1 @c Copyright (C) 1988-2014 Free Software Foundation, Inc.
2 @c This is part of the GCC manual.
3 @c For copying conditions, see the file gcc.texi.
10 @c man begin COPYRIGHT
11 Copyright @copyright{} 1988-2014 Free Software Foundation, Inc.
13 Permission is granted to copy, distribute and/or modify this document
14 under the terms of the GNU Free Documentation License, Version 1.3 or
15 any later version published by the Free Software Foundation; with the
16 Invariant Sections being ``GNU General Public License'' and ``Funding
17 Free Software'', the Front-Cover texts being (a) (see below), and with
18 the Back-Cover Texts being (b) (see below). A copy of the license is
19 included in the gfdl(7) man page.
21 (a) The FSF's Front-Cover Text is:
25 (b) The FSF's Back-Cover Text is:
27 You have freedom to copy and modify this GNU Manual, like GNU
28 software. Copies published by the Free Software Foundation raise
29 funds for GNU development.
31 @c Set file name and title for the man page.
33 @settitle GNU project C and C++ compiler
35 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
36 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
37 [@option{-W}@var{warn}@dots{}] [@option{-Wpedantic}]
38 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
39 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
40 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
41 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
43 Only the most useful options are listed here; see below for the
44 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
47 gpl(7), gfdl(7), fsf-funding(7),
48 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
49 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
50 @file{ld}, @file{binutils} and @file{gdb}.
53 For instructions on reporting bugs, see
57 See the Info entry for @command{gcc}, or
58 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
59 for contributors to GCC@.
64 @chapter GCC Command Options
65 @cindex GCC command options
66 @cindex command options
67 @cindex options, GCC command
69 @c man begin DESCRIPTION
70 When you invoke GCC, it normally does preprocessing, compilation,
71 assembly and linking. The ``overall options'' allow you to stop this
72 process at an intermediate stage. For example, the @option{-c} option
73 says not to run the linker. Then the output consists of object files
74 output by the assembler.
76 Other options are passed on to one stage of processing. Some options
77 control the preprocessor and others the compiler itself. Yet other
78 options control the assembler and linker; most of these are not
79 documented here, since you rarely need to use any of them.
81 @cindex C compilation options
82 Most of the command-line options that you can use with GCC are useful
83 for C programs; when an option is only useful with another language
84 (usually C++), the explanation says so explicitly. If the description
85 for a particular option does not mention a source language, you can use
86 that option with all supported languages.
88 @cindex C++ compilation options
89 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
90 options for compiling C++ programs.
92 @cindex grouping options
93 @cindex options, grouping
94 The @command{gcc} program accepts options and file names as operands. Many
95 options have multi-letter names; therefore multiple single-letter options
96 may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
99 @cindex order of options
100 @cindex options, order
101 You can mix options and other arguments. For the most part, the order
102 you use doesn't matter. Order does matter when you use several
103 options of the same kind; for example, if you specify @option{-L} more
104 than once, the directories are searched in the order specified. Also,
105 the placement of the @option{-l} option is significant.
107 Many options have long names starting with @samp{-f} or with
108 @samp{-W}---for example,
109 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
110 these have both positive and negative forms; the negative form of
111 @option{-ffoo} is @option{-fno-foo}. This manual documents
112 only one of these two forms, whichever one is not the default.
116 @xref{Option Index}, for an index to GCC's options.
119 * Option Summary:: Brief list of all options, without explanations.
120 * Overall Options:: Controlling the kind of output:
121 an executable, object files, assembler files,
122 or preprocessed source.
123 * Invoking G++:: Compiling C++ programs.
124 * C Dialect Options:: Controlling the variant of C language compiled.
125 * C++ Dialect Options:: Variations on C++.
126 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
128 * Language Independent Options:: Controlling how diagnostics should be
130 * Warning Options:: How picky should the compiler be?
131 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
132 * Optimize Options:: How much optimization?
133 * Preprocessor Options:: Controlling header files and macro definitions.
134 Also, getting dependency information for Make.
135 * Assembler Options:: Passing options to the assembler.
136 * Link Options:: Specifying libraries and so on.
137 * Directory Options:: Where to find header files and libraries.
138 Where to find the compiler executable files.
139 * Spec Files:: How to pass switches to sub-processes.
140 * Target Options:: Running a cross-compiler, or an old version of GCC.
141 * Submodel Options:: Specifying minor hardware or convention variations,
142 such as 68010 vs 68020.
143 * Code Gen Options:: Specifying conventions for function calls, data layout
145 * Environment Variables:: Env vars that affect GCC.
146 * Precompiled Headers:: Compiling a header once, and using it many times.
152 @section Option Summary
154 Here is a summary of all the options, grouped by type. Explanations are
155 in the following sections.
158 @item Overall Options
159 @xref{Overall Options,,Options Controlling the Kind of Output}.
160 @gccoptlist{-c -S -E -o @var{file} -no-canonical-prefixes @gol
161 -pipe -pass-exit-codes @gol
162 -x @var{language} -v -### --help@r{[}=@var{class}@r{[},@dots{}@r{]]} --target-help @gol
163 --version -wrapper @@@var{file} -fplugin=@var{file} -fplugin-arg-@var{name}=@var{arg} @gol
164 -fdump-ada-spec@r{[}-slim@r{]} -fada-spec-parent=@var{unit} -fdump-go-spec=@var{file}}
166 @item C Language Options
167 @xref{C Dialect Options,,Options Controlling C Dialect}.
168 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
169 -aux-info @var{filename} -fallow-parameterless-variadic-functions @gol
170 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
171 -fhosted -ffreestanding -fopenmp -fopenmp-simd -fms-extensions @gol
172 -fplan9-extensions -trigraphs -traditional -traditional-cpp @gol
173 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
174 -fsigned-bitfields -fsigned-char @gol
175 -funsigned-bitfields -funsigned-char}
177 @item C++ Language Options
178 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
179 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
180 -fconstexpr-depth=@var{n} -ffriend-injection @gol
181 -fno-elide-constructors @gol
182 -fno-enforce-eh-specs @gol
183 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
184 -fno-implicit-templates @gol
185 -fno-implicit-inline-templates @gol
186 -fno-implement-inlines -fms-extensions @gol
187 -fno-nonansi-builtins -fnothrow-opt -fno-operator-names @gol
188 -fno-optional-diags -fpermissive @gol
189 -fno-pretty-templates @gol
190 -frepo -fno-rtti -fstats -ftemplate-backtrace-limit=@var{n} @gol
191 -ftemplate-depth=@var{n} @gol
192 -fno-threadsafe-statics -fuse-cxa-atexit @gol
193 -fno-weak -nostdinc++ @gol
194 -fvisibility-inlines-hidden @gol
195 -fvtable-verify=@var{std|preinit|none} @gol
196 -fvtv-counts -fvtv-debug @gol
197 -fvisibility-ms-compat @gol
198 -fext-numeric-literals @gol
199 -Wabi=@var{n} -Wconversion-null -Wctor-dtor-privacy @gol
200 -Wdelete-non-virtual-dtor -Wliteral-suffix -Wnarrowing @gol
201 -Wnoexcept -Wnon-virtual-dtor -Wreorder @gol
202 -Weffc++ -Wstrict-null-sentinel @gol
203 -Wno-non-template-friend -Wold-style-cast @gol
204 -Woverloaded-virtual -Wno-pmf-conversions @gol
207 @item Objective-C and Objective-C++ Language Options
208 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
209 Objective-C and Objective-C++ Dialects}.
210 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
211 -fgnu-runtime -fnext-runtime @gol
212 -fno-nil-receivers @gol
213 -fobjc-abi-version=@var{n} @gol
214 -fobjc-call-cxx-cdtors @gol
215 -fobjc-direct-dispatch @gol
216 -fobjc-exceptions @gol
219 -fobjc-std=objc1 @gol
220 -fno-local-ivars @gol
221 -fivar-visibility=@var{public|protected|private|package} @gol
222 -freplace-objc-classes @gol
225 -Wassign-intercept @gol
226 -Wno-protocol -Wselector @gol
227 -Wstrict-selector-match @gol
228 -Wundeclared-selector}
230 @item Language Independent Options
231 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
232 @gccoptlist{-fmessage-length=@var{n} @gol
233 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
234 -fdiagnostics-color=@r{[}auto@r{|}never@r{|}always@r{]} @gol
235 -fno-diagnostics-show-option -fno-diagnostics-show-caret}
237 @item Warning Options
238 @xref{Warning Options,,Options to Request or Suppress Warnings}.
239 @gccoptlist{-fsyntax-only -fmax-errors=@var{n} -Wpedantic @gol
240 -pedantic-errors @gol
241 -w -Wextra -Wall -Waddress -Waggregate-return @gol
242 -Waggressive-loop-optimizations -Warray-bounds @gol
243 -Wno-attributes -Wno-builtin-macro-redefined @gol
244 -Wc++-compat -Wc++11-compat -Wcast-align -Wcast-qual @gol
245 -Wchar-subscripts -Wclobbered -Wcomment -Wconditionally-supported @gol
246 -Wconversion -Wcoverage-mismatch -Wdate-time -Wdelete-incomplete -Wno-cpp @gol
247 -Wno-deprecated -Wno-deprecated-declarations -Wno-designated-init @gol
248 -Wdisabled-optimization -Wno-discarded-qualifiers @gol
249 -Wno-div-by-zero -Wdouble-promotion -Wempty-body -Wenum-compare @gol
250 -Wno-endif-labels -Werror -Werror=* @gol
251 -Wfatal-errors -Wfloat-equal -Wforce-warnings -Wformat -Wformat=2 @gol
252 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
253 -Wformat-security -Wformat-signedness -Wformat-y2k @gol
254 -Wframe-larger-than=@var{len} -Wno-free-nonheap-object -Wjump-misses-init @gol
255 -Wignored-qualifiers -Wincompatible-pointer-types @gol
256 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
257 -Winit-self -Winline -Wno-int-conversion @gol
258 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
259 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
260 -Wlogical-op -Wlogical-not-parentheses -Wlong-long @gol
261 -Wmain -Wmaybe-uninitialized -Wmemset-transposed-args -Wmissing-braces @gol
262 -Wmissing-field-initializers -Wmissing-include-dirs @gol
263 -Wno-multichar -Wnonnull -Wodr -Wno-overflow -Wopenmp-simd @gol
264 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
265 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
266 -Wpointer-arith -Wno-pointer-to-int-cast @gol
267 -Wredundant-decls -Wno-return-local-addr -Wipa-opt-mismatch @gol
268 -Wreturn-type -Wself-assign -Wself-assign-non-pod -Wsequence-point -Wshadow @gol
269 -Wno-shadow-ivar -Wshadow-compatible-local -Wshadow-local @gol
270 -Wsign-compare -Wsign-conversion -Wfloat-conversion @gol
271 -Wsizeof-pointer-memaccess -Wsizeof-array-argument @gol
272 -Wstack-protector -Wstack-usage=@var{len} -Wstrict-aliasing @gol
273 -Wstrict-aliasing=n @gol -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
274 -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{]} @gol
275 -Wsuggest-final-types @gol -Wsuggest-final-methods @gol
276 -Wmissing-format-attribute @gol
277 -Wswitch -Wswitch-default -Wswitch-enum -Wswitch-bool -Wsync-nand @gol
278 -Wsystem-headers -Wtrampolines -Wtrigraphs -Wtype-limits -Wundef @gol
279 -Wuninitialized -Wunknown-pragmas -Wno-pragmas @gol
280 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
281 -Wunused-label -Wunused-local-typedefs -Wunused-parameter @gol
282 -Wno-unused-result -Wunused-value @gol -Wunused-variable @gol
283 -Wunused-but-set-parameter -Wunused-but-set-variable @gol
284 -Wuseless-cast -Wvariadic-macros -Wvector-operation-performance @gol
285 -Wvla -Wvolatile-register-var -Wwrite-strings -Wzero-as-null-pointer-constant}
287 @item C and Objective-C-only Warning Options
288 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
289 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
290 -Wold-style-declaration -Wold-style-definition @gol
291 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
292 -Wdeclaration-after-statement -Wpointer-sign}
294 @item Debugging Options
295 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
296 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
297 -fsanitize=@var{style} -fsanitize-recover @gol
298 -fsanitize-undefined-trap-on-error @gol
299 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
300 -fdisable-ipa-@var{pass_name} @gol
301 -fdisable-rtl-@var{pass_name} @gol
302 -fdisable-rtl-@var{pass-name}=@var{range-list} @gol
303 -fdisable-tree-@var{pass_name} @gol
304 -fdisable-tree-@var{pass-name}=@var{range-list} @gol
305 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
306 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
307 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
308 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
310 -fdump-statistics @gol
312 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
313 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
314 -fdump-tree-cfg -fdump-tree-alias @gol
316 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
317 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
318 -fdump-tree-gimple@r{[}-raw@r{]} @gol
319 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
320 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
321 -fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
322 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
323 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
324 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
325 -fdump-tree-nrv -fdump-tree-vect @gol
326 -fdump-tree-sink @gol
327 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
328 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
329 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
330 -fdump-tree-vtable-verify @gol
331 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
332 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
333 -fdump-final-insns=@var{file} @gol
334 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
335 -feliminate-dwarf2-dups -fno-eliminate-unused-debug-types @gol
336 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
337 -fenable-@var{kind}-@var{pass} @gol
338 -fenable-@var{kind}-@var{pass}=@var{range-list} @gol
339 -fdebug-types-section -fmem-report-wpa @gol
340 -fenable-@var{kind}-@var{pass} @gol
341 -fenable-@var{kind}-@var{pass}=@var{range-list} @gol
342 -fdebug-types-section @gol
343 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
344 -fprofile-strip=@var{suffix} -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
346 -fopt-info-@var{options}@r{[}=@var{file}@r{]} @gol
347 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
348 -fstack-usage -ftest-coverage -ftime-report -fvar-tracking @gol
349 -fvar-tracking-assignments -fvar-tracking-assignments-toggle @gol
350 -g -g@var{level} -gtoggle -gcoff -gdwarf-@var{version} @gol
351 -ggdb -grecord-gcc-switches -gno-record-gcc-switches @gol
352 -gmlt -gstabs -gstabs+ -gstrict-dwarf -gno-strict-dwarf @gol
353 -gvms -gxcoff -gxcoff+ -gz@r{[}=@var{type}@r{]} @gol
354 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
355 -fdebug-prefix-map=@var{old}=@var{new} @gol
356 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
357 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
358 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
359 -print-multi-directory -print-multi-lib -print-multi-os-directory @gol
360 -print-prog-name=@var{program} -print-search-dirs -Q @gol
361 -print-sysroot -print-sysroot-headers-suffix @gol
362 -save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
364 @item Optimization Options
365 @xref{Optimize Options,,Options that Control Optimization}.
366 @gccoptlist{-faggressive-loop-optimizations -falign-functions[=@var{n}] @gol
367 -falign-jumps[=@var{n}] @gol
368 -falign-labels[=@var{n}] -falign-loops[=@var{n}] @gol
369 -fassociative-math -fauto-inc-dec -fbranch-probabilities @gol
370 -fbranch-target-load-optimize -fbranch-target-load-optimize2 @gol
371 -fbtr-bb-exclusive -fcaller-saves @gol
372 -fcheck-data-deps -fclone-hot-version-paths @gol
373 -fcombine-stack-adjustments -fconserve-stack @gol
374 -fcompare-elim -fcprop-registers -fcrossjumping @gol
375 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules @gol
376 -fcx-limited-range @gol
377 -fdata-sections -fdce -fdelayed-branch @gol
378 -fdelete-null-pointer-checks -fdevirtualize -fdevirtualize-speculatively -fdse @gol
379 -fearly-inlining -fipa-sra -fexpensive-optimizations -ffat-lto-objects @gol
380 -ffast-math -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
381 -fforward-propagate -ffp-contract=@var{style} -ffunction-sections @gol
382 -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm -fgraphite-identity @gol
383 -fgcse-sm -fhoist-adjacent-loads -fif-conversion @gol
384 -fif-conversion2 -findirect-inlining @gol
385 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
386 -finline-small-functions -fipa-cp -fipa-cp-clone @gol
387 -fipa-pta -fipa-profile -fipa-pure-const -fipa-reference @gol
388 -fira-algorithm=@var{algorithm} @gol
389 -fira-region=@var{region} -fira-hoist-pressure @gol
390 -fira-loop-pressure -fno-ira-share-save-slots @gol
391 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
392 -fisolate-erroneous-paths-dereference -fisolate-erroneous-paths-attribute
393 -fivopts -fkeep-inline-functions -fkeep-static-consts -flive-range-shrinkage @gol
394 -floop-block -floop-interchange -floop-strip-mine -floop-nest-optimize @gol
395 -floop-parallelize-all -flto -flto-compression-level @gol
396 -flto-partition=@var{alg} -flto-report -flto-report-wpa -fmerge-all-constants @gol
397 -fmerge-constants -fmodulo-sched -fmodulo-sched-allow-regmoves @gol
398 -fmove-loop-invariants -fno-branch-count-reg @gol
399 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
400 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
401 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
402 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
403 -fomit-frame-pointer -foptimize-sibling-calls @gol
404 -fpartial-inlining -fpeel-loops -fpredictive-commoning @gol
405 -fprefetch-loop-arrays -fprofile-report @gol
406 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
407 -fprofile-generate=@var{path} -fprofile-generate-sampling @gol
408 -fprofile-use -fprofile-use=@var{path} -fprofile-values -fprofile-reorder-functions @gol
409 -freciprocal-math -free -frename-registers -freorder-blocks @gol
410 -frecord-gcc-switches-in-elf@gol
411 -freorder-blocks-and-partition -freorder-functions @gol
412 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
413 -fripa -fripa-disallow-asm-modules -fripa-disallow-opt-mismatch @gol
414 -fripa-inc-path-sub=@var{path_mapping} -fripa-no-promote-always-inline-func @gol
415 -fripa-verbose -frounding-math @gol
416 -fsched2-use-superblocks -fsched-pressure @gol
417 -fsched-spec-load -fsched-spec-load-dangerous @gol
418 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
419 -fsched-group-heuristic -fsched-critical-path-heuristic @gol
420 -fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
421 -fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
422 -fschedule-insns -fschedule-insns2 -fsection-anchors @gol
423 -fselective-scheduling -fselective-scheduling2 @gol
424 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
425 -fsemantic-interposition @gol
426 -fshrink-wrap -fsignaling-nans -fsingle-precision-constant @gol
427 -fsplit-ivs-in-unroller -fsplit-wide-types -fssa-phiopt -fstack-protector @gol
428 -fstack-protector-all -fstack-protector-strong -fstrict-aliasing @gol
429 -fstrict-overflow -fthread-jumps -ftracer -ftree-bit-ccp @gol
430 -ftree-builtin-call-dce -ftree-ccp -ftree-ch @gol
431 -ftree-coalesce-inline-vars -ftree-coalesce-vars -ftree-copy-prop @gol
432 -ftree-copyrename -ftree-dce -ftree-dominator-opts -ftree-dse @gol
433 -ftree-forwprop -ftree-fre -ftree-loop-if-convert @gol
434 -ftree-loop-if-convert-stores -ftree-loop-im @gol
435 -ftree-phiprop -ftree-loop-distribution -ftree-loop-distribute-patterns @gol
436 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
437 -ftree-loop-vectorize @gol
438 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-partial-pre -ftree-pta @gol
439 -ftree-reassoc -ftree-sink -ftree-slsr -ftree-sra @gol
440 -ftree-switch-conversion -ftree-tail-merge -ftree-ter @gol
441 -ftree-vectorize -ftree-vrp @gol
442 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
443 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
444 -fuse-caller-save -fvariable-expansion-in-unroller -fvect-cost-model -fvpt @gol
445 -fweb -fwhole-program -fwpa -fuse-ld=@var{linker} -fuse-linker-plugin @gol
446 --param @var{name}=@var{value}
447 -O -O0 -O1 -O2 -O3 -Os -Ofast -Og}
449 @item Preprocessor Options
450 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
451 @gccoptlist{-A@var{question}=@var{answer} @gol
452 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
453 -C -dD -dI -dM -dN @gol
454 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
455 -idirafter @var{dir} @gol
456 -include @var{file} -imacros @var{file} @gol
457 -iprefix @var{file} -iwithprefix @var{dir} @gol
458 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
459 -imultilib @var{dir} -isysroot @var{dir} @gol
460 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
461 -P -fdebug-cpp -ftrack-macro-expansion -fworking-directory @gol
462 -remap -trigraphs -undef -U@var{macro} @gol
463 -Wp,@var{option} -Xpreprocessor @var{option} -no-integrated-cpp}
465 @item Assembler Option
466 @xref{Assembler Options,,Passing Options to the Assembler}.
467 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
470 @xref{Link Options,,Options for Linking}.
471 @gccoptlist{@var{object-file-name} -l@var{library} @gol
472 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
473 -s -static -static-libgcc -static-libstdc++ @gol
474 -static-libasan -static-libtsan -static-liblsan -static-libubsan @gol
475 -shared -shared-libgcc -symbolic @gol
476 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
477 -u @var{symbol} -z @var{keyword}}
479 @item Directory Options
480 @xref{Directory Options,,Options for Directory Search}.
481 @gccoptlist{-B@var{prefix} -I@var{dir} -iplugindir=@var{dir} @gol
482 -iquote@var{dir} -L@var{dir} -specs=@var{file} -I- @gol
483 --sysroot=@var{dir} --no-sysroot-suffix}
485 @item Machine Dependent Options
486 @xref{Submodel Options,,Hardware Models and Configurations}.
487 @c This list is ordered alphanumerically by subsection name.
488 @c Try and put the significant identifier (CPU or system) first,
489 @c so users have a clue at guessing where the ones they want will be.
491 @emph{AArch64 Options}
492 @gccoptlist{-mabi=@var{name} -mbig-endian -mlittle-endian @gol
493 -mgeneral-regs-only @gol
494 -mcmodel=tiny -mcmodel=small -mcmodel=large @gol
496 -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
497 -mtls-dialect=desc -mtls-dialect=traditional @gol
498 -march=@var{name} -mcpu=@var{name} -mtune=@var{name}}
500 @emph{Adapteva Epiphany Options}
501 @gccoptlist{-mhalf-reg-file -mprefer-short-insn-regs @gol
502 -mbranch-cost=@var{num} -mcmove -mnops=@var{num} -msoft-cmpsf @gol
503 -msplit-lohi -mpost-inc -mpost-modify -mstack-offset=@var{num} @gol
504 -mround-nearest -mlong-calls -mshort-calls -msmall16 @gol
505 -mfp-mode=@var{mode} -mvect-double -max-vect-align=@var{num} @gol
506 -msplit-vecmove-early -m1reg-@var{reg}}
509 @gccoptlist{-mbarrel-shifter @gol
510 -mcpu=@var{cpu} -mA6 -mARC600 -mA7 -mARC700 @gol
511 -mdpfp -mdpfp-compact -mdpfp-fast -mno-dpfp-lrsr @gol
512 -mea -mno-mpy -mmul32x16 -mmul64 @gol
513 -mnorm -mspfp -mspfp-compact -mspfp-fast -msimd -msoft-float -mswap @gol
514 -mcrc -mdsp-packa -mdvbf -mlock -mmac-d16 -mmac-24 -mrtsc -mswape @gol
515 -mtelephony -mxy -misize -mannotate-align -marclinux -marclinux_prof @gol
516 -mepilogue-cfi -mlong-calls -mmedium-calls -msdata @gol
517 -mucb-mcount -mvolatile-cache @gol
518 -malign-call -mauto-modify-reg -mbbit-peephole -mno-brcc @gol
519 -mcase-vector-pcrel -mcompact-casesi -mno-cond-exec -mearly-cbranchsi @gol
520 -mexpand-adddi -mindexed-loads -mlra -mlra-priority-none @gol
521 -mlra-priority-compact mlra-priority-noncompact -mno-millicode @gol
522 -mmixed-code -mq-class -mRcq -mRcw -msize-level=@var{level} @gol
523 -mtune=@var{cpu} -mmultcost=@var{num} -munalign-prob-threshold=@var{probability}}
526 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
527 -mabi=@var{name} @gol
528 -mapcs-stack-check -mno-apcs-stack-check @gol
529 -mapcs-float -mno-apcs-float @gol
530 -mapcs-reentrant -mno-apcs-reentrant @gol
531 -msched-prolog -mno-sched-prolog @gol
532 -mlittle-endian -mbig-endian @gol
533 -mfloat-abi=@var{name} @gol
534 -mfp16-format=@var{name}
535 -mthumb-interwork -mno-thumb-interwork @gol
536 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
537 -mstructure-size-boundary=@var{n} @gol
538 -mabort-on-noreturn @gol
539 -mlong-calls -mno-long-calls @gol
540 -msingle-pic-base -mno-single-pic-base @gol
541 -mpic-register=@var{reg} @gol
542 -mnop-fun-dllimport @gol
543 -mpoke-function-name @gol
545 -mtpcs-frame -mtpcs-leaf-frame @gol
546 -mcaller-super-interworking -mcallee-super-interworking @gol
547 -mtp=@var{name} -mtls-dialect=@var{dialect} @gol
548 -mword-relocations @gol
549 -mfix-cortex-m3-ldrd @gol
550 -munaligned-access @gol
551 -mneon-for-64bits @gol
552 -mslow-flash-data @gol
556 @gccoptlist{-mmcu=@var{mcu} -maccumulate-args -mbranch-cost=@var{cost} @gol
557 -mcall-prologues -mint8 -mno-interrupts -mrelax @gol
558 -mstrict-X -mtiny-stack -Waddr-space-convert}
560 @emph{Blackfin Options}
561 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
562 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
563 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
564 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
565 -mno-id-shared-library -mshared-library-id=@var{n} @gol
566 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
567 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
568 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
572 @gccoptlist{-mbig-endian -mlittle-endian -march=@var{cpu} @gol
573 -msim -msdata=@var{sdata-type}}
576 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
577 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
578 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
579 -mstack-align -mdata-align -mconst-align @gol
580 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
581 -melf -maout -melinux -mlinux -sim -sim2 @gol
582 -mmul-bug-workaround -mno-mul-bug-workaround}
585 @gccoptlist{-mmac @gol
586 -mcr16cplus -mcr16c @gol
587 -msim -mint32 -mbit-ops
588 -mdata-model=@var{model}}
590 @emph{Darwin Options}
591 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
592 -arch_only -bind_at_load -bundle -bundle_loader @gol
593 -client_name -compatibility_version -current_version @gol
595 -dependency-file -dylib_file -dylinker_install_name @gol
596 -dynamic -dynamiclib -exported_symbols_list @gol
597 -filelist -flat_namespace -force_cpusubtype_ALL @gol
598 -force_flat_namespace -headerpad_max_install_names @gol
600 -image_base -init -install_name -keep_private_externs @gol
601 -multi_module -multiply_defined -multiply_defined_unused @gol
602 -noall_load -no_dead_strip_inits_and_terms @gol
603 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
604 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
605 -private_bundle -read_only_relocs -sectalign @gol
606 -sectobjectsymbols -whyload -seg1addr @gol
607 -sectcreate -sectobjectsymbols -sectorder @gol
608 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
609 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
610 -segprot -segs_read_only_addr -segs_read_write_addr @gol
611 -single_module -static -sub_library -sub_umbrella @gol
612 -twolevel_namespace -umbrella -undefined @gol
613 -unexported_symbols_list -weak_reference_mismatches @gol
614 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
615 -mkernel -mone-byte-bool}
617 @emph{DEC Alpha Options}
618 @gccoptlist{-mno-fp-regs -msoft-float @gol
619 -mieee -mieee-with-inexact -mieee-conformant @gol
620 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
621 -mtrap-precision=@var{mode} -mbuild-constants @gol
622 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
623 -mbwx -mmax -mfix -mcix @gol
624 -mfloat-vax -mfloat-ieee @gol
625 -mexplicit-relocs -msmall-data -mlarge-data @gol
626 -msmall-text -mlarge-text @gol
627 -mmemory-latency=@var{time}}
630 @gccoptlist{-msmall-model -mno-lsim}
633 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
634 -mhard-float -msoft-float @gol
635 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
636 -mdouble -mno-double @gol
637 -mmedia -mno-media -mmuladd -mno-muladd @gol
638 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
639 -mlinked-fp -mlong-calls -malign-labels @gol
640 -mlibrary-pic -macc-4 -macc-8 @gol
641 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
642 -moptimize-membar -mno-optimize-membar @gol
643 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
644 -mvliw-branch -mno-vliw-branch @gol
645 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
646 -mno-nested-cond-exec -mtomcat-stats @gol
650 @emph{GNU/Linux Options}
651 @gccoptlist{-mglibc -muclibc -mbionic -mandroid @gol
652 -tno-android-cc -tno-android-ld}
654 @emph{H8/300 Options}
655 @gccoptlist{-mrelax -mh -ms -mn -mexr -mno-exr -mint32 -malign-300}
658 @gccoptlist{-march=@var{architecture-type} @gol
659 -mdisable-fpregs -mdisable-indexing @gol
660 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
661 -mfixed-range=@var{register-range} @gol
662 -mjump-in-delay -mlinker-opt -mlong-calls @gol
663 -mlong-load-store -mno-disable-fpregs @gol
664 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
665 -mno-jump-in-delay -mno-long-load-store @gol
666 -mno-portable-runtime -mno-soft-float @gol
667 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
668 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
669 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
670 -munix=@var{unix-std} -nolibdld -static -threads}
672 @emph{i386 and x86-64 Options}
673 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
674 -mtune-ctrl=@var{feature-list} -mdump-tune-features -mno-default @gol
675 -mfpmath=@var{unit} @gol
676 -masm=@var{dialect} -mno-fancy-math-387 @gol
677 -mno-fp-ret-in-387 -msoft-float @gol
678 -mno-wide-multiply -mrtd -malign-double @gol
679 -mpreferred-stack-boundary=@var{num} @gol
680 -mincoming-stack-boundary=@var{num} @gol
681 -mcld -mcx16 -msahf -mmovbe -mcrc32 @gol
682 -mrecip -mrecip=@var{opt} @gol
683 -mvzeroupper -mprefer-avx128 @gol
684 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
685 -mavx2 -mavx512f -mavx512pf -mavx512er -mavx512cd -msha @gol
686 -maes -mpclmul -mfsgsbase -mrdrnd -mf16c -mfma -mprefetchwt1 @gol
687 -mclflushopt -mxsavec -mxsaves @gol
688 -msse4a -m3dnow -mpopcnt -mabm -mbmi -mtbm -mfma4 -mxop -mlzcnt @gol
689 -mbmi2 -mfxsr -mxsave -mxsaveopt -mrtm -mlwp -mthreads @gol
690 -mno-align-stringops -minline-all-stringops @gol
691 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
692 -mmemcpy-strategy=@var{strategy} -mmemset-strategy=@var{strategy}
693 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
694 -m96bit-long-double -mlong-double-64 -mlong-double-80 -mlong-double-128 @gol
695 -mregparm=@var{num} -msseregparm @gol
696 -mveclibabi=@var{type} -mvect8-ret-in-mem @gol
697 -mpc32 -mpc64 -mpc80 -mstackrealign @gol
698 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
699 -mcmodel=@var{code-model} -mabi=@var{name} -maddress-mode=@var{mode} @gol
700 -m32 -m64 -mx32 -m16 -mlarge-data-threshold=@var{num} @gol
701 -msse2avx -mfentry -m8bit-idiv @gol
702 -mavx256-split-unaligned-load -mavx256-split-unaligned-store @gol
703 -mstack-protector-guard=@var{guard}}
705 @emph{i386 and x86-64 Windows Options}
706 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll @gol
707 -mnop-fun-dllimport -mthread @gol
708 -municode -mwin32 -mwindows -fno-set-stack-executable}
711 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
712 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
713 -mconstant-gp -mauto-pic -mfused-madd @gol
714 -minline-float-divide-min-latency @gol
715 -minline-float-divide-max-throughput @gol
716 -mno-inline-float-divide @gol
717 -minline-int-divide-min-latency @gol
718 -minline-int-divide-max-throughput @gol
719 -mno-inline-int-divide @gol
720 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
721 -mno-inline-sqrt @gol
722 -mdwarf2-asm -mearly-stop-bits @gol
723 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
724 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
725 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
726 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
727 -msched-spec-ldc -msched-spec-control-ldc @gol
728 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
729 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
730 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
731 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
734 @gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
735 -msign-extend-enabled -muser-enabled}
737 @emph{M32R/D Options}
738 @gccoptlist{-m32r2 -m32rx -m32r @gol
740 -malign-loops -mno-align-loops @gol
741 -missue-rate=@var{number} @gol
742 -mbranch-cost=@var{number} @gol
743 -mmodel=@var{code-size-model-type} @gol
744 -msdata=@var{sdata-type} @gol
745 -mno-flush-func -mflush-func=@var{name} @gol
746 -mno-flush-trap -mflush-trap=@var{number} @gol
750 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
752 @emph{M680x0 Options}
753 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
754 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
755 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
756 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
757 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
758 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
759 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
760 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
764 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
765 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
766 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
767 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
768 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
771 @gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
772 -mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
773 -mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
774 -mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
777 @emph{MicroBlaze Options}
778 @gccoptlist{-msoft-float -mhard-float -msmall-divides -mcpu=@var{cpu} @gol
779 -mmemcpy -mxl-soft-mul -mxl-soft-div -mxl-barrel-shift @gol
780 -mxl-pattern-compare -mxl-stack-check -mxl-gp-opt -mno-clearbss @gol
781 -mxl-multiply-high -mxl-float-convert -mxl-float-sqrt @gol
782 -mbig-endian -mlittle-endian -mxl-reorder -mxl-mode-@var{app-model}}
785 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
786 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips32r3 -mips32r5 @gol
787 -mips64 -mips64r2 -mips64r3 -mips64r5 @gol
788 -mips16 -mno-mips16 -mflip-mips16 @gol
789 -minterlink-compressed -mno-interlink-compressed @gol
790 -minterlink-mips16 -mno-interlink-mips16 @gol
791 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
792 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
793 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
794 -mno-float -msingle-float -mdouble-float @gol
795 -mabs=@var{mode} -mnan=@var{encoding} @gol
796 -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
799 -mvirt -mno-virt @gol
801 -mmicromips -mno-micromips @gol
802 -mfpu=@var{fpu-type} @gol
803 -msmartmips -mno-smartmips @gol
804 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
805 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
806 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
807 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
808 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
809 -membedded-data -mno-embedded-data @gol
810 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
811 -mcode-readable=@var{setting} @gol
812 -msplit-addresses -mno-split-addresses @gol
813 -mexplicit-relocs -mno-explicit-relocs @gol
814 -mcheck-zero-division -mno-check-zero-division @gol
815 -mdivide-traps -mdivide-breaks @gol
816 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
817 -mmad -mno-mad -mimadd -mno-imadd -mfused-madd -mno-fused-madd -nocpp @gol
818 -mfix-24k -mno-fix-24k @gol
819 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
820 -mfix-r10000 -mno-fix-r10000 -mfix-rm7000 -mno-fix-rm7000 @gol
821 -mfix-vr4120 -mno-fix-vr4120 @gol
822 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
823 -mflush-func=@var{func} -mno-flush-func @gol
824 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
825 -mfp-exceptions -mno-fp-exceptions @gol
826 -mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
827 -mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address}
830 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
831 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
832 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
833 -mno-base-addresses -msingle-exit -mno-single-exit}
835 @emph{MN10300 Options}
836 @gccoptlist{-mmult-bug -mno-mult-bug @gol
837 -mno-am33 -mam33 -mam33-2 -mam34 @gol
838 -mtune=@var{cpu-type} @gol
839 -mreturn-pointer-on-d0 @gol
840 -mno-crt0 -mrelax -mliw -msetlb}
843 @gccoptlist{-meb -mel -mno-crt0}
845 @emph{MSP430 Options}
846 @gccoptlist{-msim -masm-hex -mmcu= -mcpu= -mlarge -msmall -mrelax @gol
850 @gccoptlist{-mbig-endian -mlittle-endian @gol
851 -mreduced-regs -mfull-regs @gol
852 -mcmov -mno-cmov @gol
853 -mperf-ext -mno-perf-ext @gol
854 -mv3push -mno-v3push @gol
855 -m16bit -mno-16bit @gol
856 -mgp-direct -mno-gp-direct @gol
857 -misr-vector-size=@var{num} @gol
858 -mcache-block-size=@var{num} @gol
859 -march=@var{arch} @gol
860 -mforce-fp-as-gp -mforbid-fp-as-gp @gol
861 -mex9 -mctor-dtor -mrelax}
863 @emph{Nios II Options}
864 @gccoptlist{-G @var{num} -mgpopt -mno-gpopt -mel -meb @gol
865 -mno-bypass-cache -mbypass-cache @gol
866 -mno-cache-volatile -mcache-volatile @gol
867 -mno-fast-sw-div -mfast-sw-div @gol
868 -mhw-mul -mno-hw-mul -mhw-mulx -mno-hw-mulx -mno-hw-div -mhw-div @gol
869 -mcustom-@var{insn}=@var{N} -mno-custom-@var{insn} @gol
870 -mcustom-fpu-cfg=@var{name} @gol
871 -mhal -msmallc -msys-crt0=@var{name} -msys-lib=@var{name}}
873 @emph{PDP-11 Options}
874 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
875 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
876 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
877 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
878 -mbranch-expensive -mbranch-cheap @gol
879 -munix-asm -mdec-asm}
881 @emph{picoChip Options}
882 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N} @gol
883 -msymbol-as-address -mno-inefficient-warnings}
885 @emph{PowerPC Options}
886 See RS/6000 and PowerPC Options.
889 @gccoptlist{-msim -mmul=none -mmul=g13 -mmul=rl78 -m64bit-doubles -m32bit-doubles}
891 @emph{RS/6000 and PowerPC Options}
892 @gccoptlist{-mcpu=@var{cpu-type} @gol
893 -mtune=@var{cpu-type} @gol
894 -mcmodel=@var{code-model} @gol
896 -maltivec -mno-altivec @gol
897 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
898 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
899 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
900 -mfprnd -mno-fprnd @gol
901 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
902 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
903 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
904 -malign-power -malign-natural @gol
905 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
906 -msingle-float -mdouble-float -msimple-fpu @gol
907 -mstring -mno-string -mupdate -mno-update @gol
908 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
909 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
910 -mstrict-align -mno-strict-align -mrelocatable @gol
911 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
912 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
913 -mdynamic-no-pic -maltivec -mswdiv -msingle-pic-base @gol
914 -mprioritize-restricted-insns=@var{priority} @gol
915 -msched-costly-dep=@var{dependence_type} @gol
916 -minsert-sched-nops=@var{scheme} @gol
917 -mcall-sysv -mcall-netbsd @gol
918 -maix-struct-return -msvr4-struct-return @gol
919 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
920 -mblock-move-inline-limit=@var{num} @gol
921 -misel -mno-isel @gol
922 -misel=yes -misel=no @gol
924 -mspe=yes -mspe=no @gol
926 -mgen-cell-microcode -mwarn-cell-microcode @gol
927 -mvrsave -mno-vrsave @gol
928 -mmulhw -mno-mulhw @gol
929 -mdlmzb -mno-dlmzb @gol
930 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
931 -mprototype -mno-prototype @gol
932 -msim -mmvme -mads -myellowknife -memb -msdata @gol
933 -msdata=@var{opt} -mvxworks -G @var{num} -pthread @gol
934 -mrecip -mrecip=@var{opt} -mno-recip -mrecip-precision @gol
935 -mno-recip-precision @gol
936 -mveclibabi=@var{type} -mfriz -mno-friz @gol
937 -mpointers-to-nested-functions -mno-pointers-to-nested-functions @gol
938 -msave-toc-indirect -mno-save-toc-indirect @gol
939 -mpower8-fusion -mno-mpower8-fusion -mpower8-vector -mno-power8-vector @gol
940 -mcrypto -mno-crypto -mdirect-move -mno-direct-move @gol
941 -mquad-memory -mno-quad-memory @gol
942 -mquad-memory-atomic -mno-quad-memory-atomic @gol
943 -mcompat-align-parm -mno-compat-align-parm}
946 @gccoptlist{-m64bit-doubles -m32bit-doubles -fpu -nofpu@gol
948 -mbig-endian-data -mlittle-endian-data @gol
951 -mas100-syntax -mno-as100-syntax@gol
953 -mmax-constant-size=@gol
956 -mno-warn-multiple-fast-interrupts@gol
957 -msave-acc-in-interrupts}
959 @emph{S/390 and zSeries Options}
960 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
961 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
962 -mlong-double-64 -mlong-double-128 @gol
963 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
964 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
965 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
966 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
967 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard @gol
968 -mhotpatch[=@var{halfwords}] -mno-hotpatch}
971 @gccoptlist{-meb -mel @gol
975 -mscore5 -mscore5u -mscore7 -mscore7d}
978 @gccoptlist{-m1 -m2 -m2e @gol
979 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
981 -m4-nofpu -m4-single-only -m4-single -m4 @gol
982 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
983 -m5-64media -m5-64media-nofpu @gol
984 -m5-32media -m5-32media-nofpu @gol
985 -m5-compact -m5-compact-nofpu @gol
986 -mb -ml -mdalign -mrelax @gol
987 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
988 -mieee -mno-ieee -mbitops -misize -minline-ic_invalidate -mpadstruct @gol
989 -mspace -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
990 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
991 -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
992 -maccumulate-outgoing-args -minvalid-symbols @gol
993 -matomic-model=@var{atomic-model} @gol
994 -mbranch-cost=@var{num} -mzdcbranch -mno-zdcbranch @gol
995 -mfused-madd -mno-fused-madd -mfsca -mno-fsca -mfsrra -mno-fsrra @gol
996 -mpretend-cmove -mtas}
998 @emph{Solaris 2 Options}
999 @gccoptlist{-mclear-hwcap -mno-clear-hwcap -mimpure-text -mno-impure-text @gol
1002 @emph{SPARC Options}
1003 @gccoptlist{-mcpu=@var{cpu-type} @gol
1004 -mtune=@var{cpu-type} @gol
1005 -mcmodel=@var{code-model} @gol
1006 -mmemory-model=@var{mem-model} @gol
1007 -m32 -m64 -mapp-regs -mno-app-regs @gol
1008 -mfaster-structs -mno-faster-structs -mflat -mno-flat @gol
1009 -mfpu -mno-fpu -mhard-float -msoft-float @gol
1010 -mhard-quad-float -msoft-quad-float @gol
1011 -mstack-bias -mno-stack-bias @gol
1012 -munaligned-doubles -mno-unaligned-doubles @gol
1013 -muser-mode -mno-user-mode @gol
1014 -mv8plus -mno-v8plus -mvis -mno-vis @gol
1015 -mvis2 -mno-vis2 -mvis3 -mno-vis3 @gol
1016 -mcbcond -mno-cbcond @gol
1017 -mfmaf -mno-fmaf -mpopc -mno-popc @gol
1018 -mfix-at697f -mfix-ut699}
1021 @gccoptlist{-mwarn-reloc -merror-reloc @gol
1022 -msafe-dma -munsafe-dma @gol
1024 -msmall-mem -mlarge-mem -mstdmain @gol
1025 -mfixed-range=@var{register-range} @gol
1027 -maddress-space-conversion -mno-address-space-conversion @gol
1028 -mcache-size=@var{cache-size} @gol
1029 -matomic-updates -mno-atomic-updates}
1031 @emph{System V Options}
1032 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
1034 @emph{TILE-Gx Options}
1035 @gccoptlist{-mcpu=CPU -m32 -m64 -mbig-endian -mlittle-endian @gol
1036 -mcmodel=@var{code-model}}
1038 @emph{TILEPro Options}
1039 @gccoptlist{-mcpu=@var{cpu} -m32}
1042 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
1043 -mprolog-function -mno-prolog-function -mspace @gol
1044 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
1045 -mapp-regs -mno-app-regs @gol
1046 -mdisable-callt -mno-disable-callt @gol
1047 -mv850e2v3 -mv850e2 -mv850e1 -mv850es @gol
1048 -mv850e -mv850 -mv850e3v5 @gol
1059 @gccoptlist{-mg -mgnu -munix}
1062 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64 @gol
1063 -mpointer-size=@var{size}}
1065 @emph{VxWorks Options}
1066 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
1067 -Xbind-lazy -Xbind-now}
1069 @emph{x86-64 Options}
1070 See i386 and x86-64 Options.
1072 @emph{Xstormy16 Options}
1075 @emph{Xtensa Options}
1076 @gccoptlist{-mconst16 -mno-const16 @gol
1077 -mfused-madd -mno-fused-madd @gol
1079 -mserialize-volatile -mno-serialize-volatile @gol
1080 -mtext-section-literals -mno-text-section-literals @gol
1081 -mtarget-align -mno-target-align @gol
1082 -mlongcalls -mno-longcalls}
1084 @emph{zSeries Options}
1085 See S/390 and zSeries Options.
1087 @item Code Generation Options
1088 @xref{Code Gen Options,,Options for Code Generation Conventions}.
1089 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
1090 -ffixed-@var{reg} -fexceptions @gol
1091 -fnon-call-exceptions -fdelete-dead-exceptions -funwind-tables @gol
1092 -fasynchronous-unwind-tables @gol
1093 -fno-gnu-unique @gol
1094 -finhibit-size-directive -finstrument-functions @gol
1095 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
1096 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
1097 -fno-common -fno-ident @gol
1098 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
1099 -fno-jump-tables @gol
1100 -frecord-gcc-switches @gol
1101 -freg-struct-return -fshort-enums @gol
1102 -fshort-double -fshort-wchar @gol
1103 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
1104 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
1105 -fno-stack-limit -fsplit-stack @gol
1106 -fleading-underscore -ftls-model=@var{model} @gol
1107 -fstack-reuse=@var{reuse_level} @gol
1108 -ftrapv -fwrapv -fbounds-check @gol
1109 -fvisibility -fstrict-volatile-bitfields -fsync-libcalls}
1113 @node Overall Options
1114 @section Options Controlling the Kind of Output
1116 Compilation can involve up to four stages: preprocessing, compilation
1117 proper, assembly and linking, always in that order. GCC is capable of
1118 preprocessing and compiling several files either into several
1119 assembler input files, or into one assembler input file; then each
1120 assembler input file produces an object file, and linking combines all
1121 the object files (those newly compiled, and those specified as input)
1122 into an executable file.
1124 @cindex file name suffix
1125 For any given input file, the file name suffix determines what kind of
1126 compilation is done:
1130 C source code that must be preprocessed.
1133 C source code that should not be preprocessed.
1136 C++ source code that should not be preprocessed.
1139 Objective-C source code. Note that you must link with the @file{libobjc}
1140 library to make an Objective-C program work.
1143 Objective-C source code that should not be preprocessed.
1147 Objective-C++ source code. Note that you must link with the @file{libobjc}
1148 library to make an Objective-C++ program work. Note that @samp{.M} refers
1149 to a literal capital M@.
1151 @item @var{file}.mii
1152 Objective-C++ source code that should not be preprocessed.
1155 C, C++, Objective-C or Objective-C++ header file to be turned into a
1156 precompiled header (default), or C, C++ header file to be turned into an
1157 Ada spec (via the @option{-fdump-ada-spec} switch).
1160 @itemx @var{file}.cp
1161 @itemx @var{file}.cxx
1162 @itemx @var{file}.cpp
1163 @itemx @var{file}.CPP
1164 @itemx @var{file}.c++
1166 C++ source code that must be preprocessed. Note that in @samp{.cxx},
1167 the last two letters must both be literally @samp{x}. Likewise,
1168 @samp{.C} refers to a literal capital C@.
1172 Objective-C++ source code that must be preprocessed.
1174 @item @var{file}.mii
1175 Objective-C++ source code that should not be preprocessed.
1179 @itemx @var{file}.hp
1180 @itemx @var{file}.hxx
1181 @itemx @var{file}.hpp
1182 @itemx @var{file}.HPP
1183 @itemx @var{file}.h++
1184 @itemx @var{file}.tcc
1185 C++ header file to be turned into a precompiled header or Ada spec.
1188 @itemx @var{file}.for
1189 @itemx @var{file}.ftn
1190 Fixed form Fortran source code that should not be preprocessed.
1193 @itemx @var{file}.FOR
1194 @itemx @var{file}.fpp
1195 @itemx @var{file}.FPP
1196 @itemx @var{file}.FTN
1197 Fixed form Fortran source code that must be preprocessed (with the traditional
1200 @item @var{file}.f90
1201 @itemx @var{file}.f95
1202 @itemx @var{file}.f03
1203 @itemx @var{file}.f08
1204 Free form Fortran source code that should not be preprocessed.
1206 @item @var{file}.F90
1207 @itemx @var{file}.F95
1208 @itemx @var{file}.F03
1209 @itemx @var{file}.F08
1210 Free form Fortran source code that must be preprocessed (with the
1211 traditional preprocessor).
1216 @c FIXME: Descriptions of Java file types.
1222 @item @var{file}.ads
1223 Ada source code file that contains a library unit declaration (a
1224 declaration of a package, subprogram, or generic, or a generic
1225 instantiation), or a library unit renaming declaration (a package,
1226 generic, or subprogram renaming declaration). Such files are also
1229 @item @var{file}.adb
1230 Ada source code file containing a library unit body (a subprogram or
1231 package body). Such files are also called @dfn{bodies}.
1233 @c GCC also knows about some suffixes for languages not yet included:
1244 @itemx @var{file}.sx
1245 Assembler code that must be preprocessed.
1248 An object file to be fed straight into linking.
1249 Any file name with no recognized suffix is treated this way.
1253 You can specify the input language explicitly with the @option{-x} option:
1256 @item -x @var{language}
1257 Specify explicitly the @var{language} for the following input files
1258 (rather than letting the compiler choose a default based on the file
1259 name suffix). This option applies to all following input files until
1260 the next @option{-x} option. Possible values for @var{language} are:
1262 c c-header cpp-output
1263 c++ c++-header c++-cpp-output
1264 objective-c objective-c-header objective-c-cpp-output
1265 objective-c++ objective-c++-header objective-c++-cpp-output
1266 assembler assembler-with-cpp
1268 f77 f77-cpp-input f95 f95-cpp-input
1274 Turn off any specification of a language, so that subsequent files are
1275 handled according to their file name suffixes (as they are if @option{-x}
1276 has not been used at all).
1278 @item -pass-exit-codes
1279 @opindex pass-exit-codes
1280 Normally the @command{gcc} program exits with the code of 1 if any
1281 phase of the compiler returns a non-success return code. If you specify
1282 @option{-pass-exit-codes}, the @command{gcc} program instead returns with
1283 the numerically highest error produced by any phase returning an error
1284 indication. The C, C++, and Fortran front ends return 4 if an internal
1285 compiler error is encountered.
1288 If you only want some of the stages of compilation, you can use
1289 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1290 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1291 @command{gcc} is to stop. Note that some combinations (for example,
1292 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1297 Compile or assemble the source files, but do not link. The linking
1298 stage simply is not done. The ultimate output is in the form of an
1299 object file for each source file.
1301 By default, the object file name for a source file is made by replacing
1302 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1304 Unrecognized input files, not requiring compilation or assembly, are
1309 Stop after the stage of compilation proper; do not assemble. The output
1310 is in the form of an assembler code file for each non-assembler input
1313 By default, the assembler file name for a source file is made by
1314 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1316 Input files that don't require compilation are ignored.
1320 Stop after the preprocessing stage; do not run the compiler proper. The
1321 output is in the form of preprocessed source code, which is sent to the
1324 Input files that don't require preprocessing are ignored.
1326 @cindex output file option
1329 Place output in file @var{file}. This applies to whatever
1330 sort of output is being produced, whether it be an executable file,
1331 an object file, an assembler file or preprocessed C code.
1333 If @option{-o} is not specified, the default is to put an executable
1334 file in @file{a.out}, the object file for
1335 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1336 assembler file in @file{@var{source}.s}, a precompiled header file in
1337 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1342 Print (on standard error output) the commands executed to run the stages
1343 of compilation. Also print the version number of the compiler driver
1344 program and of the preprocessor and the compiler proper.
1348 Like @option{-v} except the commands are not executed and arguments
1349 are quoted unless they contain only alphanumeric characters or @code{./-_}.
1350 This is useful for shell scripts to capture the driver-generated command lines.
1354 Use pipes rather than temporary files for communication between the
1355 various stages of compilation. This fails to work on some systems where
1356 the assembler is unable to read from a pipe; but the GNU assembler has
1361 Print (on the standard output) a description of the command-line options
1362 understood by @command{gcc}. If the @option{-v} option is also specified
1363 then @option{--help} is also passed on to the various processes
1364 invoked by @command{gcc}, so that they can display the command-line options
1365 they accept. If the @option{-Wextra} option has also been specified
1366 (prior to the @option{--help} option), then command-line options that
1367 have no documentation associated with them are also displayed.
1370 @opindex target-help
1371 Print (on the standard output) a description of target-specific command-line
1372 options for each tool. For some targets extra target-specific
1373 information may also be printed.
1375 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1376 Print (on the standard output) a description of the command-line
1377 options understood by the compiler that fit into all specified classes
1378 and qualifiers. These are the supported classes:
1381 @item @samp{optimizers}
1382 Display all of the optimization options supported by the
1385 @item @samp{warnings}
1386 Display all of the options controlling warning messages
1387 produced by the compiler.
1390 Display target-specific options. Unlike the
1391 @option{--target-help} option however, target-specific options of the
1392 linker and assembler are not displayed. This is because those
1393 tools do not currently support the extended @option{--help=} syntax.
1396 Display the values recognized by the @option{--param}
1399 @item @var{language}
1400 Display the options supported for @var{language}, where
1401 @var{language} is the name of one of the languages supported in this
1405 Display the options that are common to all languages.
1408 These are the supported qualifiers:
1411 @item @samp{undocumented}
1412 Display only those options that are undocumented.
1415 Display options taking an argument that appears after an equal
1416 sign in the same continuous piece of text, such as:
1417 @samp{--help=target}.
1419 @item @samp{separate}
1420 Display options taking an argument that appears as a separate word
1421 following the original option, such as: @samp{-o output-file}.
1424 Thus for example to display all the undocumented target-specific
1425 switches supported by the compiler, use:
1428 --help=target,undocumented
1431 The sense of a qualifier can be inverted by prefixing it with the
1432 @samp{^} character, so for example to display all binary warning
1433 options (i.e., ones that are either on or off and that do not take an
1434 argument) that have a description, use:
1437 --help=warnings,^joined,^undocumented
1440 The argument to @option{--help=} should not consist solely of inverted
1443 Combining several classes is possible, although this usually
1444 restricts the output so much that there is nothing to display. One
1445 case where it does work, however, is when one of the classes is
1446 @var{target}. For example, to display all the target-specific
1447 optimization options, use:
1450 --help=target,optimizers
1453 The @option{--help=} option can be repeated on the command line. Each
1454 successive use displays its requested class of options, skipping
1455 those that have already been displayed.
1457 If the @option{-Q} option appears on the command line before the
1458 @option{--help=} option, then the descriptive text displayed by
1459 @option{--help=} is changed. Instead of describing the displayed
1460 options, an indication is given as to whether the option is enabled,
1461 disabled or set to a specific value (assuming that the compiler
1462 knows this at the point where the @option{--help=} option is used).
1464 Here is a truncated example from the ARM port of @command{gcc}:
1467 % gcc -Q -mabi=2 --help=target -c
1468 The following options are target specific:
1470 -mabort-on-noreturn [disabled]
1474 The output is sensitive to the effects of previous command-line
1475 options, so for example it is possible to find out which optimizations
1476 are enabled at @option{-O2} by using:
1479 -Q -O2 --help=optimizers
1482 Alternatively you can discover which binary optimizations are enabled
1483 by @option{-O3} by using:
1486 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1487 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1488 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1491 @item -canonical-prefixes
1492 @opindex canonical-prefixes
1493 Always expand any symbolic links, resolve references to @samp{/../}
1494 or @samp{/./}, and make the path absolute when generating a relative
1497 @item -no-canonical-prefixes
1498 @opindex no-canonical-prefixes
1499 Never expand any symbolic links, resolve references to @samp{/../}
1500 or @samp{/./}, or make the path absolute when generating a relative
1501 prefix. If neither @option{-canonical-prefixes} nor
1502 @option{-nocanonical-prefixes} is given, GCC tries to set an appropriate
1503 default by looking for a target-specific subdirectory alongside the
1504 directory containing the compiler driver.
1508 Display the version number and copyrights of the invoked GCC@.
1512 Invoke all subcommands under a wrapper program. The name of the
1513 wrapper program and its parameters are passed as a comma separated
1517 gcc -c t.c -wrapper gdb,--args
1521 This invokes all subprograms of @command{gcc} under
1522 @samp{gdb --args}, thus the invocation of @command{cc1} is
1523 @samp{gdb --args cc1 @dots{}}.
1525 @item -fplugin=@var{name}.so
1527 Load the plugin code in file @var{name}.so, assumed to be a
1528 shared object to be dlopen'd by the compiler. The base name of
1529 the shared object file is used to identify the plugin for the
1530 purposes of argument parsing (See
1531 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1532 Each plugin should define the callback functions specified in the
1535 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1536 @opindex fplugin-arg
1537 Define an argument called @var{key} with a value of @var{value}
1538 for the plugin called @var{name}.
1540 @item -fdump-ada-spec@r{[}-slim@r{]}
1541 @opindex fdump-ada-spec
1542 For C and C++ source and include files, generate corresponding Ada specs.
1543 @xref{Generating Ada Bindings for C and C++ headers,,, gnat_ugn,
1544 GNAT User's Guide}, which provides detailed documentation on this feature.
1546 @item -fada-spec-parent=@var{unit}
1547 @opindex fada-spec-parent
1548 In conjunction with @option{-fdump-ada-spec@r{[}-slim@r{]}} above, generate
1549 Ada specs as child units of parent @var{unit}.
1551 @item -fdump-go-spec=@var{file}
1552 @opindex fdump-go-spec
1553 For input files in any language, generate corresponding Go
1554 declarations in @var{file}. This generates Go @code{const},
1555 @code{type}, @code{var}, and @code{func} declarations which may be a
1556 useful way to start writing a Go interface to code written in some
1559 @include @value{srcdir}/../libiberty/at-file.texi
1563 @section Compiling C++ Programs
1565 @cindex suffixes for C++ source
1566 @cindex C++ source file suffixes
1567 C++ source files conventionally use one of the suffixes @samp{.C},
1568 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1569 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1570 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1571 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1572 files with these names and compiles them as C++ programs even if you
1573 call the compiler the same way as for compiling C programs (usually
1574 with the name @command{gcc}).
1578 However, the use of @command{gcc} does not add the C++ library.
1579 @command{g++} is a program that calls GCC and automatically specifies linking
1580 against the C++ library. It treats @samp{.c},
1581 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1582 files unless @option{-x} is used. This program is also useful when
1583 precompiling a C header file with a @samp{.h} extension for use in C++
1584 compilations. On many systems, @command{g++} is also installed with
1585 the name @command{c++}.
1587 @cindex invoking @command{g++}
1588 When you compile C++ programs, you may specify many of the same
1589 command-line options that you use for compiling programs in any
1590 language; or command-line options meaningful for C and related
1591 languages; or options that are meaningful only for C++ programs.
1592 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1593 explanations of options for languages related to C@.
1594 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1595 explanations of options that are meaningful only for C++ programs.
1597 @node C Dialect Options
1598 @section Options Controlling C Dialect
1599 @cindex dialect options
1600 @cindex language dialect options
1601 @cindex options, dialect
1603 The following options control the dialect of C (or languages derived
1604 from C, such as C++, Objective-C and Objective-C++) that the compiler
1608 @cindex ANSI support
1612 In C mode, this is equivalent to @option{-std=c90}. In C++ mode, it is
1613 equivalent to @option{-std=c++98}.
1615 This turns off certain features of GCC that are incompatible with ISO
1616 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1617 such as the @code{asm} and @code{typeof} keywords, and
1618 predefined macros such as @code{unix} and @code{vax} that identify the
1619 type of system you are using. It also enables the undesirable and
1620 rarely used ISO trigraph feature. For the C compiler,
1621 it disables recognition of C++ style @samp{//} comments as well as
1622 the @code{inline} keyword.
1624 The alternate keywords @code{__asm__}, @code{__extension__},
1625 @code{__inline__} and @code{__typeof__} continue to work despite
1626 @option{-ansi}. You would not want to use them in an ISO C program, of
1627 course, but it is useful to put them in header files that might be included
1628 in compilations done with @option{-ansi}. Alternate predefined macros
1629 such as @code{__unix__} and @code{__vax__} are also available, with or
1630 without @option{-ansi}.
1632 The @option{-ansi} option does not cause non-ISO programs to be
1633 rejected gratuitously. For that, @option{-Wpedantic} is required in
1634 addition to @option{-ansi}. @xref{Warning Options}.
1636 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1637 option is used. Some header files may notice this macro and refrain
1638 from declaring certain functions or defining certain macros that the
1639 ISO standard doesn't call for; this is to avoid interfering with any
1640 programs that might use these names for other things.
1642 Functions that are normally built in but do not have semantics
1643 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1644 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1645 built-in functions provided by GCC}, for details of the functions
1650 Determine the language standard. @xref{Standards,,Language Standards
1651 Supported by GCC}, for details of these standard versions. This option
1652 is currently only supported when compiling C or C++.
1654 The compiler can accept several base standards, such as @samp{c90} or
1655 @samp{c++98}, and GNU dialects of those standards, such as
1656 @samp{gnu90} or @samp{gnu++98}. When a base standard is specified, the
1657 compiler accepts all programs following that standard plus those
1658 using GNU extensions that do not contradict it. For example,
1659 @option{-std=c90} turns off certain features of GCC that are
1660 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1661 keywords, but not other GNU extensions that do not have a meaning in
1662 ISO C90, such as omitting the middle term of a @code{?:}
1663 expression. On the other hand, when a GNU dialect of a standard is
1664 specified, all features supported by the compiler are enabled, even when
1665 those features change the meaning of the base standard. As a result, some
1666 strict-conforming programs may be rejected. The particular standard
1667 is used by @option{-Wpedantic} to identify which features are GNU
1668 extensions given that version of the standard. For example
1669 @option{-std=gnu90 -Wpedantic} warns about C++ style @samp{//}
1670 comments, while @option{-std=gnu99 -Wpedantic} does not.
1672 A value for this option must be provided; possible values are
1678 Support all ISO C90 programs (certain GNU extensions that conflict
1679 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1681 @item iso9899:199409
1682 ISO C90 as modified in amendment 1.
1688 ISO C99. This standard is substantially completely supported, modulo
1689 bugs, extended identifiers (supported except for corner cases when
1690 @option{-fextended-identifiers} is used) and floating-point issues
1691 (mainly but not entirely relating to optional C99 features from
1692 Annexes F and G). See
1693 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1694 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1699 ISO C11, the 2011 revision of the ISO C standard. This standard is
1700 substantially completely supported, modulo bugs, extended identifiers
1701 (supported except for corner cases when
1702 @option{-fextended-identifiers} is used), floating-point issues
1703 (mainly but not entirely relating to optional C11 features from
1704 Annexes F and G) and the optional Annexes K (Bounds-checking
1705 interfaces) and L (Analyzability). The name @samp{c1x} is deprecated.
1709 GNU dialect of ISO C90 (including some C99 features). This
1710 is the default for C code.
1714 GNU dialect of ISO C99. The name @samp{gnu9x} is deprecated.
1718 GNU dialect of ISO C11. This is intended to become the default in a
1719 future release of GCC. The name @samp{gnu1x} is deprecated.
1723 The 1998 ISO C++ standard plus the 2003 technical corrigendum and some
1724 additional defect reports. Same as @option{-ansi} for C++ code.
1728 GNU dialect of @option{-std=c++98}. This is the default for
1733 The 2011 ISO C++ standard plus amendments.
1734 The name @samp{c++0x} is deprecated.
1738 GNU dialect of @option{-std=c++11}.
1739 The name @samp{gnu++0x} is deprecated.
1742 The next revision of the ISO C++ standard, tentatively planned for
1743 2014. Support is highly experimental, and will almost certainly
1744 change in incompatible ways in future releases.
1747 GNU dialect of @option{-std=c++1y}. Support is highly experimental,
1748 and will almost certainly change in incompatible ways in future
1752 @item -fgnu89-inline
1753 @opindex fgnu89-inline
1754 The option @option{-fgnu89-inline} tells GCC to use the traditional
1755 GNU semantics for @code{inline} functions when in C99 mode.
1756 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1757 is accepted and ignored by GCC versions 4.1.3 up to but not including
1758 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1759 C99 mode. Using this option is roughly equivalent to adding the
1760 @code{gnu_inline} function attribute to all inline functions
1761 (@pxref{Function Attributes}).
1763 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1764 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1765 specifies the default behavior). This option was first supported in
1766 GCC 4.3. This option is not supported in @option{-std=c90} or
1767 @option{-std=gnu90} mode.
1769 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1770 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1771 in effect for @code{inline} functions. @xref{Common Predefined
1772 Macros,,,cpp,The C Preprocessor}.
1774 @item -aux-info @var{filename}
1776 Output to the given filename prototyped declarations for all functions
1777 declared and/or defined in a translation unit, including those in header
1778 files. This option is silently ignored in any language other than C@.
1780 Besides declarations, the file indicates, in comments, the origin of
1781 each declaration (source file and line), whether the declaration was
1782 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1783 @samp{O} for old, respectively, in the first character after the line
1784 number and the colon), and whether it came from a declaration or a
1785 definition (@samp{C} or @samp{F}, respectively, in the following
1786 character). In the case of function definitions, a K&R-style list of
1787 arguments followed by their declarations is also provided, inside
1788 comments, after the declaration.
1790 @item -fallow-parameterless-variadic-functions
1791 @opindex fallow-parameterless-variadic-functions
1792 Accept variadic functions without named parameters.
1794 Although it is possible to define such a function, this is not very
1795 useful as it is not possible to read the arguments. This is only
1796 supported for C as this construct is allowed by C++.
1800 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1801 keyword, so that code can use these words as identifiers. You can use
1802 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1803 instead. @option{-ansi} implies @option{-fno-asm}.
1805 In C++, this switch only affects the @code{typeof} keyword, since
1806 @code{asm} and @code{inline} are standard keywords. You may want to
1807 use the @option{-fno-gnu-keywords} flag instead, which has the same
1808 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1809 switch only affects the @code{asm} and @code{typeof} keywords, since
1810 @code{inline} is a standard keyword in ISO C99.
1813 @itemx -fno-builtin-@var{function}
1814 @opindex fno-builtin
1815 @cindex built-in functions
1816 Don't recognize built-in functions that do not begin with
1817 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1818 functions provided by GCC}, for details of the functions affected,
1819 including those which are not built-in functions when @option{-ansi} or
1820 @option{-std} options for strict ISO C conformance are used because they
1821 do not have an ISO standard meaning.
1823 GCC normally generates special code to handle certain built-in functions
1824 more efficiently; for instance, calls to @code{alloca} may become single
1825 instructions which adjust the stack directly, and calls to @code{memcpy}
1826 may become inline copy loops. The resulting code is often both smaller
1827 and faster, but since the function calls no longer appear as such, you
1828 cannot set a breakpoint on those calls, nor can you change the behavior
1829 of the functions by linking with a different library. In addition,
1830 when a function is recognized as a built-in function, GCC may use
1831 information about that function to warn about problems with calls to
1832 that function, or to generate more efficient code, even if the
1833 resulting code still contains calls to that function. For example,
1834 warnings are given with @option{-Wformat} for bad calls to
1835 @code{printf} when @code{printf} is built in and @code{strlen} is
1836 known not to modify global memory.
1838 With the @option{-fno-builtin-@var{function}} option
1839 only the built-in function @var{function} is
1840 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1841 function is named that is not built-in in this version of GCC, this
1842 option is ignored. There is no corresponding
1843 @option{-fbuiltin-@var{function}} option; if you wish to enable
1844 built-in functions selectively when using @option{-fno-builtin} or
1845 @option{-ffreestanding}, you may define macros such as:
1848 #define abs(n) __builtin_abs ((n))
1849 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1854 @cindex hosted environment
1856 Assert that compilation targets a hosted environment. This implies
1857 @option{-fbuiltin}. A hosted environment is one in which the
1858 entire standard library is available, and in which @code{main} has a return
1859 type of @code{int}. Examples are nearly everything except a kernel.
1860 This is equivalent to @option{-fno-freestanding}.
1862 @item -ffreestanding
1863 @opindex ffreestanding
1864 @cindex hosted environment
1866 Assert that compilation targets a freestanding environment. This
1867 implies @option{-fno-builtin}. A freestanding environment
1868 is one in which the standard library may not exist, and program startup may
1869 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1870 This is equivalent to @option{-fno-hosted}.
1872 @xref{Standards,,Language Standards Supported by GCC}, for details of
1873 freestanding and hosted environments.
1877 @cindex OpenMP parallel
1878 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1879 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1880 compiler generates parallel code according to the OpenMP Application
1881 Program Interface v4.0 @w{@uref{http://www.openmp.org/}}. This option
1882 implies @option{-pthread}, and thus is only supported on targets that
1883 have support for @option{-pthread}. @option{-fopenmp} implies
1884 @option{-fopenmp-simd}.
1887 @opindex fopenmp-simd
1890 Enable handling of OpenMP's SIMD directives with @code{#pragma omp}
1891 in C/C++ and @code{!$omp} in Fortran. Other OpenMP directives
1896 @cindex Enable Cilk Plus
1897 Enable the usage of Cilk Plus language extension features for C/C++.
1898 When the option @option{-fcilkplus} is specified, enable the usage of
1899 the Cilk Plus Language extension features for C/C++. The present
1900 implementation follows ABI version 1.2. This is an experimental
1901 feature that is only partially complete, and whose interface may
1902 change in future versions of GCC as the official specification
1903 changes. Currently, all features but @code{_Cilk_for} have been
1908 When the option @option{-fgnu-tm} is specified, the compiler
1909 generates code for the Linux variant of Intel's current Transactional
1910 Memory ABI specification document (Revision 1.1, May 6 2009). This is
1911 an experimental feature whose interface may change in future versions
1912 of GCC, as the official specification changes. Please note that not
1913 all architectures are supported for this feature.
1915 For more information on GCC's support for transactional memory,
1916 @xref{Enabling libitm,,The GNU Transactional Memory Library,libitm,GNU
1917 Transactional Memory Library}.
1919 Note that the transactional memory feature is not supported with
1920 non-call exceptions (@option{-fnon-call-exceptions}).
1922 @item -fms-extensions
1923 @opindex fms-extensions
1924 Accept some non-standard constructs used in Microsoft header files.
1926 In C++ code, this allows member names in structures to be similar
1927 to previous types declarations.
1936 Some cases of unnamed fields in structures and unions are only
1937 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1938 fields within structs/unions}, for details.
1940 Note that this option is off for all targets but i?86 and x86_64
1941 targets using ms-abi.
1942 @item -fplan9-extensions
1943 Accept some non-standard constructs used in Plan 9 code.
1945 This enables @option{-fms-extensions}, permits passing pointers to
1946 structures with anonymous fields to functions that expect pointers to
1947 elements of the type of the field, and permits referring to anonymous
1948 fields declared using a typedef. @xref{Unnamed Fields,,Unnamed
1949 struct/union fields within structs/unions}, for details. This is only
1950 supported for C, not C++.
1954 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1955 options for strict ISO C conformance) implies @option{-trigraphs}.
1957 @cindex traditional C language
1958 @cindex C language, traditional
1960 @itemx -traditional-cpp
1961 @opindex traditional-cpp
1962 @opindex traditional
1963 Formerly, these options caused GCC to attempt to emulate a pre-standard
1964 C compiler. They are now only supported with the @option{-E} switch.
1965 The preprocessor continues to support a pre-standard mode. See the GNU
1966 CPP manual for details.
1968 @item -fcond-mismatch
1969 @opindex fcond-mismatch
1970 Allow conditional expressions with mismatched types in the second and
1971 third arguments. The value of such an expression is void. This option
1972 is not supported for C++.
1974 @item -flax-vector-conversions
1975 @opindex flax-vector-conversions
1976 Allow implicit conversions between vectors with differing numbers of
1977 elements and/or incompatible element types. This option should not be
1980 @item -funsigned-char
1981 @opindex funsigned-char
1982 Let the type @code{char} be unsigned, like @code{unsigned char}.
1984 Each kind of machine has a default for what @code{char} should
1985 be. It is either like @code{unsigned char} by default or like
1986 @code{signed char} by default.
1988 Ideally, a portable program should always use @code{signed char} or
1989 @code{unsigned char} when it depends on the signedness of an object.
1990 But many programs have been written to use plain @code{char} and
1991 expect it to be signed, or expect it to be unsigned, depending on the
1992 machines they were written for. This option, and its inverse, let you
1993 make such a program work with the opposite default.
1995 The type @code{char} is always a distinct type from each of
1996 @code{signed char} or @code{unsigned char}, even though its behavior
1997 is always just like one of those two.
2000 @opindex fsigned-char
2001 Let the type @code{char} be signed, like @code{signed char}.
2003 Note that this is equivalent to @option{-fno-unsigned-char}, which is
2004 the negative form of @option{-funsigned-char}. Likewise, the option
2005 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
2007 @item -fsigned-bitfields
2008 @itemx -funsigned-bitfields
2009 @itemx -fno-signed-bitfields
2010 @itemx -fno-unsigned-bitfields
2011 @opindex fsigned-bitfields
2012 @opindex funsigned-bitfields
2013 @opindex fno-signed-bitfields
2014 @opindex fno-unsigned-bitfields
2015 These options control whether a bit-field is signed or unsigned, when the
2016 declaration does not use either @code{signed} or @code{unsigned}. By
2017 default, such a bit-field is signed, because this is consistent: the
2018 basic integer types such as @code{int} are signed types.
2021 @node C++ Dialect Options
2022 @section Options Controlling C++ Dialect
2024 @cindex compiler options, C++
2025 @cindex C++ options, command-line
2026 @cindex options, C++
2027 This section describes the command-line options that are only meaningful
2028 for C++ programs. You can also use most of the GNU compiler options
2029 regardless of what language your program is in. For example, you
2030 might compile a file @code{firstClass.C} like this:
2033 g++ -g -frepo -O -c firstClass.C
2037 In this example, only @option{-frepo} is an option meant
2038 only for C++ programs; you can use the other options with any
2039 language supported by GCC@.
2041 Here is a list of options that are @emph{only} for compiling C++ programs:
2045 @item -fabi-version=@var{n}
2046 @opindex fabi-version
2047 Use version @var{n} of the C++ ABI@. The default is version 0.
2049 Version 0 refers to the version conforming most closely to
2050 the C++ ABI specification. Therefore, the ABI obtained using version 0
2051 will change in different versions of G++ as ABI bugs are fixed.
2053 Version 1 is the version of the C++ ABI that first appeared in G++ 3.2.
2055 Version 2 is the version of the C++ ABI that first appeared in G++
2056 3.4, and was the default through G++ 4.9.
2058 Version 3 corrects an error in mangling a constant address as a
2061 Version 4, which first appeared in G++ 4.5, implements a standard
2062 mangling for vector types.
2064 Version 5, which first appeared in G++ 4.6, corrects the mangling of
2065 attribute const/volatile on function pointer types, decltype of a
2066 plain decl, and use of a function parameter in the declaration of
2069 Version 6, which first appeared in G++ 4.7, corrects the promotion
2070 behavior of C++11 scoped enums and the mangling of template argument
2071 packs, const/static_cast, prefix ++ and --, and a class scope function
2072 used as a template argument.
2074 Version 7, which first appeared in G++ 4.8, that treats nullptr_t as a
2075 builtin type and corrects the mangling of lambdas in default argument
2078 Version 8, which first appeared in G++ 4.9, corrects the substitution
2079 behavior of function types with function-cv-qualifiers.
2081 See also @option{-Wabi}.
2083 @item -fabi-compat-version=@var{n}
2084 @opindex fabi-compat-version
2085 Starting with GCC 4.5, on targets that support strong aliases, G++
2086 works around mangling changes by creating an alias with the correct
2087 mangled name when defining a symbol with an incorrect mangled name.
2088 This switch specifies which ABI version to use for the alias.
2090 With @option{-fabi-version=0} (the default), this defaults to 2. If
2091 another ABI version is explicitly selected, this defaults to 0.
2093 The compatibility version is also set by @option{-Wabi=@var{n}}.
2095 @item -fno-access-control
2096 @opindex fno-access-control
2097 Turn off all access checking. This switch is mainly useful for working
2098 around bugs in the access control code.
2102 Check that the pointer returned by @code{operator new} is non-null
2103 before attempting to modify the storage allocated. This check is
2104 normally unnecessary because the C++ standard specifies that
2105 @code{operator new} only returns @code{0} if it is declared
2106 @samp{throw()}, in which case the compiler always checks the
2107 return value even without this option. In all other cases, when
2108 @code{operator new} has a non-empty exception specification, memory
2109 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
2110 @samp{new (nothrow)}.
2112 @item -fconstexpr-depth=@var{n}
2113 @opindex fconstexpr-depth
2114 Set the maximum nested evaluation depth for C++11 constexpr functions
2115 to @var{n}. A limit is needed to detect endless recursion during
2116 constant expression evaluation. The minimum specified by the standard
2119 @item -fdeduce-init-list
2120 @opindex fdeduce-init-list
2121 Enable deduction of a template type parameter as
2122 @code{std::initializer_list} from a brace-enclosed initializer list, i.e.@:
2125 template <class T> auto forward(T t) -> decltype (realfn (t))
2132 forward(@{1,2@}); // call forward<std::initializer_list<int>>
2136 This deduction was implemented as a possible extension to the
2137 originally proposed semantics for the C++11 standard, but was not part
2138 of the final standard, so it is disabled by default. This option is
2139 deprecated, and may be removed in a future version of G++.
2141 @item -ffriend-injection
2142 @opindex ffriend-injection
2143 Inject friend functions into the enclosing namespace, so that they are
2144 visible outside the scope of the class in which they are declared.
2145 Friend functions were documented to work this way in the old Annotated
2146 C++ Reference Manual, and versions of G++ before 4.1 always worked
2147 that way. However, in ISO C++ a friend function that is not declared
2148 in an enclosing scope can only be found using argument dependent
2149 lookup. This option causes friends to be injected as they were in
2152 This option is for compatibility, and may be removed in a future
2155 @item -fno-elide-constructors
2156 @opindex fno-elide-constructors
2157 The C++ standard allows an implementation to omit creating a temporary
2158 that is only used to initialize another object of the same type.
2159 Specifying this option disables that optimization, and forces G++ to
2160 call the copy constructor in all cases.
2162 @item -fno-enforce-eh-specs
2163 @opindex fno-enforce-eh-specs
2164 Don't generate code to check for violation of exception specifications
2165 at run time. This option violates the C++ standard, but may be useful
2166 for reducing code size in production builds, much like defining
2167 @samp{NDEBUG}. This does not give user code permission to throw
2168 exceptions in violation of the exception specifications; the compiler
2169 still optimizes based on the specifications, so throwing an
2170 unexpected exception results in undefined behavior at run time.
2172 @item -fextern-tls-init
2173 @itemx -fno-extern-tls-init
2174 @opindex fextern-tls-init
2175 @opindex fno-extern-tls-init
2176 The C++11 and OpenMP standards allow @samp{thread_local} and
2177 @samp{threadprivate} variables to have dynamic (runtime)
2178 initialization. To support this, any use of such a variable goes
2179 through a wrapper function that performs any necessary initialization.
2180 When the use and definition of the variable are in the same
2181 translation unit, this overhead can be optimized away, but when the
2182 use is in a different translation unit there is significant overhead
2183 even if the variable doesn't actually need dynamic initialization. If
2184 the programmer can be sure that no use of the variable in a
2185 non-defining TU needs to trigger dynamic initialization (either
2186 because the variable is statically initialized, or a use of the
2187 variable in the defining TU will be executed before any uses in
2188 another TU), they can avoid this overhead with the
2189 @option{-fno-extern-tls-init} option.
2191 On targets that support symbol aliases, the default is
2192 @option{-fextern-tls-init}. On targets that do not support symbol
2193 aliases, the default is @option{-fno-extern-tls-init}.
2196 @itemx -fno-for-scope
2198 @opindex fno-for-scope
2199 If @option{-ffor-scope} is specified, the scope of variables declared in
2200 a @i{for-init-statement} is limited to the @samp{for} loop itself,
2201 as specified by the C++ standard.
2202 If @option{-fno-for-scope} is specified, the scope of variables declared in
2203 a @i{for-init-statement} extends to the end of the enclosing scope,
2204 as was the case in old versions of G++, and other (traditional)
2205 implementations of C++.
2207 If neither flag is given, the default is to follow the standard,
2208 but to allow and give a warning for old-style code that would
2209 otherwise be invalid, or have different behavior.
2211 @item -fno-gnu-keywords
2212 @opindex fno-gnu-keywords
2213 Do not recognize @code{typeof} as a keyword, so that code can use this
2214 word as an identifier. You can use the keyword @code{__typeof__} instead.
2215 @option{-ansi} implies @option{-fno-gnu-keywords}.
2217 @item -fno-implicit-templates
2218 @opindex fno-implicit-templates
2219 Never emit code for non-inline templates that are instantiated
2220 implicitly (i.e.@: by use); only emit code for explicit instantiations.
2221 @xref{Template Instantiation}, for more information.
2223 @item -fno-implicit-inline-templates
2224 @opindex fno-implicit-inline-templates
2225 Don't emit code for implicit instantiations of inline templates, either.
2226 The default is to handle inlines differently so that compiles with and
2227 without optimization need the same set of explicit instantiations.
2229 @item -fno-implement-inlines
2230 @opindex fno-implement-inlines
2231 To save space, do not emit out-of-line copies of inline functions
2232 controlled by @samp{#pragma implementation}. This causes linker
2233 errors if these functions are not inlined everywhere they are called.
2235 @item -fms-extensions
2236 @opindex fms-extensions
2237 Disable Wpedantic warnings about constructs used in MFC, such as implicit
2238 int and getting a pointer to member function via non-standard syntax.
2240 @item -fno-nonansi-builtins
2241 @opindex fno-nonansi-builtins
2242 Disable built-in declarations of functions that are not mandated by
2243 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
2244 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
2247 @opindex fnothrow-opt
2248 Treat a @code{throw()} exception specification as if it were a
2249 @code{noexcept} specification to reduce or eliminate the text size
2250 overhead relative to a function with no exception specification. If
2251 the function has local variables of types with non-trivial
2252 destructors, the exception specification actually makes the
2253 function smaller because the EH cleanups for those variables can be
2254 optimized away. The semantic effect is that an exception thrown out of
2255 a function with such an exception specification results in a call
2256 to @code{terminate} rather than @code{unexpected}.
2258 @item -fno-operator-names
2259 @opindex fno-operator-names
2260 Do not treat the operator name keywords @code{and}, @code{bitand},
2261 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
2262 synonyms as keywords.
2264 @item -fno-optional-diags
2265 @opindex fno-optional-diags
2266 Disable diagnostics that the standard says a compiler does not need to
2267 issue. Currently, the only such diagnostic issued by G++ is the one for
2268 a name having multiple meanings within a class.
2271 @opindex fpermissive
2272 Downgrade some diagnostics about nonconformant code from errors to
2273 warnings. Thus, using @option{-fpermissive} allows some
2274 nonconforming code to compile.
2276 @item -fno-pretty-templates
2277 @opindex fno-pretty-templates
2278 When an error message refers to a specialization of a function
2279 template, the compiler normally prints the signature of the
2280 template followed by the template arguments and any typedefs or
2281 typenames in the signature (e.g. @code{void f(T) [with T = int]}
2282 rather than @code{void f(int)}) so that it's clear which template is
2283 involved. When an error message refers to a specialization of a class
2284 template, the compiler omits any template arguments that match
2285 the default template arguments for that template. If either of these
2286 behaviors make it harder to understand the error message rather than
2287 easier, you can use @option{-fno-pretty-templates} to disable them.
2291 Enable automatic template instantiation at link time. This option also
2292 implies @option{-fno-implicit-templates}. @xref{Template
2293 Instantiation}, for more information.
2297 Disable generation of information about every class with virtual
2298 functions for use by the C++ run-time type identification features
2299 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
2300 of the language, you can save some space by using this flag. Note that
2301 exception handling uses the same information, but G++ generates it as
2302 needed. The @samp{dynamic_cast} operator can still be used for casts that
2303 do not require run-time type information, i.e.@: casts to @code{void *} or to
2304 unambiguous base classes.
2308 Emit statistics about front-end processing at the end of the compilation.
2309 This information is generally only useful to the G++ development team.
2311 @item -fstrict-enums
2312 @opindex fstrict-enums
2313 Allow the compiler to optimize using the assumption that a value of
2314 enumerated type can only be one of the values of the enumeration (as
2315 defined in the C++ standard; basically, a value that can be
2316 represented in the minimum number of bits needed to represent all the
2317 enumerators). This assumption may not be valid if the program uses a
2318 cast to convert an arbitrary integer value to the enumerated type.
2320 @item -ftemplate-backtrace-limit=@var{n}
2321 @opindex ftemplate-backtrace-limit
2322 Set the maximum number of template instantiation notes for a single
2323 warning or error to @var{n}. The default value is 10.
2325 @item -ftemplate-depth=@var{n}
2326 @opindex ftemplate-depth
2327 Set the maximum instantiation depth for template classes to @var{n}.
2328 A limit on the template instantiation depth is needed to detect
2329 endless recursions during template class instantiation. ANSI/ISO C++
2330 conforming programs must not rely on a maximum depth greater than 17
2331 (changed to 1024 in C++11). The default value is 900, as the compiler
2332 can run out of stack space before hitting 1024 in some situations.
2334 @item -fno-threadsafe-statics
2335 @opindex fno-threadsafe-statics
2336 Do not emit the extra code to use the routines specified in the C++
2337 ABI for thread-safe initialization of local statics. You can use this
2338 option to reduce code size slightly in code that doesn't need to be
2341 @item -fuse-cxa-atexit
2342 @opindex fuse-cxa-atexit
2343 Register destructors for objects with static storage duration with the
2344 @code{__cxa_atexit} function rather than the @code{atexit} function.
2345 This option is required for fully standards-compliant handling of static
2346 destructors, but only works if your C library supports
2347 @code{__cxa_atexit}.
2349 @item -fno-use-cxa-get-exception-ptr
2350 @opindex fno-use-cxa-get-exception-ptr
2351 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2352 causes @code{std::uncaught_exception} to be incorrect, but is necessary
2353 if the runtime routine is not available.
2355 @item -fvisibility-inlines-hidden
2356 @opindex fvisibility-inlines-hidden
2357 This switch declares that the user does not attempt to compare
2358 pointers to inline functions or methods where the addresses of the two functions
2359 are taken in different shared objects.
2361 The effect of this is that GCC may, effectively, mark inline methods with
2362 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2363 appear in the export table of a DSO and do not require a PLT indirection
2364 when used within the DSO@. Enabling this option can have a dramatic effect
2365 on load and link times of a DSO as it massively reduces the size of the
2366 dynamic export table when the library makes heavy use of templates.
2368 The behavior of this switch is not quite the same as marking the
2369 methods as hidden directly, because it does not affect static variables
2370 local to the function or cause the compiler to deduce that
2371 the function is defined in only one shared object.
2373 You may mark a method as having a visibility explicitly to negate the
2374 effect of the switch for that method. For example, if you do want to
2375 compare pointers to a particular inline method, you might mark it as
2376 having default visibility. Marking the enclosing class with explicit
2377 visibility has no effect.
2379 Explicitly instantiated inline methods are unaffected by this option
2380 as their linkage might otherwise cross a shared library boundary.
2381 @xref{Template Instantiation}.
2383 @item -fvisibility-ms-compat
2384 @opindex fvisibility-ms-compat
2385 This flag attempts to use visibility settings to make GCC's C++
2386 linkage model compatible with that of Microsoft Visual Studio.
2388 The flag makes these changes to GCC's linkage model:
2392 It sets the default visibility to @code{hidden}, like
2393 @option{-fvisibility=hidden}.
2396 Types, but not their members, are not hidden by default.
2399 The One Definition Rule is relaxed for types without explicit
2400 visibility specifications that are defined in more than one
2401 shared object: those declarations are permitted if they are
2402 permitted when this option is not used.
2405 In new code it is better to use @option{-fvisibility=hidden} and
2406 export those classes that are intended to be externally visible.
2407 Unfortunately it is possible for code to rely, perhaps accidentally,
2408 on the Visual Studio behavior.
2410 Among the consequences of these changes are that static data members
2411 of the same type with the same name but defined in different shared
2412 objects are different, so changing one does not change the other;
2413 and that pointers to function members defined in different shared
2414 objects may not compare equal. When this flag is given, it is a
2415 violation of the ODR to define types with the same name differently.
2417 @item -fvtable-verify=@var{std|preinit|none}
2418 @opindex fvtable-verify
2419 Turn on (or off, if using @option{-fvtable-verify=none}) the security
2420 feature that verifies at runtime, for every virtual call that is made, that
2421 the vtable pointer through which the call is made is valid for the type of
2422 the object, and has not been corrupted or overwritten. If an invalid vtable
2423 pointer is detected (at runtime), an error is reported and execution of the
2424 program is immediately halted.
2426 This option causes runtime data structures to be built, at program start up,
2427 for verifying the vtable pointers. The options @code{std} and @code{preinit}
2428 control the timing of when these data structures are built. In both cases the
2429 data structures are built before execution reaches 'main'. The
2430 @option{-fvtable-verify=std} causes these data structure to be built after the
2431 shared libraries have been loaded and initialized.
2432 @option{-fvtable-verify=preinit} causes them to be built before the shared
2433 libraries have been loaded and initialized.
2435 If this option appears multiple times in the compiler line, with different
2436 values specified, 'none' will take highest priority over both 'std' and
2437 'preinit'; 'preinit' will take priority over 'std'.
2440 @opindex (fvtv-debug)
2441 Causes debug versions of the runtime functions for the vtable verification
2442 feature to be called. This assumes the @option{-fvtable-verify=std} or
2443 @option{-fvtable-verify=preinit} has been used. This flag will also cause the
2444 compiler to keep track of which vtable pointers it found for each class, and
2445 record that information in the file ``vtv_set_ptr_data.log'', in the dump
2446 file directory on the user's machine.
2448 Note: This feature APPENDS data to the log file. If you want a fresh log
2449 file, be sure to delete any existing one.
2452 @opindex fvtv-counts
2453 This is a debugging flag. When used in conjunction with
2454 @option{-fvtable-verify=std} or @option{-fvtable-verify=preinit}, this
2455 causes the compiler to keep track of the total number of virtual calls
2456 it encountered and the number of verifications it inserted. It also
2457 counts the number of calls to certain runtime library functions
2458 that it inserts. This information, for each compilation unit, is written
2459 to a file named ``vtv_count_data.log'', in the dump_file directory on
2460 the user's machine. It also counts the size of the vtable pointer sets
2461 for each class, and writes this information to ``vtv_class_set_sizes.log''
2462 in the same directory.
2464 Note: This feature APPENDS data to the log files. To get a fresh log
2465 files, be sure to delete any existing ones.
2469 Do not use weak symbol support, even if it is provided by the linker.
2470 By default, G++ uses weak symbols if they are available. This
2471 option exists only for testing, and should not be used by end-users;
2472 it results in inferior code and has no benefits. This option may
2473 be removed in a future release of G++.
2477 Do not search for header files in the standard directories specific to
2478 C++, but do still search the other standard directories. (This option
2479 is used when building the C++ library.)
2482 In addition, these optimization, warning, and code generation options
2483 have meanings only for C++ programs:
2486 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2489 When an explicit @option{-fabi-version=@var{n}} option is used, causes
2490 G++ to warn when it generates code that is probably not compatible with the
2491 vendor-neutral C++ ABI@. Since G++ now defaults to
2492 @option{-fabi-version=0}, @option{-Wabi} has no effect unless either
2493 an older ABI version is selected (with @option{-fabi-version=@var{n}})
2494 or an older compatibility version is selected (with
2495 @option{-Wabi=@var{n}} or @option{-fabi-compat-version=@var{n}}).
2497 Although an effort has been made to warn about
2498 all such cases, there are probably some cases that are not warned about,
2499 even though G++ is generating incompatible code. There may also be
2500 cases where warnings are emitted even though the code that is generated
2503 You should rewrite your code to avoid these warnings if you are
2504 concerned about the fact that code generated by G++ may not be binary
2505 compatible with code generated by other compilers.
2507 @option{-Wabi} can also be used with an explicit version number to
2508 warn about compatibility with a particular @option{-fabi-version}
2509 level, e.g. @option{-Wabi=2} to warn about changes relative to
2510 @option{-fabi-version=2}. Specifying a version number also sets
2511 @option{-fabi-compat-version=@var{n}}.
2513 The known incompatibilities in @option{-fabi-version=2} (which was the
2514 default from GCC 3.4 to 4.9) include:
2519 A template with a non-type template parameter of reference type was
2520 mangled incorrectly:
2523 template <int &> struct S @{@};
2527 This was fixed in @option{-fabi-version=3}.
2530 SIMD vector types declared using @code{__attribute ((vector_size))} were
2531 mangled in a non-standard way that does not allow for overloading of
2532 functions taking vectors of different sizes.
2534 The mangling was changed in @option{-fabi-version=4}.
2537 @code{__attribute ((const))} and @code{noreturn} were mangled as type
2538 qualifiers, and @code{decltype} of a plain declaration was folded away.
2540 These mangling issues were fixed in @option{-fabi-version=5}.
2543 Scoped enumerators passed as arguments to a variadic function are
2544 promoted like unscoped enumerators, causing @samp{va_arg} to complain.
2545 On most targets this does not actually affect the parameter passing
2546 ABI, as there is no way to pass an argument smaller than @samp{int}.
2548 Also, the ABI changed the mangling of template argument packs,
2549 @code{const_cast}, @code{static_cast}, prefix increment/decrement, and
2550 a class scope function used as a template argument.
2552 These issues were corrected in @option{-fabi-version=6}.
2555 Lambdas in default argument scope were mangled incorrectly, and the
2556 ABI changed the mangling of nullptr_t.
2558 These issues were corrected in @option{-fabi-version=7}.
2561 When mangling a function type with function-cv-qualifiers, the
2562 un-qualified function type was incorrectly treated as a substitution
2565 This was fixed in @option{-fabi-version=8}.
2568 It also warns about psABI-related changes. The known psABI changes at this
2574 For SysV/x86-64, unions with @code{long double} members are
2575 passed in memory as specified in psABI. For example:
2585 @code{union U} is always passed in memory.
2589 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2590 @opindex Wctor-dtor-privacy
2591 @opindex Wno-ctor-dtor-privacy
2592 Warn when a class seems unusable because all the constructors or
2593 destructors in that class are private, and it has neither friends nor
2594 public static member functions. Also warn if there are no non-private
2595 methods, and there's at least one private member function that isn't
2596 a constructor or destructor.
2598 @item -Wdelete-non-virtual-dtor @r{(C++ and Objective-C++ only)}
2599 @opindex Wdelete-non-virtual-dtor
2600 @opindex Wno-delete-non-virtual-dtor
2601 Warn when @samp{delete} is used to destroy an instance of a class that
2602 has virtual functions and non-virtual destructor. It is unsafe to delete
2603 an instance of a derived class through a pointer to a base class if the
2604 base class does not have a virtual destructor. This warning is enabled
2607 @item -Wliteral-suffix @r{(C++ and Objective-C++ only)}
2608 @opindex Wliteral-suffix
2609 @opindex Wno-literal-suffix
2610 Warn when a string or character literal is followed by a ud-suffix which does
2611 not begin with an underscore. As a conforming extension, GCC treats such
2612 suffixes as separate preprocessing tokens in order to maintain backwards
2613 compatibility with code that uses formatting macros from @code{<inttypes.h>}.
2617 #define __STDC_FORMAT_MACROS
2618 #include <inttypes.h>
2623 printf("My int64: %"PRId64"\n", i64);
2627 In this case, @code{PRId64} is treated as a separate preprocessing token.
2629 This warning is enabled by default.
2631 @item -Wnarrowing @r{(C++ and Objective-C++ only)}
2633 @opindex Wno-narrowing
2634 Warn when a narrowing conversion prohibited by C++11 occurs within
2638 int i = @{ 2.2 @}; // error: narrowing from double to int
2641 This flag is included in @option{-Wall} and @option{-Wc++11-compat}.
2643 With @option{-std=c++11}, @option{-Wno-narrowing} suppresses the diagnostic
2644 required by the standard. Note that this does not affect the meaning
2645 of well-formed code; narrowing conversions are still considered
2646 ill-formed in SFINAE context.
2648 @item -Wnoexcept @r{(C++ and Objective-C++ only)}
2650 @opindex Wno-noexcept
2651 Warn when a noexcept-expression evaluates to false because of a call
2652 to a function that does not have a non-throwing exception
2653 specification (i.e. @samp{throw()} or @samp{noexcept}) but is known by
2654 the compiler to never throw an exception.
2656 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2657 @opindex Wnon-virtual-dtor
2658 @opindex Wno-non-virtual-dtor
2659 Warn when a class has virtual functions and an accessible non-virtual
2660 destructor itself or in an accessible polymorphic base class, in which
2661 case it is possible but unsafe to delete an instance of a derived
2662 class through a pointer to the class itself or base class. This
2663 warning is automatically enabled if @option{-Weffc++} is specified.
2665 @item -Wreorder @r{(C++ and Objective-C++ only)}
2667 @opindex Wno-reorder
2668 @cindex reordering, warning
2669 @cindex warning for reordering of member initializers
2670 Warn when the order of member initializers given in the code does not
2671 match the order in which they must be executed. For instance:
2677 A(): j (0), i (1) @{ @}
2682 The compiler rearranges the member initializers for @samp{i}
2683 and @samp{j} to match the declaration order of the members, emitting
2684 a warning to that effect. This warning is enabled by @option{-Wall}.
2686 @item -fext-numeric-literals @r{(C++ and Objective-C++ only)}
2687 @opindex fext-numeric-literals
2688 @opindex fno-ext-numeric-literals
2689 Accept imaginary, fixed-point, or machine-defined
2690 literal number suffixes as GNU extensions.
2691 When this option is turned off these suffixes are treated
2692 as C++11 user-defined literal numeric suffixes.
2693 This is on by default for all pre-C++11 dialects and all GNU dialects:
2694 @option{-std=c++98}, @option{-std=gnu++98}, @option{-std=gnu++11},
2695 @option{-std=gnu++1y}.
2696 This option is off by default
2697 for ISO C++11 onwards (@option{-std=c++11}, ...).
2700 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2703 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2706 Warn about violations of the following style guidelines from Scott Meyers'
2707 @cite{Effective C++} series of books:
2711 Define a copy constructor and an assignment operator for classes
2712 with dynamically-allocated memory.
2715 Prefer initialization to assignment in constructors.
2718 Have @code{operator=} return a reference to @code{*this}.
2721 Don't try to return a reference when you must return an object.
2724 Distinguish between prefix and postfix forms of increment and
2725 decrement operators.
2728 Never overload @code{&&}, @code{||}, or @code{,}.
2732 This option also enables @option{-Wnon-virtual-dtor}, which is also
2733 one of the effective C++ recommendations. However, the check is
2734 extended to warn about the lack of virtual destructor in accessible
2735 non-polymorphic bases classes too.
2737 When selecting this option, be aware that the standard library
2738 headers do not obey all of these guidelines; use @samp{grep -v}
2739 to filter out those warnings.
2741 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2742 @opindex Wstrict-null-sentinel
2743 @opindex Wno-strict-null-sentinel
2744 Warn about the use of an uncasted @code{NULL} as sentinel. When
2745 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2746 to @code{__null}. Although it is a null pointer constant rather than a
2747 null pointer, it is guaranteed to be of the same size as a pointer.
