1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.2 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
126 * Language Independent Options:: Controlling how diagnostics should be
128 * Warning Options:: How picky should the compiler be?
129 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
130 * Optimize Options:: How much optimization?
131 * Preprocessor Options:: Controlling header files and macro definitions.
132 Also, getting dependency information for Make.
133 * Assembler Options:: Passing options to the assembler.
134 * Link Options:: Specifying libraries and so on.
135 * Directory Options:: Where to find header files and libraries.
136 Where to find the compiler executable files.
137 * Spec Files:: How to pass switches to sub-processes.
138 * Target Options:: Running a cross-compiler, or an old version of GCC.
139 * Submodel Options:: Specifying minor hardware or convention variations,
140 such as 68010 vs 68020.
141 * Code Gen Options:: Specifying conventions for function calls, data layout
143 * Environment Variables:: Env vars that affect GCC.
144 * Precompiled Headers:: Compiling a header once, and using it many times.
145 * Running Protoize:: Automatically adding or removing function prototypes.
151 @section Option Summary
153 Here is a summary of all the options, grouped by type. Explanations are
154 in the following sections.
157 @item Overall Options
158 @xref{Overall Options,,Options Controlling the Kind of Output}.
159 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
160 -x @var{language} -v -### --help --target-help --version}
162 @item C Language Options
163 @xref{C Dialect Options,,Options Controlling C Dialect}.
164 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
165 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
166 -fhosted -ffreestanding -fms-extensions @gol
167 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
168 -fallow-single-precision -fcond-mismatch @gol
169 -fsigned-bitfields -fsigned-char @gol
170 -funsigned-bitfields -funsigned-char}
172 @item C++ Language Options
173 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
174 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
175 -fconserve-space -fno-const-strings @gol
176 -fno-elide-constructors @gol
177 -fno-enforce-eh-specs @gol
178 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
179 -fno-implicit-templates @gol
180 -fno-implicit-inline-templates @gol
181 -fno-implement-inlines -fms-extensions @gol
182 -fno-nonansi-builtins -fno-operator-names @gol
183 -fno-optional-diags -fpermissive @gol
184 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
185 -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
186 -fno-default-inline -fvisibility-inlines-hidden @gol
187 -Wabi -Wctor-dtor-privacy @gol
188 -Wnon-virtual-dtor -Wreorder @gol
189 -Weffc++ -Wno-deprecated @gol
190 -Wno-non-template-friend -Wold-style-cast @gol
191 -Woverloaded-virtual -Wno-pmf-conversions @gol
192 -Wsign-promo -Wsynth}
194 @item Objective-C and Objective-C++ Language Options
195 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
196 Objective-C and Objective-C++ Dialects}.
198 -fconstant-string-class=@var{class-name} @gol
199 -fgnu-runtime -fnext-runtime @gol
200 -fno-nil-receivers @gol
201 -fobjc-exceptions @gol
202 -freplace-objc-classes @gol
205 -Wno-protocol -Wselector -Wundeclared-selector}
207 @item Language Independent Options
208 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
209 @gccoptlist{-fmessage-length=@var{n} @gol
210 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
212 @item Warning Options
213 @xref{Warning Options,,Options to Request or Suppress Warnings}.
214 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
215 -w -Wextra -Wall -Waggregate-return @gol
216 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
217 -Wconversion -Wno-deprecated-declarations @gol
218 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
219 -Werror -Werror-implicit-function-declaration @gol
220 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
221 -Wno-format-extra-args -Wformat-nonliteral @gol
222 -Wformat-security -Wformat-y2k @gol
223 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
224 -Wimport -Wno-import -Winit-self -Winline @gol
225 -Wno-invalid-offsetof -Winvalid-pch @gol
226 -Wlarger-than-@var{len} -Wlong-long @gol
227 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
228 -Wmissing-format-attribute -Wmissing-include-dirs @gol
229 -Wmissing-noreturn @gol
230 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
231 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
232 -Wreturn-type -Wsequence-point -Wshadow @gol
233 -Wsign-compare -Wstrict-aliasing -Wstrict-aliasing=2 @gol
234 -Wswitch -Wswitch-default -Wswitch-enum @gol
235 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
236 -Wunknown-pragmas -Wunreachable-code @gol
237 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
238 -Wunused-value -Wunused-variable -Wwrite-strings @gol
241 @item C-only Warning Options
242 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
243 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
244 -Wstrict-prototypes -Wtraditional @gol
245 -Wdeclaration-after-statement}
247 @item Debugging Options
248 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
249 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
250 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
251 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
253 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
254 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
255 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
256 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
258 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
259 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
260 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
261 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
262 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
263 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
264 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
265 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
266 -fdump-tree-nrv -fdump-tree-vect @gol
267 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
268 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
269 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
270 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs -ftree-based-profiling @gol
271 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
272 -ftest-coverage -ftime-report -fvar-tracking @gol
273 -g -g@var{level} -gcoff -gdwarf-2 @gol
274 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
275 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
276 -print-multi-directory -print-multi-lib @gol
277 -print-prog-name=@var{program} -print-search-dirs -Q @gol
280 @item Optimization Options
281 @xref{Optimize Options,,Options that Control Optimization}.
282 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
283 -falign-labels=@var{n} -falign-loops=@var{n} @gol
284 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
285 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
286 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
287 -fcaller-saves -fcprop-registers @gol
288 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
289 -fdelayed-branch -fdelete-null-pointer-checks @gol
290 -fexpensive-optimizations -ffast-math -ffloat-store @gol
291 -fforce-addr -fforce-mem -ffunction-sections @gol
292 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
293 -floop-optimize -fcrossjumping -fif-conversion -fif-conversion2 @gol
294 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
295 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
296 -fmodulo-sched -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
297 -fno-default-inline -fno-defer-pop -floop-optimize2 -fmove-loop-invariants @gol
298 -fno-function-cse -fno-guess-branch-probability @gol
299 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
300 -funsafe-math-optimizations -ffinite-math-only @gol
301 -fno-trapping-math -fno-zero-initialized-in-bss @gol
302 -fomit-frame-pointer -foptimize-register-move @gol
303 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
304 -fprofile-generate -fprofile-use @gol
305 -freduce-all-givs -fregmove -frename-registers @gol
306 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
307 -frerun-cse-after-loop -frerun-loop-opt @gol
308 -frounding-math -fschedule-insns -fschedule-insns2 @gol
309 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
310 -fsched-spec-load-dangerous @gol
311 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
312 -fsched2-use-superblocks @gol
313 -fsched2-use-traces -freschedule-modulo-scheduled-loops @gol
314 -fsignaling-nans -fsingle-precision-constant @gol
315 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
316 -funroll-all-loops -funroll-loops -fpeel-loops @gol
317 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
318 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
319 -ftree-lim -fivcanon @gol
320 -ftree-dominator-opts -ftree-dse -ftree-copyrename @gol
321 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
322 --param @var{name}=@var{value}
323 -O -O0 -O1 -O2 -O3 -Os}
325 @item Preprocessor Options
326 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
327 @gccoptlist{-A@var{question}=@var{answer} @gol
328 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
329 -C -dD -dI -dM -dN @gol
330 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
331 -idirafter @var{dir} @gol
332 -include @var{file} -imacros @var{file} @gol
333 -iprefix @var{file} -iwithprefix @var{dir} @gol
334 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
335 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
336 -P -fworking-directory -remap @gol
337 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
338 -Xpreprocessor @var{option}}
340 @item Assembler Option
341 @xref{Assembler Options,,Passing Options to the Assembler}.
342 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
345 @xref{Link Options,,Options for Linking}.
346 @gccoptlist{@var{object-file-name} -l@var{library} @gol
347 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
348 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
349 -Wl,@var{option} -Xlinker @var{option} @gol
352 @item Directory Options
353 @xref{Directory Options,,Options for Directory Search}.
354 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-}
357 @c I wrote this xref this way to avoid overfull hbox. -- rms
358 @xref{Target Options}.
359 @gccoptlist{-V @var{version} -b @var{machine}}
361 @item Machine Dependent Options
362 @xref{Submodel Options,,Hardware Models and Configurations}.
363 @c This list is ordered alphanumerically by subsection name.
364 @c Try and put the significant identifier (CPU or system) first,
365 @c so users have a clue at guessing where the ones they want will be.
368 @gccoptlist{-EB -EL @gol
369 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
370 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
373 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
374 -mabi=@var{name} @gol
375 -mapcs-stack-check -mno-apcs-stack-check @gol
376 -mapcs-float -mno-apcs-float @gol
377 -mapcs-reentrant -mno-apcs-reentrant @gol
378 -msched-prolog -mno-sched-prolog @gol
379 -mlittle-endian -mbig-endian -mwords-little-endian @gol
380 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
381 -mthumb-interwork -mno-thumb-interwork @gol
382 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
383 -mstructure-size-boundary=@var{n} @gol
384 -mabort-on-noreturn @gol
385 -mlong-calls -mno-long-calls @gol
386 -msingle-pic-base -mno-single-pic-base @gol
387 -mpic-register=@var{reg} @gol
388 -mnop-fun-dllimport @gol
389 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
390 -mpoke-function-name @gol
392 -mtpcs-frame -mtpcs-leaf-frame @gol
393 -mcaller-super-interworking -mcallee-super-interworking}
396 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
397 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
400 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
401 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
402 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
403 -mstack-align -mdata-align -mconst-align @gol
404 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
405 -melf -maout -melinux -mlinux -sim -sim2 @gol
406 -mmul-bug-workaround -mno-mul-bug-workaround}
408 @emph{Darwin Options}
409 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
410 -arch_only -bind_at_load -bundle -bundle_loader @gol
411 -client_name -compatibility_version -current_version @gol
413 -dependency-file -dylib_file -dylinker_install_name @gol
414 -dynamic -dynamiclib -exported_symbols_list @gol
415 -filelist -flat_namespace -force_cpusubtype_ALL @gol
416 -force_flat_namespace -headerpad_max_install_names @gol
417 -image_base -init -install_name -keep_private_externs @gol
418 -multi_module -multiply_defined -multiply_defined_unused @gol
419 -noall_load -no_dead_strip_inits_and_terms @gol
420 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
421 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
422 -private_bundle -read_only_relocs -sectalign @gol
423 -sectobjectsymbols -whyload -seg1addr @gol
424 -sectcreate -sectobjectsymbols -sectorder @gol
425 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
426 -segprot -segs_read_only_addr -segs_read_write_addr @gol
427 -single_module -static -sub_library -sub_umbrella @gol
428 -twolevel_namespace -umbrella -undefined @gol
429 -unexported_symbols_list -weak_reference_mismatches @gol
430 -whatsloaded -F -gused -gfull -mone-byte-bool}
432 @emph{DEC Alpha Options}
433 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
434 -mieee -mieee-with-inexact -mieee-conformant @gol
435 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
436 -mtrap-precision=@var{mode} -mbuild-constants @gol
437 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
438 -mbwx -mmax -mfix -mcix @gol
439 -mfloat-vax -mfloat-ieee @gol
440 -mexplicit-relocs -msmall-data -mlarge-data @gol
441 -msmall-text -mlarge-text @gol
442 -mmemory-latency=@var{time}}
444 @emph{DEC Alpha/VMS Options}
445 @gccoptlist{-mvms-return-codes}
448 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
449 -mhard-float -msoft-float @gol
450 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
451 -mdouble -mno-double @gol
452 -mmedia -mno-media -mmuladd -mno-muladd @gol
453 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic -mlinked-fp @gol
454 -mlibrary-pic -macc-4 -macc-8 @gol
455 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
456 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
457 -mvliw-branch -mno-vliw-branch @gol
458 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
459 -mno-nested-cond-exec -mtomcat-stats @gol
462 @emph{H8/300 Options}
463 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
466 @gccoptlist{-march=@var{architecture-type} @gol
467 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
468 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
469 -mfixed-range=@var{register-range} @gol
470 -mjump-in-delay -mlinker-opt -mlong-calls @gol
471 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
472 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
473 -mno-jump-in-delay -mno-long-load-store @gol
474 -mno-portable-runtime -mno-soft-float @gol
475 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
476 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
477 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
478 -nolibdld -static -threads}
480 @emph{i386 and x86-64 Options}
481 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
482 -mfpmath=@var{unit} @gol
483 -masm=@var{dialect} -mno-fancy-math-387 @gol
484 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
485 -mno-wide-multiply -mrtd -malign-double @gol
486 -mpreferred-stack-boundary=@var{num} @gol
487 -mmmx -msse -msse2 -msse3 -m3dnow @gol
488 -mthreads -mno-align-stringops -minline-all-stringops @gol
489 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
490 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
491 -mno-red-zone -mno-tls-direct-seg-refs @gol
492 -mcmodel=@var{code-model} @gol
496 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
497 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
498 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
499 -minline-float-divide-max-throughput @gol
500 -minline-int-divide-min-latency @gol
501 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
502 -mfixed-range=@var{register-range}}
504 @emph{M32R/D Options}
505 @gccoptlist{-m32r2 -m32rx -m32r @gol
507 -malign-loops -mno-align-loops @gol
508 -missue-rate=@var{number} @gol
509 -mbranch-cost=@var{number} @gol
510 -mmodel=@var{code-size-model-type} @gol
511 -msdata=@var{sdata-type} @gol
512 -mno-flush-func -mflush-func=@var{name} @gol
513 -mno-flush-trap -mflush-trap=@var{number} @gol
516 @emph{M680x0 Options}
517 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
518 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
519 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
520 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
521 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
523 @emph{M68hc1x Options}
524 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
525 -mauto-incdec -minmax -mlong-calls -mshort @gol
526 -msoft-reg-count=@var{count}}
529 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
530 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
531 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
532 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
533 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
536 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
537 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
538 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
539 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
540 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
541 -mint64 -mlong64 -mlong32 @gol
542 -G@var{num} -membedded-data -mno-embedded-data @gol
543 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
544 -msplit-addresses -mno-split-addresses @gol
545 -mexplicit-relocs -mno-explicit-relocs @gol
546 -mcheck-zero-division -mno-check-zero-division @gol
547 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
548 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
549 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
550 -mfix-vr4120 -mno-fix-vr4120 -mfix-sb1 -mno-fix-sb1 @gol
551 -mflush-func=@var{func} -mno-flush-func @gol
552 -mbranch-likely -mno-branch-likely @gol
553 -mfp-exceptions -mno-fp-exceptions @gol
554 -mvr4130-align -mno-vr4130-align}
557 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
558 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
559 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
560 -mno-base-addresses -msingle-exit -mno-single-exit}
562 @emph{MN10300 Options}
563 @gccoptlist{-mmult-bug -mno-mult-bug @gol
564 -mam33 -mno-am33 @gol
565 -mam33-2 -mno-am33-2 @gol
569 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
570 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
571 -mregparam -mnoregparam -msb -mnosb @gol
572 -mbitfield -mnobitfield -mhimem -mnohimem}
574 @emph{PDP-11 Options}
575 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
576 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
577 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
578 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
579 -mbranch-expensive -mbranch-cheap @gol
580 -msplit -mno-split -munix-asm -mdec-asm}
582 @emph{PowerPC Options}
583 See RS/6000 and PowerPC Options.
585 @emph{RS/6000 and PowerPC Options}
586 @gccoptlist{-mcpu=@var{cpu-type} @gol
587 -mtune=@var{cpu-type} @gol
588 -mpower -mno-power -mpower2 -mno-power2 @gol
589 -mpowerpc -mpowerpc64 -mno-powerpc @gol
590 -maltivec -mno-altivec @gol
591 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
592 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
593 -mnew-mnemonics -mold-mnemonics @gol
594 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
595 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
596 -malign-power -malign-natural @gol
597 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
598 -mstring -mno-string -mupdate -mno-update @gol
599 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
600 -mstrict-align -mno-strict-align -mrelocatable @gol
601 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
602 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
603 -mdynamic-no-pic @gol
604 -mprioritize-restricted-insns=@var{priority} @gol
605 -msched-costly-dep=@var{dependence_type} @gol
606 -minsert-sched-nops=@var{scheme} @gol
607 -mcall-sysv -mcall-netbsd @gol
608 -maix-struct-return -msvr4-struct-return @gol
609 -mabi=altivec -mabi=no-altivec @gol
610 -mabi=spe -mabi=no-spe @gol
611 -misel=yes -misel=no @gol
612 -mspe=yes -mspe=no @gol
613 -mfloat-gprs=yes -mfloat-gprs=no @gol
614 -mprototype -mno-prototype @gol
615 -msim -mmvme -mads -myellowknife -memb -msdata @gol
616 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
618 @emph{S/390 and zSeries Options}
619 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
620 -mhard-float -msoft-float -mbackchain -mno-backchain -mkernel-backchain @gol
621 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
622 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
623 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
624 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
627 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
628 -m4-nofpu -m4-single-only -m4-single -m4 @gol
629 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
630 -m5-64media -m5-64media-nofpu @gol
631 -m5-32media -m5-32media-nofpu @gol
632 -m5-compact -m5-compact-nofpu @gol
633 -mb -ml -mdalign -mrelax @gol
634 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
635 -mieee -misize -mpadstruct -mspace @gol
636 -mprefergot -musermode}
639 @gccoptlist{-mcpu=@var{cpu-type} @gol
640 -mtune=@var{cpu-type} @gol
641 -mcmodel=@var{code-model} @gol
642 -m32 -m64 -mapp-regs -mno-app-regs @gol
643 -mfaster-structs -mno-faster-structs @gol
644 -mfpu -mno-fpu -mhard-float -msoft-float @gol
645 -mhard-quad-float -msoft-quad-float @gol
646 -mimpure-text -mno-impure-text -mlittle-endian @gol
647 -mstack-bias -mno-stack-bias @gol
648 -munaligned-doubles -mno-unaligned-doubles @gol
649 -mv8plus -mno-v8plus -mvis -mno-vis
652 @emph{System V Options}
653 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
655 @emph{TMS320C3x/C4x Options}
656 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
657 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
658 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
659 -mparallel-insns -mparallel-mpy -mpreserve-float}
662 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
663 -mprolog-function -mno-prolog-function -mspace @gol
664 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
665 -mapp-regs -mno-app-regs @gol
666 -mdisable-callt -mno-disable-callt @gol
672 @gccoptlist{-mg -mgnu -munix}
674 @emph{x86-64 Options}
675 See i386 and x86-64 Options.
677 @emph{Xstormy16 Options}
680 @emph{Xtensa Options}
681 @gccoptlist{-mconst16 -mno-const16 @gol
682 -mfused-madd -mno-fused-madd @gol
683 -mtext-section-literals -mno-text-section-literals @gol
684 -mtarget-align -mno-target-align @gol
685 -mlongcalls -mno-longcalls}
687 @emph{zSeries Options}
688 See S/390 and zSeries Options.
690 @item Code Generation Options
691 @xref{Code Gen Options,,Options for Code Generation Conventions}.
692 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
693 -ffixed-@var{reg} -fexceptions @gol
694 -fnon-call-exceptions -funwind-tables @gol
695 -fasynchronous-unwind-tables @gol
696 -finhibit-size-directive -finstrument-functions @gol
697 -fno-common -fno-ident @gol
698 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
699 -freg-struct-return -fshared-data -fshort-enums @gol
700 -fshort-double -fshort-wchar @gol
701 -fverbose-asm -fpack-struct -fstack-check @gol
702 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
703 -fargument-alias -fargument-noalias @gol
704 -fargument-noalias-global -fleading-underscore @gol
705 -ftls-model=@var{model} @gol
706 -ftrapv -fwrapv -fbounds-check @gol
711 * Overall Options:: Controlling the kind of output:
712 an executable, object files, assembler files,
713 or preprocessed source.
714 * C Dialect Options:: Controlling the variant of C language compiled.
715 * C++ Dialect Options:: Variations on C++.
716 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
718 * Language Independent Options:: Controlling how diagnostics should be
720 * Warning Options:: How picky should the compiler be?
721 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
722 * Optimize Options:: How much optimization?
723 * Preprocessor Options:: Controlling header files and macro definitions.
724 Also, getting dependency information for Make.
725 * Assembler Options:: Passing options to the assembler.
726 * Link Options:: Specifying libraries and so on.
727 * Directory Options:: Where to find header files and libraries.
728 Where to find the compiler executable files.
729 * Spec Files:: How to pass switches to sub-processes.
730 * Target Options:: Running a cross-compiler, or an old version of GCC.
733 @node Overall Options
734 @section Options Controlling the Kind of Output
736 Compilation can involve up to four stages: preprocessing, compilation
737 proper, assembly and linking, always in that order. GCC is capable of
738 preprocessing and compiling several files either into several
739 assembler input files, or into one assembler input file; then each
740 assembler input file produces an object file, and linking combines all
741 the object files (those newly compiled, and those specified as input)
742 into an executable file.
744 @cindex file name suffix
745 For any given input file, the file name suffix determines what kind of
750 C source code which must be preprocessed.
753 C source code which should not be preprocessed.
756 C++ source code which should not be preprocessed.
759 Objective-C source code. Note that you must link with the @file{libobjc}
760 library to make an Objective-C program work.
763 Objective-C source code which should not be preprocessed.
767 Objective-C++ source code. Note that you must link with the @file{libobjc}
768 library to make an Objective-C++ program work. Note that @samp{.M} refers
769 to a literal capital M@.
772 Objective-C++ source code which should not be preprocessed.
775 C, C++, Objective-C or Objective-C++ header file to be turned into a
780 @itemx @var{file}.cxx
781 @itemx @var{file}.cpp
782 @itemx @var{file}.CPP
783 @itemx @var{file}.c++
785 C++ source code which must be preprocessed. Note that in @samp{.cxx},
786 the last two letters must both be literally @samp{x}. Likewise,
787 @samp{.C} refers to a literal capital C@.
791 C++ header file to be turned into a precompiled header.
794 @itemx @var{file}.for
795 @itemx @var{file}.FOR
796 Fortran source code which should not be preprocessed.
799 @itemx @var{file}.fpp
800 @itemx @var{file}.FPP
801 Fortran source code which must be preprocessed (with the traditional
805 Fortran source code which must be preprocessed with a RATFOR
806 preprocessor (not included with GCC)@.
809 @itemx @var{file}.f95
810 Fortran 90/95 source code which should not be preprocessed.
812 @c FIXME: Descriptions of Java file types.
819 Ada source code file which contains a library unit declaration (a
820 declaration of a package, subprogram, or generic, or a generic
821 instantiation), or a library unit renaming declaration (a package,
822 generic, or subprogram renaming declaration). Such files are also
825 @itemx @var{file}.adb
826 Ada source code file containing a library unit body (a subprogram or
827 package body). Such files are also called @dfn{bodies}.
829 @c GCC also knows about some suffixes for languages not yet included:
838 Assembler code which must be preprocessed.
841 An object file to be fed straight into linking.
842 Any file name with no recognized suffix is treated this way.
846 You can specify the input language explicitly with the @option{-x} option:
849 @item -x @var{language}
850 Specify explicitly the @var{language} for the following input files
851 (rather than letting the compiler choose a default based on the file
852 name suffix). This option applies to all following input files until
853 the next @option{-x} option. Possible values for @var{language} are:
855 c c-header c-cpp-output
856 c++ c++-header c++-cpp-output
857 objective-c objective-c-header objective-c-cpp-output
858 objective-c++ objective-c++-header objective-c++-cpp-output
859 assembler assembler-with-cpp
861 f77 f77-cpp-input ratfor
868 Turn off any specification of a language, so that subsequent files are
869 handled according to their file name suffixes (as they are if @option{-x}
870 has not been used at all).
872 @item -pass-exit-codes
873 @opindex pass-exit-codes
874 Normally the @command{gcc} program will exit with the code of 1 if any
875 phase of the compiler returns a non-success return code. If you specify
876 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
877 numerically highest error produced by any phase that returned an error
881 If you only want some of the stages of compilation, you can use
882 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
883 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
884 @command{gcc} is to stop. Note that some combinations (for example,
885 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
890 Compile or assemble the source files, but do not link. The linking
891 stage simply is not done. The ultimate output is in the form of an
892 object file for each source file.
894 By default, the object file name for a source file is made by replacing
895 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
897 Unrecognized input files, not requiring compilation or assembly, are
902 Stop after the stage of compilation proper; do not assemble. The output
903 is in the form of an assembler code file for each non-assembler input
906 By default, the assembler file name for a source file is made by
907 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
909 Input files that don't require compilation are ignored.
913 Stop after the preprocessing stage; do not run the compiler proper. The
914 output is in the form of preprocessed source code, which is sent to the
917 Input files which don't require preprocessing are ignored.
919 @cindex output file option
922 Place output in file @var{file}. This applies regardless to whatever
923 sort of output is being produced, whether it be an executable file,
924 an object file, an assembler file or preprocessed C code.
926 If @option{-o} is not specified, the default is to put an executable
927 file in @file{a.out}, the object file for
928 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
929 assembler file in @file{@var{source}.s}, a precompiled header file in
930 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
935 Print (on standard error output) the commands executed to run the stages
936 of compilation. Also print the version number of the compiler driver
937 program and of the preprocessor and the compiler proper.
941 Like @option{-v} except the commands are not executed and all command
942 arguments are quoted. This is useful for shell scripts to capture the
943 driver-generated command lines.
947 Use pipes rather than temporary files for communication between the
948 various stages of compilation. This fails to work on some systems where
949 the assembler is unable to read from a pipe; but the GNU assembler has
954 If you are compiling multiple source files, this option tells the driver
955 to pass all the source files to the compiler at once (for those
956 languages for which the compiler can handle this). This will allow
957 intermodule analysis (IMA) to be performed by the compiler. Currently the only
958 language for which this is supported is C. If you pass source files for
959 multiple languages to the driver, using this option, the driver will invoke
960 the compiler(s) that support IMA once each, passing each compiler all the
961 source files appropriate for it. For those languages that do not support
962 IMA this option will be ignored, and the compiler will be invoked once for
963 each source file in that language. If you use this option in conjunction
964 with -save-temps, the compiler will generate multiple pre-processed files
965 (one for each source file), but only one (combined) .o or .s file.
969 Print (on the standard output) a description of the command line options
970 understood by @command{gcc}. If the @option{-v} option is also specified
971 then @option{--help} will also be passed on to the various processes
972 invoked by @command{gcc}, so that they can display the command line options
973 they accept. If the @option{-Wextra} option is also specified then command
974 line options which have no documentation associated with them will also
979 Print (on the standard output) a description of target specific command
980 line options for each tool.
984 Display the version number and copyrights of the invoked GCC.
988 @section Compiling C++ Programs
990 @cindex suffixes for C++ source
991 @cindex C++ source file suffixes
992 C++ source files conventionally use one of the suffixes @samp{.C},
993 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
994 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
995 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
996 files with these names and compiles them as C++ programs even if you
997 call the compiler the same way as for compiling C programs (usually
998 with the name @command{gcc}).
1002 However, C++ programs often require class libraries as well as a
1003 compiler that understands the C++ language---and under some
1004 circumstances, you might want to compile programs or header files from
1005 standard input, or otherwise without a suffix that flags them as C++
1006 programs. You might also like to precompile a C header file with a
1007 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1008 program that calls GCC with the default language set to C++, and
1009 automatically specifies linking against the C++ library. On many
1010 systems, @command{g++} is also installed with the name @command{c++}.
1012 @cindex invoking @command{g++}
1013 When you compile C++ programs, you may specify many of the same
1014 command-line options that you use for compiling programs in any
1015 language; or command-line options meaningful for C and related
1016 languages; or options that are meaningful only for C++ programs.
1017 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1018 explanations of options for languages related to C@.
1019 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1020 explanations of options that are meaningful only for C++ programs.
1022 @node C Dialect Options
1023 @section Options Controlling C Dialect
1024 @cindex dialect options
1025 @cindex language dialect options
1026 @cindex options, dialect
1028 The following options control the dialect of C (or languages derived
1029 from C, such as C++, Objective-C and Objective-C++) that the compiler
1033 @cindex ANSI support
1037 In C mode, support all ISO C90 programs. In C++ mode,
1038 remove GNU extensions that conflict with ISO C++.
1040 This turns off certain features of GCC that are incompatible with ISO
1041 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1042 such as the @code{asm} and @code{typeof} keywords, and
1043 predefined macros such as @code{unix} and @code{vax} that identify the
1044 type of system you are using. It also enables the undesirable and
1045 rarely used ISO trigraph feature. For the C compiler,
1046 it disables recognition of C++ style @samp{//} comments as well as
1047 the @code{inline} keyword.
1049 The alternate keywords @code{__asm__}, @code{__extension__},
1050 @code{__inline__} and @code{__typeof__} continue to work despite
1051 @option{-ansi}. You would not want to use them in an ISO C program, of
1052 course, but it is useful to put them in header files that might be included
1053 in compilations done with @option{-ansi}. Alternate predefined macros
1054 such as @code{__unix__} and @code{__vax__} are also available, with or
1055 without @option{-ansi}.
1057 The @option{-ansi} option does not cause non-ISO programs to be
1058 rejected gratuitously. For that, @option{-pedantic} is required in
1059 addition to @option{-ansi}. @xref{Warning Options}.
1061 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1062 option is used. Some header files may notice this macro and refrain
1063 from declaring certain functions or defining certain macros that the
1064 ISO standard doesn't call for; this is to avoid interfering with any
1065 programs that might use these names for other things.
1067 Functions which would normally be built in but do not have semantics
1068 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1069 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1070 built-in functions provided by GCC}, for details of the functions
1075 Determine the language standard. This option is currently only
1076 supported when compiling C or C++. A value for this option must be
1077 provided; possible values are
1082 ISO C90 (same as @option{-ansi}).
1084 @item iso9899:199409
1085 ISO C90 as modified in amendment 1.
1091 ISO C99. Note that this standard is not yet fully supported; see
1092 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1093 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1096 Default, ISO C90 plus GNU extensions (including some C99 features).
1100 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1101 this will become the default. The name @samp{gnu9x} is deprecated.
1104 The 1998 ISO C++ standard plus amendments.
1107 The same as @option{-std=c++98} plus GNU extensions. This is the
1108 default for C++ code.
1111 Even when this option is not specified, you can still use some of the
1112 features of newer standards in so far as they do not conflict with
1113 previous C standards. For example, you may use @code{__restrict__} even
1114 when @option{-std=c99} is not specified.
1116 The @option{-std} options specifying some version of ISO C have the same
1117 effects as @option{-ansi}, except that features that were not in ISO C90
1118 but are in the specified version (for example, @samp{//} comments and
1119 the @code{inline} keyword in ISO C99) are not disabled.
1121 @xref{Standards,,Language Standards Supported by GCC}, for details of
1122 these standard versions.
1124 @item -aux-info @var{filename}
1126 Output to the given filename prototyped declarations for all functions
1127 declared and/or defined in a translation unit, including those in header
1128 files. This option is silently ignored in any language other than C@.
1130 Besides declarations, the file indicates, in comments, the origin of
1131 each declaration (source file and line), whether the declaration was
1132 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1133 @samp{O} for old, respectively, in the first character after the line
1134 number and the colon), and whether it came from a declaration or a
1135 definition (@samp{C} or @samp{F}, respectively, in the following
1136 character). In the case of function definitions, a K&R-style list of
1137 arguments followed by their declarations is also provided, inside
1138 comments, after the declaration.
1142 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1143 keyword, so that code can use these words as identifiers. You can use
1144 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1145 instead. @option{-ansi} implies @option{-fno-asm}.
1147 In C++, this switch only affects the @code{typeof} keyword, since
1148 @code{asm} and @code{inline} are standard keywords. You may want to
1149 use the @option{-fno-gnu-keywords} flag instead, which has the same
1150 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1151 switch only affects the @code{asm} and @code{typeof} keywords, since
1152 @code{inline} is a standard keyword in ISO C99.
1155 @itemx -fno-builtin-@var{function}
1156 @opindex fno-builtin
1157 @cindex built-in functions
1158 Don't recognize built-in functions that do not begin with
1159 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1160 functions provided by GCC}, for details of the functions affected,
1161 including those which are not built-in functions when @option{-ansi} or
1162 @option{-std} options for strict ISO C conformance are used because they
1163 do not have an ISO standard meaning.
1165 GCC normally generates special code to handle certain built-in functions
1166 more efficiently; for instance, calls to @code{alloca} may become single
1167 instructions that adjust the stack directly, and calls to @code{memcpy}
1168 may become inline copy loops. The resulting code is often both smaller
1169 and faster, but since the function calls no longer appear as such, you
1170 cannot set a breakpoint on those calls, nor can you change the behavior
1171 of the functions by linking with a different library.
1173 With the @option{-fno-builtin-@var{function}} option
1174 only the built-in function @var{function} is
1175 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1176 function is named this is not built-in in this version of GCC, this
1177 option is ignored. There is no corresponding
1178 @option{-fbuiltin-@var{function}} option; if you wish to enable
1179 built-in functions selectively when using @option{-fno-builtin} or
1180 @option{-ffreestanding}, you may define macros such as:
1183 #define abs(n) __builtin_abs ((n))
1184 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1189 @cindex hosted environment
1191 Assert that compilation takes place in a hosted environment. This implies
1192 @option{-fbuiltin}. A hosted environment is one in which the
1193 entire standard library is available, and in which @code{main} has a return
1194 type of @code{int}. Examples are nearly everything except a kernel.
1195 This is equivalent to @option{-fno-freestanding}.
1197 @item -ffreestanding
1198 @opindex ffreestanding
1199 @cindex hosted environment
1201 Assert that compilation takes place in a freestanding environment. This
1202 implies @option{-fno-builtin}. A freestanding environment
1203 is one in which the standard library may not exist, and program startup may
1204 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1205 This is equivalent to @option{-fno-hosted}.
1207 @xref{Standards,,Language Standards Supported by GCC}, for details of
1208 freestanding and hosted environments.
1210 @item -fms-extensions
1211 @opindex fms-extensions
1212 Accept some non-standard constructs used in Microsoft header files.
1216 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1217 options for strict ISO C conformance) implies @option{-trigraphs}.
1219 @item -no-integrated-cpp
1220 @opindex no-integrated-cpp
1221 Performs a compilation in two passes: preprocessing and compiling. This
1222 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1223 @option{-B} option. The user supplied compilation step can then add in
1224 an additional preprocessing step after normal preprocessing but before
1225 compiling. The default is to use the integrated cpp (internal cpp)
1227 The semantics of this option will change if "cc1", "cc1plus", and
1228 "cc1obj" are merged.
1230 @cindex traditional C language
1231 @cindex C language, traditional
1233 @itemx -traditional-cpp
1234 @opindex traditional-cpp
1235 @opindex traditional
1236 Formerly, these options caused GCC to attempt to emulate a pre-standard
1237 C compiler. They are now only supported with the @option{-E} switch.
1238 The preprocessor continues to support a pre-standard mode. See the GNU
1239 CPP manual for details.
1241 @item -fcond-mismatch
1242 @opindex fcond-mismatch
1243 Allow conditional expressions with mismatched types in the second and
1244 third arguments. The value of such an expression is void. This option
1245 is not supported for C++.
1247 @item -funsigned-char
1248 @opindex funsigned-char
1249 Let the type @code{char} be unsigned, like @code{unsigned char}.
1251 Each kind of machine has a default for what @code{char} should
1252 be. It is either like @code{unsigned char} by default or like
1253 @code{signed char} by default.
1255 Ideally, a portable program should always use @code{signed char} or
1256 @code{unsigned char} when it depends on the signedness of an object.
1257 But many programs have been written to use plain @code{char} and
1258 expect it to be signed, or expect it to be unsigned, depending on the
1259 machines they were written for. This option, and its inverse, let you
1260 make such a program work with the opposite default.
1262 The type @code{char} is always a distinct type from each of
1263 @code{signed char} or @code{unsigned char}, even though its behavior
1264 is always just like one of those two.
1267 @opindex fsigned-char
1268 Let the type @code{char} be signed, like @code{signed char}.
1270 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1271 the negative form of @option{-funsigned-char}. Likewise, the option
1272 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1274 @item -fsigned-bitfields
1275 @itemx -funsigned-bitfields
1276 @itemx -fno-signed-bitfields
1277 @itemx -fno-unsigned-bitfields
1278 @opindex fsigned-bitfields
1279 @opindex funsigned-bitfields
1280 @opindex fno-signed-bitfields
1281 @opindex fno-unsigned-bitfields
1282 These options control whether a bit-field is signed or unsigned, when the
1283 declaration does not use either @code{signed} or @code{unsigned}. By
1284 default, such a bit-field is signed, because this is consistent: the
1285 basic integer types such as @code{int} are signed types.
1288 @node C++ Dialect Options
1289 @section Options Controlling C++ Dialect
1291 @cindex compiler options, C++
1292 @cindex C++ options, command line
1293 @cindex options, C++
1294 This section describes the command-line options that are only meaningful
1295 for C++ programs; but you can also use most of the GNU compiler options
1296 regardless of what language your program is in. For example, you
1297 might compile a file @code{firstClass.C} like this:
1300 g++ -g -frepo -O -c firstClass.C
1304 In this example, only @option{-frepo} is an option meant
1305 only for C++ programs; you can use the other options with any
1306 language supported by GCC@.
1308 Here is a list of options that are @emph{only} for compiling C++ programs:
1312 @item -fabi-version=@var{n}
1313 @opindex fabi-version
1314 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1315 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1316 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1317 the version that conforms most closely to the C++ ABI specification.
1318 Therefore, the ABI obtained using version 0 will change as ABI bugs
1321 The default is version 2.
1323 @item -fno-access-control
1324 @opindex fno-access-control
1325 Turn off all access checking. This switch is mainly useful for working
1326 around bugs in the access control code.
1330 Check that the pointer returned by @code{operator new} is non-null
1331 before attempting to modify the storage allocated. This check is
1332 normally unnecessary because the C++ standard specifies that
1333 @code{operator new} will only return @code{0} if it is declared
1334 @samp{throw()}, in which case the compiler will always check the
1335 return value even without this option. In all other cases, when
1336 @code{operator new} has a non-empty exception specification, memory
1337 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1338 @samp{new (nothrow)}.
1340 @item -fconserve-space
1341 @opindex fconserve-space
1342 Put uninitialized or runtime-initialized global variables into the
1343 common segment, as C does. This saves space in the executable at the
1344 cost of not diagnosing duplicate definitions. If you compile with this
1345 flag and your program mysteriously crashes after @code{main()} has
1346 completed, you may have an object that is being destroyed twice because
1347 two definitions were merged.
1349 This option is no longer useful on most targets, now that support has
1350 been added for putting variables into BSS without making them common.
1352 @item -fno-const-strings
1353 @opindex fno-const-strings
1354 Give string constants type @code{char *} instead of type @code{const
1355 char *}. By default, G++ uses type @code{const char *} as required by
1356 the standard. Even if you use @option{-fno-const-strings}, you cannot
1357 actually modify the value of a string constant.
1359 This option might be removed in a future release of G++. For maximum
1360 portability, you should structure your code so that it works with
1361 string constants that have type @code{const char *}.
1363 @item -fno-elide-constructors
1364 @opindex fno-elide-constructors
1365 The C++ standard allows an implementation to omit creating a temporary
1366 which is only used to initialize another object of the same type.
1367 Specifying this option disables that optimization, and forces G++ to
1368 call the copy constructor in all cases.
1370 @item -fno-enforce-eh-specs
1371 @opindex fno-enforce-eh-specs
1372 Don't check for violation of exception specifications at runtime. This
1373 option violates the C++ standard, but may be useful for reducing code
1374 size in production builds, much like defining @samp{NDEBUG}. The compiler
1375 will still optimize based on the exception specifications.
1378 @itemx -fno-for-scope
1380 @opindex fno-for-scope
1381 If @option{-ffor-scope} is specified, the scope of variables declared in
1382 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1383 as specified by the C++ standard.
1384 If @option{-fno-for-scope} is specified, the scope of variables declared in
1385 a @i{for-init-statement} extends to the end of the enclosing scope,
1386 as was the case in old versions of G++, and other (traditional)
1387 implementations of C++.
1389 The default if neither flag is given to follow the standard,
1390 but to allow and give a warning for old-style code that would
1391 otherwise be invalid, or have different behavior.
1393 @item -fno-gnu-keywords
1394 @opindex fno-gnu-keywords
1395 Do not recognize @code{typeof} as a keyword, so that code can use this
1396 word as an identifier. You can use the keyword @code{__typeof__} instead.
1397 @option{-ansi} implies @option{-fno-gnu-keywords}.
1399 @item -fno-implicit-templates
1400 @opindex fno-implicit-templates
1401 Never emit code for non-inline templates which are instantiated
1402 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1403 @xref{Template Instantiation}, for more information.
1405 @item -fno-implicit-inline-templates
1406 @opindex fno-implicit-inline-templates
1407 Don't emit code for implicit instantiations of inline templates, either.
1408 The default is to handle inlines differently so that compiles with and
1409 without optimization will need the same set of explicit instantiations.
1411 @item -fno-implement-inlines
1412 @opindex fno-implement-inlines
1413 To save space, do not emit out-of-line copies of inline functions
1414 controlled by @samp{#pragma implementation}. This will cause linker
1415 errors if these functions are not inlined everywhere they are called.
1417 @item -fms-extensions
1418 @opindex fms-extensions
1419 Disable pedantic warnings about constructs used in MFC, such as implicit
1420 int and getting a pointer to member function via non-standard syntax.
1422 @item -fno-nonansi-builtins
1423 @opindex fno-nonansi-builtins
1424 Disable built-in declarations of functions that are not mandated by
1425 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1426 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1428 @item -fno-operator-names
1429 @opindex fno-operator-names
1430 Do not treat the operator name keywords @code{and}, @code{bitand},
1431 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1432 synonyms as keywords.
1434 @item -fno-optional-diags
1435 @opindex fno-optional-diags
1436 Disable diagnostics that the standard says a compiler does not need to
1437 issue. Currently, the only such diagnostic issued by G++ is the one for
1438 a name having multiple meanings within a class.
1441 @opindex fpermissive
1442 Downgrade some diagnostics about nonconformant code from errors to
1443 warnings. Thus, using @option{-fpermissive} will allow some
1444 nonconforming code to compile.
1448 Enable automatic template instantiation at link time. This option also
1449 implies @option{-fno-implicit-templates}. @xref{Template
1450 Instantiation}, for more information.
1454 Disable generation of information about every class with virtual
1455 functions for use by the C++ runtime type identification features
1456 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1457 of the language, you can save some space by using this flag. Note that
1458 exception handling uses the same information, but it will generate it as
1463 Emit statistics about front-end processing at the end of the compilation.
1464 This information is generally only useful to the G++ development team.
1466 @item -ftemplate-depth-@var{n}
1467 @opindex ftemplate-depth
1468 Set the maximum instantiation depth for template classes to @var{n}.
1469 A limit on the template instantiation depth is needed to detect
1470 endless recursions during template class instantiation. ANSI/ISO C++
1471 conforming programs must not rely on a maximum depth greater than 17.
1473 @item -fuse-cxa-atexit
1474 @opindex fuse-cxa-atexit
1475 Register destructors for objects with static storage duration with the
1476 @code{__cxa_atexit} function rather than the @code{atexit} function.
1477 This option is required for fully standards-compliant handling of static
1478 destructors, but will only work if your C library supports
1479 @code{__cxa_atexit}.
1481 @item -fvisibility-inlines-hidden
1482 @opindex fvisibility-inlines-hidden
1483 Causes all inlined methods to be marked with
1484 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1485 appear in the export table of a DSO and do not require a PLT indirection
1486 when used within the DSO. Enabling this option can have a dramatic effect
1487 on load and link times of a DSO as it massively reduces the size of the
1488 dynamic export table when the library makes heavy use of templates. While
1489 it can cause bloating through duplication of code within each DSO where
1490 it is used, often the wastage is less than the considerable space occupied
1491 by a long symbol name in the export table which is typical when using
1492 templates and namespaces. For even more savings, combine with the
1493 @code{-fvisibility=hidden} switch.
1497 Do not use weak symbol support, even if it is provided by the linker.
1498 By default, G++ will use weak symbols if they are available. This
1499 option exists only for testing, and should not be used by end-users;
1500 it will result in inferior code and has no benefits. This option may
1501 be removed in a future release of G++.
1505 Do not search for header files in the standard directories specific to
1506 C++, but do still search the other standard directories. (This option
1507 is used when building the C++ library.)
1510 In addition, these optimization, warning, and code generation options
1511 have meanings only for C++ programs:
1514 @item -fno-default-inline
1515 @opindex fno-default-inline
1516 Do not assume @samp{inline} for functions defined inside a class scope.
1517 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1518 functions will have linkage like inline functions; they just won't be
1521 @item -Wabi @r{(C++ only)}
1523 Warn when G++ generates code that is probably not compatible with the
1524 vendor-neutral C++ ABI. Although an effort has been made to warn about
1525 all such cases, there are probably some cases that are not warned about,
1526 even though G++ is generating incompatible code. There may also be
1527 cases where warnings are emitted even though the code that is generated
1530 You should rewrite your code to avoid these warnings if you are
1531 concerned about the fact that code generated by G++ may not be binary
1532 compatible with code generated by other compilers.
1534 The known incompatibilities at this point include:
1539 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1540 pack data into the same byte as a base class. For example:
1543 struct A @{ virtual void f(); int f1 : 1; @};
1544 struct B : public A @{ int f2 : 1; @};
1548 In this case, G++ will place @code{B::f2} into the same byte
1549 as@code{A::f1}; other compilers will not. You can avoid this problem
1550 by explicitly padding @code{A} so that its size is a multiple of the
1551 byte size on your platform; that will cause G++ and other compilers to
1552 layout @code{B} identically.
1555 Incorrect handling of tail-padding for virtual bases. G++ does not use
1556 tail padding when laying out virtual bases. For example:
1559 struct A @{ virtual void f(); char c1; @};
1560 struct B @{ B(); char c2; @};
1561 struct C : public A, public virtual B @{@};
1565 In this case, G++ will not place @code{B} into the tail-padding for
1566 @code{A}; other compilers will. You can avoid this problem by
1567 explicitly padding @code{A} so that its size is a multiple of its
1568 alignment (ignoring virtual base classes); that will cause G++ and other
1569 compilers to layout @code{C} identically.
1572 Incorrect handling of bit-fields with declared widths greater than that
1573 of their underlying types, when the bit-fields appear in a union. For
1577 union U @{ int i : 4096; @};
1581 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1582 union too small by the number of bits in an @code{int}.
