1 @c Copyright (C) 1988-2015 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{} 1988-2015 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. @command{g++} accepts mostly the same options as @command{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 -fopenacc -fopenmp -fopenmp-simd @gol
172 -fms-extensions -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 -fsized-deallocation @gol
191 -fstats -ftemplate-backtrace-limit=@var{n} @gol
192 -ftemplate-depth=@var{n} @gol
193 -fno-threadsafe-statics -fuse-cxa-atexit @gol
194 -fno-weak -nostdinc++ @gol
195 -fvisibility-inlines-hidden @gol
196 -fvtable-verify=@r{[}std@r{|}preinit@r{|}none@r{]} @gol
197 -fvtv-counts -fvtv-debug @gol
198 -fvisibility-ms-compat @gol
199 -fext-numeric-literals @gol
200 -Wabi=@var{n} -Wabi-tag -Wconversion-null -Wctor-dtor-privacy @gol
201 -Wdelete-non-virtual-dtor -Wliteral-suffix -Wnarrowing @gol
202 -Wnoexcept -Wnon-virtual-dtor -Wreorder @gol
203 -Weffc++ -Wstrict-null-sentinel @gol
204 -Wno-non-template-friend -Wold-style-cast @gol
205 -Woverloaded-virtual -Wno-pmf-conversions @gol
208 @item Objective-C and Objective-C++ Language Options
209 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
210 Objective-C and Objective-C++ Dialects}.
211 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
212 -fgnu-runtime -fnext-runtime @gol
213 -fno-nil-receivers @gol
214 -fobjc-abi-version=@var{n} @gol
215 -fobjc-call-cxx-cdtors @gol
216 -fobjc-direct-dispatch @gol
217 -fobjc-exceptions @gol
220 -fobjc-std=objc1 @gol
221 -fno-local-ivars @gol
222 -fivar-visibility=@r{[}public@r{|}protected@r{|}private@r{|}package@r{]} @gol
223 -freplace-objc-classes @gol
226 -Wassign-intercept @gol
227 -Wno-protocol -Wselector @gol
228 -Wstrict-selector-match @gol
229 -Wundeclared-selector}
231 @item Language Independent Options
232 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
233 @gccoptlist{-fmessage-length=@var{n} @gol
234 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
235 -fdiagnostics-color=@r{[}auto@r{|}never@r{|}always@r{]} @gol
236 -fno-diagnostics-show-option -fno-diagnostics-show-caret}
238 @item Warning Options
239 @xref{Warning Options,,Options to Request or Suppress Warnings}.
240 @gccoptlist{-fsyntax-only -fmax-errors=@var{n} -Wpedantic @gol
241 -pedantic-errors @gol
242 -w -Wextra -Wall -Waddress -Waggregate-return @gol
243 -Waggressive-loop-optimizations -Warray-bounds -Warray-bounds=@var{n} @gol
245 -Wno-attributes -Wno-builtin-macro-redefined @gol
246 -Wc90-c99-compat -Wc99-c11-compat @gol
247 -Wc++-compat -Wc++11-compat -Wc++14-compat -Wcast-align -Wcast-qual @gol
248 -Wchar-subscripts -Wclobbered -Wcomment -Wconditionally-supported @gol
249 -Wconversion -Wcoverage-mismatch -Wdate-time -Wdelete-incomplete -Wno-cpp @gol
250 -Wno-deprecated -Wno-deprecated-declarations -Wno-designated-init @gol
251 -Wdisabled-optimization @gol
252 -Wno-discarded-qualifiers -Wno-discarded-array-qualifiers @gol
253 -Wno-div-by-zero -Wdouble-promotion -Wempty-body -Wenum-compare @gol
254 -Wno-endif-labels -Werror -Werror=* @gol
255 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
256 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
257 -Wformat-security -Wformat-signedness -Wformat-y2k @gol
258 -Wframe-larger-than=@var{len} -Wno-free-nonheap-object -Wjump-misses-init @gol
259 -Wignored-qualifiers -Wincompatible-pointer-types @gol
260 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
261 -Winit-self -Winline -Wno-int-conversion @gol
262 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
263 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
264 -Wlogical-op -Wlogical-not-parentheses -Wlong-long @gol
265 -Wmain -Wmaybe-uninitialized -Wmemset-transposed-args @gol
266 -Wmisleading-indentation -Wmissing-braces @gol
267 -Wmissing-field-initializers -Wmissing-include-dirs @gol
268 -Wno-multichar -Wnonnull -Wnormalized=@r{[}none@r{|}id@r{|}nfc@r{|}nfkc@r{]} @gol
269 -Wodr -Wno-overflow -Wopenmp-simd @gol
270 -Woverride-init-side-effects @gol
271 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
272 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
273 -Wpointer-arith -Wno-pointer-to-int-cast @gol
274 -Wredundant-decls -Wno-return-local-addr @gol
275 -Wreturn-type -Wsequence-point -Wshadow -Wno-shadow-ivar @gol
276 -Wshift-count-negative -Wshift-count-overflow -Wshift-negative-value @gol
277 -Wsign-compare -Wsign-conversion -Wfloat-conversion @gol
278 -Wsizeof-pointer-memaccess -Wsizeof-array-argument @gol
279 -Wstack-protector -Wstack-usage=@var{len} -Wstrict-aliasing @gol
280 -Wstrict-aliasing=n @gol -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
281 -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{]} @gol
282 -Wsuggest-final-types @gol -Wsuggest-final-methods -Wsuggest-override @gol
283 -Wmissing-format-attribute @gol
284 -Wswitch -Wswitch-default -Wswitch-enum -Wswitch-bool -Wsync-nand @gol
285 -Wsystem-headers -Wtrampolines -Wtrigraphs -Wtype-limits -Wundef @gol
286 -Wuninitialized -Wunknown-pragmas -Wno-pragmas @gol
287 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
288 -Wunused-label -Wunused-local-typedefs -Wunused-parameter @gol
289 -Wno-unused-result -Wunused-value @gol -Wunused-variable @gol
290 -Wunused-but-set-parameter -Wunused-but-set-variable @gol
291 -Wuseless-cast -Wvariadic-macros -Wvector-operation-performance @gol
292 -Wvla -Wvolatile-register-var -Wwrite-strings @gol
293 -Wzero-as-null-pointer-constant}
295 @item C and Objective-C-only Warning Options
296 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
297 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
298 -Wold-style-declaration -Wold-style-definition @gol
299 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
300 -Wdeclaration-after-statement -Wpointer-sign}
302 @item Debugging Options
303 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
304 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
305 -fsanitize=@var{style} -fsanitize-recover -fsanitize-recover=@var{style} @gol
306 -fasan-shadow-offset=@var{number} -fsanitize-sections=@var{s1},@var{s2},... @gol
307 -fsanitize-undefined-trap-on-error @gol
308 -fcheck-pointer-bounds -fchkp-check-incomplete-type @gol
309 -fchkp-first-field-has-own-bounds -fchkp-narrow-bounds @gol
310 -fchkp-narrow-to-innermost-array -fchkp-optimize @gol
311 -fchkp-use-fast-string-functions -fchkp-use-nochk-string-functions @gol
312 -fchkp-use-static-bounds -fchkp-use-static-const-bounds @gol
313 -fchkp-treat-zero-dynamic-size-as-infinite -fchkp-check-read @gol
314 -fchkp-check-read -fchkp-check-write -fchkp-store-bounds @gol
315 -fchkp-instrument-calls -fchkp-instrument-marked-only @gol
316 -fchkp-use-wrappers @gol
317 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
318 -fdisable-ipa-@var{pass_name} @gol
319 -fdisable-rtl-@var{pass_name} @gol
320 -fdisable-rtl-@var{pass-name}=@var{range-list} @gol
321 -fdisable-tree-@var{pass_name} @gol
322 -fdisable-tree-@var{pass-name}=@var{range-list} @gol
323 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
324 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
325 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
326 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
328 -fdump-statistics @gol
330 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
331 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
332 -fdump-tree-cfg -fdump-tree-alias @gol
334 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
335 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
336 -fdump-tree-gimple@r{[}-raw@r{]} @gol
337 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
338 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
339 -fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
340 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
341 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
342 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
343 -fdump-tree-nrv -fdump-tree-vect @gol
344 -fdump-tree-sink @gol
345 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
346 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
347 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
348 -fdump-tree-vtable-verify @gol
349 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
350 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
351 -fdump-final-insns=@var{file} @gol
352 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
353 -feliminate-dwarf2-dups -fno-eliminate-unused-debug-types @gol
354 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
355 -fenable-@var{kind}-@var{pass} @gol
356 -fenable-@var{kind}-@var{pass}=@var{range-list} @gol
357 -fdebug-types-section -fmem-report-wpa @gol
358 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
360 -fopt-info-@var{options}@r{[}=@var{file}@r{]} @gol
361 -frandom-seed=@var{number} -fsched-verbose=@var{n} @gol
362 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
363 -fstack-usage -ftest-coverage -ftime-report -fvar-tracking @gol
364 -fvar-tracking-assignments -fvar-tracking-assignments-toggle @gol
365 -g -g@var{level} -gtoggle -gcoff -gdwarf-@var{version} @gol
366 -ggdb -grecord-gcc-switches -gno-record-gcc-switches @gol
367 -gstabs -gstabs+ -gstrict-dwarf -gno-strict-dwarf @gol
368 -gvms -gxcoff -gxcoff+ -gz@r{[}=@var{type}@r{]} @gol
369 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
370 -fdebug-prefix-map=@var{old}=@var{new} @gol
371 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
372 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
373 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
374 -print-multi-directory -print-multi-lib -print-multi-os-directory @gol
375 -print-prog-name=@var{program} -print-search-dirs -Q @gol
376 -print-sysroot -print-sysroot-headers-suffix @gol
377 -save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
379 @item Optimization Options
380 @xref{Optimize Options,,Options that Control Optimization}.
381 @gccoptlist{-faggressive-loop-optimizations -falign-functions[=@var{n}] @gol
382 -falign-jumps[=@var{n}] @gol
383 -falign-labels[=@var{n}] -falign-loops[=@var{n}] @gol
384 -fassociative-math -fauto-profile -fauto-profile[=@var{path}] @gol
385 -fauto-inc-dec -fbranch-probabilities @gol
386 -fbranch-target-load-optimize -fbranch-target-load-optimize2 @gol
387 -fbtr-bb-exclusive -fcaller-saves @gol
388 -fcheck-data-deps -fcombine-stack-adjustments -fconserve-stack @gol
389 -fcompare-elim -fcprop-registers -fcrossjumping @gol
390 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules @gol
391 -fcx-limited-range @gol
392 -fdata-sections -fdce -fdelayed-branch @gol
393 -fdelete-null-pointer-checks -fdevirtualize -fdevirtualize-speculatively @gol
394 -fdevirtualize-at-ltrans -fdse @gol
395 -fearly-inlining -fipa-sra -fexpensive-optimizations -ffat-lto-objects @gol
396 -ffast-math -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
397 -fforward-propagate -ffp-contract=@var{style} -ffunction-sections @gol
398 -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm -fgraphite-identity @gol
399 -fgcse-sm -fhoist-adjacent-loads -fif-conversion @gol
400 -fif-conversion2 -findirect-inlining @gol
401 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
402 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-cp-alignment @gol
403 -fipa-pta -fipa-profile -fipa-pure-const -fipa-reference -fipa-icf @gol
404 -fira-algorithm=@var{algorithm} @gol
405 -fira-region=@var{region} -fira-hoist-pressure @gol
406 -fira-loop-pressure -fno-ira-share-save-slots @gol
407 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
408 -fisolate-erroneous-paths-dereference -fisolate-erroneous-paths-attribute @gol
409 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
410 -flive-range-shrinkage @gol
411 -floop-block -floop-interchange -floop-strip-mine @gol
412 -floop-unroll-and-jam -floop-nest-optimize @gol
413 -floop-parallelize-all -flra-remat -flto -flto-compression-level @gol
414 -flto-partition=@var{alg} -flto-report -flto-report-wpa -fmerge-all-constants @gol
415 -fmerge-constants -fmodulo-sched -fmodulo-sched-allow-regmoves @gol
416 -fmove-loop-invariants -fno-branch-count-reg @gol
417 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
418 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
419 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
420 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
421 -fomit-frame-pointer -foptimize-sibling-calls @gol
422 -fpartial-inlining -fpeel-loops -fpredictive-commoning @gol
423 -fprefetch-loop-arrays -fprofile-report @gol
424 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
425 -fprofile-generate=@var{path} @gol
426 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
427 -fprofile-reorder-functions @gol
428 -freciprocal-math -free -frename-registers -freorder-blocks @gol
429 -freorder-blocks-and-partition -freorder-functions @gol
430 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
431 -frounding-math -fsched2-use-superblocks -fsched-pressure @gol
432 -fsched-spec-load -fsched-spec-load-dangerous @gol
433 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
434 -fsched-group-heuristic -fsched-critical-path-heuristic @gol
435 -fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
436 -fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
437 -fschedule-fusion @gol
438 -fschedule-insns -fschedule-insns2 -fsection-anchors @gol
439 -fselective-scheduling -fselective-scheduling2 @gol
440 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
441 -fsemantic-interposition @gol
442 -fshrink-wrap -fsignaling-nans -fsingle-precision-constant @gol
443 -fsplit-ivs-in-unroller -fsplit-wide-types -fssa-phiopt @gol
444 -fstack-protector -fstack-protector-all -fstack-protector-strong @gol
445 -fstack-protector-explicit -fstdarg-opt -fstrict-aliasing @gol
446 -fstrict-overflow -fthread-jumps -ftracer -ftree-bit-ccp @gol
447 -ftree-builtin-call-dce -ftree-ccp -ftree-ch @gol
448 -ftree-coalesce-inline-vars -ftree-coalesce-vars -ftree-copy-prop @gol
449 -ftree-copyrename -ftree-dce -ftree-dominator-opts -ftree-dse @gol
450 -ftree-forwprop -ftree-fre -ftree-loop-if-convert @gol
451 -ftree-loop-if-convert-stores -ftree-loop-im @gol
452 -ftree-phiprop -ftree-loop-distribution -ftree-loop-distribute-patterns @gol
453 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
454 -ftree-loop-vectorize @gol
455 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-partial-pre -ftree-pta @gol
456 -ftree-reassoc -ftree-sink -ftree-slsr -ftree-sra @gol
457 -ftree-switch-conversion -ftree-tail-merge -ftree-ter @gol
458 -ftree-vectorize -ftree-vrp @gol
459 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
460 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
461 -fipa-ra -fvariable-expansion-in-unroller -fvect-cost-model -fvpt @gol
462 -fweb -fwhole-program -fwpa -fuse-linker-plugin @gol
463 --param @var{name}=@var{value}
464 -O -O0 -O1 -O2 -O3 -Os -Ofast -Og}
466 @item Preprocessor Options
467 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
468 @gccoptlist{-A@var{question}=@var{answer} @gol
469 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
470 -C -dD -dI -dM -dN @gol
471 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
472 -idirafter @var{dir} @gol
473 -include @var{file} -imacros @var{file} @gol
474 -iprefix @var{file} -iwithprefix @var{dir} @gol
475 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
476 -imultilib @var{dir} -isysroot @var{dir} @gol
477 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
478 -P -fdebug-cpp -ftrack-macro-expansion -fworking-directory @gol
479 -remap -trigraphs -undef -U@var{macro} @gol
480 -Wp,@var{option} -Xpreprocessor @var{option} -no-integrated-cpp}
482 @item Assembler Option
483 @xref{Assembler Options,,Passing Options to the Assembler}.
484 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
487 @xref{Link Options,,Options for Linking}.
488 @gccoptlist{@var{object-file-name} -fuse-ld=@var{linker} -l@var{library} @gol
489 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
490 -s -static -static-libgcc -static-libstdc++ @gol
491 -static-libasan -static-libtsan -static-liblsan -static-libubsan @gol
492 -static-libmpx -static-libmpxwrappers @gol
493 -shared -shared-libgcc -symbolic @gol
494 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
495 -u @var{symbol} -z @var{keyword}}
497 @item Directory Options
498 @xref{Directory Options,,Options for Directory Search}.
499 @gccoptlist{-B@var{prefix} -I@var{dir} -iplugindir=@var{dir} @gol
500 -iquote@var{dir} -L@var{dir} -specs=@var{file} -I- @gol
501 --sysroot=@var{dir} --no-sysroot-suffix}
503 @item Machine Dependent Options
504 @xref{Submodel Options,,Hardware Models and Configurations}.
505 @c This list is ordered alphanumerically by subsection name.
506 @c Try and put the significant identifier (CPU or system) first,
507 @c so users have a clue at guessing where the ones they want will be.
509 @emph{AArch64 Options}
510 @gccoptlist{-mabi=@var{name} -mbig-endian -mlittle-endian @gol
511 -mgeneral-regs-only @gol
512 -mcmodel=tiny -mcmodel=small -mcmodel=large @gol
514 -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
515 -mtls-dialect=desc -mtls-dialect=traditional @gol
516 -mfix-cortex-a53-835769 -mno-fix-cortex-a53-835769 @gol
517 -mfix-cortex-a53-843419 -mno-fix-cortex-a53-843419 @gol
518 -march=@var{name} -mcpu=@var{name} -mtune=@var{name}}
520 @emph{Adapteva Epiphany Options}
521 @gccoptlist{-mhalf-reg-file -mprefer-short-insn-regs @gol
522 -mbranch-cost=@var{num} -mcmove -mnops=@var{num} -msoft-cmpsf @gol
523 -msplit-lohi -mpost-inc -mpost-modify -mstack-offset=@var{num} @gol
524 -mround-nearest -mlong-calls -mshort-calls -msmall16 @gol
525 -mfp-mode=@var{mode} -mvect-double -max-vect-align=@var{num} @gol
526 -msplit-vecmove-early -m1reg-@var{reg}}
529 @gccoptlist{-mbarrel-shifter @gol
530 -mcpu=@var{cpu} -mA6 -mARC600 -mA7 -mARC700 @gol
531 -mdpfp -mdpfp-compact -mdpfp-fast -mno-dpfp-lrsr @gol
532 -mea -mno-mpy -mmul32x16 -mmul64 @gol
533 -mnorm -mspfp -mspfp-compact -mspfp-fast -msimd -msoft-float -mswap @gol
534 -mcrc -mdsp-packa -mdvbf -mlock -mmac-d16 -mmac-24 -mrtsc -mswape @gol
535 -mtelephony -mxy -misize -mannotate-align -marclinux -marclinux_prof @gol
536 -mepilogue-cfi -mlong-calls -mmedium-calls -msdata @gol
537 -mucb-mcount -mvolatile-cache @gol
538 -malign-call -mauto-modify-reg -mbbit-peephole -mno-brcc @gol
539 -mcase-vector-pcrel -mcompact-casesi -mno-cond-exec -mearly-cbranchsi @gol
540 -mexpand-adddi -mindexed-loads -mlra -mlra-priority-none @gol
541 -mlra-priority-compact mlra-priority-noncompact -mno-millicode @gol
542 -mmixed-code -mq-class -mRcq -mRcw -msize-level=@var{level} @gol
543 -mtune=@var{cpu} -mmultcost=@var{num} -munalign-prob-threshold=@var{probability}}