2748 But this use is not portable across different compilers.
2750 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2751 @opindex Wno-non-template-friend
2752 @opindex Wnon-template-friend
2753 Disable warnings when non-templatized friend functions are declared
2754 within a template. Since the advent of explicit template specification
2755 support in G++, if the name of the friend is an unqualified-id (i.e.,
2756 @samp{friend foo(int)}), the C++ language specification demands that the
2757 friend declare or define an ordinary, nontemplate function. (Section
2758 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2759 could be interpreted as a particular specialization of a templatized
2760 function. Because this non-conforming behavior is no longer the default
2761 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2762 check existing code for potential trouble spots and is on by default.
2763 This new compiler behavior can be turned off with
2764 @option{-Wno-non-template-friend}, which keeps the conformant compiler code
2765 but disables the helpful warning.
2767 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2768 @opindex Wold-style-cast
2769 @opindex Wno-old-style-cast
2770 Warn if an old-style (C-style) cast to a non-void type is used within
2771 a C++ program. The new-style casts (@samp{dynamic_cast},
2772 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2773 less vulnerable to unintended effects and much easier to search for.
2775 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2776 @opindex Woverloaded-virtual
2777 @opindex Wno-overloaded-virtual
2778 @cindex overloaded virtual function, warning
2779 @cindex warning for overloaded virtual function
2780 Warn when a function declaration hides virtual functions from a
2781 base class. For example, in:
2788 struct B: public A @{
2793 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2804 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2805 @opindex Wno-pmf-conversions
2806 @opindex Wpmf-conversions
2807 Disable the diagnostic for converting a bound pointer to member function
2810 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2811 @opindex Wsign-promo
2812 @opindex Wno-sign-promo
2813 Warn when overload resolution chooses a promotion from unsigned or
2814 enumerated type to a signed type, over a conversion to an unsigned type of
2815 the same size. Previous versions of G++ tried to preserve
2816 unsignedness, but the standard mandates the current behavior.
2819 @node Objective-C and Objective-C++ Dialect Options
2820 @section Options Controlling Objective-C and Objective-C++ Dialects
2822 @cindex compiler options, Objective-C and Objective-C++
2823 @cindex Objective-C and Objective-C++ options, command-line
2824 @cindex options, Objective-C and Objective-C++
2825 (NOTE: This manual does not describe the Objective-C and Objective-C++
2826 languages themselves. @xref{Standards,,Language Standards
2827 Supported by GCC}, for references.)
2829 This section describes the command-line options that are only meaningful
2830 for Objective-C and Objective-C++ programs. You can also use most of
2831 the language-independent GNU compiler options.
2832 For example, you might compile a file @code{some_class.m} like this:
2835 gcc -g -fgnu-runtime -O -c some_class.m
2839 In this example, @option{-fgnu-runtime} is an option meant only for
2840 Objective-C and Objective-C++ programs; you can use the other options with
2841 any language supported by GCC@.
2843 Note that since Objective-C is an extension of the C language, Objective-C
2844 compilations may also use options specific to the C front-end (e.g.,
2845 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2846 C++-specific options (e.g., @option{-Wabi}).
2848 Here is a list of options that are @emph{only} for compiling Objective-C
2849 and Objective-C++ programs:
2852 @item -fconstant-string-class=@var{class-name}
2853 @opindex fconstant-string-class
2854 Use @var{class-name} as the name of the class to instantiate for each
2855 literal string specified with the syntax @code{@@"@dots{}"}. The default
2856 class name is @code{NXConstantString} if the GNU runtime is being used, and
2857 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2858 @option{-fconstant-cfstrings} option, if also present, overrides the
2859 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2860 to be laid out as constant CoreFoundation strings.
2863 @opindex fgnu-runtime
2864 Generate object code compatible with the standard GNU Objective-C
2865 runtime. This is the default for most types of systems.
2867 @item -fnext-runtime
2868 @opindex fnext-runtime
2869 Generate output compatible with the NeXT runtime. This is the default
2870 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2871 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2874 @item -fno-nil-receivers
2875 @opindex fno-nil-receivers
2876 Assume that all Objective-C message dispatches (@code{[receiver
2877 message:arg]}) in this translation unit ensure that the receiver is
2878 not @code{nil}. This allows for more efficient entry points in the
2879 runtime to be used. This option is only available in conjunction with
2880 the NeXT runtime and ABI version 0 or 1.
2882 @item -fobjc-abi-version=@var{n}
2883 @opindex fobjc-abi-version
2884 Use version @var{n} of the Objective-C ABI for the selected runtime.
2885 This option is currently supported only for the NeXT runtime. In that
2886 case, Version 0 is the traditional (32-bit) ABI without support for
2887 properties and other Objective-C 2.0 additions. Version 1 is the
2888 traditional (32-bit) ABI with support for properties and other
2889 Objective-C 2.0 additions. Version 2 is the modern (64-bit) ABI. If
2890 nothing is specified, the default is Version 0 on 32-bit target
2891 machines, and Version 2 on 64-bit target machines.
2893 @item -fobjc-call-cxx-cdtors
2894 @opindex fobjc-call-cxx-cdtors
2895 For each Objective-C class, check if any of its instance variables is a
2896 C++ object with a non-trivial default constructor. If so, synthesize a
2897 special @code{- (id) .cxx_construct} instance method which runs
2898 non-trivial default constructors on any such instance variables, in order,
2899 and then return @code{self}. Similarly, check if any instance variable
2900 is a C++ object with a non-trivial destructor, and if so, synthesize a
2901 special @code{- (void) .cxx_destruct} method which runs
2902 all such default destructors, in reverse order.
2904 The @code{- (id) .cxx_construct} and @code{- (void) .cxx_destruct}
2905 methods thusly generated only operate on instance variables
2906 declared in the current Objective-C class, and not those inherited
2907 from superclasses. It is the responsibility of the Objective-C
2908 runtime to invoke all such methods in an object's inheritance
2909 hierarchy. The @code{- (id) .cxx_construct} methods are invoked
2910 by the runtime immediately after a new object instance is allocated;
2911 the @code{- (void) .cxx_destruct} methods are invoked immediately
2912 before the runtime deallocates an object instance.
2914 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2915 support for invoking the @code{- (id) .cxx_construct} and
2916 @code{- (void) .cxx_destruct} methods.
2918 @item -fobjc-direct-dispatch
2919 @opindex fobjc-direct-dispatch
2920 Allow fast jumps to the message dispatcher. On Darwin this is
2921 accomplished via the comm page.
2923 @item -fobjc-exceptions
2924 @opindex fobjc-exceptions
2925 Enable syntactic support for structured exception handling in
2926 Objective-C, similar to what is offered by C++ and Java. This option
2927 is required to use the Objective-C keywords @code{@@try},
2928 @code{@@throw}, @code{@@catch}, @code{@@finally} and
2929 @code{@@synchronized}. This option is available with both the GNU
2930 runtime and the NeXT runtime (but not available in conjunction with
2931 the NeXT runtime on Mac OS X 10.2 and earlier).
2935 Enable garbage collection (GC) in Objective-C and Objective-C++
2936 programs. This option is only available with the NeXT runtime; the
2937 GNU runtime has a different garbage collection implementation that
2938 does not require special compiler flags.
2940 @item -fobjc-nilcheck
2941 @opindex fobjc-nilcheck
2942 For the NeXT runtime with version 2 of the ABI, check for a nil
2943 receiver in method invocations before doing the actual method call.
2944 This is the default and can be disabled using
2945 @option{-fno-objc-nilcheck}. Class methods and super calls are never
2946 checked for nil in this way no matter what this flag is set to.
2947 Currently this flag does nothing when the GNU runtime, or an older
2948 version of the NeXT runtime ABI, is used.
2950 @item -fobjc-std=objc1
2952 Conform to the language syntax of Objective-C 1.0, the language
2953 recognized by GCC 4.0. This only affects the Objective-C additions to
2954 the C/C++ language; it does not affect conformance to C/C++ standards,
2955 which is controlled by the separate C/C++ dialect option flags. When
2956 this option is used with the Objective-C or Objective-C++ compiler,
2957 any Objective-C syntax that is not recognized by GCC 4.0 is rejected.
2958 This is useful if you need to make sure that your Objective-C code can
2959 be compiled with older versions of GCC@.
2961 @item -freplace-objc-classes
2962 @opindex freplace-objc-classes
2963 Emit a special marker instructing @command{ld(1)} not to statically link in
2964 the resulting object file, and allow @command{dyld(1)} to load it in at
2965 run time instead. This is used in conjunction with the Fix-and-Continue
2966 debugging mode, where the object file in question may be recompiled and
2967 dynamically reloaded in the course of program execution, without the need
2968 to restart the program itself. Currently, Fix-and-Continue functionality
2969 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2974 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2975 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2976 compile time) with static class references that get initialized at load time,
2977 which improves run-time performance. Specifying the @option{-fzero-link} flag
2978 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2979 to be retained. This is useful in Zero-Link debugging mode, since it allows
2980 for individual class implementations to be modified during program execution.
2981 The GNU runtime currently always retains calls to @code{objc_get_class("@dots{}")}
2982 regardless of command-line options.
2984 @item -fno-local-ivars
2985 @opindex fno-local-ivars
2986 @opindex flocal-ivars
2987 By default instance variables in Objective-C can be accessed as if
2988 they were local variables from within the methods of the class they're
2989 declared in. This can lead to shadowing between instance variables
2990 and other variables declared either locally inside a class method or
2991 globally with the same name. Specifying the @option{-fno-local-ivars}
2992 flag disables this behavior thus avoiding variable shadowing issues.
2994 @item -fivar-visibility=@var{public|protected|private|package}
2995 @opindex fivar-visibility
2996 Set the default instance variable visibility to the specified option
2997 so that instance variables declared outside the scope of any access
2998 modifier directives default to the specified visibility.
3002 Dump interface declarations for all classes seen in the source file to a
3003 file named @file{@var{sourcename}.decl}.
3005 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
3006 @opindex Wassign-intercept
3007 @opindex Wno-assign-intercept
3008 Warn whenever an Objective-C assignment is being intercepted by the
3011 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
3012 @opindex Wno-protocol
3014 If a class is declared to implement a protocol, a warning is issued for
3015 every method in the protocol that is not implemented by the class. The
3016 default behavior is to issue a warning for every method not explicitly
3017 implemented in the class, even if a method implementation is inherited
3018 from the superclass. If you use the @option{-Wno-protocol} option, then
3019 methods inherited from the superclass are considered to be implemented,
3020 and no warning is issued for them.
3022 @item -Wselector @r{(Objective-C and Objective-C++ only)}
3024 @opindex Wno-selector
3025 Warn if multiple methods of different types for the same selector are
3026 found during compilation. The check is performed on the list of methods
3027 in the final stage of compilation. Additionally, a check is performed
3028 for each selector appearing in a @code{@@selector(@dots{})}
3029 expression, and a corresponding method for that selector has been found
3030 during compilation. Because these checks scan the method table only at
3031 the end of compilation, these warnings are not produced if the final
3032 stage of compilation is not reached, for example because an error is
3033 found during compilation, or because the @option{-fsyntax-only} option is
3036 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
3037 @opindex Wstrict-selector-match
3038 @opindex Wno-strict-selector-match
3039 Warn if multiple methods with differing argument and/or return types are
3040 found for a given selector when attempting to send a message using this
3041 selector to a receiver of type @code{id} or @code{Class}. When this flag
3042 is off (which is the default behavior), the compiler omits such warnings
3043 if any differences found are confined to types that share the same size
3046 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
3047 @opindex Wundeclared-selector
3048 @opindex Wno-undeclared-selector
3049 Warn if a @code{@@selector(@dots{})} expression referring to an
3050 undeclared selector is found. A selector is considered undeclared if no
3051 method with that name has been declared before the
3052 @code{@@selector(@dots{})} expression, either explicitly in an
3053 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
3054 an @code{@@implementation} section. This option always performs its
3055 checks as soon as a @code{@@selector(@dots{})} expression is found,
3056 while @option{-Wselector} only performs its checks in the final stage of
3057 compilation. This also enforces the coding style convention
3058 that methods and selectors must be declared before being used.
3060 @item -print-objc-runtime-info
3061 @opindex print-objc-runtime-info
3062 Generate C header describing the largest structure that is passed by
3067 @node Language Independent Options
3068 @section Options to Control Diagnostic Messages Formatting
3069 @cindex options to control diagnostics formatting
3070 @cindex diagnostic messages
3071 @cindex message formatting
3073 Traditionally, diagnostic messages have been formatted irrespective of
3074 the output device's aspect (e.g.@: its width, @dots{}). You can use the
3075 options described below
3076 to control the formatting algorithm for diagnostic messages,
3077 e.g.@: how many characters per line, how often source location
3078 information should be reported. Note that some language front ends may not
3079 honor these options.
3082 @item -fmessage-length=@var{n}
3083 @opindex fmessage-length
3084 Try to format error messages so that they fit on lines of about @var{n}
3085 characters. The default is 72 characters for @command{g++} and 0 for the rest of
3086 the front ends supported by GCC@. If @var{n} is zero, then no
3087 line-wrapping is done; each error message appears on a single
3090 @item -fdiagnostics-show-location=once
3091 @opindex fdiagnostics-show-location
3092 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
3093 reporter to emit source location information @emph{once}; that is, in
3094 case the message is too long to fit on a single physical line and has to
3095 be wrapped, the source location won't be emitted (as prefix) again,
3096 over and over, in subsequent continuation lines. This is the default
3099 @item -fdiagnostics-show-location=every-line
3100 Only meaningful in line-wrapping mode. Instructs the diagnostic
3101 messages reporter to emit the same source location information (as
3102 prefix) for physical lines that result from the process of breaking
3103 a message which is too long to fit on a single line.
3105 @item -fdiagnostics-color[=@var{WHEN}]
3106 @itemx -fno-diagnostics-color
3107 @opindex fdiagnostics-color
3108 @cindex highlight, color, colour
3109 @vindex GCC_COLORS @r{environment variable}
3110 Use color in diagnostics. @var{WHEN} is @samp{never}, @samp{always},
3111 or @samp{auto}. The default is @samp{never} if @env{GCC_COLORS} environment
3112 variable isn't present in the environment, and @samp{auto} otherwise.
3113 @samp{auto} means to use color only when the standard error is a terminal.
3114 The forms @option{-fdiagnostics-color} and @option{-fno-diagnostics-color} are
3115 aliases for @option{-fdiagnostics-color=always} and
3116 @option{-fdiagnostics-color=never}, respectively.
3118 The colors are defined by the environment variable @env{GCC_COLORS}.
3119 Its value is a colon-separated list of capabilities and Select Graphic
3120 Rendition (SGR) substrings. SGR commands are interpreted by the
3121 terminal or terminal emulator. (See the section in the documentation
3122 of your text terminal for permitted values and their meanings as
3123 character attributes.) These substring values are integers in decimal
3124 representation and can be concatenated with semicolons.
3125 Common values to concatenate include
3127 @samp{4} for underline,
3129 @samp{7} for inverse,
3130 @samp{39} for default foreground color,
3131 @samp{30} to @samp{37} for foreground colors,
3132 @samp{90} to @samp{97} for 16-color mode foreground colors,
3133 @samp{38;5;0} to @samp{38;5;255}
3134 for 88-color and 256-color modes foreground colors,
3135 @samp{49} for default background color,
3136 @samp{40} to @samp{47} for background colors,
3137 @samp{100} to @samp{107} for 16-color mode background colors,
3138 and @samp{48;5;0} to @samp{48;5;255}
3139 for 88-color and 256-color modes background colors.
3141 The default @env{GCC_COLORS} is
3142 @samp{error=01;31:warning=01;35:note=01;36:caret=01;32:locus=01:quote=01}
3143 where @samp{01;31} is bold red, @samp{01;35} is bold magenta,
3144 @samp{01;36} is bold cyan, @samp{01;32} is bold green and
3145 @samp{01} is bold. Setting @env{GCC_COLORS} to the empty
3146 string disables colors.
3147 Supported capabilities are as follows.
3151 @vindex error GCC_COLORS @r{capability}
3152 SGR substring for error: markers.
3155 @vindex warning GCC_COLORS @r{capability}
3156 SGR substring for warning: markers.
3159 @vindex note GCC_COLORS @r{capability}
3160 SGR substring for note: markers.
3163 @vindex caret GCC_COLORS @r{capability}
3164 SGR substring for caret line.
3167 @vindex locus GCC_COLORS @r{capability}
3168 SGR substring for location information, @samp{file:line} or
3169 @samp{file:line:column} etc.
3172 @vindex quote GCC_COLORS @r{capability}
3173 SGR substring for information printed within quotes.
3176 @item -fno-diagnostics-show-option
3177 @opindex fno-diagnostics-show-option
3178 @opindex fdiagnostics-show-option
3179 By default, each diagnostic emitted includes text indicating the
3180 command-line option that directly controls the diagnostic (if such an
3181 option is known to the diagnostic machinery). Specifying the
3182 @option{-fno-diagnostics-show-option} flag suppresses that behavior.
3184 @item -fno-diagnostics-show-caret
3185 @opindex fno-diagnostics-show-caret
3186 @opindex fdiagnostics-show-caret
3187 By default, each diagnostic emitted includes the original source line
3188 and a caret '^' indicating the column. This option suppresses this
3193 @node Warning Options
3194 @section Options to Request or Suppress Warnings
3195 @cindex options to control warnings
3196 @cindex warning messages
3197 @cindex messages, warning
3198 @cindex suppressing warnings
3200 Warnings are diagnostic messages that report constructions that
3201 are not inherently erroneous but that are risky or suggest there
3202 may have been an error.
3204 The following language-independent options do not enable specific
3205 warnings but control the kinds of diagnostics produced by GCC@.
3208 @cindex syntax checking
3210 @opindex fsyntax-only
3211 Check the code for syntax errors, but don't do anything beyond that.
3213 @item -fmax-errors=@var{n}
3214 @opindex fmax-errors
3215 Limits the maximum number of error messages to @var{n}, at which point
3216 GCC bails out rather than attempting to continue processing the source
3217 code. If @var{n} is 0 (the default), there is no limit on the number
3218 of error messages produced. If @option{-Wfatal-errors} is also
3219 specified, then @option{-Wfatal-errors} takes precedence over this
3224 Inhibit all warning messages.
3229 Make all warnings into errors.
3234 Make the specified warning into an error. The specifier for a warning
3235 is appended; for example @option{-Werror=switch} turns the warnings
3236 controlled by @option{-Wswitch} into errors. This switch takes a
3237 negative form, to be used to negate @option{-Werror} for specific
3238 warnings; for example @option{-Wno-error=switch} makes
3239 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3242 The warning message for each controllable warning includes the
3243 option that controls the warning. That option can then be used with
3244 @option{-Werror=} and @option{-Wno-error=} as described above.
3245 (Printing of the option in the warning message can be disabled using the
3246 @option{-fno-diagnostics-show-option} flag.)
3248 Note that specifying @option{-Werror=}@var{foo} automatically implies
3249 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3252 @item -Wfatal-errors
3253 @opindex Wfatal-errors
3254 @opindex Wno-fatal-errors
3255 This option causes the compiler to abort compilation on the first error
3256 occurred rather than trying to keep going and printing further error
3259 @item -Wforce-warnings
3260 @opindex Wforce-warnings
3261 This option causes the compiler to ignore @option{-Werror} and
3262 @option{-Werror=} and suppress promotion of warnings to errors.
3266 You can request many specific warnings with options beginning with
3267 @samp{-W}, for example @option{-Wimplicit} to request warnings on
3268 implicit declarations. Each of these specific warning options also
3269 has a negative form beginning @samp{-Wno-} to turn off warnings; for
3270 example, @option{-Wno-implicit}. This manual lists only one of the
3271 two forms, whichever is not the default. For further
3272 language-specific options also refer to @ref{C++ Dialect Options} and
3273 @ref{Objective-C and Objective-C++ Dialect Options}.
3275 When an unrecognized warning option is requested (e.g.,
3276 @option{-Wunknown-warning}), GCC emits a diagnostic stating
3277 that the option is not recognized. However, if the @option{-Wno-} form
3278 is used, the behavior is slightly different: no diagnostic is
3279 produced for @option{-Wno-unknown-warning} unless other diagnostics
3280 are being produced. This allows the use of new @option{-Wno-} options
3281 with old compilers, but if something goes wrong, the compiler
3282 warns that an unrecognized option is present.
3289 Issue all the warnings demanded by strict ISO C and ISO C++;
3290 reject all programs that use forbidden extensions, and some other
3291 programs that do not follow ISO C and ISO C++. For ISO C, follows the
3292 version of the ISO C standard specified by any @option{-std} option used.
3294 Valid ISO C and ISO C++ programs should compile properly with or without
3295 this option (though a rare few require @option{-ansi} or a
3296 @option{-std} option specifying the required version of ISO C)@. However,
3297 without this option, certain GNU extensions and traditional C and C++
3298 features are supported as well. With this option, they are rejected.
3300 @option{-Wpedantic} does not cause warning messages for use of the
3301 alternate keywords whose names begin and end with @samp{__}. Pedantic
3302 warnings are also disabled in the expression that follows
3303 @code{__extension__}. However, only system header files should use
3304 these escape routes; application programs should avoid them.
3305 @xref{Alternate Keywords}.
3307 Some users try to use @option{-Wpedantic} to check programs for strict ISO
3308 C conformance. They soon find that it does not do quite what they want:
3309 it finds some non-ISO practices, but not all---only those for which
3310 ISO C @emph{requires} a diagnostic, and some others for which
3311 diagnostics have been added.
3313 A feature to report any failure to conform to ISO C might be useful in
3314 some instances, but would require considerable additional work and would
3315 be quite different from @option{-Wpedantic}. We don't have plans to
3316 support such a feature in the near future.
3318 Where the standard specified with @option{-std} represents a GNU
3319 extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
3320 corresponding @dfn{base standard}, the version of ISO C on which the GNU
3321 extended dialect is based. Warnings from @option{-Wpedantic} are given
3322 where they are required by the base standard. (It does not make sense
3323 for such warnings to be given only for features not in the specified GNU
3324 C dialect, since by definition the GNU dialects of C include all
3325 features the compiler supports with the given option, and there would be
3326 nothing to warn about.)
3328 @item -pedantic-errors
3329 @opindex pedantic-errors
3330 Like @option{-Wpedantic}, except that errors are produced rather than
3336 This enables all the warnings about constructions that some users
3337 consider questionable, and that are easy to avoid (or modify to
3338 prevent the warning), even in conjunction with macros. This also
3339 enables some language-specific warnings described in @ref{C++ Dialect
3340 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
3342 @option{-Wall} turns on the following warning flags:
3344 @gccoptlist{-Waddress @gol
3345 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
3347 -Wchar-subscripts @gol
3348 -Wenum-compare @r{(in C/ObjC; this is on by default in C++)} @gol
3349 -Wimplicit-int @r{(C and Objective-C only)} @gol
3350 -Wimplicit-function-declaration @r{(C and Objective-C only)} @gol
3353 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
3354 -Wmaybe-uninitialized @gol
3355 -Wmissing-braces @r{(only for C/ObjC)} @gol
3362 -Wripa-opt-mismatch @gol
3363 -Wsequence-point @gol
3364 -Wsign-compare @r{(only in C++)} @gol
3365 -Wstrict-aliasing @gol
3366 -Wstrict-overflow=1 @gol
3369 -Wuninitialized @gol
3370 -Wunknown-pragmas @gol
3371 -Wunused-function @gol
3374 -Wunused-variable @gol
3375 -Wvolatile-register-var @gol
3378 Note that some warning flags are not implied by @option{-Wall}. Some of
3379 them warn about constructions that users generally do not consider
3380 questionable, but which occasionally you might wish to check for;
3381 others warn about constructions that are necessary or hard to avoid in
3382 some cases, and there is no simple way to modify the code to suppress
3383 the warning. Some of them are enabled by @option{-Wextra} but many of
3384 them must be enabled individually.
3390 This enables some extra warning flags that are not enabled by
3391 @option{-Wall}. (This option used to be called @option{-W}. The older
3392 name is still supported, but the newer name is more descriptive.)
3394 @gccoptlist{-Wclobbered @gol
3396 -Wignored-qualifiers @gol
3397 -Wmissing-field-initializers @gol
3398 -Wmissing-parameter-type @r{(C only)} @gol
3399 -Wold-style-declaration @r{(C only)} @gol
3400 -Woverride-init @gol
3403 -Wuninitialized @gol
3404 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
3405 -Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
3408 The option @option{-Wextra} also prints warning messages for the
3414 A pointer is compared against integer zero with @samp{<}, @samp{<=},
3415 @samp{>}, or @samp{>=}.
3418 (C++ only) An enumerator and a non-enumerator both appear in a
3419 conditional expression.
3422 (C++ only) Ambiguous virtual bases.
3425 (C++ only) Subscripting an array that has been declared @samp{register}.
3428 (C++ only) Taking the address of a variable that has been declared
3432 (C++ only) A base class is not initialized in a derived class's copy
3437 @item -Wchar-subscripts
3438 @opindex Wchar-subscripts
3439 @opindex Wno-char-subscripts
3440 Warn if an array subscript has type @code{char}. This is a common cause
3441 of error, as programmers often forget that this type is signed on some
3443 This warning is enabled by @option{-Wall}.
3447 @opindex Wno-comment
3448 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
3449 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
3450 This warning is enabled by @option{-Wall}.
3452 @item -Wno-coverage-mismatch
3453 @opindex Wno-coverage-mismatch
3454 Warn if feedback profiles do not match when using the
3455 @option{-fprofile-use} option.
3456 If a source file is changed between compiling with @option{-fprofile-gen} and
3457 with @option{-fprofile-use}, the files with the profile feedback can fail
3458 to match the source file and GCC cannot use the profile feedback
3459 information. By default, this warning is enabled and is treated as an
3460 error. @option{-Wno-coverage-mismatch} can be used to disable the
3461 warning or @option{-Wno-error=coverage-mismatch} can be used to
3462 disable the error. Disabling the error for this warning can result in
3463 poorly optimized code and is useful only in the
3464 case of very minor changes such as bug fixes to an existing code-base.
3465 Completely disabling the warning is not recommended.
3468 @r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
3470 Suppress warning messages emitted by @code{#warning} directives.
3472 @item -Wdouble-promotion @r{(C, C++, Objective-C and Objective-C++ only)}
3473 @opindex Wdouble-promotion
3474 @opindex Wno-double-promotion
3475 Give a warning when a value of type @code{float} is implicitly
3476 promoted to @code{double}. CPUs with a 32-bit ``single-precision''
3477 floating-point unit implement @code{float} in hardware, but emulate
3478 @code{double} in software. On such a machine, doing computations
3479 using @code{double} values is much more expensive because of the
3480 overhead required for software emulation.
3482 It is easy to accidentally do computations with @code{double} because
3483 floating-point literals are implicitly of type @code{double}. For
3487 float area(float radius)
3489 return 3.14159 * radius * radius;
3493 the compiler performs the entire computation with @code{double}
3494 because the floating-point literal is a @code{double}.
3497 @itemx -Wformat=@var{n}
3500 @opindex ffreestanding
3501 @opindex fno-builtin
3503 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
3504 the arguments supplied have types appropriate to the format string
3505 specified, and that the conversions specified in the format string make
3506 sense. This includes standard functions, and others specified by format
3507 attributes (@pxref{Function Attributes}), in the @code{printf},
3508 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
3509 not in the C standard) families (or other target-specific families).
3510 Which functions are checked without format attributes having been
3511 specified depends on the standard version selected, and such checks of
3512 functions without the attribute specified are disabled by
3513 @option{-ffreestanding} or @option{-fno-builtin}.
3515 The formats are checked against the format features supported by GNU
3516 libc version 2.2. These include all ISO C90 and C99 features, as well
3517 as features from the Single Unix Specification and some BSD and GNU
3518 extensions. Other library implementations may not support all these
3519 features; GCC does not support warning about features that go beyond a
3520 particular library's limitations. However, if @option{-Wpedantic} is used
3521 with @option{-Wformat}, warnings are given about format features not
3522 in the selected standard version (but not for @code{strfmon} formats,
3523 since those are not in any version of the C standard). @xref{C Dialect
3524 Options,,Options Controlling C Dialect}.
3531 Option @option{-Wformat} is equivalent to @option{-Wformat=1}, and
3532 @option{-Wno-format} is equivalent to @option{-Wformat=0}. Since
3533 @option{-Wformat} also checks for null format arguments for several
3534 functions, @option{-Wformat} also implies @option{-Wnonnull}. Some
3535 aspects of this level of format checking can be disabled by the
3536 options: @option{-Wno-format-contains-nul},
3537 @option{-Wno-format-extra-args}, and @option{-Wno-format-zero-length}.
3538 @option{-Wformat} is enabled by @option{-Wall}.
3540 @item -Wno-format-contains-nul
3541 @opindex Wno-format-contains-nul
3542 @opindex Wformat-contains-nul
3543 If @option{-Wformat} is specified, do not warn about format strings that
3546 @item -Wno-format-extra-args
3547 @opindex Wno-format-extra-args
3548 @opindex Wformat-extra-args
3549 If @option{-Wformat} is specified, do not warn about excess arguments to a
3550 @code{printf} or @code{scanf} format function. The C standard specifies
3551 that such arguments are ignored.
3553 Where the unused arguments lie between used arguments that are
3554 specified with @samp{$} operand number specifications, normally
3555 warnings are still given, since the implementation could not know what
3556 type to pass to @code{va_arg} to skip the unused arguments. However,
3557 in the case of @code{scanf} formats, this option suppresses the
3558 warning if the unused arguments are all pointers, since the Single
3559 Unix Specification says that such unused arguments are allowed.
3561 @item -Wno-format-zero-length
3562 @opindex Wno-format-zero-length
3563 @opindex Wformat-zero-length
3564 If @option{-Wformat} is specified, do not warn about zero-length formats.
3565 The C standard specifies that zero-length formats are allowed.
3570 Enable @option{-Wformat} plus additional format checks. Currently
3571 equivalent to @option{-Wformat -Wformat-nonliteral -Wformat-security
3572 -Wformat-signedness -Wformat-y2k}.
3574 @item -Wformat-nonliteral
3575 @opindex Wformat-nonliteral
3576 @opindex Wno-format-nonliteral
3577 If @option{-Wformat} is specified, also warn if the format string is not a
3578 string literal and so cannot be checked, unless the format function
3579 takes its format arguments as a @code{va_list}.
3581 @item -Wformat-security
3582 @opindex Wformat-security
3583 @opindex Wno-format-security
3584 If @option{-Wformat} is specified, also warn about uses of format
3585 functions that represent possible security problems. At present, this
3586 warns about calls to @code{printf} and @code{scanf} functions where the
3587 format string is not a string literal and there are no format arguments,
3588 as in @code{printf (foo);}. This may be a security hole if the format
3589 string came from untrusted input and contains @samp{%n}. (This is
3590 currently a subset of what @option{-Wformat-nonliteral} warns about, but
3591 in future warnings may be added to @option{-Wformat-security} that are not
3592 included in @option{-Wformat-nonliteral}.)
3594 @item -Wformat-signedness
3595 @opindex Wformat-signedness
3596 @opindex Wno-format-signedness
3597 If @option{-Wformat} is specified, also warn if the format string
3598 requires an unsigned argument and the argument is signed and vice versa.
3601 @opindex Wformat-y2k
3602 @opindex Wno-format-y2k
3603 If @option{-Wformat} is specified, also warn about @code{strftime}
3604 formats that may yield only a two-digit year.
3607 @item -Wnonnull @r{(C and Objective-C only)}
3609 @opindex Wno-nonnull
3610 Warn about passing a null pointer for arguments marked as
3611 requiring a non-null value by the @code{nonnull} function attribute.
3613 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3614 can be disabled with the @option{-Wno-nonnull} option.
3616 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3618 @opindex Wno-init-self
3619 Warn about uninitialized variables that are initialized with themselves.
3620 Note this option can only be used with the @option{-Wuninitialized} option.
3622 For example, GCC warns about @code{i} being uninitialized in the
3623 following snippet only when @option{-Winit-self} has been specified:
3634 This warning is enabled by @option{-Wall} in C++.
3636 @item -Wimplicit-int @r{(C and Objective-C only)}
3637 @opindex Wimplicit-int
3638 @opindex Wno-implicit-int
3639 Warn when a declaration does not specify a type.
3640 This warning is enabled by @option{-Wall}.
3642 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3643 @opindex Wimplicit-function-declaration
3644 @opindex Wno-implicit-function-declaration
3645 Give a warning whenever a function is used before being declared. In
3646 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3647 enabled by default and it is made into an error by
3648 @option{-pedantic-errors}. This warning is also enabled by
3651 @item -Wimplicit @r{(C and Objective-C only)}
3653 @opindex Wno-implicit
3654 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3655 This warning is enabled by @option{-Wall}.
3657 @item -Wignored-qualifiers @r{(C and C++ only)}
3658 @opindex Wignored-qualifiers
3659 @opindex Wno-ignored-qualifiers
3660 Warn if the return type of a function has a type qualifier
3661 such as @code{const}. For ISO C such a type qualifier has no effect,
3662 since the value returned by a function is not an lvalue.
3663 For C++, the warning is only emitted for scalar types or @code{void}.
3664 ISO C prohibits qualified @code{void} return types on function
3665 definitions, so such return types always receive a warning
3666 even without this option.
3668 This warning is also enabled by @option{-Wextra}.
3673 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3674 a function with external linkage, returning int, taking either zero
3675 arguments, two, or three arguments of appropriate types. This warning
3676 is enabled by default in C++ and is enabled by either @option{-Wall}
3677 or @option{-Wpedantic}.
3679 @item -Wmissing-braces
3680 @opindex Wmissing-braces
3681 @opindex Wno-missing-braces
3682 Warn if an aggregate or union initializer is not fully bracketed. In
3683 the following example, the initializer for @samp{a} is not fully
3684 bracketed, but that for @samp{b} is fully bracketed. This warning is
3685 enabled by @option{-Wall} in C.
3688 int a[2][2] = @{ 0, 1, 2, 3 @};
3689 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3692 This warning is enabled by @option{-Wall}.
3694 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3695 @opindex Wmissing-include-dirs
3696 @opindex Wno-missing-include-dirs
3697 Warn if a user-supplied include directory does not exist.
3700 @opindex Wparentheses
3701 @opindex Wno-parentheses
3702 Warn if parentheses are omitted in certain contexts, such
3703 as when there is an assignment in a context where a truth value
3704 is expected, or when operators are nested whose precedence people
3705 often get confused about.
3707 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3708 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3709 interpretation from that of ordinary mathematical notation.
3711 Also warn about constructions where there may be confusion to which
3712 @code{if} statement an @code{else} branch belongs. Here is an example of
3727 In C/C++, every @code{else} branch belongs to the innermost possible
3728 @code{if} statement, which in this example is @code{if (b)}. This is
3729 often not what the programmer expected, as illustrated in the above
3730 example by indentation the programmer chose. When there is the
3731 potential for this confusion, GCC issues a warning when this flag
3732 is specified. To eliminate the warning, add explicit braces around
3733 the innermost @code{if} statement so there is no way the @code{else}
3734 can belong to the enclosing @code{if}. The resulting code
3751 Also warn for dangerous uses of the GNU extension to
3752 @code{?:} with omitted middle operand. When the condition
3753 in the @code{?}: operator is a boolean expression, the omitted value is
3754 always 1. Often programmers expect it to be a value computed
3755 inside the conditional expression instead.
3757 This warning is enabled by @option{-Wall}.
3759 @item -Wsequence-point
3760 @opindex Wsequence-point
3761 @opindex Wno-sequence-point
3762 Warn about code that may have undefined semantics because of violations
3763 of sequence point rules in the C and C++ standards.
3765 The C and C++ standards define the order in which expressions in a C/C++
3766 program are evaluated in terms of @dfn{sequence points}, which represent
3767 a partial ordering between the execution of parts of the program: those
3768 executed before the sequence point, and those executed after it. These
3769 occur after the evaluation of a full expression (one which is not part
3770 of a larger expression), after the evaluation of the first operand of a
3771 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3772 function is called (but after the evaluation of its arguments and the
3773 expression denoting the called function), and in certain other places.
3774 Other than as expressed by the sequence point rules, the order of
3775 evaluation of subexpressions of an expression is not specified. All
3776 these rules describe only a partial order rather than a total order,
3777 since, for example, if two functions are called within one expression
3778 with no sequence point between them, the order in which the functions
3779 are called is not specified. However, the standards committee have
3780 ruled that function calls do not overlap.
3782 It is not specified when between sequence points modifications to the
3783 values of objects take effect. Programs whose behavior depends on this
3784 have undefined behavior; the C and C++ standards specify that ``Between
3785 the previous and next sequence point an object shall have its stored
3786 value modified at most once by the evaluation of an expression.
3787 Furthermore, the prior value shall be read only to determine the value
3788 to be stored.''. If a program breaks these rules, the results on any
3789 particular implementation are entirely unpredictable.
3791 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3792 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3793 diagnosed by this option, and it may give an occasional false positive
3794 result, but in general it has been found fairly effective at detecting
3795 this sort of problem in programs.
3797 The standard is worded confusingly, therefore there is some debate
3798 over the precise meaning of the sequence point rules in subtle cases.
3799 Links to discussions of the problem, including proposed formal
3800 definitions, may be found on the GCC readings page, at
3801 @uref{http://gcc.gnu.org/@/readings.html}.
3803 This warning is enabled by @option{-Wall} for C and C++.
3806 @opindex Wself-assign
3807 @opindex Wno-self-assign
3808 Warn about self-assignment and self-initialization. This warning is intended
3809 for detecting accidental self-assignment due to typos, and therefore does
3810 not warn on a statement that is semantically a self-assignment after
3811 constant folding. Here is an example of what will trigger a self-assign
3812 warning and what will not:
3819 int x = x; /* warn */
3823 i = i + 0; /* not warn */
3824 f = f / 1; /* not warn */
3825 a[1] = a[1]; /* warn */
3826 i += 0; /* not warn */
3831 In C++ it will not warn on self-assignment of non-POD variables unless
3832 @option{-Wself-assign-non-pod} is also enabled.
3834 @item -Wself-assign-non-pod
3835 @opindex Wself-assign-non-pod
3836 @opindex Wno-self-assign-non-pod
3837 Warn about self-assignment of non-POD variables. This is a C++-specific
3838 warning and only effective when @option{-Wself-assign} is enabled.
3840 There are cases where self-assignment might be intentional. For example,
3841 a C++ programmer might write code to test whether an overloaded
3842 @code{operator=} works when the same object is assigned to itself.
3843 One way to work around the self-assign warning in such cases when this flag
3844 is enabled is using the functional form @code{object.operator=(object)}
3845 instead of the assignment form @code{object = object}, as shown in the
3854 t.operator=(t); // not warn
3860 @item -Wno-return-local-addr
3861 @opindex Wno-return-local-addr
3862 @opindex Wreturn-local-addr
3863 Do not warn about returning a pointer (or in C++, a reference) to a
3864 variable that goes out of scope after the function returns.
3867 @opindex Wreturn-type
3868 @opindex Wno-return-type
3869 Warn whenever a function is defined with a return type that defaults
3870 to @code{int}. Also warn about any @code{return} statement with no
3871 return value in a function whose return type is not @code{void}
3872 (falling off the end of the function body is considered returning
3873 without a value), and about a @code{return} statement with an
3874 expression in a function whose return type is @code{void}.
3876 For C++, a function without return type always produces a diagnostic
3877 message, even when @option{-Wno-return-type} is specified. The only
3878 exceptions are @samp{main} and functions defined in system headers.
3880 This warning is enabled by @option{-Wall}.
3882 @item -Wripa-opt-mismatch
3883 @opindex Wripa-opt-mismatch
3884 @opindex Wno-ripa-opt-mismatch
3885 When doing an FDO build with @option{-fprofile-use} and @option{-fripa},
3886 warn if importing an axuiliary module that was built with a different
3887 GCC command line during the profile-generate phase than the primary
3890 This warning is enabled by @option{-Wall}.
3895 Warn whenever a @code{switch} statement has an index of enumerated type
3896 and lacks a @code{case} for one or more of the named codes of that
3897 enumeration. (The presence of a @code{default} label prevents this
3898 warning.) @code{case} labels outside the enumeration range also
3899 provoke warnings when this option is used (even if there is a
3900 @code{default} label).
3901 This warning is enabled by @option{-Wall}.
3903 @item -Wswitch-default
3904 @opindex Wswitch-default
3905 @opindex Wno-switch-default
3906 Warn whenever a @code{switch} statement does not have a @code{default}
3910 @opindex Wswitch-enum
3911 @opindex Wno-switch-enum
3912 Warn whenever a @code{switch} statement has an index of enumerated type
3913 and lacks a @code{case} for one or more of the named codes of that
3914 enumeration. @code{case} labels outside the enumeration range also
3915 provoke warnings when this option is used. The only difference
3916 between @option{-Wswitch} and this option is that this option gives a
3917 warning about an omitted enumeration code even if there is a
3918 @code{default} label.
3921 @opindex Wswitch-bool
3922 @opindex Wno-switch-bool
3923 Warn whenever a @code{switch} statement has an index of boolean type.
3924 It is possible to suppress this warning by casting the controlling
3925 expression to a type other than @code{bool}. For example:
3928 switch ((int) (a == 4))
3934 This warning is enabled by default for C and C++ programs.
3936 @item -Wsync-nand @r{(C and C++ only)}
3938 @opindex Wno-sync-nand
3939 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3940 built-in functions are used. These functions changed semantics in GCC 4.4.
3944 @opindex Wno-trigraphs
3945 Warn if any trigraphs are encountered that might change the meaning of
3946 the program (trigraphs within comments are not warned about).
3947 This warning is enabled by @option{-Wall}.
3949 @item -Wunused-but-set-parameter
3950 @opindex Wunused-but-set-parameter
3951 @opindex Wno-unused-but-set-parameter
3952 Warn whenever a function parameter is assigned to, but otherwise unused
3953 (aside from its declaration).
3955 To suppress this warning use the @samp{unused} attribute
3956 (@pxref{Variable Attributes}).
3958 This warning is also enabled by @option{-Wunused} together with
3961 @item -Wunused-but-set-variable
3962 @opindex Wunused-but-set-variable
3963 @opindex Wno-unused-but-set-variable
3964 Warn whenever a local variable is assigned to, but otherwise unused
3965 (aside from its declaration).
3966 This warning is enabled by @option{-Wall}.
3968 To suppress this warning use the @samp{unused} attribute
3969 (@pxref{Variable Attributes}).
3971 This warning is also enabled by @option{-Wunused}, which is enabled
3974 @item -Wunused-function
3975 @opindex Wunused-function
3976 @opindex Wno-unused-function
3977 Warn whenever a static function is declared but not defined or a
3978 non-inline static function is unused.
3979 This warning is enabled by @option{-Wall}.
3981 @item -Wunused-label
3982 @opindex Wunused-label
3983 @opindex Wno-unused-label
3984 Warn whenever a label is declared but not used.
3985 This warning is enabled by @option{-Wall}.
3987 To suppress this warning use the @samp{unused} attribute
3988 (@pxref{Variable Attributes}).
3990 @item -Wunused-local-typedefs @r{(C, Objective-C, C++ and Objective-C++ only)}
3991 @opindex Wunused-local-typedefs
3992 Warn when a typedef locally defined in a function is not used.
3993 This warning is enabled by @option{-Wall}.
3995 @item -Wunused-parameter
3996 @opindex Wunused-parameter
3997 @opindex Wno-unused-parameter
3998 Warn whenever a function parameter is unused aside from its declaration.
4000 To suppress this warning use the @samp{unused} attribute
4001 (@pxref{Variable Attributes}).
4003 @item -Wno-unused-result
4004 @opindex Wunused-result
4005 @opindex Wno-unused-result
4006 Do not warn if a caller of a function marked with attribute
4007 @code{warn_unused_result} (@pxref{Function Attributes}) does not use
4008 its return value. The default is @option{-Wunused-result}.
4010 @item -Wunused-variable
4011 @opindex Wunused-variable
4012 @opindex Wno-unused-variable
4013 Warn whenever a local variable or non-constant static variable is unused
4014 aside from its declaration.
4015 This warning is enabled by @option{-Wall}.
4017 To suppress this warning use the @samp{unused} attribute
4018 (@pxref{Variable Attributes}).
4020 Note that a classic way to avoid @option{-Wunused-variable} warning is
4021 using @code{x = x}, but that does not work with @option{-Wself-assign}.
4022 Use @code{(void) x} or @code{static_cast<void>(x)} instead.
4024 @item -Wunused-value
4025 @opindex Wunused-value
4026 @opindex Wno-unused-value
4027 Warn whenever a statement computes a result that is explicitly not
4028 used. To suppress this warning cast the unused expression to
4029 @samp{void}. This includes an expression-statement or the left-hand
4030 side of a comma expression that contains no side effects. For example,
4031 an expression such as @samp{x[i,j]} causes a warning, while
4032 @samp{x[(void)i,j]} does not.
4034 This warning is enabled by @option{-Wall}.
4039 All the above @option{-Wunused} options combined.
4041 In order to get a warning about an unused function parameter, you must
4042 either specify @option{-Wextra -Wunused} (note that @option{-Wall} implies
4043 @option{-Wunused}), or separately specify @option{-Wunused-parameter}.
4045 @item -Wuninitialized
4046 @opindex Wuninitialized
4047 @opindex Wno-uninitialized
4048 Warn if an automatic variable is used without first being initialized
4049 or if a variable may be clobbered by a @code{setjmp} call. In C++,
4050 warn if a non-static reference or non-static @samp{const} member
4051 appears in a class without constructors.
4053 If you want to warn about code that uses the uninitialized value of the
4054 variable in its own initializer, use the @option{-Winit-self} option.
4056 These warnings occur for individual uninitialized or clobbered
4057 elements of structure, union or array variables as well as for
4058 variables that are uninitialized or clobbered as a whole. They do
4059 not occur for variables or elements declared @code{volatile}. Because
4060 these warnings depend on optimization, the exact variables or elements
4061 for which there are warnings depends on the precise optimization
4062 options and version of GCC used.
4064 Note that there may be no warning about a variable that is used only
4065 to compute a value that itself is never used, because such
4066 computations may be deleted by data flow analysis before the warnings
4069 @item -Wmaybe-uninitialized
4070 @opindex Wmaybe-uninitialized
4071 @opindex Wno-maybe-uninitialized
4072 For an automatic variable, if there exists a path from the function
4073 entry to a use of the variable that is initialized, but there exist
4074 some other paths for which the variable is not initialized, the compiler
4075 emits a warning if it cannot prove the uninitialized paths are not
4076 executed at run time. These warnings are made optional because GCC is
4077 not smart enough to see all the reasons why the code might be correct
4078 in spite of appearing to have an error. Here is one example of how
4099 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
4100 always initialized, but GCC doesn't know this. To suppress the
4101 warning, you need to provide a default case with assert(0) or
4104 @cindex @code{longjmp} warnings
4105 This option also warns when a non-volatile automatic variable might be
4106 changed by a call to @code{longjmp}. These warnings as well are possible
4107 only in optimizing compilation.
4109 The compiler sees only the calls to @code{setjmp}. It cannot know
4110 where @code{longjmp} will be called; in fact, a signal handler could
4111 call it at any point in the code. As a result, you may get a warning
4112 even when there is in fact no problem because @code{longjmp} cannot
4113 in fact be called at the place that would cause a problem.
4115 Some spurious warnings can be avoided if you declare all the functions
4116 you use that never return as @code{noreturn}. @xref{Function
4119 This warning is enabled by @option{-Wall} or @option{-Wextra}.
4121 @item -Wunknown-pragmas
4122 @opindex Wunknown-pragmas
4123 @opindex Wno-unknown-pragmas
4124 @cindex warning for unknown pragmas
4125 @cindex unknown pragmas, warning
4126 @cindex pragmas, warning of unknown
4127 Warn when a @code{#pragma} directive is encountered that is not understood by
4128 GCC@. If this command-line option is used, warnings are even issued
4129 for unknown pragmas in system header files. This is not the case if
4130 the warnings are only enabled by the @option{-Wall} command-line option.
4133 @opindex Wno-pragmas
4135 Do not warn about misuses of pragmas, such as incorrect parameters,
4136 invalid syntax, or conflicts between pragmas. See also
4137 @option{-Wunknown-pragmas}.
4139 @item -Wstrict-aliasing
4140 @opindex Wstrict-aliasing
4141 @opindex Wno-strict-aliasing
4142 This option is only active when @option{-fstrict-aliasing} is active.
4143 It warns about code that might break the strict aliasing rules that the
4144 compiler is using for optimization. The warning does not catch all
4145 cases, but does attempt to catch the more common pitfalls. It is
4146 included in @option{-Wall}.
4147 It is equivalent to @option{-Wstrict-aliasing=3}
4149 @item -Wstrict-aliasing=n
4150 @opindex Wstrict-aliasing=n
4151 This option is only active when @option{-fstrict-aliasing} is active.
4152 It warns about code that might break the strict aliasing rules that the
4153 compiler is using for optimization.
4154 Higher levels correspond to higher accuracy (fewer false positives).
4155 Higher levels also correspond to more effort, similar to the way @option{-O}
4157 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=3}.
4159 Level 1: Most aggressive, quick, least accurate.
4160 Possibly useful when higher levels
4161 do not warn but @option{-fstrict-aliasing} still breaks the code, as it has very few
4162 false negatives. However, it has many false positives.
4163 Warns for all pointer conversions between possibly incompatible types,
4164 even if never dereferenced. Runs in the front end only.
4166 Level 2: Aggressive, quick, not too precise.
4167 May still have many false positives (not as many as level 1 though),
4168 and few false negatives (but possibly more than level 1).
4169 Unlike level 1, it only warns when an address is taken. Warns about
4170 incomplete types. Runs in the front end only.
4172 Level 3 (default for @option{-Wstrict-aliasing}):
4173 Should have very few false positives and few false
4174 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
4175 Takes care of the common pun+dereference pattern in the front end:
4176 @code{*(int*)&some_float}.
4177 If optimization is enabled, it also runs in the back end, where it deals
4178 with multiple statement cases using flow-sensitive points-to information.
4179 Only warns when the converted pointer is dereferenced.
4180 Does not warn about incomplete types.
4182 @item -Wstrict-overflow
4183 @itemx -Wstrict-overflow=@var{n}
4184 @opindex Wstrict-overflow
4185 @opindex Wno-strict-overflow
4186 This option is only active when @option{-fstrict-overflow} is active.
4187 It warns about cases where the compiler optimizes based on the
4188 assumption that signed overflow does not occur. Note that it does not
4189 warn about all cases where the code might overflow: it only warns
4190 about cases where the compiler implements some optimization. Thus
4191 this warning depends on the optimization level.
4193 An optimization that assumes that signed overflow does not occur is
4194 perfectly safe if the values of the variables involved are such that
4195 overflow never does, in fact, occur. Therefore this warning can
4196 easily give a false positive: a warning about code that is not
4197 actually a problem. To help focus on important issues, several
4198 warning levels are defined. No warnings are issued for the use of
4199 undefined signed overflow when estimating how many iterations a loop
4200 requires, in particular when determining whether a loop will be
4204 @item -Wstrict-overflow=1
4205 Warn about cases that are both questionable and easy to avoid. For
4206 example, with @option{-fstrict-overflow}, the compiler simplifies
4207 @code{x + 1 > x} to @code{1}. This level of
4208 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
4209 are not, and must be explicitly requested.
4211 @item -Wstrict-overflow=2
4212 Also warn about other cases where a comparison is simplified to a
4213 constant. For example: @code{abs (x) >= 0}. This can only be
4214 simplified when @option{-fstrict-overflow} is in effect, because
4215 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
4216 zero. @option{-Wstrict-overflow} (with no level) is the same as
4217 @option{-Wstrict-overflow=2}.
4219 @item -Wstrict-overflow=3
4220 Also warn about other cases where a comparison is simplified. For
4221 example: @code{x + 1 > 1} is simplified to @code{x > 0}.
4223 @item -Wstrict-overflow=4
4224 Also warn about other simplifications not covered by the above cases.
4225 For example: @code{(x * 10) / 5} is simplified to @code{x * 2}.
4227 @item -Wstrict-overflow=5
4228 Also warn about cases where the compiler reduces the magnitude of a
4229 constant involved in a comparison. For example: @code{x + 2 > y} is
4230 simplified to @code{x + 1 >= y}. This is reported only at the
4231 highest warning level because this simplification applies to many
4232 comparisons, so this warning level gives a very large number of
4236 @item -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{]}
4237 @opindex Wsuggest-attribute=
4238 @opindex Wno-suggest-attribute=
4239 Warn for cases where adding an attribute may be beneficial. The
4240 attributes currently supported are listed below.
4243 @item -Wsuggest-attribute=pure
4244 @itemx -Wsuggest-attribute=const
4245 @itemx -Wsuggest-attribute=noreturn
4246 @opindex Wsuggest-attribute=pure
4247 @opindex Wno-suggest-attribute=pure
4248 @opindex Wsuggest-attribute=const
4249 @opindex Wno-suggest-attribute=const
4250 @opindex Wsuggest-attribute=noreturn
4251 @opindex Wno-suggest-attribute=noreturn
4253 Warn about functions that might be candidates for attributes
4254 @code{pure}, @code{const} or @code{noreturn}. The compiler only warns for
4255 functions visible in other compilation units or (in the case of @code{pure} and
4256 @code{const}) if it cannot prove that the function returns normally. A function
4257 returns normally if it doesn't contain an infinite loop or return abnormally
4258 by throwing, calling @code{abort()} or trapping. This analysis requires option
4259 @option{-fipa-pure-const}, which is enabled by default at @option{-O} and
4260 higher. Higher optimization levels improve the accuracy of the analysis.
4262 @item -Wsuggest-attribute=format
4263 @itemx -Wmissing-format-attribute
4264 @opindex Wsuggest-attribute=format
4265 @opindex Wmissing-format-attribute
4266 @opindex Wno-suggest-attribute=format
4267 @opindex Wno-missing-format-attribute
4271 Warn about function pointers that might be candidates for @code{format}
4272 attributes. Note these are only possible candidates, not absolute ones.
4273 GCC guesses that function pointers with @code{format} attributes that
4274 are used in assignment, initialization, parameter passing or return
4275 statements should have a corresponding @code{format} attribute in the
4276 resulting type. I.e.@: the left-hand side of the assignment or
4277 initialization, the type of the parameter variable, or the return type
4278 of the containing function respectively should also have a @code{format}
4279 attribute to avoid the warning.
4281 GCC also warns about function definitions that might be
4282 candidates for @code{format} attributes. Again, these are only
4283 possible candidates. GCC guesses that @code{format} attributes
4284 might be appropriate for any function that calls a function like
4285 @code{vprintf} or @code{vscanf}, but this might not always be the
4286 case, and some functions for which @code{format} attributes are
4287 appropriate may not be detected.
4290 @item -Wsuggest-final-types
4291 @opindex Wno-suggest-final-types
4292 @opindex Wsuggest-final-types
4293 Warn about types with virtual methods where code quality would be improved
4294 if the type was declared with C++11 final specifier, or, if possible,
4295 declared in anonymous namespace. This allows GCC to devritualize more aggressively
4296 the polymorphic calls. This warning is more effective with link time optimization,
4297 where the information about the class hiearchy graph is more complete.
4299 @item -Wsuggest-final-methods
4300 @opindex Wno-suggest-final-methods
4301 @opindex Wsuggest-final-methods
4302 Warn about virtual methods where code quality would be improved if the method
4303 was declared with C++11 final specifier, or, if possible, its type was declared
4304 in the anonymous namespace or with final specifier. This warning is more
4305 effective with link time optimization, where the information about the class
4306 hiearchy graph is more complete. It is recommended to first consider suggestins
4307 of @option{-Wsuggest-final-types} and then rebuild with new annotations.
4309 @item -Warray-bounds
4310 @opindex Wno-array-bounds
4311 @opindex Warray-bounds
4312 This option is only active when @option{-ftree-vrp} is active
4313 (default for @option{-O2} and above). It warns about subscripts to arrays
4314 that are always out of bounds. This warning is enabled by @option{-Wall}.
4316 @item -Wno-discarded-qualifiers @r{(C and Objective-C only)}
4317 @opindex Wno-discarded-qualifiers
4318 @opindex Wdiscarded-qualifiers
4319 Do not warn if type qualifiers on pointers are being discarded.
4320 Typically, the compiler will warn if a @code{const char *} variable is
4321 passed to a function that takes @code{char *} parameter. This option
4322 can be used to suppress such a warning.
4324 @item -Wno-incompatible-pointer-types @r{(C and Objective-C only)}
4325 @opindex Wno-incompatible-pointer-types
4326 @opindex Wincompatible-pointer-types
4327 Do not warn when there is a conversion between pointers that have incompatible
4328 types. This warning is for cases not covered by @option{-Wno-pointer-sign},
4329 which warns for pointer argument passing or assignment with different signedness
4331 @item -Wno-int-conversion @r{(C and Objective-C only)}
4332 @opindex Wno-int-conversion
4333 @opindex Wint-conversion
4334 Do not warn about incompatible integer to pointer and pointer to integer
4335 conversions. This warning is about implicit conversions; for explicit
4336 conversions the warnings @option{-Wno-int-to-pointer-cast} and
4337 @option{-Wno-pointer-to-int-cast} may be used.
4339 @item -Wno-div-by-zero
4340 @opindex Wno-div-by-zero
4341 @opindex Wdiv-by-zero
4342 Do not warn about compile-time integer division by zero. Floating-point
4343 division by zero is not warned about, as it can be a legitimate way of
4344 obtaining infinities and NaNs.
4346 @item -Wsystem-headers
4347 @opindex Wsystem-headers
4348 @opindex Wno-system-headers
4349 @cindex warnings from system headers
4350 @cindex system headers, warnings from
4351 Print warning messages for constructs found in system header files.
4352 Warnings from system headers are normally suppressed, on the assumption
4353 that they usually do not indicate real problems and would only make the
4354 compiler output harder to read. Using this command-line option tells
4355 GCC to emit warnings from system headers as if they occurred in user
4356 code. However, note that using @option{-Wall} in conjunction with this
4357 option does @emph{not} warn about unknown pragmas in system
4358 headers---for that, @option{-Wunknown-pragmas} must also be used.
4361 @opindex Wtrampolines
4362 @opindex Wno-trampolines
4363 Warn about trampolines generated for pointers to nested functions.
4365 A trampoline is a small piece of data or code that is created at run
4366 time on the stack when the address of a nested function is taken, and
4367 is used to call the nested function indirectly. For some targets, it
4368 is made up of data only and thus requires no special treatment. But,
4369 for most targets, it is made up of code and thus requires the stack
4370 to be made executable in order for the program to work properly.
4373 @opindex Wfloat-equal
4374 @opindex Wno-float-equal
4375 Warn if floating-point values are used in equality comparisons.
4377 The idea behind this is that sometimes it is convenient (for the
4378 programmer) to consider floating-point values as approximations to
4379 infinitely precise real numbers. If you are doing this, then you need
4380 to compute (by analyzing the code, or in some other way) the maximum or
4381 likely maximum error that the computation introduces, and allow for it
4382 when performing comparisons (and when producing output, but that's a
4383 different problem). In particular, instead of testing for equality, you
4384 should check to see whether the two values have ranges that overlap; and
4385 this is done with the relational operators, so equality comparisons are
4388 @item -Wtraditional @r{(C and Objective-C only)}
4389 @opindex Wtraditional
4390 @opindex Wno-traditional
4391 Warn about certain constructs that behave differently in traditional and
4392 ISO C@. Also warn about ISO C constructs that have no traditional C
4393 equivalent, and/or problematic constructs that should be avoided.
4397 Macro parameters that appear within string literals in the macro body.
4398 In traditional C macro replacement takes place within string literals,
4399 but in ISO C it does not.
4402 In traditional C, some preprocessor directives did not exist.
4403 Traditional preprocessors only considered a line to be a directive
4404 if the @samp{#} appeared in column 1 on the line. Therefore
4405 @option{-Wtraditional} warns about directives that traditional C
4406 understands but ignores because the @samp{#} does not appear as the
4407 first character on the line. It also suggests you hide directives like
4408 @samp{#pragma} not understood by traditional C by indenting them. Some
4409 traditional implementations do not recognize @samp{#elif}, so this option
4410 suggests avoiding it altogether.
4413 A function-like macro that appears without arguments.
4416 The unary plus operator.
4419 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating-point
4420 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
4421 constants.) Note, these suffixes appear in macros defined in the system
4422 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
4423 Use of these macros in user code might normally lead to spurious
4424 warnings, however GCC's integrated preprocessor has enough context to
4425 avoid warning in these cases.
4428 A function declared external in one block and then used after the end of
4432 A @code{switch} statement has an operand of type @code{long}.
4435 A non-@code{static} function declaration follows a @code{static} one.
4436 This construct is not accepted by some traditional C compilers.
4439 The ISO type of an integer constant has a different width or
4440 signedness from its traditional type. This warning is only issued if
4441 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
4442 typically represent bit patterns, are not warned about.
4445 Usage of ISO string concatenation is detected.
4448 Initialization of automatic aggregates.
4451 Identifier conflicts with labels. Traditional C lacks a separate
4452 namespace for labels.
4455 Initialization of unions. If the initializer is zero, the warning is
4456 omitted. This is done under the assumption that the zero initializer in
4457 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
4458 initializer warnings and relies on default initialization to zero in the
4462 Conversions by prototypes between fixed/floating-point values and vice
4463 versa. The absence of these prototypes when compiling with traditional
4464 C causes serious problems. This is a subset of the possible
4465 conversion warnings; for the full set use @option{-Wtraditional-conversion}.
4468 Use of ISO C style function definitions. This warning intentionally is
4469 @emph{not} issued for prototype declarations or variadic functions
4470 because these ISO C features appear in your code when using
4471 libiberty's traditional C compatibility macros, @code{PARAMS} and
4472 @code{VPARAMS}. This warning is also bypassed for nested functions
4473 because that feature is already a GCC extension and thus not relevant to
4474 traditional C compatibility.
4477 @item -Wtraditional-conversion @r{(C and Objective-C only)}
4478 @opindex Wtraditional-conversion
4479 @opindex Wno-traditional-conversion
4480 Warn if a prototype causes a type conversion that is different from what
4481 would happen to the same argument in the absence of a prototype. This
4482 includes conversions of fixed point to floating and vice versa, and
4483 conversions changing the width or signedness of a fixed-point argument
4484 except when the same as the default promotion.
4486 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
4487 @opindex Wdeclaration-after-statement
4488 @opindex Wno-declaration-after-statement
4489 Warn when a declaration is found after a statement in a block. This
4490 construct, known from C++, was introduced with ISO C99 and is by default
4491 allowed in GCC@. It is not supported by ISO C90 and was not supported by
4492 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
4497 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
4499 @item -Wno-endif-labels
4500 @opindex Wno-endif-labels
4501 @opindex Wendif-labels
4502 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
4507 Warn whenever a local variable or type declaration shadows another
4508 variable, parameter, type, class member (in C++), or instance variable
4509 (in Objective-C) or whenever a built-in function is shadowed. Note
4510 that in C++, the compiler warns if a local variable shadows an
4511 explicit typedef, but not if it shadows a struct/class/enum.
4513 @item -Wshadow-local
4514 @opindex Wshadow-local
4515 @opindex Wno-shadow-local
4516 Warn when a local variable shadows another local variable or parameter.
4518 @item -Wshadow-compatible-local
4519 @opindex Wshadow-compatible-local
4520 @opindex Wno-shadow-compatible-local
4521 Warn when a local variable shadows another local variable or parameter
4522 whose type is compatible with that of the shadowing variable. In C++,
4523 type compatibility here means the type of the shadowing variable can be
4524 converted to that of the shadowed variable. The creation of this flag
4525 (in addition to @option{-Wshadow-local}) is based on the idea that when
4526 a local variable shadows another one of incompatible type, it is most
4527 likely intentional, not a bug or typo, as shown in the following example:
4531 for (SomeIterator i = SomeObj.begin(); i != SomeObj.end(); ++i)
4533 for (int i = 0; i < N; ++i)
4542 Since the two variable @code{i} in the example above have incompatible types,
4543 enabling only @option{-Wshadow-compatible-local} will not emit a warning.
4544 Because their types are incompatible, if a programmer accidentally uses one
4545 in place of the other, type checking will catch that and emit an error or
4546 warning. So not warning (about shadowing) in this case will not lead to
4547 undetected bugs. Use of this flag instead of @option{-Wshadow-local} can
4548 possibly reduce the number of warnings triggered by intentional shadowing.
4550 @item -Wno-shadow-ivar @r{(Objective-C only)}
4551 @opindex Wno-shadow-ivar
4552 @opindex Wshadow-ivar
4553 Do not warn whenever a local variable shadows an instance variable in an
4556 @item -Wlarger-than=@var{len}
4557 @opindex Wlarger-than=@var{len}
4558 @opindex Wlarger-than-@var{len}
4559 Warn whenever an object of larger than @var{len} bytes is defined.
4561 @item -Wframe-larger-than=@var{len}
4562 @opindex Wframe-larger-than
4563 Warn if the size of a function frame is larger than @var{len} bytes.
4564 The computation done to determine the stack frame size is approximate
4565 and not conservative.
4566 The actual requirements may be somewhat greater than @var{len}
4567 even if you do not get a warning. In addition, any space allocated
4568 via @code{alloca}, variable-length arrays, or related constructs
4569 is not included by the compiler when determining
4570 whether or not to issue a warning.
4572 @item -Wno-free-nonheap-object
4573 @opindex Wno-free-nonheap-object
4574 @opindex Wfree-nonheap-object
4575 Do not warn when attempting to free an object that was not allocated
4578 @item -Wstack-usage=@var{len}
4579 @opindex Wstack-usage
4580 Warn if the stack usage of a function might be larger than @var{len} bytes.
4581 The computation done to determine the stack usage is conservative.
4582 Any space allocated via @code{alloca}, variable-length arrays, or related
4583 constructs is included by the compiler when determining whether or not to
4586 The message is in keeping with the output of @option{-fstack-usage}.
4590 If the stack usage is fully static but exceeds the specified amount, it's:
4593 warning: stack usage is 1120 bytes
4596 If the stack usage is (partly) dynamic but bounded, it's:
4599 warning: stack usage might be 1648 bytes
4602 If the stack usage is (partly) dynamic and not bounded, it's:
4605 warning: stack usage might be unbounded
4609 @item -Wunsafe-loop-optimizations
4610 @opindex Wunsafe-loop-optimizations
4611 @opindex Wno-unsafe-loop-optimizations
4612 Warn if the loop cannot be optimized because the compiler cannot
4613 assume anything on the bounds of the loop indices. With
4614 @option{-funsafe-loop-optimizations} warn if the compiler makes
4617 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
4618 @opindex Wno-pedantic-ms-format
4619 @opindex Wpedantic-ms-format
4620 When used in combination with @option{-Wformat}
4621 and @option{-pedantic} without GNU extensions, this option
4622 disables the warnings about non-ISO @code{printf} / @code{scanf} format
4623 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets,
4624 which depend on the MS runtime.
4626 @item -Wpointer-arith
4627 @opindex Wpointer-arith
4628 @opindex Wno-pointer-arith
4629 Warn about anything that depends on the ``size of'' a function type or
4630 of @code{void}. GNU C assigns these types a size of 1, for
4631 convenience in calculations with @code{void *} pointers and pointers
4632 to functions. In C++, warn also when an arithmetic operation involves
4633 @code{NULL}. This warning is also enabled by @option{-Wpedantic}.
4636 @opindex Wtype-limits
4637 @opindex Wno-type-limits
4638 Warn if a comparison is always true or always false due to the limited
4639 range of the data type, but do not warn for constant expressions. For
4640 example, warn if an unsigned variable is compared against zero with
4641 @samp{<} or @samp{>=}. This warning is also enabled by
4644 @item -Wbad-function-cast @r{(C and Objective-C only)}
4645 @opindex Wbad-function-cast
4646 @opindex Wno-bad-function-cast
4647 Warn whenever a function call is cast to a non-matching type.
4648 For example, warn if @code{int malloc()} is cast to @code{anything *}.
4650 @item -Wc++-compat @r{(C and Objective-C only)}
4651 Warn about ISO C constructs that are outside of the common subset of
4652 ISO C and ISO C++, e.g.@: request for implicit conversion from
4653 @code{void *} to a pointer to non-@code{void} type.
4655 @item -Wc++11-compat @r{(C++ and Objective-C++ only)}
4656 Warn about C++ constructs whose meaning differs between ISO C++ 1998
4657 and ISO C++ 2011, e.g., identifiers in ISO C++ 1998 that are keywords
4658 in ISO C++ 2011. This warning turns on @option{-Wnarrowing} and is
4659 enabled by @option{-Wall}.
4663 @opindex Wno-cast-qual
4664 Warn whenever a pointer is cast so as to remove a type qualifier from
4665 the target type. For example, warn if a @code{const char *} is cast
4666 to an ordinary @code{char *}.
4668 Also warn when making a cast that introduces a type qualifier in an
4669 unsafe way. For example, casting @code{char **} to @code{const char **}
4670 is unsafe, as in this example:
4673 /* p is char ** value. */
4674 const char **q = (const char **) p;
4675 /* Assignment of readonly string to const char * is OK. */
4677 /* Now char** pointer points to read-only memory. */
4682 @opindex Wcast-align
4683 @opindex Wno-cast-align
4684 Warn whenever a pointer is cast such that the required alignment of the
4685 target is increased. For example, warn if a @code{char *} is cast to
4686 an @code{int *} on machines where integers can only be accessed at
4687 two- or four-byte boundaries.
4689 @item -Wwrite-strings
4690 @opindex Wwrite-strings
4691 @opindex Wno-write-strings
4692 When compiling C, give string constants the type @code{const
4693 char[@var{length}]} so that copying the address of one into a
4694 non-@code{const} @code{char *} pointer produces a warning. These
4695 warnings help you find at compile time code that can try to write
4696 into a string constant, but only if you have been very careful about
4697 using @code{const} in declarations and prototypes. Otherwise, it is
4698 just a nuisance. This is why we did not make @option{-Wall} request
4701 When compiling C++, warn about the deprecated conversion from string
4702 literals to @code{char *}. This warning is enabled by default for C++
4707 @opindex Wno-clobbered
4708 Warn for variables that might be changed by @samp{longjmp} or
4709 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
4711 @item -Wconditionally-supported @r{(C++ and Objective-C++ only)}
4712 @opindex Wconditionally-supported
4713 @opindex Wno-conditionally-supported
4714 Warn for conditionally-supported (C++11 [intro.defs]) constructs.
4717 @opindex Wconversion
4718 @opindex Wno-conversion
4719 Warn for implicit conversions that may alter a value. This includes
4720 conversions between real and integer, like @code{abs (x)} when
4721 @code{x} is @code{double}; conversions between signed and unsigned,
4722 like @code{unsigned ui = -1}; and conversions to smaller types, like
4723 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
4724 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
4725 changed by the conversion like in @code{abs (2.0)}. Warnings about
4726 conversions between signed and unsigned integers can be disabled by
4727 using @option{-Wno-sign-conversion}.
4729 For C++, also warn for confusing overload resolution for user-defined
4730 conversions; and conversions that never use a type conversion
4731 operator: conversions to @code{void}, the same type, a base class or a
4732 reference to them. Warnings about conversions between signed and
4733 unsigned integers are disabled by default in C++ unless
4734 @option{-Wsign-conversion} is explicitly enabled.
4736 @item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
4737 @opindex Wconversion-null
4738 @opindex Wno-conversion-null
4739 Do not warn for conversions between @code{NULL} and non-pointer
4740 types. @option{-Wconversion-null} is enabled by default.
4742 @item -Wzero-as-null-pointer-constant @r{(C++ and Objective-C++ only)}
4743 @opindex Wzero-as-null-pointer-constant
4744 @opindex Wno-zero-as-null-pointer-constant
4745 Warn when a literal '0' is used as null pointer constant. This can
4746 be useful to facilitate the conversion to @code{nullptr} in C++11.
4750 @opindex Wno-date-time
4751 Warn when macros @code{__TIME__}, @code{__DATE__} or @code{__TIMESTAMP__}
4752 are encountered as they might prevent bit-wise-identical reproducible
4755 @item -Wdelete-incomplete @r{(C++ and Objective-C++ only)}
4756 @opindex Wdelete-incomplete
4757 @opindex Wno-delete-incomplete
4758 Warn when deleting a pointer to incomplete type, which may cause
4759 undefined behavior at runtime. This warning is enabled by default.
4761 @item -Wuseless-cast @r{(C++ and Objective-C++ only)}
4762 @opindex Wuseless-cast
4763 @opindex Wno-useless-cast
4764 Warn when an expression is casted to its own type.
4767 @opindex Wempty-body
4768 @opindex Wno-empty-body
4769 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
4770 while} statement. This warning is also enabled by @option{-Wextra}.
4772 @item -Wenum-compare
4773 @opindex Wenum-compare
4774 @opindex Wno-enum-compare
4775 Warn about a comparison between values of different enumerated types.
4776 In C++ enumeral mismatches in conditional expressions are also
4777 diagnosed and the warning is enabled by default. In C this warning is
4778 enabled by @option{-Wall}.
4780 @item -Wjump-misses-init @r{(C, Objective-C only)}
4781 @opindex Wjump-misses-init
4782 @opindex Wno-jump-misses-init
4783 Warn if a @code{goto} statement or a @code{switch} statement jumps
4784 forward across the initialization of a variable, or jumps backward to a
4785 label after the variable has been initialized. This only warns about
4786 variables that are initialized when they are declared. This warning is
4787 only supported for C and Objective-C; in C++ this sort of branch is an
4790 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
4791 can be disabled with the @option{-Wno-jump-misses-init} option.
4793 @item -Wsign-compare
4794 @opindex Wsign-compare
4795 @opindex Wno-sign-compare
4796 @cindex warning for comparison of signed and unsigned values
4797 @cindex comparison of signed and unsigned values, warning
4798 @cindex signed and unsigned values, comparison warning
4799 Warn when a comparison between signed and unsigned values could produce
4800 an incorrect result when the signed value is converted to unsigned.
4801 This warning is also enabled by @option{-Wextra}; to get the other warnings
4802 of @option{-Wextra} without this warning, use @option{-Wextra -Wno-sign-compare}.
4804 @item -Wsign-conversion
4805 @opindex Wsign-conversion
4806 @opindex Wno-sign-conversion
4807 Warn for implicit conversions that may change the sign of an integer
4808 value, like assigning a signed integer expression to an unsigned
4809 integer variable. An explicit cast silences the warning. In C, this
4810 option is enabled also by @option{-Wconversion}.
4812 @item -Wfloat-conversion
4813 @opindex Wfloat-conversion
4814 @opindex Wno-float-conversion
4815 Warn for implicit conversions that reduce the precision of a real value.
4816 This includes conversions from real to integer, and from higher precision
4817 real to lower precision real values. This option is also enabled by
4818 @option{-Wconversion}.
4820 @item -Wsizeof-pointer-memaccess
4821 @opindex Wsizeof-pointer-memaccess
4822 @opindex Wno-sizeof-pointer-memaccess
4823 Warn for suspicious length parameters to certain string and memory built-in
4824 functions if the argument uses @code{sizeof}. This warning warns e.g.@:
4825 about @code{memset (ptr, 0, sizeof (ptr));} if @code{ptr} is not an array,
4826 but a pointer, and suggests a possible fix, or about
4827 @code{memcpy (&foo, ptr, sizeof (&foo));}. This warning is enabled by
4830 @item -Wsizeof-array-argument
4831 @opindex Wsizeof-array-argument
4832 @opindex Wno-sizeof-array-argument
4833 Warn when the @code{sizeof} operator is applied to a parameter that is
4834 declared as an array in a function definition. This warning is enabled by
4835 default for C and C++ programs.
4837 @item -Wmemset-transposed-args
4838 @opindex Wmemset-transposed-args
4839 @opindex Wno-memset-transposed-args
4840 Warn for suspicious calls to the @code{memset} built-in function, if the
4841 second argument is not zero and the third argument is zero. This warns e.g.@
4842 about @code{memset (buf, sizeof buf, 0)} where most probably
4843 @code{memset (buf, 0, sizeof buf)} was meant instead. The diagnostics
4844 is only emitted if the third argument is literal zero, if it is some expression
4845 that is folded to zero, or e.g. a cast of zero to some type etc., it
4846 is far less likely that user has mistakenly exchanged the arguments and
4847 no warning is emitted. This warning is enabled by @option{-Wall}.
4851 @opindex Wno-address
4852 Warn about suspicious uses of memory addresses. These include using
4853 the address of a function in a conditional expression, such as
4854 @code{void func(void); if (func)}, and comparisons against the memory
4855 address of a string literal, such as @code{if (x == "abc")}. Such
4856 uses typically indicate a programmer error: the address of a function
4857 always evaluates to true, so their use in a conditional usually
4858 indicate that the programmer forgot the parentheses in a function
4859 call; and comparisons against string literals result in unspecified
4860 behavior and are not portable in C, so they usually indicate that the
4861 programmer intended to use @code{strcmp}. This warning is enabled by
4865 @opindex Wlogical-op
4866 @opindex Wno-logical-op
4867 Warn about suspicious uses of logical operators in expressions.
4868 This includes using logical operators in contexts where a
4869 bit-wise operator is likely to be expected.
4871 @item -Wlogical-not-parentheses
4872 @opindex Wlogical-not-parentheses
4873 @opindex Wno-logical-not-parentheses
4874 Warn about logical not used on the left hand side operand of a comparison.
4875 This option does not warn if the LHS or RHS operand is of a boolean or
4876 a vector type. Its purpose is to detect suspicious code like the following:
4880 if (!a > 1) @{ @dots{} @}
4883 It is possible to suppress the warning by wrapping the LHS into
4886 if ((!a) > 1) @{ @dots{} @}
4889 @item -Waggregate-return
4890 @opindex Waggregate-return
4891 @opindex Wno-aggregate-return
4892 Warn if any functions that return structures or unions are defined or
4893 called. (In languages where you can return an array, this also elicits
4896 @item -Wno-aggressive-loop-optimizations
4897 @opindex Wno-aggressive-loop-optimizations
4898 @opindex Waggressive-loop-optimizations
4899 Warn if in a loop with constant number of iterations the compiler detects
4900 undefined behavior in some statement during one or more of the iterations.
4902 @item -Wno-attributes
4903 @opindex Wno-attributes
4904 @opindex Wattributes
4905 Do not warn if an unexpected @code{__attribute__} is used, such as
4906 unrecognized attributes, function attributes applied to variables,
4907 etc. This does not stop errors for incorrect use of supported
4910 @item -Wno-builtin-macro-redefined
4911 @opindex Wno-builtin-macro-redefined
4912 @opindex Wbuiltin-macro-redefined
4913 Do not warn if certain built-in macros are redefined. This suppresses
4914 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
4915 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
4917 @item -Wstrict-prototypes @r{(C and Objective-C only)}
4918 @opindex Wstrict-prototypes
4919 @opindex Wno-strict-prototypes
4920 Warn if a function is declared or defined without specifying the
4921 argument types. (An old-style function definition is permitted without
4922 a warning if preceded by a declaration that specifies the argument
4925 @item -Wold-style-declaration @r{(C and Objective-C only)}
4926 @opindex Wold-style-declaration
4927 @opindex Wno-old-style-declaration
4928 Warn for obsolescent usages, according to the C Standard, in a
4929 declaration. For example, warn if storage-class specifiers like
4930 @code{static} are not the first things in a declaration. This warning
4931 is also enabled by @option{-Wextra}.
4933 @item -Wold-style-definition @r{(C and Objective-C only)}
4934 @opindex Wold-style-definition
4935 @opindex Wno-old-style-definition
4936 Warn if an old-style function definition is used. A warning is given
4937 even if there is a previous prototype.
4939 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
4940 @opindex Wmissing-parameter-type
4941 @opindex Wno-missing-parameter-type
4942 A function parameter is declared without a type specifier in K&R-style
4949 This warning is also enabled by @option{-Wextra}.
4951 @item -Wmissing-prototypes @r{(C and Objective-C only)}
4952 @opindex Wmissing-prototypes
4953 @opindex Wno-missing-prototypes
4954 Warn if a global function is defined without a previous prototype
4955 declaration. This warning is issued even if the definition itself
4956 provides a prototype. Use this option to detect global functions
4957 that do not have a matching prototype declaration in a header file.
4958 This option is not valid for C++ because all function declarations
4959 provide prototypes and a non-matching declaration will declare an
4960 overload rather than conflict with an earlier declaration.
4961 Use @option{-Wmissing-declarations} to detect missing declarations in C++.
4963 @item -Wmissing-declarations
4964 @opindex Wmissing-declarations
4965 @opindex Wno-missing-declarations
4966 Warn if a global function is defined without a previous declaration.
4967 Do so even if the definition itself provides a prototype.
4968 Use this option to detect global functions that are not declared in
4969 header files. In C, no warnings are issued for functions with previous
4970 non-prototype declarations; use @option{-Wmissing-prototypes} to detect
4971 missing prototypes. In C++, no warnings are issued for function templates,
4972 or for inline functions, or for functions in anonymous namespaces.
4974 @item -Wmissing-field-initializers
4975 @opindex Wmissing-field-initializers
4976 @opindex Wno-missing-field-initializers
4980 Warn if a structure's initializer has some fields missing. For
4981 example, the following code causes such a warning, because
4982 @code{x.h} is implicitly zero:
4985 struct s @{ int f, g, h; @};
4986 struct s x = @{ 3, 4 @};
4989 This option does not warn about designated initializers, so the following
4990 modification does not trigger a warning:
4993 struct s @{ int f, g, h; @};
4994 struct s x = @{ .f = 3, .g = 4 @};
4997 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
4998 warnings without this one, use @option{-Wextra -Wno-missing-field-initializers}.
5000 @item -Wno-multichar
5001 @opindex Wno-multichar
5003 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
5004 Usually they indicate a typo in the user's code, as they have
5005 implementation-defined values, and should not be used in portable code.
5007 @item -Wnormalized=<none|id|nfc|nfkc>
5008 @opindex Wnormalized=
5011 @cindex character set, input normalization
5012 In ISO C and ISO C++, two identifiers are different if they are
5013 different sequences of characters. However, sometimes when characters
5014 outside the basic ASCII character set are used, you can have two
5015 different character sequences that look the same. To avoid confusion,
5016 the ISO 10646 standard sets out some @dfn{normalization rules} which
5017 when applied ensure that two sequences that look the same are turned into
5018 the same sequence. GCC can warn you if you are using identifiers that
5019 have not been normalized; this option controls that warning.
5021 There are four levels of warning supported by GCC@. The default is
5022 @option{-Wnormalized=nfc}, which warns about any identifier that is
5023 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
5024 recommended form for most uses.
5026 Unfortunately, there are some characters allowed in identifiers by
5027 ISO C and ISO C++ that, when turned into NFC, are not allowed in
5028 identifiers. That is, there's no way to use these symbols in portable
5029 ISO C or C++ and have all your identifiers in NFC@.
5030 @option{-Wnormalized=id} suppresses the warning for these characters.
5031 It is hoped that future versions of the standards involved will correct
5032 this, which is why this option is not the default.
5034 You can switch the warning off for all characters by writing
5035 @option{-Wnormalized=none}. You should only do this if you
5036 are using some other normalization scheme (like ``D''), because
5037 otherwise you can easily create bugs that are literally impossible to see.
5039 Some characters in ISO 10646 have distinct meanings but look identical
5040 in some fonts or display methodologies, especially once formatting has
5041 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
5042 LETTER N'', displays just like a regular @code{n} that has been
5043 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
5044 normalization scheme to convert all these into a standard form as
5045 well, and GCC warns if your code is not in NFKC if you use
5046 @option{-Wnormalized=nfkc}. This warning is comparable to warning
5047 about every identifier that contains the letter O because it might be
5048 confused with the digit 0, and so is not the default, but may be
5049 useful as a local coding convention if the programming environment
5050 cannot be fixed to display these characters distinctly.
5052 @item -Wno-deprecated
5053 @opindex Wno-deprecated
5054 @opindex Wdeprecated
5055 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
5057 @item -Wno-deprecated-declarations
5058 @opindex Wno-deprecated-declarations
5059 @opindex Wdeprecated-declarations
5060 Do not warn about uses of functions (@pxref{Function Attributes}),
5061 variables (@pxref{Variable Attributes}), and types (@pxref{Type
5062 Attributes}) marked as deprecated by using the @code{deprecated}
5066 @opindex Wno-overflow
5068 Do not warn about compile-time overflow in constant expressions.
5073 Warn about One Definition Rule violations during link-time optimization.
5074 Requires @option{-flto-odr-type-merging} to be enabled. Enabled by default.
5077 @opindex Wopenm-simd
5078 Warn if the vectorizer cost model overrides the OpenMP or the Cilk Plus
5079 simd directive set by user. The @option{-fsimd-cost-model=unlimited} can
5080 be used to relax the cost model.
5082 @item -Woverride-init @r{(C and Objective-C only)}
5083 @opindex Woverride-init
5084 @opindex Wno-override-init
5088 Warn if an initialized field without side effects is overridden when
5089 using designated initializers (@pxref{Designated Inits, , Designated
5092 This warning is included in @option{-Wextra}. To get other
5093 @option{-Wextra} warnings without this one, use @option{-Wextra
5094 -Wno-override-init}.
5099 Warn if a structure is given the packed attribute, but the packed
5100 attribute has no effect on the layout or size of the structure.
5101 Such structures may be mis-aligned for little benefit. For
5102 instance, in this code, the variable @code{f.x} in @code{struct bar}
5103 is misaligned even though @code{struct bar} does not itself
5104 have the packed attribute:
5111 @} __attribute__((packed));
5119 @item -Wpacked-bitfield-compat
5120 @opindex Wpacked-bitfield-compat
5121 @opindex Wno-packed-bitfield-compat
5122 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
5123 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
5124 the change can lead to differences in the structure layout. GCC
5125 informs you when the offset of such a field has changed in GCC 4.4.
5126 For example there is no longer a 4-bit padding between field @code{a}
5127 and @code{b} in this structure:
5134 @} __attribute__ ((packed));
5137 This warning is enabled by default. Use
5138 @option{-Wno-packed-bitfield-compat} to disable this warning.
5143 Warn if padding is included in a structure, either to align an element
5144 of the structure or to align the whole structure. Sometimes when this
5145 happens it is possible to rearrange the fields of the structure to
5146 reduce the padding and so make the structure smaller.
5148 @item -Wredundant-decls
5149 @opindex Wredundant-decls
5150 @opindex Wno-redundant-decls
5151 Warn if anything is declared more than once in the same scope, even in
5152 cases where multiple declaration is valid and changes nothing.
5154 @item -Wnested-externs @r{(C and Objective-C only)}
5155 @opindex Wnested-externs
5156 @opindex Wno-nested-externs
5157 Warn if an @code{extern} declaration is encountered within a function.
5159 @item -Wno-inherited-variadic-ctor
5160 @opindex Winherited-variadic-ctor
5161 @opindex Wno-inherited-variadic-ctor
5162 Suppress warnings about use of C++11 inheriting constructors when the
5163 base class inherited from has a C variadic constructor; the warning is
5164 on by default because the ellipsis is not inherited.
5169 Warn if a function that is declared as inline cannot be inlined.
5170 Even with this option, the compiler does not warn about failures to
5171 inline functions declared in system headers.
5173 The compiler uses a variety of heuristics to determine whether or not
5174 to inline a function. For example, the compiler takes into account
5175 the size of the function being inlined and the amount of inlining
5176 that has already been done in the current function. Therefore,
5177 seemingly insignificant changes in the source program can cause the
5178 warnings produced by @option{-Winline} to appear or disappear.
5180 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
5181 @opindex Wno-invalid-offsetof
5182 @opindex Winvalid-offsetof
5183 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
5184 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
5185 to a non-POD type is undefined. In existing C++ implementations,
5186 however, @samp{offsetof} typically gives meaningful results even when
5187 applied to certain kinds of non-POD types (such as a simple
5188 @samp{struct} that fails to be a POD type only by virtue of having a
5189 constructor). This flag is for users who are aware that they are
5190 writing nonportable code and who have deliberately chosen to ignore the
5193 The restrictions on @samp{offsetof} may be relaxed in a future version
5194 of the C++ standard.
5196 @item -Wno-int-to-pointer-cast
5197 @opindex Wno-int-to-pointer-cast
5198 @opindex Wint-to-pointer-cast
5199 Suppress warnings from casts to pointer type of an integer of a
5200 different size. In C++, casting to a pointer type of smaller size is
5201 an error. @option{Wint-to-pointer-cast} is enabled by default.
5204 When importing auxiliary modules during profile-use, check current
5205 memory consumption after parsing each auxiliary module. If it exceeds
5206 this limit (specified in kb), don't import any more auxiliary modules.
5207 Specifying a value of 0 means don't enforce this limit. This parameter
5208 is only useful when using @option{-fprofile-use} and @option{-fripa}.
5210 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
5211 @opindex Wno-pointer-to-int-cast
5212 @opindex Wpointer-to-int-cast
5213 Suppress warnings from casts from a pointer to an integer type of a
5217 @opindex Winvalid-pch
5218 @opindex Wno-invalid-pch
5219 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
5220 the search path but can't be used.
5224 @opindex Wno-long-long
5225 Warn if @samp{long long} type is used. This is enabled by either
5226 @option{-Wpedantic} or @option{-Wtraditional} in ISO C90 and C++98
5227 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
5229 @item -Wvariadic-macros
5230 @opindex Wvariadic-macros
5231 @opindex Wno-variadic-macros
5232 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
5233 alternate syntax when in pedantic ISO C99 mode. This is default.
5234 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
5238 @opindex Wno-varargs
5239 Warn upon questionable usage of the macros used to handle variable
5240 arguments like @samp{va_start}. This is default. To inhibit the
5241 warning messages, use @option{-Wno-varargs}.
5243 @item -Wvector-operation-performance
5244 @opindex Wvector-operation-performance
5245 @opindex Wno-vector-operation-performance
5246 Warn if vector operation is not implemented via SIMD capabilities of the
5247 architecture. Mainly useful for the performance tuning.
5248 Vector operation can be implemented @code{piecewise}, which means that the
5249 scalar operation is performed on every vector element;
5250 @code{in parallel}, which means that the vector operation is implemented
5251 using scalars of wider type, which normally is more performance efficient;
5252 and @code{as a single scalar}, which means that vector fits into a
5255 @item -Wno-virtual-move-assign
5256 @opindex Wvirtual-move-assign
5257 @opindex Wno-virtual-move-assign
5258 Suppress warnings about inheriting from a virtual base with a
5259 non-trivial C++11 move assignment operator. This is dangerous because
5260 if the virtual base is reachable along more than one path, it will be
5261 moved multiple times, which can mean both objects end up in the
5262 moved-from state. If the move assignment operator is written to avoid
5263 moving from a moved-from object, this warning can be disabled.
5268 Warn if variable length array is used in the code.
5269 @option{-Wno-vla} prevents the @option{-Wpedantic} warning of
5270 the variable length array.
5272 @item -Wvolatile-register-var
5273 @opindex Wvolatile-register-var
5274 @opindex Wno-volatile-register-var
5275 Warn if a register variable is declared volatile. The volatile
5276 modifier does not inhibit all optimizations that may eliminate reads
5277 and/or writes to register variables. This warning is enabled by
5280 @item -Wdisabled-optimization
5281 @opindex Wdisabled-optimization
5282 @opindex Wno-disabled-optimization
5283 Warn if a requested optimization pass is disabled. This warning does
5284 not generally indicate that there is anything wrong with your code; it
5285 merely indicates that GCC's optimizers are unable to handle the code
5286 effectively. Often, the problem is that your code is too big or too
5287 complex; GCC refuses to optimize programs when the optimization
5288 itself is likely to take inordinate amounts of time.
5290 @item -Wpointer-sign @r{(C and Objective-C only)}
5291 @opindex Wpointer-sign
5292 @opindex Wno-pointer-sign
5293 Warn for pointer argument passing or assignment with different signedness.
5294 This option is only supported for C and Objective-C@. It is implied by
5295 @option{-Wall} and by @option{-Wpedantic}, which can be disabled with
5296 @option{-Wno-pointer-sign}.
5298 @item -Wstack-protector
5299 @opindex Wstack-protector
5300 @opindex Wno-stack-protector
5301 This option is only active when @option{-fstack-protector} is active. It
5302 warns about functions that are not protected against stack smashing.
5304 @item -Woverlength-strings
5305 @opindex Woverlength-strings
5306 @opindex Wno-overlength-strings
5307 Warn about string constants that are longer than the ``minimum
5308 maximum'' length specified in the C standard. Modern compilers
5309 generally allow string constants that are much longer than the
5310 standard's minimum limit, but very portable programs should avoid
5311 using longer strings.
5313 The limit applies @emph{after} string constant concatenation, and does
5314 not count the trailing NUL@. In C90, the limit was 509 characters; in
5315 C99, it was raised to 4095. C++98 does not specify a normative
5316 minimum maximum, so we do not diagnose overlength strings in C++@.
5318 This option is implied by @option{-Wpedantic}, and can be disabled with
5319 @option{-Wno-overlength-strings}.
5321 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
5322 @opindex Wunsuffixed-float-constants
5324 Issue a warning for any floating constant that does not have
5325 a suffix. When used together with @option{-Wsystem-headers} it
5326 warns about such constants in system header files. This can be useful
5327 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
5328 from the decimal floating-point extension to C99.
5330 @item -Wno-designated-init @r{(C and Objective-C only)}
5331 Suppress warnings when a positional initializer is used to initialize
5332 a structure that has been marked with the @code{designated_init}
5337 @node Debugging Options
5338 @section Options for Debugging Your Program or GCC
5339 @cindex options, debugging
5340 @cindex debugging information options
5342 GCC has various special options that are used for debugging
5343 either your program or GCC:
5348 Produce debugging information in the operating system's native format
5349 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
5352 On most systems that use stabs format, @option{-g} enables use of extra
5353 debugging information that only GDB can use; this extra information
5354 makes debugging work better in GDB but probably makes other debuggers
5356 refuse to read the program. If you want to control for certain whether
5357 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
5358 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
5360 GCC allows you to use @option{-g} with
5361 @option{-O}. The shortcuts taken by optimized code may occasionally
5362 produce surprising results: some variables you declared may not exist
5363 at all; flow of control may briefly move where you did not expect it;
5364 some statements may not be executed because they compute constant
5365 results or their values are already at hand; some statements may
5366 execute in different places because they have been moved out of loops.
5368 Nevertheless it proves possible to debug optimized output. This makes
5369 it reasonable to use the optimizer for programs that might have bugs.
5371 The following options are useful when GCC is generated with the
5372 capability for more than one debugging format.
5375 @opindex gsplit-dwarf
5376 Separate as much dwarf debugging information as possible into a
5377 separate output file with the extension .dwo. This option allows
5378 the build system to avoid linking files with debug information. To
5379 be useful, this option requires a debugger capable of reading .dwo
5384 Produce debugging information for use by GDB@. This means to use the
5385 most expressive format available (DWARF 2, stabs, or the native format
5386 if neither of those are supported), including GDB extensions if at all
5391 Generate dwarf .debug_pubnames and .debug_pubtypes sections.
5393 @item -ggnu-pubnames
5394 @opindex ggnu-pubnames
5395 Generate .debug_pubnames and .debug_pubtypes sections in a format
5396 suitable for conversion into a GDB@ index. This option is only useful
5397 with a linker that can produce GDB@ index version 7.
5401 Produce debugging information in stabs format (if that is supported),
5402 without GDB extensions. This is the format used by DBX on most BSD
5403 systems. On MIPS, Alpha and System V Release 4 systems this option
5404 produces stabs debugging output that is not understood by DBX or SDB@.
5405 On System V Release 4 systems this option requires the GNU assembler.
5407 @item -feliminate-unused-debug-symbols
5408 @opindex feliminate-unused-debug-symbols
5409 Produce debugging information in stabs format (if that is supported),
5410 for only symbols that are actually used.
5412 @item -femit-class-debug-always
5413 Instead of emitting debugging information for a C++ class in only one
5414 object file, emit it in all object files using the class. This option
5415 should be used only with debuggers that are unable to handle the way GCC
5416 normally emits debugging information for classes because using this
5417 option increases the size of debugging information by as much as a
5420 @item -fdebug-types-section
5421 @opindex fdebug-types-section
5422 @opindex fno-debug-types-section
5423 When using DWARF Version 4 or higher, type DIEs can be put into
5424 their own @code{.debug_types} section instead of making them part of the
5425 @code{.debug_info} section. It is more efficient to put them in a separate
5426 comdat sections since the linker can then remove duplicates.
5427 But not all DWARF consumers support @code{.debug_types} sections yet
5428 and on some objects @code{.debug_types} produces larger instead of smaller
5429 debugging information.
5433 Produce debugging information in stabs format (if that is supported),
5434 using GNU extensions understood only by the GNU debugger (GDB)@. The
5435 use of these extensions is likely to make other debuggers crash or
5436 refuse to read the program.
5440 Produce debugging information in COFF format (if that is supported).
5441 This is the format used by SDB on most System V systems prior to
5446 Produce debugging information in XCOFF format (if that is supported).
5447 This is the format used by the DBX debugger on IBM RS/6000 systems.
5451 Produce debugging information in XCOFF format (if that is supported),
5452 using GNU extensions understood only by the GNU debugger (GDB)@. The
5453 use of these extensions is likely to make other debuggers crash or
5454 refuse to read the program, and may cause assemblers other than the GNU
5455 assembler (GAS) to fail with an error.
5457 @item -gdwarf-@var{version}
5458 @opindex gdwarf-@var{version}
5459 Produce debugging information in DWARF format (if that is supported).
5460 The value of @var{version} may be either 2, 3 or 4; the default version
5461 for most targets is 4.
5463 Note that with DWARF Version 2, some ports require and always
5464 use some non-conflicting DWARF 3 extensions in the unwind tables.
5466 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
5467 for maximum benefit.
5469 @item -grecord-gcc-switches
5470 @opindex grecord-gcc-switches
5471 This switch causes the command-line options used to invoke the
5472 compiler that may affect code generation to be appended to the
5473 DW_AT_producer attribute in DWARF debugging information. The options
5474 are concatenated with spaces separating them from each other and from
5475 the compiler version. See also @option{-frecord-gcc-switches} for another
5476 way of storing compiler options into the object file. This is the default.
5478 @item -gno-record-gcc-switches
5479 @opindex gno-record-gcc-switches
5480 Disallow appending command-line options to the DW_AT_producer attribute
5481 in DWARF debugging information.
5483 @item -gstrict-dwarf
5484 @opindex gstrict-dwarf
5485 Disallow using extensions of later DWARF standard version than selected
5486 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
5487 DWARF extensions from later standard versions is allowed.
5489 @item -gno-strict-dwarf
5490 @opindex gno-strict-dwarf
5491 Allow using extensions of later DWARF standard version than selected with
5492 @option{-gdwarf-@var{version}}.
5494 @item -gz@r{[}=@var{type}@r{]}
5496 Produce compressed debug sections in DWARF format, if that is supported.
5497 If @var{type} is not given, the default type depends on the capabilities
5498 of the assembler and linker used. @var{type} may be one of
5499 @option{none} (don't compress debug sections), @option{zlib} (use zlib
5500 compression in ELF gABI format), or @option{zlib-gnu} (use zlib
5501 compression in traditional GNU format). If the linker doesn't support
5502 writing compressed debug sections, the option is rejected. Otherwise,
5503 if the assembler does not support them, @option{-gz} is silently ignored
5504 when producing object files.