1585 Empty classes can be placed at incorrect offsets. For example:
1595 struct C : public B, public A @{@};
1599 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1600 it should be placed at offset zero. G++ mistakenly believes that the
1601 @code{A} data member of @code{B} is already at offset zero.
1604 Names of template functions whose types involve @code{typename} or
1605 template template parameters can be mangled incorrectly.
1608 template <typename Q>
1609 void f(typename Q::X) @{@}
1611 template <template <typename> class Q>
1612 void f(typename Q<int>::X) @{@}
1616 Instantiations of these templates may be mangled incorrectly.
1620 @item -Wctor-dtor-privacy @r{(C++ only)}
1621 @opindex Wctor-dtor-privacy
1622 Warn when a class seems unusable because all the constructors or
1623 destructors in that class are private, and it has neither friends nor
1624 public static member functions.
1626 @item -Wnon-virtual-dtor @r{(C++ only)}
1627 @opindex Wnon-virtual-dtor
1628 Warn when a class appears to be polymorphic, thereby requiring a virtual
1629 destructor, yet it declares a non-virtual one.
1630 This warning is enabled by @option{-Wall}.
1632 @item -Wreorder @r{(C++ only)}
1634 @cindex reordering, warning
1635 @cindex warning for reordering of member initializers
1636 Warn when the order of member initializers given in the code does not
1637 match the order in which they must be executed. For instance:
1643 A(): j (0), i (1) @{ @}
1647 The compiler will rearrange the member initializers for @samp{i}
1648 and @samp{j} to match the declaration order of the members, emitting
1649 a warning to that effect. This warning is enabled by @option{-Wall}.
1652 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1655 @item -Weffc++ @r{(C++ only)}
1657 Warn about violations of the following style guidelines from Scott Meyers'
1658 @cite{Effective C++} book:
1662 Item 11: Define a copy constructor and an assignment operator for classes
1663 with dynamically allocated memory.
1666 Item 12: Prefer initialization to assignment in constructors.
1669 Item 14: Make destructors virtual in base classes.
1672 Item 15: Have @code{operator=} return a reference to @code{*this}.
1675 Item 23: Don't try to return a reference when you must return an object.
1679 Also warn about violations of the following style guidelines from
1680 Scott Meyers' @cite{More Effective C++} book:
1684 Item 6: Distinguish between prefix and postfix forms of increment and
1685 decrement operators.
1688 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1692 When selecting this option, be aware that the standard library
1693 headers do not obey all of these guidelines; use @samp{grep -v}
1694 to filter out those warnings.
1696 @item -Wno-deprecated @r{(C++ only)}
1697 @opindex Wno-deprecated
1698 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1700 @item -Wno-non-template-friend @r{(C++ only)}
1701 @opindex Wno-non-template-friend
1702 Disable warnings when non-templatized friend functions are declared
1703 within a template. Since the advent of explicit template specification
1704 support in G++, if the name of the friend is an unqualified-id (i.e.,
1705 @samp{friend foo(int)}), the C++ language specification demands that the
1706 friend declare or define an ordinary, nontemplate function. (Section
1707 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1708 could be interpreted as a particular specialization of a templatized
1709 function. Because this non-conforming behavior is no longer the default
1710 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1711 check existing code for potential trouble spots and is on by default.
1712 This new compiler behavior can be turned off with
1713 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1714 but disables the helpful warning.
1716 @item -Wold-style-cast @r{(C++ only)}
1717 @opindex Wold-style-cast
1718 Warn if an old-style (C-style) cast to a non-void type is used within
1719 a C++ program. The new-style casts (@samp{static_cast},
1720 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1721 unintended effects and much easier to search for.
1723 @item -Woverloaded-virtual @r{(C++ only)}
1724 @opindex Woverloaded-virtual
1725 @cindex overloaded virtual fn, warning
1726 @cindex warning for overloaded virtual fn
1727 Warn when a function declaration hides virtual functions from a
1728 base class. For example, in:
1735 struct B: public A @{
1740 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1748 will fail to compile.
1750 @item -Wno-pmf-conversions @r{(C++ only)}
1751 @opindex Wno-pmf-conversions
1752 Disable the diagnostic for converting a bound pointer to member function
1755 @item -Wsign-promo @r{(C++ only)}
1756 @opindex Wsign-promo
1757 Warn when overload resolution chooses a promotion from unsigned or
1758 enumerated type to a signed type, over a conversion to an unsigned type of
1759 the same size. Previous versions of G++ would try to preserve
1760 unsignedness, but the standard mandates the current behavior.
1762 @item -Wsynth @r{(C++ only)}
1764 @cindex warning for synthesized methods
1765 @cindex synthesized methods, warning
1766 Warn when G++'s synthesis behavior does not match that of cfront. For
1772 A& operator = (int);
1782 In this example, G++ will synthesize a default @samp{A& operator =
1783 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1786 @node Objective-C and Objective-C++ Dialect Options
1787 @section Options Controlling Objective-C and Objective-C++ Dialects
1789 @cindex compiler options, Objective-C and Objective-C++
1790 @cindex Objective-C and Objective-C++ options, command line
1791 @cindex options, Objective-C and Objective-C++
1792 (NOTE: This manual does not describe the Objective-C and Objective-C++
1793 languages themselves. See @xref{Standards,,Language Standards
1794 Supported by GCC}, for references.)
1796 This section describes the command-line options that are only meaningful
1797 for Objective-C and Objective-C++ programs, but you can also use most of
1798 the language-independent GNU compiler options.
1799 For example, you might compile a file @code{some_class.m} like this:
1802 gcc -g -fgnu-runtime -O -c some_class.m
1806 In this example, @option{-fgnu-runtime} is an option meant only for
1807 Objective-C and Objective-C++ programs; you can use the other options with
1808 any language supported by GCC@.
1810 Note that since Objective-C is an extension of the C language, Objective-C
1811 compilations may also use options specific to the C front-end (e.g.,
1812 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1813 C++-specific options (e.g., @option{-Wabi}).
1815 Here is a list of options that are @emph{only} for compiling Objective-C
1816 and Objective-C++ programs:
1819 @item -fconstant-string-class=@var{class-name}
1820 @opindex fconstant-string-class
1821 Use @var{class-name} as the name of the class to instantiate for each
1822 literal string specified with the syntax @code{@@"@dots{}"}. The default
1823 class name is @code{NXConstantString} if the GNU runtime is being used, and
1824 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1825 @option{-fconstant-cfstrings} option, if also present, will override the
1826 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1827 to be laid out as constant CoreFoundation strings.
1830 @opindex fgnu-runtime
1831 Generate object code compatible with the standard GNU Objective-C
1832 runtime. This is the default for most types of systems.
1834 @item -fnext-runtime
1835 @opindex fnext-runtime
1836 Generate output compatible with the NeXT runtime. This is the default
1837 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1838 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1841 @item -fno-nil-receivers
1842 @opindex fno-nil-receivers
1843 Assume that all Objective-C message dispatches (e.g.,
1844 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1845 is not @code{nil}. This allows for more efficient entry points in the runtime
1846 to be used. Currently, this option is only available in conjunction with
1847 the NeXT runtime on Mac OS X 10.3 and later.
1849 @item -fobjc-exceptions
1850 @opindex fobjc-exceptions
1851 Enable syntactic support for structured exception handling in Objective-C,
1852 similar to what is offered by C++ and Java. Currently, this option is only
1853 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1861 @@catch (AnObjCClass *exc) @{
1868 @@catch (AnotherClass *exc) @{
1871 @@catch (id allOthers) @{
1881 The @code{@@throw} statement may appear anywhere in an Objective-C or
1882 Objective-C++ program; when used inside of a @code{@@catch} block, the
1883 @code{@@throw} may appear without an argument (as shown above), in which case
1884 the object caught by the @code{@@catch} will be rethrown.
1886 Note that only (pointers to) Objective-C objects may be thrown and
1887 caught using this scheme. When an object is thrown, it will be caught
1888 by the nearest @code{@@catch} clause capable of handling objects of that type,
1889 analogously to how @code{catch} blocks work in C++ and Java. A
1890 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1891 any and all Objective-C exceptions not caught by previous @code{@@catch}
1894 The @code{@@finally} clause, if present, will be executed upon exit from the
1895 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1896 regardless of whether any exceptions are thrown, caught or rethrown
1897 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1898 of the @code{finally} clause in Java.
1900 There are several caveats to using the new exception mechanism:
1904 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1905 idioms provided by the @code{NSException} class, the new
1906 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1907 systems, due to additional functionality needed in the (NeXT) Objective-C
1911 As mentioned above, the new exceptions do not support handling
1912 types other than Objective-C objects. Furthermore, when used from
1913 Objective-C++, the Objective-C exception model does not interoperate with C++
1914 exceptions at this time. This means you cannot @code{@@throw} an exception
1915 from Objective-C and @code{catch} it in C++, or vice versa
1916 (i.e., @code{throw @dots{} @@catch}).
1919 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1920 blocks for thread-safe execution:
1923 @@synchronized (ObjCClass *guard) @{
1928 Upon entering the @code{@@synchronized} block, a thread of execution shall
1929 first check whether a lock has been placed on the corresponding @code{guard}
1930 object by another thread. If it has, the current thread shall wait until
1931 the other thread relinquishes its lock. Once @code{guard} becomes available,
1932 the current thread will place its own lock on it, execute the code contained in
1933 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1934 making @code{guard} available to other threads).
1936 Unlike Java, Objective-C does not allow for entire methods to be marked
1937 @code{@@synchronized}. Note that throwing exceptions out of
1938 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1939 to be unlocked properly.
1941 @item -freplace-objc-classes
1942 @opindex freplace-objc-classes
1943 Emit a special marker instructing @command{ld(1)} not to statically link in
1944 the resulting object file, and allow @command{dyld(1)} to load it in at
1945 run time instead. This is used in conjunction with the Fix-and-Continue
1946 debugging mode, where the object file in question may be recompiled and
1947 dynamically reloaded in the course of program execution, without the need
1948 to restart the program itself. Currently, Fix-and-Continue functionality
1949 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1954 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1955 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1956 compile time) with static class references that get initialized at load time,
1957 which improves run-time performance. Specifying the @option{-fzero-link} flag
1958 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1959 to be retained. This is useful in Zero-Link debugging mode, since it allows
1960 for individual class implementations to be modified during program execution.
1964 Dump interface declarations for all classes seen in the source file to a
1965 file named @file{@var{sourcename}.decl}.
1968 @opindex Wno-protocol
1969 If a class is declared to implement a protocol, a warning is issued for
1970 every method in the protocol that is not implemented by the class. The
1971 default behavior is to issue a warning for every method not explicitly
1972 implemented in the class, even if a method implementation is inherited
1973 from the superclass. If you use the @code{-Wno-protocol} option, then
1974 methods inherited from the superclass are considered to be implemented,
1975 and no warning is issued for them.
1979 Warn if multiple methods of different types for the same selector are
1980 found during compilation. The check is performed on the list of methods
1981 in the final stage of compilation. Additionally, a check is performed
1982 for each selector appearing in a @code{@@selector(@dots{})}
1983 expression, and a corresponding method for that selector has been found
1984 during compilation. Because these checks scan the method table only at
1985 the end of compilation, these warnings are not produced if the final
1986 stage of compilation is not reached, for example because an error is
1987 found during compilation, or because the @code{-fsyntax-only} option is
1990 @item -Wundeclared-selector
1991 @opindex Wundeclared-selector
1992 Warn if a @code{@@selector(@dots{})} expression referring to an
1993 undeclared selector is found. A selector is considered undeclared if no
1994 method with that name has been declared before the
1995 @code{@@selector(@dots{})} expression, either explicitly in an
1996 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1997 an @code{@@implementation} section. This option always performs its
1998 checks as soon as a @code{@@selector(@dots{})} expression is found,
1999 while @code{-Wselector} only performs its checks in the final stage of
2000 compilation. This also enforces the coding style convention
2001 that methods and selectors must be declared before being used.
2003 @item -print-objc-runtime-info
2004 @opindex print-objc-runtime-info
2005 Generate C header describing the largest structure that is passed by
2010 @node Language Independent Options
2011 @section Options to Control Diagnostic Messages Formatting
2012 @cindex options to control diagnostics formatting
2013 @cindex diagnostic messages
2014 @cindex message formatting
2016 Traditionally, diagnostic messages have been formatted irrespective of
2017 the output device's aspect (e.g.@: its width, @dots{}). The options described
2018 below can be used to control the diagnostic messages formatting
2019 algorithm, e.g.@: how many characters per line, how often source location
2020 information should be reported. Right now, only the C++ front end can
2021 honor these options. However it is expected, in the near future, that
2022 the remaining front ends would be able to digest them correctly.
2025 @item -fmessage-length=@var{n}
2026 @opindex fmessage-length
2027 Try to format error messages so that they fit on lines of about @var{n}
2028 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2029 the front ends supported by GCC@. If @var{n} is zero, then no
2030 line-wrapping will be done; each error message will appear on a single
2033 @opindex fdiagnostics-show-location
2034 @item -fdiagnostics-show-location=once
2035 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2036 reporter to emit @emph{once} source location information; that is, in
2037 case the message is too long to fit on a single physical line and has to
2038 be wrapped, the source location won't be emitted (as prefix) again,
2039 over and over, in subsequent continuation lines. This is the default
2042 @item -fdiagnostics-show-location=every-line
2043 Only meaningful in line-wrapping mode. Instructs the diagnostic
2044 messages reporter to emit the same source location information (as
2045 prefix) for physical lines that result from the process of breaking
2046 a message which is too long to fit on a single line.
2050 @node Warning Options
2051 @section Options to Request or Suppress Warnings
2052 @cindex options to control warnings
2053 @cindex warning messages
2054 @cindex messages, warning
2055 @cindex suppressing warnings
2057 Warnings are diagnostic messages that report constructions which
2058 are not inherently erroneous but which are risky or suggest there
2059 may have been an error.
2061 You can request many specific warnings with options beginning @samp{-W},
2062 for example @option{-Wimplicit} to request warnings on implicit
2063 declarations. Each of these specific warning options also has a
2064 negative form beginning @samp{-Wno-} to turn off warnings;
2065 for example, @option{-Wno-implicit}. This manual lists only one of the
2066 two forms, whichever is not the default.
2068 The following options control the amount and kinds of warnings produced
2069 by GCC; for further, language-specific options also refer to
2070 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2074 @cindex syntax checking
2076 @opindex fsyntax-only
2077 Check the code for syntax errors, but don't do anything beyond that.
2081 Issue all the warnings demanded by strict ISO C and ISO C++;
2082 reject all programs that use forbidden extensions, and some other
2083 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2084 version of the ISO C standard specified by any @option{-std} option used.
2086 Valid ISO C and ISO C++ programs should compile properly with or without
2087 this option (though a rare few will require @option{-ansi} or a
2088 @option{-std} option specifying the required version of ISO C)@. However,
2089 without this option, certain GNU extensions and traditional C and C++
2090 features are supported as well. With this option, they are rejected.
2092 @option{-pedantic} does not cause warning messages for use of the
2093 alternate keywords whose names begin and end with @samp{__}. Pedantic
2094 warnings are also disabled in the expression that follows
2095 @code{__extension__}. However, only system header files should use
2096 these escape routes; application programs should avoid them.
2097 @xref{Alternate Keywords}.
2099 Some users try to use @option{-pedantic} to check programs for strict ISO
2100 C conformance. They soon find that it does not do quite what they want:
2101 it finds some non-ISO practices, but not all---only those for which
2102 ISO C @emph{requires} a diagnostic, and some others for which
2103 diagnostics have been added.
2105 A feature to report any failure to conform to ISO C might be useful in
2106 some instances, but would require considerable additional work and would
2107 be quite different from @option{-pedantic}. We don't have plans to
2108 support such a feature in the near future.
2110 Where the standard specified with @option{-std} represents a GNU
2111 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2112 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2113 extended dialect is based. Warnings from @option{-pedantic} are given
2114 where they are required by the base standard. (It would not make sense
2115 for such warnings to be given only for features not in the specified GNU
2116 C dialect, since by definition the GNU dialects of C include all
2117 features the compiler supports with the given option, and there would be
2118 nothing to warn about.)
2120 @item -pedantic-errors
2121 @opindex pedantic-errors
2122 Like @option{-pedantic}, except that errors are produced rather than
2127 Inhibit all warning messages.
2131 Inhibit warning messages about the use of @samp{#import}.
2133 @item -Wchar-subscripts
2134 @opindex Wchar-subscripts
2135 Warn if an array subscript has type @code{char}. This is a common cause
2136 of error, as programmers often forget that this type is signed on some
2141 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2142 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2144 @item -Wfatal-errors
2145 @opindex Wfatal-errors
2146 This option causes the compiler to abort compilation on the first error
2147 occurred rather than trying to keep going and printing further error
2152 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2153 the arguments supplied have types appropriate to the format string
2154 specified, and that the conversions specified in the format string make
2155 sense. This includes standard functions, and others specified by format
2156 attributes (@pxref{Function Attributes}), in the @code{printf},
2157 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2158 not in the C standard) families (or other target-specific families).
2160 The formats are checked against the format features supported by GNU
2161 libc version 2.2. These include all ISO C90 and C99 features, as well
2162 as features from the Single Unix Specification and some BSD and GNU
2163 extensions. Other library implementations may not support all these
2164 features; GCC does not support warning about features that go beyond a
2165 particular library's limitations. However, if @option{-pedantic} is used
2166 with @option{-Wformat}, warnings will be given about format features not
2167 in the selected standard version (but not for @code{strfmon} formats,
2168 since those are not in any version of the C standard). @xref{C Dialect
2169 Options,,Options Controlling C Dialect}.
2171 Since @option{-Wformat} also checks for null format arguments for
2172 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2174 @option{-Wformat} is included in @option{-Wall}. For more control over some
2175 aspects of format checking, the options @option{-Wformat-y2k},
2176 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2177 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2178 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2181 @opindex Wformat-y2k
2182 If @option{-Wformat} is specified, also warn about @code{strftime}
2183 formats which may yield only a two-digit year.
2185 @item -Wno-format-extra-args
2186 @opindex Wno-format-extra-args
2187 If @option{-Wformat} is specified, do not warn about excess arguments to a
2188 @code{printf} or @code{scanf} format function. The C standard specifies
2189 that such arguments are ignored.
2191 Where the unused arguments lie between used arguments that are
2192 specified with @samp{$} operand number specifications, normally
2193 warnings are still given, since the implementation could not know what
2194 type to pass to @code{va_arg} to skip the unused arguments. However,
2195 in the case of @code{scanf} formats, this option will suppress the
2196 warning if the unused arguments are all pointers, since the Single
2197 Unix Specification says that such unused arguments are allowed.
2199 @item -Wno-format-zero-length
2200 @opindex Wno-format-zero-length
2201 If @option{-Wformat} is specified, do not warn about zero-length formats.
2202 The C standard specifies that zero-length formats are allowed.
2204 @item -Wformat-nonliteral
2205 @opindex Wformat-nonliteral
2206 If @option{-Wformat} is specified, also warn if the format string is not a
2207 string literal and so cannot be checked, unless the format function
2208 takes its format arguments as a @code{va_list}.
2210 @item -Wformat-security
2211 @opindex Wformat-security
2212 If @option{-Wformat} is specified, also warn about uses of format
2213 functions that represent possible security problems. At present, this
2214 warns about calls to @code{printf} and @code{scanf} functions where the
2215 format string is not a string literal and there are no format arguments,
2216 as in @code{printf (foo);}. This may be a security hole if the format
2217 string came from untrusted input and contains @samp{%n}. (This is
2218 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2219 in future warnings may be added to @option{-Wformat-security} that are not
2220 included in @option{-Wformat-nonliteral}.)
2224 Enable @option{-Wformat} plus format checks not included in
2225 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2226 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2230 Warn about passing a null pointer for arguments marked as
2231 requiring a non-null value by the @code{nonnull} function attribute.
2233 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2234 can be disabled with the @option{-Wno-nonnull} option.
2236 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2238 Warn about uninitialized variables which are initialized with themselves.
2239 Note this option can only be used with the @option{-Wuninitialized} option,
2240 which in turn only works with @option{-O1} and above.
2242 For example, GCC will warn about @code{i} being uninitialized in the
2243 following snippet only when @option{-Winit-self} has been specified:
2254 @item -Wimplicit-int
2255 @opindex Wimplicit-int
2256 Warn when a declaration does not specify a type.
2258 @item -Wimplicit-function-declaration
2259 @itemx -Werror-implicit-function-declaration
2260 @opindex Wimplicit-function-declaration
2261 @opindex Werror-implicit-function-declaration
2262 Give a warning (or error) whenever a function is used before being
2267 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2271 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2272 function with external linkage, returning int, taking either zero
2273 arguments, two, or three arguments of appropriate types.
2275 @item -Wmissing-braces
2276 @opindex Wmissing-braces
2277 Warn if an aggregate or union initializer is not fully bracketed. In
2278 the following example, the initializer for @samp{a} is not fully
2279 bracketed, but that for @samp{b} is fully bracketed.
2282 int a[2][2] = @{ 0, 1, 2, 3 @};
2283 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2286 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2287 @opindex Wmissing-include-dirs
2288 Warn if a user-supplied include directory does not exist.
2291 @opindex Wparentheses
2292 Warn if parentheses are omitted in certain contexts, such
2293 as when there is an assignment in a context where a truth value
2294 is expected, or when operators are nested whose precedence people
2295 often get confused about. Only the warning for an assignment used as
2296 a truth value is supported when compiling C++; the other warnings are
2297 only supported when compiling C@.
2299 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2300 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2301 interpretation from that of ordinary mathematical notation.
2303 Also warn about constructions where there may be confusion to which
2304 @code{if} statement an @code{else} branch belongs. Here is an example of
2319 In C, every @code{else} branch belongs to the innermost possible @code{if}
2320 statement, which in this example is @code{if (b)}. This is often not
2321 what the programmer expected, as illustrated in the above example by
2322 indentation the programmer chose. When there is the potential for this
2323 confusion, GCC will issue a warning when this flag is specified.
2324 To eliminate the warning, add explicit braces around the innermost
2325 @code{if} statement so there is no way the @code{else} could belong to
2326 the enclosing @code{if}. The resulting code would look like this:
2342 @item -Wsequence-point
2343 @opindex Wsequence-point
2344 Warn about code that may have undefined semantics because of violations
2345 of sequence point rules in the C standard.
2347 The C standard defines the order in which expressions in a C program are
2348 evaluated in terms of @dfn{sequence points}, which represent a partial
2349 ordering between the execution of parts of the program: those executed
2350 before the sequence point, and those executed after it. These occur
2351 after the evaluation of a full expression (one which is not part of a
2352 larger expression), after the evaluation of the first operand of a
2353 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2354 function is called (but after the evaluation of its arguments and the
2355 expression denoting the called function), and in certain other places.
2356 Other than as expressed by the sequence point rules, the order of
2357 evaluation of subexpressions of an expression is not specified. All
2358 these rules describe only a partial order rather than a total order,
2359 since, for example, if two functions are called within one expression
2360 with no sequence point between them, the order in which the functions
2361 are called is not specified. However, the standards committee have
2362 ruled that function calls do not overlap.
2364 It is not specified when between sequence points modifications to the
2365 values of objects take effect. Programs whose behavior depends on this
2366 have undefined behavior; the C standard specifies that ``Between the
2367 previous and next sequence point an object shall have its stored value
2368 modified at most once by the evaluation of an expression. Furthermore,
2369 the prior value shall be read only to determine the value to be
2370 stored.''. If a program breaks these rules, the results on any
2371 particular implementation are entirely unpredictable.
2373 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2374 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2375 diagnosed by this option, and it may give an occasional false positive
2376 result, but in general it has been found fairly effective at detecting
2377 this sort of problem in programs.
2379 The present implementation of this option only works for C programs. A
2380 future implementation may also work for C++ programs.
2382 The C standard is worded confusingly, therefore there is some debate
2383 over the precise meaning of the sequence point rules in subtle cases.
2384 Links to discussions of the problem, including proposed formal
2385 definitions, may be found on the GCC readings page, at
2386 @w{@uref{http://gcc.gnu.org/readings.html}}.
2389 @opindex Wreturn-type
2390 Warn whenever a function is defined with a return-type that defaults to
2391 @code{int}. Also warn about any @code{return} statement with no
2392 return-value in a function whose return-type is not @code{void}.
2394 For C, also warn if the return type of a function has a type qualifier
2395 such as @code{const}. Such a type qualifier has no effect, since the
2396 value returned by a function is not an lvalue. ISO C prohibits
2397 qualified @code{void} return types on function definitions, so such
2398 return types always receive a warning even without this option.
2400 For C++, a function without return type always produces a diagnostic
2401 message, even when @option{-Wno-return-type} is specified. The only
2402 exceptions are @samp{main} and functions defined in system headers.
2406 Warn whenever a @code{switch} statement has an index of enumerated type
2407 and lacks a @code{case} for one or more of the named codes of that
2408 enumeration. (The presence of a @code{default} label prevents this
2409 warning.) @code{case} labels outside the enumeration range also
2410 provoke warnings when this option is used.
2412 @item -Wswitch-default
2413 @opindex Wswitch-switch
2414 Warn whenever a @code{switch} statement does not have a @code{default}
2418 @opindex Wswitch-enum
2419 Warn whenever a @code{switch} statement has an index of enumerated type
2420 and lacks a @code{case} for one or more of the named codes of that
2421 enumeration. @code{case} labels outside the enumeration range also
2422 provoke warnings when this option is used.
2426 Warn if any trigraphs are encountered that might change the meaning of
2427 the program (trigraphs within comments are not warned about).
2429 @item -Wunused-function
2430 @opindex Wunused-function
2431 Warn whenever a static function is declared but not defined or a
2432 non\-inline static function is unused.
2434 @item -Wunused-label
2435 @opindex Wunused-label
2436 Warn whenever a label is declared but not used.
2438 To suppress this warning use the @samp{unused} attribute
2439 (@pxref{Variable Attributes}).
2441 @item -Wunused-parameter
2442 @opindex Wunused-parameter
2443 Warn whenever a function parameter is unused aside from its declaration.
2445 To suppress this warning use the @samp{unused} attribute
2446 (@pxref{Variable Attributes}).
2448 @item -Wunused-variable
2449 @opindex Wunused-variable
2450 Warn whenever a local variable or non-constant static variable is unused
2451 aside from its declaration
2453 To suppress this warning use the @samp{unused} attribute
2454 (@pxref{Variable Attributes}).
2456 @item -Wunused-value
2457 @opindex Wunused-value
2458 Warn whenever a statement computes a result that is explicitly not used.
2460 To suppress this warning cast the expression to @samp{void}.
2464 All the above @option{-Wunused} options combined.
2466 In order to get a warning about an unused function parameter, you must
2467 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2468 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2470 @item -Wuninitialized
2471 @opindex Wuninitialized
2472 Warn if an automatic variable is used without first being initialized or
2473 if a variable may be clobbered by a @code{setjmp} call.
2475 These warnings are possible only in optimizing compilation,
2476 because they require data flow information that is computed only
2477 when optimizing. If you don't specify @option{-O}, you simply won't
2480 If you want to warn about code which uses the uninitialized value of the
2481 variable in its own initializer, use the @option{-Winit-self} option.
2483 These warnings occur only for variables that are candidates for
2484 register allocation. Therefore, they do not occur for a variable that
2485 is declared @code{volatile}, or whose address is taken, or whose size
2486 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2487 structures, unions or arrays, even when they are in registers.
2489 Note that there may be no warning about a variable that is used only
2490 to compute a value that itself is never used, because such
2491 computations may be deleted by data flow analysis before the warnings
2494 These warnings are made optional because GCC is not smart
2495 enough to see all the reasons why the code might be correct
2496 despite appearing to have an error. Here is one example of how
2517 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2518 always initialized, but GCC doesn't know this. Here is
2519 another common case:
2524 if (change_y) save_y = y, y = new_y;
2526 if (change_y) y = save_y;
2531 This has no bug because @code{save_y} is used only if it is set.
2533 @cindex @code{longjmp} warnings
2534 This option also warns when a non-volatile automatic variable might be
2535 changed by a call to @code{longjmp}. These warnings as well are possible
2536 only in optimizing compilation.
2538 The compiler sees only the calls to @code{setjmp}. It cannot know
2539 where @code{longjmp} will be called; in fact, a signal handler could
2540 call it at any point in the code. As a result, you may get a warning
2541 even when there is in fact no problem because @code{longjmp} cannot
2542 in fact be called at the place which would cause a problem.
2544 Some spurious warnings can be avoided if you declare all the functions
2545 you use that never return as @code{noreturn}. @xref{Function
2548 @item -Wunknown-pragmas
2549 @opindex Wunknown-pragmas
2550 @cindex warning for unknown pragmas
2551 @cindex unknown pragmas, warning
2552 @cindex pragmas, warning of unknown
2553 Warn when a #pragma directive is encountered which is not understood by
2554 GCC@. If this command line option is used, warnings will even be issued
2555 for unknown pragmas in system header files. This is not the case if
2556 the warnings were only enabled by the @option{-Wall} command line option.
2558 @item -Wstrict-aliasing
2559 @opindex Wstrict-aliasing
2560 This option is only active when @option{-fstrict-aliasing} is active.
2561 It warns about code which might break the strict aliasing rules that the
2562 compiler is using for optimization. The warning does not catch all
2563 cases, but does attempt to catch the more common pitfalls. It is
2564 included in @option{-Wall}.
2566 @item -Wstrict-aliasing=2
2567 @opindex Wstrict-aliasing=2
2568 This option is only active when @option{-fstrict-aliasing} is active.
2569 It warns about all code which might break the strict aliasing rules that the
2570 compiler is using for optimization. This warning catches all cases, but
2571 it will also give a warning for some ambiguous cases that are safe.
2575 All of the above @samp{-W} options combined. This enables all the
2576 warnings about constructions that some users consider questionable, and
2577 that are easy to avoid (or modify to prevent the warning), even in
2578 conjunction with macros. This also enables some language-specific
2579 warnings described in @ref{C++ Dialect Options} and
2580 @ref{Objective-C and Objective-C++ Dialect Options}.
2583 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2584 Some of them warn about constructions that users generally do not
2585 consider questionable, but which occasionally you might wish to check
2586 for; others warn about constructions that are necessary or hard to avoid
2587 in some cases, and there is no simple way to modify the code to suppress
2594 (This option used to be called @option{-W}. The older name is still
2595 supported, but the newer name is more descriptive.) Print extra warning
2596 messages for these events:
2600 A function can return either with or without a value. (Falling
2601 off the end of the function body is considered returning without
2602 a value.) For example, this function would evoke such a
2616 An expression-statement or the left-hand side of a comma expression
2617 contains no side effects.
2618 To suppress the warning, cast the unused expression to void.
2619 For example, an expression such as @samp{x[i,j]} will cause a warning,
2620 but @samp{x[(void)i,j]} will not.
2623 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2626 Storage-class specifiers like @code{static} are not the first things in
2627 a declaration. According to the C Standard, this usage is obsolescent.
2630 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2634 A comparison between signed and unsigned values could produce an
2635 incorrect result when the signed value is converted to unsigned.
2636 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2639 An aggregate has an initializer which does not initialize all members.
2640 This warning can be independently controlled by
2641 @option{-Wmissing-field-initializers}.
2644 A function parameter is declared without a type specifier in K&R-style
2652 An empty body occurs in an @samp{if} or @samp{else} statement.
2655 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2656 @samp{>}, or @samp{>=}.
2659 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2662 Any of several floating-point events that often indicate errors, such as
2663 overflow, underflow, loss of precision, etc.
2665 @item @r{(C++ only)}
2666 An enumerator and a non-enumerator both appear in a conditional expression.
2668 @item @r{(C++ only)}
2669 A non-static reference or non-static @samp{const} member appears in a
2670 class without constructors.
2672 @item @r{(C++ only)}
2673 Ambiguous virtual bases.
2675 @item @r{(C++ only)}
2676 Subscripting an array which has been declared @samp{register}.
2678 @item @r{(C++ only)}
2679 Taking the address of a variable which has been declared @samp{register}.
2681 @item @r{(C++ only)}
2682 A base class is not initialized in a derived class' copy constructor.
2685 @item -Wno-div-by-zero
2686 @opindex Wno-div-by-zero
2687 @opindex Wdiv-by-zero
2688 Do not warn about compile-time integer division by zero. Floating point
2689 division by zero is not warned about, as it can be a legitimate way of
2690 obtaining infinities and NaNs.
2692 @item -Wsystem-headers
2693 @opindex Wsystem-headers
2694 @cindex warnings from system headers
2695 @cindex system headers, warnings from
2696 Print warning messages for constructs found in system header files.
2697 Warnings from system headers are normally suppressed, on the assumption
2698 that they usually do not indicate real problems and would only make the
2699 compiler output harder to read. Using this command line option tells
2700 GCC to emit warnings from system headers as if they occurred in user
2701 code. However, note that using @option{-Wall} in conjunction with this
2702 option will @emph{not} warn about unknown pragmas in system
2703 headers---for that, @option{-Wunknown-pragmas} must also be used.
2706 @opindex Wfloat-equal
2707 Warn if floating point values are used in equality comparisons.
2709 The idea behind this is that sometimes it is convenient (for the
2710 programmer) to consider floating-point values as approximations to
2711 infinitely precise real numbers. If you are doing this, then you need
2712 to compute (by analyzing the code, or in some other way) the maximum or
2713 likely maximum error that the computation introduces, and allow for it
2714 when performing comparisons (and when producing output, but that's a
2715 different problem). In particular, instead of testing for equality, you
2716 would check to see whether the two values have ranges that overlap; and
2717 this is done with the relational operators, so equality comparisons are
2720 @item -Wtraditional @r{(C only)}
2721 @opindex Wtraditional
2722 Warn about certain constructs that behave differently in traditional and
2723 ISO C@. Also warn about ISO C constructs that have no traditional C
2724 equivalent, and/or problematic constructs which should be avoided.
2728 Macro parameters that appear within string literals in the macro body.
2729 In traditional C macro replacement takes place within string literals,
2730 but does not in ISO C@.
2733 In traditional C, some preprocessor directives did not exist.
2734 Traditional preprocessors would only consider a line to be a directive
2735 if the @samp{#} appeared in column 1 on the line. Therefore
2736 @option{-Wtraditional} warns about directives that traditional C
2737 understands but would ignore because the @samp{#} does not appear as the
2738 first character on the line. It also suggests you hide directives like
2739 @samp{#pragma} not understood by traditional C by indenting them. Some
2740 traditional implementations would not recognize @samp{#elif}, so it
2741 suggests avoiding it altogether.
2744 A function-like macro that appears without arguments.
2747 The unary plus operator.
2750 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2751 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2752 constants.) Note, these suffixes appear in macros defined in the system
2753 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2754 Use of these macros in user code might normally lead to spurious
2755 warnings, however GCC's integrated preprocessor has enough context to
2756 avoid warning in these cases.
2759 A function declared external in one block and then used after the end of
2763 A @code{switch} statement has an operand of type @code{long}.
2766 A non-@code{static} function declaration follows a @code{static} one.
2767 This construct is not accepted by some traditional C compilers.
2770 The ISO type of an integer constant has a different width or
2771 signedness from its traditional type. This warning is only issued if
2772 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2773 typically represent bit patterns, are not warned about.
2776 Usage of ISO string concatenation is detected.
2779 Initialization of automatic aggregates.
2782 Identifier conflicts with labels. Traditional C lacks a separate
2783 namespace for labels.
2786 Initialization of unions. If the initializer is zero, the warning is
2787 omitted. This is done under the assumption that the zero initializer in
2788 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2789 initializer warnings and relies on default initialization to zero in the
2793 Conversions by prototypes between fixed/floating point values and vice
2794 versa. The absence of these prototypes when compiling with traditional
2795 C would cause serious problems. This is a subset of the possible
2796 conversion warnings, for the full set use @option{-Wconversion}.
2799 Use of ISO C style function definitions. This warning intentionally is
2800 @emph{not} issued for prototype declarations or variadic functions
2801 because these ISO C features will appear in your code when using
2802 libiberty's traditional C compatibility macros, @code{PARAMS} and
2803 @code{VPARAMS}. This warning is also bypassed for nested functions
2804 because that feature is already a GCC extension and thus not relevant to
2805 traditional C compatibility.
2808 @item -Wdeclaration-after-statement @r{(C only)}
2809 @opindex Wdeclaration-after-statement
2810 Warn when a declaration is found after a statement in a block. This
2811 construct, known from C++, was introduced with ISO C99 and is by default
2812 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2813 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2817 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2819 @item -Wendif-labels
2820 @opindex Wendif-labels
2821 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2825 Warn whenever a local variable shadows another local variable, parameter or
2826 global variable or whenever a built-in function is shadowed.
2828 @item -Wlarger-than-@var{len}
2829 @opindex Wlarger-than
2830 Warn whenever an object of larger than @var{len} bytes is defined.
2832 @item -Wpointer-arith
2833 @opindex Wpointer-arith
2834 Warn about anything that depends on the ``size of'' a function type or
2835 of @code{void}. GNU C assigns these types a size of 1, for
2836 convenience in calculations with @code{void *} pointers and pointers
2839 @item -Wbad-function-cast @r{(C only)}
2840 @opindex Wbad-function-cast
2841 Warn whenever a function call is cast to a non-matching type.
2842 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2846 Warn whenever a pointer is cast so as to remove a type qualifier from
2847 the target type. For example, warn if a @code{const char *} is cast
2848 to an ordinary @code{char *}.
2851 @opindex Wcast-align
2852 Warn whenever a pointer is cast such that the required alignment of the
2853 target is increased. For example, warn if a @code{char *} is cast to
2854 an @code{int *} on machines where integers can only be accessed at
2855 two- or four-byte boundaries.
2857 @item -Wwrite-strings
2858 @opindex Wwrite-strings
2859 When compiling C, give string constants the type @code{const
2860 char[@var{length}]} so that
2861 copying the address of one into a non-@code{const} @code{char *}
2862 pointer will get a warning; when compiling C++, warn about the
2863 deprecated conversion from string constants to @code{char *}.
2864 These warnings will help you find at
2865 compile time code that can try to write into a string constant, but
2866 only if you have been very careful about using @code{const} in
2867 declarations and prototypes. Otherwise, it will just be a nuisance;
2868 this is why we did not make @option{-Wall} request these warnings.
2871 @opindex Wconversion
2872 Warn if a prototype causes a type conversion that is different from what
2873 would happen to the same argument in the absence of a prototype. This
2874 includes conversions of fixed point to floating and vice versa, and
2875 conversions changing the width or signedness of a fixed point argument
2876 except when the same as the default promotion.
2878 Also, warn if a negative integer constant expression is implicitly
2879 converted to an unsigned type. For example, warn about the assignment
2880 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2881 casts like @code{(unsigned) -1}.
2883 @item -Wsign-compare
2884 @opindex Wsign-compare
2885 @cindex warning for comparison of signed and unsigned values
2886 @cindex comparison of signed and unsigned values, warning
2887 @cindex signed and unsigned values, comparison warning
2888 Warn when a comparison between signed and unsigned values could produce
2889 an incorrect result when the signed value is converted to unsigned.
2890 This warning is also enabled by @option{-Wextra}; to get the other warnings
2891 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2893 @item -Waggregate-return
2894 @opindex Waggregate-return
2895 Warn if any functions that return structures or unions are defined or
2896 called. (In languages where you can return an array, this also elicits
2899 @item -Wstrict-prototypes @r{(C only)}
2900 @opindex Wstrict-prototypes
2901 Warn if a function is declared or defined without specifying the
2902 argument types. (An old-style function definition is permitted without
2903 a warning if preceded by a declaration which specifies the argument
2906 @item -Wold-style-definition @r{(C only)}
2907 @opindex Wold-style-definition
2908 Warn if an old-style function definition is used. A warning is given
2909 even if there is a previous prototype.
2911 @item -Wmissing-prototypes @r{(C only)}
2912 @opindex Wmissing-prototypes
2913 Warn if a global function is defined without a previous prototype
2914 declaration. This warning is issued even if the definition itself
2915 provides a prototype. The aim is to detect global functions that fail
2916 to be declared in header files.
2918 @item -Wmissing-declarations @r{(C only)}
2919 @opindex Wmissing-declarations
2920 Warn if a global function is defined without a previous declaration.
2921 Do so even if the definition itself provides a prototype.
2922 Use this option to detect global functions that are not declared in
2925 @item -Wmissing-field-initializers
2926 @opindex Wmissing-field-initializers
2929 Warn if a structure's initializer has some fields missing. For
2930 example, the following code would cause such a warning, because
2931 @code{x.h} is implicitly zero:
2934 struct s @{ int f, g, h; @};
2935 struct s x = @{ 3, 4 @};
2938 This option does not warn about designated initializers, so the following
2939 modification would not trigger a warning:
2942 struct s @{ int f, g, h; @};
2943 struct s x = @{ .f = 3, .g = 4 @};
2946 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
2947 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
2949 @item -Wmissing-noreturn
2950 @opindex Wmissing-noreturn
2951 Warn about functions which might be candidates for attribute @code{noreturn}.
2952 Note these are only possible candidates, not absolute ones. Care should
2953 be taken to manually verify functions actually do not ever return before
2954 adding the @code{noreturn} attribute, otherwise subtle code generation
2955 bugs could be introduced. You will not get a warning for @code{main} in
2956 hosted C environments.
2958 @item -Wmissing-format-attribute
2959 @opindex Wmissing-format-attribute
2961 If @option{-Wformat} is enabled, also warn about functions which might be
2962 candidates for @code{format} attributes. Note these are only possible
2963 candidates, not absolute ones. GCC will guess that @code{format}
2964 attributes might be appropriate for any function that calls a function
2965 like @code{vprintf} or @code{vscanf}, but this might not always be the
2966 case, and some functions for which @code{format} attributes are
2967 appropriate may not be detected. This option has no effect unless
2968 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2970 @item -Wno-multichar
2971 @opindex Wno-multichar
2973 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2974 Usually they indicate a typo in the user's code, as they have
2975 implementation-defined values, and should not be used in portable code.
2977 @item -Wno-deprecated-declarations
2978 @opindex Wno-deprecated-declarations
2979 Do not warn about uses of functions, variables, and types marked as
2980 deprecated by using the @code{deprecated} attribute.
2981 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2982 @pxref{Type Attributes}.)
2986 Warn if a structure is given the packed attribute, but the packed
2987 attribute has no effect on the layout or size of the structure.
2988 Such structures may be mis-aligned for little benefit. For
2989 instance, in this code, the variable @code{f.x} in @code{struct bar}
2990 will be misaligned even though @code{struct bar} does not itself
2991 have the packed attribute:
2998 @} __attribute__((packed));
3008 Warn if padding is included in a structure, either to align an element
3009 of the structure or to align the whole structure. Sometimes when this
3010 happens it is possible to rearrange the fields of the structure to
3011 reduce the padding and so make the structure smaller.
3013 @item -Wredundant-decls
3014 @opindex Wredundant-decls
3015 Warn if anything is declared more than once in the same scope, even in
3016 cases where multiple declaration is valid and changes nothing.
3018 @item -Wnested-externs @r{(C only)}
3019 @opindex Wnested-externs
3020 Warn if an @code{extern} declaration is encountered within a function.
3022 @item -Wunreachable-code
3023 @opindex Wunreachable-code
3024 Warn if the compiler detects that code will never be executed.
3026 This option is intended to warn when the compiler detects that at
3027 least a whole line of source code will never be executed, because
3028 some condition is never satisfied or because it is after a
3029 procedure that never returns.
3031 It is possible for this option to produce a warning even though there
3032 are circumstances under which part of the affected line can be executed,
3033 so care should be taken when removing apparently-unreachable code.
3035 For instance, when a function is inlined, a warning may mean that the
3036 line is unreachable in only one inlined copy of the function.
3038 This option is not made part of @option{-Wall} because in a debugging
3039 version of a program there is often substantial code which checks
3040 correct functioning of the program and is, hopefully, unreachable
3041 because the program does work. Another common use of unreachable
3042 code is to provide behavior which is selectable at compile-time.
3046 Warn if a function can not be inlined and it was declared as inline.
3047 Even with this option, the compiler will not warn about failures to
3048 inline functions declared in system headers.
3050 The compiler uses a variety of heuristics to determine whether or not
3051 to inline a function. For example, the compiler takes into account
3052 the size of the function being inlined and the the amount of inlining
3053 that has already been done in the current function. Therefore,
3054 seemingly insignificant changes in the source program can cause the
3055 warnings produced by @option{-Winline} to appear or disappear.
3057 @item -Wno-invalid-offsetof @r{(C++ only)}
3058 @opindex Wno-invalid-offsetof
3059 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3060 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3061 to a non-POD type is undefined. In existing C++ implementations,
3062 however, @samp{offsetof} typically gives meaningful results even when
3063 applied to certain kinds of non-POD types. (Such as a simple
3064 @samp{struct} that fails to be a POD type only by virtue of having a
3065 constructor.) This flag is for users who are aware that they are
3066 writing nonportable code and who have deliberately chosen to ignore the
3069 The restrictions on @samp{offsetof} may be relaxed in a future version
3070 of the C++ standard.
3073 @opindex Winvalid-pch
3074 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3075 the search path but can't be used.
3079 @opindex Wno-long-long
3080 Warn if @samp{long long} type is used. This is default. To inhibit
3081 the warning messages, use @option{-Wno-long-long}. Flags
3082 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3083 only when @option{-pedantic} flag is used.
3085 @item -Wvariadic-macros
3086 @opindex Wvariadic-macros
3087 @opindex Wno-variadic-macros
3088 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3089 alternate syntax when in pedantic ISO C99 mode. This is default.
3090 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3092 @item -Wdisabled-optimization
3093 @opindex Wdisabled-optimization
3094 Warn if a requested optimization pass is disabled. This warning does
3095 not generally indicate that there is anything wrong with your code; it
3096 merely indicates that GCC's optimizers were unable to handle the code
3097 effectively. Often, the problem is that your code is too big or too
3098 complex; GCC will refuse to optimize programs when the optimization
3099 itself is likely to take inordinate amounts of time.