546 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
547 -mabi=@var{name} @gol
548 -mapcs-stack-check -mno-apcs-stack-check @gol
549 -mapcs-float -mno-apcs-float @gol
550 -mapcs-reentrant -mno-apcs-reentrant @gol
551 -msched-prolog -mno-sched-prolog @gol
552 -mlittle-endian -mbig-endian @gol
553 -mfloat-abi=@var{name} @gol
554 -mfp16-format=@var{name}
555 -mthumb-interwork -mno-thumb-interwork @gol
556 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
557 -mtune=@var{name} -mprint-tune-info @gol
558 -mstructure-size-boundary=@var{n} @gol
559 -mabort-on-noreturn @gol
560 -mlong-calls -mno-long-calls @gol
561 -msingle-pic-base -mno-single-pic-base @gol
562 -mpic-register=@var{reg} @gol
563 -mnop-fun-dllimport @gol
564 -mpoke-function-name @gol
566 -mtpcs-frame -mtpcs-leaf-frame @gol
567 -mcaller-super-interworking -mcallee-super-interworking @gol
568 -mtp=@var{name} -mtls-dialect=@var{dialect} @gol
569 -mword-relocations @gol
570 -mfix-cortex-m3-ldrd @gol
571 -munaligned-access @gol
572 -mneon-for-64bits @gol
573 -mslow-flash-data @gol
574 -masm-syntax-unified @gol
578 @gccoptlist{-mmcu=@var{mcu} -maccumulate-args -mbranch-cost=@var{cost} @gol
579 -mcall-prologues -mint8 -mn_flash=@var{size} -mno-interrupts @gol
580 -mrelax -mrmw -mstrict-X -mtiny-stack -nodevicelib -Waddr-space-convert}
582 @emph{Blackfin Options}
583 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
584 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
585 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
586 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
587 -mno-id-shared-library -mshared-library-id=@var{n} @gol
588 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
589 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
590 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
594 @gccoptlist{-mbig-endian -mlittle-endian -march=@var{cpu} @gol
595 -msim -msdata=@var{sdata-type}}
598 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
599 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
600 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
601 -mstack-align -mdata-align -mconst-align @gol
602 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
603 -melf -maout -melinux -mlinux -sim -sim2 @gol
604 -mmul-bug-workaround -mno-mul-bug-workaround}
607 @gccoptlist{-mmac @gol
608 -mcr16cplus -mcr16c @gol
609 -msim -mint32 -mbit-ops
610 -mdata-model=@var{model}}
612 @emph{Darwin Options}
613 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
614 -arch_only -bind_at_load -bundle -bundle_loader @gol
615 -client_name -compatibility_version -current_version @gol
617 -dependency-file -dylib_file -dylinker_install_name @gol
618 -dynamic -dynamiclib -exported_symbols_list @gol
619 -filelist -flat_namespace -force_cpusubtype_ALL @gol
620 -force_flat_namespace -headerpad_max_install_names @gol
622 -image_base -init -install_name -keep_private_externs @gol
623 -multi_module -multiply_defined -multiply_defined_unused @gol
624 -noall_load -no_dead_strip_inits_and_terms @gol
625 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
626 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
627 -private_bundle -read_only_relocs -sectalign @gol
628 -sectobjectsymbols -whyload -seg1addr @gol
629 -sectcreate -sectobjectsymbols -sectorder @gol
630 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
631 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
632 -segprot -segs_read_only_addr -segs_read_write_addr @gol
633 -single_module -static -sub_library -sub_umbrella @gol
634 -twolevel_namespace -umbrella -undefined @gol
635 -unexported_symbols_list -weak_reference_mismatches @gol
636 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
637 -mkernel -mone-byte-bool}
639 @emph{DEC Alpha Options}
640 @gccoptlist{-mno-fp-regs -msoft-float @gol
641 -mieee -mieee-with-inexact -mieee-conformant @gol
642 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
643 -mtrap-precision=@var{mode} -mbuild-constants @gol
644 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
645 -mbwx -mmax -mfix -mcix @gol
646 -mfloat-vax -mfloat-ieee @gol
647 -mexplicit-relocs -msmall-data -mlarge-data @gol
648 -msmall-text -mlarge-text @gol
649 -mmemory-latency=@var{time}}
652 @gccoptlist{-msmall-model -mno-lsim}
655 @gccoptlist{-msim -mlra}
658 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
659 -mhard-float -msoft-float @gol
660 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
661 -mdouble -mno-double @gol
662 -mmedia -mno-media -mmuladd -mno-muladd @gol
663 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
664 -mlinked-fp -mlong-calls -malign-labels @gol
665 -mlibrary-pic -macc-4 -macc-8 @gol
666 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
667 -moptimize-membar -mno-optimize-membar @gol
668 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
669 -mvliw-branch -mno-vliw-branch @gol
670 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
671 -mno-nested-cond-exec -mtomcat-stats @gol
675 @emph{GNU/Linux Options}
676 @gccoptlist{-mglibc -muclibc -mmusl -mbionic -mandroid @gol
677 -tno-android-cc -tno-android-ld}
679 @emph{H8/300 Options}
680 @gccoptlist{-mrelax -mh -ms -mn -mexr -mno-exr -mint32 -malign-300}
683 @gccoptlist{-march=@var{architecture-type} @gol
684 -mdisable-fpregs -mdisable-indexing @gol
685 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
686 -mfixed-range=@var{register-range} @gol
687 -mjump-in-delay -mlinker-opt -mlong-calls @gol
688 -mlong-load-store -mno-disable-fpregs @gol
689 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
690 -mno-jump-in-delay -mno-long-load-store @gol
691 -mno-portable-runtime -mno-soft-float @gol
692 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
693 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
694 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
695 -munix=@var{unix-std} -nolibdld -static -threads}
698 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
699 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
700 -mconstant-gp -mauto-pic -mfused-madd @gol
701 -minline-float-divide-min-latency @gol
702 -minline-float-divide-max-throughput @gol
703 -mno-inline-float-divide @gol
704 -minline-int-divide-min-latency @gol
705 -minline-int-divide-max-throughput @gol
706 -mno-inline-int-divide @gol
707 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
708 -mno-inline-sqrt @gol
709 -mdwarf2-asm -mearly-stop-bits @gol
710 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
711 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
712 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
713 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
714 -msched-spec-ldc -msched-spec-control-ldc @gol
715 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
716 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
717 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
718 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
721 @gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
722 -msign-extend-enabled -muser-enabled}
724 @emph{M32R/D Options}
725 @gccoptlist{-m32r2 -m32rx -m32r @gol
727 -malign-loops -mno-align-loops @gol
728 -missue-rate=@var{number} @gol
729 -mbranch-cost=@var{number} @gol
730 -mmodel=@var{code-size-model-type} @gol
731 -msdata=@var{sdata-type} @gol
732 -mno-flush-func -mflush-func=@var{name} @gol
733 -mno-flush-trap -mflush-trap=@var{number} @gol
737 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
739 @emph{M680x0 Options}
740 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune} @gol
741 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
742 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
743 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
744 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
745 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
746 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
747 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
751 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
752 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
753 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
754 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
755 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
758 @gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
759 -mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
760 -mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
761 -mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
764 @emph{MicroBlaze Options}
765 @gccoptlist{-msoft-float -mhard-float -msmall-divides -mcpu=@var{cpu} @gol
766 -mmemcpy -mxl-soft-mul -mxl-soft-div -mxl-barrel-shift @gol
767 -mxl-pattern-compare -mxl-stack-check -mxl-gp-opt -mno-clearbss @gol
768 -mxl-multiply-high -mxl-float-convert -mxl-float-sqrt @gol
769 -mbig-endian -mlittle-endian -mxl-reorder -mxl-mode-@var{app-model}}
772 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
773 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips32r3 -mips32r5 @gol
774 -mips32r6 -mips64 -mips64r2 -mips64r3 -mips64r5 -mips64r6 @gol
775 -mips16 -mno-mips16 -mflip-mips16 @gol
776 -minterlink-compressed -mno-interlink-compressed @gol
777 -minterlink-mips16 -mno-interlink-mips16 @gol
778 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
779 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
780 -mgp32 -mgp64 -mfp32 -mfpxx -mfp64 -mhard-float -msoft-float @gol
781 -mno-float -msingle-float -mdouble-float @gol
782 -modd-spreg -mno-odd-spreg @gol
783 -mabs=@var{mode} -mnan=@var{encoding} @gol
784 -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
787 -mvirt -mno-virt @gol
789 -mmicromips -mno-micromips @gol
790 -mfpu=@var{fpu-type} @gol
791 -msmartmips -mno-smartmips @gol
792 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
793 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
794 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
795 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
796 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
797 -membedded-data -mno-embedded-data @gol
798 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
799 -mcode-readable=@var{setting} @gol
800 -msplit-addresses -mno-split-addresses @gol
801 -mexplicit-relocs -mno-explicit-relocs @gol
802 -mcheck-zero-division -mno-check-zero-division @gol
803 -mdivide-traps -mdivide-breaks @gol
804 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
805 -mmad -mno-mad -mimadd -mno-imadd -mfused-madd -mno-fused-madd -nocpp @gol
806 -mfix-24k -mno-fix-24k @gol
807 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
808 -mfix-r10000 -mno-fix-r10000 -mfix-rm7000 -mno-fix-rm7000 @gol
809 -mfix-vr4120 -mno-fix-vr4120 @gol
810 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
811 -mflush-func=@var{func} -mno-flush-func @gol
812 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
813 -mfp-exceptions -mno-fp-exceptions @gol
814 -mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
815 -mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address}
818 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
819 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
820 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
821 -mno-base-addresses -msingle-exit -mno-single-exit}
823 @emph{MN10300 Options}
824 @gccoptlist{-mmult-bug -mno-mult-bug @gol
825 -mno-am33 -mam33 -mam33-2 -mam34 @gol
826 -mtune=@var{cpu-type} @gol
827 -mreturn-pointer-on-d0 @gol
828 -mno-crt0 -mrelax -mliw -msetlb}
831 @gccoptlist{-meb -mel -mmul.x -mno-crt0}
833 @emph{MSP430 Options}
834 @gccoptlist{-msim -masm-hex -mmcu= -mcpu= -mlarge -msmall -mrelax @gol
835 -mcode-region= -mdata-region= @gol
839 @gccoptlist{-mbig-endian -mlittle-endian @gol
840 -mreduced-regs -mfull-regs @gol
841 -mcmov -mno-cmov @gol
842 -mperf-ext -mno-perf-ext @gol
843 -mv3push -mno-v3push @gol
844 -m16bit -mno-16bit @gol
845 -misr-vector-size=@var{num} @gol
846 -mcache-block-size=@var{num} @gol
847 -march=@var{arch} @gol
848 -mcmodel=@var{code-model} @gol
851 @emph{Nios II Options}
852 @gccoptlist{-G @var{num} -mgpopt=@var{option} -mgpopt -mno-gpopt @gol
854 -mno-bypass-cache -mbypass-cache @gol
855 -mno-cache-volatile -mcache-volatile @gol
856 -mno-fast-sw-div -mfast-sw-div @gol
857 -mhw-mul -mno-hw-mul -mhw-mulx -mno-hw-mulx -mno-hw-div -mhw-div @gol
858 -mcustom-@var{insn}=@var{N} -mno-custom-@var{insn} @gol
859 -mcustom-fpu-cfg=@var{name} @gol
860 -mhal -msmallc -msys-crt0=@var{name} -msys-lib=@var{name}}
862 @emph{Nvidia PTX Options}
863 @gccoptlist{-m32 -m64 -mmainkernel}
865 @emph{PDP-11 Options}
866 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
867 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
868 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
869 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
870 -mbranch-expensive -mbranch-cheap @gol
871 -munix-asm -mdec-asm}
873 @emph{picoChip Options}
874 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N} @gol
875 -msymbol-as-address -mno-inefficient-warnings}
877 @emph{PowerPC Options}
878 See RS/6000 and PowerPC Options.
881 @gccoptlist{-msim -mmul=none -mmul=g13 -mmul=g14 -mallregs @gol
882 -mcpu=g10 -mcpu=g13 -mcpu=g14 -mg10 -mg13 -mg14 @gol
883 -m64bit-doubles -m32bit-doubles}
885 @emph{RS/6000 and PowerPC Options}
886 @gccoptlist{-mcpu=@var{cpu-type} @gol
887 -mtune=@var{cpu-type} @gol
888 -mcmodel=@var{code-model} @gol
890 -maltivec -mno-altivec @gol
891 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
892 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
893 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
894 -mfprnd -mno-fprnd @gol
895 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
896 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
897 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
898 -malign-power -malign-natural @gol
899 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
900 -msingle-float -mdouble-float -msimple-fpu @gol
901 -mstring -mno-string -mupdate -mno-update @gol
902 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
903 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
904 -mstrict-align -mno-strict-align -mrelocatable @gol
905 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
906 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
907 -mdynamic-no-pic -maltivec -mswdiv -msingle-pic-base @gol
908 -mprioritize-restricted-insns=@var{priority} @gol
909 -msched-costly-dep=@var{dependence_type} @gol
910 -minsert-sched-nops=@var{scheme} @gol
911 -mcall-sysv -mcall-netbsd @gol
912 -maix-struct-return -msvr4-struct-return @gol
913 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
914 -mblock-move-inline-limit=@var{num} @gol
915 -misel -mno-isel @gol
916 -misel=yes -misel=no @gol
918 -mspe=yes -mspe=no @gol
920 -mgen-cell-microcode -mwarn-cell-microcode @gol
921 -mvrsave -mno-vrsave @gol
922 -mmulhw -mno-mulhw @gol
923 -mdlmzb -mno-dlmzb @gol
924 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
925 -mprototype -mno-prototype @gol
926 -msim -mmvme -mads -myellowknife -memb -msdata @gol
927 -msdata=@var{opt} -mvxworks -G @var{num} -pthread @gol
928 -mrecip -mrecip=@var{opt} -mno-recip -mrecip-precision @gol
929 -mno-recip-precision @gol
930 -mveclibabi=@var{type} -mfriz -mno-friz @gol
931 -mpointers-to-nested-functions -mno-pointers-to-nested-functions @gol
932 -msave-toc-indirect -mno-save-toc-indirect @gol
933 -mpower8-fusion -mno-mpower8-fusion -mpower8-vector -mno-power8-vector @gol
934 -mcrypto -mno-crypto -mdirect-move -mno-direct-move @gol
935 -mquad-memory -mno-quad-memory @gol
936 -mquad-memory-atomic -mno-quad-memory-atomic @gol
937 -mcompat-align-parm -mno-compat-align-parm @gol
938 -mupper-regs-df -mno-upper-regs-df -mupper-regs-sf -mno-upper-regs-sf @gol
939 -mupper-regs -mno-upper-regs}
942 @gccoptlist{-m64bit-doubles -m32bit-doubles -fpu -nofpu@gol
944 -mbig-endian-data -mlittle-endian-data @gol
947 -mas100-syntax -mno-as100-syntax@gol
949 -mmax-constant-size=@gol
952 -mallow-string-insns -mno-allow-string-insns@gol
953 -mno-warn-multiple-fast-interrupts@gol
954 -msave-acc-in-interrupts}
956 @emph{S/390 and zSeries Options}
957 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
958 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
959 -mlong-double-64 -mlong-double-128 @gol
960 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
961 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
962 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
963 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
964 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard @gol
965 -mhotpatch=@var{halfwords},@var{halfwords}}
968 @gccoptlist{-meb -mel @gol
972 -mscore5 -mscore5u -mscore7 -mscore7d}
975 @gccoptlist{-m1 -m2 -m2e @gol
976 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
978 -m4-nofpu -m4-single-only -m4-single -m4 @gol
979 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
980 -m5-64media -m5-64media-nofpu @gol
981 -m5-32media -m5-32media-nofpu @gol
982 -m5-compact -m5-compact-nofpu @gol
983 -mb -ml -mdalign -mrelax @gol
984 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
985 -mieee -mno-ieee -mbitops -misize -minline-ic_invalidate -mpadstruct @gol
986 -mspace -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
987 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
988 -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
989 -maccumulate-outgoing-args -minvalid-symbols @gol
990 -matomic-model=@var{atomic-model} @gol
991 -mbranch-cost=@var{num} -mzdcbranch -mno-zdcbranch @gol
992 -mcbranch-force-delay-slot @gol
993 -mfused-madd -mno-fused-madd -mfsca -mno-fsca -mfsrra -mno-fsrra @gol
994 -mpretend-cmove -mtas}
996 @emph{Solaris 2 Options}
997 @gccoptlist{-mclear-hwcap -mno-clear-hwcap -mimpure-text -mno-impure-text @gol
1000 @emph{SPARC Options}
1001 @gccoptlist{-mcpu=@var{cpu-type} @gol
1002 -mtune=@var{cpu-type} @gol
1003 -mcmodel=@var{code-model} @gol
1004 -mmemory-model=@var{mem-model} @gol
1005 -m32 -m64 -mapp-regs -mno-app-regs @gol
1006 -mfaster-structs -mno-faster-structs -mflat -mno-flat @gol
1007 -mfpu -mno-fpu -mhard-float -msoft-float @gol
1008 -mhard-quad-float -msoft-quad-float @gol
1009 -mstack-bias -mno-stack-bias @gol
1010 -munaligned-doubles -mno-unaligned-doubles @gol
1011 -muser-mode -mno-user-mode @gol
1012 -mv8plus -mno-v8plus -mvis -mno-vis @gol
1013 -mvis2 -mno-vis2 -mvis3 -mno-vis3 @gol
1014 -mcbcond -mno-cbcond @gol
1015 -mfmaf -mno-fmaf -mpopc -mno-popc @gol
1016 -mfix-at697f -mfix-ut699}
1019 @gccoptlist{-mwarn-reloc -merror-reloc @gol
1020 -msafe-dma -munsafe-dma @gol
1022 -msmall-mem -mlarge-mem -mstdmain @gol
1023 -mfixed-range=@var{register-range} @gol
1025 -maddress-space-conversion -mno-address-space-conversion @gol
1026 -mcache-size=@var{cache-size} @gol
1027 -matomic-updates -mno-atomic-updates}
1029 @emph{System V Options}
1030 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
1032 @emph{TILE-Gx Options}
1033 @gccoptlist{-mcpu=CPU -m32 -m64 -mbig-endian -mlittle-endian @gol
1034 -mcmodel=@var{code-model}}
1036 @emph{TILEPro Options}
1037 @gccoptlist{-mcpu=@var{cpu} -m32}
1040 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
1041 -mprolog-function -mno-prolog-function -mspace @gol
1042 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
1043 -mapp-regs -mno-app-regs @gol
1044 -mdisable-callt -mno-disable-callt @gol
1045 -mv850e2v3 -mv850e2 -mv850e1 -mv850es @gol
1046 -mv850e -mv850 -mv850e3v5 @gol
1057 @gccoptlist{-mg -mgnu -munix}
1059 @emph{Visium Options}
1060 @gccoptlist{-mdebug -msim -mfpu -mno-fpu -mhard-float -msoft-float @gol
1061 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} -msv-mode -muser-mode}
1064 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64 @gol
1065 -mpointer-size=@var{size}}
1067 @emph{VxWorks Options}
1068 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
1069 -Xbind-lazy -Xbind-now}
1072 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
1073 -mtune-ctrl=@var{feature-list} -mdump-tune-features -mno-default @gol
1074 -mfpmath=@var{unit} @gol
1075 -masm=@var{dialect} -mno-fancy-math-387 @gol
1076 -mno-fp-ret-in-387 -msoft-float @gol
1077 -mno-wide-multiply -mrtd -malign-double @gol
1078 -mpreferred-stack-boundary=@var{num} @gol
1079 -mincoming-stack-boundary=@var{num} @gol
1080 -mcld -mcx16 -msahf -mmovbe -mcrc32 @gol
1081 -mrecip -mrecip=@var{opt} @gol
1082 -mvzeroupper -mprefer-avx128 @gol
1083 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
1084 -mavx2 -mavx512f -mavx512pf -mavx512er -mavx512cd -msha @gol
1085 -maes -mpclmul -mfsgsbase -mrdrnd -mf16c -mfma -mprefetchwt1 @gol
1086 -mclflushopt -mxsavec -mxsaves @gol
1087 -msse4a -m3dnow -mpopcnt -mabm -mbmi -mtbm -mfma4 -mxop -mlzcnt @gol
1088 -mbmi2 -mfxsr -mxsave -mxsaveopt -mrtm -mlwp -mmpx -mmwaitx -mthreads @gol
1089 -mno-align-stringops -minline-all-stringops @gol
1090 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
1091 -mmemcpy-strategy=@var{strategy} -mmemset-strategy=@var{strategy} @gol
1092 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
1093 -m96bit-long-double -mlong-double-64 -mlong-double-80 -mlong-double-128 @gol
1094 -mregparm=@var{num} -msseregparm @gol
1095 -mveclibabi=@var{type} -mvect8-ret-in-mem @gol
1096 -mpc32 -mpc64 -mpc80 -mstackrealign @gol
1097 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
1098 -mcmodel=@var{code-model} -mabi=@var{name} -maddress-mode=@var{mode} @gol
1099 -m32 -m64 -mx32 -m16 -miamcu -mlarge-data-threshold=@var{num} @gol
1100 -msse2avx -mfentry -mrecord-mcount -mnop-mcount -m8bit-idiv @gol
1101 -mavx256-split-unaligned-load -mavx256-split-unaligned-store @gol
1102 -malign-data=@var{type} -mstack-protector-guard=@var{guard}}
1104 @emph{x86 Windows Options}
1105 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll @gol
1106 -mnop-fun-dllimport -mthread @gol
1107 -municode -mwin32 -mwindows -fno-set-stack-executable}
1109 @emph{Xstormy16 Options}
1112 @emph{Xtensa Options}
1113 @gccoptlist{-mconst16 -mno-const16 @gol
1114 -mfused-madd -mno-fused-madd @gol
1116 -mserialize-volatile -mno-serialize-volatile @gol
1117 -mtext-section-literals -mno-text-section-literals @gol
1118 -mtarget-align -mno-target-align @gol
1119 -mlongcalls -mno-longcalls}
1121 @emph{zSeries Options}
1122 See S/390 and zSeries Options.
1124 @item Code Generation Options
1125 @xref{Code Gen Options,,Options for Code Generation Conventions}.
1126 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
1127 -ffixed-@var{reg} -fexceptions @gol
1128 -fnon-call-exceptions -fdelete-dead-exceptions -funwind-tables @gol
1129 -fasynchronous-unwind-tables @gol
1130 -fno-gnu-unique @gol
1131 -finhibit-size-directive -finstrument-functions @gol
1132 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
1133 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
1134 -fno-common -fno-ident @gol
1135 -fpcc-struct-return -fpic -fPIC -fpie -fPIE -fno-plt @gol
1136 -fno-jump-tables @gol
1137 -frecord-gcc-switches @gol
1138 -freg-struct-return -fshort-enums @gol
1139 -fshort-double -fshort-wchar @gol
1140 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
1141 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
1142 -fno-stack-limit -fsplit-stack @gol
1143 -fleading-underscore -ftls-model=@var{model} @gol
1144 -fstack-reuse=@var{reuse_level} @gol
1145 -ftrapv -fwrapv -fbounds-check @gol
1146 -fvisibility=@r{[}default@r{|}internal@r{|}hidden@r{|}protected@r{]} @gol
1147 -fstrict-volatile-bitfields -fsync-libcalls}
1151 @node Overall Options
1152 @section Options Controlling the Kind of Output
1154 Compilation can involve up to four stages: preprocessing, compilation
1155 proper, assembly and linking, always in that order. GCC is capable of
1156 preprocessing and compiling several files either into several
1157 assembler input files, or into one assembler input file; then each
1158 assembler input file produces an object file, and linking combines all
1159 the object files (those newly compiled, and those specified as input)
1160 into an executable file.
1162 @cindex file name suffix
1163 For any given input file, the file name suffix determines what kind of
1164 compilation is done:
1168 C source code that must be preprocessed.
1171 C source code that should not be preprocessed.
1174 C++ source code that should not be preprocessed.
1177 Objective-C source code. Note that you must link with the @file{libobjc}
1178 library to make an Objective-C program work.
1181 Objective-C source code that should not be preprocessed.
1185 Objective-C++ source code. Note that you must link with the @file{libobjc}
1186 library to make an Objective-C++ program work. Note that @samp{.M} refers
1187 to a literal capital M@.
1189 @item @var{file}.mii
1190 Objective-C++ source code that should not be preprocessed.
1193 C, C++, Objective-C or Objective-C++ header file to be turned into a
1194 precompiled header (default), or C, C++ header file to be turned into an
1195 Ada spec (via the @option{-fdump-ada-spec} switch).
1198 @itemx @var{file}.cp
1199 @itemx @var{file}.cxx
1200 @itemx @var{file}.cpp
1201 @itemx @var{file}.CPP
1202 @itemx @var{file}.c++
1204 C++ source code that must be preprocessed. Note that in @samp{.cxx},
1205 the last two letters must both be literally @samp{x}. Likewise,
1206 @samp{.C} refers to a literal capital C@.
1210 Objective-C++ source code that must be preprocessed.
1212 @item @var{file}.mii
1213 Objective-C++ source code that should not be preprocessed.
1217 @itemx @var{file}.hp
1218 @itemx @var{file}.hxx
1219 @itemx @var{file}.hpp
1220 @itemx @var{file}.HPP
1221 @itemx @var{file}.h++
1222 @itemx @var{file}.tcc
1223 C++ header file to be turned into a precompiled header or Ada spec.
1226 @itemx @var{file}.for
1227 @itemx @var{file}.ftn
1228 Fixed form Fortran source code that should not be preprocessed.
1231 @itemx @var{file}.FOR
1232 @itemx @var{file}.fpp
1233 @itemx @var{file}.FPP
1234 @itemx @var{file}.FTN
1235 Fixed form Fortran source code that must be preprocessed (with the traditional
1238 @item @var{file}.f90
1239 @itemx @var{file}.f95
1240 @itemx @var{file}.f03
1241 @itemx @var{file}.f08
1242 Free form Fortran source code that should not be preprocessed.
1244 @item @var{file}.F90
1245 @itemx @var{file}.F95
1246 @itemx @var{file}.F03
1247 @itemx @var{file}.F08
1248 Free form Fortran source code that must be preprocessed (with the
1249 traditional preprocessor).
1254 @c FIXME: Descriptions of Java file types.
1260 @item @var{file}.ads
1261 Ada source code file that contains a library unit declaration (a
1262 declaration of a package, subprogram, or generic, or a generic
1263 instantiation), or a library unit renaming declaration (a package,
1264 generic, or subprogram renaming declaration). Such files are also
1267 @item @var{file}.adb
1268 Ada source code file containing a library unit body (a subprogram or
1269 package body). Such files are also called @dfn{bodies}.
1271 @c GCC also knows about some suffixes for languages not yet included:
1282 @itemx @var{file}.sx
1283 Assembler code that must be preprocessed.
1286 An object file to be fed straight into linking.
1287 Any file name with no recognized suffix is treated this way.
1291 You can specify the input language explicitly with the @option{-x} option:
1294 @item -x @var{language}
1295 Specify explicitly the @var{language} for the following input files
1296 (rather than letting the compiler choose a default based on the file
1297 name suffix). This option applies to all following input files until
1298 the next @option{-x} option. Possible values for @var{language} are:
1300 c c-header cpp-output
1301 c++ c++-header c++-cpp-output
1302 objective-c objective-c-header objective-c-cpp-output
1303 objective-c++ objective-c++-header objective-c++-cpp-output
1304 assembler assembler-with-cpp
1306 f77 f77-cpp-input f95 f95-cpp-input
1312 Turn off any specification of a language, so that subsequent files are
1313 handled according to their file name suffixes (as they are if @option{-x}
1314 has not been used at all).
1316 @item -pass-exit-codes
1317 @opindex pass-exit-codes
1318 Normally the @command{gcc} program exits with the code of 1 if any
1319 phase of the compiler returns a non-success return code. If you specify
1320 @option{-pass-exit-codes}, the @command{gcc} program instead returns with
1321 the numerically highest error produced by any phase returning an error
1322 indication. The C, C++, and Fortran front ends return 4 if an internal
1323 compiler error is encountered.
1326 If you only want some of the stages of compilation, you can use
1327 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1328 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1329 @command{gcc} is to stop. Note that some combinations (for example,
1330 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1335 Compile or assemble the source files, but do not link. The linking
1336 stage simply is not done. The ultimate output is in the form of an
1337 object file for each source file.
1339 By default, the object file name for a source file is made by replacing
1340 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1342 Unrecognized input files, not requiring compilation or assembly, are
1347 Stop after the stage of compilation proper; do not assemble. The output
1348 is in the form of an assembler code file for each non-assembler input
1351 By default, the assembler file name for a source file is made by
1352 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1354 Input files that don't require compilation are ignored.
1358 Stop after the preprocessing stage; do not run the compiler proper. The
1359 output is in the form of preprocessed source code, which is sent to the
1362 Input files that don't require preprocessing are ignored.
1364 @cindex output file option
1367 Place output in file @var{file}. This applies to whatever
1368 sort of output is being produced, whether it be an executable file,
1369 an object file, an assembler file or preprocessed C code.
1371 If @option{-o} is not specified, the default is to put an executable
1372 file in @file{a.out}, the object file for
1373 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1374 assembler file in @file{@var{source}.s}, a precompiled header file in
1375 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1380 Print (on standard error output) the commands executed to run the stages
1381 of compilation. Also print the version number of the compiler driver
1382 program and of the preprocessor and the compiler proper.
1386 Like @option{-v} except the commands are not executed and arguments
1387 are quoted unless they contain only alphanumeric characters or @code{./-_}.
1388 This is useful for shell scripts to capture the driver-generated command lines.
1392 Use pipes rather than temporary files for communication between the
1393 various stages of compilation. This fails to work on some systems where
1394 the assembler is unable to read from a pipe; but the GNU assembler has
1399 Print (on the standard output) a description of the command-line options
1400 understood by @command{gcc}. If the @option{-v} option is also specified
1401 then @option{--help} is also passed on to the various processes
1402 invoked by @command{gcc}, so that they can display the command-line options
1403 they accept. If the @option{-Wextra} option has also been specified
1404 (prior to the @option{--help} option), then command-line options that
1405 have no documentation associated with them are also displayed.
1408 @opindex target-help
1409 Print (on the standard output) a description of target-specific command-line
1410 options for each tool. For some targets extra target-specific
1411 information may also be printed.
1413 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1414 Print (on the standard output) a description of the command-line
1415 options understood by the compiler that fit into all specified classes
1416 and qualifiers. These are the supported classes:
1419 @item @samp{optimizers}
1420 Display all of the optimization options supported by the
1423 @item @samp{warnings}
1424 Display all of the options controlling warning messages
1425 produced by the compiler.
1428 Display target-specific options. Unlike the
1429 @option{--target-help} option however, target-specific options of the
1430 linker and assembler are not displayed. This is because those
1431 tools do not currently support the extended @option{--help=} syntax.
1434 Display the values recognized by the @option{--param}
1437 @item @var{language}
1438 Display the options supported for @var{language}, where
1439 @var{language} is the name of one of the languages supported in this
1443 Display the options that are common to all languages.
1446 These are the supported qualifiers:
1449 @item @samp{undocumented}
1450 Display only those options that are undocumented.
1453 Display options taking an argument that appears after an equal
1454 sign in the same continuous piece of text, such as:
1455 @samp{--help=target}.
1457 @item @samp{separate}
1458 Display options taking an argument that appears as a separate word
1459 following the original option, such as: @samp{-o output-file}.
1462 Thus for example to display all the undocumented target-specific
1463 switches supported by the compiler, use:
1466 --help=target,undocumented
1469 The sense of a qualifier can be inverted by prefixing it with the
1470 @samp{^} character, so for example to display all binary warning
1471 options (i.e., ones that are either on or off and that do not take an
1472 argument) that have a description, use:
1475 --help=warnings,^joined,^undocumented
1478 The argument to @option{--help=} should not consist solely of inverted
1481 Combining several classes is possible, although this usually
1482 restricts the output so much that there is nothing to display. One
1483 case where it does work, however, is when one of the classes is
1484 @var{target}. For example, to display all the target-specific
1485 optimization options, use:
1488 --help=target,optimizers
1491 The @option{--help=} option can be repeated on the command line. Each
1492 successive use displays its requested class of options, skipping
1493 those that have already been displayed.