5508 Produce debugging information in Alpha/VMS debug format (if that is
5509 supported). This is the format used by DEBUG on Alpha/VMS systems.
5512 @itemx -ggdb@var{level}
5513 @itemx -gstabs@var{level}
5514 @itemx -gcoff@var{level}
5515 @itemx -gxcoff@var{level}
5516 @itemx -gvms@var{level}
5517 Request debugging information and also use @var{level} to specify how
5518 much information. The default level is 2.
5520 Level 0 produces no debug information at all. Thus, @option{-g0} negates
5523 Level 1 produces minimal information, enough for making backtraces in
5524 parts of the program that you don't plan to debug. This includes
5525 descriptions of functions and external variables, and line number
5526 tables, but no information about local variables.
5528 Level 3 includes extra information, such as all the macro definitions
5529 present in the program. Some debuggers support macro expansion when
5530 you use @option{-g3}.
5532 @option{-gdwarf-2} does not accept a concatenated debug level, because
5533 GCC used to support an option @option{-gdwarf} that meant to generate
5534 debug information in version 1 of the DWARF format (which is very
5535 different from version 2), and it would have been too confusing. That
5536 debug format is long obsolete, but the option cannot be changed now.
5537 Instead use an additional @option{-g@var{level}} option to change the
5538 debug level for DWARF.
5542 Turn off generation of debug info, if leaving out this option
5543 generates it, or turn it on at level 2 otherwise. The position of this
5544 argument in the command line does not matter; it takes effect after all
5545 other options are processed, and it does so only once, no matter how
5546 many times it is given. This is mainly intended to be used with
5547 @option{-fcompare-debug}.
5549 @item -fsanitize=address
5550 @opindex fsanitize=address
5551 Enable AddressSanitizer, a fast memory error detector.
5552 Memory access instructions will be instrumented to detect
5553 out-of-bounds and use-after-free bugs.
5554 See @uref{http://code.google.com/p/address-sanitizer/} for
5555 more details. The run-time behavior can be influenced using the
5556 @env{ASAN_OPTIONS} environment variable; see
5557 @url{https://code.google.com/p/address-sanitizer/wiki/Flags#Run-time_flags} for
5558 a list of supported options.
5560 @item -fsanitize=kernel-address
5561 @opindex fsanitize=kernel-address
5562 Enable AddressSanitizer for Linux kernel.
5563 See @uref{http://code.google.com/p/address-sanitizer/wiki/AddressSanitizerForKernel} for more details.
5565 @item -fsanitize=thread
5566 @opindex fsanitize=thread
5567 Enable ThreadSanitizer, a fast data race detector.
5568 Memory access instructions will be instrumented to detect
5569 data race bugs. See @uref{http://code.google.com/p/thread-sanitizer/} for more
5570 details. The run-time behavior can be influenced using the @env{TSAN_OPTIONS}
5571 environment variable; see
5572 @url{https://code.google.com/p/thread-sanitizer/wiki/Flags} for a list of
5575 @item -fsanitize=leak
5576 @opindex fsanitize=leak
5577 Enable LeakSanitizer, a memory leak detector.
5578 This option only matters for linking of executables and if neither
5579 @option{-fsanitize=address} nor @option{-fsanitize=thread} is used. In that
5580 case it will link the executable against a library that overrides @code{malloc}
5581 and other allocator functions. See
5582 @uref{https://code.google.com/p/address-sanitizer/wiki/LeakSanitizer} for more
5583 details. The run-time behavior can be influenced using the
5584 @env{LSAN_OPTIONS} environment variable.
5586 @item -fsanitize=undefined
5587 @opindex fsanitize=undefined
5588 Enable UndefinedBehaviorSanitizer, a fast undefined behavior detector.
5589 Various computations will be instrumented to detect undefined behavior
5590 at runtime. Current suboptions are:
5594 @item -fsanitize=shift
5595 @opindex fsanitize=shift
5596 This option enables checking that the result of a shift operation is
5597 not undefined. Note that what exactly is considered undefined differs
5598 slightly between C and C++, as well as between ISO C90 and C99, etc.
5600 @item -fsanitize=integer-divide-by-zero
5601 @opindex fsanitize=integer-divide-by-zero
5602 Detect integer division by zero as well as @code{INT_MIN / -1} division.
5604 @item -fsanitize=unreachable
5605 @opindex fsanitize=unreachable
5606 With this option, the compiler will turn the @code{__builtin_unreachable}
5607 call into a diagnostics message call instead. When reaching the
5608 @code{__builtin_unreachable} call, the behavior is undefined.
5610 @item -fsanitize=vla-bound
5611 @opindex fsanitize=vla-bound
5612 This option instructs the compiler to check that the size of a variable
5613 length array is positive. This option does not have any effect in
5614 @option{-std=c++1y} mode, as the standard requires the exception be thrown
5617 @item -fsanitize=null
5618 @opindex fsanitize=null
5619 This option enables pointer checking. Particularly, the application
5620 built with this option turned on will issue an error message when it
5621 tries to dereference a NULL pointer, or if a reference (possibly an
5622 rvalue reference) is bound to a NULL pointer, or if a method is invoked
5623 on an object pointed by a NULL pointer.
5625 @item -fsanitize=return
5626 @opindex fsanitize=return
5627 This option enables return statement checking. Programs
5628 built with this option turned on will issue an error message
5629 when the end of a non-void function is reached without actually
5630 returning a value. This option works in C++ only.
5632 @item -fsanitize=signed-integer-overflow
5633 @opindex fsanitize=signed-integer-overflow
5634 This option enables signed integer overflow checking. We check that
5635 the result of @code{+}, @code{*}, and both unary and binary @code{-}
5636 does not overflow in the signed arithmetics. Note, integer promotion
5637 rules must be taken into account. That is, the following is not an
5640 signed char a = SCHAR_MAX;
5644 @item -fsanitize=bounds
5645 @opindex fsanitize=bounds
5646 This option enables instrumentation of array bounds. Various out of bounds
5647 accesses are detected. Flexible array members and initializers of variables
5648 with static storage are not instrumented.
5650 @item -fsanitize=alignment
5651 @opindex fsanitize=alignment
5653 This option enables checking of alignment of pointers when they are
5654 dereferenced, or when a reference is bound to insufficiently aligned target,
5655 or when a method or constructor is invoked on insufficiently aligned object.
5657 @item -fsanitize=float-divide-by-zero
5658 @opindex fsanitize=float-divide-by-zero
5659 Detect floating-point division by zero. Unlike other similar options,
5660 @option{-fsanitize=float-divide-by-zero} is not enabled by
5661 @option{-fsanitize=undefined}, since floating-point division by zero can
5662 be a legitimate way of obtaining infinities and NaNs.
5664 @item -fsanitize=float-cast-overflow
5665 @opindex fsanitize=float-cast-overflow
5666 This option enables floating-point type to integer conversion checking.
5667 We check that the result of the conversion does not overflow.
5668 This option does not work well with @code{FE_INVALID} exceptions enabled.
5672 While @option{-ftrapv} causes traps for signed overflows to be emitted,
5673 @option{-fsanitize=undefined} gives a diagnostic message.
5674 This currently works only for the C family of languages.
5676 @item -fsanitize-recover
5677 @opindex fsanitize-recover
5678 By default @option{-fsanitize=undefined} sanitization (and its suboptions
5679 except for @option{-fsanitize=unreachable} and @option{-fsanitize=return})
5680 after reporting undefined behavior attempts to continue running the
5681 program as if no undefined behavior happened. This means multiple undefined
5682 behavior runtime errors can be reported in a single program run, and the exit
5683 code of the program may indicate success even when undefined behavior
5684 has been reported. The @option{-fno-sanitize-recover} can be used to alter
5685 this behavior, only the first detected undefined behavior will be reported
5686 and program will exit after that with non-zero exit code.
5688 @item -fsanitize-undefined-trap-on-error
5689 @opindex fsanitize-undefined-trap-on-error
5690 The @option{-fsanitize-undefined-trap-on-error} instructs the compiler to
5691 report undefined behavior using @code{__builtin_trap ()} rather than
5692 a @code{libubsan} library routine. The advantage of this is that the
5693 @code{libubsan} library is not needed and will not be linked in, so this
5694 is usable even for use in freestanding environments.
5696 @item -fdump-final-insns@r{[}=@var{file}@r{]}
5697 @opindex fdump-final-insns
5698 Dump the final internal representation (RTL) to @var{file}. If the
5699 optional argument is omitted (or if @var{file} is @code{.}), the name
5700 of the dump file is determined by appending @code{.gkd} to the
5701 compilation output file name.
5703 @item -fcompare-debug@r{[}=@var{opts}@r{]}
5704 @opindex fcompare-debug
5705 @opindex fno-compare-debug
5706 If no error occurs during compilation, run the compiler a second time,
5707 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
5708 passed to the second compilation. Dump the final internal
5709 representation in both compilations, and print an error if they differ.
5711 If the equal sign is omitted, the default @option{-gtoggle} is used.
5713 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
5714 and nonzero, implicitly enables @option{-fcompare-debug}. If
5715 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
5716 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
5719 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
5720 is equivalent to @option{-fno-compare-debug}, which disables the dumping
5721 of the final representation and the second compilation, preventing even
5722 @env{GCC_COMPARE_DEBUG} from taking effect.
5724 To verify full coverage during @option{-fcompare-debug} testing, set
5725 @env{GCC_COMPARE_DEBUG} to say @samp{-fcompare-debug-not-overridden},
5726 which GCC rejects as an invalid option in any actual compilation
5727 (rather than preprocessing, assembly or linking). To get just a
5728 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
5729 not overridden} will do.
5731 @item -fcompare-debug-second
5732 @opindex fcompare-debug-second
5733 This option is implicitly passed to the compiler for the second
5734 compilation requested by @option{-fcompare-debug}, along with options to
5735 silence warnings, and omitting other options that would cause
5736 side-effect compiler outputs to files or to the standard output. Dump
5737 files and preserved temporary files are renamed so as to contain the
5738 @code{.gk} additional extension during the second compilation, to avoid
5739 overwriting those generated by the first.
5741 When this option is passed to the compiler driver, it causes the
5742 @emph{first} compilation to be skipped, which makes it useful for little
5743 other than debugging the compiler proper.
5745 @item -feliminate-dwarf2-dups
5746 @opindex feliminate-dwarf2-dups
5747 Compress DWARF 2 debugging information by eliminating duplicated
5748 information about each symbol. This option only makes sense when
5749 generating DWARF 2 debugging information with @option{-gdwarf-2}.
5751 @item -femit-struct-debug-baseonly
5752 @opindex femit-struct-debug-baseonly
5753 Emit debug information for struct-like types
5754 only when the base name of the compilation source file
5755 matches the base name of file in which the struct is defined.
5757 This option substantially reduces the size of debugging information,
5758 but at significant potential loss in type information to the debugger.
5759 See @option{-femit-struct-debug-reduced} for a less aggressive option.
5760 See @option{-femit-struct-debug-detailed} for more detailed control.
5762 This option works only with DWARF 2.
5764 @item -femit-struct-debug-reduced
5765 @opindex femit-struct-debug-reduced
5766 Emit debug information for struct-like types
5767 only when the base name of the compilation source file
5768 matches the base name of file in which the type is defined,
5769 unless the struct is a template or defined in a system header.
5771 This option significantly reduces the size of debugging information,
5772 with some potential loss in type information to the debugger.
5773 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
5774 See @option{-femit-struct-debug-detailed} for more detailed control.
5776 This option works only with DWARF 2.
5778 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
5779 Specify the struct-like types
5780 for which the compiler generates debug information.
5781 The intent is to reduce duplicate struct debug information
5782 between different object files within the same program.
5784 This option is a detailed version of
5785 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
5786 which serves for most needs.
5788 A specification has the syntax@*
5789 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
5791 The optional first word limits the specification to
5792 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
5793 A struct type is used directly when it is the type of a variable, member.
5794 Indirect uses arise through pointers to structs.
5795 That is, when use of an incomplete struct is valid, the use is indirect.
5797 @samp{struct one direct; struct two * indirect;}.
5799 The optional second word limits the specification to
5800 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
5801 Generic structs are a bit complicated to explain.
5802 For C++, these are non-explicit specializations of template classes,
5803 or non-template classes within the above.
5804 Other programming languages have generics,
5805 but @option{-femit-struct-debug-detailed} does not yet implement them.
5807 The third word specifies the source files for those
5808 structs for which the compiler should emit debug information.
5809 The values @samp{none} and @samp{any} have the normal meaning.
5810 The value @samp{base} means that
5811 the base of name of the file in which the type declaration appears
5812 must match the base of the name of the main compilation file.
5813 In practice, this means that when compiling @file{foo.c}, debug information
5814 is generated for types declared in that file and @file{foo.h},
5815 but not other header files.
5816 The value @samp{sys} means those types satisfying @samp{base}
5817 or declared in system or compiler headers.
5819 You may need to experiment to determine the best settings for your application.
5821 The default is @option{-femit-struct-debug-detailed=all}.
5823 This option works only with DWARF 2.
5825 @item -fno-merge-debug-strings
5826 @opindex fmerge-debug-strings
5827 @opindex fno-merge-debug-strings
5828 Direct the linker to not merge together strings in the debugging
5829 information that are identical in different object files. Merging is
5830 not supported by all assemblers or linkers. Merging decreases the size
5831 of the debug information in the output file at the cost of increasing
5832 link processing time. Merging is enabled by default.
5834 @item -fdebug-prefix-map=@var{old}=@var{new}
5835 @opindex fdebug-prefix-map
5836 When compiling files in directory @file{@var{old}}, record debugging
5837 information describing them as in @file{@var{new}} instead.
5839 @item -fno-dwarf2-cfi-asm
5840 @opindex fdwarf2-cfi-asm
5841 @opindex fno-dwarf2-cfi-asm
5842 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
5843 instead of using GAS @code{.cfi_*} directives.
5845 @cindex @command{prof}
5848 Generate extra code to write profile information suitable for the
5849 analysis program @command{prof}. You must use this option when compiling
5850 the source files you want data about, and you must also use it when
5853 @cindex @command{gprof}
5856 Generate extra code to write profile information suitable for the
5857 analysis program @command{gprof}. You must use this option when compiling
5858 the source files you want data about, and you must also use it when
5863 Makes the compiler print out each function name as it is compiled, and
5864 print some statistics about each pass when it finishes.
5867 @opindex ftime-report
5868 Makes the compiler print some statistics about the time consumed by each
5869 pass when it finishes.
5872 @opindex fmem-report
5873 Makes the compiler print some statistics about permanent memory
5874 allocation when it finishes.
5876 @item -fmem-report-wpa
5877 @opindex fmem-report-wpa
5878 Makes the compiler print some statistics about permanent memory
5879 allocation for the WPA phase only.
5881 @item -fpre-ipa-mem-report
5882 @opindex fpre-ipa-mem-report
5883 @item -fpost-ipa-mem-report
5884 @opindex fpost-ipa-mem-report
5885 Makes the compiler print some statistics about permanent memory
5886 allocation before or after interprocedural optimization.
5888 @item -fprofile-report
5889 @opindex fprofile-report
5890 Makes the compiler print some statistics about consistency of the
5891 (estimated) profile and effect of individual passes.
5894 @opindex fstack-usage
5895 Makes the compiler output stack usage information for the program, on a
5896 per-function basis. The filename for the dump is made by appending
5897 @file{.su} to the @var{auxname}. @var{auxname} is generated from the name of
5898 the output file, if explicitly specified and it is not an executable,
5899 otherwise it is the basename of the source file. An entry is made up
5904 The name of the function.
5908 One or more qualifiers: @code{static}, @code{dynamic}, @code{bounded}.
5911 The qualifier @code{static} means that the function manipulates the stack
5912 statically: a fixed number of bytes are allocated for the frame on function
5913 entry and released on function exit; no stack adjustments are otherwise made
5914 in the function. The second field is this fixed number of bytes.
5916 The qualifier @code{dynamic} means that the function manipulates the stack
5917 dynamically: in addition to the static allocation described above, stack
5918 adjustments are made in the body of the function, for example to push/pop
5919 arguments around function calls. If the qualifier @code{bounded} is also
5920 present, the amount of these adjustments is bounded at compile time and
5921 the second field is an upper bound of the total amount of stack used by
5922 the function. If it is not present, the amount of these adjustments is
5923 not bounded at compile time and the second field only represents the
5926 @item -fprofile-arcs
5927 @opindex fprofile-arcs
5928 Add code so that program flow @dfn{arcs} are instrumented. During
5929 execution the program records how many times each branch and call is
5930 executed and how many times it is taken or returns. When the compiled
5931 program exits it saves this data to a file called
5932 @file{@var{auxname}.gcda} for each source file. The data may be used for
5933 profile-directed optimizations (@option{-fbranch-probabilities}), or for
5934 test coverage analysis (@option{-ftest-coverage}). Each object file's
5935 @var{auxname} is generated from the name of the output file, if
5936 explicitly specified and it is not the final executable, otherwise it is
5937 the basename of the source file. In both cases any suffix is removed
5938 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
5939 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
5940 @xref{Cross-profiling}.
5942 @cindex @command{gcov}
5946 This option is used to compile and link code instrumented for coverage
5947 analysis. The option is a synonym for @option{-fprofile-arcs}
5948 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
5949 linking). See the documentation for those options for more details.
5954 Compile the source files with @option{-fprofile-arcs} plus optimization
5955 and code generation options. For test coverage analysis, use the
5956 additional @option{-ftest-coverage} option. You do not need to profile
5957 every source file in a program.
5960 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
5961 (the latter implies the former).
5964 Run the program on a representative workload to generate the arc profile
5965 information. This may be repeated any number of times. You can run
5966 concurrent instances of your program, and provided that the file system
5967 supports locking, the data files will be correctly updated. Also
5968 @code{fork} calls are detected and correctly handled (double counting
5972 For profile-directed optimizations, compile the source files again with
5973 the same optimization and code generation options plus
5974 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
5975 Control Optimization}).
5978 For test coverage analysis, use @command{gcov} to produce human readable
5979 information from the @file{.gcno} and @file{.gcda} files. Refer to the
5980 @command{gcov} documentation for further information.
5984 With @option{-fprofile-arcs}, for each function of your program GCC
5985 creates a program flow graph, then finds a spanning tree for the graph.
5986 Only arcs that are not on the spanning tree have to be instrumented: the
5987 compiler adds code to count the number of times that these arcs are
5988 executed. When an arc is the only exit or only entrance to a block, the
5989 instrumentation code can be added to the block; otherwise, a new basic
5990 block must be created to hold the instrumentation code.
5993 @item -ftest-coverage
5994 @opindex ftest-coverage
5995 Produce a notes file that the @command{gcov} code-coverage utility
5996 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
5997 show program coverage. Each source file's note file is called
5998 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
5999 above for a description of @var{auxname} and instructions on how to
6000 generate test coverage data. Coverage data matches the source files
6001 more closely if you do not optimize.
6003 @item -fdbg-cnt-list
6004 @opindex fdbg-cnt-list
6005 Print the name and the counter upper bound for all debug counters.
6008 @item -fdbg-cnt=@var{counter-value-list}
6010 Set the internal debug counter upper bound. @var{counter-value-list}
6011 is a comma-separated list of @var{name}:@var{value} pairs
6012 which sets the upper bound of each debug counter @var{name} to @var{value}.
6013 All debug counters have the initial upper bound of @code{UINT_MAX};
6014 thus @code{dbg_cnt()} returns true always unless the upper bound
6015 is set by this option.
6016 For example, with @option{-fdbg-cnt=dce:10,tail_call:0},
6017 @code{dbg_cnt(dce)} returns true only for first 10 invocations.
6019 @item -fenable-@var{kind}-@var{pass}
6020 @itemx -fdisable-@var{kind}-@var{pass}=@var{range-list}
6024 This is a set of options that are used to explicitly disable/enable
6025 optimization passes. These options are intended for use for debugging GCC.
6026 Compiler users should use regular options for enabling/disabling
6031 @item -fdisable-ipa-@var{pass}
6032 Disable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
6033 statically invoked in the compiler multiple times, the pass name should be
6034 appended with a sequential number starting from 1.
6036 @item -fdisable-rtl-@var{pass}
6037 @itemx -fdisable-rtl-@var{pass}=@var{range-list}
6038 Disable RTL pass @var{pass}. @var{pass} is the pass name. If the same pass is
6039 statically invoked in the compiler multiple times, the pass name should be
6040 appended with a sequential number starting from 1. @var{range-list} is a
6041 comma-separated list of function ranges or assembler names. Each range is a number
6042 pair separated by a colon. The range is inclusive in both ends. If the range
6043 is trivial, the number pair can be simplified as a single number. If the
6044 function's call graph node's @var{uid} falls within one of the specified ranges,
6045 the @var{pass} is disabled for that function. The @var{uid} is shown in the
6046 function header of a dump file, and the pass names can be dumped by using
6047 option @option{-fdump-passes}.
6049 @item -fdisable-tree-@var{pass}
6050 @itemx -fdisable-tree-@var{pass}=@var{range-list}
6051 Disable tree pass @var{pass}. See @option{-fdisable-rtl} for the description of
6054 @item -fenable-ipa-@var{pass}
6055 Enable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
6056 statically invoked in the compiler multiple times, the pass name should be
6057 appended with a sequential number starting from 1.
6059 @item -fenable-rtl-@var{pass}
6060 @itemx -fenable-rtl-@var{pass}=@var{range-list}
6061 Enable RTL pass @var{pass}. See @option{-fdisable-rtl} for option argument
6062 description and examples.
6064 @item -fenable-tree-@var{pass}
6065 @itemx -fenable-tree-@var{pass}=@var{range-list}
6066 Enable tree pass @var{pass}. See @option{-fdisable-rtl} for the description
6067 of option arguments.
6071 Here are some examples showing uses of these options.
6075 # disable ccp1 for all functions
6077 # disable complete unroll for function whose cgraph node uid is 1
6078 -fenable-tree-cunroll=1
6079 # disable gcse2 for functions at the following ranges [1,1],
6080 # [300,400], and [400,1000]
6081 # disable gcse2 for functions foo and foo2
6082 -fdisable-rtl-gcse2=foo,foo2
6083 # disable early inlining
6084 -fdisable-tree-einline
6085 # disable ipa inlining
6086 -fdisable-ipa-inline
6087 # enable tree full unroll
6088 -fenable-tree-unroll
6092 @item -d@var{letters}
6093 @itemx -fdump-rtl-@var{pass}
6094 @itemx -fdump-rtl-@var{pass}=@var{filename}
6096 @opindex fdump-rtl-@var{pass}
6097 Says to make debugging dumps during compilation at times specified by
6098 @var{letters}. This is used for debugging the RTL-based passes of the
6099 compiler. The file names for most of the dumps are made by appending
6100 a pass number and a word to the @var{dumpname}, and the files are
6101 created in the directory of the output file. In case of
6102 @option{=@var{filename}} option, the dump is output on the given file
6103 instead of the pass numbered dump files. Note that the pass number is
6104 computed statically as passes get registered into the pass manager.
6105 Thus the numbering is not related to the dynamic order of execution of
6106 passes. In particular, a pass installed by a plugin could have a
6107 number over 200 even if it executed quite early. @var{dumpname} is
6108 generated from the name of the output file, if explicitly specified
6109 and it is not an executable, otherwise it is the basename of the
6110 source file. These switches may have different effects when
6111 @option{-E} is used for preprocessing.
6113 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
6114 @option{-d} option @var{letters}. Here are the possible
6115 letters for use in @var{pass} and @var{letters}, and their meanings:
6119 @item -fdump-rtl-alignments
6120 @opindex fdump-rtl-alignments
6121 Dump after branch alignments have been computed.
6123 @item -fdump-rtl-asmcons
6124 @opindex fdump-rtl-asmcons
6125 Dump after fixing rtl statements that have unsatisfied in/out constraints.
6127 @item -fdump-rtl-auto_inc_dec
6128 @opindex fdump-rtl-auto_inc_dec
6129 Dump after auto-inc-dec discovery. This pass is only run on
6130 architectures that have auto inc or auto dec instructions.
6132 @item -fdump-rtl-barriers
6133 @opindex fdump-rtl-barriers
6134 Dump after cleaning up the barrier instructions.
6136 @item -fdump-rtl-bbpart
6137 @opindex fdump-rtl-bbpart
6138 Dump after partitioning hot and cold basic blocks.
6140 @item -fdump-rtl-bbro
6141 @opindex fdump-rtl-bbro
6142 Dump after block reordering.
6144 @item -fdump-rtl-btl1
6145 @itemx -fdump-rtl-btl2
6146 @opindex fdump-rtl-btl2
6147 @opindex fdump-rtl-btl2
6148 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
6149 after the two branch
6150 target load optimization passes.
6152 @item -fdump-rtl-bypass
6153 @opindex fdump-rtl-bypass
6154 Dump after jump bypassing and control flow optimizations.
6156 @item -fdump-rtl-combine
6157 @opindex fdump-rtl-combine
6158 Dump after the RTL instruction combination pass.
6160 @item -fdump-rtl-compgotos
6161 @opindex fdump-rtl-compgotos
6162 Dump after duplicating the computed gotos.
6164 @item -fdump-rtl-ce1
6165 @itemx -fdump-rtl-ce2
6166 @itemx -fdump-rtl-ce3
6167 @opindex fdump-rtl-ce1
6168 @opindex fdump-rtl-ce2
6169 @opindex fdump-rtl-ce3
6170 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
6171 @option{-fdump-rtl-ce3} enable dumping after the three
6172 if conversion passes.
6174 @item -fdump-rtl-cprop_hardreg
6175 @opindex fdump-rtl-cprop_hardreg
6176 Dump after hard register copy propagation.
6178 @item -fdump-rtl-csa
6179 @opindex fdump-rtl-csa
6180 Dump after combining stack adjustments.
6182 @item -fdump-rtl-cse1
6183 @itemx -fdump-rtl-cse2
6184 @opindex fdump-rtl-cse1
6185 @opindex fdump-rtl-cse2
6186 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
6187 the two common subexpression elimination passes.
6189 @item -fdump-rtl-dce
6190 @opindex fdump-rtl-dce
6191 Dump after the standalone dead code elimination passes.
6193 @item -fdump-rtl-dbr
6194 @opindex fdump-rtl-dbr
6195 Dump after delayed branch scheduling.
6197 @item -fdump-rtl-dce1
6198 @itemx -fdump-rtl-dce2
6199 @opindex fdump-rtl-dce1
6200 @opindex fdump-rtl-dce2
6201 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
6202 the two dead store elimination passes.
6205 @opindex fdump-rtl-eh
6206 Dump after finalization of EH handling code.
6208 @item -fdump-rtl-eh_ranges
6209 @opindex fdump-rtl-eh_ranges
6210 Dump after conversion of EH handling range regions.
6212 @item -fdump-rtl-expand
6213 @opindex fdump-rtl-expand
6214 Dump after RTL generation.
6216 @item -fdump-rtl-fwprop1
6217 @itemx -fdump-rtl-fwprop2
6218 @opindex fdump-rtl-fwprop1
6219 @opindex fdump-rtl-fwprop2
6220 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
6221 dumping after the two forward propagation passes.
6223 @item -fdump-rtl-gcse1
6224 @itemx -fdump-rtl-gcse2
6225 @opindex fdump-rtl-gcse1
6226 @opindex fdump-rtl-gcse2
6227 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
6228 after global common subexpression elimination.
6230 @item -fdump-rtl-init-regs
6231 @opindex fdump-rtl-init-regs
6232 Dump after the initialization of the registers.
6234 @item -fdump-rtl-initvals
6235 @opindex fdump-rtl-initvals
6236 Dump after the computation of the initial value sets.
6238 @item -fdump-rtl-into_cfglayout
6239 @opindex fdump-rtl-into_cfglayout
6240 Dump after converting to cfglayout mode.
6242 @item -fdump-rtl-ira
6243 @opindex fdump-rtl-ira
6244 Dump after iterated register allocation.
6246 @item -fdump-rtl-jump
6247 @opindex fdump-rtl-jump
6248 Dump after the second jump optimization.
6250 @item -fdump-rtl-loop2
6251 @opindex fdump-rtl-loop2
6252 @option{-fdump-rtl-loop2} enables dumping after the rtl
6253 loop optimization passes.
6255 @item -fdump-rtl-mach
6256 @opindex fdump-rtl-mach
6257 Dump after performing the machine dependent reorganization pass, if that
6260 @item -fdump-rtl-mode_sw
6261 @opindex fdump-rtl-mode_sw
6262 Dump after removing redundant mode switches.
6264 @item -fdump-rtl-rnreg
6265 @opindex fdump-rtl-rnreg
6266 Dump after register renumbering.
6268 @item -fdump-rtl-outof_cfglayout
6269 @opindex fdump-rtl-outof_cfglayout
6270 Dump after converting from cfglayout mode.
6272 @item -fdump-rtl-peephole2
6273 @opindex fdump-rtl-peephole2
6274 Dump after the peephole pass.
6276 @item -fdump-rtl-postreload
6277 @opindex fdump-rtl-postreload
6278 Dump after post-reload optimizations.
6280 @item -fdump-rtl-pro_and_epilogue
6281 @opindex fdump-rtl-pro_and_epilogue
6282 Dump after generating the function prologues and epilogues.
6284 @item -fdump-rtl-sched1
6285 @itemx -fdump-rtl-sched2
6286 @opindex fdump-rtl-sched1
6287 @opindex fdump-rtl-sched2
6288 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
6289 after the basic block scheduling passes.
6291 @item -fdump-rtl-ree
6292 @opindex fdump-rtl-ree
6293 Dump after sign/zero extension elimination.
6295 @item -fdump-rtl-seqabstr
6296 @opindex fdump-rtl-seqabstr
6297 Dump after common sequence discovery.
6299 @item -fdump-rtl-shorten
6300 @opindex fdump-rtl-shorten
6301 Dump after shortening branches.
6303 @item -fdump-rtl-sibling
6304 @opindex fdump-rtl-sibling
6305 Dump after sibling call optimizations.
6307 @item -fdump-rtl-split1
6308 @itemx -fdump-rtl-split2
6309 @itemx -fdump-rtl-split3
6310 @itemx -fdump-rtl-split4
6311 @itemx -fdump-rtl-split5
6312 @opindex fdump-rtl-split1
6313 @opindex fdump-rtl-split2
6314 @opindex fdump-rtl-split3
6315 @opindex fdump-rtl-split4
6316 @opindex fdump-rtl-split5
6317 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
6318 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
6319 @option{-fdump-rtl-split5} enable dumping after five rounds of
6320 instruction splitting.
6322 @item -fdump-rtl-sms
6323 @opindex fdump-rtl-sms
6324 Dump after modulo scheduling. This pass is only run on some
6327 @item -fdump-rtl-stack
6328 @opindex fdump-rtl-stack
6329 Dump after conversion from GCC's ``flat register file'' registers to the
6330 x87's stack-like registers. This pass is only run on x86 variants.
6332 @item -fdump-rtl-subreg1
6333 @itemx -fdump-rtl-subreg2
6334 @opindex fdump-rtl-subreg1
6335 @opindex fdump-rtl-subreg2
6336 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
6337 the two subreg expansion passes.
6339 @item -fdump-rtl-unshare
6340 @opindex fdump-rtl-unshare
6341 Dump after all rtl has been unshared.
6343 @item -fdump-rtl-vartrack
6344 @opindex fdump-rtl-vartrack
6345 Dump after variable tracking.
6347 @item -fdump-rtl-vregs
6348 @opindex fdump-rtl-vregs
6349 Dump after converting virtual registers to hard registers.
6351 @item -fdump-rtl-web
6352 @opindex fdump-rtl-web
6353 Dump after live range splitting.
6355 @item -fdump-rtl-regclass
6356 @itemx -fdump-rtl-subregs_of_mode_init
6357 @itemx -fdump-rtl-subregs_of_mode_finish
6358 @itemx -fdump-rtl-dfinit
6359 @itemx -fdump-rtl-dfinish
6360 @opindex fdump-rtl-regclass
6361 @opindex fdump-rtl-subregs_of_mode_init
6362 @opindex fdump-rtl-subregs_of_mode_finish
6363 @opindex fdump-rtl-dfinit
6364 @opindex fdump-rtl-dfinish
6365 These dumps are defined but always produce empty files.
6368 @itemx -fdump-rtl-all
6370 @opindex fdump-rtl-all
6371 Produce all the dumps listed above.
6375 Annotate the assembler output with miscellaneous debugging information.
6379 Dump all macro definitions, at the end of preprocessing, in addition to
6384 Produce a core dump whenever an error occurs.
6388 Annotate the assembler output with a comment indicating which
6389 pattern and alternative is used. The length of each instruction is
6394 Dump the RTL in the assembler output as a comment before each instruction.
6395 Also turns on @option{-dp} annotation.
6399 Just generate RTL for a function instead of compiling it. Usually used
6400 with @option{-fdump-rtl-expand}.
6404 @opindex fdump-noaddr
6405 When doing debugging dumps, suppress address output. This makes it more
6406 feasible to use diff on debugging dumps for compiler invocations with
6407 different compiler binaries and/or different
6408 text / bss / data / heap / stack / dso start locations.
6410 @item -fdump-unnumbered
6411 @opindex fdump-unnumbered
6412 When doing debugging dumps, suppress instruction numbers and address output.
6413 This makes it more feasible to use diff on debugging dumps for compiler
6414 invocations with different options, in particular with and without
6417 @item -fdump-unnumbered-links
6418 @opindex fdump-unnumbered-links
6419 When doing debugging dumps (see @option{-d} option above), suppress
6420 instruction numbers for the links to the previous and next instructions
6423 @item -fdump-translation-unit @r{(C++ only)}
6424 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
6425 @opindex fdump-translation-unit
6426 Dump a representation of the tree structure for the entire translation
6427 unit to a file. The file name is made by appending @file{.tu} to the
6428 source file name, and the file is created in the same directory as the
6429 output file. If the @samp{-@var{options}} form is used, @var{options}
6430 controls the details of the dump as described for the
6431 @option{-fdump-tree} options.
6433 @item -fdump-class-hierarchy @r{(C++ only)}
6434 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
6435 @opindex fdump-class-hierarchy
6436 Dump a representation of each class's hierarchy and virtual function
6437 table layout to a file. The file name is made by appending
6438 @file{.class} to the source file name, and the file is created in the
6439 same directory as the output file. If the @samp{-@var{options}} form
6440 is used, @var{options} controls the details of the dump as described
6441 for the @option{-fdump-tree} options.
6443 @item -fdump-ipa-@var{switch}
6445 Control the dumping at various stages of inter-procedural analysis
6446 language tree to a file. The file name is generated by appending a
6447 switch specific suffix to the source file name, and the file is created
6448 in the same directory as the output file. The following dumps are
6453 Enables all inter-procedural analysis dumps.
6456 Dumps information about call-graph optimization, unused function removal,
6457 and inlining decisions.
6460 Dump after function inlining.
6465 @opindex fdump-passes
6466 Dump the list of optimization passes that are turned on and off by
6467 the current command-line options.
6469 @item -fdump-statistics-@var{option}
6470 @opindex fdump-statistics
6471 Enable and control dumping of pass statistics in a separate file. The
6472 file name is generated by appending a suffix ending in
6473 @samp{.statistics} to the source file name, and the file is created in
6474 the same directory as the output file. If the @samp{-@var{option}}
6475 form is used, @samp{-stats} causes counters to be summed over the
6476 whole compilation unit while @samp{-details} dumps every event as
6477 the passes generate them. The default with no option is to sum
6478 counters for each function compiled.
6480 @item -fdump-tree-@var{switch}
6481 @itemx -fdump-tree-@var{switch}-@var{options}
6482 @itemx -fdump-tree-@var{switch}-@var{options}=@var{filename}
6484 Control the dumping at various stages of processing the intermediate
6485 language tree to a file. The file name is generated by appending a
6486 switch-specific suffix to the source file name, and the file is
6487 created in the same directory as the output file. In case of
6488 @option{=@var{filename}} option, the dump is output on the given file
6489 instead of the auto named dump files. If the @samp{-@var{options}}
6490 form is used, @var{options} is a list of @samp{-} separated options
6491 which control the details of the dump. Not all options are applicable
6492 to all dumps; those that are not meaningful are ignored. The
6493 following options are available
6497 Print the address of each node. Usually this is not meaningful as it
6498 changes according to the environment and source file. Its primary use
6499 is for tying up a dump file with a debug environment.
6501 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
6502 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
6503 use working backward from mangled names in the assembly file.
6505 When dumping front-end intermediate representations, inhibit dumping
6506 of members of a scope or body of a function merely because that scope
6507 has been reached. Only dump such items when they are directly reachable
6510 When dumping pretty-printed trees, this option inhibits dumping the
6511 bodies of control structures.
6513 When dumping RTL, print the RTL in slim (condensed) form instead of
6514 the default LISP-like representation.
6516 Print a raw representation of the tree. By default, trees are
6517 pretty-printed into a C-like representation.
6519 Enable more detailed dumps (not honored by every dump option). Also
6520 include information from the optimization passes.
6522 Enable dumping various statistics about the pass (not honored by every dump
6525 Enable showing basic block boundaries (disabled in raw dumps).
6527 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
6528 dump a representation of the control flow graph suitable for viewing with
6529 GraphViz to @file{@var{file}.@var{passid}.@var{pass}.dot}. Each function in
6530 the file is pretty-printed as a subgraph, so that GraphViz can render them
6531 all in a single plot.
6533 This option currently only works for RTL dumps, and the RTL is always
6534 dumped in slim form.
6536 Enable showing virtual operands for every statement.
6538 Enable showing line numbers for statements.
6540 Enable showing the unique ID (@code{DECL_UID}) for each variable.
6542 Enable showing the tree dump for each statement.
6544 Enable showing the EH region number holding each statement.
6546 Enable showing scalar evolution analysis details.
6548 Enable showing optimization information (only available in certain
6551 Enable showing missed optimization information (only available in certain
6554 Enable other detailed optimization information (only available in
6556 @item =@var{filename}
6557 Instead of an auto named dump file, output into the given file
6558 name. The file names @file{stdout} and @file{stderr} are treated
6559 specially and are considered already open standard streams. For
6563 gcc -O2 -ftree-vectorize -fdump-tree-vect-blocks=foo.dump
6564 -fdump-tree-pre=stderr file.c
6567 outputs vectorizer dump into @file{foo.dump}, while the PRE dump is
6568 output on to @file{stderr}. If two conflicting dump filenames are
6569 given for the same pass, then the latter option overrides the earlier
6573 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
6574 and @option{lineno}.
6577 Turn on all optimization options, i.e., @option{optimized},
6578 @option{missed}, and @option{note}.
6581 The following tree dumps are possible:
6585 @opindex fdump-tree-original
6586 Dump before any tree based optimization, to @file{@var{file}.original}.
6589 @opindex fdump-tree-optimized
6590 Dump after all tree based optimization, to @file{@var{file}.optimized}.
6593 @opindex fdump-tree-gimple
6594 Dump each function before and after the gimplification pass to a file. The
6595 file name is made by appending @file{.gimple} to the source file name.
6598 @opindex fdump-tree-cfg
6599 Dump the control flow graph of each function to a file. The file name is
6600 made by appending @file{.cfg} to the source file name.
6603 @opindex fdump-tree-ch
6604 Dump each function after copying loop headers. The file name is made by
6605 appending @file{.ch} to the source file name.
6608 @opindex fdump-tree-ssa
6609 Dump SSA related information to a file. The file name is made by appending
6610 @file{.ssa} to the source file name.
6613 @opindex fdump-tree-alias
6614 Dump aliasing information for each function. The file name is made by
6615 appending @file{.alias} to the source file name.
6618 @opindex fdump-tree-ccp
6619 Dump each function after CCP@. The file name is made by appending
6620 @file{.ccp} to the source file name.
6623 @opindex fdump-tree-storeccp
6624 Dump each function after STORE-CCP@. The file name is made by appending
6625 @file{.storeccp} to the source file name.
6628 @opindex fdump-tree-pre
6629 Dump trees after partial redundancy elimination. The file name is made
6630 by appending @file{.pre} to the source file name.
6633 @opindex fdump-tree-fre
6634 Dump trees after full redundancy elimination. The file name is made
6635 by appending @file{.fre} to the source file name.
6638 @opindex fdump-tree-copyprop
6639 Dump trees after copy propagation. The file name is made
6640 by appending @file{.copyprop} to the source file name.
6642 @item store_copyprop
6643 @opindex fdump-tree-store_copyprop
6644 Dump trees after store copy-propagation. The file name is made
6645 by appending @file{.store_copyprop} to the source file name.
6648 @opindex fdump-tree-dce
6649 Dump each function after dead code elimination. The file name is made by
6650 appending @file{.dce} to the source file name.
6653 @opindex fdump-tree-sra
6654 Dump each function after performing scalar replacement of aggregates. The
6655 file name is made by appending @file{.sra} to the source file name.
6658 @opindex fdump-tree-sink
6659 Dump each function after performing code sinking. The file name is made
6660 by appending @file{.sink} to the source file name.
6663 @opindex fdump-tree-dom
6664 Dump each function after applying dominator tree optimizations. The file
6665 name is made by appending @file{.dom} to the source file name.
6668 @opindex fdump-tree-dse
6669 Dump each function after applying dead store elimination. The file
6670 name is made by appending @file{.dse} to the source file name.
6673 @opindex fdump-tree-phiopt
6674 Dump each function after optimizing PHI nodes into straightline code. The file
6675 name is made by appending @file{.phiopt} to the source file name.
6678 @opindex fdump-tree-forwprop
6679 Dump each function after forward propagating single use variables. The file
6680 name is made by appending @file{.forwprop} to the source file name.
6683 @opindex fdump-tree-copyrename
6684 Dump each function after applying the copy rename optimization. The file
6685 name is made by appending @file{.copyrename} to the source file name.
6688 @opindex fdump-tree-nrv
6689 Dump each function after applying the named return value optimization on
6690 generic trees. The file name is made by appending @file{.nrv} to the source
6694 @opindex fdump-tree-vect
6695 Dump each function after applying vectorization of loops. The file name is
6696 made by appending @file{.vect} to the source file name.
6699 @opindex fdump-tree-slp
6700 Dump each function after applying vectorization of basic blocks. The file name
6701 is made by appending @file{.slp} to the source file name.
6704 @opindex fdump-tree-vrp
6705 Dump each function after Value Range Propagation (VRP). The file name
6706 is made by appending @file{.vrp} to the source file name.
6709 @opindex fdump-tree-all
6710 Enable all the available tree dumps with the flags provided in this option.
6714 @itemx -fopt-info-@var{options}
6715 @itemx -fopt-info-@var{options}=@var{filename}
6717 Controls optimization dumps from various optimization passes. If the
6718 @samp{-@var{options}} form is used, @var{options} is a list of
6719 @samp{-} separated options to select the dump details and
6720 optimizations. If @var{options} is not specified, it defaults to
6721 @option{optimized} for details and @option{optall} for optimization
6722 groups. If the @var{filename} is not specified, it defaults to
6723 @file{stderr}. Note that the output @var{filename} will be overwritten
6724 in case of multiple translation units. If a combined output from
6725 multiple translation units is desired, @file{stderr} should be used
6728 The options can be divided into two groups, 1) options describing the
6729 verbosity of the dump, and 2) options describing which optimizations
6730 should be included. The options from both the groups can be freely
6731 mixed as they are non-overlapping. However, in case of any conflicts,
6732 the latter options override the earlier options on the command
6733 line. Though multiple -fopt-info options are accepted, only one of
6734 them can have @option{=filename}. If other filenames are provided then
6735 all but the first one are ignored.
6737 The dump verbosity has the following options
6741 Print information when an optimization is successfully applied. It is
6742 up to a pass to decide which information is relevant. For example, the
6743 vectorizer passes print the source location of loops which got
6744 successfully vectorized.
6746 Print information about missed optimizations. Individual passes
6747 control which information to include in the output. For example,
6750 gcc -O2 -ftree-vectorize -fopt-info-vec-missed
6753 will print information about missed optimization opportunities from
6754 vectorization passes on stderr.
6756 Print verbose information about optimizations, such as certain
6757 transformations, more detailed messages about decisions etc.
6759 Print detailed optimization information. This includes
6760 @var{optimized}, @var{missed}, and @var{note}.
6763 The second set of options describes a group of optimizations and may
6764 include one or more of the following.
6768 Enable dumps from all interprocedural optimizations.
6770 Enable dumps from all loop optimizations.
6772 Enable dumps from all inlining optimizations.
6774 Enable dumps from all vectorization optimizations.
6776 Enable dumps from all optimizations. This is a superset of
6777 the optimization groups listed above.
6782 gcc -O3 -fopt-info-missed=missed.all
6785 outputs missed optimization report from all the passes into
6790 gcc -O3 -fopt-info-inline-optimized-missed=inline.txt
6793 will output information about missed optimizations as well as
6794 optimized locations from all the inlining passes into
6797 If the @var{filename} is provided, then the dumps from all the
6798 applicable optimizations are concatenated into the @file{filename}.
6799 Otherwise the dump is output onto @file{stderr}. If @var{options} is
6800 omitted, it defaults to @option{all-optall}, which means dump all
6801 available optimization info from all the passes. In the following
6802 example, all optimization info is output on to @file{stderr}.
6808 Note that @option{-fopt-info-vec-missed} behaves the same as
6809 @option{-fopt-info-missed-vec}.
6811 As another example, consider
6814 gcc -fopt-info-vec-missed=vec.miss -fopt-info-loop-optimized=loop.opt
6817 Here the two output filenames @file{vec.miss} and @file{loop.opt} are
6818 in conflict since only one output file is allowed. In this case, only
6819 the first option takes effect and the subsequent options are
6820 ignored. Thus only the @file{vec.miss} is produced which contains
6821 dumps from the vectorizer about missed opportunities.
6823 @item -frandom-seed=@var{string}
6824 @opindex frandom-seed
6825 This option provides a seed that GCC uses in place of
6826 random numbers in generating certain symbol names
6827 that have to be different in every compiled file. It is also used to
6828 place unique stamps in coverage data files and the object files that
6829 produce them. You can use the @option{-frandom-seed} option to produce
6830 reproducibly identical object files.
6832 The @var{string} should be different for every file you compile.
6834 @item -fsched-verbose=@var{n}
6835 @opindex fsched-verbose
6836 On targets that use instruction scheduling, this option controls the
6837 amount of debugging output the scheduler prints. This information is
6838 written to standard error, unless @option{-fdump-rtl-sched1} or
6839 @option{-fdump-rtl-sched2} is specified, in which case it is output
6840 to the usual dump listing file, @file{.sched1} or @file{.sched2}
6841 respectively. However for @var{n} greater than nine, the output is
6842 always printed to standard error.
6844 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
6845 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
6846 For @var{n} greater than one, it also output basic block probabilities,
6847 detailed ready list information and unit/insn info. For @var{n} greater
6848 than two, it includes RTL at abort point, control-flow and regions info.
6849 And for @var{n} over four, @option{-fsched-verbose} also includes
6853 @itemx -save-temps=cwd
6855 Store the usual ``temporary'' intermediate files permanently; place them
6856 in the current directory and name them based on the source file. Thus,
6857 compiling @file{foo.c} with @option{-c -save-temps} produces files
6858 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
6859 preprocessed @file{foo.i} output file even though the compiler now
6860 normally uses an integrated preprocessor.
6862 When used in combination with the @option{-x} command-line option,
6863 @option{-save-temps} is sensible enough to avoid over writing an
6864 input source file with the same extension as an intermediate file.
6865 The corresponding intermediate file may be obtained by renaming the
6866 source file before using @option{-save-temps}.
6868 If you invoke GCC in parallel, compiling several different source
6869 files that share a common base name in different subdirectories or the
6870 same source file compiled for multiple output destinations, it is
6871 likely that the different parallel compilers will interfere with each
6872 other, and overwrite the temporary files. For instance:
6875 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
6876 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
6879 may result in @file{foo.i} and @file{foo.o} being written to
6880 simultaneously by both compilers.
6882 @item -save-temps=obj
6883 @opindex save-temps=obj
6884 Store the usual ``temporary'' intermediate files permanently. If the
6885 @option{-o} option is used, the temporary files are based on the
6886 object file. If the @option{-o} option is not used, the
6887 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
6892 gcc -save-temps=obj -c foo.c
6893 gcc -save-temps=obj -c bar.c -o dir/xbar.o
6894 gcc -save-temps=obj foobar.c -o dir2/yfoobar
6898 creates @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
6899 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
6900 @file{dir2/yfoobar.o}.
6902 @item -time@r{[}=@var{file}@r{]}
6904 Report the CPU time taken by each subprocess in the compilation
6905 sequence. For C source files, this is the compiler proper and assembler
6906 (plus the linker if linking is done).
6908 Without the specification of an output file, the output looks like this:
6915 The first number on each line is the ``user time'', that is time spent
6916 executing the program itself. The second number is ``system time'',
6917 time spent executing operating system routines on behalf of the program.
6918 Both numbers are in seconds.
6920 With the specification of an output file, the output is appended to the
6921 named file, and it looks like this:
6924 0.12 0.01 cc1 @var{options}
6925 0.00 0.01 as @var{options}
6928 The ``user time'' and the ``system time'' are moved before the program
6929 name, and the options passed to the program are displayed, so that one
6930 can later tell what file was being compiled, and with which options.
6932 @item -fvar-tracking
6933 @opindex fvar-tracking
6934 Run variable tracking pass. It computes where variables are stored at each
6935 position in code. Better debugging information is then generated
6936 (if the debugging information format supports this information).
6938 It is enabled by default when compiling with optimization (@option{-Os},
6939 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
6940 the debug info format supports it.
6942 @item -fvar-tracking-assignments
6943 @opindex fvar-tracking-assignments
6944 @opindex fno-var-tracking-assignments
6945 Annotate assignments to user variables early in the compilation and
6946 attempt to carry the annotations over throughout the compilation all the
6947 way to the end, in an attempt to improve debug information while
6948 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
6950 It can be enabled even if var-tracking is disabled, in which case
6951 annotations are created and maintained, but discarded at the end.
6953 @item -fvar-tracking-assignments-toggle
6954 @opindex fvar-tracking-assignments-toggle
6955 @opindex fno-var-tracking-assignments-toggle
6956 Toggle @option{-fvar-tracking-assignments}, in the same way that
6957 @option{-gtoggle} toggles @option{-g}.
6959 @item -print-file-name=@var{library}
6960 @opindex print-file-name
6961 Print the full absolute name of the library file @var{library} that
6962 would be used when linking---and don't do anything else. With this
6963 option, GCC does not compile or link anything; it just prints the
6966 @item -print-multi-directory
6967 @opindex print-multi-directory
6968 Print the directory name corresponding to the multilib selected by any
6969 other switches present in the command line. This directory is supposed
6970 to exist in @env{GCC_EXEC_PREFIX}.
6972 @item -print-multi-lib
6973 @opindex print-multi-lib
6974 Print the mapping from multilib directory names to compiler switches
6975 that enable them. The directory name is separated from the switches by
6976 @samp{;}, and each switch starts with an @samp{@@} instead of the
6977 @samp{-}, without spaces between multiple switches. This is supposed to
6978 ease shell processing.
6980 @item -print-multi-os-directory
6981 @opindex print-multi-os-directory
6982 Print the path to OS libraries for the selected
6983 multilib, relative to some @file{lib} subdirectory. If OS libraries are
6984 present in the @file{lib} subdirectory and no multilibs are used, this is
6985 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
6986 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
6987 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
6988 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
6990 @item -print-multiarch
6991 @opindex print-multiarch
6992 Print the path to OS libraries for the selected multiarch,
6993 relative to some @file{lib} subdirectory.
6995 @item -print-prog-name=@var{program}
6996 @opindex print-prog-name
6997 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
6999 @item -print-libgcc-file-name
7000 @opindex print-libgcc-file-name
7001 Same as @option{-print-file-name=libgcc.a}.
7003 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
7004 but you do want to link with @file{libgcc.a}. You can do:
7007 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
7010 @item -print-search-dirs
7011 @opindex print-search-dirs
7012 Print the name of the configured installation directory and a list of
7013 program and library directories @command{gcc} searches---and don't do anything else.
7015 This is useful when @command{gcc} prints the error message
7016 @samp{installation problem, cannot exec cpp0: No such file or directory}.
7017 To resolve this you either need to put @file{cpp0} and the other compiler
7018 components where @command{gcc} expects to find them, or you can set the environment
7019 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
7020 Don't forget the trailing @samp{/}.
7021 @xref{Environment Variables}.
7023 @item -print-sysroot
7024 @opindex print-sysroot
7025 Print the target sysroot directory that is used during
7026 compilation. This is the target sysroot specified either at configure
7027 time or using the @option{--sysroot} option, possibly with an extra
7028 suffix that depends on compilation options. If no target sysroot is
7029 specified, the option prints nothing.
7031 @item -print-sysroot-headers-suffix
7032 @opindex print-sysroot-headers-suffix
7033 Print the suffix added to the target sysroot when searching for
7034 headers, or give an error if the compiler is not configured with such
7035 a suffix---and don't do anything else.
7038 @opindex dumpmachine
7039 Print the compiler's target machine (for example,
7040 @samp{i686-pc-linux-gnu})---and don't do anything else.
7043 @opindex dumpversion
7044 Print the compiler version (for example, @samp{3.0})---and don't do
7049 Print the compiler's built-in specs---and don't do anything else. (This
7050 is used when GCC itself is being built.) @xref{Spec Files}.
7052 @item -fno-eliminate-unused-debug-types
7053 @opindex feliminate-unused-debug-types
7054 @opindex fno-eliminate-unused-debug-types
7055 Normally, when producing DWARF 2 output, GCC avoids producing debug symbol
7056 output for types that are nowhere used in the source file being compiled.
7057 Sometimes it is useful to have GCC emit debugging
7058 information for all types declared in a compilation
7059 unit, regardless of whether or not they are actually used
7060 in that compilation unit, for example
7061 if, in the debugger, you want to cast a value to a type that is
7062 not actually used in your program (but is declared). More often,
7063 however, this results in a significant amount of wasted space.
7066 @node Optimize Options
7067 @section Options That Control Optimization
7068 @cindex optimize options
7069 @cindex options, optimization
7071 These options control various sorts of optimizations.
7073 Without any optimization option, the compiler's goal is to reduce the
7074 cost of compilation and to make debugging produce the expected
7075 results. Statements are independent: if you stop the program with a
7076 breakpoint between statements, you can then assign a new value to any
7077 variable or change the program counter to any other statement in the
7078 function and get exactly the results you expect from the source
7081 Turning on optimization flags makes the compiler attempt to improve
7082 the performance and/or code size at the expense of compilation time
7083 and possibly the ability to debug the program.
7085 The compiler performs optimization based on the knowledge it has of the
7086 program. Compiling multiple files at once to a single output file mode allows
7087 the compiler to use information gained from all of the files when compiling
7090 Not all optimizations are controlled directly by a flag. Only
7091 optimizations that have a flag are listed in this section.
7093 Most optimizations are only enabled if an @option{-O} level is set on
7094 the command line. Otherwise they are disabled, even if individual
7095 optimization flags are specified.
7097 Depending on the target and how GCC was configured, a slightly different
7098 set of optimizations may be enabled at each @option{-O} level than
7099 those listed here. You can invoke GCC with @option{-Q --help=optimizers}
7100 to find out the exact set of optimizations that are enabled at each level.
7101 @xref{Overall Options}, for examples.
7108 Optimize. Optimizing compilation takes somewhat more time, and a lot
7109 more memory for a large function.
7111 With @option{-O}, the compiler tries to reduce code size and execution
7112 time, without performing any optimizations that take a great deal of
7115 @option{-O} turns on the following optimization flags:
7118 -fbranch-count-reg @gol
7119 -fcombine-stack-adjustments @gol
7121 -fcprop-registers @gol
7124 -fdelayed-branch @gol
7126 -fforward-propagate @gol
7127 -fguess-branch-probability @gol
7128 -fif-conversion2 @gol
7129 -fif-conversion @gol
7130 -finline-functions-called-once @gol
7131 -fipa-pure-const @gol
7133 -fipa-reference @gol
7134 -fmerge-constants @gol
7135 -fmove-loop-invariants @gol
7137 -fsplit-wide-types @gol
7142 -ftree-copy-prop @gol
7143 -ftree-copyrename @gol
7145 -ftree-dominator-opts @gol
7147 -ftree-forwprop @gol
7157 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
7158 where doing so does not interfere with debugging.
7162 Optimize even more. GCC performs nearly all supported optimizations
7163 that do not involve a space-speed tradeoff.
7164 As compared to @option{-O}, this option increases both compilation time
7165 and the performance of the generated code.
7167 @option{-O2} turns on all optimization flags specified by @option{-O}. It
7168 also turns on the following optimization flags:
7169 @gccoptlist{-fthread-jumps @gol
7170 -falign-functions -falign-jumps @gol
7171 -falign-loops -falign-labels @gol
7174 -fcse-follow-jumps -fcse-skip-blocks @gol
7175 -fdelete-null-pointer-checks @gol
7176 -fdevirtualize -fdevirtualize-speculatively @gol
7177 -fexpensive-optimizations @gol
7178 -fgcse -fgcse-lm @gol
7179 -fhoist-adjacent-loads @gol
7180 -finline-small-functions @gol
7181 -findirect-inlining @gol
7184 -fisolate-erroneous-paths-dereference @gol
7185 -foptimize-sibling-calls @gol
7186 -foptimize-strlen @gol
7187 -fpartial-inlining @gol
7189 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
7190 -frerun-cse-after-loop @gol
7191 -fsched-interblock -fsched-spec @gol
7192 -fschedule-insns -fschedule-insns2 @gol
7193 -fstrict-aliasing -fstrict-overflow @gol
7194 -ftree-builtin-call-dce @gol
7195 -ftree-switch-conversion -ftree-tail-merge @gol
7200 Please note the warning under @option{-fgcse} about
7201 invoking @option{-O2} on programs that use computed gotos.
7205 Optimize yet more. @option{-O3} turns on all optimizations specified
7206 by @option{-O2} and also turns on the @option{-finline-functions},
7207 @option{-funswitch-loops}, @option{-fpredictive-commoning},
7208 @option{-fgcse-after-reload}, @option{-ftree-loop-vectorize},
7209 @option{-ftree-loop-distribute-patterns},
7210 @option{-ftree-slp-vectorize}, @option{-fvect-cost-model},
7211 @option{-ftree-partial-pre} and @option{-fipa-cp-clone} options.
7215 Reduce compilation time and make debugging produce the expected
7216 results. This is the default.
7220 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
7221 do not typically increase code size. It also performs further
7222 optimizations designed to reduce code size.
7224 @option{-Os} disables the following optimization flags:
7225 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
7226 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
7227 -fprefetch-loop-arrays}
7231 Disregard strict standards compliance. @option{-Ofast} enables all
7232 @option{-O3} optimizations. It also enables optimizations that are not
7233 valid for all standard-compliant programs.
7234 It turns on @option{-ffast-math} and the Fortran-specific
7235 @option{-fno-protect-parens} and @option{-fstack-arrays}.
7239 Optimize debugging experience. @option{-Og} enables optimizations
7240 that do not interfere with debugging. It should be the optimization
7241 level of choice for the standard edit-compile-debug cycle, offering
7242 a reasonable level of optimization while maintaining fast compilation
7243 and a good debugging experience.
7245 If you use multiple @option{-O} options, with or without level numbers,
7246 the last such option is the one that is effective.
7249 Options of the form @option{-f@var{flag}} specify machine-independent
7250 flags. Most flags have both positive and negative forms; the negative
7251 form of @option{-ffoo} is @option{-fno-foo}. In the table
7252 below, only one of the forms is listed---the one you typically
7253 use. You can figure out the other form by either removing @samp{no-}
7256 The following options control specific optimizations. They are either
7257 activated by @option{-O} options or are related to ones that are. You
7258 can use the following flags in the rare cases when ``fine-tuning'' of
7259 optimizations to be performed is desired.
7262 @item -fno-defer-pop
7263 @opindex fno-defer-pop
7264 Always pop the arguments to each function call as soon as that function
7265 returns. For machines that must pop arguments after a function call,
7266 the compiler normally lets arguments accumulate on the stack for several
7267 function calls and pops them all at once.
7269 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7271 @item -fforward-propagate
7272 @opindex fforward-propagate
7273 Perform a forward propagation pass on RTL@. The pass tries to combine two
7274 instructions and checks if the result can be simplified. If loop unrolling
7275 is active, two passes are performed and the second is scheduled after
7278 This option is enabled by default at optimization levels @option{-O},
7279 @option{-O2}, @option{-O3}, @option{-Os}.
7281 @item -ffp-contract=@var{style}
7282 @opindex ffp-contract
7283 @option{-ffp-contract=off} disables floating-point expression contraction.
7284 @option{-ffp-contract=fast} enables floating-point expression contraction
7285 such as forming of fused multiply-add operations if the target has
7286 native support for them.
7287 @option{-ffp-contract=on} enables floating-point expression contraction
7288 if allowed by the language standard. This is currently not implemented
7289 and treated equal to @option{-ffp-contract=off}.
7291 The default is @option{-ffp-contract=fast}.
7293 @item -fomit-frame-pointer
7294 @opindex fomit-frame-pointer
7295 Don't keep the frame pointer in a register for functions that
7296 don't need one. This avoids the instructions to save, set up and
7297 restore frame pointers; it also makes an extra register available
7298 in many functions. @strong{It also makes debugging impossible on
7301 On some machines, such as the VAX, this flag has no effect, because
7302 the standard calling sequence automatically handles the frame pointer
7303 and nothing is saved by pretending it doesn't exist. The
7304 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
7305 whether a target machine supports this flag. @xref{Registers,,Register
7306 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
7308 Starting with GCC version 4.6, the default setting (when not optimizing for
7309 size) for 32-bit GNU/Linux x86 and 32-bit Darwin x86 targets has been changed to
7310 @option{-fomit-frame-pointer}. The default can be reverted to
7311 @option{-fno-omit-frame-pointer} by configuring GCC with the
7312 @option{--enable-frame-pointer} configure option.
7314 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7316 @item -foptimize-sibling-calls
7317 @opindex foptimize-sibling-calls
7318 Optimize sibling and tail recursive calls.
7320 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7322 @item -foptimize-strlen
7323 @opindex foptimize-strlen
7324 Optimize various standard C string functions (e.g. @code{strlen},
7325 @code{strchr} or @code{strcpy}) and
7326 their _FORTIFY_SOURCE counterparts into faster alternatives.
7328 Enabled at levels @option{-O2}, @option{-O3}.
7332 Do not expand any functions inline apart from those marked with
7333 the @code{always_inline} attribute. This is the default when not
7336 Single functions can be exempted from inlining by marking them
7337 with the @code{noinline} attribute.
7339 @item -finline-small-functions
7340 @opindex finline-small-functions
7341 Integrate functions into their callers when their body is smaller than expected
7342 function call code (so overall size of program gets smaller). The compiler
7343 heuristically decides which functions are simple enough to be worth integrating
7344 in this way. This inlining applies to all functions, even those not declared
7347 Enabled at level @option{-O2}.
7349 @item -findirect-inlining
7350 @opindex findirect-inlining
7351 Inline also indirect calls that are discovered to be known at compile
7352 time thanks to previous inlining. This option has any effect only
7353 when inlining itself is turned on by the @option{-finline-functions}
7354 or @option{-finline-small-functions} options.
7356 Enabled at level @option{-O2}.
7358 @item -finline-functions
7359 @opindex finline-functions
7360 Consider all functions for inlining, even if they are not declared inline.
7361 The compiler heuristically decides which functions are worth integrating
7364 If all calls to a given function are integrated, and the function is
7365 declared @code{static}, then the function is normally not output as
7366 assembler code in its own right.
7368 Enabled at level @option{-O3}.
7370 @item -finline-functions-called-once
7371 @opindex finline-functions-called-once
7372 Consider all @code{static} functions called once for inlining into their
7373 caller even if they are not marked @code{inline}. If a call to a given
7374 function is integrated, then the function is not output as assembler code
7377 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
7379 @item -fearly-inlining
7380 @opindex fearly-inlining
7381 Inline functions marked by @code{always_inline} and functions whose body seems
7382 smaller than the function call overhead early before doing
7383 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
7384 makes profiling significantly cheaper and usually inlining faster on programs
7385 having large chains of nested wrapper functions.
7391 Perform interprocedural scalar replacement of aggregates, removal of
7392 unused parameters and replacement of parameters passed by reference
7393 by parameters passed by value.
7395 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
7397 @item -finline-limit=@var{n}
7398 @opindex finline-limit
7399 By default, GCC limits the size of functions that can be inlined. This flag
7400 allows coarse control of this limit. @var{n} is the size of functions that
7401 can be inlined in number of pseudo instructions.
7403 Inlining is actually controlled by a number of parameters, which may be
7404 specified individually by using @option{--param @var{name}=@var{value}}.
7405 The @option{-finline-limit=@var{n}} option sets some of these parameters
7409 @item max-inline-insns-single
7410 is set to @var{n}/2.
7411 @item max-inline-insns-auto
7412 is set to @var{n}/2.
7415 See below for a documentation of the individual
7416 parameters controlling inlining and for the defaults of these parameters.
7418 @emph{Note:} there may be no value to @option{-finline-limit} that results
7419 in default behavior.
7421 @emph{Note:} pseudo instruction represents, in this particular context, an
7422 abstract measurement of function's size. In no way does it represent a count
7423 of assembly instructions and as such its exact meaning might change from one
7424 release to an another.
7426 @item -fno-keep-inline-dllexport
7427 @opindex -fno-keep-inline-dllexport
7428 This is a more fine-grained version of @option{-fkeep-inline-functions},
7429 which applies only to functions that are declared using the @code{dllexport}
7430 attribute or declspec (@xref{Function Attributes,,Declaring Attributes of
7433 @item -fkeep-inline-functions
7434 @opindex fkeep-inline-functions
7435 In C, emit @code{static} functions that are declared @code{inline}
7436 into the object file, even if the function has been inlined into all
7437 of its callers. This switch does not affect functions using the
7438 @code{extern inline} extension in GNU C90@. In C++, emit any and all
7439 inline functions into the object file.
7441 @item -fkeep-static-consts
7442 @opindex fkeep-static-consts
7443 Emit variables declared @code{static const} when optimization isn't turned
7444 on, even if the variables aren't referenced.
7446 GCC enables this option by default. If you want to force the compiler to
7447 check if a variable is referenced, regardless of whether or not
7448 optimization is turned on, use the @option{-fno-keep-static-consts} option.
7450 @item -fmerge-constants
7451 @opindex fmerge-constants
7452 Attempt to merge identical constants (string constants and floating-point
7453 constants) across compilation units.
7455 This option is the default for optimized compilation if the assembler and
7456 linker support it. Use @option{-fno-merge-constants} to inhibit this
7459 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7461 @item -fmerge-all-constants
7462 @opindex fmerge-all-constants
7463 Attempt to merge identical constants and identical variables.
7465 This option implies @option{-fmerge-constants}. In addition to
7466 @option{-fmerge-constants} this considers e.g.@: even constant initialized
7467 arrays or initialized constant variables with integral or floating-point
7468 types. Languages like C or C++ require each variable, including multiple
7469 instances of the same variable in recursive calls, to have distinct locations,
7470 so using this option results in non-conforming
7473 @item -fmodulo-sched
7474 @opindex fmodulo-sched
7475 Perform swing modulo scheduling immediately before the first scheduling
7476 pass. This pass looks at innermost loops and reorders their
7477 instructions by overlapping different iterations.
7479 @item -fmodulo-sched-allow-regmoves
7480 @opindex fmodulo-sched-allow-regmoves
7481 Perform more aggressive SMS-based modulo scheduling with register moves
7482 allowed. By setting this flag certain anti-dependences edges are
7483 deleted, which triggers the generation of reg-moves based on the
7484 life-range analysis. This option is effective only with
7485 @option{-fmodulo-sched} enabled.
7487 @item -fno-branch-count-reg
7488 @opindex fno-branch-count-reg
7489 Do not use ``decrement and branch'' instructions on a count register,
7490 but instead generate a sequence of instructions that decrement a
7491 register, compare it against zero, then branch based upon the result.
7492 This option is only meaningful on architectures that support such
7493 instructions, which include x86, PowerPC, IA-64 and S/390.
7495 Enabled by default at -O1 and higher.
7497 The default is @option{-fbranch-count-reg}.
7499 @item -fno-function-cse
7500 @opindex fno-function-cse
7501 Do not put function addresses in registers; make each instruction that
7502 calls a constant function contain the function's address explicitly.
7504 This option results in less efficient code, but some strange hacks
7505 that alter the assembler output may be confused by the optimizations
7506 performed when this option is not used.
7508 The default is @option{-ffunction-cse}
7510 @item -fno-zero-initialized-in-bss
7511 @opindex fno-zero-initialized-in-bss
7512 If the target supports a BSS section, GCC by default puts variables that
7513 are initialized to zero into BSS@. This can save space in the resulting
7516 This option turns off this behavior because some programs explicitly
7517 rely on variables going to the data section---e.g., so that the
7518 resulting executable can find the beginning of that section and/or make
7519 assumptions based on that.
7521 The default is @option{-fzero-initialized-in-bss}.
7523 @item -fthread-jumps
7524 @opindex fthread-jumps
7525 Perform optimizations that check to see if a jump branches to a
7526 location where another comparison subsumed by the first is found. If
7527 so, the first branch is redirected to either the destination of the
7528 second branch or a point immediately following it, depending on whether
7529 the condition is known to be true or false.
7531 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7533 @item -fsplit-wide-types
7534 @opindex fsplit-wide-types
7535 When using a type that occupies multiple registers, such as @code{long
7536 long} on a 32-bit system, split the registers apart and allocate them
7537 independently. This normally generates better code for those types,
7538 but may make debugging more difficult.
7540 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
7543 @item -fcse-follow-jumps
7544 @opindex fcse-follow-jumps
7545 In common subexpression elimination (CSE), scan through jump instructions
7546 when the target of the jump is not reached by any other path. For
7547 example, when CSE encounters an @code{if} statement with an
7548 @code{else} clause, CSE follows the jump when the condition
7551 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7553 @item -fcse-skip-blocks
7554 @opindex fcse-skip-blocks
7555 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
7556 follow jumps that conditionally skip over blocks. When CSE
7557 encounters a simple @code{if} statement with no else clause,
7558 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
7559 body of the @code{if}.
7561 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7563 @item -frerun-cse-after-loop
7564 @opindex frerun-cse-after-loop
7565 Re-run common subexpression elimination after loop optimizations are
7568 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7572 Perform a global common subexpression elimination pass.
7573 This pass also performs global constant and copy propagation.
7575 @emph{Note:} When compiling a program using computed gotos, a GCC
7576 extension, you may get better run-time performance if you disable
7577 the global common subexpression elimination pass by adding
7578 @option{-fno-gcse} to the command line.
7580 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7584 When @option{-fgcse-lm} is enabled, global common subexpression elimination
7585 attempts to move loads that are only killed by stores into themselves. This
7586 allows a loop containing a load/store sequence to be changed to a load outside
7587 the loop, and a copy/store within the loop.
7589 Enabled by default when @option{-fgcse} is enabled.
7593 When @option{-fgcse-sm} is enabled, a store motion pass is run after
7594 global common subexpression elimination. This pass attempts to move
7595 stores out of loops. When used in conjunction with @option{-fgcse-lm},
7596 loops containing a load/store sequence can be changed to a load before
7597 the loop and a store after the loop.
7599 Not enabled at any optimization level.
7603 When @option{-fgcse-las} is enabled, the global common subexpression
7604 elimination pass eliminates redundant loads that come after stores to the
7605 same memory location (both partial and full redundancies).
7607 Not enabled at any optimization level.
7609 @item -fgcse-after-reload
7610 @opindex fgcse-after-reload
7611 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
7612 pass is performed after reload. The purpose of this pass is to clean up
7615 @item -faggressive-loop-optimizations
7616 @opindex faggressive-loop-optimizations
7617 This option tells the loop optimizer to use language constraints to
7618 derive bounds for the number of iterations of a loop. This assumes that
7619 loop code does not invoke undefined behavior by for example causing signed
7620 integer overflows or out-of-bound array accesses. The bounds for the
7621 number of iterations of a loop are used to guide loop unrolling and peeling
7622 and loop exit test optimizations.
7623 This option is enabled by default.
7625 @item -funsafe-loop-optimizations
7626 @opindex funsafe-loop-optimizations
7627 This option tells the loop optimizer to assume that loop indices do not
7628 overflow, and that loops with nontrivial exit condition are not
7629 infinite. This enables a wider range of loop optimizations even if
7630 the loop optimizer itself cannot prove that these assumptions are valid.
7631 If you use @option{-Wunsafe-loop-optimizations}, the compiler warns you
7632 if it finds this kind of loop.
7634 @item -fcrossjumping
7635 @opindex fcrossjumping
7636 Perform cross-jumping transformation.
7637 This transformation unifies equivalent code and saves code size. The
7638 resulting code may or may not perform better than without cross-jumping.
7640 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7642 @item -fauto-inc-dec
7643 @opindex fauto-inc-dec
7644 Combine increments or decrements of addresses with memory accesses.
7645 This pass is always skipped on architectures that do not have
7646 instructions to support this. Enabled by default at @option{-O} and
7647 higher on architectures that support this.
7651 Perform dead code elimination (DCE) on RTL@.
7652 Enabled by default at @option{-O} and higher.
7656 Perform dead store elimination (DSE) on RTL@.
7657 Enabled by default at @option{-O} and higher.
7659 @item -fif-conversion
7660 @opindex fif-conversion
7661 Attempt to transform conditional jumps into branch-less equivalents. This
7662 includes use of conditional moves, min, max, set flags and abs instructions, and
7663 some tricks doable by standard arithmetics. The use of conditional execution
7664 on chips where it is available is controlled by @code{if-conversion2}.
7666 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7668 @item -fif-conversion2
7669 @opindex fif-conversion2
7670 Use conditional execution (where available) to transform conditional jumps into
7671 branch-less equivalents.
7673 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7675 @item -fdeclone-ctor-dtor
7676 @opindex fdeclone-ctor-dtor
7677 The C++ ABI requires multiple entry points for constructors and
7678 destructors: one for a base subobject, one for a complete object, and
7679 one for a virtual destructor that calls operator delete afterwards.
7680 For a hierarchy with virtual bases, the base and complete variants are
7681 clones, which means two copies of the function. With this option, the
7682 base and complete variants are changed to be thunks that call a common
7685 Enabled by @option{-Os}.
7687 @item -fdelete-null-pointer-checks
7688 @opindex fdelete-null-pointer-checks
7689 Assume that programs cannot safely dereference null pointers, and that
7690 no code or data element resides there. This enables simple constant
7691 folding optimizations at all optimization levels. In addition, other
7692 optimization passes in GCC use this flag to control global dataflow
7693 analyses that eliminate useless checks for null pointers; these assume
7694 that if a pointer is checked after it has already been dereferenced,
7697 Note however that in some environments this assumption is not true.
7698 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
7699 for programs that depend on that behavior.
7701 Some targets, especially embedded ones, disable this option at all levels.
7702 Otherwise it is enabled at all levels: @option{-O0}, @option{-O1},
7703 @option{-O2}, @option{-O3}, @option{-Os}. Passes that use the information
7704 are enabled independently at different optimization levels.
7706 @item -fdevirtualize
7707 @opindex fdevirtualize
7708 Attempt to convert calls to virtual functions to direct calls. This
7709 is done both within a procedure and interprocedurally as part of
7710 indirect inlining (@code{-findirect-inlining}) and interprocedural constant
7711 propagation (@option{-fipa-cp}).
7712 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7714 @item -fdevirtualize-speculatively
7715 @opindex fdevirtualize-speculatively
7716 Attempt to convert calls to virtual functions to speculative direct calls.
7717 Based on the analysis of the type inheritance graph, determine for a given call
7718 the set of likely targets. If the set is small, preferably of size 1, change
7719 the call into an conditional deciding on direct and indirect call. The
7720 speculative calls enable more optimizations, such as inlining. When they seem
7721 useless after further optimization, they are converted back into original form.
7723 @item -fexpensive-optimizations
7724 @opindex fexpensive-optimizations
7725 Perform a number of minor optimizations that are relatively expensive.
7727 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7731 Attempt to remove redundant extension instructions. This is especially
7732 helpful for the x86-64 architecture, which implicitly zero-extends in 64-bit
7733 registers after writing to their lower 32-bit half.
7735 Enabled for Alpha, AArch64 and x86 at levels @option{-O2},
7736 @option{-O3}, @option{-Os}.
7738 @item -flive-range-shrinkage
7739 @opindex flive-range-shrinkage
7740 Attempt to decrease register pressure through register live range
7741 shrinkage. This is helpful for fast processors with small or moderate
7744 @item -fira-algorithm=@var{algorithm}
7745 Use the specified coloring algorithm for the integrated register
7746 allocator. The @var{algorithm} argument can be @samp{priority}, which
7747 specifies Chow's priority coloring, or @samp{CB}, which specifies
7748 Chaitin-Briggs coloring. Chaitin-Briggs coloring is not implemented
7749 for all architectures, but for those targets that do support it, it is
7750 the default because it generates better code.
7752 @item -fira-region=@var{region}
7753 Use specified regions for the integrated register allocator. The
7754 @var{region} argument should be one of the following:
7759 Use all loops as register allocation regions.
7760 This can give the best results for machines with a small and/or
7761 irregular register set.
7764 Use all loops except for loops with small register pressure
7765 as the regions. This value usually gives
7766 the best results in most cases and for most architectures,
7767 and is enabled by default when compiling with optimization for speed
7768 (@option{-O}, @option{-O2}, @dots{}).
7771 Use all functions as a single region.
7772 This typically results in the smallest code size, and is enabled by default for
7773 @option{-Os} or @option{-O0}.
7777 @item -fira-hoist-pressure
7778 @opindex fira-hoist-pressure
7779 Use IRA to evaluate register pressure in the code hoisting pass for
7780 decisions to hoist expressions. This option usually results in smaller
7781 code, but it can slow the compiler down.
7783 This option is enabled at level @option{-Os} for all targets.
7785 @item -fira-loop-pressure
7786 @opindex fira-loop-pressure
7787 Use IRA to evaluate register pressure in loops for decisions to move
7788 loop invariants. This option usually results in generation
7789 of faster and smaller code on machines with large register files (>= 32
7790 registers), but it can slow the compiler down.
7792 This option is enabled at level @option{-O3} for some targets.
7794 @item -fno-ira-share-save-slots
7795 @opindex fno-ira-share-save-slots
7796 Disable sharing of stack slots used for saving call-used hard
7797 registers living through a call. Each hard register gets a
7798 separate stack slot, and as a result function stack frames are
7801 @item -fno-ira-share-spill-slots
7802 @opindex fno-ira-share-spill-slots
7803 Disable sharing of stack slots allocated for pseudo-registers. Each
7804 pseudo-register that does not get a hard register gets a separate
7805 stack slot, and as a result function stack frames are larger.
7807 @item -fira-verbose=@var{n}
7808 @opindex fira-verbose
7809 Control the verbosity of the dump file for the integrated register allocator.
7810 The default value is 5. If the value @var{n} is greater or equal to 10,
7811 the dump output is sent to stderr using the same format as @var{n} minus 10.
7813 @item -fdelayed-branch
7814 @opindex fdelayed-branch
7815 If supported for the target machine, attempt to reorder instructions
7816 to exploit instruction slots available after delayed branch
7819 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7821 @item -fschedule-insns
7822 @opindex fschedule-insns
7823 If supported for the target machine, attempt to reorder instructions to
7824 eliminate execution stalls due to required data being unavailable. This
7825 helps machines that have slow floating point or memory load instructions
7826 by allowing other instructions to be issued until the result of the load
7827 or floating-point instruction is required.
7829 Enabled at levels @option{-O2}, @option{-O3}.
7831 @item -fschedule-insns2
7832 @opindex fschedule-insns2
7833 Similar to @option{-fschedule-insns}, but requests an additional pass of
7834 instruction scheduling after register allocation has been done. This is
7835 especially useful on machines with a relatively small number of
7836 registers and where memory load instructions take more than one cycle.
7838 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7840 @item -fno-sched-interblock
7841 @opindex fno-sched-interblock
7842 Don't schedule instructions across basic blocks. This is normally
7843 enabled by default when scheduling before register allocation, i.e.@:
7844 with @option{-fschedule-insns} or at @option{-O2} or higher.
7846 @item -fno-sched-spec
7847 @opindex fno-sched-spec
7848 Don't allow speculative motion of non-load instructions. This is normally
7849 enabled by default when scheduling before register allocation, i.e.@:
7850 with @option{-fschedule-insns} or at @option{-O2} or higher.
7852 @item -fsched-pressure
7853 @opindex fsched-pressure
7854 Enable register pressure sensitive insn scheduling before register
7855 allocation. This only makes sense when scheduling before register
7856 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
7857 @option{-O2} or higher. Usage of this option can improve the
7858 generated code and decrease its size by preventing register pressure
7859 increase above the number of available hard registers and subsequent
7860 spills in register allocation.
7862 @item -fsched-spec-load
7863 @opindex fsched-spec-load
7864 Allow speculative motion of some load instructions. This only makes
7865 sense when scheduling before register allocation, i.e.@: with
7866 @option{-fschedule-insns} or at @option{-O2} or higher.
7868 @item -fsched-spec-load-dangerous
7869 @opindex fsched-spec-load-dangerous
7870 Allow speculative motion of more load instructions. This only makes
7871 sense when scheduling before register allocation, i.e.@: with
7872 @option{-fschedule-insns} or at @option{-O2} or higher.
7874 @item -fsched-stalled-insns
7875 @itemx -fsched-stalled-insns=@var{n}
7876 @opindex fsched-stalled-insns
7877 Define how many insns (if any) can be moved prematurely from the queue
7878 of stalled insns into the ready list during the second scheduling pass.
7879 @option{-fno-sched-stalled-insns} means that no insns are moved
7880 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
7881 on how many queued insns can be moved prematurely.
7882 @option{-fsched-stalled-insns} without a value is equivalent to
7883 @option{-fsched-stalled-insns=1}.
7885 @item -fsched-stalled-insns-dep
7886 @itemx -fsched-stalled-insns-dep=@var{n}
7887 @opindex fsched-stalled-insns-dep
7888 Define how many insn groups (cycles) are examined for a dependency
7889 on a stalled insn that is a candidate for premature removal from the queue
7890 of stalled insns. This has an effect only during the second scheduling pass,
7891 and only if @option{-fsched-stalled-insns} is used.
7892 @option{-fno-sched-stalled-insns-dep} is equivalent to
7893 @option{-fsched-stalled-insns-dep=0}.
7894 @option{-fsched-stalled-insns-dep} without a value is equivalent to
7895 @option{-fsched-stalled-insns-dep=1}.
7897 @item -fsched2-use-superblocks
7898 @opindex fsched2-use-superblocks
7899 When scheduling after register allocation, use superblock scheduling.
7900 This allows motion across basic block boundaries,
7901 resulting in faster schedules. This option is experimental, as not all machine
7902 descriptions used by GCC model the CPU closely enough to avoid unreliable
7903 results from the algorithm.
7905 This only makes sense when scheduling after register allocation, i.e.@: with
7906 @option{-fschedule-insns2} or at @option{-O2} or higher.
7908 @item -fsched-group-heuristic
7909 @opindex fsched-group-heuristic
7910 Enable the group heuristic in the scheduler. This heuristic favors
7911 the instruction that belongs to a schedule group. This is enabled
7912 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
7913 or @option{-fschedule-insns2} or at @option{-O2} or higher.
7915 @item -fsched-critical-path-heuristic
7916 @opindex fsched-critical-path-heuristic
7917 Enable the critical-path heuristic in the scheduler. This heuristic favors
7918 instructions on the critical path. This is enabled by default when
7919 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
7920 or @option{-fschedule-insns2} or at @option{-O2} or higher.
7922 @item -fsched-spec-insn-heuristic
7923 @opindex fsched-spec-insn-heuristic
7924 Enable the speculative instruction heuristic in the scheduler. This
7925 heuristic favors speculative instructions with greater dependency weakness.
7926 This is enabled by default when scheduling is enabled, i.e.@:
7927 with @option{-fschedule-insns} or @option{-fschedule-insns2}
7928 or at @option{-O2} or higher.
7930 @item -fsched-rank-heuristic
7931 @opindex fsched-rank-heuristic
7932 Enable the rank heuristic in the scheduler. This heuristic favors
7933 the instruction belonging to a basic block with greater size or frequency.
7934 This is enabled by default when scheduling is enabled, i.e.@:
7935 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
7936 at @option{-O2} or higher.
7938 @item -fsched-last-insn-heuristic
7939 @opindex fsched-last-insn-heuristic
7940 Enable the last-instruction heuristic in the scheduler. This heuristic
7941 favors the instruction that is less dependent on the last instruction
7942 scheduled. This is enabled by default when scheduling is enabled,
7943 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
7944 at @option{-O2} or higher.
7946 @item -fsched-dep-count-heuristic
7947 @opindex fsched-dep-count-heuristic
7948 Enable the dependent-count heuristic in the scheduler. This heuristic
7949 favors the instruction that has more instructions depending on it.
7950 This is enabled by default when scheduling is enabled, i.e.@:
7951 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
7952 at @option{-O2} or higher.
7954 @item -freschedule-modulo-scheduled-loops
7955 @opindex freschedule-modulo-scheduled-loops
7956 Modulo scheduling is performed before traditional scheduling. If a loop
7957 is modulo scheduled, later scheduling passes may change its schedule.
7958 Use this option to control that behavior.
7960 @item -fselective-scheduling
7961 @opindex fselective-scheduling
7962 Schedule instructions using selective scheduling algorithm. Selective
7963 scheduling runs instead of the first scheduler pass.
7965 @item -fselective-scheduling2
7966 @opindex fselective-scheduling2
7967 Schedule instructions using selective scheduling algorithm. Selective
7968 scheduling runs instead of the second scheduler pass.
7970 @item -fsel-sched-pipelining
7971 @opindex fsel-sched-pipelining
7972 Enable software pipelining of innermost loops during selective scheduling.
7973 This option has no effect unless one of @option{-fselective-scheduling} or
7974 @option{-fselective-scheduling2} is turned on.
7976 @item -fsel-sched-pipelining-outer-loops
7977 @opindex fsel-sched-pipelining-outer-loops
7978 When pipelining loops during selective scheduling, also pipeline outer loops.
7979 This option has no effect unless @option{-fsel-sched-pipelining} is turned on.
7981 @item -fsemantic-interposition
7982 @opindex fsemantic-interposition
7983 Some object formats, like ELF, allow interposing of symbols by dynamic linker.
7984 This means that for symbols exported from the DSO compiler can not perform
7985 inter-procedural propagation, inlining and other optimizations in anticipation
7986 that the function or variable in question may change. While this feature is
7987 useful, for example, to rewrite memory allocation functions by a debugging
7988 implementation, it is expensive in the terms of code quality.
7989 With @option{-fno-semantic-inteposition} compiler assumest that if interposition
7990 happens for functions the overwritting function will have
7991 precisely same semantics (and side effects). Similarly if interposition happens
7992 for variables, the constructor of the variable will be the same. The flag
7993 has no effect for functions explicitly declared inline, where
7994 interposition changing semantic is never allowed and for symbols explicitly
7998 @opindex fshrink-wrap
7999 Emit function prologues only before parts of the function that need it,
8000 rather than at the top of the function. This flag is enabled by default at
8001 @option{-O} and higher.
8003 @item -fcaller-saves
8004 @opindex fcaller-saves
8005 Enable allocation of values to registers that are clobbered by
8006 function calls, by emitting extra instructions to save and restore the
8007 registers around such calls. Such allocation is done only when it
8008 seems to result in better code.
8010 This option is always enabled by default on certain machines, usually
8011 those which have no call-preserved registers to use instead.
8013 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8015 @item -fcombine-stack-adjustments
8016 @opindex fcombine-stack-adjustments
8017 Tracks stack adjustments (pushes and pops) and stack memory references
8018 and then tries to find ways to combine them.
8020 Enabled by default at @option{-O1} and higher.
8022 @item -fuse-caller-save
8023 Use caller save registers for allocation if those registers are not used by
8024 any called function. In that case it is not necessary to save and restore
8025 them around calls. This is only possible if called functions are part of
8026 same compilation unit as current function and they are compiled before it.
8028 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8030 @item -fconserve-stack
8031 @opindex fconserve-stack
8032 Attempt to minimize stack usage. The compiler attempts to use less
8033 stack space, even if that makes the program slower. This option
8034 implies setting the @option{large-stack-frame} parameter to 100
8035 and the @option{large-stack-frame-growth} parameter to 400.
8037 @item -ftree-reassoc
8038 @opindex ftree-reassoc
8039 Perform reassociation on trees. This flag is enabled by default
8040 at @option{-O} and higher.
8044 Perform partial redundancy elimination (PRE) on trees. This flag is
8045 enabled by default at @option{-O2} and @option{-O3}.
8047 @item -ftree-partial-pre
8048 @opindex ftree-partial-pre
8049 Make partial redundancy elimination (PRE) more aggressive. This flag is
8050 enabled by default at @option{-O3}.
8052 @item -ftree-forwprop
8053 @opindex ftree-forwprop
8054 Perform forward propagation on trees. This flag is enabled by default
8055 at @option{-O} and higher.
8059 Perform full redundancy elimination (FRE) on trees. The difference
8060 between FRE and PRE is that FRE only considers expressions
8061 that are computed on all paths leading to the redundant computation.
8062 This analysis is faster than PRE, though it exposes fewer redundancies.
8063 This flag is enabled by default at @option{-O} and higher.
8065 @item -ftree-phiprop
8066 @opindex ftree-phiprop
8067 Perform hoisting of loads from conditional pointers on trees. This
8068 pass is enabled by default at @option{-O} and higher.
8070 @item -fhoist-adjacent-loads
8071 @opindex hoist-adjacent-loads
8072 Speculatively hoist loads from both branches of an if-then-else if the
8073 loads are from adjacent locations in the same structure and the target
8074 architecture has a conditional move instruction. This flag is enabled
8075 by default at @option{-O2} and higher.
8077 @item -ftree-copy-prop
8078 @opindex ftree-copy-prop
8079 Perform copy propagation on trees. This pass eliminates unnecessary
8080 copy operations. This flag is enabled by default at @option{-O} and
8083 @item -fipa-pure-const
8084 @opindex fipa-pure-const
8085 Discover which functions are pure or constant.
8086 Enabled by default at @option{-O} and higher.
8088 @item -fipa-reference
8089 @opindex fipa-reference
8090 Discover which static variables do not escape the
8092 Enabled by default at @option{-O} and higher.
8096 Perform interprocedural pointer analysis and interprocedural modification
8097 and reference analysis. This option can cause excessive memory and
8098 compile-time usage on large compilation units. It is not enabled by
8099 default at any optimization level.
8102 @opindex fipa-profile
8103 Perform interprocedural profile propagation. The functions called only from
8104 cold functions are marked as cold. Also functions executed once (such as
8105 @code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
8106 functions and loop less parts of functions executed once are then optimized for
8108 Enabled by default at @option{-O} and higher.
8112 Perform interprocedural constant propagation.
8113 This optimization analyzes the program to determine when values passed
8114 to functions are constants and then optimizes accordingly.
8115 This optimization can substantially increase performance
8116 if the application has constants passed to functions.
8117 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
8119 @item -fipa-cp-clone
8120 @opindex fipa-cp-clone
8121 Perform function cloning to make interprocedural constant propagation stronger.
8122 When enabled, interprocedural constant propagation performs function cloning
8123 when externally visible function can be called with constant arguments.
8124 Because this optimization can create multiple copies of functions,
8125 it may significantly increase code size
8126 (see @option{--param ipcp-unit-growth=@var{value}}).
8127 This flag is enabled by default at @option{-O3}.
8129 @item -fisolate-erroneous-paths-dereference
8130 Detect paths which trigger erroneous or undefined behaviour due to
8131 dereferencing a NULL pointer. Isolate those paths from the main control
8132 flow and turn the statement with erroneous or undefined behaviour into a trap.
8134 @item -fisolate-erroneous-paths-attribute
8135 Detect paths which trigger erroneous or undefined behaviour due a NULL value
8136 being used in a way which is forbidden by a @code{returns_nonnull} or @code{nonnull}
8137 attribute. Isolate those paths from the main control flow and turn the
8138 statement with erroneous or undefined behaviour into a trap. This is not
8139 currently enabled, but may be enabled by @code{-O2} in the future.
8143 Perform forward store motion on trees. This flag is
8144 enabled by default at @option{-O} and higher.
8146 @item -ftree-bit-ccp
8147 @opindex ftree-bit-ccp
8148 Perform sparse conditional bit constant propagation on trees and propagate
8149 pointer alignment information.
8150 This pass only operates on local scalar variables and is enabled by default
8151 at @option{-O} and higher. It requires that @option{-ftree-ccp} is enabled.
8155 Perform sparse conditional constant propagation (CCP) on trees. This
8156 pass only operates on local scalar variables and is enabled by default
8157 at @option{-O} and higher.
8160 @opindex fssa-phiopt
8161 Perform pattern matching on SSA PHI nodes to optimize conditional
8162 code. This pass is enabled by default at @option{-O} and higher.
8164 @item -ftree-switch-conversion
8165 Perform conversion of simple initializations in a switch to
8166 initializations from a scalar array. This flag is enabled by default
8167 at @option{-O2} and higher.
8169 @item -ftree-tail-merge
8170 Look for identical code sequences. When found, replace one with a jump to the
8171 other. This optimization is known as tail merging or cross jumping. This flag
8172 is enabled by default at @option{-O2} and higher. The compilation time
8174 be limited using @option{max-tail-merge-comparisons} parameter and
8175 @option{max-tail-merge-iterations} parameter.
8179 Perform dead code elimination (DCE) on trees. This flag is enabled by
8180 default at @option{-O} and higher.
8182 @item -ftree-builtin-call-dce
8183 @opindex ftree-builtin-call-dce
8184 Perform conditional dead code elimination (DCE) for calls to built-in functions
8185 that may set @code{errno} but are otherwise side-effect free. This flag is
8186 enabled by default at @option{-O2} and higher if @option{-Os} is not also
8189 @item -ftree-dominator-opts
8190 @opindex ftree-dominator-opts
8191 Perform a variety of simple scalar cleanups (constant/copy
8192 propagation, redundancy elimination, range propagation and expression
8193 simplification) based on a dominator tree traversal. This also
8194 performs jump threading (to reduce jumps to jumps). This flag is
8195 enabled by default at @option{-O} and higher.
8199 Perform dead store elimination (DSE) on trees. A dead store is a store into
8200 a memory location that is later overwritten by another store without
8201 any intervening loads. In this case the earlier store can be deleted. This
8202 flag is enabled by default at @option{-O} and higher.
8206 Perform loop header copying on trees. This is beneficial since it increases
8207 effectiveness of code motion optimizations. It also saves one jump. This flag
8208 is enabled by default at @option{-O} and higher. It is not enabled
8209 for @option{-Os}, since it usually increases code size.
8211 @item -ftree-loop-optimize
8212 @opindex ftree-loop-optimize
8213 Perform loop optimizations on trees. This flag is enabled by default
8214 at @option{-O} and higher.
8216 @item -ftree-loop-linear
8217 @opindex ftree-loop-linear
8218 Perform loop interchange transformations on tree. Same as
8219 @option{-floop-interchange}. To use this code transformation, GCC has
8220 to be configured with @option{--with-ppl} and @option{--with-cloog} to
8221 enable the Graphite loop transformation infrastructure.
8223 @item -floop-interchange
8224 @opindex floop-interchange
8225 Perform loop interchange transformations on loops. Interchanging two
8226 nested loops switches the inner and outer loops. For example, given a
8231 A(J, I) = A(J, I) * C
8235 loop interchange transforms the loop as if it were written:
8239 A(J, I) = A(J, I) * C
8243 which can be beneficial when @code{N} is larger than the caches,
8244 because in Fortran, the elements of an array are stored in memory
8245 contiguously by column, and the original loop iterates over rows,
8246 potentially creating at each access a cache miss. This optimization
8247 applies to all the languages supported by GCC and is not limited to
8248 Fortran. To use this code transformation, GCC has to be configured
8249 with @option{--with-ppl} and @option{--with-cloog} to enable the
8250 Graphite loop transformation infrastructure.
8252 @item -floop-strip-mine
8253 @opindex floop-strip-mine
8254 Perform loop strip mining transformations on loops. Strip mining
8255 splits a loop into two nested loops. The outer loop has strides
8256 equal to the strip size and the inner loop has strides of the
8257 original loop within a strip. The strip length can be changed
8258 using the @option{loop-block-tile-size} parameter. For example,
8265 loop strip mining transforms the loop as if it were written:
8268 DO I = II, min (II + 50, N)
8273 This optimization applies to all the languages supported by GCC and is
8274 not limited to Fortran. To use this code transformation, GCC has to
8275 be configured with @option{--with-ppl} and @option{--with-cloog} to
8276 enable the Graphite loop transformation infrastructure.
8279 @opindex floop-block
8280 Perform loop blocking transformations on loops. Blocking strip mines
8281 each loop in the loop nest such that the memory accesses of the
8282 element loops fit inside caches. The strip length can be changed
8283 using the @option{loop-block-tile-size} parameter. For example, given
8288 A(J, I) = B(I) + C(J)
8292 loop blocking transforms the loop as if it were written:
8296 DO I = II, min (II + 50, N)
8297 DO J = JJ, min (JJ + 50, M)
8298 A(J, I) = B(I) + C(J)
8304 which can be beneficial when @code{M} is larger than the caches,
8305 because the innermost loop iterates over a smaller amount of data
8306 which can be kept in the caches. This optimization applies to all the
8307 languages supported by GCC and is not limited to Fortran. To use this
8308 code transformation, GCC has to be configured with @option{--with-ppl}
8309 and @option{--with-cloog} to enable the Graphite loop transformation
8312 @item -fgraphite-identity
8313 @opindex fgraphite-identity
8314 Enable the identity transformation for graphite. For every SCoP we generate
8315 the polyhedral representation and transform it back to gimple. Using
8316 @option{-fgraphite-identity} we can check the costs or benefits of the
8317 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
8318 are also performed by the code generator CLooG, like index splitting and
8319 dead code elimination in loops.
8321 @item -floop-nest-optimize
8322 @opindex floop-nest-optimize
8323 Enable the ISL based loop nest optimizer. This is a generic loop nest
8324 optimizer based on the Pluto optimization algorithms. It calculates a loop
8325 structure optimized for data-locality and parallelism. This option
8328 @item -floop-parallelize-all
8329 @opindex floop-parallelize-all
8330 Use the Graphite data dependence analysis to identify loops that can
8331 be parallelized. Parallelize all the loops that can be analyzed to
8332 not contain loop carried dependences without checking that it is
8333 profitable to parallelize the loops.
8335 @item -fcheck-data-deps
8336 @opindex fcheck-data-deps
8337 Compare the results of several data dependence analyzers. This option
8338 is used for debugging the data dependence analyzers.
8340 @item -ftree-loop-if-convert
8341 Attempt to transform conditional jumps in the innermost loops to
8342 branch-less equivalents. The intent is to remove control-flow from
8343 the innermost loops in order to improve the ability of the
8344 vectorization pass to handle these loops. This is enabled by default
8345 if vectorization is enabled.
8347 @item -ftree-loop-if-convert-stores
8348 Attempt to also if-convert conditional jumps containing memory writes.
8349 This transformation can be unsafe for multi-threaded programs as it
8350 transforms conditional memory writes into unconditional memory writes.
8353 for (i = 0; i < N; i++)
8359 for (i = 0; i < N; i++)
8360 A[i] = cond ? expr : A[i];
8362 potentially producing data races.