3103 Make all warnings into errors.
3106 @node Debugging Options
3107 @section Options for Debugging Your Program or GCC
3108 @cindex options, debugging
3109 @cindex debugging information options
3111 GCC has various special options that are used for debugging
3112 either your program or GCC:
3117 Produce debugging information in the operating system's native format
3118 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3121 On most systems that use stabs format, @option{-g} enables use of extra
3122 debugging information that only GDB can use; this extra information
3123 makes debugging work better in GDB but will probably make other debuggers
3125 refuse to read the program. If you want to control for certain whether
3126 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3127 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3129 GCC allows you to use @option{-g} with
3130 @option{-O}. The shortcuts taken by optimized code may occasionally
3131 produce surprising results: some variables you declared may not exist
3132 at all; flow of control may briefly move where you did not expect it;
3133 some statements may not be executed because they compute constant
3134 results or their values were already at hand; some statements may
3135 execute in different places because they were moved out of loops.
3137 Nevertheless it proves possible to debug optimized output. This makes
3138 it reasonable to use the optimizer for programs that might have bugs.
3140 The following options are useful when GCC is generated with the
3141 capability for more than one debugging format.
3145 Produce debugging information for use by GDB@. This means to use the
3146 most expressive format available (DWARF 2, stabs, or the native format
3147 if neither of those are supported), including GDB extensions if at all
3152 Produce debugging information in stabs format (if that is supported),
3153 without GDB extensions. This is the format used by DBX on most BSD
3154 systems. On MIPS, Alpha and System V Release 4 systems this option
3155 produces stabs debugging output which is not understood by DBX or SDB@.
3156 On System V Release 4 systems this option requires the GNU assembler.
3158 @item -feliminate-unused-debug-symbols
3159 @opindex feliminate-unused-debug-symbols
3160 Produce debugging information in stabs format (if that is supported),
3161 for only symbols that are actually used.
3165 Produce debugging information in stabs format (if that is supported),
3166 using GNU extensions understood only by the GNU debugger (GDB)@. The
3167 use of these extensions is likely to make other debuggers crash or
3168 refuse to read the program.
3172 Produce debugging information in COFF format (if that is supported).
3173 This is the format used by SDB on most System V systems prior to
3178 Produce debugging information in XCOFF format (if that is supported).
3179 This is the format used by the DBX debugger on IBM RS/6000 systems.
3183 Produce debugging information in XCOFF format (if that is supported),
3184 using GNU extensions understood only by the GNU debugger (GDB)@. The
3185 use of these extensions is likely to make other debuggers crash or
3186 refuse to read the program, and may cause assemblers other than the GNU
3187 assembler (GAS) to fail with an error.
3191 Produce debugging information in DWARF version 2 format (if that is
3192 supported). This is the format used by DBX on IRIX 6. With this
3193 option, GCC uses features of DWARF version 3 when they are useful;
3194 version 3 is upward compatible with version 2, but may still cause
3195 problems for older debuggers.
3199 Produce debugging information in VMS debug format (if that is
3200 supported). This is the format used by DEBUG on VMS systems.
3203 @itemx -ggdb@var{level}
3204 @itemx -gstabs@var{level}
3205 @itemx -gcoff@var{level}
3206 @itemx -gxcoff@var{level}
3207 @itemx -gvms@var{level}
3208 Request debugging information and also use @var{level} to specify how
3209 much information. The default level is 2.
3211 Level 1 produces minimal information, enough for making backtraces in
3212 parts of the program that you don't plan to debug. This includes
3213 descriptions of functions and external variables, but no information
3214 about local variables and no line numbers.
3216 Level 3 includes extra information, such as all the macro definitions
3217 present in the program. Some debuggers support macro expansion when
3218 you use @option{-g3}.
3220 @option{-gdwarf-2} does not accept a concatenated debug level, because
3221 GCC used to support an option @option{-gdwarf} that meant to generate
3222 debug information in version 1 of the DWARF format (which is very
3223 different from version 2), and it would have been too confusing. That
3224 debug format is long obsolete, but the option cannot be changed now.
3225 Instead use an additional @option{-g@var{level}} option to change the
3226 debug level for DWARF2.
3228 @item -feliminate-dwarf2-dups
3229 @opindex feliminate-dwarf2-dups
3230 Compress DWARF2 debugging information by eliminating duplicated
3231 information about each symbol. This option only makes sense when
3232 generating DWARF2 debugging information with @option{-gdwarf-2}.
3234 @cindex @command{prof}
3237 Generate extra code to write profile information suitable for the
3238 analysis program @command{prof}. You must use this option when compiling
3239 the source files you want data about, and you must also use it when
3242 @cindex @command{gprof}
3245 Generate extra code to write profile information suitable for the
3246 analysis program @command{gprof}. You must use this option when compiling
3247 the source files you want data about, and you must also use it when
3252 Makes the compiler print out each function name as it is compiled, and
3253 print some statistics about each pass when it finishes.
3256 @opindex ftime-report
3257 Makes the compiler print some statistics about the time consumed by each
3258 pass when it finishes.
3261 @opindex fmem-report
3262 Makes the compiler print some statistics about permanent memory
3263 allocation when it finishes.
3265 @item -fprofile-arcs
3266 @opindex fprofile-arcs
3267 Add code so that program flow @dfn{arcs} are instrumented. During
3268 execution the program records how many times each branch and call is
3269 executed and how many times it is taken or returns. When the compiled
3270 program exits it saves this data to a file called
3271 @file{@var{auxname}.gcda} for each source file. The data may be used for
3272 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3273 test coverage analysis (@option{-ftest-coverage}). Each object file's
3274 @var{auxname} is generated from the name of the output file, if
3275 explicitly specified and it is not the final executable, otherwise it is
3276 the basename of the source file. In both cases any suffix is removed
3277 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3278 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3283 Compile the source files with @option{-fprofile-arcs} plus optimization
3284 and code generation options. For test coverage analysis, use the
3285 additional @option{-ftest-coverage} option. You do not need to profile
3286 every source file in a program.
3289 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3290 (the latter implies the former).
3293 Run the program on a representative workload to generate the arc profile
3294 information. This may be repeated any number of times. You can run
3295 concurrent instances of your program, and provided that the file system
3296 supports locking, the data files will be correctly updated. Also
3297 @code{fork} calls are detected and correctly handled (double counting
3301 For profile-directed optimizations, compile the source files again with
3302 the same optimization and code generation options plus
3303 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3304 Control Optimization}).
3307 For test coverage analysis, use @command{gcov} to produce human readable
3308 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3309 @command{gcov} documentation for further information.
3313 With @option{-fprofile-arcs}, for each function of your program GCC
3314 creates a program flow graph, then finds a spanning tree for the graph.
3315 Only arcs that are not on the spanning tree have to be instrumented: the
3316 compiler adds code to count the number of times that these arcs are
3317 executed. When an arc is the only exit or only entrance to a block, the
3318 instrumentation code can be added to the block; otherwise, a new basic
3319 block must be created to hold the instrumentation code.
3321 @item -ftree-based-profiling
3322 @opindex ftree-based-profiling
3323 This option is used in addition to @option{-fprofile-arcs} or
3324 @option{-fbranch-probabilities} to control whether those optimizations
3325 are performed on a tree-based or rtl-based internal representation.
3326 If you use this option when compiling with @option{-fprofile-arcs},
3327 you must also use it when compiling later with @option{-fbranch-probabilities}.
3328 Currently the tree-based optimization is in an early stage of
3329 development, and this option is recommended only for those people
3330 working on improving it.
3333 @item -ftest-coverage
3334 @opindex ftest-coverage
3335 Produce a notes file that the @command{gcov} code-coverage utility
3336 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3337 show program coverage. Each source file's note file is called
3338 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3339 above for a description of @var{auxname} and instructions on how to
3340 generate test coverage data. Coverage data will match the source files
3341 more closely, if you do not optimize.
3343 @item -d@var{letters}
3345 Says to make debugging dumps during compilation at times specified by
3346 @var{letters}. This is used for debugging the compiler. The file names
3347 for most of the dumps are made by appending a pass number and a word to
3348 the @var{dumpname}. @var{dumpname} is generated from the name of the
3349 output file, if explicitly specified and it is not an executable,
3350 otherwise it is the basename of the source file. In both cases any
3351 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3352 Here are the possible letters for use in @var{letters}, and their
3358 Annotate the assembler output with miscellaneous debugging information.
3361 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3364 Dump after block reordering, to @file{@var{file}.32.bbro}.
3367 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3370 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3371 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3374 Dump after branch target load optimization, to to @file{@var{file}.33.btl}.
3375 Also dump after delayed branch scheduling, to @file{@var{file}.37.dbr}.
3378 Dump all macro definitions, at the end of preprocessing, in addition to
3382 Dump after the third if conversion, to @file{@var{file}.31.ce3}.
3385 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3386 Also dump after life analysis, to @file{@var{file}.19.life}.
3389 Dump after global register allocation, to @file{@var{file}.26.greg}.
3392 Dump after GCSE, to @file{@var{file}.08.gcse}.
3393 Also dump after jump bypassing and control flow optimizations, to
3394 @file{@var{file}.10.bypass}.
3397 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3400 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3403 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3406 Dump after conversion from registers to stack, to @file{@var{file}.35.stack}.
3409 Dump after local register allocation, to @file{@var{file}.25.lreg}.
3412 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3413 @file{@var{file}.16.loop2}.
3416 Dump after modulo scheduling, to @file{@var{file}.23.sms}.
3419 Dump after performing the machine dependent reorganization pass, to
3420 @file{@var{file}.36.mach}.
3423 Dump after register renumbering, to @file{@var{file}.30.rnreg}.
3426 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3429 Dump after post-reload optimizations, to @file{@var{file}.27.postreload}.
3432 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3435 Dump after the second scheduling pass, to @file{@var{file}.34.sched2}.
3438 Dump after CSE (including the jump optimization that sometimes follows
3439 CSE), to @file{@var{file}.06.cse}.
3442 Dump after the first scheduling pass, to @file{@var{file}.24.sched}.
3445 Dump after the second CSE pass (including the jump optimization that
3446 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3449 Dump after running tracer, to @file{@var{file}.15.tracer}.
3452 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3455 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3458 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3459 Also dump after variable tracking, to @file{@var{file}.35.vartrack}.
3462 Dump after the second flow pass, to @file{@var{file}.28.flow2}.
3465 Dump after the peephole pass, to @file{@var{file}.29.peephole2}.
3468 Dump after constructing the web, to @file{@var{file}.17.web}.
3471 Produce all the dumps listed above.
3474 Produce a core dump whenever an error occurs.
3477 Print statistics on memory usage, at the end of the run, to
3481 Annotate the assembler output with a comment indicating which
3482 pattern and alternative was used. The length of each instruction is
3486 Dump the RTL in the assembler output as a comment before each instruction.
3487 Also turns on @option{-dp} annotation.
3490 For each of the other indicated dump files (except for
3491 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3492 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3495 Just generate RTL for a function instead of compiling it. Usually used
3499 Dump debugging information during parsing, to standard error.
3502 @item -fdump-unnumbered
3503 @opindex fdump-unnumbered
3504 When doing debugging dumps (see @option{-d} option above), suppress instruction
3505 numbers and line number note output. This makes it more feasible to
3506 use diff on debugging dumps for compiler invocations with different
3507 options, in particular with and without @option{-g}.
3509 @item -fdump-translation-unit @r{(C and C++ only)}
3510 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3511 @opindex fdump-translation-unit
3512 Dump a representation of the tree structure for the entire translation
3513 unit to a file. The file name is made by appending @file{.tu} to the
3514 source file name. If the @samp{-@var{options}} form is used, @var{options}
3515 controls the details of the dump as described for the
3516 @option{-fdump-tree} options.
3518 @item -fdump-class-hierarchy @r{(C++ only)}
3519 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3520 @opindex fdump-class-hierarchy
3521 Dump a representation of each class's hierarchy and virtual function
3522 table layout to a file. The file name is made by appending @file{.class}
3523 to the source file name. If the @samp{-@var{options}} form is used,
3524 @var{options} controls the details of the dump as described for the
3525 @option{-fdump-tree} options.
3527 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3528 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3530 Control the dumping at various stages of processing the intermediate
3531 language tree to a file. The file name is generated by appending a switch
3532 specific suffix to the source file name. If the @samp{-@var{options}}
3533 form is used, @var{options} is a list of @samp{-} separated options that
3534 control the details of the dump. Not all options are applicable to all
3535 dumps, those which are not meaningful will be ignored. The following
3536 options are available
3540 Print the address of each node. Usually this is not meaningful as it
3541 changes according to the environment and source file. Its primary use
3542 is for tying up a dump file with a debug environment.
3544 Inhibit dumping of members of a scope or body of a function merely
3545 because that scope has been reached. Only dump such items when they
3546 are directly reachable by some other path. When dumping pretty-printed
3547 trees, this option inhibits dumping the bodies of control structures.
3549 Print a raw representation of the tree. By default, trees are
3550 pretty-printed into a C-like representation.
3552 Enable more detailed dumps (not honored by every dump option).
3554 Enable dumping various statistics about the pass (not honored by every dump
3557 Enable showing basic block boundaries (disabled in raw dumps).
3559 Enable showing virtual operands for every statement.
3561 Enable showing line numbers for statements.
3563 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3565 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3568 The following tree dumps are possible:
3572 Dump before any tree based optimization, to @file{@var{file}.original}.
3575 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3578 Dump after function inlining, to @file{@var{file}.inlined}.
3581 @opindex fdump-tree-gimple
3582 Dump each function before and after the gimplification pass to a file. The
3583 file name is made by appending @file{.gimple} to the source file name.
3586 @opindex fdump-tree-cfg
3587 Dump the control flow graph of each function to a file. The file name is
3588 made by appending @file{.cfg} to the source file name.
3591 @opindex fdump-tree-vcg
3592 Dump the control flow graph of each function to a file in VCG format. The
3593 file name is made by appending @file{.vcg} to the source file name. Note
3594 that if the file contains more than one function, the generated file cannot
3595 be used directly by VCG. You will need to cut and paste each function's
3596 graph into its own separate file first.
3599 @opindex fdump-tree-ch
3600 Dump each function after copying loop headers. The file name is made by
3601 appending @file{.ch} to the source file name.
3604 @opindex fdump-tree-ssa
3605 Dump SSA related information to a file. The file name is made by appending
3606 @file{.ssa} to the source file name.
3609 @opindex fdump-tree-alias
3610 Dump aliasing information for each function. The file name is made by
3611 appending @file{.alias} to the source file name.
3614 @opindex fdump-tree-ccp
3615 Dump each function after CCP. The file name is made by appending
3616 @file{.ccp} to the source file name.
3619 @opindex fdump-tree-pre
3620 Dump trees after partial redundancy elimination. The file name is made
3621 by appending @file{.pre} to the source file name.
3624 @opindex fdump-tree-fre
3625 Dump trees after full redundancy elimination. The file name is made
3626 by appending @file{.fre} to the source file name.
3629 @opindex fdump-tree-dce
3630 Dump each function after dead code elimination. The file name is made by
3631 appending @file{.dce} to the source file name.
3634 @opindex fdump-tree-mudflap
3635 Dump each function after adding mudflap instrumentation. The file name is
3636 made by appending @file{.mudflap} to the source file name.
3639 @opindex fdump-tree-sra
3640 Dump each function after performing scalar replacement of aggregates. The
3641 file name is made by appending @file{.sra} to the source file name.
3644 @opindex fdump-tree-dom
3645 Dump each function after applying dominator tree optimizations. The file
3646 name is made by appending @file{.dom} to the source file name.
3649 @opindex fdump-tree-dse
3650 Dump each function after applying dead store elimination. The file
3651 name is made by appending @file{.dse} to the source file name.
3654 @opindex fdump-tree-phiopt
3655 Dump each function after optimizing PHI nodes into straightline code. The file
3656 name is made by appending @file{.phiopt} to the source file name.
3659 @opindex fdump-tree-forwprop
3660 Dump each function after forward propagating single use variables. The file
3661 name is made by appending @file{.forwprop} to the source file name.
3664 @opindex fdump-tree-copyrename
3665 Dump each function after applying the copy rename optimization. The file
3666 name is made by appending @file{.copyrename} to the source file name.
3669 @opindex fdump-tree-nrv
3670 Dump each function after applying the named return value optimization on
3671 generic trees. The file name is made by appending @file{.nrv} to the source
3675 @opindex fdump-tree-vect
3676 Dump each function after applying vectorization of loops. The file name is
3677 made by appending @file{.vect} to the source file name.
3680 @opindex fdump-tree-all
3681 Enable all the available tree dumps with the flags provided in this option.
3684 @item -frandom-seed=@var{string}
3685 @opindex frandom-string
3686 This option provides a seed that GCC uses when it would otherwise use
3687 random numbers. It is used to generate certain symbol names
3688 that have to be different in every compiled file. It is also used to
3689 place unique stamps in coverage data files and the object files that
3690 produce them. You can use the @option{-frandom-seed} option to produce
3691 reproducibly identical object files.
3693 The @var{string} should be different for every file you compile.
3695 @item -fsched-verbose=@var{n}
3696 @opindex fsched-verbose
3697 On targets that use instruction scheduling, this option controls the
3698 amount of debugging output the scheduler prints. This information is
3699 written to standard error, unless @option{-dS} or @option{-dR} is
3700 specified, in which case it is output to the usual dump
3701 listing file, @file{.sched} or @file{.sched2} respectively. However
3702 for @var{n} greater than nine, the output is always printed to standard
3705 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3706 same information as @option{-dRS}. For @var{n} greater than one, it
3707 also output basic block probabilities, detailed ready list information
3708 and unit/insn info. For @var{n} greater than two, it includes RTL
3709 at abort point, control-flow and regions info. And for @var{n} over
3710 four, @option{-fsched-verbose} also includes dependence info.
3714 Store the usual ``temporary'' intermediate files permanently; place them
3715 in the current directory and name them based on the source file. Thus,
3716 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3717 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3718 preprocessed @file{foo.i} output file even though the compiler now
3719 normally uses an integrated preprocessor.
3723 Report the CPU time taken by each subprocess in the compilation
3724 sequence. For C source files, this is the compiler proper and assembler
3725 (plus the linker if linking is done). The output looks like this:
3732 The first number on each line is the ``user time,'' that is time spent
3733 executing the program itself. The second number is ``system time,''
3734 time spent executing operating system routines on behalf of the program.
3735 Both numbers are in seconds.
3737 @item -fvar-tracking
3738 @opindex fvar-tracking
3739 Run variable tracking pass. It computes where variables are stored at each
3740 position in code. Better debugging information is then generated
3741 (if the debugging information format supports this information).
3743 It is enabled by default when compiling with optimization (@option{-Os},
3744 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3745 the debug info format supports it.
3747 @item -print-file-name=@var{library}
3748 @opindex print-file-name
3749 Print the full absolute name of the library file @var{library} that
3750 would be used when linking---and don't do anything else. With this
3751 option, GCC does not compile or link anything; it just prints the
3754 @item -print-multi-directory
3755 @opindex print-multi-directory
3756 Print the directory name corresponding to the multilib selected by any
3757 other switches present in the command line. This directory is supposed
3758 to exist in @env{GCC_EXEC_PREFIX}.
3760 @item -print-multi-lib
3761 @opindex print-multi-lib
3762 Print the mapping from multilib directory names to compiler switches
3763 that enable them. The directory name is separated from the switches by
3764 @samp{;}, and each switch starts with an @samp{@@} instead of the
3765 @samp{-}, without spaces between multiple switches. This is supposed to
3766 ease shell-processing.
3768 @item -print-prog-name=@var{program}
3769 @opindex print-prog-name
3770 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3772 @item -print-libgcc-file-name
3773 @opindex print-libgcc-file-name
3774 Same as @option{-print-file-name=libgcc.a}.
3776 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3777 but you do want to link with @file{libgcc.a}. You can do
3780 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3783 @item -print-search-dirs
3784 @opindex print-search-dirs
3785 Print the name of the configured installation directory and a list of
3786 program and library directories @command{gcc} will search---and don't do anything else.
3788 This is useful when @command{gcc} prints the error message
3789 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3790 To resolve this you either need to put @file{cpp0} and the other compiler
3791 components where @command{gcc} expects to find them, or you can set the environment
3792 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3793 Don't forget the trailing '/'.
3794 @xref{Environment Variables}.
3797 @opindex dumpmachine
3798 Print the compiler's target machine (for example,
3799 @samp{i686-pc-linux-gnu})---and don't do anything else.
3802 @opindex dumpversion
3803 Print the compiler version (for example, @samp{3.0})---and don't do
3808 Print the compiler's built-in specs---and don't do anything else. (This
3809 is used when GCC itself is being built.) @xref{Spec Files}.
3811 @item -feliminate-unused-debug-types
3812 @opindex feliminate-unused-debug-types
3813 Normally, when producing DWARF2 output, GCC will emit debugging
3814 information for all types declared in a compilation
3815 unit, regardless of whether or not they are actually used
3816 in that compilation unit. Sometimes this is useful, such as
3817 if, in the debugger, you want to cast a value to a type that is
3818 not actually used in your program (but is declared). More often,
3819 however, this results in a significant amount of wasted space.
3820 With this option, GCC will avoid producing debug symbol output
3821 for types that are nowhere used in the source file being compiled.
3824 @node Optimize Options
3825 @section Options That Control Optimization
3826 @cindex optimize options
3827 @cindex options, optimization
3829 These options control various sorts of optimizations.
3831 Without any optimization option, the compiler's goal is to reduce the
3832 cost of compilation and to make debugging produce the expected
3833 results. Statements are independent: if you stop the program with a
3834 breakpoint between statements, you can then assign a new value to any
3835 variable or change the program counter to any other statement in the
3836 function and get exactly the results you would expect from the source
3839 Turning on optimization flags makes the compiler attempt to improve
3840 the performance and/or code size at the expense of compilation time
3841 and possibly the ability to debug the program.
3843 The compiler performs optimization based on the knowledge it has of
3844 the program. Optimization levels @option{-O2} and above, in
3845 particular, enable @emph{unit-at-a-time} mode, which allows the
3846 compiler to consider information gained from later functions in
3847 the file when compiling a function. Compiling multiple files at
3848 once to a single output file in @emph{unit-at-a-time} mode allows
3849 the compiler to use information gained from all of the files when
3850 compiling each of them.
3852 Not all optimizations are controlled directly by a flag. Only
3853 optimizations that have a flag are listed.
3860 Optimize. Optimizing compilation takes somewhat more time, and a lot
3861 more memory for a large function.
3863 With @option{-O}, the compiler tries to reduce code size and execution
3864 time, without performing any optimizations that take a great deal of
3867 @option{-O} turns on the following optimization flags:
3868 @gccoptlist{-fdefer-pop @gol
3869 -fmerge-constants @gol
3871 -floop-optimize @gol
3872 -fif-conversion @gol
3873 -fif-conversion2 @gol
3874 -fdelayed-branch @gol
3875 -fguess-branch-probability @gol
3878 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3879 where doing so does not interfere with debugging.
3883 Optimize even more. GCC performs nearly all supported optimizations
3884 that do not involve a space-speed tradeoff. The compiler does not
3885 perform loop unrolling or function inlining when you specify @option{-O2}.
3886 As compared to @option{-O}, this option increases both compilation time
3887 and the performance of the generated code.
3889 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3890 also turns on the following optimization flags:
3891 @gccoptlist{-fforce-mem @gol
3892 -foptimize-sibling-calls @gol
3893 -fstrength-reduce @gol
3894 -fcse-follow-jumps -fcse-skip-blocks @gol
3895 -frerun-cse-after-loop -frerun-loop-opt @gol
3896 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3897 -fdelete-null-pointer-checks @gol
3898 -fexpensive-optimizations @gol
3900 -fschedule-insns -fschedule-insns2 @gol
3901 -fsched-interblock -fsched-spec @gol
3904 -freorder-blocks -freorder-functions @gol
3905 -fstrict-aliasing @gol
3906 -funit-at-a-time @gol
3907 -falign-functions -falign-jumps @gol
3908 -falign-loops -falign-labels @gol
3911 Please note the warning under @option{-fgcse} about
3912 invoking @option{-O2} on programs that use computed gotos.
3916 Optimize yet more. @option{-O3} turns on all optimizations specified by
3917 @option{-O2} and also turns on the @option{-finline-functions},
3918 @option{-fweb} and @option{-fgcse-after-reload} options.
3922 Do not optimize. This is the default.
3926 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3927 do not typically increase code size. It also performs further
3928 optimizations designed to reduce code size.
3930 @option{-Os} disables the following optimization flags:
3931 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3932 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
3934 If you use multiple @option{-O} options, with or without level numbers,
3935 the last such option is the one that is effective.
3938 Options of the form @option{-f@var{flag}} specify machine-independent
3939 flags. Most flags have both positive and negative forms; the negative
3940 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3941 below, only one of the forms is listed---the one you typically will
3942 use. You can figure out the other form by either removing @samp{no-}
3945 The following options control specific optimizations. They are either
3946 activated by @option{-O} options or are related to ones that are. You
3947 can use the following flags in the rare cases when ``fine-tuning'' of
3948 optimizations to be performed is desired.
3951 @item -fno-default-inline
3952 @opindex fno-default-inline
3953 Do not make member functions inline by default merely because they are
3954 defined inside the class scope (C++ only). Otherwise, when you specify
3955 @w{@option{-O}}, member functions defined inside class scope are compiled
3956 inline by default; i.e., you don't need to add @samp{inline} in front of
3957 the member function name.
3959 @item -fno-defer-pop
3960 @opindex fno-defer-pop
3961 Always pop the arguments to each function call as soon as that function
3962 returns. For machines which must pop arguments after a function call,
3963 the compiler normally lets arguments accumulate on the stack for several
3964 function calls and pops them all at once.
3966 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3970 Force memory operands to be copied into registers before doing
3971 arithmetic on them. This produces better code by making all memory
3972 references potential common subexpressions. When they are not common
3973 subexpressions, instruction combination should eliminate the separate
3976 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3979 @opindex fforce-addr
3980 Force memory address constants to be copied into registers before
3981 doing arithmetic on them. This may produce better code just as
3982 @option{-fforce-mem} may.
3984 @item -fomit-frame-pointer
3985 @opindex fomit-frame-pointer
3986 Don't keep the frame pointer in a register for functions that
3987 don't need one. This avoids the instructions to save, set up and
3988 restore frame pointers; it also makes an extra register available
3989 in many functions. @strong{It also makes debugging impossible on
3992 On some machines, such as the VAX, this flag has no effect, because
3993 the standard calling sequence automatically handles the frame pointer
3994 and nothing is saved by pretending it doesn't exist. The
3995 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3996 whether a target machine supports this flag. @xref{Registers,,Register
3997 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3999 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4001 @item -foptimize-sibling-calls
4002 @opindex foptimize-sibling-calls
4003 Optimize sibling and tail recursive calls.
4005 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4009 Don't pay attention to the @code{inline} keyword. Normally this option
4010 is used to keep the compiler from expanding any functions inline.
4011 Note that if you are not optimizing, no functions can be expanded inline.
4013 @item -finline-functions
4014 @opindex finline-functions
4015 Integrate all simple functions into their callers. The compiler
4016 heuristically decides which functions are simple enough to be worth
4017 integrating in this way.
4019 If all calls to a given function are integrated, and the function is
4020 declared @code{static}, then the function is normally not output as
4021 assembler code in its own right.
4023 Enabled at level @option{-O3}.
4025 @item -finline-limit=@var{n}
4026 @opindex finline-limit
4027 By default, GCC limits the size of functions that can be inlined. This flag
4028 allows the control of this limit for functions that are explicitly marked as
4029 inline (i.e., marked with the inline keyword or defined within the class
4030 definition in c++). @var{n} is the size of functions that can be inlined in
4031 number of pseudo instructions (not counting parameter handling). The default
4032 value of @var{n} is 600.
4033 Increasing this value can result in more inlined code at
4034 the cost of compilation time and memory consumption. Decreasing usually makes
4035 the compilation faster and less code will be inlined (which presumably
4036 means slower programs). This option is particularly useful for programs that
4037 use inlining heavily such as those based on recursive templates with C++.
4039 Inlining is actually controlled by a number of parameters, which may be
4040 specified individually by using @option{--param @var{name}=@var{value}}.
4041 The @option{-finline-limit=@var{n}} option sets some of these parameters
4045 @item max-inline-insns-single
4046 is set to @var{n}/2.
4047 @item max-inline-insns-auto
4048 is set to @var{n}/2.
4049 @item min-inline-insns
4050 is set to 130 or @var{n}/4, whichever is smaller.
4051 @item max-inline-insns-rtl
4055 See below for a documentation of the individual
4056 parameters controlling inlining.
4058 @emph{Note:} pseudo instruction represents, in this particular context, an
4059 abstract measurement of function's size. In no way, it represents a count
4060 of assembly instructions and as such its exact meaning might change from one
4061 release to an another.
4063 @item -fkeep-inline-functions
4064 @opindex fkeep-inline-functions
4065 In C, emit @code{static} functions that are declared @code{inline}
4066 into the object file, even if the function has been inlined into all
4067 of its callers. This switch does not affect functions using the
4068 @code{extern inline} extension in GNU C. In C++, emit any and all
4069 inline functions into the object file.
4071 @item -fkeep-static-consts
4072 @opindex fkeep-static-consts
4073 Emit variables declared @code{static const} when optimization isn't turned
4074 on, even if the variables aren't referenced.
4076 GCC enables this option by default. If you want to force the compiler to
4077 check if the variable was referenced, regardless of whether or not
4078 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4080 @item -fmerge-constants
4081 Attempt to merge identical constants (string constants and floating point
4082 constants) across compilation units.
4084 This option is the default for optimized compilation if the assembler and
4085 linker support it. Use @option{-fno-merge-constants} to inhibit this
4088 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4090 @item -fmerge-all-constants
4091 Attempt to merge identical constants and identical variables.
4093 This option implies @option{-fmerge-constants}. In addition to
4094 @option{-fmerge-constants} this considers e.g. even constant initialized
4095 arrays or initialized constant variables with integral or floating point
4096 types. Languages like C or C++ require each non-automatic variable to
4097 have distinct location, so using this option will result in non-conforming
4100 @item -fmodulo-sched
4101 @opindex fmodulo-sched
4102 Perform swing modulo scheduling immediately before the first scheduling
4103 pass. This pass looks at innermost loops and reorders their
4104 instructions by overlapping different iterations.
4108 Use a graph coloring register allocator. Currently this option is meant
4109 only for testing. Users should not specify this option, since it is not
4110 yet ready for production use.
4112 @item -fno-branch-count-reg
4113 @opindex fno-branch-count-reg
4114 Do not use ``decrement and branch'' instructions on a count register,
4115 but instead generate a sequence of instructions that decrement a
4116 register, compare it against zero, then branch based upon the result.
4117 This option is only meaningful on architectures that support such
4118 instructions, which include x86, PowerPC, IA-64 and S/390.
4120 The default is @option{-fbranch-count-reg}, enabled when
4121 @option{-fstrength-reduce} is enabled.
4123 @item -fno-function-cse
4124 @opindex fno-function-cse
4125 Do not put function addresses in registers; make each instruction that
4126 calls a constant function contain the function's address explicitly.
4128 This option results in less efficient code, but some strange hacks
4129 that alter the assembler output may be confused by the optimizations
4130 performed when this option is not used.
4132 The default is @option{-ffunction-cse}
4134 @item -fno-zero-initialized-in-bss
4135 @opindex fno-zero-initialized-in-bss
4136 If the target supports a BSS section, GCC by default puts variables that
4137 are initialized to zero into BSS@. This can save space in the resulting
4140 This option turns off this behavior because some programs explicitly
4141 rely on variables going to the data section. E.g., so that the
4142 resulting executable can find the beginning of that section and/or make
4143 assumptions based on that.
4145 The default is @option{-fzero-initialized-in-bss}.
4147 @item -fbounds-check
4148 @opindex fbounds-check
4149 For front-ends that support it, generate additional code to check that
4150 indices used to access arrays are within the declared range. This is
4151 currently only supported by the Java and Fortran front-ends, where
4152 this option defaults to true and false respectively.
4154 @item -fmudflap -fmudflapth -fmudflapir
4158 @cindex bounds checking
4160 For front-ends that support it (C and C++), instrument all risky
4161 pointer/array dereferencing operations, some standard library
4162 string/heap functions, and some other associated constructs with
4163 range/validity tests. Modules so instrumented should be immune to
4164 buffer overflows, invalid heap use, and some other classes of C/C++
4165 programming errors. The instrumentation relies on a separate runtime
4166 library (@file{libmudflap}), which will be linked into a program if
4167 @option{-fmudflap} is given at link time. Run-time behavior of the
4168 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4169 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4172 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4173 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4174 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4175 instrumentation should ignore pointer reads. This produces less
4176 instrumentation (and therefore faster execution) and still provides
4177 some protection against outright memory corrupting writes, but allows
4178 erroneously read data to propagate within a program.
4180 @item -fstrength-reduce
4181 @opindex fstrength-reduce
4182 Perform the optimizations of loop strength reduction and
4183 elimination of iteration variables.
4185 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4187 @item -fthread-jumps
4188 @opindex fthread-jumps
4189 Perform optimizations where we check to see if a jump branches to a
4190 location where another comparison subsumed by the first is found. If
4191 so, the first branch is redirected to either the destination of the
4192 second branch or a point immediately following it, depending on whether
4193 the condition is known to be true or false.
4195 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4197 @item -fcse-follow-jumps
4198 @opindex fcse-follow-jumps
4199 In common subexpression elimination, scan through jump instructions
4200 when the target of the jump is not reached by any other path. For
4201 example, when CSE encounters an @code{if} statement with an
4202 @code{else} clause, CSE will follow the jump when the condition
4205 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4207 @item -fcse-skip-blocks
4208 @opindex fcse-skip-blocks
4209 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4210 follow jumps which conditionally skip over blocks. When CSE
4211 encounters a simple @code{if} statement with no else clause,
4212 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4213 body of the @code{if}.
4215 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4217 @item -frerun-cse-after-loop
4218 @opindex frerun-cse-after-loop
4219 Re-run common subexpression elimination after loop optimizations has been
4222 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4224 @item -frerun-loop-opt
4225 @opindex frerun-loop-opt
4226 Run the loop optimizer twice.
4228 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4232 Perform a global common subexpression elimination pass.
4233 This pass also performs global constant and copy propagation.
4235 @emph{Note:} When compiling a program using computed gotos, a GCC
4236 extension, you may get better runtime performance if you disable
4237 the global common subexpression elimination pass by adding
4238 @option{-fno-gcse} to the command line.
4240 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4244 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4245 attempt to move loads which are only killed by stores into themselves. This
4246 allows a loop containing a load/store sequence to be changed to a load outside
4247 the loop, and a copy/store within the loop.
4249 Enabled by default when gcse is enabled.
4253 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4254 global common subexpression elimination. This pass will attempt to move
4255 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4256 loops containing a load/store sequence can be changed to a load before
4257 the loop and a store after the loop.
4259 Enabled by default when gcse is enabled.
4263 When @option{-fgcse-las} is enabled, the global common subexpression
4264 elimination pass eliminates redundant loads that come after stores to the
4265 same memory location (both partial and full redundancies).
4267 Enabled by default when gcse is enabled.
4269 @item -fgcse-after-reload
4270 @opindex fgcse-after-reload
4271 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4272 pass is performed after reload. The purpose of this pass is to cleanup
4275 @item -floop-optimize
4276 @opindex floop-optimize
4277 Perform loop optimizations: move constant expressions out of loops, simplify
4278 exit test conditions and optionally do strength-reduction and loop unrolling as
4281 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4283 @item -floop-optimize2
4284 @opindex floop-optimize2
4285 Perform loop optimizations using the new loop optimizer. The optimizations
4286 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4289 @item -fcrossjumping
4290 @opindex crossjumping
4291 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4292 resulting code may or may not perform better than without cross-jumping.
4294 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4296 @item -fif-conversion
4297 @opindex if-conversion
4298 Attempt to transform conditional jumps into branch-less equivalents. This
4299 include use of conditional moves, min, max, set flags and abs instructions, and
4300 some tricks doable by standard arithmetics. The use of conditional execution
4301 on chips where it is available is controlled by @code{if-conversion2}.
4303 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4305 @item -fif-conversion2
4306 @opindex if-conversion2
4307 Use conditional execution (where available) to transform conditional jumps into
4308 branch-less equivalents.
4310 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4312 @item -fdelete-null-pointer-checks
4313 @opindex fdelete-null-pointer-checks
4314 Use global dataflow analysis to identify and eliminate useless checks
4315 for null pointers. The compiler assumes that dereferencing a null
4316 pointer would have halted the program. If a pointer is checked after
4317 it has already been dereferenced, it cannot be null.
4319 In some environments, this assumption is not true, and programs can
4320 safely dereference null pointers. Use
4321 @option{-fno-delete-null-pointer-checks} to disable this optimization
4322 for programs which depend on that behavior.
4324 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4326 @item -fexpensive-optimizations
4327 @opindex fexpensive-optimizations
4328 Perform a number of minor optimizations that are relatively expensive.
4330 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4332 @item -foptimize-register-move
4334 @opindex foptimize-register-move
4336 Attempt to reassign register numbers in move instructions and as
4337 operands of other simple instructions in order to maximize the amount of
4338 register tying. This is especially helpful on machines with two-operand
4341 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4344 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4346 @item -fdelayed-branch
4347 @opindex fdelayed-branch
4348 If supported for the target machine, attempt to reorder instructions
4349 to exploit instruction slots available after delayed branch
4352 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4354 @item -fschedule-insns
4355 @opindex fschedule-insns
4356 If supported for the target machine, attempt to reorder instructions to
4357 eliminate execution stalls due to required data being unavailable. This
4358 helps machines that have slow floating point or memory load instructions
4359 by allowing other instructions to be issued until the result of the load
4360 or floating point instruction is required.
4362 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4364 @item -fschedule-insns2
4365 @opindex fschedule-insns2
4366 Similar to @option{-fschedule-insns}, but requests an additional pass of
4367 instruction scheduling after register allocation has been done. This is
4368 especially useful on machines with a relatively small number of
4369 registers and where memory load instructions take more than one cycle.
4371 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4373 @item -fno-sched-interblock
4374 @opindex fno-sched-interblock
4375 Don't schedule instructions across basic blocks. This is normally
4376 enabled by default when scheduling before register allocation, i.e.@:
4377 with @option{-fschedule-insns} or at @option{-O2} or higher.
4379 @item -fno-sched-spec
4380 @opindex fno-sched-spec
4381 Don't allow speculative motion of non-load instructions. This is normally
4382 enabled by default when scheduling before register allocation, i.e.@:
4383 with @option{-fschedule-insns} or at @option{-O2} or higher.
4385 @item -fsched-spec-load
4386 @opindex fsched-spec-load
4387 Allow speculative motion of some load instructions. This only makes
4388 sense when scheduling before register allocation, i.e.@: with
4389 @option{-fschedule-insns} or at @option{-O2} or higher.
4391 @item -fsched-spec-load-dangerous
4392 @opindex fsched-spec-load-dangerous
4393 Allow speculative motion of more load instructions. This only makes
4394 sense when scheduling before register allocation, i.e.@: with
4395 @option{-fschedule-insns} or at @option{-O2} or higher.
4397 @item -fsched-stalled-insns=@var{n}
4398 @opindex fsched-stalled-insns
4399 Define how many insns (if any) can be moved prematurely from the queue
4400 of stalled insns into the ready list, during the second scheduling pass.
4402 @item -fsched-stalled-insns-dep=@var{n}
4403 @opindex fsched-stalled-insns-dep
4404 Define how many insn groups (cycles) will be examined for a dependency
4405 on a stalled insn that is candidate for premature removal from the queue
4406 of stalled insns. Has an effect only during the second scheduling pass,
4407 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4409 @item -fsched2-use-superblocks
4410 @opindex fsched2-use-superblocks
4411 When scheduling after register allocation, do use superblock scheduling
4412 algorithm. Superblock scheduling allows motion across basic block boundaries
4413 resulting on faster schedules. This option is experimental, as not all machine
4414 descriptions used by GCC model the CPU closely enough to avoid unreliable
4415 results from the algorithm.
4417 This only makes sense when scheduling after register allocation, i.e.@: with
4418 @option{-fschedule-insns2} or at @option{-O2} or higher.
4420 @item -fsched2-use-traces
4421 @opindex fsched2-use-traces
4422 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4423 allocation and additionally perform code duplication in order to increase the
4424 size of superblocks using tracer pass. See @option{-ftracer} for details on
4427 This mode should produce faster but significantly longer programs. Also
4428 without @code{-fbranch-probabilities} the traces constructed may not match the
4429 reality and hurt the performance. This only makes
4430 sense when scheduling after register allocation, i.e.@: with
4431 @option{-fschedule-insns2} or at @option{-O2} or higher.
4433 @item -freschedule-modulo-scheduled-loops
4434 @opindex fscheduling-in-modulo-scheduled-loops
4435 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4436 we may want to prevent the later scheduling passes from changing its schedule, we use this
4437 option to control that.
4439 @item -fcaller-saves
4440 @opindex fcaller-saves
4441 Enable values to be allocated in registers that will be clobbered by
4442 function calls, by emitting extra instructions to save and restore the
4443 registers around such calls. Such allocation is done only when it
4444 seems to result in better code than would otherwise be produced.
4446 This option is always enabled by default on certain machines, usually
4447 those which have no call-preserved registers to use instead.
4449 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4452 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4453 enabled by default at -O and higher.
4456 Perform Full Redundancy Elimination (FRE) on trees. The difference
4457 between FRE and PRE is that FRE only considers expressions
4458 that are computed on all paths leading to the redundant computation.
4459 This analysis faster than PRE, though it exposes fewer redundancies.
4460 This flag is enabled by default at -O and higher.
4463 Perform sparse conditional constant propagation (CCP) on trees. This flag
4464 is enabled by default at -O and higher.
4467 Perform dead code elimination (DCE) on trees. This flag is enabled by
4468 default at -O and higher.
4470 @item -ftree-dominator-opts
4471 Perform dead code elimination (DCE) on trees. This flag is enabled by
4472 default at -O and higher.
4475 Perform loop header copying on trees. This is beneficial since it increases
4476 effectivity of code motion optimizations. It also saves one jump. This flag
4477 is enabled by default at -O and higher. It is not enabled for -Os, since it
4478 usually increases code size.
4480 @item -ftree-loop-optimize
4481 Perform loop optimizations on trees. This flag is enabled by default at -O
4485 Perform loop invariant motion on trees. This pass moves only invartiants that
4486 would be hard to handle on rtl level (function calls, operations that expand to
4487 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
4488 operands of conditions that are invariant out of the loop, so that we can use
4489 just trivial invariantness analysis in loop unswitching. The pass also includes
4493 Create a canonical counter for number of iterations in the loop for that
4494 determining number of iterations requires complicated analysis. Later
4495 optimizations then may determine the number easily. Useful especially
4496 in connection with unrolling.
4499 Perform scalar replacement of aggregates. This pass replaces structure
4500 references with scalars to prevent committing structures to memory too
4501 early. This flag is enabled by default at -O and higher.
4503 @item -ftree-copyrename
4504 Perform copy renaming on trees. This pass attempts to rename compiler
4505 temporaries to other variables at copy locations, usually resulting in
4506 variable names which more closely resemble the original variables. This flag
4507 is enabled by default at -O and higher.
4510 Perform temporary expression replacement during the SSA->normal phase. Single
4511 use/single def temporaries are replaced at their use location with their
4512 defining expression. This results in non-GIMPLE code, but gives the expanders
4513 much more complex trees to work on resulting in better RTL generation. This is
4514 enabled by default at -O and higher.
4517 Perform live range splitting during the SSA->normal phase. Distinct live
4518 ranges of a variable are split into unique variables, allowing for better
4519 optimization later. This is enabled by default at -O and higher.
4521 @item -ftree-vectorize
4522 Perform loop vectorization on trees.
4526 Perform tail duplication to enlarge superblock size. This transformation
4527 simplifies the control flow of the function allowing other optimizations to do
4530 @item -funroll-loops
4531 @opindex funroll-loops
4532 Unroll loops whose number of iterations can be determined at compile
4533 time or upon entry to the loop. @option{-funroll-loops} implies both
4534 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4535 option makes code larger, and may or may not make it run faster.
4537 @item -funroll-all-loops
4538 @opindex funroll-all-loops
4539 Unroll all loops, even if their number of iterations is uncertain when
4540 the loop is entered. This usually makes programs run more slowly.
4541 @option{-funroll-all-loops} implies the same options as
4542 @option{-funroll-loops},
4544 @item -fprefetch-loop-arrays
4545 @opindex fprefetch-loop-arrays
4546 If supported by the target machine, generate instructions to prefetch
4547 memory to improve the performance of loops that access large arrays.
4549 @item -fmove-all-movables
4550 @opindex fmove-all-movables
4551 Forces all invariant computations in loops to be moved
4554 @item -freduce-all-givs
4555 @opindex freduce-all-givs
4556 Forces all general-induction variables in loops to be
4559 @emph{Note:} When compiling programs written in Fortran,
4560 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4561 by default when you use the optimizer.
4563 These options may generate better or worse code; results are highly
4564 dependent on the structure of loops within the source code.
4566 These two options are intended to be removed someday, once
4567 they have helped determine the efficacy of various
4568 approaches to improving loop optimizations.
4570 Please contact @w{@email{gcc@@gcc.gnu.org}}, and describe how use of
4571 these options affects the performance of your production code.
4572 Examples of code that runs @emph{slower} when these options are
4573 @emph{enabled} are very valuable.
4576 @itemx -fno-peephole2
4577 @opindex fno-peephole
4578 @opindex fno-peephole2
4579 Disable any machine-specific peephole optimizations. The difference
4580 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4581 are implemented in the compiler; some targets use one, some use the
4582 other, a few use both.
4584 @option{-fpeephole} is enabled by default.