1495 If the @option{-Q} option appears on the command line before the
1496 @option{--help=} option, then the descriptive text displayed by
1497 @option{--help=} is changed. Instead of describing the displayed
1498 options, an indication is given as to whether the option is enabled,
1499 disabled or set to a specific value (assuming that the compiler
1500 knows this at the point where the @option{--help=} option is used).
1502 Here is a truncated example from the ARM port of @command{gcc}:
1505 % gcc -Q -mabi=2 --help=target -c
1506 The following options are target specific:
1508 -mabort-on-noreturn [disabled]
1512 The output is sensitive to the effects of previous command-line
1513 options, so for example it is possible to find out which optimizations
1514 are enabled at @option{-O2} by using:
1517 -Q -O2 --help=optimizers
1520 Alternatively you can discover which binary optimizations are enabled
1521 by @option{-O3} by using:
1524 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1525 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1526 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1529 @item -no-canonical-prefixes
1530 @opindex no-canonical-prefixes
1531 Do not expand any symbolic links, resolve references to @samp{/../}
1532 or @samp{/./}, or make the path absolute when generating a relative
1537 Display the version number and copyrights of the invoked GCC@.
1541 Invoke all subcommands under a wrapper program. The name of the
1542 wrapper program and its parameters are passed as a comma separated
1546 gcc -c t.c -wrapper gdb,--args
1550 This invokes all subprograms of @command{gcc} under
1551 @samp{gdb --args}, thus the invocation of @command{cc1} is
1552 @samp{gdb --args cc1 @dots{}}.
1554 @item -fplugin=@var{name}.so
1556 Load the plugin code in file @var{name}.so, assumed to be a
1557 shared object to be dlopen'd by the compiler. The base name of
1558 the shared object file is used to identify the plugin for the
1559 purposes of argument parsing (See
1560 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1561 Each plugin should define the callback functions specified in the
1564 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1565 @opindex fplugin-arg
1566 Define an argument called @var{key} with a value of @var{value}
1567 for the plugin called @var{name}.
1569 @item -fdump-ada-spec@r{[}-slim@r{]}
1570 @opindex fdump-ada-spec
1571 For C and C++ source and include files, generate corresponding Ada specs.
1572 @xref{Generating Ada Bindings for C and C++ headers,,, gnat_ugn,
1573 GNAT User's Guide}, which provides detailed documentation on this feature.
1575 @item -fada-spec-parent=@var{unit}
1576 @opindex fada-spec-parent
1577 In conjunction with @option{-fdump-ada-spec@r{[}-slim@r{]}} above, generate
1578 Ada specs as child units of parent @var{unit}.
1580 @item -fdump-go-spec=@var{file}
1581 @opindex fdump-go-spec
1582 For input files in any language, generate corresponding Go
1583 declarations in @var{file}. This generates Go @code{const},
1584 @code{type}, @code{var}, and @code{func} declarations which may be a
1585 useful way to start writing a Go interface to code written in some
1588 @include @value{srcdir}/../libiberty/at-file.texi
1592 @section Compiling C++ Programs
1594 @cindex suffixes for C++ source
1595 @cindex C++ source file suffixes
1596 C++ source files conventionally use one of the suffixes @samp{.C},
1597 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1598 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1599 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1600 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1601 files with these names and compiles them as C++ programs even if you
1602 call the compiler the same way as for compiling C programs (usually
1603 with the name @command{gcc}).
1607 However, the use of @command{gcc} does not add the C++ library.
1608 @command{g++} is a program that calls GCC and automatically specifies linking
1609 against the C++ library. It treats @samp{.c},
1610 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1611 files unless @option{-x} is used. This program is also useful when
1612 precompiling a C header file with a @samp{.h} extension for use in C++
1613 compilations. On many systems, @command{g++} is also installed with
1614 the name @command{c++}.
1616 @cindex invoking @command{g++}
1617 When you compile C++ programs, you may specify many of the same
1618 command-line options that you use for compiling programs in any
1619 language; or command-line options meaningful for C and related
1620 languages; or options that are meaningful only for C++ programs.
1621 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1622 explanations of options for languages related to C@.
1623 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1624 explanations of options that are meaningful only for C++ programs.
1626 @node C Dialect Options
1627 @section Options Controlling C Dialect
1628 @cindex dialect options
1629 @cindex language dialect options
1630 @cindex options, dialect
1632 The following options control the dialect of C (or languages derived
1633 from C, such as C++, Objective-C and Objective-C++) that the compiler
1637 @cindex ANSI support
1641 In C mode, this is equivalent to @option{-std=c90}. In C++ mode, it is
1642 equivalent to @option{-std=c++98}.
1644 This turns off certain features of GCC that are incompatible with ISO
1645 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1646 such as the @code{asm} and @code{typeof} keywords, and
1647 predefined macros such as @code{unix} and @code{vax} that identify the
1648 type of system you are using. It also enables the undesirable and
1649 rarely used ISO trigraph feature. For the C compiler,
1650 it disables recognition of C++ style @samp{//} comments as well as
1651 the @code{inline} keyword.
1653 The alternate keywords @code{__asm__}, @code{__extension__},
1654 @code{__inline__} and @code{__typeof__} continue to work despite
1655 @option{-ansi}. You would not want to use them in an ISO C program, of
1656 course, but it is useful to put them in header files that might be included
1657 in compilations done with @option{-ansi}. Alternate predefined macros
1658 such as @code{__unix__} and @code{__vax__} are also available, with or
1659 without @option{-ansi}.
1661 The @option{-ansi} option does not cause non-ISO programs to be
1662 rejected gratuitously. For that, @option{-Wpedantic} is required in
1663 addition to @option{-ansi}. @xref{Warning Options}.
1665 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1666 option is used. Some header files may notice this macro and refrain
1667 from declaring certain functions or defining certain macros that the
1668 ISO standard doesn't call for; this is to avoid interfering with any
1669 programs that might use these names for other things.
1671 Functions that are normally built in but do not have semantics
1672 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1673 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1674 built-in functions provided by GCC}, for details of the functions
1679 Determine the language standard. @xref{Standards,,Language Standards
1680 Supported by GCC}, for details of these standard versions. This option
1681 is currently only supported when compiling C or C++.
1683 The compiler can accept several base standards, such as @samp{c90} or
1684 @samp{c++98}, and GNU dialects of those standards, such as
1685 @samp{gnu90} or @samp{gnu++98}. When a base standard is specified, the
1686 compiler accepts all programs following that standard plus those
1687 using GNU extensions that do not contradict it. For example,
1688 @option{-std=c90} turns off certain features of GCC that are
1689 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1690 keywords, but not other GNU extensions that do not have a meaning in
1691 ISO C90, such as omitting the middle term of a @code{?:}
1692 expression. On the other hand, when a GNU dialect of a standard is
1693 specified, all features supported by the compiler are enabled, even when
1694 those features change the meaning of the base standard. As a result, some
1695 strict-conforming programs may be rejected. The particular standard
1696 is used by @option{-Wpedantic} to identify which features are GNU
1697 extensions given that version of the standard. For example
1698 @option{-std=gnu90 -Wpedantic} warns about C++ style @samp{//}
1699 comments, while @option{-std=gnu99 -Wpedantic} does not.
1701 A value for this option must be provided; possible values are
1707 Support all ISO C90 programs (certain GNU extensions that conflict
1708 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1710 @item iso9899:199409
1711 ISO C90 as modified in amendment 1.
1717 ISO C99. This standard is substantially completely supported, modulo
1718 bugs and floating-point issues
1719 (mainly but not entirely relating to optional C99 features from
1720 Annexes F and G). See
1721 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1722 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1727 ISO C11, the 2011 revision of the ISO C standard. This standard is
1728 substantially completely supported, modulo bugs, floating-point issues
1729 (mainly but not entirely relating to optional C11 features from
1730 Annexes F and G) and the optional Annexes K (Bounds-checking
1731 interfaces) and L (Analyzability). The name @samp{c1x} is deprecated.
1735 GNU dialect of ISO C90 (including some C99 features).
1739 GNU dialect of ISO C99. The name @samp{gnu9x} is deprecated.
1743 GNU dialect of ISO C11. This is the default for C code.
1744 The name @samp{gnu1x} is deprecated.
1748 The 1998 ISO C++ standard plus the 2003 technical corrigendum and some
1749 additional defect reports. Same as @option{-ansi} for C++ code.
1753 GNU dialect of @option{-std=c++98}. This is the default for
1758 The 2011 ISO C++ standard plus amendments.
1759 The name @samp{c++0x} is deprecated.
1763 GNU dialect of @option{-std=c++11}.
1764 The name @samp{gnu++0x} is deprecated.
1768 The 2014 ISO C++ standard plus amendments.
1769 The name @samp{c++1y} is deprecated.
1773 GNU dialect of @option{-std=c++14}.
1774 The name @samp{gnu++1y} is deprecated.
1777 The next revision of the ISO C++ standard, tentatively planned for
1778 2017. Support is highly experimental, and will almost certainly
1779 change in incompatible ways in future releases.
1782 GNU dialect of @option{-std=c++1z}. Support is highly experimental,
1783 and will almost certainly change in incompatible ways in future
1787 @item -fgnu89-inline
1788 @opindex fgnu89-inline
1789 The option @option{-fgnu89-inline} tells GCC to use the traditional
1790 GNU semantics for @code{inline} functions when in C99 mode.
1791 @xref{Inline,,An Inline Function is As Fast As a Macro}.
1792 Using this option is roughly equivalent to adding the
1793 @code{gnu_inline} function attribute to all inline functions
1794 (@pxref{Function Attributes}).
1796 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1797 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1798 specifies the default behavior).
1799 This option is not supported in @option{-std=c90} or
1800 @option{-std=gnu90} mode.
1802 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1803 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1804 in effect for @code{inline} functions. @xref{Common Predefined
1805 Macros,,,cpp,The C Preprocessor}.
1807 @item -aux-info @var{filename}
1809 Output to the given filename prototyped declarations for all functions
1810 declared and/or defined in a translation unit, including those in header
1811 files. This option is silently ignored in any language other than C@.
1813 Besides declarations, the file indicates, in comments, the origin of
1814 each declaration (source file and line), whether the declaration was
1815 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1816 @samp{O} for old, respectively, in the first character after the line
1817 number and the colon), and whether it came from a declaration or a
1818 definition (@samp{C} or @samp{F}, respectively, in the following
1819 character). In the case of function definitions, a K&R-style list of
1820 arguments followed by their declarations is also provided, inside
1821 comments, after the declaration.
1823 @item -fallow-parameterless-variadic-functions
1824 @opindex fallow-parameterless-variadic-functions
1825 Accept variadic functions without named parameters.
1827 Although it is possible to define such a function, this is not very
1828 useful as it is not possible to read the arguments. This is only
1829 supported for C as this construct is allowed by C++.
1833 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1834 keyword, so that code can use these words as identifiers. You can use
1835 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1836 instead. @option{-ansi} implies @option{-fno-asm}.
1838 In C++, this switch only affects the @code{typeof} keyword, since
1839 @code{asm} and @code{inline} are standard keywords. You may want to
1840 use the @option{-fno-gnu-keywords} flag instead, which has the same
1841 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1842 switch only affects the @code{asm} and @code{typeof} keywords, since
1843 @code{inline} is a standard keyword in ISO C99.
1846 @itemx -fno-builtin-@var{function}
1847 @opindex fno-builtin
1848 @cindex built-in functions
1849 Don't recognize built-in functions that do not begin with
1850 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1851 functions provided by GCC}, for details of the functions affected,
1852 including those which are not built-in functions when @option{-ansi} or
1853 @option{-std} options for strict ISO C conformance are used because they
1854 do not have an ISO standard meaning.
1856 GCC normally generates special code to handle certain built-in functions
1857 more efficiently; for instance, calls to @code{alloca} may become single
1858 instructions which adjust the stack directly, and calls to @code{memcpy}
1859 may become inline copy loops. The resulting code is often both smaller
1860 and faster, but since the function calls no longer appear as such, you
1861 cannot set a breakpoint on those calls, nor can you change the behavior
1862 of the functions by linking with a different library. In addition,
1863 when a function is recognized as a built-in function, GCC may use
1864 information about that function to warn about problems with calls to
1865 that function, or to generate more efficient code, even if the
1866 resulting code still contains calls to that function. For example,
1867 warnings are given with @option{-Wformat} for bad calls to
1868 @code{printf} when @code{printf} is built in and @code{strlen} is
1869 known not to modify global memory.
1871 With the @option{-fno-builtin-@var{function}} option
1872 only the built-in function @var{function} is
1873 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1874 function is named that is not built-in in this version of GCC, this
1875 option is ignored. There is no corresponding
1876 @option{-fbuiltin-@var{function}} option; if you wish to enable
1877 built-in functions selectively when using @option{-fno-builtin} or
1878 @option{-ffreestanding}, you may define macros such as:
1881 #define abs(n) __builtin_abs ((n))
1882 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1887 @cindex hosted environment
1889 Assert that compilation targets a hosted environment. This implies
1890 @option{-fbuiltin}. A hosted environment is one in which the
1891 entire standard library is available, and in which @code{main} has a return
1892 type of @code{int}. Examples are nearly everything except a kernel.
1893 This is equivalent to @option{-fno-freestanding}.
1895 @item -ffreestanding
1896 @opindex ffreestanding
1897 @cindex hosted environment
1899 Assert that compilation targets a freestanding environment. This
1900 implies @option{-fno-builtin}. A freestanding environment
1901 is one in which the standard library may not exist, and program startup may
1902 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1903 This is equivalent to @option{-fno-hosted}.
1905 @xref{Standards,,Language Standards Supported by GCC}, for details of
1906 freestanding and hosted environments.
1910 @cindex OpenACC accelerator programming
1911 Enable handling of OpenACC directives @code{#pragma acc} in C/C++ and
1912 @code{!$acc} in Fortran. When @option{-fopenacc} is specified, the
1913 compiler generates accelerated code according to the OpenACC Application
1914 Programming Interface v2.0 @w{@uref{http://www.openacc.org/}}. This option
1915 implies @option{-pthread}, and thus is only supported on targets that
1916 have support for @option{-pthread}.
1918 Note that this is an experimental feature, incomplete, and subject to
1919 change in future versions of GCC. See
1920 @w{@uref{https://gcc.gnu.org/wiki/OpenACC}} for more information.
1924 @cindex OpenMP parallel
1925 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1926 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1927 compiler generates parallel code according to the OpenMP Application
1928 Program Interface v4.0 @w{@uref{http://www.openmp.org/}}. This option
1929 implies @option{-pthread}, and thus is only supported on targets that
1930 have support for @option{-pthread}. @option{-fopenmp} implies
1931 @option{-fopenmp-simd}.
1934 @opindex fopenmp-simd
1937 Enable handling of OpenMP's SIMD directives with @code{#pragma omp}
1938 in C/C++ and @code{!$omp} in Fortran. Other OpenMP directives
1943 @cindex Enable Cilk Plus
1944 Enable the usage of Cilk Plus language extension features for C/C++.
1945 When the option @option{-fcilkplus} is specified, enable the usage of
1946 the Cilk Plus Language extension features for C/C++. The present
1947 implementation follows ABI version 1.2. This is an experimental
1948 feature that is only partially complete, and whose interface may
1949 change in future versions of GCC as the official specification
1950 changes. Currently, all features but @code{_Cilk_for} have been
1955 When the option @option{-fgnu-tm} is specified, the compiler
1956 generates code for the Linux variant of Intel's current Transactional
1957 Memory ABI specification document (Revision 1.1, May 6 2009). This is
1958 an experimental feature whose interface may change in future versions
1959 of GCC, as the official specification changes. Please note that not
1960 all architectures are supported for this feature.
1962 For more information on GCC's support for transactional memory,
1963 @xref{Enabling libitm,,The GNU Transactional Memory Library,libitm,GNU
1964 Transactional Memory Library}.
1966 Note that the transactional memory feature is not supported with
1967 non-call exceptions (@option{-fnon-call-exceptions}).
1969 @item -fms-extensions
1970 @opindex fms-extensions
1971 Accept some non-standard constructs used in Microsoft header files.
1973 In C++ code, this allows member names in structures to be similar
1974 to previous types declarations.
1983 Some cases of unnamed fields in structures and unions are only
1984 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1985 fields within structs/unions}, for details.
1987 Note that this option is off for all targets but x86
1988 targets using ms-abi.
1990 @item -fplan9-extensions
1991 @opindex fplan9-extensions
1992 Accept some non-standard constructs used in Plan 9 code.
1994 This enables @option{-fms-extensions}, permits passing pointers to
1995 structures with anonymous fields to functions that expect pointers to
1996 elements of the type of the field, and permits referring to anonymous
1997 fields declared using a typedef. @xref{Unnamed Fields,,Unnamed
1998 struct/union fields within structs/unions}, for details. This is only
1999 supported for C, not C++.
2003 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
2004 options for strict ISO C conformance) implies @option{-trigraphs}.
2006 @cindex traditional C language
2007 @cindex C language, traditional
2009 @itemx -traditional-cpp
2010 @opindex traditional-cpp
2011 @opindex traditional
2012 Formerly, these options caused GCC to attempt to emulate a pre-standard
2013 C compiler. They are now only supported with the @option{-E} switch.
2014 The preprocessor continues to support a pre-standard mode. See the GNU
2015 CPP manual for details.
2017 @item -fcond-mismatch
2018 @opindex fcond-mismatch
2019 Allow conditional expressions with mismatched types in the second and
2020 third arguments. The value of such an expression is void. This option
2021 is not supported for C++.
2023 @item -flax-vector-conversions
2024 @opindex flax-vector-conversions
2025 Allow implicit conversions between vectors with differing numbers of
2026 elements and/or incompatible element types. This option should not be
2029 @item -funsigned-char
2030 @opindex funsigned-char
2031 Let the type @code{char} be unsigned, like @code{unsigned char}.
2033 Each kind of machine has a default for what @code{char} should
2034 be. It is either like @code{unsigned char} by default or like
2035 @code{signed char} by default.
2037 Ideally, a portable program should always use @code{signed char} or
2038 @code{unsigned char} when it depends on the signedness of an object.
2039 But many programs have been written to use plain @code{char} and
2040 expect it to be signed, or expect it to be unsigned, depending on the
2041 machines they were written for. This option, and its inverse, let you
2042 make such a program work with the opposite default.
2044 The type @code{char} is always a distinct type from each of
2045 @code{signed char} or @code{unsigned char}, even though its behavior
2046 is always just like one of those two.
2049 @opindex fsigned-char
2050 Let the type @code{char} be signed, like @code{signed char}.
2052 Note that this is equivalent to @option{-fno-unsigned-char}, which is
2053 the negative form of @option{-funsigned-char}. Likewise, the option
2054 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
2056 @item -fsigned-bitfields
2057 @itemx -funsigned-bitfields
2058 @itemx -fno-signed-bitfields
2059 @itemx -fno-unsigned-bitfields
2060 @opindex fsigned-bitfields
2061 @opindex funsigned-bitfields
2062 @opindex fno-signed-bitfields
2063 @opindex fno-unsigned-bitfields
2064 These options control whether a bit-field is signed or unsigned, when the
2065 declaration does not use either @code{signed} or @code{unsigned}. By
2066 default, such a bit-field is signed, because this is consistent: the
2067 basic integer types such as @code{int} are signed types.
2070 @node C++ Dialect Options
2071 @section Options Controlling C++ Dialect
2073 @cindex compiler options, C++
2074 @cindex C++ options, command-line
2075 @cindex options, C++
2076 This section describes the command-line options that are only meaningful
2077 for C++ programs. You can also use most of the GNU compiler options
2078 regardless of what language your program is in. For example, you
2079 might compile a file @file{firstClass.C} like this:
2082 g++ -g -frepo -O -c firstClass.C
2086 In this example, only @option{-frepo} is an option meant
2087 only for C++ programs; you can use the other options with any
2088 language supported by GCC@.
2090 Here is a list of options that are @emph{only} for compiling C++ programs:
2094 @item -fabi-version=@var{n}
2095 @opindex fabi-version
2096 Use version @var{n} of the C++ ABI@. The default is version 0.
2098 Version 0 refers to the version conforming most closely to
2099 the C++ ABI specification. Therefore, the ABI obtained using version 0
2100 will change in different versions of G++ as ABI bugs are fixed.
2102 Version 1 is the version of the C++ ABI that first appeared in G++ 3.2.
2104 Version 2 is the version of the C++ ABI that first appeared in G++
2105 3.4, and was the default through G++ 4.9.
2107 Version 3 corrects an error in mangling a constant address as a
2110 Version 4, which first appeared in G++ 4.5, implements a standard
2111 mangling for vector types.
2113 Version 5, which first appeared in G++ 4.6, corrects the mangling of
2114 attribute const/volatile on function pointer types, decltype of a
2115 plain decl, and use of a function parameter in the declaration of
2118 Version 6, which first appeared in G++ 4.7, corrects the promotion
2119 behavior of C++11 scoped enums and the mangling of template argument
2120 packs, const/static_cast, prefix ++ and --, and a class scope function
2121 used as a template argument.
2123 Version 7, which first appeared in G++ 4.8, that treats nullptr_t as a
2124 builtin type and corrects the mangling of lambdas in default argument
2127 Version 8, which first appeared in G++ 4.9, corrects the substitution
2128 behavior of function types with function-cv-qualifiers.
2130 See also @option{-Wabi}.
2132 @item -fabi-compat-version=@var{n}
2133 @opindex fabi-compat-version
2134 On targets that support strong aliases, G++
2135 works around mangling changes by creating an alias with the correct
2136 mangled name when defining a symbol with an incorrect mangled name.
2137 This switch specifies which ABI version to use for the alias.
2139 With @option{-fabi-version=0} (the default), this defaults to 2. If
2140 another ABI version is explicitly selected, this defaults to 0.
2142 The compatibility version is also set by @option{-Wabi=@var{n}}.
2144 @item -fno-access-control
2145 @opindex fno-access-control
2146 Turn off all access checking. This switch is mainly useful for working
2147 around bugs in the access control code.
2151 Check that the pointer returned by @code{operator new} is non-null
2152 before attempting to modify the storage allocated. This check is
2153 normally unnecessary because the C++ standard specifies that
2154 @code{operator new} only returns @code{0} if it is declared
2155 @code{throw()}, in which case the compiler always checks the
2156 return value even without this option. In all other cases, when
2157 @code{operator new} has a non-empty exception specification, memory
2158 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
2159 @samp{new (nothrow)}.
2161 @item -fconstexpr-depth=@var{n}
2162 @opindex fconstexpr-depth
2163 Set the maximum nested evaluation depth for C++11 constexpr functions
2164 to @var{n}. A limit is needed to detect endless recursion during
2165 constant expression evaluation. The minimum specified by the standard
2168 @item -fdeduce-init-list
2169 @opindex fdeduce-init-list
2170 Enable deduction of a template type parameter as
2171 @code{std::initializer_list} from a brace-enclosed initializer list, i.e.@:
2174 template <class T> auto forward(T t) -> decltype (realfn (t))
2181 forward(@{1,2@}); // call forward<std::initializer_list<int>>
2185 This deduction was implemented as a possible extension to the
2186 originally proposed semantics for the C++11 standard, but was not part
2187 of the final standard, so it is disabled by default. This option is
2188 deprecated, and may be removed in a future version of G++.
2190 @item -ffriend-injection
2191 @opindex ffriend-injection
2192 Inject friend functions into the enclosing namespace, so that they are
2193 visible outside the scope of the class in which they are declared.
2194 Friend functions were documented to work this way in the old Annotated
2195 C++ Reference Manual.
2196 However, in ISO C++ a friend function that is not declared
2197 in an enclosing scope can only be found using argument dependent
2198 lookup. GCC defaults to the standard behavior.
2200 This option is for compatibility, and may be removed in a future
2203 @item -fno-elide-constructors
2204 @opindex fno-elide-constructors
2205 The C++ standard allows an implementation to omit creating a temporary
2206 that is only used to initialize another object of the same type.
2207 Specifying this option disables that optimization, and forces G++ to
2208 call the copy constructor in all cases.
2210 @item -fno-enforce-eh-specs
2211 @opindex fno-enforce-eh-specs
2212 Don't generate code to check for violation of exception specifications
2213 at run time. This option violates the C++ standard, but may be useful
2214 for reducing code size in production builds, much like defining
2215 @code{NDEBUG}. This does not give user code permission to throw
2216 exceptions in violation of the exception specifications; the compiler
2217 still optimizes based on the specifications, so throwing an
2218 unexpected exception results in undefined behavior at run time.
2220 @item -fextern-tls-init
2221 @itemx -fno-extern-tls-init
2222 @opindex fextern-tls-init
2223 @opindex fno-extern-tls-init
2224 The C++11 and OpenMP standards allow @code{thread_local} and
2225 @code{threadprivate} variables to have dynamic (runtime)
2226 initialization. To support this, any use of such a variable goes
2227 through a wrapper function that performs any necessary initialization.
2228 When the use and definition of the variable are in the same
2229 translation unit, this overhead can be optimized away, but when the
2230 use is in a different translation unit there is significant overhead
2231 even if the variable doesn't actually need dynamic initialization. If
2232 the programmer can be sure that no use of the variable in a
2233 non-defining TU needs to trigger dynamic initialization (either
2234 because the variable is statically initialized, or a use of the
2235 variable in the defining TU will be executed before any uses in
2236 another TU), they can avoid this overhead with the
2237 @option{-fno-extern-tls-init} option.
2239 On targets that support symbol aliases, the default is
2240 @option{-fextern-tls-init}. On targets that do not support symbol
2241 aliases, the default is @option{-fno-extern-tls-init}.
2244 @itemx -fno-for-scope
2246 @opindex fno-for-scope
2247 If @option{-ffor-scope} is specified, the scope of variables declared in
2248 a @i{for-init-statement} is limited to the @code{for} loop itself,
2249 as specified by the C++ standard.