8364 @item -ftree-loop-distribution
8365 Perform loop distribution. This flag can improve cache performance on
8366 big loop bodies and allow further loop optimizations, like
8367 parallelization or vectorization, to take place. For example, the loop
8384 @item -ftree-loop-distribute-patterns
8385 Perform loop distribution of patterns that can be code generated with
8386 calls to a library. This flag is enabled by default at @option{-O3}.
8388 This pass distributes the initialization loops and generates a call to
8389 memset zero. For example, the loop
8405 and the initialization loop is transformed into a call to memset zero.
8407 @item -ftree-loop-im
8408 @opindex ftree-loop-im
8409 Perform loop invariant motion on trees. This pass moves only invariants that
8410 are hard to handle at RTL level (function calls, operations that expand to
8411 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
8412 operands of conditions that are invariant out of the loop, so that we can use
8413 just trivial invariantness analysis in loop unswitching. The pass also includes
8416 @item -ftree-loop-ivcanon
8417 @opindex ftree-loop-ivcanon
8418 Create a canonical counter for number of iterations in loops for which
8419 determining number of iterations requires complicated analysis. Later
8420 optimizations then may determine the number easily. Useful especially
8421 in connection with unrolling.
8425 Perform induction variable optimizations (strength reduction, induction
8426 variable merging and induction variable elimination) on trees.
8428 @item -ftree-parallelize-loops=n
8429 @opindex ftree-parallelize-loops
8430 Parallelize loops, i.e., split their iteration space to run in n threads.
8431 This is only possible for loops whose iterations are independent
8432 and can be arbitrarily reordered. The optimization is only
8433 profitable on multiprocessor machines, for loops that are CPU-intensive,
8434 rather than constrained e.g.@: by memory bandwidth. This option
8435 implies @option{-pthread}, and thus is only supported on targets
8436 that have support for @option{-pthread}.
8440 Perform function-local points-to analysis on trees. This flag is
8441 enabled by default at @option{-O} and higher.
8445 Perform scalar replacement of aggregates. This pass replaces structure
8446 references with scalars to prevent committing structures to memory too
8447 early. This flag is enabled by default at @option{-O} and higher.
8449 @item -ftree-copyrename
8450 @opindex ftree-copyrename
8451 Perform copy renaming on trees. This pass attempts to rename compiler
8452 temporaries to other variables at copy locations, usually resulting in
8453 variable names which more closely resemble the original variables. This flag
8454 is enabled by default at @option{-O} and higher.
8456 @item -ftree-coalesce-inlined-vars
8457 @opindex ftree-coalesce-inlined-vars
8458 Tell the copyrename pass (see @option{-ftree-copyrename}) to attempt to
8459 combine small user-defined variables too, but only if they were inlined
8460 from other functions. It is a more limited form of
8461 @option{-ftree-coalesce-vars}. This may harm debug information of such
8462 inlined variables, but it will keep variables of the inlined-into
8463 function apart from each other, such that they are more likely to
8464 contain the expected values in a debugging session. This was the
8465 default in GCC versions older than 4.7.
8467 @item -ftree-coalesce-vars
8468 @opindex ftree-coalesce-vars
8469 Tell the copyrename pass (see @option{-ftree-copyrename}) to attempt to
8470 combine small user-defined variables too, instead of just compiler
8471 temporaries. This may severely limit the ability to debug an optimized
8472 program compiled with @option{-fno-var-tracking-assignments}. In the
8473 negated form, this flag prevents SSA coalescing of user variables,
8474 including inlined ones. This option is enabled by default.
8478 Perform temporary expression replacement during the SSA->normal phase. Single
8479 use/single def temporaries are replaced at their use location with their
8480 defining expression. This results in non-GIMPLE code, but gives the expanders
8481 much more complex trees to work on resulting in better RTL generation. This is
8482 enabled by default at @option{-O} and higher.
8486 Perform straight-line strength reduction on trees. This recognizes related
8487 expressions involving multiplications and replaces them by less expensive
8488 calculations when possible. This is enabled by default at @option{-O} and
8491 @item -ftree-vectorize
8492 @opindex ftree-vectorize
8493 Perform vectorization on trees. This flag enables @option{-ftree-loop-vectorize}
8494 and @option{-ftree-slp-vectorize} if not explicitly specified.
8496 @item -ftree-loop-vectorize
8497 @opindex ftree-loop-vectorize
8498 Perform loop vectorization on trees. This flag is enabled by default at
8499 @option{-O3} and when @option{-ftree-vectorize} is enabled.
8501 @item -ftree-slp-vectorize
8502 @opindex ftree-slp-vectorize
8503 Perform basic block vectorization on trees. This flag is enabled by default at
8504 @option{-O3} and when @option{-ftree-vectorize} is enabled.
8506 @item -fvect-cost-model=@var{model}
8507 @opindex fvect-cost-model
8508 Alter the cost model used for vectorization. The @var{model} argument
8509 should be one of @code{unlimited}, @code{dynamic} or @code{cheap}.
8510 With the @code{unlimited} model the vectorized code-path is assumed
8511 to be profitable while with the @code{dynamic} model a runtime check
8512 will guard the vectorized code-path to enable it only for iteration
8513 counts that will likely execute faster than when executing the original
8514 scalar loop. The @code{cheap} model will disable vectorization of
8515 loops where doing so would be cost prohibitive for example due to
8516 required runtime checks for data dependence or alignment but otherwise
8517 is equal to the @code{dynamic} model.
8518 The default cost model depends on other optimization flags and is
8519 either @code{dynamic} or @code{cheap}.
8521 @item -fsimd-cost-model=@var{model}
8522 @opindex fsimd-cost-model
8523 Alter the cost model used for vectorization of loops marked with the OpenMP
8524 or Cilk Plus simd directive. The @var{model} argument should be one of
8525 @code{unlimited}, @code{dynamic}, @code{cheap}. All values of @var{model}
8526 have the same meaning as described in @option{-fvect-cost-model} and by
8527 default a cost model defined with @option{-fvect-cost-model} is used.
8531 Perform Value Range Propagation on trees. This is similar to the
8532 constant propagation pass, but instead of values, ranges of values are
8533 propagated. This allows the optimizers to remove unnecessary range
8534 checks like array bound checks and null pointer checks. This is
8535 enabled by default at @option{-O2} and higher. Null pointer check
8536 elimination is only done if @option{-fdelete-null-pointer-checks} is
8541 Perform tail duplication to enlarge superblock size. This transformation
8542 simplifies the control flow of the function allowing other optimizations to do
8545 @item -funroll-loops
8546 @opindex funroll-loops
8547 Unroll loops whose number of iterations can be determined at compile
8548 time or upon entry to the loop. @option{-funroll-loops} implies
8549 @option{-frerun-cse-after-loop}. This option makes code larger,
8550 and may or may not make it run faster.
8552 @item -funroll-all-loops
8553 @opindex funroll-all-loops
8554 Unroll all loops, even if their number of iterations is uncertain when
8555 the loop is entered. This usually makes programs run more slowly.
8556 @option{-funroll-all-loops} implies the same options as
8557 @option{-funroll-loops},
8559 @item -fsplit-ivs-in-unroller
8560 @opindex fsplit-ivs-in-unroller
8561 Enables expression of values of induction variables in later iterations
8562 of the unrolled loop using the value in the first iteration. This breaks
8563 long dependency chains, thus improving efficiency of the scheduling passes.
8565 A combination of @option{-fweb} and CSE is often sufficient to obtain the
8566 same effect. However, that is not reliable in cases where the loop body
8567 is more complicated than a single basic block. It also does not work at all
8568 on some architectures due to restrictions in the CSE pass.
8570 This optimization is enabled by default.
8572 @item -fvariable-expansion-in-unroller
8573 @opindex fvariable-expansion-in-unroller
8574 With this option, the compiler creates multiple copies of some
8575 local variables when unrolling a loop, which can result in superior code.
8577 @item -fpartial-inlining
8578 @opindex fpartial-inlining
8579 Inline parts of functions. This option has any effect only
8580 when inlining itself is turned on by the @option{-finline-functions}
8581 or @option{-finline-small-functions} options.
8583 Enabled at level @option{-O2}.
8585 @item -fpredictive-commoning
8586 @opindex fpredictive-commoning
8587 Perform predictive commoning optimization, i.e., reusing computations
8588 (especially memory loads and stores) performed in previous
8589 iterations of loops.
8591 This option is enabled at level @option{-O3}.
8593 @item -fprefetch-loop-arrays
8594 @opindex fprefetch-loop-arrays
8595 If supported by the target machine, generate instructions to prefetch
8596 memory to improve the performance of loops that access large arrays.
8598 This option may generate better or worse code; results are highly
8599 dependent on the structure of loops within the source code.
8601 Disabled at level @option{-Os}.
8604 @itemx -fno-peephole2
8605 @opindex fno-peephole
8606 @opindex fno-peephole2
8607 Disable any machine-specific peephole optimizations. The difference
8608 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
8609 are implemented in the compiler; some targets use one, some use the
8610 other, a few use both.
8612 @option{-fpeephole} is enabled by default.
8613 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8615 @item -fno-guess-branch-probability
8616 @opindex fno-guess-branch-probability
8617 Do not guess branch probabilities using heuristics.
8619 GCC uses heuristics to guess branch probabilities if they are
8620 not provided by profiling feedback (@option{-fprofile-arcs}). These
8621 heuristics are based on the control flow graph. If some branch probabilities
8622 are specified by @samp{__builtin_expect}, then the heuristics are
8623 used to guess branch probabilities for the rest of the control flow graph,
8624 taking the @samp{__builtin_expect} info into account. The interactions
8625 between the heuristics and @samp{__builtin_expect} can be complex, and in
8626 some cases, it may be useful to disable the heuristics so that the effects
8627 of @samp{__builtin_expect} are easier to understand.
8629 The default is @option{-fguess-branch-probability} at levels
8630 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8632 @item -freorder-blocks
8633 @opindex freorder-blocks
8634 Reorder basic blocks in the compiled function in order to reduce number of
8635 taken branches and improve code locality.
8637 Enabled at levels @option{-O2}, @option{-O3}.
8639 @item -freorder-blocks-and-partition
8640 @opindex freorder-blocks-and-partition
8641 In addition to reordering basic blocks in the compiled function, in order
8642 to reduce number of taken branches, partitions hot and cold basic blocks
8643 into separate sections of the assembly and .o files, to improve
8644 paging and cache locality performance.
8646 This optimization is automatically turned off in the presence of
8647 exception handling, for linkonce sections, for functions with a user-defined
8648 section attribute and on any architecture that does not support named
8651 Enabled for x86 at levels @option{-O2}, @option{-O3}.
8653 @item -freorder-functions
8654 @opindex freorder-functions
8655 Reorder functions in the object file in order to
8656 improve code locality. This is implemented by using special
8657 subsections @code{.text.hot} for most frequently executed functions and
8658 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
8659 the linker so object file format must support named sections and linker must
8660 place them in a reasonable way.
8662 Also profile feedback must be available to make this option effective. See
8663 @option{-fprofile-arcs} for details.
8665 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8667 @item -freorder-functions=@var{alg}
8668 @opindex freorder-functions
8669 This invokes a linker plugin that globally reorders functions in the final binary according to the algorithm specified. Currently, only one type of reordering is supported, @code{callgraph}. This algorithm places functions that are connected by hot callgraph edges closer. This option also enables @option{-ffunction-sections}. Also, profile feedback must be available to makes this option effective. This option emits callgraph edge profile information in special sections named .gnu.callgraph.text. The params variable "gnu-cgraph-section-edge-threshold" can be used to only list edges above a certain threshold.
8671 @item -fstrict-aliasing
8672 @opindex fstrict-aliasing
8673 Allow the compiler to assume the strictest aliasing rules applicable to
8674 the language being compiled. For C (and C++), this activates
8675 optimizations based on the type of expressions. In particular, an
8676 object of one type is assumed never to reside at the same address as an
8677 object of a different type, unless the types are almost the same. For
8678 example, an @code{unsigned int} can alias an @code{int}, but not a
8679 @code{void*} or a @code{double}. A character type may alias any other
8682 @anchor{Type-punning}Pay special attention to code like this:
8695 The practice of reading from a different union member than the one most
8696 recently written to (called ``type-punning'') is common. Even with
8697 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
8698 is accessed through the union type. So, the code above works as
8699 expected. @xref{Structures unions enumerations and bit-fields
8700 implementation}. However, this code might not:
8711 Similarly, access by taking the address, casting the resulting pointer
8712 and dereferencing the result has undefined behavior, even if the cast
8713 uses a union type, e.g.:
8717 return ((union a_union *) &d)->i;
8721 The @option{-fstrict-aliasing} option is enabled at levels
8722 @option{-O2}, @option{-O3}, @option{-Os}.
8724 @item -fstrict-overflow
8725 @opindex fstrict-overflow
8726 Allow the compiler to assume strict signed overflow rules, depending
8727 on the language being compiled. For C (and C++) this means that
8728 overflow when doing arithmetic with signed numbers is undefined, which
8729 means that the compiler may assume that it does not happen. This
8730 permits various optimizations. For example, the compiler assumes
8731 that an expression like @code{i + 10 > i} is always true for
8732 signed @code{i}. This assumption is only valid if signed overflow is
8733 undefined, as the expression is false if @code{i + 10} overflows when
8734 using twos complement arithmetic. When this option is in effect any
8735 attempt to determine whether an operation on signed numbers
8736 overflows must be written carefully to not actually involve overflow.
8738 This option also allows the compiler to assume strict pointer
8739 semantics: given a pointer to an object, if adding an offset to that
8740 pointer does not produce a pointer to the same object, the addition is
8741 undefined. This permits the compiler to conclude that @code{p + u >
8742 p} is always true for a pointer @code{p} and unsigned integer
8743 @code{u}. This assumption is only valid because pointer wraparound is
8744 undefined, as the expression is false if @code{p + u} overflows using
8745 twos complement arithmetic.
8747 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
8748 that integer signed overflow is fully defined: it wraps. When
8749 @option{-fwrapv} is used, there is no difference between
8750 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
8751 integers. With @option{-fwrapv} certain types of overflow are
8752 permitted. For example, if the compiler gets an overflow when doing
8753 arithmetic on constants, the overflowed value can still be used with
8754 @option{-fwrapv}, but not otherwise.
8756 The @option{-fstrict-overflow} option is enabled at levels
8757 @option{-O2}, @option{-O3}, @option{-Os}.
8759 @item -falign-functions
8760 @itemx -falign-functions=@var{n}
8761 @opindex falign-functions
8762 Align the start of functions to the next power-of-two greater than
8763 @var{n}, skipping up to @var{n} bytes. For instance,
8764 @option{-falign-functions=32} aligns functions to the next 32-byte
8765 boundary, but @option{-falign-functions=24} aligns to the next
8766 32-byte boundary only if this can be done by skipping 23 bytes or less.
8768 @option{-fno-align-functions} and @option{-falign-functions=1} are
8769 equivalent and mean that functions are not aligned.
8771 Some assemblers only support this flag when @var{n} is a power of two;
8772 in that case, it is rounded up.
8774 If @var{n} is not specified or is zero, use a machine-dependent default.
8776 Enabled at levels @option{-O2}, @option{-O3}.
8778 @item -falign-labels
8779 @itemx -falign-labels=@var{n}
8780 @opindex falign-labels
8781 Align all branch targets to a power-of-two boundary, skipping up to
8782 @var{n} bytes like @option{-falign-functions}. This option can easily
8783 make code slower, because it must insert dummy operations for when the
8784 branch target is reached in the usual flow of the code.
8786 @option{-fno-align-labels} and @option{-falign-labels=1} are
8787 equivalent and mean that labels are not aligned.
8789 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
8790 are greater than this value, then their values are used instead.
8792 If @var{n} is not specified or is zero, use a machine-dependent default
8793 which is very likely to be @samp{1}, meaning no alignment.
8795 Enabled at levels @option{-O2}, @option{-O3}.
8798 @itemx -falign-loops=@var{n}
8799 @opindex falign-loops
8800 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
8801 like @option{-falign-functions}. If the loops are
8802 executed many times, this makes up for any execution of the dummy
8805 @option{-fno-align-loops} and @option{-falign-loops=1} are
8806 equivalent and mean that loops are not aligned.
8808 If @var{n} is not specified or is zero, use a machine-dependent default.
8810 Enabled at levels @option{-O2}, @option{-O3}.
8813 @itemx -falign-jumps=@var{n}
8814 @opindex falign-jumps
8815 Align branch targets to a power-of-two boundary, for branch targets
8816 where the targets can only be reached by jumping, skipping up to @var{n}
8817 bytes like @option{-falign-functions}. In this case, no dummy operations
8820 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
8821 equivalent and mean that loops are not aligned.
8823 If @var{n} is not specified or is zero, use a machine-dependent default.
8825 Enabled at levels @option{-O2}, @option{-O3}.
8827 @item -funit-at-a-time
8828 @opindex funit-at-a-time
8829 This option is left for compatibility reasons. @option{-funit-at-a-time}
8830 has no effect, while @option{-fno-unit-at-a-time} implies
8831 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
8835 @item -fno-toplevel-reorder
8836 @opindex fno-toplevel-reorder
8837 Do not reorder top-level functions, variables, and @code{asm}
8838 statements. Output them in the same order that they appear in the
8839 input file. When this option is used, unreferenced static variables
8840 are not removed. This option is intended to support existing code
8841 that relies on a particular ordering. For new code, it is better to
8842 use attributes when possible.
8844 Enabled at level @option{-O0}. When disabled explicitly, it also implies
8845 @option{-fno-section-anchors}, which is otherwise enabled at @option{-O0} on some
8850 Constructs webs as commonly used for register allocation purposes and assign
8851 each web individual pseudo register. This allows the register allocation pass
8852 to operate on pseudos directly, but also strengthens several other optimization
8853 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
8854 however, make debugging impossible, since variables no longer stay in a
8857 Enabled by default with @option{-funroll-loops}.
8859 @item -fwhole-program
8860 @opindex fwhole-program
8861 Assume that the current compilation unit represents the whole program being
8862 compiled. All public functions and variables with the exception of @code{main}
8863 and those merged by attribute @code{externally_visible} become static functions
8864 and in effect are optimized more aggressively by interprocedural optimizers.
8866 This option should not be used in combination with @code{-flto}.
8867 Instead relying on a linker plugin should provide safer and more precise
8870 @item -flto[=@var{n}]
8872 This option runs the standard link-time optimizer. When invoked
8873 with source code, it generates GIMPLE (one of GCC's internal
8874 representations) and writes it to special ELF sections in the object
8875 file. When the object files are linked together, all the function
8876 bodies are read from these ELF sections and instantiated as if they
8877 had been part of the same translation unit.
8879 To use the link-time optimizer, @option{-flto} and optimization
8880 options should be specified at compile time and during the final link.
8884 gcc -c -O2 -flto foo.c
8885 gcc -c -O2 -flto bar.c
8886 gcc -o myprog -flto -O2 foo.o bar.o
8889 The first two invocations to GCC save a bytecode representation
8890 of GIMPLE into special ELF sections inside @file{foo.o} and
8891 @file{bar.o}. The final invocation reads the GIMPLE bytecode from
8892 @file{foo.o} and @file{bar.o}, merges the two files into a single
8893 internal image, and compiles the result as usual. Since both
8894 @file{foo.o} and @file{bar.o} are merged into a single image, this
8895 causes all the interprocedural analyses and optimizations in GCC to
8896 work across the two files as if they were a single one. This means,
8897 for example, that the inliner is able to inline functions in
8898 @file{bar.o} into functions in @file{foo.o} and vice-versa.
8900 Another (simpler) way to enable link-time optimization is:
8903 gcc -o myprog -flto -O2 foo.c bar.c
8906 The above generates bytecode for @file{foo.c} and @file{bar.c},
8907 merges them together into a single GIMPLE representation and optimizes
8908 them as usual to produce @file{myprog}.
8910 The only important thing to keep in mind is that to enable link-time
8911 optimizations you need to use the GCC driver to perform the link-step.
8912 GCC then automatically performs link-time optimization if any of the
8913 objects involved were compiled with the @option{-flto}. You generally
8914 should specify the optimization options to be used for link-time
8915 optimization though GCC will try to be clever at guessing an
8916 optimization level to use from the options used at compile-time
8917 if you fail to specify one at link-time. You can always override
8918 the automatic decision to do link-time optimization at link-time
8919 by passing @option{-fno-lto} to the link command.
8921 To make whole program optimization effective, it is necessary to make
8922 certain whole program assumptions. The compiler needs to know
8923 what functions and variables can be accessed by libraries and runtime
8924 outside of the link-time optimized unit. When supported by the linker,
8925 the linker plugin (see @option{-fuse-linker-plugin}) passes information
8926 to the compiler about used and externally visible symbols. When
8927 the linker plugin is not available, @option{-fwhole-program} should be
8928 used to allow the compiler to make these assumptions, which leads
8929 to more aggressive optimization decisions.
8931 When @option{-fuse-linker-plugin} is not enabled then, when a file is
8932 compiled with @option{-flto}, the generated object file is larger than
8933 a regular object file because it contains GIMPLE bytecodes and the usual
8934 final code (see @option{-ffat-lto-objects}. This means that
8935 object files with LTO information can be linked as normal object
8936 files; if @option{-fno-lto} is passed to the linker, no
8937 interprocedural optimizations are applied. Note that when
8938 @option{-fno-fat-lto-objects} is enabled the compile-stage is faster
8939 but you cannot perform a regular, non-LTO link on them.
8941 Additionally, the optimization flags used to compile individual files
8942 are not necessarily related to those used at link time. For instance,
8945 gcc -c -O0 -ffat-lto-objects -flto foo.c
8946 gcc -c -O0 -ffat-lto-objects -flto bar.c
8947 gcc -o myprog -O3 foo.o bar.o
8950 This produces individual object files with unoptimized assembler
8951 code, but the resulting binary @file{myprog} is optimized at
8952 @option{-O3}. If, instead, the final binary is generated with
8953 @option{-fno-lto}, then @file{myprog} is not optimized.
8955 When producing the final binary, GCC only
8956 applies link-time optimizations to those files that contain bytecode.
8957 Therefore, you can mix and match object files and libraries with
8958 GIMPLE bytecodes and final object code. GCC automatically selects
8959 which files to optimize in LTO mode and which files to link without
8962 There are some code generation flags preserved by GCC when
8963 generating bytecodes, as they need to be used during the final link
8964 stage. Generally options specified at link-time override those
8965 specified at compile-time.
8967 If you do not specify an optimization level option @option{-O} at
8968 link-time then GCC will compute one based on the optimization levels
8969 used when compiling the object files. The highest optimization
8970 level will win here.
8972 Currently, the following options and their setting are take from
8973 the first object file that explicitely specified it:
8974 @option{-fPIC}, @option{-fpic}, @option{-fpie}, @option{-fcommon},
8975 @option{-fexceptions}, @option{-fnon-call-exceptions}, @option{-fgnu-tm}
8976 and all the @option{-m} target flags.
8978 Certain ABI changing flags are required to match in all compilation-units
8979 and trying to override this at link-time with a conflicting value
8980 is ignored. This includes options such as @option{-freg-struct-return}
8981 and @option{-fpcc-struct-return}.
8983 Other options such as @option{-ffp-contract}, @option{-fno-strict-overflow},
8984 @option{-fwrapv}, @option{-fno-trapv} or @option{-fno-strict-aliasing}
8985 are passed through to the link stage and merged conservatively for
8986 conflicting translation units. Specifically
8987 @option{-fno-strict-overflow}, @option{-fwrapv} and @option{-fno-trapv} take
8988 precedence and for example @option{-ffp-contract=off} takes precedence
8989 over @option{-ffp-contract=fast}. You can override them at linke-time.
8991 It is recommended that you compile all the files participating in the
8992 same link with the same options and also specify those options at
8995 If LTO encounters objects with C linkage declared with incompatible
8996 types in separate translation units to be linked together (undefined
8997 behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
8998 issued. The behavior is still undefined at run time. Similar
8999 diagnostics may be raised for other languages.
9001 Another feature of LTO is that it is possible to apply interprocedural
9002 optimizations on files written in different languages:
9007 gfortran -c -flto baz.f90
9008 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
9011 Notice that the final link is done with @command{g++} to get the C++
9012 runtime libraries and @option{-lgfortran} is added to get the Fortran
9013 runtime libraries. In general, when mixing languages in LTO mode, you
9014 should use the same link command options as when mixing languages in a
9015 regular (non-LTO) compilation.
9017 If object files containing GIMPLE bytecode are stored in a library archive, say
9018 @file{libfoo.a}, it is possible to extract and use them in an LTO link if you
9019 are using a linker with plugin support. To create static libraries suitable
9020 for LTO, use @command{gcc-ar} and @command{gcc-ranlib} instead of @command{ar}
9021 and @code{ranlib}; to show the symbols of object files with GIMPLE bytecode, use
9022 @command{gcc-nm}. Those commands require that @command{ar}, @command{ranlib}
9023 and @command{nm} have been compiled with plugin support. At link time, use the the
9024 flag @option{-fuse-linker-plugin} to ensure that the library participates in
9025 the LTO optimization process:
9028 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
9031 With the linker plugin enabled, the linker extracts the needed
9032 GIMPLE files from @file{libfoo.a} and passes them on to the running GCC
9033 to make them part of the aggregated GIMPLE image to be optimized.
9035 If you are not using a linker with plugin support and/or do not
9036 enable the linker plugin, then the objects inside @file{libfoo.a}
9037 are extracted and linked as usual, but they do not participate
9038 in the LTO optimization process. In order to make a static library suitable
9039 for both LTO optimization and usual linkage, compile its object files with
9040 @option{-flto} @code{-ffat-lto-objects}.
9042 Link-time optimizations do not require the presence of the whole program to
9043 operate. If the program does not require any symbols to be exported, it is
9044 possible to combine @option{-flto} and @option{-fwhole-program} to allow
9045 the interprocedural optimizers to use more aggressive assumptions which may
9046 lead to improved optimization opportunities.
9047 Use of @option{-fwhole-program} is not needed when linker plugin is
9048 active (see @option{-fuse-linker-plugin}).
9050 The current implementation of LTO makes no
9051 attempt to generate bytecode that is portable between different
9052 types of hosts. The bytecode files are versioned and there is a
9053 strict version check, so bytecode files generated in one version of
9054 GCC will not work with an older or newer version of GCC.
9056 Link-time optimization does not work well with generation of debugging
9057 information. Combining @option{-flto} with
9058 @option{-g} is currently experimental and expected to produce unexpected
9061 If you specify the optional @var{n}, the optimization and code
9062 generation done at link time is executed in parallel using @var{n}
9063 parallel jobs by utilizing an installed @command{make} program. The
9064 environment variable @env{MAKE} may be used to override the program
9065 used. The default value for @var{n} is 1.
9067 You can also specify @option{-flto=jobserver} to use GNU make's
9068 job server mode to determine the number of parallel jobs. This
9069 is useful when the Makefile calling GCC is already executing in parallel.
9070 You must prepend a @samp{+} to the command recipe in the parent Makefile
9071 for this to work. This option likely only works if @env{MAKE} is
9074 @item -flto-partition=@var{alg}
9075 @opindex flto-partition
9076 Specify the partitioning algorithm used by the link-time optimizer.
9077 The value is either @code{1to1} to specify a partitioning mirroring
9078 the original source files or @code{balanced} to specify partitioning
9079 into equally sized chunks (whenever possible) or @code{max} to create
9080 new partition for every symbol where possible. Specifying @code{none}
9081 as an algorithm disables partitioning and streaming completely.
9082 The default value is @code{balanced}. While @code{1to1} can be used
9083 as an workaround for various code ordering issues, the @code{max}
9084 partitioning is intended for internal testing only.
9085 The value @code{one} specifies that exactly one partition should be
9086 used while the value @code{none} bypasses partitioning and executes
9087 the link-time optimization step directly from the WPA phase.
9089 @item -flto-compression-level=@var{n}
9090 This option specifies the level of compression used for intermediate
9091 language written to LTO object files, and is only meaningful in
9092 conjunction with LTO mode (@option{-flto}). Valid
9093 values are 0 (no compression) to 9 (maximum compression). Values
9094 outside this range are clamped to either 0 or 9. If the option is not
9095 given, a default balanced compression setting is used.
9098 Prints a report with internal details on the workings of the link-time
9099 optimizer. The contents of this report vary from version to version.
9100 It is meant to be useful to GCC developers when processing object
9101 files in LTO mode (via @option{-flto}).
9103 Disabled by default.
9105 @item -flto-report-wpa
9106 Like @option{-flto-report}, but only print for the WPA phase of Link
9109 @item -fuse-linker-plugin
9110 Enables the use of a linker plugin during link-time optimization. This
9111 option relies on plugin support in the linker, which is available in gold
9112 or in GNU ld 2.21 or newer.
9114 This option enables the extraction of object files with GIMPLE bytecode out
9115 of library archives. This improves the quality of optimization by exposing
9116 more code to the link-time optimizer. This information specifies what
9117 symbols can be accessed externally (by non-LTO object or during dynamic
9118 linking). Resulting code quality improvements on binaries (and shared
9119 libraries that use hidden visibility) are similar to @code{-fwhole-program}.
9120 See @option{-flto} for a description of the effect of this flag and how to
9123 This option is enabled by default when LTO support in GCC is enabled
9124 and GCC was configured for use with
9125 a linker supporting plugins (GNU ld 2.21 or newer or gold).
9127 @item -ffat-lto-objects
9128 @opindex ffat-lto-objects
9129 Fat LTO objects are object files that contain both the intermediate language
9130 and the object code. This makes them usable for both LTO linking and normal
9131 linking. This option is effective only when compiling with @option{-flto}
9132 and is ignored at link time.
9134 @option{-fno-fat-lto-objects} improves compilation time over plain LTO, but
9135 requires the complete toolchain to be aware of LTO. It requires a linker with
9136 linker plugin support for basic functionality. Additionally,
9137 @command{nm}, @command{ar} and @command{ranlib}
9138 need to support linker plugins to allow a full-featured build environment
9139 (capable of building static libraries etc). GCC provides the @command{gcc-ar},
9140 @command{gcc-nm}, @command{gcc-ranlib} wrappers to pass the right options
9141 to these tools. With non fat LTO makefiles need to be modified to use them.
9143 The default is @option{-fno-fat-lto-objects} on targets with linker plugin
9146 @item -fcompare-elim
9147 @opindex fcompare-elim
9148 After register allocation and post-register allocation instruction splitting,
9149 identify arithmetic instructions that compute processor flags similar to a
9150 comparison operation based on that arithmetic. If possible, eliminate the
9151 explicit comparison operation.
9153 This pass only applies to certain targets that cannot explicitly represent
9154 the comparison operation before register allocation is complete.
9156 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
9159 @opindex fuse-ld=bfd
9160 Use the @command{bfd} linker instead of the default linker.
9163 @opindex fuse-ld=gold
9164 Use the @command{gold} linker instead of the default linker.
9166 @item -fcprop-registers
9167 @opindex fcprop-registers
9168 After register allocation and post-register allocation instruction splitting,
9169 perform a copy-propagation pass to try to reduce scheduling dependencies
9170 and occasionally eliminate the copy.
9172 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
9174 @item -fprofile-correction
9175 @opindex fprofile-correction
9176 Profiles collected using an instrumented binary for multi-threaded programs may
9177 be inconsistent due to missed counter updates. When this option is specified,
9178 GCC uses heuristics to correct or smooth out such inconsistencies. By
9179 default, GCC emits an error message when an inconsistent profile is detected.
9181 @item -fprofile-dir=@var{path}
9182 @opindex fprofile-dir
9184 Set the directory to search for the profile data files in to @var{path}.
9185 This option affects only the profile data generated by
9186 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
9187 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
9188 and its related options. Both absolute and relative paths can be used.
9189 By default, GCC uses the current directory as @var{path}, thus the
9190 profile data file appears in the same directory as the object file.
9192 @item -fprofile-generate
9193 @itemx -fprofile-generate=@var{path}
9194 @opindex fprofile-generate
9196 Enable options usually used for instrumenting application to produce
9197 profile useful for later recompilation with profile feedback based
9198 optimization. You must use @option{-fprofile-generate} both when
9199 compiling and when linking your program.
9201 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
9203 If @var{path} is specified, GCC looks at the @var{path} to find
9204 the profile feedback data files. See @option{-fprofile-dir}.
9206 @item -fprofile-generate-sampling
9207 @opindex -fprofile-generate-sampling
9209 Enable sampling for instrumented binaries. Instead of recording every event,
9210 record only every N-th event, where N (the sampling period) can be set either
9211 at compile time using
9212 @option{--param profile-generate-sampling-period=@var{value}}, or at
9213 execution start time through environment variable @samp{GCOV_SAMPLING_PERIOD}.
9215 At this time sampling applies only to branch counters. A sampling period of 100
9216 decreases instrumentated binary slowdown from up to 20x for heavily threaded
9217 applications down to around 2x. @option{-fprofile-correction} is always
9218 needed with sampling.
9221 @itemx -fprofile-use=@var{path}
9222 @opindex fprofile-use
9223 Enable profile feedback directed optimizations, and optimizations
9224 generally profitable only with profile feedback available.
9226 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
9227 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}, @code{-ftree-vectorize},
9228 @code{ftree-loop-distribute-patterns}
9230 By default, GCC emits an error message if the feedback profiles do not
9231 match the source code. This error can be turned into a warning by using
9232 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
9235 If @var{path} is specified, GCC looks at the @var{path} to find
9236 the profile feedback data files. See @option{-fprofile-dir}.
9238 @item -fprofile-strip=@var{base_suffix}
9239 @opindex fprofile-strip
9241 The option causes the compiler to strip @var{base_suffix} from the
9242 base file name and use that to find the gcda file.
9246 Perform dynamic inter-procedural analysis. This is used in conjunction with
9247 the @option{-fprofile-generate} and @option{-fprofile-use} options.
9248 During the @option{-fprofile-generate} phase, this flag turns on some additional
9249 instrumentation code that enables dynamic call-graph analysis.
9250 During the @option{-fprofile-use} phase, this flag enables cross-module
9251 optimizations such as inlining.
9253 @item -fripa-disallow-asm-modules
9254 @opindex fripa-disallow-asm-modules
9255 During profile-gen, if this flag is enabled, and the module has asm statements,
9256 arrange so that a bit recording this information will be set in the profile
9258 During profile-use, if this flag is enabled, and the same bit in auxiliary
9259 module's profile feedback data is set, don't import this auxiliary module.
9260 If this is the primary module, don't export it.
9262 @item -fripa-disallow-opt-mismatch
9263 @opindex fripa-disallow-opt-mismatch
9264 Don't import an auxiliary module, if the GCC command line options used for this
9265 auxiliary module during the profile-generate stage were different from those used
9266 for the primary module. Note that any mismatches in warning-related options are
9267 ignored for this comparison.
9269 @item -fripa-inc-path-sub=@var{path_mapping}
9270 @opindex fripa-inc-path-sub
9271 This option allows LIPO profile data associated with one compiler to be usable
9272 by another compiler at different installation path. A user can use this option
9273 to specify an include search path mapping so that the compiler can convert include
9274 search paths stored in gcda files into exisiting paths. The mapping is specified
9275 as a comma separated sub path pairs in the form: OLD_SUBPATH:NEW_SUBPATH.
9277 @item -fripa-no-promote-always-inline-func
9278 @opindex fripa-no-promote-always-inline-func
9279 Do not promote static functions with always inline attribute in LIPO compilation.
9281 @item -fripa-verbose
9282 @opindex fripa-verbose
9283 Enable printing of verbose information about dynamic inter-procedural optimizations.
9284 This is used in conjunction with the @option{-fripa}.
9286 @item -frecord-gcc-switches-in-elf
9287 @opindex frecord-gcc-switches-in-elf
9288 Record the command line options in the .gnu.switches.text elf section for sample
9289 based LIPO to do module grouping.
9292 The following options control compiler behavior regarding floating-point
9293 arithmetic. These options trade off between speed and
9294 correctness. All must be specifically enabled.
9298 @opindex ffloat-store
9299 Do not store floating-point variables in registers, and inhibit other
9300 options that might change whether a floating-point value is taken from a
9303 @cindex floating-point precision
9304 This option prevents undesirable excess precision on machines such as
9305 the 68000 where the floating registers (of the 68881) keep more
9306 precision than a @code{double} is supposed to have. Similarly for the
9307 x86 architecture. For most programs, the excess precision does only
9308 good, but a few programs rely on the precise definition of IEEE floating
9309 point. Use @option{-ffloat-store} for such programs, after modifying
9310 them to store all pertinent intermediate computations into variables.
9312 @item -fexcess-precision=@var{style}
9313 @opindex fexcess-precision
9314 This option allows further control over excess precision on machines
9315 where floating-point registers have more precision than the IEEE
9316 @code{float} and @code{double} types and the processor does not
9317 support operations rounding to those types. By default,
9318 @option{-fexcess-precision=fast} is in effect; this means that
9319 operations are carried out in the precision of the registers and that
9320 it is unpredictable when rounding to the types specified in the source
9321 code takes place. When compiling C, if
9322 @option{-fexcess-precision=standard} is specified then excess
9323 precision follows the rules specified in ISO C99; in particular,
9324 both casts and assignments cause values to be rounded to their
9325 semantic types (whereas @option{-ffloat-store} only affects
9326 assignments). This option is enabled by default for C if a strict
9327 conformance option such as @option{-std=c99} is used.
9330 @option{-fexcess-precision=standard} is not implemented for languages
9331 other than C, and has no effect if
9332 @option{-funsafe-math-optimizations} or @option{-ffast-math} is
9333 specified. On the x86, it also has no effect if @option{-mfpmath=sse}
9334 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
9335 semantics apply without excess precision, and in the latter, rounding
9340 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
9341 @option{-ffinite-math-only}, @option{-fno-rounding-math},
9342 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
9344 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
9346 This option is not turned on by any @option{-O} option besides
9347 @option{-Ofast} since it can result in incorrect output for programs
9348 that depend on an exact implementation of IEEE or ISO rules/specifications
9349 for math functions. It may, however, yield faster code for programs
9350 that do not require the guarantees of these specifications.
9352 @item -fno-math-errno
9353 @opindex fno-math-errno
9354 Do not set @code{errno} after calling math functions that are executed
9355 with a single instruction, e.g., @code{sqrt}. A program that relies on
9356 IEEE exceptions for math error handling may want to use this flag
9357 for speed while maintaining IEEE arithmetic compatibility.
9359 This option is not turned on by any @option{-O} option since
9360 it can result in incorrect output for programs that depend on
9361 an exact implementation of IEEE or ISO rules/specifications for
9362 math functions. It may, however, yield faster code for programs
9363 that do not require the guarantees of these specifications.
9365 The default is @option{-fmath-errno}.
9367 On Darwin systems, the math library never sets @code{errno}. There is
9368 therefore no reason for the compiler to consider the possibility that
9369 it might, and @option{-fno-math-errno} is the default.
9371 @item -funsafe-math-optimizations
9372 @opindex funsafe-math-optimizations
9374 Allow optimizations for floating-point arithmetic that (a) assume
9375 that arguments and results are valid and (b) may violate IEEE or
9376 ANSI standards. When used at link-time, it may include libraries
9377 or startup files that change the default FPU control word or other
9378 similar optimizations.
9380 This option is not turned on by any @option{-O} option since
9381 it can result in incorrect output for programs that depend on
9382 an exact implementation of IEEE or ISO rules/specifications for
9383 math functions. It may, however, yield faster code for programs
9384 that do not require the guarantees of these specifications.
9385 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
9386 @option{-fassociative-math} and @option{-freciprocal-math}.
9388 The default is @option{-fno-unsafe-math-optimizations}.
9390 @item -fassociative-math
9391 @opindex fassociative-math
9393 Allow re-association of operands in series of floating-point operations.
9394 This violates the ISO C and C++ language standard by possibly changing
9395 computation result. NOTE: re-ordering may change the sign of zero as
9396 well as ignore NaNs and inhibit or create underflow or overflow (and
9397 thus cannot be used on code that relies on rounding behavior like
9398 @code{(x + 2**52) - 2**52}. May also reorder floating-point comparisons
9399 and thus may not be used when ordered comparisons are required.
9400 This option requires that both @option{-fno-signed-zeros} and
9401 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
9402 much sense with @option{-frounding-math}. For Fortran the option
9403 is automatically enabled when both @option{-fno-signed-zeros} and
9404 @option{-fno-trapping-math} are in effect.
9406 The default is @option{-fno-associative-math}.
9408 @item -freciprocal-math
9409 @opindex freciprocal-math
9411 Allow the reciprocal of a value to be used instead of dividing by
9412 the value if this enables optimizations. For example @code{x / y}
9413 can be replaced with @code{x * (1/y)}, which is useful if @code{(1/y)}
9414 is subject to common subexpression elimination. Note that this loses
9415 precision and increases the number of flops operating on the value.
9417 The default is @option{-fno-reciprocal-math}.
9419 @item -ffinite-math-only
9420 @opindex ffinite-math-only
9421 Allow optimizations for floating-point arithmetic that assume
9422 that arguments and results are not NaNs or +-Infs.
9424 This option is not turned on by any @option{-O} option since
9425 it can result in incorrect output for programs that depend on
9426 an exact implementation of IEEE or ISO rules/specifications for
9427 math functions. It may, however, yield faster code for programs
9428 that do not require the guarantees of these specifications.
9430 The default is @option{-fno-finite-math-only}.
9432 @item -fno-signed-zeros
9433 @opindex fno-signed-zeros
9434 Allow optimizations for floating-point arithmetic that ignore the
9435 signedness of zero. IEEE arithmetic specifies the behavior of
9436 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
9437 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
9438 This option implies that the sign of a zero result isn't significant.
9440 The default is @option{-fsigned-zeros}.
9442 @item -fno-trapping-math
9443 @opindex fno-trapping-math
9444 Compile code assuming that floating-point operations cannot generate
9445 user-visible traps. These traps include division by zero, overflow,
9446 underflow, inexact result and invalid operation. This option requires
9447 that @option{-fno-signaling-nans} be in effect. Setting this option may
9448 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
9450 This option should never be turned on by any @option{-O} option since
9451 it can result in incorrect output for programs that depend on
9452 an exact implementation of IEEE or ISO rules/specifications for
9455 The default is @option{-ftrapping-math}.
9457 @item -frounding-math
9458 @opindex frounding-math
9459 Disable transformations and optimizations that assume default floating-point
9460 rounding behavior. This is round-to-zero for all floating point
9461 to integer conversions, and round-to-nearest for all other arithmetic
9462 truncations. This option should be specified for programs that change
9463 the FP rounding mode dynamically, or that may be executed with a
9464 non-default rounding mode. This option disables constant folding of
9465 floating-point expressions at compile time (which may be affected by
9466 rounding mode) and arithmetic transformations that are unsafe in the
9467 presence of sign-dependent rounding modes.
9469 The default is @option{-fno-rounding-math}.
9471 This option is experimental and does not currently guarantee to
9472 disable all GCC optimizations that are affected by rounding mode.
9473 Future versions of GCC may provide finer control of this setting
9474 using C99's @code{FENV_ACCESS} pragma. This command-line option
9475 will be used to specify the default state for @code{FENV_ACCESS}.
9477 @item -fsignaling-nans
9478 @opindex fsignaling-nans
9479 Compile code assuming that IEEE signaling NaNs may generate user-visible
9480 traps during floating-point operations. Setting this option disables
9481 optimizations that may change the number of exceptions visible with
9482 signaling NaNs. This option implies @option{-ftrapping-math}.
9484 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
9487 The default is @option{-fno-signaling-nans}.
9489 This option is experimental and does not currently guarantee to
9490 disable all GCC optimizations that affect signaling NaN behavior.
9492 @item -fsingle-precision-constant
9493 @opindex fsingle-precision-constant
9494 Treat floating-point constants as single precision instead of
9495 implicitly converting them to double-precision constants.
9497 @item -fcx-limited-range
9498 @opindex fcx-limited-range
9499 When enabled, this option states that a range reduction step is not
9500 needed when performing complex division. Also, there is no checking
9501 whether the result of a complex multiplication or division is @code{NaN
9502 + I*NaN}, with an attempt to rescue the situation in that case. The
9503 default is @option{-fno-cx-limited-range}, but is enabled by
9504 @option{-ffast-math}.
9506 This option controls the default setting of the ISO C99
9507 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
9510 @item -fcx-fortran-rules
9511 @opindex fcx-fortran-rules
9512 Complex multiplication and division follow Fortran rules. Range
9513 reduction is done as part of complex division, but there is no checking
9514 whether the result of a complex multiplication or division is @code{NaN
9515 + I*NaN}, with an attempt to rescue the situation in that case.
9517 The default is @option{-fno-cx-fortran-rules}.
9519 @item min-mcf-cancel-iters
9520 The minimum number of iterations of negative cycle cancellation during
9521 MCF profile correction before early termination. This parameter is
9522 only useful when using @option{-fprofile-correction}.
9527 The following options control optimizations that may improve
9528 performance, but are not enabled by any @option{-O} options. This
9529 section includes experimental options that may produce broken code.
9532 @item -fbranch-probabilities
9533 @opindex fbranch-probabilities
9534 After running a program compiled with @option{-fprofile-arcs}
9535 (@pxref{Debugging Options,, Options for Debugging Your Program or
9536 @command{gcc}}), you can compile it a second time using
9537 @option{-fbranch-probabilities}, to improve optimizations based on
9538 the number of times each branch was taken. When a program
9539 compiled with @option{-fprofile-arcs} exits, it saves arc execution
9540 counts to a file called @file{@var{sourcename}.gcda} for each source
9541 file. The information in this data file is very dependent on the
9542 structure of the generated code, so you must use the same source code
9543 and the same optimization options for both compilations.
9545 With @option{-fbranch-probabilities}, GCC puts a
9546 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
9547 These can be used to improve optimization. Currently, they are only
9548 used in one place: in @file{reorg.c}, instead of guessing which path a
9549 branch is most likely to take, the @samp{REG_BR_PROB} values are used to
9550 exactly determine which path is taken more often.
9552 @item -fprofile-values
9553 @opindex fprofile-values
9554 If combined with @option{-fprofile-arcs}, it adds code so that some
9555 data about values of expressions in the program is gathered.
9557 With @option{-fbranch-probabilities}, it reads back the data gathered
9558 from profiling values of expressions for usage in optimizations.
9560 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
9562 @item -fprofile-reorder-functions
9563 @opindex fprofile-reorder-functions
9564 Function reordering based on profile instrumentation collects
9565 first time of execution of a function and orders these functions
9568 Enabled with @option{-fprofile-use}.
9572 If combined with @option{-fprofile-arcs}, this option instructs the compiler
9573 to add code to gather information about values of expressions.
9575 With @option{-fbranch-probabilities}, it reads back the data gathered
9576 and actually performs the optimizations based on them.
9577 Currently the optimizations include specialization of division operations
9578 using the knowledge about the value of the denominator.
9580 @item -frename-registers
9581 @opindex frename-registers
9582 Attempt to avoid false dependencies in scheduled code by making use
9583 of registers left over after register allocation. This optimization
9584 most benefits processors with lots of registers. Depending on the
9585 debug information format adopted by the target, however, it can
9586 make debugging impossible, since variables no longer stay in
9587 a ``home register''.
9589 Enabled by default with @option{-funroll-loops} and @option{-fpeel-loops}.
9593 Perform tail duplication to enlarge superblock size. This transformation
9594 simplifies the control flow of the function allowing other optimizations to do
9597 Enabled with @option{-fprofile-use}.
9599 @item -funroll-loops
9600 @opindex funroll-loops
9601 Unroll loops whose number of iterations can be determined at compile time or
9602 upon entry to the loop. @option{-funroll-loops} implies
9603 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
9604 It also turns on complete loop peeling (i.e.@: complete removal of loops with
9605 a small constant number of iterations). This option makes code larger, and may
9606 or may not make it run faster.
9608 Enabled with @option{-fprofile-use}.
9610 @item -funroll-all-loops
9611 @opindex funroll-all-loops
9612 Unroll all loops, even if their number of iterations is uncertain when
9613 the loop is entered. This usually makes programs run more slowly.
9614 @option{-funroll-all-loops} implies the same options as
9615 @option{-funroll-loops}.
9617 @item -funroll-codesize-limit
9618 @opindex funroll-codesize-limit
9619 Limit loop unrolling of non-const non-FP loops in a profile feedback compilation
9620 under estimates of a large code footprint. Enabled by default with
9621 @option{-fprofile-use}. Code size and execution weight thresholds are controlled
9622 by the @option{unrollpeel-codesize-threshold} and
9623 @option{unrollpeel-hotness-threshold} parameters.
9626 @opindex fpeel-loops
9627 Peels loops for which there is enough information that they do not
9628 roll much (from profile feedback). It also turns on complete loop peeling
9629 (i.e.@: complete removal of loops with small constant number of iterations).
9631 Enabled with @option{-fprofile-use}.
9633 @item -fpeel-codesize-limit
9634 @opindex fpeel-codesize-limit
9635 Limit loop peeling of non-const non-FP loops in a profile feedback compilation
9636 under estimates of a large code footprint. Enabled by default with
9637 @option{-fprofile-use}. Code size and execution weight thresholds are controlled
9638 by the @option{unrollpeel-codesize-threshold} and
9639 @option{unrollpeel-hotness-threshold} parameters.
9641 @item -fmove-loop-invariants
9642 @opindex fmove-loop-invariants
9643 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
9644 at level @option{-O1}
9646 @item -funswitch-loops
9647 @opindex funswitch-loops
9648 Move branches with loop invariant conditions out of the loop, with duplicates
9649 of the loop on both branches (modified according to result of the condition).
9651 @item -ffunction-sections
9652 @itemx -fdata-sections
9653 @opindex ffunction-sections
9654 @opindex fdata-sections
9655 Place each function or data item into its own section in the output
9656 file if the target supports arbitrary sections. The name of the
9657 function or the name of the data item determines the section's name
9660 Use these options on systems where the linker can perform optimizations
9661 to improve locality of reference in the instruction space. Most systems
9662 using the ELF object format and SPARC processors running Solaris 2 have
9663 linkers with such optimizations. AIX may have these optimizations in
9666 Only use these options when there are significant benefits from doing
9667 so. When you specify these options, the assembler and linker
9668 create larger object and executable files and are also slower.
9669 You cannot use @code{gprof} on all systems if you
9670 specify this option, and you may have problems with debugging if
9671 you specify both this option and @option{-g}.
9673 @item -fbranch-target-load-optimize
9674 @opindex fbranch-target-load-optimize
9675 Perform branch target register load optimization before prologue / epilogue
9677 The use of target registers can typically be exposed only during reload,
9678 thus hoisting loads out of loops and doing inter-block scheduling needs
9679 a separate optimization pass.
9681 @item -fbranch-target-load-optimize2
9682 @opindex fbranch-target-load-optimize2
9683 Perform branch target register load optimization after prologue / epilogue
9686 @item -fbtr-bb-exclusive
9687 @opindex fbtr-bb-exclusive
9688 When performing branch target register load optimization, don't reuse
9689 branch target registers within any basic block.
9691 @item -fstack-protector
9692 @opindex fstack-protector
9693 Emit extra code to check for buffer overflows, such as stack smashing
9694 attacks. This is done by adding a guard variable to functions with
9695 vulnerable objects. This includes functions that call @code{alloca}, and
9696 functions with buffers larger than 8 bytes. The guards are initialized
9697 when a function is entered and then checked when the function exits.
9698 If a guard check fails, an error message is printed and the program exits.
9700 @item -fstack-protector-all
9701 @opindex fstack-protector-all
9702 Like @option{-fstack-protector} except that all functions are protected.
9704 @item -fstack-protector-strong
9705 @opindex fstack-protector-strong
9706 Like @option{-fstack-protector} but includes additional functions to
9707 be protected --- those that have local array definitions, or have
9708 references to local frame addresses.
9710 @item -fsection-anchors
9711 @opindex fsection-anchors
9712 Try to reduce the number of symbolic address calculations by using
9713 shared ``anchor'' symbols to address nearby objects. This transformation
9714 can help to reduce the number of GOT entries and GOT accesses on some
9717 For example, the implementation of the following function @code{foo}:
9721 int foo (void) @{ return a + b + c; @}
9725 usually calculates the addresses of all three variables, but if you
9726 compile it with @option{-fsection-anchors}, it accesses the variables
9727 from a common anchor point instead. The effect is similar to the
9728 following pseudocode (which isn't valid C):
9733 register int *xr = &x;
9734 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
9738 Not all targets support this option.
9740 @item --param @var{name}=@var{value}
9742 In some places, GCC uses various constants to control the amount of
9743 optimization that is done. For example, GCC does not inline functions
9744 that contain more than a certain number of instructions. You can
9745 control some of these constants on the command line using the
9746 @option{--param} option.
9748 The names of specific parameters, and the meaning of the values, are
9749 tied to the internals of the compiler, and are subject to change
9750 without notice in future releases.
9752 In each case, the @var{value} is an integer. The allowable choices for
9756 @item predictable-branch-outcome
9757 When branch is predicted to be taken with probability lower than this threshold
9758 (in percent), then it is considered well predictable. The default is 10.
9760 @item max-crossjump-edges
9761 The maximum number of incoming edges to consider for cross-jumping.
9762 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
9763 the number of edges incoming to each block. Increasing values mean
9764 more aggressive optimization, making the compilation time increase with
9765 probably small improvement in executable size.
9767 @item min-crossjump-insns
9768 The minimum number of instructions that must be matched at the end
9769 of two blocks before cross-jumping is performed on them. This
9770 value is ignored in the case where all instructions in the block being
9771 cross-jumped from are matched. The default value is 5.
9773 @item max-grow-copy-bb-insns
9774 The maximum code size expansion factor when copying basic blocks
9775 instead of jumping. The expansion is relative to a jump instruction.
9776 The default value is 8.
9778 @item max-goto-duplication-insns
9779 The maximum number of instructions to duplicate to a block that jumps
9780 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
9781 passes, GCC factors computed gotos early in the compilation process,
9782 and unfactors them as late as possible. Only computed jumps at the
9783 end of a basic blocks with no more than max-goto-duplication-insns are
9784 unfactored. The default value is 8.
9786 @item max-delay-slot-insn-search
9787 The maximum number of instructions to consider when looking for an
9788 instruction to fill a delay slot. If more than this arbitrary number of
9789 instructions are searched, the time savings from filling the delay slot
9790 are minimal, so stop searching. Increasing values mean more
9791 aggressive optimization, making the compilation time increase with probably
9792 small improvement in execution time.
9794 @item max-delay-slot-live-search
9795 When trying to fill delay slots, the maximum number of instructions to
9796 consider when searching for a block with valid live register
9797 information. Increasing this arbitrarily chosen value means more
9798 aggressive optimization, increasing the compilation time. This parameter
9799 should be removed when the delay slot code is rewritten to maintain the
9802 @item max-gcse-memory
9803 The approximate maximum amount of memory that can be allocated in
9804 order to perform the global common subexpression elimination
9805 optimization. If more memory than specified is required, the
9806 optimization is not done.
9808 @item max-gcse-insertion-ratio
9809 If the ratio of expression insertions to deletions is larger than this value
9810 for any expression, then RTL PRE inserts or removes the expression and thus
9811 leaves partially redundant computations in the instruction stream. The default value is 20.
9813 @item max-pending-list-length
9814 The maximum number of pending dependencies scheduling allows
9815 before flushing the current state and starting over. Large functions
9816 with few branches or calls can create excessively large lists which
9817 needlessly consume memory and resources.
9819 @item max-modulo-backtrack-attempts
9820 The maximum number of backtrack attempts the scheduler should make
9821 when modulo scheduling a loop. Larger values can exponentially increase
9824 @item max-inline-insns-single
9825 Several parameters control the tree inliner used in GCC@.
9826 This number sets the maximum number of instructions (counted in GCC's
9827 internal representation) in a single function that the tree inliner
9828 considers for inlining. This only affects functions declared
9829 inline and methods implemented in a class declaration (C++).
9830 The default value is 400.
9832 @item max-inline-insns-auto
9833 When you use @option{-finline-functions} (included in @option{-O3}),
9834 a lot of functions that would otherwise not be considered for inlining
9835 by the compiler are investigated. To those functions, a different
9836 (more restrictive) limit compared to functions declared inline can
9838 The default value is 40.
9840 @item inline-min-speedup
9841 When estimated performance improvement of caller + callee runtime exceeds this
9842 threshold (in precent), the function can be inlined regardless the limit on
9843 @option{--param max-inline-insns-single} and @option{--param
9844 max-inline-insns-auto}.
9846 @item large-function-insns
9847 The limit specifying really large functions. For functions larger than this
9848 limit after inlining, inlining is constrained by
9849 @option{--param large-function-growth}. This parameter is useful primarily
9850 to avoid extreme compilation time caused by non-linear algorithms used by the
9852 The default value is 2700.
9854 @item large-function-growth
9855 Specifies maximal growth of large function caused by inlining in percents.
9856 The default value is 100 which limits large function growth to 2.0 times
9859 @item large-unit-insns
9860 The limit specifying large translation unit. Growth caused by inlining of
9861 units larger than this limit is limited by @option{--param inline-unit-growth}.
9862 For small units this might be too tight.
9863 For example, consider a unit consisting of function A
9864 that is inline and B that just calls A three times. If B is small relative to
9865 A, the growth of unit is 300\% and yet such inlining is very sane. For very
9866 large units consisting of small inlineable functions, however, the overall unit
9867 growth limit is needed to avoid exponential explosion of code size. Thus for
9868 smaller units, the size is increased to @option{--param large-unit-insns}
9869 before applying @option{--param inline-unit-growth}. The default is 10000.
9871 @item inline-unit-growth
9872 Specifies maximal overall growth of the compilation unit caused by inlining.
9873 The default value is 30 which limits unit growth to 1.3 times the original
9874 size. Cold functions (either marked cold via an attribute or by profile
9875 feedback) are not accounted into the unit size.
9877 @item ipcp-unit-growth
9878 Specifies maximal overall growth of the compilation unit caused by
9879 interprocedural constant propagation. The default value is 10 which limits
9880 unit growth to 1.1 times the original size.
9882 @item large-stack-frame
9883 The limit specifying large stack frames. While inlining the algorithm is trying
9884 to not grow past this limit too much. The default value is 256 bytes.
9886 @item large-stack-frame-growth
9887 Specifies maximal growth of large stack frames caused by inlining in percents.
9888 The default value is 1000 which limits large stack frame growth to 11 times
9891 @item max-inline-insns-recursive
9892 @itemx max-inline-insns-recursive-auto
9893 Specifies the maximum number of instructions an out-of-line copy of a
9894 self-recursive inline
9895 function can grow into by performing recursive inlining.
9897 For functions declared inline, @option{--param max-inline-insns-recursive} is
9898 taken into account. For functions not declared inline, recursive inlining
9899 happens only when @option{-finline-functions} (included in @option{-O3}) is
9900 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
9901 default value is 450.
9903 @item max-inline-recursive-depth
9904 @itemx max-inline-recursive-depth-auto
9905 Specifies the maximum recursion depth used for recursive inlining.
9907 For functions declared inline, @option{--param max-inline-recursive-depth} is
9908 taken into account. For functions not declared inline, recursive inlining
9909 happens only when @option{-finline-functions} (included in @option{-O3}) is
9910 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
9913 @item min-inline-recursive-probability
9914 Recursive inlining is profitable only for function having deep recursion
9915 in average and can hurt for function having little recursion depth by
9916 increasing the prologue size or complexity of function body to other
9919 When profile feedback is available (see @option{-fprofile-generate}) the actual
9920 recursion depth can be guessed from probability that function recurses via a
9921 given call expression. This parameter limits inlining only to call expressions
9922 whose probability exceeds the given threshold (in percents).
9923 The default value is 10.
9925 @item early-inlining-insns
9926 Specify growth that the early inliner can make. In effect it increases
9927 the amount of inlining for code having a large abstraction penalty.
9928 The default value is 10.
9930 @item max-early-inliner-iterations
9931 @itemx max-early-inliner-iterations
9932 Limit of iterations of the early inliner. This basically bounds
9933 the number of nested indirect calls the early inliner can resolve.
9934 Deeper chains are still handled by late inlining.
9936 @item comdat-sharing-probability
9937 @itemx comdat-sharing-probability
9938 Probability (in percent) that C++ inline function with comdat visibility
9939 are shared across multiple compilation units. The default value is 20.
9941 @item profile-func-internal-id
9942 @itemx profile-func-internal-id
9943 A parameter to control whether to use function internal id in profile
9944 database lookup. If the value is 0, the compiler will use id that
9945 is based on function assembler name and filename, which makes old profile
9946 data more tolerant to source changes such as function reordering etc.
9947 The default value is 0.
9949 @item min-vect-loop-bound
9950 The minimum number of iterations under which loops are not vectorized
9951 when @option{-ftree-vectorize} is used. The number of iterations after
9952 vectorization needs to be greater than the value specified by this option
9953 to allow vectorization. The default value is 0.
9955 @item gcse-cost-distance-ratio
9956 Scaling factor in calculation of maximum distance an expression
9957 can be moved by GCSE optimizations. This is currently supported only in the
9958 code hoisting pass. The bigger the ratio, the more aggressive code hoisting
9959 is with simple expressions, i.e., the expressions that have cost
9960 less than @option{gcse-unrestricted-cost}. Specifying 0 disables
9961 hoisting of simple expressions. The default value is 10.
9963 @item gcse-unrestricted-cost
9964 Cost, roughly measured as the cost of a single typical machine
9965 instruction, at which GCSE optimizations do not constrain
9966 the distance an expression can travel. This is currently
9967 supported only in the code hoisting pass. The lesser the cost,
9968 the more aggressive code hoisting is. Specifying 0
9969 allows all expressions to travel unrestricted distances.
9970 The default value is 3.
9972 @item max-hoist-depth
9973 The depth of search in the dominator tree for expressions to hoist.
9974 This is used to avoid quadratic behavior in hoisting algorithm.
9975 The value of 0 does not limit on the search, but may slow down compilation
9976 of huge functions. The default value is 30.
9978 @item max-tail-merge-comparisons
9979 The maximum amount of similar bbs to compare a bb with. This is used to
9980 avoid quadratic behavior in tree tail merging. The default value is 10.
9982 @item max-tail-merge-iterations
9983 The maximum amount of iterations of the pass over the function. This is used to
9984 limit compilation time in tree tail merging. The default value is 2.
9986 @item max-unrolled-insns
9987 The maximum number of instructions that a loop may have to be unrolled.
9988 If a loop is unrolled, this parameter also determines how many times
9989 the loop code is unrolled.
9991 @item max-average-unrolled-insns
9992 The maximum number of instructions biased by probabilities of their execution
9993 that a loop may have to be unrolled. If a loop is unrolled,
9994 this parameter also determines how many times the loop code is unrolled.
9996 @item max-unroll-times
9997 The maximum number of unrollings of a single loop.
9999 @item max-peeled-insns
10000 The maximum number of instructions that a loop may have to be peeled.
10001 If a loop is peeled, this parameter also determines how many times
10002 the loop code is peeled.
10004 @item max-peel-times
10005 The maximum number of peelings of a single loop.
10007 @item max-peel-branches
10008 The maximum number of branches on the hot path through the peeled sequence.
10010 @item max-completely-peeled-insns
10011 The maximum number of insns of a completely peeled loop.
10013 @item max-completely-peel-times
10014 The maximum number of iterations of a loop to be suitable for complete peeling.
10016 @item max-completely-peel-loop-nest-depth
10017 The maximum depth of a loop nest suitable for complete peeling.
10019 @item unrollpeel-codesize-threshold
10020 Maximum profile-based code size footprint estimate for loop unrolling and
10023 @item unrollpeel-hotness-threshold
10024 Maximum ratio of total execution count to loop entry block count under which
10025 most profile-based code size estimates will be ignored for loop unrolling and
10028 @item max-unswitch-insns
10029 The maximum number of insns of an unswitched loop.
10031 @item max-unswitch-level
10032 The maximum number of branches unswitched in a single loop.
10034 @item min-iter-unroll-with-branches
10035 Minimum iteration count to ignore branch effects when unrolling.
10037 @item unroll-outer-loop-branch-budget
10038 Maximum number of branches allowed in hot outer loop region after unroll.
10040 @item lim-expensive
10041 The minimum cost of an expensive expression in the loop invariant motion.
10043 @item iv-consider-all-candidates-bound
10044 Bound on number of candidates for induction variables, below which
10045 all candidates are considered for each use in induction variable
10046 optimizations. If there are more candidates than this,
10047 only the most relevant ones are considered to avoid quadratic time complexity.
10049 @item iv-max-considered-uses
10050 The induction variable optimizations give up on loops that contain more
10051 induction variable uses.
10053 @item iv-always-prune-cand-set-bound
10054 If the number of candidates in the set is smaller than this value,
10055 always try to remove unnecessary ivs from the set
10056 when adding a new one.
10058 @item scev-max-expr-size
10059 Bound on size of expressions used in the scalar evolutions analyzer.
10060 Large expressions slow the analyzer.
10062 @item scev-max-expr-complexity
10063 Bound on the complexity of the expressions in the scalar evolutions analyzer.
10064 Complex expressions slow the analyzer.
10066 @item omega-max-vars
10067 The maximum number of variables in an Omega constraint system.
10068 The default value is 128.
10070 @item omega-max-geqs
10071 The maximum number of inequalities in an Omega constraint system.
10072 The default value is 256.
10074 @item omega-max-eqs
10075 The maximum number of equalities in an Omega constraint system.
10076 The default value is 128.
10078 @item omega-max-wild-cards
10079 The maximum number of wildcard variables that the Omega solver is
10080 able to insert. The default value is 18.
10082 @item omega-hash-table-size
10083 The size of the hash table in the Omega solver. The default value is
10086 @item omega-max-keys
10087 The maximal number of keys used by the Omega solver. The default
10090 @item omega-eliminate-redundant-constraints
10091 When set to 1, use expensive methods to eliminate all redundant
10092 constraints. The default value is 0.
10094 @item vect-max-version-for-alignment-checks
10095 The maximum number of run-time checks that can be performed when
10096 doing loop versioning for alignment in the vectorizer.
10098 @item vect-max-version-for-alias-checks
10099 The maximum number of run-time checks that can be performed when
10100 doing loop versioning for alias in the vectorizer.
10102 @item vect-max-peeling-for-alignment
10103 The maximum number of loop peels to enhance access alignment
10104 for vectorizer. Value -1 means 'no limit'.
10106 @item max-iterations-to-track
10107 The maximum number of iterations of a loop the brute-force algorithm
10108 for analysis of the number of iterations of the loop tries to evaluate.
10110 @item hot-bb-count-ws-permille
10111 A basic block profile count is considered hot if it contributes to
10112 the given permillage (i.e. 0...1000) of the entire profiled execution.
10114 @item hot-bb-frequency-fraction
10115 Select fraction of the entry block frequency of executions of basic block in
10116 function given basic block needs to have to be considered hot.
10118 @item max-predicted-iterations
10119 The maximum number of loop iterations we predict statically. This is useful
10120 in cases where a function contains a single loop with known bound and
10121 another loop with unknown bound.
10122 The known number of iterations is predicted correctly, while
10123 the unknown number of iterations average to roughly 10. This means that the
10124 loop without bounds appears artificially cold relative to the other one.
10126 @item builtin-expect-probability
10127 Control the probability of the expression having the specified value. This
10128 parameter takes a percentage (i.e. 0 ... 100) as input.
10129 The default probability of 90 is obtained empirically.
10131 @item align-threshold
10133 Select fraction of the maximal frequency of executions of a basic block in
10134 a function to align the basic block.
10136 @item align-loop-iterations
10138 A loop expected to iterate at least the selected number of iterations is
10141 @item tracer-dynamic-coverage
10142 @itemx tracer-dynamic-coverage-feedback
10144 This value is used to limit superblock formation once the given percentage of
10145 executed instructions is covered. This limits unnecessary code size
10148 The @option{tracer-dynamic-coverage-feedback} is used only when profile
10149 feedback is available. The real profiles (as opposed to statically estimated
10150 ones) are much less balanced allowing the threshold to be larger value.
10152 @item tracer-max-code-growth
10153 Stop tail duplication once code growth has reached given percentage. This is
10154 a rather artificial limit, as most of the duplicates are eliminated later in
10155 cross jumping, so it may be set to much higher values than is the desired code
10158 @item tracer-min-branch-ratio
10160 Stop reverse growth when the reverse probability of best edge is less than this
10161 threshold (in percent).
10163 @item tracer-min-branch-ratio
10164 @itemx tracer-min-branch-ratio-feedback
10166 Stop forward growth if the best edge has probability lower than this
10169 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
10170 compilation for profile feedback and one for compilation without. The value
10171 for compilation with profile feedback needs to be more conservative (higher) in
10172 order to make tracer effective.
10174 @item max-cse-path-length
10176 The maximum number of basic blocks on path that CSE considers.
10179 @item max-cse-insns
10180 The maximum number of instructions CSE processes before flushing.
10181 The default is 1000.
10183 @item ggc-min-expand
10185 GCC uses a garbage collector to manage its own memory allocation. This
10186 parameter specifies the minimum percentage by which the garbage
10187 collector's heap should be allowed to expand between collections.
10188 Tuning this may improve compilation speed; it has no effect on code
10191 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
10192 RAM >= 1GB@. If @code{getrlimit} is available, the notion of ``RAM'' is
10193 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
10194 GCC is not able to calculate RAM on a particular platform, the lower
10195 bound of 30% is used. Setting this parameter and
10196 @option{ggc-min-heapsize} to zero causes a full collection to occur at
10197 every opportunity. This is extremely slow, but can be useful for
10200 @item ggc-min-heapsize
10202 Minimum size of the garbage collector's heap before it begins bothering
10203 to collect garbage. The first collection occurs after the heap expands
10204 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
10205 tuning this may improve compilation speed, and has no effect on code
10208 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit that
10209 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
10210 with a lower bound of 4096 (four megabytes) and an upper bound of
10211 131072 (128 megabytes). If GCC is not able to calculate RAM on a
10212 particular platform, the lower bound is used. Setting this parameter
10213 very large effectively disables garbage collection. Setting this
10214 parameter and @option{ggc-min-expand} to zero causes a full collection
10215 to occur at every opportunity.
10217 @item max-reload-search-insns
10218 The maximum number of instruction reload should look backward for equivalent
10219 register. Increasing values mean more aggressive optimization, making the
10220 compilation time increase with probably slightly better performance.
10221 The default value is 100.
10223 @item max-cselib-memory-locations
10224 The maximum number of memory locations cselib should take into account.
10225 Increasing values mean more aggressive optimization, making the compilation time
10226 increase with probably slightly better performance. The default value is 500.
10228 @item reorder-blocks-duplicate
10229 @itemx reorder-blocks-duplicate-feedback
10231 Used by the basic block reordering pass to decide whether to use unconditional
10232 branch or duplicate the code on its destination. Code is duplicated when its
10233 estimated size is smaller than this value multiplied by the estimated size of
10234 unconditional jump in the hot spots of the program.
10236 The @option{reorder-block-duplicate-feedback} is used only when profile
10237 feedback is available. It may be set to higher values than
10238 @option{reorder-block-duplicate} since information about the hot spots is more
10241 @item max-sched-ready-insns
10242 The maximum number of instructions ready to be issued the scheduler should
10243 consider at any given time during the first scheduling pass. Increasing
10244 values mean more thorough searches, making the compilation time increase
10245 with probably little benefit. The default value is 100.
10247 @item max-sched-region-blocks
10248 The maximum number of blocks in a region to be considered for
10249 interblock scheduling. The default value is 10.
10251 @item max-pipeline-region-blocks
10252 The maximum number of blocks in a region to be considered for
10253 pipelining in the selective scheduler. The default value is 15.
10255 @item max-sched-region-insns
10256 The maximum number of insns in a region to be considered for
10257 interblock scheduling. The default value is 100.
10259 @item max-pipeline-region-insns
10260 The maximum number of insns in a region to be considered for
10261 pipelining in the selective scheduler. The default value is 200.
10263 @item min-spec-prob
10264 The minimum probability (in percents) of reaching a source block
10265 for interblock speculative scheduling. The default value is 40.
10267 @item max-sched-extend-regions-iters
10268 The maximum number of iterations through CFG to extend regions.
10269 A value of 0 (the default) disables region extensions.
10271 @item max-sched-insn-conflict-delay
10272 The maximum conflict delay for an insn to be considered for speculative motion.
10273 The default value is 3.
10275 @item sched-spec-prob-cutoff
10276 The minimal probability of speculation success (in percents), so that
10277 speculative insns are scheduled.
10278 The default value is 40.
10280 @item sched-spec-state-edge-prob-cutoff
10281 The minimum probability an edge must have for the scheduler to save its
10283 The default value is 10.
10285 @item sched-mem-true-dep-cost
10286 Minimal distance (in CPU cycles) between store and load targeting same
10287 memory locations. The default value is 1.
10289 @item selsched-max-lookahead
10290 The maximum size of the lookahead window of selective scheduling. It is a
10291 depth of search for available instructions.
10292 The default value is 50.
10294 @item selsched-max-sched-times
10295 The maximum number of times that an instruction is scheduled during
10296 selective scheduling. This is the limit on the number of iterations
10297 through which the instruction may be pipelined. The default value is 2.
10299 @item selsched-max-insns-to-rename
10300 The maximum number of best instructions in the ready list that are considered
10301 for renaming in the selective scheduler. The default value is 2.
10304 The minimum value of stage count that swing modulo scheduler
10305 generates. The default value is 2.
10307 @item max-last-value-rtl
10308 The maximum size measured as number of RTLs that can be recorded in an expression
10309 in combiner for a pseudo register as last known value of that register. The default
10312 @item max-combine-insns
10313 The maximum number of instructions the RTL combiner tries to combine.
10314 The default value is 2 at @option{-Og} and 4 otherwise.
10316 @item integer-share-limit
10317 Small integer constants can use a shared data structure, reducing the
10318 compiler's memory usage and increasing its speed. This sets the maximum
10319 value of a shared integer constant. The default value is 256.
10321 @item ssp-buffer-size
10322 The minimum size of buffers (i.e.@: arrays) that receive stack smashing
10323 protection when @option{-fstack-protection} is used.
10325 @item min-size-for-stack-sharing
10326 The minimum size of variables taking part in stack slot sharing when not
10327 optimizing. The default value is 32.
10329 @item max-jump-thread-duplication-stmts
10330 Maximum number of statements allowed in a block that needs to be
10331 duplicated when threading jumps.
10333 @item max-fields-for-field-sensitive
10334 Maximum number of fields in a structure treated in
10335 a field sensitive manner during pointer analysis. The default is zero
10336 for @option{-O0} and @option{-O1},
10337 and 100 for @option{-Os}, @option{-O2}, and @option{-O3}.
10339 @item prefetch-latency
10340 Estimate on average number of instructions that are executed before
10341 prefetch finishes. The distance prefetched ahead is proportional
10342 to this constant. Increasing this number may also lead to less
10343 streams being prefetched (see @option{simultaneous-prefetches}).
10345 @item simultaneous-prefetches
10346 Maximum number of prefetches that can run at the same time.
10348 @item l1-cache-line-size
10349 The size of cache line in L1 cache, in bytes.
10351 @item l1-cache-size
10352 The size of L1 cache, in kilobytes.
10354 @item l2-cache-size
10355 The size of L2 cache, in kilobytes.
10357 @item min-insn-to-prefetch-ratio
10358 The minimum ratio between the number of instructions and the
10359 number of prefetches to enable prefetching in a loop.
10361 @item prefetch-min-insn-to-mem-ratio
10362 The minimum ratio between the number of instructions and the
10363 number of memory references to enable prefetching in a loop.
10365 @item use-canonical-types
10366 Whether the compiler should use the ``canonical'' type system. By
10367 default, this should always be 1, which uses a more efficient internal
10368 mechanism for comparing types in C++ and Objective-C++. However, if
10369 bugs in the canonical type system are causing compilation failures,
10370 set this value to 0 to disable canonical types.
10372 @item switch-conversion-max-branch-ratio
10373 Switch initialization conversion refuses to create arrays that are
10374 bigger than @option{switch-conversion-max-branch-ratio} times the number of
10375 branches in the switch.
10377 @item max-partial-antic-length
10378 Maximum length of the partial antic set computed during the tree
10379 partial redundancy elimination optimization (@option{-ftree-pre}) when
10380 optimizing at @option{-O3} and above. For some sorts of source code
10381 the enhanced partial redundancy elimination optimization can run away,
10382 consuming all of the memory available on the host machine. This
10383 parameter sets a limit on the length of the sets that are computed,
10384 which prevents the runaway behavior. Setting a value of 0 for
10385 this parameter allows an unlimited set length.
10387 @item sccvn-max-scc-size
10388 Maximum size of a strongly connected component (SCC) during SCCVN
10389 processing. If this limit is hit, SCCVN processing for the whole
10390 function is not done and optimizations depending on it are
10391 disabled. The default maximum SCC size is 10000.
10393 @item sccvn-max-alias-queries-per-access
10394 Maximum number of alias-oracle queries we perform when looking for
10395 redundancies for loads and stores. If this limit is hit the search
10396 is aborted and the load or store is not considered redundant. The
10397 number of queries is algorithmically limited to the number of
10398 stores on all paths from the load to the function entry.
10399 The default maxmimum number of queries is 1000.
10401 @item ira-max-loops-num
10402 IRA uses regional register allocation by default. If a function
10403 contains more loops than the number given by this parameter, only at most
10404 the given number of the most frequently-executed loops form regions
10405 for regional register allocation. The default value of the
10408 @item ira-max-conflict-table-size
10409 Although IRA uses a sophisticated algorithm to compress the conflict
10410 table, the table can still require excessive amounts of memory for
10411 huge functions. If the conflict table for a function could be more
10412 than the size in MB given by this parameter, the register allocator
10413 instead uses a faster, simpler, and lower-quality
10414 algorithm that does not require building a pseudo-register conflict table.
10415 The default value of the parameter is 2000.
10417 @item ira-loop-reserved-regs
10418 IRA can be used to evaluate more accurate register pressure in loops
10419 for decisions to move loop invariants (see @option{-O3}). The number
10420 of available registers reserved for some other purposes is given
10421 by this parameter. The default value of the parameter is 2, which is
10422 the minimal number of registers needed by typical instructions.
10423 This value is the best found from numerous experiments.
10425 @item loop-invariant-max-bbs-in-loop
10426 Loop invariant motion can be very expensive, both in compilation time and
10427 in amount of needed compile-time memory, with very large loops. Loops
10428 with more basic blocks than this parameter won't have loop invariant
10429 motion optimization performed on them. The default value of the
10430 parameter is 1000 for @option{-O1} and 10000 for @option{-O2} and above.
10432 @item loop-max-datarefs-for-datadeps
10433 Building data dapendencies is expensive for very large loops. This
10434 parameter limits the number of data references in loops that are
10435 considered for data dependence analysis. These large loops are no
10436 handled by the optimizations using loop data dependencies.
10437 The default value is 1000.
10439 @item max-vartrack-size
10440 Sets a maximum number of hash table slots to use during variable
10441 tracking dataflow analysis of any function. If this limit is exceeded
10442 with variable tracking at assignments enabled, analysis for that
10443 function is retried without it, after removing all debug insns from
10444 the function. If the limit is exceeded even without debug insns, var
10445 tracking analysis is completely disabled for the function. Setting
10446 the parameter to zero makes it unlimited.
10448 @item max-vartrack-expr-depth
10449 Sets a maximum number of recursion levels when attempting to map
10450 variable names or debug temporaries to value expressions. This trades
10451 compilation time for more complete debug information. If this is set too
10452 low, value expressions that are available and could be represented in
10453 debug information may end up not being used; setting this higher may
10454 enable the compiler to find more complex debug expressions, but compile
10455 time and memory use may grow. The default is 12.
10457 @item min-nondebug-insn-uid
10458 Use uids starting at this parameter for nondebug insns. The range below
10459 the parameter is reserved exclusively for debug insns created by
10460 @option{-fvar-tracking-assignments}, but debug insns may get
10461 (non-overlapping) uids above it if the reserved range is exhausted.
10463 @item ipa-sra-ptr-growth-factor
10464 IPA-SRA replaces a pointer to an aggregate with one or more new
10465 parameters only when their cumulative size is less or equal to
10466 @option{ipa-sra-ptr-growth-factor} times the size of the original
10469 @item tm-max-aggregate-size
10470 When making copies of thread-local variables in a transaction, this
10471 parameter specifies the size in bytes after which variables are
10472 saved with the logging functions as opposed to save/restore code
10473 sequence pairs. This option only applies when using
10476 @item graphite-max-nb-scop-params
10477 To avoid exponential effects in the Graphite loop transforms, the
10478 number of parameters in a Static Control Part (SCoP) is bounded. The
10479 default value is 10 parameters. A variable whose value is unknown at
10480 compilation time and defined outside a SCoP is a parameter of the SCoP.
10482 @item graphite-max-bbs-per-function
10483 To avoid exponential effects in the detection of SCoPs, the size of
10484 the functions analyzed by Graphite is bounded. The default value is
10487 @item loop-block-tile-size
10488 Loop blocking or strip mining transforms, enabled with
10489 @option{-floop-block} or @option{-floop-strip-mine}, strip mine each
10490 loop in the loop nest by a given number of iterations. The strip
10491 length can be changed using the @option{loop-block-tile-size}
10492 parameter. The default value is 51 iterations.
10494 @item ipa-cp-value-list-size
10495 IPA-CP attempts to track all possible values and types passed to a function's
10496 parameter in order to propagate them and perform devirtualization.
10497 @option{ipa-cp-value-list-size} is the maximum number of values and types it
10498 stores per one formal parameter of a function.
10500 @item ipa-cp-eval-threshold
10501 IPA-CP calculates its own score of cloning profitability heuristics
10502 and performs those cloning opportunities with scores that exceed
10503 @option{ipa-cp-eval-threshold}.
10505 @item ipa-max-agg-items
10506 IPA-CP is also capable to propagate a number of scalar values passed
10507 in an aggregate. @option{ipa-max-agg-items} controls the maximum
10508 number of such values per one parameter.
10510 @item ipa-cp-loop-hint-bonus
10511 When IPA-CP determines that a cloning candidate would make the number
10512 of iterations of a loop known, it adds a bonus of
10513 @option{ipa-cp-loop-hint-bonus} to the profitability score of
10516 @item ipa-cp-array-index-hint-bonus
10517 When IPA-CP determines that a cloning candidate would make the index of
10518 an array access known, it adds a bonus of
10519 @option{ipa-cp-array-index-hint-bonus} to the profitability
10520 score of the candidate.
10522 @item ipa-max-aa-steps
10523 During its analysis of function bodies, IPA-CP employs alias analysis
10524 in order to track values pointed to by function parameters. In order
10525 not spend too much time analyzing huge functions, it will give up and
10526 consider all memory clobbered after examining
10527 @option{ipa-max-aa-steps} statements modifying memory.
10529 @item lto-partitions
10530 Specify desired number of partitions produced during WHOPR compilation.
10531 The number of partitions should exceed the number of CPUs used for compilation.
10532 The default value is 32.
10534 @item lto-minpartition
10535 Size of minimal partition for WHOPR (in estimated instructions).
10536 This prevents expenses of splitting very small programs into too many
10539 @item cxx-max-namespaces-for-diagnostic-help
10540 The maximum number of namespaces to consult for suggestions when C++
10541 name lookup fails for an identifier. The default is 1000.
10543 @item sink-frequency-threshold
10544 The maximum relative execution frequency (in percents) of the target block
10545 relative to a statement's original block to allow statement sinking of a
10546 statement. Larger numbers result in more aggressive statement sinking.
10547 The default value is 75. A small positive adjustment is applied for
10548 statements with memory operands as those are even more profitable so sink.
10550 @item max-stores-to-sink
10551 The maximum number of conditional stores paires that can be sunk. Set to 0
10552 if either vectorization (@option{-ftree-vectorize}) or if-conversion
10553 (@option{-ftree-loop-if-convert}) is disabled. The default is 2.
10555 @item allow-store-data-races
10556 Allow optimizers to introduce new data races on stores.
10557 Set to 1 to allow, otherwise to 0. This option is enabled by default
10558 at optimization level @option{-Ofast}.
10560 @item case-values-threshold
10561 The smallest number of different values for which it is best to use a
10562 jump-table instead of a tree of conditional branches. If the value is
10563 0, use the default for the machine. The default is 0.
10565 @item tree-reassoc-width
10566 Set the maximum number of instructions executed in parallel in
10567 reassociated tree. This parameter overrides target dependent
10568 heuristics used by default if has non zero value.
10570 @item sched-pressure-algorithm
10571 Choose between the two available implementations of
10572 @option{-fsched-pressure}. Algorithm 1 is the original implementation
10573 and is the more likely to prevent instructions from being reordered.
10574 Algorithm 2 was designed to be a compromise between the relatively
10575 conservative approach taken by algorithm 1 and the rather aggressive
10576 approach taken by the default scheduler. It relies more heavily on
10577 having a regular register file and accurate register pressure classes.
10578 See @file{haifa-sched.c} in the GCC sources for more details.
10580 The default choice depends on the target.
10582 @item max-slsr-cand-scan
10583 Set the maximum number of existing candidates that will be considered when
10584 seeking a basis for a new straight-line strength reduction candidate.
10586 @item coverage-callback
10587 Set to 1 to instrument a callback function together with
10588 arc counter update. The name of the callback function
10589 is "__coverage_callback".
10591 @item coverage-exec_once
10592 Set to 1 to update each arc counter only once.
10595 Enable buffer overflow detection for global objects. This kind
10596 of protection is enabled by default if you are using
10597 @option{-fsanitize=address} option.
10598 To disable global objects protection use @option{--param asan-globals=0}.
10601 Enable buffer overflow detection for stack objects. This kind of
10602 protection is enabled by default when using@option{-fsanitize=address}.
10603 To disable stack protection use @option{--param asan-stack=0} option.
10605 @item asan-instrument-reads
10606 Enable buffer overflow detection for memory reads. This kind of
10607 protection is enabled by default when using @option{-fsanitize=address}.
10608 To disable memory reads protection use
10609 @option{--param asan-instrument-reads=0}.
10611 @item asan-instrument-writes
10612 Enable buffer overflow detection for memory writes. This kind of
10613 protection is enabled by default when using @option{-fsanitize=address}.
10614 To disable memory writes protection use
10615 @option{--param asan-instrument-writes=0} option.
10617 @item asan-memintrin
10618 Enable detection for built-in functions. This kind of protection
10619 is enabled by default when using @option{-fsanitize=address}.
10620 To disable built-in functions protection use
10621 @option{--param asan-memintrin=0}.
10623 @item asan-use-after-return
10624 Enable detection of use-after-return. This kind of protection
10625 is enabled by default when using @option{-fsanitize=address} option.
10626 To disable use-after-return detection use
10627 @option{--param asan-use-after-return=0}.
10629 @item asan-instrumentation-with-call-threshold
10630 Once number of memory accesses in function becomes greater
10631 or equal than this number, use callbacks instead of
10632 generating inline code. E.g. to disable inline code use
10633 @option{--param asan-instrumentation-with-call-threshold=0}.
10638 @node Preprocessor Options
10639 @section Options Controlling the Preprocessor
10640 @cindex preprocessor options
10641 @cindex options, preprocessor
10643 These options control the C preprocessor, which is run on each C source
10644 file before actual compilation.
10646 If you use the @option{-E} option, nothing is done except preprocessing.
10647 Some of these options make sense only together with @option{-E} because
10648 they cause the preprocessor output to be unsuitable for actual
10652 @item -Wp,@var{option}
10654 You can use @option{-Wp,@var{option}} to bypass the compiler driver
10655 and pass @var{option} directly through to the preprocessor. If
10656 @var{option} contains commas, it is split into multiple options at the
10657 commas. However, many options are modified, translated or interpreted
10658 by the compiler driver before being passed to the preprocessor, and
10659 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
10660 interface is undocumented and subject to change, so whenever possible
10661 you should avoid using @option{-Wp} and let the driver handle the
10664 @item -Xpreprocessor @var{option}
10665 @opindex Xpreprocessor
10666 Pass @var{option} as an option to the preprocessor. You can use this to
10667 supply system-specific preprocessor options that GCC does not
10670 If you want to pass an option that takes an argument, you must use
10671 @option{-Xpreprocessor} twice, once for the option and once for the argument.
10673 @item -no-integrated-cpp
10674 @opindex no-integrated-cpp
10675 Perform preprocessing as a separate pass before compilation.
10676 By default, GCC performs preprocessing as an integrated part of
10677 input tokenization and parsing.
10678 If this option is provided, the appropriate language front end
10679 (@command{cc1}, @command{cc1plus}, or @command{cc1obj} for C, C++,
10680 and Objective-C, respectively) is instead invoked twice,
10681 once for preprocessing only and once for actual compilation
10682 of the preprocessed input.
10683 This option may be useful in conjunction with the @option{-B} or
10684 @option{-wrapper} options to specify an alternate preprocessor or
10685 perform additional processing of the program source between
10686 normal preprocessing and compilation.
10689 @include cppopts.texi
10691 @node Assembler Options
10692 @section Passing Options to the Assembler
10694 @c prevent bad page break with this line
10695 You can pass options to the assembler.
10698 @item -Wa,@var{option}
10700 Pass @var{option} as an option to the assembler. If @var{option}
10701 contains commas, it is split into multiple options at the commas.
10703 @item -Xassembler @var{option}
10704 @opindex Xassembler
10705 Pass @var{option} as an option to the assembler. You can use this to
10706 supply system-specific assembler options that GCC does not
10709 If you want to pass an option that takes an argument, you must use
10710 @option{-Xassembler} twice, once for the option and once for the argument.
10712 @item profile-generate-sampling-rate
10713 Set the sampling rate with @option{-fprofile-generate-sampling}.
10718 @section Options for Linking
10719 @cindex link options
10720 @cindex options, linking
10722 These options come into play when the compiler links object files into
10723 an executable output file. They are meaningless if the compiler is
10724 not doing a link step.
10728 @item @var{object-file-name}
10729 A file name that does not end in a special recognized suffix is
10730 considered to name an object file or library. (Object files are
10731 distinguished from libraries by the linker according to the file
10732 contents.) If linking is done, these object files are used as input
10741 If any of these options is used, then the linker is not run, and
10742 object file names should not be used as arguments. @xref{Overall
10746 @item -l@var{library}
10747 @itemx -l @var{library}
10749 Search the library named @var{library} when linking. (The second
10750 alternative with the library as a separate argument is only for
10751 POSIX compliance and is not recommended.)
10753 It makes a difference where in the command you write this option; the
10754 linker searches and processes libraries and object files in the order they
10755 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
10756 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
10757 to functions in @samp{z}, those functions may not be loaded.
10759 The linker searches a standard list of directories for the library,
10760 which is actually a file named @file{lib@var{library}.a}. The linker
10761 then uses this file as if it had been specified precisely by name.
10763 The directories searched include several standard system directories
10764 plus any that you specify with @option{-L}.
10766 Normally the files found this way are library files---archive files
10767 whose members are object files. The linker handles an archive file by
10768 scanning through it for members which define symbols that have so far
10769 been referenced but not defined. But if the file that is found is an
10770 ordinary object file, it is linked in the usual fashion. The only
10771 difference between using an @option{-l} option and specifying a file name
10772 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
10773 and searches several directories.
10777 You need this special case of the @option{-l} option in order to
10778 link an Objective-C or Objective-C++ program.
10780 @item -nostartfiles
10781 @opindex nostartfiles
10782 Do not use the standard system startup files when linking.
10783 The standard system libraries are used normally, unless @option{-nostdlib}
10784 or @option{-nodefaultlibs} is used.
10786 @item -nodefaultlibs
10787 @opindex nodefaultlibs
10788 Do not use the standard system libraries when linking.
10789 Only the libraries you specify are passed to the linker, and options
10790 specifying linkage of the system libraries, such as @code{-static-libgcc}
10791 or @code{-shared-libgcc}, are ignored.
10792 The standard startup files are used normally, unless @option{-nostartfiles}
10795 The compiler may generate calls to @code{memcmp},
10796 @code{memset}, @code{memcpy} and @code{memmove}.
10797 These entries are usually resolved by entries in
10798 libc. These entry points should be supplied through some other
10799 mechanism when this option is specified.
10803 Do not use the standard system startup files or libraries when linking.
10804 No startup files and only the libraries you specify are passed to
10805 the linker, and options specifying linkage of the system libraries, such as
10806 @code{-static-libgcc} or @code{-shared-libgcc}, are ignored.
10808 The compiler may generate calls to @code{memcmp}, @code{memset},
10809 @code{memcpy} and @code{memmove}.
10810 These entries are usually resolved by entries in
10811 libc. These entry points should be supplied through some other
10812 mechanism when this option is specified.
10814 @cindex @option{-lgcc}, use with @option{-nostdlib}
10815 @cindex @option{-nostdlib} and unresolved references
10816 @cindex unresolved references and @option{-nostdlib}
10817 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
10818 @cindex @option{-nodefaultlibs} and unresolved references
10819 @cindex unresolved references and @option{-nodefaultlibs}
10820 One of the standard libraries bypassed by @option{-nostdlib} and
10821 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
10822 which GCC uses to overcome shortcomings of particular machines, or special
10823 needs for some languages.
10824 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
10825 Collection (GCC) Internals},
10826 for more discussion of @file{libgcc.a}.)
10827 In most cases, you need @file{libgcc.a} even when you want to avoid
10828 other standard libraries. In other words, when you specify @option{-nostdlib}
10829 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
10830 This ensures that you have no unresolved references to internal GCC
10831 library subroutines.
10832 (An example of such an internal subroutine is @samp{__main}, used to ensure C++
10833 constructors are called; @pxref{Collect2,,@code{collect2}, gccint,
10834 GNU Compiler Collection (GCC) Internals}.)
10838 Produce a position independent executable on targets that support it.
10839 For predictable results, you must also specify the same set of options
10840 used for compilation (@option{-fpie}, @option{-fPIE},
10841 or model suboptions) when you specify this linker option.
10845 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
10846 that support it. This instructs the linker to add all symbols, not
10847 only used ones, to the dynamic symbol table. This option is needed
10848 for some uses of @code{dlopen} or to allow obtaining backtraces
10849 from within a program.
10853 Remove all symbol table and relocation information from the executable.
10857 On systems that support dynamic linking, this prevents linking with the shared
10858 libraries. On other systems, this option has no effect.
10862 Produce a shared object which can then be linked with other objects to
10863 form an executable. Not all systems support this option. For predictable
10864 results, you must also specify the same set of options used for compilation
10865 (@option{-fpic}, @option{-fPIC}, or model suboptions) when
10866 you specify this linker option.@footnote{On some systems, @samp{gcc -shared}
10867 needs to build supplementary stub code for constructors to work. On
10868 multi-libbed systems, @samp{gcc -shared} must select the correct support
10869 libraries to link against. Failing to supply the correct flags may lead
10870 to subtle defects. Supplying them in cases where they are not necessary
10873 @item -shared-libgcc
10874 @itemx -static-libgcc
10875 @opindex shared-libgcc
10876 @opindex static-libgcc
10877 On systems that provide @file{libgcc} as a shared library, these options
10878 force the use of either the shared or static version, respectively.
10879 If no shared version of @file{libgcc} was built when the compiler was
10880 configured, these options have no effect.
10882 There are several situations in which an application should use the
10883 shared @file{libgcc} instead of the static version. The most common
10884 of these is when the application wishes to throw and catch exceptions
10885 across different shared libraries. In that case, each of the libraries
10886 as well as the application itself should use the shared @file{libgcc}.
10888 Therefore, the G++ and GCJ drivers automatically add
10889 @option{-shared-libgcc} whenever you build a shared library or a main
10890 executable, because C++ and Java programs typically use exceptions, so
10891 this is the right thing to do.
10893 If, instead, you use the GCC driver to create shared libraries, you may
10894 find that they are not always linked with the shared @file{libgcc}.
10895 If GCC finds, at its configuration time, that you have a non-GNU linker
10896 or a GNU linker that does not support option @option{--eh-frame-hdr},
10897 it links the shared version of @file{libgcc} into shared libraries
10898 by default. Otherwise, it takes advantage of the linker and optimizes
10899 away the linking with the shared version of @file{libgcc}, linking with
10900 the static version of libgcc by default. This allows exceptions to
10901 propagate through such shared libraries, without incurring relocation
10902 costs at library load time.
10904 However, if a library or main executable is supposed to throw or catch
10905 exceptions, you must link it using the G++ or GCJ driver, as appropriate
10906 for the languages used in the program, or using the option
10907 @option{-shared-libgcc}, such that it is linked with the shared
10910 @item -static-libasan
10911 @opindex static-libasan
10912 When the @option{-fsanitize=address} option is used to link a program,
10913 the GCC driver automatically links against @option{libasan}. If
10914 @file{libasan} is available as a shared library, and the @option{-static}
10915 option is not used, then this links against the shared version of
10916 @file{libasan}. The @option{-static-libasan} option directs the GCC
10917 driver to link @file{libasan} statically, without necessarily linking
10918 other libraries statically.
10920 @item -static-libtsan
10921 @opindex static-libtsan
10922 When the @option{-fsanitize=thread} option is used to link a program,
10923 the GCC driver automatically links against @option{libtsan}. If
10924 @file{libtsan} is available as a shared library, and the @option{-static}
10925 option is not used, then this links against the shared version of
10926 @file{libtsan}. The @option{-static-libtsan} option directs the GCC
10927 driver to link @file{libtsan} statically, without necessarily linking
10928 other libraries statically.
10930 @item -static-liblsan
10931 @opindex static-liblsan
10932 When the @option{-fsanitize=leak} option is used to link a program,
10933 the GCC driver automatically links against @option{liblsan}. If
10934 @file{liblsan} is available as a shared library, and the @option{-static}
10935 option is not used, then this links against the shared version of
10936 @file{liblsan}. The @option{-static-liblsan} option directs the GCC
10937 driver to link @file{liblsan} statically, without necessarily linking
10938 other libraries statically.
10940 @item -static-libubsan
10941 @opindex static-libubsan
10942 When the @option{-fsanitize=undefined} option is used to link a program,
10943 the GCC driver automatically links against @option{libubsan}. If
10944 @file{libubsan} is available as a shared library, and the @option{-static}
10945 option is not used, then this links against the shared version of
10946 @file{libubsan}. The @option{-static-libubsan} option directs the GCC
10947 driver to link @file{libubsan} statically, without necessarily linking
10948 other libraries statically.
10950 @item -static-libstdc++
10951 @opindex static-libstdc++
10952 When the @command{g++} program is used to link a C++ program, it
10953 normally automatically links against @option{libstdc++}. If
10954 @file{libstdc++} is available as a shared library, and the
10955 @option{-static} option is not used, then this links against the
10956 shared version of @file{libstdc++}. That is normally fine. However, it
10957 is sometimes useful to freeze the version of @file{libstdc++} used by
10958 the program without going all the way to a fully static link. The
10959 @option{-static-libstdc++} option directs the @command{g++} driver to
10960 link @file{libstdc++} statically, without necessarily linking other
10961 libraries statically.
10965 Bind references to global symbols when building a shared object. Warn
10966 about any unresolved references (unless overridden by the link editor
10967 option @option{-Xlinker -z -Xlinker defs}). Only a few systems support
10970 @item -T @var{script}
10972 @cindex linker script
10973 Use @var{script} as the linker script. This option is supported by most
10974 systems using the GNU linker. On some targets, such as bare-board
10975 targets without an operating system, the @option{-T} option may be required
10976 when linking to avoid references to undefined symbols.