4585 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4587 @item -fno-guess-branch-probability
4588 @opindex fno-guess-branch-probability
4589 Do not guess branch probabilities using a randomized model.
4591 Sometimes GCC will opt to use a randomized model to guess branch
4592 probabilities, when none are available from either profiling feedback
4593 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4594 different runs of the compiler on the same program may produce different
4597 In a hard real-time system, people don't want different runs of the
4598 compiler to produce code that has different behavior; minimizing
4599 non-determinism is of paramount import. This switch allows users to
4600 reduce non-determinism, possibly at the expense of inferior
4603 The default is @option{-fguess-branch-probability} at levels
4604 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4606 @item -freorder-blocks
4607 @opindex freorder-blocks
4608 Reorder basic blocks in the compiled function in order to reduce number of
4609 taken branches and improve code locality.
4611 Enabled at levels @option{-O2}, @option{-O3}.
4613 @item -freorder-blocks-and-partition
4614 @opindex freorder-blocks-and-partition
4615 In addition to reordering basic blocks in the compiled function, in order
4616 to reduce number of taken branches, partitions hot and cold basic blocks
4617 into separate sections of the assembly and .o files, to improve
4618 paging and cache locality performance.
4620 @item -freorder-functions
4621 @opindex freorder-functions
4622 Reorder basic blocks in the compiled function in order to reduce number of
4623 taken branches and improve code locality. This is implemented by using special
4624 subsections @code{.text.hot} for most frequently executed functions and
4625 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4626 the linker so object file format must support named sections and linker must
4627 place them in a reasonable way.
4629 Also profile feedback must be available in to make this option effective. See
4630 @option{-fprofile-arcs} for details.
4632 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4634 @item -fstrict-aliasing
4635 @opindex fstrict-aliasing
4636 Allows the compiler to assume the strictest aliasing rules applicable to
4637 the language being compiled. For C (and C++), this activates
4638 optimizations based on the type of expressions. In particular, an
4639 object of one type is assumed never to reside at the same address as an
4640 object of a different type, unless the types are almost the same. For
4641 example, an @code{unsigned int} can alias an @code{int}, but not a
4642 @code{void*} or a @code{double}. A character type may alias any other
4645 Pay special attention to code like this:
4658 The practice of reading from a different union member than the one most
4659 recently written to (called ``type-punning'') is common. Even with
4660 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4661 is accessed through the union type. So, the code above will work as
4662 expected. However, this code might not:
4673 Every language that wishes to perform language-specific alias analysis
4674 should define a function that computes, given an @code{tree}
4675 node, an alias set for the node. Nodes in different alias sets are not
4676 allowed to alias. For an example, see the C front-end function
4677 @code{c_get_alias_set}.
4679 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4681 @item -falign-functions
4682 @itemx -falign-functions=@var{n}
4683 @opindex falign-functions
4684 Align the start of functions to the next power-of-two greater than
4685 @var{n}, skipping up to @var{n} bytes. For instance,
4686 @option{-falign-functions=32} aligns functions to the next 32-byte
4687 boundary, but @option{-falign-functions=24} would align to the next
4688 32-byte boundary only if this can be done by skipping 23 bytes or less.
4690 @option{-fno-align-functions} and @option{-falign-functions=1} are
4691 equivalent and mean that functions will not be aligned.
4693 Some assemblers only support this flag when @var{n} is a power of two;
4694 in that case, it is rounded up.
4696 If @var{n} is not specified or is zero, use a machine-dependent default.
4698 Enabled at levels @option{-O2}, @option{-O3}.
4700 @item -falign-labels
4701 @itemx -falign-labels=@var{n}
4702 @opindex falign-labels
4703 Align all branch targets to a power-of-two boundary, skipping up to
4704 @var{n} bytes like @option{-falign-functions}. This option can easily
4705 make code slower, because it must insert dummy operations for when the
4706 branch target is reached in the usual flow of the code.
4708 @option{-fno-align-labels} and @option{-falign-labels=1} are
4709 equivalent and mean that labels will not be aligned.
4711 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4712 are greater than this value, then their values are used instead.
4714 If @var{n} is not specified or is zero, use a machine-dependent default
4715 which is very likely to be @samp{1}, meaning no alignment.
4717 Enabled at levels @option{-O2}, @option{-O3}.
4720 @itemx -falign-loops=@var{n}
4721 @opindex falign-loops
4722 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4723 like @option{-falign-functions}. The hope is that the loop will be
4724 executed many times, which will make up for any execution of the dummy
4727 @option{-fno-align-loops} and @option{-falign-loops=1} are
4728 equivalent and mean that loops will not be aligned.
4730 If @var{n} is not specified or is zero, use a machine-dependent default.
4732 Enabled at levels @option{-O2}, @option{-O3}.
4735 @itemx -falign-jumps=@var{n}
4736 @opindex falign-jumps
4737 Align branch targets to a power-of-two boundary, for branch targets
4738 where the targets can only be reached by jumping, skipping up to @var{n}
4739 bytes like @option{-falign-functions}. In this case, no dummy operations
4742 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4743 equivalent and mean that loops will not be aligned.
4745 If @var{n} is not specified or is zero, use a machine-dependent default.
4747 Enabled at levels @option{-O2}, @option{-O3}.
4749 @item -funit-at-a-time
4750 @opindex funit-at-a-time
4751 Parse the whole compilation unit before starting to produce code.
4752 This allows some extra optimizations to take place but consumes
4753 more memory (in general). There are some compatibility issues
4754 with @emph{unit-at-at-time} mode:
4757 enabling @emph{unit-at-a-time} mode may change the order
4758 in which functions, variables, and top-level @code{asm} statements
4759 are emitted, and will likely break code relying on some particular
4760 ordering. The majority of such top-level @code{asm} statements,
4761 though, can be replaced by @code{section} attributes.
4764 @emph{unit-at-a-time} mode removes unreferenced static variables
4765 and functions are removed. This may result in undefined references
4766 when an @code{asm} statement refers directly to variables or functions
4767 that are otherwise unused. In that case either the variable/function
4768 shall be listed as an operand of the @code{asm} statement operand or,
4769 in the case of top-level @code{asm} statements the attribute @code{used}
4770 shall be used on the declaration.
4773 Static functions now can use non-standard passing conventions that
4774 may break @code{asm} statements calling functions directly. Again,
4775 attribute @code{used} will prevent this behavior.
4778 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
4779 but this scheme may not be supported by future releases of GCC.
4781 Enabled at levels @option{-O2}, @option{-O3}.
4785 Constructs webs as commonly used for register allocation purposes and assign
4786 each web individual pseudo register. This allows the register allocation pass
4787 to operate on pseudos directly, but also strengthens several other optimization
4788 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4789 however, make debugging impossible, since variables will no longer stay in a
4792 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
4793 on targets where the default format for debugging information supports
4796 @item -fno-cprop-registers
4797 @opindex fno-cprop-registers
4798 After register allocation and post-register allocation instruction splitting,
4799 we perform a copy-propagation pass to try to reduce scheduling dependencies
4800 and occasionally eliminate the copy.
4802 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4804 @item -fprofile-generate
4805 @opindex fprofile-generate
4807 Enable options usually used for instrumenting application to produce
4808 profile useful for later recompilation with profile feedback based
4809 optimization. You must use @code{-fprofile-generate} both when
4810 compiling and when linking your program.
4812 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
4815 @opindex fprofile-use
4816 Enable profile feedback directed optimizations, and optimizations
4817 generally profitable only with profile feedback available.
4819 The following options are enabled: @code{-fbranch-probabilities},
4820 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4824 The following options control compiler behavior regarding floating
4825 point arithmetic. These options trade off between speed and
4826 correctness. All must be specifically enabled.
4830 @opindex ffloat-store
4831 Do not store floating point variables in registers, and inhibit other
4832 options that might change whether a floating point value is taken from a
4835 @cindex floating point precision
4836 This option prevents undesirable excess precision on machines such as
4837 the 68000 where the floating registers (of the 68881) keep more
4838 precision than a @code{double} is supposed to have. Similarly for the
4839 x86 architecture. For most programs, the excess precision does only
4840 good, but a few programs rely on the precise definition of IEEE floating
4841 point. Use @option{-ffloat-store} for such programs, after modifying
4842 them to store all pertinent intermediate computations into variables.
4846 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4847 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4848 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4850 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4852 This option should never be turned on by any @option{-O} option since
4853 it can result in incorrect output for programs which depend on
4854 an exact implementation of IEEE or ISO rules/specifications for
4857 @item -fno-math-errno
4858 @opindex fno-math-errno
4859 Do not set ERRNO after calling math functions that are executed
4860 with a single instruction, e.g., sqrt. A program that relies on
4861 IEEE exceptions for math error handling may want to use this flag
4862 for speed while maintaining IEEE arithmetic compatibility.
4864 This option should never be turned on by any @option{-O} option since
4865 it can result in incorrect output for programs which depend on
4866 an exact implementation of IEEE or ISO rules/specifications for
4869 The default is @option{-fmath-errno}.
4871 @item -funsafe-math-optimizations
4872 @opindex funsafe-math-optimizations
4873 Allow optimizations for floating-point arithmetic that (a) assume
4874 that arguments and results are valid and (b) may violate IEEE or
4875 ANSI standards. When used at link-time, it may include libraries
4876 or startup files that change the default FPU control word or other
4877 similar optimizations.
4879 This option should never be turned on by any @option{-O} option since
4880 it can result in incorrect output for programs which depend on
4881 an exact implementation of IEEE or ISO rules/specifications for
4884 The default is @option{-fno-unsafe-math-optimizations}.
4886 @item -ffinite-math-only
4887 @opindex ffinite-math-only
4888 Allow optimizations for floating-point arithmetic that assume
4889 that arguments and results are not NaNs or +-Infs.
4891 This option should never be turned on by any @option{-O} option since
4892 it can result in incorrect output for programs which depend on
4893 an exact implementation of IEEE or ISO rules/specifications.
4895 The default is @option{-fno-finite-math-only}.
4897 @item -fno-trapping-math
4898 @opindex fno-trapping-math
4899 Compile code assuming that floating-point operations cannot generate
4900 user-visible traps. These traps include division by zero, overflow,
4901 underflow, inexact result and invalid operation. This option implies
4902 @option{-fno-signaling-nans}. Setting this option may allow faster
4903 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4905 This option should never be turned on by any @option{-O} option since
4906 it can result in incorrect output for programs which depend on
4907 an exact implementation of IEEE or ISO rules/specifications for
4910 The default is @option{-ftrapping-math}.
4912 @item -frounding-math
4913 @opindex frounding-math
4914 Disable transformations and optimizations that assume default floating
4915 point rounding behavior. This is round-to-zero for all floating point
4916 to integer conversions, and round-to-nearest for all other arithmetic
4917 truncations. This option should be specified for programs that change
4918 the FP rounding mode dynamically, or that may be executed with a
4919 non-default rounding mode. This option disables constant folding of
4920 floating point expressions at compile-time (which may be affected by
4921 rounding mode) and arithmetic transformations that are unsafe in the
4922 presence of sign-dependent rounding modes.
4924 The default is @option{-fno-rounding-math}.
4926 This option is experimental and does not currently guarantee to
4927 disable all GCC optimizations that are affected by rounding mode.
4928 Future versions of GCC may provide finer control of this setting
4929 using C99's @code{FENV_ACCESS} pragma. This command line option
4930 will be used to specify the default state for @code{FENV_ACCESS}.
4932 @item -fsignaling-nans
4933 @opindex fsignaling-nans
4934 Compile code assuming that IEEE signaling NaNs may generate user-visible
4935 traps during floating-point operations. Setting this option disables
4936 optimizations that may change the number of exceptions visible with
4937 signaling NaNs. This option implies @option{-ftrapping-math}.
4939 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4942 The default is @option{-fno-signaling-nans}.
4944 This option is experimental and does not currently guarantee to
4945 disable all GCC optimizations that affect signaling NaN behavior.
4947 @item -fsingle-precision-constant
4948 @opindex fsingle-precision-constant
4949 Treat floating point constant as single precision constant instead of
4950 implicitly converting it to double precision constant.
4955 The following options control optimizations that may improve
4956 performance, but are not enabled by any @option{-O} options. This
4957 section includes experimental options that may produce broken code.
4960 @item -fbranch-probabilities
4961 @opindex fbranch-probabilities
4962 After running a program compiled with @option{-fprofile-arcs}
4963 (@pxref{Debugging Options,, Options for Debugging Your Program or
4964 @command{gcc}}), you can compile it a second time using
4965 @option{-fbranch-probabilities}, to improve optimizations based on
4966 the number of times each branch was taken. When the program
4967 compiled with @option{-fprofile-arcs} exits it saves arc execution
4968 counts to a file called @file{@var{sourcename}.gcda} for each source
4969 file The information in this data file is very dependent on the
4970 structure of the generated code, so you must use the same source code
4971 and the same optimization options for both compilations.
4973 With @option{-fbranch-probabilities}, GCC puts a
4974 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4975 These can be used to improve optimization. Currently, they are only
4976 used in one place: in @file{reorg.c}, instead of guessing which path a
4977 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4978 exactly determine which path is taken more often.
4980 @item -fprofile-values
4981 @opindex fprofile-values
4982 If combined with @option{-fprofile-arcs}, it adds code so that some
4983 data about values of expressions in the program is gathered.
4985 With @option{-fbranch-probabilities}, it reads back the data gathered
4986 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4987 notes to instructions for their later usage in optimizations.
4989 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
4993 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4994 a code to gather information about values of expressions.
4996 With @option{-fbranch-probabilities}, it reads back the data gathered
4997 and actually performs the optimizations based on them.
4998 Currently the optimizations include specialization of division operation
4999 using the knowledge about the value of the denominator.
5001 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5003 @item -frename-registers
5004 @opindex frename-registers
5005 Attempt to avoid false dependencies in scheduled code by making use
5006 of registers left over after register allocation. This optimization
5007 will most benefit processors with lots of registers. Depending on the
5008 debug information format adopted by the target, however, it can
5009 make debugging impossible, since variables will no longer stay in
5010 a ``home register''.
5012 Not enabled by default at any level because it has known bugs.
5016 Use a graph coloring register allocator. Currently this option is meant
5017 for testing, so we are interested to hear about miscompilations with
5022 Perform tail duplication to enlarge superblock size. This transformation
5023 simplifies the control flow of the function allowing other optimizations to do
5026 Enabled with @option{-fprofile-use}.
5028 @item -funroll-loops
5029 @opindex funroll-loops
5030 Unroll loops whose number of iterations can be determined at compile time or
5031 upon entry to the loop. @option{-funroll-loops} implies
5032 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
5033 (i.e. complete removal of loops with small constant number of iterations).
5034 This option makes code larger, and may or may not make it run faster.
5036 Enabled with @option{-fprofile-use}.
5038 @item -funroll-all-loops
5039 @opindex funroll-all-loops
5040 Unroll all loops, even if their number of iterations is uncertain when
5041 the loop is entered. This usually makes programs run more slowly.
5042 @option{-funroll-all-loops} implies the same options as
5043 @option{-funroll-loops}.
5046 @opindex fpeel-loops
5047 Peels the loops for that there is enough information that they do not
5048 roll much (from profile feedback). It also turns on complete loop peeling
5049 (i.e. complete removal of loops with small constant number of iterations).
5051 Enabled with @option{-fprofile-use}.
5053 @item -fmove-loop-invariants
5054 @opindex fmove-loop-invariants
5055 Enables the loop invariant motion pass in the new loop optimizer. Enabled
5056 at level @option{-O1}
5058 @item -funswitch-loops
5059 @opindex funswitch-loops
5060 Move branches with loop invariant conditions out of the loop, with duplicates
5061 of the loop on both branches (modified according to result of the condition).
5063 @item -fold-unroll-loops
5064 @opindex fold-unroll-loops
5065 Unroll loops whose number of iterations can be determined at compile
5066 time or upon entry to the loop, using the old loop unroller whose loop
5067 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
5068 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
5069 option makes code larger, and may or may not make it run faster.
5071 @item -fold-unroll-all-loops
5072 @opindex fold-unroll-all-loops
5073 Unroll all loops, even if their number of iterations is uncertain when
5074 the loop is entered. This is done using the old loop unroller whose loop
5075 recognition is based on notes from frontend. This usually makes programs run more slowly.
5076 @option{-fold-unroll-all-loops} implies the same options as
5077 @option{-fold-unroll-loops}.
5079 @item -fprefetch-loop-arrays
5080 @opindex fprefetch-loop-arrays
5081 If supported by the target machine, generate instructions to prefetch
5082 memory to improve the performance of loops that access large arrays.
5084 Disabled at level @option{-Os}.
5086 @item -ffunction-sections
5087 @itemx -fdata-sections
5088 @opindex ffunction-sections
5089 @opindex fdata-sections
5090 Place each function or data item into its own section in the output
5091 file if the target supports arbitrary sections. The name of the
5092 function or the name of the data item determines the section's name
5095 Use these options on systems where the linker can perform optimizations
5096 to improve locality of reference in the instruction space. Most systems
5097 using the ELF object format and SPARC processors running Solaris 2 have
5098 linkers with such optimizations. AIX may have these optimizations in
5101 Only use these options when there are significant benefits from doing
5102 so. When you specify these options, the assembler and linker will
5103 create larger object and executable files and will also be slower.
5104 You will not be able to use @code{gprof} on all systems if you
5105 specify this option and you may have problems with debugging if
5106 you specify both this option and @option{-g}.
5108 @item -fbranch-target-load-optimize
5109 @opindex fbranch-target-load-optimize
5110 Perform branch target register load optimization before prologue / epilogue
5112 The use of target registers can typically be exposed only during reload,
5113 thus hoisting loads out of loops and doing inter-block scheduling needs
5114 a separate optimization pass.
5116 @item -fbranch-target-load-optimize2
5117 @opindex fbranch-target-load-optimize2
5118 Perform branch target register load optimization after prologue / epilogue
5121 @item -fbtr-bb-exclusive
5122 @opindex fbtr-bb-exclusive
5123 When performing branch target register load optimization, don't reuse
5124 branch target registers in within any basic block.
5126 @item --param @var{name}=@var{value}
5128 In some places, GCC uses various constants to control the amount of
5129 optimization that is done. For example, GCC will not inline functions
5130 that contain more that a certain number of instructions. You can
5131 control some of these constants on the command-line using the
5132 @option{--param} option.
5134 The names of specific parameters, and the meaning of the values, are
5135 tied to the internals of the compiler, and are subject to change
5136 without notice in future releases.
5138 In each case, the @var{value} is an integer. The allowable choices for
5139 @var{name} are given in the following table:
5142 @item max-crossjump-edges
5143 The maximum number of incoming edges to consider for crossjumping.
5144 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5145 the number of edges incoming to each block. Increasing values mean
5146 more aggressive optimization, making the compile time increase with
5147 probably small improvement in executable size.
5149 @item min-crossjump-insns
5150 The minimum number of instructions which must be matched at the end
5151 of two blocks before crossjumping will be performed on them. This
5152 value is ignored in the case where all instructions in the block being
5153 crossjumped from are matched. The default value is 5.
5155 @item max-delay-slot-insn-search
5156 The maximum number of instructions to consider when looking for an
5157 instruction to fill a delay slot. If more than this arbitrary number of
5158 instructions is searched, the time savings from filling the delay slot
5159 will be minimal so stop searching. Increasing values mean more
5160 aggressive optimization, making the compile time increase with probably
5161 small improvement in executable run time.
5163 @item max-delay-slot-live-search
5164 When trying to fill delay slots, the maximum number of instructions to
5165 consider when searching for a block with valid live register
5166 information. Increasing this arbitrarily chosen value means more
5167 aggressive optimization, increasing the compile time. This parameter
5168 should be removed when the delay slot code is rewritten to maintain the
5171 @item max-gcse-memory
5172 The approximate maximum amount of memory that will be allocated in
5173 order to perform the global common subexpression elimination
5174 optimization. If more memory than specified is required, the
5175 optimization will not be done.
5177 @item max-gcse-passes
5178 The maximum number of passes of GCSE to run. The default is 1.
5180 @item max-pending-list-length
5181 The maximum number of pending dependencies scheduling will allow
5182 before flushing the current state and starting over. Large functions
5183 with few branches or calls can create excessively large lists which
5184 needlessly consume memory and resources.
5186 @item max-inline-insns-single
5187 Several parameters control the tree inliner used in gcc.
5188 This number sets the maximum number of instructions (counted in GCC's
5189 internal representation) in a single function that the tree inliner
5190 will consider for inlining. This only affects functions declared
5191 inline and methods implemented in a class declaration (C++).
5192 The default value is 500.
5194 @item max-inline-insns-auto
5195 When you use @option{-finline-functions} (included in @option{-O3}),
5196 a lot of functions that would otherwise not be considered for inlining
5197 by the compiler will be investigated. To those functions, a different
5198 (more restrictive) limit compared to functions declared inline can
5200 The default value is 120.
5202 @item large-function-insns
5203 The limit specifying really large functions. For functions greater than this
5204 limit inlining is constrained by @option{--param large-function-growth}.
5205 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
5206 algorithms used by the backend.
5207 This parameter is ignored when @option{-funit-at-a-time} is not used.
5208 The default value is 3000.
5210 @item large-function-growth
5211 Specifies maximal growth of large function caused by inlining in percents.
5212 This parameter is ignored when @option{-funit-at-a-time} is not used.
5213 The default value is 200.
5215 @item inline-unit-growth
5216 Specifies maximal overall growth of the compilation unit caused by inlining.
5217 This parameter is ignored when @option{-funit-at-a-time} is not used.
5218 The default value is 150.
5220 @item max-inline-insns-recursive
5221 @itemx max-inline-insns-recursive-auto
5222 Specifies maximum number of instructions out-of-line copy of self recursive inline
5223 function can grow into by performing recursive inlining.
5225 For functions declared inline @option{--param max-inline-insns-recursive} is
5226 taken into acount. For function not declared inline, recursive inlining
5227 happens only when @option{-finline-functions} (included in @option{-O3}) is
5228 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5229 default value is 500.
5231 @item max-inline-recursive-depth
5232 @itemx max-inline-recursive-depth-auto
5233 Specifies maximum recursion depth used by the recursive inlining.
5235 For functions declared inline @option{--param max-inline-recursive-depth} is
5236 taken into acount. For function not declared inline, recursive inlining
5237 happens only when @option{-finline-functions} (included in @option{-O3}) is
5238 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5239 default value is 500.
5241 @item max-inline-insns-rtl
5242 For languages that use the RTL inliner (this happens at a later stage
5243 than tree inlining), you can set the maximum allowable size (counted
5244 in RTL instructions) for the RTL inliner with this parameter.
5245 The default value is 600.
5247 @item max-unrolled-insns
5248 The maximum number of instructions that a loop should have if that loop
5249 is unrolled, and if the loop is unrolled, it determines how many times
5250 the loop code is unrolled.
5252 @item max-average-unrolled-insns
5253 The maximum number of instructions biased by probabilities of their execution
5254 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5255 it determines how many times the loop code is unrolled.
5257 @item max-unroll-times
5258 The maximum number of unrollings of a single loop.
5260 @item max-peeled-insns
5261 The maximum number of instructions that a loop should have if that loop
5262 is peeled, and if the loop is peeled, it determines how many times
5263 the loop code is peeled.
5265 @item max-peel-times
5266 The maximum number of peelings of a single loop.
5268 @item max-completely-peeled-insns
5269 The maximum number of insns of a completely peeled loop.
5271 @item max-completely-peel-times
5272 The maximum number of iterations of a loop to be suitable for complete peeling.
5274 @item max-unswitch-insns
5275 The maximum number of insns of an unswitched loop.
5277 @item max-unswitch-level
5278 The maximum number of branches unswitched in a single loop.
5281 The minimum cost of an expensive expression in the loop invariant motion.
5283 @item max-iterations-to-track
5285 The maximum number of iterations of a loop the brute force algorithm
5286 for analysis of # of iterations of the loop tries to evaluate.
5288 @item hot-bb-count-fraction
5289 Select fraction of the maximal count of repetitions of basic block in program
5290 given basic block needs to have to be considered hot.
5292 @item hot-bb-frequency-fraction
5293 Select fraction of the maximal frequency of executions of basic block in
5294 function given basic block needs to have to be considered hot
5296 @item tracer-dynamic-coverage
5297 @itemx tracer-dynamic-coverage-feedback
5299 This value is used to limit superblock formation once the given percentage of
5300 executed instructions is covered. This limits unnecessary code size
5303 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5304 feedback is available. The real profiles (as opposed to statically estimated
5305 ones) are much less balanced allowing the threshold to be larger value.
5307 @item tracer-max-code-growth
5308 Stop tail duplication once code growth has reached given percentage. This is
5309 rather hokey argument, as most of the duplicates will be eliminated later in
5310 cross jumping, so it may be set to much higher values than is the desired code
5313 @item tracer-min-branch-ratio
5315 Stop reverse growth when the reverse probability of best edge is less than this
5316 threshold (in percent).
5318 @item tracer-min-branch-ratio
5319 @itemx tracer-min-branch-ratio-feedback
5321 Stop forward growth if the best edge do have probability lower than this
5324 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5325 compilation for profile feedback and one for compilation without. The value
5326 for compilation with profile feedback needs to be more conservative (higher) in
5327 order to make tracer effective.
5329 @item max-cse-path-length
5331 Maximum number of basic blocks on path that cse considers. The default is 10.
5333 @item global-var-threshold
5335 Counts the number of function calls (N) and the number of
5336 call-clobbered variables (V). If NxV is larger than this limit, a
5337 single artificial variable will be created to represent all the
5338 call-clobbered variables at function call sites. This artificial
5339 variable will then be made to alias every call-clobbered variable.
5340 (done as int * size_t on the host machine; beware overflow).
5342 @item max-aliased-vops
5344 Maxiumum number of virtual operands allowed to represent aliases
5345 before triggering the alias grouping heuristic. Alias grouping
5346 reduces compile times and memory consumption needed for aliasing at
5347 the expense of precision loss in alias information.
5349 @item ggc-min-expand
5351 GCC uses a garbage collector to manage its own memory allocation. This
5352 parameter specifies the minimum percentage by which the garbage
5353 collector's heap should be allowed to expand between collections.
5354 Tuning this may improve compilation speed; it has no effect on code
5357 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
5358 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
5359 the smallest of actual RAM and RLIMIT_DATA or RLIMIT_AS. If
5360 GCC is not able to calculate RAM on a particular platform, the lower
5361 bound of 30% is used. Setting this parameter and
5362 @option{ggc-min-heapsize} to zero causes a full collection to occur at
5363 every opportunity. This is extremely slow, but can be useful for
5366 @item ggc-min-heapsize
5368 Minimum size of the garbage collector's heap before it begins bothering
5369 to collect garbage. The first collection occurs after the heap expands
5370 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
5371 tuning this may improve compilation speed, and has no effect on code
5374 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
5375 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
5376 with a lower bound of 4096 (four megabytes) and an upper bound of
5377 131072 (128 megabytes). If GCC is not able to calculate RAM on a
5378 particular platform, the lower bound is used. Setting this parameter
5379 very large effectively disables garbage collection. Setting this
5380 parameter and @option{ggc-min-expand} to zero causes a full collection
5381 to occur at every opportunity.
5383 @item max-reload-search-insns
5384 The maximum number of instruction reload should look backward for equivalent
5385 register. Increasing values mean more aggressive optimization, making the
5386 compile time increase with probably slightly better performance. The default
5389 @item max-cselib-memory-location
5390 The maximum number of memory locations cselib should take into acount.
5391 Increasing values mean more aggressive optimization, making the compile time
5392 increase with probably slightly better performance. The default value is 500.
5394 @item reorder-blocks-duplicate
5395 @itemx reorder-blocks-duplicate-feedback
5397 Used by basic block reordering pass to decide whether to use unconditional
5398 branch or duplicate the code on its destination. Code is duplicated when its
5399 estimated size is smaller than this value multiplied by the estimated size of
5400 unconditional jump in the hot spots of the program.
5402 The @option{reorder-block-duplicate-feedback} is used only when profile
5403 feedback is available and may be set to higher values than
5404 @option{reorder-block-duplicate} since information about the hot spots is more
5407 @item max-sched-region-blocks
5408 The maximum number of blocks in a region to be considered for
5409 interblock scheduling. The default value is 10.
5411 @item max-sched-region-insns
5412 The maximum number of insns in a region to be considered for
5413 interblock scheduling. The default value is 100.
5415 @item integer-share-limit
5416 Small integer constants can use a shared data structure, reducing the
5417 compiler's memory usage and increasing its speed. This sets the maximum
5418 value of a shared integer constant's. The default value is 256.
5423 @node Preprocessor Options
5424 @section Options Controlling the Preprocessor
5425 @cindex preprocessor options
5426 @cindex options, preprocessor
5428 These options control the C preprocessor, which is run on each C source
5429 file before actual compilation.
5431 If you use the @option{-E} option, nothing is done except preprocessing.
5432 Some of these options make sense only together with @option{-E} because
5433 they cause the preprocessor output to be unsuitable for actual
5438 You can use @option{-Wp,@var{option}} to bypass the compiler driver
5439 and pass @var{option} directly through to the preprocessor. If
5440 @var{option} contains commas, it is split into multiple options at the
5441 commas. However, many options are modified, translated or interpreted
5442 by the compiler driver before being passed to the preprocessor, and
5443 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
5444 interface is undocumented and subject to change, so whenever possible
5445 you should avoid using @option{-Wp} and let the driver handle the
5448 @item -Xpreprocessor @var{option}
5449 @opindex preprocessor
5450 Pass @var{option} as an option to the preprocessor. You can use this to
5451 supply system-specific preprocessor options which GCC does not know how to
5454 If you want to pass an option that takes an argument, you must use
5455 @option{-Xpreprocessor} twice, once for the option and once for the argument.
5458 @include cppopts.texi
5460 @node Assembler Options
5461 @section Passing Options to the Assembler
5463 @c prevent bad page break with this line
5464 You can pass options to the assembler.
5467 @item -Wa,@var{option}
5469 Pass @var{option} as an option to the assembler. If @var{option}
5470 contains commas, it is split into multiple options at the commas.
5472 @item -Xassembler @var{option}
5474 Pass @var{option} as an option to the assembler. You can use this to
5475 supply system-specific assembler options which GCC does not know how to
5478 If you want to pass an option that takes an argument, you must use
5479 @option{-Xassembler} twice, once for the option and once for the argument.
5484 @section Options for Linking
5485 @cindex link options
5486 @cindex options, linking
5488 These options come into play when the compiler links object files into
5489 an executable output file. They are meaningless if the compiler is
5490 not doing a link step.
5494 @item @var{object-file-name}
5495 A file name that does not end in a special recognized suffix is
5496 considered to name an object file or library. (Object files are
5497 distinguished from libraries by the linker according to the file
5498 contents.) If linking is done, these object files are used as input
5507 If any of these options is used, then the linker is not run, and
5508 object file names should not be used as arguments. @xref{Overall
5512 @item -l@var{library}
5513 @itemx -l @var{library}
5515 Search the library named @var{library} when linking. (The second
5516 alternative with the library as a separate argument is only for
5517 POSIX compliance and is not recommended.)
5519 It makes a difference where in the command you write this option; the
5520 linker searches and processes libraries and object files in the order they
5521 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5522 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5523 to functions in @samp{z}, those functions may not be loaded.
5525 The linker searches a standard list of directories for the library,
5526 which is actually a file named @file{lib@var{library}.a}. The linker
5527 then uses this file as if it had been specified precisely by name.
5529 The directories searched include several standard system directories
5530 plus any that you specify with @option{-L}.
5532 Normally the files found this way are library files---archive files
5533 whose members are object files. The linker handles an archive file by
5534 scanning through it for members which define symbols that have so far
5535 been referenced but not defined. But if the file that is found is an
5536 ordinary object file, it is linked in the usual fashion. The only
5537 difference between using an @option{-l} option and specifying a file name
5538 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5539 and searches several directories.
5543 You need this special case of the @option{-l} option in order to
5544 link an Objective-C or Objective-C++ program.
5547 @opindex nostartfiles
5548 Do not use the standard system startup files when linking.
5549 The standard system libraries are used normally, unless @option{-nostdlib}
5550 or @option{-nodefaultlibs} is used.
5552 @item -nodefaultlibs
5553 @opindex nodefaultlibs
5554 Do not use the standard system libraries when linking.
5555 Only the libraries you specify will be passed to the linker.
5556 The standard startup files are used normally, unless @option{-nostartfiles}
5557 is used. The compiler may generate calls to @code{memcmp},
5558 @code{memset}, @code{memcpy} and @code{memmove}.
5559 These entries are usually resolved by entries in
5560 libc. These entry points should be supplied through some other
5561 mechanism when this option is specified.
5565 Do not use the standard system startup files or libraries when linking.
5566 No startup files and only the libraries you specify will be passed to
5567 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
5568 @code{memcpy} and @code{memmove}.
5569 These entries are usually resolved by entries in
5570 libc. These entry points should be supplied through some other
5571 mechanism when this option is specified.
5573 @cindex @option{-lgcc}, use with @option{-nostdlib}
5574 @cindex @option{-nostdlib} and unresolved references
5575 @cindex unresolved references and @option{-nostdlib}
5576 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5577 @cindex @option{-nodefaultlibs} and unresolved references
5578 @cindex unresolved references and @option{-nodefaultlibs}
5579 One of the standard libraries bypassed by @option{-nostdlib} and
5580 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5581 that GCC uses to overcome shortcomings of particular machines, or special
5582 needs for some languages.
5583 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5584 Collection (GCC) Internals},
5585 for more discussion of @file{libgcc.a}.)
5586 In most cases, you need @file{libgcc.a} even when you want to avoid
5587 other standard libraries. In other words, when you specify @option{-nostdlib}
5588 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5589 This ensures that you have no unresolved references to internal GCC
5590 library subroutines. (For example, @samp{__main}, used to ensure C++
5591 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5592 GNU Compiler Collection (GCC) Internals}.)
5596 Produce a position independent executable on targets which support it.
5597 For predictable results, you must also specify the same set of options
5598 that were used to generate code (@option{-fpie}, @option{-fPIE},
5599 or model suboptions) when you specify this option.
5603 Remove all symbol table and relocation information from the executable.
5607 On systems that support dynamic linking, this prevents linking with the shared
5608 libraries. On other systems, this option has no effect.
5612 Produce a shared object which can then be linked with other objects to
5613 form an executable. Not all systems support this option. For predictable
5614 results, you must also specify the same set of options that were used to
5615 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5616 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5617 needs to build supplementary stub code for constructors to work. On
5618 multi-libbed systems, @samp{gcc -shared} must select the correct support
5619 libraries to link against. Failing to supply the correct flags may lead
5620 to subtle defects. Supplying them in cases where they are not necessary
5623 @item -shared-libgcc
5624 @itemx -static-libgcc
5625 @opindex shared-libgcc
5626 @opindex static-libgcc
5627 On systems that provide @file{libgcc} as a shared library, these options
5628 force the use of either the shared or static version respectively.
5629 If no shared version of @file{libgcc} was built when the compiler was
5630 configured, these options have no effect.
5632 There are several situations in which an application should use the
5633 shared @file{libgcc} instead of the static version. The most common
5634 of these is when the application wishes to throw and catch exceptions
5635 across different shared libraries. In that case, each of the libraries
5636 as well as the application itself should use the shared @file{libgcc}.
5638 Therefore, the G++ and GCJ drivers automatically add
5639 @option{-shared-libgcc} whenever you build a shared library or a main
5640 executable, because C++ and Java programs typically use exceptions, so
5641 this is the right thing to do.
5643 If, instead, you use the GCC driver to create shared libraries, you may
5644 find that they will not always be linked with the shared @file{libgcc}.
5645 If GCC finds, at its configuration time, that you have a non-GNU linker
5646 or a GNU linker that does not support option @option{--eh-frame-hdr},
5647 it will link the shared version of @file{libgcc} into shared libraries
5648 by default. Otherwise, it will take advantage of the linker and optimize
5649 away the linking with the shared version of @file{libgcc}, linking with
5650 the static version of libgcc by default. This allows exceptions to
5651 propagate through such shared libraries, without incurring relocation
5652 costs at library load time.
5654 However, if a library or main executable is supposed to throw or catch
5655 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5656 for the languages used in the program, or using the option
5657 @option{-shared-libgcc}, such that it is linked with the shared
5662 Bind references to global symbols when building a shared object. Warn
5663 about any unresolved references (unless overridden by the link editor
5664 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5667 @item -Xlinker @var{option}
5669 Pass @var{option} as an option to the linker. You can use this to
5670 supply system-specific linker options which GCC does not know how to
5673 If you want to pass an option that takes an argument, you must use
5674 @option{-Xlinker} twice, once for the option and once for the argument.
5675 For example, to pass @option{-assert definitions}, you must write
5676 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5677 @option{-Xlinker "-assert definitions"}, because this passes the entire
5678 string as a single argument, which is not what the linker expects.
5680 @item -Wl,@var{option}
5682 Pass @var{option} as an option to the linker. If @var{option} contains
5683 commas, it is split into multiple options at the commas.
5685 @item -u @var{symbol}
5687 Pretend the symbol @var{symbol} is undefined, to force linking of
5688 library modules to define it. You can use @option{-u} multiple times with
5689 different symbols to force loading of additional library modules.
5692 @node Directory Options
5693 @section Options for Directory Search
5694 @cindex directory options
5695 @cindex options, directory search
5698 These options specify directories to search for header files, for
5699 libraries and for parts of the compiler:
5704 Add the directory @var{dir} to the head of the list of directories to be
5705 searched for header files. This can be used to override a system header
5706 file, substituting your own version, since these directories are
5707 searched before the system header file directories. However, you should
5708 not use this option to add directories that contain vendor-supplied
5709 system header files (use @option{-isystem} for that). If you use more than
5710 one @option{-I} option, the directories are scanned in left-to-right
5711 order; the standard system directories come after.
5713 If a standard system include directory, or a directory specified with
5714 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5715 option will be ignored. The directory will still be searched but as a
5716 system directory at its normal position in the system include chain.
5717 This is to ensure that GCC's procedure to fix buggy system headers and
5718 the ordering for the include_next directive are not inadvertently changed.
5719 If you really need to change the search order for system directories,
5720 use the @option{-nostdinc} and/or @option{-isystem} options.
5722 @item -iquote@var{dir}
5724 Add the directory @var{dir} to the head of the list of directories to
5725 be searched for header files only for the case of @samp{#include
5726 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
5727 otherwise just like @option{-I}.
5731 Add directory @var{dir} to the list of directories to be searched
5734 @item -B@var{prefix}
5736 This option specifies where to find the executables, libraries,
5737 include files, and data files of the compiler itself.
5739 The compiler driver program runs one or more of the subprograms
5740 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5741 @var{prefix} as a prefix for each program it tries to run, both with and
5742 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5744 For each subprogram to be run, the compiler driver first tries the
5745 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5746 was not specified, the driver tries two standard prefixes, which are
5747 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5748 those results in a file name that is found, the unmodified program
5749 name is searched for using the directories specified in your
5750 @env{PATH} environment variable.
5752 The compiler will check to see if the path provided by the @option{-B}
5753 refers to a directory, and if necessary it will add a directory
5754 separator character at the end of the path.
5756 @option{-B} prefixes that effectively specify directory names also apply
5757 to libraries in the linker, because the compiler translates these
5758 options into @option{-L} options for the linker. They also apply to
5759 includes files in the preprocessor, because the compiler translates these
5760 options into @option{-isystem} options for the preprocessor. In this case,
5761 the compiler appends @samp{include} to the prefix.
5763 The run-time support file @file{libgcc.a} can also be searched for using
5764 the @option{-B} prefix, if needed. If it is not found there, the two
5765 standard prefixes above are tried, and that is all. The file is left
5766 out of the link if it is not found by those means.
5768 Another way to specify a prefix much like the @option{-B} prefix is to use
5769 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5772 As a special kludge, if the path provided by @option{-B} is
5773 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5774 9, then it will be replaced by @file{[dir/]include}. This is to help
5775 with boot-strapping the compiler.
5777 @item -specs=@var{file}
5779 Process @var{file} after the compiler reads in the standard @file{specs}
5780 file, in order to override the defaults that the @file{gcc} driver
5781 program uses when determining what switches to pass to @file{cc1},
5782 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5783 @option{-specs=@var{file}} can be specified on the command line, and they
5784 are processed in order, from left to right.
5788 This option has been deprecated. Please use @option{-iquote} instead for
5789 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
5790 Any directories you specify with @option{-I} options before the @option{-I-}
5791 option are searched only for the case of @samp{#include "@var{file}"};
5792 they are not searched for @samp{#include <@var{file}>}.
5794 If additional directories are specified with @option{-I} options after
5795 the @option{-I-}, these directories are searched for all @samp{#include}
5796 directives. (Ordinarily @emph{all} @option{-I} directories are used
5799 In addition, the @option{-I-} option inhibits the use of the current
5800 directory (where the current input file came from) as the first search
5801 directory for @samp{#include "@var{file}"}. There is no way to
5802 override this effect of @option{-I-}. With @option{-I.} you can specify
5803 searching the directory which was current when the compiler was
5804 invoked. That is not exactly the same as what the preprocessor does
5805 by default, but it is often satisfactory.
5807 @option{-I-} does not inhibit the use of the standard system directories
5808 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5815 @section Specifying subprocesses and the switches to pass to them
5818 @command{gcc} is a driver program. It performs its job by invoking a
5819 sequence of other programs to do the work of compiling, assembling and
5820 linking. GCC interprets its command-line parameters and uses these to
5821 deduce which programs it should invoke, and which command-line options
5822 it ought to place on their command lines. This behavior is controlled
5823 by @dfn{spec strings}. In most cases there is one spec string for each
5824 program that GCC can invoke, but a few programs have multiple spec
5825 strings to control their behavior. The spec strings built into GCC can
5826 be overridden by using the @option{-specs=} command-line switch to specify
5829 @dfn{Spec files} are plaintext files that are used to construct spec
5830 strings. They consist of a sequence of directives separated by blank
5831 lines. The type of directive is determined by the first non-whitespace
5832 character on the line and it can be one of the following:
5835 @item %@var{command}
5836 Issues a @var{command} to the spec file processor. The commands that can
5840 @item %include <@var{file}>
5842 Search for @var{file} and insert its text at the current point in the
5845 @item %include_noerr <@var{file}>
5846 @cindex %include_noerr
5847 Just like @samp{%include}, but do not generate an error message if the include
5848 file cannot be found.
5850 @item %rename @var{old_name} @var{new_name}
5852 Rename the spec string @var{old_name} to @var{new_name}.
5856 @item *[@var{spec_name}]:
5857 This tells the compiler to create, override or delete the named spec
5858 string. All lines after this directive up to the next directive or
5859 blank line are considered to be the text for the spec string. If this
5860 results in an empty string then the spec will be deleted. (Or, if the
5861 spec did not exist, then nothing will happened.) Otherwise, if the spec
5862 does not currently exist a new spec will be created. If the spec does
5863 exist then its contents will be overridden by the text of this
5864 directive, unless the first character of that text is the @samp{+}
5865 character, in which case the text will be appended to the spec.
5867 @item [@var{suffix}]:
5868 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5869 and up to the next directive or blank line are considered to make up the
5870 spec string for the indicated suffix. When the compiler encounters an
5871 input file with the named suffix, it will processes the spec string in
5872 order to work out how to compile that file. For example:
5879 This says that any input file whose name ends in @samp{.ZZ} should be
5880 passed to the program @samp{z-compile}, which should be invoked with the
5881 command-line switch @option{-input} and with the result of performing the
5882 @samp{%i} substitution. (See below.)
5884 As an alternative to providing a spec string, the text that follows a
5885 suffix directive can be one of the following:
5888 @item @@@var{language}
5889 This says that the suffix is an alias for a known @var{language}. This is
5890 similar to using the @option{-x} command-line switch to GCC to specify a
5891 language explicitly. For example:
5898 Says that .ZZ files are, in fact, C++ source files.
5901 This causes an error messages saying:
5904 @var{name} compiler not installed on this system.
5908 GCC already has an extensive list of suffixes built into it.
5909 This directive will add an entry to the end of the list of suffixes, but
5910 since the list is searched from the end backwards, it is effectively
5911 possible to override earlier entries using this technique.
5915 GCC has the following spec strings built into it. Spec files can
5916 override these strings or create their own. Note that individual
5917 targets can also add their own spec strings to this list.
5920 asm Options to pass to the assembler
5921 asm_final Options to pass to the assembler post-processor
5922 cpp Options to pass to the C preprocessor
5923 cc1 Options to pass to the C compiler
5924 cc1plus Options to pass to the C++ compiler
5925 endfile Object files to include at the end of the link
5926 link Options to pass to the linker
5927 lib Libraries to include on the command line to the linker
5928 libgcc Decides which GCC support library to pass to the linker
5929 linker Sets the name of the linker
5930 predefines Defines to be passed to the C preprocessor
5931 signed_char Defines to pass to CPP to say whether @code{char} is signed
5933 startfile Object files to include at the start of the link
5936 Here is a small example of a spec file:
5942 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5945 This example renames the spec called @samp{lib} to @samp{old_lib} and
5946 then overrides the previous definition of @samp{lib} with a new one.
5947 The new definition adds in some extra command-line options before
5948 including the text of the old definition.
5950 @dfn{Spec strings} are a list of command-line options to be passed to their
5951 corresponding program. In addition, the spec strings can contain
5952 @samp{%}-prefixed sequences to substitute variable text or to
5953 conditionally insert text into the command line. Using these constructs
5954 it is possible to generate quite complex command lines.
5956 Here is a table of all defined @samp{%}-sequences for spec
5957 strings. Note that spaces are not generated automatically around the
5958 results of expanding these sequences. Therefore you can concatenate them
5959 together or combine them with constant text in a single argument.
5963 Substitute one @samp{%} into the program name or argument.
5966 Substitute the name of the input file being processed.
5969 Substitute the basename of the input file being processed.
5970 This is the substring up to (and not including) the last period
5971 and not including the directory.
5974 This is the same as @samp{%b}, but include the file suffix (text after
5978 Marks the argument containing or following the @samp{%d} as a
5979 temporary file name, so that that file will be deleted if GCC exits
5980 successfully. Unlike @samp{%g}, this contributes no text to the
5983 @item %g@var{suffix}
5984 Substitute a file name that has suffix @var{suffix} and is chosen
5985 once per compilation, and mark the argument in the same way as
5986 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5987 name is now chosen in a way that is hard to predict even when previously
5988 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5989 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5990 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5991 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5992 was simply substituted with a file name chosen once per compilation,
5993 without regard to any appended suffix (which was therefore treated
5994 just like ordinary text), making such attacks more likely to succeed.