2250 If @option{-fno-for-scope} is specified, the scope of variables declared in
2251 a @i{for-init-statement} extends to the end of the enclosing scope,
2252 as was the case in old versions of G++, and other (traditional)
2253 implementations of C++.
2255 If neither flag is given, the default is to follow the standard,
2256 but to allow and give a warning for old-style code that would
2257 otherwise be invalid, or have different behavior.
2259 @item -fno-gnu-keywords
2260 @opindex fno-gnu-keywords
2261 Do not recognize @code{typeof} as a keyword, so that code can use this
2262 word as an identifier. You can use the keyword @code{__typeof__} instead.
2263 @option{-ansi} implies @option{-fno-gnu-keywords}.
2265 @item -fno-implicit-templates
2266 @opindex fno-implicit-templates
2267 Never emit code for non-inline templates that are instantiated
2268 implicitly (i.e.@: by use); only emit code for explicit instantiations.
2269 @xref{Template Instantiation}, for more information.
2271 @item -fno-implicit-inline-templates
2272 @opindex fno-implicit-inline-templates
2273 Don't emit code for implicit instantiations of inline templates, either.
2274 The default is to handle inlines differently so that compiles with and
2275 without optimization need the same set of explicit instantiations.
2277 @item -fno-implement-inlines
2278 @opindex fno-implement-inlines
2279 To save space, do not emit out-of-line copies of inline functions
2280 controlled by @code{#pragma implementation}. This causes linker
2281 errors if these functions are not inlined everywhere they are called.
2283 @item -fms-extensions
2284 @opindex fms-extensions
2285 Disable Wpedantic warnings about constructs used in MFC, such as implicit
2286 int and getting a pointer to member function via non-standard syntax.
2288 @item -fno-nonansi-builtins
2289 @opindex fno-nonansi-builtins
2290 Disable built-in declarations of functions that are not mandated by
2291 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
2292 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
2295 @opindex fnothrow-opt
2296 Treat a @code{throw()} exception specification as if it were a
2297 @code{noexcept} specification to reduce or eliminate the text size
2298 overhead relative to a function with no exception specification. If
2299 the function has local variables of types with non-trivial
2300 destructors, the exception specification actually makes the
2301 function smaller because the EH cleanups for those variables can be
2302 optimized away. The semantic effect is that an exception thrown out of
2303 a function with such an exception specification results in a call
2304 to @code{terminate} rather than @code{unexpected}.
2306 @item -fno-operator-names
2307 @opindex fno-operator-names
2308 Do not treat the operator name keywords @code{and}, @code{bitand},
2309 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
2310 synonyms as keywords.
2312 @item -fno-optional-diags
2313 @opindex fno-optional-diags
2314 Disable diagnostics that the standard says a compiler does not need to
2315 issue. Currently, the only such diagnostic issued by G++ is the one for
2316 a name having multiple meanings within a class.
2319 @opindex fpermissive
2320 Downgrade some diagnostics about nonconformant code from errors to
2321 warnings. Thus, using @option{-fpermissive} allows some
2322 nonconforming code to compile.
2324 @item -fno-pretty-templates
2325 @opindex fno-pretty-templates
2326 When an error message refers to a specialization of a function
2327 template, the compiler normally prints the signature of the
2328 template followed by the template arguments and any typedefs or
2329 typenames in the signature (e.g. @code{void f(T) [with T = int]}
2330 rather than @code{void f(int)}) so that it's clear which template is
2331 involved. When an error message refers to a specialization of a class
2332 template, the compiler omits any template arguments that match
2333 the default template arguments for that template. If either of these
2334 behaviors make it harder to understand the error message rather than
2335 easier, you can use @option{-fno-pretty-templates} to disable them.
2339 Enable automatic template instantiation at link time. This option also
2340 implies @option{-fno-implicit-templates}. @xref{Template
2341 Instantiation}, for more information.
2345 Disable generation of information about every class with virtual
2346 functions for use by the C++ run-time type identification features
2347 (@code{dynamic_cast} and @code{typeid}). If you don't use those parts
2348 of the language, you can save some space by using this flag. Note that
2349 exception handling uses the same information, but G++ generates it as
2350 needed. The @code{dynamic_cast} operator can still be used for casts that
2351 do not require run-time type information, i.e.@: casts to @code{void *} or to
2352 unambiguous base classes.
2354 @item -fsized-deallocation
2355 @opindex fsized-deallocation
2356 Enable the built-in global declarations
2358 void operator delete (void *, std::size_t) noexcept;
2359 void operator delete[] (void *, std::size_t) noexcept;
2361 as introduced in C++14. This is useful for user-defined replacement
2362 deallocation functions that, for example, use the size of the object
2363 to make deallocation faster. Enabled by default under
2364 @option{-std=c++14} and above. The flag @option{-Wsized-deallocation}
2365 warns about places that might want to add a definition.
2369 Emit statistics about front-end processing at the end of the compilation.
2370 This information is generally only useful to the G++ development team.
2372 @item -fstrict-enums
2373 @opindex fstrict-enums
2374 Allow the compiler to optimize using the assumption that a value of
2375 enumerated type can only be one of the values of the enumeration (as
2376 defined in the C++ standard; basically, a value that can be
2377 represented in the minimum number of bits needed to represent all the
2378 enumerators). This assumption may not be valid if the program uses a
2379 cast to convert an arbitrary integer value to the enumerated type.
2381 @item -ftemplate-backtrace-limit=@var{n}
2382 @opindex ftemplate-backtrace-limit
2383 Set the maximum number of template instantiation notes for a single
2384 warning or error to @var{n}. The default value is 10.
2386 @item -ftemplate-depth=@var{n}
2387 @opindex ftemplate-depth
2388 Set the maximum instantiation depth for template classes to @var{n}.
2389 A limit on the template instantiation depth is needed to detect
2390 endless recursions during template class instantiation. ANSI/ISO C++
2391 conforming programs must not rely on a maximum depth greater than 17
2392 (changed to 1024 in C++11). The default value is 900, as the compiler
2393 can run out of stack space before hitting 1024 in some situations.
2395 @item -fno-threadsafe-statics
2396 @opindex fno-threadsafe-statics
2397 Do not emit the extra code to use the routines specified in the C++
2398 ABI for thread-safe initialization of local statics. You can use this
2399 option to reduce code size slightly in code that doesn't need to be
2402 @item -fuse-cxa-atexit
2403 @opindex fuse-cxa-atexit
2404 Register destructors for objects with static storage duration with the
2405 @code{__cxa_atexit} function rather than the @code{atexit} function.
2406 This option is required for fully standards-compliant handling of static
2407 destructors, but only works if your C library supports
2408 @code{__cxa_atexit}.
2410 @item -fno-use-cxa-get-exception-ptr
2411 @opindex fno-use-cxa-get-exception-ptr
2412 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2413 causes @code{std::uncaught_exception} to be incorrect, but is necessary
2414 if the runtime routine is not available.
2416 @item -fvisibility-inlines-hidden
2417 @opindex fvisibility-inlines-hidden
2418 This switch declares that the user does not attempt to compare
2419 pointers to inline functions or methods where the addresses of the two functions
2420 are taken in different shared objects.
2422 The effect of this is that GCC may, effectively, mark inline methods with
2423 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2424 appear in the export table of a DSO and do not require a PLT indirection
2425 when used within the DSO@. Enabling this option can have a dramatic effect
2426 on load and link times of a DSO as it massively reduces the size of the
2427 dynamic export table when the library makes heavy use of templates.
2429 The behavior of this switch is not quite the same as marking the
2430 methods as hidden directly, because it does not affect static variables
2431 local to the function or cause the compiler to deduce that
2432 the function is defined in only one shared object.
2434 You may mark a method as having a visibility explicitly to negate the
2435 effect of the switch for that method. For example, if you do want to
2436 compare pointers to a particular inline method, you might mark it as
2437 having default visibility. Marking the enclosing class with explicit
2438 visibility has no effect.
2440 Explicitly instantiated inline methods are unaffected by this option
2441 as their linkage might otherwise cross a shared library boundary.
2442 @xref{Template Instantiation}.
2444 @item -fvisibility-ms-compat
2445 @opindex fvisibility-ms-compat
2446 This flag attempts to use visibility settings to make GCC's C++
2447 linkage model compatible with that of Microsoft Visual Studio.
2449 The flag makes these changes to GCC's linkage model:
2453 It sets the default visibility to @code{hidden}, like
2454 @option{-fvisibility=hidden}.
2457 Types, but not their members, are not hidden by default.
2460 The One Definition Rule is relaxed for types without explicit
2461 visibility specifications that are defined in more than one
2462 shared object: those declarations are permitted if they are
2463 permitted when this option is not used.
2466 In new code it is better to use @option{-fvisibility=hidden} and
2467 export those classes that are intended to be externally visible.
2468 Unfortunately it is possible for code to rely, perhaps accidentally,
2469 on the Visual Studio behavior.
2471 Among the consequences of these changes are that static data members
2472 of the same type with the same name but defined in different shared
2473 objects are different, so changing one does not change the other;
2474 and that pointers to function members defined in different shared
2475 objects may not compare equal. When this flag is given, it is a
2476 violation of the ODR to define types with the same name differently.
2478 @item -fvtable-verify=@r{[}std@r{|}preinit@r{|}none@r{]}
2479 @opindex fvtable-verify
2480 Turn on (or off, if using @option{-fvtable-verify=none}) the security
2481 feature that verifies at run time, for every virtual call, that
2482 the vtable pointer through which the call is made is valid for the type of
2483 the object, and has not been corrupted or overwritten. If an invalid vtable
2484 pointer is detected at run time, an error is reported and execution of the
2485 program is immediately halted.
2487 This option causes run-time data structures to be built at program startup,
2488 which are used for verifying the vtable pointers.
2489 The options @samp{std} and @samp{preinit}
2490 control the timing of when these data structures are built. In both cases the
2491 data structures are built before execution reaches @code{main}. Using
2492 @option{-fvtable-verify=std} causes the data structures to be built after
2493 shared libraries have been loaded and initialized.
2494 @option{-fvtable-verify=preinit} causes them to be built before shared
2495 libraries have been loaded and initialized.
2497 If this option appears multiple times in the command line with different
2498 values specified, @samp{none} takes highest priority over both @samp{std} and
2499 @samp{preinit}; @samp{preinit} takes priority over @samp{std}.
2503 When used in conjunction with @option{-fvtable-verify=std} or
2504 @option{-fvtable-verify=preinit}, causes debug versions of the
2505 runtime functions for the vtable verification feature to be called.
2506 This flag also causes the compiler to log information about which
2507 vtable pointers it finds for each class.
2508 This information is written to a file named @file{vtv_set_ptr_data.log}
2509 in the directory named by the environment variable @env{VTV_LOGS_DIR}
2510 if that is defined or the current working directory otherwise.
2512 Note: This feature @emph{appends} data to the log file. If you want a fresh log
2513 file, be sure to delete any existing one.
2516 @opindex fvtv-counts
2517 This is a debugging flag. When used in conjunction with
2518 @option{-fvtable-verify=std} or @option{-fvtable-verify=preinit}, this
2519 causes the compiler to keep track of the total number of virtual calls
2520 it encounters and the number of verifications it inserts. It also
2521 counts the number of calls to certain run-time library functions
2522 that it inserts and logs this information for each compilation unit.
2523 The compiler writes this information to a file named
2524 @file{vtv_count_data.log} in the directory named by the environment
2525 variable @env{VTV_LOGS_DIR} if that is defined or the current working
2526 directory otherwise. It also counts the size of the vtable pointer sets
2527 for each class, and writes this information to @file{vtv_class_set_sizes.log}
2528 in the same directory.
2530 Note: This feature @emph{appends} data to the log files. To get fresh log
2531 files, be sure to delete any existing ones.
2535 Do not use weak symbol support, even if it is provided by the linker.
2536 By default, G++ uses weak symbols if they are available. This
2537 option exists only for testing, and should not be used by end-users;
2538 it results in inferior code and has no benefits. This option may
2539 be removed in a future release of G++.
2543 Do not search for header files in the standard directories specific to
2544 C++, but do still search the other standard directories. (This option
2545 is used when building the C++ library.)
2548 In addition, these optimization, warning, and code generation options
2549 have meanings only for C++ programs:
2552 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2555 When an explicit @option{-fabi-version=@var{n}} option is used, causes
2556 G++ to warn when it generates code that is probably not compatible with the
2557 vendor-neutral C++ ABI@. Since G++ now defaults to
2558 @option{-fabi-version=0}, @option{-Wabi} has no effect unless either
2559 an older ABI version is selected (with @option{-fabi-version=@var{n}})
2560 or an older compatibility version is selected (with
2561 @option{-Wabi=@var{n}} or @option{-fabi-compat-version=@var{n}}).
2563 Although an effort has been made to warn about
2564 all such cases, there are probably some cases that are not warned about,
2565 even though G++ is generating incompatible code. There may also be
2566 cases where warnings are emitted even though the code that is generated
2569 You should rewrite your code to avoid these warnings if you are
2570 concerned about the fact that code generated by G++ may not be binary
2571 compatible with code generated by other compilers.
2573 @option{-Wabi} can also be used with an explicit version number to
2574 warn about compatibility with a particular @option{-fabi-version}
2575 level, e.g. @option{-Wabi=2} to warn about changes relative to
2576 @option{-fabi-version=2}. Specifying a version number also sets
2577 @option{-fabi-compat-version=@var{n}}.
2579 The known incompatibilities in @option{-fabi-version=2} (which was the
2580 default from GCC 3.4 to 4.9) include:
2585 A template with a non-type template parameter of reference type was
2586 mangled incorrectly:
2589 template <int &> struct S @{@};
2593 This was fixed in @option{-fabi-version=3}.
2596 SIMD vector types declared using @code{__attribute ((vector_size))} were
2597 mangled in a non-standard way that does not allow for overloading of
2598 functions taking vectors of different sizes.
2600 The mangling was changed in @option{-fabi-version=4}.
2603 @code{__attribute ((const))} and @code{noreturn} were mangled as type
2604 qualifiers, and @code{decltype} of a plain declaration was folded away.
2606 These mangling issues were fixed in @option{-fabi-version=5}.
2609 Scoped enumerators passed as arguments to a variadic function are
2610 promoted like unscoped enumerators, causing @code{va_arg} to complain.
2611 On most targets this does not actually affect the parameter passing
2612 ABI, as there is no way to pass an argument smaller than @code{int}.
2614 Also, the ABI changed the mangling of template argument packs,
2615 @code{const_cast}, @code{static_cast}, prefix increment/decrement, and
2616 a class scope function used as a template argument.
2618 These issues were corrected in @option{-fabi-version=6}.
2621 Lambdas in default argument scope were mangled incorrectly, and the
2622 ABI changed the mangling of @code{nullptr_t}.
2624 These issues were corrected in @option{-fabi-version=7}.
2627 When mangling a function type with function-cv-qualifiers, the
2628 un-qualified function type was incorrectly treated as a substitution
2631 This was fixed in @option{-fabi-version=8}.
2634 It also warns about psABI-related changes. The known psABI changes at this
2640 For SysV/x86-64, unions with @code{long double} members are
2641 passed in memory as specified in psABI. For example:
2651 @code{union U} is always passed in memory.
2655 @item -Wabi-tag @r{(C++ and Objective-C++ only)}
2658 Warn when a type with an ABI tag is used in a context that does not
2659 have that ABI tag. See @ref{C++ Attributes} for more information
2662 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2663 @opindex Wctor-dtor-privacy
2664 @opindex Wno-ctor-dtor-privacy
2665 Warn when a class seems unusable because all the constructors or
2666 destructors in that class are private, and it has neither friends nor
2667 public static member functions. Also warn if there are no non-private
2668 methods, and there's at least one private member function that isn't
2669 a constructor or destructor.
2671 @item -Wdelete-non-virtual-dtor @r{(C++ and Objective-C++ only)}
2672 @opindex Wdelete-non-virtual-dtor
2673 @opindex Wno-delete-non-virtual-dtor
2674 Warn when @code{delete} is used to destroy an instance of a class that
2675 has virtual functions and non-virtual destructor. It is unsafe to delete
2676 an instance of a derived class through a pointer to a base class if the
2677 base class does not have a virtual destructor. This warning is enabled
2680 @item -Wliteral-suffix @r{(C++ and Objective-C++ only)}
2681 @opindex Wliteral-suffix
2682 @opindex Wno-literal-suffix
2683 Warn when a string or character literal is followed by a ud-suffix which does
2684 not begin with an underscore. As a conforming extension, GCC treats such
2685 suffixes as separate preprocessing tokens in order to maintain backwards
2686 compatibility with code that uses formatting macros from @code{<inttypes.h>}.
2690 #define __STDC_FORMAT_MACROS
2691 #include <inttypes.h>
2696 printf("My int64: %" PRId64"\n", i64);
2700 In this case, @code{PRId64} is treated as a separate preprocessing token.
2702 This warning is enabled by default.
2704 @item -Wlto-type-mismatch
2705 @opindex Wlto-type-mismatch
2706 @opindex Wno-lto-type-mistmach
2708 During the link-time optimization warn about type mismatches in between
2709 global declarations from different compilation units.
2710 Requires @option{-flto} to be enabled. Enabled by default.
2712 @item -Wnarrowing @r{(C++ and Objective-C++ only)}
2714 @opindex Wno-narrowing
2715 Warn when a narrowing conversion prohibited by C++11 occurs within
2719 int i = @{ 2.2 @}; // error: narrowing from double to int
2722 This flag is included in @option{-Wall} and @option{-Wc++11-compat}.
2724 With @option{-std=c++11}, @option{-Wno-narrowing} suppresses the diagnostic
2725 required by the standard. Note that this does not affect the meaning
2726 of well-formed code; narrowing conversions are still considered
2727 ill-formed in SFINAE context.
2729 @item -Wnoexcept @r{(C++ and Objective-C++ only)}
2731 @opindex Wno-noexcept
2732 Warn when a noexcept-expression evaluates to false because of a call
2733 to a function that does not have a non-throwing exception
2734 specification (i.e. @code{throw()} or @code{noexcept}) but is known by
2735 the compiler to never throw an exception.
2737 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2738 @opindex Wnon-virtual-dtor
2739 @opindex Wno-non-virtual-dtor
2740 Warn when a class has virtual functions and an accessible non-virtual
2741 destructor itself or in an accessible polymorphic base class, in which
2742 case it is possible but unsafe to delete an instance of a derived
2743 class through a pointer to the class itself or base class. This
2744 warning is automatically enabled if @option{-Weffc++} is specified.
2746 @item -Wreorder @r{(C++ and Objective-C++ only)}
2748 @opindex Wno-reorder
2749 @cindex reordering, warning
2750 @cindex warning for reordering of member initializers
2751 Warn when the order of member initializers given in the code does not
2752 match the order in which they must be executed. For instance:
2758 A(): j (0), i (1) @{ @}
2763 The compiler rearranges the member initializers for @code{i}
2764 and @code{j} to match the declaration order of the members, emitting
2765 a warning to that effect. This warning is enabled by @option{-Wall}.
2767 @item -fext-numeric-literals @r{(C++ and Objective-C++ only)}
2768 @opindex fext-numeric-literals
2769 @opindex fno-ext-numeric-literals
2770 Accept imaginary, fixed-point, or machine-defined
2771 literal number suffixes as GNU extensions.
2772 When this option is turned off these suffixes are treated
2773 as C++11 user-defined literal numeric suffixes.
2774 This is on by default for all pre-C++11 dialects and all GNU dialects:
2775 @option{-std=c++98}, @option{-std=gnu++98}, @option{-std=gnu++11},
2776 @option{-std=gnu++14}.
2777 This option is off by default
2778 for ISO C++11 onwards (@option{-std=c++11}, ...).
2781 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2784 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2787 Warn about violations of the following style guidelines from Scott Meyers'
2788 @cite{Effective C++} series of books:
2792 Define a copy constructor and an assignment operator for classes
2793 with dynamically-allocated memory.
2796 Prefer initialization to assignment in constructors.
2799 Have @code{operator=} return a reference to @code{*this}.
2802 Don't try to return a reference when you must return an object.
2805 Distinguish between prefix and postfix forms of increment and
2806 decrement operators.
2809 Never overload @code{&&}, @code{||}, or @code{,}.
2813 This option also enables @option{-Wnon-virtual-dtor}, which is also
2814 one of the effective C++ recommendations. However, the check is
2815 extended to warn about the lack of virtual destructor in accessible
2816 non-polymorphic bases classes too.
2818 When selecting this option, be aware that the standard library
2819 headers do not obey all of these guidelines; use @samp{grep -v}
2820 to filter out those warnings.
2822 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2823 @opindex Wstrict-null-sentinel
2824 @opindex Wno-strict-null-sentinel
2825 Warn about the use of an uncasted @code{NULL} as sentinel. When
2826 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2827 to @code{__null}. Although it is a null pointer constant rather than a
2828 null pointer, it is guaranteed to be of the same size as a pointer.
2829 But this use is not portable across different compilers.
2831 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2832 @opindex Wno-non-template-friend
2833 @opindex Wnon-template-friend
2834 Disable warnings when non-templatized friend functions are declared
2835 within a template. Since the advent of explicit template specification
2836 support in G++, if the name of the friend is an unqualified-id (i.e.,
2837 @samp{friend foo(int)}), the C++ language specification demands that the
2838 friend declare or define an ordinary, nontemplate function. (Section
2839 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2840 could be interpreted as a particular specialization of a templatized
2841 function. Because this non-conforming behavior is no longer the default
2842 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2843 check existing code for potential trouble spots and is on by default.
2844 This new compiler behavior can be turned off with
2845 @option{-Wno-non-template-friend}, which keeps the conformant compiler code
2846 but disables the helpful warning.
2848 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2849 @opindex Wold-style-cast
2850 @opindex Wno-old-style-cast
2851 Warn if an old-style (C-style) cast to a non-void type is used within
2852 a C++ program. The new-style casts (@code{dynamic_cast},
2853 @code{static_cast}, @code{reinterpret_cast}, and @code{const_cast}) are
2854 less vulnerable to unintended effects and much easier to search for.
2856 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2857 @opindex Woverloaded-virtual
2858 @opindex Wno-overloaded-virtual
2859 @cindex overloaded virtual function, warning
2860 @cindex warning for overloaded virtual function
2861 Warn when a function declaration hides virtual functions from a
2862 base class. For example, in:
2869 struct B: public A @{
2874 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2885 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2886 @opindex Wno-pmf-conversions
2887 @opindex Wpmf-conversions
2888 Disable the diagnostic for converting a bound pointer to member function
2891 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2892 @opindex Wsign-promo
2893 @opindex Wno-sign-promo
2894 Warn when overload resolution chooses a promotion from unsigned or
2895 enumerated type to a signed type, over a conversion to an unsigned type of
2896 the same size. Previous versions of G++ tried to preserve
2897 unsignedness, but the standard mandates the current behavior.
2899 @item -Wno-terminate @r{(C++ and Objective-C++ only)}
2901 @opindex Wno-terminate
2902 Disable the warning about a throw-expression that will immediately
2903 result in a call to @code{terminate}.
2906 @node Objective-C and Objective-C++ Dialect Options
2907 @section Options Controlling Objective-C and Objective-C++ Dialects
2909 @cindex compiler options, Objective-C and Objective-C++
2910 @cindex Objective-C and Objective-C++ options, command-line
2911 @cindex options, Objective-C and Objective-C++
2912 (NOTE: This manual does not describe the Objective-C and Objective-C++
2913 languages themselves. @xref{Standards,,Language Standards
2914 Supported by GCC}, for references.)
2916 This section describes the command-line options that are only meaningful
2917 for Objective-C and Objective-C++ programs. You can also use most of
2918 the language-independent GNU compiler options.
2919 For example, you might compile a file @file{some_class.m} like this:
2922 gcc -g -fgnu-runtime -O -c some_class.m
2926 In this example, @option{-fgnu-runtime} is an option meant only for
2927 Objective-C and Objective-C++ programs; you can use the other options with
2928 any language supported by GCC@.
2930 Note that since Objective-C is an extension of the C language, Objective-C
2931 compilations may also use options specific to the C front-end (e.g.,
2932 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2933 C++-specific options (e.g., @option{-Wabi}).
2935 Here is a list of options that are @emph{only} for compiling Objective-C
2936 and Objective-C++ programs:
2939 @item -fconstant-string-class=@var{class-name}
2940 @opindex fconstant-string-class
2941 Use @var{class-name} as the name of the class to instantiate for each
2942 literal string specified with the syntax @code{@@"@dots{}"}. The default
2943 class name is @code{NXConstantString} if the GNU runtime is being used, and
2944 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2945 @option{-fconstant-cfstrings} option, if also present, overrides the
2946 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2947 to be laid out as constant CoreFoundation strings.
2950 @opindex fgnu-runtime
2951 Generate object code compatible with the standard GNU Objective-C
2952 runtime. This is the default for most types of systems.
2954 @item -fnext-runtime
2955 @opindex fnext-runtime
2956 Generate output compatible with the NeXT runtime. This is the default
2957 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2958 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2961 @item -fno-nil-receivers
2962 @opindex fno-nil-receivers
2963 Assume that all Objective-C message dispatches (@code{[receiver
2964 message:arg]}) in this translation unit ensure that the receiver is
2965 not @code{nil}. This allows for more efficient entry points in the
2966 runtime to be used. This option is only available in conjunction with
2967 the NeXT runtime and ABI version 0 or 1.
2969 @item -fobjc-abi-version=@var{n}
2970 @opindex fobjc-abi-version
2971 Use version @var{n} of the Objective-C ABI for the selected runtime.
2972 This option is currently supported only for the NeXT runtime. In that
2973 case, Version 0 is the traditional (32-bit) ABI without support for
2974 properties and other Objective-C 2.0 additions. Version 1 is the
2975 traditional (32-bit) ABI with support for properties and other
2976 Objective-C 2.0 additions. Version 2 is the modern (64-bit) ABI. If
2977 nothing is specified, the default is Version 0 on 32-bit target
2978 machines, and Version 2 on 64-bit target machines.