10978 @item -Xclang-only @var{option}
10979 @opindex Xclang-only
10980 Ignore @var{option}. This is used by some custom drivers to pass options
10981 to Clang but not GCC.
10983 @item -Xlinker @var{option}
10985 Pass @var{option} as an option to the linker. You can use this to
10986 supply system-specific linker options that GCC does not recognize.
10988 If you want to pass an option that takes a separate argument, you must use
10989 @option{-Xlinker} twice, once for the option and once for the argument.
10990 For example, to pass @option{-assert definitions}, you must write
10991 @option{-Xlinker -assert -Xlinker definitions}. It does not work to write
10992 @option{-Xlinker "-assert definitions"}, because this passes the entire
10993 string as a single argument, which is not what the linker expects.
10995 When using the GNU linker, it is usually more convenient to pass
10996 arguments to linker options using the @option{@var{option}=@var{value}}
10997 syntax than as separate arguments. For example, you can specify
10998 @option{-Xlinker -Map=output.map} rather than
10999 @option{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
11000 this syntax for command-line options.
11002 @item -Wl,@var{option}
11004 Pass @var{option} as an option to the linker. If @var{option} contains
11005 commas, it is split into multiple options at the commas. You can use this
11006 syntax to pass an argument to the option.
11007 For example, @option{-Wl,-Map,output.map} passes @option{-Map output.map} to the
11008 linker. When using the GNU linker, you can also get the same effect with
11009 @option{-Wl,-Map=output.map}.
11011 @item -u @var{symbol}
11013 Pretend the symbol @var{symbol} is undefined, to force linking of
11014 library modules to define it. You can use @option{-u} multiple times with
11015 different symbols to force loading of additional library modules.
11017 @item -z @var{keyword}
11019 @option{-z} is passed directly on to the linker along with the keyword
11020 @var{keyword}. See the section in the documentation of your linker for
11021 permitted values and their meanings.
11024 @node Directory Options
11025 @section Options for Directory Search
11026 @cindex directory options
11027 @cindex options, directory search
11028 @cindex search path
11030 These options specify directories to search for header files, for
11031 libraries and for parts of the compiler:
11036 Add the directory @var{dir} to the head of the list of directories to be
11037 searched for header files. This can be used to override a system header
11038 file, substituting your own version, since these directories are
11039 searched before the system header file directories. However, you should
11040 not use this option to add directories that contain vendor-supplied
11041 system header files (use @option{-isystem} for that). If you use more than
11042 one @option{-I} option, the directories are scanned in left-to-right
11043 order; the standard system directories come after.
11045 If a standard system include directory, or a directory specified with
11046 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
11047 option is ignored. The directory is still searched but as a
11048 system directory at its normal position in the system include chain.
11049 This is to ensure that GCC's procedure to fix buggy system headers and
11050 the ordering for the @code{include_next} directive are not inadvertently changed.
11051 If you really need to change the search order for system directories,
11052 use the @option{-nostdinc} and/or @option{-isystem} options.
11054 @item -iplugindir=@var{dir}
11055 @opindex iplugindir=
11056 Set the directory to search for plugins that are passed
11057 by @option{-fplugin=@var{name}} instead of
11058 @option{-fplugin=@var{path}/@var{name}.so}. This option is not meant
11059 to be used by the user, but only passed by the driver.
11061 @item -iquote@var{dir}
11063 Add the directory @var{dir} to the head of the list of directories to
11064 be searched for header files only for the case of @samp{#include
11065 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
11066 otherwise just like @option{-I}.
11070 Add directory @var{dir} to the list of directories to be searched
11073 @item -B@var{prefix}
11075 This option specifies where to find the executables, libraries,
11076 include files, and data files of the compiler itself.
11078 The compiler driver program runs one or more of the subprograms
11079 @command{cpp}, @command{cc1}, @command{as} and @command{ld}. It tries
11080 @var{prefix} as a prefix for each program it tries to run, both with and
11081 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
11083 For each subprogram to be run, the compiler driver first tries the
11084 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
11085 is not specified, the driver tries two standard prefixes,
11086 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
11087 those results in a file name that is found, the unmodified program
11088 name is searched for using the directories specified in your
11089 @env{PATH} environment variable.
11091 The compiler checks to see if the path provided by the @option{-B}
11092 refers to a directory, and if necessary it adds a directory
11093 separator character at the end of the path.
11095 @option{-B} prefixes that effectively specify directory names also apply
11096 to libraries in the linker, because the compiler translates these
11097 options into @option{-L} options for the linker. They also apply to
11098 include files in the preprocessor, because the compiler translates these
11099 options into @option{-isystem} options for the preprocessor. In this case,
11100 the compiler appends @samp{include} to the prefix.
11102 The runtime support file @file{libgcc.a} can also be searched for using
11103 the @option{-B} prefix, if needed. If it is not found there, the two
11104 standard prefixes above are tried, and that is all. The file is left
11105 out of the link if it is not found by those means.
11107 Another way to specify a prefix much like the @option{-B} prefix is to use
11108 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
11111 As a special kludge, if the path provided by @option{-B} is
11112 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
11113 9, then it is replaced by @file{[dir/]include}. This is to help
11114 with boot-strapping the compiler.
11116 @item -specs=@var{file}
11118 Process @var{file} after the compiler reads in the standard @file{specs}
11119 file, in order to override the defaults which the @command{gcc} driver
11120 program uses when determining what switches to pass to @command{cc1},
11121 @command{cc1plus}, @command{as}, @command{ld}, etc. More than one
11122 @option{-specs=@var{file}} can be specified on the command line, and they
11123 are processed in order, from left to right.
11125 @item --sysroot=@var{dir}
11127 Use @var{dir} as the logical root directory for headers and libraries.
11128 For example, if the compiler normally searches for headers in
11129 @file{/usr/include} and libraries in @file{/usr/lib}, it instead
11130 searches @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
11132 If you use both this option and the @option{-isysroot} option, then
11133 the @option{--sysroot} option applies to libraries, but the
11134 @option{-isysroot} option applies to header files.
11136 The GNU linker (beginning with version 2.16) has the necessary support
11137 for this option. If your linker does not support this option, the
11138 header file aspect of @option{--sysroot} still works, but the
11139 library aspect does not.
11141 @item --no-sysroot-suffix
11142 @opindex no-sysroot-suffix
11143 For some targets, a suffix is added to the root directory specified
11144 with @option{--sysroot}, depending on the other options used, so that
11145 headers may for example be found in
11146 @file{@var{dir}/@var{suffix}/usr/include} instead of
11147 @file{@var{dir}/usr/include}. This option disables the addition of
11152 This option has been deprecated. Please use @option{-iquote} instead for
11153 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
11154 Any directories you specify with @option{-I} options before the @option{-I-}
11155 option are searched only for the case of @samp{#include "@var{file}"};
11156 they are not searched for @samp{#include <@var{file}>}.
11158 If additional directories are specified with @option{-I} options after
11159 the @option{-I-}, these directories are searched for all @samp{#include}
11160 directives. (Ordinarily @emph{all} @option{-I} directories are used
11163 In addition, the @option{-I-} option inhibits the use of the current
11164 directory (where the current input file came from) as the first search
11165 directory for @samp{#include "@var{file}"}. There is no way to
11166 override this effect of @option{-I-}. With @option{-I.} you can specify
11167 searching the directory that is current when the compiler is
11168 invoked. That is not exactly the same as what the preprocessor does
11169 by default, but it is often satisfactory.
11171 @option{-I-} does not inhibit the use of the standard system directories
11172 for header files. Thus, @option{-I-} and @option{-nostdinc} are
11179 @section Specifying subprocesses and the switches to pass to them
11182 @command{gcc} is a driver program. It performs its job by invoking a
11183 sequence of other programs to do the work of compiling, assembling and
11184 linking. GCC interprets its command-line parameters and uses these to
11185 deduce which programs it should invoke, and which command-line options
11186 it ought to place on their command lines. This behavior is controlled
11187 by @dfn{spec strings}. In most cases there is one spec string for each
11188 program that GCC can invoke, but a few programs have multiple spec
11189 strings to control their behavior. The spec strings built into GCC can
11190 be overridden by using the @option{-specs=} command-line switch to specify
11193 @dfn{Spec files} are plaintext files that are used to construct spec
11194 strings. They consist of a sequence of directives separated by blank
11195 lines. The type of directive is determined by the first non-whitespace
11196 character on the line, which can be one of the following:
11199 @item %@var{command}
11200 Issues a @var{command} to the spec file processor. The commands that can
11204 @item %include <@var{file}>
11205 @cindex @code{%include}
11206 Search for @var{file} and insert its text at the current point in the
11209 @item %include_noerr <@var{file}>
11210 @cindex @code{%include_noerr}
11211 Just like @samp{%include}, but do not generate an error message if the include
11212 file cannot be found.
11214 @item %rename @var{old_name} @var{new_name}
11215 @cindex @code{%rename}
11216 Rename the spec string @var{old_name} to @var{new_name}.
11220 @item *[@var{spec_name}]:
11221 This tells the compiler to create, override or delete the named spec
11222 string. All lines after this directive up to the next directive or
11223 blank line are considered to be the text for the spec string. If this
11224 results in an empty string then the spec is deleted. (Or, if the
11225 spec did not exist, then nothing happens.) Otherwise, if the spec
11226 does not currently exist a new spec is created. If the spec does
11227 exist then its contents are overridden by the text of this
11228 directive, unless the first character of that text is the @samp{+}
11229 character, in which case the text is appended to the spec.
11231 @item [@var{suffix}]:
11232 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
11233 and up to the next directive or blank line are considered to make up the
11234 spec string for the indicated suffix. When the compiler encounters an
11235 input file with the named suffix, it processes the spec string in
11236 order to work out how to compile that file. For example:
11240 z-compile -input %i
11243 This says that any input file whose name ends in @samp{.ZZ} should be
11244 passed to the program @samp{z-compile}, which should be invoked with the
11245 command-line switch @option{-input} and with the result of performing the
11246 @samp{%i} substitution. (See below.)
11248 As an alternative to providing a spec string, the text following a
11249 suffix directive can be one of the following:
11252 @item @@@var{language}
11253 This says that the suffix is an alias for a known @var{language}. This is
11254 similar to using the @option{-x} command-line switch to GCC to specify a
11255 language explicitly. For example:
11262 Says that .ZZ files are, in fact, C++ source files.
11265 This causes an error messages saying:
11268 @var{name} compiler not installed on this system.
11272 GCC already has an extensive list of suffixes built into it.
11273 This directive adds an entry to the end of the list of suffixes, but
11274 since the list is searched from the end backwards, it is effectively
11275 possible to override earlier entries using this technique.
11279 GCC has the following spec strings built into it. Spec files can
11280 override these strings or create their own. Note that individual
11281 targets can also add their own spec strings to this list.
11284 asm Options to pass to the assembler
11285 asm_final Options to pass to the assembler post-processor
11286 cpp Options to pass to the C preprocessor
11287 cc1 Options to pass to the C compiler
11288 cc1plus Options to pass to the C++ compiler
11289 endfile Object files to include at the end of the link
11290 link Options to pass to the linker
11291 lib Libraries to include on the command line to the linker
11292 libgcc Decides which GCC support library to pass to the linker
11293 linker Sets the name of the linker
11294 predefines Defines to be passed to the C preprocessor
11295 signed_char Defines to pass to CPP to say whether @code{char} is signed
11297 startfile Object files to include at the start of the link
11300 Here is a small example of a spec file:
11303 %rename lib old_lib
11306 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
11309 This example renames the spec called @samp{lib} to @samp{old_lib} and
11310 then overrides the previous definition of @samp{lib} with a new one.
11311 The new definition adds in some extra command-line options before
11312 including the text of the old definition.
11314 @dfn{Spec strings} are a list of command-line options to be passed to their
11315 corresponding program. In addition, the spec strings can contain
11316 @samp{%}-prefixed sequences to substitute variable text or to
11317 conditionally insert text into the command line. Using these constructs
11318 it is possible to generate quite complex command lines.
11320 Here is a table of all defined @samp{%}-sequences for spec
11321 strings. Note that spaces are not generated automatically around the
11322 results of expanding these sequences. Therefore you can concatenate them
11323 together or combine them with constant text in a single argument.
11327 Substitute one @samp{%} into the program name or argument.
11330 Substitute the name of the input file being processed.
11333 Substitute the basename of the input file being processed.
11334 This is the substring up to (and not including) the last period
11335 and not including the directory.
11338 This is the same as @samp{%b}, but include the file suffix (text after
11342 Marks the argument containing or following the @samp{%d} as a
11343 temporary file name, so that that file is deleted if GCC exits
11344 successfully. Unlike @samp{%g}, this contributes no text to the
11347 @item %g@var{suffix}
11348 Substitute a file name that has suffix @var{suffix} and is chosen
11349 once per compilation, and mark the argument in the same way as
11350 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
11351 name is now chosen in a way that is hard to predict even when previously
11352 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
11353 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
11354 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
11355 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
11356 was simply substituted with a file name chosen once per compilation,
11357 without regard to any appended suffix (which was therefore treated
11358 just like ordinary text), making such attacks more likely to succeed.
11360 @item %u@var{suffix}
11361 Like @samp{%g}, but generates a new temporary file name
11362 each time it appears instead of once per compilation.
11364 @item %U@var{suffix}
11365 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
11366 new one if there is no such last file name. In the absence of any
11367 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
11368 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
11369 involves the generation of two distinct file names, one
11370 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
11371 simply substituted with a file name chosen for the previous @samp{%u},
11372 without regard to any appended suffix.
11374 @item %j@var{suffix}
11375 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
11376 writable, and if @option{-save-temps} is not used;
11377 otherwise, substitute the name
11378 of a temporary file, just like @samp{%u}. This temporary file is not
11379 meant for communication between processes, but rather as a junk
11380 disposal mechanism.
11382 @item %|@var{suffix}
11383 @itemx %m@var{suffix}
11384 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
11385 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
11386 all. These are the two most common ways to instruct a program that it
11387 should read from standard input or write to standard output. If you
11388 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
11389 construct: see for example @file{f/lang-specs.h}.
11391 @item %.@var{SUFFIX}
11392 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
11393 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
11394 terminated by the next space or %.
11397 Marks the argument containing or following the @samp{%w} as the
11398 designated output file of this compilation. This puts the argument
11399 into the sequence of arguments that @samp{%o} substitutes.
11402 Substitutes the names of all the output files, with spaces
11403 automatically placed around them. You should write spaces
11404 around the @samp{%o} as well or the results are undefined.
11405 @samp{%o} is for use in the specs for running the linker.
11406 Input files whose names have no recognized suffix are not compiled
11407 at all, but they are included among the output files, so they are
11411 Substitutes the suffix for object files. Note that this is
11412 handled specially when it immediately follows @samp{%g, %u, or %U},
11413 because of the need for those to form complete file names. The
11414 handling is such that @samp{%O} is treated exactly as if it had already
11415 been substituted, except that @samp{%g, %u, and %U} do not currently
11416 support additional @var{suffix} characters following @samp{%O} as they do
11417 following, for example, @samp{.o}.
11420 Substitutes the standard macro predefinitions for the
11421 current target machine. Use this when running @code{cpp}.
11424 Like @samp{%p}, but puts @samp{__} before and after the name of each
11425 predefined macro, except for macros that start with @samp{__} or with
11426 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
11430 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
11431 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
11432 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
11433 and @option{-imultilib} as necessary.
11436 Current argument is the name of a library or startup file of some sort.
11437 Search for that file in a standard list of directories and substitute
11438 the full name found. The current working directory is included in the
11439 list of directories scanned.
11442 Current argument is the name of a linker script. Search for that file
11443 in the current list of directories to scan for libraries. If the file
11444 is located insert a @option{--script} option into the command line
11445 followed by the full path name found. If the file is not found then
11446 generate an error message. Note: the current working directory is not
11450 Print @var{str} as an error message. @var{str} is terminated by a newline.
11451 Use this when inconsistent options are detected.
11453 @item %(@var{name})
11454 Substitute the contents of spec string @var{name} at this point.
11456 @item %x@{@var{option}@}
11457 Accumulate an option for @samp{%X}.
11460 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
11464 Output the accumulated assembler options specified by @option{-Wa}.
11467 Output the accumulated preprocessor options specified by @option{-Wp}.
11470 Process the @code{asm} spec. This is used to compute the
11471 switches to be passed to the assembler.
11474 Process the @code{asm_final} spec. This is a spec string for
11475 passing switches to an assembler post-processor, if such a program is
11479 Process the @code{link} spec. This is the spec for computing the
11480 command line passed to the linker. Typically it makes use of the
11481 @samp{%L %G %S %D and %E} sequences.
11484 Dump out a @option{-L} option for each directory that GCC believes might
11485 contain startup files. If the target supports multilibs then the
11486 current multilib directory is prepended to each of these paths.
11489 Process the @code{lib} spec. This is a spec string for deciding which
11490 libraries are included on the command line to the linker.
11493 Process the @code{libgcc} spec. This is a spec string for deciding
11494 which GCC support library is included on the command line to the linker.
11497 Process the @code{startfile} spec. This is a spec for deciding which
11498 object files are the first ones passed to the linker. Typically
11499 this might be a file named @file{crt0.o}.
11502 Process the @code{endfile} spec. This is a spec string that specifies
11503 the last object files that are passed to the linker.
11506 Process the @code{cpp} spec. This is used to construct the arguments
11507 to be passed to the C preprocessor.
11510 Process the @code{cc1} spec. This is used to construct the options to be
11511 passed to the actual C compiler (@samp{cc1}).
11514 Process the @code{cc1plus} spec. This is used to construct the options to be
11515 passed to the actual C++ compiler (@samp{cc1plus}).
11518 Substitute the variable part of a matched option. See below.
11519 Note that each comma in the substituted string is replaced by
11523 Remove all occurrences of @code{-S} from the command line. Note---this
11524 command is position dependent. @samp{%} commands in the spec string
11525 before this one see @code{-S}, @samp{%} commands in the spec string
11526 after this one do not.
11528 @item %:@var{function}(@var{args})
11529 Call the named function @var{function}, passing it @var{args}.
11530 @var{args} is first processed as a nested spec string, then split
11531 into an argument vector in the usual fashion. The function returns
11532 a string which is processed as if it had appeared literally as part
11533 of the current spec.
11535 The following built-in spec functions are provided:
11538 @item @code{getenv}
11539 The @code{getenv} spec function takes two arguments: an environment
11540 variable name and a string. If the environment variable is not
11541 defined, a fatal error is issued. Otherwise, the return value is the
11542 value of the environment variable concatenated with the string. For
11543 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
11546 %:getenv(TOPDIR /include)
11549 expands to @file{/path/to/top/include}.
11551 @item @code{if-exists}
11552 The @code{if-exists} spec function takes one argument, an absolute
11553 pathname to a file. If the file exists, @code{if-exists} returns the
11554 pathname. Here is a small example of its usage:
11558 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
11561 @item @code{if-exists-else}
11562 The @code{if-exists-else} spec function is similar to the @code{if-exists}
11563 spec function, except that it takes two arguments. The first argument is
11564 an absolute pathname to a file. If the file exists, @code{if-exists-else}
11565 returns the pathname. If it does not exist, it returns the second argument.
11566 This way, @code{if-exists-else} can be used to select one file or another,
11567 based on the existence of the first. Here is a small example of its usage:
11571 crt0%O%s %:if-exists(crti%O%s) \
11572 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
11575 @item @code{replace-outfile}
11576 The @code{replace-outfile} spec function takes two arguments. It looks for the
11577 first argument in the outfiles array and replaces it with the second argument. Here
11578 is a small example of its usage:
11581 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
11584 @item @code{remove-outfile}
11585 The @code{remove-outfile} spec function takes one argument. It looks for the
11586 first argument in the outfiles array and removes it. Here is a small example
11590 %:remove-outfile(-lm)
11593 @item @code{pass-through-libs}
11594 The @code{pass-through-libs} spec function takes any number of arguments. It
11595 finds any @option{-l} options and any non-options ending in @file{.a} (which it
11596 assumes are the names of linker input library archive files) and returns a
11597 result containing all the found arguments each prepended by
11598 @option{-plugin-opt=-pass-through=} and joined by spaces. This list is
11599 intended to be passed to the LTO linker plugin.
11602 %:pass-through-libs(%G %L %G)
11605 @item @code{print-asm-header}
11606 The @code{print-asm-header} function takes no arguments and simply
11607 prints a banner like:
11613 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
11616 It is used to separate compiler options from assembler options
11617 in the @option{--target-help} output.
11620 @item %@{@code{S}@}
11621 Substitutes the @code{-S} switch, if that switch is given to GCC@.
11622 If that switch is not specified, this substitutes nothing. Note that
11623 the leading dash is omitted when specifying this option, and it is
11624 automatically inserted if the substitution is performed. Thus the spec
11625 string @samp{%@{foo@}} matches the command-line option @option{-foo}
11626 and outputs the command-line option @option{-foo}.
11628 @item %W@{@code{S}@}
11629 Like %@{@code{S}@} but mark last argument supplied within as a file to be
11630 deleted on failure.
11632 @item %@{@code{S}*@}
11633 Substitutes all the switches specified to GCC whose names start
11634 with @code{-S}, but which also take an argument. This is used for
11635 switches like @option{-o}, @option{-D}, @option{-I}, etc.
11636 GCC considers @option{-o foo} as being
11637 one switch whose name starts with @samp{o}. %@{o*@} substitutes this
11638 text, including the space. Thus two arguments are generated.
11640 @item %@{@code{S}*&@code{T}*@}
11641 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
11642 (the order of @code{S} and @code{T} in the spec is not significant).
11643 There can be any number of ampersand-separated variables; for each the
11644 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
11646 @item %@{@code{S}:@code{X}@}
11647 Substitutes @code{X}, if the @option{-S} switch is given to GCC@.
11649 @item %@{!@code{S}:@code{X}@}
11650 Substitutes @code{X}, if the @option{-S} switch is @emph{not} given to GCC@.
11652 @item %@{@code{S}*:@code{X}@}
11653 Substitutes @code{X} if one or more switches whose names start with
11654 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
11655 once, no matter how many such switches appeared. However, if @code{%*}
11656 appears somewhere in @code{X}, then @code{X} is substituted once
11657 for each matching switch, with the @code{%*} replaced by the part of
11658 that switch matching the @code{*}.
11660 If @code{%*} appears as the last part of a spec sequence then a space
11661 will be added after the end of the last substitution. If there is more
11662 text in the sequence however then a space will not be generated. This
11663 allows the @code{%*} substitution to be used as part of a larger
11664 string. For example, a spec string like this:
11667 %@{mcu=*:--script=%*/memory.ld@}
11670 when matching an option like @code{-mcu=newchip} will produce:
11673 --script=newchip/memory.ld
11676 @item %@{.@code{S}:@code{X}@}
11677 Substitutes @code{X}, if processing a file with suffix @code{S}.
11679 @item %@{!.@code{S}:@code{X}@}
11680 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
11682 @item %@{,@code{S}:@code{X}@}
11683 Substitutes @code{X}, if processing a file for language @code{S}.
11685 @item %@{!,@code{S}:@code{X}@}
11686 Substitutes @code{X}, if not processing a file for language @code{S}.
11688 @item %@{@code{S}|@code{P}:@code{X}@}
11689 Substitutes @code{X} if either @code{-S} or @code{-P} is given to
11690 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
11691 @code{*} sequences as well, although they have a stronger binding than
11692 the @samp{|}. If @code{%*} appears in @code{X}, all of the
11693 alternatives must be starred, and only the first matching alternative
11696 For example, a spec string like this:
11699 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
11703 outputs the following command-line options from the following input
11704 command-line options:
11709 -d fred.c -foo -baz -boggle
11710 -d jim.d -bar -baz -boggle
11713 @item %@{S:X; T:Y; :D@}
11715 If @code{S} is given to GCC, substitutes @code{X}; else if @code{T} is
11716 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
11717 be as many clauses as you need. This may be combined with @code{.},
11718 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
11721 When importing auxiliary modules during profile-use, check current
11722 memory consumption after parsing each auxiliary module. If it exceeds
11723 this limit (specified in kb), don't import any more auxiliary modules.
11724 Specifying a value of 0 means don't enforce this limit. This parameter
11725 is only useful when using @option{-fprofile-use} and @option{-fripa}.
11729 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
11730 construct may contain other nested @samp{%} constructs or spaces, or
11731 even newlines. They are processed as usual, as described above.
11732 Trailing white space in @code{X} is ignored. White space may also
11733 appear anywhere on the left side of the colon in these constructs,
11734 except between @code{.} or @code{*} and the corresponding word.
11736 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
11737 handled specifically in these constructs. If another value of
11738 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
11739 @option{-W} switch is found later in the command line, the earlier
11740 switch value is ignored, except with @{@code{S}*@} where @code{S} is
11741 just one letter, which passes all matching options.
11743 The character @samp{|} at the beginning of the predicate text is used to
11744 indicate that a command should be piped to the following command, but
11745 only if @option{-pipe} is specified.
11747 It is built into GCC which switches take arguments and which do not.
11748 (You might think it would be useful to generalize this to allow each
11749 compiler's spec to say which switches take arguments. But this cannot
11750 be done in a consistent fashion. GCC cannot even decide which input
11751 files have been specified without knowing which switches take arguments,
11752 and it must know which input files to compile in order to tell which
11755 GCC also knows implicitly that arguments starting in @option{-l} are to be
11756 treated as compiler output files, and passed to the linker in their
11757 proper position among the other output files.
11759 @c man begin OPTIONS
11761 @node Target Options
11762 @section Specifying Target Machine and Compiler Version
11763 @cindex target options
11764 @cindex cross compiling
11765 @cindex specifying machine version
11766 @cindex specifying compiler version and target machine
11767 @cindex compiler version, specifying
11768 @cindex target machine, specifying
11770 The usual way to run GCC is to run the executable called @command{gcc}, or
11771 @command{@var{machine}-gcc} when cross-compiling, or
11772 @command{@var{machine}-gcc-@var{version}} to run a version other than the
11773 one that was installed last.
11775 @node Submodel Options
11776 @section Hardware Models and Configurations
11777 @cindex submodel options
11778 @cindex specifying hardware config
11779 @cindex hardware models and configurations, specifying
11780 @cindex machine dependent options
11782 Each target machine types can have its own
11783 special options, starting with @samp{-m}, to choose among various
11784 hardware models or configurations---for example, 68010 vs 68020,
11785 floating coprocessor or none. A single installed version of the
11786 compiler can compile for any model or configuration, according to the
11789 Some configurations of the compiler also support additional special
11790 options, usually for compatibility with other compilers on the same
11793 @c This list is ordered alphanumerically by subsection name.
11794 @c It should be the same order and spelling as these options are listed
11795 @c in Machine Dependent Options
11798 * AArch64 Options::
11799 * Adapteva Epiphany Options::
11803 * Blackfin Options::
11808 * DEC Alpha Options::
11811 * GNU/Linux Options::
11814 * i386 and x86-64 Options::
11815 * i386 and x86-64 Windows Options::
11823 * MicroBlaze Options::
11826 * MN10300 Options::
11830 * Nios II Options::
11832 * picoChip Options::
11833 * PowerPC Options::
11835 * RS/6000 and PowerPC Options::
11837 * S/390 and zSeries Options::
11840 * Solaris 2 Options::
11843 * System V Options::
11844 * TILE-Gx Options::
11845 * TILEPro Options::
11849 * VxWorks Options::
11851 * Xstormy16 Options::
11853 * zSeries Options::
11856 @node AArch64 Options
11857 @subsection AArch64 Options
11858 @cindex AArch64 Options
11860 These options are defined for AArch64 implementations:
11864 @item -mabi=@var{name}
11866 Generate code for the specified data model. Permissible values
11867 are @samp{ilp32} for SysV-like data model where int, long int and pointer
11868 are 32-bit, and @samp{lp64} for SysV-like data model where int is 32-bit,
11869 but long int and pointer are 64-bit.
11871 The default depends on the specific target configuration. Note that
11872 the LP64 and ILP32 ABIs are not link-compatible; you must compile your
11873 entire program with the same ABI, and link with a compatible set of libraries.
11876 @opindex mbig-endian
11877 Generate big-endian code. This is the default when GCC is configured for an
11878 @samp{aarch64_be-*-*} target.
11880 @item -mgeneral-regs-only
11881 @opindex mgeneral-regs-only
11882 Generate code which uses only the general registers.
11884 @item -mlittle-endian
11885 @opindex mlittle-endian
11886 Generate little-endian code. This is the default when GCC is configured for an
11887 @samp{aarch64-*-*} but not an @samp{aarch64_be-*-*} target.
11889 @item -mcmodel=tiny
11890 @opindex mcmodel=tiny
11891 Generate code for the tiny code model. The program and its statically defined
11892 symbols must be within 1GB of each other. Pointers are 64 bits. Programs can
11893 be statically or dynamically linked. This model is not fully implemented and
11894 mostly treated as @samp{small}.
11896 @item -mcmodel=small
11897 @opindex mcmodel=small
11898 Generate code for the small code model. The program and its statically defined
11899 symbols must be within 4GB of each other. Pointers are 64 bits. Programs can
11900 be statically or dynamically linked. This is the default code model.
11902 @item -mcmodel=large
11903 @opindex mcmodel=large
11904 Generate code for the large code model. This makes no assumptions about
11905 addresses and sizes of sections. Pointers are 64 bits. Programs can be
11906 statically linked only.
11908 @item -mstrict-align
11909 @opindex mstrict-align
11910 Do not assume that unaligned memory references will be handled by the system.
11912 @item -momit-leaf-frame-pointer
11913 @itemx -mno-omit-leaf-frame-pointer
11914 @opindex momit-leaf-frame-pointer
11915 @opindex mno-omit-leaf-frame-pointer
11916 Omit or keep the frame pointer in leaf functions. The former behaviour is the
11919 @item -mtls-dialect=desc
11920 @opindex mtls-dialect=desc
11921 Use TLS descriptors as the thread-local storage mechanism for dynamic accesses
11922 of TLS variables. This is the default.
11924 @item -mtls-dialect=traditional
11925 @opindex mtls-dialect=traditional
11926 Use traditional TLS as the thread-local storage mechanism for dynamic accesses
11929 @item -march=@var{name}
11931 Specify the name of the target architecture, optionally suffixed by one or
11932 more feature modifiers. This option has the form
11933 @option{-march=@var{arch}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}, where the
11934 only permissible value for @var{arch} is @samp{armv8-a}. The permissible
11935 values for @var{feature} are documented in the sub-section below.
11937 Where conflicting feature modifiers are specified, the right-most feature is
11940 GCC uses this name to determine what kind of instructions it can emit when
11941 generating assembly code.
11943 Where @option{-march} is specified without either of @option{-mtune}
11944 or @option{-mcpu} also being specified, the code will be tuned to perform
11945 well across a range of target processors implementing the target
11948 @item -mtune=@var{name}
11950 Specify the name of the target processor for which GCC should tune the
11951 performance of the code. Permissible values for this option are:
11952 @samp{generic}, @samp{cortex-a53}, @samp{cortex-a57}.
11954 Additionally, this option can specify that GCC should tune the performance
11955 of the code for a big.LITTLE system. The only permissible value is
11956 @samp{cortex-a57.cortex-a53}.
11958 Where none of @option{-mtune=}, @option{-mcpu=} or @option{-march=}
11959 are specified, the code will be tuned to perform well across a range
11960 of target processors.
11962 This option cannot be suffixed by feature modifiers.
11964 @item -mcpu=@var{name}
11966 Specify the name of the target processor, optionally suffixed by one or more
11967 feature modifiers. This option has the form
11968 @option{-mcpu=@var{cpu}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}, where the
11969 permissible values for @var{cpu} are the same as those available for
11972 The permissible values for @var{feature} are documented in the sub-section
11975 Where conflicting feature modifiers are specified, the right-most feature is
11978 GCC uses this name to determine what kind of instructions it can emit when
11979 generating assembly code (as if by @option{-march}) and to determine
11980 the target processor for which to tune for performance (as if
11981 by @option{-mtune}). Where this option is used in conjunction
11982 with @option{-march} or @option{-mtune}, those options take precedence
11983 over the appropriate part of this option.
11986 @subsubsection @option{-march} and @option{-mcpu} feature modifiers
11987 @cindex @option{-march} feature modifiers
11988 @cindex @option{-mcpu} feature modifiers
11989 Feature modifiers used with @option{-march} and @option{-mcpu} can be one
11994 Enable CRC extension.
11996 Enable Crypto extension. This implies Advanced SIMD is enabled.
11998 Enable floating-point instructions.
12000 Enable Advanced SIMD instructions. This implies floating-point instructions
12001 are enabled. This is the default for all current possible values for options
12002 @option{-march} and @option{-mcpu=}.
12005 @node Adapteva Epiphany Options
12006 @subsection Adapteva Epiphany Options
12008 These @samp{-m} options are defined for Adapteva Epiphany:
12011 @item -mhalf-reg-file
12012 @opindex mhalf-reg-file
12013 Don't allocate any register in the range @code{r32}@dots{}@code{r63}.
12014 That allows code to run on hardware variants that lack these registers.
12016 @item -mprefer-short-insn-regs
12017 @opindex mprefer-short-insn-regs
12018 Preferrentially allocate registers that allow short instruction generation.
12019 This can result in increased instruction count, so this may either reduce or
12020 increase overall code size.
12022 @item -mbranch-cost=@var{num}
12023 @opindex mbranch-cost
12024 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12025 This cost is only a heuristic and is not guaranteed to produce
12026 consistent results across releases.
12030 Enable the generation of conditional moves.
12032 @item -mnops=@var{num}
12034 Emit @var{num} NOPs before every other generated instruction.
12036 @item -mno-soft-cmpsf
12037 @opindex mno-soft-cmpsf
12038 For single-precision floating-point comparisons, emit an @code{fsub} instruction
12039 and test the flags. This is faster than a software comparison, but can
12040 get incorrect results in the presence of NaNs, or when two different small
12041 numbers are compared such that their difference is calculated as zero.
12042 The default is @option{-msoft-cmpsf}, which uses slower, but IEEE-compliant,
12043 software comparisons.
12045 @item -mstack-offset=@var{num}
12046 @opindex mstack-offset
12047 Set the offset between the top of the stack and the stack pointer.
12048 E.g., a value of 8 means that the eight bytes in the range @code{sp+0@dots{}sp+7}
12049 can be used by leaf functions without stack allocation.
12050 Values other than @samp{8} or @samp{16} are untested and unlikely to work.
12051 Note also that this option changes the ABI; compiling a program with a
12052 different stack offset than the libraries have been compiled with
12053 generally does not work.
12054 This option can be useful if you want to evaluate if a different stack
12055 offset would give you better code, but to actually use a different stack
12056 offset to build working programs, it is recommended to configure the
12057 toolchain with the appropriate @option{--with-stack-offset=@var{num}} option.
12059 @item -mno-round-nearest
12060 @opindex mno-round-nearest
12061 Make the scheduler assume that the rounding mode has been set to
12062 truncating. The default is @option{-mround-nearest}.
12065 @opindex mlong-calls
12066 If not otherwise specified by an attribute, assume all calls might be beyond
12067 the offset range of the @code{b} / @code{bl} instructions, and therefore load the
12068 function address into a register before performing a (otherwise direct) call.
12069 This is the default.
12071 @item -mshort-calls
12072 @opindex short-calls
12073 If not otherwise specified by an attribute, assume all direct calls are
12074 in the range of the @code{b} / @code{bl} instructions, so use these instructions
12075 for direct calls. The default is @option{-mlong-calls}.
12079 Assume addresses can be loaded as 16-bit unsigned values. This does not
12080 apply to function addresses for which @option{-mlong-calls} semantics
12083 @item -mfp-mode=@var{mode}
12085 Set the prevailing mode of the floating-point unit.
12086 This determines the floating-point mode that is provided and expected
12087 at function call and return time. Making this mode match the mode you
12088 predominantly need at function start can make your programs smaller and
12089 faster by avoiding unnecessary mode switches.
12091 @var{mode} can be set to one the following values:
12095 Any mode at function entry is valid, and retained or restored when
12096 the function returns, and when it calls other functions.
12097 This mode is useful for compiling libraries or other compilation units
12098 you might want to incorporate into different programs with different
12099 prevailing FPU modes, and the convenience of being able to use a single
12100 object file outweighs the size and speed overhead for any extra
12101 mode switching that might be needed, compared with what would be needed
12102 with a more specific choice of prevailing FPU mode.
12105 This is the mode used for floating-point calculations with
12106 truncating (i.e.@: round towards zero) rounding mode. That includes
12107 conversion from floating point to integer.
12109 @item round-nearest
12110 This is the mode used for floating-point calculations with
12111 round-to-nearest-or-even rounding mode.
12114 This is the mode used to perform integer calculations in the FPU, e.g.@:
12115 integer multiply, or integer multiply-and-accumulate.
12118 The default is @option{-mfp-mode=caller}
12120 @item -mnosplit-lohi
12121 @itemx -mno-postinc
12122 @itemx -mno-postmodify
12123 @opindex mnosplit-lohi
12124 @opindex mno-postinc
12125 @opindex mno-postmodify
12126 Code generation tweaks that disable, respectively, splitting of 32-bit
12127 loads, generation of post-increment addresses, and generation of
12128 post-modify addresses. The defaults are @option{msplit-lohi},
12129 @option{-mpost-inc}, and @option{-mpost-modify}.
12131 @item -mnovect-double
12132 @opindex mno-vect-double
12133 Change the preferred SIMD mode to SImode. The default is
12134 @option{-mvect-double}, which uses DImode as preferred SIMD mode.
12136 @item -max-vect-align=@var{num}
12137 @opindex max-vect-align
12138 The maximum alignment for SIMD vector mode types.
12139 @var{num} may be 4 or 8. The default is 8.
12140 Note that this is an ABI change, even though many library function
12141 interfaces are unaffected if they don't use SIMD vector modes
12142 in places that affect size and/or alignment of relevant types.
12144 @item -msplit-vecmove-early
12145 @opindex msplit-vecmove-early
12146 Split vector moves into single word moves before reload. In theory this
12147 can give better register allocation, but so far the reverse seems to be
12148 generally the case.
12150 @item -m1reg-@var{reg}
12152 Specify a register to hold the constant @minus{}1, which makes loading small negative
12153 constants and certain bitmasks faster.
12154 Allowable values for @var{reg} are @samp{r43} and @samp{r63},
12155 which specify use of that register as a fixed register,
12156 and @samp{none}, which means that no register is used for this
12157 purpose. The default is @option{-m1reg-none}.
12162 @subsection ARC Options
12163 @cindex ARC options
12165 The following options control the architecture variant for which code
12168 @c architecture variants
12171 @item -mbarrel-shifter
12172 @opindex mbarrel-shifter
12173 Generate instructions supported by barrel shifter. This is the default
12174 unless @samp{-mcpu=ARC601} is in effect.
12176 @item -mcpu=@var{cpu}
12178 Set architecture type, register usage, and instruction scheduling
12179 parameters for @var{cpu}. There are also shortcut alias options
12180 available for backward compatibility and convenience. Supported
12181 values for @var{cpu} are
12187 Compile for ARC600. Aliases: @option{-mA6}, @option{-mARC600}.
12191 Compile for ARC601. Alias: @option{-mARC601}.
12196 Compile for ARC700. Aliases: @option{-mA7}, @option{-mARC700}.
12197 This is the default when configured with @samp{--with-cpu=arc700}@.
12202 @itemx -mdpfp-compact
12203 @opindex mdpfp-compact
12204 FPX: Generate Double Precision FPX instructions, tuned for the compact
12208 @opindex mdpfp-fast
12209 FPX: Generate Double Precision FPX instructions, tuned for the fast
12212 @item -mno-dpfp-lrsr
12213 @opindex mno-dpfp-lrsr
12214 Disable LR and SR instructions from using FPX extension aux registers.
12218 Generate Extended arithmetic instructions. Currently only
12219 @code{divaw}, @code{adds}, @code{subs}, and @code{sat16} are
12220 supported. This is always enabled for @samp{-mcpu=ARC700}.
12224 Do not generate mpy instructions for ARC700.
12228 Generate 32x16 bit multiply and mac instructions.
12232 Generate mul64 and mulu64 instructions. Only valid for @samp{-mcpu=ARC600}.
12236 Generate norm instruction. This is the default if @samp{-mcpu=ARC700}
12241 @itemx -mspfp-compact
12242 @opindex mspfp-compact
12243 FPX: Generate Single Precision FPX instructions, tuned for the compact
12247 @opindex mspfp-fast
12248 FPX: Generate Single Precision FPX instructions, tuned for the fast
12253 Enable generation of ARC SIMD instructions via target-specific
12254 builtins. Only valid for @samp{-mcpu=ARC700}.
12257 @opindex msoft-float
12258 This option ignored; it is provided for compatibility purposes only.
12259 Software floating point code is emitted by default, and this default
12260 can overridden by FPX options; @samp{mspfp}, @samp{mspfp-compact}, or
12261 @samp{mspfp-fast} for single precision, and @samp{mdpfp},
12262 @samp{mdpfp-compact}, or @samp{mdpfp-fast} for double precision.
12266 Generate swap instructions.
12270 The following options are passed through to the assembler, and also
12271 define preprocessor macro symbols.
12273 @c Flags used by the assembler, but for which we define preprocessor
12274 @c macro symbols as well.
12277 @opindex mdsp-packa
12278 Passed down to the assembler to enable the DSP Pack A extensions.
12279 Also sets the preprocessor symbol @code{__Xdsp_packa}.
12283 Passed down to the assembler to enable the dual viterbi butterfly
12284 extension. Also sets the preprocessor symbol @code{__Xdvbf}.
12286 @c ARC700 4.10 extension instruction
12289 Passed down to the assembler to enable the Locked Load/Store
12290 Conditional extension. Also sets the preprocessor symbol
12295 Passed down to the assembler. Also sets the preprocessor symbol
12296 @code{__Xxmac_d16}.
12300 Passed down to the assembler. Also sets the preprocessor symbol
12303 @c ARC700 4.10 extension instruction
12306 Passed down to the assembler to enable the 64-bit Time-Stamp Counter
12307 extension instruction. Also sets the preprocessor symbol
12310 @c ARC700 4.10 extension instruction
12313 Passed down to the assembler to enable the swap byte ordering
12314 extension instruction. Also sets the preprocessor symbol
12318 @opindex mtelephony
12319 Passed down to the assembler to enable dual and single operand
12320 instructions for telephony. Also sets the preprocessor symbol
12321 @code{__Xtelephony}.
12325 Passed down to the assembler to enable the XY Memory extension. Also
12326 sets the preprocessor symbol @code{__Xxy}.
12330 The following options control how the assembly code is annotated:
12332 @c Assembly annotation options
12336 Annotate assembler instructions with estimated addresses.
12338 @item -mannotate-align
12339 @opindex mannotate-align
12340 Explain what alignment considerations lead to the decision to make an
12341 instruction short or long.
12345 The following options are passed through to the linker:
12347 @c options passed through to the linker
12351 Passed through to the linker, to specify use of the @code{arclinux} emulation.
12352 This option is enabled by default in tool chains built for
12353 @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets
12354 when profiling is not requested.
12356 @item -marclinux_prof
12357 @opindex marclinux_prof
12358 Passed through to the linker, to specify use of the
12359 @code{arclinux_prof} emulation. This option is enabled by default in
12360 tool chains built for @w{@code{arc-linux-uclibc}} and
12361 @w{@code{arceb-linux-uclibc}} targets when profiling is requested.
12365 The following options control the semantics of generated code:
12367 @c semantically relevant code generation options
12369 @item -mepilogue-cfi
12370 @opindex mepilogue-cfi
12371 Enable generation of call frame information for epilogues.
12373 @item -mno-epilogue-cfi
12374 @opindex mno-epilogue-cfi
12375 Disable generation of call frame information for epilogues.
12378 @opindex mlong-calls
12379 Generate call insns as register indirect calls, thus providing access
12380 to the full 32-bit address range.
12382 @item -mmedium-calls
12383 @opindex mmedium-calls
12384 Don't use less than 25 bit addressing range for calls, which is the
12385 offset available for an unconditional branch-and-link
12386 instruction. Conditional execution of function calls is suppressed, to
12387 allow use of the 25-bit range, rather than the 21-bit range with
12388 conditional branch-and-link. This is the default for tool chains built
12389 for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets.
12393 Do not generate sdata references. This is the default for tool chains
12394 built for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}}
12398 @opindex mucb-mcount
12399 Instrument with mcount calls as used in UCB code. I.e. do the
12400 counting in the callee, not the caller. By default ARC instrumentation
12401 counts in the caller.
12403 @item -mvolatile-cache
12404 @opindex mvolatile-cache
12405 Use ordinarily cached memory accesses for volatile references. This is the
12408 @item -mno-volatile-cache
12409 @opindex mno-volatile-cache
12410 Enable cache bypass for volatile references.
12414 The following options fine tune code generation:
12415 @c code generation tuning options
12418 @opindex malign-call
12419 Do alignment optimizations for call instructions.
12421 @item -mauto-modify-reg
12422 @opindex mauto-modify-reg
12423 Enable the use of pre/post modify with register displacement.
12425 @item -mbbit-peephole
12426 @opindex mbbit-peephole
12427 Enable bbit peephole2.
12431 This option disables a target-specific pass in @file{arc_reorg} to
12432 generate @code{BRcc} instructions. It has no effect on @code{BRcc}
12433 generation driven by the combiner pass.
12435 @item -mcase-vector-pcrel
12436 @opindex mcase-vector-pcrel
12437 Use pc-relative switch case tables - this enables case table shortening.
12438 This is the default for @option{-Os}.
12440 @item -mcompact-casesi
12441 @opindex mcompact-casesi
12442 Enable compact casesi pattern.
12443 This is the default for @option{-Os}.
12445 @item -mno-cond-exec
12446 @opindex mno-cond-exec
12447 Disable ARCompact specific pass to generate conditional execution instructions.
12448 Due to delay slot scheduling and interactions between operand numbers,
12449 literal sizes, instruction lengths, and the support for conditional execution,
12450 the target-independent pass to generate conditional execution is often lacking,
12451 so the ARC port has kept a special pass around that tries to find more
12452 conditional execution generating opportunities after register allocation,
12453 branch shortening, and delay slot scheduling have been done. This pass
12454 generally, but not always, improves performance and code size, at the cost of
12455 extra compilation time, which is why there is an option to switch it off.
12456 If you have a problem with call instructions exceeding their allowable
12457 offset range because they are conditionalized, you should consider using
12458 @option{-mmedium-calls} instead.
12460 @item -mearly-cbranchsi
12461 @opindex mearly-cbranchsi
12462 Enable pre-reload use of the cbranchsi pattern.
12464 @item -mexpand-adddi
12465 @opindex mexpand-adddi
12466 Expand @code{adddi3} and @code{subdi3} at rtl generation time into
12467 @code{add.f}, @code{adc} etc.
12469 @item -mindexed-loads
12470 @opindex mindexed-loads
12471 Enable the use of indexed loads. This can be problematic because some
12472 optimizers will then assume the that indexed stores exist, which is not
12477 Enable Local Register Allocation. This is still experimental for ARC,
12478 so by default the compiler uses standard reload
12479 (i.e. @samp{-mno-lra}).
12481 @item -mlra-priority-none
12482 @opindex mlra-priority-none
12483 Don't indicate any priority for target registers.
12485 @item -mlra-priority-compact
12486 @opindex mlra-priority-compact
12487 Indicate target register priority for r0..r3 / r12..r15.
12489 @item -mlra-priority-noncompact
12490 @opindex mlra-priority-noncompact
12491 Reduce target regsiter priority for r0..r3 / r12..r15.
12493 @item -mno-millicode
12494 @opindex mno-millicode
12495 When optimizing for size (using @option{-Os}), prologues and epilogues
12496 that have to save or restore a large number of registers are often
12497 shortened by using call to a special function in libgcc; this is
12498 referred to as a @emph{millicode} call. As these calls can pose
12499 performance issues, and/or cause linking issues when linking in a
12500 nonstandard way, this option is provided to turn off millicode call
12504 @opindex mmixed-code
12505 Tweak register allocation to help 16-bit instruction generation.
12506 This generally has the effect of decreasing the average instruction size
12507 while increasing the instruction count.
12511 Enable 'q' instruction alternatives.
12512 This is the default for @option{-Os}.
12516 Enable Rcq constraint handling - most short code generation depends on this.
12517 This is the default.
12521 Enable Rcw constraint handling - ccfsm condexec mostly depends on this.
12522 This is the default.
12524 @item -msize-level=@var{level}
12525 @opindex msize-level
12526 Fine-tune size optimization with regards to instruction lengths and alignment.
12527 The recognized values for @var{level} are:
12530 No size optimization. This level is deprecated and treated like @samp{1}.
12533 Short instructions are used opportunistically.
12536 In addition, alignment of loops and of code after barriers are dropped.
12539 In addition, optional data alignment is dropped, and the option @option{Os} is enabled.
12543 This defaults to @samp{3} when @option{-Os} is in effect. Otherwise,
12544 the behavior when this is not set is equivalent to level @samp{1}.
12546 @item -mtune=@var{cpu}
12548 Set instruction scheduling parameters for @var{cpu}, overriding any implied
12549 by @option{-mcpu=}.
12551 Supported values for @var{cpu} are
12555 Tune for ARC600 cpu.
12558 Tune for ARC601 cpu.
12561 Tune for ARC700 cpu with standard multiplier block.
12564 Tune for ARC700 cpu with XMAC block.
12567 Tune for ARC725D cpu.
12570 Tune for ARC750D cpu.
12574 @item -mmultcost=@var{num}
12576 Cost to assume for a multiply instruction, with @samp{4} being equal to a
12577 normal instruction.
12579 @item -munalign-prob-threshold=@var{probability}
12580 @opindex munalign-prob-threshold
12581 Set probability threshold for unaligning branches.
12582 When tuning for @samp{ARC700} and optimizing for speed, branches without
12583 filled delay slot are preferably emitted unaligned and long, unless
12584 profiling indicates that the probability for the branch to be taken
12585 is below @var{probability}. @xref{Cross-profiling}.
12586 The default is (REG_BR_PROB_BASE/2), i.e.@: 5000.
12590 The following options are maintained for backward compatibility, but
12591 are now deprecated and will be removed in a future release:
12593 @c Deprecated options
12601 @opindex mbig-endian
12604 Compile code for big endian targets. Use of these options is now
12605 deprecated. Users wanting big-endian code, should use the
12606 @w{@code{arceb-elf32}} and @w{@code{arceb-linux-uclibc}} targets when
12607 building the tool chain, for which big-endian is the default.
12609 @item -mlittle-endian
12610 @opindex mlittle-endian
12613 Compile code for little endian targets. Use of these options is now
12614 deprecated. Users wanting little-endian code should use the
12615 @w{@code{arc-elf32}} and @w{@code{arc-linux-uclibc}} targets when
12616 building the tool chain, for which little-endian is the default.
12618 @item -mbarrel_shifter
12619 @opindex mbarrel_shifter
12620 Replaced by @samp{-mbarrel-shifter}
12622 @item -mdpfp_compact
12623 @opindex mdpfp_compact
12624 Replaced by @samp{-mdpfp-compact}
12627 @opindex mdpfp_fast
12628 Replaced by @samp{-mdpfp-fast}
12631 @opindex mdsp_packa
12632 Replaced by @samp{-mdsp-packa}
12636 Replaced by @samp{-mea}
12640 Replaced by @samp{-mmac-24}
12644 Replaced by @samp{-mmac-d16}
12646 @item -mspfp_compact
12647 @opindex mspfp_compact
12648 Replaced by @samp{-mspfp-compact}
12651 @opindex mspfp_fast
12652 Replaced by @samp{-mspfp-fast}
12654 @item -mtune=@var{cpu}
12656 Values @samp{arc600}, @samp{arc601}, @samp{arc700} and
12657 @samp{arc700-xmac} for @var{cpu} are replaced by @samp{ARC600},
12658 @samp{ARC601}, @samp{ARC700} and @samp{ARC700-xmac} respectively
12660 @item -multcost=@var{num}
12662 Replaced by @samp{-mmultcost}.
12667 @subsection ARM Options
12668 @cindex ARM options
12670 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
12674 @item -mabi=@var{name}
12676 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
12677 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
12680 @opindex mapcs-frame
12681 Generate a stack frame that is compliant with the ARM Procedure Call
12682 Standard for all functions, even if this is not strictly necessary for
12683 correct execution of the code. Specifying @option{-fomit-frame-pointer}
12684 with this option causes the stack frames not to be generated for
12685 leaf functions. The default is @option{-mno-apcs-frame}.
12689 This is a synonym for @option{-mapcs-frame}.
12692 @c not currently implemented
12693 @item -mapcs-stack-check
12694 @opindex mapcs-stack-check
12695 Generate code to check the amount of stack space available upon entry to
12696 every function (that actually uses some stack space). If there is
12697 insufficient space available then either the function
12698 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} is
12699 called, depending upon the amount of stack space required. The runtime
12700 system is required to provide these functions. The default is
12701 @option{-mno-apcs-stack-check}, since this produces smaller code.
12703 @c not currently implemented
12705 @opindex mapcs-float
12706 Pass floating-point arguments using the floating-point registers. This is
12707 one of the variants of the APCS@. This option is recommended if the
12708 target hardware has a floating-point unit or if a lot of floating-point
12709 arithmetic is going to be performed by the code. The default is
12710 @option{-mno-apcs-float}, since the size of integer-only code is
12711 slightly increased if @option{-mapcs-float} is used.
12713 @c not currently implemented
12714 @item -mapcs-reentrant
12715 @opindex mapcs-reentrant
12716 Generate reentrant, position-independent code. The default is
12717 @option{-mno-apcs-reentrant}.
12720 @item -mthumb-interwork
12721 @opindex mthumb-interwork
12722 Generate code that supports calling between the ARM and Thumb
12723 instruction sets. Without this option, on pre-v5 architectures, the
12724 two instruction sets cannot be reliably used inside one program. The
12725 default is @option{-mno-thumb-interwork}, since slightly larger code
12726 is generated when @option{-mthumb-interwork} is specified. In AAPCS
12727 configurations this option is meaningless.
12729 @item -mno-sched-prolog
12730 @opindex mno-sched-prolog
12731 Prevent the reordering of instructions in the function prologue, or the
12732 merging of those instruction with the instructions in the function's
12733 body. This means that all functions start with a recognizable set
12734 of instructions (or in fact one of a choice from a small set of
12735 different function prologues), and this information can be used to
12736 locate the start of functions inside an executable piece of code. The
12737 default is @option{-msched-prolog}.
12739 @item -mfloat-abi=@var{name}
12740 @opindex mfloat-abi
12741 Specifies which floating-point ABI to use. Permissible values
12742 are: @samp{soft}, @samp{softfp} and @samp{hard}.
12744 Specifying @samp{soft} causes GCC to generate output containing
12745 library calls for floating-point operations.
12746 @samp{softfp} allows the generation of code using hardware floating-point
12747 instructions, but still uses the soft-float calling conventions.
12748 @samp{hard} allows generation of floating-point instructions
12749 and uses FPU-specific calling conventions.
12751 The default depends on the specific target configuration. Note that
12752 the hard-float and soft-float ABIs are not link-compatible; you must
12753 compile your entire program with the same ABI, and link with a
12754 compatible set of libraries.
12756 @item -mlittle-endian
12757 @opindex mlittle-endian
12758 Generate code for a processor running in little-endian mode. This is
12759 the default for all standard configurations.
12762 @opindex mbig-endian
12763 Generate code for a processor running in big-endian mode; the default is
12764 to compile code for a little-endian processor.
12766 @item -march=@var{name}
12768 This specifies the name of the target ARM architecture. GCC uses this
12769 name to determine what kind of instructions it can emit when generating
12770 assembly code. This option can be used in conjunction with or instead
12771 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
12772 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
12773 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
12774 @samp{armv6}, @samp{armv6j},
12775 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
12776 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m}, @samp{armv7e-m},
12777 @samp{armv7ve}, @samp{armv8-a}, @samp{armv8-a+crc},
12778 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
12780 @option{-march=armv7ve} is the armv7-a architecture with virtualization
12783 @option{-march=armv8-a+crc} enables code generation for the ARMv8-A
12784 architecture together with the optional CRC32 extensions.
12786 @option{-march=native} causes the compiler to auto-detect the architecture
12787 of the build computer. At present, this feature is only supported on
12788 Linux, and not all architectures are recognized. If the auto-detect is
12789 unsuccessful the option has no effect.
12791 @item -mtune=@var{name}
12793 This option specifies the name of the target ARM processor for
12794 which GCC should tune the performance of the code.
12795 For some ARM implementations better performance can be obtained by using
12797 Permissible names are: @samp{arm2}, @samp{arm250},
12798 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
12799 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
12800 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
12801 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
12803 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
12804 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
12805 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
12806 @samp{strongarm1110},
12807 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
12808 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
12809 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
12810 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
12811 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
12812 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
12813 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
12814 @samp{cortex-a5}, @samp{cortex-a7}, @samp{cortex-a8}, @samp{cortex-a9},
12815 @samp{cortex-a12}, @samp{cortex-a15}, @samp{cortex-a53}, @samp{cortex-a57},
12817 @samp{cortex-r4f}, @samp{cortex-r5}, @samp{cortex-r7}, @samp{cortex-m4},
12821 @samp{cortex-m0plus},
12822 @samp{marvell-pj4},
12823 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312},
12824 @samp{fa526}, @samp{fa626},
12825 @samp{fa606te}, @samp{fa626te}, @samp{fmp626}, @samp{fa726te}.
12827 Additionally, this option can specify that GCC should tune the performance
12828 of the code for a big.LITTLE system. Permissible names are:
12829 @samp{cortex-a15.cortex-a7}, @samp{cortex-a57.cortex-a53}.
12831 @option{-mtune=generic-@var{arch}} specifies that GCC should tune the
12832 performance for a blend of processors within architecture @var{arch}.
12833 The aim is to generate code that run well on the current most popular
12834 processors, balancing between optimizations that benefit some CPUs in the
12835 range, and avoiding performance pitfalls of other CPUs. The effects of
12836 this option may change in future GCC versions as CPU models come and go.
12838 @option{-mtune=native} causes the compiler to auto-detect the CPU
12839 of the build computer. At present, this feature is only supported on
12840 Linux, and not all architectures are recognized. If the auto-detect is
12841 unsuccessful the option has no effect.
12843 @item -mcpu=@var{name}
12845 This specifies the name of the target ARM processor. GCC uses this name
12846 to derive the name of the target ARM architecture (as if specified
12847 by @option{-march}) and the ARM processor type for which to tune for
12848 performance (as if specified by @option{-mtune}). Where this option
12849 is used in conjunction with @option{-march} or @option{-mtune},
12850 those options take precedence over the appropriate part of this option.
12852 Permissible names for this option are the same as those for
12855 @option{-mcpu=generic-@var{arch}} is also permissible, and is
12856 equivalent to @option{-march=@var{arch} -mtune=generic-@var{arch}}.
12857 See @option{-mtune} for more information.
12859 @option{-mcpu=native} causes the compiler to auto-detect the CPU
12860 of the build computer. At present, this feature is only supported on
12861 Linux, and not all architectures are recognized. If the auto-detect is
12862 unsuccessful the option has no effect.
12864 @item -mfpu=@var{name}
12866 This specifies what floating-point hardware (or hardware emulation) is
12867 available on the target. Permissible names are: @samp{vfp}, @samp{vfpv3},
12868 @samp{vfpv3-fp16}, @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd},
12869 @samp{vfpv3xd-fp16}, @samp{neon}, @samp{neon-fp16}, @samp{vfpv4},
12870 @samp{vfpv4-d16}, @samp{fpv4-sp-d16}, @samp{neon-vfpv4},
12871 @samp{fp-armv8}, @samp{neon-fp-armv8}, and @samp{crypto-neon-fp-armv8}.
12873 If @option{-msoft-float} is specified this specifies the format of
12874 floating-point values.
12876 If the selected floating-point hardware includes the NEON extension
12877 (e.g. @option{-mfpu}=@samp{neon}), note that floating-point
12878 operations are not generated by GCC's auto-vectorization pass unless
12879 @option{-funsafe-math-optimizations} is also specified. This is
12880 because NEON hardware does not fully implement the IEEE 754 standard for
12881 floating-point arithmetic (in particular denormal values are treated as
12882 zero), so the use of NEON instructions may lead to a loss of precision.
12884 @item -mfp16-format=@var{name}
12885 @opindex mfp16-format
12886 Specify the format of the @code{__fp16} half-precision floating-point type.
12887 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
12888 the default is @samp{none}, in which case the @code{__fp16} type is not
12889 defined. @xref{Half-Precision}, for more information.
12891 @item -mstructure-size-boundary=@var{n}
12892 @opindex mstructure-size-boundary
12893 The sizes of all structures and unions are rounded up to a multiple
12894 of the number of bits set by this option. Permissible values are 8, 32
12895 and 64. The default value varies for different toolchains. For the COFF
12896 targeted toolchain the default value is 8. A value of 64 is only allowed
12897 if the underlying ABI supports it.
12899 Specifying a larger number can produce faster, more efficient code, but
12900 can also increase the size of the program. Different values are potentially
12901 incompatible. Code compiled with one value cannot necessarily expect to
12902 work with code or libraries compiled with another value, if they exchange
12903 information using structures or unions.
12905 @item -mabort-on-noreturn
12906 @opindex mabort-on-noreturn
12907 Generate a call to the function @code{abort} at the end of a
12908 @code{noreturn} function. It is executed if the function tries to
12912 @itemx -mno-long-calls
12913 @opindex mlong-calls
12914 @opindex mno-long-calls
12915 Tells the compiler to perform function calls by first loading the
12916 address of the function into a register and then performing a subroutine
12917 call on this register. This switch is needed if the target function
12918 lies outside of the 64-megabyte addressing range of the offset-based
12919 version of subroutine call instruction.
12921 Even if this switch is enabled, not all function calls are turned
12922 into long calls. The heuristic is that static functions, functions
12923 that have the @samp{short-call} attribute, functions that are inside
12924 the scope of a @samp{#pragma no_long_calls} directive, and functions whose
12925 definitions have already been compiled within the current compilation
12926 unit are not turned into long calls. The exceptions to this rule are
12927 that weak function definitions, functions with the @samp{long-call}
12928 attribute or the @samp{section} attribute, and functions that are within
12929 the scope of a @samp{#pragma long_calls} directive are always
12930 turned into long calls.
12932 This feature is not enabled by default. Specifying
12933 @option{-mno-long-calls} restores the default behavior, as does
12934 placing the function calls within the scope of a @samp{#pragma
12935 long_calls_off} directive. Note these switches have no effect on how
12936 the compiler generates code to handle function calls via function
12939 @item -msingle-pic-base
12940 @opindex msingle-pic-base
12941 Treat the register used for PIC addressing as read-only, rather than
12942 loading it in the prologue for each function. The runtime system is
12943 responsible for initializing this register with an appropriate value
12944 before execution begins.
12946 @item -mpic-register=@var{reg}
12947 @opindex mpic-register
12948 Specify the register to be used for PIC addressing.
12949 For standard PIC base case, the default will be any suitable register
12950 determined by compiler. For single PIC base case, the default is
12951 @samp{R9} if target is EABI based or stack-checking is enabled,
12952 otherwise the default is @samp{R10}.
12954 @item -mpic-data-is-text-relative
12955 @opindex mpic-data-is-text-relative
12956 Assume that each data segments are relative to text segment at load time.
12957 Therefore, it permits addressing data using PC-relative operations.
12958 This option is on by default for targets other than VxWorks RTP.
12960 @item -mpoke-function-name
12961 @opindex mpoke-function-name
12962 Write the name of each function into the text section, directly
12963 preceding the function prologue. The generated code is similar to this:
12967 .ascii "arm_poke_function_name", 0
12970 .word 0xff000000 + (t1 - t0)
12971 arm_poke_function_name
12973 stmfd sp!, @{fp, ip, lr, pc@}
12977 When performing a stack backtrace, code can inspect the value of
12978 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
12979 location @code{pc - 12} and the top 8 bits are set, then we know that
12980 there is a function name embedded immediately preceding this location
12981 and has length @code{((pc[-3]) & 0xff000000)}.
12988 Select between generating code that executes in ARM and Thumb
12989 states. The default for most configurations is to generate code
12990 that executes in ARM state, but the default can be changed by
12991 configuring GCC with the @option{--with-mode=}@var{state}
12995 @opindex mtpcs-frame
12996 Generate a stack frame that is compliant with the Thumb Procedure Call
12997 Standard for all non-leaf functions. (A leaf function is one that does
12998 not call any other functions.) The default is @option{-mno-tpcs-frame}.
13000 @item -mtpcs-leaf-frame
13001 @opindex mtpcs-leaf-frame
13002 Generate a stack frame that is compliant with the Thumb Procedure Call
13003 Standard for all leaf functions. (A leaf function is one that does
13004 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
13006 @item -mcallee-super-interworking
13007 @opindex mcallee-super-interworking
13008 Gives all externally visible functions in the file being compiled an ARM
13009 instruction set header which switches to Thumb mode before executing the
13010 rest of the function. This allows these functions to be called from
13011 non-interworking code. This option is not valid in AAPCS configurations
13012 because interworking is enabled by default.
13014 @item -mcaller-super-interworking
13015 @opindex mcaller-super-interworking
13016 Allows calls via function pointers (including virtual functions) to
13017 execute correctly regardless of whether the target code has been
13018 compiled for interworking or not. There is a small overhead in the cost
13019 of executing a function pointer if this option is enabled. This option
13020 is not valid in AAPCS configurations because interworking is enabled
13023 @item -mtp=@var{name}
13025 Specify the access model for the thread local storage pointer. The valid
13026 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
13027 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
13028 (supported in the arm6k architecture), and @option{auto}, which uses the
13029 best available method for the selected processor. The default setting is
13032 @item -mtls-dialect=@var{dialect}
13033 @opindex mtls-dialect
13034 Specify the dialect to use for accessing thread local storage. Two
13035 @var{dialect}s are supported---@samp{gnu} and @samp{gnu2}. The
13036 @samp{gnu} dialect selects the original GNU scheme for supporting
13037 local and global dynamic TLS models. The @samp{gnu2} dialect
13038 selects the GNU descriptor scheme, which provides better performance
13039 for shared libraries. The GNU descriptor scheme is compatible with
13040 the original scheme, but does require new assembler, linker and
13041 library support. Initial and local exec TLS models are unaffected by
13042 this option and always use the original scheme.
13044 @item -mword-relocations
13045 @opindex mword-relocations
13046 Only generate absolute relocations on word-sized values (i.e. R_ARM_ABS32).
13047 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
13048 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
13051 @item -mfix-cortex-m3-ldrd
13052 @opindex mfix-cortex-m3-ldrd
13053 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
13054 with overlapping destination and base registers are used. This option avoids
13055 generating these instructions. This option is enabled by default when
13056 @option{-mcpu=cortex-m3} is specified.
13058 @item -munaligned-access
13059 @itemx -mno-unaligned-access
13060 @opindex munaligned-access
13061 @opindex mno-unaligned-access
13062 Enables (or disables) reading and writing of 16- and 32- bit values
13063 from addresses that are not 16- or 32- bit aligned. By default
13064 unaligned access is disabled for all pre-ARMv6 and all ARMv6-M
13065 architectures, and enabled for all other architectures. If unaligned
13066 access is not enabled then words in packed data structures will be
13067 accessed a byte at a time.
13069 The ARM attribute @code{Tag_CPU_unaligned_access} will be set in the
13070 generated object file to either true or false, depending upon the
13071 setting of this option. If unaligned access is enabled then the
13072 preprocessor symbol @code{__ARM_FEATURE_UNALIGNED} will also be
13075 @item -mneon-for-64bits
13076 @opindex mneon-for-64bits
13077 Enables using Neon to handle scalar 64-bits operations. This is
13078 disabled by default since the cost of moving data from core registers
13081 @item -mslow-flash-data
13082 @opindex mslow-flash-data
13083 Assume loading data from flash is slower than fetching instruction.
13084 Therefore literal load is minimized for better performance.
13085 This option is only supported when compiling for ARMv7 M-profile and
13088 @item -mrestrict-it
13089 @opindex mrestrict-it
13090 Restricts generation of IT blocks to conform to the rules of ARMv8.
13091 IT blocks can only contain a single 16-bit instruction from a select
13092 set of instructions. This option is on by default for ARMv8 Thumb mode.
13096 @subsection AVR Options
13097 @cindex AVR Options
13099 These options are defined for AVR implementations:
13102 @item -mmcu=@var{mcu}
13104 Specify Atmel AVR instruction set architectures (ISA) or MCU type.
13106 The default for this option is@tie{}@code{avr2}.
13108 GCC supports the following AVR devices and ISAs:
13110 @include avr-mmcu.texi
13112 @item -maccumulate-args
13113 @opindex maccumulate-args
13114 Accumulate outgoing function arguments and acquire/release the needed
13115 stack space for outgoing function arguments once in function
13116 prologue/epilogue. Without this option, outgoing arguments are pushed
13117 before calling a function and popped afterwards.
13119 Popping the arguments after the function call can be expensive on
13120 AVR so that accumulating the stack space might lead to smaller
13121 executables because arguments need not to be removed from the
13122 stack after such a function call.
13124 This option can lead to reduced code size for functions that perform
13125 several calls to functions that get their arguments on the stack like
13126 calls to printf-like functions.
13128 @item -mbranch-cost=@var{cost}
13129 @opindex mbranch-cost
13130 Set the branch costs for conditional branch instructions to
13131 @var{cost}. Reasonable values for @var{cost} are small, non-negative
13132 integers. The default branch cost is 0.
13134 @item -mcall-prologues
13135 @opindex mcall-prologues
13136 Functions prologues/epilogues are expanded as calls to appropriate
13137 subroutines. Code size is smaller.
13141 Assume @code{int} to be 8-bit integer. This affects the sizes of all types: a
13142 @code{char} is 1 byte, an @code{int} is 1 byte, a @code{long} is 2 bytes,
13143 and @code{long long} is 4 bytes. Please note that this option does not
13144 conform to the C standards, but it results in smaller code
13147 @item -mno-interrupts
13148 @opindex mno-interrupts
13149 Generated code is not compatible with hardware interrupts.
13150 Code size is smaller.
13154 Try to replace @code{CALL} resp.@: @code{JMP} instruction by the shorter
13155 @code{RCALL} resp.@: @code{RJMP} instruction if applicable.
13156 Setting @code{-mrelax} just adds the @code{--relax} option to the
13157 linker command line when the linker is called.
13159 Jump relaxing is performed by the linker because jump offsets are not
13160 known before code is located. Therefore, the assembler code generated by the
13161 compiler is the same, but the instructions in the executable may
13162 differ from instructions in the assembler code.
13164 Relaxing must be turned on if linker stubs are needed, see the
13165 section on @code{EIND} and linker stubs below.
13169 Treat the stack pointer register as an 8-bit register,
13170 i.e.@: assume the high byte of the stack pointer is zero.
13171 In general, you don't need to set this option by hand.
13173 This option is used internally by the compiler to select and
13174 build multilibs for architectures @code{avr2} and @code{avr25}.
13175 These architectures mix devices with and without @code{SPH}.
13176 For any setting other than @code{-mmcu=avr2} or @code{-mmcu=avr25}
13177 the compiler driver will add or remove this option from the compiler
13178 proper's command line, because the compiler then knows if the device
13179 or architecture has an 8-bit stack pointer and thus no @code{SPH}
13184 Use address register @code{X} in a way proposed by the hardware. This means
13185 that @code{X} is only used in indirect, post-increment or
13186 pre-decrement addressing.
13188 Without this option, the @code{X} register may be used in the same way
13189 as @code{Y} or @code{Z} which then is emulated by additional
13191 For example, loading a value with @code{X+const} addressing with a
13192 small non-negative @code{const < 64} to a register @var{Rn} is
13196 adiw r26, const ; X += const
13197 ld @var{Rn}, X ; @var{Rn} = *X
13198 sbiw r26, const ; X -= const
13202 @opindex mtiny-stack
13203 Only change the lower 8@tie{}bits of the stack pointer.
13205 @item -Waddr-space-convert
13206 @opindex Waddr-space-convert
13207 Warn about conversions between address spaces in the case where the
13208 resulting address space is not contained in the incoming address space.
13211 @subsubsection @code{EIND} and Devices with more than 128 Ki Bytes of Flash
13212 @cindex @code{EIND}
13213 Pointers in the implementation are 16@tie{}bits wide.
13214 The address of a function or label is represented as word address so
13215 that indirect jumps and calls can target any code address in the
13216 range of 64@tie{}Ki words.
13218 In order to facilitate indirect jump on devices with more than 128@tie{}Ki
13219 bytes of program memory space, there is a special function register called
13220 @code{EIND} that serves as most significant part of the target address
13221 when @code{EICALL} or @code{EIJMP} instructions are used.
13223 Indirect jumps and calls on these devices are handled as follows by
13224 the compiler and are subject to some limitations:
13229 The compiler never sets @code{EIND}.
13232 The compiler uses @code{EIND} implicitely in @code{EICALL}/@code{EIJMP}
13233 instructions or might read @code{EIND} directly in order to emulate an
13234 indirect call/jump by means of a @code{RET} instruction.
13237 The compiler assumes that @code{EIND} never changes during the startup
13238 code or during the application. In particular, @code{EIND} is not
13239 saved/restored in function or interrupt service routine
13243 For indirect calls to functions and computed goto, the linker
13244 generates @emph{stubs}. Stubs are jump pads sometimes also called
13245 @emph{trampolines}. Thus, the indirect call/jump jumps to such a stub.
13246 The stub contains a direct jump to the desired address.
13249 Linker relaxation must be turned on so that the linker will generate
13250 the stubs correctly an all situaltion. See the compiler option
13251 @code{-mrelax} and the linler option @code{--relax}.
13252 There are corner cases where the linker is supposed to generate stubs
13253 but aborts without relaxation and without a helpful error message.
13256 The default linker script is arranged for code with @code{EIND = 0}.
13257 If code is supposed to work for a setup with @code{EIND != 0}, a custom
13258 linker script has to be used in order to place the sections whose
13259 name start with @code{.trampolines} into the segment where @code{EIND}
13263 The startup code from libgcc never sets @code{EIND}.
13264 Notice that startup code is a blend of code from libgcc and AVR-LibC.
13265 For the impact of AVR-LibC on @code{EIND}, see the
13266 @w{@uref{http://nongnu.org/avr-libc/user-manual/,AVR-LibC user manual}}.
13269 It is legitimate for user-specific startup code to set up @code{EIND}
13270 early, for example by means of initialization code located in
13271 section @code{.init3}. Such code runs prior to general startup code
13272 that initializes RAM and calls constructors, but after the bit
13273 of startup code from AVR-LibC that sets @code{EIND} to the segment
13274 where the vector table is located.
13276 #include <avr/io.h>
13279 __attribute__((section(".init3"),naked,used,no_instrument_function))
13280 init3_set_eind (void)
13282 __asm volatile ("ldi r24,pm_hh8(__trampolines_start)\n\t"
13283 "out %i0,r24" :: "n" (&EIND) : "r24","memory");
13288 The @code{__trampolines_start} symbol is defined in the linker script.
13291 Stubs are generated automatically by the linker if
13292 the following two conditions are met:
13295 @item The address of a label is taken by means of the @code{gs} modifier
13296 (short for @emph{generate stubs}) like so:
13298 LDI r24, lo8(gs(@var{func}))
13299 LDI r25, hi8(gs(@var{func}))
13301 @item The final location of that label is in a code segment
13302 @emph{outside} the segment where the stubs are located.
13306 The compiler emits such @code{gs} modifiers for code labels in the
13307 following situations:
13309 @item Taking address of a function or code label.
13310 @item Computed goto.
13311 @item If prologue-save function is used, see @option{-mcall-prologues}
13312 command-line option.
13313 @item Switch/case dispatch tables. If you do not want such dispatch
13314 tables you can specify the @option{-fno-jump-tables} command-line option.
13315 @item C and C++ constructors/destructors called during startup/shutdown.
13316 @item If the tools hit a @code{gs()} modifier explained above.
13320 Jumping to non-symbolic addresses like so is @emph{not} supported:
13325 /* Call function at word address 0x2 */
13326 return ((int(*)(void)) 0x2)();
13330 Instead, a stub has to be set up, i.e.@: the function has to be called
13331 through a symbol (@code{func_4} in the example):
13336 extern int func_4 (void);
13338 /* Call function at byte address 0x4 */
13343 and the application be linked with @code{-Wl,--defsym,func_4=0x4}.
13344 Alternatively, @code{func_4} can be defined in the linker script.
13347 @subsubsection Handling of the @code{RAMPD}, @code{RAMPX}, @code{RAMPY} and @code{RAMPZ} Special Function Registers
13348 @cindex @code{RAMPD}
13349 @cindex @code{RAMPX}
13350 @cindex @code{RAMPY}
13351 @cindex @code{RAMPZ}
13352 Some AVR devices support memories larger than the 64@tie{}KiB range
13353 that can be accessed with 16-bit pointers. To access memory locations
13354 outside this 64@tie{}KiB range, the contentent of a @code{RAMP}
13355 register is used as high part of the address:
13356 The @code{X}, @code{Y}, @code{Z} address register is concatenated
13357 with the @code{RAMPX}, @code{RAMPY}, @code{RAMPZ} special function
13358 register, respectively, to get a wide address. Similarly,
13359 @code{RAMPD} is used together with direct addressing.
13363 The startup code initializes the @code{RAMP} special function
13364 registers with zero.
13367 If a @ref{AVR Named Address Spaces,named address space} other than
13368 generic or @code{__flash} is used, then @code{RAMPZ} is set
13369 as needed before the operation.
13372 If the device supports RAM larger than 64@tie{}KiB and the compiler
13373 needs to change @code{RAMPZ} to accomplish an operation, @code{RAMPZ}
13374 is reset to zero after the operation.
13377 If the device comes with a specific @code{RAMP} register, the ISR
13378 prologue/epilogue saves/restores that SFR and initializes it with
13379 zero in case the ISR code might (implicitly) use it.
13382 RAM larger than 64@tie{}KiB is not supported by GCC for AVR targets.
13383 If you use inline assembler to read from locations outside the
13384 16-bit address range and change one of the @code{RAMP} registers,
13385 you must reset it to zero after the access.
13389 @subsubsection AVR Built-in Macros
13391 GCC defines several built-in macros so that the user code can test
13392 for the presence or absence of features. Almost any of the following
13393 built-in macros are deduced from device capabilities and thus
13394 triggered by the @code{-mmcu=} command-line option.
13396 For even more AVR-specific built-in macros see
13397 @ref{AVR Named Address Spaces} and @ref{AVR Built-in Functions}.
13402 Build-in macro that resolves to a decimal number that identifies the
13403 architecture and depends on the @code{-mmcu=@var{mcu}} option.
13404 Possible values are:
13406 @code{2}, @code{25}, @code{3}, @code{31}, @code{35},
13407 @code{4}, @code{5}, @code{51}, @code{6}, @code{102}, @code{104},
13408 @code{105}, @code{106}, @code{107}
13410 for @var{mcu}=@code{avr2}, @code{avr25}, @code{avr3},
13411 @code{avr31}, @code{avr35}, @code{avr4}, @code{avr5}, @code{avr51},
13412 @code{avr6}, @code{avrxmega2}, @code{avrxmega4}, @code{avrxmega5},
13413 @code{avrxmega6}, @code{avrxmega7}, respectively.
13414 If @var{mcu} specifies a device, this built-in macro is set
13415 accordingly. For example, with @code{-mmcu=atmega8} the macro will be
13416 defined to @code{4}.
13418 @item __AVR_@var{Device}__
13419 Setting @code{-mmcu=@var{device}} defines this built-in macro which reflects
13420 the device's name. For example, @code{-mmcu=atmega8} defines the
13421 built-in macro @code{__AVR_ATmega8__}, @code{-mmcu=attiny261a} defines
13422 @code{__AVR_ATtiny261A__}, etc.
13424 The built-in macros' names follow
13425 the scheme @code{__AVR_@var{Device}__} where @var{Device} is
13426 the device name as from the AVR user manual. The difference between
13427 @var{Device} in the built-in macro and @var{device} in
13428 @code{-mmcu=@var{device}} is that the latter is always lowercase.
13430 If @var{device} is not a device but only a core architecture like
13431 @code{avr51}, this macro will not be defined.
13433 @item __AVR_DEVICE_NAME__
13434 Setting @code{-mmcu=@var{device}} defines this built-in macro to
13435 the device's name. For example, with @code{-mmcu=atmega8} the macro
13436 will be defined to @code{atmega8}.
13438 If @var{device} is not a device but only a core architecture like
13439 @code{avr51}, this macro will not be defined.
13441 @item __AVR_XMEGA__
13442 The device / architecture belongs to the XMEGA family of devices.
13444 @item __AVR_HAVE_ELPM__
13445 The device has the the @code{ELPM} instruction.
13447 @item __AVR_HAVE_ELPMX__
13448 The device has the @code{ELPM R@var{n},Z} and @code{ELPM
13449 R@var{n},Z+} instructions.
13451 @item __AVR_HAVE_MOVW__
13452 The device has the @code{MOVW} instruction to perform 16-bit
13453 register-register moves.
13455 @item __AVR_HAVE_LPMX__
13456 The device has the @code{LPM R@var{n},Z} and
13457 @code{LPM R@var{n},Z+} instructions.
13459 @item __AVR_HAVE_MUL__
13460 The device has a hardware multiplier.
13462 @item __AVR_HAVE_JMP_CALL__
13463 The device has the @code{JMP} and @code{CALL} instructions.
13464 This is the case for devices with at least 16@tie{}KiB of program
13467 @item __AVR_HAVE_EIJMP_EICALL__
13468 @itemx __AVR_3_BYTE_PC__
13469 The device has the @code{EIJMP} and @code{EICALL} instructions.
13470 This is the case for devices with more than 128@tie{}KiB of program memory.
13471 This also means that the program counter
13472 (PC) is 3@tie{}bytes wide.
13474 @item __AVR_2_BYTE_PC__
13475 The program counter (PC) is 2@tie{}bytes wide. This is the case for devices
13476 with up to 128@tie{}KiB of program memory.
13478 @item __AVR_HAVE_8BIT_SP__
13479 @itemx __AVR_HAVE_16BIT_SP__
13480 The stack pointer (SP) register is treated as 8-bit respectively
13481 16-bit register by the compiler.
13482 The definition of these macros is affected by @code{-mtiny-stack}.
13484 @item __AVR_HAVE_SPH__
13486 The device has the SPH (high part of stack pointer) special function
13487 register or has an 8-bit stack pointer, respectively.
13488 The definition of these macros is affected by @code{-mmcu=} and
13489 in the cases of @code{-mmcu=avr2} and @code{-mmcu=avr25} also
13492 @item __AVR_HAVE_RAMPD__
13493 @itemx __AVR_HAVE_RAMPX__
13494 @itemx __AVR_HAVE_RAMPY__
13495 @itemx __AVR_HAVE_RAMPZ__
13496 The device has the @code{RAMPD}, @code{RAMPX}, @code{RAMPY},
13497 @code{RAMPZ} special function register, respectively.
13499 @item __NO_INTERRUPTS__
13500 This macro reflects the @code{-mno-interrupts} command line option.
13502 @item __AVR_ERRATA_SKIP__
13503 @itemx __AVR_ERRATA_SKIP_JMP_CALL__
13504 Some AVR devices (AT90S8515, ATmega103) must not skip 32-bit
13505 instructions because of a hardware erratum. Skip instructions are
13506 @code{SBRS}, @code{SBRC}, @code{SBIS}, @code{SBIC} and @code{CPSE}.
13507 The second macro is only defined if @code{__AVR_HAVE_JMP_CALL__} is also
13510 @item __AVR_ISA_RMW__
13511 The device has Read-Modify-Write instructions (XCH, LAC, LAS and LAT).
13513 @item __AVR_SFR_OFFSET__=@var{offset}
13514 Instructions that can address I/O special function registers directly
13515 like @code{IN}, @code{OUT}, @code{SBI}, etc.@: may use a different
13516 address as if addressed by an instruction to access RAM like @code{LD}
13517 or @code{STS}. This offset depends on the device architecture and has
13518 to be subtracted from the RAM address in order to get the
13519 respective I/O@tie{}address.
13521 @item __WITH_AVRLIBC__
13522 The compiler is configured to be used together with AVR-Libc.
13523 See the @code{--with-avrlibc} configure option.
13527 @node Blackfin Options
13528 @subsection Blackfin Options
13529 @cindex Blackfin Options
13532 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
13534 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
13535 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
13536 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
13537 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
13538 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
13539 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
13540 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
13541 @samp{bf561}, @samp{bf592}.
13543 The optional @var{sirevision} specifies the silicon revision of the target
13544 Blackfin processor. Any workarounds available for the targeted silicon revision
13545 are enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
13546 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
13547 are enabled. The @code{__SILICON_REVISION__} macro is defined to two
13548 hexadecimal digits representing the major and minor numbers in the silicon
13549 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
13550 is not defined. If @var{sirevision} is @samp{any}, the
13551 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
13552 If this optional @var{sirevision} is not used, GCC assumes the latest known
13553 silicon revision of the targeted Blackfin processor.
13555 GCC defines a preprocessor macro for the specified @var{cpu}.
13556 For the @samp{bfin-elf} toolchain, this option causes the hardware BSP
13557 provided by libgloss to be linked in if @option{-msim} is not given.
13559 Without this option, @samp{bf532} is used as the processor by default.
13561 Note that support for @samp{bf561} is incomplete. For @samp{bf561},
13562 only the preprocessor macro is defined.
13566 Specifies that the program will be run on the simulator. This causes
13567 the simulator BSP provided by libgloss to be linked in. This option
13568 has effect only for @samp{bfin-elf} toolchain.
13569 Certain other options, such as @option{-mid-shared-library} and
13570 @option{-mfdpic}, imply @option{-msim}.
13572 @item -momit-leaf-frame-pointer
13573 @opindex momit-leaf-frame-pointer
13574 Don't keep the frame pointer in a register for leaf functions. This
13575 avoids the instructions to save, set up and restore frame pointers and
13576 makes an extra register available in leaf functions. The option
13577 @option{-fomit-frame-pointer} removes the frame pointer for all functions,
13578 which might make debugging harder.
13580 @item -mspecld-anomaly
13581 @opindex mspecld-anomaly
13582 When enabled, the compiler ensures that the generated code does not
13583 contain speculative loads after jump instructions. If this option is used,
13584 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
13586 @item -mno-specld-anomaly
13587 @opindex mno-specld-anomaly
13588 Don't generate extra code to prevent speculative loads from occurring.
13590 @item -mcsync-anomaly
13591 @opindex mcsync-anomaly
13592 When enabled, the compiler ensures that the generated code does not
13593 contain CSYNC or SSYNC instructions too soon after conditional branches.
13594 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
13596 @item -mno-csync-anomaly
13597 @opindex mno-csync-anomaly
13598 Don't generate extra code to prevent CSYNC or SSYNC instructions from
13599 occurring too soon after a conditional branch.
13603 When enabled, the compiler is free to take advantage of the knowledge that
13604 the entire program fits into the low 64k of memory.
13607 @opindex mno-low-64k
13608 Assume that the program is arbitrarily large. This is the default.
13610 @item -mstack-check-l1
13611 @opindex mstack-check-l1
13612 Do stack checking using information placed into L1 scratchpad memory by the
13615 @item -mid-shared-library
13616 @opindex mid-shared-library
13617 Generate code that supports shared libraries via the library ID method.
13618 This allows for execute in place and shared libraries in an environment
13619 without virtual memory management. This option implies @option{-fPIC}.
13620 With a @samp{bfin-elf} target, this option implies @option{-msim}.
13622 @item -mno-id-shared-library
13623 @opindex mno-id-shared-library
13624 Generate code that doesn't assume ID-based shared libraries are being used.
13625 This is the default.
13627 @item -mleaf-id-shared-library
13628 @opindex mleaf-id-shared-library
13629 Generate code that supports shared libraries via the library ID method,
13630 but assumes that this library or executable won't link against any other
13631 ID shared libraries. That allows the compiler to use faster code for jumps
13634 @item -mno-leaf-id-shared-library
13635 @opindex mno-leaf-id-shared-library
13636 Do not assume that the code being compiled won't link against any ID shared
13637 libraries. Slower code is generated for jump and call insns.
13639 @item -mshared-library-id=n
13640 @opindex mshared-library-id
13641 Specifies the identification number of the ID-based shared library being
13642 compiled. Specifying a value of 0 generates more compact code; specifying
13643 other values forces the allocation of that number to the current
13644 library but is no more space- or time-efficient than omitting this option.
13648 Generate code that allows the data segment to be located in a different
13649 area of memory from the text segment. This allows for execute in place in
13650 an environment without virtual memory management by eliminating relocations
13651 against the text section.
13653 @item -mno-sep-data
13654 @opindex mno-sep-data
13655 Generate code that assumes that the data segment follows the text segment.
13656 This is the default.
13659 @itemx -mno-long-calls
13660 @opindex mlong-calls
13661 @opindex mno-long-calls
13662 Tells the compiler to perform function calls by first loading the
13663 address of the function into a register and then performing a subroutine
13664 call on this register. This switch is needed if the target function
13665 lies outside of the 24-bit addressing range of the offset-based
13666 version of subroutine call instruction.
13668 This feature is not enabled by default. Specifying
13669 @option{-mno-long-calls} restores the default behavior. Note these
13670 switches have no effect on how the compiler generates code to handle
13671 function calls via function pointers.
13675 Link with the fast floating-point library. This library relaxes some of
13676 the IEEE floating-point standard's rules for checking inputs against
13677 Not-a-Number (NAN), in the interest of performance.
13680 @opindex minline-plt
13681 Enable inlining of PLT entries in function calls to functions that are
13682 not known to bind locally. It has no effect without @option{-mfdpic}.
13685 @opindex mmulticore
13686 Build a standalone application for multicore Blackfin processors.
13687 This option causes proper start files and link scripts supporting
13688 multicore to be used, and defines the macro @code{__BFIN_MULTICORE}.
13689 It can only be used with @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}.
13691 This option can be used with @option{-mcorea} or @option{-mcoreb}, which
13692 selects the one-application-per-core programming model. Without
13693 @option{-mcorea} or @option{-mcoreb}, the single-application/dual-core
13694 programming model is used. In this model, the main function of Core B
13695 should be named as @code{coreb_main}.
13697 If this option is not used, the single-core application programming
13702 Build a standalone application for Core A of BF561 when using
13703 the one-application-per-core programming model. Proper start files
13704 and link scripts are used to support Core A, and the macro
13705 @code{__BFIN_COREA} is defined.
13706 This option can only be used in conjunction with @option{-mmulticore}.
13710 Build a standalone application for Core B of BF561 when using
13711 the one-application-per-core programming model. Proper start files
13712 and link scripts are used to support Core B, and the macro
13713 @code{__BFIN_COREB} is defined. When this option is used, @code{coreb_main}
13714 should be used instead of @code{main}.
13715 This option can only be used in conjunction with @option{-mmulticore}.
13719 Build a standalone application for SDRAM. Proper start files and
13720 link scripts are used to put the application into SDRAM, and the macro
13721 @code{__BFIN_SDRAM} is defined.
13722 The loader should initialize SDRAM before loading the application.
13726 Assume that ICPLBs are enabled at run time. This has an effect on certain
13727 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
13728 are enabled; for standalone applications the default is off.
13732 @subsection C6X Options
13733 @cindex C6X Options
13736 @item -march=@var{name}
13738 This specifies the name of the target architecture. GCC uses this
13739 name to determine what kind of instructions it can emit when generating
13740 assembly code. Permissible names are: @samp{c62x},
13741 @samp{c64x}, @samp{c64x+}, @samp{c67x}, @samp{c67x+}, @samp{c674x}.
13744 @opindex mbig-endian
13745 Generate code for a big-endian target.
13747 @item -mlittle-endian
13748 @opindex mlittle-endian
13749 Generate code for a little-endian target. This is the default.
13753 Choose startup files and linker script suitable for the simulator.
13755 @item -msdata=default
13756 @opindex msdata=default
13757 Put small global and static data in the @samp{.neardata} section,
13758 which is pointed to by register @code{B14}. Put small uninitialized
13759 global and static data in the @samp{.bss} section, which is adjacent
13760 to the @samp{.neardata} section. Put small read-only data into the
13761 @samp{.rodata} section. The corresponding sections used for large
13762 pieces of data are @samp{.fardata}, @samp{.far} and @samp{.const}.
13765 @opindex msdata=all
13766 Put all data, not just small objects, into the sections reserved for
13767 small data, and use addressing relative to the @code{B14} register to
13771 @opindex msdata=none
13772 Make no use of the sections reserved for small data, and use absolute
13773 addresses to access all data. Put all initialized global and static
13774 data in the @samp{.fardata} section, and all uninitialized data in the
13775 @samp{.far} section. Put all constant data into the @samp{.const}
13780 @subsection CRIS Options
13781 @cindex CRIS Options
13783 These options are defined specifically for the CRIS ports.
13786 @item -march=@var{architecture-type}
13787 @itemx -mcpu=@var{architecture-type}
13790 Generate code for the specified architecture. The choices for
13791 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
13792 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
13793 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
13796 @item -mtune=@var{architecture-type}
13798 Tune to @var{architecture-type} everything applicable about the generated
13799 code, except for the ABI and the set of available instructions. The
13800 choices for @var{architecture-type} are the same as for
13801 @option{-march=@var{architecture-type}}.
13803 @item -mmax-stack-frame=@var{n}
13804 @opindex mmax-stack-frame
13805 Warn when the stack frame of a function exceeds @var{n} bytes.
13811 The options @option{-metrax4} and @option{-metrax100} are synonyms for
13812 @option{-march=v3} and @option{-march=v8} respectively.
13814 @item -mmul-bug-workaround
13815 @itemx -mno-mul-bug-workaround
13816 @opindex mmul-bug-workaround
13817 @opindex mno-mul-bug-workaround
13818 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
13819 models where it applies. This option is active by default.
13823 Enable CRIS-specific verbose debug-related information in the assembly
13824 code. This option also has the effect of turning off the @samp{#NO_APP}
13825 formatted-code indicator to the assembler at the beginning of the
13830 Do not use condition-code results from previous instruction; always emit
13831 compare and test instructions before use of condition codes.
13833 @item -mno-side-effects
13834 @opindex mno-side-effects
13835 Do not emit instructions with side effects in addressing modes other than
13838 @item -mstack-align
13839 @itemx -mno-stack-align
13840 @itemx -mdata-align
13841 @itemx -mno-data-align
13842 @itemx -mconst-align
13843 @itemx -mno-const-align
13844 @opindex mstack-align
13845 @opindex mno-stack-align
13846 @opindex mdata-align
13847 @opindex mno-data-align
13848 @opindex mconst-align
13849 @opindex mno-const-align
13850 These options (@samp{no-} options) arrange (eliminate arrangements) for the
13851 stack frame, individual data and constants to be aligned for the maximum
13852 single data access size for the chosen CPU model. The default is to
13853 arrange for 32-bit alignment. ABI details such as structure layout are
13854 not affected by these options.
13862 Similar to the stack- data- and const-align options above, these options
13863 arrange for stack frame, writable data and constants to all be 32-bit,
13864 16-bit or 8-bit aligned. The default is 32-bit alignment.
13866 @item -mno-prologue-epilogue
13867 @itemx -mprologue-epilogue
13868 @opindex mno-prologue-epilogue
13869 @opindex mprologue-epilogue
13870 With @option{-mno-prologue-epilogue}, the normal function prologue and
13871 epilogue which set up the stack frame are omitted and no return
13872 instructions or return sequences are generated in the code. Use this
13873 option only together with visual inspection of the compiled code: no
13874 warnings or errors are generated when call-saved registers must be saved,
13875 or storage for local variables needs to be allocated.
13879 @opindex mno-gotplt
13881 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
13882 instruction sequences that load addresses for functions from the PLT part
13883 of the GOT rather than (traditional on other architectures) calls to the
13884 PLT@. The default is @option{-mgotplt}.
13888 Legacy no-op option only recognized with the cris-axis-elf and
13889 cris-axis-linux-gnu targets.
13893 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
13897 This option, recognized for the cris-axis-elf, arranges
13898 to link with input-output functions from a simulator library. Code,
13899 initialized data and zero-initialized data are allocated consecutively.
13903 Like @option{-sim}, but pass linker options to locate initialized data at
13904 0x40000000 and zero-initialized data at 0x80000000.
13908 @subsection CR16 Options
13909 @cindex CR16 Options
13911 These options are defined specifically for the CR16 ports.
13917 Enable the use of multiply-accumulate instructions. Disabled by default.
13921 @opindex mcr16cplus
13923 Generate code for CR16C or CR16C+ architecture. CR16C+ architecture
13928 Links the library libsim.a which is in compatible with simulator. Applicable
13929 to ELF compiler only.
13933 Choose integer type as 32-bit wide.
13937 Generates @code{sbit}/@code{cbit} instructions for bit manipulations.
13939 @item -mdata-model=@var{model}
13940 @opindex mdata-model
13941 Choose a data model. The choices for @var{model} are @samp{near},
13942 @samp{far} or @samp{medium}. @samp{medium} is default.
13943 However, @samp{far} is not valid with @option{-mcr16c}, as the
13944 CR16C architecture does not support the far data model.
13947 @node Darwin Options
13948 @subsection Darwin Options
13949 @cindex Darwin options
13951 These options are defined for all architectures running the Darwin operating
13954 FSF GCC on Darwin does not create ``fat'' object files; it creates
13955 an object file for the single architecture that GCC was built to
13956 target. Apple's GCC on Darwin does create ``fat'' files if multiple
13957 @option{-arch} options are used; it does so by running the compiler or
13958 linker multiple times and joining the results together with
13961 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
13962 @samp{i686}) is determined by the flags that specify the ISA
13963 that GCC is targeting, like @option{-mcpu} or @option{-march}. The
13964 @option{-force_cpusubtype_ALL} option can be used to override this.
13966 The Darwin tools vary in their behavior when presented with an ISA
13967 mismatch. The assembler, @file{as}, only permits instructions to
13968 be used that are valid for the subtype of the file it is generating,
13969 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
13970 The linker for shared libraries, @file{/usr/bin/libtool}, fails
13971 and prints an error if asked to create a shared library with a less
13972 restrictive subtype than its input files (for instance, trying to put
13973 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
13974 for executables, @command{ld}, quietly gives the executable the most
13975 restrictive subtype of any of its input files.
13980 Add the framework directory @var{dir} to the head of the list of
13981 directories to be searched for header files. These directories are
13982 interleaved with those specified by @option{-I} options and are
13983 scanned in a left-to-right order.
13985 A framework directory is a directory with frameworks in it. A
13986 framework is a directory with a @file{Headers} and/or
13987 @file{PrivateHeaders} directory contained directly in it that ends
13988 in @file{.framework}. The name of a framework is the name of this
13989 directory excluding the @file{.framework}. Headers associated with
13990 the framework are found in one of those two directories, with
13991 @file{Headers} being searched first. A subframework is a framework
13992 directory that is in a framework's @file{Frameworks} directory.
13993 Includes of subframework headers can only appear in a header of a
13994 framework that contains the subframework, or in a sibling subframework
13995 header. Two subframeworks are siblings if they occur in the same
13996 framework. A subframework should not have the same name as a
13997 framework; a warning is issued if this is violated. Currently a
13998 subframework cannot have subframeworks; in the future, the mechanism
13999 may be extended to support this. The standard frameworks can be found
14000 in @file{/System/Library/Frameworks} and
14001 @file{/Library/Frameworks}. An example include looks like
14002 @code{#include <Framework/header.h>}, where @file{Framework} denotes
14003 the name of the framework and @file{header.h} is found in the
14004 @file{PrivateHeaders} or @file{Headers} directory.