5996 @item %u@var{suffix}
5997 Like @samp{%g}, but generates a new temporary file name even if
5998 @samp{%u@var{suffix}} was already seen.
6000 @item %U@var{suffix}
6001 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6002 new one if there is no such last file name. In the absence of any
6003 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6004 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6005 would involve the generation of two distinct file names, one
6006 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6007 simply substituted with a file name chosen for the previous @samp{%u},
6008 without regard to any appended suffix.
6010 @item %j@var{suffix}
6011 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6012 writable, and if save-temps is off; otherwise, substitute the name
6013 of a temporary file, just like @samp{%u}. This temporary file is not
6014 meant for communication between processes, but rather as a junk
6017 @item %|@var{suffix}
6018 @itemx %m@var{suffix}
6019 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6020 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6021 all. These are the two most common ways to instruct a program that it
6022 should read from standard input or write to standard output. If you
6023 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6024 construct: see for example @file{f/lang-specs.h}.
6026 @item %.@var{SUFFIX}
6027 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6028 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6029 terminated by the next space or %.
6032 Marks the argument containing or following the @samp{%w} as the
6033 designated output file of this compilation. This puts the argument
6034 into the sequence of arguments that @samp{%o} will substitute later.
6037 Substitutes the names of all the output files, with spaces
6038 automatically placed around them. You should write spaces
6039 around the @samp{%o} as well or the results are undefined.
6040 @samp{%o} is for use in the specs for running the linker.
6041 Input files whose names have no recognized suffix are not compiled
6042 at all, but they are included among the output files, so they will
6046 Substitutes the suffix for object files. Note that this is
6047 handled specially when it immediately follows @samp{%g, %u, or %U},
6048 because of the need for those to form complete file names. The
6049 handling is such that @samp{%O} is treated exactly as if it had already
6050 been substituted, except that @samp{%g, %u, and %U} do not currently
6051 support additional @var{suffix} characters following @samp{%O} as they would
6052 following, for example, @samp{.o}.
6055 Substitutes the standard macro predefinitions for the
6056 current target machine. Use this when running @code{cpp}.
6059 Like @samp{%p}, but puts @samp{__} before and after the name of each
6060 predefined macro, except for macros that start with @samp{__} or with
6061 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6065 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6066 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
6067 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6071 Current argument is the name of a library or startup file of some sort.
6072 Search for that file in a standard list of directories and substitute
6073 the full name found.
6076 Print @var{str} as an error message. @var{str} is terminated by a newline.
6077 Use this when inconsistent options are detected.
6080 Substitute the contents of spec string @var{name} at this point.
6083 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6085 @item %x@{@var{option}@}
6086 Accumulate an option for @samp{%X}.
6089 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6093 Output the accumulated assembler options specified by @option{-Wa}.
6096 Output the accumulated preprocessor options specified by @option{-Wp}.
6099 Process the @code{asm} spec. This is used to compute the
6100 switches to be passed to the assembler.
6103 Process the @code{asm_final} spec. This is a spec string for
6104 passing switches to an assembler post-processor, if such a program is
6108 Process the @code{link} spec. This is the spec for computing the
6109 command line passed to the linker. Typically it will make use of the
6110 @samp{%L %G %S %D and %E} sequences.
6113 Dump out a @option{-L} option for each directory that GCC believes might
6114 contain startup files. If the target supports multilibs then the
6115 current multilib directory will be prepended to each of these paths.
6118 Output the multilib directory with directory separators replaced with
6119 @samp{_}. If multilib directories are not set, or the multilib directory is
6120 @file{.} then this option emits nothing.
6123 Process the @code{lib} spec. This is a spec string for deciding which
6124 libraries should be included on the command line to the linker.
6127 Process the @code{libgcc} spec. This is a spec string for deciding
6128 which GCC support library should be included on the command line to the linker.
6131 Process the @code{startfile} spec. This is a spec for deciding which
6132 object files should be the first ones passed to the linker. Typically
6133 this might be a file named @file{crt0.o}.
6136 Process the @code{endfile} spec. This is a spec string that specifies
6137 the last object files that will be passed to the linker.
6140 Process the @code{cpp} spec. This is used to construct the arguments
6141 to be passed to the C preprocessor.
6144 Process the @code{cc1} spec. This is used to construct the options to be
6145 passed to the actual C compiler (@samp{cc1}).
6148 Process the @code{cc1plus} spec. This is used to construct the options to be
6149 passed to the actual C++ compiler (@samp{cc1plus}).
6152 Substitute the variable part of a matched option. See below.
6153 Note that each comma in the substituted string is replaced by
6157 Remove all occurrences of @code{-S} from the command line. Note---this
6158 command is position dependent. @samp{%} commands in the spec string
6159 before this one will see @code{-S}, @samp{%} commands in the spec string
6160 after this one will not.
6162 @item %:@var{function}(@var{args})
6163 Call the named function @var{function}, passing it @var{args}.
6164 @var{args} is first processed as a nested spec string, then split
6165 into an argument vector in the usual fashion. The function returns
6166 a string which is processed as if it had appeared literally as part
6167 of the current spec.
6169 The following built-in spec functions are provided:
6172 @item @code{if-exists}
6173 The @code{if-exists} spec function takes one argument, an absolute
6174 pathname to a file. If the file exists, @code{if-exists} returns the
6175 pathname. Here is a small example of its usage:
6179 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6182 @item @code{if-exists-else}
6183 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6184 spec function, except that it takes two arguments. The first argument is
6185 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6186 returns the pathname. If it does not exist, it returns the second argument.
6187 This way, @code{if-exists-else} can be used to select one file or another,
6188 based on the existence of the first. Here is a small example of its usage:
6192 crt0%O%s %:if-exists(crti%O%s) \
6193 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6198 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6199 If that switch was not specified, this substitutes nothing. Note that
6200 the leading dash is omitted when specifying this option, and it is
6201 automatically inserted if the substitution is performed. Thus the spec
6202 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6203 and would output the command line option @option{-foo}.
6205 @item %W@{@code{S}@}
6206 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6209 @item %@{@code{S}*@}
6210 Substitutes all the switches specified to GCC whose names start
6211 with @code{-S}, but which also take an argument. This is used for
6212 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6213 GCC considers @option{-o foo} as being
6214 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6215 text, including the space. Thus two arguments would be generated.
6217 @item %@{@code{S}*&@code{T}*@}
6218 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6219 (the order of @code{S} and @code{T} in the spec is not significant).
6220 There can be any number of ampersand-separated variables; for each the
6221 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6223 @item %@{@code{S}:@code{X}@}
6224 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6226 @item %@{!@code{S}:@code{X}@}
6227 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6229 @item %@{@code{S}*:@code{X}@}
6230 Substitutes @code{X} if one or more switches whose names start with
6231 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6232 once, no matter how many such switches appeared. However, if @code{%*}
6233 appears somewhere in @code{X}, then @code{X} will be substituted once
6234 for each matching switch, with the @code{%*} replaced by the part of
6235 that switch that matched the @code{*}.
6237 @item %@{.@code{S}:@code{X}@}
6238 Substitutes @code{X}, if processing a file with suffix @code{S}.
6240 @item %@{!.@code{S}:@code{X}@}
6241 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6243 @item %@{@code{S}|@code{P}:@code{X}@}
6244 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6245 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6246 although they have a stronger binding than the @samp{|}. If @code{%*}
6247 appears in @code{X}, all of the alternatives must be starred, and only
6248 the first matching alternative is substituted.
6250 For example, a spec string like this:
6253 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6256 will output the following command-line options from the following input
6257 command-line options:
6262 -d fred.c -foo -baz -boggle
6263 -d jim.d -bar -baz -boggle
6266 @item %@{S:X; T:Y; :D@}
6268 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6269 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6270 be as many clauses as you need. This may be combined with @code{.},
6271 @code{!}, @code{|}, and @code{*} as needed.
6276 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6277 construct may contain other nested @samp{%} constructs or spaces, or
6278 even newlines. They are processed as usual, as described above.
6279 Trailing white space in @code{X} is ignored. White space may also
6280 appear anywhere on the left side of the colon in these constructs,
6281 except between @code{.} or @code{*} and the corresponding word.
6283 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6284 handled specifically in these constructs. If another value of
6285 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
6286 @option{-W} switch is found later in the command line, the earlier
6287 switch value is ignored, except with @{@code{S}*@} where @code{S} is
6288 just one letter, which passes all matching options.
6290 The character @samp{|} at the beginning of the predicate text is used to
6291 indicate that a command should be piped to the following command, but
6292 only if @option{-pipe} is specified.
6294 It is built into GCC which switches take arguments and which do not.
6295 (You might think it would be useful to generalize this to allow each
6296 compiler's spec to say which switches take arguments. But this cannot
6297 be done in a consistent fashion. GCC cannot even decide which input
6298 files have been specified without knowing which switches take arguments,
6299 and it must know which input files to compile in order to tell which
6302 GCC also knows implicitly that arguments starting in @option{-l} are to be
6303 treated as compiler output files, and passed to the linker in their
6304 proper position among the other output files.
6306 @c man begin OPTIONS
6308 @node Target Options
6309 @section Specifying Target Machine and Compiler Version
6310 @cindex target options
6311 @cindex cross compiling
6312 @cindex specifying machine version
6313 @cindex specifying compiler version and target machine
6314 @cindex compiler version, specifying
6315 @cindex target machine, specifying
6317 The usual way to run GCC is to run the executable called @file{gcc}, or
6318 @file{<machine>-gcc} when cross-compiling, or
6319 @file{<machine>-gcc-<version>} to run a version other than the one that
6320 was installed last. Sometimes this is inconvenient, so GCC provides
6321 options that will switch to another cross-compiler or version.
6324 @item -b @var{machine}
6326 The argument @var{machine} specifies the target machine for compilation.
6328 The value to use for @var{machine} is the same as was specified as the
6329 machine type when configuring GCC as a cross-compiler. For
6330 example, if a cross-compiler was configured with @samp{configure
6331 i386v}, meaning to compile for an 80386 running System V, then you
6332 would specify @option{-b i386v} to run that cross compiler.
6334 @item -V @var{version}
6336 The argument @var{version} specifies which version of GCC to run.
6337 This is useful when multiple versions are installed. For example,
6338 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
6341 The @option{-V} and @option{-b} options work by running the
6342 @file{<machine>-gcc-<version>} executable, so there's no real reason to
6343 use them if you can just run that directly.
6345 @node Submodel Options
6346 @section Hardware Models and Configurations
6347 @cindex submodel options
6348 @cindex specifying hardware config
6349 @cindex hardware models and configurations, specifying
6350 @cindex machine dependent options
6352 Earlier we discussed the standard option @option{-b} which chooses among
6353 different installed compilers for completely different target
6354 machines, such as VAX vs.@: 68000 vs.@: 80386.
6356 In addition, each of these target machine types can have its own
6357 special options, starting with @samp{-m}, to choose among various
6358 hardware models or configurations---for example, 68010 vs 68020,
6359 floating coprocessor or none. A single installed version of the
6360 compiler can compile for any model or configuration, according to the
6363 Some configurations of the compiler also support additional special
6364 options, usually for compatibility with other compilers on the same
6367 These options are defined by the macro @code{TARGET_SWITCHES} in the
6368 machine description. The default for the options is also defined by
6369 that macro, which enables you to change the defaults.
6371 @c This list is ordered alphanumerically by subsection name.
6372 @c It should be the same order and spelling as these options are listed
6373 @c in Machine Dependent Options
6381 * DEC Alpha Options::
6382 * DEC Alpha/VMS Options::
6386 * i386 and x86-64 Options::
6398 * RS/6000 and PowerPC Options::
6399 * S/390 and zSeries Options::
6402 * System V Options::
6403 * TMS320C3x/C4x Options::
6407 * Xstormy16 Options::
6413 @subsection ARC Options
6416 These options are defined for ARC implementations:
6421 Compile code for little endian mode. This is the default.
6425 Compile code for big endian mode.
6428 @opindex mmangle-cpu
6429 Prepend the name of the cpu to all public symbol names.
6430 In multiple-processor systems, there are many ARC variants with different
6431 instruction and register set characteristics. This flag prevents code
6432 compiled for one cpu to be linked with code compiled for another.
6433 No facility exists for handling variants that are ``almost identical''.
6434 This is an all or nothing option.
6436 @item -mcpu=@var{cpu}
6438 Compile code for ARC variant @var{cpu}.
6439 Which variants are supported depend on the configuration.
6440 All variants support @option{-mcpu=base}, this is the default.
6442 @item -mtext=@var{text-section}
6443 @itemx -mdata=@var{data-section}
6444 @itemx -mrodata=@var{readonly-data-section}
6448 Put functions, data, and readonly data in @var{text-section},
6449 @var{data-section}, and @var{readonly-data-section} respectively
6450 by default. This can be overridden with the @code{section} attribute.
6451 @xref{Variable Attributes}.
6456 @subsection ARM Options
6459 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6463 @item -mabi=@var{name}
6465 Generate code for the specified ABI. Permissible values are: @samp{apcs-gnu},
6466 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6469 @opindex mapcs-frame
6470 Generate a stack frame that is compliant with the ARM Procedure Call
6471 Standard for all functions, even if this is not strictly necessary for
6472 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6473 with this option will cause the stack frames not to be generated for
6474 leaf functions. The default is @option{-mno-apcs-frame}.
6478 This is a synonym for @option{-mapcs-frame}.
6481 @c not currently implemented
6482 @item -mapcs-stack-check
6483 @opindex mapcs-stack-check
6484 Generate code to check the amount of stack space available upon entry to
6485 every function (that actually uses some stack space). If there is
6486 insufficient space available then either the function
6487 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6488 called, depending upon the amount of stack space required. The run time
6489 system is required to provide these functions. The default is
6490 @option{-mno-apcs-stack-check}, since this produces smaller code.
6492 @c not currently implemented
6494 @opindex mapcs-float
6495 Pass floating point arguments using the float point registers. This is
6496 one of the variants of the APCS@. This option is recommended if the
6497 target hardware has a floating point unit or if a lot of floating point
6498 arithmetic is going to be performed by the code. The default is
6499 @option{-mno-apcs-float}, since integer only code is slightly increased in
6500 size if @option{-mapcs-float} is used.
6502 @c not currently implemented
6503 @item -mapcs-reentrant
6504 @opindex mapcs-reentrant
6505 Generate reentrant, position independent code. The default is
6506 @option{-mno-apcs-reentrant}.
6509 @item -mthumb-interwork
6510 @opindex mthumb-interwork
6511 Generate code which supports calling between the ARM and Thumb
6512 instruction sets. Without this option the two instruction sets cannot
6513 be reliably used inside one program. The default is
6514 @option{-mno-thumb-interwork}, since slightly larger code is generated
6515 when @option{-mthumb-interwork} is specified.
6517 @item -mno-sched-prolog
6518 @opindex mno-sched-prolog
6519 Prevent the reordering of instructions in the function prolog, or the
6520 merging of those instruction with the instructions in the function's
6521 body. This means that all functions will start with a recognizable set
6522 of instructions (or in fact one of a choice from a small set of
6523 different function prologues), and this information can be used to
6524 locate the start if functions inside an executable piece of code. The
6525 default is @option{-msched-prolog}.
6528 @opindex mhard-float
6529 Generate output containing floating point instructions. This is the
6533 @opindex msoft-float
6534 Generate output containing library calls for floating point.
6535 @strong{Warning:} the requisite libraries are not available for all ARM
6536 targets. Normally the facilities of the machine's usual C compiler are
6537 used, but this cannot be done directly in cross-compilation. You must make
6538 your own arrangements to provide suitable library functions for
6541 @option{-msoft-float} changes the calling convention in the output file;
6542 therefore, it is only useful if you compile @emph{all} of a program with
6543 this option. In particular, you need to compile @file{libgcc.a}, the
6544 library that comes with GCC, with @option{-msoft-float} in order for
6547 @item -mfloat-abi=@var{name}
6549 Specifies which ABI to use for floating point values. Permissible values
6550 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6552 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6553 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6554 of floating point instructions, but still uses the soft-float calling
6557 @item -mlittle-endian
6558 @opindex mlittle-endian
6559 Generate code for a processor running in little-endian mode. This is
6560 the default for all standard configurations.
6563 @opindex mbig-endian
6564 Generate code for a processor running in big-endian mode; the default is
6565 to compile code for a little-endian processor.
6567 @item -mwords-little-endian
6568 @opindex mwords-little-endian
6569 This option only applies when generating code for big-endian processors.
6570 Generate code for a little-endian word order but a big-endian byte
6571 order. That is, a byte order of the form @samp{32107654}. Note: this
6572 option should only be used if you require compatibility with code for
6573 big-endian ARM processors generated by versions of the compiler prior to
6576 @item -mcpu=@var{name}
6578 This specifies the name of the target ARM processor. GCC uses this name
6579 to determine what kind of instructions it can emit when generating
6580 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6581 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6582 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6583 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6584 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6585 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6586 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6587 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6588 @samp{arm920t}, @samp{arm922t}, @samp{arm946es}, @samp{arm966es},
6589 @samp{arm968es}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6590 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6591 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
6592 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6595 @itemx -mtune=@var{name}
6597 This option is very similar to the @option{-mcpu=} option, except that
6598 instead of specifying the actual target processor type, and hence
6599 restricting which instructions can be used, it specifies that GCC should
6600 tune the performance of the code as if the target were of the type
6601 specified in this option, but still choosing the instructions that it
6602 will generate based on the cpu specified by a @option{-mcpu=} option.
6603 For some ARM implementations better performance can be obtained by using
6606 @item -march=@var{name}
6608 This specifies the name of the target ARM architecture. GCC uses this
6609 name to determine what kind of instructions it can emit when generating
6610 assembly code. This option can be used in conjunction with or instead
6611 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6612 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6613 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6614 @samp{iwmmxt}, @samp{ep9312}.
6616 @item -mfpu=@var{name}
6617 @itemx -mfpe=@var{number}
6618 @itemx -mfp=@var{number}
6622 This specifies what floating point hardware (or hardware emulation) is
6623 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6624 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6625 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6626 with older versions of GCC@.
6628 If @option{-msoft-float} is specified this specifies the format of
6629 floating point values.
6631 @item -mstructure-size-boundary=@var{n}
6632 @opindex mstructure-size-boundary
6633 The size of all structures and unions will be rounded up to a multiple
6634 of the number of bits set by this option. Permissible values are 8, 32
6635 and 64. The default value varies for different toolchains. For the COFF
6636 targeted toolchain the default value is 8. A value of 64 is only allowed
6637 if the underlying ABI supports it.
6639 Specifying the larger number can produce faster, more efficient code, but
6640 can also increase the size of the program. Different values are potentially
6641 incompatible. Code compiled with one value cannot necessarily expect to
6642 work with code or libraries compiled with another value, if they exchange
6643 information using structures or unions.
6645 @item -mabort-on-noreturn
6646 @opindex mabort-on-noreturn
6647 Generate a call to the function @code{abort} at the end of a
6648 @code{noreturn} function. It will be executed if the function tries to
6652 @itemx -mno-long-calls
6653 @opindex mlong-calls
6654 @opindex mno-long-calls
6655 Tells the compiler to perform function calls by first loading the
6656 address of the function into a register and then performing a subroutine
6657 call on this register. This switch is needed if the target function
6658 will lie outside of the 64 megabyte addressing range of the offset based
6659 version of subroutine call instruction.
6661 Even if this switch is enabled, not all function calls will be turned
6662 into long calls. The heuristic is that static functions, functions
6663 which have the @samp{short-call} attribute, functions that are inside
6664 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6665 definitions have already been compiled within the current compilation
6666 unit, will not be turned into long calls. The exception to this rule is
6667 that weak function definitions, functions with the @samp{long-call}
6668 attribute or the @samp{section} attribute, and functions that are within
6669 the scope of a @samp{#pragma long_calls} directive, will always be
6670 turned into long calls.
6672 This feature is not enabled by default. Specifying
6673 @option{-mno-long-calls} will restore the default behavior, as will
6674 placing the function calls within the scope of a @samp{#pragma
6675 long_calls_off} directive. Note these switches have no effect on how
6676 the compiler generates code to handle function calls via function
6679 @item -mnop-fun-dllimport
6680 @opindex mnop-fun-dllimport
6681 Disable support for the @code{dllimport} attribute.
6683 @item -msingle-pic-base
6684 @opindex msingle-pic-base
6685 Treat the register used for PIC addressing as read-only, rather than
6686 loading it in the prologue for each function. The run-time system is
6687 responsible for initializing this register with an appropriate value
6688 before execution begins.
6690 @item -mpic-register=@var{reg}
6691 @opindex mpic-register
6692 Specify the register to be used for PIC addressing. The default is R10
6693 unless stack-checking is enabled, when R9 is used.
6695 @item -mcirrus-fix-invalid-insns
6696 @opindex mcirrus-fix-invalid-insns
6697 @opindex mno-cirrus-fix-invalid-insns
6698 Insert NOPs into the instruction stream to in order to work around
6699 problems with invalid Maverick instruction combinations. This option
6700 is only valid if the @option{-mcpu=ep9312} option has been used to
6701 enable generation of instructions for the Cirrus Maverick floating
6702 point co-processor. This option is not enabled by default, since the
6703 problem is only present in older Maverick implementations. The default
6704 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6707 @item -mpoke-function-name
6708 @opindex mpoke-function-name
6709 Write the name of each function into the text section, directly
6710 preceding the function prologue. The generated code is similar to this:
6714 .ascii "arm_poke_function_name", 0
6717 .word 0xff000000 + (t1 - t0)
6718 arm_poke_function_name
6720 stmfd sp!, @{fp, ip, lr, pc@}
6724 When performing a stack backtrace, code can inspect the value of
6725 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6726 location @code{pc - 12} and the top 8 bits are set, then we know that
6727 there is a function name embedded immediately preceding this location
6728 and has length @code{((pc[-3]) & 0xff000000)}.
6732 Generate code for the 16-bit Thumb instruction set. The default is to
6733 use the 32-bit ARM instruction set.
6736 @opindex mtpcs-frame
6737 Generate a stack frame that is compliant with the Thumb Procedure Call
6738 Standard for all non-leaf functions. (A leaf function is one that does
6739 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6741 @item -mtpcs-leaf-frame
6742 @opindex mtpcs-leaf-frame
6743 Generate a stack frame that is compliant with the Thumb Procedure Call
6744 Standard for all leaf functions. (A leaf function is one that does
6745 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6747 @item -mcallee-super-interworking
6748 @opindex mcallee-super-interworking
6749 Gives all externally visible functions in the file being compiled an ARM
6750 instruction set header which switches to Thumb mode before executing the
6751 rest of the function. This allows these functions to be called from
6752 non-interworking code.
6754 @item -mcaller-super-interworking
6755 @opindex mcaller-super-interworking
6756 Allows calls via function pointers (including virtual functions) to
6757 execute correctly regardless of whether the target code has been
6758 compiled for interworking or not. There is a small overhead in the cost
6759 of executing a function pointer if this option is enabled.
6764 @subsection AVR Options
6767 These options are defined for AVR implementations:
6770 @item -mmcu=@var{mcu}
6772 Specify ATMEL AVR instruction set or MCU type.
6774 Instruction set avr1 is for the minimal AVR core, not supported by the C
6775 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
6776 attiny11, attiny12, attiny15, attiny28).
6778 Instruction set avr2 (default) is for the classic AVR core with up to
6779 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
6780 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
6781 at90c8534, at90s8535).
6783 Instruction set avr3 is for the classic AVR core with up to 128K program
6784 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
6786 Instruction set avr4 is for the enhanced AVR core with up to 8K program
6787 memory space (MCU types: atmega8, atmega83, atmega85).
6789 Instruction set avr5 is for the enhanced AVR core with up to 128K program
6790 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
6791 atmega64, atmega128, at43usb355, at94k).
6795 Output instruction sizes to the asm file.
6797 @item -minit-stack=@var{N}
6798 @opindex minit-stack
6799 Specify the initial stack address, which may be a symbol or numeric value,
6800 @samp{__stack} is the default.
6802 @item -mno-interrupts
6803 @opindex mno-interrupts
6804 Generated code is not compatible with hardware interrupts.
6805 Code size will be smaller.
6807 @item -mcall-prologues
6808 @opindex mcall-prologues
6809 Functions prologues/epilogues expanded as call to appropriate
6810 subroutines. Code size will be smaller.
6812 @item -mno-tablejump
6813 @opindex mno-tablejump
6814 Do not generate tablejump insns which sometimes increase code size.
6817 @opindex mtiny-stack
6818 Change only the low 8 bits of the stack pointer.
6822 Assume int to be 8 bit integer. This affects the sizes of all types: A
6823 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
6824 and long long will be 4 bytes. Please note that this option does not
6825 comply to the C standards, but it will provide you with smaller code
6830 @subsection CRIS Options
6831 @cindex CRIS Options
6833 These options are defined specifically for the CRIS ports.
6836 @item -march=@var{architecture-type}
6837 @itemx -mcpu=@var{architecture-type}
6840 Generate code for the specified architecture. The choices for
6841 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
6842 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
6843 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
6846 @item -mtune=@var{architecture-type}
6848 Tune to @var{architecture-type} everything applicable about the generated
6849 code, except for the ABI and the set of available instructions. The
6850 choices for @var{architecture-type} are the same as for
6851 @option{-march=@var{architecture-type}}.
6853 @item -mmax-stack-frame=@var{n}
6854 @opindex mmax-stack-frame
6855 Warn when the stack frame of a function exceeds @var{n} bytes.
6857 @item -melinux-stacksize=@var{n}
6858 @opindex melinux-stacksize
6859 Only available with the @samp{cris-axis-aout} target. Arranges for
6860 indications in the program to the kernel loader that the stack of the
6861 program should be set to @var{n} bytes.
6867 The options @option{-metrax4} and @option{-metrax100} are synonyms for
6868 @option{-march=v3} and @option{-march=v8} respectively.
6870 @item -mmul-bug-workaround
6871 @itemx -mno-mul-bug-workaround
6872 @opindex mmul-bug-workaround
6873 @opindex mno-mul-bug-workaround
6874 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
6875 models where it applies. This option is active by default.
6879 Enable CRIS-specific verbose debug-related information in the assembly
6880 code. This option also has the effect to turn off the @samp{#NO_APP}
6881 formatted-code indicator to the assembler at the beginning of the
6886 Do not use condition-code results from previous instruction; always emit
6887 compare and test instructions before use of condition codes.
6889 @item -mno-side-effects
6890 @opindex mno-side-effects
6891 Do not emit instructions with side-effects in addressing modes other than
6895 @itemx -mno-stack-align
6897 @itemx -mno-data-align
6898 @itemx -mconst-align
6899 @itemx -mno-const-align
6900 @opindex mstack-align
6901 @opindex mno-stack-align
6902 @opindex mdata-align
6903 @opindex mno-data-align
6904 @opindex mconst-align
6905 @opindex mno-const-align
6906 These options (no-options) arranges (eliminate arrangements) for the
6907 stack-frame, individual data and constants to be aligned for the maximum
6908 single data access size for the chosen CPU model. The default is to
6909 arrange for 32-bit alignment. ABI details such as structure layout are
6910 not affected by these options.
6918 Similar to the stack- data- and const-align options above, these options
6919 arrange for stack-frame, writable data and constants to all be 32-bit,
6920 16-bit or 8-bit aligned. The default is 32-bit alignment.
6922 @item -mno-prologue-epilogue
6923 @itemx -mprologue-epilogue
6924 @opindex mno-prologue-epilogue
6925 @opindex mprologue-epilogue
6926 With @option{-mno-prologue-epilogue}, the normal function prologue and
6927 epilogue that sets up the stack-frame are omitted and no return
6928 instructions or return sequences are generated in the code. Use this
6929 option only together with visual inspection of the compiled code: no
6930 warnings or errors are generated when call-saved registers must be saved,
6931 or storage for local variable needs to be allocated.
6937 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
6938 instruction sequences that load addresses for functions from the PLT part
6939 of the GOT rather than (traditional on other architectures) calls to the
6940 PLT. The default is @option{-mgotplt}.
6944 Legacy no-op option only recognized with the cris-axis-aout target.
6948 Legacy no-op option only recognized with the cris-axis-elf and
6949 cris-axis-linux-gnu targets.
6953 Only recognized with the cris-axis-aout target, where it selects a
6954 GNU/linux-like multilib, include files and instruction set for
6959 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
6963 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
6964 to link with input-output functions from a simulator library. Code,
6965 initialized data and zero-initialized data are allocated consecutively.
6969 Like @option{-sim}, but pass linker options to locate initialized data at
6970 0x40000000 and zero-initialized data at 0x80000000.
6973 @node Darwin Options
6974 @subsection Darwin Options
6975 @cindex Darwin options
6977 These options are defined for all architectures running the Darwin operating
6978 system. They are useful for compatibility with other Mac OS compilers.
6983 Add the framework directory @var{dir} to the head of the list of
6984 directories to be searched for header files. These directories are
6985 interleaved with those specified by @option{-I} options and are
6986 scanned in a left-to-right order.
6988 A framework directory is a directory with frameworks in it. A
6989 framework is a directory with a @samp{"Headers"} and/or
6990 @samp{"PrivateHeaders"} directory contained directly in it that ends
6991 in @samp{".framework"}. The name of a framework is the name of this
6992 directory excluding the @samp{".framework"}. Headers associated with
6993 the framework are found in one of those two directories, with
6994 @samp{"Headers"} being searched first. A subframework is a framework
6995 directory that is in a framework's @samp{"Frameworks"} directory.
6996 Includes of subframework headers can only appear in a header of a
6997 framework that contains the subframework, or in a sibling subframework
6998 header. Two subframeworks are siblings if they occur in the same
6999 framework. A subframework should not have the same name as a
7000 framework, a warning will be issued if this is violated. Currently a
7001 subframework cannot have subframeworks, in the future, the mechanism
7002 may be extended to support this. The standard frameworks can be found
7003 in @samp{"/System/Library/Frameworks"} and
7004 @samp{"/Library/Frameworks"}. An example include looks like
7005 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7006 the name of the framework and header.h is found in the
7007 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7011 Emit debugging information for symbols that are used. For STABS
7012 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7013 This is by default ON.
7017 Emit debugging information for all symbols and types.
7019 @item -mone-byte-bool
7020 @opindex -mone-byte-bool
7021 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
7022 By default @samp{sizeof(bool)} is @samp{4} when compiling for
7023 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
7024 option has no effect on x86.
7026 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
7027 to generate code that is not binary compatible with code generated
7028 without that switch. Using this switch may require recompiling all
7029 other modules in a program, including system libraries. Use this
7030 switch to conform to a non-default data model.
7032 @item -mfix-and-continue
7033 @itemx -ffix-and-continue
7034 @itemx -findirect-data
7035 @opindex mfix-and-continue
7036 @opindex ffix-and-continue
7037 @opindex findirect-data
7038 Generate code suitable for fast turn around development. Needed to
7039 enable gdb to dynamically load @code{.o} files into already running
7040 programs. @option{-findirect-data} and @option{-ffix-and-continue}
7041 are provided for backwards compatibility.
7045 Loads all members of static archive libraries.
7046 See man ld(1) for more information.
7048 @item -arch_errors_fatal
7049 @opindex arch_errors_fatal
7050 Cause the errors having to do with files that have the wrong architecture
7054 @opindex bind_at_load
7055 Causes the output file to be marked such that the dynamic linker will
7056 bind all undefined references when the file is loaded or launched.
7060 Produce a Mach-o bundle format file.
7061 See man ld(1) for more information.
7063 @item -bundle_loader @var{executable}
7064 @opindex bundle_loader
7065 This specifies the @var{executable} that will be loading the build
7066 output file being linked. See man ld(1) for more information.
7068 @item -allowable_client @var{client_name}
7072 @itemx -compatibility_version
7073 @itemx -current_version
7075 @itemx -dependency-file
7077 @itemx -dylinker_install_name
7080 @itemx -exported_symbols_list
7082 @itemx -flat_namespace
7083 @itemx -force_cpusubtype_ALL
7084 @itemx -force_flat_namespace
7085 @itemx -headerpad_max_install_names
7088 @itemx -install_name
7089 @itemx -keep_private_externs
7090 @itemx -multi_module
7091 @itemx -multiply_defined
7092 @itemx -multiply_defined_unused
7094 @itemx -no_dead_strip_inits_and_terms
7095 @itemx -nofixprebinding
7098 @itemx -noseglinkedit
7099 @itemx -pagezero_size
7101 @itemx -prebind_all_twolevel_modules
7102 @itemx -private_bundle
7103 @itemx -read_only_relocs
7105 @itemx -sectobjectsymbols
7109 @itemx -sectobjectsymbols
7111 @itemx -seg_addr_table
7112 @itemx -seg_addr_table_filename
7115 @itemx -segs_read_only_addr
7116 @itemx -segs_read_write_addr
7117 @itemx -single_module
7120 @itemx -sub_umbrella
7121 @itemx -twolevel_namespace
7124 @itemx -unexported_symbols_list
7125 @itemx -weak_reference_mismatches
7128 @opindex allowable_client
7130 @opindex client_name
7131 @opindex compatibility_version
7132 @opindex current_version
7134 @opindex dependency-file
7136 @opindex dylinker_install_name
7139 @opindex exported_symbols_list
7141 @opindex flat_namespace
7142 @opindex force_cpusubtype_ALL
7143 @opindex force_flat_namespace
7144 @opindex headerpad_max_install_names
7147 @opindex install_name
7148 @opindex keep_private_externs
7149 @opindex multi_module
7150 @opindex multiply_defined
7151 @opindex multiply_defined_unused
7153 @opindex no_dead_strip_inits_and_terms
7154 @opindex nofixprebinding
7155 @opindex nomultidefs
7157 @opindex noseglinkedit
7158 @opindex pagezero_size
7160 @opindex prebind_all_twolevel_modules
7161 @opindex private_bundle
7162 @opindex read_only_relocs
7164 @opindex sectobjectsymbols
7168 @opindex sectobjectsymbols
7170 @opindex seg_addr_table
7171 @opindex seg_addr_table_filename
7172 @opindex seglinkedit
7174 @opindex segs_read_only_addr
7175 @opindex segs_read_write_addr
7176 @opindex single_module
7178 @opindex sub_library
7179 @opindex sub_umbrella
7180 @opindex twolevel_namespace
7183 @opindex unexported_symbols_list
7184 @opindex weak_reference_mismatches
7185 @opindex whatsloaded
7187 These options are available for Darwin linker. Darwin linker man page
7188 describes them in detail.
7191 @node DEC Alpha Options
7192 @subsection DEC Alpha Options
7194 These @samp{-m} options are defined for the DEC Alpha implementations:
7197 @item -mno-soft-float
7199 @opindex mno-soft-float
7200 @opindex msoft-float
7201 Use (do not use) the hardware floating-point instructions for
7202 floating-point operations. When @option{-msoft-float} is specified,
7203 functions in @file{libgcc.a} will be used to perform floating-point
7204 operations. Unless they are replaced by routines that emulate the
7205 floating-point operations, or compiled in such a way as to call such
7206 emulations routines, these routines will issue floating-point
7207 operations. If you are compiling for an Alpha without floating-point
7208 operations, you must ensure that the library is built so as not to call
7211 Note that Alpha implementations without floating-point operations are
7212 required to have floating-point registers.
7217 @opindex mno-fp-regs
7218 Generate code that uses (does not use) the floating-point register set.
7219 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7220 register set is not used, floating point operands are passed in integer
7221 registers as if they were integers and floating-point results are passed
7222 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
7223 so any function with a floating-point argument or return value called by code
7224 compiled with @option{-mno-fp-regs} must also be compiled with that
7227 A typical use of this option is building a kernel that does not use,
7228 and hence need not save and restore, any floating-point registers.
7232 The Alpha architecture implements floating-point hardware optimized for
7233 maximum performance. It is mostly compliant with the IEEE floating
7234 point standard. However, for full compliance, software assistance is
7235 required. This option generates code fully IEEE compliant code
7236 @emph{except} that the @var{inexact-flag} is not maintained (see below).
7237 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
7238 defined during compilation. The resulting code is less efficient but is
7239 able to correctly support denormalized numbers and exceptional IEEE
7240 values such as not-a-number and plus/minus infinity. Other Alpha
7241 compilers call this option @option{-ieee_with_no_inexact}.
7243 @item -mieee-with-inexact
7244 @opindex mieee-with-inexact
7245 This is like @option{-mieee} except the generated code also maintains
7246 the IEEE @var{inexact-flag}. Turning on this option causes the
7247 generated code to implement fully-compliant IEEE math. In addition to
7248 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
7249 macro. On some Alpha implementations the resulting code may execute
7250 significantly slower than the code generated by default. Since there is
7251 very little code that depends on the @var{inexact-flag}, you should
7252 normally not specify this option. Other Alpha compilers call this
7253 option @option{-ieee_with_inexact}.
7255 @item -mfp-trap-mode=@var{trap-mode}
7256 @opindex mfp-trap-mode
7257 This option controls what floating-point related traps are enabled.
7258 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
7259 The trap mode can be set to one of four values:
7263 This is the default (normal) setting. The only traps that are enabled
7264 are the ones that cannot be disabled in software (e.g., division by zero
7268 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7272 Like @samp{su}, but the instructions are marked to be safe for software
7273 completion (see Alpha architecture manual for details).
7276 Like @samp{su}, but inexact traps are enabled as well.
7279 @item -mfp-rounding-mode=@var{rounding-mode}
7280 @opindex mfp-rounding-mode
7281 Selects the IEEE rounding mode. Other Alpha compilers call this option
7282 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
7287 Normal IEEE rounding mode. Floating point numbers are rounded towards
7288 the nearest machine number or towards the even machine number in case
7292 Round towards minus infinity.
7295 Chopped rounding mode. Floating point numbers are rounded towards zero.
7298 Dynamic rounding mode. A field in the floating point control register
7299 (@var{fpcr}, see Alpha architecture reference manual) controls the
7300 rounding mode in effect. The C library initializes this register for
7301 rounding towards plus infinity. Thus, unless your program modifies the
7302 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7305 @item -mtrap-precision=@var{trap-precision}
7306 @opindex mtrap-precision
7307 In the Alpha architecture, floating point traps are imprecise. This
7308 means without software assistance it is impossible to recover from a
7309 floating trap and program execution normally needs to be terminated.
7310 GCC can generate code that can assist operating system trap handlers
7311 in determining the exact location that caused a floating point trap.
7312 Depending on the requirements of an application, different levels of
7313 precisions can be selected:
7317 Program precision. This option is the default and means a trap handler
7318 can only identify which program caused a floating point exception.
7321 Function precision. The trap handler can determine the function that
7322 caused a floating point exception.
7325 Instruction precision. The trap handler can determine the exact
7326 instruction that caused a floating point exception.
7329 Other Alpha compilers provide the equivalent options called
7330 @option{-scope_safe} and @option{-resumption_safe}.
7332 @item -mieee-conformant
7333 @opindex mieee-conformant
7334 This option marks the generated code as IEEE conformant. You must not
7335 use this option unless you also specify @option{-mtrap-precision=i} and either
7336 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
7337 is to emit the line @samp{.eflag 48} in the function prologue of the
7338 generated assembly file. Under DEC Unix, this has the effect that
7339 IEEE-conformant math library routines will be linked in.
7341 @item -mbuild-constants
7342 @opindex mbuild-constants
7343 Normally GCC examines a 32- or 64-bit integer constant to
7344 see if it can construct it from smaller constants in two or three
7345 instructions. If it cannot, it will output the constant as a literal and
7346 generate code to load it from the data segment at runtime.
7348 Use this option to require GCC to construct @emph{all} integer constants
7349 using code, even if it takes more instructions (the maximum is six).
7351 You would typically use this option to build a shared library dynamic
7352 loader. Itself a shared library, it must relocate itself in memory
7353 before it can find the variables and constants in its own data segment.
7359 Select whether to generate code to be assembled by the vendor-supplied
7360 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
7378 Indicate whether GCC should generate code to use the optional BWX,
7379 CIX, FIX and MAX instruction sets. The default is to use the instruction
7380 sets supported by the CPU type specified via @option{-mcpu=} option or that
7381 of the CPU on which GCC was built if none was specified.
7386 @opindex mfloat-ieee
7387 Generate code that uses (does not use) VAX F and G floating point
7388 arithmetic instead of IEEE single and double precision.
7390 @item -mexplicit-relocs
7391 @itemx -mno-explicit-relocs
7392 @opindex mexplicit-relocs
7393 @opindex mno-explicit-relocs
7394 Older Alpha assemblers provided no way to generate symbol relocations
7395 except via assembler macros. Use of these macros does not allow
7396 optimal instruction scheduling. GNU binutils as of version 2.12
7397 supports a new syntax that allows the compiler to explicitly mark
7398 which relocations should apply to which instructions. This option
7399 is mostly useful for debugging, as GCC detects the capabilities of
7400 the assembler when it is built and sets the default accordingly.
7404 @opindex msmall-data
7405 @opindex mlarge-data
7406 When @option{-mexplicit-relocs} is in effect, static data is
7407 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
7408 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
7409 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
7410 16-bit relocations off of the @code{$gp} register. This limits the
7411 size of the small data area to 64KB, but allows the variables to be
7412 directly accessed via a single instruction.
7414 The default is @option{-mlarge-data}. With this option the data area
7415 is limited to just below 2GB. Programs that require more than 2GB of
7416 data must use @code{malloc} or @code{mmap} to allocate the data in the
7417 heap instead of in the program's data segment.
7419 When generating code for shared libraries, @option{-fpic} implies
7420 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
7424 @opindex msmall-text
7425 @opindex mlarge-text
7426 When @option{-msmall-text} is used, the compiler assumes that the
7427 code of the entire program (or shared library) fits in 4MB, and is
7428 thus reachable with a branch instruction. When @option{-msmall-data}
7429 is used, the compiler can assume that all local symbols share the
7430 same @code{$gp} value, and thus reduce the number of instructions
7431 required for a function call from 4 to 1.
7433 The default is @option{-mlarge-text}.
7435 @item -mcpu=@var{cpu_type}
7437 Set the instruction set and instruction scheduling parameters for
7438 machine type @var{cpu_type}. You can specify either the @samp{EV}
7439 style name or the corresponding chip number. GCC supports scheduling
7440 parameters for the EV4, EV5 and EV6 family of processors and will
7441 choose the default values for the instruction set from the processor
7442 you specify. If you do not specify a processor type, GCC will default
7443 to the processor on which the compiler was built.
7445 Supported values for @var{cpu_type} are
7451 Schedules as an EV4 and has no instruction set extensions.
7455 Schedules as an EV5 and has no instruction set extensions.
7459 Schedules as an EV5 and supports the BWX extension.
7464 Schedules as an EV5 and supports the BWX and MAX extensions.
7468 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
7472 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
7475 @item -mtune=@var{cpu_type}
7477 Set only the instruction scheduling parameters for machine type
7478 @var{cpu_type}. The instruction set is not changed.
7480 @item -mmemory-latency=@var{time}
7481 @opindex mmemory-latency
7482 Sets the latency the scheduler should assume for typical memory
7483 references as seen by the application. This number is highly
7484 dependent on the memory access patterns used by the application
7485 and the size of the external cache on the machine.
7487 Valid options for @var{time} are
7491 A decimal number representing clock cycles.
7497 The compiler contains estimates of the number of clock cycles for
7498 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7499 (also called Dcache, Scache, and Bcache), as well as to main memory.
7500 Note that L3 is only valid for EV5.
7505 @node DEC Alpha/VMS Options
7506 @subsection DEC Alpha/VMS Options
7508 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
7511 @item -mvms-return-codes
7512 @opindex mvms-return-codes
7513 Return VMS condition codes from main. The default is to return POSIX
7514 style condition (e.g.@ error) codes.
7518 @subsection FRV Options
7525 Only use the first 32 general purpose registers.
7530 Use all 64 general purpose registers.
7535 Use only the first 32 floating point registers.
7540 Use all 64 floating point registers
7543 @opindex mhard-float
7545 Use hardware instructions for floating point operations.
7548 @opindex msoft-float
7550 Use library routines for floating point operations.
7555 Dynamically allocate condition code registers.
7560 Do not try to dynamically allocate condition code registers, only
7561 use @code{icc0} and @code{fcc0}.
7566 Change ABI to use double word insns.
7571 Do not use double word instructions.
7576 Use floating point double instructions.
7581 Do not use floating point double instructions.
7586 Use media instructions.
7591 Do not use media instructions.
7596 Use multiply and add/subtract instructions.
7601 Do not use multiply and add/subtract instructions.
7604 @opindex mlibrary-pic
7606 Generate position-independent EABI code.
7611 Use only the first four media accumulator registers.
7616 Use all eight media accumulator registers.
7621 Pack VLIW instructions.
7626 Do not pack VLIW instructions.
7631 Do not mark ABI switches in e_flags.
7636 Enable the use of conditional-move instructions (default).
7638 This switch is mainly for debugging the compiler and will likely be removed
7639 in a future version.
7641 @item -mno-cond-move
7642 @opindex mno-cond-move
7644 Disable the use of conditional-move instructions.
7646 This switch is mainly for debugging the compiler and will likely be removed
7647 in a future version.
7652 Enable the use of conditional set instructions (default).
7654 This switch is mainly for debugging the compiler and will likely be removed
7655 in a future version.
7660 Disable the use of conditional set instructions.
7662 This switch is mainly for debugging the compiler and will likely be removed
7663 in a future version.
7668 Enable the use of conditional execution (default).
7670 This switch is mainly for debugging the compiler and will likely be removed
7671 in a future version.
7673 @item -mno-cond-exec
7674 @opindex mno-cond-exec
7676 Disable the use of conditional execution.
7678 This switch is mainly for debugging the compiler and will likely be removed
7679 in a future version.
7682 @opindex mvliw-branch
7684 Run a pass to pack branches into VLIW instructions (default).
7686 This switch is mainly for debugging the compiler and will likely be removed
7687 in a future version.
7689 @item -mno-vliw-branch
7690 @opindex mno-vliw-branch
7692 Do not run a pass to pack branches into VLIW instructions.
7694 This switch is mainly for debugging the compiler and will likely be removed
7695 in a future version.