2980 @item -fobjc-call-cxx-cdtors
2981 @opindex fobjc-call-cxx-cdtors
2982 For each Objective-C class, check if any of its instance variables is a
2983 C++ object with a non-trivial default constructor. If so, synthesize a
2984 special @code{- (id) .cxx_construct} instance method which runs
2985 non-trivial default constructors on any such instance variables, in order,
2986 and then return @code{self}. Similarly, check if any instance variable
2987 is a C++ object with a non-trivial destructor, and if so, synthesize a
2988 special @code{- (void) .cxx_destruct} method which runs
2989 all such default destructors, in reverse order.
2991 The @code{- (id) .cxx_construct} and @code{- (void) .cxx_destruct}
2992 methods thusly generated only operate on instance variables
2993 declared in the current Objective-C class, and not those inherited
2994 from superclasses. It is the responsibility of the Objective-C
2995 runtime to invoke all such methods in an object's inheritance
2996 hierarchy. The @code{- (id) .cxx_construct} methods are invoked
2997 by the runtime immediately after a new object instance is allocated;
2998 the @code{- (void) .cxx_destruct} methods are invoked immediately
2999 before the runtime deallocates an object instance.
3001 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
3002 support for invoking the @code{- (id) .cxx_construct} and
3003 @code{- (void) .cxx_destruct} methods.
3005 @item -fobjc-direct-dispatch
3006 @opindex fobjc-direct-dispatch
3007 Allow fast jumps to the message dispatcher. On Darwin this is
3008 accomplished via the comm page.
3010 @item -fobjc-exceptions
3011 @opindex fobjc-exceptions
3012 Enable syntactic support for structured exception handling in
3013 Objective-C, similar to what is offered by C++ and Java. This option
3014 is required to use the Objective-C keywords @code{@@try},
3015 @code{@@throw}, @code{@@catch}, @code{@@finally} and
3016 @code{@@synchronized}. This option is available with both the GNU
3017 runtime and the NeXT runtime (but not available in conjunction with
3018 the NeXT runtime on Mac OS X 10.2 and earlier).
3022 Enable garbage collection (GC) in Objective-C and Objective-C++
3023 programs. This option is only available with the NeXT runtime; the
3024 GNU runtime has a different garbage collection implementation that
3025 does not require special compiler flags.
3027 @item -fobjc-nilcheck
3028 @opindex fobjc-nilcheck
3029 For the NeXT runtime with version 2 of the ABI, check for a nil
3030 receiver in method invocations before doing the actual method call.
3031 This is the default and can be disabled using
3032 @option{-fno-objc-nilcheck}. Class methods and super calls are never
3033 checked for nil in this way no matter what this flag is set to.
3034 Currently this flag does nothing when the GNU runtime, or an older
3035 version of the NeXT runtime ABI, is used.
3037 @item -fobjc-std=objc1
3039 Conform to the language syntax of Objective-C 1.0, the language
3040 recognized by GCC 4.0. This only affects the Objective-C additions to
3041 the C/C++ language; it does not affect conformance to C/C++ standards,
3042 which is controlled by the separate C/C++ dialect option flags. When
3043 this option is used with the Objective-C or Objective-C++ compiler,
3044 any Objective-C syntax that is not recognized by GCC 4.0 is rejected.
3045 This is useful if you need to make sure that your Objective-C code can
3046 be compiled with older versions of GCC@.
3048 @item -freplace-objc-classes
3049 @opindex freplace-objc-classes
3050 Emit a special marker instructing @command{ld(1)} not to statically link in
3051 the resulting object file, and allow @command{dyld(1)} to load it in at
3052 run time instead. This is used in conjunction with the Fix-and-Continue
3053 debugging mode, where the object file in question may be recompiled and
3054 dynamically reloaded in the course of program execution, without the need
3055 to restart the program itself. Currently, Fix-and-Continue functionality
3056 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
3061 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
3062 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
3063 compile time) with static class references that get initialized at load time,
3064 which improves run-time performance. Specifying the @option{-fzero-link} flag
3065 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
3066 to be retained. This is useful in Zero-Link debugging mode, since it allows
3067 for individual class implementations to be modified during program execution.
3068 The GNU runtime currently always retains calls to @code{objc_get_class("@dots{}")}
3069 regardless of command-line options.
3071 @item -fno-local-ivars
3072 @opindex fno-local-ivars
3073 @opindex flocal-ivars
3074 By default instance variables in Objective-C can be accessed as if
3075 they were local variables from within the methods of the class they're
3076 declared in. This can lead to shadowing between instance variables
3077 and other variables declared either locally inside a class method or
3078 globally with the same name. Specifying the @option{-fno-local-ivars}
3079 flag disables this behavior thus avoiding variable shadowing issues.
3081 @item -fivar-visibility=@r{[}public@r{|}protected@r{|}private@r{|}package@r{]}
3082 @opindex fivar-visibility
3083 Set the default instance variable visibility to the specified option
3084 so that instance variables declared outside the scope of any access
3085 modifier directives default to the specified visibility.
3089 Dump interface declarations for all classes seen in the source file to a
3090 file named @file{@var{sourcename}.decl}.
3092 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
3093 @opindex Wassign-intercept
3094 @opindex Wno-assign-intercept
3095 Warn whenever an Objective-C assignment is being intercepted by the
3098 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
3099 @opindex Wno-protocol
3101 If a class is declared to implement a protocol, a warning is issued for
3102 every method in the protocol that is not implemented by the class. The
3103 default behavior is to issue a warning for every method not explicitly
3104 implemented in the class, even if a method implementation is inherited
3105 from the superclass. If you use the @option{-Wno-protocol} option, then
3106 methods inherited from the superclass are considered to be implemented,
3107 and no warning is issued for them.
3109 @item -Wselector @r{(Objective-C and Objective-C++ only)}
3111 @opindex Wno-selector
3112 Warn if multiple methods of different types for the same selector are
3113 found during compilation. The check is performed on the list of methods
3114 in the final stage of compilation. Additionally, a check is performed
3115 for each selector appearing in a @code{@@selector(@dots{})}
3116 expression, and a corresponding method for that selector has been found
3117 during compilation. Because these checks scan the method table only at
3118 the end of compilation, these warnings are not produced if the final
3119 stage of compilation is not reached, for example because an error is
3120 found during compilation, or because the @option{-fsyntax-only} option is
3123 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
3124 @opindex Wstrict-selector-match
3125 @opindex Wno-strict-selector-match
3126 Warn if multiple methods with differing argument and/or return types are
3127 found for a given selector when attempting to send a message using this
3128 selector to a receiver of type @code{id} or @code{Class}. When this flag
3129 is off (which is the default behavior), the compiler omits such warnings
3130 if any differences found are confined to types that share the same size
3133 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
3134 @opindex Wundeclared-selector
3135 @opindex Wno-undeclared-selector
3136 Warn if a @code{@@selector(@dots{})} expression referring to an
3137 undeclared selector is found. A selector is considered undeclared if no
3138 method with that name has been declared before the
3139 @code{@@selector(@dots{})} expression, either explicitly in an
3140 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
3141 an @code{@@implementation} section. This option always performs its
3142 checks as soon as a @code{@@selector(@dots{})} expression is found,
3143 while @option{-Wselector} only performs its checks in the final stage of
3144 compilation. This also enforces the coding style convention
3145 that methods and selectors must be declared before being used.
3147 @item -print-objc-runtime-info
3148 @opindex print-objc-runtime-info
3149 Generate C header describing the largest structure that is passed by
3154 @node Language Independent Options
3155 @section Options to Control Diagnostic Messages Formatting
3156 @cindex options to control diagnostics formatting
3157 @cindex diagnostic messages
3158 @cindex message formatting
3160 Traditionally, diagnostic messages have been formatted irrespective of
3161 the output device's aspect (e.g.@: its width, @dots{}). You can use the
3162 options described below
3163 to control the formatting algorithm for diagnostic messages,
3164 e.g.@: how many characters per line, how often source location
3165 information should be reported. Note that some language front ends may not
3166 honor these options.
3169 @item -fmessage-length=@var{n}
3170 @opindex fmessage-length
3171 Try to format error messages so that they fit on lines of about
3172 @var{n} characters. If @var{n} is zero, then no line-wrapping is
3173 done; each error message appears on a single line. This is the
3174 default for all front ends.
3176 @item -fdiagnostics-show-location=once
3177 @opindex fdiagnostics-show-location
3178 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
3179 reporter to emit source location information @emph{once}; that is, in
3180 case the message is too long to fit on a single physical line and has to
3181 be wrapped, the source location won't be emitted (as prefix) again,
3182 over and over, in subsequent continuation lines. This is the default
3185 @item -fdiagnostics-show-location=every-line
3186 Only meaningful in line-wrapping mode. Instructs the diagnostic
3187 messages reporter to emit the same source location information (as
3188 prefix) for physical lines that result from the process of breaking
3189 a message which is too long to fit on a single line.
3191 @item -fdiagnostics-color[=@var{WHEN}]
3192 @itemx -fno-diagnostics-color
3193 @opindex fdiagnostics-color
3194 @cindex highlight, color, colour
3195 @vindex GCC_COLORS @r{environment variable}
3196 Use color in diagnostics. @var{WHEN} is @samp{never}, @samp{always},
3197 or @samp{auto}. The default depends on how the compiler has been configured,
3198 it can be any of the above @var{WHEN} options or also @samp{never}
3199 if @env{GCC_COLORS} environment variable isn't present in the environment,
3200 and @samp{auto} otherwise.
3201 @samp{auto} means to use color only when the standard error is a terminal.
3202 The forms @option{-fdiagnostics-color} and @option{-fno-diagnostics-color} are
3203 aliases for @option{-fdiagnostics-color=always} and
3204 @option{-fdiagnostics-color=never}, respectively.
3206 The colors are defined by the environment variable @env{GCC_COLORS}.
3207 Its value is a colon-separated list of capabilities and Select Graphic
3208 Rendition (SGR) substrings. SGR commands are interpreted by the
3209 terminal or terminal emulator. (See the section in the documentation
3210 of your text terminal for permitted values and their meanings as
3211 character attributes.) These substring values are integers in decimal
3212 representation and can be concatenated with semicolons.
3213 Common values to concatenate include
3215 @samp{4} for underline,
3217 @samp{7} for inverse,
3218 @samp{39} for default foreground color,
3219 @samp{30} to @samp{37} for foreground colors,
3220 @samp{90} to @samp{97} for 16-color mode foreground colors,
3221 @samp{38;5;0} to @samp{38;5;255}
3222 for 88-color and 256-color modes foreground colors,
3223 @samp{49} for default background color,
3224 @samp{40} to @samp{47} for background colors,
3225 @samp{100} to @samp{107} for 16-color mode background colors,
3226 and @samp{48;5;0} to @samp{48;5;255}
3227 for 88-color and 256-color modes background colors.
3229 The default @env{GCC_COLORS} is
3231 error=01;31:warning=01;35:note=01;36:caret=01;32:locus=01:quote=01
3234 where @samp{01;31} is bold red, @samp{01;35} is bold magenta,
3235 @samp{01;36} is bold cyan, @samp{01;32} is bold green and
3236 @samp{01} is bold. Setting @env{GCC_COLORS} to the empty
3237 string disables colors.
3238 Supported capabilities are as follows.
3242 @vindex error GCC_COLORS @r{capability}
3243 SGR substring for error: markers.
3246 @vindex warning GCC_COLORS @r{capability}
3247 SGR substring for warning: markers.
3250 @vindex note GCC_COLORS @r{capability}
3251 SGR substring for note: markers.
3254 @vindex caret GCC_COLORS @r{capability}
3255 SGR substring for caret line.
3258 @vindex locus GCC_COLORS @r{capability}
3259 SGR substring for location information, @samp{file:line} or
3260 @samp{file:line:column} etc.
3263 @vindex quote GCC_COLORS @r{capability}
3264 SGR substring for information printed within quotes.
3267 @item -fno-diagnostics-show-option
3268 @opindex fno-diagnostics-show-option
3269 @opindex fdiagnostics-show-option
3270 By default, each diagnostic emitted includes text indicating the
3271 command-line option that directly controls the diagnostic (if such an
3272 option is known to the diagnostic machinery). Specifying the
3273 @option{-fno-diagnostics-show-option} flag suppresses that behavior.
3275 @item -fno-diagnostics-show-caret
3276 @opindex fno-diagnostics-show-caret
3277 @opindex fdiagnostics-show-caret
3278 By default, each diagnostic emitted includes the original source line
3279 and a caret '^' indicating the column. This option suppresses this
3280 information. The source line is truncated to @var{n} characters, if
3281 the @option{-fmessage-length=n} option is given. When the output is done
3282 to the terminal, the width is limited to the width given by the
3283 @env{COLUMNS} environment variable or, if not set, to the terminal width.
3287 @node Warning Options
3288 @section Options to Request or Suppress Warnings
3289 @cindex options to control warnings
3290 @cindex warning messages
3291 @cindex messages, warning
3292 @cindex suppressing warnings
3294 Warnings are diagnostic messages that report constructions that
3295 are not inherently erroneous but that are risky or suggest there
3296 may have been an error.
3298 The following language-independent options do not enable specific
3299 warnings but control the kinds of diagnostics produced by GCC@.
3302 @cindex syntax checking
3304 @opindex fsyntax-only
3305 Check the code for syntax errors, but don't do anything beyond that.
3307 @item -fmax-errors=@var{n}
3308 @opindex fmax-errors
3309 Limits the maximum number of error messages to @var{n}, at which point
3310 GCC bails out rather than attempting to continue processing the source
3311 code. If @var{n} is 0 (the default), there is no limit on the number
3312 of error messages produced. If @option{-Wfatal-errors} is also
3313 specified, then @option{-Wfatal-errors} takes precedence over this
3318 Inhibit all warning messages.
3323 Make all warnings into errors.
3328 Make the specified warning into an error. The specifier for a warning
3329 is appended; for example @option{-Werror=switch} turns the warnings
3330 controlled by @option{-Wswitch} into errors. This switch takes a
3331 negative form, to be used to negate @option{-Werror} for specific
3332 warnings; for example @option{-Wno-error=switch} makes
3333 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3336 The warning message for each controllable warning includes the
3337 option that controls the warning. That option can then be used with
3338 @option{-Werror=} and @option{-Wno-error=} as described above.
3339 (Printing of the option in the warning message can be disabled using the
3340 @option{-fno-diagnostics-show-option} flag.)
3342 Note that specifying @option{-Werror=}@var{foo} automatically implies
3343 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3346 @item -Wfatal-errors
3347 @opindex Wfatal-errors
3348 @opindex Wno-fatal-errors
3349 This option causes the compiler to abort compilation on the first error
3350 occurred rather than trying to keep going and printing further error
3355 You can request many specific warnings with options beginning with
3356 @samp{-W}, for example @option{-Wimplicit} to request warnings on
3357 implicit declarations. Each of these specific warning options also
3358 has a negative form beginning @samp{-Wno-} to turn off warnings; for
3359 example, @option{-Wno-implicit}. This manual lists only one of the
3360 two forms, whichever is not the default. For further
3361 language-specific options also refer to @ref{C++ Dialect Options} and
3362 @ref{Objective-C and Objective-C++ Dialect Options}.
3364 Some options, such as @option{-Wall} and @option{-Wextra}, turn on other
3365 options, such as @option{-Wunused}, which may turn on further options,
3366 such as @option{-Wunused-value}. The combined effect of positive and
3367 negative forms is that more specific options have priority over less
3368 specific ones, independently of their position in the command-line. For
3369 options of the same specificity, the last one takes effect. Options
3370 enabled or disabled via pragmas (@pxref{Diagnostic Pragmas}) take effect
3371 as if they appeared at the end of the command-line.
3373 When an unrecognized warning option is requested (e.g.,
3374 @option{-Wunknown-warning}), GCC emits a diagnostic stating
3375 that the option is not recognized. However, if the @option{-Wno-} form
3376 is used, the behavior is slightly different: no diagnostic is
3377 produced for @option{-Wno-unknown-warning} unless other diagnostics
3378 are being produced. This allows the use of new @option{-Wno-} options
3379 with old compilers, but if something goes wrong, the compiler
3380 warns that an unrecognized option is present.
3387 Issue all the warnings demanded by strict ISO C and ISO C++;
3388 reject all programs that use forbidden extensions, and some other
3389 programs that do not follow ISO C and ISO C++. For ISO C, follows the
3390 version of the ISO C standard specified by any @option{-std} option used.
3392 Valid ISO C and ISO C++ programs should compile properly with or without
3393 this option (though a rare few require @option{-ansi} or a
3394 @option{-std} option specifying the required version of ISO C)@. However,
3395 without this option, certain GNU extensions and traditional C and C++
3396 features are supported as well. With this option, they are rejected.
3398 @option{-Wpedantic} does not cause warning messages for use of the
3399 alternate keywords whose names begin and end with @samp{__}. Pedantic
3400 warnings are also disabled in the expression that follows
3401 @code{__extension__}. However, only system header files should use
3402 these escape routes; application programs should avoid them.
3403 @xref{Alternate Keywords}.
3405 Some users try to use @option{-Wpedantic} to check programs for strict ISO
3406 C conformance. They soon find that it does not do quite what they want:
3407 it finds some non-ISO practices, but not all---only those for which
3408 ISO C @emph{requires} a diagnostic, and some others for which
3409 diagnostics have been added.
3411 A feature to report any failure to conform to ISO C might be useful in
3412 some instances, but would require considerable additional work and would
3413 be quite different from @option{-Wpedantic}. We don't have plans to
3414 support such a feature in the near future.
3416 Where the standard specified with @option{-std} represents a GNU
3417 extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
3418 corresponding @dfn{base standard}, the version of ISO C on which the GNU
3419 extended dialect is based. Warnings from @option{-Wpedantic} are given
3420 where they are required by the base standard. (It does not make sense
3421 for such warnings to be given only for features not in the specified GNU
3422 C dialect, since by definition the GNU dialects of C include all
3423 features the compiler supports with the given option, and there would be
3424 nothing to warn about.)
3426 @item -pedantic-errors
3427 @opindex pedantic-errors
3428 Give an error whenever the @dfn{base standard} (see @option{-Wpedantic})
3429 requires a diagnostic, in some cases where there is undefined behavior
3430 at compile-time and in some other cases that do not prevent compilation
3431 of programs that are valid according to the standard. This is not
3432 equivalent to @option{-Werror=pedantic}, since there are errors enabled
3433 by this option and not enabled by the latter and vice versa.
3438 This enables all the warnings about constructions that some users
3439 consider questionable, and that are easy to avoid (or modify to
3440 prevent the warning), even in conjunction with macros. This also
3441 enables some language-specific warnings described in @ref{C++ Dialect
3442 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
3444 @option{-Wall} turns on the following warning flags:
3446 @gccoptlist{-Waddress @gol
3447 -Warray-bounds=1 @r{(only with} @option{-O2}@r{)} @gol
3448 -Wc++11-compat -Wc++14-compat@gol
3449 -Wchar-subscripts @gol
3450 -Wenum-compare @r{(in C/ObjC; this is on by default in C++)} @gol
3451 -Wimplicit-int @r{(C and Objective-C only)} @gol
3452 -Wimplicit-function-declaration @r{(C and Objective-C only)} @gol
3455 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
3456 -Wmaybe-uninitialized @gol
3457 -Wmissing-braces @r{(only for C/ObjC)} @gol
3464 -Wsequence-point @gol
3465 -Wsign-compare @r{(only in C++)} @gol
3466 -Wstrict-aliasing @gol
3467 -Wstrict-overflow=1 @gol
3470 -Wuninitialized @gol
3471 -Wunknown-pragmas @gol
3472 -Wunused-function @gol
3475 -Wunused-variable @gol
3476 -Wvolatile-register-var @gol
3479 Note that some warning flags are not implied by @option{-Wall}. Some of
3480 them warn about constructions that users generally do not consider
3481 questionable, but which occasionally you might wish to check for;
3482 others warn about constructions that are necessary or hard to avoid in
3483 some cases, and there is no simple way to modify the code to suppress
3484 the warning. Some of them are enabled by @option{-Wextra} but many of
3485 them must be enabled individually.
3491 This enables some extra warning flags that are not enabled by
3492 @option{-Wall}. (This option used to be called @option{-W}. The older
3493 name is still supported, but the newer name is more descriptive.)
3495 @gccoptlist{-Wclobbered @gol
3497 -Wignored-qualifiers @gol
3498 -Wmissing-field-initializers @gol
3499 -Wmissing-parameter-type @r{(C only)} @gol
3500 -Wold-style-declaration @r{(C only)} @gol
3501 -Woverride-init @gol
3504 -Wuninitialized @gol
3505 -Wshift-negative-value @gol
3506 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
3507 -Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
3510 The option @option{-Wextra} also prints warning messages for the
3516 A pointer is compared against integer zero with @code{<}, @code{<=},
3517 @code{>}, or @code{>=}.
3520 (C++ only) An enumerator and a non-enumerator both appear in a
3521 conditional expression.
3524 (C++ only) Ambiguous virtual bases.
3527 (C++ only) Subscripting an array that has been declared @code{register}.
3530 (C++ only) Taking the address of a variable that has been declared
3534 (C++ only) A base class is not initialized in a derived class's copy
3539 @item -Wchar-subscripts
3540 @opindex Wchar-subscripts
3541 @opindex Wno-char-subscripts
3542 Warn if an array subscript has type @code{char}. This is a common cause
3543 of error, as programmers often forget that this type is signed on some
3545 This warning is enabled by @option{-Wall}.
3549 @opindex Wno-comment
3550 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
3551 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
3552 This warning is enabled by @option{-Wall}.
3554 @item -Wno-coverage-mismatch
3555 @opindex Wno-coverage-mismatch
3556 Warn if feedback profiles do not match when using the
3557 @option{-fprofile-use} option.
3558 If a source file is changed between compiling with @option{-fprofile-gen} and
3559 with @option{-fprofile-use}, the files with the profile feedback can fail
3560 to match the source file and GCC cannot use the profile feedback
3561 information. By default, this warning is enabled and is treated as an
3562 error. @option{-Wno-coverage-mismatch} can be used to disable the
3563 warning or @option{-Wno-error=coverage-mismatch} can be used to
3564 disable the error. Disabling the error for this warning can result in
3565 poorly optimized code and is useful only in the
3566 case of very minor changes such as bug fixes to an existing code-base.
3567 Completely disabling the warning is not recommended.
3570 @r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
3572 Suppress warning messages emitted by @code{#warning} directives.
3574 @item -Wdouble-promotion @r{(C, C++, Objective-C and Objective-C++ only)}
3575 @opindex Wdouble-promotion
3576 @opindex Wno-double-promotion
3577 Give a warning when a value of type @code{float} is implicitly
3578 promoted to @code{double}. CPUs with a 32-bit ``single-precision''
3579 floating-point unit implement @code{float} in hardware, but emulate
3580 @code{double} in software. On such a machine, doing computations
3581 using @code{double} values is much more expensive because of the
3582 overhead required for software emulation.
3584 It is easy to accidentally do computations with @code{double} because
3585 floating-point literals are implicitly of type @code{double}. For
3589 float area(float radius)
3591 return 3.14159 * radius * radius;
3595 the compiler performs the entire computation with @code{double}
3596 because the floating-point literal is a @code{double}.
3599 @itemx -Wformat=@var{n}
3602 @opindex ffreestanding
3603 @opindex fno-builtin
3605 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
3606 the arguments supplied have types appropriate to the format string
3607 specified, and that the conversions specified in the format string make
3608 sense. This includes standard functions, and others specified by format
3609 attributes (@pxref{Function Attributes}), in the @code{printf},
3610 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
3611 not in the C standard) families (or other target-specific families).
3612 Which functions are checked without format attributes having been
3613 specified depends on the standard version selected, and such checks of
3614 functions without the attribute specified are disabled by
3615 @option{-ffreestanding} or @option{-fno-builtin}.
3617 The formats are checked against the format features supported by GNU
3618 libc version 2.2. These include all ISO C90 and C99 features, as well
3619 as features from the Single Unix Specification and some BSD and GNU
3620 extensions. Other library implementations may not support all these
3621 features; GCC does not support warning about features that go beyond a
3622 particular library's limitations. However, if @option{-Wpedantic} is used
3623 with @option{-Wformat}, warnings are given about format features not
3624 in the selected standard version (but not for @code{strfmon} formats,
3625 since those are not in any version of the C standard). @xref{C Dialect
3626 Options,,Options Controlling C Dialect}.
3633 Option @option{-Wformat} is equivalent to @option{-Wformat=1}, and
3634 @option{-Wno-format} is equivalent to @option{-Wformat=0}. Since
3635 @option{-Wformat} also checks for null format arguments for several
3636 functions, @option{-Wformat} also implies @option{-Wnonnull}. Some
3637 aspects of this level of format checking can be disabled by the
3638 options: @option{-Wno-format-contains-nul},
3639 @option{-Wno-format-extra-args}, and @option{-Wno-format-zero-length}.
3640 @option{-Wformat} is enabled by @option{-Wall}.
3642 @item -Wno-format-contains-nul
3643 @opindex Wno-format-contains-nul
3644 @opindex Wformat-contains-nul
3645 If @option{-Wformat} is specified, do not warn about format strings that
3648 @item -Wno-format-extra-args
3649 @opindex Wno-format-extra-args
3650 @opindex Wformat-extra-args
3651 If @option{-Wformat} is specified, do not warn about excess arguments to a
3652 @code{printf} or @code{scanf} format function. The C standard specifies
3653 that such arguments are ignored.