14006 @item -iframework@var{dir}
14007 @opindex iframework
14008 Like @option{-F} except the directory is a treated as a system
14009 directory. The main difference between this @option{-iframework} and
14010 @option{-F} is that with @option{-iframework} the compiler does not
14011 warn about constructs contained within header files found via
14012 @var{dir}. This option is valid only for the C family of languages.
14016 Emit debugging information for symbols that are used. For stabs
14017 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
14018 This is by default ON@.
14022 Emit debugging information for all symbols and types.
14024 @item -mmacosx-version-min=@var{version}
14025 The earliest version of MacOS X that this executable will run on
14026 is @var{version}. Typical values of @var{version} include @code{10.1},
14027 @code{10.2}, and @code{10.3.9}.
14029 If the compiler was built to use the system's headers by default,
14030 then the default for this option is the system version on which the
14031 compiler is running, otherwise the default is to make choices that
14032 are compatible with as many systems and code bases as possible.
14036 Enable kernel development mode. The @option{-mkernel} option sets
14037 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
14038 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
14039 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
14040 applicable. This mode also sets @option{-mno-altivec},
14041 @option{-msoft-float}, @option{-fno-builtin} and
14042 @option{-mlong-branch} for PowerPC targets.
14044 @item -mone-byte-bool
14045 @opindex mone-byte-bool
14046 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
14047 By default @samp{sizeof(bool)} is @samp{4} when compiling for
14048 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
14049 option has no effect on x86.
14051 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
14052 to generate code that is not binary compatible with code generated
14053 without that switch. Using this switch may require recompiling all
14054 other modules in a program, including system libraries. Use this
14055 switch to conform to a non-default data model.
14057 @item -mfix-and-continue
14058 @itemx -ffix-and-continue
14059 @itemx -findirect-data
14060 @opindex mfix-and-continue
14061 @opindex ffix-and-continue
14062 @opindex findirect-data
14063 Generate code suitable for fast turnaround development, such as to
14064 allow GDB to dynamically load @code{.o} files into already-running
14065 programs. @option{-findirect-data} and @option{-ffix-and-continue}
14066 are provided for backwards compatibility.
14070 Loads all members of static archive libraries.
14071 See man ld(1) for more information.
14073 @item -arch_errors_fatal
14074 @opindex arch_errors_fatal
14075 Cause the errors having to do with files that have the wrong architecture
14078 @item -bind_at_load
14079 @opindex bind_at_load
14080 Causes the output file to be marked such that the dynamic linker will
14081 bind all undefined references when the file is loaded or launched.
14085 Produce a Mach-o bundle format file.
14086 See man ld(1) for more information.
14088 @item -bundle_loader @var{executable}
14089 @opindex bundle_loader
14090 This option specifies the @var{executable} that will load the build
14091 output file being linked. See man ld(1) for more information.
14094 @opindex dynamiclib
14095 When passed this option, GCC produces a dynamic library instead of
14096 an executable when linking, using the Darwin @file{libtool} command.
14098 @item -force_cpusubtype_ALL
14099 @opindex force_cpusubtype_ALL
14100 This causes GCC's output file to have the @var{ALL} subtype, instead of
14101 one controlled by the @option{-mcpu} or @option{-march} option.
14103 @item -allowable_client @var{client_name}
14104 @itemx -client_name
14105 @itemx -compatibility_version
14106 @itemx -current_version
14108 @itemx -dependency-file
14110 @itemx -dylinker_install_name
14112 @itemx -exported_symbols_list
14115 @itemx -flat_namespace
14116 @itemx -force_flat_namespace
14117 @itemx -headerpad_max_install_names
14120 @itemx -install_name
14121 @itemx -keep_private_externs
14122 @itemx -multi_module
14123 @itemx -multiply_defined
14124 @itemx -multiply_defined_unused
14127 @itemx -no_dead_strip_inits_and_terms
14128 @itemx -nofixprebinding
14129 @itemx -nomultidefs
14131 @itemx -noseglinkedit
14132 @itemx -pagezero_size
14134 @itemx -prebind_all_twolevel_modules
14135 @itemx -private_bundle
14137 @itemx -read_only_relocs
14139 @itemx -sectobjectsymbols
14143 @itemx -sectobjectsymbols
14146 @itemx -segs_read_only_addr
14148 @itemx -segs_read_write_addr
14149 @itemx -seg_addr_table
14150 @itemx -seg_addr_table_filename
14151 @itemx -seglinkedit
14153 @itemx -segs_read_only_addr
14154 @itemx -segs_read_write_addr
14155 @itemx -single_module
14157 @itemx -sub_library
14159 @itemx -sub_umbrella
14160 @itemx -twolevel_namespace
14163 @itemx -unexported_symbols_list
14164 @itemx -weak_reference_mismatches
14165 @itemx -whatsloaded
14166 @opindex allowable_client
14167 @opindex client_name
14168 @opindex compatibility_version
14169 @opindex current_version
14170 @opindex dead_strip
14171 @opindex dependency-file
14172 @opindex dylib_file
14173 @opindex dylinker_install_name
14175 @opindex exported_symbols_list
14177 @opindex flat_namespace
14178 @opindex force_flat_namespace
14179 @opindex headerpad_max_install_names
14180 @opindex image_base
14182 @opindex install_name
14183 @opindex keep_private_externs
14184 @opindex multi_module
14185 @opindex multiply_defined
14186 @opindex multiply_defined_unused
14187 @opindex noall_load
14188 @opindex no_dead_strip_inits_and_terms
14189 @opindex nofixprebinding
14190 @opindex nomultidefs
14192 @opindex noseglinkedit
14193 @opindex pagezero_size
14195 @opindex prebind_all_twolevel_modules
14196 @opindex private_bundle
14197 @opindex read_only_relocs
14199 @opindex sectobjectsymbols
14202 @opindex sectcreate
14203 @opindex sectobjectsymbols
14206 @opindex segs_read_only_addr
14207 @opindex segs_read_write_addr
14208 @opindex seg_addr_table
14209 @opindex seg_addr_table_filename
14210 @opindex seglinkedit
14212 @opindex segs_read_only_addr
14213 @opindex segs_read_write_addr
14214 @opindex single_module
14216 @opindex sub_library
14217 @opindex sub_umbrella
14218 @opindex twolevel_namespace
14221 @opindex unexported_symbols_list
14222 @opindex weak_reference_mismatches
14223 @opindex whatsloaded
14224 These options are passed to the Darwin linker. The Darwin linker man page
14225 describes them in detail.
14228 @node DEC Alpha Options
14229 @subsection DEC Alpha Options
14231 These @samp{-m} options are defined for the DEC Alpha implementations:
14234 @item -mno-soft-float
14235 @itemx -msoft-float
14236 @opindex mno-soft-float
14237 @opindex msoft-float
14238 Use (do not use) the hardware floating-point instructions for
14239 floating-point operations. When @option{-msoft-float} is specified,
14240 functions in @file{libgcc.a} are used to perform floating-point
14241 operations. Unless they are replaced by routines that emulate the
14242 floating-point operations, or compiled in such a way as to call such
14243 emulations routines, these routines issue floating-point
14244 operations. If you are compiling for an Alpha without floating-point
14245 operations, you must ensure that the library is built so as not to call
14248 Note that Alpha implementations without floating-point operations are
14249 required to have floating-point registers.
14252 @itemx -mno-fp-regs
14254 @opindex mno-fp-regs
14255 Generate code that uses (does not use) the floating-point register set.
14256 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
14257 register set is not used, floating-point operands are passed in integer
14258 registers as if they were integers and floating-point results are passed
14259 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
14260 so any function with a floating-point argument or return value called by code
14261 compiled with @option{-mno-fp-regs} must also be compiled with that
14264 A typical use of this option is building a kernel that does not use,
14265 and hence need not save and restore, any floating-point registers.
14269 The Alpha architecture implements floating-point hardware optimized for
14270 maximum performance. It is mostly compliant with the IEEE floating-point
14271 standard. However, for full compliance, software assistance is
14272 required. This option generates code fully IEEE-compliant code
14273 @emph{except} that the @var{inexact-flag} is not maintained (see below).
14274 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
14275 defined during compilation. The resulting code is less efficient but is
14276 able to correctly support denormalized numbers and exceptional IEEE
14277 values such as not-a-number and plus/minus infinity. Other Alpha
14278 compilers call this option @option{-ieee_with_no_inexact}.
14280 @item -mieee-with-inexact
14281 @opindex mieee-with-inexact
14282 This is like @option{-mieee} except the generated code also maintains
14283 the IEEE @var{inexact-flag}. Turning on this option causes the
14284 generated code to implement fully-compliant IEEE math. In addition to
14285 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
14286 macro. On some Alpha implementations the resulting code may execute
14287 significantly slower than the code generated by default. Since there is
14288 very little code that depends on the @var{inexact-flag}, you should
14289 normally not specify this option. Other Alpha compilers call this
14290 option @option{-ieee_with_inexact}.
14292 @item -mfp-trap-mode=@var{trap-mode}
14293 @opindex mfp-trap-mode
14294 This option controls what floating-point related traps are enabled.
14295 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
14296 The trap mode can be set to one of four values:
14300 This is the default (normal) setting. The only traps that are enabled
14301 are the ones that cannot be disabled in software (e.g., division by zero
14305 In addition to the traps enabled by @samp{n}, underflow traps are enabled
14309 Like @samp{u}, but the instructions are marked to be safe for software
14310 completion (see Alpha architecture manual for details).
14313 Like @samp{su}, but inexact traps are enabled as well.
14316 @item -mfp-rounding-mode=@var{rounding-mode}
14317 @opindex mfp-rounding-mode
14318 Selects the IEEE rounding mode. Other Alpha compilers call this option
14319 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
14324 Normal IEEE rounding mode. Floating-point numbers are rounded towards
14325 the nearest machine number or towards the even machine number in case
14329 Round towards minus infinity.
14332 Chopped rounding mode. Floating-point numbers are rounded towards zero.
14335 Dynamic rounding mode. A field in the floating-point control register
14336 (@var{fpcr}, see Alpha architecture reference manual) controls the
14337 rounding mode in effect. The C library initializes this register for
14338 rounding towards plus infinity. Thus, unless your program modifies the
14339 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
14342 @item -mtrap-precision=@var{trap-precision}
14343 @opindex mtrap-precision
14344 In the Alpha architecture, floating-point traps are imprecise. This
14345 means without software assistance it is impossible to recover from a
14346 floating trap and program execution normally needs to be terminated.
14347 GCC can generate code that can assist operating system trap handlers
14348 in determining the exact location that caused a floating-point trap.
14349 Depending on the requirements of an application, different levels of
14350 precisions can be selected:
14354 Program precision. This option is the default and means a trap handler
14355 can only identify which program caused a floating-point exception.
14358 Function precision. The trap handler can determine the function that
14359 caused a floating-point exception.
14362 Instruction precision. The trap handler can determine the exact
14363 instruction that caused a floating-point exception.
14366 Other Alpha compilers provide the equivalent options called
14367 @option{-scope_safe} and @option{-resumption_safe}.
14369 @item -mieee-conformant
14370 @opindex mieee-conformant
14371 This option marks the generated code as IEEE conformant. You must not
14372 use this option unless you also specify @option{-mtrap-precision=i} and either
14373 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
14374 is to emit the line @samp{.eflag 48} in the function prologue of the
14375 generated assembly file.
14377 @item -mbuild-constants
14378 @opindex mbuild-constants
14379 Normally GCC examines a 32- or 64-bit integer constant to
14380 see if it can construct it from smaller constants in two or three
14381 instructions. If it cannot, it outputs the constant as a literal and
14382 generates code to load it from the data segment at run time.
14384 Use this option to require GCC to construct @emph{all} integer constants
14385 using code, even if it takes more instructions (the maximum is six).
14387 You typically use this option to build a shared library dynamic
14388 loader. Itself a shared library, it must relocate itself in memory
14389 before it can find the variables and constants in its own data segment.
14407 Indicate whether GCC should generate code to use the optional BWX,
14408 CIX, FIX and MAX instruction sets. The default is to use the instruction
14409 sets supported by the CPU type specified via @option{-mcpu=} option or that
14410 of the CPU on which GCC was built if none is specified.
14413 @itemx -mfloat-ieee
14414 @opindex mfloat-vax
14415 @opindex mfloat-ieee
14416 Generate code that uses (does not use) VAX F and G floating-point
14417 arithmetic instead of IEEE single and double precision.
14419 @item -mexplicit-relocs
14420 @itemx -mno-explicit-relocs
14421 @opindex mexplicit-relocs
14422 @opindex mno-explicit-relocs
14423 Older Alpha assemblers provided no way to generate symbol relocations
14424 except via assembler macros. Use of these macros does not allow
14425 optimal instruction scheduling. GNU binutils as of version 2.12
14426 supports a new syntax that allows the compiler to explicitly mark
14427 which relocations should apply to which instructions. This option
14428 is mostly useful for debugging, as GCC detects the capabilities of
14429 the assembler when it is built and sets the default accordingly.
14432 @itemx -mlarge-data
14433 @opindex msmall-data
14434 @opindex mlarge-data
14435 When @option{-mexplicit-relocs} is in effect, static data is
14436 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
14437 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
14438 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
14439 16-bit relocations off of the @code{$gp} register. This limits the
14440 size of the small data area to 64KB, but allows the variables to be
14441 directly accessed via a single instruction.
14443 The default is @option{-mlarge-data}. With this option the data area
14444 is limited to just below 2GB@. Programs that require more than 2GB of
14445 data must use @code{malloc} or @code{mmap} to allocate the data in the
14446 heap instead of in the program's data segment.
14448 When generating code for shared libraries, @option{-fpic} implies
14449 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
14452 @itemx -mlarge-text
14453 @opindex msmall-text
14454 @opindex mlarge-text
14455 When @option{-msmall-text} is used, the compiler assumes that the
14456 code of the entire program (or shared library) fits in 4MB, and is
14457 thus reachable with a branch instruction. When @option{-msmall-data}
14458 is used, the compiler can assume that all local symbols share the
14459 same @code{$gp} value, and thus reduce the number of instructions
14460 required for a function call from 4 to 1.
14462 The default is @option{-mlarge-text}.
14464 @item -mcpu=@var{cpu_type}
14466 Set the instruction set and instruction scheduling parameters for
14467 machine type @var{cpu_type}. You can specify either the @samp{EV}
14468 style name or the corresponding chip number. GCC supports scheduling
14469 parameters for the EV4, EV5 and EV6 family of processors and
14470 chooses the default values for the instruction set from the processor
14471 you specify. If you do not specify a processor type, GCC defaults
14472 to the processor on which the compiler was built.
14474 Supported values for @var{cpu_type} are
14480 Schedules as an EV4 and has no instruction set extensions.
14484 Schedules as an EV5 and has no instruction set extensions.
14488 Schedules as an EV5 and supports the BWX extension.
14493 Schedules as an EV5 and supports the BWX and MAX extensions.
14497 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
14501 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
14504 Native toolchains also support the value @samp{native},
14505 which selects the best architecture option for the host processor.
14506 @option{-mcpu=native} has no effect if GCC does not recognize
14509 @item -mtune=@var{cpu_type}
14511 Set only the instruction scheduling parameters for machine type
14512 @var{cpu_type}. The instruction set is not changed.
14514 Native toolchains also support the value @samp{native},
14515 which selects the best architecture option for the host processor.
14516 @option{-mtune=native} has no effect if GCC does not recognize
14519 @item -mmemory-latency=@var{time}
14520 @opindex mmemory-latency
14521 Sets the latency the scheduler should assume for typical memory
14522 references as seen by the application. This number is highly
14523 dependent on the memory access patterns used by the application
14524 and the size of the external cache on the machine.
14526 Valid options for @var{time} are
14530 A decimal number representing clock cycles.
14536 The compiler contains estimates of the number of clock cycles for
14537 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
14538 (also called Dcache, Scache, and Bcache), as well as to main memory.
14539 Note that L3 is only valid for EV5.
14545 @subsection FR30 Options
14546 @cindex FR30 Options
14548 These options are defined specifically for the FR30 port.
14552 @item -msmall-model
14553 @opindex msmall-model
14554 Use the small address space model. This can produce smaller code, but
14555 it does assume that all symbolic values and addresses fit into a
14560 Assume that runtime support has been provided and so there is no need
14561 to include the simulator library (@file{libsim.a}) on the linker
14567 @subsection FRV Options
14568 @cindex FRV Options
14574 Only use the first 32 general-purpose registers.
14579 Use all 64 general-purpose registers.
14584 Use only the first 32 floating-point registers.
14589 Use all 64 floating-point registers.
14592 @opindex mhard-float
14594 Use hardware instructions for floating-point operations.
14597 @opindex msoft-float
14599 Use library routines for floating-point operations.
14604 Dynamically allocate condition code registers.
14609 Do not try to dynamically allocate condition code registers, only
14610 use @code{icc0} and @code{fcc0}.
14615 Change ABI to use double word insns.
14620 Do not use double word instructions.
14625 Use floating-point double instructions.
14628 @opindex mno-double
14630 Do not use floating-point double instructions.
14635 Use media instructions.
14640 Do not use media instructions.
14645 Use multiply and add/subtract instructions.
14648 @opindex mno-muladd
14650 Do not use multiply and add/subtract instructions.
14655 Select the FDPIC ABI, which uses function descriptors to represent
14656 pointers to functions. Without any PIC/PIE-related options, it
14657 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
14658 assumes GOT entries and small data are within a 12-bit range from the
14659 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
14660 are computed with 32 bits.
14661 With a @samp{bfin-elf} target, this option implies @option{-msim}.
14664 @opindex minline-plt
14666 Enable inlining of PLT entries in function calls to functions that are
14667 not known to bind locally. It has no effect without @option{-mfdpic}.
14668 It's enabled by default if optimizing for speed and compiling for
14669 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
14670 optimization option such as @option{-O3} or above is present in the
14676 Assume a large TLS segment when generating thread-local code.
14681 Do not assume a large TLS segment when generating thread-local code.
14686 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
14687 that is known to be in read-only sections. It's enabled by default,
14688 except for @option{-fpic} or @option{-fpie}: even though it may help
14689 make the global offset table smaller, it trades 1 instruction for 4.
14690 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
14691 one of which may be shared by multiple symbols, and it avoids the need
14692 for a GOT entry for the referenced symbol, so it's more likely to be a
14693 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
14695 @item -multilib-library-pic
14696 @opindex multilib-library-pic
14698 Link with the (library, not FD) pic libraries. It's implied by
14699 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
14700 @option{-fpic} without @option{-mfdpic}. You should never have to use
14704 @opindex mlinked-fp
14706 Follow the EABI requirement of always creating a frame pointer whenever
14707 a stack frame is allocated. This option is enabled by default and can
14708 be disabled with @option{-mno-linked-fp}.
14711 @opindex mlong-calls
14713 Use indirect addressing to call functions outside the current
14714 compilation unit. This allows the functions to be placed anywhere
14715 within the 32-bit address space.
14717 @item -malign-labels
14718 @opindex malign-labels
14720 Try to align labels to an 8-byte boundary by inserting NOPs into the
14721 previous packet. This option only has an effect when VLIW packing
14722 is enabled. It doesn't create new packets; it merely adds NOPs to
14725 @item -mlibrary-pic
14726 @opindex mlibrary-pic
14728 Generate position-independent EABI code.
14733 Use only the first four media accumulator registers.
14738 Use all eight media accumulator registers.
14743 Pack VLIW instructions.
14748 Do not pack VLIW instructions.
14751 @opindex mno-eflags
14753 Do not mark ABI switches in e_flags.
14756 @opindex mcond-move
14758 Enable the use of conditional-move instructions (default).
14760 This switch is mainly for debugging the compiler and will likely be removed
14761 in a future version.
14763 @item -mno-cond-move
14764 @opindex mno-cond-move
14766 Disable the use of conditional-move instructions.
14768 This switch is mainly for debugging the compiler and will likely be removed
14769 in a future version.
14774 Enable the use of conditional set instructions (default).
14776 This switch is mainly for debugging the compiler and will likely be removed
14777 in a future version.
14782 Disable the use of conditional set instructions.
14784 This switch is mainly for debugging the compiler and will likely be removed
14785 in a future version.
14788 @opindex mcond-exec
14790 Enable the use of conditional execution (default).
14792 This switch is mainly for debugging the compiler and will likely be removed
14793 in a future version.
14795 @item -mno-cond-exec
14796 @opindex mno-cond-exec
14798 Disable the use of conditional execution.
14800 This switch is mainly for debugging the compiler and will likely be removed
14801 in a future version.
14803 @item -mvliw-branch
14804 @opindex mvliw-branch
14806 Run a pass to pack branches into VLIW instructions (default).
14808 This switch is mainly for debugging the compiler and will likely be removed
14809 in a future version.
14811 @item -mno-vliw-branch
14812 @opindex mno-vliw-branch
14814 Do not run a pass to pack branches into VLIW instructions.
14816 This switch is mainly for debugging the compiler and will likely be removed
14817 in a future version.
14819 @item -mmulti-cond-exec
14820 @opindex mmulti-cond-exec
14822 Enable optimization of @code{&&} and @code{||} in conditional execution
14825 This switch is mainly for debugging the compiler and will likely be removed
14826 in a future version.
14828 @item -mno-multi-cond-exec
14829 @opindex mno-multi-cond-exec
14831 Disable optimization of @code{&&} and @code{||} in conditional execution.
14833 This switch is mainly for debugging the compiler and will likely be removed
14834 in a future version.
14836 @item -mnested-cond-exec
14837 @opindex mnested-cond-exec
14839 Enable nested conditional execution optimizations (default).
14841 This switch is mainly for debugging the compiler and will likely be removed
14842 in a future version.
14844 @item -mno-nested-cond-exec
14845 @opindex mno-nested-cond-exec
14847 Disable nested conditional execution optimizations.
14849 This switch is mainly for debugging the compiler and will likely be removed
14850 in a future version.
14852 @item -moptimize-membar
14853 @opindex moptimize-membar
14855 This switch removes redundant @code{membar} instructions from the
14856 compiler-generated code. It is enabled by default.
14858 @item -mno-optimize-membar
14859 @opindex mno-optimize-membar
14861 This switch disables the automatic removal of redundant @code{membar}
14862 instructions from the generated code.
14864 @item -mtomcat-stats
14865 @opindex mtomcat-stats
14867 Cause gas to print out tomcat statistics.
14869 @item -mcpu=@var{cpu}
14872 Select the processor type for which to generate code. Possible values are
14873 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
14874 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
14878 @node GNU/Linux Options
14879 @subsection GNU/Linux Options
14881 These @samp{-m} options are defined for GNU/Linux targets:
14886 Use the GNU C library. This is the default except
14887 on @samp{*-*-linux-*uclibc*} and @samp{*-*-linux-*android*} targets.
14891 Use uClibc C library. This is the default on
14892 @samp{*-*-linux-*uclibc*} targets.
14896 Use Bionic C library. This is the default on
14897 @samp{*-*-linux-*android*} targets.
14901 Compile code compatible with Android platform. This is the default on
14902 @samp{*-*-linux-*android*} targets.
14904 When compiling, this option enables @option{-mbionic}, @option{-fPIC},
14905 @option{-fno-exceptions} and @option{-fno-rtti} by default. When linking,
14906 this option makes the GCC driver pass Android-specific options to the linker.
14907 Finally, this option causes the preprocessor macro @code{__ANDROID__}
14910 @item -tno-android-cc
14911 @opindex tno-android-cc
14912 Disable compilation effects of @option{-mandroid}, i.e., do not enable
14913 @option{-mbionic}, @option{-fPIC}, @option{-fno-exceptions} and
14914 @option{-fno-rtti} by default.
14916 @item -tno-android-ld
14917 @opindex tno-android-ld
14918 Disable linking effects of @option{-mandroid}, i.e., pass standard Linux
14919 linking options to the linker.
14923 @node H8/300 Options
14924 @subsection H8/300 Options
14926 These @samp{-m} options are defined for the H8/300 implementations:
14931 Shorten some address references at link time, when possible; uses the
14932 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
14933 ld, Using ld}, for a fuller description.
14937 Generate code for the H8/300H@.
14941 Generate code for the H8S@.
14945 Generate code for the H8S and H8/300H in the normal mode. This switch
14946 must be used either with @option{-mh} or @option{-ms}.
14950 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
14954 Extended registers are stored on stack before execution of function
14955 with monitor attribute. Default option is @option{-mexr}.
14956 This option is valid only for H8S targets.
14960 Extended registers are not stored on stack before execution of function
14961 with monitor attribute. Default option is @option{-mno-exr}.
14962 This option is valid only for H8S targets.
14966 Make @code{int} data 32 bits by default.
14969 @opindex malign-300
14970 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
14971 The default for the H8/300H and H8S is to align longs and floats on
14973 @option{-malign-300} causes them to be aligned on 2-byte boundaries.
14974 This option has no effect on the H8/300.
14978 @subsection HPPA Options
14979 @cindex HPPA Options
14981 These @samp{-m} options are defined for the HPPA family of computers:
14984 @item -march=@var{architecture-type}
14986 Generate code for the specified architecture. The choices for
14987 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
14988 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
14989 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
14990 architecture option for your machine. Code compiled for lower numbered
14991 architectures runs on higher numbered architectures, but not the
14994 @item -mpa-risc-1-0
14995 @itemx -mpa-risc-1-1
14996 @itemx -mpa-risc-2-0
14997 @opindex mpa-risc-1-0
14998 @opindex mpa-risc-1-1
14999 @opindex mpa-risc-2-0
15000 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
15002 @item -mjump-in-delay
15003 @opindex mjump-in-delay
15004 Fill delay slots of function calls with unconditional jump instructions
15005 by modifying the return pointer for the function call to be the target
15006 of the conditional jump.
15008 @item -mdisable-fpregs
15009 @opindex mdisable-fpregs
15010 Prevent floating-point registers from being used in any manner. This is
15011 necessary for compiling kernels that perform lazy context switching of
15012 floating-point registers. If you use this option and attempt to perform
15013 floating-point operations, the compiler aborts.
15015 @item -mdisable-indexing
15016 @opindex mdisable-indexing
15017 Prevent the compiler from using indexing address modes. This avoids some
15018 rather obscure problems when compiling MIG generated code under MACH@.
15020 @item -mno-space-regs
15021 @opindex mno-space-regs
15022 Generate code that assumes the target has no space registers. This allows
15023 GCC to generate faster indirect calls and use unscaled index address modes.
15025 Such code is suitable for level 0 PA systems and kernels.
15027 @item -mfast-indirect-calls
15028 @opindex mfast-indirect-calls
15029 Generate code that assumes calls never cross space boundaries. This
15030 allows GCC to emit code that performs faster indirect calls.
15032 This option does not work in the presence of shared libraries or nested
15035 @item -mfixed-range=@var{register-range}
15036 @opindex mfixed-range
15037 Generate code treating the given register range as fixed registers.
15038 A fixed register is one that the register allocator cannot use. This is
15039 useful when compiling kernel code. A register range is specified as
15040 two registers separated by a dash. Multiple register ranges can be
15041 specified separated by a comma.
15043 @item -mlong-load-store
15044 @opindex mlong-load-store
15045 Generate 3-instruction load and store sequences as sometimes required by
15046 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
15049 @item -mportable-runtime
15050 @opindex mportable-runtime
15051 Use the portable calling conventions proposed by HP for ELF systems.
15055 Enable the use of assembler directives only GAS understands.
15057 @item -mschedule=@var{cpu-type}
15059 Schedule code according to the constraints for the machine type
15060 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
15061 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
15062 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
15063 proper scheduling option for your machine. The default scheduling is
15067 @opindex mlinker-opt
15068 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
15069 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
15070 linkers in which they give bogus error messages when linking some programs.
15073 @opindex msoft-float
15074 Generate output containing library calls for floating point.
15075 @strong{Warning:} the requisite libraries are not available for all HPPA
15076 targets. Normally the facilities of the machine's usual C compiler are
15077 used, but this cannot be done directly in cross-compilation. You must make
15078 your own arrangements to provide suitable library functions for
15081 @option{-msoft-float} changes the calling convention in the output file;
15082 therefore, it is only useful if you compile @emph{all} of a program with
15083 this option. In particular, you need to compile @file{libgcc.a}, the
15084 library that comes with GCC, with @option{-msoft-float} in order for
15089 Generate the predefine, @code{_SIO}, for server IO@. The default is
15090 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
15091 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
15092 options are available under HP-UX and HI-UX@.
15096 Use options specific to GNU @command{ld}.
15097 This passes @option{-shared} to @command{ld} when
15098 building a shared library. It is the default when GCC is configured,
15099 explicitly or implicitly, with the GNU linker. This option does not
15100 affect which @command{ld} is called; it only changes what parameters
15101 are passed to that @command{ld}.
15102 The @command{ld} that is called is determined by the
15103 @option{--with-ld} configure option, GCC's program search path, and
15104 finally by the user's @env{PATH}. The linker used by GCC can be printed
15105 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
15106 on the 64-bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
15110 Use options specific to HP @command{ld}.
15111 This passes @option{-b} to @command{ld} when building
15112 a shared library and passes @option{+Accept TypeMismatch} to @command{ld} on all
15113 links. It is the default when GCC is configured, explicitly or
15114 implicitly, with the HP linker. This option does not affect
15115 which @command{ld} is called; it only changes what parameters are passed to that
15117 The @command{ld} that is called is determined by the @option{--with-ld}
15118 configure option, GCC's program search path, and finally by the user's
15119 @env{PATH}. The linker used by GCC can be printed using @samp{which
15120 `gcc -print-prog-name=ld`}. This option is only available on the 64-bit
15121 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
15124 @opindex mno-long-calls
15125 Generate code that uses long call sequences. This ensures that a call
15126 is always able to reach linker generated stubs. The default is to generate
15127 long calls only when the distance from the call site to the beginning
15128 of the function or translation unit, as the case may be, exceeds a
15129 predefined limit set by the branch type being used. The limits for
15130 normal calls are 7,600,000 and 240,000 bytes, respectively for the
15131 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
15134 Distances are measured from the beginning of functions when using the
15135 @option{-ffunction-sections} option, or when using the @option{-mgas}
15136 and @option{-mno-portable-runtime} options together under HP-UX with
15139 It is normally not desirable to use this option as it degrades
15140 performance. However, it may be useful in large applications,
15141 particularly when partial linking is used to build the application.
15143 The types of long calls used depends on the capabilities of the
15144 assembler and linker, and the type of code being generated. The
15145 impact on systems that support long absolute calls, and long pic
15146 symbol-difference or pc-relative calls should be relatively small.
15147 However, an indirect call is used on 32-bit ELF systems in pic code
15148 and it is quite long.
15150 @item -munix=@var{unix-std}
15152 Generate compiler predefines and select a startfile for the specified
15153 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
15154 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
15155 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
15156 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
15157 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
15160 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
15161 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
15162 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
15163 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
15164 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
15165 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
15167 It is @emph{important} to note that this option changes the interfaces
15168 for various library routines. It also affects the operational behavior
15169 of the C library. Thus, @emph{extreme} care is needed in using this
15172 Library code that is intended to operate with more than one UNIX
15173 standard must test, set and restore the variable @var{__xpg4_extended_mask}
15174 as appropriate. Most GNU software doesn't provide this capability.
15178 Suppress the generation of link options to search libdld.sl when the
15179 @option{-static} option is specified on HP-UX 10 and later.
15183 The HP-UX implementation of setlocale in libc has a dependency on
15184 libdld.sl. There isn't an archive version of libdld.sl. Thus,
15185 when the @option{-static} option is specified, special link options
15186 are needed to resolve this dependency.
15188 On HP-UX 10 and later, the GCC driver adds the necessary options to
15189 link with libdld.sl when the @option{-static} option is specified.
15190 This causes the resulting binary to be dynamic. On the 64-bit port,
15191 the linkers generate dynamic binaries by default in any case. The
15192 @option{-nolibdld} option can be used to prevent the GCC driver from
15193 adding these link options.
15197 Add support for multithreading with the @dfn{dce thread} library
15198 under HP-UX@. This option sets flags for both the preprocessor and
15202 @node i386 and x86-64 Options
15203 @subsection Intel 386 and AMD x86-64 Options
15204 @cindex i386 Options
15205 @cindex x86-64 Options
15206 @cindex Intel 386 Options
15207 @cindex AMD x86-64 Options
15209 These @samp{-m} options are defined for the i386 and x86-64 family of
15214 @item -march=@var{cpu-type}
15216 Generate instructions for the machine type @var{cpu-type}. In contrast to
15217 @option{-mtune=@var{cpu-type}}, which merely tunes the generated code
15218 for the specified @var{cpu-type}, @option{-march=@var{cpu-type}} allows GCC
15219 to generate code that may not run at all on processors other than the one
15220 indicated. Specifying @option{-march=@var{cpu-type}} implies
15221 @option{-mtune=@var{cpu-type}}.
15223 The choices for @var{cpu-type} are:
15227 This selects the CPU to generate code for at compilation time by determining
15228 the processor type of the compiling machine. Using @option{-march=native}
15229 enables all instruction subsets supported by the local machine (hence
15230 the result might not run on different machines). Using @option{-mtune=native}
15231 produces code optimized for the local machine under the constraints
15232 of the selected instruction set.
15235 Original Intel i386 CPU@.
15238 Intel i486 CPU@. (No scheduling is implemented for this chip.)
15242 Intel Pentium CPU with no MMX support.
15245 Intel Pentium MMX CPU, based on Pentium core with MMX instruction set support.
15248 Intel Pentium Pro CPU@.
15251 When used with @option{-march}, the Pentium Pro
15252 instruction set is used, so the code runs on all i686 family chips.
15253 When used with @option{-mtune}, it has the same meaning as @samp{generic}.
15256 Intel Pentium II CPU, based on Pentium Pro core with MMX instruction set
15261 Intel Pentium III CPU, based on Pentium Pro core with MMX and SSE instruction
15265 Intel Pentium M; low-power version of Intel Pentium III CPU
15266 with MMX, SSE and SSE2 instruction set support. Used by Centrino notebooks.
15270 Intel Pentium 4 CPU with MMX, SSE and SSE2 instruction set support.
15273 Improved version of Intel Pentium 4 CPU with MMX, SSE, SSE2 and SSE3 instruction
15277 Improved version of Intel Pentium 4 CPU with 64-bit extensions, MMX, SSE,
15278 SSE2 and SSE3 instruction set support.
15281 Intel Core 2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
15282 instruction set support.
15285 Intel Nehalem CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
15286 SSE4.1, SSE4.2 and POPCNT instruction set support.
15289 Intel Westmere CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
15290 SSE4.1, SSE4.2, POPCNT, AES and PCLMUL instruction set support.
15293 Intel Sandy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
15294 SSE4.1, SSE4.2, POPCNT, AVX, AES and PCLMUL instruction set support.
15297 Intel Ivy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
15298 SSE4.1, SSE4.2, POPCNT, AVX, AES, PCLMUL, FSGSBASE, RDRND and F16C
15299 instruction set support.
15302 Intel Haswell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
15303 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
15304 BMI, BMI2 and F16C instruction set support.
15307 Intel Broadwell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
15308 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
15309 BMI, BMI2, F16C, RDSEED, ADCX and PREFETCHW instruction set support.
15312 Intel Bonnell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3 and SSSE3
15313 instruction set support.
15316 Intel Silvermont CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
15317 SSE4.1, SSE4.2, POPCNT, AES, PCLMUL and RDRND instruction set support.
15320 AMD K6 CPU with MMX instruction set support.
15324 Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
15327 @itemx athlon-tbird
15328 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
15334 Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
15335 instruction set support.
15341 Processors based on the AMD K8 core with x86-64 instruction set support,
15342 including the AMD Opteron, Athlon 64, and Athlon 64 FX processors.
15343 (This supersets MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit
15344 instruction set extensions.)
15347 @itemx opteron-sse3
15348 @itemx athlon64-sse3
15349 Improved versions of AMD K8 cores with SSE3 instruction set support.
15353 CPUs based on AMD Family 10h cores with x86-64 instruction set support. (This
15354 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
15355 instruction set extensions.)
15358 CPUs based on AMD Family 15h cores with x86-64 instruction set support. (This
15359 supersets FMA4, AVX, XOP, LWP, AES, PCL_MUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A,
15360 SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set extensions.)
15362 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
15363 supersets BMI, TBM, F16C, FMA, FMA4, AVX, XOP, LWP, AES, PCL_MUL, CX16, MMX,
15364 SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set
15367 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
15368 supersets BMI, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, XOP, LWP, AES,
15369 PCL_MUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and
15370 64-bit instruction set extensions.
15372 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
15373 supersets BMI, BMI2, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, AVX2, XOP, LWP,
15374 AES, PCL_MUL, CX16, MOVBE, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1,
15375 SSE4.2, ABM and 64-bit instruction set extensions.
15378 CPUs based on AMD Family 14h cores with x86-64 instruction set support. (This
15379 supersets MMX, SSE, SSE2, SSE3, SSSE3, SSE4A, CX16, ABM and 64-bit
15380 instruction set extensions.)
15383 CPUs based on AMD Family 16h cores with x86-64 instruction set support. This
15384 includes MOVBE, F16C, BMI, AVX, PCL_MUL, AES, SSE4.2, SSE4.1, CX16, ABM,
15385 SSE4A, SSSE3, SSE3, SSE2, SSE, MMX and 64-bit instruction set extensions.
15388 IDT WinChip C6 CPU, dealt in same way as i486 with additional MMX instruction
15392 IDT WinChip 2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
15393 instruction set support.
15396 VIA C3 CPU with MMX and 3DNow!@: instruction set support. (No scheduling is
15397 implemented for this chip.)
15400 VIA C3-2 (Nehemiah/C5XL) CPU with MMX and SSE instruction set support.
15402 implemented for this chip.)
15405 AMD Geode embedded processor with MMX and 3DNow!@: instruction set support.
15408 @item -mtune=@var{cpu-type}
15410 Tune to @var{cpu-type} everything applicable about the generated code, except
15411 for the ABI and the set of available instructions.
15412 While picking a specific @var{cpu-type} schedules things appropriately
15413 for that particular chip, the compiler does not generate any code that
15414 cannot run on the default machine type unless you use a
15415 @option{-march=@var{cpu-type}} option.
15416 For example, if GCC is configured for i686-pc-linux-gnu
15417 then @option{-mtune=pentium4} generates code that is tuned for Pentium 4
15418 but still runs on i686 machines.
15420 The choices for @var{cpu-type} are the same as for @option{-march}.
15421 In addition, @option{-mtune} supports 2 extra choices for @var{cpu-type}:
15425 Produce code optimized for the most common IA32/@/AMD64/@/EM64T processors.
15426 If you know the CPU on which your code will run, then you should use
15427 the corresponding @option{-mtune} or @option{-march} option instead of
15428 @option{-mtune=generic}. But, if you do not know exactly what CPU users
15429 of your application will have, then you should use this option.
15431 As new processors are deployed in the marketplace, the behavior of this
15432 option will change. Therefore, if you upgrade to a newer version of
15433 GCC, code generation controlled by this option will change to reflect
15435 that are most common at the time that version of GCC is released.
15437 There is no @option{-march=generic} option because @option{-march}
15438 indicates the instruction set the compiler can use, and there is no
15439 generic instruction set applicable to all processors. In contrast,
15440 @option{-mtune} indicates the processor (or, in this case, collection of
15441 processors) for which the code is optimized.
15444 Produce code optimized for the most current Intel processors, which are
15445 Haswell and Silvermont for this version of GCC. If you know the CPU
15446 on which your code will run, then you should use the corresponding
15447 @option{-mtune} or @option{-march} option instead of @option{-mtune=intel}.
15448 But, if you want your application performs better on both Haswell and
15449 Silvermont, then you should use this option.
15451 As new Intel processors are deployed in the marketplace, the behavior of
15452 this option will change. Therefore, if you upgrade to a newer version of
15453 GCC, code generation controlled by this option will change to reflect
15454 the most current Intel processors at the time that version of GCC is
15457 There is no @option{-march=intel} option because @option{-march} indicates
15458 the instruction set the compiler can use, and there is no common
15459 instruction set applicable to all processors. In contrast,
15460 @option{-mtune} indicates the processor (or, in this case, collection of
15461 processors) for which the code is optimized.
15464 @item -mcpu=@var{cpu-type}
15466 A deprecated synonym for @option{-mtune}.
15468 @item -mfpmath=@var{unit}
15470 Generate floating-point arithmetic for selected unit @var{unit}. The choices
15471 for @var{unit} are:
15475 Use the standard 387 floating-point coprocessor present on the majority of chips and
15476 emulated otherwise. Code compiled with this option runs almost everywhere.
15477 The temporary results are computed in 80-bit precision instead of the precision
15478 specified by the type, resulting in slightly different results compared to most
15479 of other chips. See @option{-ffloat-store} for more detailed description.
15481 This is the default choice for i386 compiler.
15484 Use scalar floating-point instructions present in the SSE instruction set.
15485 This instruction set is supported by Pentium III and newer chips,
15486 and in the AMD line
15487 by Athlon-4, Athlon XP and Athlon MP chips. The earlier version of the SSE
15488 instruction set supports only single-precision arithmetic, thus the double and
15489 extended-precision arithmetic are still done using 387. A later version, present
15490 only in Pentium 4 and AMD x86-64 chips, supports double-precision
15493 For the i386 compiler, you must use @option{-march=@var{cpu-type}}, @option{-msse}
15494 or @option{-msse2} switches to enable SSE extensions and make this option
15495 effective. For the x86-64 compiler, these extensions are enabled by default.
15497 The resulting code should be considerably faster in the majority of cases and avoid
15498 the numerical instability problems of 387 code, but may break some existing
15499 code that expects temporaries to be 80 bits.
15501 This is the default choice for the x86-64 compiler.
15506 Attempt to utilize both instruction sets at once. This effectively doubles the
15507 amount of available registers, and on chips with separate execution units for
15508 387 and SSE the execution resources too. Use this option with care, as it is
15509 still experimental, because the GCC register allocator does not model separate
15510 functional units well, resulting in unstable performance.
15513 @item -masm=@var{dialect}
15514 @opindex masm=@var{dialect}
15515 Output assembly instructions using selected @var{dialect}. Supported
15516 choices are @samp{intel} or @samp{att} (the default). Darwin does
15517 not support @samp{intel}.
15520 @itemx -mno-ieee-fp
15522 @opindex mno-ieee-fp
15523 Control whether or not the compiler uses IEEE floating-point
15524 comparisons. These correctly handle the case where the result of a
15525 comparison is unordered.
15528 @opindex msoft-float
15529 Generate output containing library calls for floating point.
15531 @strong{Warning:} the requisite libraries are not part of GCC@.
15532 Normally the facilities of the machine's usual C compiler are used, but
15533 this can't be done directly in cross-compilation. You must make your
15534 own arrangements to provide suitable library functions for
15537 On machines where a function returns floating-point results in the 80387
15538 register stack, some floating-point opcodes may be emitted even if
15539 @option{-msoft-float} is used.
15541 @item -mno-fp-ret-in-387
15542 @opindex mno-fp-ret-in-387
15543 Do not use the FPU registers for return values of functions.
15545 The usual calling convention has functions return values of types
15546 @code{float} and @code{double} in an FPU register, even if there
15547 is no FPU@. The idea is that the operating system should emulate
15550 The option @option{-mno-fp-ret-in-387} causes such values to be returned
15551 in ordinary CPU registers instead.
15553 @item -mno-fancy-math-387
15554 @opindex mno-fancy-math-387
15555 Some 387 emulators do not support the @code{sin}, @code{cos} and
15556 @code{sqrt} instructions for the 387. Specify this option to avoid
15557 generating those instructions. This option is the default on FreeBSD,
15558 OpenBSD and NetBSD@. This option is overridden when @option{-march}
15559 indicates that the target CPU always has an FPU and so the
15560 instruction does not need emulation. These
15561 instructions are not generated unless you also use the
15562 @option{-funsafe-math-optimizations} switch.
15564 @item -malign-double
15565 @itemx -mno-align-double
15566 @opindex malign-double
15567 @opindex mno-align-double
15568 Control whether GCC aligns @code{double}, @code{long double}, and
15569 @code{long long} variables on a two-word boundary or a one-word
15570 boundary. Aligning @code{double} variables on a two-word boundary
15571 produces code that runs somewhat faster on a Pentium at the
15572 expense of more memory.
15574 On x86-64, @option{-malign-double} is enabled by default.
15576 @strong{Warning:} if you use the @option{-malign-double} switch,
15577 structures containing the above types are aligned differently than
15578 the published application binary interface specifications for the 386
15579 and are not binary compatible with structures in code compiled
15580 without that switch.
15582 @item -m96bit-long-double
15583 @itemx -m128bit-long-double
15584 @opindex m96bit-long-double
15585 @opindex m128bit-long-double
15586 These switches control the size of @code{long double} type. The i386
15587 application binary interface specifies the size to be 96 bits,
15588 so @option{-m96bit-long-double} is the default in 32-bit mode.
15590 Modern architectures (Pentium and newer) prefer @code{long double}
15591 to be aligned to an 8- or 16-byte boundary. In arrays or structures
15592 conforming to the ABI, this is not possible. So specifying
15593 @option{-m128bit-long-double} aligns @code{long double}
15594 to a 16-byte boundary by padding the @code{long double} with an additional
15597 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
15598 its ABI specifies that @code{long double} is aligned on 16-byte boundary.
15600 Notice that neither of these options enable any extra precision over the x87
15601 standard of 80 bits for a @code{long double}.
15603 @strong{Warning:} if you override the default value for your target ABI, this
15604 changes the size of
15605 structures and arrays containing @code{long double} variables,
15606 as well as modifying the function calling convention for functions taking
15607 @code{long double}. Hence they are not binary-compatible
15608 with code compiled without that switch.
15610 @item -mlong-double-64
15611 @itemx -mlong-double-80
15612 @itemx -mlong-double-128
15613 @opindex mlong-double-64
15614 @opindex mlong-double-80
15615 @opindex mlong-double-128
15616 These switches control the size of @code{long double} type. A size
15617 of 64 bits makes the @code{long double} type equivalent to the @code{double}
15618 type. This is the default for 32-bit Bionic C library. A size
15619 of 128 bits makes the @code{long double} type equivalent to the
15620 @code{__float128} type. This is the default for 64-bit Bionic C library.
15622 @strong{Warning:} if you override the default value for your target ABI, this
15623 changes the size of
15624 structures and arrays containing @code{long double} variables,
15625 as well as modifying the function calling convention for functions taking
15626 @code{long double}. Hence they are not binary-compatible
15627 with code compiled without that switch.
15629 @item -mlarge-data-threshold=@var{threshold}
15630 @opindex mlarge-data-threshold
15631 When @option{-mcmodel=medium} is specified, data objects larger than
15632 @var{threshold} are placed in the large data section. This value must be the
15633 same across all objects linked into the binary, and defaults to 65535.
15637 Use a different function-calling convention, in which functions that
15638 take a fixed number of arguments return with the @code{ret @var{num}}
15639 instruction, which pops their arguments while returning. This saves one
15640 instruction in the caller since there is no need to pop the arguments
15643 You can specify that an individual function is called with this calling
15644 sequence with the function attribute @samp{stdcall}. You can also
15645 override the @option{-mrtd} option by using the function attribute
15646 @samp{cdecl}. @xref{Function Attributes}.
15648 @strong{Warning:} this calling convention is incompatible with the one
15649 normally used on Unix, so you cannot use it if you need to call
15650 libraries compiled with the Unix compiler.
15652 Also, you must provide function prototypes for all functions that
15653 take variable numbers of arguments (including @code{printf});
15654 otherwise incorrect code is generated for calls to those
15657 In addition, seriously incorrect code results if you call a
15658 function with too many arguments. (Normally, extra arguments are
15659 harmlessly ignored.)
15661 @item -mregparm=@var{num}
15663 Control how many registers are used to pass integer arguments. By
15664 default, no registers are used to pass arguments, and at most 3
15665 registers can be used. You can control this behavior for a specific
15666 function by using the function attribute @samp{regparm}.
15667 @xref{Function Attributes}.
15669 @strong{Warning:} if you use this switch, and
15670 @var{num} is nonzero, then you must build all modules with the same
15671 value, including any libraries. This includes the system libraries and
15675 @opindex msseregparm
15676 Use SSE register passing conventions for float and double arguments
15677 and return values. You can control this behavior for a specific
15678 function by using the function attribute @samp{sseregparm}.
15679 @xref{Function Attributes}.
15681 @strong{Warning:} if you use this switch then you must build all
15682 modules with the same value, including any libraries. This includes
15683 the system libraries and startup modules.
15685 @item -mvect8-ret-in-mem
15686 @opindex mvect8-ret-in-mem
15687 Return 8-byte vectors in memory instead of MMX registers. This is the
15688 default on Solaris@tie{}8 and 9 and VxWorks to match the ABI of the Sun
15689 Studio compilers until version 12. Later compiler versions (starting
15690 with Studio 12 Update@tie{}1) follow the ABI used by other x86 targets, which
15691 is the default on Solaris@tie{}10 and later. @emph{Only} use this option if
15692 you need to remain compatible with existing code produced by those
15693 previous compiler versions or older versions of GCC@.
15702 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
15703 is specified, the significands of results of floating-point operations are
15704 rounded to 24 bits (single precision); @option{-mpc64} rounds the
15705 significands of results of floating-point operations to 53 bits (double
15706 precision) and @option{-mpc80} rounds the significands of results of
15707 floating-point operations to 64 bits (extended double precision), which is
15708 the default. When this option is used, floating-point operations in higher
15709 precisions are not available to the programmer without setting the FPU
15710 control word explicitly.
15712 Setting the rounding of floating-point operations to less than the default
15713 80 bits can speed some programs by 2% or more. Note that some mathematical
15714 libraries assume that extended-precision (80-bit) floating-point operations
15715 are enabled by default; routines in such libraries could suffer significant
15716 loss of accuracy, typically through so-called ``catastrophic cancellation'',
15717 when this option is used to set the precision to less than extended precision.
15719 @item -mstackrealign
15720 @opindex mstackrealign
15721 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
15722 option generates an alternate prologue and epilogue that realigns the
15723 run-time stack if necessary. This supports mixing legacy codes that keep
15724 4-byte stack alignment with modern codes that keep 16-byte stack alignment for
15725 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
15726 applicable to individual functions.
15728 @item -mpreferred-stack-boundary=@var{num}
15729 @opindex mpreferred-stack-boundary
15730 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
15731 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
15732 the default is 4 (16 bytes or 128 bits).
15734 @strong{Warning:} When generating code for the x86-64 architecture with
15735 SSE extensions disabled, @option{-mpreferred-stack-boundary=3} can be
15736 used to keep the stack boundary aligned to 8 byte boundary. Since
15737 x86-64 ABI require 16 byte stack alignment, this is ABI incompatible and
15738 intended to be used in controlled environment where stack space is
15739 important limitation. This option will lead to wrong code when functions
15740 compiled with 16 byte stack alignment (such as functions from a standard
15741 library) are called with misaligned stack. In this case, SSE
15742 instructions may lead to misaligned memory access traps. In addition,
15743 variable arguments will be handled incorrectly for 16 byte aligned
15744 objects (including x87 long double and __int128), leading to wrong
15745 results. You must build all modules with
15746 @option{-mpreferred-stack-boundary=3}, including any libraries. This
15747 includes the system libraries and startup modules.
15749 @item -mincoming-stack-boundary=@var{num}
15750 @opindex mincoming-stack-boundary
15751 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
15752 boundary. If @option{-mincoming-stack-boundary} is not specified,
15753 the one specified by @option{-mpreferred-stack-boundary} is used.
15755 On Pentium and Pentium Pro, @code{double} and @code{long double} values
15756 should be aligned to an 8-byte boundary (see @option{-malign-double}) or
15757 suffer significant run time performance penalties. On Pentium III, the
15758 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
15759 properly if it is not 16-byte aligned.
15761 To ensure proper alignment of this values on the stack, the stack boundary
15762 must be as aligned as that required by any value stored on the stack.
15763 Further, every function must be generated such that it keeps the stack
15764 aligned. Thus calling a function compiled with a higher preferred
15765 stack boundary from a function compiled with a lower preferred stack
15766 boundary most likely misaligns the stack. It is recommended that
15767 libraries that use callbacks always use the default setting.
15769 This extra alignment does consume extra stack space, and generally
15770 increases code size. Code that is sensitive to stack space usage, such
15771 as embedded systems and operating system kernels, may want to reduce the
15772 preferred alignment to @option{-mpreferred-stack-boundary=2}.
15796 @itemx -mno-avx512f
15799 @itemx -mno-avx512pf
15801 @itemx -mno-avx512er
15803 @itemx -mno-avx512cd
15811 @itemx -mno-clflsuhopt
15814 @itemx -mno-fsgsbase
15821 @itemx -mprefetchwt1
15822 @itemx -mno-prefetchwt1
15860 These switches enable or disable the use of instructions in the MMX, SSE,
15861 SSE2, SSE3, SSSE3, SSE4.1, AVX, AVX2, AVX512F, AVX512PF, AVX512ER, AVX512CD,
15862 SHA, AES, PCLMUL, FSGSBASE, RDRND, F16C, FMA, SSE4A, FMA4, XOP, LWP, ABM,
15863 BMI, BMI2, FXSR, XSAVE, XSAVEOPT, LZCNT, RTM, or 3DNow!@:
15864 extended instruction sets.
15865 These extensions are also available as built-in functions: see
15866 @ref{X86 Built-in Functions}, for details of the functions enabled and
15867 disabled by these switches.
15869 To generate SSE/SSE2 instructions automatically from floating-point
15870 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
15872 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
15873 generates new AVX instructions or AVX equivalence for all SSEx instructions
15876 These options enable GCC to use these extended instructions in
15877 generated code, even without @option{-mfpmath=sse}. Applications that
15878 perform run-time CPU detection must compile separate files for each
15879 supported architecture, using the appropriate flags. In particular,
15880 the file containing the CPU detection code should be compiled without
15883 @item -mdump-tune-features
15884 @opindex mdump-tune-features
15885 This option instructs GCC to dump the names of the x86 performance
15886 tuning features and default settings. The names can be used in
15887 @option{-mtune-ctrl=@var{feature-list}}.
15889 @item -mtune-ctrl=@var{feature-list}
15890 @opindex mtune-ctrl=@var{feature-list}
15891 This option is used to do fine grain control of x86 code generation features.
15892 @var{feature-list} is a comma separated list of @var{feature} names. See also
15893 @option{-mdump-tune-features}. When specified, the @var{feature} will be turned
15894 on if it is not preceded with @code{^}, otherwise, it will be turned off.
15895 @option{-mtune-ctrl=@var{feature-list}} is intended to be used by GCC
15896 developers. Using it may lead to code paths not covered by testing and can
15897 potentially result in compiler ICEs or runtime errors.
15900 @opindex mno-default
15901 This option instructs GCC to turn off all tunable features. See also
15902 @option{-mtune-ctrl=@var{feature-list}} and @option{-mdump-tune-features}.
15906 This option instructs GCC to emit a @code{cld} instruction in the prologue
15907 of functions that use string instructions. String instructions depend on
15908 the DF flag to select between autoincrement or autodecrement mode. While the
15909 ABI specifies the DF flag to be cleared on function entry, some operating
15910 systems violate this specification by not clearing the DF flag in their
15911 exception dispatchers. The exception handler can be invoked with the DF flag
15912 set, which leads to wrong direction mode when string instructions are used.
15913 This option can be enabled by default on 32-bit x86 targets by configuring
15914 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
15915 instructions can be suppressed with the @option{-mno-cld} compiler option
15919 @opindex mvzeroupper
15920 This option instructs GCC to emit a @code{vzeroupper} instruction
15921 before a transfer of control flow out of the function to minimize
15922 the AVX to SSE transition penalty as well as remove unnecessary @code{zeroupper}
15925 @item -mprefer-avx128
15926 @opindex mprefer-avx128
15927 This option instructs GCC to use 128-bit AVX instructions instead of
15928 256-bit AVX instructions in the auto-vectorizer.
15932 This option enables GCC to generate @code{CMPXCHG16B} instructions.
15933 @code{CMPXCHG16B} allows for atomic operations on 128-bit double quadword
15934 (or oword) data types.
15935 This is useful for high-resolution counters that can be updated
15936 by multiple processors (or cores). This instruction is generated as part of
15937 atomic built-in functions: see @ref{__sync Builtins} or
15938 @ref{__atomic Builtins} for details.
15942 This option enables generation of @code{SAHF} instructions in 64-bit code.
15943 Early Intel Pentium 4 CPUs with Intel 64 support,
15944 prior to the introduction of Pentium 4 G1 step in December 2005,
15945 lacked the @code{LAHF} and @code{SAHF} instructions
15946 which were supported by AMD64.
15947 These are load and store instructions, respectively, for certain status flags.
15948 In 64-bit mode, the @code{SAHF} instruction is used to optimize @code{fmod},
15949 @code{drem}, and @code{remainder} built-in functions;
15950 see @ref{Other Builtins} for details.
15954 This option enables use of the @code{movbe} instruction to implement
15955 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
15959 This option enables built-in functions @code{__builtin_ia32_crc32qi},
15960 @code{__builtin_ia32_crc32hi}, @code{__builtin_ia32_crc32si} and
15961 @code{__builtin_ia32_crc32di} to generate the @code{crc32} machine instruction.
15965 This option enables use of @code{RCPSS} and @code{RSQRTSS} instructions
15966 (and their vectorized variants @code{RCPPS} and @code{RSQRTPS})
15967 with an additional Newton-Raphson step
15968 to increase precision instead of @code{DIVSS} and @code{SQRTSS}
15969 (and their vectorized
15970 variants) for single-precision floating-point arguments. These instructions
15971 are generated only when @option{-funsafe-math-optimizations} is enabled
15972 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
15973 Note that while the throughput of the sequence is higher than the throughput
15974 of the non-reciprocal instruction, the precision of the sequence can be
15975 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
15977 Note that GCC implements @code{1.0f/sqrtf(@var{x})} in terms of @code{RSQRTSS}
15978 (or @code{RSQRTPS}) already with @option{-ffast-math} (or the above option
15979 combination), and doesn't need @option{-mrecip}.
15981 Also note that GCC emits the above sequence with additional Newton-Raphson step
15982 for vectorized single-float division and vectorized @code{sqrtf(@var{x})}
15983 already with @option{-ffast-math} (or the above option combination), and
15984 doesn't need @option{-mrecip}.
15986 @item -mrecip=@var{opt}
15987 @opindex mrecip=opt
15988 This option controls which reciprocal estimate instructions
15989 may be used. @var{opt} is a comma-separated list of options, which may
15990 be preceded by a @samp{!} to invert the option:
15994 Enable all estimate instructions.
15997 Enable the default instructions, equivalent to @option{-mrecip}.
16000 Disable all estimate instructions, equivalent to @option{-mno-recip}.
16003 Enable the approximation for scalar division.
16006 Enable the approximation for vectorized division.
16009 Enable the approximation for scalar square root.
16012 Enable the approximation for vectorized square root.
16015 So, for example, @option{-mrecip=all,!sqrt} enables
16016 all of the reciprocal approximations, except for square root.
16018 @item -mveclibabi=@var{type}
16019 @opindex mveclibabi
16020 Specifies the ABI type to use for vectorizing intrinsics using an
16021 external library. Supported values for @var{type} are @samp{svml}
16022 for the Intel short
16023 vector math library and @samp{acml} for the AMD math core library.
16024 To use this option, both @option{-ftree-vectorize} and
16025 @option{-funsafe-math-optimizations} have to be enabled, and an SVML or ACML
16026 ABI-compatible library must be specified at link time.
16028 GCC currently emits calls to @code{vmldExp2},
16029 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
16030 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
16031 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
16032 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
16033 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
16034 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
16035 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
16036 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
16037 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
16038 function type when @option{-mveclibabi=svml} is used, and @code{__vrd2_sin},
16039 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
16040 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
16041 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
16042 @code{__vrs4_log10f} and @code{__vrs4_powf} for the corresponding function type
16043 when @option{-mveclibabi=acml} is used.
16045 @item -mabi=@var{name}
16047 Generate code for the specified calling convention. Permissible values
16048 are @samp{sysv} for the ABI used on GNU/Linux and other systems, and
16049 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
16050 ABI when targeting Microsoft Windows and the SysV ABI on all other systems.
16051 You can control this behavior for a specific function by
16052 using the function attribute @samp{ms_abi}/@samp{sysv_abi}.
16053 @xref{Function Attributes}.
16055 @item -mtls-dialect=@var{type}
16056 @opindex mtls-dialect
16057 Generate code to access thread-local storage using the @samp{gnu} or
16058 @samp{gnu2} conventions. @samp{gnu} is the conservative default;
16059 @samp{gnu2} is more efficient, but it may add compile- and run-time
16060 requirements that cannot be satisfied on all systems.
16063 @itemx -mno-push-args
16064 @opindex mpush-args
16065 @opindex mno-push-args
16066 Use PUSH operations to store outgoing parameters. This method is shorter
16067 and usually equally fast as method using SUB/MOV operations and is enabled
16068 by default. In some cases disabling it may improve performance because of
16069 improved scheduling and reduced dependencies.
16071 @item -maccumulate-outgoing-args
16072 @opindex maccumulate-outgoing-args
16073 If enabled, the maximum amount of space required for outgoing arguments is
16074 computed in the function prologue. This is faster on most modern CPUs
16075 because of reduced dependencies, improved scheduling and reduced stack usage
16076 when the preferred stack boundary is not equal to 2. The drawback is a notable
16077 increase in code size. This switch implies @option{-mno-push-args}.
16081 Support thread-safe exception handling on MinGW. Programs that rely
16082 on thread-safe exception handling must compile and link all code with the
16083 @option{-mthreads} option. When compiling, @option{-mthreads} defines
16084 @code{-D_MT}; when linking, it links in a special thread helper library
16085 @option{-lmingwthrd} which cleans up per-thread exception-handling data.
16087 @item -mno-align-stringops
16088 @opindex mno-align-stringops
16089 Do not align the destination of inlined string operations. This switch reduces
16090 code size and improves performance in case the destination is already aligned,
16091 but GCC doesn't know about it.
16093 @item -minline-all-stringops
16094 @opindex minline-all-stringops
16095 By default GCC inlines string operations only when the destination is
16096 known to be aligned to least a 4-byte boundary.
16097 This enables more inlining and increases code
16098 size, but may improve performance of code that depends on fast
16099 @code{memcpy}, @code{strlen},
16100 and @code{memset} for short lengths.
16102 @item -minline-stringops-dynamically
16103 @opindex minline-stringops-dynamically
16104 For string operations of unknown size, use run-time checks with
16105 inline code for small blocks and a library call for large blocks.
16107 @item -mstringop-strategy=@var{alg}
16108 @opindex mstringop-strategy=@var{alg}
16109 Override the internal decision heuristic for the particular algorithm to use
16110 for inlining string operations. The allowed values for @var{alg} are:
16116 Expand using i386 @code{rep} prefix of the specified size.
16120 @itemx unrolled_loop
16121 Expand into an inline loop.
16124 Always use a library call.
16127 @item -mmemcpy-strategy=@var{strategy}
16128 @opindex mmemcpy-strategy=@var{strategy}
16129 Override the internal decision heuristic to decide if @code{__builtin_memcpy}
16130 should be inlined and what inline algorithm to use when the expected size
16131 of the copy operation is known. @var{strategy}
16132 is a comma-separated list of @var{alg}:@var{max_size}:@var{dest_align} triplets.
16133 @var{alg} is specified in @option{-mstringop-strategy}, @var{max_size} specifies
16134 the max byte size with which inline algorithm @var{alg} is allowed. For the last
16135 triplet, the @var{max_size} must be @code{-1}. The @var{max_size} of the triplets
16136 in the list must be specified in increasing order. The minimal byte size for
16137 @var{alg} is @code{0} for the first triplet and @code{@var{max_size} + 1} of the
16140 @item -mmemset-strategy=@var{strategy}
16141 @opindex mmemset-strategy=@var{strategy}
16142 The option is similar to @option{-mmemcpy-strategy=} except that it is to control
16143 @code{__builtin_memset} expansion.
16145 @item -momit-leaf-frame-pointer
16146 @opindex momit-leaf-frame-pointer
16147 Don't keep the frame pointer in a register for leaf functions. This
16148 avoids the instructions to save, set up, and restore frame pointers and
16149 makes an extra register available in leaf functions. The option
16150 @option{-fomit-leaf-frame-pointer} removes the frame pointer for leaf functions,
16151 which might make debugging harder.
16153 @item -mtls-direct-seg-refs
16154 @itemx -mno-tls-direct-seg-refs
16155 @opindex mtls-direct-seg-refs
16156 Controls whether TLS variables may be accessed with offsets from the
16157 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
16158 or whether the thread base pointer must be added. Whether or not this
16159 is valid depends on the operating system, and whether it maps the
16160 segment to cover the entire TLS area.
16162 For systems that use the GNU C Library, the default is on.
16165 @itemx -mno-sse2avx
16167 Specify that the assembler should encode SSE instructions with VEX
16168 prefix. The option @option{-mavx} turns this on by default.
16173 If profiling is active (@option{-pg}), put the profiling
16174 counter call before the prologue.
16175 Note: On x86 architectures the attribute @code{ms_hook_prologue}
16176 isn't possible at the moment for @option{-mfentry} and @option{-pg}.
16179 @itemx -mno-8bit-idiv
16181 On some processors, like Intel Atom, 8-bit unsigned integer divide is
16182 much faster than 32-bit/64-bit integer divide. This option generates a
16183 run-time check. If both dividend and divisor are within range of 0
16184 to 255, 8-bit unsigned integer divide is used instead of
16185 32-bit/64-bit integer divide.
16187 @item -mavx256-split-unaligned-load
16188 @itemx -mavx256-split-unaligned-store
16189 @opindex avx256-split-unaligned-load
16190 @opindex avx256-split-unaligned-store
16191 Split 32-byte AVX unaligned load and store.
16193 @item -mstack-protector-guard=@var{guard}
16194 @opindex mstack-protector-guard=@var{guard}
16195 Generate stack protection code using canary at @var{guard}. Supported
16196 locations are @samp{global} for global canary or @samp{tls} for per-thread
16197 canary in the TLS block (the default). This option has effect only when
16198 @option{-fstack-protector} or @option{-fstack-protector-all} is specified.
16202 These @samp{-m} switches are supported in addition to the above
16203 on x86-64 processors in 64-bit environments.
16214 Generate code for a 16-bit, 32-bit or 64-bit environment.
16215 The @option{-m32} option sets @code{int}, @code{long}, and pointer types
16217 generates code that runs on any i386 system.
16219 The @option{-m64} option sets @code{int} to 32 bits and @code{long} and pointer
16220 types to 64 bits, and generates code for the x86-64 architecture.
16221 For Darwin only the @option{-m64} option also turns off the @option{-fno-pic}
16222 and @option{-mdynamic-no-pic} options.
16224 The @option{-mx32} option sets @code{int}, @code{long}, and pointer types
16226 generates code for the x86-64 architecture.
16228 The @option{-m16} option is the same as @option{-m32}, except for that
16229 it outputs the @code{.code16gcc} assembly directive at the beginning of
16230 the assembly output so that the binary can run in 16-bit mode.
16232 @item -mno-red-zone
16233 @opindex mno-red-zone
16234 Do not use a so-called ``red zone'' for x86-64 code. The red zone is mandated
16235 by the x86-64 ABI; it is a 128-byte area beyond the location of the
16236 stack pointer that is not modified by signal or interrupt handlers
16237 and therefore can be used for temporary data without adjusting the stack
16238 pointer. The flag @option{-mno-red-zone} disables this red zone.
16240 @item -mcmodel=small
16241 @opindex mcmodel=small
16242 Generate code for the small code model: the program and its symbols must
16243 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
16244 Programs can be statically or dynamically linked. This is the default
16247 @item -mcmodel=kernel
16248 @opindex mcmodel=kernel
16249 Generate code for the kernel code model. The kernel runs in the
16250 negative 2 GB of the address space.
16251 This model has to be used for Linux kernel code.
16253 @item -mcmodel=medium
16254 @opindex mcmodel=medium
16255 Generate code for the medium model: the program is linked in the lower 2
16256 GB of the address space. Small symbols are also placed there. Symbols
16257 with sizes larger than @option{-mlarge-data-threshold} are put into
16258 large data or BSS sections and can be located above 2GB. Programs can
16259 be statically or dynamically linked.
16261 @item -mcmodel=large
16262 @opindex mcmodel=large
16263 Generate code for the large model. This model makes no assumptions
16264 about addresses and sizes of sections.
16266 @item -maddress-mode=long
16267 @opindex maddress-mode=long
16268 Generate code for long address mode. This is only supported for 64-bit
16269 and x32 environments. It is the default address mode for 64-bit
16272 @item -maddress-mode=short
16273 @opindex maddress-mode=short
16274 Generate code for short address mode. This is only supported for 32-bit
16275 and x32 environments. It is the default address mode for 32-bit and
16279 @node i386 and x86-64 Windows Options
16280 @subsection i386 and x86-64 Windows Options
16281 @cindex i386 and x86-64 Windows Options
16283 These additional options are available for Microsoft Windows targets:
16289 specifies that a console application is to be generated, by
16290 instructing the linker to set the PE header subsystem type
16291 required for console applications.
16292 This option is available for Cygwin and MinGW targets and is
16293 enabled by default on those targets.
16297 This option is available for Cygwin and MinGW targets. It
16298 specifies that a DLL---a dynamic link library---is to be
16299 generated, enabling the selection of the required runtime
16300 startup object and entry point.
16302 @item -mnop-fun-dllimport
16303 @opindex mnop-fun-dllimport
16304 This option is available for Cygwin and MinGW targets. It
16305 specifies that the @code{dllimport} attribute should be ignored.
16309 This option is available for MinGW targets. It specifies
16310 that MinGW-specific thread support is to be used.
16314 This option is available for MinGW-w64 targets. It causes
16315 the @code{UNICODE} preprocessor macro to be predefined, and
16316 chooses Unicode-capable runtime startup code.
16320 This option is available for Cygwin and MinGW targets. It
16321 specifies that the typical Microsoft Windows predefined macros are to
16322 be set in the pre-processor, but does not influence the choice
16323 of runtime library/startup code.
16327 This option is available for Cygwin and MinGW targets. It
16328 specifies that a GUI application is to be generated by
16329 instructing the linker to set the PE header subsystem type
16332 @item -fno-set-stack-executable
16333 @opindex fno-set-stack-executable
16334 This option is available for MinGW targets. It specifies that
16335 the executable flag for the stack used by nested functions isn't
16336 set. This is necessary for binaries running in kernel mode of
16337 Microsoft Windows, as there the User32 API, which is used to set executable
16338 privileges, isn't available.
16340 @item -fwritable-relocated-rdata
16341 @opindex fno-writable-relocated-rdata
16342 This option is available for MinGW and Cygwin targets. It specifies
16343 that relocated-data in read-only section is put into .data
16344 section. This is a necessary for older runtimes not supporting
16345 modification of .rdata sections for pseudo-relocation.
16347 @item -mpe-aligned-commons
16348 @opindex mpe-aligned-commons
16349 This option is available for Cygwin and MinGW targets. It
16350 specifies that the GNU extension to the PE file format that
16351 permits the correct alignment of COMMON variables should be
16352 used when generating code. It is enabled by default if
16353 GCC detects that the target assembler found during configuration
16354 supports the feature.
16357 See also under @ref{i386 and x86-64 Options} for standard options.
16359 @node IA-64 Options
16360 @subsection IA-64 Options
16361 @cindex IA-64 Options
16363 These are the @samp{-m} options defined for the Intel IA-64 architecture.
16367 @opindex mbig-endian
16368 Generate code for a big-endian target. This is the default for HP-UX@.
16370 @item -mlittle-endian
16371 @opindex mlittle-endian
16372 Generate code for a little-endian target. This is the default for AIX5
16378 @opindex mno-gnu-as
16379 Generate (or don't) code for the GNU assembler. This is the default.
16380 @c Also, this is the default if the configure option @option{--with-gnu-as}
16386 @opindex mno-gnu-ld
16387 Generate (or don't) code for the GNU linker. This is the default.
16388 @c Also, this is the default if the configure option @option{--with-gnu-ld}
16393 Generate code that does not use a global pointer register. The result
16394 is not position independent code, and violates the IA-64 ABI@.
16396 @item -mvolatile-asm-stop
16397 @itemx -mno-volatile-asm-stop
16398 @opindex mvolatile-asm-stop
16399 @opindex mno-volatile-asm-stop
16400 Generate (or don't) a stop bit immediately before and after volatile asm
16403 @item -mregister-names
16404 @itemx -mno-register-names
16405 @opindex mregister-names
16406 @opindex mno-register-names
16407 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
16408 the stacked registers. This may make assembler output more readable.
16414 Disable (or enable) optimizations that use the small data section. This may
16415 be useful for working around optimizer bugs.
16417 @item -mconstant-gp
16418 @opindex mconstant-gp
16419 Generate code that uses a single constant global pointer value. This is
16420 useful when compiling kernel code.
16424 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
16425 This is useful when compiling firmware code.
16427 @item -minline-float-divide-min-latency
16428 @opindex minline-float-divide-min-latency
16429 Generate code for inline divides of floating-point values
16430 using the minimum latency algorithm.
16432 @item -minline-float-divide-max-throughput
16433 @opindex minline-float-divide-max-throughput
16434 Generate code for inline divides of floating-point values
16435 using the maximum throughput algorithm.
16437 @item -mno-inline-float-divide
16438 @opindex mno-inline-float-divide
16439 Do not generate inline code for divides of floating-point values.
16441 @item -minline-int-divide-min-latency
16442 @opindex minline-int-divide-min-latency
16443 Generate code for inline divides of integer values
16444 using the minimum latency algorithm.
16446 @item -minline-int-divide-max-throughput
16447 @opindex minline-int-divide-max-throughput
16448 Generate code for inline divides of integer values
16449 using the maximum throughput algorithm.
16451 @item -mno-inline-int-divide
16452 @opindex mno-inline-int-divide
16453 Do not generate inline code for divides of integer values.
16455 @item -minline-sqrt-min-latency
16456 @opindex minline-sqrt-min-latency
16457 Generate code for inline square roots
16458 using the minimum latency algorithm.
16460 @item -minline-sqrt-max-throughput
16461 @opindex minline-sqrt-max-throughput
16462 Generate code for inline square roots
16463 using the maximum throughput algorithm.
16465 @item -mno-inline-sqrt
16466 @opindex mno-inline-sqrt
16467 Do not generate inline code for @code{sqrt}.
16470 @itemx -mno-fused-madd
16471 @opindex mfused-madd
16472 @opindex mno-fused-madd
16473 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
16474 instructions. The default is to use these instructions.
16476 @item -mno-dwarf2-asm
16477 @itemx -mdwarf2-asm
16478 @opindex mno-dwarf2-asm
16479 @opindex mdwarf2-asm
16480 Don't (or do) generate assembler code for the DWARF 2 line number debugging
16481 info. This may be useful when not using the GNU assembler.
16483 @item -mearly-stop-bits
16484 @itemx -mno-early-stop-bits
16485 @opindex mearly-stop-bits
16486 @opindex mno-early-stop-bits
16487 Allow stop bits to be placed earlier than immediately preceding the
16488 instruction that triggered the stop bit. This can improve instruction
16489 scheduling, but does not always do so.
16491 @item -mfixed-range=@var{register-range}
16492 @opindex mfixed-range
16493 Generate code treating the given register range as fixed registers.
16494 A fixed register is one that the register allocator cannot use. This is
16495 useful when compiling kernel code. A register range is specified as
16496 two registers separated by a dash. Multiple register ranges can be
16497 specified separated by a comma.
16499 @item -mtls-size=@var{tls-size}
16501 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
16504 @item -mtune=@var{cpu-type}
16506 Tune the instruction scheduling for a particular CPU, Valid values are
16507 @samp{itanium}, @samp{itanium1}, @samp{merced}, @samp{itanium2},
16508 and @samp{mckinley}.
16514 Generate code for a 32-bit or 64-bit environment.
16515 The 32-bit environment sets int, long and pointer to 32 bits.
16516 The 64-bit environment sets int to 32 bits and long and pointer
16517 to 64 bits. These are HP-UX specific flags.
16519 @item -mno-sched-br-data-spec
16520 @itemx -msched-br-data-spec
16521 @opindex mno-sched-br-data-spec
16522 @opindex msched-br-data-spec
16523 (Dis/En)able data speculative scheduling before reload.
16524 This results in generation of @code{ld.a} instructions and
16525 the corresponding check instructions (@code{ld.c} / @code{chk.a}).
16526 The default is 'disable'.
16528 @item -msched-ar-data-spec
16529 @itemx -mno-sched-ar-data-spec
16530 @opindex msched-ar-data-spec
16531 @opindex mno-sched-ar-data-spec
16532 (En/Dis)able data speculative scheduling after reload.
16533 This results in generation of @code{ld.a} instructions and
16534 the corresponding check instructions (@code{ld.c} / @code{chk.a}).
16535 The default is 'enable'.
16537 @item -mno-sched-control-spec
16538 @itemx -msched-control-spec
16539 @opindex mno-sched-control-spec
16540 @opindex msched-control-spec
16541 (Dis/En)able control speculative scheduling. This feature is
16542 available only during region scheduling (i.e.@: before reload).
16543 This results in generation of the @code{ld.s} instructions and
16544 the corresponding check instructions @code{chk.s}.
16545 The default is 'disable'.
16547 @item -msched-br-in-data-spec
16548 @itemx -mno-sched-br-in-data-spec
16549 @opindex msched-br-in-data-spec
16550 @opindex mno-sched-br-in-data-spec
16551 (En/Dis)able speculative scheduling of the instructions that
16552 are dependent on the data speculative loads before reload.
16553 This is effective only with @option{-msched-br-data-spec} enabled.
16554 The default is 'enable'.
16556 @item -msched-ar-in-data-spec
16557 @itemx -mno-sched-ar-in-data-spec
16558 @opindex msched-ar-in-data-spec
16559 @opindex mno-sched-ar-in-data-spec
16560 (En/Dis)able speculative scheduling of the instructions that
16561 are dependent on the data speculative loads after reload.
16562 This is effective only with @option{-msched-ar-data-spec} enabled.
16563 The default is 'enable'.
16565 @item -msched-in-control-spec
16566 @itemx -mno-sched-in-control-spec
16567 @opindex msched-in-control-spec
16568 @opindex mno-sched-in-control-spec
16569 (En/Dis)able speculative scheduling of the instructions that
16570 are dependent on the control speculative loads.
16571 This is effective only with @option{-msched-control-spec} enabled.
16572 The default is 'enable'.
16574 @item -mno-sched-prefer-non-data-spec-insns
16575 @itemx -msched-prefer-non-data-spec-insns
16576 @opindex mno-sched-prefer-non-data-spec-insns
16577 @opindex msched-prefer-non-data-spec-insns
16578 If enabled, data-speculative instructions are chosen for schedule
16579 only if there are no other choices at the moment. This makes
16580 the use of the data speculation much more conservative.
16581 The default is 'disable'.
16583 @item -mno-sched-prefer-non-control-spec-insns
16584 @itemx -msched-prefer-non-control-spec-insns
16585 @opindex mno-sched-prefer-non-control-spec-insns
16586 @opindex msched-prefer-non-control-spec-insns
16587 If enabled, control-speculative instructions are chosen for schedule
16588 only if there are no other choices at the moment. This makes
16589 the use of the control speculation much more conservative.
16590 The default is 'disable'.
16592 @item -mno-sched-count-spec-in-critical-path
16593 @itemx -msched-count-spec-in-critical-path
16594 @opindex mno-sched-count-spec-in-critical-path
16595 @opindex msched-count-spec-in-critical-path
16596 If enabled, speculative dependencies are considered during
16597 computation of the instructions priorities. This makes the use of the
16598 speculation a bit more conservative.
16599 The default is 'disable'.
16601 @item -msched-spec-ldc
16602 @opindex msched-spec-ldc
16603 Use a simple data speculation check. This option is on by default.
16605 @item -msched-control-spec-ldc
16606 @opindex msched-spec-ldc
16607 Use a simple check for control speculation. This option is on by default.
16609 @item -msched-stop-bits-after-every-cycle
16610 @opindex msched-stop-bits-after-every-cycle
16611 Place a stop bit after every cycle when scheduling. This option is on
16614 @item -msched-fp-mem-deps-zero-cost
16615 @opindex msched-fp-mem-deps-zero-cost
16616 Assume that floating-point stores and loads are not likely to cause a conflict
16617 when placed into the same instruction group. This option is disabled by
16620 @item -msel-sched-dont-check-control-spec
16621 @opindex msel-sched-dont-check-control-spec
16622 Generate checks for control speculation in selective scheduling.
16623 This flag is disabled by default.
16625 @item -msched-max-memory-insns=@var{max-insns}
16626 @opindex msched-max-memory-insns
16627 Limit on the number of memory insns per instruction group, giving lower
16628 priority to subsequent memory insns attempting to schedule in the same
16629 instruction group. Frequently useful to prevent cache bank conflicts.
16630 The default value is 1.
16632 @item -msched-max-memory-insns-hard-limit
16633 @opindex msched-max-memory-insns-hard-limit
16634 Makes the limit specified by @option{msched-max-memory-insns} a hard limit,
16635 disallowing more than that number in an instruction group.
16636 Otherwise, the limit is ``soft'', meaning that non-memory operations
16637 are preferred when the limit is reached, but memory operations may still
16643 @subsection LM32 Options
16644 @cindex LM32 options
16646 These @option{-m} options are defined for the LatticeMico32 architecture:
16649 @item -mbarrel-shift-enabled
16650 @opindex mbarrel-shift-enabled
16651 Enable barrel-shift instructions.
16653 @item -mdivide-enabled
16654 @opindex mdivide-enabled
16655 Enable divide and modulus instructions.
16657 @item -mmultiply-enabled
16658 @opindex multiply-enabled
16659 Enable multiply instructions.
16661 @item -msign-extend-enabled
16662 @opindex msign-extend-enabled
16663 Enable sign extend instructions.
16665 @item -muser-enabled
16666 @opindex muser-enabled
16667 Enable user-defined instructions.
16672 @subsection M32C Options
16673 @cindex M32C options
16676 @item -mcpu=@var{name}
16678 Select the CPU for which code is generated. @var{name} may be one of
16679 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
16680 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
16681 the M32C/80 series.
16685 Specifies that the program will be run on the simulator. This causes
16686 an alternate runtime library to be linked in which supports, for
16687 example, file I/O@. You must not use this option when generating
16688 programs that will run on real hardware; you must provide your own
16689 runtime library for whatever I/O functions are needed.
16691 @item -memregs=@var{number}
16693 Specifies the number of memory-based pseudo-registers GCC uses
16694 during code generation. These pseudo-registers are used like real
16695 registers, so there is a tradeoff between GCC's ability to fit the
16696 code into available registers, and the performance penalty of using
16697 memory instead of registers. Note that all modules in a program must
16698 be compiled with the same value for this option. Because of that, you
16699 must not use this option with GCC's default runtime libraries.
16703 @node M32R/D Options
16704 @subsection M32R/D Options
16705 @cindex M32R/D options
16707 These @option{-m} options are defined for Renesas M32R/D architectures:
16712 Generate code for the M32R/2@.
16716 Generate code for the M32R/X@.
16720 Generate code for the M32R@. This is the default.
16722 @item -mmodel=small
16723 @opindex mmodel=small
16724 Assume all objects live in the lower 16MB of memory (so that their addresses
16725 can be loaded with the @code{ld24} instruction), and assume all subroutines
16726 are reachable with the @code{bl} instruction.
16727 This is the default.
16729 The addressability of a particular object can be set with the
16730 @code{model} attribute.
16732 @item -mmodel=medium
16733 @opindex mmodel=medium
16734 Assume objects may be anywhere in the 32-bit address space (the compiler
16735 generates @code{seth/add3} instructions to load their addresses), and
16736 assume all subroutines are reachable with the @code{bl} instruction.
16738 @item -mmodel=large
16739 @opindex mmodel=large
16740 Assume objects may be anywhere in the 32-bit address space (the compiler
16741 generates @code{seth/add3} instructions to load their addresses), and
16742 assume subroutines may not be reachable with the @code{bl} instruction
16743 (the compiler generates the much slower @code{seth/add3/jl}
16744 instruction sequence).
16747 @opindex msdata=none
16748 Disable use of the small data area. Variables are put into
16749 one of @samp{.data}, @samp{.bss}, or @samp{.rodata} (unless the
16750 @code{section} attribute has been specified).
16751 This is the default.
16753 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
16754 Objects may be explicitly put in the small data area with the
16755 @code{section} attribute using one of these sections.
16757 @item -msdata=sdata
16758 @opindex msdata=sdata
16759 Put small global and static data in the small data area, but do not
16760 generate special code to reference them.
16763 @opindex msdata=use
16764 Put small global and static data in the small data area, and generate
16765 special instructions to reference them.
16769 @cindex smaller data references
16770 Put global and static objects less than or equal to @var{num} bytes
16771 into the small data or BSS sections instead of the normal data or BSS
16772 sections. The default value of @var{num} is 8.
16773 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
16774 for this option to have any effect.
16776 All modules should be compiled with the same @option{-G @var{num}} value.
16777 Compiling with different values of @var{num} may or may not work; if it
16778 doesn't the linker gives an error message---incorrect code is not
16783 Makes the M32R-specific code in the compiler display some statistics
16784 that might help in debugging programs.
16786 @item -malign-loops
16787 @opindex malign-loops
16788 Align all loops to a 32-byte boundary.
16790 @item -mno-align-loops
16791 @opindex mno-align-loops
16792 Do not enforce a 32-byte alignment for loops. This is the default.
16794 @item -missue-rate=@var{number}
16795 @opindex missue-rate=@var{number}
16796 Issue @var{number} instructions per cycle. @var{number} can only be 1
16799 @item -mbranch-cost=@var{number}
16800 @opindex mbranch-cost=@var{number}
16801 @var{number} can only be 1 or 2. If it is 1 then branches are
16802 preferred over conditional code, if it is 2, then the opposite applies.
16804 @item -mflush-trap=@var{number}
16805 @opindex mflush-trap=@var{number}
16806 Specifies the trap number to use to flush the cache. The default is
16807 12. Valid numbers are between 0 and 15 inclusive.
16809 @item -mno-flush-trap
16810 @opindex mno-flush-trap
16811 Specifies that the cache cannot be flushed by using a trap.
16813 @item -mflush-func=@var{name}
16814 @opindex mflush-func=@var{name}
16815 Specifies the name of the operating system function to call to flush
16816 the cache. The default is @emph{_flush_cache}, but a function call
16817 is only used if a trap is not available.
16819 @item -mno-flush-func
16820 @opindex mno-flush-func
16821 Indicates that there is no OS function for flushing the cache.
16825 @node M680x0 Options
16826 @subsection M680x0 Options
16827 @cindex M680x0 options
16829 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
16830 The default settings depend on which architecture was selected when
16831 the compiler was configured; the defaults for the most common choices
16835 @item -march=@var{arch}
16837 Generate code for a specific M680x0 or ColdFire instruction set
16838 architecture. Permissible values of @var{arch} for M680x0
16839 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
16840 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
16841 architectures are selected according to Freescale's ISA classification
16842 and the permissible values are: @samp{isaa}, @samp{isaaplus},
16843 @samp{isab} and @samp{isac}.
16845 GCC defines a macro @samp{__mcf@var{arch}__} whenever it is generating
16846 code for a ColdFire target. The @var{arch} in this macro is one of the
16847 @option{-march} arguments given above.
16849 When used together, @option{-march} and @option{-mtune} select code
16850 that runs on a family of similar processors but that is optimized
16851 for a particular microarchitecture.
16853 @item -mcpu=@var{cpu}
16855 Generate code for a specific M680x0 or ColdFire processor.
16856 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
16857 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
16858 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
16859 below, which also classifies the CPUs into families:
16861 @multitable @columnfractions 0.20 0.80
16862 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
16863 @item @samp{51} @tab @samp{51} @samp{51ac} @samp{51ag} @samp{51cn} @samp{51em} @samp{51je} @samp{51jf} @samp{51jg} @samp{51jm} @samp{51mm} @samp{51qe} @samp{51qm}
16864 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
16865 @item @samp{5206e} @tab @samp{5206e}
16866 @item @samp{5208} @tab @samp{5207} @samp{5208}
16867 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
16868 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
16869 @item @samp{5216} @tab @samp{5214} @samp{5216}
16870 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
16871 @item @samp{5225} @tab @samp{5224} @samp{5225}
16872 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
16873 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
16874 @item @samp{5249} @tab @samp{5249}
16875 @item @samp{5250} @tab @samp{5250}
16876 @item @samp{5271} @tab @samp{5270} @samp{5271}
16877 @item @samp{5272} @tab @samp{5272}
16878 @item @samp{5275} @tab @samp{5274} @samp{5275}
16879 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
16880 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
16881 @item @samp{5307} @tab @samp{5307}
16882 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
16883 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
16884 @item @samp{5407} @tab @samp{5407}
16885 @item @samp{5475} @tab @samp{5470} @samp{5471} @samp{5472} @samp{5473} @samp{5474} @samp{5475} @samp{547x} @samp{5480} @samp{5481} @samp{5482} @samp{5483} @samp{5484} @samp{5485}
16888 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
16889 @var{arch} is compatible with @var{cpu}. Other combinations of
16890 @option{-mcpu} and @option{-march} are rejected.
16892 GCC defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
16893 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
16894 where the value of @var{family} is given by the table above.
16896 @item -mtune=@var{tune}
16898 Tune the code for a particular microarchitecture within the
16899 constraints set by @option{-march} and @option{-mcpu}.
16900 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
16901 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
16902 and @samp{cpu32}. The ColdFire microarchitectures
16903 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
16905 You can also use @option{-mtune=68020-40} for code that needs
16906 to run relatively well on 68020, 68030 and 68040 targets.
16907 @option{-mtune=68020-60} is similar but includes 68060 targets
16908 as well. These two options select the same tuning decisions as
16909 @option{-m68020-40} and @option{-m68020-60} respectively.
16911 GCC defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
16912 when tuning for 680x0 architecture @var{arch}. It also defines
16913 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
16914 option is used. If GCC is tuning for a range of architectures,
16915 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
16916 it defines the macros for every architecture in the range.
16918 GCC also defines the macro @samp{__m@var{uarch}__} when tuning for
16919 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
16920 of the arguments given above.
16926 Generate output for a 68000. This is the default
16927 when the compiler is configured for 68000-based systems.
16928 It is equivalent to @option{-march=68000}.
16930 Use this option for microcontrollers with a 68000 or EC000 core,
16931 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
16935 Generate output for a 68010. This is the default
16936 when the compiler is configured for 68010-based systems.
16937 It is equivalent to @option{-march=68010}.
16943 Generate output for a 68020. This is the default
16944 when the compiler is configured for 68020-based systems.
16945 It is equivalent to @option{-march=68020}.
16949 Generate output for a 68030. This is the default when the compiler is
16950 configured for 68030-based systems. It is equivalent to
16951 @option{-march=68030}.
16955 Generate output for a 68040. This is the default when the compiler is
16956 configured for 68040-based systems. It is equivalent to
16957 @option{-march=68040}.
16959 This option inhibits the use of 68881/68882 instructions that have to be
16960 emulated by software on the 68040. Use this option if your 68040 does not
16961 have code to emulate those instructions.
16965 Generate output for a 68060. This is the default when the compiler is
16966 configured for 68060-based systems. It is equivalent to
16967 @option{-march=68060}.
16969 This option inhibits the use of 68020 and 68881/68882 instructions that
16970 have to be emulated by software on the 68060. Use this option if your 68060
16971 does not have code to emulate those instructions.
16975 Generate output for a CPU32. This is the default
16976 when the compiler is configured for CPU32-based systems.
16977 It is equivalent to @option{-march=cpu32}.
16979 Use this option for microcontrollers with a
16980 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
16981 68336, 68340, 68341, 68349 and 68360.
16985 Generate output for a 520X ColdFire CPU@. This is the default
16986 when the compiler is configured for 520X-based systems.
16987 It is equivalent to @option{-mcpu=5206}, and is now deprecated
16988 in favor of that option.
16990 Use this option for microcontroller with a 5200 core, including
16991 the MCF5202, MCF5203, MCF5204 and MCF5206.
16995 Generate output for a 5206e ColdFire CPU@. The option is now
16996 deprecated in favor of the equivalent @option{-mcpu=5206e}.
17000 Generate output for a member of the ColdFire 528X family.
17001 The option is now deprecated in favor of the equivalent
17002 @option{-mcpu=528x}.
17006 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
17007 in favor of the equivalent @option{-mcpu=5307}.
17011 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
17012 in favor of the equivalent @option{-mcpu=5407}.
17016 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
17017 This includes use of hardware floating-point instructions.
17018 The option is equivalent to @option{-mcpu=547x}, and is now
17019 deprecated in favor of that option.
17023 Generate output for a 68040, without using any of the new instructions.
17024 This results in code that can run relatively efficiently on either a
17025 68020/68881 or a 68030 or a 68040. The generated code does use the
17026 68881 instructions that are emulated on the 68040.
17028 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
17032 Generate output for a 68060, without using any of the new instructions.
17033 This results in code that can run relatively efficiently on either a
17034 68020/68881 or a 68030 or a 68040. The generated code does use the
17035 68881 instructions that are emulated on the 68060.
17037 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
17041 @opindex mhard-float
17043 Generate floating-point instructions. This is the default for 68020
17044 and above, and for ColdFire devices that have an FPU@. It defines the
17045 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
17046 on ColdFire targets.
17049 @opindex msoft-float
17050 Do not generate floating-point instructions; use library calls instead.
17051 This is the default for 68000, 68010, and 68832 targets. It is also
17052 the default for ColdFire devices that have no FPU.
17058 Generate (do not generate) ColdFire hardware divide and remainder
17059 instructions. If @option{-march} is used without @option{-mcpu},
17060 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
17061 architectures. Otherwise, the default is taken from the target CPU
17062 (either the default CPU, or the one specified by @option{-mcpu}). For
17063 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
17064 @option{-mcpu=5206e}.
17066 GCC defines the macro @samp{__mcfhwdiv__} when this option is enabled.
17070 Consider type @code{int} to be 16 bits wide, like @code{short int}.
17071 Additionally, parameters passed on the stack are also aligned to a
17072 16-bit boundary even on targets whose API mandates promotion to 32-bit.
17076 Do not consider type @code{int} to be 16 bits wide. This is the default.
17079 @itemx -mno-bitfield
17080 @opindex mnobitfield
17081 @opindex mno-bitfield
17082 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
17083 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
17087 Do use the bit-field instructions. The @option{-m68020} option implies
17088 @option{-mbitfield}. This is the default if you use a configuration
17089 designed for a 68020.
17093 Use a different function-calling convention, in which functions
17094 that take a fixed number of arguments return with the @code{rtd}
17095 instruction, which pops their arguments while returning. This
17096 saves one instruction in the caller since there is no need to pop
17097 the arguments there.
17099 This calling convention is incompatible with the one normally
17100 used on Unix, so you cannot use it if you need to call libraries
17101 compiled with the Unix compiler.
17103 Also, you must provide function prototypes for all functions that
17104 take variable numbers of arguments (including @code{printf});
17105 otherwise incorrect code is generated for calls to those
17108 In addition, seriously incorrect code results if you call a
17109 function with too many arguments. (Normally, extra arguments are
17110 harmlessly ignored.)
17112 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
17113 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
17117 Do not use the calling conventions selected by @option{-mrtd}.
17118 This is the default.
17121 @itemx -mno-align-int
17122 @opindex malign-int
17123 @opindex mno-align-int
17124 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
17125 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
17126 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
17127 Aligning variables on 32-bit boundaries produces code that runs somewhat
17128 faster on processors with 32-bit busses at the expense of more memory.
17130 @strong{Warning:} if you use the @option{-malign-int} switch, GCC
17131 aligns structures containing the above types differently than
17132 most published application binary interface specifications for the m68k.
17136 Use the pc-relative addressing mode of the 68000 directly, instead of
17137 using a global offset table. At present, this option implies @option{-fpic},
17138 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
17139 not presently supported with @option{-mpcrel}, though this could be supported for
17140 68020 and higher processors.
17142 @item -mno-strict-align
17143 @itemx -mstrict-align
17144 @opindex mno-strict-align
17145 @opindex mstrict-align
17146 Do not (do) assume that unaligned memory references are handled by
17150 Generate code that allows the data segment to be located in a different
17151 area of memory from the text segment. This allows for execute-in-place in
17152 an environment without virtual memory management. This option implies
17155 @item -mno-sep-data
17156 Generate code that assumes that the data segment follows the text segment.
17157 This is the default.
17159 @item -mid-shared-library
17160 Generate code that supports shared libraries via the library ID method.
17161 This allows for execute-in-place and shared libraries in an environment
17162 without virtual memory management. This option implies @option{-fPIC}.
17164 @item -mno-id-shared-library
17165 Generate code that doesn't assume ID-based shared libraries are being used.
17166 This is the default.
17168 @item -mshared-library-id=n
17169 Specifies the identification number of the ID-based shared library being
17170 compiled. Specifying a value of 0 generates more compact code; specifying
17171 other values forces the allocation of that number to the current
17172 library, but is no more space- or time-efficient than omitting this option.
17178 When generating position-independent code for ColdFire, generate code
17179 that works if the GOT has more than 8192 entries. This code is
17180 larger and slower than code generated without this option. On M680x0
17181 processors, this option is not needed; @option{-fPIC} suffices.
17183 GCC normally uses a single instruction to load values from the GOT@.
17184 While this is relatively efficient, it only works if the GOT
17185 is smaller than about 64k. Anything larger causes the linker
17186 to report an error such as:
17188 @cindex relocation truncated to fit (ColdFire)
17190 relocation truncated to fit: R_68K_GOT16O foobar
17193 If this happens, you should recompile your code with @option{-mxgot}.
17194 It should then work with very large GOTs. However, code generated with
17195 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
17196 the value of a global symbol.
17198 Note that some linkers, including newer versions of the GNU linker,
17199 can create multiple GOTs and sort GOT entries. If you have such a linker,
17200 you should only need to use @option{-mxgot} when compiling a single
17201 object file that accesses more than 8192 GOT entries. Very few do.
17203 These options have no effect unless GCC is generating
17204 position-independent code.
17208 @node MCore Options
17209 @subsection MCore Options
17210 @cindex MCore options
17212 These are the @samp{-m} options defined for the Motorola M*Core
17218 @itemx -mno-hardlit
17220 @opindex mno-hardlit
17221 Inline constants into the code stream if it can be done in two
17222 instructions or less.
17228 Use the divide instruction. (Enabled by default).
17230 @item -mrelax-immediate
17231 @itemx -mno-relax-immediate
17232 @opindex mrelax-immediate
17233 @opindex mno-relax-immediate
17234 Allow arbitrary-sized immediates in bit operations.
17236 @item -mwide-bitfields
17237 @itemx -mno-wide-bitfields
17238 @opindex mwide-bitfields
17239 @opindex mno-wide-bitfields
17240 Always treat bit-fields as @code{int}-sized.
17242 @item -m4byte-functions
17243 @itemx -mno-4byte-functions
17244 @opindex m4byte-functions
17245 @opindex mno-4byte-functions
17246 Force all functions to be aligned to a 4-byte boundary.
17248 @item -mcallgraph-data
17249 @itemx -mno-callgraph-data
17250 @opindex mcallgraph-data
17251 @opindex mno-callgraph-data
17252 Emit callgraph information.
17255 @itemx -mno-slow-bytes
17256 @opindex mslow-bytes
17257 @opindex mno-slow-bytes
17258 Prefer word access when reading byte quantities.
17260 @item -mlittle-endian
17261 @itemx -mbig-endian
17262 @opindex mlittle-endian
17263 @opindex mbig-endian
17264 Generate code for a little-endian target.
17270 Generate code for the 210 processor.
17274 Assume that runtime support has been provided and so omit the
17275 simulator library (@file{libsim.a)} from the linker command line.
