, jump tables
12856 do not require a GOT and this option is not needed.
12858 @item -ffixed-@var{reg}
12860 Treat the register named @var{reg} as a fixed register; generated code
12861 should never refer to it (except perhaps as a stack pointer, frame
12862 pointer or in some other fixed role).
12864 @var{reg} must be the name of a register. The register names accepted
12865 are machine-specific and are defined in the @code{REGISTER_NAMES}
12866 macro in the machine description macro file.
12868 This flag does not have a negative form, because it specifies a
12871 @item -fcall-used-@var{reg}
12872 @opindex fcall-used
12873 Treat the register named @var{reg} as an allocable register that is
12874 clobbered by function calls. It may be allocated for temporaries or
12875 variables that do not live across a call. Functions compiled this way
12876 will not save and restore the register @var{reg}.
12878 It is an error to used this flag with the frame pointer or stack pointer.
12879 Use of this flag for other registers that have fixed pervasive roles in
12880 the machine's execution model will produce disastrous results.
12882 This flag does not have a negative form, because it specifies a
12885 @item -fcall-saved-@var{reg}
12886 @opindex fcall-saved
12887 Treat the register named @var{reg} as an allocable register saved by
12888 functions. It may be allocated even for temporaries or variables that
12889 live across a call. Functions compiled this way will save and restore
12890 the register @var{reg} if they use it.
12892 It is an error to used this flag with the frame pointer or stack pointer.
12893 Use of this flag for other registers that have fixed pervasive roles in
12894 the machine's execution model will produce disastrous results.
12896 A different sort of disaster will result from the use of this flag for
12897 a register in which function values may be returned.
12899 This flag does not have a negative form, because it specifies a
12902 @item -fpack-struct[=@var{n}]
12903 @opindex fpack-struct
12904 Without a value specified, pack all structure members together without
12905 holes. When a value is specified (which must be a small power of two), pack
12906 structure members according to this value, representing the maximum
12907 alignment (that is, objects with default alignment requirements larger than
12908 this will be output potentially unaligned at the next fitting location.
12910 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
12911 code that is not binary compatible with code generated without that switch.
12912 Additionally, it makes the code suboptimal.
12913 Use it to conform to a non-default application binary interface.
12915 @item -finstrument-functions
12916 @opindex finstrument-functions
12917 Generate instrumentation calls for entry and exit to functions. Just
12918 after function entry and just before function exit, the following
12919 profiling functions will be called with the address of the current
12920 function and its call site. (On some platforms,
12921 @code{__builtin_return_address} does not work beyond the current
12922 function, so the call site information may not be available to the
12923 profiling functions otherwise.)
12926 void __cyg_profile_func_enter (void *this_fn,
12928 void __cyg_profile_func_exit (void *this_fn,
12932 The first argument is the address of the start of the current function,
12933 which may be looked up exactly in the symbol table.
12935 This instrumentation is also done for functions expanded inline in other
12936 functions. The profiling calls will indicate where, conceptually, the
12937 inline function is entered and exited. This means that addressable
12938 versions of such functions must be available. If all your uses of a
12939 function are expanded inline, this may mean an additional expansion of
12940 code size. If you use @samp{extern inline} in your C code, an
12941 addressable version of such functions must be provided. (This is
12942 normally the case anyways, but if you get lucky and the optimizer always
12943 expands the functions inline, you might have gotten away without
12944 providing static copies.)
12946 A function may be given the attribute @code{no_instrument_function}, in
12947 which case this instrumentation will not be done. This can be used, for
12948 example, for the profiling functions listed above, high-priority
12949 interrupt routines, and any functions from which the profiling functions
12950 cannot safely be called (perhaps signal handlers, if the profiling
12951 routines generate output or allocate memory).