3655 Where the unused arguments lie between used arguments that are
3656 specified with @samp{$} operand number specifications, normally
3657 warnings are still given, since the implementation could not know what
3658 type to pass to @code{va_arg} to skip the unused arguments. However,
3659 in the case of @code{scanf} formats, this option suppresses the
3660 warning if the unused arguments are all pointers, since the Single
3661 Unix Specification says that such unused arguments are allowed.
3663 @item -Wno-format-zero-length
3664 @opindex Wno-format-zero-length
3665 @opindex Wformat-zero-length
3666 If @option{-Wformat} is specified, do not warn about zero-length formats.
3667 The C standard specifies that zero-length formats are allowed.
3672 Enable @option{-Wformat} plus additional format checks. Currently
3673 equivalent to @option{-Wformat -Wformat-nonliteral -Wformat-security
3676 @item -Wformat-nonliteral
3677 @opindex Wformat-nonliteral
3678 @opindex Wno-format-nonliteral
3679 If @option{-Wformat} is specified, also warn if the format string is not a
3680 string literal and so cannot be checked, unless the format function
3681 takes its format arguments as a @code{va_list}.
3683 @item -Wformat-security
3684 @opindex Wformat-security
3685 @opindex Wno-format-security
3686 If @option{-Wformat} is specified, also warn about uses of format
3687 functions that represent possible security problems. At present, this
3688 warns about calls to @code{printf} and @code{scanf} functions where the
3689 format string is not a string literal and there are no format arguments,
3690 as in @code{printf (foo);}. This may be a security hole if the format
3691 string came from untrusted input and contains @samp{%n}. (This is
3692 currently a subset of what @option{-Wformat-nonliteral} warns about, but
3693 in future warnings may be added to @option{-Wformat-security} that are not
3694 included in @option{-Wformat-nonliteral}.)
3696 @item -Wformat-signedness
3697 @opindex Wformat-signedness
3698 @opindex Wno-format-signedness
3699 If @option{-Wformat} is specified, also warn if the format string
3700 requires an unsigned argument and the argument is signed and vice versa.
3703 @opindex Wformat-y2k
3704 @opindex Wno-format-y2k
3705 If @option{-Wformat} is specified, also warn about @code{strftime}
3706 formats that may yield only a two-digit year.
3711 @opindex Wno-nonnull
3712 Warn about passing a null pointer for arguments marked as
3713 requiring a non-null value by the @code{nonnull} function attribute.
3715 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3716 can be disabled with the @option{-Wno-nonnull} option.
3718 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3720 @opindex Wno-init-self
3721 Warn about uninitialized variables that are initialized with themselves.
3722 Note this option can only be used with the @option{-Wuninitialized} option.
3724 For example, GCC warns about @code{i} being uninitialized in the
3725 following snippet only when @option{-Winit-self} has been specified:
3736 This warning is enabled by @option{-Wall} in C++.
3738 @item -Wimplicit-int @r{(C and Objective-C only)}
3739 @opindex Wimplicit-int
3740 @opindex Wno-implicit-int
3741 Warn when a declaration does not specify a type.
3742 This warning is enabled by @option{-Wall}.
3744 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3745 @opindex Wimplicit-function-declaration
3746 @opindex Wno-implicit-function-declaration
3747 Give a warning whenever a function is used before being declared. In
3748 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3749 enabled by default and it is made into an error by
3750 @option{-pedantic-errors}. This warning is also enabled by
3753 @item -Wimplicit @r{(C and Objective-C only)}
3755 @opindex Wno-implicit
3756 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3757 This warning is enabled by @option{-Wall}.
3759 @item -Wignored-qualifiers @r{(C and C++ only)}
3760 @opindex Wignored-qualifiers
3761 @opindex Wno-ignored-qualifiers
3762 Warn if the return type of a function has a type qualifier
3763 such as @code{const}. For ISO C such a type qualifier has no effect,
3764 since the value returned by a function is not an lvalue.
3765 For C++, the warning is only emitted for scalar types or @code{void}.
3766 ISO C prohibits qualified @code{void} return types on function
3767 definitions, so such return types always receive a warning
3768 even without this option.
3770 This warning is also enabled by @option{-Wextra}.
3775 Warn if the type of @code{main} is suspicious. @code{main} should be
3776 a function with external linkage, returning int, taking either zero
3777 arguments, two, or three arguments of appropriate types. This warning
3778 is enabled by default in C++ and is enabled by either @option{-Wall}
3779 or @option{-Wpedantic}.
3781 @item -Wmisleading-indentation @r{(C and C++ only)}
3782 @opindex Wmisleading-indentation
3783 @opindex Wno-misleading-indentation
3784 Warn when the indentation of the code does not reflect the block structure.
3785 Specifically, a warning is issued for @code{if}, @code{else}, @code{while}, and
3786 @code{for} clauses with a guarded statement that does not use braces,
3787 followed by an unguarded statement with the same indentation.
3789 This warning is disabled by default.
3791 In the following example, the call to ``bar'' is misleadingly indented as
3792 if it were guarded by the ``if'' conditional.
3795 if (some_condition ())
3797 bar (); /* Gotcha: this is not guarded by the "if". */
3800 In the case of mixed tabs and spaces, the warning uses the
3801 @option{-ftabstop=} option to determine if the statements line up
3804 The warning is not issued for code involving multiline preprocessor logic
3805 such as the following example.
3810 #if SOME_CONDITION_THAT_DOES_NOT_HOLD
3816 The warning is not issued after a @code{#line} directive, since this
3817 typically indicates autogenerated code, and no assumptions can be made
3818 about the layout of the file that the directive references.
3820 @item -Wmissing-braces
3821 @opindex Wmissing-braces
3822 @opindex Wno-missing-braces
3823 Warn if an aggregate or union initializer is not fully bracketed. In
3824 the following example, the initializer for @code{a} is not fully
3825 bracketed, but that for @code{b} is fully bracketed. This warning is
3826 enabled by @option{-Wall} in C.
3829 int a[2][2] = @{ 0, 1, 2, 3 @};
3830 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3833 This warning is enabled by @option{-Wall}.
3835 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3836 @opindex Wmissing-include-dirs
3837 @opindex Wno-missing-include-dirs
3838 Warn if a user-supplied include directory does not exist.
3841 @opindex Wparentheses
3842 @opindex Wno-parentheses
3843 Warn if parentheses are omitted in certain contexts, such
3844 as when there is an assignment in a context where a truth value
3845 is expected, or when operators are nested whose precedence people
3846 often get confused about.
3848 Also warn if a comparison like @code{x<=y<=z} appears; this is
3849 equivalent to @code{(x<=y ? 1 : 0) <= z}, which is a different
3850 interpretation from that of ordinary mathematical notation.
3852 Also warn about constructions where there may be confusion to which
3853 @code{if} statement an @code{else} branch belongs. Here is an example of
3868 In C/C++, every @code{else} branch belongs to the innermost possible
3869 @code{if} statement, which in this example is @code{if (b)}. This is
3870 often not what the programmer expected, as illustrated in the above
3871 example by indentation the programmer chose. When there is the
3872 potential for this confusion, GCC issues a warning when this flag
3873 is specified. To eliminate the warning, add explicit braces around
3874 the innermost @code{if} statement so there is no way the @code{else}
3875 can belong to the enclosing @code{if}. The resulting code
3892 Also warn for dangerous uses of the GNU extension to
3893 @code{?:} with omitted middle operand. When the condition
3894 in the @code{?}: operator is a boolean expression, the omitted value is
3895 always 1. Often programmers expect it to be a value computed
3896 inside the conditional expression instead.
3898 This warning is enabled by @option{-Wall}.
3900 @item -Wsequence-point
3901 @opindex Wsequence-point
3902 @opindex Wno-sequence-point
3903 Warn about code that may have undefined semantics because of violations
3904 of sequence point rules in the C and C++ standards.
3906 The C and C++ standards define the order in which expressions in a C/C++
3907 program are evaluated in terms of @dfn{sequence points}, which represent
3908 a partial ordering between the execution of parts of the program: those
3909 executed before the sequence point, and those executed after it. These
3910 occur after the evaluation of a full expression (one which is not part
3911 of a larger expression), after the evaluation of the first operand of a
3912 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3913 function is called (but after the evaluation of its arguments and the
3914 expression denoting the called function), and in certain other places.
3915 Other than as expressed by the sequence point rules, the order of
3916 evaluation of subexpressions of an expression is not specified. All
3917 these rules describe only a partial order rather than a total order,
3918 since, for example, if two functions are called within one expression
3919 with no sequence point between them, the order in which the functions
3920 are called is not specified. However, the standards committee have
3921 ruled that function calls do not overlap.
3923 It is not specified when between sequence points modifications to the
3924 values of objects take effect. Programs whose behavior depends on this
3925 have undefined behavior; the C and C++ standards specify that ``Between
3926 the previous and next sequence point an object shall have its stored
3927 value modified at most once by the evaluation of an expression.
3928 Furthermore, the prior value shall be read only to determine the value
3929 to be stored.''. If a program breaks these rules, the results on any
3930 particular implementation are entirely unpredictable.
3932 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3933 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3934 diagnosed by this option, and it may give an occasional false positive
3935 result, but in general it has been found fairly effective at detecting
3936 this sort of problem in programs.
3938 The standard is worded confusingly, therefore there is some debate
3939 over the precise meaning of the sequence point rules in subtle cases.
3940 Links to discussions of the problem, including proposed formal
3941 definitions, may be found on the GCC readings page, at
3942 @uref{http://gcc.gnu.org/@/readings.html}.
3944 This warning is enabled by @option{-Wall} for C and C++.
3946 @item -Wno-return-local-addr
3947 @opindex Wno-return-local-addr
3948 @opindex Wreturn-local-addr
3949 Do not warn about returning a pointer (or in C++, a reference) to a
3950 variable that goes out of scope after the function returns.
3953 @opindex Wreturn-type
3954 @opindex Wno-return-type
3955 Warn whenever a function is defined with a return type that defaults
3956 to @code{int}. Also warn about any @code{return} statement with no
3957 return value in a function whose return type is not @code{void}
3958 (falling off the end of the function body is considered returning
3959 without a value), and about a @code{return} statement with an
3960 expression in a function whose return type is @code{void}.
3962 For C++, a function without return type always produces a diagnostic
3963 message, even when @option{-Wno-return-type} is specified. The only
3964 exceptions are @code{main} and functions defined in system headers.
3966 This warning is enabled by @option{-Wall}.
3968 @item -Wshift-count-negative
3969 @opindex Wshift-count-negative
3970 @opindex Wno-shift-count-negative
3971 Warn if shift count is negative. This warning is enabled by default.
3973 @item -Wshift-count-overflow
3974 @opindex Wshift-count-overflow
3975 @opindex Wno-shift-count-overflow
3976 Warn if shift count >= width of type. This warning is enabled by default.
3978 @item -Wshift-negative-value
3979 @opindex Wshift-negative-value
3980 @opindex Wno-shift-negative-value
3981 Warn if left shifting a negative value. This warning is enabled by
3982 @option{-Wextra} in C99 and C++11 modes (and newer).
3987 Warn whenever a @code{switch} statement has an index of enumerated type
3988 and lacks a @code{case} for one or more of the named codes of that
3989 enumeration. (The presence of a @code{default} label prevents this
3990 warning.) @code{case} labels outside the enumeration range also
3991 provoke warnings when this option is used (even if there is a
3992 @code{default} label).
3993 This warning is enabled by @option{-Wall}.
3995 @item -Wswitch-default
3996 @opindex Wswitch-default
3997 @opindex Wno-switch-default
3998 Warn whenever a @code{switch} statement does not have a @code{default}
4002 @opindex Wswitch-enum
4003 @opindex Wno-switch-enum
4004 Warn whenever a @code{switch} statement has an index of enumerated type
4005 and lacks a @code{case} for one or more of the named codes of that
4006 enumeration. @code{case} labels outside the enumeration range also
4007 provoke warnings when this option is used. The only difference
4008 between @option{-Wswitch} and this option is that this option gives a
4009 warning about an omitted enumeration code even if there is a
4010 @code{default} label.
4013 @opindex Wswitch-bool
4014 @opindex Wno-switch-bool
4015 Warn whenever a @code{switch} statement has an index of boolean type
4016 and the case values are outside the range of a boolean type.
4017 It is possible to suppress this warning by casting the controlling
4018 expression to a type other than @code{bool}. For example:
4021 switch ((int) (a == 4))
4027 This warning is enabled by default for C and C++ programs.
4029 @item -Wsync-nand @r{(C and C++ only)}
4031 @opindex Wno-sync-nand
4032 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
4033 built-in functions are used. These functions changed semantics in GCC 4.4.
4037 @opindex Wno-trigraphs
4038 Warn if any trigraphs are encountered that might change the meaning of
4039 the program (trigraphs within comments are not warned about).
4040 This warning is enabled by @option{-Wall}.
4042 @item -Wunused-but-set-parameter
4043 @opindex Wunused-but-set-parameter
4044 @opindex Wno-unused-but-set-parameter
4045 Warn whenever a function parameter is assigned to, but otherwise unused
4046 (aside from its declaration).
4048 To suppress this warning use the @code{unused} attribute
4049 (@pxref{Variable Attributes}).
4051 This warning is also enabled by @option{-Wunused} together with
4054 @item -Wunused-but-set-variable
4055 @opindex Wunused-but-set-variable
4056 @opindex Wno-unused-but-set-variable
4057 Warn whenever a local variable is assigned to, but otherwise unused
4058 (aside from its declaration).
4059 This warning is enabled by @option{-Wall}.
4061 To suppress this warning use the @code{unused} attribute
4062 (@pxref{Variable Attributes}).
4064 This warning is also enabled by @option{-Wunused}, which is enabled
4067 @item -Wunused-function
4068 @opindex Wunused-function
4069 @opindex Wno-unused-function
4070 Warn whenever a static function is declared but not defined or a
4071 non-inline static function is unused.
4072 This warning is enabled by @option{-Wall}.
4074 @item -Wunused-label
4075 @opindex Wunused-label
4076 @opindex Wno-unused-label
4077 Warn whenever a label is declared but not used.
4078 This warning is enabled by @option{-Wall}.
4080 To suppress this warning use the @code{unused} attribute
4081 (@pxref{Variable Attributes}).
4083 @item -Wunused-local-typedefs @r{(C, Objective-C, C++ and Objective-C++ only)}
4084 @opindex Wunused-local-typedefs
4085 Warn when a typedef locally defined in a function is not used.
4086 This warning is enabled by @option{-Wall}.
4088 @item -Wunused-parameter
4089 @opindex Wunused-parameter
4090 @opindex Wno-unused-parameter
4091 Warn whenever a function parameter is unused aside from its declaration.
4093 To suppress this warning use the @code{unused} attribute
4094 (@pxref{Variable Attributes}).
4096 @item -Wno-unused-result
4097 @opindex Wunused-result
4098 @opindex Wno-unused-result
4099 Do not warn if a caller of a function marked with attribute
4100 @code{warn_unused_result} (@pxref{Function Attributes}) does not use
4101 its return value. The default is @option{-Wunused-result}.
4103 @item -Wunused-variable
4104 @opindex Wunused-variable
4105 @opindex Wno-unused-variable
4106 Warn whenever a local variable or non-constant static variable is unused
4107 aside from its declaration.
4108 This warning is enabled by @option{-Wall}.
4110 To suppress this warning use the @code{unused} attribute
4111 (@pxref{Variable Attributes}).
4113 @item -Wunused-value
4114 @opindex Wunused-value
4115 @opindex Wno-unused-value
4116 Warn whenever a statement computes a result that is explicitly not
4117 used. To suppress this warning cast the unused expression to
4118 @code{void}. This includes an expression-statement or the left-hand
4119 side of a comma expression that contains no side effects. For example,
4120 an expression such as @code{x[i,j]} causes a warning, while
4121 @code{x[(void)i,j]} does not.
4123 This warning is enabled by @option{-Wall}.
4128 All the above @option{-Wunused} options combined.
4130 In order to get a warning about an unused function parameter, you must
4131 either specify @option{-Wextra -Wunused} (note that @option{-Wall} implies
4132 @option{-Wunused}), or separately specify @option{-Wunused-parameter}.
4134 @item -Wuninitialized
4135 @opindex Wuninitialized
4136 @opindex Wno-uninitialized
4137 Warn if an automatic variable is used without first being initialized
4138 or if a variable may be clobbered by a @code{setjmp} call. In C++,
4139 warn if a non-static reference or non-static @code{const} member
4140 appears in a class without constructors.
4142 If you want to warn about code that uses the uninitialized value of the
4143 variable in its own initializer, use the @option{-Winit-self} option.
4145 These warnings occur for individual uninitialized or clobbered
4146 elements of structure, union or array variables as well as for
4147 variables that are uninitialized or clobbered as a whole. They do
4148 not occur for variables or elements declared @code{volatile}. Because
4149 these warnings depend on optimization, the exact variables or elements
4150 for which there are warnings depends on the precise optimization
4151 options and version of GCC used.
4153 Note that there may be no warning about a variable that is used only
4154 to compute a value that itself is never used, because such
4155 computations may be deleted by data flow analysis before the warnings
4158 @item -Wmaybe-uninitialized
4159 @opindex Wmaybe-uninitialized
4160 @opindex Wno-maybe-uninitialized
4161 For an automatic variable, if there exists a path from the function
4162 entry to a use of the variable that is initialized, but there exist
4163 some other paths for which the variable is not initialized, the compiler
4164 emits a warning if it cannot prove the uninitialized paths are not
4165 executed at run time. These warnings are made optional because GCC is
4166 not smart enough to see all the reasons why the code might be correct
4167 in spite of appearing to have an error. Here is one example of how
4188 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
4189 always initialized, but GCC doesn't know this. To suppress the
4190 warning, you need to provide a default case with assert(0) or
4193 @cindex @code{longjmp} warnings
4194 This option also warns when a non-volatile automatic variable might be
4195 changed by a call to @code{longjmp}. These warnings as well are possible
4196 only in optimizing compilation.
4198 The compiler sees only the calls to @code{setjmp}. It cannot know
4199 where @code{longjmp} will be called; in fact, a signal handler could
4200 call it at any point in the code. As a result, you may get a warning
4201 even when there is in fact no problem because @code{longjmp} cannot
4202 in fact be called at the place that would cause a problem.
4204 Some spurious warnings can be avoided if you declare all the functions
4205 you use that never return as @code{noreturn}. @xref{Function
4208 This warning is enabled by @option{-Wall} or @option{-Wextra}.
4210 @item -Wunknown-pragmas
4211 @opindex Wunknown-pragmas
4212 @opindex Wno-unknown-pragmas
4213 @cindex warning for unknown pragmas
4214 @cindex unknown pragmas, warning
4215 @cindex pragmas, warning of unknown
4216 Warn when a @code{#pragma} directive is encountered that is not understood by
4217 GCC@. If this command-line option is used, warnings are even issued
4218 for unknown pragmas in system header files. This is not the case if
4219 the warnings are only enabled by the @option{-Wall} command-line option.
4222 @opindex Wno-pragmas
4224 Do not warn about misuses of pragmas, such as incorrect parameters,
4225 invalid syntax, or conflicts between pragmas. See also
4226 @option{-Wunknown-pragmas}.
4228 @item -Wstrict-aliasing
4229 @opindex Wstrict-aliasing
4230 @opindex Wno-strict-aliasing
4231 This option is only active when @option{-fstrict-aliasing} is active.
4232 It warns about code that might break the strict aliasing rules that the
4233 compiler is using for optimization. The warning does not catch all
4234 cases, but does attempt to catch the more common pitfalls. It is
4235 included in @option{-Wall}.
4236 It is equivalent to @option{-Wstrict-aliasing=3}
4238 @item -Wstrict-aliasing=n
4239 @opindex Wstrict-aliasing=n
4240 This option is only active when @option{-fstrict-aliasing} is active.
4241 It warns about code that might break the strict aliasing rules that the
4242 compiler is using for optimization.
4243 Higher levels correspond to higher accuracy (fewer false positives).
4244 Higher levels also correspond to more effort, similar to the way @option{-O}
4246 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=3}.
4248 Level 1: Most aggressive, quick, least accurate.
4249 Possibly useful when higher levels
4250 do not warn but @option{-fstrict-aliasing} still breaks the code, as it has very few
4251 false negatives. However, it has many false positives.
4252 Warns for all pointer conversions between possibly incompatible types,
4253 even if never dereferenced. Runs in the front end only.
4255 Level 2: Aggressive, quick, not too precise.
4256 May still have many false positives (not as many as level 1 though),
4257 and few false negatives (but possibly more than level 1).
4258 Unlike level 1, it only warns when an address is taken. Warns about
4259 incomplete types. Runs in the front end only.
4261 Level 3 (default for @option{-Wstrict-aliasing}):
4262 Should have very few false positives and few false
4263 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
4264 Takes care of the common pun+dereference pattern in the front end:
4265 @code{*(int*)&some_float}.
4266 If optimization is enabled, it also runs in the back end, where it deals
4267 with multiple statement cases using flow-sensitive points-to information.
4268 Only warns when the converted pointer is dereferenced.
4269 Does not warn about incomplete types.
4271 @item -Wstrict-overflow
4272 @itemx -Wstrict-overflow=@var{n}
4273 @opindex Wstrict-overflow
4274 @opindex Wno-strict-overflow
4275 This option is only active when @option{-fstrict-overflow} is active.
4276 It warns about cases where the compiler optimizes based on the
4277 assumption that signed overflow does not occur. Note that it does not
4278 warn about all cases where the code might overflow: it only warns
4279 about cases where the compiler implements some optimization. Thus
4280 this warning depends on the optimization level.
4282 An optimization that assumes that signed overflow does not occur is
4283 perfectly safe if the values of the variables involved are such that
4284 overflow never does, in fact, occur. Therefore this warning can
4285 easily give a false positive: a warning about code that is not
4286 actually a problem. To help focus on important issues, several
4287 warning levels are defined. No warnings are issued for the use of
4288 undefined signed overflow when estimating how many iterations a loop
4289 requires, in particular when determining whether a loop will be
4293 @item -Wstrict-overflow=1
4294 Warn about cases that are both questionable and easy to avoid. For
4295 example, with @option{-fstrict-overflow}, the compiler simplifies
4296 @code{x + 1 > x} to @code{1}. This level of
4297 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
4298 are not, and must be explicitly requested.
4300 @item -Wstrict-overflow=2
4301 Also warn about other cases where a comparison is simplified to a
4302 constant. For example: @code{abs (x) >= 0}. This can only be
4303 simplified when @option{-fstrict-overflow} is in effect, because
4304 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
4305 zero. @option{-Wstrict-overflow} (with no level) is the same as
4306 @option{-Wstrict-overflow=2}.
4308 @item -Wstrict-overflow=3
4309 Also warn about other cases where a comparison is simplified. For
4310 example: @code{x + 1 > 1} is simplified to @code{x > 0}.
4312 @item -Wstrict-overflow=4
4313 Also warn about other simplifications not covered by the above cases.
4314 For example: @code{(x * 10) / 5} is simplified to @code{x * 2}.
4316 @item -Wstrict-overflow=5
4317 Also warn about cases where the compiler reduces the magnitude of a
4318 constant involved in a comparison. For example: @code{x + 2 > y} is
4319 simplified to @code{x + 1 >= y}. This is reported only at the
4320 highest warning level because this simplification applies to many
4321 comparisons, so this warning level gives a very large number of
4325 @item -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{]}
4326 @opindex Wsuggest-attribute=
4327 @opindex Wno-suggest-attribute=
4328 Warn for cases where adding an attribute may be beneficial. The
4329 attributes currently supported are listed below.
4332 @item -Wsuggest-attribute=pure
4333 @itemx -Wsuggest-attribute=const
4334 @itemx -Wsuggest-attribute=noreturn
4335 @opindex Wsuggest-attribute=pure
4336 @opindex Wno-suggest-attribute=pure
4337 @opindex Wsuggest-attribute=const
4338 @opindex Wno-suggest-attribute=const
4339 @opindex Wsuggest-attribute=noreturn
4340 @opindex Wno-suggest-attribute=noreturn
4342 Warn about functions that might be candidates for attributes
4343 @code{pure}, @code{const} or @code{noreturn}. The compiler only warns for
4344 functions visible in other compilation units or (in the case of @code{pure} and
4345 @code{const}) if it cannot prove that the function returns normally. A function
4346 returns normally if it doesn't contain an infinite loop or return abnormally
4347 by throwing, calling @code{abort} or trapping. This analysis requires option
4348 @option{-fipa-pure-const}, which is enabled by default at @option{-O} and
4349 higher. Higher optimization levels improve the accuracy of the analysis.
4351 @item -Wsuggest-attribute=format
4352 @itemx -Wmissing-format-attribute
4353 @opindex Wsuggest-attribute=format
4354 @opindex Wmissing-format-attribute
4355 @opindex Wno-suggest-attribute=format
4356 @opindex Wno-missing-format-attribute
4360 Warn about function pointers that might be candidates for @code{format}
4361 attributes. Note these are only possible candidates, not absolute ones.
4362 GCC guesses that function pointers with @code{format} attributes that
4363 are used in assignment, initialization, parameter passing or return
4364 statements should have a corresponding @code{format} attribute in the
4365 resulting type. I.e.@: the left-hand side of the assignment or
4366 initialization, the type of the parameter variable, or the return type
4367 of the containing function respectively should also have a @code{format}
4368 attribute to avoid the warning.
4370 GCC also warns about function definitions that might be
4371 candidates for @code{format} attributes. Again, these are only
4372 possible candidates. GCC guesses that @code{format} attributes
4373 might be appropriate for any function that calls a function like
4374 @code{vprintf} or @code{vscanf}, but this might not always be the
4375 case, and some functions for which @code{format} attributes are
4376 appropriate may not be detected.