17277 @item -mstack-increment=@var{size}
17278 @opindex mstack-increment
17279 Set the maximum amount for a single stack increment operation. Large
17280 values can increase the speed of programs that contain functions
17281 that need a large amount of stack space, but they can also trigger a
17282 segmentation fault if the stack is extended too much. The default
17288 @subsection MeP Options
17289 @cindex MeP options
17295 Enables the @code{abs} instruction, which is the absolute difference
17296 between two registers.
17300 Enables all the optional instructions---average, multiply, divide, bit
17301 operations, leading zero, absolute difference, min/max, clip, and
17307 Enables the @code{ave} instruction, which computes the average of two
17310 @item -mbased=@var{n}
17312 Variables of size @var{n} bytes or smaller are placed in the
17313 @code{.based} section by default. Based variables use the @code{$tp}
17314 register as a base register, and there is a 128-byte limit to the
17315 @code{.based} section.
17319 Enables the bit operation instructions---bit test (@code{btstm}), set
17320 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
17321 test-and-set (@code{tas}).
17323 @item -mc=@var{name}
17325 Selects which section constant data is placed in. @var{name} may
17326 be @code{tiny}, @code{near}, or @code{far}.
17330 Enables the @code{clip} instruction. Note that @code{-mclip} is not
17331 useful unless you also provide @code{-mminmax}.
17333 @item -mconfig=@var{name}
17335 Selects one of the built-in core configurations. Each MeP chip has
17336 one or more modules in it; each module has a core CPU and a variety of
17337 coprocessors, optional instructions, and peripherals. The
17338 @code{MeP-Integrator} tool, not part of GCC, provides these
17339 configurations through this option; using this option is the same as
17340 using all the corresponding command-line options. The default
17341 configuration is @code{default}.
17345 Enables the coprocessor instructions. By default, this is a 32-bit
17346 coprocessor. Note that the coprocessor is normally enabled via the
17347 @code{-mconfig=} option.
17351 Enables the 32-bit coprocessor's instructions.
17355 Enables the 64-bit coprocessor's instructions.
17359 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
17363 Causes constant variables to be placed in the @code{.near} section.
17367 Enables the @code{div} and @code{divu} instructions.
17371 Generate big-endian code.
17375 Generate little-endian code.
17377 @item -mio-volatile
17378 @opindex mio-volatile
17379 Tells the compiler that any variable marked with the @code{io}
17380 attribute is to be considered volatile.
17384 Causes variables to be assigned to the @code{.far} section by default.
17388 Enables the @code{leadz} (leading zero) instruction.
17392 Causes variables to be assigned to the @code{.near} section by default.
17396 Enables the @code{min} and @code{max} instructions.
17400 Enables the multiplication and multiply-accumulate instructions.
17404 Disables all the optional instructions enabled by @code{-mall-opts}.
17408 Enables the @code{repeat} and @code{erepeat} instructions, used for
17409 low-overhead looping.
17413 Causes all variables to default to the @code{.tiny} section. Note
17414 that there is a 65536-byte limit to this section. Accesses to these
17415 variables use the @code{%gp} base register.
17419 Enables the saturation instructions. Note that the compiler does not
17420 currently generate these itself, but this option is included for
17421 compatibility with other tools, like @code{as}.
17425 Link the SDRAM-based runtime instead of the default ROM-based runtime.
17429 Link the simulator run-time libraries.
17433 Link the simulator runtime libraries, excluding built-in support
17434 for reset and exception vectors and tables.
17438 Causes all functions to default to the @code{.far} section. Without
17439 this option, functions default to the @code{.near} section.
17441 @item -mtiny=@var{n}
17443 Variables that are @var{n} bytes or smaller are allocated to the
17444 @code{.tiny} section. These variables use the @code{$gp} base
17445 register. The default for this option is 4, but note that there's a
17446 65536-byte limit to the @code{.tiny} section.
17450 @node MicroBlaze Options
17451 @subsection MicroBlaze Options
17452 @cindex MicroBlaze Options
17457 @opindex msoft-float
17458 Use software emulation for floating point (default).
17461 @opindex mhard-float
17462 Use hardware floating-point instructions.
17466 Do not optimize block moves, use @code{memcpy}.
17468 @item -mno-clearbss
17469 @opindex mno-clearbss
17470 This option is deprecated. Use @option{-fno-zero-initialized-in-bss} instead.
17472 @item -mcpu=@var{cpu-type}
17474 Use features of, and schedule code for, the given CPU.
17475 Supported values are in the format @samp{v@var{X}.@var{YY}.@var{Z}},
17476 where @var{X} is a major version, @var{YY} is the minor version, and
17477 @var{Z} is compatibility code. Example values are @samp{v3.00.a},
17478 @samp{v4.00.b}, @samp{v5.00.a}, @samp{v5.00.b}, @samp{v5.00.b}, @samp{v6.00.a}.
17480 @item -mxl-soft-mul
17481 @opindex mxl-soft-mul
17482 Use software multiply emulation (default).
17484 @item -mxl-soft-div
17485 @opindex mxl-soft-div
17486 Use software emulation for divides (default).
17488 @item -mxl-barrel-shift
17489 @opindex mxl-barrel-shift
17490 Use the hardware barrel shifter.
17492 @item -mxl-pattern-compare
17493 @opindex mxl-pattern-compare
17494 Use pattern compare instructions.
17496 @item -msmall-divides
17497 @opindex msmall-divides
17498 Use table lookup optimization for small signed integer divisions.
17500 @item -mxl-stack-check
17501 @opindex mxl-stack-check
17502 This option is deprecated. Use @option{-fstack-check} instead.
17505 @opindex mxl-gp-opt
17506 Use GP-relative @code{.sdata}/@code{.sbss} sections.
17508 @item -mxl-multiply-high
17509 @opindex mxl-multiply-high
17510 Use multiply high instructions for high part of 32x32 multiply.
17512 @item -mxl-float-convert
17513 @opindex mxl-float-convert
17514 Use hardware floating-point conversion instructions.
17516 @item -mxl-float-sqrt
17517 @opindex mxl-float-sqrt
17518 Use hardware floating-point square root instruction.
17521 @opindex mbig-endian
17522 Generate code for a big-endian target.
17524 @item -mlittle-endian
17525 @opindex mlittle-endian
17526 Generate code for a little-endian target.
17529 @opindex mxl-reorder
17530 Use reorder instructions (swap and byte reversed load/store).
17532 @item -mxl-mode-@var{app-model}
17533 Select application model @var{app-model}. Valid models are
17536 normal executable (default), uses startup code @file{crt0.o}.
17539 for use with Xilinx Microprocessor Debugger (XMD) based
17540 software intrusive debug agent called xmdstub. This uses startup file
17541 @file{crt1.o} and sets the start address of the program to 0x800.
17544 for applications that are loaded using a bootloader.
17545 This model uses startup file @file{crt2.o} which does not contain a processor
17546 reset vector handler. This is suitable for transferring control on a
17547 processor reset to the bootloader rather than the application.
17550 for applications that do not require any of the
17551 MicroBlaze vectors. This option may be useful for applications running
17552 within a monitoring application. This model uses @file{crt3.o} as a startup file.
17555 Option @option{-xl-mode-@var{app-model}} is a deprecated alias for
17556 @option{-mxl-mode-@var{app-model}}.
17561 @subsection MIPS Options
17562 @cindex MIPS options
17568 Generate big-endian code.
17572 Generate little-endian code. This is the default for @samp{mips*el-*-*}
17575 @item -march=@var{arch}
17577 Generate code that runs on @var{arch}, which can be the name of a
17578 generic MIPS ISA, or the name of a particular processor.
17580 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
17581 @samp{mips32}, @samp{mips32r2}, @samp{mips32r3}, @samp{mips32r5},
17582 @samp{mips64}, @samp{mips64r2}, @samp{mips64r3} and @samp{mips64r5}.
17583 The processor names are:
17584 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
17585 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
17586 @samp{5kc}, @samp{5kf},
17588 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
17589 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
17590 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1}, @samp{34kn},
17591 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
17592 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
17593 @samp{loongson2e}, @samp{loongson2f}, @samp{loongson3a},
17595 @samp{m14k}, @samp{m14kc}, @samp{m14ke}, @samp{m14kec},
17596 @samp{octeon}, @samp{octeon+}, @samp{octeon2},
17599 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
17600 @samp{r4600}, @samp{r4650}, @samp{r4700}, @samp{r6000}, @samp{r8000},
17601 @samp{rm7000}, @samp{rm9000},
17602 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
17605 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
17606 @samp{vr5000}, @samp{vr5400}, @samp{vr5500},
17607 @samp{xlr} and @samp{xlp}.
17608 The special value @samp{from-abi} selects the
17609 most compatible architecture for the selected ABI (that is,
17610 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
17612 The native Linux/GNU toolchain also supports the value @samp{native},
17613 which selects the best architecture option for the host processor.
17614 @option{-march=native} has no effect if GCC does not recognize
17617 In processor names, a final @samp{000} can be abbreviated as @samp{k}
17618 (for example, @option{-march=r2k}). Prefixes are optional, and
17619 @samp{vr} may be written @samp{r}.
17621 Names of the form @samp{@var{n}f2_1} refer to processors with
17622 FPUs clocked at half the rate of the core, names of the form
17623 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
17624 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
17625 processors with FPUs clocked a ratio of 3:2 with respect to the core.
17626 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
17627 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
17628 accepted as synonyms for @samp{@var{n}f1_1}.
17630 GCC defines two macros based on the value of this option. The first
17631 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
17632 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
17633 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
17634 For example, @option{-march=r2000} sets @samp{_MIPS_ARCH}
17635 to @samp{"r2000"} and defines the macro @samp{_MIPS_ARCH_R2000}.
17637 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
17638 above. In other words, it has the full prefix and does not
17639 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
17640 the macro names the resolved architecture (either @samp{"mips1"} or
17641 @samp{"mips3"}). It names the default architecture when no
17642 @option{-march} option is given.
17644 @item -mtune=@var{arch}
17646 Optimize for @var{arch}. Among other things, this option controls
17647 the way instructions are scheduled, and the perceived cost of arithmetic
17648 operations. The list of @var{arch} values is the same as for
17651 When this option is not used, GCC optimizes for the processor
17652 specified by @option{-march}. By using @option{-march} and
17653 @option{-mtune} together, it is possible to generate code that
17654 runs on a family of processors, but optimize the code for one
17655 particular member of that family.
17657 @option{-mtune} defines the macros @samp{_MIPS_TUNE} and
17658 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
17659 @option{-march} ones described above.
17663 Equivalent to @option{-march=mips1}.
17667 Equivalent to @option{-march=mips2}.
17671 Equivalent to @option{-march=mips3}.
17675 Equivalent to @option{-march=mips4}.
17679 Equivalent to @option{-march=mips32}.
17683 Equivalent to @option{-march=mips32r3}.
17687 Equivalent to @option{-march=mips32r5}.
17691 Equivalent to @option{-march=mips64}.
17695 Equivalent to @option{-march=mips64r2}.
17699 Equivalent to @option{-march=mips64r3}.
17703 Equivalent to @option{-march=mips64r5}.
17708 @opindex mno-mips16
17709 Generate (do not generate) MIPS16 code. If GCC is targeting a
17710 MIPS32 or MIPS64 architecture, it makes use of the MIPS16e ASE@.
17712 MIPS16 code generation can also be controlled on a per-function basis
17713 by means of @code{mips16} and @code{nomips16} attributes.
17714 @xref{Function Attributes}, for more information.
17716 @item -mflip-mips16
17717 @opindex mflip-mips16
17718 Generate MIPS16 code on alternating functions. This option is provided
17719 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
17720 not intended for ordinary use in compiling user code.
17722 @item -minterlink-compressed
17723 @item -mno-interlink-compressed
17724 @opindex minterlink-compressed
17725 @opindex mno-interlink-compressed
17726 Require (do not require) that code using the standard (uncompressed) MIPS ISA
17727 be link-compatible with MIPS16 and microMIPS code, and vice versa.
17729 For example, code using the standard ISA encoding cannot jump directly
17730 to MIPS16 or microMIPS code; it must either use a call or an indirect jump.
17731 @option{-minterlink-compressed} therefore disables direct jumps unless GCC
17732 knows that the target of the jump is not compressed.
17734 @item -minterlink-mips16
17735 @itemx -mno-interlink-mips16
17736 @opindex minterlink-mips16
17737 @opindex mno-interlink-mips16
17738 Aliases of @option{-minterlink-compressed} and
17739 @option{-mno-interlink-compressed}. These options predate the microMIPS ASE
17740 and are retained for backwards compatibility.
17752 Generate code for the given ABI@.
17754 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
17755 generates 64-bit code when you select a 64-bit architecture, but you
17756 can use @option{-mgp32} to get 32-bit code instead.
17758 For information about the O64 ABI, see
17759 @uref{http://gcc.gnu.org/@/projects/@/mipso64-abi.html}.
17761 GCC supports a variant of the o32 ABI in which floating-point registers
17762 are 64 rather than 32 bits wide. You can select this combination with
17763 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @code{mthc1}
17764 and @code{mfhc1} instructions and is therefore only supported for
17765 MIPS32R2, MIPS32R3 and MIPS32R5 processors.
17767 The register assignments for arguments and return values remain the
17768 same, but each scalar value is passed in a single 64-bit register
17769 rather than a pair of 32-bit registers. For example, scalar
17770 floating-point values are returned in @samp{$f0} only, not a
17771 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
17772 remains the same, but all 64 bits are saved.
17775 @itemx -mno-abicalls
17777 @opindex mno-abicalls
17778 Generate (do not generate) code that is suitable for SVR4-style
17779 dynamic objects. @option{-mabicalls} is the default for SVR4-based
17784 Generate (do not generate) code that is fully position-independent,
17785 and that can therefore be linked into shared libraries. This option
17786 only affects @option{-mabicalls}.
17788 All @option{-mabicalls} code has traditionally been position-independent,
17789 regardless of options like @option{-fPIC} and @option{-fpic}. However,
17790 as an extension, the GNU toolchain allows executables to use absolute
17791 accesses for locally-binding symbols. It can also use shorter GP
17792 initialization sequences and generate direct calls to locally-defined
17793 functions. This mode is selected by @option{-mno-shared}.
17795 @option{-mno-shared} depends on binutils 2.16 or higher and generates
17796 objects that can only be linked by the GNU linker. However, the option
17797 does not affect the ABI of the final executable; it only affects the ABI
17798 of relocatable objects. Using @option{-mno-shared} generally makes
17799 executables both smaller and quicker.
17801 @option{-mshared} is the default.
17807 Assume (do not assume) that the static and dynamic linkers
17808 support PLTs and copy relocations. This option only affects
17809 @option{-mno-shared -mabicalls}. For the n64 ABI, this option
17810 has no effect without @option{-msym32}.
17812 You can make @option{-mplt} the default by configuring
17813 GCC with @option{--with-mips-plt}. The default is
17814 @option{-mno-plt} otherwise.
17820 Lift (do not lift) the usual restrictions on the size of the global
17823 GCC normally uses a single instruction to load values from the GOT@.
17824 While this is relatively efficient, it only works if the GOT
17825 is smaller than about 64k. Anything larger causes the linker
17826 to report an error such as:
17828 @cindex relocation truncated to fit (MIPS)
17830 relocation truncated to fit: R_MIPS_GOT16 foobar
17833 If this happens, you should recompile your code with @option{-mxgot}.
17834 This works with very large GOTs, although the code is also
17835 less efficient, since it takes three instructions to fetch the
17836 value of a global symbol.
17838 Note that some linkers can create multiple GOTs. If you have such a
17839 linker, you should only need to use @option{-mxgot} when a single object
17840 file accesses more than 64k's worth of GOT entries. Very few do.
17842 These options have no effect unless GCC is generating position
17847 Assume that general-purpose registers are 32 bits wide.
17851 Assume that general-purpose registers are 64 bits wide.
17855 Assume that floating-point registers are 32 bits wide.
17859 Assume that floating-point registers are 64 bits wide.
17862 @opindex mhard-float
17863 Use floating-point coprocessor instructions.
17866 @opindex msoft-float
17867 Do not use floating-point coprocessor instructions. Implement
17868 floating-point calculations using library calls instead.
17872 Equivalent to @option{-msoft-float}, but additionally asserts that the
17873 program being compiled does not perform any floating-point operations.
17874 This option is presently supported only by some bare-metal MIPS
17875 configurations, where it may select a special set of libraries
17876 that lack all floating-point support (including, for example, the
17877 floating-point @code{printf} formats).
17878 If code compiled with @code{-mno-float} accidentally contains
17879 floating-point operations, it is likely to suffer a link-time
17880 or run-time failure.
17882 @item -msingle-float
17883 @opindex msingle-float
17884 Assume that the floating-point coprocessor only supports single-precision
17887 @item -mdouble-float
17888 @opindex mdouble-float
17889 Assume that the floating-point coprocessor supports double-precision
17890 operations. This is the default.
17893 @itemx -mabs=legacy
17895 @opindex mabs=legacy
17896 These options control the treatment of the special not-a-number (NaN)
17897 IEEE 754 floating-point data with the @code{abs.@i{fmt}} and
17898 @code{neg.@i{fmt}} machine instructions.
17900 By default or when the @option{-mabs=legacy} is used the legacy
17901 treatment is selected. In this case these instructions are considered
17902 arithmetic and avoided where correct operation is required and the
17903 input operand might be a NaN. A longer sequence of instructions that
17904 manipulate the sign bit of floating-point datum manually is used
17905 instead unless the @option{-ffinite-math-only} option has also been
17908 The @option{-mabs=2008} option selects the IEEE 754-2008 treatment. In
17909 this case these instructions are considered non-arithmetic and therefore
17910 operating correctly in all cases, including in particular where the
17911 input operand is a NaN. These instructions are therefore always used
17912 for the respective operations.
17915 @itemx -mnan=legacy
17917 @opindex mnan=legacy
17918 These options control the encoding of the special not-a-number (NaN)
17919 IEEE 754 floating-point data.
17921 The @option{-mnan=legacy} option selects the legacy encoding. In this
17922 case quiet NaNs (qNaNs) are denoted by the first bit of their trailing
17923 significand field being 0, whereas signalling NaNs (sNaNs) are denoted
17924 by the first bit of their trailing significand field being 1.
17926 The @option{-mnan=2008} option selects the IEEE 754-2008 encoding. In
17927 this case qNaNs are denoted by the first bit of their trailing
17928 significand field being 1, whereas sNaNs are denoted by the first bit of
17929 their trailing significand field being 0.
17931 The default is @option{-mnan=legacy} unless GCC has been configured with
17932 @option{--with-nan=2008}.
17938 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
17939 implement atomic memory built-in functions. When neither option is
17940 specified, GCC uses the instructions if the target architecture
17943 @option{-mllsc} is useful if the runtime environment can emulate the
17944 instructions and @option{-mno-llsc} can be useful when compiling for
17945 nonstandard ISAs. You can make either option the default by
17946 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
17947 respectively. @option{--with-llsc} is the default for some
17948 configurations; see the installation documentation for details.
17954 Use (do not use) revision 1 of the MIPS DSP ASE@.
17955 @xref{MIPS DSP Built-in Functions}. This option defines the
17956 preprocessor macro @samp{__mips_dsp}. It also defines
17957 @samp{__mips_dsp_rev} to 1.
17963 Use (do not use) revision 2 of the MIPS DSP ASE@.
17964 @xref{MIPS DSP Built-in Functions}. This option defines the
17965 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
17966 It also defines @samp{__mips_dsp_rev} to 2.
17969 @itemx -mno-smartmips
17970 @opindex msmartmips
17971 @opindex mno-smartmips
17972 Use (do not use) the MIPS SmartMIPS ASE.
17974 @item -mpaired-single
17975 @itemx -mno-paired-single
17976 @opindex mpaired-single
17977 @opindex mno-paired-single
17978 Use (do not use) paired-single floating-point instructions.
17979 @xref{MIPS Paired-Single Support}. This option requires
17980 hardware floating-point support to be enabled.
17986 Use (do not use) MIPS Digital Media Extension instructions.
17987 This option can only be used when generating 64-bit code and requires
17988 hardware floating-point support to be enabled.
17993 @opindex mno-mips3d
17994 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
17995 The option @option{-mips3d} implies @option{-mpaired-single}.
17998 @itemx -mno-micromips
17999 @opindex mmicromips
18000 @opindex mno-mmicromips
18001 Generate (do not generate) microMIPS code.
18003 MicroMIPS code generation can also be controlled on a per-function basis
18004 by means of @code{micromips} and @code{nomicromips} attributes.
18005 @xref{Function Attributes}, for more information.
18011 Use (do not use) MT Multithreading instructions.
18017 Use (do not use) the MIPS MCU ASE instructions.
18023 Use (do not use) the MIPS Enhanced Virtual Addressing instructions.
18029 Use (do not use) the MIPS Virtualization Application Specific instructions.
18035 Use (do not use) the MIPS eXtended Physical Address (XPA) instructions.
18039 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
18040 an explanation of the default and the way that the pointer size is
18045 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
18047 The default size of @code{int}s, @code{long}s and pointers depends on
18048 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
18049 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
18050 32-bit @code{long}s. Pointers are the same size as @code{long}s,
18051 or the same size as integer registers, whichever is smaller.
18057 Assume (do not assume) that all symbols have 32-bit values, regardless
18058 of the selected ABI@. This option is useful in combination with
18059 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
18060 to generate shorter and faster references to symbolic addresses.
18064 Put definitions of externally-visible data in a small data section
18065 if that data is no bigger than @var{num} bytes. GCC can then generate
18066 more efficient accesses to the data; see @option{-mgpopt} for details.
18068 The default @option{-G} option depends on the configuration.
18070 @item -mlocal-sdata
18071 @itemx -mno-local-sdata
18072 @opindex mlocal-sdata
18073 @opindex mno-local-sdata
18074 Extend (do not extend) the @option{-G} behavior to local data too,
18075 such as to static variables in C@. @option{-mlocal-sdata} is the
18076 default for all configurations.
18078 If the linker complains that an application is using too much small data,
18079 you might want to try rebuilding the less performance-critical parts with
18080 @option{-mno-local-sdata}. You might also want to build large
18081 libraries with @option{-mno-local-sdata}, so that the libraries leave
18082 more room for the main program.
18084 @item -mextern-sdata
18085 @itemx -mno-extern-sdata
18086 @opindex mextern-sdata
18087 @opindex mno-extern-sdata
18088 Assume (do not assume) that externally-defined data is in
18089 a small data section if the size of that data is within the @option{-G} limit.
18090 @option{-mextern-sdata} is the default for all configurations.
18092 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
18093 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
18094 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
18095 is placed in a small data section. If @var{Var} is defined by another
18096 module, you must either compile that module with a high-enough
18097 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
18098 definition. If @var{Var} is common, you must link the application
18099 with a high-enough @option{-G} setting.
18101 The easiest way of satisfying these restrictions is to compile
18102 and link every module with the same @option{-G} option. However,
18103 you may wish to build a library that supports several different
18104 small data limits. You can do this by compiling the library with
18105 the highest supported @option{-G} setting and additionally using
18106 @option{-mno-extern-sdata} to stop the library from making assumptions
18107 about externally-defined data.
18113 Use (do not use) GP-relative accesses for symbols that are known to be
18114 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
18115 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
18118 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
18119 might not hold the value of @code{_gp}. For example, if the code is
18120 part of a library that might be used in a boot monitor, programs that
18121 call boot monitor routines pass an unknown value in @code{$gp}.
18122 (In such situations, the boot monitor itself is usually compiled
18123 with @option{-G0}.)
18125 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
18126 @option{-mno-extern-sdata}.
18128 @item -membedded-data
18129 @itemx -mno-embedded-data
18130 @opindex membedded-data
18131 @opindex mno-embedded-data
18132 Allocate variables to the read-only data section first if possible, then
18133 next in the small data section if possible, otherwise in data. This gives
18134 slightly slower code than the default, but reduces the amount of RAM required
18135 when executing, and thus may be preferred for some embedded systems.
18137 @item -muninit-const-in-rodata
18138 @itemx -mno-uninit-const-in-rodata
18139 @opindex muninit-const-in-rodata
18140 @opindex mno-uninit-const-in-rodata
18141 Put uninitialized @code{const} variables in the read-only data section.
18142 This option is only meaningful in conjunction with @option{-membedded-data}.
18144 @item -mcode-readable=@var{setting}
18145 @opindex mcode-readable
18146 Specify whether GCC may generate code that reads from executable sections.
18147 There are three possible settings:
18150 @item -mcode-readable=yes
18151 Instructions may freely access executable sections. This is the
18154 @item -mcode-readable=pcrel
18155 MIPS16 PC-relative load instructions can access executable sections,
18156 but other instructions must not do so. This option is useful on 4KSc
18157 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
18158 It is also useful on processors that can be configured to have a dual
18159 instruction/data SRAM interface and that, like the M4K, automatically
18160 redirect PC-relative loads to the instruction RAM.
18162 @item -mcode-readable=no
18163 Instructions must not access executable sections. This option can be
18164 useful on targets that are configured to have a dual instruction/data
18165 SRAM interface but that (unlike the M4K) do not automatically redirect
18166 PC-relative loads to the instruction RAM.
18169 @item -msplit-addresses
18170 @itemx -mno-split-addresses
18171 @opindex msplit-addresses
18172 @opindex mno-split-addresses
18173 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
18174 relocation operators. This option has been superseded by
18175 @option{-mexplicit-relocs} but is retained for backwards compatibility.
18177 @item -mexplicit-relocs
18178 @itemx -mno-explicit-relocs
18179 @opindex mexplicit-relocs
18180 @opindex mno-explicit-relocs
18181 Use (do not use) assembler relocation operators when dealing with symbolic
18182 addresses. The alternative, selected by @option{-mno-explicit-relocs},
18183 is to use assembler macros instead.
18185 @option{-mexplicit-relocs} is the default if GCC was configured
18186 to use an assembler that supports relocation operators.
18188 @item -mcheck-zero-division
18189 @itemx -mno-check-zero-division
18190 @opindex mcheck-zero-division
18191 @opindex mno-check-zero-division
18192 Trap (do not trap) on integer division by zero.
18194 The default is @option{-mcheck-zero-division}.
18196 @item -mdivide-traps
18197 @itemx -mdivide-breaks
18198 @opindex mdivide-traps
18199 @opindex mdivide-breaks
18200 MIPS systems check for division by zero by generating either a
18201 conditional trap or a break instruction. Using traps results in
18202 smaller code, but is only supported on MIPS II and later. Also, some
18203 versions of the Linux kernel have a bug that prevents trap from
18204 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
18205 allow conditional traps on architectures that support them and
18206 @option{-mdivide-breaks} to force the use of breaks.
18208 The default is usually @option{-mdivide-traps}, but this can be
18209 overridden at configure time using @option{--with-divide=breaks}.
18210 Divide-by-zero checks can be completely disabled using
18211 @option{-mno-check-zero-division}.
18216 @opindex mno-memcpy
18217 Force (do not force) the use of @code{memcpy()} for non-trivial block
18218 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
18219 most constant-sized copies.
18222 @itemx -mno-long-calls
18223 @opindex mlong-calls
18224 @opindex mno-long-calls
18225 Disable (do not disable) use of the @code{jal} instruction. Calling
18226 functions using @code{jal} is more efficient but requires the caller
18227 and callee to be in the same 256 megabyte segment.
18229 This option has no effect on abicalls code. The default is
18230 @option{-mno-long-calls}.
18236 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
18237 instructions, as provided by the R4650 ISA@.
18243 Enable (disable) use of the @code{madd} and @code{msub} integer
18244 instructions. The default is @option{-mimadd} on architectures
18245 that support @code{madd} and @code{msub} except for the 74k
18246 architecture where it was found to generate slower code.
18249 @itemx -mno-fused-madd
18250 @opindex mfused-madd
18251 @opindex mno-fused-madd
18252 Enable (disable) use of the floating-point multiply-accumulate
18253 instructions, when they are available. The default is
18254 @option{-mfused-madd}.
18256 On the R8000 CPU when multiply-accumulate instructions are used,
18257 the intermediate product is calculated to infinite precision
18258 and is not subject to the FCSR Flush to Zero bit. This may be
18259 undesirable in some circumstances. On other processors the result
18260 is numerically identical to the equivalent computation using
18261 separate multiply, add, subtract and negate instructions.
18265 Tell the MIPS assembler to not run its preprocessor over user
18266 assembler files (with a @samp{.s} suffix) when assembling them.
18271 @opindex mno-fix-24k
18272 Work around the 24K E48 (lost data on stores during refill) errata.
18273 The workarounds are implemented by the assembler rather than by GCC@.
18276 @itemx -mno-fix-r4000
18277 @opindex mfix-r4000
18278 @opindex mno-fix-r4000
18279 Work around certain R4000 CPU errata:
18282 A double-word or a variable shift may give an incorrect result if executed
18283 immediately after starting an integer division.
18285 A double-word or a variable shift may give an incorrect result if executed
18286 while an integer multiplication is in progress.
18288 An integer division may give an incorrect result if started in a delay slot
18289 of a taken branch or a jump.
18293 @itemx -mno-fix-r4400
18294 @opindex mfix-r4400
18295 @opindex mno-fix-r4400
18296 Work around certain R4400 CPU errata:
18299 A double-word or a variable shift may give an incorrect result if executed
18300 immediately after starting an integer division.
18304 @itemx -mno-fix-r10000
18305 @opindex mfix-r10000
18306 @opindex mno-fix-r10000
18307 Work around certain R10000 errata:
18310 @code{ll}/@code{sc} sequences may not behave atomically on revisions
18311 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
18314 This option can only be used if the target architecture supports
18315 branch-likely instructions. @option{-mfix-r10000} is the default when
18316 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
18320 @itemx -mno-fix-rm7000
18321 @opindex mfix-rm7000
18322 Work around the RM7000 @code{dmult}/@code{dmultu} errata. The
18323 workarounds are implemented by the assembler rather than by GCC@.
18326 @itemx -mno-fix-vr4120
18327 @opindex mfix-vr4120
18328 Work around certain VR4120 errata:
18331 @code{dmultu} does not always produce the correct result.
18333 @code{div} and @code{ddiv} do not always produce the correct result if one
18334 of the operands is negative.
18336 The workarounds for the division errata rely on special functions in
18337 @file{libgcc.a}. At present, these functions are only provided by
18338 the @code{mips64vr*-elf} configurations.
18340 Other VR4120 errata require a NOP to be inserted between certain pairs of
18341 instructions. These errata are handled by the assembler, not by GCC itself.
18344 @opindex mfix-vr4130
18345 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
18346 workarounds are implemented by the assembler rather than by GCC,
18347 although GCC avoids using @code{mflo} and @code{mfhi} if the
18348 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
18349 instructions are available instead.
18352 @itemx -mno-fix-sb1
18354 Work around certain SB-1 CPU core errata.
18355 (This flag currently works around the SB-1 revision 2
18356 ``F1'' and ``F2'' floating-point errata.)
18358 @item -mr10k-cache-barrier=@var{setting}
18359 @opindex mr10k-cache-barrier
18360 Specify whether GCC should insert cache barriers to avoid the
18361 side-effects of speculation on R10K processors.
18363 In common with many processors, the R10K tries to predict the outcome
18364 of a conditional branch and speculatively executes instructions from
18365 the ``taken'' branch. It later aborts these instructions if the
18366 predicted outcome is wrong. However, on the R10K, even aborted
18367 instructions can have side effects.
18369 This problem only affects kernel stores and, depending on the system,
18370 kernel loads. As an example, a speculatively-executed store may load
18371 the target memory into cache and mark the cache line as dirty, even if
18372 the store itself is later aborted. If a DMA operation writes to the
18373 same area of memory before the ``dirty'' line is flushed, the cached
18374 data overwrites the DMA-ed data. See the R10K processor manual
18375 for a full description, including other potential problems.
18377 One workaround is to insert cache barrier instructions before every memory
18378 access that might be speculatively executed and that might have side
18379 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
18380 controls GCC's implementation of this workaround. It assumes that
18381 aborted accesses to any byte in the following regions does not have
18386 the memory occupied by the current function's stack frame;
18389 the memory occupied by an incoming stack argument;
18392 the memory occupied by an object with a link-time-constant address.
18395 It is the kernel's responsibility to ensure that speculative
18396 accesses to these regions are indeed safe.
18398 If the input program contains a function declaration such as:
18404 then the implementation of @code{foo} must allow @code{j foo} and
18405 @code{jal foo} to be executed speculatively. GCC honors this
18406 restriction for functions it compiles itself. It expects non-GCC
18407 functions (such as hand-written assembly code) to do the same.
18409 The option has three forms:
18412 @item -mr10k-cache-barrier=load-store
18413 Insert a cache barrier before a load or store that might be
18414 speculatively executed and that might have side effects even
18417 @item -mr10k-cache-barrier=store
18418 Insert a cache barrier before a store that might be speculatively
18419 executed and that might have side effects even if aborted.
18421 @item -mr10k-cache-barrier=none
18422 Disable the insertion of cache barriers. This is the default setting.
18425 @item -mflush-func=@var{func}
18426 @itemx -mno-flush-func
18427 @opindex mflush-func
18428 Specifies the function to call to flush the I and D caches, or to not
18429 call any such function. If called, the function must take the same
18430 arguments as the common @code{_flush_func()}, that is, the address of the
18431 memory range for which the cache is being flushed, the size of the
18432 memory range, and the number 3 (to flush both caches). The default
18433 depends on the target GCC was configured for, but commonly is either
18434 @samp{_flush_func} or @samp{__cpu_flush}.
18436 @item mbranch-cost=@var{num}
18437 @opindex mbranch-cost
18438 Set the cost of branches to roughly @var{num} ``simple'' instructions.
18439 This cost is only a heuristic and is not guaranteed to produce
18440 consistent results across releases. A zero cost redundantly selects
18441 the default, which is based on the @option{-mtune} setting.
18443 @item -mbranch-likely
18444 @itemx -mno-branch-likely
18445 @opindex mbranch-likely
18446 @opindex mno-branch-likely
18447 Enable or disable use of Branch Likely instructions, regardless of the
18448 default for the selected architecture. By default, Branch Likely
18449 instructions may be generated if they are supported by the selected
18450 architecture. An exception is for the MIPS32 and MIPS64 architectures
18451 and processors that implement those architectures; for those, Branch
18452 Likely instructions are not be generated by default because the MIPS32
18453 and MIPS64 architectures specifically deprecate their use.
18455 @item -mfp-exceptions
18456 @itemx -mno-fp-exceptions
18457 @opindex mfp-exceptions
18458 Specifies whether FP exceptions are enabled. This affects how
18459 FP instructions are scheduled for some processors.
18460 The default is that FP exceptions are
18463 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
18464 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
18467 @item -mvr4130-align
18468 @itemx -mno-vr4130-align
18469 @opindex mvr4130-align
18470 The VR4130 pipeline is two-way superscalar, but can only issue two
18471 instructions together if the first one is 8-byte aligned. When this
18472 option is enabled, GCC aligns pairs of instructions that it
18473 thinks should execute in parallel.
18475 This option only has an effect when optimizing for the VR4130.
18476 It normally makes code faster, but at the expense of making it bigger.
18477 It is enabled by default at optimization level @option{-O3}.
18482 Enable (disable) generation of @code{synci} instructions on
18483 architectures that support it. The @code{synci} instructions (if
18484 enabled) are generated when @code{__builtin___clear_cache()} is
18487 This option defaults to @code{-mno-synci}, but the default can be
18488 overridden by configuring with @code{--with-synci}.
18490 When compiling code for single processor systems, it is generally safe
18491 to use @code{synci}. However, on many multi-core (SMP) systems, it
18492 does not invalidate the instruction caches on all cores and may lead
18493 to undefined behavior.
18495 @item -mrelax-pic-calls
18496 @itemx -mno-relax-pic-calls
18497 @opindex mrelax-pic-calls
18498 Try to turn PIC calls that are normally dispatched via register
18499 @code{$25} into direct calls. This is only possible if the linker can
18500 resolve the destination at link-time and if the destination is within
18501 range for a direct call.
18503 @option{-mrelax-pic-calls} is the default if GCC was configured to use
18504 an assembler and a linker that support the @code{.reloc} assembly
18505 directive and @code{-mexplicit-relocs} is in effect. With
18506 @code{-mno-explicit-relocs}, this optimization can be performed by the
18507 assembler and the linker alone without help from the compiler.
18509 @item -mmcount-ra-address
18510 @itemx -mno-mcount-ra-address
18511 @opindex mmcount-ra-address
18512 @opindex mno-mcount-ra-address
18513 Emit (do not emit) code that allows @code{_mcount} to modify the
18514 calling function's return address. When enabled, this option extends
18515 the usual @code{_mcount} interface with a new @var{ra-address}
18516 parameter, which has type @code{intptr_t *} and is passed in register
18517 @code{$12}. @code{_mcount} can then modify the return address by
18518 doing both of the following:
18521 Returning the new address in register @code{$31}.
18523 Storing the new address in @code{*@var{ra-address}},
18524 if @var{ra-address} is nonnull.
18527 The default is @option{-mno-mcount-ra-address}.
18532 @subsection MMIX Options
18533 @cindex MMIX Options
18535 These options are defined for the MMIX:
18539 @itemx -mno-libfuncs
18541 @opindex mno-libfuncs
18542 Specify that intrinsic library functions are being compiled, passing all
18543 values in registers, no matter the size.
18546 @itemx -mno-epsilon
18548 @opindex mno-epsilon
18549 Generate floating-point comparison instructions that compare with respect
18550 to the @code{rE} epsilon register.
18552 @item -mabi=mmixware
18554 @opindex mabi=mmixware
18556 Generate code that passes function parameters and return values that (in
18557 the called function) are seen as registers @code{$0} and up, as opposed to
18558 the GNU ABI which uses global registers @code{$231} and up.
18560 @item -mzero-extend
18561 @itemx -mno-zero-extend
18562 @opindex mzero-extend
18563 @opindex mno-zero-extend
18564 When reading data from memory in sizes shorter than 64 bits, use (do not
18565 use) zero-extending load instructions by default, rather than
18566 sign-extending ones.
18569 @itemx -mno-knuthdiv
18571 @opindex mno-knuthdiv
18572 Make the result of a division yielding a remainder have the same sign as
18573 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
18574 remainder follows the sign of the dividend. Both methods are
18575 arithmetically valid, the latter being almost exclusively used.
18577 @item -mtoplevel-symbols
18578 @itemx -mno-toplevel-symbols
18579 @opindex mtoplevel-symbols
18580 @opindex mno-toplevel-symbols
18581 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
18582 code can be used with the @code{PREFIX} assembly directive.
18586 Generate an executable in the ELF format, rather than the default
18587 @samp{mmo} format used by the @command{mmix} simulator.
18589 @item -mbranch-predict
18590 @itemx -mno-branch-predict
18591 @opindex mbranch-predict
18592 @opindex mno-branch-predict
18593 Use (do not use) the probable-branch instructions, when static branch
18594 prediction indicates a probable branch.
18596 @item -mbase-addresses
18597 @itemx -mno-base-addresses
18598 @opindex mbase-addresses
18599 @opindex mno-base-addresses
18600 Generate (do not generate) code that uses @emph{base addresses}. Using a
18601 base address automatically generates a request (handled by the assembler
18602 and the linker) for a constant to be set up in a global register. The
18603 register is used for one or more base address requests within the range 0
18604 to 255 from the value held in the register. The generally leads to short
18605 and fast code, but the number of different data items that can be
18606 addressed is limited. This means that a program that uses lots of static
18607 data may require @option{-mno-base-addresses}.
18609 @item -msingle-exit
18610 @itemx -mno-single-exit
18611 @opindex msingle-exit
18612 @opindex mno-single-exit
18613 Force (do not force) generated code to have a single exit point in each
18617 @node MN10300 Options
18618 @subsection MN10300 Options
18619 @cindex MN10300 options
18621 These @option{-m} options are defined for Matsushita MN10300 architectures:
18626 Generate code to avoid bugs in the multiply instructions for the MN10300
18627 processors. This is the default.
18629 @item -mno-mult-bug
18630 @opindex mno-mult-bug
18631 Do not generate code to avoid bugs in the multiply instructions for the
18632 MN10300 processors.
18636 Generate code using features specific to the AM33 processor.
18640 Do not generate code using features specific to the AM33 processor. This
18645 Generate code using features specific to the AM33/2.0 processor.
18649 Generate code using features specific to the AM34 processor.
18651 @item -mtune=@var{cpu-type}
18653 Use the timing characteristics of the indicated CPU type when
18654 scheduling instructions. This does not change the targeted processor
18655 type. The CPU type must be one of @samp{mn10300}, @samp{am33},
18656 @samp{am33-2} or @samp{am34}.
18658 @item -mreturn-pointer-on-d0
18659 @opindex mreturn-pointer-on-d0
18660 When generating a function that returns a pointer, return the pointer
18661 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
18662 only in @code{a0}, and attempts to call such functions without a prototype
18663 result in errors. Note that this option is on by default; use
18664 @option{-mno-return-pointer-on-d0} to disable it.
18668 Do not link in the C run-time initialization object file.
18672 Indicate to the linker that it should perform a relaxation optimization pass
18673 to shorten branches, calls and absolute memory addresses. This option only
18674 has an effect when used on the command line for the final link step.
18676 This option makes symbolic debugging impossible.
18680 Allow the compiler to generate @emph{Long Instruction Word}
18681 instructions if the target is the @samp{AM33} or later. This is the
18682 default. This option defines the preprocessor macro @samp{__LIW__}.
18686 Do not allow the compiler to generate @emph{Long Instruction Word}
18687 instructions. This option defines the preprocessor macro
18692 Allow the compiler to generate the @emph{SETLB} and @emph{Lcc}
18693 instructions if the target is the @samp{AM33} or later. This is the
18694 default. This option defines the preprocessor macro @samp{__SETLB__}.
18698 Do not allow the compiler to generate @emph{SETLB} or @emph{Lcc}
18699 instructions. This option defines the preprocessor macro
18700 @samp{__NO_SETLB__}.
18704 @node Moxie Options
18705 @subsection Moxie Options
18706 @cindex Moxie Options
18712 Generate big-endian code. This is the default for @samp{moxie-*-*}
18717 Generate little-endian code.
18721 Do not link in the C run-time initialization object file.
18725 @node MSP430 Options
18726 @subsection MSP430 Options
18727 @cindex MSP430 Options
18729 These options are defined for the MSP430:
18735 Force assembly output to always use hex constants. Normally such
18736 constants are signed decimals, but this option is available for
18737 testsuite and/or aesthetic purposes.
18741 Select the MCU to target. This is used to create a C preprocessor
18742 symbol based upon the MCU name, converted to upper case and pre- and
18743 post- fixed with @code{__}. This in turn will be used by the
18744 @code{msp430.h} header file to select an MCU specific supplimentary
18747 The option also sets the ISA to use. If the MCU name is one that is
18748 known to only support the 430 ISA then that is selected, otherwise the
18749 430X ISA is selected. A generic MCU name of @code{msp430} can also be
18750 used to select the 430 ISA. Similarly the generic @code{msp430x} MCU
18751 name will select the 430X ISA.
18753 In addition an MCU specific linker script will be added to the linker
18754 command line. The script's name is the name of the MCU with
18755 @code{.ld} appended. Thus specifying @option{-mmcu=xxx} on the gcc
18756 command line will define the C preprocessor symbol @code{__XXX__} and
18757 cause the linker to search for a script called @file{xxx.ld}.
18759 This option is also passed on to the assembler.
18763 Specifies the ISA to use. Accepted values are @code{msp430},
18764 @code{msp430x} and @code{msp430xv2}. This option is deprecated. The
18765 @option{-mmcu=} option should be used to select the ISA.
18769 Link to the simulator runtime libraries and linker script. Overrides
18770 any scripts that would be selected by the @option{-mmcu=} option.
18774 Use large-model addressing (20-bit pointers, 32-bit @code{size_t}).
18778 Use small-model addressing (16-bit pointers, 16-bit @code{size_t}).
18782 This option is passed to the assembler and linker, and allows the
18783 linker to perform certain optimizations that cannot be done until
18788 Describes the type of hardware multiply supported by the target.
18789 Accepted values are @code{none} for no hardware multiply, @code{16bit}
18790 for the original 16-bit-only multiply supported by early MCUs.
18791 @code{32bit} for the 16/32-bit multiply supported by later MCUs and
18792 @code{f5series} for the 16/32-bit multiply supported by F5-series MCUs.
18793 A value of @code{auto} can also be given. This tells GCC to deduce
18794 the hardware multiply support based upon the MCU name provided by the
18795 @option{-mmcu} option. If no @option{-mmcu} option is specified then
18796 @code{32bit} hardware multiply support is assumed. @code{auto} is the
18799 Hardware multiplies are normally performed by calling a library
18800 routine. This saves space in the generated code. When compiling at
18801 @code{-O3} or higher however the hardware multiplier is invoked
18802 inline. This makes for bigger, but faster code.
18804 The hardware multiply routines disable interrupts whilst running and
18805 restore the previous interrupt state when they finish. This makes
18806 them safe to use inside interrupt handlers as well as in normal code.
18810 @node NDS32 Options
18811 @subsection NDS32 Options
18812 @cindex NDS32 Options
18814 These options are defined for NDS32 implementations:
18819 @opindex mbig-endian
18820 Generate code in big-endian mode.
18822 @item -mlittle-endian
18823 @opindex mlittle-endian
18824 Generate code in little-endian mode.
18826 @item -mreduced-regs
18827 @opindex mreduced-regs
18828 Use reduced-set registers for register allocation.
18831 @opindex mfull-regs
18832 Use full-set registers for register allocation.
18836 Generate conditional move instructions.
18840 Do not generate conditional move instructions.
18844 Generate performance extension instructions.
18846 @item -mno-perf-ext
18847 @opindex mno-perf-ext
18848 Do not generate performance extension instructions.
18852 Generate v3 push25/pop25 instructions.
18855 @opindex mno-v3push
18856 Do not generate v3 push25/pop25 instructions.
18860 Generate 16-bit instructions.
18863 @opindex mno-16-bit
18864 Do not generate 16-bit instructions.
18867 @opindex mgp-direct
18868 Generate GP base instructions directly.
18870 @item -mno-gp-direct
18871 @opindex mno-gp-direct
18872 Do no generate GP base instructions directly.
18874 @item -misr-vector-size=@var{num}
18875 @opindex misr-vector-size
18876 Specify the size of each interrupt vector, which must be 4 or 16.
18878 @item -mcache-block-size=@var{num}
18879 @opindex mcache-block-size
18880 Specify the size of each cache block,
18881 which must be a power of 2 between 4 and 512.
18883 @item -march=@var{arch}
18885 Specify the name of the target architecture.
18887 @item -mforce-fp-as-gp
18888 @opindex mforce-fp-as-gp
18889 Prevent $fp being allocated during register allocation so that compiler
18890 is able to force performing fp-as-gp optimization.
18892 @item -mforbid-fp-as-gp
18893 @opindex mforbid-fp-as-gp
18894 Forbid using $fp to access static and global variables.
18895 This option strictly forbids fp-as-gp optimization
18896 regardless of @option{-mforce-fp-as-gp}.
18900 Use special directives to guide linker doing ex9 optimization.
18903 @opindex mctor-dtor
18904 Enable constructor/destructor feature.
18908 Guide linker to relax instructions.
18912 @node Nios II Options
18913 @subsection Nios II Options
18914 @cindex Nios II options
18915 @cindex Altera Nios II options
18917 These are the options defined for the Altera Nios II processor.
18923 @cindex smaller data references
18924 Put global and static objects less than or equal to @var{num} bytes
18925 into the small data or BSS sections instead of the normal data or BSS
18926 sections. The default value of @var{num} is 8.
18932 Generate (do not generate) GP-relative accesses for objects in the
18933 small data or BSS sections. The default is @option{-mgpopt} except
18934 when @option{-fpic} or @option{-fPIC} is specified to generate
18935 position-independent code. Note that the Nios II ABI does not permit
18936 GP-relative accesses from shared libraries.
18938 You may need to specify @option{-mno-gpopt} explicitly when building
18939 programs that include large amounts of small data, including large
18940 GOT data sections. In this case, the 16-bit offset for GP-relative
18941 addressing may not be large enough to allow access to the entire
18942 small data section.
18948 Generate little-endian (default) or big-endian (experimental) code,
18951 @item -mbypass-cache
18952 @itemx -mno-bypass-cache
18953 @opindex mno-bypass-cache
18954 @opindex mbypass-cache
18955 Force all load and store instructions to always bypass cache by
18956 using I/O variants of the instructions. The default is not to
18959 @item -mno-cache-volatile
18960 @itemx -mcache-volatile
18961 @opindex mcache-volatile
18962 @opindex mno-cache-volatile
18963 Volatile memory access bypass the cache using the I/O variants of
18964 the load and store instructions. The default is not to bypass the cache.
18966 @item -mno-fast-sw-div
18967 @itemx -mfast-sw-div
18968 @opindex mno-fast-sw-div
18969 @opindex mfast-sw-div
18970 Do not use table-based fast divide for small numbers. The default
18971 is to use the fast divide at @option{-O3} and above.
18975 @itemx -mno-hw-mulx
18979 @opindex mno-hw-mul
18981 @opindex mno-hw-mulx
18983 @opindex mno-hw-div
18985 Enable or disable emitting @code{mul}, @code{mulx} and @code{div} family of
18986 instructions by the compiler. The default is to emit @code{mul}
18987 and not emit @code{div} and @code{mulx}.
18989 @item -mcustom-@var{insn}=@var{N}
18990 @itemx -mno-custom-@var{insn}
18991 @opindex mcustom-@var{insn}
18992 @opindex mno-custom-@var{insn}
18993 Each @option{-mcustom-@var{insn}=@var{N}} option enables use of a
18994 custom instruction with encoding @var{N} when generating code that uses
18995 @var{insn}. For example, @code{-mcustom-fadds=253} generates custom
18996 instruction 253 for single-precision floating-point add operations instead
18997 of the default behavior of using a library call.
18999 The following values of @var{insn} are supported. Except as otherwise
19000 noted, floating-point operations are expected to be implemented with
19001 normal IEEE 754 semantics and correspond directly to the C operators or the
19002 equivalent GCC built-in functions (@pxref{Other Builtins}).
19004 Single-precision floating point:
19007 @item @samp{fadds}, @samp{fsubs}, @samp{fdivs}, @samp{fmuls}
19008 Binary arithmetic operations.
19014 Unary absolute value.
19016 @item @samp{fcmpeqs}, @samp{fcmpges}, @samp{fcmpgts}, @samp{fcmples}, @samp{fcmplts}, @samp{fcmpnes}
19017 Comparison operations.
19019 @item @samp{fmins}, @samp{fmaxs}
19020 Floating-point minimum and maximum. These instructions are only
19021 generated if @option{-ffinite-math-only} is specified.
19023 @item @samp{fsqrts}
19024 Unary square root operation.
19026 @item @samp{fcoss}, @samp{fsins}, @samp{ftans}, @samp{fatans}, @samp{fexps}, @samp{flogs}
19027 Floating-point trigonometric and exponential functions. These instructions
19028 are only generated if @option{-funsafe-math-optimizations} is also specified.
19032 Double-precision floating point:
19035 @item @samp{faddd}, @samp{fsubd}, @samp{fdivd}, @samp{fmuld}
19036 Binary arithmetic operations.
19042 Unary absolute value.
19044 @item @samp{fcmpeqd}, @samp{fcmpged}, @samp{fcmpgtd}, @samp{fcmpled}, @samp{fcmpltd}, @samp{fcmpned}
19045 Comparison operations.
19047 @item @samp{fmind}, @samp{fmaxd}
19048 Double-precision minimum and maximum. These instructions are only
19049 generated if @option{-ffinite-math-only} is specified.
19051 @item @samp{fsqrtd}
19052 Unary square root operation.
19054 @item @samp{fcosd}, @samp{fsind}, @samp{ftand}, @samp{fatand}, @samp{fexpd}, @samp{flogd}
19055 Double-precision trigonometric and exponential functions. These instructions
19056 are only generated if @option{-funsafe-math-optimizations} is also specified.
19062 @item @samp{fextsd}
19063 Conversion from single precision to double precision.
19065 @item @samp{ftruncds}
19066 Conversion from double precision to single precision.
19068 @item @samp{fixsi}, @samp{fixsu}, @samp{fixdi}, @samp{fixdu}
19069 Conversion from floating point to signed or unsigned integer types, with
19070 truncation towards zero.
19073 Conversion from single-precision floating point to signed integer,
19074 rounding to the nearest integer and ties away from zero.
19075 This corresponds to the @code{__builtin_lroundf} function when
19076 @option{-fno-math-errno} is used.
19078 @item @samp{floatis}, @samp{floatus}, @samp{floatid}, @samp{floatud}
19079 Conversion from signed or unsigned integer types to floating-point types.
19083 In addition, all of the following transfer instructions for internal
19084 registers X and Y must be provided to use any of the double-precision
19085 floating-point instructions. Custom instructions taking two
19086 double-precision source operands expect the first operand in the
19087 64-bit register X. The other operand (or only operand of a unary
19088 operation) is given to the custom arithmetic instruction with the
19089 least significant half in source register @var{src1} and the most
19090 significant half in @var{src2}. A custom instruction that returns a
19091 double-precision result returns the most significant 32 bits in the
19092 destination register and the other half in 32-bit register Y.
19093 GCC automatically generates the necessary code sequences to write
19094 register X and/or read register Y when double-precision floating-point
19095 instructions are used.
19100 Write @var{src1} into the least significant half of X and @var{src2} into
19101 the most significant half of X.
19104 Write @var{src1} into Y.
19106 @item @samp{frdxhi}, @samp{frdxlo}
19107 Read the most or least (respectively) significant half of X and store it in
19111 Read the value of Y and store it into @var{dest}.
19114 Note that you can gain more local control over generation of Nios II custom
19115 instructions by using the @code{target("custom-@var{insn}=@var{N}")}
19116 and @code{target("no-custom-@var{insn}")} function attributes
19117 (@pxref{Function Attributes})
19118 or pragmas (@pxref{Function Specific Option Pragmas}).
19120 @item -mcustom-fpu-cfg=@var{name}
19121 @opindex mcustom-fpu-cfg
19123 This option enables a predefined, named set of custom instruction encodings
19124 (see @option{-mcustom-@var{insn}} above).
19125 Currently, the following sets are defined:
19127 @option{-mcustom-fpu-cfg=60-1} is equivalent to:
19128 @gccoptlist{-mcustom-fmuls=252 @gol
19129 -mcustom-fadds=253 @gol
19130 -mcustom-fsubs=254 @gol
19131 -fsingle-precision-constant}
19133 @option{-mcustom-fpu-cfg=60-2} is equivalent to:
19134 @gccoptlist{-mcustom-fmuls=252 @gol
19135 -mcustom-fadds=253 @gol
19136 -mcustom-fsubs=254 @gol
19137 -mcustom-fdivs=255 @gol
19138 -fsingle-precision-constant}
19140 @option{-mcustom-fpu-cfg=72-3} is equivalent to:
19141 @gccoptlist{-mcustom-floatus=243 @gol
19142 -mcustom-fixsi=244 @gol
19143 -mcustom-floatis=245 @gol
19144 -mcustom-fcmpgts=246 @gol
19145 -mcustom-fcmples=249 @gol
19146 -mcustom-fcmpeqs=250 @gol
19147 -mcustom-fcmpnes=251 @gol
19148 -mcustom-fmuls=252 @gol
19149 -mcustom-fadds=253 @gol
19150 -mcustom-fsubs=254 @gol
19151 -mcustom-fdivs=255 @gol
19152 -fsingle-precision-constant}
19154 Custom instruction assignments given by individual
19155 @option{-mcustom-@var{insn}=} options override those given by
19156 @option{-mcustom-fpu-cfg=}, regardless of the
19157 order of the options on the command line.
19159 Note that you can gain more local control over selection of a FPU
19160 configuration by using the @code{target("custom-fpu-cfg=@var{name}")}
19161 function attribute (@pxref{Function Attributes})
19162 or pragma (@pxref{Function Specific Option Pragmas}).
19166 These additional @samp{-m} options are available for the Altera Nios II
19167 ELF (bare-metal) target:
19173 Link with HAL BSP. This suppresses linking with the GCC-provided C runtime
19174 startup and termination code, and is typically used in conjunction with
19175 @option{-msys-crt0=} to specify the location of the alternate startup code
19176 provided by the HAL BSP.
19180 Link with a limited version of the C library, @option{-lsmallc}, rather than
19183 @item -msys-crt0=@var{startfile}
19185 @var{startfile} is the file name of the startfile (crt0) to use
19186 when linking. This option is only useful in conjunction with @option{-mhal}.
19188 @item -msys-lib=@var{systemlib}
19190 @var{systemlib} is the library name of the library that provides
19191 low-level system calls required by the C library,
19192 e.g. @code{read} and @code{write}.
19193 This option is typically used to link with a library provided by a HAL BSP.
19197 @node PDP-11 Options
19198 @subsection PDP-11 Options
19199 @cindex PDP-11 Options
19201 These options are defined for the PDP-11:
19206 Use hardware FPP floating point. This is the default. (FIS floating
19207 point on the PDP-11/40 is not supported.)
19210 @opindex msoft-float
19211 Do not use hardware floating point.
19215 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
19219 Return floating-point results in memory. This is the default.
19223 Generate code for a PDP-11/40.
19227 Generate code for a PDP-11/45. This is the default.
19231 Generate code for a PDP-11/10.
19233 @item -mbcopy-builtin
19234 @opindex mbcopy-builtin
19235 Use inline @code{movmemhi} patterns for copying memory. This is the
19240 Do not use inline @code{movmemhi} patterns for copying memory.
19246 Use 16-bit @code{int}. This is the default.
19252 Use 32-bit @code{int}.
19255 @itemx -mno-float32
19257 @opindex mno-float32
19258 Use 64-bit @code{float}. This is the default.
19261 @itemx -mno-float64
19263 @opindex mno-float64
19264 Use 32-bit @code{float}.
19268 Use @code{abshi2} pattern. This is the default.
19272 Do not use @code{abshi2} pattern.
19274 @item -mbranch-expensive
19275 @opindex mbranch-expensive
19276 Pretend that branches are expensive. This is for experimenting with
19277 code generation only.
19279 @item -mbranch-cheap
19280 @opindex mbranch-cheap
19281 Do not pretend that branches are expensive. This is the default.
19285 Use Unix assembler syntax. This is the default when configured for
19286 @samp{pdp11-*-bsd}.
19290 Use DEC assembler syntax. This is the default when configured for any
19291 PDP-11 target other than @samp{pdp11-*-bsd}.
19294 @node picoChip Options
19295 @subsection picoChip Options
19296 @cindex picoChip options
19298 These @samp{-m} options are defined for picoChip implementations:
19302 @item -mae=@var{ae_type}
19304 Set the instruction set, register set, and instruction scheduling
19305 parameters for array element type @var{ae_type}. Supported values
19306 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
19308 @option{-mae=ANY} selects a completely generic AE type. Code
19309 generated with this option runs on any of the other AE types. The
19310 code is not as efficient as it would be if compiled for a specific
19311 AE type, and some types of operation (e.g., multiplication) do not
19312 work properly on all types of AE.
19314 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
19315 for compiled code, and is the default.
19317 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
19318 option may suffer from poor performance of byte (char) manipulation,
19319 since the DSP AE does not provide hardware support for byte load/stores.
19321 @item -msymbol-as-address
19322 Enable the compiler to directly use a symbol name as an address in a
19323 load/store instruction, without first loading it into a
19324 register. Typically, the use of this option generates larger
19325 programs, which run faster than when the option isn't used. However, the
19326 results vary from program to program, so it is left as a user option,
19327 rather than being permanently enabled.
19329 @item -mno-inefficient-warnings
19330 Disables warnings about the generation of inefficient code. These
19331 warnings can be generated, for example, when compiling code that
19332 performs byte-level memory operations on the MAC AE type. The MAC AE has
19333 no hardware support for byte-level memory operations, so all byte
19334 load/stores must be synthesized from word load/store operations. This is
19335 inefficient and a warning is generated to indicate
19336 that you should rewrite the code to avoid byte operations, or to target
19337 an AE type that has the necessary hardware support. This option disables
19342 @node PowerPC Options
19343 @subsection PowerPC Options
19344 @cindex PowerPC options
19346 These are listed under @xref{RS/6000 and PowerPC Options}.
19349 @subsection RL78 Options
19350 @cindex RL78 Options
19356 Links in additional target libraries to support operation within a
19363 Specifies the type of hardware multiplication support to be used. The
19364 default is @code{none}, which uses software multiplication functions.
19365 The @code{g13} option is for the hardware multiply/divide peripheral
19366 only on the RL78/G13 targets. The @code{rl78} option is for the
19367 standard hardware multiplication defined in the RL78 software manual.
19369 @item -m64bit-doubles
19370 @itemx -m32bit-doubles
19371 @opindex m64bit-doubles
19372 @opindex m32bit-doubles
19373 Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
19374 or 32 bits (@option{-m32bit-doubles}) in size. The default is
19375 @option{-m32bit-doubles}.
19379 @node RS/6000 and PowerPC Options
19380 @subsection IBM RS/6000 and PowerPC Options
19381 @cindex RS/6000 and PowerPC Options
19382 @cindex IBM RS/6000 and PowerPC Options
19384 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
19386 @item -mpowerpc-gpopt
19387 @itemx -mno-powerpc-gpopt
19388 @itemx -mpowerpc-gfxopt
19389 @itemx -mno-powerpc-gfxopt
19392 @itemx -mno-powerpc64
19396 @itemx -mno-popcntb
19398 @itemx -mno-popcntd
19407 @itemx -mno-hard-dfp
19408 @opindex mpowerpc-gpopt
19409 @opindex mno-powerpc-gpopt
19410 @opindex mpowerpc-gfxopt
19411 @opindex mno-powerpc-gfxopt
19412 @opindex mpowerpc64
19413 @opindex mno-powerpc64
19417 @opindex mno-popcntb
19419 @opindex mno-popcntd
19425 @opindex mno-mfpgpr
19427 @opindex mno-hard-dfp
19428 You use these options to specify which instructions are available on the
19429 processor you are using. The default value of these options is
19430 determined when configuring GCC@. Specifying the
19431 @option{-mcpu=@var{cpu_type}} overrides the specification of these
19432 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
19433 rather than the options listed above.
19435 Specifying @option{-mpowerpc-gpopt} allows
19436 GCC to use the optional PowerPC architecture instructions in the
19437 General Purpose group, including floating-point square root. Specifying
19438 @option{-mpowerpc-gfxopt} allows GCC to
19439 use the optional PowerPC architecture instructions in the Graphics
19440 group, including floating-point select.
19442 The @option{-mmfcrf} option allows GCC to generate the move from
19443 condition register field instruction implemented on the POWER4
19444 processor and other processors that support the PowerPC V2.01
19446 The @option{-mpopcntb} option allows GCC to generate the popcount and
19447 double-precision FP reciprocal estimate instruction implemented on the
19448 POWER5 processor and other processors that support the PowerPC V2.02
19450 The @option{-mpopcntd} option allows GCC to generate the popcount
19451 instruction implemented on the POWER7 processor and other processors
19452 that support the PowerPC V2.06 architecture.
19453 The @option{-mfprnd} option allows GCC to generate the FP round to
19454 integer instructions implemented on the POWER5+ processor and other
19455 processors that support the PowerPC V2.03 architecture.
19456 The @option{-mcmpb} option allows GCC to generate the compare bytes
19457 instruction implemented on the POWER6 processor and other processors
19458 that support the PowerPC V2.05 architecture.
19459 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
19460 general-purpose register instructions implemented on the POWER6X
19461 processor and other processors that support the extended PowerPC V2.05
19463 The @option{-mhard-dfp} option allows GCC to generate the decimal
19464 floating-point instructions implemented on some POWER processors.
19466 The @option{-mpowerpc64} option allows GCC to generate the additional
19467 64-bit instructions that are found in the full PowerPC64 architecture
19468 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
19469 @option{-mno-powerpc64}.
19471 @item -mcpu=@var{cpu_type}
19473 Set architecture type, register usage, and
19474 instruction scheduling parameters for machine type @var{cpu_type}.
19475 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
19476 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
19477 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
19478 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
19479 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
19480 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
19481 @samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{e5500},
19482 @samp{e6500}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
19483 @samp{titan}, @samp{power3}, @samp{power4}, @samp{power5}, @samp{power5+},
19484 @samp{power6}, @samp{power6x}, @samp{power7}, @samp{power8}, @samp{powerpc},
19485 @samp{powerpc64}, and @samp{rs64}.
19487 @option{-mcpu=powerpc}, and @option{-mcpu=powerpc64} specify pure 32-bit
19488 PowerPC and 64-bit PowerPC architecture machine
19489 types, with an appropriate, generic processor model assumed for
19490 scheduling purposes.
19492 The other options specify a specific processor. Code generated under
19493 those options runs best on that processor, and may not run at all on
19496 The @option{-mcpu} options automatically enable or disable the
19499 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
19500 -mpopcntb -mpopcntd -mpowerpc64 @gol
19501 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
19502 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx @gol
19503 -mcrypto -mdirect-move -mpower8-fusion -mpower8-vector @gol
19504 -mquad-memory -mquad-memory-atomic}
19506 The particular options set for any particular CPU varies between
19507 compiler versions, depending on what setting seems to produce optimal
19508 code for that CPU; it doesn't necessarily reflect the actual hardware's
19509 capabilities. If you wish to set an individual option to a particular
19510 value, you may specify it after the @option{-mcpu} option, like
19511 @option{-mcpu=970 -mno-altivec}.
19513 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
19514 not enabled or disabled by the @option{-mcpu} option at present because
19515 AIX does not have full support for these options. You may still
19516 enable or disable them individually if you're sure it'll work in your
19519 @item -mtune=@var{cpu_type}
19521 Set the instruction scheduling parameters for machine type
19522 @var{cpu_type}, but do not set the architecture type or register usage,
19523 as @option{-mcpu=@var{cpu_type}} does. The same
19524 values for @var{cpu_type} are used for @option{-mtune} as for
19525 @option{-mcpu}. If both are specified, the code generated uses the
19526 architecture and registers set by @option{-mcpu}, but the
19527 scheduling parameters set by @option{-mtune}.
19529 @item -mcmodel=small
19530 @opindex mcmodel=small
19531 Generate PowerPC64 code for the small model: The TOC is limited to
19534 @item -mcmodel=medium
19535 @opindex mcmodel=medium
19536 Generate PowerPC64 code for the medium model: The TOC and other static
19537 data may be up to a total of 4G in size.
19539 @item -mcmodel=large
19540 @opindex mcmodel=large
19541 Generate PowerPC64 code for the large model: The TOC may be up to 4G
19542 in size. Other data and code is only limited by the 64-bit address
19546 @itemx -mno-altivec
19548 @opindex mno-altivec
19549 Generate code that uses (does not use) AltiVec instructions, and also
19550 enable the use of built-in functions that allow more direct access to
19551 the AltiVec instruction set. You may also need to set
19552 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
19555 When @option{-maltivec} is used, rather than @option{-maltivec=le} or
19556 @option{-maltivec=be}, the element order for Altivec intrinsics such
19557 as @code{vec_splat}, @code{vec_extract}, and @code{vec_insert} will
19558 match array element order corresponding to the endianness of the
19559 target. That is, element zero identifies the leftmost element in a
19560 vector register when targeting a big-endian platform, and identifies
19561 the rightmost element in a vector register when targeting a
19562 little-endian platform.
19565 @opindex maltivec=be
19566 Generate Altivec instructions using big-endian element order,
19567 regardless of whether the target is big- or little-endian. This is
19568 the default when targeting a big-endian platform.
19570 The element order is used to interpret element numbers in Altivec
19571 intrinsics such as @code{vec_splat}, @code{vec_extract}, and
19572 @code{vec_insert}. By default, these will match array element order
19573 corresponding to the endianness for the target.
19576 @opindex maltivec=le
19577 Generate Altivec instructions using little-endian element order,
19578 regardless of whether the target is big- or little-endian. This is
19579 the default when targeting a little-endian platform. This option is
19580 currently ignored when targeting a big-endian platform.
19582 The element order is used to interpret element numbers in Altivec
19583 intrinsics such as @code{vec_splat}, @code{vec_extract}, and
19584 @code{vec_insert}. By default, these will match array element order
19585 corresponding to the endianness for the target.
19590 @opindex mno-vrsave
19591 Generate VRSAVE instructions when generating AltiVec code.
19593 @item -mgen-cell-microcode
19594 @opindex mgen-cell-microcode
19595 Generate Cell microcode instructions.
19597 @item -mwarn-cell-microcode
19598 @opindex mwarn-cell-microcode
19599 Warn when a Cell microcode instruction is emitted. An example
19600 of a Cell microcode instruction is a variable shift.
19603 @opindex msecure-plt
19604 Generate code that allows @command{ld} and @command{ld.so}
19605 to build executables and shared
19606 libraries with non-executable @code{.plt} and @code{.got} sections.
19608 32-bit SYSV ABI option.
19612 Generate code that uses a BSS @code{.plt} section that @command{ld.so}
19614 requires @code{.plt} and @code{.got}
19615 sections that are both writable and executable.
19616 This is a PowerPC 32-bit SYSV ABI option.
19622 This switch enables or disables the generation of ISEL instructions.
19624 @item -misel=@var{yes/no}
19625 This switch has been deprecated. Use @option{-misel} and
19626 @option{-mno-isel} instead.
19632 This switch enables or disables the generation of SPE simd
19638 @opindex mno-paired
19639 This switch enables or disables the generation of PAIRED simd
19642 @item -mspe=@var{yes/no}
19643 This option has been deprecated. Use @option{-mspe} and
19644 @option{-mno-spe} instead.
19650 Generate code that uses (does not use) vector/scalar (VSX)
19651 instructions, and also enable the use of built-in functions that allow
19652 more direct access to the VSX instruction set.
19657 @opindex mno-crypto
19658 Enable the use (disable) of the built-in functions that allow direct
19659 access to the cryptographic instructions that were added in version
19660 2.07 of the PowerPC ISA.
19662 @item -mdirect-move
19663 @itemx -mno-direct-move
19664 @opindex mdirect-move
19665 @opindex mno-direct-move
19666 Generate code that uses (does not use) the instructions to move data
19667 between the general purpose registers and the vector/scalar (VSX)
19668 registers that were added in version 2.07 of the PowerPC ISA.
19670 @item -mpower8-fusion
19671 @itemx -mno-power8-fusion
19672 @opindex mpower8-fusion
19673 @opindex mno-power8-fusion
19674 Generate code that keeps (does not keeps) some integer operations
19675 adjacent so that the instructions can be fused together on power8 and
19678 @item -mpower8-vector
19679 @itemx -mno-power8-vector
19680 @opindex mpower8-vector
19681 @opindex mno-power8-vector
19682 Generate code that uses (does not use) the vector and scalar
19683 instructions that were added in version 2.07 of the PowerPC ISA. Also
19684 enable the use of built-in functions that allow more direct access to
19685 the vector instructions.
19687 @item -mquad-memory
19688 @itemx -mno-quad-memory
19689 @opindex mquad-memory
19690 @opindex mno-quad-memory
19691 Generate code that uses (does not use) the non-atomic quad word memory
19692 instructions. The @option{-mquad-memory} option requires use of
19695 @item -mquad-memory-atomic
19696 @itemx -mno-quad-memory-atomic
19697 @opindex mquad-memory-atomic
19698 @opindex mno-quad-memory-atomic
19699 Generate code that uses (does not use) the atomic quad word memory
19700 instructions. The @option{-mquad-memory-atomic} option requires use of
19703 @item -mfloat-gprs=@var{yes/single/double/no}
19704 @itemx -mfloat-gprs
19705 @opindex mfloat-gprs
19706 This switch enables or disables the generation of floating-point
19707 operations on the general-purpose registers for architectures that
19710 The argument @var{yes} or @var{single} enables the use of
19711 single-precision floating-point operations.
19713 The argument @var{double} enables the use of single and
19714 double-precision floating-point operations.
19716 The argument @var{no} disables floating-point operations on the
19717 general-purpose registers.
19719 This option is currently only available on the MPC854x.
19725 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
19726 targets (including GNU/Linux). The 32-bit environment sets int, long
19727 and pointer to 32 bits and generates code that runs on any PowerPC
19728 variant. The 64-bit environment sets int to 32 bits and long and
19729 pointer to 64 bits, and generates code for PowerPC64, as for
19730 @option{-mpowerpc64}.
19733 @itemx -mno-fp-in-toc
19734 @itemx -mno-sum-in-toc
19735 @itemx -mminimal-toc
19737 @opindex mno-fp-in-toc
19738 @opindex mno-sum-in-toc
19739 @opindex mminimal-toc
19740 Modify generation of the TOC (Table Of Contents), which is created for
19741 every executable file. The @option{-mfull-toc} option is selected by
19742 default. In that case, GCC allocates at least one TOC entry for
19743 each unique non-automatic variable reference in your program. GCC
19744 also places floating-point constants in the TOC@. However, only
19745 16,384 entries are available in the TOC@.
19747 If you receive a linker error message that saying you have overflowed
19748 the available TOC space, you can reduce the amount of TOC space used
19749 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
19750 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
19751 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
19752 generate code to calculate the sum of an address and a constant at
19753 run time instead of putting that sum into the TOC@. You may specify one
19754 or both of these options. Each causes GCC to produce very slightly
19755 slower and larger code at the expense of conserving TOC space.
19757 If you still run out of space in the TOC even when you specify both of
19758 these options, specify @option{-mminimal-toc} instead. This option causes
19759 GCC to make only one TOC entry for every file. When you specify this
19760 option, GCC produces code that is slower and larger but which
19761 uses extremely little TOC space. You may wish to use this option
19762 only on files that contain less frequently-executed code.
19768 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
19769 @code{long} type, and the infrastructure needed to support them.
19770 Specifying @option{-maix64} implies @option{-mpowerpc64},
19771 while @option{-maix32} disables the 64-bit ABI and
19772 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
19775 @itemx -mno-xl-compat
19776 @opindex mxl-compat
19777 @opindex mno-xl-compat
19778 Produce code that conforms more closely to IBM XL compiler semantics
19779 when using AIX-compatible ABI@. Pass floating-point arguments to
19780 prototyped functions beyond the register save area (RSA) on the stack
19781 in addition to argument FPRs. Do not assume that most significant
19782 double in 128-bit long double value is properly rounded when comparing
19783 values and converting to double. Use XL symbol names for long double
19786 The AIX calling convention was extended but not initially documented to
19787 handle an obscure K&R C case of calling a function that takes the
19788 address of its arguments with fewer arguments than declared. IBM XL
19789 compilers access floating-point arguments that do not fit in the
19790 RSA from the stack when a subroutine is compiled without
19791 optimization. Because always storing floating-point arguments on the
19792 stack is inefficient and rarely needed, this option is not enabled by
19793 default and only is necessary when calling subroutines compiled by IBM
19794 XL compilers without optimization.
19798 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
19799 application written to use message passing with special startup code to
19800 enable the application to run. The system must have PE installed in the
19801 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
19802 must be overridden with the @option{-specs=} option to specify the
19803 appropriate directory location. The Parallel Environment does not
19804 support threads, so the @option{-mpe} option and the @option{-pthread}
19805 option are incompatible.
19807 @item -malign-natural
19808 @itemx -malign-power
19809 @opindex malign-natural
19810 @opindex malign-power
19811 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
19812 @option{-malign-natural} overrides the ABI-defined alignment of larger
19813 types, such as floating-point doubles, on their natural size-based boundary.
19814 The option @option{-malign-power} instructs GCC to follow the ABI-specified
19815 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
19817 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
19821 @itemx -mhard-float
19822 @opindex msoft-float
19823 @opindex mhard-float
19824 Generate code that does not use (uses) the floating-point register set.
19825 Software floating-point emulation is provided if you use the
19826 @option{-msoft-float} option, and pass the option to GCC when linking.
19828 @item -msingle-float
19829 @itemx -mdouble-float
19830 @opindex msingle-float
19831 @opindex mdouble-float
19832 Generate code for single- or double-precision floating-point operations.
19833 @option{-mdouble-float} implies @option{-msingle-float}.
19836 @opindex msimple-fpu
19837 Do not generate @code{sqrt} and @code{div} instructions for hardware
19838 floating-point unit.
19840 @item -mfpu=@var{name}
19842 Specify type of floating-point unit. Valid values for @var{name} are
19843 @samp{sp_lite} (equivalent to @option{-msingle-float -msimple-fpu}),
19844 @samp{dp_lite} (equivalent to @option{-mdouble-float -msimple-fpu}),
19845 @samp{sp_full} (equivalent to @option{-msingle-float}),
19846 and @samp{dp_full} (equivalent to @option{-mdouble-float}).
19849 @opindex mxilinx-fpu
19850 Perform optimizations for the floating-point unit on Xilinx PPC 405/440.
19853 @itemx -mno-multiple
19855 @opindex mno-multiple
19856 Generate code that uses (does not use) the load multiple word
19857 instructions and the store multiple word instructions. These
19858 instructions are generated by default on POWER systems, and not
19859 generated on PowerPC systems. Do not use @option{-mmultiple} on little-endian
19860 PowerPC systems, since those instructions do not work when the
19861 processor is in little-endian mode. The exceptions are PPC740 and
19862 PPC750 which permit these instructions in little-endian mode.
19867 @opindex mno-string
19868 Generate code that uses (does not use) the load string instructions
19869 and the store string word instructions to save multiple registers and
19870 do small block moves. These instructions are generated by default on
19871 POWER systems, and not generated on PowerPC systems. Do not use
19872 @option{-mstring} on little-endian PowerPC systems, since those
19873 instructions do not work when the processor is in little-endian mode.
19874 The exceptions are PPC740 and PPC750 which permit these instructions
19875 in little-endian mode.
19880 @opindex mno-update
19881 Generate code that uses (does not use) the load or store instructions
19882 that update the base register to the address of the calculated memory
19883 location. These instructions are generated by default. If you use
19884 @option{-mno-update}, there is a small window between the time that the
19885 stack pointer is updated and the address of the previous frame is
19886 stored, which means code that walks the stack frame across interrupts or
19887 signals may get corrupted data.
19889 @item -mavoid-indexed-addresses
19890 @itemx -mno-avoid-indexed-addresses
19891 @opindex mavoid-indexed-addresses
19892 @opindex mno-avoid-indexed-addresses
19893 Generate code that tries to avoid (not avoid) the use of indexed load
19894 or store instructions. These instructions can incur a performance
19895 penalty on Power6 processors in certain situations, such as when
19896 stepping through large arrays that cross a 16M boundary. This option
19897 is enabled by default when targeting Power6 and disabled otherwise.
19900 @itemx -mno-fused-madd
19901 @opindex mfused-madd
19902 @opindex mno-fused-madd
19903 Generate code that uses (does not use) the floating-point multiply and
19904 accumulate instructions. These instructions are generated by default
19905 if hardware floating point is used. The machine-dependent
19906 @option{-mfused-madd} option is now mapped to the machine-independent
19907 @option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
19908 mapped to @option{-ffp-contract=off}.
19914 Generate code that uses (does not use) the half-word multiply and
19915 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
19916 These instructions are generated by default when targeting those
19923 Generate code that uses (does not use) the string-search @samp{dlmzb}
19924 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
19925 generated by default when targeting those processors.
19927 @item -mno-bit-align
19929 @opindex mno-bit-align
19930 @opindex mbit-align
19931 On System V.4 and embedded PowerPC systems do not (do) force structures
19932 and unions that contain bit-fields to be aligned to the base type of the
19935 For example, by default a structure containing nothing but 8
19936 @code{unsigned} bit-fields of length 1 is aligned to a 4-byte
19937 boundary and has a size of 4 bytes. By using @option{-mno-bit-align},
19938 the structure is aligned to a 1-byte boundary and is 1 byte in
19941 @item -mno-strict-align
19942 @itemx -mstrict-align
19943 @opindex mno-strict-align
19944 @opindex mstrict-align
19945 On System V.4 and embedded PowerPC systems do not (do) assume that
19946 unaligned memory references are handled by the system.
19948 @item -mrelocatable
19949 @itemx -mno-relocatable
19950 @opindex mrelocatable
19951 @opindex mno-relocatable
19952 Generate code that allows (does not allow) a static executable to be
19953 relocated to a different address at run time. A simple embedded
19954 PowerPC system loader should relocate the entire contents of
19955 @code{.got2} and 4-byte locations listed in the @code{.fixup} section,
19956 a table of 32-bit addresses generated by this option. For this to
19957 work, all objects linked together must be compiled with
19958 @option{-mrelocatable} or @option{-mrelocatable-lib}.
19959 @option{-mrelocatable} code aligns the stack to an 8-byte boundary.
19961 @item -mrelocatable-lib
19962 @itemx -mno-relocatable-lib
19963 @opindex mrelocatable-lib
19964 @opindex mno-relocatable-lib
19965 Like @option{-mrelocatable}, @option{-mrelocatable-lib} generates a
19966 @code{.fixup} section to allow static executables to be relocated at
19967 run time, but @option{-mrelocatable-lib} does not use the smaller stack
19968 alignment of @option{-mrelocatable}. Objects compiled with
19969 @option{-mrelocatable-lib} may be linked with objects compiled with
19970 any combination of the @option{-mrelocatable} options.
19976 On System V.4 and embedded PowerPC systems do not (do) assume that
19977 register 2 contains a pointer to a global area pointing to the addresses
19978 used in the program.
19981 @itemx -mlittle-endian
19983 @opindex mlittle-endian
19984 On System V.4 and embedded PowerPC systems compile code for the
19985 processor in little-endian mode. The @option{-mlittle-endian} option is
19986 the same as @option{-mlittle}.
19989 @itemx -mbig-endian
19991 @opindex mbig-endian
19992 On System V.4 and embedded PowerPC systems compile code for the
19993 processor in big-endian mode. The @option{-mbig-endian} option is
19994 the same as @option{-mbig}.
19996 @item -mdynamic-no-pic
19997 @opindex mdynamic-no-pic
19998 On Darwin and Mac OS X systems, compile code so that it is not
19999 relocatable, but that its external references are relocatable. The
20000 resulting code is suitable for applications, but not shared
20003 @item -msingle-pic-base
20004 @opindex msingle-pic-base
20005 Treat the register used for PIC addressing as read-only, rather than
20006 loading it in the prologue for each function. The runtime system is
20007 responsible for initializing this register with an appropriate value
20008 before execution begins.
20010 @item -mprioritize-restricted-insns=@var{priority}
20011 @opindex mprioritize-restricted-insns
20012 This option controls the priority that is assigned to
20013 dispatch-slot restricted instructions during the second scheduling
20014 pass. The argument @var{priority} takes the value @samp{0}, @samp{1},
20015 or @samp{2} to assign no, highest, or second-highest (respectively)
20016 priority to dispatch-slot restricted
20019 @item -msched-costly-dep=@var{dependence_type}
20020 @opindex msched-costly-dep
20021 This option controls which dependences are considered costly
20022 by the target during instruction scheduling. The argument
20023 @var{dependence_type} takes one of the following values:
20027 No dependence is costly.
20030 All dependences are costly.
20032 @item @samp{true_store_to_load}
20033 A true dependence from store to load is costly.
20035 @item @samp{store_to_load}
20036 Any dependence from store to load is costly.
20039 Any dependence for which the latency is greater than or equal to
20040 @var{number} is costly.
20043 @item -minsert-sched-nops=@var{scheme}
20044 @opindex minsert-sched-nops
20045 This option controls which NOP insertion scheme is used during
20046 the second scheduling pass. The argument @var{scheme} takes one of the
20054 Pad with NOPs any dispatch group that has vacant issue slots,
20055 according to the scheduler's grouping.
20057 @item @samp{regroup_exact}
20058 Insert NOPs to force costly dependent insns into
20059 separate groups. Insert exactly as many NOPs as needed to force an insn
20060 to a new group, according to the estimated processor grouping.
20063 Insert NOPs to force costly dependent insns into
20064 separate groups. Insert @var{number} NOPs to force an insn to a new group.
20068 @opindex mcall-sysv
20069 On System V.4 and embedded PowerPC systems compile code using calling
20070 conventions that adhere to the March 1995 draft of the System V
20071 Application Binary Interface, PowerPC processor supplement. This is the
20072 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
20074 @item -mcall-sysv-eabi
20076 @opindex mcall-sysv-eabi
20077 @opindex mcall-eabi
20078 Specify both @option{-mcall-sysv} and @option{-meabi} options.
20080 @item -mcall-sysv-noeabi
20081 @opindex mcall-sysv-noeabi
20082 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
20084 @item -mcall-aixdesc
20086 On System V.4 and embedded PowerPC systems compile code for the AIX
20090 @opindex mcall-linux
20091 On System V.4 and embedded PowerPC systems compile code for the
20092 Linux-based GNU system.
20094 @item -mcall-freebsd
20095 @opindex mcall-freebsd
20096 On System V.4 and embedded PowerPC systems compile code for the
20097 FreeBSD operating system.
20099 @item -mcall-netbsd
20100 @opindex mcall-netbsd
20101 On System V.4 and embedded PowerPC systems compile code for the
20102 NetBSD operating system.
20104 @item -mcall-openbsd
20105 @opindex mcall-netbsd
20106 On System V.4 and embedded PowerPC systems compile code for the
20107 OpenBSD operating system.
20109 @item -maix-struct-return
20110 @opindex maix-struct-return
20111 Return all structures in memory (as specified by the AIX ABI)@.
20113 @item -msvr4-struct-return
20114 @opindex msvr4-struct-return
20115 Return structures smaller than 8 bytes in registers (as specified by the
20118 @item -mabi=@var{abi-type}
20120 Extend the current ABI with a particular extension, or remove such extension.
20121 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
20122 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble},
20123 @var{elfv1}, @var{elfv2}@.
20127 Extend the current ABI with SPE ABI extensions. This does not change
20128 the default ABI, instead it adds the SPE ABI extensions to the current
20132 @opindex mabi=no-spe
20133 Disable Book-E SPE ABI extensions for the current ABI@.
20135 @item -mabi=ibmlongdouble
20136 @opindex mabi=ibmlongdouble
20137 Change the current ABI to use IBM extended-precision long double.
20138 This is a PowerPC 32-bit SYSV ABI option.
20140 @item -mabi=ieeelongdouble
20141 @opindex mabi=ieeelongdouble
20142 Change the current ABI to use IEEE extended-precision long double.
20143 This is a PowerPC 32-bit Linux ABI option.
20146 @opindex mabi=elfv1
20147 Change the current ABI to use the ELFv1 ABI.
20148 This is the default ABI for big-endian PowerPC 64-bit Linux.
20149 Overriding the default ABI requires special system support and is
20150 likely to fail in spectacular ways.
20153 @opindex mabi=elfv2
20154 Change the current ABI to use the ELFv2 ABI.
20155 This is the default ABI for little-endian PowerPC 64-bit Linux.
20156 Overriding the default ABI requires special system support and is
20157 likely to fail in spectacular ways.
20160 @itemx -mno-prototype
20161 @opindex mprototype
20162 @opindex mno-prototype
20163 On System V.4 and embedded PowerPC systems assume that all calls to
20164 variable argument functions are properly prototyped. Otherwise, the
20165 compiler must insert an instruction before every non-prototyped call to
20166 set or clear bit 6 of the condition code register (@var{CR}) to
20167 indicate whether floating-point values are passed in the floating-point
20168 registers in case the function takes variable arguments. With
20169 @option{-mprototype}, only calls to prototyped variable argument functions
20170 set or clear the bit.
20174 On embedded PowerPC systems, assume that the startup module is called
20175 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
20176 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
20181 On embedded PowerPC systems, assume that the startup module is called
20182 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
20187 On embedded PowerPC systems, assume that the startup module is called
20188 @file{crt0.o} and the standard C libraries are @file{libads.a} and
20191 @item -myellowknife
20192 @opindex myellowknife
20193 On embedded PowerPC systems, assume that the startup module is called
20194 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
20199 On System V.4 and embedded PowerPC systems, specify that you are
20200 compiling for a VxWorks system.
20204 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
20205 header to indicate that @samp{eabi} extended relocations are used.
20211 On System V.4 and embedded PowerPC systems do (do not) adhere to the
20212 Embedded Applications Binary Interface (EABI), which is a set of
20213 modifications to the System V.4 specifications. Selecting @option{-meabi}
20214 means that the stack is aligned to an 8-byte boundary, a function
20215 @code{__eabi} is called from @code{main} to set up the EABI
20216 environment, and the @option{-msdata} option can use both @code{r2} and
20217 @code{r13} to point to two separate small data areas. Selecting
20218 @option{-mno-eabi} means that the stack is aligned to a 16-byte boundary,
20219 no EABI initialization function is called from @code{main}, and the
20220 @option{-msdata} option only uses @code{r13} to point to a single
20221 small data area. The @option{-meabi} option is on by default if you
20222 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
20225 @opindex msdata=eabi
20226 On System V.4 and embedded PowerPC systems, put small initialized
20227 @code{const} global and static data in the @samp{.sdata2} section, which
20228 is pointed to by register @code{r2}. Put small initialized
20229 non-@code{const} global and static data in the @samp{.sdata} section,
20230 which is pointed to by register @code{r13}. Put small uninitialized
20231 global and static data in the @samp{.sbss} section, which is adjacent to
20232 the @samp{.sdata} section. The @option{-msdata=eabi} option is
20233 incompatible with the @option{-mrelocatable} option. The
20234 @option{-msdata=eabi} option also sets the @option{-memb} option.
20237 @opindex msdata=sysv
20238 On System V.4 and embedded PowerPC systems, put small global and static
20239 data in the @samp{.sdata} section, which is pointed to by register
20240 @code{r13}. Put small uninitialized global and static data in the
20241 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
20242 The @option{-msdata=sysv} option is incompatible with the
20243 @option{-mrelocatable} option.
20245 @item -msdata=default
20247 @opindex msdata=default
20249 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
20250 compile code the same as @option{-msdata=eabi}, otherwise compile code the
20251 same as @option{-msdata=sysv}.
20254 @opindex msdata=data
20255 On System V.4 and embedded PowerPC systems, put small global
20256 data in the @samp{.sdata} section. Put small uninitialized global
20257 data in the @samp{.sbss} section. Do not use register @code{r13}
20258 to address small data however. This is the default behavior unless
20259 other @option{-msdata} options are used.
20263 @opindex msdata=none
20265 On embedded PowerPC systems, put all initialized global and static data
20266 in the @samp{.data} section, and all uninitialized data in the
20267 @samp{.bss} section.
20269 @item -mblock-move-inline-limit=@var{num}
20270 @opindex mblock-move-inline-limit
20271 Inline all block moves (such as calls to @code{memcpy} or structure
20272 copies) less than or equal to @var{num} bytes. The minimum value for
20273 @var{num} is 32 bytes on 32-bit targets and 64 bytes on 64-bit
20274 targets. The default value is target-specific.
20278 @cindex smaller data references (PowerPC)
20279 @cindex .sdata/.sdata2 references (PowerPC)
20280 On embedded PowerPC systems, put global and static items less than or
20281 equal to @var{num} bytes into the small data or BSS sections instead of
20282 the normal data or BSS section. By default, @var{num} is 8. The
20283 @option{-G @var{num}} switch is also passed to the linker.
20284 All modules should be compiled with the same @option{-G @var{num}} value.
20287 @itemx -mno-regnames
20289 @opindex mno-regnames
20290 On System V.4 and embedded PowerPC systems do (do not) emit register
20291 names in the assembly language output using symbolic forms.
20294 @itemx -mno-longcall
20296 @opindex mno-longcall
20297 By default assume that all calls are far away so that a longer and more
20298 expensive calling sequence is required. This is required for calls
20299 farther than 32 megabytes (33,554,432 bytes) from the current location.
20300 A short call is generated if the compiler knows
20301 the call cannot be that far away. This setting can be overridden by
20302 the @code{shortcall} function attribute, or by @code{#pragma
20305 Some linkers are capable of detecting out-of-range calls and generating
20306 glue code on the fly. On these systems, long calls are unnecessary and
20307 generate slower code. As of this writing, the AIX linker can do this,
20308 as can the GNU linker for PowerPC/64. It is planned to add this feature
20309 to the GNU linker for 32-bit PowerPC systems as well.
20311 On Darwin/PPC systems, @code{#pragma longcall} generates @code{jbsr
20312 callee, L42}, plus a @dfn{branch island} (glue code). The two target
20313 addresses represent the callee and the branch island. The
20314 Darwin/PPC linker prefers the first address and generates a @code{bl
20315 callee} if the PPC @code{bl} instruction reaches the callee directly;
20316 otherwise, the linker generates @code{bl L42} to call the branch
20317 island. The branch island is appended to the body of the
20318 calling function; it computes the full 32-bit address of the callee
20321 On Mach-O (Darwin) systems, this option directs the compiler emit to
20322 the glue for every direct call, and the Darwin linker decides whether
20323 to use or discard it.
20325 In the future, GCC may ignore all longcall specifications
20326 when the linker is known to generate glue.
20328 @item -mtls-markers
20329 @itemx -mno-tls-markers
20330 @opindex mtls-markers
20331 @opindex mno-tls-markers
20332 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
20333 specifying the function argument. The relocation allows the linker to
20334 reliably associate function call with argument setup instructions for
20335 TLS optimization, which in turn allows GCC to better schedule the
20340 Adds support for multithreading with the @dfn{pthreads} library.
20341 This option sets flags for both the preprocessor and linker.
20346 This option enables use of the reciprocal estimate and
20347 reciprocal square root estimate instructions with additional
20348 Newton-Raphson steps to increase precision instead of doing a divide or
20349 square root and divide for floating-point arguments. You should use
20350 the @option{-ffast-math} option when using @option{-mrecip} (or at
20351 least @option{-funsafe-math-optimizations},
20352 @option{-finite-math-only}, @option{-freciprocal-math} and
20353 @option{-fno-trapping-math}). Note that while the throughput of the
20354 sequence is generally higher than the throughput of the non-reciprocal
20355 instruction, the precision of the sequence can be decreased by up to 2
20356 ulp (i.e.@: the inverse of 1.0 equals 0.99999994) for reciprocal square
20359 @item -mrecip=@var{opt}
20360 @opindex mrecip=opt
20361 This option controls which reciprocal estimate instructions
20362 may be used. @var{opt} is a comma-separated list of options, which may
20363 be preceded by a @code{!} to invert the option:
20364 @code{all}: enable all estimate instructions,
20365 @code{default}: enable the default instructions, equivalent to @option{-mrecip},
20366 @code{none}: disable all estimate instructions, equivalent to @option{-mno-recip};
20367 @code{div}: enable the reciprocal approximation instructions for both single and double precision;
20368 @code{divf}: enable the single-precision reciprocal approximation instructions;
20369 @code{divd}: enable the double-precision reciprocal approximation instructions;
20370 @code{rsqrt}: enable the reciprocal square root approximation instructions for both single and double precision;
20371 @code{rsqrtf}: enable the single-precision reciprocal square root approximation instructions;
20372 @code{rsqrtd}: enable the double-precision reciprocal square root approximation instructions;
20374 So, for example, @option{-mrecip=all,!rsqrtd} enables
20375 all of the reciprocal estimate instructions, except for the
20376 @code{FRSQRTE}, @code{XSRSQRTEDP}, and @code{XVRSQRTEDP} instructions
20377 which handle the double-precision reciprocal square root calculations.
20379 @item -mrecip-precision
20380 @itemx -mno-recip-precision
20381 @opindex mrecip-precision
20382 Assume (do not assume) that the reciprocal estimate instructions
20383 provide higher-precision estimates than is mandated by the PowerPC
20384 ABI. Selecting @option{-mcpu=power6}, @option{-mcpu=power7} or
20385 @option{-mcpu=power8} automatically selects @option{-mrecip-precision}.
20386 The double-precision square root estimate instructions are not generated by
20387 default on low-precision machines, since they do not provide an
20388 estimate that converges after three steps.
20390 @item -mveclibabi=@var{type}
20391 @opindex mveclibabi
20392 Specifies the ABI type to use for vectorizing intrinsics using an
20393 external library. The only type supported at present is @code{mass},
20394 which specifies to use IBM's Mathematical Acceleration Subsystem
20395 (MASS) libraries for vectorizing intrinsics using external libraries.
20396 GCC currently emits calls to @code{acosd2}, @code{acosf4},
20397 @code{acoshd2}, @code{acoshf4}, @code{asind2}, @code{asinf4},
20398 @code{asinhd2}, @code{asinhf4}, @code{atan2d2}, @code{atan2f4},
20399 @code{atand2}, @code{atanf4}, @code{atanhd2}, @code{atanhf4},
20400 @code{cbrtd2}, @code{cbrtf4}, @code{cosd2}, @code{cosf4},
20401 @code{coshd2}, @code{coshf4}, @code{erfcd2}, @code{erfcf4},
20402 @code{erfd2}, @code{erff4}, @code{exp2d2}, @code{exp2f4},
20403 @code{expd2}, @code{expf4}, @code{expm1d2}, @code{expm1f4},
20404 @code{hypotd2}, @code{hypotf4}, @code{lgammad2}, @code{lgammaf4},
20405 @code{log10d2}, @code{log10f4}, @code{log1pd2}, @code{log1pf4},
20406 @code{log2d2}, @code{log2f4}, @code{logd2}, @code{logf4},
20407 @code{powd2}, @code{powf4}, @code{sind2}, @code{sinf4}, @code{sinhd2},
20408 @code{sinhf4}, @code{sqrtd2}, @code{sqrtf4}, @code{tand2},
20409 @code{tanf4}, @code{tanhd2}, and @code{tanhf4} when generating code
20410 for power7. Both @option{-ftree-vectorize} and
20411 @option{-funsafe-math-optimizations} must also be enabled. The MASS
20412 libraries must be specified at link time.
20417 Generate (do not generate) the @code{friz} instruction when the
20418 @option{-funsafe-math-optimizations} option is used to optimize
20419 rounding of floating-point values to 64-bit integer and back to floating
20420 point. The @code{friz} instruction does not return the same value if
20421 the floating-point number is too large to fit in an integer.
20423 @item -mpointers-to-nested-functions
20424 @itemx -mno-pointers-to-nested-functions
20425 @opindex mpointers-to-nested-functions
20426 Generate (do not generate) code to load up the static chain register
20427 (@var{r11}) when calling through a pointer on AIX and 64-bit Linux
20428 systems where a function pointer points to a 3-word descriptor giving
20429 the function address, TOC value to be loaded in register @var{r2}, and
20430 static chain value to be loaded in register @var{r11}. The
20431 @option{-mpointers-to-nested-functions} is on by default. You cannot
20432 call through pointers to nested functions or pointers
20433 to functions compiled in other languages that use the static chain if
20434 you use the @option{-mno-pointers-to-nested-functions}.
20436 @item -msave-toc-indirect
20437 @itemx -mno-save-toc-indirect
20438 @opindex msave-toc-indirect
20439 Generate (do not generate) code to save the TOC value in the reserved
20440 stack location in the function prologue if the function calls through
20441 a pointer on AIX and 64-bit Linux systems. If the TOC value is not
20442 saved in the prologue, it is saved just before the call through the
20443 pointer. The @option{-mno-save-toc-indirect} option is the default.