12953 @item -fstack-check
12954 @opindex fstack-check
12955 Generate code to verify that you do not go beyond the boundary of the
12956 stack. You should specify this flag if you are running in an
12957 environment with multiple threads, but only rarely need to specify it in
12958 a single-threaded environment since stack overflow is automatically
12959 detected on nearly all systems if there is only one stack.
12961 Note that this switch does not actually cause checking to be done; the
12962 operating system must do that. The switch causes generation of code
12963 to ensure that the operating system sees the stack being extended.
12965 @item -fstack-limit-register=@var{reg}
12966 @itemx -fstack-limit-symbol=@var{sym}
12967 @itemx -fno-stack-limit
12968 @opindex fstack-limit-register
12969 @opindex fstack-limit-symbol
12970 @opindex fno-stack-limit
12971 Generate code to ensure that the stack does not grow beyond a certain value,
12972 either the value of a register or the address of a symbol. If the stack
12973 would grow beyond the value, a signal is raised. For most targets,
12974 the signal is raised before the stack overruns the boundary, so
12975 it is possible to catch the signal without taking special precautions.
12977 For instance, if the stack starts at absolute address @samp{0x80000000}
12978 and grows downwards, you can use the flags
12979 @option{-fstack-limit-symbol=__stack_limit} and
12980 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
12981 of 128KB@. Note that this may only work with the GNU linker.
12983 @cindex aliasing of parameters
12984 @cindex parameters, aliased
12985 @item -fargument-alias
12986 @itemx -fargument-noalias
12987 @itemx -fargument-noalias-global
12988 @opindex fargument-alias
12989 @opindex fargument-noalias
12990 @opindex fargument-noalias-global
12991 Specify the possible relationships among parameters and between
12992 parameters and global data.
12994 @option{-fargument-alias} specifies that arguments (parameters) may
12995 alias each other and may alias global storage.@*
12996 @option{-fargument-noalias} specifies that arguments do not alias
12997 each other, but may alias global storage.@*
12998 @option{-fargument-noalias-global} specifies that arguments do not
12999 alias each other and do not alias global storage.
13001 Each language will automatically use whatever option is required by
13002 the language standard. You should not need to use these options yourself.
13004 @item -fleading-underscore
13005 @opindex fleading-underscore
13006 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13007 change the way C symbols are represented in the object file. One use
13008 is to help link with legacy assembly code.
13010 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13011 generate code that is not binary compatible with code generated without that
13012 switch. Use it to conform to a non-default application binary interface.
13013 Not all targets provide complete support for this switch.
13015 @item -ftls-model=@var{model}
13016 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13017 The @var{model} argument should be one of @code{global-dynamic},
13018 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13020 The default without @option{-fpic} is @code{initial-exec}; with
13021 @option{-fpic} the default is @code{global-dynamic}.
13023 @item -fvisibility=@var{default|internal|hidden|protected}
13024 @opindex fvisibility
13025 Set the default ELF image symbol visibility to the specified option---all
13026 symbols will be marked with this unless overridden within the code.
13027 Using this feature can very substantially improve linking and
13028 load times of shared object libraries, produce more optimized
13029 code, provide near-perfect API export and prevent symbol clashes.
13030 It is @strong{strongly} recommended that you use this in any shared objects
13033 Despite the nomenclature, @code{default} always means public ie;
13034 available to be linked against from outside the shared object.
13035 @code{protected} and @code{internal} are pretty useless in real-world
13036 usage so the only other commonly used option will be @code{hidden}.
13037 The default if @option{-fvisibility} isn't specified is
13038 @code{default}, i.e., make every
13039 symbol public---this causes the same behavior as previous versions of
13042 A good explanation of the benefits offered by ensuring ELF
13043 symbols have the correct visibility is given by ``How To Write
13044 Shared Libraries'' by Ulrich Drepper (which can be found at
13045 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13046 solution made possible by this option to marking things hidden when
13047 the default is public is to make the default hidden and mark things
13048 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13049 and @code{__attribute__ ((visibility("default")))} instead of
13050 @code{__declspec(dllexport)} you get almost identical semantics with
13051 identical syntax. This is a great boon to those working with
13052 cross-platform projects.