4379 @item -Wsuggest-final-types
4380 @opindex Wno-suggest-final-types
4381 @opindex Wsuggest-final-types
4382 Warn about types with virtual methods where code quality would be improved
4383 if the type were declared with the C++11 @code{final} specifier,
4385 declared in an anonymous namespace. This allows GCC to more aggressively
4386 devirtualize the polymorphic calls. This warning is more effective with link
4387 time optimization, where the information about the class hierarchy graph is
4390 @item -Wsuggest-final-methods
4391 @opindex Wno-suggest-final-methods
4392 @opindex Wsuggest-final-methods
4393 Warn about virtual methods where code quality would be improved if the method
4394 were declared with the C++11 @code{final} specifier,
4395 or, if possible, its type were
4396 declared in an anonymous namespace or with the @code{final} specifier.
4398 more effective with link time optimization, where the information about the
4399 class hierarchy graph is more complete. It is recommended to first consider
4400 suggestions of @option{-Wsuggest-final-types} and then rebuild with new
4403 @item -Wsuggest-override
4404 Warn about overriding virtual functions that are not marked with the override
4407 @item -Warray-bounds
4408 @itemx -Warray-bounds=@var{n}
4409 @opindex Wno-array-bounds
4410 @opindex Warray-bounds
4411 This option is only active when @option{-ftree-vrp} is active
4412 (default for @option{-O2} and above). It warns about subscripts to arrays
4413 that are always out of bounds. This warning is enabled by @option{-Wall}.
4416 @item -Warray-bounds=1
4417 This is the warning level of @option{-Warray-bounds} and is enabled
4418 by @option{-Wall}; higher levels are not, and must be explicitly requested.
4420 @item -Warray-bounds=2
4421 This warning level also warns about out of bounds access for
4422 arrays at the end of a struct and for arrays accessed through
4423 pointers. This warning level may give a larger number of
4424 false positives and is deactivated by default.
4427 @item -Wbool-compare
4428 @opindex Wno-bool-compare
4429 @opindex Wbool-compare
4430 Warn about boolean expression compared with an integer value different from
4431 @code{true}/@code{false}. For instance, the following comparison is
4436 if ((n > 1) == 2) @{ @dots{} @}
4438 This warning is enabled by @option{-Wall}.
4440 @item -Wno-discarded-qualifiers @r{(C and Objective-C only)}
4441 @opindex Wno-discarded-qualifiers
4442 @opindex Wdiscarded-qualifiers
4443 Do not warn if type qualifiers on pointers are being discarded.
4444 Typically, the compiler warns if a @code{const char *} variable is
4445 passed to a function that takes a @code{char *} parameter. This option
4446 can be used to suppress such a warning.
4448 @item -Wno-discarded-array-qualifiers @r{(C and Objective-C only)}
4449 @opindex Wno-discarded-array-qualifiers
4450 @opindex Wdiscarded-array-qualifiers
4451 Do not warn if type qualifiers on arrays which are pointer targets
4452 are being discarded. Typically, the compiler warns if a
4453 @code{const int (*)[]} variable is passed to a function that
4454 takes a @code{int (*)[]} parameter. This option can be used to
4455 suppress such a warning.
4457 @item -Wno-incompatible-pointer-types @r{(C and Objective-C only)}
4458 @opindex Wno-incompatible-pointer-types
4459 @opindex Wincompatible-pointer-types
4460 Do not warn when there is a conversion between pointers that have incompatible
4461 types. This warning is for cases not covered by @option{-Wno-pointer-sign},
4462 which warns for pointer argument passing or assignment with different
4465 @item -Wno-int-conversion @r{(C and Objective-C only)}
4466 @opindex Wno-int-conversion
4467 @opindex Wint-conversion
4468 Do not warn about incompatible integer to pointer and pointer to integer
4469 conversions. This warning is about implicit conversions; for explicit
4470 conversions the warnings @option{-Wno-int-to-pointer-cast} and
4471 @option{-Wno-pointer-to-int-cast} may be used.
4473 @item -Wno-div-by-zero
4474 @opindex Wno-div-by-zero
4475 @opindex Wdiv-by-zero
4476 Do not warn about compile-time integer division by zero. Floating-point
4477 division by zero is not warned about, as it can be a legitimate way of
4478 obtaining infinities and NaNs.
4480 @item -Wsystem-headers
4481 @opindex Wsystem-headers
4482 @opindex Wno-system-headers
4483 @cindex warnings from system headers
4484 @cindex system headers, warnings from
4485 Print warning messages for constructs found in system header files.
4486 Warnings from system headers are normally suppressed, on the assumption
4487 that they usually do not indicate real problems and would only make the
4488 compiler output harder to read. Using this command-line option tells
4489 GCC to emit warnings from system headers as if they occurred in user
4490 code. However, note that using @option{-Wall} in conjunction with this
4491 option does @emph{not} warn about unknown pragmas in system
4492 headers---for that, @option{-Wunknown-pragmas} must also be used.
4495 @opindex Wtrampolines
4496 @opindex Wno-trampolines
4497 Warn about trampolines generated for pointers to nested functions.
4498 A trampoline is a small piece of data or code that is created at run
4499 time on the stack when the address of a nested function is taken, and is
4500 used to call the nested function indirectly. For some targets, it is
4501 made up of data only and thus requires no special treatment. But, for
4502 most targets, it is made up of code and thus requires the stack to be
4503 made executable in order for the program to work properly.
4506 @opindex Wfloat-equal
4507 @opindex Wno-float-equal
4508 Warn if floating-point values are used in equality comparisons.
4510 The idea behind this is that sometimes it is convenient (for the
4511 programmer) to consider floating-point values as approximations to
4512 infinitely precise real numbers. If you are doing this, then you need
4513 to compute (by analyzing the code, or in some other way) the maximum or
4514 likely maximum error that the computation introduces, and allow for it
4515 when performing comparisons (and when producing output, but that's a
4516 different problem). In particular, instead of testing for equality, you
4517 should check to see whether the two values have ranges that overlap; and
4518 this is done with the relational operators, so equality comparisons are
4521 @item -Wtraditional @r{(C and Objective-C only)}
4522 @opindex Wtraditional
4523 @opindex Wno-traditional
4524 Warn about certain constructs that behave differently in traditional and
4525 ISO C@. Also warn about ISO C constructs that have no traditional C
4526 equivalent, and/or problematic constructs that should be avoided.
4530 Macro parameters that appear within string literals in the macro body.
4531 In traditional C macro replacement takes place within string literals,
4532 but in ISO C it does not.
4535 In traditional C, some preprocessor directives did not exist.
4536 Traditional preprocessors only considered a line to be a directive
4537 if the @samp{#} appeared in column 1 on the line. Therefore
4538 @option{-Wtraditional} warns about directives that traditional C
4539 understands but ignores because the @samp{#} does not appear as the
4540 first character on the line. It also suggests you hide directives like
4541 @code{#pragma} not understood by traditional C by indenting them. Some
4542 traditional implementations do not recognize @code{#elif}, so this option
4543 suggests avoiding it altogether.
4546 A function-like macro that appears without arguments.
4549 The unary plus operator.
4552 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating-point
4553 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
4554 constants.) Note, these suffixes appear in macros defined in the system
4555 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
4556 Use of these macros in user code might normally lead to spurious
4557 warnings, however GCC's integrated preprocessor has enough context to
4558 avoid warning in these cases.
4561 A function declared external in one block and then used after the end of
4565 A @code{switch} statement has an operand of type @code{long}.
4568 A non-@code{static} function declaration follows a @code{static} one.
4569 This construct is not accepted by some traditional C compilers.
4572 The ISO type of an integer constant has a different width or
4573 signedness from its traditional type. This warning is only issued if
4574 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
4575 typically represent bit patterns, are not warned about.
4578 Usage of ISO string concatenation is detected.
4581 Initialization of automatic aggregates.
4584 Identifier conflicts with labels. Traditional C lacks a separate
4585 namespace for labels.
4588 Initialization of unions. If the initializer is zero, the warning is
4589 omitted. This is done under the assumption that the zero initializer in
4590 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
4591 initializer warnings and relies on default initialization to zero in the
4595 Conversions by prototypes between fixed/floating-point values and vice
4596 versa. The absence of these prototypes when compiling with traditional
4597 C causes serious problems. This is a subset of the possible
4598 conversion warnings; for the full set use @option{-Wtraditional-conversion}.
4601 Use of ISO C style function definitions. This warning intentionally is
4602 @emph{not} issued for prototype declarations or variadic functions
4603 because these ISO C features appear in your code when using
4604 libiberty's traditional C compatibility macros, @code{PARAMS} and
4605 @code{VPARAMS}. This warning is also bypassed for nested functions
4606 because that feature is already a GCC extension and thus not relevant to
4607 traditional C compatibility.
4610 @item -Wtraditional-conversion @r{(C and Objective-C only)}
4611 @opindex Wtraditional-conversion
4612 @opindex Wno-traditional-conversion
4613 Warn if a prototype causes a type conversion that is different from what
4614 would happen to the same argument in the absence of a prototype. This
4615 includes conversions of fixed point to floating and vice versa, and
4616 conversions changing the width or signedness of a fixed-point argument
4617 except when the same as the default promotion.
4619 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
4620 @opindex Wdeclaration-after-statement
4621 @opindex Wno-declaration-after-statement
4622 Warn when a declaration is found after a statement in a block. This
4623 construct, known from C++, was introduced with ISO C99 and is by default
4624 allowed in GCC@. It is not supported by ISO C90. @xref{Mixed Declarations}.
4629 Warn if an undefined identifier is evaluated in an @code{#if} directive.
4631 @item -Wno-endif-labels
4632 @opindex Wno-endif-labels
4633 @opindex Wendif-labels
4634 Do not warn whenever an @code{#else} or an @code{#endif} are followed by text.
4639 Warn whenever a local variable or type declaration shadows another
4640 variable, parameter, type, class member (in C++), or instance variable
4641 (in Objective-C) or whenever a built-in function is shadowed. Note
4642 that in C++, the compiler warns if a local variable shadows an
4643 explicit typedef, but not if it shadows a struct/class/enum.
4645 @item -Wno-shadow-ivar @r{(Objective-C only)}
4646 @opindex Wno-shadow-ivar
4647 @opindex Wshadow-ivar
4648 Do not warn whenever a local variable shadows an instance variable in an
4651 @item -Wlarger-than=@var{len}
4652 @opindex Wlarger-than=@var{len}
4653 @opindex Wlarger-than-@var{len}
4654 Warn whenever an object of larger than @var{len} bytes is defined.
4656 @item -Wframe-larger-than=@var{len}
4657 @opindex Wframe-larger-than
4658 Warn if the size of a function frame is larger than @var{len} bytes.
4659 The computation done to determine the stack frame size is approximate
4660 and not conservative.
4661 The actual requirements may be somewhat greater than @var{len}
4662 even if you do not get a warning. In addition, any space allocated
4663 via @code{alloca}, variable-length arrays, or related constructs
4664 is not included by the compiler when determining
4665 whether or not to issue a warning.
4667 @item -Wno-free-nonheap-object
4668 @opindex Wno-free-nonheap-object
4669 @opindex Wfree-nonheap-object
4670 Do not warn when attempting to free an object that was not allocated
4673 @item -Wstack-usage=@var{len}
4674 @opindex Wstack-usage
4675 Warn if the stack usage of a function might be larger than @var{len} bytes.
4676 The computation done to determine the stack usage is conservative.
4677 Any space allocated via @code{alloca}, variable-length arrays, or related
4678 constructs is included by the compiler when determining whether or not to
4681 The message is in keeping with the output of @option{-fstack-usage}.
4685 If the stack usage is fully static but exceeds the specified amount, it's:
4688 warning: stack usage is 1120 bytes
4691 If the stack usage is (partly) dynamic but bounded, it's:
4694 warning: stack usage might be 1648 bytes
4697 If the stack usage is (partly) dynamic and not bounded, it's:
4700 warning: stack usage might be unbounded
4704 @item -Wunsafe-loop-optimizations
4705 @opindex Wunsafe-loop-optimizations
4706 @opindex Wno-unsafe-loop-optimizations
4707 Warn if the loop cannot be optimized because the compiler cannot
4708 assume anything on the bounds of the loop indices. With
4709 @option{-funsafe-loop-optimizations} warn if the compiler makes
4712 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
4713 @opindex Wno-pedantic-ms-format
4714 @opindex Wpedantic-ms-format
4715 When used in combination with @option{-Wformat}
4716 and @option{-pedantic} without GNU extensions, this option
4717 disables the warnings about non-ISO @code{printf} / @code{scanf} format
4718 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets,
4719 which depend on the MS runtime.
4721 @item -Wpointer-arith
4722 @opindex Wpointer-arith
4723 @opindex Wno-pointer-arith
4724 Warn about anything that depends on the ``size of'' a function type or
4725 of @code{void}. GNU C assigns these types a size of 1, for
4726 convenience in calculations with @code{void *} pointers and pointers
4727 to functions. In C++, warn also when an arithmetic operation involves
4728 @code{NULL}. This warning is also enabled by @option{-Wpedantic}.
4731 @opindex Wtype-limits
4732 @opindex Wno-type-limits
4733 Warn if a comparison is always true or always false due to the limited
4734 range of the data type, but do not warn for constant expressions. For
4735 example, warn if an unsigned variable is compared against zero with
4736 @code{<} or @code{>=}. This warning is also enabled by
4739 @item -Wbad-function-cast @r{(C and Objective-C only)}
4740 @opindex Wbad-function-cast
4741 @opindex Wno-bad-function-cast
4742 Warn when a function call is cast to a non-matching type.
4743 For example, warn if a call to a function returning an integer type
4744 is cast to a pointer type.
4746 @item -Wc90-c99-compat @r{(C and Objective-C only)}
4747 @opindex Wc90-c99-compat
4748 @opindex Wno-c90-c99-compat
4749 Warn about features not present in ISO C90, but present in ISO C99.
4750 For instance, warn about use of variable length arrays, @code{long long}
4751 type, @code{bool} type, compound literals, designated initializers, and so
4752 on. This option is independent of the standards mode. Warnings are disabled
4753 in the expression that follows @code{__extension__}.
4755 @item -Wc99-c11-compat @r{(C and Objective-C only)}
4756 @opindex Wc99-c11-compat
4757 @opindex Wno-c99-c11-compat
4758 Warn about features not present in ISO C99, but present in ISO C11.
4759 For instance, warn about use of anonymous structures and unions,
4760 @code{_Atomic} type qualifier, @code{_Thread_local} storage-class specifier,
4761 @code{_Alignas} specifier, @code{Alignof} operator, @code{_Generic} keyword,
4762 and so on. This option is independent of the standards mode. Warnings are
4763 disabled in the expression that follows @code{__extension__}.
4765 @item -Wc++-compat @r{(C and Objective-C only)}
4766 @opindex Wc++-compat
4767 Warn about ISO C constructs that are outside of the common subset of
4768 ISO C and ISO C++, e.g.@: request for implicit conversion from
4769 @code{void *} to a pointer to non-@code{void} type.
4771 @item -Wc++11-compat @r{(C++ and Objective-C++ only)}
4772 @opindex Wc++11-compat
4773 Warn about C++ constructs whose meaning differs between ISO C++ 1998
4774 and ISO C++ 2011, e.g., identifiers in ISO C++ 1998 that are keywords
4775 in ISO C++ 2011. This warning turns on @option{-Wnarrowing} and is
4776 enabled by @option{-Wall}.
4778 @item -Wc++14-compat @r{(C++ and Objective-C++ only)}
4779 @opindex Wc++14-compat
4780 Warn about C++ constructs whose meaning differs between ISO C++ 2011
4781 and ISO C++ 2014. This warning is enabled by @option{-Wall}.
4785 @opindex Wno-cast-qual
4786 Warn whenever a pointer is cast so as to remove a type qualifier from
4787 the target type. For example, warn if a @code{const char *} is cast
4788 to an ordinary @code{char *}.
4790 Also warn when making a cast that introduces a type qualifier in an
4791 unsafe way. For example, casting @code{char **} to @code{const char **}
4792 is unsafe, as in this example:
4795 /* p is char ** value. */
4796 const char **q = (const char **) p;
4797 /* Assignment of readonly string to const char * is OK. */
4799 /* Now char** pointer points to read-only memory. */
4804 @opindex Wcast-align
4805 @opindex Wno-cast-align
4806 Warn whenever a pointer is cast such that the required alignment of the
4807 target is increased. For example, warn if a @code{char *} is cast to
4808 an @code{int *} on machines where integers can only be accessed at
4809 two- or four-byte boundaries.
4811 @item -Wwrite-strings
4812 @opindex Wwrite-strings
4813 @opindex Wno-write-strings
4814 When compiling C, give string constants the type @code{const
4815 char[@var{length}]} so that copying the address of one into a
4816 non-@code{const} @code{char *} pointer produces a warning. These
4817 warnings help you find at compile time code that can try to write
4818 into a string constant, but only if you have been very careful about
4819 using @code{const} in declarations and prototypes. Otherwise, it is
4820 just a nuisance. This is why we did not make @option{-Wall} request
4823 When compiling C++, warn about the deprecated conversion from string
4824 literals to @code{char *}. This warning is enabled by default for C++
4829 @opindex Wno-clobbered
4830 Warn for variables that might be changed by @code{longjmp} or
4831 @code{vfork}. This warning is also enabled by @option{-Wextra}.
4833 @item -Wconditionally-supported @r{(C++ and Objective-C++ only)}
4834 @opindex Wconditionally-supported
4835 @opindex Wno-conditionally-supported
4836 Warn for conditionally-supported (C++11 [intro.defs]) constructs.
4839 @opindex Wconversion
4840 @opindex Wno-conversion
4841 Warn for implicit conversions that may alter a value. This includes
4842 conversions between real and integer, like @code{abs (x)} when
4843 @code{x} is @code{double}; conversions between signed and unsigned,
4844 like @code{unsigned ui = -1}; and conversions to smaller types, like
4845 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
4846 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
4847 changed by the conversion like in @code{abs (2.0)}. Warnings about
4848 conversions between signed and unsigned integers can be disabled by
4849 using @option{-Wno-sign-conversion}.
4851 For C++, also warn for confusing overload resolution for user-defined
4852 conversions; and conversions that never use a type conversion
4853 operator: conversions to @code{void}, the same type, a base class or a
4854 reference to them. Warnings about conversions between signed and
4855 unsigned integers are disabled by default in C++ unless
4856 @option{-Wsign-conversion} is explicitly enabled.
4858 @item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
4859 @opindex Wconversion-null
4860 @opindex Wno-conversion-null
4861 Do not warn for conversions between @code{NULL} and non-pointer
4862 types. @option{-Wconversion-null} is enabled by default.
4864 @item -Wzero-as-null-pointer-constant @r{(C++ and Objective-C++ only)}
4865 @opindex Wzero-as-null-pointer-constant
4866 @opindex Wno-zero-as-null-pointer-constant
4867 Warn when a literal '0' is used as null pointer constant. This can
4868 be useful to facilitate the conversion to @code{nullptr} in C++11.
4872 @opindex Wno-date-time
4873 Warn when macros @code{__TIME__}, @code{__DATE__} or @code{__TIMESTAMP__}
4874 are encountered as they might prevent bit-wise-identical reproducible
4877 @item -Wdelete-incomplete @r{(C++ and Objective-C++ only)}
4878 @opindex Wdelete-incomplete
4879 @opindex Wno-delete-incomplete
4880 Warn when deleting a pointer to incomplete type, which may cause
4881 undefined behavior at runtime. This warning is enabled by default.
4883 @item -Wuseless-cast @r{(C++ and Objective-C++ only)}
4884 @opindex Wuseless-cast
4885 @opindex Wno-useless-cast
4886 Warn when an expression is casted to its own type.
4889 @opindex Wempty-body
4890 @opindex Wno-empty-body
4891 Warn if an empty body occurs in an @code{if}, @code{else} or @code{do
4892 while} statement. This warning is also enabled by @option{-Wextra}.
4894 @item -Wenum-compare
4895 @opindex Wenum-compare
4896 @opindex Wno-enum-compare
4897 Warn about a comparison between values of different enumerated types.
4898 In C++ enumeral mismatches in conditional expressions are also
4899 diagnosed and the warning is enabled by default. In C this warning is
4900 enabled by @option{-Wall}.
4902 @item -Wjump-misses-init @r{(C, Objective-C only)}
4903 @opindex Wjump-misses-init
4904 @opindex Wno-jump-misses-init
4905 Warn if a @code{goto} statement or a @code{switch} statement jumps
4906 forward across the initialization of a variable, or jumps backward to a
4907 label after the variable has been initialized. This only warns about
4908 variables that are initialized when they are declared. This warning is
4909 only supported for C and Objective-C; in C++ this sort of branch is an
4912 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
4913 can be disabled with the @option{-Wno-jump-misses-init} option.
4915 @item -Wsign-compare
4916 @opindex Wsign-compare
4917 @opindex Wno-sign-compare
4918 @cindex warning for comparison of signed and unsigned values
4919 @cindex comparison of signed and unsigned values, warning
4920 @cindex signed and unsigned values, comparison warning
4921 Warn when a comparison between signed and unsigned values could produce
4922 an incorrect result when the signed value is converted to unsigned.
4923 This warning is also enabled by @option{-Wextra}; to get the other warnings
4924 of @option{-Wextra} without this warning, use @option{-Wextra -Wno-sign-compare}.
4926 @item -Wsign-conversion
4927 @opindex Wsign-conversion
4928 @opindex Wno-sign-conversion
4929 Warn for implicit conversions that may change the sign of an integer
4930 value, like assigning a signed integer expression to an unsigned
4931 integer variable. An explicit cast silences the warning. In C, this
4932 option is enabled also by @option{-Wconversion}.
4934 @item -Wfloat-conversion
4935 @opindex Wfloat-conversion
4936 @opindex Wno-float-conversion
4937 Warn for implicit conversions that reduce the precision of a real value.
4938 This includes conversions from real to integer, and from higher precision
4939 real to lower precision real values. This option is also enabled by
4940 @option{-Wconversion}.
4942 @item -Wsized-deallocation @r{(C++ and Objective-C++ only)}
4943 @opindex Wsized-deallocation
4944 @opindex Wno-sized-deallocation
4945 Warn about a definition of an unsized deallocation function
4947 void operator delete (void *) noexcept;
4948 void operator delete[] (void *) noexcept;
4950 without a definition of the corresponding sized deallocation function
4952 void operator delete (void *, std::size_t) noexcept;
4953 void operator delete[] (void *, std::size_t) noexcept;
4955 or vice versa. Enabled by @option{-Wextra} along with
4956 @option{-fsized-deallocation}.
4958 @item -Wsizeof-pointer-memaccess
4959 @opindex Wsizeof-pointer-memaccess
4960 @opindex Wno-sizeof-pointer-memaccess
4961 Warn for suspicious length parameters to certain string and memory built-in
4962 functions if the argument uses @code{sizeof}. This warning warns e.g.@:
4963 about @code{memset (ptr, 0, sizeof (ptr));} if @code{ptr} is not an array,
4964 but a pointer, and suggests a possible fix, or about
4965 @code{memcpy (&foo, ptr, sizeof (&foo));}. This warning is enabled by
4968 @item -Wsizeof-array-argument
4969 @opindex Wsizeof-array-argument
4970 @opindex Wno-sizeof-array-argument
4971 Warn when the @code{sizeof} operator is applied to a parameter that is
4972 declared as an array in a function definition. This warning is enabled by
4973 default for C and C++ programs.
4975 @item -Wmemset-transposed-args
4976 @opindex Wmemset-transposed-args
4977 @opindex Wno-memset-transposed-args
4978 Warn for suspicious calls to the @code{memset} built-in function, if the
4979 second argument is not zero and the third argument is zero. This warns e.g.@
4980 about @code{memset (buf, sizeof buf, 0)} where most probably
4981 @code{memset (buf, 0, sizeof buf)} was meant instead. The diagnostics
4982 is only emitted if the third argument is literal zero. If it is some
4983 expression that is folded to zero, a cast of zero to some type, etc.,
4984 it is far less likely that the user has mistakenly exchanged the arguments
4985 and no warning is emitted. This warning is enabled by @option{-Wall}.
4989 @opindex Wno-address
4990 Warn about suspicious uses of memory addresses. These include using
4991 the address of a function in a conditional expression, such as
4992 @code{void func(void); if (func)}, and comparisons against the memory
4993 address of a string literal, such as @code{if (x == "abc")}. Such
4994 uses typically indicate a programmer error: the address of a function
4995 always evaluates to true, so their use in a conditional usually
4996 indicate that the programmer forgot the parentheses in a function
4997 call; and comparisons against string literals result in unspecified
4998 behavior and are not portable in C, so they usually indicate that the
4999 programmer intended to use @code{strcmp}. This warning is enabled by
5003 @opindex Wlogical-op
5004 @opindex Wno-logical-op
5005 Warn about suspicious uses of logical operators in expressions.
5006 This includes using logical operators in contexts where a
5007 bit-wise operator is likely to be expected. Also warns when
5008 the operands of a logical operator are the same:
5011 if (a < 0 && a < 0) @{ @dots{} @}
5014 @item -Wlogical-not-parentheses
5015 @opindex Wlogical-not-parentheses
5016 @opindex Wno-logical-not-parentheses
5017 Warn about logical not used on the left hand side operand of a comparison.
5018 This option does not warn if the RHS operand is of a boolean type. Its
5019 purpose is to detect suspicious code like the following:
5023 if (!a > 1) @{ @dots{} @}
5026 It is possible to suppress the warning by wrapping the LHS into
5029 if ((!a) > 1) @{ @dots{} @}
5032 This warning is enabled by @option{-Wall}.
5034 @item -Waggregate-return
5035 @opindex Waggregate-return
5036 @opindex Wno-aggregate-return
5037 Warn if any functions that return structures or unions are defined or
5038 called. (In languages where you can return an array, this also elicits
5041 @item -Wno-aggressive-loop-optimizations
5042 @opindex Wno-aggressive-loop-optimizations
5043 @opindex Waggressive-loop-optimizations
5044 Warn if in a loop with constant number of iterations the compiler detects
5045 undefined behavior in some statement during one or more of the iterations.