20445 @item -mcompat-align-parm
20446 @itemx -mno-compat-align-parm
20447 @opindex mcompat-align-parm
20448 Generate (do not generate) code to pass structure parameters with a
20449 maximum alignment of 64 bits, for compatibility with older versions
20452 Older versions of GCC (prior to 4.9.0) incorrectly did not align a
20453 structure parameter on a 128-bit boundary when that structure contained
20454 a member requiring 128-bit alignment. This is corrected in more
20455 recent versions of GCC. This option may be used to generate code
20456 that is compatible with functions compiled with older versions of
20459 The @option{-mno-compat-align-parm} option is the default.
20463 @subsection RX Options
20466 These command-line options are defined for RX targets:
20469 @item -m64bit-doubles
20470 @itemx -m32bit-doubles
20471 @opindex m64bit-doubles
20472 @opindex m32bit-doubles
20473 Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
20474 or 32 bits (@option{-m32bit-doubles}) in size. The default is
20475 @option{-m32bit-doubles}. @emph{Note} RX floating-point hardware only
20476 works on 32-bit values, which is why the default is
20477 @option{-m32bit-doubles}.
20483 Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
20484 floating-point hardware. The default is enabled for the @var{RX600}
20485 series and disabled for the @var{RX200} series.
20487 Floating-point instructions are only generated for 32-bit floating-point
20488 values, however, so the FPU hardware is not used for doubles if the
20489 @option{-m64bit-doubles} option is used.
20491 @emph{Note} If the @option{-fpu} option is enabled then
20492 @option{-funsafe-math-optimizations} is also enabled automatically.
20493 This is because the RX FPU instructions are themselves unsafe.
20495 @item -mcpu=@var{name}
20497 Selects the type of RX CPU to be targeted. Currently three types are
20498 supported, the generic @var{RX600} and @var{RX200} series hardware and
20499 the specific @var{RX610} CPU. The default is @var{RX600}.
20501 The only difference between @var{RX600} and @var{RX610} is that the
20502 @var{RX610} does not support the @code{MVTIPL} instruction.
20504 The @var{RX200} series does not have a hardware floating-point unit
20505 and so @option{-nofpu} is enabled by default when this type is
20508 @item -mbig-endian-data
20509 @itemx -mlittle-endian-data
20510 @opindex mbig-endian-data
20511 @opindex mlittle-endian-data
20512 Store data (but not code) in the big-endian format. The default is
20513 @option{-mlittle-endian-data}, i.e.@: to store data in the little-endian
20516 @item -msmall-data-limit=@var{N}
20517 @opindex msmall-data-limit
20518 Specifies the maximum size in bytes of global and static variables
20519 which can be placed into the small data area. Using the small data
20520 area can lead to smaller and faster code, but the size of area is
20521 limited and it is up to the programmer to ensure that the area does
20522 not overflow. Also when the small data area is used one of the RX's
20523 registers (usually @code{r13}) is reserved for use pointing to this
20524 area, so it is no longer available for use by the compiler. This
20525 could result in slower and/or larger code if variables are pushed onto
20526 the stack instead of being held in this register.
20528 Note, common variables (variables that have not been initialized) and
20529 constants are not placed into the small data area as they are assigned
20530 to other sections in the output executable.
20532 The default value is zero, which disables this feature. Note, this
20533 feature is not enabled by default with higher optimization levels
20534 (@option{-O2} etc) because of the potentially detrimental effects of
20535 reserving a register. It is up to the programmer to experiment and
20536 discover whether this feature is of benefit to their program. See the
20537 description of the @option{-mpid} option for a description of how the
20538 actual register to hold the small data area pointer is chosen.
20544 Use the simulator runtime. The default is to use the libgloss
20545 board-specific runtime.
20547 @item -mas100-syntax
20548 @itemx -mno-as100-syntax
20549 @opindex mas100-syntax
20550 @opindex mno-as100-syntax
20551 When generating assembler output use a syntax that is compatible with
20552 Renesas's AS100 assembler. This syntax can also be handled by the GAS
20553 assembler, but it has some restrictions so it is not generated by default.
20555 @item -mmax-constant-size=@var{N}
20556 @opindex mmax-constant-size
20557 Specifies the maximum size, in bytes, of a constant that can be used as
20558 an operand in a RX instruction. Although the RX instruction set does
20559 allow constants of up to 4 bytes in length to be used in instructions,
20560 a longer value equates to a longer instruction. Thus in some
20561 circumstances it can be beneficial to restrict the size of constants
20562 that are used in instructions. Constants that are too big are instead
20563 placed into a constant pool and referenced via register indirection.
20565 The value @var{N} can be between 0 and 4. A value of 0 (the default)
20566 or 4 means that constants of any size are allowed.
20570 Enable linker relaxation. Linker relaxation is a process whereby the
20571 linker attempts to reduce the size of a program by finding shorter
20572 versions of various instructions. Disabled by default.
20574 @item -mint-register=@var{N}
20575 @opindex mint-register
20576 Specify the number of registers to reserve for fast interrupt handler
20577 functions. The value @var{N} can be between 0 and 4. A value of 1
20578 means that register @code{r13} is reserved for the exclusive use
20579 of fast interrupt handlers. A value of 2 reserves @code{r13} and
20580 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
20581 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
20582 A value of 0, the default, does not reserve any registers.
20584 @item -msave-acc-in-interrupts
20585 @opindex msave-acc-in-interrupts
20586 Specifies that interrupt handler functions should preserve the
20587 accumulator register. This is only necessary if normal code might use
20588 the accumulator register, for example because it performs 64-bit
20589 multiplications. The default is to ignore the accumulator as this
20590 makes the interrupt handlers faster.
20596 Enables the generation of position independent data. When enabled any
20597 access to constant data is done via an offset from a base address
20598 held in a register. This allows the location of constant data to be
20599 determined at run time without requiring the executable to be
20600 relocated, which is a benefit to embedded applications with tight
20601 memory constraints. Data that can be modified is not affected by this
20604 Note, using this feature reserves a register, usually @code{r13}, for
20605 the constant data base address. This can result in slower and/or
20606 larger code, especially in complicated functions.
20608 The actual register chosen to hold the constant data base address
20609 depends upon whether the @option{-msmall-data-limit} and/or the
20610 @option{-mint-register} command-line options are enabled. Starting
20611 with register @code{r13} and proceeding downwards, registers are
20612 allocated first to satisfy the requirements of @option{-mint-register},
20613 then @option{-mpid} and finally @option{-msmall-data-limit}. Thus it
20614 is possible for the small data area register to be @code{r8} if both
20615 @option{-mint-register=4} and @option{-mpid} are specified on the
20618 By default this feature is not enabled. The default can be restored
20619 via the @option{-mno-pid} command-line option.
20621 @item -mno-warn-multiple-fast-interrupts
20622 @itemx -mwarn-multiple-fast-interrupts
20623 @opindex mno-warn-multiple-fast-interrupts
20624 @opindex mwarn-multiple-fast-interrupts
20625 Prevents GCC from issuing a warning message if it finds more than one
20626 fast interrupt handler when it is compiling a file. The default is to
20627 issue a warning for each extra fast interrupt handler found, as the RX
20628 only supports one such interrupt.
20632 @emph{Note:} The generic GCC command-line option @option{-ffixed-@var{reg}}
20633 has special significance to the RX port when used with the
20634 @code{interrupt} function attribute. This attribute indicates a
20635 function intended to process fast interrupts. GCC ensures
20636 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
20637 and/or @code{r13} and only provided that the normal use of the
20638 corresponding registers have been restricted via the
20639 @option{-ffixed-@var{reg}} or @option{-mint-register} command-line
20642 @node S/390 and zSeries Options
20643 @subsection S/390 and zSeries Options
20644 @cindex S/390 and zSeries Options
20646 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
20650 @itemx -msoft-float
20651 @opindex mhard-float
20652 @opindex msoft-float
20653 Use (do not use) the hardware floating-point instructions and registers
20654 for floating-point operations. When @option{-msoft-float} is specified,
20655 functions in @file{libgcc.a} are used to perform floating-point
20656 operations. When @option{-mhard-float} is specified, the compiler
20657 generates IEEE floating-point instructions. This is the default.
20660 @itemx -mno-hard-dfp
20662 @opindex mno-hard-dfp
20663 Use (do not use) the hardware decimal-floating-point instructions for
20664 decimal-floating-point operations. When @option{-mno-hard-dfp} is
20665 specified, functions in @file{libgcc.a} are used to perform
20666 decimal-floating-point operations. When @option{-mhard-dfp} is
20667 specified, the compiler generates decimal-floating-point hardware
20668 instructions. This is the default for @option{-march=z9-ec} or higher.
20670 @item -mlong-double-64
20671 @itemx -mlong-double-128
20672 @opindex mlong-double-64
20673 @opindex mlong-double-128
20674 These switches control the size of @code{long double} type. A size
20675 of 64 bits makes the @code{long double} type equivalent to the @code{double}
20676 type. This is the default.
20679 @itemx -mno-backchain
20680 @opindex mbackchain
20681 @opindex mno-backchain
20682 Store (do not store) the address of the caller's frame as backchain pointer
20683 into the callee's stack frame.
20684 A backchain may be needed to allow debugging using tools that do not understand
20685 DWARF 2 call frame information.
20686 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
20687 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
20688 the backchain is placed into the topmost word of the 96/160 byte register
20691 In general, code compiled with @option{-mbackchain} is call-compatible with
20692 code compiled with @option{-mmo-backchain}; however, use of the backchain
20693 for debugging purposes usually requires that the whole binary is built with
20694 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
20695 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
20696 to build a linux kernel use @option{-msoft-float}.
20698 The default is to not maintain the backchain.
20700 @item -mpacked-stack
20701 @itemx -mno-packed-stack
20702 @opindex mpacked-stack
20703 @opindex mno-packed-stack
20704 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
20705 specified, the compiler uses the all fields of the 96/160 byte register save
20706 area only for their default purpose; unused fields still take up stack space.
20707 When @option{-mpacked-stack} is specified, register save slots are densely
20708 packed at the top of the register save area; unused space is reused for other
20709 purposes, allowing for more efficient use of the available stack space.
20710 However, when @option{-mbackchain} is also in effect, the topmost word of
20711 the save area is always used to store the backchain, and the return address
20712 register is always saved two words below the backchain.
20714 As long as the stack frame backchain is not used, code generated with
20715 @option{-mpacked-stack} is call-compatible with code generated with
20716 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
20717 S/390 or zSeries generated code that uses the stack frame backchain at run
20718 time, not just for debugging purposes. Such code is not call-compatible
20719 with code compiled with @option{-mpacked-stack}. Also, note that the
20720 combination of @option{-mbackchain},
20721 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
20722 to build a linux kernel use @option{-msoft-float}.
20724 The default is to not use the packed stack layout.
20727 @itemx -mno-small-exec
20728 @opindex msmall-exec
20729 @opindex mno-small-exec
20730 Generate (or do not generate) code using the @code{bras} instruction
20731 to do subroutine calls.
20732 This only works reliably if the total executable size does not
20733 exceed 64k. The default is to use the @code{basr} instruction instead,
20734 which does not have this limitation.
20740 When @option{-m31} is specified, generate code compliant to the
20741 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
20742 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
20743 particular to generate 64-bit instructions. For the @samp{s390}
20744 targets, the default is @option{-m31}, while the @samp{s390x}
20745 targets default to @option{-m64}.
20751 When @option{-mzarch} is specified, generate code using the
20752 instructions available on z/Architecture.
20753 When @option{-mesa} is specified, generate code using the
20754 instructions available on ESA/390. Note that @option{-mesa} is
20755 not possible with @option{-m64}.
20756 When generating code compliant to the GNU/Linux for S/390 ABI,
20757 the default is @option{-mesa}. When generating code compliant
20758 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
20764 Generate (or do not generate) code using the @code{mvcle} instruction
20765 to perform block moves. When @option{-mno-mvcle} is specified,
20766 use a @code{mvc} loop instead. This is the default unless optimizing for
20773 Print (or do not print) additional debug information when compiling.
20774 The default is to not print debug information.
20776 @item -march=@var{cpu-type}
20778 Generate code that runs on @var{cpu-type}, which is the name of a system
20779 representing a certain processor type. Possible values for
20780 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
20781 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
20782 When generating code using the instructions available on z/Architecture,
20783 the default is @option{-march=z900}. Otherwise, the default is
20784 @option{-march=g5}.
20786 @item -mtune=@var{cpu-type}
20788 Tune to @var{cpu-type} everything applicable about the generated code,
20789 except for the ABI and the set of available instructions.
20790 The list of @var{cpu-type} values is the same as for @option{-march}.
20791 The default is the value used for @option{-march}.
20794 @itemx -mno-tpf-trace
20795 @opindex mtpf-trace
20796 @opindex mno-tpf-trace
20797 Generate code that adds (does not add) in TPF OS specific branches to trace
20798 routines in the operating system. This option is off by default, even
20799 when compiling for the TPF OS@.
20802 @itemx -mno-fused-madd
20803 @opindex mfused-madd
20804 @opindex mno-fused-madd
20805 Generate code that uses (does not use) the floating-point multiply and
20806 accumulate instructions. These instructions are generated by default if
20807 hardware floating point is used.
20809 @item -mwarn-framesize=@var{framesize}
20810 @opindex mwarn-framesize
20811 Emit a warning if the current function exceeds the given frame size. Because
20812 this is a compile-time check it doesn't need to be a real problem when the program
20813 runs. It is intended to identify functions that most probably cause
20814 a stack overflow. It is useful to be used in an environment with limited stack
20815 size e.g.@: the linux kernel.
20817 @item -mwarn-dynamicstack
20818 @opindex mwarn-dynamicstack
20819 Emit a warning if the function calls @code{alloca} or uses dynamically-sized
20820 arrays. This is generally a bad idea with a limited stack size.
20822 @item -mstack-guard=@var{stack-guard}
20823 @itemx -mstack-size=@var{stack-size}
20824 @opindex mstack-guard
20825 @opindex mstack-size
20826 If these options are provided the S/390 back end emits additional instructions in
20827 the function prologue that trigger a trap if the stack size is @var{stack-guard}
20828 bytes above the @var{stack-size} (remember that the stack on S/390 grows downward).
20829 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
20830 the frame size of the compiled function is chosen.
20831 These options are intended to be used to help debugging stack overflow problems.
20832 The additionally emitted code causes only little overhead and hence can also be
20833 used in production-like systems without greater performance degradation. The given
20834 values have to be exact powers of 2 and @var{stack-size} has to be greater than
20835 @var{stack-guard} without exceeding 64k.
20836 In order to be efficient the extra code makes the assumption that the stack starts
20837 at an address aligned to the value given by @var{stack-size}.
20838 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
20840 @item -mhotpatch[=@var{halfwords}]
20841 @itemx -mno-hotpatch
20843 If the hotpatch option is enabled, a ``hot-patching'' function
20844 prologue is generated for all functions in the compilation unit.
20845 The funtion label is prepended with the given number of two-byte
20846 Nop instructions (@var{halfwords}, maximum 1000000) or 12 Nop
20847 instructions if no argument is present. Functions with a
20848 hot-patching prologue are never inlined automatically, and a
20849 hot-patching prologue is never generated for functions functions
20850 that are explicitly inline.
20852 This option can be overridden for individual functions with the
20853 @code{hotpatch} attribute.
20856 @node Score Options
20857 @subsection Score Options
20858 @cindex Score Options
20860 These options are defined for Score implementations:
20865 Compile code for big-endian mode. This is the default.
20869 Compile code for little-endian mode.
20873 Disable generation of @code{bcnz} instructions.
20877 Enable generation of unaligned load and store instructions.
20881 Enable the use of multiply-accumulate instructions. Disabled by default.
20885 Specify the SCORE5 as the target architecture.
20889 Specify the SCORE5U of the target architecture.
20893 Specify the SCORE7 as the target architecture. This is the default.
20897 Specify the SCORE7D as the target architecture.
20901 @subsection SH Options
20903 These @samp{-m} options are defined for the SH implementations:
20908 Generate code for the SH1.
20912 Generate code for the SH2.
20915 Generate code for the SH2e.
20919 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
20920 that the floating-point unit is not used.
20922 @item -m2a-single-only
20923 @opindex m2a-single-only
20924 Generate code for the SH2a-FPU, in such a way that no double-precision
20925 floating-point operations are used.
20928 @opindex m2a-single
20929 Generate code for the SH2a-FPU assuming the floating-point unit is in
20930 single-precision mode by default.
20934 Generate code for the SH2a-FPU assuming the floating-point unit is in
20935 double-precision mode by default.
20939 Generate code for the SH3.
20943 Generate code for the SH3e.
20947 Generate code for the SH4 without a floating-point unit.
20949 @item -m4-single-only
20950 @opindex m4-single-only
20951 Generate code for the SH4 with a floating-point unit that only
20952 supports single-precision arithmetic.
20956 Generate code for the SH4 assuming the floating-point unit is in
20957 single-precision mode by default.
20961 Generate code for the SH4.
20965 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
20966 floating-point unit is not used.
20968 @item -m4a-single-only
20969 @opindex m4a-single-only
20970 Generate code for the SH4a, in such a way that no double-precision
20971 floating-point operations are used.
20974 @opindex m4a-single
20975 Generate code for the SH4a assuming the floating-point unit is in
20976 single-precision mode by default.
20980 Generate code for the SH4a.
20984 Same as @option{-m4a-nofpu}, except that it implicitly passes
20985 @option{-dsp} to the assembler. GCC doesn't generate any DSP
20986 instructions at the moment.
20990 Compile code for the processor in big-endian mode.
20994 Compile code for the processor in little-endian mode.
20998 Align doubles at 64-bit boundaries. Note that this changes the calling
20999 conventions, and thus some functions from the standard C library do
21000 not work unless you recompile it first with @option{-mdalign}.
21004 Shorten some address references at link time, when possible; uses the
21005 linker option @option{-relax}.
21009 Use 32-bit offsets in @code{switch} tables. The default is to use
21014 Enable the use of bit manipulation instructions on SH2A.
21018 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
21019 alignment constraints.
21023 Comply with the calling conventions defined by Renesas.
21027 Comply with the calling conventions defined by Renesas.
21031 Comply with the calling conventions defined for GCC before the Renesas
21032 conventions were available. This option is the default for all
21033 targets of the SH toolchain.
21036 @opindex mnomacsave
21037 Mark the @code{MAC} register as call-clobbered, even if
21038 @option{-mhitachi} is given.
21044 Control the IEEE compliance of floating-point comparisons, which affects the
21045 handling of cases where the result of a comparison is unordered. By default
21046 @option{-mieee} is implicitly enabled. If @option{-ffinite-math-only} is
21047 enabled @option{-mno-ieee} is implicitly set, which results in faster
21048 floating-point greater-equal and less-equal comparisons. The implcit settings
21049 can be overridden by specifying either @option{-mieee} or @option{-mno-ieee}.
21051 @item -minline-ic_invalidate
21052 @opindex minline-ic_invalidate
21053 Inline code to invalidate instruction cache entries after setting up
21054 nested function trampolines.
21055 This option has no effect if @option{-musermode} is in effect and the selected
21056 code generation option (e.g. @option{-m4}) does not allow the use of the @code{icbi}
21058 If the selected code generation option does not allow the use of the @code{icbi}
21059 instruction, and @option{-musermode} is not in effect, the inlined code
21060 manipulates the instruction cache address array directly with an associative
21061 write. This not only requires privileged mode at run time, but it also
21062 fails if the cache line had been mapped via the TLB and has become unmapped.
21066 Dump instruction size and location in the assembly code.
21069 @opindex mpadstruct
21070 This option is deprecated. It pads structures to multiple of 4 bytes,
21071 which is incompatible with the SH ABI@.
21073 @item -matomic-model=@var{model}
21074 @opindex matomic-model=@var{model}
21075 Sets the model of atomic operations and additional parameters as a comma
21076 separated list. For details on the atomic built-in functions see
21077 @ref{__atomic Builtins}. The following models and parameters are supported:
21082 Disable compiler generated atomic sequences and emit library calls for atomic
21083 operations. This is the default if the target is not @code{sh-*-linux*}.
21086 Generate GNU/Linux compatible gUSA software atomic sequences for the atomic
21087 built-in functions. The generated atomic sequences require additional support
21088 from the interrupt/exception handling code of the system and are only suitable
21089 for SH3* and SH4* single-core systems. This option is enabled by default when
21090 the target is @code{sh-*-linux*} and SH3* or SH4*. When the target is SH4A,
21091 this option will also partially utilize the hardware atomic instructions
21092 @code{movli.l} and @code{movco.l} to create more efficient code, unless
21093 @samp{strict} is specified.
21096 Generate software atomic sequences that use a variable in the thread control
21097 block. This is a variation of the gUSA sequences which can also be used on
21098 SH1* and SH2* targets. The generated atomic sequences require additional
21099 support from the interrupt/exception handling code of the system and are only
21100 suitable for single-core systems. When using this model, the @samp{gbr-offset=}
21101 parameter has to be specified as well.
21104 Generate software atomic sequences that temporarily disable interrupts by
21105 setting @code{SR.IMASK = 1111}. This model works only when the program runs
21106 in privileged mode and is only suitable for single-core systems. Additional
21107 support from the interrupt/exception handling code of the system is not
21108 required. This model is enabled by default when the target is
21109 @code{sh-*-linux*} and SH1* or SH2*.
21112 Generate hardware atomic sequences using the @code{movli.l} and @code{movco.l}
21113 instructions only. This is only available on SH4A and is suitable for
21114 multi-core systems. Since the hardware instructions support only 32 bit atomic
21115 variables access to 8 or 16 bit variables is emulated with 32 bit accesses.
21116 Code compiled with this option will also be compatible with other software
21117 atomic model interrupt/exception handling systems if executed on an SH4A
21118 system. Additional support from the interrupt/exception handling code of the
21119 system is not required for this model.
21122 This parameter specifies the offset in bytes of the variable in the thread
21123 control block structure that should be used by the generated atomic sequences
21124 when the @samp{soft-tcb} model has been selected. For other models this
21125 parameter is ignored. The specified value must be an integer multiple of four
21126 and in the range 0-1020.
21129 This parameter prevents mixed usage of multiple atomic models, even though they
21130 would be compatible, and will make the compiler generate atomic sequences of the
21131 specified model only.
21137 Generate the @code{tas.b} opcode for @code{__atomic_test_and_set}.
21138 Notice that depending on the particular hardware and software configuration
21139 this can degrade overall performance due to the operand cache line flushes
21140 that are implied by the @code{tas.b} instruction. On multi-core SH4A
21141 processors the @code{tas.b} instruction must be used with caution since it
21142 can result in data corruption for certain cache configurations.
21146 Optimize for space instead of speed. Implied by @option{-Os}.
21149 @opindex mprefergot
21150 When generating position-independent code, emit function calls using
21151 the Global Offset Table instead of the Procedure Linkage Table.
21155 Don't generate privileged mode only code. This option
21156 implies @option{-mno-inline-ic_invalidate}
21157 if the inlined code would not work in user mode.
21158 This is the default when the target is @code{sh-*-linux*}.
21160 @item -multcost=@var{number}
21161 @opindex multcost=@var{number}
21162 Set the cost to assume for a multiply insn.
21164 @item -mdiv=@var{strategy}
21165 @opindex mdiv=@var{strategy}
21166 Set the division strategy to be used for integer division operations.
21167 For SHmedia @var{strategy} can be one of:
21172 Performs the operation in floating point. This has a very high latency,
21173 but needs only a few instructions, so it might be a good choice if
21174 your code has enough easily-exploitable ILP to allow the compiler to
21175 schedule the floating-point instructions together with other instructions.
21176 Division by zero causes a floating-point exception.
21179 Uses integer operations to calculate the inverse of the divisor,
21180 and then multiplies the dividend with the inverse. This strategy allows
21181 CSE and hoisting of the inverse calculation. Division by zero calculates
21182 an unspecified result, but does not trap.
21185 A variant of @samp{inv} where, if no CSE or hoisting opportunities
21186 have been found, or if the entire operation has been hoisted to the same
21187 place, the last stages of the inverse calculation are intertwined with the
21188 final multiply to reduce the overall latency, at the expense of using a few
21189 more instructions, and thus offering fewer scheduling opportunities with
21193 Calls a library function that usually implements the @samp{inv:minlat}
21195 This gives high code density for @code{m5-*media-nofpu} compilations.
21198 Uses a different entry point of the same library function, where it
21199 assumes that a pointer to a lookup table has already been set up, which
21200 exposes the pointer load to CSE and code hoisting optimizations.
21205 Use the @samp{inv} algorithm for initial
21206 code generation, but if the code stays unoptimized, revert to the @samp{call},
21207 @samp{call2}, or @samp{fp} strategies, respectively. Note that the
21208 potentially-trapping side effect of division by zero is carried by a
21209 separate instruction, so it is possible that all the integer instructions
21210 are hoisted out, but the marker for the side effect stays where it is.
21211 A recombination to floating-point operations or a call is not possible
21216 Variants of the @samp{inv:minlat} strategy. In the case
21217 that the inverse calculation is not separated from the multiply, they speed
21218 up division where the dividend fits into 20 bits (plus sign where applicable)
21219 by inserting a test to skip a number of operations in this case; this test
21220 slows down the case of larger dividends. @samp{inv20u} assumes the case of a such
21221 a small dividend to be unlikely, and @samp{inv20l} assumes it to be likely.
21225 For targets other than SHmedia @var{strategy} can be one of:
21230 Calls a library function that uses the single-step division instruction
21231 @code{div1} to perform the operation. Division by zero calculates an
21232 unspecified result and does not trap. This is the default except for SH4,
21233 SH2A and SHcompact.
21236 Calls a library function that performs the operation in double precision
21237 floating point. Division by zero causes a floating-point exception. This is
21238 the default for SHcompact with FPU. Specifying this for targets that do not
21239 have a double precision FPU will default to @code{call-div1}.
21242 Calls a library function that uses a lookup table for small divisors and
21243 the @code{div1} instruction with case distinction for larger divisors. Division
21244 by zero calculates an unspecified result and does not trap. This is the default
21245 for SH4. Specifying this for targets that do not have dynamic shift
21246 instructions will default to @code{call-div1}.
21250 When a division strategy has not been specified the default strategy will be
21251 selected based on the current target. For SH2A the default strategy is to
21252 use the @code{divs} and @code{divu} instructions instead of library function
21255 @item -maccumulate-outgoing-args
21256 @opindex maccumulate-outgoing-args
21257 Reserve space once for outgoing arguments in the function prologue rather
21258 than around each call. Generally beneficial for performance and size. Also
21259 needed for unwinding to avoid changing the stack frame around conditional code.
21261 @item -mdivsi3_libfunc=@var{name}
21262 @opindex mdivsi3_libfunc=@var{name}
21263 Set the name of the library function used for 32-bit signed division to
21265 This only affects the name used in the @samp{call} and @samp{inv:call}
21266 division strategies, and the compiler still expects the same
21267 sets of input/output/clobbered registers as if this option were not present.
21269 @item -mfixed-range=@var{register-range}
21270 @opindex mfixed-range
21271 Generate code treating the given register range as fixed registers.
21272 A fixed register is one that the register allocator can not use. This is
21273 useful when compiling kernel code. A register range is specified as
21274 two registers separated by a dash. Multiple register ranges can be
21275 specified separated by a comma.
21277 @item -mindexed-addressing
21278 @opindex mindexed-addressing
21279 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
21280 This is only safe if the hardware and/or OS implement 32-bit wrap-around
21281 semantics for the indexed addressing mode. The architecture allows the
21282 implementation of processors with 64-bit MMU, which the OS could use to
21283 get 32-bit addressing, but since no current hardware implementation supports
21284 this or any other way to make the indexed addressing mode safe to use in
21285 the 32-bit ABI, the default is @option{-mno-indexed-addressing}.
21287 @item -mgettrcost=@var{number}
21288 @opindex mgettrcost=@var{number}
21289 Set the cost assumed for the @code{gettr} instruction to @var{number}.
21290 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
21294 Assume @code{pt*} instructions won't trap. This generally generates
21295 better-scheduled code, but is unsafe on current hardware.
21296 The current architecture
21297 definition says that @code{ptabs} and @code{ptrel} trap when the target
21299 This has the unintentional effect of making it unsafe to schedule these
21300 instructions before a branch, or hoist them out of a loop. For example,
21301 @code{__do_global_ctors}, a part of @file{libgcc}
21302 that runs constructors at program
21303 startup, calls functions in a list which is delimited by @minus{}1. With the
21304 @option{-mpt-fixed} option, the @code{ptabs} is done before testing against @minus{}1.
21305 That means that all the constructors run a bit more quickly, but when
21306 the loop comes to the end of the list, the program crashes because @code{ptabs}
21307 loads @minus{}1 into a target register.
21309 Since this option is unsafe for any
21310 hardware implementing the current architecture specification, the default
21311 is @option{-mno-pt-fixed}. Unless specified explicitly with
21312 @option{-mgettrcost}, @option{-mno-pt-fixed} also implies @option{-mgettrcost=100};
21313 this deters register allocation from using target registers for storing
21316 @item -minvalid-symbols
21317 @opindex minvalid-symbols
21318 Assume symbols might be invalid. Ordinary function symbols generated by
21319 the compiler are always valid to load with
21320 @code{movi}/@code{shori}/@code{ptabs} or
21321 @code{movi}/@code{shori}/@code{ptrel},
21322 but with assembler and/or linker tricks it is possible
21323 to generate symbols that cause @code{ptabs} or @code{ptrel} to trap.
21324 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
21325 It prevents cross-basic-block CSE, hoisting and most scheduling
21326 of symbol loads. The default is @option{-mno-invalid-symbols}.
21328 @item -mbranch-cost=@var{num}
21329 @opindex mbranch-cost=@var{num}
21330 Assume @var{num} to be the cost for a branch instruction. Higher numbers
21331 make the compiler try to generate more branch-free code if possible.
21332 If not specified the value is selected depending on the processor type that
21333 is being compiled for.
21336 @itemx -mno-zdcbranch
21337 @opindex mzdcbranch
21338 @opindex mno-zdcbranch
21339 Assume (do not assume) that zero displacement conditional branch instructions
21340 @code{bt} and @code{bf} are fast. If @option{-mzdcbranch} is specified, the
21341 compiler will try to prefer zero displacement branch code sequences. This is
21342 enabled by default when generating code for SH4 and SH4A. It can be explicitly
21343 disabled by specifying @option{-mno-zdcbranch}.
21346 @itemx -mno-fused-madd
21347 @opindex mfused-madd
21348 @opindex mno-fused-madd
21349 Generate code that uses (does not use) the floating-point multiply and
21350 accumulate instructions. These instructions are generated by default
21351 if hardware floating point is used. The machine-dependent
21352 @option{-mfused-madd} option is now mapped to the machine-independent
21353 @option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
21354 mapped to @option{-ffp-contract=off}.
21360 Allow or disallow the compiler to emit the @code{fsca} instruction for sine
21361 and cosine approximations. The option @code{-mfsca} must be used in
21362 combination with @code{-funsafe-math-optimizations}. It is enabled by default
21363 when generating code for SH4A. Using @code{-mno-fsca} disables sine and cosine
21364 approximations even if @code{-funsafe-math-optimizations} is in effect.
21370 Allow or disallow the compiler to emit the @code{fsrra} instruction for
21371 reciprocal square root approximations. The option @code{-mfsrra} must be used
21372 in combination with @code{-funsafe-math-optimizations} and
21373 @code{-ffinite-math-only}. It is enabled by default when generating code for
21374 SH4A. Using @code{-mno-fsrra} disables reciprocal square root approximations
21375 even if @code{-funsafe-math-optimizations} and @code{-ffinite-math-only} are
21378 @item -mpretend-cmove
21379 @opindex mpretend-cmove
21380 Prefer zero-displacement conditional branches for conditional move instruction
21381 patterns. This can result in faster code on the SH4 processor.
21385 @node Solaris 2 Options
21386 @subsection Solaris 2 Options
21387 @cindex Solaris 2 options
21389 These @samp{-m} options are supported on Solaris 2:
21392 @item -mclear-hwcap
21393 @opindex mclear-hwcap
21394 @option{-mclear-hwcap} tells the compiler to remove the hardware
21395 capabilities generated by the Solaris assembler. This is only necessary
21396 when object files use ISA extensions not supported by the current
21397 machine, but check at runtime whether or not to use them.
21399 @item -mimpure-text
21400 @opindex mimpure-text
21401 @option{-mimpure-text}, used in addition to @option{-shared}, tells
21402 the compiler to not pass @option{-z text} to the linker when linking a
21403 shared object. Using this option, you can link position-dependent
21404 code into a shared object.
21406 @option{-mimpure-text} suppresses the ``relocations remain against
21407 allocatable but non-writable sections'' linker error message.
21408 However, the necessary relocations trigger copy-on-write, and the
21409 shared object is not actually shared across processes. Instead of
21410 using @option{-mimpure-text}, you should compile all source code with
21411 @option{-fpic} or @option{-fPIC}.
21415 These switches are supported in addition to the above on Solaris 2:
21420 Add support for multithreading using the POSIX threads library. This
21421 option sets flags for both the preprocessor and linker. This option does
21422 not affect the thread safety of object code produced by the compiler or
21423 that of libraries supplied with it.
21427 This is a synonym for @option{-pthreads}.
21430 @node SPARC Options
21431 @subsection SPARC Options
21432 @cindex SPARC options
21434 These @samp{-m} options are supported on the SPARC:
21437 @item -mno-app-regs
21439 @opindex mno-app-regs
21441 Specify @option{-mapp-regs} to generate output using the global registers
21442 2 through 4, which the SPARC SVR4 ABI reserves for applications. Like the
21443 global register 1, each global register 2 through 4 is then treated as an
21444 allocable register that is clobbered by function calls. This is the default.
21446 To be fully SVR4 ABI-compliant at the cost of some performance loss,
21447 specify @option{-mno-app-regs}. You should compile libraries and system
21448 software with this option.
21454 With @option{-mflat}, the compiler does not generate save/restore instructions
21455 and uses a ``flat'' or single register window model. This model is compatible
21456 with the regular register window model. The local registers and the input
21457 registers (0--5) are still treated as ``call-saved'' registers and are
21458 saved on the stack as needed.
21460 With @option{-mno-flat} (the default), the compiler generates save/restore
21461 instructions (except for leaf functions). This is the normal operating mode.
21464 @itemx -mhard-float
21466 @opindex mhard-float
21467 Generate output containing floating-point instructions. This is the
21471 @itemx -msoft-float
21473 @opindex msoft-float
21474 Generate output containing library calls for floating point.
21475 @strong{Warning:} the requisite libraries are not available for all SPARC
21476 targets. Normally the facilities of the machine's usual C compiler are
21477 used, but this cannot be done directly in cross-compilation. You must make
21478 your own arrangements to provide suitable library functions for
21479 cross-compilation. The embedded targets @samp{sparc-*-aout} and
21480 @samp{sparclite-*-*} do provide software floating-point support.
21482 @option{-msoft-float} changes the calling convention in the output file;
21483 therefore, it is only useful if you compile @emph{all} of a program with
21484 this option. In particular, you need to compile @file{libgcc.a}, the
21485 library that comes with GCC, with @option{-msoft-float} in order for
21488 @item -mhard-quad-float
21489 @opindex mhard-quad-float
21490 Generate output containing quad-word (long double) floating-point
21493 @item -msoft-quad-float
21494 @opindex msoft-quad-float
21495 Generate output containing library calls for quad-word (long double)
21496 floating-point instructions. The functions called are those specified
21497 in the SPARC ABI@. This is the default.
21499 As of this writing, there are no SPARC implementations that have hardware
21500 support for the quad-word floating-point instructions. They all invoke
21501 a trap handler for one of these instructions, and then the trap handler
21502 emulates the effect of the instruction. Because of the trap handler overhead,
21503 this is much slower than calling the ABI library routines. Thus the
21504 @option{-msoft-quad-float} option is the default.
21506 @item -mno-unaligned-doubles
21507 @itemx -munaligned-doubles
21508 @opindex mno-unaligned-doubles
21509 @opindex munaligned-doubles
21510 Assume that doubles have 8-byte alignment. This is the default.
21512 With @option{-munaligned-doubles}, GCC assumes that doubles have 8-byte
21513 alignment only if they are contained in another type, or if they have an
21514 absolute address. Otherwise, it assumes they have 4-byte alignment.
21515 Specifying this option avoids some rare compatibility problems with code
21516 generated by other compilers. It is not the default because it results
21517 in a performance loss, especially for floating-point code.
21520 @itemx -mno-user-mode
21521 @opindex muser-mode
21522 @opindex mno-user-mode
21523 Do not generate code that can only run in supervisor mode. This is relevant
21524 only for the @code{casa} instruction emitted for the LEON3 processor. The
21525 default is @option{-mno-user-mode}.
21527 @item -mno-faster-structs
21528 @itemx -mfaster-structs
21529 @opindex mno-faster-structs
21530 @opindex mfaster-structs
21531 With @option{-mfaster-structs}, the compiler assumes that structures
21532 should have 8-byte alignment. This enables the use of pairs of
21533 @code{ldd} and @code{std} instructions for copies in structure
21534 assignment, in place of twice as many @code{ld} and @code{st} pairs.
21535 However, the use of this changed alignment directly violates the SPARC
21536 ABI@. Thus, it's intended only for use on targets where the developer
21537 acknowledges that their resulting code is not directly in line with
21538 the rules of the ABI@.
21540 @item -mcpu=@var{cpu_type}
21542 Set the instruction set, register set, and instruction scheduling parameters
21543 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
21544 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{hypersparc},
21545 @samp{leon}, @samp{leon3}, @samp{sparclite}, @samp{f930}, @samp{f934},
21546 @samp{sparclite86x}, @samp{sparclet}, @samp{tsc701}, @samp{v9},
21547 @samp{ultrasparc}, @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2},
21548 @samp{niagara3} and @samp{niagara4}.
21550 Native Solaris and GNU/Linux toolchains also support the value @samp{native},
21551 which selects the best architecture option for the host processor.
21552 @option{-mcpu=native} has no effect if GCC does not recognize
21555 Default instruction scheduling parameters are used for values that select
21556 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
21557 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
21559 Here is a list of each supported architecture and their supported
21567 supersparc, hypersparc, leon, leon3
21570 f930, f934, sparclite86x
21576 ultrasparc, ultrasparc3, niagara, niagara2, niagara3, niagara4
21579 By default (unless configured otherwise), GCC generates code for the V7
21580 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
21581 additionally optimizes it for the Cypress CY7C602 chip, as used in the
21582 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
21583 SPARCStation 1, 2, IPX etc.
21585 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
21586 architecture. The only difference from V7 code is that the compiler emits
21587 the integer multiply and integer divide instructions which exist in SPARC-V8
21588 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
21589 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
21592 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
21593 the SPARC architecture. This adds the integer multiply, integer divide step
21594 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
21595 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
21596 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
21597 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
21598 MB86934 chip, which is the more recent SPARClite with FPU@.
21600 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
21601 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
21602 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
21603 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
21604 optimizes it for the TEMIC SPARClet chip.
21606 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
21607 architecture. This adds 64-bit integer and floating-point move instructions,
21608 3 additional floating-point condition code registers and conditional move
21609 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
21610 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
21611 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
21612 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
21613 @option{-mcpu=niagara}, the compiler additionally optimizes it for
21614 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
21615 additionally optimizes it for Sun UltraSPARC T2 chips. With
21616 @option{-mcpu=niagara3}, the compiler additionally optimizes it for Sun
21617 UltraSPARC T3 chips. With @option{-mcpu=niagara4}, the compiler
21618 additionally optimizes it for Sun UltraSPARC T4 chips.
21620 @item -mtune=@var{cpu_type}
21622 Set the instruction scheduling parameters for machine type
21623 @var{cpu_type}, but do not set the instruction set or register set that the
21624 option @option{-mcpu=@var{cpu_type}} does.
21626 The same values for @option{-mcpu=@var{cpu_type}} can be used for
21627 @option{-mtune=@var{cpu_type}}, but the only useful values are those
21628 that select a particular CPU implementation. Those are @samp{cypress},
21629 @samp{supersparc}, @samp{hypersparc}, @samp{leon}, @samp{leon3}, @samp{f930},
21630 @samp{f934}, @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
21631 @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2}, @samp{niagara3} and
21632 @samp{niagara4}. With native Solaris and GNU/Linux toolchains, @samp{native}
21638 @opindex mno-v8plus
21639 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
21640 difference from the V8 ABI is that the global and out registers are
21641 considered 64 bits wide. This is enabled by default on Solaris in 32-bit
21642 mode for all SPARC-V9 processors.
21648 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
21649 Visual Instruction Set extensions. The default is @option{-mno-vis}.
21655 With @option{-mvis2}, GCC generates code that takes advantage of
21656 version 2.0 of the UltraSPARC Visual Instruction Set extensions. The
21657 default is @option{-mvis2} when targeting a cpu that supports such
21658 instructions, such as UltraSPARC-III and later. Setting @option{-mvis2}
21659 also sets @option{-mvis}.
21665 With @option{-mvis3}, GCC generates code that takes advantage of
21666 version 3.0 of the UltraSPARC Visual Instruction Set extensions. The
21667 default is @option{-mvis3} when targeting a cpu that supports such
21668 instructions, such as niagara-3 and later. Setting @option{-mvis3}
21669 also sets @option{-mvis2} and @option{-mvis}.
21674 @opindex mno-cbcond
21675 With @option{-mcbcond}, GCC generates code that takes advantage of
21676 compare-and-branch instructions, as defined in the Sparc Architecture 2011.
21677 The default is @option{-mcbcond} when targeting a cpu that supports such
21678 instructions, such as niagara-4 and later.
21684 With @option{-mpopc}, GCC generates code that takes advantage of the UltraSPARC
21685 population count instruction. The default is @option{-mpopc}
21686 when targeting a cpu that supports such instructions, such as Niagara-2 and
21693 With @option{-mfmaf}, GCC generates code that takes advantage of the UltraSPARC
21694 Fused Multiply-Add Floating-point extensions. The default is @option{-mfmaf}
21695 when targeting a cpu that supports such instructions, such as Niagara-3 and
21699 @opindex mfix-at697f
21700 Enable the documented workaround for the single erratum of the Atmel AT697F
21701 processor (which corresponds to erratum #13 of the AT697E processor).
21704 @opindex mfix-ut699
21705 Enable the documented workarounds for the floating-point errata and the data
21706 cache nullify errata of the UT699 processor.
21709 These @samp{-m} options are supported in addition to the above
21710 on SPARC-V9 processors in 64-bit environments:
21717 Generate code for a 32-bit or 64-bit environment.
21718 The 32-bit environment sets int, long and pointer to 32 bits.
21719 The 64-bit environment sets int to 32 bits and long and pointer
21722 @item -mcmodel=@var{which}
21724 Set the code model to one of
21728 The Medium/Low code model: 64-bit addresses, programs
21729 must be linked in the low 32 bits of memory. Programs can be statically
21730 or dynamically linked.
21733 The Medium/Middle code model: 64-bit addresses, programs
21734 must be linked in the low 44 bits of memory, the text and data segments must
21735 be less than 2GB in size and the data segment must be located within 2GB of
21739 The Medium/Anywhere code model: 64-bit addresses, programs
21740 may be linked anywhere in memory, the text and data segments must be less
21741 than 2GB in size and the data segment must be located within 2GB of the
21745 The Medium/Anywhere code model for embedded systems:
21746 64-bit addresses, the text and data segments must be less than 2GB in
21747 size, both starting anywhere in memory (determined at link time). The
21748 global register %g4 points to the base of the data segment. Programs
21749 are statically linked and PIC is not supported.
21752 @item -mmemory-model=@var{mem-model}
21753 @opindex mmemory-model
21754 Set the memory model in force on the processor to one of
21758 The default memory model for the processor and operating system.
21761 Relaxed Memory Order
21764 Partial Store Order
21770 Sequential Consistency
21773 These memory models are formally defined in Appendix D of the Sparc V9
21774 architecture manual, as set in the processor's @code{PSTATE.MM} field.
21777 @itemx -mno-stack-bias
21778 @opindex mstack-bias
21779 @opindex mno-stack-bias
21780 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
21781 frame pointer if present, are offset by @minus{}2047 which must be added back
21782 when making stack frame references. This is the default in 64-bit mode.
21783 Otherwise, assume no such offset is present.
21787 @subsection SPU Options
21788 @cindex SPU options
21790 These @samp{-m} options are supported on the SPU:
21794 @itemx -merror-reloc
21795 @opindex mwarn-reloc
21796 @opindex merror-reloc
21798 The loader for SPU does not handle dynamic relocations. By default, GCC
21799 gives an error when it generates code that requires a dynamic
21800 relocation. @option{-mno-error-reloc} disables the error,
21801 @option{-mwarn-reloc} generates a warning instead.
21804 @itemx -munsafe-dma
21806 @opindex munsafe-dma
21808 Instructions that initiate or test completion of DMA must not be
21809 reordered with respect to loads and stores of the memory that is being
21811 With @option{-munsafe-dma} you must use the @code{volatile} keyword to protect
21812 memory accesses, but that can lead to inefficient code in places where the
21813 memory is known to not change. Rather than mark the memory as volatile,
21814 you can use @option{-msafe-dma} to tell the compiler to treat
21815 the DMA instructions as potentially affecting all memory.
21817 @item -mbranch-hints
21818 @opindex mbranch-hints
21820 By default, GCC generates a branch hint instruction to avoid
21821 pipeline stalls for always-taken or probably-taken branches. A hint
21822 is not generated closer than 8 instructions away from its branch.
21823 There is little reason to disable them, except for debugging purposes,
21824 or to make an object a little bit smaller.
21828 @opindex msmall-mem
21829 @opindex mlarge-mem
21831 By default, GCC generates code assuming that addresses are never larger
21832 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
21833 a full 32-bit address.
21838 By default, GCC links against startup code that assumes the SPU-style
21839 main function interface (which has an unconventional parameter list).
21840 With @option{-mstdmain}, GCC links your program against startup
21841 code that assumes a C99-style interface to @code{main}, including a
21842 local copy of @code{argv} strings.
21844 @item -mfixed-range=@var{register-range}
21845 @opindex mfixed-range
21846 Generate code treating the given register range as fixed registers.
21847 A fixed register is one that the register allocator cannot use. This is
21848 useful when compiling kernel code. A register range is specified as
21849 two registers separated by a dash. Multiple register ranges can be
21850 specified separated by a comma.
21856 Compile code assuming that pointers to the PPU address space accessed
21857 via the @code{__ea} named address space qualifier are either 32 or 64
21858 bits wide. The default is 32 bits. As this is an ABI-changing option,
21859 all object code in an executable must be compiled with the same setting.
21861 @item -maddress-space-conversion
21862 @itemx -mno-address-space-conversion
21863 @opindex maddress-space-conversion
21864 @opindex mno-address-space-conversion
21865 Allow/disallow treating the @code{__ea} address space as superset
21866 of the generic address space. This enables explicit type casts
21867 between @code{__ea} and generic pointer as well as implicit
21868 conversions of generic pointers to @code{__ea} pointers. The
21869 default is to allow address space pointer conversions.
21871 @item -mcache-size=@var{cache-size}
21872 @opindex mcache-size
21873 This option controls the version of libgcc that the compiler links to an
21874 executable and selects a software-managed cache for accessing variables
21875 in the @code{__ea} address space with a particular cache size. Possible
21876 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
21877 and @samp{128}. The default cache size is 64KB.
21879 @item -matomic-updates
21880 @itemx -mno-atomic-updates
21881 @opindex matomic-updates
21882 @opindex mno-atomic-updates
21883 This option controls the version of libgcc that the compiler links to an
21884 executable and selects whether atomic updates to the software-managed
21885 cache of PPU-side variables are used. If you use atomic updates, changes
21886 to a PPU variable from SPU code using the @code{__ea} named address space
21887 qualifier do not interfere with changes to other PPU variables residing
21888 in the same cache line from PPU code. If you do not use atomic updates,
21889 such interference may occur; however, writing back cache lines is
21890 more efficient. The default behavior is to use atomic updates.
21893 @itemx -mdual-nops=@var{n}
21894 @opindex mdual-nops
21895 By default, GCC inserts nops to increase dual issue when it expects
21896 it to increase performance. @var{n} can be a value from 0 to 10. A
21897 smaller @var{n} inserts fewer nops. 10 is the default, 0 is the
21898 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
21900 @item -mhint-max-nops=@var{n}
21901 @opindex mhint-max-nops
21902 Maximum number of nops to insert for a branch hint. A branch hint must
21903 be at least 8 instructions away from the branch it is affecting. GCC
21904 inserts up to @var{n} nops to enforce this, otherwise it does not
21905 generate the branch hint.
21907 @item -mhint-max-distance=@var{n}
21908 @opindex mhint-max-distance
21909 The encoding of the branch hint instruction limits the hint to be within
21910 256 instructions of the branch it is affecting. By default, GCC makes
21911 sure it is within 125.
21914 @opindex msafe-hints
21915 Work around a hardware bug that causes the SPU to stall indefinitely.
21916 By default, GCC inserts the @code{hbrp} instruction to make sure
21917 this stall won't happen.
21921 @node System V Options
21922 @subsection Options for System V
21924 These additional options are available on System V Release 4 for
21925 compatibility with other compilers on those systems:
21930 Create a shared object.
21931 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
21935 Identify the versions of each tool used by the compiler, in a
21936 @code{.ident} assembler directive in the output.
21940 Refrain from adding @code{.ident} directives to the output file (this is
21943 @item -YP,@var{dirs}
21945 Search the directories @var{dirs}, and no others, for libraries
21946 specified with @option{-l}.
21948 @item -Ym,@var{dir}
21950 Look in the directory @var{dir} to find the M4 preprocessor.
21951 The assembler uses this option.
21952 @c This is supposed to go with a -Yd for predefined M4 macro files, but
21953 @c the generic assembler that comes with Solaris takes just -Ym.
21956 @node TILE-Gx Options
21957 @subsection TILE-Gx Options
21958 @cindex TILE-Gx options
21960 These @samp{-m} options are supported on the TILE-Gx:
21963 @item -mcmodel=small
21964 @opindex mcmodel=small
21965 Generate code for the small model. The distance for direct calls is
21966 limited to 500M in either direction. PC-relative addresses are 32
21967 bits. Absolute addresses support the full address range.
21969 @item -mcmodel=large
21970 @opindex mcmodel=large
21971 Generate code for the large model. There is no limitation on call
21972 distance, pc-relative addresses, or absolute addresses.
21974 @item -mcpu=@var{name}
21976 Selects the type of CPU to be targeted. Currently the only supported
21977 type is @samp{tilegx}.
21983 Generate code for a 32-bit or 64-bit environment. The 32-bit
21984 environment sets int, long, and pointer to 32 bits. The 64-bit
21985 environment sets int to 32 bits and long and pointer to 64 bits.
21988 @itemx -mlittle-endian
21989 @opindex mbig-endian
21990 @opindex mlittle-endian
21991 Generate code in big/little endian mode, respectively.
21994 @node TILEPro Options
21995 @subsection TILEPro Options
21996 @cindex TILEPro options
21998 These @samp{-m} options are supported on the TILEPro:
22001 @item -mcpu=@var{name}
22003 Selects the type of CPU to be targeted. Currently the only supported
22004 type is @samp{tilepro}.
22008 Generate code for a 32-bit environment, which sets int, long, and
22009 pointer to 32 bits. This is the only supported behavior so the flag
22010 is essentially ignored.
22014 @subsection V850 Options
22015 @cindex V850 Options
22017 These @samp{-m} options are defined for V850 implementations:
22021 @itemx -mno-long-calls
22022 @opindex mlong-calls
22023 @opindex mno-long-calls
22024 Treat all calls as being far away (near). If calls are assumed to be
22025 far away, the compiler always loads the function's address into a
22026 register, and calls indirect through the pointer.
22032 Do not optimize (do optimize) basic blocks that use the same index
22033 pointer 4 or more times to copy pointer into the @code{ep} register, and
22034 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
22035 option is on by default if you optimize.
22037 @item -mno-prolog-function
22038 @itemx -mprolog-function
22039 @opindex mno-prolog-function
22040 @opindex mprolog-function
22041 Do not use (do use) external functions to save and restore registers
22042 at the prologue and epilogue of a function. The external functions
22043 are slower, but use less code space if more than one function saves
22044 the same number of registers. The @option{-mprolog-function} option
22045 is on by default if you optimize.
22049 Try to make the code as small as possible. At present, this just turns
22050 on the @option{-mep} and @option{-mprolog-function} options.
22052 @item -mtda=@var{n}
22054 Put static or global variables whose size is @var{n} bytes or less into
22055 the tiny data area that register @code{ep} points to. The tiny data
22056 area can hold up to 256 bytes in total (128 bytes for byte references).
22058 @item -msda=@var{n}
22060 Put static or global variables whose size is @var{n} bytes or less into
22061 the small data area that register @code{gp} points to. The small data
22062 area can hold up to 64 kilobytes.
22064 @item -mzda=@var{n}
22066 Put static or global variables whose size is @var{n} bytes or less into
22067 the first 32 kilobytes of memory.
22071 Specify that the target processor is the V850.
22075 Specify that the target processor is the V850E3V5. The preprocessor
22076 constant @samp{__v850e3v5__} is defined if this option is used.
22080 Specify that the target processor is the V850E3V5. This is an alias for
22081 the @option{-mv850e3v5} option.
22085 Specify that the target processor is the V850E2V3. The preprocessor
22086 constant @samp{__v850e2v3__} is defined if this option is used.
22090 Specify that the target processor is the V850E2. The preprocessor
22091 constant @samp{__v850e2__} is defined if this option is used.
22095 Specify that the target processor is the V850E1. The preprocessor
22096 constants @samp{__v850e1__} and @samp{__v850e__} are defined if
22097 this option is used.
22101 Specify that the target processor is the V850ES. This is an alias for
22102 the @option{-mv850e1} option.
22106 Specify that the target processor is the V850E@. The preprocessor
22107 constant @samp{__v850e__} is defined if this option is used.
22109 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
22110 nor @option{-mv850e2} nor @option{-mv850e2v3} nor @option{-mv850e3v5}
22111 are defined then a default target processor is chosen and the
22112 relevant @samp{__v850*__} preprocessor constant is defined.
22114 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
22115 defined, regardless of which processor variant is the target.
22117 @item -mdisable-callt
22118 @itemx -mno-disable-callt
22119 @opindex mdisable-callt
22120 @opindex mno-disable-callt
22121 This option suppresses generation of the @code{CALLT} instruction for the
22122 v850e, v850e1, v850e2, v850e2v3 and v850e3v5 flavors of the v850
22125 This option is enabled by default when the RH850 ABI is
22126 in use (see @option{-mrh850-abi}), and disabled by default when the
22127 GCC ABI is in use. If @code{CALLT} instructions are being generated
22128 then the C preprocessor symbol @code{__V850_CALLT__} will be defined.
22134 Pass on (or do not pass on) the @option{-mrelax} command line option
22138 @itemx -mno-long-jumps
22139 @opindex mlong-jumps
22140 @opindex mno-long-jumps
22141 Disable (or re-enable) the generation of PC-relative jump instructions.
22144 @itemx -mhard-float
22145 @opindex msoft-float
22146 @opindex mhard-float
22147 Disable (or re-enable) the generation of hardware floating point
22148 instructions. This option is only significant when the target
22149 architecture is @samp{V850E2V3} or higher. If hardware floating point
22150 instructions are being generated then the C preprocessor symbol
22151 @code{__FPU_OK__} will be defined, otherwise the symbol
22152 @code{__NO_FPU__} will be defined.
22156 Enables the use of the e3v5 LOOP instruction. The use of this
22157 instruction is not enabled by default when the e3v5 architecture is
22158 selected because its use is still experimental.
22162 @opindex mrh850-abi
22164 Enables support for the RH850 version of the V850 ABI. This is the
22165 default. With this version of the ABI the following rules apply:
22169 Integer sized structures and unions are returned via a memory pointer
22170 rather than a register.
22173 Large structures and unions (more than 8 bytes in size) are passed by
22177 Functions are aligned to 16-bit boundaries.
22180 The @option{-m8byte-align} command line option is supported.
22183 The @option{-mdisable-callt} command line option is enabled by
22184 default. The @option{-mno-disable-callt} command line option is not
22188 When this version of the ABI is enabled the C preprocessor symbol
22189 @code{__V850_RH850_ABI__} is defined.
22193 Enables support for the old GCC version of the V850 ABI. With this
22194 version of the ABI the following rules apply:
22198 Integer sized structures and unions are returned in register @code{r10}.
22201 Large structures and unions (more than 8 bytes in size) are passed by
22205 Functions are aligned to 32-bit boundaries, unless optimizing for
22209 The @option{-m8byte-align} command line option is not supported.
22212 The @option{-mdisable-callt} command line option is supported but not
22213 enabled by default.
22216 When this version of the ABI is enabled the C preprocessor symbol
22217 @code{__V850_GCC_ABI__} is defined.
22219 @item -m8byte-align
22220 @itemx -mno-8byte-align
22221 @opindex m8byte-align
22222 @opindex mno-8byte-align
22223 Enables support for @code{doubles} and @code{long long} types to be
22224 aligned on 8-byte boundaries. The default is to restrict the
22225 alignment of all objects to at most 4-bytes. When
22226 @option{-m8byte-align} is in effect the C preprocessor symbol
22227 @code{__V850_8BYTE_ALIGN__} will be defined.
22230 @opindex mbig-switch
22231 Generate code suitable for big switch tables. Use this option only if
22232 the assembler/linker complain about out of range branches within a switch
22237 This option causes r2 and r5 to be used in the code generated by
22238 the compiler. This setting is the default.
22240 @item -mno-app-regs
22241 @opindex mno-app-regs
22242 This option causes r2 and r5 to be treated as fixed registers.
22247 @subsection VAX Options
22248 @cindex VAX options
22250 These @samp{-m} options are defined for the VAX:
22255 Do not output certain jump instructions (@code{aobleq} and so on)
22256 that the Unix assembler for the VAX cannot handle across long
22261 Do output those jump instructions, on the assumption that the
22262 GNU assembler is being used.
22266 Output code for G-format floating-point numbers instead of D-format.
22270 @subsection VMS Options
22272 These @samp{-m} options are defined for the VMS implementations:
22275 @item -mvms-return-codes
22276 @opindex mvms-return-codes
22277 Return VMS condition codes from @code{main}. The default is to return POSIX-style
22278 condition (e.g.@ error) codes.
22280 @item -mdebug-main=@var{prefix}
22281 @opindex mdebug-main=@var{prefix}
22282 Flag the first routine whose name starts with @var{prefix} as the main
22283 routine for the debugger.
22287 Default to 64-bit memory allocation routines.
22289 @item -mpointer-size=@var{size}
22290 @opindex -mpointer-size=@var{size}
22291 Set the default size of pointers. Possible options for @var{size} are
22292 @samp{32} or @samp{short} for 32 bit pointers, @samp{64} or @samp{long}
22293 for 64 bit pointers, and @samp{no} for supporting only 32 bit pointers.
22294 The later option disables @code{pragma pointer_size}.
22297 @node VxWorks Options
22298 @subsection VxWorks Options
22299 @cindex VxWorks Options
22301 The options in this section are defined for all VxWorks targets.
22302 Options specific to the target hardware are listed with the other
22303 options for that target.
22308 GCC can generate code for both VxWorks kernels and real time processes
22309 (RTPs). This option switches from the former to the latter. It also
22310 defines the preprocessor macro @code{__RTP__}.
22313 @opindex non-static
22314 Link an RTP executable against shared libraries rather than static
22315 libraries. The options @option{-static} and @option{-shared} can
22316 also be used for RTPs (@pxref{Link Options}); @option{-static}
22323 These options are passed down to the linker. They are defined for
22324 compatibility with Diab.
22327 @opindex Xbind-lazy
22328 Enable lazy binding of function calls. This option is equivalent to
22329 @option{-Wl,-z,now} and is defined for compatibility with Diab.
22333 Disable lazy binding of function calls. This option is the default and
22334 is defined for compatibility with Diab.
22337 @node x86-64 Options
22338 @subsection x86-64 Options
22339 @cindex x86-64 options
22341 These are listed under @xref{i386 and x86-64 Options}.
22343 @node Xstormy16 Options
22344 @subsection Xstormy16 Options
22345 @cindex Xstormy16 Options
22347 These options are defined for Xstormy16:
22352 Choose startup files and linker script suitable for the simulator.
22355 @node Xtensa Options
22356 @subsection Xtensa Options
22357 @cindex Xtensa Options
22359 These options are supported for Xtensa targets:
22363 @itemx -mno-const16
22365 @opindex mno-const16
22366 Enable or disable use of @code{CONST16} instructions for loading
22367 constant values. The @code{CONST16} instruction is currently not a
22368 standard option from Tensilica. When enabled, @code{CONST16}
22369 instructions are always used in place of the standard @code{L32R}
22370 instructions. The use of @code{CONST16} is enabled by default only if
22371 the @code{L32R} instruction is not available.
22374 @itemx -mno-fused-madd
22375 @opindex mfused-madd
22376 @opindex mno-fused-madd
22377 Enable or disable use of fused multiply/add and multiply/subtract
22378 instructions in the floating-point option. This has no effect if the
22379 floating-point option is not also enabled. Disabling fused multiply/add
22380 and multiply/subtract instructions forces the compiler to use separate
22381 instructions for the multiply and add/subtract operations. This may be
22382 desirable in some cases where strict IEEE 754-compliant results are
22383 required: the fused multiply add/subtract instructions do not round the
22384 intermediate result, thereby producing results with @emph{more} bits of
22385 precision than specified by the IEEE standard. Disabling fused multiply
22386 add/subtract instructions also ensures that the program output is not
22387 sensitive to the compiler's ability to combine multiply and add/subtract
22390 @item -mserialize-volatile
22391 @itemx -mno-serialize-volatile
22392 @opindex mserialize-volatile
22393 @opindex mno-serialize-volatile
22394 When this option is enabled, GCC inserts @code{MEMW} instructions before
22395 @code{volatile} memory references to guarantee sequential consistency.
22396 The default is @option{-mserialize-volatile}. Use
22397 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
22399 @item -mforce-no-pic
22400 @opindex mforce-no-pic
22401 For targets, like GNU/Linux, where all user-mode Xtensa code must be
22402 position-independent code (PIC), this option disables PIC for compiling
22405 @item -mtext-section-literals
22406 @itemx -mno-text-section-literals
22407 @opindex mtext-section-literals
22408 @opindex mno-text-section-literals
22409 Control the treatment of literal pools. The default is
22410 @option{-mno-text-section-literals}, which places literals in a separate
22411 section in the output file. This allows the literal pool to be placed
22412 in a data RAM/ROM, and it also allows the linker to combine literal
22413 pools from separate object files to remove redundant literals and
22414 improve code size. With @option{-mtext-section-literals}, the literals
22415 are interspersed in the text section in order to keep them as close as
22416 possible to their references. This may be necessary for large assembly
22419 @item -mtarget-align
22420 @itemx -mno-target-align
22421 @opindex mtarget-align
22422 @opindex mno-target-align
22423 When this option is enabled, GCC instructs the assembler to
22424 automatically align instructions to reduce branch penalties at the
22425 expense of some code density. The assembler attempts to widen density
22426 instructions to align branch targets and the instructions following call
22427 instructions. If there are not enough preceding safe density
22428 instructions to align a target, no widening is performed. The
22429 default is @option{-mtarget-align}. These options do not affect the
22430 treatment of auto-aligned instructions like @code{LOOP}, which the
22431 assembler always aligns, either by widening density instructions or
22432 by inserting NOP instructions.
22435 @itemx -mno-longcalls
22436 @opindex mlongcalls
22437 @opindex mno-longcalls
22438 When this option is enabled, GCC instructs the assembler to translate
22439 direct calls to indirect calls unless it can determine that the target
22440 of a direct call is in the range allowed by the call instruction. This
22441 translation typically occurs for calls to functions in other source
22442 files. Specifically, the assembler translates a direct @code{CALL}
22443 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
22444 The default is @option{-mno-longcalls}. This option should be used in
22445 programs where the call target can potentially be out of range. This
22446 option is implemented in the assembler, not the compiler, so the
22447 assembly code generated by GCC still shows direct call
22448 instructions---look at the disassembled object code to see the actual
22449 instructions. Note that the assembler uses an indirect call for
22450 every cross-file call, not just those that really are out of range.
22453 @node zSeries Options
22454 @subsection zSeries Options
22455 @cindex zSeries options
22457 These are listed under @xref{S/390 and zSeries Options}.
22459 @node Code Gen Options
22460 @section Options for Code Generation Conventions
22461 @cindex code generation conventions
22462 @cindex options, code generation
22463 @cindex run-time options
22465 These machine-independent options control the interface conventions
22466 used in code generation.
22468 Most of them have both positive and negative forms; the negative form
22469 of @option{-ffoo} is @option{-fno-foo}. In the table below, only
22470 one of the forms is listed---the one that is not the default. You
22471 can figure out the other form by either removing @samp{no-} or adding
22475 @item -fbounds-check
22476 @opindex fbounds-check
22477 For front ends that support it, generate additional code to check that
22478 indices used to access arrays are within the declared range. This is
22479 currently only supported by the Java and Fortran front ends, where
22480 this option defaults to true and false respectively.
22482 @item -fstack-reuse=@var{reuse-level}
22483 @opindex fstack_reuse
22484 This option controls stack space reuse for user declared local/auto variables
22485 and compiler generated temporaries. @var{reuse_level} can be @samp{all},
22486 @samp{named_vars}, or @samp{none}. @samp{all} enables stack reuse for all
22487 local variables and temporaries, @samp{named_vars} enables the reuse only for
22488 user defined local variables with names, and @samp{none} disables stack reuse
22489 completely. The default value is @samp{all}. The option is needed when the
22490 program extends the lifetime of a scoped local variable or a compiler generated
22491 temporary beyond the end point defined by the language. When a lifetime of
22492 a variable ends, and if the variable lives in memory, the optimizing compiler
22493 has the freedom to reuse its stack space with other temporaries or scoped
22494 local variables whose live range does not overlap with it. Legacy code extending
22495 local lifetime will likely to break with the stack reuse optimization.
22514 if (*p == 10) // out of scope use of local1
22525 A(int k) : i(k), j(k) @{ @}
22532 void foo(const A& ar)
22539 foo(A(10)); // temp object's lifetime ends when foo returns
22545 ap->i+= 10; // ap references out of scope temp whose space
22546 // is reused with a. What is the value of ap->i?
22551 The lifetime of a compiler generated temporary is well defined by the C++
22552 standard. When a lifetime of a temporary ends, and if the temporary lives
22553 in memory, the optimizing compiler has the freedom to reuse its stack
22554 space with other temporaries or scoped local variables whose live range
22555 does not overlap with it. However some of the legacy code relies on
22556 the behavior of older compilers in which temporaries' stack space is
22557 not reused, the aggressive stack reuse can lead to runtime errors. This
22558 option is used to control the temporary stack reuse optimization.
22562 This option generates traps for signed overflow on addition, subtraction,
22563 multiplication operations.
22567 This option instructs the compiler to assume that signed arithmetic
22568 overflow of addition, subtraction and multiplication wraps around
22569 using twos-complement representation. This flag enables some optimizations
22570 and disables others. This option is enabled by default for the Java
22571 front end, as required by the Java language specification.
22574 @opindex fexceptions
22575 Enable exception handling. Generates extra code needed to propagate
22576 exceptions. For some targets, this implies GCC generates frame
22577 unwind information for all functions, which can produce significant data
22578 size overhead, although it does not affect execution. If you do not
22579 specify this option, GCC enables it by default for languages like
22580 C++ that normally require exception handling, and disables it for
22581 languages like C that do not normally require it. However, you may need
22582 to enable this option when compiling C code that needs to interoperate
22583 properly with exception handlers written in C++. You may also wish to
22584 disable this option if you are compiling older C++ programs that don't
22585 use exception handling.
22587 @item -fnon-call-exceptions
22588 @opindex fnon-call-exceptions
22589 Generate code that allows trapping instructions to throw exceptions.
22590 Note that this requires platform-specific runtime support that does
22591 not exist everywhere. Moreover, it only allows @emph{trapping}
22592 instructions to throw exceptions, i.e.@: memory references or floating-point
22593 instructions. It does not allow exceptions to be thrown from
22594 arbitrary signal handlers such as @code{SIGALRM}.
22596 @item -fdelete-dead-exceptions
22597 @opindex fdelete-dead-exceptions
22598 Consider that instructions that may throw exceptions but don't otherwise
22599 contribute to the execution of the program can be optimized away.
22600 This option is enabled by default for the Ada front end, as permitted by
22601 the Ada language specification.
22602 Optimization passes that cause dead exceptions to be removed are enabled independently at different optimization levels.
22604 @item -funwind-tables
22605 @opindex funwind-tables
22606 Similar to @option{-fexceptions}, except that it just generates any needed
22607 static data, but does not affect the generated code in any other way.
22608 You normally do not need to enable this option; instead, a language processor
22609 that needs this handling enables it on your behalf.
22611 @item -fasynchronous-unwind-tables
22612 @opindex fasynchronous-unwind-tables
22613 Generate unwind table in DWARF 2 format, if supported by target machine. The
22614 table is exact at each instruction boundary, so it can be used for stack
22615 unwinding from asynchronous events (such as debugger or garbage collector).
22617 @item -fno-gnu-unique
22618 @opindex fno-gnu-unique
22619 On systems with recent GNU assembler and C library, the C++ compiler
22620 uses the @code{STB_GNU_UNIQUE} binding to make sure that definitions
22621 of template static data members and static local variables in inline
22622 functions are unique even in the presence of @code{RTLD_LOCAL}; this
22623 is necessary to avoid problems with a library used by two different
22624 @code{RTLD_LOCAL} plugins depending on a definition in one of them and
22625 therefore disagreeing with the other one about the binding of the
22626 symbol. But this causes @code{dlclose} to be ignored for affected
22627 DSOs; if your program relies on reinitialization of a DSO via
22628 @code{dlclose} and @code{dlopen}, you can use
22629 @option{-fno-gnu-unique}.
22631 @item -fpcc-struct-return
22632 @opindex fpcc-struct-return
22633 Return ``short'' @code{struct} and @code{union} values in memory like
22634 longer ones, rather than in registers. This convention is less
22635 efficient, but it has the advantage of allowing intercallability between
22636 GCC-compiled files and files compiled with other compilers, particularly
22637 the Portable C Compiler (pcc).
22639 The precise convention for returning structures in memory depends
22640 on the target configuration macros.
22642 Short structures and unions are those whose size and alignment match
22643 that of some integer type.
22645 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
22646 switch is not binary compatible with code compiled with the
22647 @option{-freg-struct-return} switch.
22648 Use it to conform to a non-default application binary interface.
22650 @item -freg-struct-return
22651 @opindex freg-struct-return
22652 Return @code{struct} and @code{union} values in registers when possible.
22653 This is more efficient for small structures than
22654 @option{-fpcc-struct-return}.
22656 If you specify neither @option{-fpcc-struct-return} nor
22657 @option{-freg-struct-return}, GCC defaults to whichever convention is
22658 standard for the target. If there is no standard convention, GCC
22659 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
22660 the principal compiler. In those cases, we can choose the standard, and
22661 we chose the more efficient register return alternative.
22663 @strong{Warning:} code compiled with the @option{-freg-struct-return}
22664 switch is not binary compatible with code compiled with the
22665 @option{-fpcc-struct-return} switch.
22666 Use it to conform to a non-default application binary interface.
22668 @item -fshort-enums
22669 @opindex fshort-enums
22670 Allocate to an @code{enum} type only as many bytes as it needs for the
22671 declared range of possible values. Specifically, the @code{enum} type
22672 is equivalent to the smallest integer type that has enough room.
22674 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
22675 code that is not binary compatible with code generated without that switch.
22676 Use it to conform to a non-default application binary interface.
22678 @item -fshort-double
22679 @opindex fshort-double
22680 Use the same size for @code{double} as for @code{float}.
22682 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
22683 code that is not binary compatible with code generated without that switch.
22684 Use it to conform to a non-default application binary interface.
22686 @item -fshort-wchar
22687 @opindex fshort-wchar
22688 Override the underlying type for @samp{wchar_t} to be @samp{short
22689 unsigned int} instead of the default for the target. This option is
22690 useful for building programs to run under WINE@.
22692 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
22693 code that is not binary compatible with code generated without that switch.
22694 Use it to conform to a non-default application binary interface.
22697 @opindex fno-common
22698 In C code, controls the placement of uninitialized global variables.
22699 Unix C compilers have traditionally permitted multiple definitions of
22700 such variables in different compilation units by placing the variables
22702 This is the behavior specified by @option{-fcommon}, and is the default
22703 for GCC on most targets.
22704 On the other hand, this behavior is not required by ISO C, and on some
22705 targets may carry a speed or code size penalty on variable references.
22706 The @option{-fno-common} option specifies that the compiler should place
22707 uninitialized global variables in the data section of the object file,
22708 rather than generating them as common blocks.
22709 This has the effect that if the same variable is declared
22710 (without @code{extern}) in two different compilations,
22711 you get a multiple-definition error when you link them.
22712 In this case, you must compile with @option{-fcommon} instead.
22713 Compiling with @option{-fno-common} is useful on targets for which
22714 it provides better performance, or if you wish to verify that the
22715 program will work on other systems that always treat uninitialized
22716 variable declarations this way.
22720 Ignore the @samp{#ident} directive.
22722 @item -finhibit-size-directive
22723 @opindex finhibit-size-directive
22724 Don't output a @code{.size} assembler directive, or anything else that
22725 would cause trouble if the function is split in the middle, and the
22726 two halves are placed at locations far apart in memory. This option is
22727 used when compiling @file{crtstuff.c}; you should not need to use it
22730 @item -fverbose-asm
22731 @opindex fverbose-asm
22732 Put extra commentary information in the generated assembly code to
22733 make it more readable. This option is generally only of use to those
22734 who actually need to read the generated assembly code (perhaps while
22735 debugging the compiler itself).
22737 @option{-fno-verbose-asm}, the default, causes the
22738 extra information to be omitted and is useful when comparing two assembler
22741 @item -frecord-gcc-switches
22742 @opindex frecord-gcc-switches
22743 This switch causes the command line used to invoke the
22744 compiler to be recorded into the object file that is being created.
22745 This switch is only implemented on some targets and the exact format
22746 of the recording is target and binary file format dependent, but it
22747 usually takes the form of a section containing ASCII text. This
22748 switch is related to the @option{-fverbose-asm} switch, but that
22749 switch only records information in the assembler output file as
22750 comments, so it never reaches the object file.
22751 See also @option{-grecord-gcc-switches} for another
22752 way of storing compiler options into the object file.
22756 @cindex global offset table
22758 Generate position-independent code (PIC) suitable for use in a shared
22759 library, if supported for the target machine. Such code accesses all
22760 constant addresses through a global offset table (GOT)@. The dynamic
22761 loader resolves the GOT entries when the program starts (the dynamic
22762 loader is not part of GCC; it is part of the operating system). If
22763 the GOT size for the linked executable exceeds a machine-specific
22764 maximum size, you get an error message from the linker indicating that
22765 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
22766 instead. (These maximums are 8k on the SPARC and 32k
22767 on the m68k and RS/6000. The 386 has no such limit.)
22769 Position-independent code requires special support, and therefore works
22770 only on certain machines. For the 386, GCC supports PIC for System V
22771 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
22772 position-independent.
22774 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
22779 If supported for the target machine, emit position-independent code,
22780 suitable for dynamic linking and avoiding any limit on the size of the
22781 global offset table. This option makes a difference on the m68k,
22782 PowerPC and SPARC@.
22784 Position-independent code requires special support, and therefore works
22785 only on certain machines.
22787 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
22794 These options are similar to @option{-fpic} and @option{-fPIC}, but
22795 generated position independent code can be only linked into executables.
22796 Usually these options are used when @option{-pie} GCC option is
22797 used during linking.
22799 @option{-fpie} and @option{-fPIE} both define the macros
22800 @code{__pie__} and @code{__PIE__}. The macros have the value 1
22801 for @option{-fpie} and 2 for @option{-fPIE}.
22803 @item -fno-jump-tables
22804 @opindex fno-jump-tables
22805 Do not use jump tables for switch statements even where it would be
22806 more efficient than other code generation strategies. This option is
22807 of use in conjunction with @option{-fpic} or @option{-fPIC} for
22808 building code that forms part of a dynamic linker and cannot
22809 reference the address of a jump table. On some targets, jump tables
22810 do not require a GOT and this option is not needed.
22812 @item -ffixed-@var{reg}
22814 Treat the register named @var{reg} as a fixed register; generated code
22815 should never refer to it (except perhaps as a stack pointer, frame
22816 pointer or in some other fixed role).
22818 @var{reg} must be the name of a register. The register names accepted
22819 are machine-specific and are defined in the @code{REGISTER_NAMES}
22820 macro in the machine description macro file.
22822 This flag does not have a negative form, because it specifies a
22825 @item -fcall-used-@var{reg}
22826 @opindex fcall-used
22827 Treat the register named @var{reg} as an allocable register that is
22828 clobbered by function calls. It may be allocated for temporaries or
22829 variables that do not live across a call. Functions compiled this way
22830 do not save and restore the register @var{reg}.
22832 It is an error to use this flag with the frame pointer or stack pointer.
22833 Use of this flag for other registers that have fixed pervasive roles in
22834 the machine's execution model produces disastrous results.
22836 This flag does not have a negative form, because it specifies a
22839 @item -fcall-saved-@var{reg}
22840 @opindex fcall-saved
22841 Treat the register named @var{reg} as an allocable register saved by
22842 functions. It may be allocated even for temporaries or variables that
22843 live across a call. Functions compiled this way save and restore
22844 the register @var{reg} if they use it.
22846 It is an error to use this flag with the frame pointer or stack pointer.
22847 Use of this flag for other registers that have fixed pervasive roles in
22848 the machine's execution model produces disastrous results.
22850 A different sort of disaster results from the use of this flag for
22851 a register in which function values may be returned.
22853 This flag does not have a negative form, because it specifies a
22856 @item -fpack-struct[=@var{n}]
22857 @opindex fpack-struct
22858 Without a value specified, pack all structure members together without
22859 holes. When a value is specified (which must be a small power of two), pack
22860 structure members according to this value, representing the maximum
22861 alignment (that is, objects with default alignment requirements larger than
22862 this are output potentially unaligned at the next fitting location.
22864 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
22865 code that is not binary compatible with code generated without that switch.
22866 Additionally, it makes the code suboptimal.
22867 Use it to conform to a non-default application binary interface.
22869 @item -finstrument-functions
22870 @opindex finstrument-functions
22871 Generate instrumentation calls for entry and exit to functions. Just
22872 after function entry and just before function exit, the following
22873 profiling functions are called with the address of the current
22874 function and its call site. (On some platforms,
22875 @code{__builtin_return_address} does not work beyond the current
22876 function, so the call site information may not be available to the
22877 profiling functions otherwise.)
22880 void __cyg_profile_func_enter (void *this_fn,
22882 void __cyg_profile_func_exit (void *this_fn,
22886 The first argument is the address of the start of the current function,
22887 which may be looked up exactly in the symbol table.
22889 This instrumentation is also done for functions expanded inline in other
22890 functions. The profiling calls indicate where, conceptually, the
22891 inline function is entered and exited. This means that addressable
22892 versions of such functions must be available. If all your uses of a
22893 function are expanded inline, this may mean an additional expansion of
22894 code size. If you use @samp{extern inline} in your C code, an
22895 addressable version of such functions must be provided. (This is
22896 normally the case anyway, but if you get lucky and the optimizer always
22897 expands the functions inline, you might have gotten away without
22898 providing static copies.)
22900 A function may be given the attribute @code{no_instrument_function}, in
22901 which case this instrumentation is not done. This can be used, for
22902 example, for the profiling functions listed above, high-priority
22903 interrupt routines, and any functions from which the profiling functions
22904 cannot safely be called (perhaps signal handlers, if the profiling
22905 routines generate output or allocate memory).
22907 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
22908 @opindex finstrument-functions-exclude-file-list
22910 Set the list of functions that are excluded from instrumentation (see
22911 the description of @code{-finstrument-functions}). If the file that
22912 contains a function definition matches with one of @var{file}, then
22913 that function is not instrumented. The match is done on substrings:
22914 if the @var{file} parameter is a substring of the file name, it is
22915 considered to be a match.
22920 -finstrument-functions-exclude-file-list=/bits/stl,include/sys
22924 excludes any inline function defined in files whose pathnames
22925 contain @code{/bits/stl} or @code{include/sys}.
22927 If, for some reason, you want to include letter @code{','} in one of
22928 @var{sym}, write @code{'\,'}. For example,
22929 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
22930 (note the single quote surrounding the option).
22932 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
22933 @opindex finstrument-functions-exclude-function-list
22935 This is similar to @code{-finstrument-functions-exclude-file-list},
22936 but this option sets the list of function names to be excluded from
22937 instrumentation. The function name to be matched is its user-visible
22938 name, such as @code{vector<int> blah(const vector<int> &)}, not the
22939 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
22940 match is done on substrings: if the @var{sym} parameter is a substring
22941 of the function name, it is considered to be a match. For C99 and C++
22942 extended identifiers, the function name must be given in UTF-8, not
22943 using universal character names.
22945 @item -fstack-check
22946 @opindex fstack-check
22947 Generate code to verify that you do not go beyond the boundary of the
22948 stack. You should specify this flag if you are running in an
22949 environment with multiple threads, but you only rarely need to specify it in
22950 a single-threaded environment since stack overflow is automatically
22951 detected on nearly all systems if there is only one stack.
22953 Note that this switch does not actually cause checking to be done; the
22954 operating system or the language runtime must do that. The switch causes
22955 generation of code to ensure that they see the stack being extended.
22957 You can additionally specify a string parameter: @code{no} means no
22958 checking, @code{generic} means force the use of old-style checking,
22959 @code{specific} means use the best checking method and is equivalent
22960 to bare @option{-fstack-check}.
22962 Old-style checking is a generic mechanism that requires no specific
22963 target support in the compiler but comes with the following drawbacks:
22967 Modified allocation strategy for large objects: they are always
22968 allocated dynamically if their size exceeds a fixed threshold.
22971 Fixed limit on the size of the static frame of functions: when it is
22972 topped by a particular function, stack checking is not reliable and
22973 a warning is issued by the compiler.
22976 Inefficiency: because of both the modified allocation strategy and the
22977 generic implementation, code performance is hampered.
22980 Note that old-style stack checking is also the fallback method for
22981 @code{specific} if no target support has been added in the compiler.
22983 @item -fstack-limit-register=@var{reg}
22984 @itemx -fstack-limit-symbol=@var{sym}
22985 @itemx -fno-stack-limit
22986 @opindex fstack-limit-register
22987 @opindex fstack-limit-symbol
22988 @opindex fno-stack-limit
22989 Generate code to ensure that the stack does not grow beyond a certain value,
22990 either the value of a register or the address of a symbol. If a larger
22991 stack is required, a signal is raised at run time. For most targets,
22992 the signal is raised before the stack overruns the boundary, so
22993 it is possible to catch the signal without taking special precautions.
22995 For instance, if the stack starts at absolute address @samp{0x80000000}
22996 and grows downwards, you can use the flags
22997 @option{-fstack-limit-symbol=__stack_limit} and
22998 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
22999 of 128KB@. Note that this may only work with the GNU linker.
23001 @item -fsplit-stack
23002 @opindex fsplit-stack
23003 Generate code to automatically split the stack before it overflows.
23004 The resulting program has a discontiguous stack which can only
23005 overflow if the program is unable to allocate any more memory. This
23006 is most useful when running threaded programs, as it is no longer
23007 necessary to calculate a good stack size to use for each thread. This
23008 is currently only implemented for the i386 and x86_64 back ends running
23011 When code compiled with @option{-fsplit-stack} calls code compiled
23012 without @option{-fsplit-stack}, there may not be much stack space
23013 available for the latter code to run. If compiling all code,
23014 including library code, with @option{-fsplit-stack} is not an option,
23015 then the linker can fix up these calls so that the code compiled
23016 without @option{-fsplit-stack} always has a large stack. Support for
23017 this is implemented in the gold linker in GNU binutils release 2.21
23020 @item -fleading-underscore
23021 @opindex fleading-underscore
23022 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
23023 change the way C symbols are represented in the object file. One use
23024 is to help link with legacy assembly code.
23026 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
23027 generate code that is not binary compatible with code generated without that
23028 switch. Use it to conform to a non-default application binary interface.
23029 Not all targets provide complete support for this switch.
23031 @item -ftls-model=@var{model}
23032 @opindex ftls-model
23033 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
23034 The @var{model} argument should be one of @code{global-dynamic},
23035 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
23036 Note that the choice is subject to optimization: the compiler may use
23037 a more efficient model for symbols not visible outside of the translation
23038 unit, or if @option{-fpic} is not given on the command line.
23040 The default without @option{-fpic} is @code{initial-exec}; with
23041 @option{-fpic} the default is @code{global-dynamic}.
23043 @item -fvisibility=@var{default|internal|hidden|protected}
23044 @opindex fvisibility
23045 Set the default ELF image symbol visibility to the specified option---all
23046 symbols are marked with this unless overridden within the code.
23047 Using this feature can very substantially improve linking and
23048 load times of shared object libraries, produce more optimized
23049 code, provide near-perfect API export and prevent symbol clashes.
23050 It is @strong{strongly} recommended that you use this in any shared objects
23053 Despite the nomenclature, @code{default} always means public; i.e.,
23054 available to be linked against from outside the shared object.
23055 @code{protected} and @code{internal} are pretty useless in real-world
23056 usage so the only other commonly used option is @code{hidden}.
23057 The default if @option{-fvisibility} isn't specified is
23058 @code{default}, i.e., make every
23059 symbol public---this causes the same behavior as previous versions of
23062 A good explanation of the benefits offered by ensuring ELF
23063 symbols have the correct visibility is given by ``How To Write
23064 Shared Libraries'' by Ulrich Drepper (which can be found at
23065 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
23066 solution made possible by this option to marking things hidden when
23067 the default is public is to make the default hidden and mark things
23068 public. This is the norm with DLLs on Windows and with @option{-fvisibility=hidden}
23069 and @code{__attribute__ ((visibility("default")))} instead of
23070 @code{__declspec(dllexport)} you get almost identical semantics with
23071 identical syntax. This is a great boon to those working with
23072 cross-platform projects.
23074 For those adding visibility support to existing code, you may find
23075 @samp{#pragma GCC visibility} of use. This works by you enclosing
23076 the declarations you wish to set visibility for with (for example)
23077 @samp{#pragma GCC visibility push(hidden)} and
23078 @samp{#pragma GCC visibility pop}.
23079 Bear in mind that symbol visibility should be viewed @strong{as
23080 part of the API interface contract} and thus all new code should
23081 always specify visibility when it is not the default; i.e., declarations
23082 only for use within the local DSO should @strong{always} be marked explicitly
23083 as hidden as so to avoid PLT indirection overheads---making this
23084 abundantly clear also aids readability and self-documentation of the code.
23085 Note that due to ISO C++ specification requirements, @code{operator new} and
23086 @code{operator delete} must always be of default visibility.
23088 Be aware that headers from outside your project, in particular system
23089 headers and headers from any other library you use, may not be
23090 expecting to be compiled with visibility other than the default. You
23091 may need to explicitly say @samp{#pragma GCC visibility push(default)}
23092 before including any such headers.
23094 @samp{extern} declarations are not affected by @option{-fvisibility}, so
23095 a lot of code can be recompiled with @option{-fvisibility=hidden} with
23096 no modifications. However, this means that calls to @code{extern}
23097 functions with no explicit visibility use the PLT, so it is more
23098 effective to use @code{__attribute ((visibility))} and/or
23099 @code{#pragma GCC visibility} to tell the compiler which @code{extern}
23100 declarations should be treated as hidden.
23102 Note that @option{-fvisibility} does affect C++ vague linkage
23103 entities. This means that, for instance, an exception class that is
23104 be thrown between DSOs must be explicitly marked with default
23105 visibility so that the @samp{type_info} nodes are unified between
23108 An overview of these techniques, their benefits and how to use them
23109 is at @uref{http://gcc.gnu.org/@/wiki/@/Visibility}.
23111 @item -fstrict-volatile-bitfields
23112 @opindex fstrict-volatile-bitfields
23113 This option should be used if accesses to volatile bit-fields (or other
23114 structure fields, although the compiler usually honors those types
23115 anyway) should use a single access of the width of the
23116 field's type, aligned to a natural alignment if possible. For
23117 example, targets with memory-mapped peripheral registers might require
23118 all such accesses to be 16 bits wide; with this flag you can
23119 declare all peripheral bit-fields as @code{unsigned short} (assuming short
23120 is 16 bits on these targets) to force GCC to use 16-bit accesses
23121 instead of, perhaps, a more efficient 32-bit access.
23123 If this option is disabled, the compiler uses the most efficient
23124 instruction. In the previous example, that might be a 32-bit load
23125 instruction, even though that accesses bytes that do not contain
23126 any portion of the bit-field, or memory-mapped registers unrelated to
23127 the one being updated.
23129 In some cases, such as when the @code{packed} attribute is applied to a
23130 structure field, it may not be possible to access the field with a single
23131 read or write that is correctly aligned for the target machine. In this
23132 case GCC falls back to generating multiple accesses rather than code that
23133 will fault or truncate the result at run time.
23135 Note: Due to restrictions of the C/C++11 memory model, write accesses are
23136 not allowed to touch non bit-field members. It is therefore recommended
23137 to define all bits of the field's type as bit-field members.
23139 The default value of this option is determined by the application binary
23140 interface for the target processor.
23142 @item -fsync-libcalls
23143 @opindex fsync-libcalls
23144 This option controls whether any out-of-line instance of the @code{__sync}
23145 family of functions may be used to implement the C++11 @code{__atomic}
23146 family of functions.
23148 The default value of this option is enabled, thus the only useful form
23149 of the option is @option{-fno-sync-libcalls}. This option is used in
23150 the implementation of the @file{libatomic} runtime library.
23156 @node Environment Variables
23157 @section Environment Variables Affecting GCC
23158 @cindex environment variables
23160 @c man begin ENVIRONMENT
23161 This section describes several environment variables that affect how GCC
23162 operates. Some of them work by specifying directories or prefixes to use
23163 when searching for various kinds of files. Some are used to specify other
23164 aspects of the compilation environment.
23166 Note that you can also specify places to search using options such as
23167 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
23168 take precedence over places specified using environment variables, which
23169 in turn take precedence over those specified by the configuration of GCC@.
23170 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
23171 GNU Compiler Collection (GCC) Internals}.
23176 @c @itemx LC_COLLATE
23178 @c @itemx LC_MONETARY
23179 @c @itemx LC_NUMERIC
23184 @c @findex LC_COLLATE
23185 @findex LC_MESSAGES
23186 @c @findex LC_MONETARY
23187 @c @findex LC_NUMERIC
23191 These environment variables control the way that GCC uses
23192 localization information which allows GCC to work with different
23193 national conventions. GCC inspects the locale categories
23194 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
23195 so. These locale categories can be set to any value supported by your
23196 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
23197 Kingdom encoded in UTF-8.
23199 The @env{LC_CTYPE} environment variable specifies character
23200 classification. GCC uses it to determine the character boundaries in
23201 a string; this is needed for some multibyte encodings that contain quote
23202 and escape characters that are otherwise interpreted as a string
23205 The @env{LC_MESSAGES} environment variable specifies the language to
23206 use in diagnostic messages.
23208 If the @env{LC_ALL} environment variable is set, it overrides the value
23209 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
23210 and @env{LC_MESSAGES} default to the value of the @env{LANG}
23211 environment variable. If none of these variables are set, GCC
23212 defaults to traditional C English behavior.
23216 If @env{TMPDIR} is set, it specifies the directory to use for temporary
23217 files. GCC uses temporary files to hold the output of one stage of
23218 compilation which is to be used as input to the next stage: for example,
23219 the output of the preprocessor, which is the input to the compiler
23222 @item GCC_COMPARE_DEBUG
23223 @findex GCC_COMPARE_DEBUG
23224 Setting @env{GCC_COMPARE_DEBUG} is nearly equivalent to passing
23225 @option{-fcompare-debug} to the compiler driver. See the documentation
23226 of this option for more details.
23228 @item GCC_EXEC_PREFIX
23229 @findex GCC_EXEC_PREFIX
23230 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
23231 names of the subprograms executed by the compiler. No slash is added
23232 when this prefix is combined with the name of a subprogram, but you can
23233 specify a prefix that ends with a slash if you wish.
23235 If @env{GCC_EXEC_PREFIX} is not set, GCC attempts to figure out
23236 an appropriate prefix to use based on the pathname it is invoked with.
23238 If GCC cannot find the subprogram using the specified prefix, it
23239 tries looking in the usual places for the subprogram.
23241 The default value of @env{GCC_EXEC_PREFIX} is
23242 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
23243 the installed compiler. In many cases @var{prefix} is the value
23244 of @code{prefix} when you ran the @file{configure} script.
23246 Other prefixes specified with @option{-B} take precedence over this prefix.
23248 This prefix is also used for finding files such as @file{crt0.o} that are
23251 In addition, the prefix is used in an unusual way in finding the
23252 directories to search for header files. For each of the standard
23253 directories whose name normally begins with @samp{/usr/local/lib/gcc}
23254 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
23255 replacing that beginning with the specified prefix to produce an
23256 alternate directory name. Thus, with @option{-Bfoo/}, GCC searches
23257 @file{foo/bar} just before it searches the standard directory
23258 @file{/usr/local/lib/bar}.
23259 If a standard directory begins with the configured
23260 @var{prefix} then the value of @var{prefix} is replaced by
23261 @env{GCC_EXEC_PREFIX} when looking for header files.
23263 @item COMPILER_PATH
23264 @findex COMPILER_PATH
23265 The value of @env{COMPILER_PATH} is a colon-separated list of
23266 directories, much like @env{PATH}. GCC tries the directories thus
23267 specified when searching for subprograms, if it can't find the
23268 subprograms using @env{GCC_EXEC_PREFIX}.
23271 @findex LIBRARY_PATH
23272 The value of @env{LIBRARY_PATH} is a colon-separated list of
23273 directories, much like @env{PATH}. When configured as a native compiler,
23274 GCC tries the directories thus specified when searching for special
23275 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
23276 using GCC also uses these directories when searching for ordinary
23277 libraries for the @option{-l} option (but directories specified with
23278 @option{-L} come first).
23282 @cindex locale definition
23283 This variable is used to pass locale information to the compiler. One way in
23284 which this information is used is to determine the character set to be used
23285 when character literals, string literals and comments are parsed in C and C++.
23286 When the compiler is configured to allow multibyte characters,
23287 the following values for @env{LANG} are recognized:
23291 Recognize JIS characters.
23293 Recognize SJIS characters.
23295 Recognize EUCJP characters.
23298 If @env{LANG} is not defined, or if it has some other value, then the
23299 compiler uses @code{mblen} and @code{mbtowc} as defined by the default locale to
23300 recognize and translate multibyte characters.
23304 Some additional environment variables affect the behavior of the
23307 @include cppenv.texi
23311 @node Precompiled Headers
23312 @section Using Precompiled Headers
23313 @cindex precompiled headers
23314 @cindex speed of compilation
23316 Often large projects have many header files that are included in every
23317 source file. The time the compiler takes to process these header files
23318 over and over again can account for nearly all of the time required to
23319 build the project. To make builds faster, GCC allows you to
23320 @dfn{precompile} a header file.
23322 To create a precompiled header file, simply compile it as you would any
23323 other file, if necessary using the @option{-x} option to make the driver
23324 treat it as a C or C++ header file. You may want to use a
23325 tool like @command{make} to keep the precompiled header up-to-date when
23326 the headers it contains change.
23328 A precompiled header file is searched for when @code{#include} is
23329 seen in the compilation. As it searches for the included file
23330 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
23331 compiler looks for a precompiled header in each directory just before it
23332 looks for the include file in that directory. The name searched for is
23333 the name specified in the @code{#include} with @samp{.gch} appended. If
23334 the precompiled header file can't be used, it is ignored.
23336 For instance, if you have @code{#include "all.h"}, and you have
23337 @file{all.h.gch} in the same directory as @file{all.h}, then the
23338 precompiled header file is used if possible, and the original
23339 header is used otherwise.
23341 Alternatively, you might decide to put the precompiled header file in a
23342 directory and use @option{-I} to ensure that directory is searched
23343 before (or instead of) the directory containing the original header.
23344 Then, if you want to check that the precompiled header file is always
23345 used, you can put a file of the same name as the original header in this
23346 directory containing an @code{#error} command.
23348 This also works with @option{-include}. So yet another way to use
23349 precompiled headers, good for projects not designed with precompiled
23350 header files in mind, is to simply take most of the header files used by
23351 a project, include them from another header file, precompile that header
23352 file, and @option{-include} the precompiled header. If the header files
23353 have guards against multiple inclusion, they are skipped because
23354 they've already been included (in the precompiled header).
23356 If you need to precompile the same header file for different
23357 languages, targets, or compiler options, you can instead make a
23358 @emph{directory} named like @file{all.h.gch}, and put each precompiled
23359 header in the directory, perhaps using @option{-o}. It doesn't matter
23360 what you call the files in the directory; every precompiled header in
23361 the directory is considered. The first precompiled header
23362 encountered in the directory that is valid for this compilation is
23363 used; they're searched in no particular order.
23365 There are many other possibilities, limited only by your imagination,
23366 good sense, and the constraints of your build system.
23368 A precompiled header file can be used only when these conditions apply:
23372 Only one precompiled header can be used in a particular compilation.
23375 A precompiled header can't be used once the first C token is seen. You
23376 can have preprocessor directives before a precompiled header; you cannot
23377 include a precompiled header from inside another header.
23380 The precompiled header file must be produced for the same language as
23381 the current compilation. You can't use a C precompiled header for a C++
23385 The precompiled header file must have been produced by the same compiler
23386 binary as the current compilation is using.
23389 Any macros defined before the precompiled header is included must
23390 either be defined in the same way as when the precompiled header was
23391 generated, or must not affect the precompiled header, which usually
23392 means that they don't appear in the precompiled header at all.
23394 The @option{-D} option is one way to define a macro before a
23395 precompiled header is included; using a @code{#define} can also do it.
23396 There are also some options that define macros implicitly, like
23397 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
23400 @item If debugging information is output when using the precompiled
23401 header, using @option{-g} or similar, the same kind of debugging information
23402 must have been output when building the precompiled header. However,
23403 a precompiled header built using @option{-g} can be used in a compilation
23404 when no debugging information is being output.
23406 @item The same @option{-m} options must generally be used when building
23407 and using the precompiled header. @xref{Submodel Options},
23408 for any cases where this rule is relaxed.
23410 @item Each of the following options must be the same when building and using
23411 the precompiled header:
23413 @gccoptlist{-fexceptions}
23416 Some other command-line options starting with @option{-f},
23417 @option{-p}, or @option{-O} must be defined in the same way as when
23418 the precompiled header was generated. At present, it's not clear
23419 which options are safe to change and which are not; the safest choice
23420 is to use exactly the same options when generating and using the
23421 precompiled header. The following are known to be safe:
23423 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
23424 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
23425 -fsched-verbose=@var{number} -fschedule-insns -fvisibility= @gol
23430 For all of these except the last, the compiler automatically
23431 ignores the precompiled header if the conditions aren't met. If you
23432 find an option combination that doesn't work and doesn't cause the
23433 precompiled header to be ignored, please consider filing a bug report,
23436 If you do use differing options when generating and using the
23437 precompiled header, the actual behavior is a mixture of the
23438 behavior for the options. For instance, if you use @option{-g} to
23439 generate the precompiled header but not when using it, you may or may
23440 not get debugging information for routines in the precompiled header.