13054 For those adding visibility support to existing code, you may find
13055 @samp{#pragma GCC visibility} of use. This works by you enclosing
13056 the declarations you wish to set visibility for with (for example)
13057 @samp{#pragma GCC visibility push(hidden)} and
13058 @samp{#pragma GCC visibility pop}.
13059 Bear in mind that symbol visibility should be viewed @strong{as
13060 part of the API interface contract} and thus all new code should
13061 always specify visibility when it is not the default ie; declarations
13062 only for use within the local DSO should @strong{always} be marked explicitly
13063 as hidden as so to avoid PLT indirection overheads---making this
13064 abundantly clear also aids readability and self-documentation of the code.
13065 Note that due to ISO C++ specification requirements, operator new and
13066 operator delete must always be of default visibility.
13068 An overview of these techniques, their benefits and how to use them
13069 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13075 @node Environment Variables
13076 @section Environment Variables Affecting GCC
13077 @cindex environment variables
13079 @c man begin ENVIRONMENT
13080 This section describes several environment variables that affect how GCC
13081 operates. Some of them work by specifying directories or prefixes to use
13082 when searching for various kinds of files. Some are used to specify other
13083 aspects of the compilation environment.
13085 Note that you can also specify places to search using options such as
13086 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13087 take precedence over places specified using environment variables, which
13088 in turn take precedence over those specified by the configuration of GCC@.
13089 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13090 GNU Compiler Collection (GCC) Internals}.
13095 @c @itemx LC_COLLATE
13097 @c @itemx LC_MONETARY
13098 @c @itemx LC_NUMERIC
13103 @c @findex LC_COLLATE
13104 @findex LC_MESSAGES
13105 @c @findex LC_MONETARY
13106 @c @findex LC_NUMERIC
13110 These environment variables control the way that GCC uses
13111 localization information that allow GCC to work with different
13112 national conventions. GCC inspects the locale categories
13113 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13114 so. These locale categories can be set to any value supported by your
13115 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13116 Kingdom encoded in UTF-8.
13118 The @env{LC_CTYPE} environment variable specifies character
13119 classification. GCC uses it to determine the character boundaries in
13120 a string; this is needed for some multibyte encodings that contain quote
13121 and escape characters that would otherwise be interpreted as a string
13124 The @env{LC_MESSAGES} environment variable specifies the language to
13125 use in diagnostic messages.
13127 If the @env{LC_ALL} environment variable is set, it overrides the value
13128 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13129 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13130 environment variable. If none of these variables are set, GCC
13131 defaults to traditional C English behavior.
13135 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13136 files. GCC uses temporary files to hold the output of one stage of
13137 compilation which is to be used as input to the next stage: for example,
13138 the output of the preprocessor, which is the input to the compiler
13141 @item GCC_EXEC_PREFIX
13142 @findex GCC_EXEC_PREFIX
13143 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13144 names of the subprograms executed by the compiler. No slash is added
13145 when this prefix is combined with the name of a subprogram, but you can
13146 specify a prefix that ends with a slash if you wish.
13148 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13149 an appropriate prefix to use based on the pathname it was invoked with.
13151 If GCC cannot find the subprogram using the specified prefix, it
13152 tries looking in the usual places for the subprogram.
13154 The default value of @env{GCC_EXEC_PREFIX} is
13155 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13156 of @code{prefix} when you ran the @file{configure} script.
13158 Other prefixes specified with @option{-B} take precedence over this prefix.
13160 This prefix is also used for finding files such as @file{crt0.o} that are
13163 In addition, the prefix is used in an unusual way in finding the
13164 directories to search for header files. For each of the standard
13165 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13166 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13167 replacing that beginning with the specified prefix to produce an
13168 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13169 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13170 These alternate directories are searched first; the standard directories
13173 @item COMPILER_PATH
13174 @findex COMPILER_PATH
13175 The value of @env{COMPILER_PATH} is a colon-separated list of
13176 directories, much like @env{PATH}. GCC tries the directories thus
13177 specified when searching for subprograms, if it can't find the
13178 subprograms using @env{GCC_EXEC_PREFIX}.