7697 @item -mmulti-cond-exec
7698 @opindex mmulti-cond-exec
7700 Enable optimization of @code{&&} and @code{||} in conditional execution
7703 This switch is mainly for debugging the compiler and will likely be removed
7704 in a future version.
7706 @item -mno-multi-cond-exec
7707 @opindex mno-multi-cond-exec
7709 Disable optimization of @code{&&} and @code{||} in conditional execution.
7711 This switch is mainly for debugging the compiler and will likely be removed
7712 in a future version.
7714 @item -mnested-cond-exec
7715 @opindex mnested-cond-exec
7717 Enable nested conditional execution optimizations (default).
7719 This switch is mainly for debugging the compiler and will likely be removed
7720 in a future version.
7722 @item -mno-nested-cond-exec
7723 @opindex mno-nested-cond-exec
7725 Disable nested conditional execution optimizations.
7727 This switch is mainly for debugging the compiler and will likely be removed
7728 in a future version.
7730 @item -mtomcat-stats
7731 @opindex mtomcat-stats
7733 Cause gas to print out tomcat statistics.
7735 @item -mcpu=@var{cpu}
7738 Select the processor type for which to generate code. Possible values are
7739 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
7744 @node H8/300 Options
7745 @subsection H8/300 Options
7747 These @samp{-m} options are defined for the H8/300 implementations:
7752 Shorten some address references at link time, when possible; uses the
7753 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
7754 ld, Using ld}, for a fuller description.
7758 Generate code for the H8/300H@.
7762 Generate code for the H8S@.
7766 Generate code for the H8S and H8/300H in the normal mode. This switch
7767 must be used either with -mh or -ms.
7771 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
7775 Make @code{int} data 32 bits by default.
7779 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
7780 The default for the H8/300H and H8S is to align longs and floats on 4
7782 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
7783 This option has no effect on the H8/300.
7787 @subsection HPPA Options
7788 @cindex HPPA Options
7790 These @samp{-m} options are defined for the HPPA family of computers:
7793 @item -march=@var{architecture-type}
7795 Generate code for the specified architecture. The choices for
7796 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
7797 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
7798 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
7799 architecture option for your machine. Code compiled for lower numbered
7800 architectures will run on higher numbered architectures, but not the
7803 PA 2.0 support currently requires gas snapshot 19990413 or later. The
7804 next release of binutils (current is 2.9.1) will probably contain PA 2.0
7808 @itemx -mpa-risc-1-1
7809 @itemx -mpa-risc-2-0
7810 @opindex mpa-risc-1-0
7811 @opindex mpa-risc-1-1
7812 @opindex mpa-risc-2-0
7813 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
7816 @opindex mbig-switch
7817 Generate code suitable for big switch tables. Use this option only if
7818 the assembler/linker complain about out of range branches within a switch
7821 @item -mjump-in-delay
7822 @opindex mjump-in-delay
7823 Fill delay slots of function calls with unconditional jump instructions
7824 by modifying the return pointer for the function call to be the target
7825 of the conditional jump.
7827 @item -mdisable-fpregs
7828 @opindex mdisable-fpregs
7829 Prevent floating point registers from being used in any manner. This is
7830 necessary for compiling kernels which perform lazy context switching of
7831 floating point registers. If you use this option and attempt to perform
7832 floating point operations, the compiler will abort.
7834 @item -mdisable-indexing
7835 @opindex mdisable-indexing
7836 Prevent the compiler from using indexing address modes. This avoids some
7837 rather obscure problems when compiling MIG generated code under MACH@.
7839 @item -mno-space-regs
7840 @opindex mno-space-regs
7841 Generate code that assumes the target has no space registers. This allows
7842 GCC to generate faster indirect calls and use unscaled index address modes.
7844 Such code is suitable for level 0 PA systems and kernels.
7846 @item -mfast-indirect-calls
7847 @opindex mfast-indirect-calls
7848 Generate code that assumes calls never cross space boundaries. This
7849 allows GCC to emit code which performs faster indirect calls.
7851 This option will not work in the presence of shared libraries or nested
7854 @item -mfixed-range=@var{register-range}
7855 @opindex mfixed-range
7856 Generate code treating the given register range as fixed registers.
7857 A fixed register is one that the register allocator can not use. This is
7858 useful when compiling kernel code. A register range is specified as
7859 two registers separated by a dash. Multiple register ranges can be
7860 specified separated by a comma.
7862 @item -mlong-load-store
7863 @opindex mlong-load-store
7864 Generate 3-instruction load and store sequences as sometimes required by
7865 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
7868 @item -mportable-runtime
7869 @opindex mportable-runtime
7870 Use the portable calling conventions proposed by HP for ELF systems.
7874 Enable the use of assembler directives only GAS understands.
7876 @item -mschedule=@var{cpu-type}
7878 Schedule code according to the constraints for the machine type
7879 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
7880 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
7881 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
7882 proper scheduling option for your machine. The default scheduling is
7886 @opindex mlinker-opt
7887 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
7888 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
7889 linkers in which they give bogus error messages when linking some programs.
7892 @opindex msoft-float
7893 Generate output containing library calls for floating point.
7894 @strong{Warning:} the requisite libraries are not available for all HPPA
7895 targets. Normally the facilities of the machine's usual C compiler are
7896 used, but this cannot be done directly in cross-compilation. You must make
7897 your own arrangements to provide suitable library functions for
7898 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
7899 does provide software floating point support.
7901 @option{-msoft-float} changes the calling convention in the output file;
7902 therefore, it is only useful if you compile @emph{all} of a program with
7903 this option. In particular, you need to compile @file{libgcc.a}, the
7904 library that comes with GCC, with @option{-msoft-float} in order for
7909 Generate the predefine, @code{_SIO}, for server IO. The default is
7910 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
7911 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
7912 options are available under HP-UX and HI-UX.
7916 Use GNU ld specific options. This passes @option{-shared} to ld when
7917 building a shared library. It is the default when GCC is configured,
7918 explicitly or implicitly, with the GNU linker. This option does not
7919 have any affect on which ld is called, it only changes what parameters
7920 are passed to that ld. The ld that is called is determined by the
7921 @option{--with-ld} configure option, GCC's program search path, and
7922 finally by the user's @env{PATH}. The linker used by GCC can be printed
7923 using @samp{which `gcc -print-prog-name=ld`}.
7927 Use HP ld specific options. This passes @option{-b} to ld when building
7928 a shared library and passes @option{+Accept TypeMismatch} to ld on all
7929 links. It is the default when GCC is configured, explicitly or
7930 implicitly, with the HP linker. This option does not have any affect on
7931 which ld is called, it only changes what parameters are passed to that
7932 ld. The ld that is called is determined by the @option{--with-ld}
7933 configure option, GCC's program search path, and finally by the user's
7934 @env{PATH}. The linker used by GCC can be printed using @samp{which
7935 `gcc -print-prog-name=ld`}.
7940 Select the FDPIC ABI, that uses function descriptors to represent
7941 pointers to functions. Without any PIC/PIE-related options, it
7942 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
7943 assumes GOT entries and small data are within a 12-bit range from the
7944 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
7945 are computed with 32 bits.
7948 @opindex minline-plt
7950 Enable inlining of PLT entries in function calls to functions that are
7951 not known to bind locally. It has no effect without @option{-mfdpic}.
7952 It's enabled by default if optimizing for speed and compiling for
7953 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
7954 optimization option such as @option{-O3} or above is present in the
7960 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
7961 that is known to be in read-only sections. It's enabled by default,
7962 except for @option{-fpic} or @option{-fpie}: even though it may help
7963 make the global offset table smaller, it trades 1 instruction for 4.
7964 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
7965 one of which may be shared by multiple symbols, and it avoids the need
7966 for a GOT entry for the referenced symbol, so it's more likely to be a
7967 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
7969 @item -multilib-library-pic
7970 @opindex multilib-library-pic
7972 Link with the (library, not FD) pic libraries. It's implied by
7973 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
7974 @option{-fpic} without @option{-mfdpic}. You should never have to use
7980 Follow the EABI requirement of always creating a frame pointer whenever
7981 a stack frame is allocated. This option is enabled by default and can
7982 be disabled with @option{-mno-linked-fp}.
7985 @opindex mno-long-calls
7986 Generate code that uses long call sequences. This ensures that a call
7987 is always able to reach linker generated stubs. The default is to generate
7988 long calls only when the distance from the call site to the beginning
7989 of the function or translation unit, as the case may be, exceeds a
7990 predefined limit set by the branch type being used. The limits for
7991 normal calls are 7,600,000 and 240,000 bytes, respectively for the
7992 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
7995 Distances are measured from the beginning of functions when using the
7996 @option{-ffunction-sections} option, or when using the @option{-mgas}
7997 and @option{-mno-portable-runtime} options together under HP-UX with
8000 It is normally not desirable to use this option as it will degrade
8001 performance. However, it may be useful in large applications,
8002 particularly when partial linking is used to build the application.
8004 The types of long calls used depends on the capabilities of the
8005 assembler and linker, and the type of code being generated. The
8006 impact on systems that support long absolute calls, and long pic
8007 symbol-difference or pc-relative calls should be relatively small.
8008 However, an indirect call is used on 32-bit ELF systems in pic code
8009 and it is quite long.
8013 Suppress the generation of link options to search libdld.sl when the
8014 @option{-static} option is specified on HP-UX 10 and later.
8018 The HP-UX implementation of setlocale in libc has a dependency on
8019 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8020 when the @option{-static} option is specified, special link options
8021 are needed to resolve this dependency.
8023 On HP-UX 10 and later, the GCC driver adds the necessary options to
8024 link with libdld.sl when the @option{-static} option is specified.
8025 This causes the resulting binary to be dynamic. On the 64-bit port,
8026 the linkers generate dynamic binaries by default in any case. The
8027 @option{-nolibdld} option can be used to prevent the GCC driver from
8028 adding these link options.
8032 Add support for multithreading with the @dfn{dce thread} library
8033 under HP-UX. This option sets flags for both the preprocessor and
8037 @node i386 and x86-64 Options
8038 @subsection Intel 386 and AMD x86-64 Options
8039 @cindex i386 Options
8040 @cindex x86-64 Options
8041 @cindex Intel 386 Options
8042 @cindex AMD x86-64 Options
8044 These @samp{-m} options are defined for the i386 and x86-64 family of
8048 @item -mtune=@var{cpu-type}
8050 Tune to @var{cpu-type} everything applicable about the generated code, except
8051 for the ABI and the set of available instructions. The choices for
8055 Original Intel's i386 CPU.
8057 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8059 Intel Pentium CPU with no MMX support.
8061 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8062 @item i686, pentiumpro
8063 Intel PentiumPro CPU.
8065 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8066 @item pentium3, pentium3m
8067 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8070 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8071 support. Used by Centrino notebooks.
8072 @item pentium4, pentium4m
8073 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8075 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8078 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8079 SSE2 and SSE3 instruction set support.
8081 AMD K6 CPU with MMX instruction set support.
8083 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8084 @item athlon, athlon-tbird
8085 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8087 @item athlon-4, athlon-xp, athlon-mp
8088 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8089 instruction set support.
8090 @item k8, opteron, athlon64, athlon-fx
8091 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8092 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8094 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8097 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8098 instruction set support.
8100 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8101 implemented for this chip.)
8103 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8104 implemented for this chip.)
8107 While picking a specific @var{cpu-type} will schedule things appropriately
8108 for that particular chip, the compiler will not generate any code that
8109 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8112 @item -march=@var{cpu-type}
8114 Generate instructions for the machine type @var{cpu-type}. The choices
8115 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8116 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8118 @item -mcpu=@var{cpu-type}
8120 A deprecated synonym for @option{-mtune}.
8129 @opindex mpentiumpro
8130 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8131 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8132 These synonyms are deprecated.
8134 @item -mfpmath=@var{unit}
8136 Generate floating point arithmetics for selected unit @var{unit}. The choices
8141 Use the standard 387 floating point coprocessor present majority of chips and
8142 emulated otherwise. Code compiled with this option will run almost everywhere.
8143 The temporary results are computed in 80bit precision instead of precision
8144 specified by the type resulting in slightly different results compared to most
8145 of other chips. See @option{-ffloat-store} for more detailed description.
8147 This is the default choice for i386 compiler.
8150 Use scalar floating point instructions present in the SSE instruction set.
8151 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8152 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8153 instruction set supports only single precision arithmetics, thus the double and
8154 extended precision arithmetics is still done using 387. Later version, present
8155 only in Pentium4 and the future AMD x86-64 chips supports double precision
8158 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8159 @option{-msse2} switches to enable SSE extensions and make this option
8160 effective. For x86-64 compiler, these extensions are enabled by default.
8162 The resulting code should be considerably faster in the majority of cases and avoid
8163 the numerical instability problems of 387 code, but may break some existing
8164 code that expects temporaries to be 80bit.
8166 This is the default choice for the x86-64 compiler.
8169 Attempt to utilize both instruction sets at once. This effectively double the
8170 amount of available registers and on chips with separate execution units for
8171 387 and SSE the execution resources too. Use this option with care, as it is
8172 still experimental, because the GCC register allocator does not model separate
8173 functional units well resulting in instable performance.
8176 @item -masm=@var{dialect}
8177 @opindex masm=@var{dialect}
8178 Output asm instructions using selected @var{dialect}. Supported choices are
8179 @samp{intel} or @samp{att} (the default one).
8184 @opindex mno-ieee-fp
8185 Control whether or not the compiler uses IEEE floating point
8186 comparisons. These handle correctly the case where the result of a
8187 comparison is unordered.
8190 @opindex msoft-float
8191 Generate output containing library calls for floating point.
8192 @strong{Warning:} the requisite libraries are not part of GCC@.
8193 Normally the facilities of the machine's usual C compiler are used, but
8194 this can't be done directly in cross-compilation. You must make your
8195 own arrangements to provide suitable library functions for
8198 On machines where a function returns floating point results in the 80387
8199 register stack, some floating point opcodes may be emitted even if
8200 @option{-msoft-float} is used.
8202 @item -mno-fp-ret-in-387
8203 @opindex mno-fp-ret-in-387
8204 Do not use the FPU registers for return values of functions.
8206 The usual calling convention has functions return values of types
8207 @code{float} and @code{double} in an FPU register, even if there
8208 is no FPU@. The idea is that the operating system should emulate
8211 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8212 in ordinary CPU registers instead.
8214 @item -mno-fancy-math-387
8215 @opindex mno-fancy-math-387
8216 Some 387 emulators do not support the @code{sin}, @code{cos} and
8217 @code{sqrt} instructions for the 387. Specify this option to avoid
8218 generating those instructions. This option is the default on FreeBSD,
8219 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8220 indicates that the target cpu will always have an FPU and so the
8221 instruction will not need emulation. As of revision 2.6.1, these
8222 instructions are not generated unless you also use the
8223 @option{-funsafe-math-optimizations} switch.
8225 @item -malign-double
8226 @itemx -mno-align-double
8227 @opindex malign-double
8228 @opindex mno-align-double
8229 Control whether GCC aligns @code{double}, @code{long double}, and
8230 @code{long long} variables on a two word boundary or a one word
8231 boundary. Aligning @code{double} variables on a two word boundary will
8232 produce code that runs somewhat faster on a @samp{Pentium} at the
8233 expense of more memory.
8235 @strong{Warning:} if you use the @option{-malign-double} switch,
8236 structures containing the above types will be aligned differently than
8237 the published application binary interface specifications for the 386
8238 and will not be binary compatible with structures in code compiled
8239 without that switch.
8241 @item -m96bit-long-double
8242 @itemx -m128bit-long-double
8243 @opindex m96bit-long-double
8244 @opindex m128bit-long-double
8245 These switches control the size of @code{long double} type. The i386
8246 application binary interface specifies the size to be 96 bits,
8247 so @option{-m96bit-long-double} is the default in 32 bit mode.
8249 Modern architectures (Pentium and newer) would prefer @code{long double}
8250 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8251 conforming to the ABI, this would not be possible. So specifying a
8252 @option{-m128bit-long-double} will align @code{long double}
8253 to a 16 byte boundary by padding the @code{long double} with an additional
8256 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8257 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8259 Notice that neither of these options enable any extra precision over the x87
8260 standard of 80 bits for a @code{long double}.
8262 @strong{Warning:} if you override the default value for your target ABI, the
8263 structures and arrays containing @code{long double} variables will change
8264 their size as well as function calling convention for function taking
8265 @code{long double} will be modified. Hence they will not be binary
8266 compatible with arrays or structures in code compiled without that switch.
8270 @itemx -mno-svr3-shlib
8271 @opindex msvr3-shlib
8272 @opindex mno-svr3-shlib
8273 Control whether GCC places uninitialized local variables into the
8274 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8275 into @code{bss}. These options are meaningful only on System V Release 3.
8279 Use a different function-calling convention, in which functions that
8280 take a fixed number of arguments return with the @code{ret} @var{num}
8281 instruction, which pops their arguments while returning. This saves one
8282 instruction in the caller since there is no need to pop the arguments
8285 You can specify that an individual function is called with this calling
8286 sequence with the function attribute @samp{stdcall}. You can also
8287 override the @option{-mrtd} option by using the function attribute
8288 @samp{cdecl}. @xref{Function Attributes}.
8290 @strong{Warning:} this calling convention is incompatible with the one
8291 normally used on Unix, so you cannot use it if you need to call
8292 libraries compiled with the Unix compiler.
8294 Also, you must provide function prototypes for all functions that
8295 take variable numbers of arguments (including @code{printf});
8296 otherwise incorrect code will be generated for calls to those
8299 In addition, seriously incorrect code will result if you call a
8300 function with too many arguments. (Normally, extra arguments are
8301 harmlessly ignored.)
8303 @item -mregparm=@var{num}
8305 Control how many registers are used to pass integer arguments. By
8306 default, no registers are used to pass arguments, and at most 3
8307 registers can be used. You can control this behavior for a specific
8308 function by using the function attribute @samp{regparm}.
8309 @xref{Function Attributes}.
8311 @strong{Warning:} if you use this switch, and
8312 @var{num} is nonzero, then you must build all modules with the same
8313 value, including any libraries. This includes the system libraries and
8316 @item -mpreferred-stack-boundary=@var{num}
8317 @opindex mpreferred-stack-boundary
8318 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8319 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8320 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8321 size (@option{-Os}), in which case the default is the minimum correct
8322 alignment (4 bytes for x86, and 8 bytes for x86-64).
8324 On Pentium and PentiumPro, @code{double} and @code{long double} values
8325 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8326 suffer significant run time performance penalties. On Pentium III, the
8327 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8328 penalties if it is not 16 byte aligned.
8330 To ensure proper alignment of this values on the stack, the stack boundary
8331 must be as aligned as that required by any value stored on the stack.
8332 Further, every function must be generated such that it keeps the stack
8333 aligned. Thus calling a function compiled with a higher preferred
8334 stack boundary from a function compiled with a lower preferred stack
8335 boundary will most likely misalign the stack. It is recommended that
8336 libraries that use callbacks always use the default setting.
8338 This extra alignment does consume extra stack space, and generally
8339 increases code size. Code that is sensitive to stack space usage, such
8340 as embedded systems and operating system kernels, may want to reduce the
8341 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8359 These switches enable or disable the use of built-in functions that allow
8360 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8363 @xref{X86 Built-in Functions}, for details of the functions enabled
8364 and disabled by these switches.
8366 To have SSE/SSE2 instructions generated automatically from floating-point
8367 code, see @option{-mfpmath=sse}.
8370 @itemx -mno-push-args
8372 @opindex mno-push-args
8373 Use PUSH operations to store outgoing parameters. This method is shorter
8374 and usually equally fast as method using SUB/MOV operations and is enabled
8375 by default. In some cases disabling it may improve performance because of
8376 improved scheduling and reduced dependencies.
8378 @item -maccumulate-outgoing-args
8379 @opindex maccumulate-outgoing-args
8380 If enabled, the maximum amount of space required for outgoing arguments will be
8381 computed in the function prologue. This is faster on most modern CPUs
8382 because of reduced dependencies, improved scheduling and reduced stack usage
8383 when preferred stack boundary is not equal to 2. The drawback is a notable
8384 increase in code size. This switch implies @option{-mno-push-args}.
8388 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8389 on thread-safe exception handling must compile and link all code with the
8390 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8391 @option{-D_MT}; when linking, it links in a special thread helper library
8392 @option{-lmingwthrd} which cleans up per thread exception handling data.
8394 @item -mno-align-stringops
8395 @opindex mno-align-stringops
8396 Do not align destination of inlined string operations. This switch reduces
8397 code size and improves performance in case the destination is already aligned,
8398 but GCC doesn't know about it.
8400 @item -minline-all-stringops
8401 @opindex minline-all-stringops
8402 By default GCC inlines string operations only when destination is known to be
8403 aligned at least to 4 byte boundary. This enables more inlining, increase code
8404 size, but may improve performance of code that depends on fast memcpy, strlen
8405 and memset for short lengths.
8407 @item -momit-leaf-frame-pointer
8408 @opindex momit-leaf-frame-pointer
8409 Don't keep the frame pointer in a register for leaf functions. This
8410 avoids the instructions to save, set up and restore frame pointers and
8411 makes an extra register available in leaf functions. The option
8412 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8413 which might make debugging harder.
8415 @item -mtls-direct-seg-refs
8416 @itemx -mno-tls-direct-seg-refs
8417 @opindex mtls-direct-seg-refs
8418 Controls whether TLS variables may be accessed with offsets from the
8419 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8420 or whether the thread base pointer must be added. Whether or not this
8421 is legal depends on the operating system, and whether it maps the
8422 segment to cover the entire TLS area.
8424 For systems that use GNU libc, the default is on.
8427 These @samp{-m} switches are supported in addition to the above
8428 on AMD x86-64 processors in 64-bit environments.
8435 Generate code for a 32-bit or 64-bit environment.
8436 The 32-bit environment sets int, long and pointer to 32 bits and
8437 generates code that runs on any i386 system.
8438 The 64-bit environment sets int to 32 bits and long and pointer
8439 to 64 bits and generates code for AMD's x86-64 architecture.
8442 @opindex no-red-zone
8443 Do not use a so called red zone for x86-64 code. The red zone is mandated
8444 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8445 stack pointer that will not be modified by signal or interrupt handlers
8446 and therefore can be used for temporary data without adjusting the stack
8447 pointer. The flag @option{-mno-red-zone} disables this red zone.
8449 @item -mcmodel=small
8450 @opindex mcmodel=small
8451 Generate code for the small code model: the program and its symbols must
8452 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8453 Programs can be statically or dynamically linked. This is the default
8456 @item -mcmodel=kernel
8457 @opindex mcmodel=kernel
8458 Generate code for the kernel code model. The kernel runs in the
8459 negative 2 GB of the address space.
8460 This model has to be used for Linux kernel code.
8462 @item -mcmodel=medium
8463 @opindex mcmodel=medium
8464 Generate code for the medium model: The program is linked in the lower 2
8465 GB of the address space but symbols can be located anywhere in the
8466 address space. Programs can be statically or dynamically linked, but
8467 building of shared libraries are not supported with the medium model.
8469 @item -mcmodel=large
8470 @opindex mcmodel=large
8471 Generate code for the large model: This model makes no assumptions
8472 about addresses and sizes of sections. Currently GCC does not implement
8477 @subsection IA-64 Options
8478 @cindex IA-64 Options
8480 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8484 @opindex mbig-endian
8485 Generate code for a big endian target. This is the default for HP-UX@.
8487 @item -mlittle-endian
8488 @opindex mlittle-endian
8489 Generate code for a little endian target. This is the default for AIX5
8496 Generate (or don't) code for the GNU assembler. This is the default.
8497 @c Also, this is the default if the configure option @option{--with-gnu-as}
8504 Generate (or don't) code for the GNU linker. This is the default.
8505 @c Also, this is the default if the configure option @option{--with-gnu-ld}
8510 Generate code that does not use a global pointer register. The result
8511 is not position independent code, and violates the IA-64 ABI@.
8513 @item -mvolatile-asm-stop
8514 @itemx -mno-volatile-asm-stop
8515 @opindex mvolatile-asm-stop
8516 @opindex mno-volatile-asm-stop
8517 Generate (or don't) a stop bit immediately before and after volatile asm
8522 Generate code that works around Itanium B step errata.
8524 @item -mregister-names
8525 @itemx -mno-register-names
8526 @opindex mregister-names
8527 @opindex mno-register-names
8528 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8529 the stacked registers. This may make assembler output more readable.
8535 Disable (or enable) optimizations that use the small data section. This may
8536 be useful for working around optimizer bugs.
8539 @opindex mconstant-gp
8540 Generate code that uses a single constant global pointer value. This is
8541 useful when compiling kernel code.
8545 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
8546 This is useful when compiling firmware code.
8548 @item -minline-float-divide-min-latency
8549 @opindex minline-float-divide-min-latency
8550 Generate code for inline divides of floating point values
8551 using the minimum latency algorithm.
8553 @item -minline-float-divide-max-throughput
8554 @opindex minline-float-divide-max-throughput
8555 Generate code for inline divides of floating point values
8556 using the maximum throughput algorithm.
8558 @item -minline-int-divide-min-latency
8559 @opindex minline-int-divide-min-latency
8560 Generate code for inline divides of integer values
8561 using the minimum latency algorithm.
8563 @item -minline-int-divide-max-throughput
8564 @opindex minline-int-divide-max-throughput
8565 Generate code for inline divides of integer values
8566 using the maximum throughput algorithm.
8568 @item -mno-dwarf2-asm
8570 @opindex mno-dwarf2-asm
8571 @opindex mdwarf2-asm
8572 Don't (or do) generate assembler code for the DWARF2 line number debugging
8573 info. This may be useful when not using the GNU assembler.
8575 @item -mfixed-range=@var{register-range}
8576 @opindex mfixed-range
8577 Generate code treating the given register range as fixed registers.
8578 A fixed register is one that the register allocator can not use. This is
8579 useful when compiling kernel code. A register range is specified as
8580 two registers separated by a dash. Multiple register ranges can be
8581 specified separated by a comma.
8583 @item -mearly-stop-bits
8584 @itemx -mno-early-stop-bits
8585 @opindex mearly-stop-bits
8586 @opindex mno-early-stop-bits
8587 Allow stop bits to be placed earlier than immediately preceding the
8588 instruction that triggered the stop bit. This can improve instruction
8589 scheduling, but does not always do so.
8592 @node M32R/D Options
8593 @subsection M32R/D Options
8594 @cindex M32R/D options
8596 These @option{-m} options are defined for Renesas M32R/D architectures:
8601 Generate code for the M32R/2@.
8605 Generate code for the M32R/X@.
8609 Generate code for the M32R@. This is the default.
8612 @opindex mmodel=small
8613 Assume all objects live in the lower 16MB of memory (so that their addresses
8614 can be loaded with the @code{ld24} instruction), and assume all subroutines
8615 are reachable with the @code{bl} instruction.
8616 This is the default.
8618 The addressability of a particular object can be set with the
8619 @code{model} attribute.
8621 @item -mmodel=medium
8622 @opindex mmodel=medium
8623 Assume objects may be anywhere in the 32-bit address space (the compiler
8624 will generate @code{seth/add3} instructions to load their addresses), and
8625 assume all subroutines are reachable with the @code{bl} instruction.
8628 @opindex mmodel=large
8629 Assume objects may be anywhere in the 32-bit address space (the compiler
8630 will generate @code{seth/add3} instructions to load their addresses), and
8631 assume subroutines may not be reachable with the @code{bl} instruction
8632 (the compiler will generate the much slower @code{seth/add3/jl}
8633 instruction sequence).
8636 @opindex msdata=none
8637 Disable use of the small data area. Variables will be put into
8638 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
8639 @code{section} attribute has been specified).
8640 This is the default.
8642 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
8643 Objects may be explicitly put in the small data area with the
8644 @code{section} attribute using one of these sections.
8647 @opindex msdata=sdata
8648 Put small global and static data in the small data area, but do not
8649 generate special code to reference them.
8653 Put small global and static data in the small data area, and generate
8654 special instructions to reference them.
8658 @cindex smaller data references
8659 Put global and static objects less than or equal to @var{num} bytes
8660 into the small data or bss sections instead of the normal data or bss
8661 sections. The default value of @var{num} is 8.
8662 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
8663 for this option to have any effect.
8665 All modules should be compiled with the same @option{-G @var{num}} value.
8666 Compiling with different values of @var{num} may or may not work; if it
8667 doesn't the linker will give an error message---incorrect code will not be
8672 Makes the M32R specific code in the compiler display some statistics
8673 that might help in debugging programs.
8676 @opindex malign-loops
8677 Align all loops to a 32-byte boundary.
8679 @item -mno-align-loops
8680 @opindex mno-align-loops
8681 Do not enforce a 32-byte alignment for loops. This is the default.
8683 @item -missue-rate=@var{number}
8684 @opindex missue-rate=@var{number}
8685 Issue @var{number} instructions per cycle. @var{number} can only be 1
8688 @item -mbranch-cost=@var{number}
8689 @opindex mbranch-cost=@var{number}
8690 @var{number} can only be 1 or 2. If it is 1 then branches will be
8691 preferred over conditional code, if it is 2, then the opposite will
8694 @item -mflush-trap=@var{number}
8695 @opindex mflush-trap=@var{number}
8696 Specifies the trap number to use to flush the cache. The default is
8697 12. Valid numbers are between 0 and 15 inclusive.
8699 @item -mno-flush-trap
8700 @opindex mno-flush-trap
8701 Specifies that the cache cannot be flushed by using a trap.
8703 @item -mflush-func=@var{name}
8704 @opindex mflush-func=@var{name}
8705 Specifies the name of the operating system function to call to flush
8706 the cache. The default is @emph{_flush_cache}, but a function call
8707 will only be used if a trap is not available.
8709 @item -mno-flush-func
8710 @opindex mno-flush-func
8711 Indicates that there is no OS function for flushing the cache.
8715 @node M680x0 Options
8716 @subsection M680x0 Options
8717 @cindex M680x0 options
8719 These are the @samp{-m} options defined for the 68000 series. The default
8720 values for these options depends on which style of 68000 was selected when
8721 the compiler was configured; the defaults for the most common choices are
8729 Generate output for a 68000. This is the default
8730 when the compiler is configured for 68000-based systems.
8732 Use this option for microcontrollers with a 68000 or EC000 core,
8733 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
8739 Generate output for a 68020. This is the default
8740 when the compiler is configured for 68020-based systems.
8744 Generate output containing 68881 instructions for floating point.
8745 This is the default for most 68020 systems unless @option{--nfp} was
8746 specified when the compiler was configured.
8750 Generate output for a 68030. This is the default when the compiler is
8751 configured for 68030-based systems.
8755 Generate output for a 68040. This is the default when the compiler is
8756 configured for 68040-based systems.
8758 This option inhibits the use of 68881/68882 instructions that have to be
8759 emulated by software on the 68040. Use this option if your 68040 does not
8760 have code to emulate those instructions.
8764 Generate output for a 68060. This is the default when the compiler is
8765 configured for 68060-based systems.
8767 This option inhibits the use of 68020 and 68881/68882 instructions that
8768 have to be emulated by software on the 68060. Use this option if your 68060
8769 does not have code to emulate those instructions.
8773 Generate output for a CPU32. This is the default
8774 when the compiler is configured for CPU32-based systems.
8776 Use this option for microcontrollers with a
8777 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
8778 68336, 68340, 68341, 68349 and 68360.
8782 Generate output for a 520X ``coldfire'' family cpu. This is the default
8783 when the compiler is configured for 520X-based systems.
8785 Use this option for microcontroller with a 5200 core, including
8786 the MCF5202, MCF5203, MCF5204 and MCF5202.
8791 Generate output for a 68040, without using any of the new instructions.
8792 This results in code which can run relatively efficiently on either a
8793 68020/68881 or a 68030 or a 68040. The generated code does use the
8794 68881 instructions that are emulated on the 68040.
8798 Generate output for a 68060, without using any of the new instructions.
8799 This results in code which can run relatively efficiently on either a
8800 68020/68881 or a 68030 or a 68040. The generated code does use the
8801 68881 instructions that are emulated on the 68060.
8804 @opindex msoft-float
8805 Generate output containing library calls for floating point.
8806 @strong{Warning:} the requisite libraries are not available for all m68k
8807 targets. Normally the facilities of the machine's usual C compiler are
8808 used, but this can't be done directly in cross-compilation. You must
8809 make your own arrangements to provide suitable library functions for
8810 cross-compilation. The embedded targets @samp{m68k-*-aout} and
8811 @samp{m68k-*-coff} do provide software floating point support.
8815 Consider type @code{int} to be 16 bits wide, like @code{short int}.
8816 Additionally, parameters passed on the stack are also aligned to a
8817 16-bit boundary even on targets whose API mandates promotion to 32-bit.
8820 @opindex mnobitfield
8821 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
8822 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
8826 Do use the bit-field instructions. The @option{-m68020} option implies
8827 @option{-mbitfield}. This is the default if you use a configuration
8828 designed for a 68020.
8832 Use a different function-calling convention, in which functions
8833 that take a fixed number of arguments return with the @code{rtd}
8834 instruction, which pops their arguments while returning. This
8835 saves one instruction in the caller since there is no need to pop
8836 the arguments there.
8838 This calling convention is incompatible with the one normally
8839 used on Unix, so you cannot use it if you need to call libraries
8840 compiled with the Unix compiler.
8842 Also, you must provide function prototypes for all functions that
8843 take variable numbers of arguments (including @code{printf});
8844 otherwise incorrect code will be generated for calls to those
8847 In addition, seriously incorrect code will result if you call a
8848 function with too many arguments. (Normally, extra arguments are
8849 harmlessly ignored.)
8851 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
8852 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
8855 @itemx -mno-align-int
8857 @opindex mno-align-int
8858 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
8859 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
8860 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
8861 Aligning variables on 32-bit boundaries produces code that runs somewhat
8862 faster on processors with 32-bit busses at the expense of more memory.
8864 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
8865 align structures containing the above types differently than
8866 most published application binary interface specifications for the m68k.
8870 Use the pc-relative addressing mode of the 68000 directly, instead of
8871 using a global offset table. At present, this option implies @option{-fpic},
8872 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
8873 not presently supported with @option{-mpcrel}, though this could be supported for
8874 68020 and higher processors.
8876 @item -mno-strict-align
8877 @itemx -mstrict-align
8878 @opindex mno-strict-align
8879 @opindex mstrict-align
8880 Do not (do) assume that unaligned memory references will be handled by
8884 Generate code that allows the data segment to be located in a different
8885 area of memory from the text segment. This allows for execute in place in
8886 an environment without virtual memory management. This option implies -fPIC.
8889 Generate code that assumes that the data segment follows the text segment.
8890 This is the default.
8892 @item -mid-shared-library
8893 Generate code that supports shared libraries via the library ID method.
8894 This allows for execute in place and shared libraries in an environment
8895 without virtual memory management. This option implies -fPIC.
8897 @item -mno-id-shared-library
8898 Generate code that doesn't assume ID based shared libraries are being used.
8899 This is the default.
8901 @item -mshared-library-id=n
8902 Specified the identification number of the ID based shared library being
8903 compiled. Specifying a value of 0 will generate more compact code, specifying
8904 other values will force the allocation of that number to the current
8905 library but is no more space or time efficient than omitting this option.
8909 @node M68hc1x Options
8910 @subsection M68hc1x Options
8911 @cindex M68hc1x options
8913 These are the @samp{-m} options defined for the 68hc11 and 68hc12
8914 microcontrollers. The default values for these options depends on
8915 which style of microcontroller was selected when the compiler was configured;
8916 the defaults for the most common choices are given below.
8923 Generate output for a 68HC11. This is the default
8924 when the compiler is configured for 68HC11-based systems.
8930 Generate output for a 68HC12. This is the default
8931 when the compiler is configured for 68HC12-based systems.
8937 Generate output for a 68HCS12.
8940 @opindex mauto-incdec
8941 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
8948 Enable the use of 68HC12 min and max instructions.
8951 @itemx -mno-long-calls
8952 @opindex mlong-calls
8953 @opindex mno-long-calls
8954 Treat all calls as being far away (near). If calls are assumed to be
8955 far away, the compiler will use the @code{call} instruction to
8956 call a function and the @code{rtc} instruction for returning.
8960 Consider type @code{int} to be 16 bits wide, like @code{short int}.
8962 @item -msoft-reg-count=@var{count}
8963 @opindex msoft-reg-count
8964 Specify the number of pseudo-soft registers which are used for the
8965 code generation. The maximum number is 32. Using more pseudo-soft
8966 register may or may not result in better code depending on the program.
8967 The default is 4 for 68HC11 and 2 for 68HC12.
8972 @subsection MCore Options
8973 @cindex MCore options
8975 These are the @samp{-m} options defined for the Motorola M*Core
8983 @opindex mno-hardlit
8984 Inline constants into the code stream if it can be done in two
8985 instructions or less.
8991 Use the divide instruction. (Enabled by default).
8993 @item -mrelax-immediate
8994 @itemx -mno-relax-immediate
8995 @opindex mrelax-immediate
8996 @opindex mno-relax-immediate
8997 Allow arbitrary sized immediates in bit operations.
8999 @item -mwide-bitfields
9000 @itemx -mno-wide-bitfields
9001 @opindex mwide-bitfields
9002 @opindex mno-wide-bitfields
9003 Always treat bit-fields as int-sized.
9005 @item -m4byte-functions
9006 @itemx -mno-4byte-functions
9007 @opindex m4byte-functions
9008 @opindex mno-4byte-functions
9009 Force all functions to be aligned to a four byte boundary.
9011 @item -mcallgraph-data
9012 @itemx -mno-callgraph-data
9013 @opindex mcallgraph-data
9014 @opindex mno-callgraph-data
9015 Emit callgraph information.
9018 @itemx -mno-slow-bytes
9019 @opindex mslow-bytes
9020 @opindex mno-slow-bytes
9021 Prefer word access when reading byte quantities.
9023 @item -mlittle-endian
9025 @opindex mlittle-endian
9026 @opindex mbig-endian
9027 Generate code for a little endian target.
9033 Generate code for the 210 processor.
9037 @subsection MIPS Options
9038 @cindex MIPS options
9044 Generate big-endian code.
9048 Generate little-endian code. This is the default for @samp{mips*el-*-*}
9051 @item -march=@var{arch}
9053 Generate code that will run on @var{arch}, which can be the name of a
9054 generic MIPS ISA, or the name of a particular processor.
9056 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
9057 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
9058 The processor names are:
9059 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
9061 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
9062 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
9066 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
9067 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
9068 The special value @samp{from-abi} selects the
9069 most compatible architecture for the selected ABI (that is,
9070 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
9072 In processor names, a final @samp{000} can be abbreviated as @samp{k}
9073 (for example, @samp{-march=r2k}). Prefixes are optional, and
9074 @samp{vr} may be written @samp{r}.
9076 GCC defines two macros based on the value of this option. The first
9077 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
9078 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
9079 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
9080 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
9081 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
9083 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
9084 above. In other words, it will have the full prefix and will not
9085 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
9086 the macro names the resolved architecture (either @samp{"mips1"} or
9087 @samp{"mips3"}). It names the default architecture when no
9088 @option{-march} option is given.
9090 @item -mtune=@var{arch}
9092 Optimize for @var{arch}. Among other things, this option controls
9093 the way instructions are scheduled, and the perceived cost of arithmetic
9094 operations. The list of @var{arch} values is the same as for
9097 When this option is not used, GCC will optimize for the processor
9098 specified by @option{-march}. By using @option{-march} and
9099 @option{-mtune} together, it is possible to generate code that will
9100 run on a family of processors, but optimize the code for one
9101 particular member of that family.
9103 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
9104 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
9105 @samp{-march} ones described above.
9109 Equivalent to @samp{-march=mips1}.
9113 Equivalent to @samp{-march=mips2}.
9117 Equivalent to @samp{-march=mips3}.
9121 Equivalent to @samp{-march=mips4}.
9125 Equivalent to @samp{-march=mips32}.
9129 Equivalent to @samp{-march=mips32r2}.
9133 Equivalent to @samp{-march=mips64}.
9139 Use (do not use) the MIPS16 ISA.
9151 Generate code for the given ABI@.
9153 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
9154 generates 64-bit code when you select a 64-bit architecture, but you
9155 can use @option{-mgp32} to get 32-bit code instead.
9157 For information about the O64 ABI, see
9158 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
9161 @itemx -mno-abicalls
9163 @opindex mno-abicalls
9164 Generate (do not generate) SVR4-style position-independent code.
9165 @option{-mabicalls} is the default for SVR4-based systems.
9171 Lift (do not lift) the usual restrictions on the size of the global
9174 GCC normally uses a single instruction to load values from the GOT.
9175 While this is relatively efficient, it will only work if the GOT
9176 is smaller than about 64k. Anything larger will cause the linker
9177 to report an error such as:
9179 @cindex relocation truncated to fit (MIPS)
9181 relocation truncated to fit: R_MIPS_GOT16 foobar
9184 If this happens, you should recompile your code with @option{-mxgot}.
9185 It should then work with very large GOTs, although it will also be
9186 less efficient, since it will take three instructions to fetch the
9187 value of a global symbol.
9189 Note that some linkers can create multiple GOTs. If you have such a
9190 linker, you should only need to use @option{-mxgot} when a single object
9191 file accesses more than 64k's worth of GOT entries. Very few do.
9193 These options have no effect unless GCC is generating position
9198 Assume that general-purpose registers are 32 bits wide.
9202 Assume that general-purpose registers are 64 bits wide.
9206 Assume that floating-point registers are 32 bits wide.
9210 Assume that floating-point registers are 64 bits wide.
9213 @opindex mhard-float
9214 Use floating-point coprocessor instructions.
9217 @opindex msoft-float
9218 Do not use floating-point coprocessor instructions. Implement
9219 floating-point calculations using library calls instead.
9221 @item -msingle-float
9222 @opindex msingle-float
9223 Assume that the floating-point coprocessor only supports single-precision
9226 @itemx -mdouble-float
9227 @opindex mdouble-float
9228 Assume that the floating-point coprocessor supports double-precision
9229 operations. This is the default.
9233 Force @code{int} and @code{long} types to be 64 bits wide. See
9234 @option{-mlong32} for an explanation of the default and the way
9235 that the pointer size is determined.
9239 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
9240 an explanation of the default and the way that the pointer size is
9245 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
9247 The default size of @code{int}s, @code{long}s and pointers depends on
9248 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
9249 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
9250 32-bit @code{long}s. Pointers are the same size as @code{long}s,
9251 or the same size as integer registers, whichever is smaller.
9255 @cindex smaller data references (MIPS)
9256 @cindex gp-relative references (MIPS)
9257 Put global and static items less than or equal to @var{num} bytes into
9258 the small data or bss section instead of the normal data or bss section.
9259 This allows the data to be accessed using a single instruction.
9261 All modules should be compiled with the same @option{-G @var{num}}
9264 @item -membedded-data
9265 @itemx -mno-embedded-data
9266 @opindex membedded-data
9267 @opindex mno-embedded-data
9268 Allocate variables to the read-only data section first if possible, then
9269 next in the small data section if possible, otherwise in data. This gives
9270 slightly slower code than the default, but reduces the amount of RAM required
9271 when executing, and thus may be preferred for some embedded systems.
9273 @item -muninit-const-in-rodata
9274 @itemx -mno-uninit-const-in-rodata
9275 @opindex muninit-const-in-rodata
9276 @opindex mno-uninit-const-in-rodata
9277 Put uninitialized @code{const} variables in the read-only data section.
9278 This option is only meaningful in conjunction with @option{-membedded-data}.
9280 @item -msplit-addresses
9281 @itemx -mno-split-addresses
9282 @opindex msplit-addresses
9283 @opindex mno-split-addresses
9284 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
9285 relocation operators. This option has been superceded by
9286 @option{-mexplicit-relocs} but is retained for backwards compatibility.
9288 @item -mexplicit-relocs
9289 @itemx -mno-explicit-relocs
9290 @opindex mexplicit-relocs
9291 @opindex mno-explicit-relocs
9292 Use (do not use) assembler relocation operators when dealing with symbolic
9293 addresses. The alternative, selected by @option{-mno-explicit-relocs},
9294 is to use assembler macros instead.
9296 @option{-mexplicit-relocs} is the default if GCC was configured
9297 to use an assembler that supports relocation operators.
9299 @item -mcheck-zero-division
9300 @itemx -mno-check-zero-division
9301 @opindex mcheck-zero-division
9302 @opindex mno-check-zero-division
9303 Trap (do not trap) on integer division by zero. The default is
9304 @option{-mcheck-zero-division}.
9310 Force (do not force) the use of @code{memcpy()} for non-trivial block
9311 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
9312 most constant-sized copies.
9315 @itemx -mno-long-calls
9316 @opindex mlong-calls
9317 @opindex mno-long-calls
9318 Disable (do not disable) use of the @code{jal} instruction. Calling
9319 functions using @code{jal} is more efficient but requires the caller
9320 and callee to be in the same 256 megabyte segment.
9322 This option has no effect on abicalls code. The default is
9323 @option{-mno-long-calls}.
9329 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
9330 instructions, as provided by the R4650 ISA.
9333 @itemx -mno-fused-madd
9334 @opindex mfused-madd
9335 @opindex mno-fused-madd
9336 Enable (disable) use of the floating point multiply-accumulate
9337 instructions, when they are available. The default is
9338 @option{-mfused-madd}.
9340 When multiply-accumulate instructions are used, the intermediate
9341 product is calculated to infinite precision and is not subject to
9342 the FCSR Flush to Zero bit. This may be undesirable in some
9347 Tell the MIPS assembler to not run its preprocessor over user
9348 assembler files (with a @samp{.s} suffix) when assembling them.
9351 @itemx -mno-fix-r4000
9353 @opindex mno-fix-r4000
9354 Work around certain R4000 CPU errata:
9357 A double-word or a variable shift may give an incorrect result if executed
9358 immediately after starting an integer division.
9360 A double-word or a variable shift may give an incorrect result if executed
9361 while an integer multiplication is in progress.
9363 An integer division may give an incorrect result if started in a delay slot
9364 of a taken branch or a jump.
9368 @itemx -mno-fix-r4400
9370 @opindex mno-fix-r4400
9371 Work around certain R4400 CPU errata:
9374 A double-word or a variable shift may give an incorrect result if executed
9375 immediately after starting an integer division.