13181 @findex LIBRARY_PATH
13182 The value of @env{LIBRARY_PATH} is a colon-separated list of
13183 directories, much like @env{PATH}. When configured as a native compiler,
13184 GCC tries the directories thus specified when searching for special
13185 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13186 using GCC also uses these directories when searching for ordinary
13187 libraries for the @option{-l} option (but directories specified with
13188 @option{-L} come first).
13192 @cindex locale definition
13193 This variable is used to pass locale information to the compiler. One way in
13194 which this information is used is to determine the character set to be used
13195 when character literals, string literals and comments are parsed in C and C++.
13196 When the compiler is configured to allow multibyte characters,
13197 the following values for @env{LANG} are recognized:
13201 Recognize JIS characters.
13203 Recognize SJIS characters.
13205 Recognize EUCJP characters.
13208 If @env{LANG} is not defined, or if it has some other value, then the
13209 compiler will use mblen and mbtowc as defined by the default locale to
13210 recognize and translate multibyte characters.
13214 Some additional environments variables affect the behavior of the
13217 @include cppenv.texi
13221 @node Precompiled Headers
13222 @section Using Precompiled Headers
13223 @cindex precompiled headers
13224 @cindex speed of compilation
13226 Often large projects have many header files that are included in every
13227 source file. The time the compiler takes to process these header files
13228 over and over again can account for nearly all of the time required to
13229 build the project. To make builds faster, GCC allows users to
13230 `precompile' a header file; then, if builds can use the precompiled
13231 header file they will be much faster.
13233 To create a precompiled header file, simply compile it as you would any
13234 other file, if necessary using the @option{-x} option to make the driver
13235 treat it as a C or C++ header file. You will probably want to use a
13236 tool like @command{make} to keep the precompiled header up-to-date when
13237 the headers it contains change.
13239 A precompiled header file will be searched for when @code{#include} is
13240 seen in the compilation. As it searches for the included file
13241 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13242 compiler looks for a precompiled header in each directory just before it
13243 looks for the include file in that directory. The name searched for is
13244 the name specified in the @code{#include} with @samp{.gch} appended. If
13245 the precompiled header file can't be used, it is ignored.
13247 For instance, if you have @code{#include "all.h"}, and you have
13248 @file{all.h.gch} in the same directory as @file{all.h}, then the
13249 precompiled header file will be used if possible, and the original
13250 header will be used otherwise.
13252 Alternatively, you might decide to put the precompiled header file in a
13253 directory and use @option{-I} to ensure that directory is searched
13254 before (or instead of) the directory containing the original header.
13255 Then, if you want to check that the precompiled header file is always
13256 used, you can put a file of the same name as the original header in this
13257 directory containing an @code{#error} command.
13259 This also works with @option{-include}. So yet another way to use
13260 precompiled headers, good for projects not designed with precompiled
13261 header files in mind, is to simply take most of the header files used by
13262 a project, include them from another header file, precompile that header
13263 file, and @option{-include} the precompiled header. If the header files
13264 have guards against multiple inclusion, they will be skipped because
13265 they've already been included (in the precompiled header).
13267 If you need to precompile the same header file for different
13268 languages, targets, or compiler options, you can instead make a
13269 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13270 header in the directory, perhaps using @option{-o}. It doesn't matter
13271 what you call the files in the directory, every precompiled header in
13272 the directory will be considered. The first precompiled header
13273 encountered in the directory that is valid for this compilation will
13274 be used; they're searched in no particular order.
13276 There are many other possibilities, limited only by your imagination,
13277 good sense, and the constraints of your build system.
13279 A precompiled header file can be used only when these conditions apply:
13283 Only one precompiled header can be used in a particular compilation.
13286 A precompiled header can't be used once the first C token is seen. You
13287 can have preprocessor directives before a precompiled header; you can
13288 even include a precompiled header from inside another header, so long as
13289 there are no C tokens before the @code{#include}.
13292 The precompiled header file must be produced for the same language as
13293 the current compilation. You can't use a C precompiled header for a C++
13297 The precompiled header file must have been produced by the same compiler
13298 binary as the current compilation is using.
13301 Any macros defined before the precompiled header is included must
13302 either be defined in the same way as when the precompiled header was
13303 generated, or must not affect the precompiled header, which usually
13304 means that they don't appear in the precompiled header at all.
13306 The @option{-D} option is one way to define a macro before a
13307 precompiled header is included; using a @code{#define} can also do it.
13308 There are also some options that define macros implicitly, like
13309 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13312 @item If debugging information is output when using the precompiled
13313 header, using @option{-g} or similar, the same kind of debugging information
13314 must have been output when building the precompiled header. However,
13315 a precompiled header built using @option{-g} can be used in a compilation
13316 when no debugging information is being output.
13318 @item The same @option{-m} options must generally be used when building
13319 and using the precompiled header. @xref{Submodel Options},
13320 for any cases where this rule is relaxed.
13322 @item Each of the following options must be the same when building and using
13323 the precompiled header:
13325 @gccoptlist{-fexceptions -funit-at-a-time}
13328 Some other command-line options starting with @option{-f},
13329 @option{-p}, or @option{-O} must be defined in the same way as when
13330 the precompiled header was generated. At present, it's not clear
13331 which options are safe to change and which are not; the safest choice
13332 is to use exactly the same options when generating and using the
13333 precompiled header. The following are known to be safe:
13335 @gccoptlist{-fmessage-length= -fpreprocessed
13336 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13337 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13342 For all of these except the last, the compiler will automatically
13343 ignore the precompiled header if the conditions aren't met. If you
13344 find an option combination that doesn't work and doesn't cause the
13345 precompiled header to be ignored, please consider filing a bug report,
13348 If you do use differing options when generating and using the
13349 precompiled header, the actual behavior will be a mixture of the
13350 behavior for the options. For instance, if you use @option{-g} to
13351 generate the precompiled header but not when using it, you may or may
13352 not get debugging information for routines in the precompiled header.
13354 @node Running Protoize
13355 @section Running Protoize
13357 The program @code{protoize} is an optional part of GCC@. You can use
13358 it to add prototypes to a program, thus converting the program to ISO
13359 C in one respect. The companion program @code{unprotoize} does the
13360 reverse: it removes argument types from any prototypes that are found.
13362 When you run these programs, you must specify a set of source files as
13363 command line arguments. The conversion programs start out by compiling
13364 these files to see what functions they define. The information gathered
13365 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
13367 After scanning comes actual conversion. The specified files are all
13368 eligible to be converted; any files they include (whether sources or
13369 just headers) are eligible as well.
13371 But not all the eligible files are converted. By default,
13372 @code{protoize} and @code{unprotoize} convert only source and header
13373 files in the current directory. You can specify additional directories
13374 whose files should be converted with the @option{-d @var{directory}}
13375 option. You can also specify particular files to exclude with the
13376 @option{-x @var{file}} option. A file is converted if it is eligible, its
13377 directory name matches one of the specified directory names, and its
13378 name within the directory has not been excluded.
13380 Basic conversion with @code{protoize} consists of rewriting most
13381 function definitions and function declarations to specify the types of
13382 the arguments. The only ones not rewritten are those for varargs
13385 @code{protoize} optionally inserts prototype declarations at the
13386 beginning of the source file, to make them available for any calls that
13387 precede the function's definition. Or it can insert prototype
13388 declarations with block scope in the blocks where undeclared functions
13391 Basic conversion with @code{unprotoize} consists of rewriting most
13392 function declarations to remove any argument types, and rewriting
13393 function definitions to the old-style pre-ISO form.