9379 @itemx -mno-fix-vr4120
9380 @opindex mfix-vr4120
9381 Work around certain VR4120 errata:
9384 @code{dmultu} does not always produce the correct result.
9386 @code{div} and @code{ddiv} do not always produce the correct result if one
9387 of the operands is negative.
9389 The workarounds for the division errata rely on special functions in
9390 @file{libgcc.a}. At present, these functions are only provided by
9391 the @code{mips64vr*-elf} configurations.
9393 Other VR4120 errata require a nop to be inserted between certain pairs of
9394 instructions. These errata are handled by the assembler, not by GCC itself.
9399 Work around certain SB-1 CPU core errata.
9400 (This flag currently works around the SB-1 revision 2
9401 ``F1'' and ``F2'' floating point errata.)
9403 @item -mflush-func=@var{func}
9404 @itemx -mno-flush-func
9405 @opindex mflush-func
9406 Specifies the function to call to flush the I and D caches, or to not
9407 call any such function. If called, the function must take the same
9408 arguments as the common @code{_flush_func()}, that is, the address of the
9409 memory range for which the cache is being flushed, the size of the
9410 memory range, and the number 3 (to flush both caches). The default
9411 depends on the target GCC was configured for, but commonly is either
9412 @samp{_flush_func} or @samp{__cpu_flush}.
9414 @item -mbranch-likely
9415 @itemx -mno-branch-likely
9416 @opindex mbranch-likely
9417 @opindex mno-branch-likely
9418 Enable or disable use of Branch Likely instructions, regardless of the
9419 default for the selected architecture. By default, Branch Likely
9420 instructions may be generated if they are supported by the selected
9421 architecture. An exception is for the MIPS32 and MIPS64 architectures
9422 and processors which implement those architectures; for those, Branch
9423 Likely instructions will not be generated by default because the MIPS32
9424 and MIPS64 architectures specifically deprecate their use.
9426 @item -mfp-exceptions
9427 @itemx -mno-fp-exceptions
9428 @opindex mfp-exceptions
9429 Specifies whether FP exceptions are enabled. This affects how we schedule
9430 FP instructions for some processors. The default is that FP exceptions are
9433 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
9434 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
9437 @item -mvr4130-align
9438 @itemx -mno-vr4130-align
9439 @opindex mvr4130-align
9440 The VR4130 pipeline is two-way superscalar, but can only issue two
9441 instructions together if the first one is 8-byte aligned. When this
9442 option is enabled, GCC will align pairs of instructions that it
9443 thinks should execute in parallel.
9445 This option only has an effect when optimizing for the VR4130.
9446 It normally makes code faster, but at the expense of making it bigger.
9447 It is enabled by default at optimization level @option{-O3}.
9451 @subsection MMIX Options
9452 @cindex MMIX Options
9454 These options are defined for the MMIX:
9458 @itemx -mno-libfuncs
9460 @opindex mno-libfuncs
9461 Specify that intrinsic library functions are being compiled, passing all
9462 values in registers, no matter the size.
9467 @opindex mno-epsilon
9468 Generate floating-point comparison instructions that compare with respect
9469 to the @code{rE} epsilon register.
9471 @item -mabi=mmixware
9473 @opindex mabi-mmixware
9475 Generate code that passes function parameters and return values that (in
9476 the called function) are seen as registers @code{$0} and up, as opposed to
9477 the GNU ABI which uses global registers @code{$231} and up.
9480 @itemx -mno-zero-extend
9481 @opindex mzero-extend
9482 @opindex mno-zero-extend
9483 When reading data from memory in sizes shorter than 64 bits, use (do not
9484 use) zero-extending load instructions by default, rather than
9485 sign-extending ones.
9488 @itemx -mno-knuthdiv
9490 @opindex mno-knuthdiv
9491 Make the result of a division yielding a remainder have the same sign as
9492 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9493 remainder follows the sign of the dividend. Both methods are
9494 arithmetically valid, the latter being almost exclusively used.
9496 @item -mtoplevel-symbols
9497 @itemx -mno-toplevel-symbols
9498 @opindex mtoplevel-symbols
9499 @opindex mno-toplevel-symbols
9500 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9501 code can be used with the @code{PREFIX} assembly directive.
9505 Generate an executable in the ELF format, rather than the default
9506 @samp{mmo} format used by the @command{mmix} simulator.
9508 @item -mbranch-predict
9509 @itemx -mno-branch-predict
9510 @opindex mbranch-predict
9511 @opindex mno-branch-predict
9512 Use (do not use) the probable-branch instructions, when static branch
9513 prediction indicates a probable branch.
9515 @item -mbase-addresses
9516 @itemx -mno-base-addresses
9517 @opindex mbase-addresses
9518 @opindex mno-base-addresses
9519 Generate (do not generate) code that uses @emph{base addresses}. Using a
9520 base address automatically generates a request (handled by the assembler
9521 and the linker) for a constant to be set up in a global register. The
9522 register is used for one or more base address requests within the range 0
9523 to 255 from the value held in the register. The generally leads to short
9524 and fast code, but the number of different data items that can be
9525 addressed is limited. This means that a program that uses lots of static
9526 data may require @option{-mno-base-addresses}.
9529 @itemx -mno-single-exit
9530 @opindex msingle-exit
9531 @opindex mno-single-exit
9532 Force (do not force) generated code to have a single exit point in each
9536 @node MN10300 Options
9537 @subsection MN10300 Options
9538 @cindex MN10300 options
9540 These @option{-m} options are defined for Matsushita MN10300 architectures:
9545 Generate code to avoid bugs in the multiply instructions for the MN10300
9546 processors. This is the default.
9549 @opindex mno-mult-bug
9550 Do not generate code to avoid bugs in the multiply instructions for the
9555 Generate code which uses features specific to the AM33 processor.
9559 Do not generate code which uses features specific to the AM33 processor. This
9564 Do not link in the C run-time initialization object file.
9568 Indicate to the linker that it should perform a relaxation optimization pass
9569 to shorten branches, calls and absolute memory addresses. This option only
9570 has an effect when used on the command line for the final link step.
9572 This option makes symbolic debugging impossible.
9576 @subsection NS32K Options
9577 @cindex NS32K options
9579 These are the @samp{-m} options defined for the 32000 series. The default
9580 values for these options depends on which style of 32000 was selected when
9581 the compiler was configured; the defaults for the most common choices are
9589 Generate output for a 32032. This is the default
9590 when the compiler is configured for 32032 and 32016 based systems.
9596 Generate output for a 32332. This is the default
9597 when the compiler is configured for 32332-based systems.
9603 Generate output for a 32532. This is the default
9604 when the compiler is configured for 32532-based systems.
9608 Generate output containing 32081 instructions for floating point.
9609 This is the default for all systems.
9613 Generate output containing 32381 instructions for floating point. This
9614 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9615 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9619 Try and generate multiply-add floating point instructions @code{polyF}
9620 and @code{dotF}. This option is only available if the @option{-m32381}
9621 option is in effect. Using these instructions requires changes to
9622 register allocation which generally has a negative impact on
9623 performance. This option should only be enabled when compiling code
9624 particularly likely to make heavy use of multiply-add instructions.
9627 @opindex mnomulti-add
9628 Do not try and generate multiply-add floating point instructions
9629 @code{polyF} and @code{dotF}. This is the default on all platforms.
9632 @opindex msoft-float
9633 Generate output containing library calls for floating point.
9634 @strong{Warning:} the requisite libraries may not be available.
9636 @item -mieee-compare
9637 @itemx -mno-ieee-compare
9638 @opindex mieee-compare
9639 @opindex mno-ieee-compare
9640 Control whether or not the compiler uses IEEE floating point
9641 comparisons. These handle correctly the case where the result of a
9642 comparison is unordered.
9643 @strong{Warning:} the requisite kernel support may not be available.
9646 @opindex mnobitfield
9647 Do not use the bit-field instructions. On some machines it is faster to
9648 use shifting and masking operations. This is the default for the pc532.
9652 Do use the bit-field instructions. This is the default for all platforms
9657 Use a different function-calling convention, in which functions
9658 that take a fixed number of arguments return pop their
9659 arguments on return with the @code{ret} instruction.
9661 This calling convention is incompatible with the one normally
9662 used on Unix, so you cannot use it if you need to call libraries
9663 compiled with the Unix compiler.
9665 Also, you must provide function prototypes for all functions that
9666 take variable numbers of arguments (including @code{printf});
9667 otherwise incorrect code will be generated for calls to those
9670 In addition, seriously incorrect code will result if you call a
9671 function with too many arguments. (Normally, extra arguments are
9672 harmlessly ignored.)
9674 This option takes its name from the 680x0 @code{rtd} instruction.
9679 Use a different function-calling convention where the first two arguments
9680 are passed in registers.
9682 This calling convention is incompatible with the one normally
9683 used on Unix, so you cannot use it if you need to call libraries
9684 compiled with the Unix compiler.
9687 @opindex mnoregparam
9688 Do not pass any arguments in registers. This is the default for all
9693 It is OK to use the sb as an index register which is always loaded with
9694 zero. This is the default for the pc532-netbsd target.
9698 The sb register is not available for use or has not been initialized to
9699 zero by the run time system. This is the default for all targets except
9700 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9701 @option{-fpic} is set.
9705 Many ns32000 series addressing modes use displacements of up to 512MB@.
9706 If an address is above 512MB then displacements from zero can not be used.
9707 This option causes code to be generated which can be loaded above 512MB@.
9708 This may be useful for operating systems or ROM code.
9712 Assume code will be loaded in the first 512MB of virtual address space.
9713 This is the default for all platforms.
9717 @node PDP-11 Options
9718 @subsection PDP-11 Options
9719 @cindex PDP-11 Options
9721 These options are defined for the PDP-11:
9726 Use hardware FPP floating point. This is the default. (FIS floating
9727 point on the PDP-11/40 is not supported.)
9730 @opindex msoft-float
9731 Do not use hardware floating point.
9735 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
9739 Return floating-point results in memory. This is the default.
9743 Generate code for a PDP-11/40.
9747 Generate code for a PDP-11/45. This is the default.
9751 Generate code for a PDP-11/10.
9753 @item -mbcopy-builtin
9754 @opindex bcopy-builtin
9755 Use inline @code{movmemhi} patterns for copying memory. This is the
9760 Do not use inline @code{movmemhi} patterns for copying memory.
9766 Use 16-bit @code{int}. This is the default.
9772 Use 32-bit @code{int}.
9777 @opindex mno-float32
9778 Use 64-bit @code{float}. This is the default.
9783 @opindex mno-float64
9784 Use 32-bit @code{float}.
9788 Use @code{abshi2} pattern. This is the default.
9792 Do not use @code{abshi2} pattern.
9794 @item -mbranch-expensive
9795 @opindex mbranch-expensive
9796 Pretend that branches are expensive. This is for experimenting with
9797 code generation only.
9799 @item -mbranch-cheap
9800 @opindex mbranch-cheap
9801 Do not pretend that branches are expensive. This is the default.
9805 Generate code for a system with split I&D.
9809 Generate code for a system without split I&D. This is the default.
9813 Use Unix assembler syntax. This is the default when configured for
9818 Use DEC assembler syntax. This is the default when configured for any
9819 PDP-11 target other than @samp{pdp11-*-bsd}.
9822 @node PowerPC Options
9823 @subsection PowerPC Options
9824 @cindex PowerPC options
9826 These are listed under @xref{RS/6000 and PowerPC Options}.
9828 @node RS/6000 and PowerPC Options
9829 @subsection IBM RS/6000 and PowerPC Options
9830 @cindex RS/6000 and PowerPC Options
9831 @cindex IBM RS/6000 and PowerPC Options
9833 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
9841 @itemx -mpowerpc-gpopt
9842 @itemx -mno-powerpc-gpopt
9843 @itemx -mpowerpc-gfxopt
9844 @itemx -mno-powerpc-gfxopt
9846 @itemx -mno-powerpc64
9852 @opindex mno-powerpc
9853 @opindex mpowerpc-gpopt
9854 @opindex mno-powerpc-gpopt
9855 @opindex mpowerpc-gfxopt
9856 @opindex mno-powerpc-gfxopt
9858 @opindex mno-powerpc64
9859 GCC supports two related instruction set architectures for the
9860 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
9861 instructions supported by the @samp{rios} chip set used in the original
9862 RS/6000 systems and the @dfn{PowerPC} instruction set is the
9863 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
9864 the IBM 4xx microprocessors.
9866 Neither architecture is a subset of the other. However there is a
9867 large common subset of instructions supported by both. An MQ
9868 register is included in processors supporting the POWER architecture.
9870 You use these options to specify which instructions are available on the
9871 processor you are using. The default value of these options is
9872 determined when configuring GCC@. Specifying the
9873 @option{-mcpu=@var{cpu_type}} overrides the specification of these
9874 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
9875 rather than the options listed above.
9877 The @option{-mpower} option allows GCC to generate instructions that
9878 are found only in the POWER architecture and to use the MQ register.
9879 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
9880 to generate instructions that are present in the POWER2 architecture but
9881 not the original POWER architecture.
9883 The @option{-mpowerpc} option allows GCC to generate instructions that
9884 are found only in the 32-bit subset of the PowerPC architecture.
9885 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
9886 GCC to use the optional PowerPC architecture instructions in the
9887 General Purpose group, including floating-point square root. Specifying
9888 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
9889 use the optional PowerPC architecture instructions in the Graphics
9890 group, including floating-point select.
9892 The @option{-mpowerpc64} option allows GCC to generate the additional
9893 64-bit instructions that are found in the full PowerPC64 architecture
9894 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
9895 @option{-mno-powerpc64}.
9897 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
9898 will use only the instructions in the common subset of both
9899 architectures plus some special AIX common-mode calls, and will not use
9900 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
9901 permits GCC to use any instruction from either architecture and to
9902 allow use of the MQ register; specify this for the Motorola MPC601.
9904 @item -mnew-mnemonics
9905 @itemx -mold-mnemonics
9906 @opindex mnew-mnemonics
9907 @opindex mold-mnemonics
9908 Select which mnemonics to use in the generated assembler code. With
9909 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
9910 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
9911 assembler mnemonics defined for the POWER architecture. Instructions
9912 defined in only one architecture have only one mnemonic; GCC uses that
9913 mnemonic irrespective of which of these options is specified.
9915 GCC defaults to the mnemonics appropriate for the architecture in
9916 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
9917 value of these option. Unless you are building a cross-compiler, you
9918 should normally not specify either @option{-mnew-mnemonics} or
9919 @option{-mold-mnemonics}, but should instead accept the default.
9921 @item -mcpu=@var{cpu_type}
9923 Set architecture type, register usage, choice of mnemonics, and
9924 instruction scheduling parameters for machine type @var{cpu_type}.
9925 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
9926 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
9927 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
9928 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
9929 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
9930 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
9931 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
9932 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
9933 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
9935 @option{-mcpu=common} selects a completely generic processor. Code
9936 generated under this option will run on any POWER or PowerPC processor.
9937 GCC will use only the instructions in the common subset of both
9938 architectures, and will not use the MQ register. GCC assumes a generic
9939 processor model for scheduling purposes.
9941 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
9942 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
9943 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
9944 types, with an appropriate, generic processor model assumed for
9945 scheduling purposes.
9947 The other options specify a specific processor. Code generated under
9948 those options will run best on that processor, and may not run at all on
9951 The @option{-mcpu} options automatically enable or disable the
9952 following options: @option{-maltivec}, @option{-mhard-float},
9953 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
9954 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
9955 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
9956 @option{-mstring}. The particular options set for any particular CPU
9957 will vary between compiler versions, depending on what setting seems
9958 to produce optimal code for that CPU; it doesn't necessarily reflect
9959 the actual hardware's capabilities. If you wish to set an individual
9960 option to a particular value, you may specify it after the
9961 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
9963 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
9964 not enabled or disabled by the @option{-mcpu} option at present, since
9965 AIX does not have full support for these options. You may still
9966 enable or disable them individually if you're sure it'll work in your
9969 @item -mtune=@var{cpu_type}
9971 Set the instruction scheduling parameters for machine type
9972 @var{cpu_type}, but do not set the architecture type, register usage, or
9973 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
9974 values for @var{cpu_type} are used for @option{-mtune} as for
9975 @option{-mcpu}. If both are specified, the code generated will use the
9976 architecture, registers, and mnemonics set by @option{-mcpu}, but the
9977 scheduling parameters set by @option{-mtune}.
9982 @opindex mno-altivec
9983 These switches enable or disable the use of built-in functions that
9984 allow access to the AltiVec instruction set. You may also need to set
9985 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
9990 Extend the current ABI with SPE ABI extensions. This does not change
9991 the default ABI, instead it adds the SPE ABI extensions to the current
9995 @opindex mabi=no-spe
9996 Disable Booke SPE ABI extensions for the current ABI.
9998 @item -misel=@var{yes/no}
10001 This switch enables or disables the generation of ISEL instructions.
10003 @item -mspe=@var{yes/no}
10006 This switch enables or disables the generation of SPE simd
10009 @item -mfloat-gprs=@var{yes/no}
10010 @itemx -mfloat-gprs
10011 @opindex mfloat-gprs
10012 This switch enables or disables the generation of floating point
10013 operations on the general purpose registers for architectures that
10014 support it. This option is currently only available on the MPC8540.
10020 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
10021 targets (including GNU/Linux). The 32-bit environment sets int, long
10022 and pointer to 32 bits and generates code that runs on any PowerPC
10023 variant. The 64-bit environment sets int to 32 bits and long and
10024 pointer to 64 bits, and generates code for PowerPC64, as for
10025 @option{-mpowerpc64}.
10028 @itemx -mno-fp-in-toc
10029 @itemx -mno-sum-in-toc
10030 @itemx -mminimal-toc
10032 @opindex mno-fp-in-toc
10033 @opindex mno-sum-in-toc
10034 @opindex mminimal-toc
10035 Modify generation of the TOC (Table Of Contents), which is created for
10036 every executable file. The @option{-mfull-toc} option is selected by
10037 default. In that case, GCC will allocate at least one TOC entry for
10038 each unique non-automatic variable reference in your program. GCC
10039 will also place floating-point constants in the TOC@. However, only
10040 16,384 entries are available in the TOC@.
10042 If you receive a linker error message that saying you have overflowed
10043 the available TOC space, you can reduce the amount of TOC space used
10044 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
10045 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
10046 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
10047 generate code to calculate the sum of an address and a constant at
10048 run-time instead of putting that sum into the TOC@. You may specify one
10049 or both of these options. Each causes GCC to produce very slightly
10050 slower and larger code at the expense of conserving TOC space.
10052 If you still run out of space in the TOC even when you specify both of
10053 these options, specify @option{-mminimal-toc} instead. This option causes
10054 GCC to make only one TOC entry for every file. When you specify this
10055 option, GCC will produce code that is slower and larger but which
10056 uses extremely little TOC space. You may wish to use this option
10057 only on files that contain less frequently executed code.
10063 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
10064 @code{long} type, and the infrastructure needed to support them.
10065 Specifying @option{-maix64} implies @option{-mpowerpc64} and
10066 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
10067 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
10070 @itemx -mno-xl-call
10072 @opindex mno-xl-call
10073 On AIX, pass floating-point arguments to prototyped functions beyond the
10074 register save area (RSA) on the stack in addition to argument FPRs. The
10075 AIX calling convention was extended but not initially documented to
10076 handle an obscure K&R C case of calling a function that takes the
10077 address of its arguments with fewer arguments than declared. AIX XL
10078 compilers access floating point arguments which do not fit in the
10079 RSA from the stack when a subroutine is compiled without
10080 optimization. Because always storing floating-point arguments on the
10081 stack is inefficient and rarely needed, this option is not enabled by
10082 default and only is necessary when calling subroutines compiled by AIX
10083 XL compilers without optimization.
10087 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
10088 application written to use message passing with special startup code to
10089 enable the application to run. The system must have PE installed in the
10090 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
10091 must be overridden with the @option{-specs=} option to specify the
10092 appropriate directory location. The Parallel Environment does not
10093 support threads, so the @option{-mpe} option and the @option{-pthread}
10094 option are incompatible.
10096 @item -malign-natural
10097 @itemx -malign-power
10098 @opindex malign-natural
10099 @opindex malign-power
10100 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
10101 @option{-malign-natural} overrides the ABI-defined alignment of larger
10102 types, such as floating-point doubles, on their natural size-based boundary.
10103 The option @option{-malign-power} instructs GCC to follow the ABI-specified
10104 alignment rules. GCC defaults to the standard alignment defined in the ABI.
10107 @itemx -mhard-float
10108 @opindex msoft-float
10109 @opindex mhard-float
10110 Generate code that does not use (uses) the floating-point register set.
10111 Software floating point emulation is provided if you use the
10112 @option{-msoft-float} option, and pass the option to GCC when linking.
10115 @itemx -mno-multiple
10117 @opindex mno-multiple
10118 Generate code that uses (does not use) the load multiple word
10119 instructions and the store multiple word instructions. These
10120 instructions are generated by default on POWER systems, and not
10121 generated on PowerPC systems. Do not use @option{-mmultiple} on little
10122 endian PowerPC systems, since those instructions do not work when the
10123 processor is in little endian mode. The exceptions are PPC740 and
10124 PPC750 which permit the instructions usage in little endian mode.
10129 @opindex mno-string
10130 Generate code that uses (does not use) the load string instructions
10131 and the store string word instructions to save multiple registers and
10132 do small block moves. These instructions are generated by default on
10133 POWER systems, and not generated on PowerPC systems. Do not use
10134 @option{-mstring} on little endian PowerPC systems, since those
10135 instructions do not work when the processor is in little endian mode.
10136 The exceptions are PPC740 and PPC750 which permit the instructions
10137 usage in little endian mode.
10142 @opindex mno-update
10143 Generate code that uses (does not use) the load or store instructions
10144 that update the base register to the address of the calculated memory
10145 location. These instructions are generated by default. If you use
10146 @option{-mno-update}, there is a small window between the time that the
10147 stack pointer is updated and the address of the previous frame is
10148 stored, which means code that walks the stack frame across interrupts or
10149 signals may get corrupted data.
10152 @itemx -mno-fused-madd
10153 @opindex mfused-madd
10154 @opindex mno-fused-madd
10155 Generate code that uses (does not use) the floating point multiply and
10156 accumulate instructions. These instructions are generated by default if
10157 hardware floating is used.
10159 @item -mno-bit-align
10161 @opindex mno-bit-align
10162 @opindex mbit-align
10163 On System V.4 and embedded PowerPC systems do not (do) force structures
10164 and unions that contain bit-fields to be aligned to the base type of the
10167 For example, by default a structure containing nothing but 8
10168 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
10169 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
10170 the structure would be aligned to a 1 byte boundary and be one byte in
10173 @item -mno-strict-align
10174 @itemx -mstrict-align
10175 @opindex mno-strict-align
10176 @opindex mstrict-align
10177 On System V.4 and embedded PowerPC systems do not (do) assume that
10178 unaligned memory references will be handled by the system.
10180 @item -mrelocatable
10181 @itemx -mno-relocatable
10182 @opindex mrelocatable
10183 @opindex mno-relocatable
10184 On embedded PowerPC systems generate code that allows (does not allow)
10185 the program to be relocated to a different address at runtime. If you
10186 use @option{-mrelocatable} on any module, all objects linked together must
10187 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
10189 @item -mrelocatable-lib
10190 @itemx -mno-relocatable-lib
10191 @opindex mrelocatable-lib
10192 @opindex mno-relocatable-lib
10193 On embedded PowerPC systems generate code that allows (does not allow)
10194 the program to be relocated to a different address at runtime. Modules
10195 compiled with @option{-mrelocatable-lib} can be linked with either modules
10196 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
10197 with modules compiled with the @option{-mrelocatable} options.
10203 On System V.4 and embedded PowerPC systems do not (do) assume that
10204 register 2 contains a pointer to a global area pointing to the addresses
10205 used in the program.
10208 @itemx -mlittle-endian
10210 @opindex mlittle-endian
10211 On System V.4 and embedded PowerPC systems compile code for the
10212 processor in little endian mode. The @option{-mlittle-endian} option is
10213 the same as @option{-mlittle}.
10216 @itemx -mbig-endian
10218 @opindex mbig-endian
10219 On System V.4 and embedded PowerPC systems compile code for the
10220 processor in big endian mode. The @option{-mbig-endian} option is
10221 the same as @option{-mbig}.
10223 @item -mdynamic-no-pic
10224 @opindex mdynamic-no-pic
10225 On Darwin and Mac OS X systems, compile code so that it is not
10226 relocatable, but that its external references are relocatable. The
10227 resulting code is suitable for applications, but not shared
10230 @item -mprioritize-restricted-insns=@var{priority}
10231 @opindex mprioritize-restricted-insns
10232 This option controls the priority that is assigned to
10233 dispatch-slot restricted instructions during the second scheduling
10234 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
10235 @var{no/highest/second-highest} priority to dispatch slot restricted
10238 @item -msched-costly-dep=@var{dependence_type}
10239 @opindex msched-costly-dep
10240 This option controls which dependences are considered costly
10241 by the target during instruction scheduling. The argument
10242 @var{dependence_type} takes one of the following values:
10243 @var{no}: no dependence is costly,
10244 @var{all}: all dependences are costly,
10245 @var{true_store_to_load}: a true dependence from store to load is costly,
10246 @var{store_to_load}: any dependence from store to load is costly,
10247 @var{number}: any dependence which latency >= @var{number} is costly.
10249 @item -minsert-sched-nops=@var{scheme}
10250 @opindex minsert-sched-nops
10251 This option controls which nop insertion scheme will be used during
10252 the second scheduling pass. The argument @var{scheme} takes one of the
10254 @var{no}: Don't insert nops.
10255 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
10256 according to the scheduler's grouping.
10257 @var{regroup_exact}: Insert nops to force costly dependent insns into
10258 separate groups. Insert exactly as many nops as needed to force an insn
10259 to a new group, according to the estimated processor grouping.
10260 @var{number}: Insert nops to force costly dependent insns into
10261 separate groups. Insert @var{number} nops to force an insn to a new group.
10264 @opindex mcall-sysv
10265 On System V.4 and embedded PowerPC systems compile code using calling
10266 conventions that adheres to the March 1995 draft of the System V
10267 Application Binary Interface, PowerPC processor supplement. This is the
10268 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
10270 @item -mcall-sysv-eabi
10271 @opindex mcall-sysv-eabi
10272 Specify both @option{-mcall-sysv} and @option{-meabi} options.
10274 @item -mcall-sysv-noeabi
10275 @opindex mcall-sysv-noeabi
10276 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
10278 @item -mcall-solaris
10279 @opindex mcall-solaris
10280 On System V.4 and embedded PowerPC systems compile code for the Solaris
10284 @opindex mcall-linux
10285 On System V.4 and embedded PowerPC systems compile code for the
10286 Linux-based GNU system.
10290 On System V.4 and embedded PowerPC systems compile code for the
10291 Hurd-based GNU system.
10293 @item -mcall-netbsd
10294 @opindex mcall-netbsd
10295 On System V.4 and embedded PowerPC systems compile code for the
10296 NetBSD operating system.
10298 @item -maix-struct-return
10299 @opindex maix-struct-return
10300 Return all structures in memory (as specified by the AIX ABI)@.
10302 @item -msvr4-struct-return
10303 @opindex msvr4-struct-return
10304 Return structures smaller than 8 bytes in registers (as specified by the
10307 @item -mabi=altivec
10308 @opindex mabi=altivec
10309 Extend the current ABI with AltiVec ABI extensions. This does not
10310 change the default ABI, instead it adds the AltiVec ABI extensions to
10313 @item -mabi=no-altivec
10314 @opindex mabi=no-altivec
10315 Disable AltiVec ABI extensions for the current ABI.
10318 @itemx -mno-prototype
10319 @opindex mprototype
10320 @opindex mno-prototype
10321 On System V.4 and embedded PowerPC systems assume that all calls to
10322 variable argument functions are properly prototyped. Otherwise, the
10323 compiler must insert an instruction before every non prototyped call to
10324 set or clear bit 6 of the condition code register (@var{CR}) to
10325 indicate whether floating point values were passed in the floating point
10326 registers in case the function takes a variable arguments. With
10327 @option{-mprototype}, only calls to prototyped variable argument functions
10328 will set or clear the bit.
10332 On embedded PowerPC systems, assume that the startup module is called
10333 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
10334 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
10339 On embedded PowerPC systems, assume that the startup module is called
10340 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
10345 On embedded PowerPC systems, assume that the startup module is called
10346 @file{crt0.o} and the standard C libraries are @file{libads.a} and
10349 @item -myellowknife
10350 @opindex myellowknife
10351 On embedded PowerPC systems, assume that the startup module is called
10352 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
10357 On System V.4 and embedded PowerPC systems, specify that you are
10358 compiling for a VxWorks system.
10362 Specify that you are compiling for the WindISS simulation environment.
10366 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
10367 header to indicate that @samp{eabi} extended relocations are used.
10373 On System V.4 and embedded PowerPC systems do (do not) adhere to the
10374 Embedded Applications Binary Interface (eabi) which is a set of
10375 modifications to the System V.4 specifications. Selecting @option{-meabi}
10376 means that the stack is aligned to an 8 byte boundary, a function
10377 @code{__eabi} is called to from @code{main} to set up the eabi
10378 environment, and the @option{-msdata} option can use both @code{r2} and
10379 @code{r13} to point to two separate small data areas. Selecting
10380 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
10381 do not call an initialization function from @code{main}, and the
10382 @option{-msdata} option will only use @code{r13} to point to a single
10383 small data area. The @option{-meabi} option is on by default if you
10384 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
10387 @opindex msdata=eabi
10388 On System V.4 and embedded PowerPC systems, put small initialized
10389 @code{const} global and static data in the @samp{.sdata2} section, which
10390 is pointed to by register @code{r2}. Put small initialized
10391 non-@code{const} global and static data in the @samp{.sdata} section,
10392 which is pointed to by register @code{r13}. Put small uninitialized
10393 global and static data in the @samp{.sbss} section, which is adjacent to
10394 the @samp{.sdata} section. The @option{-msdata=eabi} option is
10395 incompatible with the @option{-mrelocatable} option. The
10396 @option{-msdata=eabi} option also sets the @option{-memb} option.
10399 @opindex msdata=sysv
10400 On System V.4 and embedded PowerPC systems, put small global and static
10401 data in the @samp{.sdata} section, which is pointed to by register
10402 @code{r13}. Put small uninitialized global and static data in the
10403 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
10404 The @option{-msdata=sysv} option is incompatible with the
10405 @option{-mrelocatable} option.
10407 @item -msdata=default
10409 @opindex msdata=default
10411 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
10412 compile code the same as @option{-msdata=eabi}, otherwise compile code the
10413 same as @option{-msdata=sysv}.
10416 @opindex msdata-data
10417 On System V.4 and embedded PowerPC systems, put small global and static
10418 data in the @samp{.sdata} section. Put small uninitialized global and
10419 static data in the @samp{.sbss} section. Do not use register @code{r13}
10420 to address small data however. This is the default behavior unless
10421 other @option{-msdata} options are used.
10425 @opindex msdata=none
10427 On embedded PowerPC systems, put all initialized global and static data
10428 in the @samp{.data} section, and all uninitialized data in the
10429 @samp{.bss} section.
10433 @cindex smaller data references (PowerPC)
10434 @cindex .sdata/.sdata2 references (PowerPC)
10435 On embedded PowerPC systems, put global and static items less than or
10436 equal to @var{num} bytes into the small data or bss sections instead of
10437 the normal data or bss section. By default, @var{num} is 8. The
10438 @option{-G @var{num}} switch is also passed to the linker.
10439 All modules should be compiled with the same @option{-G @var{num}} value.
10442 @itemx -mno-regnames
10444 @opindex mno-regnames
10445 On System V.4 and embedded PowerPC systems do (do not) emit register
10446 names in the assembly language output using symbolic forms.
10449 @itemx -mno-longcall
10451 @opindex mno-longcall
10452 Default to making all function calls indirectly, using a register, so
10453 that functions which reside further than 32 megabytes (33,554,432
10454 bytes) from the current location can be called. This setting can be
10455 overridden by the @code{shortcall} function attribute, or by
10456 @code{#pragma longcall(0)}.
10458 Some linkers are capable of detecting out-of-range calls and generating
10459 glue code on the fly. On these systems, long calls are unnecessary and
10460 generate slower code. As of this writing, the AIX linker can do this,
10461 as can the GNU linker for PowerPC/64. It is planned to add this feature
10462 to the GNU linker for 32-bit PowerPC systems as well.
10464 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
10465 callee, L42'', plus a ``branch island'' (glue code). The two target
10466 addresses represent the callee and the ``branch island.'' The
10467 Darwin/PPC linker will prefer the first address and generate a ``bl
10468 callee'' if the PPC ``bl'' instruction will reach the callee directly;
10469 otherwise, the linker will generate ``bl L42'' to call the ``branch
10470 island.'' The ``branch island'' is appended to the body of the
10471 calling function; it computes the full 32-bit address of the callee
10474 On Mach-O (Darwin) systems, this option directs the compiler emit to
10475 the glue for every direct call, and the Darwin linker decides whether
10476 to use or discard it.
10478 In the future, we may cause GCC to ignore all longcall specifications
10479 when the linker is known to generate glue.
10483 Adds support for multithreading with the @dfn{pthreads} library.
10484 This option sets flags for both the preprocessor and linker.
10488 @node S/390 and zSeries Options
10489 @subsection S/390 and zSeries Options
10490 @cindex S/390 and zSeries Options
10492 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10496 @itemx -msoft-float
10497 @opindex mhard-float
10498 @opindex msoft-float
10499 Use (do not use) the hardware floating-point instructions and registers
10500 for floating-point operations. When @option{-msoft-float} is specified,
10501 functions in @file{libgcc.a} will be used to perform floating-point
10502 operations. When @option{-mhard-float} is specified, the compiler
10503 generates IEEE floating-point instructions. This is the default.
10506 @itemx -mno-backchain
10507 @itemx -mkernel-backchain
10508 @opindex mbackchain
10509 @opindex mno-backchain
10510 @opindex mkernel-backchain
10511 In order to provide a backchain the address of the caller's frame
10512 is stored within the callee's stack frame.
10513 A backchain may be needed to allow debugging using tools that do not understand
10514 DWARF-2 call frame information.
10515 For @option{-mno-backchain} no backchain is maintained at all which is the
10517 If one of the other options is present the backchain pointer is placed either
10518 on top of the stack frame (@option{-mkernel-backchain}) or on
10519 the bottom (@option{-mbackchain}).
10520 Beside the different backchain location @option{-mkernel-backchain}
10521 also changes stack frame layout breaking the ABI. This option
10522 is intended to be used for code which internally needs a backchain but has
10523 to get by with a limited stack size e.g. the linux kernel.
10524 Internal unwinding code not using DWARF-2 info has to be able to locate the
10525 return address of a function. That will be eased be the fact that
10526 the return address of a function is placed two words below the backchain
10530 @itemx -mno-small-exec
10531 @opindex msmall-exec
10532 @opindex mno-small-exec
10533 Generate (or do not generate) code using the @code{bras} instruction
10534 to do subroutine calls.
10535 This only works reliably if the total executable size does not
10536 exceed 64k. The default is to use the @code{basr} instruction instead,
10537 which does not have this limitation.
10543 When @option{-m31} is specified, generate code compliant to the
10544 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10545 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10546 particular to generate 64-bit instructions. For the @samp{s390}
10547 targets, the default is @option{-m31}, while the @samp{s390x}
10548 targets default to @option{-m64}.
10554 When @option{-mzarch} is specified, generate code using the
10555 instructions available on z/Architecture.
10556 When @option{-mesa} is specified, generate code using the
10557 instructions available on ESA/390. Note that @option{-mesa} is
10558 not possible with @option{-m64}.
10559 When generating code compliant to the GNU/Linux for S/390 ABI,
10560 the default is @option{-mesa}. When generating code compliant
10561 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10567 Generate (or do not generate) code using the @code{mvcle} instruction
10568 to perform block moves. When @option{-mno-mvcle} is specified,
10569 use a @code{mvc} loop instead. This is the default.
10575 Print (or do not print) additional debug information when compiling.
10576 The default is to not print debug information.
10578 @item -march=@var{cpu-type}
10580 Generate code that will run on @var{cpu-type}, which is the name of a system
10581 representing a certain processor type. Possible values for
10582 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10583 When generating code using the instructions available on z/Architecture,
10584 the default is @option{-march=z900}. Otherwise, the default is
10585 @option{-march=g5}.
10587 @item -mtune=@var{cpu-type}
10589 Tune to @var{cpu-type} everything applicable about the generated code,
10590 except for the ABI and the set of available instructions.
10591 The list of @var{cpu-type} values is the same as for @option{-march}.
10592 The default is the value used for @option{-march}.
10595 @itemx -mno-tpf-trace
10596 @opindex mtpf-trace
10597 @opindex mno-tpf-trace
10598 Generate code that adds (does not add) in TPF OS specific branches to trace
10599 routines in the operating system. This option is off by default, even
10600 when compiling for the TPF OS.
10603 @itemx -mno-fused-madd
10604 @opindex mfused-madd
10605 @opindex mno-fused-madd
10606 Generate code that uses (does not use) the floating point multiply and
10607 accumulate instructions. These instructions are generated by default if
10608 hardware floating point is used.
10610 @item -mwarn-framesize=@var{framesize}
10611 @opindex mwarn-framesize
10612 Emit a warning if the current function exceeds the given frame size. Because
10613 this is a compile time check it doesn't need to be a real problem when the program
10614 runs. It is intended to identify functions which most probably cause
10615 a stack overflow. It is useful to be used in an environment with limited stack
10616 size e.g. the linux kernel.
10618 @item -mwarn-dynamicstack
10619 @opindex mwarn-dynamicstack
10620 Emit a warning if the function calls alloca or uses dynamically
10621 sized arrays. This is generally a bad idea with a limited stack size.
10623 @item -mstack-guard=@var{stack-guard}
10624 @item -mstack-size=@var{stack-size}
10625 @opindex mstack-guard
10626 @opindex mstack-size
10627 These arguments always have to be used in conjunction. If they are present the s390
10628 back end emits additional instructions in the function prologue which trigger a trap
10629 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
10630 (remember that the stack on s390 grows downward). These options are intended to
10631 be used to help debugging stack overflow problems. The additionally emitted code
10632 cause only little overhead and hence can also be used in production like systems
10633 without greater performance degradation. The given values have to be exact
10634 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard}.
10635 In order to be efficient the extra code makes the assumption that the stack starts
10636 at an address aligned to the value given by @var{stack-size}.
10640 @subsection SH Options
10642 These @samp{-m} options are defined for the SH implementations:
10647 Generate code for the SH1.
10651 Generate code for the SH2.
10654 Generate code for the SH2e.
10658 Generate code for the SH3.
10662 Generate code for the SH3e.
10666 Generate code for the SH4 without a floating-point unit.
10668 @item -m4-single-only
10669 @opindex m4-single-only
10670 Generate code for the SH4 with a floating-point unit that only
10671 supports single-precision arithmetic.
10675 Generate code for the SH4 assuming the floating-point unit is in
10676 single-precision mode by default.
10680 Generate code for the SH4.
10684 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
10685 floating-point unit is not used.
10687 @item -m4a-single-only
10688 @opindex m4a-single-only
10689 Generate code for the SH4a, in such a way that no double-precision
10690 floating point operations are used.
10693 @opindex m4a-single
10694 Generate code for the SH4a assuming the floating-point unit is in
10695 single-precision mode by default.
10699 Generate code for the SH4a.
10703 Same as @option{-m4a-nofpu}, except that it implicitly passes
10704 @option{-dsp} to the assembler. GCC doesn't generate any DSP
10705 instructions at the moment.
10709 Compile code for the processor in big endian mode.
10713 Compile code for the processor in little endian mode.
10717 Align doubles at 64-bit boundaries. Note that this changes the calling
10718 conventions, and thus some functions from the standard C library will
10719 not work unless you recompile it first with @option{-mdalign}.
10723 Shorten some address references at link time, when possible; uses the
10724 linker option @option{-relax}.
10728 Use 32-bit offsets in @code{switch} tables. The default is to use
10733 Enable the use of the instruction @code{fmovd}.
10737 Comply with the calling conventions defined by Renesas.
10741 Comply with the calling conventions defined by Renesas.
10745 Comply with the calling conventions defined for GCC before the Renesas
10746 conventions were available. This option is the default for all
10747 targets of the SH toolchain except for @samp{sh-symbianelf}.
10750 @opindex mnomacsave
10751 Mark the @code{MAC} register as call-clobbered, even if
10752 @option{-mhitachi} is given.
10756 Increase IEEE-compliance of floating-point code.
10760 Dump instruction size and location in the assembly code.
10763 @opindex mpadstruct
10764 This option is deprecated. It pads structures to multiple of 4 bytes,
10765 which is incompatible with the SH ABI@.
10769 Optimize for space instead of speed. Implied by @option{-Os}.
10772 @opindex mprefergot
10773 When generating position-independent code, emit function calls using
10774 the Global Offset Table instead of the Procedure Linkage Table.
10778 Generate a library function call to invalidate instruction cache
10779 entries, after fixing up a trampoline. This library function call
10780 doesn't assume it can write to the whole memory address space. This
10781 is the default when the target is @code{sh-*-linux*}.
10784 @node SPARC Options
10785 @subsection SPARC Options
10786 @cindex SPARC options
10788 These @samp{-m} options are supported on the SPARC:
10791 @item -mno-app-regs
10793 @opindex mno-app-regs
10795 Specify @option{-mapp-regs} to generate output using the global registers
10796 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
10799 To be fully SVR4 ABI compliant at the cost of some performance loss,
10800 specify @option{-mno-app-regs}. You should compile libraries and system
10801 software with this option.
10804 @itemx -mhard-float
10806 @opindex mhard-float
10807 Generate output containing floating point instructions. This is the
10811 @itemx -msoft-float
10813 @opindex msoft-float
10814 Generate output containing library calls for floating point.
10815 @strong{Warning:} the requisite libraries are not available for all SPARC
10816 targets. Normally the facilities of the machine's usual C compiler are
10817 used, but this cannot be done directly in cross-compilation. You must make
10818 your own arrangements to provide suitable library functions for
10819 cross-compilation. The embedded targets @samp{sparc-*-aout} and
10820 @samp{sparclite-*-*} do provide software floating point support.
10822 @option{-msoft-float} changes the calling convention in the output file;
10823 therefore, it is only useful if you compile @emph{all} of a program with
10824 this option. In particular, you need to compile @file{libgcc.a}, the
10825 library that comes with GCC, with @option{-msoft-float} in order for
10828 @item -mhard-quad-float
10829 @opindex mhard-quad-float
10830 Generate output containing quad-word (long double) floating point
10833 @item -msoft-quad-float
10834 @opindex msoft-quad-float
10835 Generate output containing library calls for quad-word (long double)
10836 floating point instructions. The functions called are those specified
10837 in the SPARC ABI@. This is the default.
10839 As of this writing, there are no SPARC implementations that have hardware
10840 support for the quad-word floating point instructions. They all invoke
10841 a trap handler for one of these instructions, and then the trap handler
10842 emulates the effect of the instruction. Because of the trap handler overhead,
10843 this is much slower than calling the ABI library routines. Thus the
10844 @option{-msoft-quad-float} option is the default.
10846 @item -mno-unaligned-doubles
10847 @itemx -munaligned-doubles
10848 @opindex mno-unaligned-doubles
10849 @opindex munaligned-doubles
10850 Assume that doubles have 8 byte alignment. This is the default.
10852 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
10853 alignment only if they are contained in another type, or if they have an
10854 absolute address. Otherwise, it assumes they have 4 byte alignment.
10855 Specifying this option avoids some rare compatibility problems with code
10856 generated by other compilers. It is not the default because it results
10857 in a performance loss, especially for floating point code.
10859 @item -mno-faster-structs
10860 @itemx -mfaster-structs
10861 @opindex mno-faster-structs
10862 @opindex mfaster-structs
10863 With @option{-mfaster-structs}, the compiler assumes that structures
10864 should have 8 byte alignment. This enables the use of pairs of
10865 @code{ldd} and @code{std} instructions for copies in structure
10866 assignment, in place of twice as many @code{ld} and @code{st} pairs.
10867 However, the use of this changed alignment directly violates the SPARC
10868 ABI@. Thus, it's intended only for use on targets where the developer
10869 acknowledges that their resulting code will not be directly in line with
10870 the rules of the ABI@.
10872 @item -mimpure-text
10873 @opindex mimpure-text
10874 @option{-mimpure-text}, used in addition to @option{-shared}, tells
10875 the compiler to not pass @option{-z text} to the linker when linking a
10876 shared object. Using this option, you can link position-dependent
10877 code into a shared object.
10879 @option{-mimpure-text} suppresses the ``relocations remain against
10880 allocatable but non-writable sections'' linker error message.
10881 However, the necessary relocations will trigger copy-on-write, and the
10882 shared object is not actually shared across processes. Instead of
10883 using @option{-mimpure-text}, you should compile all source code with
10884 @option{-fpic} or @option{-fPIC}.
10886 This option is only available on SunOS and Solaris.
10888 @item -mcpu=@var{cpu_type}
10890 Set the instruction set, register set, and instruction scheduling parameters
10891 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
10892 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
10893 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
10894 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
10895 @samp{ultrasparc3}.
10897 Default instruction scheduling parameters are used for values that select
10898 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
10899 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
10901 Here is a list of each supported architecture and their supported
10906 v8: supersparc, hypersparc
10907 sparclite: f930, f934, sparclite86x
10909 v9: ultrasparc, ultrasparc3
10912 By default (unless configured otherwise), GCC generates code for the V7
10913 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
10914 additionally optimizes it for the Cypress CY7C602 chip, as used in the
10915 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
10916 SPARCStation 1, 2, IPX etc.
10918 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
10919 architecture. The only difference from V7 code is that the compiler emits
10920 the integer multiply and integer divide instructions which exist in SPARC-V8
10921 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
10922 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
10925 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
10926 the SPARC architecture. This adds the integer multiply, integer divide step
10927 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
10928 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
10929 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
10930 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
10931 MB86934 chip, which is the more recent SPARClite with FPU.
10933 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
10934 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
10935 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
10936 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
10937 optimizes it for the TEMIC SPARClet chip.
10939 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
10940 architecture. This adds 64-bit integer and floating-point move instructions,
10941 3 additional floating-point condition code registers and conditional move
10942 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
10943 optimizes it for the Sun UltraSPARC I/II chips. With
10944 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
10945 Sun UltraSPARC III chip.
10947 @item -mtune=@var{cpu_type}
10949 Set the instruction scheduling parameters for machine type
10950 @var{cpu_type}, but do not set the instruction set or register set that the
10951 option @option{-mcpu=@var{cpu_type}} would.
10953 The same values for @option{-mcpu=@var{cpu_type}} can be used for
10954 @option{-mtune=@var{cpu_type}}, but the only useful values are those
10955 that select a particular cpu implementation. Those are @samp{cypress},
10956 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
10957 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
10958 @samp{ultrasparc3}.
10963 @opindex mno-v8plus
10964 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
10965 difference from the V8 ABI is that the global and out registers are
10966 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
10967 mode for all SPARC-V9 processors.
10973 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
10974 Visual Instruction Set extensions. The default is @option{-mno-vis}.
10977 These @samp{-m} options are supported in addition to the above
10978 on SPARC-V9 processors in 64-bit environments:
10981 @item -mlittle-endian
10982 @opindex mlittle-endian
10983 Generate code for a processor running in little-endian mode. It is only
10984 available for a few configurations and most notably not on Solaris.
10990 Generate code for a 32-bit or 64-bit environment.
10991 The 32-bit environment sets int, long and pointer to 32 bits.
10992 The 64-bit environment sets int to 32 bits and long and pointer
10995 @item -mcmodel=medlow
10996 @opindex mcmodel=medlow
10997 Generate code for the Medium/Low code model: 64-bit addresses, programs
10998 must be linked in the low 32 bits of memory. Programs can be statically
10999 or dynamically linked.
11001 @item -mcmodel=medmid
11002 @opindex mcmodel=medmid
11003 Generate code for the Medium/Middle code model: 64-bit addresses, programs
11004 must be linked in the low 44 bits of memory, the text and data segments must
11005 be less than 2GB in size and the data segment must be located within 2GB of
11008 @item -mcmodel=medany
11009 @opindex mcmodel=medany
11010 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
11011 may be linked anywhere in memory, the text and data segments must be less
11012 than 2GB in size and the data segment must be located within 2GB of the
11015 @item -mcmodel=embmedany
11016 @opindex mcmodel=embmedany
11017 Generate code for the Medium/Anywhere code model for embedded systems:
11018 64-bit addresses, the text and data segments must be less than 2GB in
11019 size, both starting anywhere in memory (determined at link time). The
11020 global register %g4 points to the base of the data segment. Programs
11021 are statically linked and PIC is not supported.
11024 @itemx -mno-stack-bias
11025 @opindex mstack-bias
11026 @opindex mno-stack-bias
11027 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
11028 frame pointer if present, are offset by @minus{}2047 which must be added back
11029 when making stack frame references. This is the default in 64-bit mode.
11030 Otherwise, assume no such offset is present.
11033 These switches are supported in addition to the above on Solaris:
11038 Add support for multithreading using the Solaris threads library. This
11039 option sets flags for both the preprocessor and linker. This option does
11040 not affect the thread safety of object code produced by the compiler or
11041 that of libraries supplied with it.
11045 Add support for multithreading using the POSIX threads library. This
11046 option sets flags for both the preprocessor and linker. This option does
11047 not affect the thread safety of object code produced by the compiler or
11048 that of libraries supplied with it.
11051 @node System V Options
11052 @subsection Options for System V
11054 These additional options are available on System V Release 4 for
11055 compatibility with other compilers on those systems:
11060 Create a shared object.
11061 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
11065 Identify the versions of each tool used by the compiler, in a
11066 @code{.ident} assembler directive in the output.
11070 Refrain from adding @code{.ident} directives to the output file (this is
11073 @item -YP,@var{dirs}
11075 Search the directories @var{dirs}, and no others, for libraries
11076 specified with @option{-l}.
11078 @item -Ym,@var{dir}
11080 Look in the directory @var{dir} to find the M4 preprocessor.
11081 The assembler uses this option.
11082 @c This is supposed to go with a -Yd for predefined M4 macro files, but
11083 @c the generic assembler that comes with Solaris takes just -Ym.
11086 @node TMS320C3x/C4x Options
11087 @subsection TMS320C3x/C4x Options
11088 @cindex TMS320C3x/C4x Options
11090 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
11094 @item -mcpu=@var{cpu_type}
11096 Set the instruction set, register set, and instruction scheduling
11097 parameters for machine type @var{cpu_type}. Supported values for
11098 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
11099 @samp{c44}. The default is @samp{c40} to generate code for the
11104 @itemx -msmall-memory
11106 @opindex mbig-memory
11108 @opindex msmall-memory
11110 Generates code for the big or small memory model. The small memory
11111 model assumed that all data fits into one 64K word page. At run-time
11112 the data page (DP) register must be set to point to the 64K page
11113 containing the .bss and .data program sections. The big memory model is
11114 the default and requires reloading of the DP register for every direct
11121 Allow (disallow) allocation of general integer operands into the block
11122 count register BK@.
11128 Enable (disable) generation of code using decrement and branch,
11129 DBcond(D), instructions. This is enabled by default for the C4x. To be
11130 on the safe side, this is disabled for the C3x, since the maximum
11131 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
11132 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
11133 that it can utilize the decrement and branch instruction, but will give
11134 up if there is more than one memory reference in the loop. Thus a loop
11135 where the loop counter is decremented can generate slightly more
11136 efficient code, in cases where the RPTB instruction cannot be utilized.
11138 @item -mdp-isr-reload
11140 @opindex mdp-isr-reload
11142 Force the DP register to be saved on entry to an interrupt service
11143 routine (ISR), reloaded to point to the data section, and restored on
11144 exit from the ISR@. This should not be required unless someone has
11145 violated the small memory model by modifying the DP register, say within
11152 For the C3x use the 24-bit MPYI instruction for integer multiplies
11153 instead of a library call to guarantee 32-bit results. Note that if one
11154 of the operands is a constant, then the multiplication will be performed
11155 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
11156 then squaring operations are performed inline instead of a library call.
11159 @itemx -mno-fast-fix
11161 @opindex mno-fast-fix
11162 The C3x/C4x FIX instruction to convert a floating point value to an
11163 integer value chooses the nearest integer less than or equal to the
11164 floating point value rather than to the nearest integer. Thus if the
11165 floating point number is negative, the result will be incorrectly
11166 truncated an additional code is necessary to detect and correct this
11167 case. This option can be used to disable generation of the additional
11168 code required to correct the result.
11174 Enable (disable) generation of repeat block sequences using the RPTB
11175 instruction for zero overhead looping. The RPTB construct is only used
11176 for innermost loops that do not call functions or jump across the loop
11177 boundaries. There is no advantage having nested RPTB loops due to the
11178 overhead required to save and restore the RC, RS, and RE registers.
11179 This is enabled by default with @option{-O2}.
11181 @item -mrpts=@var{count}
11185 Enable (disable) the use of the single instruction repeat instruction
11186 RPTS@. If a repeat block contains a single instruction, and the loop
11187 count can be guaranteed to be less than the value @var{count}, GCC will
11188 emit a RPTS instruction instead of a RPTB@. If no value is specified,
11189 then a RPTS will be emitted even if the loop count cannot be determined
11190 at compile time. Note that the repeated instruction following RPTS does
11191 not have to be reloaded from memory each iteration, thus freeing up the
11192 CPU buses for operands. However, since interrupts are blocked by this
11193 instruction, it is disabled by default.
11195 @item -mloop-unsigned
11196 @itemx -mno-loop-unsigned
11197 @opindex mloop-unsigned
11198 @opindex mno-loop-unsigned
11199 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
11200 is @math{2^{31} + 1} since these instructions test if the iteration count is
11201 negative to terminate the loop. If the iteration count is unsigned
11202 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
11203 exceeded. This switch allows an unsigned iteration count.
11207 Try to emit an assembler syntax that the TI assembler (asm30) is happy
11208 with. This also enforces compatibility with the API employed by the TI
11209 C3x C compiler. For example, long doubles are passed as structures
11210 rather than in floating point registers.
11216 Generate code that uses registers (stack) for passing arguments to functions.
11217 By default, arguments are passed in registers where possible rather
11218 than by pushing arguments on to the stack.
11220 @item -mparallel-insns
11221 @itemx -mno-parallel-insns
11222 @opindex mparallel-insns
11223 @opindex mno-parallel-insns
11224 Allow the generation of parallel instructions. This is enabled by
11225 default with @option{-O2}.
11227 @item -mparallel-mpy
11228 @itemx -mno-parallel-mpy
11229 @opindex mparallel-mpy
11230 @opindex mno-parallel-mpy
11231 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
11232 provided @option{-mparallel-insns} is also specified. These instructions have
11233 tight register constraints which can pessimize the code generation
11234 of large functions.
11239 @subsection V850 Options
11240 @cindex V850 Options
11242 These @samp{-m} options are defined for V850 implementations:
11246 @itemx -mno-long-calls
11247 @opindex mlong-calls
11248 @opindex mno-long-calls
11249 Treat all calls as being far away (near). If calls are assumed to be
11250 far away, the compiler will always load the functions address up into a
11251 register, and call indirect through the pointer.
11257 Do not optimize (do optimize) basic blocks that use the same index
11258 pointer 4 or more times to copy pointer into the @code{ep} register, and
11259 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
11260 option is on by default if you optimize.
11262 @item -mno-prolog-function
11263 @itemx -mprolog-function
11264 @opindex mno-prolog-function
11265 @opindex mprolog-function
11266 Do not use (do use) external functions to save and restore registers
11267 at the prologue and epilogue of a function. The external functions
11268 are slower, but use less code space if more than one function saves
11269 the same number of registers. The @option{-mprolog-function} option
11270 is on by default if you optimize.
11274 Try to make the code as small as possible. At present, this just turns
11275 on the @option{-mep} and @option{-mprolog-function} options.
11277 @item -mtda=@var{n}
11279 Put static or global variables whose size is @var{n} bytes or less into
11280 the tiny data area that register @code{ep} points to. The tiny data
11281 area can hold up to 256 bytes in total (128 bytes for byte references).
11283 @item -msda=@var{n}
11285 Put static or global variables whose size is @var{n} bytes or less into
11286 the small data area that register @code{gp} points to. The small data
11287 area can hold up to 64 kilobytes.
11289 @item -mzda=@var{n}
11291 Put static or global variables whose size is @var{n} bytes or less into
11292 the first 32 kilobytes of memory.
11296 Specify that the target processor is the V850.
11299 @opindex mbig-switch
11300 Generate code suitable for big switch tables. Use this option only if
11301 the assembler/linker complain about out of range branches within a switch
11306 This option will cause r2 and r5 to be used in the code generated by
11307 the compiler. This setting is the default.
11309 @item -mno-app-regs
11310 @opindex mno-app-regs
11311 This option will cause r2 and r5 to be treated as fixed registers.
11315 Specify that the target processor is the V850E1. The preprocessor
11316 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
11317 this option is used.
11321 Specify that the target processor is the V850E. The preprocessor
11322 constant @samp{__v850e__} will be defined if this option is used.
11324 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
11325 are defined then a default target processor will be chosen and the
11326 relevant @samp{__v850*__} preprocessor constant will be defined.
11328 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
11329 defined, regardless of which processor variant is the target.
11331 @item -mdisable-callt
11332 @opindex mdisable-callt
11333 This option will suppress generation of the CALLT instruction for the
11334 v850e and v850e1 flavors of the v850 architecture. The default is
11335 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
11340 @subsection VAX Options
11341 @cindex VAX options
11343 These @samp{-m} options are defined for the VAX:
11348 Do not output certain jump instructions (@code{aobleq} and so on)
11349 that the Unix assembler for the VAX cannot handle across long
11354 Do output those jump instructions, on the assumption that you
11355 will assemble with the GNU assembler.
11359 Output code for g-format floating point numbers instead of d-format.
11362 @node x86-64 Options
11363 @subsection x86-64 Options
11364 @cindex x86-64 options
11366 These are listed under @xref{i386 and x86-64 Options}.
11368 @node Xstormy16 Options
11369 @subsection Xstormy16 Options
11370 @cindex Xstormy16 Options
11372 These options are defined for Xstormy16:
11377 Choose startup files and linker script suitable for the simulator.
11380 @node Xtensa Options
11381 @subsection Xtensa Options
11382 @cindex Xtensa Options
11384 These options are supported for Xtensa targets:
11388 @itemx -mno-const16
11390 @opindex mno-const16
11391 Enable or disable use of @code{CONST16} instructions for loading
11392 constant values. The @code{CONST16} instruction is currently not a
11393 standard option from Tensilica. When enabled, @code{CONST16}
11394 instructions are always used in place of the standard @code{L32R}
11395 instructions. The use of @code{CONST16} is enabled by default only if
11396 the @code{L32R} instruction is not available.
11399 @itemx -mno-fused-madd
11400 @opindex mfused-madd
11401 @opindex mno-fused-madd
11402 Enable or disable use of fused multiply/add and multiply/subtract
11403 instructions in the floating-point option. This has no effect if the
11404 floating-point option is not also enabled. Disabling fused multiply/add
11405 and multiply/subtract instructions forces the compiler to use separate
11406 instructions for the multiply and add/subtract operations. This may be
11407 desirable in some cases where strict IEEE 754-compliant results are
11408 required: the fused multiply add/subtract instructions do not round the
11409 intermediate result, thereby producing results with @emph{more} bits of
11410 precision than specified by the IEEE standard. Disabling fused multiply
11411 add/subtract instructions also ensures that the program output is not
11412 sensitive to the compiler's ability to combine multiply and add/subtract
11415 @item -mtext-section-literals
11416 @itemx -mno-text-section-literals
11417 @opindex mtext-section-literals
11418 @opindex mno-text-section-literals
11419 Control the treatment of literal pools. The default is
11420 @option{-mno-text-section-literals}, which places literals in a separate
11421 section in the output file. This allows the literal pool to be placed
11422 in a data RAM/ROM, and it also allows the linker to combine literal
11423 pools from separate object files to remove redundant literals and
11424 improve code size. With @option{-mtext-section-literals}, the literals
11425 are interspersed in the text section in order to keep them as close as
11426 possible to their references. This may be necessary for large assembly
11429 @item -mtarget-align
11430 @itemx -mno-target-align
11431 @opindex mtarget-align
11432 @opindex mno-target-align
11433 When this option is enabled, GCC instructs the assembler to
11434 automatically align instructions to reduce branch penalties at the
11435 expense of some code density. The assembler attempts to widen density
11436 instructions to align branch targets and the instructions following call
11437 instructions. If there are not enough preceding safe density
11438 instructions to align a target, no widening will be performed. The
11439 default is @option{-mtarget-align}. These options do not affect the
11440 treatment of auto-aligned instructions like @code{LOOP}, which the
11441 assembler will always align, either by widening density instructions or
11442 by inserting no-op instructions.
11445 @itemx -mno-longcalls
11446 @opindex mlongcalls
11447 @opindex mno-longcalls
11448 When this option is enabled, GCC instructs the assembler to translate
11449 direct calls to indirect calls unless it can determine that the target
11450 of a direct call is in the range allowed by the call instruction. This
11451 translation typically occurs for calls to functions in other source
11452 files. Specifically, the assembler translates a direct @code{CALL}
11453 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11454 The default is @option{-mno-longcalls}. This option should be used in
11455 programs where the call target can potentially be out of range. This
11456 option is implemented in the assembler, not the compiler, so the
11457 assembly code generated by GCC will still show direct call
11458 instructions---look at the disassembled object code to see the actual
11459 instructions. Note that the assembler will use an indirect call for
11460 every cross-file call, not just those that really will be out of range.
11463 @node zSeries Options
11464 @subsection zSeries Options
11465 @cindex zSeries options
11467 These are listed under @xref{S/390 and zSeries Options}.
11469 @node Code Gen Options
11470 @section Options for Code Generation Conventions
11471 @cindex code generation conventions
11472 @cindex options, code generation
11473 @cindex run-time options
11475 These machine-independent options control the interface conventions
11476 used in code generation.
11478 Most of them have both positive and negative forms; the negative form
11479 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11480 one of the forms is listed---the one which is not the default. You
11481 can figure out the other form by either removing @samp{no-} or adding
11485 @item -fbounds-check
11486 @opindex fbounds-check
11487 For front-ends that support it, generate additional code to check that
11488 indices used to access arrays are within the declared range. This is
11489 currently only supported by the Java and Fortran 77 front-ends, where
11490 this option defaults to true and false respectively.
11494 This option generates traps for signed overflow on addition, subtraction,
11495 multiplication operations.
11499 This option instructs the compiler to assume that signed arithmetic
11500 overflow of addition, subtraction and multiplication wraps around
11501 using twos-complement representation. This flag enables some optimizations
11502 and disables other. This option is enabled by default for the Java
11503 front-end, as required by the Java language specification.
11506 @opindex fexceptions
11507 Enable exception handling. Generates extra code needed to propagate
11508 exceptions. For some targets, this implies GCC will generate frame
11509 unwind information for all functions, which can produce significant data
11510 size overhead, although it does not affect execution. If you do not
11511 specify this option, GCC will enable it by default for languages like
11512 C++ which normally require exception handling, and disable it for
11513 languages like C that do not normally require it. However, you may need
11514 to enable this option when compiling C code that needs to interoperate
11515 properly with exception handlers written in C++. You may also wish to
11516 disable this option if you are compiling older C++ programs that don't
11517 use exception handling.
11519 @item -fnon-call-exceptions
11520 @opindex fnon-call-exceptions
11521 Generate code that allows trapping instructions to throw exceptions.
11522 Note that this requires platform-specific runtime support that does
11523 not exist everywhere. Moreover, it only allows @emph{trapping}
11524 instructions to throw exceptions, i.e.@: memory references or floating
11525 point instructions. It does not allow exceptions to be thrown from
11526 arbitrary signal handlers such as @code{SIGALRM}.
11528 @item -funwind-tables
11529 @opindex funwind-tables
11530 Similar to @option{-fexceptions}, except that it will just generate any needed
11531 static data, but will not affect the generated code in any other way.
11532 You will normally not enable this option; instead, a language processor
11533 that needs this handling would enable it on your behalf.
11535 @item -fasynchronous-unwind-tables
11536 @opindex fasynchronous-unwind-tables
11537 Generate unwind table in dwarf2 format, if supported by target machine. The
11538 table is exact at each instruction boundary, so it can be used for stack
11539 unwinding from asynchronous events (such as debugger or garbage collector).
11541 @item -fpcc-struct-return
11542 @opindex fpcc-struct-return
11543 Return ``short'' @code{struct} and @code{union} values in memory like
11544 longer ones, rather than in registers. This convention is less
11545 efficient, but it has the advantage of allowing intercallability between
11546 GCC-compiled files and files compiled with other compilers, particularly
11547 the Portable C Compiler (pcc).
11549 The precise convention for returning structures in memory depends
11550 on the target configuration macros.
11552 Short structures and unions are those whose size and alignment match
11553 that of some integer type.
11555 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11556 switch is not binary compatible with code compiled with the
11557 @option{-freg-struct-return} switch.
11558 Use it to conform to a non-default application binary interface.
11560 @item -freg-struct-return
11561 @opindex freg-struct-return
11562 Return @code{struct} and @code{union} values in registers when possible.
11563 This is more efficient for small structures than
11564 @option{-fpcc-struct-return}.
11566 If you specify neither @option{-fpcc-struct-return} nor
11567 @option{-freg-struct-return}, GCC defaults to whichever convention is
11568 standard for the target. If there is no standard convention, GCC
11569 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11570 the principal compiler. In those cases, we can choose the standard, and
11571 we chose the more efficient register return alternative.
11573 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11574 switch is not binary compatible with code compiled with the
11575 @option{-fpcc-struct-return} switch.
11576 Use it to conform to a non-default application binary interface.
11578 @item -fshort-enums
11579 @opindex fshort-enums
11580 Allocate to an @code{enum} type only as many bytes as it needs for the
11581 declared range of possible values. Specifically, the @code{enum} type
11582 will be equivalent to the smallest integer type which has enough room.
11584 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11585 code that is not binary compatible with code generated without that switch.
11586 Use it to conform to a non-default application binary interface.
11588 @item -fshort-double
11589 @opindex fshort-double
11590 Use the same size for @code{double} as for @code{float}.
11592 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11593 code that is not binary compatible with code generated without that switch.
11594 Use it to conform to a non-default application binary interface.
11596 @item -fshort-wchar
11597 @opindex fshort-wchar
11598 Override the underlying type for @samp{wchar_t} to be @samp{short
11599 unsigned int} instead of the default for the target. This option is
11600 useful for building programs to run under WINE@.
11602 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11603 code that is not binary compatible with code generated without that switch.
11604 Use it to conform to a non-default application binary interface.
11606 @item -fshared-data
11607 @opindex fshared-data
11608 Requests that the data and non-@code{const} variables of this
11609 compilation be shared data rather than private data. The distinction
11610 makes sense only on certain operating systems, where shared data is
11611 shared between processes running the same program, while private data
11612 exists in one copy per process.
11615 @opindex fno-common
11616 In C, allocate even uninitialized global variables in the data section of the
11617 object file, rather than generating them as common blocks. This has the
11618 effect that if the same variable is declared (without @code{extern}) in
11619 two different compilations, you will get an error when you link them.
11620 The only reason this might be useful is if you wish to verify that the
11621 program will work on other systems which always work this way.
11625 Ignore the @samp{#ident} directive.
11627 @item -finhibit-size-directive
11628 @opindex finhibit-size-directive
11629 Don't output a @code{.size} assembler directive, or anything else that
11630 would cause trouble if the function is split in the middle, and the
11631 two halves are placed at locations far apart in memory. This option is
11632 used when compiling @file{crtstuff.c}; you should not need to use it
11635 @item -fverbose-asm
11636 @opindex fverbose-asm
11637 Put extra commentary information in the generated assembly code to
11638 make it more readable. This option is generally only of use to those
11639 who actually need to read the generated assembly code (perhaps while
11640 debugging the compiler itself).
11642 @option{-fno-verbose-asm}, the default, causes the
11643 extra information to be omitted and is useful when comparing two assembler
11648 @cindex global offset table
11650 Generate position-independent code (PIC) suitable for use in a shared
11651 library, if supported for the target machine. Such code accesses all
11652 constant addresses through a global offset table (GOT)@. The dynamic
11653 loader resolves the GOT entries when the program starts (the dynamic
11654 loader is not part of GCC; it is part of the operating system). If
11655 the GOT size for the linked executable exceeds a machine-specific
11656 maximum size, you get an error message from the linker indicating that
11657 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11658 instead. (These maximums are 8k on the SPARC and 32k
11659 on the m68k and RS/6000. The 386 has no such limit.)
11661 Position-independent code requires special support, and therefore works
11662 only on certain machines. For the 386, GCC supports PIC for System V
11663 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11664 position-independent.
11668 If supported for the target machine, emit position-independent code,
11669 suitable for dynamic linking and avoiding any limit on the size of the
11670 global offset table. This option makes a difference on the m68k
11673 Position-independent code requires special support, and therefore works
11674 only on certain machines.
11680 These options are similar to @option{-fpic} and @option{-fPIC}, but
11681 generated position independent code can be only linked into executables.
11682 Usually these options are used when @option{-pie} GCC option will be
11683 used during linking.
11685 @item -ffixed-@var{reg}
11687 Treat the register named @var{reg} as a fixed register; generated code
11688 should never refer to it (except perhaps as a stack pointer, frame
11689 pointer or in some other fixed role).
11691 @var{reg} must be the name of a register. The register names accepted
11692 are machine-specific and are defined in the @code{REGISTER_NAMES}
11693 macro in the machine description macro file.
11695 This flag does not have a negative form, because it specifies a
11698 @item -fcall-used-@var{reg}
11699 @opindex fcall-used
11700 Treat the register named @var{reg} as an allocable register that is
11701 clobbered by function calls. It may be allocated for temporaries or
11702 variables that do not live across a call. Functions compiled this way
11703 will not save and restore the register @var{reg}.
11705 It is an error to used this flag with the frame pointer or stack pointer.
11706 Use of this flag for other registers that have fixed pervasive roles in
11707 the machine's execution model will produce disastrous results.
11709 This flag does not have a negative form, because it specifies a
11712 @item -fcall-saved-@var{reg}
11713 @opindex fcall-saved
11714 Treat the register named @var{reg} as an allocable register saved by
11715 functions. It may be allocated even for temporaries or variables that
11716 live across a call. Functions compiled this way will save and restore
11717 the register @var{reg} if they use it.
11719 It is an error to used this flag with the frame pointer or stack pointer.
11720 Use of this flag for other registers that have fixed pervasive roles in
11721 the machine's execution model will produce disastrous results.
11723 A different sort of disaster will result from the use of this flag for
11724 a register in which function values may be returned.
11726 This flag does not have a negative form, because it specifies a
11729 @item -fpack-struct
11730 @opindex fpack-struct
11731 Pack all structure members together without holes.
11733 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11734 code that is not binary compatible with code generated without that switch.
11735 Additionally, it makes the code suboptimal.
11736 Use it to conform to a non-default application binary interface.
11738 @item -finstrument-functions
11739 @opindex finstrument-functions
11740 Generate instrumentation calls for entry and exit to functions. Just
11741 after function entry and just before function exit, the following
11742 profiling functions will be called with the address of the current
11743 function and its call site. (On some platforms,
11744 @code{__builtin_return_address} does not work beyond the current
11745 function, so the call site information may not be available to the
11746 profiling functions otherwise.)
11749 void __cyg_profile_func_enter (void *this_fn,
11751 void __cyg_profile_func_exit (void *this_fn,
11755 The first argument is the address of the start of the current function,
11756 which may be looked up exactly in the symbol table.
11758 This instrumentation is also done for functions expanded inline in other
11759 functions. The profiling calls will indicate where, conceptually, the
11760 inline function is entered and exited. This means that addressable
11761 versions of such functions must be available. If all your uses of a
11762 function are expanded inline, this may mean an additional expansion of
11763 code size. If you use @samp{extern inline} in your C code, an
11764 addressable version of such functions must be provided. (This is
11765 normally the case anyways, but if you get lucky and the optimizer always
11766 expands the functions inline, you might have gotten away without
11767 providing static copies.)
11769 A function may be given the attribute @code{no_instrument_function}, in
11770 which case this instrumentation will not be done. This can be used, for
11771 example, for the profiling functions listed above, high-priority
11772 interrupt routines, and any functions from which the profiling functions
11773 cannot safely be called (perhaps signal handlers, if the profiling
11774 routines generate output or allocate memory).
11776 @item -fstack-check
11777 @opindex fstack-check
11778 Generate code to verify that you do not go beyond the boundary of the
11779 stack. You should specify this flag if you are running in an
11780 environment with multiple threads, but only rarely need to specify it in
11781 a single-threaded environment since stack overflow is automatically
11782 detected on nearly all systems if there is only one stack.
11784 Note that this switch does not actually cause checking to be done; the
11785 operating system must do that. The switch causes generation of code
11786 to ensure that the operating system sees the stack being extended.
11788 @item -fstack-limit-register=@var{reg}
11789 @itemx -fstack-limit-symbol=@var{sym}
11790 @itemx -fno-stack-limit
11791 @opindex fstack-limit-register
11792 @opindex fstack-limit-symbol
11793 @opindex fno-stack-limit
11794 Generate code to ensure that the stack does not grow beyond a certain value,
11795 either the value of a register or the address of a symbol. If the stack
11796 would grow beyond the value, a signal is raised. For most targets,
11797 the signal is raised before the stack overruns the boundary, so
11798 it is possible to catch the signal without taking special precautions.
11800 For instance, if the stack starts at absolute address @samp{0x80000000}
11801 and grows downwards, you can use the flags
11802 @option{-fstack-limit-symbol=__stack_limit} and
11803 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11804 of 128KB@. Note that this may only work with the GNU linker.
11806 @cindex aliasing of parameters
11807 @cindex parameters, aliased
11808 @item -fargument-alias
11809 @itemx -fargument-noalias
11810 @itemx -fargument-noalias-global
11811 @opindex fargument-alias
11812 @opindex fargument-noalias
11813 @opindex fargument-noalias-global
11814 Specify the possible relationships among parameters and between
11815 parameters and global data.
11817 @option{-fargument-alias} specifies that arguments (parameters) may
11818 alias each other and may alias global storage.@*
11819 @option{-fargument-noalias} specifies that arguments do not alias
11820 each other, but may alias global storage.@*
11821 @option{-fargument-noalias-global} specifies that arguments do not
11822 alias each other and do not alias global storage.
11824 Each language will automatically use whatever option is required by
11825 the language standard. You should not need to use these options yourself.
11827 @item -fleading-underscore
11828 @opindex fleading-underscore
11829 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11830 change the way C symbols are represented in the object file. One use
11831 is to help link with legacy assembly code.
11833 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11834 generate code that is not binary compatible with code generated without that
11835 switch. Use it to conform to a non-default application binary interface.
11836 Not all targets provide complete support for this switch.
11838 @item -ftls-model=@var{model}
11839 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11840 The @var{model} argument should be one of @code{global-dynamic},
11841 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11843 The default without @option{-fpic} is @code{initial-exec}; with
11844 @option{-fpic} the default is @code{global-dynamic}.
11846 @item -fvisibility=@var{default|internal|hidden|protected}
11847 @opindex fvisibility
11848 Set the default ELF image symbol visibility to the specified option - all
11849 symbols will be marked with this unless overrided within the code.
11850 Using this feature can very substantially improve linking and
11851 load times of shared object libraries, produce more optimised
11852 code, provide near-perfect API export and prevent symbol clashes.
11853 It is @strong{strongly} recommended that you use this in any shared objects
11856 Despite the nomenclature, @code{default} always means public ie;
11857 available to be linked against from outside the shared object.
11858 @code{protected} and @code{internal} are pretty useless in real-world
11859 usage so the only other commonly used option will be @code{hidden}.
11860 The default if -fvisibility isn't specified is @code{default} ie; make every
11861 symbol public - this causes the same behaviour as previous versions of
11864 A good explanation of the benefits offered by ensuring ELF
11865 symbols have the correct visibility is given by ``How To Write
11866 Shared Libraries'' by Ulrich Drepper (which can be found at
11867 @w{@uref{http://people.redhat.com/~drepper/}}) - however a superior
11868 solution made possible by this option to marking things hidden when
11869 the default is public is to make the default hidden and mark things
11870 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
11871 and @code{__attribute__ ((visibility("default")))} instead of
11872 @code{__declspec(dllexport)} you get almost identical semantics with
11873 identical syntax. This is a great boon to those working with
11874 cross-platform projects.
11876 For those adding visibility support to existing code, you may find
11877 @samp{#pragma GCC visibility} of use. This works by you enclosing
11878 the declarations you wish to set visibility for with (for example)
11879 @samp{#pragma GCC visibility push(hidden)} and
11880 @samp{#pragma GCC visibility pop}. These can be nested up to sixteen
11881 times. Bear in mind that symbol visibility should be viewed @strong{as
11882 part of the API interface contract} and thus all new code should
11883 always specify visibility when it is not the default ie; declarations
11884 only for use within the local DSO should @strong{always} be marked explicitly
11885 as hidden as so to avoid PLT indirection overheads - making this
11886 abundantly clear also aids readability and self-documentation of the code.
11887 Note that due to ISO C++ specification requirements, operator new and
11888 operator delete must always be of default visibility.
11890 An overview of these techniques, their benefits and how to use them
11891 is at @w{@uref{http://www.nedprod.com/programs/gccvisibility.html}}.
11897 @node Environment Variables
11898 @section Environment Variables Affecting GCC
11899 @cindex environment variables
11901 @c man begin ENVIRONMENT
11902 This section describes several environment variables that affect how GCC
11903 operates. Some of them work by specifying directories or prefixes to use
11904 when searching for various kinds of files. Some are used to specify other
11905 aspects of the compilation environment.
11907 Note that you can also specify places to search using options such as
11908 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11909 take precedence over places specified using environment variables, which
11910 in turn take precedence over those specified by the configuration of GCC@.
11911 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11912 GNU Compiler Collection (GCC) Internals}.
11917 @c @itemx LC_COLLATE
11919 @c @itemx LC_MONETARY
11920 @c @itemx LC_NUMERIC
11925 @c @findex LC_COLLATE
11926 @findex LC_MESSAGES
11927 @c @findex LC_MONETARY
11928 @c @findex LC_NUMERIC
11932 These environment variables control the way that GCC uses
11933 localization information that allow GCC to work with different
11934 national conventions. GCC inspects the locale categories
11935 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11936 so. These locale categories can be set to any value supported by your
11937 installation. A typical value is @samp{en_UK} for English in the United
11940 The @env{LC_CTYPE} environment variable specifies character
11941 classification. GCC uses it to determine the character boundaries in
11942 a string; this is needed for some multibyte encodings that contain quote
11943 and escape characters that would otherwise be interpreted as a string
11946 The @env{LC_MESSAGES} environment variable specifies the language to
11947 use in diagnostic messages.
11949 If the @env{LC_ALL} environment variable is set, it overrides the value
11950 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11951 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11952 environment variable. If none of these variables are set, GCC
11953 defaults to traditional C English behavior.
11957 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11958 files. GCC uses temporary files to hold the output of one stage of
11959 compilation which is to be used as input to the next stage: for example,
11960 the output of the preprocessor, which is the input to the compiler
11963 @item GCC_EXEC_PREFIX
11964 @findex GCC_EXEC_PREFIX
11965 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11966 names of the subprograms executed by the compiler. No slash is added
11967 when this prefix is combined with the name of a subprogram, but you can
11968 specify a prefix that ends with a slash if you wish.
11970 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11971 an appropriate prefix to use based on the pathname it was invoked with.
11973 If GCC cannot find the subprogram using the specified prefix, it
11974 tries looking in the usual places for the subprogram.
11976 The default value of @env{GCC_EXEC_PREFIX} is
11977 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11978 of @code{prefix} when you ran the @file{configure} script.
11980 Other prefixes specified with @option{-B} take precedence over this prefix.
11982 This prefix is also used for finding files such as @file{crt0.o} that are
11985 In addition, the prefix is used in an unusual way in finding the
11986 directories to search for header files. For each of the standard
11987 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11988 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11989 replacing that beginning with the specified prefix to produce an
11990 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11991 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11992 These alternate directories are searched first; the standard directories
11995 @item COMPILER_PATH
11996 @findex COMPILER_PATH
11997 The value of @env{COMPILER_PATH} is a colon-separated list of
11998 directories, much like @env{PATH}. GCC tries the directories thus
11999 specified when searching for subprograms, if it can't find the
12000 subprograms using @env{GCC_EXEC_PREFIX}.
12003 @findex LIBRARY_PATH
12004 The value of @env{LIBRARY_PATH} is a colon-separated list of
12005 directories, much like @env{PATH}. When configured as a native compiler,
12006 GCC tries the directories thus specified when searching for special
12007 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
12008 using GCC also uses these directories when searching for ordinary
12009 libraries for the @option{-l} option (but directories specified with
12010 @option{-L} come first).
12014 @cindex locale definition
12015 This variable is used to pass locale information to the compiler. One way in
12016 which this information is used is to determine the character set to be used
12017 when character literals, string literals and comments are parsed in C and C++.
12018 When the compiler is configured to allow multibyte characters,
12019 the following values for @env{LANG} are recognized:
12023 Recognize JIS characters.
12025 Recognize SJIS characters.
12027 Recognize EUCJP characters.
12030 If @env{LANG} is not defined, or if it has some other value, then the
12031 compiler will use mblen and mbtowc as defined by the default locale to
12032 recognize and translate multibyte characters.
12036 Some additional environments variables affect the behavior of the
12039 @include cppenv.texi
12043 @node Precompiled Headers
12044 @section Using Precompiled Headers
12045 @cindex precompiled headers
12046 @cindex speed of compilation
12048 Often large projects have many header files that are included in every
12049 source file. The time the compiler takes to process these header files
12050 over and over again can account for nearly all of the time required to
12051 build the project. To make builds faster, GCC allows users to
12052 `precompile' a header file; then, if builds can use the precompiled
12053 header file they will be much faster.
12055 @strong{Caution:} There are a few known situations where GCC will
12056 crash when trying to use a precompiled header. If you have trouble
12057 with a precompiled header, you should remove the precompiled header
12058 and compile without it. In addition, please use GCC's on-line
12059 defect-tracking system to report any problems you encounter with
12060 precompiled headers. @xref{Bugs}.
12062 To create a precompiled header file, simply compile it as you would any
12063 other file, if necessary using the @option{-x} option to make the driver
12064 treat it as a C or C++ header file. You will probably want to use a
12065 tool like @command{make} to keep the precompiled header up-to-date when
12066 the headers it contains change.
12068 A precompiled header file will be searched for when @code{#include} is
12069 seen in the compilation. As it searches for the included file
12070 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
12071 compiler looks for a precompiled header in each directory just before it
12072 looks for the include file in that directory. The name searched for is
12073 the name specified in the @code{#include} with @samp{.gch} appended. If
12074 the precompiled header file can't be used, it is ignored.
12076 For instance, if you have @code{#include "all.h"}, and you have
12077 @file{all.h.gch} in the same directory as @file{all.h}, then the
12078 precompiled header file will be used if possible, and the original
12079 header will be used otherwise.
12081 Alternatively, you might decide to put the precompiled header file in a
12082 directory and use @option{-I} to ensure that directory is searched
12083 before (or instead of) the directory containing the original header.
12084 Then, if you want to check that the precompiled header file is always
12085 used, you can put a file of the same name as the original header in this
12086 directory containing an @code{#error} command.
12088 This also works with @option{-include}. So yet another way to use
12089 precompiled headers, good for projects not designed with precompiled
12090 header files in mind, is to simply take most of the header files used by
12091 a project, include them from another header file, precompile that header
12092 file, and @option{-include} the precompiled header. If the header files
12093 have guards against multiple inclusion, they will be skipped because
12094 they've already been included (in the precompiled header).
12096 If you need to precompile the same header file for different
12097 languages, targets, or compiler options, you can instead make a
12098 @emph{directory} named like @file{all.h.gch}, and put each precompiled
12099 header in the directory, perhaps using @option{-o}. It doesn't matter
12100 what you call the files in the directory, every precompiled header in
12101 the directory will be considered. The first precompiled header
12102 encountered in the directory that is valid for this compilation will
12103 be used; they're searched in no particular order.
12105 There are many other possibilities, limited only by your imagination,
12106 good sense, and the constraints of your build system.
12108 A precompiled header file can be used only when these conditions apply:
12112 Only one precompiled header can be used in a particular compilation.
12115 A precompiled header can't be used once the first C token is seen. You
12116 can have preprocessor directives before a precompiled header; you can
12117 even include a precompiled header from inside another header, so long as
12118 there are no C tokens before the @code{#include}.
12121 The precompiled header file must be produced for the same language as
12122 the current compilation. You can't use a C precompiled header for a C++
12126 The precompiled header file must be produced by the same compiler
12127 version and configuration as the current compilation is using.
12128 The easiest way to guarantee this is to use the same compiler binary
12129 for creating and using precompiled headers.
12132 Any macros defined before the precompiled header is included must
12133 either be defined in the same way as when the precompiled header was
12134 generated, or must not affect the precompiled header, which usually
12135 means that the they don't appear in the precompiled header at all.
12137 The @option{-D} option is one way to define a macro before a
12138 precompiled header is included; using a @code{#define} can also do it.
12139 There are also some options that define macros implicitly, like
12140 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
12143 @item If debugging information is output when using the precompiled
12144 header, using @option{-g} or similar, the same kind of debugging information
12145 must have been output when building the precompiled header. However,
12146 a precompiled header built using @option{-g} can be used in a compilation
12147 when no debugging information is being output.
12149 @item The same @option{-m} options must generally be used when building
12150 and using the precompiled header. @xref{Submodel Options},
12151 for any cases where this rule is relaxed.
12153 @item Each of the following options must be the same when building and using
12154 the precompiled header:
12156 @gccoptlist{-fexceptions -funit-at-a-time}
12159 Some other command-line options starting with @option{-f},
12160 @option{-p}, or @option{-O} must be defined in the same way as when
12161 the precompiled header was generated. At present, it's not clear
12162 which options are safe to change and which are not; the safest choice
12163 is to use exactly the same options when generating and using the
12164 precompiled header. The following are known to be safe:
12166 @gccoptlist{-fpreprocessed -pedantic-errors}
12170 For all of these except the last, the compiler will automatically
12171 ignore the precompiled header if the conditions aren't met. If you
12172 find an option combination that doesn't work and doesn't cause the
12173 precompiled header to be ignored, please consider filing a bug report,
12176 If you do use differing options when generating and using the
12177 precompiled header, the actual behaviour will be a mixture of the
12178 behaviour for the options. For instance, if you use @option{-g} to
12179 generate the precompiled header but not when using it, you may or may
12180 not get debugging information for routines in the precompiled header.
12182 @node Running Protoize
12183 @section Running Protoize
12185 The program @code{protoize} is an optional part of GCC@. You can use
12186 it to add prototypes to a program, thus converting the program to ISO
12187 C in one respect. The companion program @code{unprotoize} does the
12188 reverse: it removes argument types from any prototypes that are found.
12190 When you run these programs, you must specify a set of source files as
12191 command line arguments. The conversion programs start out by compiling
12192 these files to see what functions they define. The information gathered
12193 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
12195 After scanning comes actual conversion. The specified files are all
12196 eligible to be converted; any files they include (whether sources or
12197 just headers) are eligible as well.
12199 But not all the eligible files are converted. By default,
12200 @code{protoize} and @code{unprotoize} convert only source and header
12201 files in the current directory. You can specify additional directories
12202 whose files should be converted with the @option{-d @var{directory}}
12203 option. You can also specify particular files to exclude with the
12204 @option{-x @var{file}} option. A file is converted if it is eligible, its
12205 directory name matches one of the specified directory names, and its
12206 name within the directory has not been excluded.
12208 Basic conversion with @code{protoize} consists of rewriting most
12209 function definitions and function declarations to specify the types of
12210 the arguments. The only ones not rewritten are those for varargs
12213 @code{protoize} optionally inserts prototype declarations at the
12214 beginning of the source file, to make them available for any calls that
12215 precede the function's definition. Or it can insert prototype
12216 declarations with block scope in the blocks where undeclared functions
12219 Basic conversion with @code{unprotoize} consists of rewriting most
12220 function declarations to remove any argument types, and rewriting
12221 function definitions to the old-style pre-ISO form.
12223 Both conversion programs print a warning for any function declaration or
12224 definition that they can't convert. You can suppress these warnings
12227 The output from @code{protoize} or @code{unprotoize} replaces the
12228 original source file. The original file is renamed to a name ending
12229 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
12230 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
12231 for DOS) file already exists, then the source file is simply discarded.
12233 @code{protoize} and @code{unprotoize} both depend on GCC itself to
12234 scan the program and collect information about the functions it uses.
12235 So neither of these programs will work until GCC is installed.
12237 Here is a table of the options you can use with @code{protoize} and
12238 @code{unprotoize}. Each option works with both programs unless
12242 @item -B @var{directory}
12243 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
12244 usual directory (normally @file{/usr/local/lib}). This file contains
12245 prototype information about standard system functions. This option
12246 applies only to @code{protoize}.
12248 @item -c @var{compilation-options}
12249 Use @var{compilation-options} as the options when running @command{gcc} to
12250 produce the @samp{.X} files. The special option @option{-aux-info} is
12251 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
12253 Note that the compilation options must be given as a single argument to
12254 @code{protoize} or @code{unprotoize}. If you want to specify several
12255 @command{gcc} options, you must quote the entire set of compilation options
12256 to make them a single word in the shell.
12258 There are certain @command{gcc} arguments that you cannot use, because they
12259 would produce the wrong kind of output. These include @option{-g},
12260 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
12261 the @var{compilation-options}, they are ignored.
12264 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
12265 systems) instead of @samp{.c}. This is convenient if you are converting
12266 a C program to C++. This option applies only to @code{protoize}.
12269 Add explicit global declarations. This means inserting explicit
12270 declarations at the beginning of each source file for each function
12271 that is called in the file and was not declared. These declarations
12272 precede the first function definition that contains a call to an
12273 undeclared function. This option applies only to @code{protoize}.
12275 @item -i @var{string}
12276 Indent old-style parameter declarations with the string @var{string}.
12277 This option applies only to @code{protoize}.
12279 @code{unprotoize} converts prototyped function definitions to old-style
12280 function definitions, where the arguments are declared between the
12281 argument list and the initial @samp{@{}. By default, @code{unprotoize}
12282 uses five spaces as the indentation. If you want to indent with just
12283 one space instead, use @option{-i " "}.
12286 Keep the @samp{.X} files. Normally, they are deleted after conversion
12290 Add explicit local declarations. @code{protoize} with @option{-l} inserts
12291 a prototype declaration for each function in each block which calls the
12292 function without any declaration. This option applies only to
12296 Make no real changes. This mode just prints information about the conversions
12297 that would have been done without @option{-n}.
12300 Make no @samp{.save} files. The original files are simply deleted.
12301 Use this option with caution.
12303 @item -p @var{program}
12304 Use the program @var{program} as the compiler. Normally, the name
12305 @file{gcc} is used.
12308 Work quietly. Most warnings are suppressed.
12311 Print the version number, just like @option{-v} for @command{gcc}.
12314 If you need special compiler options to compile one of your program's
12315 source files, then you should generate that file's @samp{.X} file
12316 specially, by running @command{gcc} on that source file with the
12317 appropriate options and the option @option{-aux-info}. Then run
12318 @code{protoize} on the entire set of files. @code{protoize} will use
12319 the existing @samp{.X} file because it is newer than the source file.
12323 gcc -Dfoo=bar file1.c -aux-info file1.X
12328 You need to include the special files along with the rest in the
12329 @code{protoize} command, even though their @samp{.X} files already
12330 exist, because otherwise they won't get converted.
12332 @xref{Protoize Caveats}, for more information on how to use
12333 @code{protoize} successfully.