13395 Both conversion programs print a warning for any function declaration or
13396 definition that they can't convert. You can suppress these warnings
13399 The output from @code{protoize} or @code{unprotoize} replaces the
13400 original source file. The original file is renamed to a name ending
13401 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
13402 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
13403 for DOS) file already exists, then the source file is simply discarded.
13405 @code{protoize} and @code{unprotoize} both depend on GCC itself to
13406 scan the program and collect information about the functions it uses.
13407 So neither of these programs will work until GCC is installed.
13409 Here is a table of the options you can use with @code{protoize} and
13410 @code{unprotoize}. Each option works with both programs unless
13414 @item -B @var{directory}
13415 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
13416 usual directory (normally @file{/usr/local/lib}). This file contains
13417 prototype information about standard system functions. This option
13418 applies only to @code{protoize}.
13420 @item -c @var{compilation-options}
13421 Use @var{compilation-options} as the options when running @command{gcc} to
13422 produce the @samp{.X} files. The special option @option{-aux-info} is
13423 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
13425 Note that the compilation options must be given as a single argument to
13426 @code{protoize} or @code{unprotoize}. If you want to specify several
13427 @command{gcc} options, you must quote the entire set of compilation options
13428 to make them a single word in the shell.
13430 There are certain @command{gcc} arguments that you cannot use, because they
13431 would produce the wrong kind of output. These include @option{-g},
13432 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
13433 the @var{compilation-options}, they are ignored.
13436 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
13437 systems) instead of @samp{.c}. This is convenient if you are converting
13438 a C program to C++. This option applies only to @code{protoize}.
13441 Add explicit global declarations. This means inserting explicit
13442 declarations at the beginning of each source file for each function
13443 that is called in the file and was not declared. These declarations
13444 precede the first function definition that contains a call to an
13445 undeclared function. This option applies only to @code{protoize}.
13447 @item -i @var{string}
13448 Indent old-style parameter declarations with the string @var{string}.
13449 This option applies only to @code{protoize}.
13451 @code{unprotoize} converts prototyped function definitions to old-style
13452 function definitions, where the arguments are declared between the
13453 argument list and the initial @samp{@{}. By default, @code{unprotoize}
13454 uses five spaces as the indentation. If you want to indent with just
13455 one space instead, use @option{-i " "}.
13458 Keep the @samp{.X} files. Normally, they are deleted after conversion
13462 Add explicit local declarations. @code{protoize} with @option{-l} inserts
13463 a prototype declaration for each function in each block which calls the
13464 function without any declaration. This option applies only to
13468 Make no real changes. This mode just prints information about the conversions
13469 that would have been done without @option{-n}.
13472 Make no @samp{.save} files. The original files are simply deleted.
13473 Use this option with caution.
13475 @item -p @var{program}
13476 Use the program @var{program} as the compiler. Normally, the name
13477 @file{gcc} is used.
13480 Work quietly. Most warnings are suppressed.
13483 Print the version number, just like @option{-v} for @command{gcc}.
13486 If you need special compiler options to compile one of your program's
13487 source files, then you should generate that file's @samp{.X} file
13488 specially, by running @command{gcc} on that source file with the
13489 appropriate options and the option @option{-aux-info}. Then run
13490 @code{protoize} on the entire set of files. @code{protoize} will use
13491 the existing @samp{.X} file because it is newer than the source file.
13495 gcc -Dfoo=bar file1.c -aux-info file1.X
13500 You need to include the special files along with the rest in the
13501 @code{protoize} command, even though their @samp{.X} files already
13502 exist, because otherwise they won't get converted.
13504 @xref{Protoize Caveats}, for more information on how to use
13505 @code{protoize} successfully.