5047 @item -Wno-attributes
5048 @opindex Wno-attributes
5049 @opindex Wattributes
5050 Do not warn if an unexpected @code{__attribute__} is used, such as
5051 unrecognized attributes, function attributes applied to variables,
5052 etc. This does not stop errors for incorrect use of supported
5055 @item -Wno-builtin-macro-redefined
5056 @opindex Wno-builtin-macro-redefined
5057 @opindex Wbuiltin-macro-redefined
5058 Do not warn if certain built-in macros are redefined. This suppresses
5059 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
5060 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
5062 @item -Wstrict-prototypes @r{(C and Objective-C only)}
5063 @opindex Wstrict-prototypes
5064 @opindex Wno-strict-prototypes
5065 Warn if a function is declared or defined without specifying the
5066 argument types. (An old-style function definition is permitted without
5067 a warning if preceded by a declaration that specifies the argument
5070 @item -Wold-style-declaration @r{(C and Objective-C only)}
5071 @opindex Wold-style-declaration
5072 @opindex Wno-old-style-declaration
5073 Warn for obsolescent usages, according to the C Standard, in a
5074 declaration. For example, warn if storage-class specifiers like
5075 @code{static} are not the first things in a declaration. This warning
5076 is also enabled by @option{-Wextra}.
5078 @item -Wold-style-definition @r{(C and Objective-C only)}
5079 @opindex Wold-style-definition
5080 @opindex Wno-old-style-definition
5081 Warn if an old-style function definition is used. A warning is given
5082 even if there is a previous prototype.
5084 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
5085 @opindex Wmissing-parameter-type
5086 @opindex Wno-missing-parameter-type
5087 A function parameter is declared without a type specifier in K&R-style
5094 This warning is also enabled by @option{-Wextra}.
5096 @item -Wmissing-prototypes @r{(C and Objective-C only)}
5097 @opindex Wmissing-prototypes
5098 @opindex Wno-missing-prototypes
5099 Warn if a global function is defined without a previous prototype
5100 declaration. This warning is issued even if the definition itself
5101 provides a prototype. Use this option to detect global functions
5102 that do not have a matching prototype declaration in a header file.
5103 This option is not valid for C++ because all function declarations
5104 provide prototypes and a non-matching declaration declares an
5105 overload rather than conflict with an earlier declaration.
5106 Use @option{-Wmissing-declarations} to detect missing declarations in C++.
5108 @item -Wmissing-declarations
5109 @opindex Wmissing-declarations
5110 @opindex Wno-missing-declarations
5111 Warn if a global function is defined without a previous declaration.
5112 Do so even if the definition itself provides a prototype.
5113 Use this option to detect global functions that are not declared in
5114 header files. In C, no warnings are issued for functions with previous
5115 non-prototype declarations; use @option{-Wmissing-prototypes} to detect
5116 missing prototypes. In C++, no warnings are issued for function templates,
5117 or for inline functions, or for functions in anonymous namespaces.
5119 @item -Wmissing-field-initializers
5120 @opindex Wmissing-field-initializers
5121 @opindex Wno-missing-field-initializers
5125 Warn if a structure's initializer has some fields missing. For
5126 example, the following code causes such a warning, because
5127 @code{x.h} is implicitly zero:
5130 struct s @{ int f, g, h; @};
5131 struct s x = @{ 3, 4 @};
5134 This option does not warn about designated initializers, so the following
5135 modification does not trigger a warning:
5138 struct s @{ int f, g, h; @};
5139 struct s x = @{ .f = 3, .g = 4 @};
5142 In C++ this option does not warn either about the empty @{ @}
5143 initializer, for example:
5146 struct s @{ int f, g, h; @};
5150 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
5151 warnings without this one, use @option{-Wextra -Wno-missing-field-initializers}.
5153 @item -Wno-multichar
5154 @opindex Wno-multichar
5156 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
5157 Usually they indicate a typo in the user's code, as they have
5158 implementation-defined values, and should not be used in portable code.
5160 @item -Wnormalized@r{[}=@r{<}none@r{|}id@r{|}nfc@r{|}nfkc@r{>]}
5161 @opindex Wnormalized=
5162 @opindex Wnormalized
5163 @opindex Wno-normalized
5166 @cindex character set, input normalization
5167 In ISO C and ISO C++, two identifiers are different if they are
5168 different sequences of characters. However, sometimes when characters
5169 outside the basic ASCII character set are used, you can have two
5170 different character sequences that look the same. To avoid confusion,
5171 the ISO 10646 standard sets out some @dfn{normalization rules} which
5172 when applied ensure that two sequences that look the same are turned into
5173 the same sequence. GCC can warn you if you are using identifiers that
5174 have not been normalized; this option controls that warning.
5176 There are four levels of warning supported by GCC@. The default is
5177 @option{-Wnormalized=nfc}, which warns about any identifier that is
5178 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
5179 recommended form for most uses. It is equivalent to
5180 @option{-Wnormalized}.
5182 Unfortunately, there are some characters allowed in identifiers by
5183 ISO C and ISO C++ that, when turned into NFC, are not allowed in
5184 identifiers. That is, there's no way to use these symbols in portable
5185 ISO C or C++ and have all your identifiers in NFC@.
5186 @option{-Wnormalized=id} suppresses the warning for these characters.
5187 It is hoped that future versions of the standards involved will correct
5188 this, which is why this option is not the default.
5190 You can switch the warning off for all characters by writing
5191 @option{-Wnormalized=none} or @option{-Wno-normalized}. You should
5192 only do this if you are using some other normalization scheme (like
5193 ``D''), because otherwise you can easily create bugs that are
5194 literally impossible to see.
5196 Some characters in ISO 10646 have distinct meanings but look identical
5197 in some fonts or display methodologies, especially once formatting has
5198 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
5199 LETTER N'', displays just like a regular @code{n} that has been
5200 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
5201 normalization scheme to convert all these into a standard form as
5202 well, and GCC warns if your code is not in NFKC if you use
5203 @option{-Wnormalized=nfkc}. This warning is comparable to warning
5204 about every identifier that contains the letter O because it might be
5205 confused with the digit 0, and so is not the default, but may be
5206 useful as a local coding convention if the programming environment
5207 cannot be fixed to display these characters distinctly.
5209 @item -Wno-deprecated
5210 @opindex Wno-deprecated
5211 @opindex Wdeprecated
5212 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
5214 @item -Wno-deprecated-declarations
5215 @opindex Wno-deprecated-declarations
5216 @opindex Wdeprecated-declarations
5217 Do not warn about uses of functions (@pxref{Function Attributes}),
5218 variables (@pxref{Variable Attributes}), and types (@pxref{Type
5219 Attributes}) marked as deprecated by using the @code{deprecated}
5223 @opindex Wno-overflow
5225 Do not warn about compile-time overflow in constant expressions.
5230 Warn about One Definition Rule violations during link-time optimization.
5231 Requires @option{-flto-odr-type-merging} to be enabled. Enabled by default.
5234 @opindex Wopenm-simd
5235 Warn if the vectorizer cost model overrides the OpenMP or the Cilk Plus
5236 simd directive set by user. The @option{-fsimd-cost-model=unlimited}
5237 option can be used to relax the cost model.
5239 @item -Woverride-init @r{(C and Objective-C only)}
5240 @opindex Woverride-init
5241 @opindex Wno-override-init
5245 Warn if an initialized field without side effects is overridden when
5246 using designated initializers (@pxref{Designated Inits, , Designated
5249 This warning is included in @option{-Wextra}. To get other
5250 @option{-Wextra} warnings without this one, use @option{-Wextra
5251 -Wno-override-init}.
5253 @item -Woverride-init-side-effects @r{(C and Objective-C only)}
5254 @opindex Woverride-init-side-effects
5255 @opindex Wno-override-init-side-effects
5256 Warn if an initialized field with side effects is overridden when
5257 using designated initializers (@pxref{Designated Inits, , Designated
5258 Initializers}). This warning is enabled by default.
5263 Warn if a structure is given the packed attribute, but the packed
5264 attribute has no effect on the layout or size of the structure.
5265 Such structures may be mis-aligned for little benefit. For
5266 instance, in this code, the variable @code{f.x} in @code{struct bar}
5267 is misaligned even though @code{struct bar} does not itself
5268 have the packed attribute:
5275 @} __attribute__((packed));
5283 @item -Wpacked-bitfield-compat
5284 @opindex Wpacked-bitfield-compat
5285 @opindex Wno-packed-bitfield-compat
5286 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
5287 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
5288 the change can lead to differences in the structure layout. GCC
5289 informs you when the offset of such a field has changed in GCC 4.4.
5290 For example there is no longer a 4-bit padding between field @code{a}
5291 and @code{b} in this structure:
5298 @} __attribute__ ((packed));
5301 This warning is enabled by default. Use
5302 @option{-Wno-packed-bitfield-compat} to disable this warning.
5307 Warn if padding is included in a structure, either to align an element
5308 of the structure or to align the whole structure. Sometimes when this
5309 happens it is possible to rearrange the fields of the structure to
5310 reduce the padding and so make the structure smaller.
5312 @item -Wredundant-decls
5313 @opindex Wredundant-decls
5314 @opindex Wno-redundant-decls
5315 Warn if anything is declared more than once in the same scope, even in
5316 cases where multiple declaration is valid and changes nothing.
5318 @item -Wnested-externs @r{(C and Objective-C only)}
5319 @opindex Wnested-externs
5320 @opindex Wno-nested-externs
5321 Warn if an @code{extern} declaration is encountered within a function.
5323 @item -Wno-inherited-variadic-ctor
5324 @opindex Winherited-variadic-ctor
5325 @opindex Wno-inherited-variadic-ctor
5326 Suppress warnings about use of C++11 inheriting constructors when the
5327 base class inherited from has a C variadic constructor; the warning is
5328 on by default because the ellipsis is not inherited.
5333 Warn if a function that is declared as inline cannot be inlined.
5334 Even with this option, the compiler does not warn about failures to
5335 inline functions declared in system headers.
5337 The compiler uses a variety of heuristics to determine whether or not
5338 to inline a function. For example, the compiler takes into account
5339 the size of the function being inlined and the amount of inlining
5340 that has already been done in the current function. Therefore,
5341 seemingly insignificant changes in the source program can cause the
5342 warnings produced by @option{-Winline} to appear or disappear.
5344 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
5345 @opindex Wno-invalid-offsetof
5346 @opindex Winvalid-offsetof
5347 Suppress warnings from applying the @code{offsetof} macro to a non-POD
5348 type. According to the 2014 ISO C++ standard, applying @code{offsetof}
5349 to a non-standard-layout type is undefined. In existing C++ implementations,
5350 however, @code{offsetof} typically gives meaningful results.
5351 This flag is for users who are aware that they are
5352 writing nonportable code and who have deliberately chosen to ignore the
5355 The restrictions on @code{offsetof} may be relaxed in a future version
5356 of the C++ standard.
5358 @item -Wno-int-to-pointer-cast
5359 @opindex Wno-int-to-pointer-cast
5360 @opindex Wint-to-pointer-cast
5361 Suppress warnings from casts to pointer type of an integer of a
5362 different size. In C++, casting to a pointer type of smaller size is
5363 an error. @option{Wint-to-pointer-cast} is enabled by default.
5366 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
5367 @opindex Wno-pointer-to-int-cast
5368 @opindex Wpointer-to-int-cast
5369 Suppress warnings from casts from a pointer to an integer type of a
5373 @opindex Winvalid-pch
5374 @opindex Wno-invalid-pch
5375 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
5376 the search path but can't be used.
5380 @opindex Wno-long-long
5381 Warn if @code{long long} type is used. This is enabled by either
5382 @option{-Wpedantic} or @option{-Wtraditional} in ISO C90 and C++98
5383 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
5385 @item -Wvariadic-macros
5386 @opindex Wvariadic-macros
5387 @opindex Wno-variadic-macros
5388 Warn if variadic macros are used in ISO C90 mode, or if the GNU
5389 alternate syntax is used in ISO C99 mode. This is enabled by either
5390 @option{-Wpedantic} or @option{-Wtraditional}. To inhibit the warning
5391 messages, use @option{-Wno-variadic-macros}.
5395 @opindex Wno-varargs
5396 Warn upon questionable usage of the macros used to handle variable
5397 arguments like @code{va_start}. This is default. To inhibit the
5398 warning messages, use @option{-Wno-varargs}.
5400 @item -Wvector-operation-performance
5401 @opindex Wvector-operation-performance
5402 @opindex Wno-vector-operation-performance
5403 Warn if vector operation is not implemented via SIMD capabilities of the
5404 architecture. Mainly useful for the performance tuning.
5405 Vector operation can be implemented @code{piecewise}, which means that the
5406 scalar operation is performed on every vector element;
5407 @code{in parallel}, which means that the vector operation is implemented
5408 using scalars of wider type, which normally is more performance efficient;
5409 and @code{as a single scalar}, which means that vector fits into a
5412 @item -Wno-virtual-move-assign
5413 @opindex Wvirtual-move-assign
5414 @opindex Wno-virtual-move-assign
5415 Suppress warnings about inheriting from a virtual base with a
5416 non-trivial C++11 move assignment operator. This is dangerous because
5417 if the virtual base is reachable along more than one path, it is
5418 moved multiple times, which can mean both objects end up in the
5419 moved-from state. If the move assignment operator is written to avoid
5420 moving from a moved-from object, this warning can be disabled.
5425 Warn if variable length array is used in the code.
5426 @option{-Wno-vla} prevents the @option{-Wpedantic} warning of
5427 the variable length array.
5429 @item -Wvolatile-register-var
5430 @opindex Wvolatile-register-var
5431 @opindex Wno-volatile-register-var
5432 Warn if a register variable is declared volatile. The volatile
5433 modifier does not inhibit all optimizations that may eliminate reads
5434 and/or writes to register variables. This warning is enabled by
5437 @item -Wdisabled-optimization
5438 @opindex Wdisabled-optimization
5439 @opindex Wno-disabled-optimization
5440 Warn if a requested optimization pass is disabled. This warning does
5441 not generally indicate that there is anything wrong with your code; it
5442 merely indicates that GCC's optimizers are unable to handle the code
5443 effectively. Often, the problem is that your code is too big or too
5444 complex; GCC refuses to optimize programs when the optimization
5445 itself is likely to take inordinate amounts of time.
5447 @item -Wpointer-sign @r{(C and Objective-C only)}
5448 @opindex Wpointer-sign
5449 @opindex Wno-pointer-sign
5450 Warn for pointer argument passing or assignment with different signedness.
5451 This option is only supported for C and Objective-C@. It is implied by
5452 @option{-Wall} and by @option{-Wpedantic}, which can be disabled with
5453 @option{-Wno-pointer-sign}.
5455 @item -Wstack-protector
5456 @opindex Wstack-protector
5457 @opindex Wno-stack-protector
5458 This option is only active when @option{-fstack-protector} is active. It
5459 warns about functions that are not protected against stack smashing.
5461 @item -Woverlength-strings
5462 @opindex Woverlength-strings
5463 @opindex Wno-overlength-strings
5464 Warn about string constants that are longer than the ``minimum
5465 maximum'' length specified in the C standard. Modern compilers
5466 generally allow string constants that are much longer than the
5467 standard's minimum limit, but very portable programs should avoid
5468 using longer strings.
5470 The limit applies @emph{after} string constant concatenation, and does
5471 not count the trailing NUL@. In C90, the limit was 509 characters; in
5472 C99, it was raised to 4095. C++98 does not specify a normative
5473 minimum maximum, so we do not diagnose overlength strings in C++@.
5475 This option is implied by @option{-Wpedantic}, and can be disabled with
5476 @option{-Wno-overlength-strings}.
5478 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
5479 @opindex Wunsuffixed-float-constants
5481 Issue a warning for any floating constant that does not have
5482 a suffix. When used together with @option{-Wsystem-headers} it
5483 warns about such constants in system header files. This can be useful
5484 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
5485 from the decimal floating-point extension to C99.
5487 @item -Wno-designated-init @r{(C and Objective-C only)}
5488 Suppress warnings when a positional initializer is used to initialize
5489 a structure that has been marked with the @code{designated_init}
5494 @node Debugging Options
5495 @section Options for Debugging Your Program or GCC
5496 @cindex options, debugging
5497 @cindex debugging information options
5499 GCC has various special options that are used for debugging
5500 either your program or GCC:
5505 Produce debugging information in the operating system's native format
5506 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
5509 On most systems that use stabs format, @option{-g} enables use of extra
5510 debugging information that only GDB can use; this extra information
5511 makes debugging work better in GDB but probably makes other debuggers
5513 refuse to read the program. If you want to control for certain whether
5514 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
5515 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
5517 GCC allows you to use @option{-g} with
5518 @option{-O}. The shortcuts taken by optimized code may occasionally
5519 produce surprising results: some variables you declared may not exist
5520 at all; flow of control may briefly move where you did not expect it;
5521 some statements may not be executed because they compute constant
5522 results or their values are already at hand; some statements may
5523 execute in different places because they have been moved out of loops.
5525 Nevertheless it proves possible to debug optimized output. This makes
5526 it reasonable to use the optimizer for programs that might have bugs.
5528 The following options are useful when GCC is generated with the
5529 capability for more than one debugging format.
5532 @opindex gsplit-dwarf
5533 Separate as much dwarf debugging information as possible into a
5534 separate output file with the extension .dwo. This option allows
5535 the build system to avoid linking files with debug information. To
5536 be useful, this option requires a debugger capable of reading .dwo
5541 Produce debugging information for use by GDB@. This means to use the
5542 most expressive format available (DWARF 2, stabs, or the native format
5543 if neither of those are supported), including GDB extensions if at all
5548 Generate dwarf .debug_pubnames and .debug_pubtypes sections.
5550 @item -ggnu-pubnames
5551 @opindex ggnu-pubnames
5552 Generate .debug_pubnames and .debug_pubtypes sections in a format
5553 suitable for conversion into a GDB@ index. This option is only useful
5554 with a linker that can produce GDB@ index version 7.
5558 Produce debugging information in stabs format (if that is supported),
5559 without GDB extensions. This is the format used by DBX on most BSD
5560 systems. On MIPS, Alpha and System V Release 4 systems this option
5561 produces stabs debugging output that is not understood by DBX or SDB@.
5562 On System V Release 4 systems this option requires the GNU assembler.
5564 @item -feliminate-unused-debug-symbols
5565 @opindex feliminate-unused-debug-symbols
5566 Produce debugging information in stabs format (if that is supported),
5567 for only symbols that are actually used.
5569 @item -femit-class-debug-always
5570 @opindex femit-class-debug-always
5571 Instead of emitting debugging information for a C++ class in only one
5572 object file, emit it in all object files using the class. This option
5573 should be used only with debuggers that are unable to handle the way GCC
5574 normally emits debugging information for classes because using this
5575 option increases the size of debugging information by as much as a
5578 @item -fdebug-types-section
5579 @opindex fdebug-types-section
5580 @opindex fno-debug-types-section
5581 When using DWARF Version 4 or higher, type DIEs can be put into
5582 their own @code{.debug_types} section instead of making them part of the
5583 @code{.debug_info} section. It is more efficient to put them in a separate
5584 comdat sections since the linker can then remove duplicates.
5585 But not all DWARF consumers support @code{.debug_types} sections yet
5586 and on some objects @code{.debug_types} produces larger instead of smaller
5587 debugging information.
5591 Produce debugging information in stabs format (if that is supported),
5592 using GNU extensions understood only by the GNU debugger (GDB)@. The
5593 use of these extensions is likely to make other debuggers crash or
5594 refuse to read the program.
5598 Produce debugging information in COFF format (if that is supported).
5599 This is the format used by SDB on most System V systems prior to
5604 Produce debugging information in XCOFF format (if that is supported).
5605 This is the format used by the DBX debugger on IBM RS/6000 systems.
5609 Produce debugging information in XCOFF format (if that is supported),
5610 using GNU extensions understood only by the GNU debugger (GDB)@. The
5611 use of these extensions is likely to make other debuggers crash or
5612 refuse to read the program, and may cause assemblers other than the GNU
5613 assembler (GAS) to fail with an error.
5615 @item -gdwarf-@var{version}
5616 @opindex gdwarf-@var{version}
5617 Produce debugging information in DWARF format (if that is supported).
5618 The value of @var{version} may be either 2, 3, 4 or 5; the default version
5619 for most targets is 4. DWARF Version 5 is only experimental.
5621 Note that with DWARF Version 2, some ports require and always
5622 use some non-conflicting DWARF 3 extensions in the unwind tables.
5624 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
5625 for maximum benefit.
5627 @item -grecord-gcc-switches
5628 @opindex grecord-gcc-switches
5629 This switch causes the command-line options used to invoke the
5630 compiler that may affect code generation to be appended to the
5631 DW_AT_producer attribute in DWARF debugging information. The options
5632 are concatenated with spaces separating them from each other and from
5633 the compiler version. See also @option{-frecord-gcc-switches} for another
5634 way of storing compiler options into the object file. This is the default.
5636 @item -gno-record-gcc-switches
5637 @opindex gno-record-gcc-switches
5638 Disallow appending command-line options to the DW_AT_producer attribute
5639 in DWARF debugging information.
5641 @item -gstrict-dwarf
5642 @opindex gstrict-dwarf
5643 Disallow using extensions of later DWARF standard version than selected
5644 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
5645 DWARF extensions from later standard versions is allowed.
5647 @item -gno-strict-dwarf
5648 @opindex gno-strict-dwarf
5649 Allow using extensions of later DWARF standard version than selected with
5650 @option{-gdwarf-@var{version}}.
5652 @item -gz@r{[}=@var{type}@r{]}
5654 Produce compressed debug sections in DWARF format, if that is supported.
5655 If @var{type} is not given, the default type depends on the capabilities
5656 of the assembler and linker used. @var{type} may be one of
5657 @samp{none} (don't compress debug sections), @samp{zlib} (use zlib
5658 compression in ELF gABI format), or @samp{zlib-gnu} (use zlib
5659 compression in traditional GNU format). If the linker doesn't support
5660 writing compressed debug sections, the option is rejected. Otherwise,
5661 if the assembler does not support them, @option{-gz} is silently ignored
5662 when producing object files.
5666 Produce debugging information in Alpha/VMS debug format (if that is
5667 supported). This is the format used by DEBUG on Alpha/VMS systems.
5670 @itemx -ggdb@var{level}
5671 @itemx -gstabs@var{level}
5672 @itemx -gcoff@var{level}
5673 @itemx -gxcoff@var{level}
5674 @itemx -gvms@var{level}
5675 Request debugging information and also use @var{level} to specify how
5676 much information. The default level is 2.
5678 Level 0 produces no debug information at all. Thus, @option{-g0} negates
5681 Level 1 produces minimal information, enough for making backtraces in
5682 parts of the program that you don't plan to debug. This includes
5683 descriptions of functions and external variables, and line number
5684 tables, but no information about local variables.
5686 Level 3 includes extra information, such as all the macro definitions
5687 present in the program. Some debuggers support macro expansion when
5688 you use @option{-g3}.
5690 @option{-gdwarf-2} does not accept a concatenated debug level, because
5691 GCC used to support an option @option{-gdwarf} that meant to generate
5692 debug information in version 1 of the DWARF format (which is very
5693 different from version 2), and it would have been too confusing. That
5694 debug format is long obsolete, but the option cannot be changed now.
5695 Instead use an additional @option{-g@var{level}} option to change the
5696 debug level for DWARF.
5700 Turn off generation of debug info, if leaving out this option
5701 generates it, or turn it on at level 2 otherwise. The position of this
5702 argument in the command line does not matter; it takes effect after all
5703 other options are processed, and it does so only once, no matter how
5704 many times it is given. This is mainly intended to be used with
5705 @option{-fcompare-debug}.
5707 @item -fsanitize=address
5708 @opindex fsanitize=address
5709 Enable AddressSanitizer, a fast memory error detector.
5710 Memory access instructions are instrumented to detect
5711 out-of-bounds and use-after-free bugs.
5712 See @uref{http://code.google.com/p/address-sanitizer/} for
5713 more details. The run-time behavior can be influenced using the
5714 @env{ASAN_OPTIONS} environment variable; see
5715 @url{https://code.google.com/p/address-sanitizer/wiki/Flags#Run-time_flags} for
5716 a list of supported options.
5718 @item -fsanitize=kernel-address
5719 @opindex fsanitize=kernel-address
5720 Enable AddressSanitizer for Linux kernel.
5721 See @uref{http://code.google.com/p/address-sanitizer/wiki/AddressSanitizerForKernel} for more details.
5723 @item -fsanitize=thread
5724 @opindex fsanitize=thread
5725 Enable ThreadSanitizer, a fast data race detector.
5726 Memory access instructions are instrumented to detect
5727 data race bugs. See @uref{http://code.google.com/p/thread-sanitizer/} for more
5728 details. The run-time behavior can be influenced using the @env{TSAN_OPTIONS}
5729 environment variable; see
5730 @url{https://code.google.com/p/thread-sanitizer/wiki/Flags} for a list of
5733 @item -fsanitize=leak
5734 @opindex fsanitize=leak
5735 Enable LeakSanitizer, a memory leak detector.
5736 This option only matters for linking of executables and if neither
5737 @option{-fsanitize=address} nor @option{-fsanitize=thread} is used. In that
5738 case the executable is linked against a library that overrides @code{malloc}
5739 and other allocator functions. See
5740 @uref{https://code.google.com/p/address-sanitizer/wiki/LeakSanitizer} for more
5741 details. The run-time behavior can be influenced using the
5742 @env{LSAN_OPTIONS} environment variable.
5744 @item -fsanitize=undefined
5745 @opindex fsanitize=undefined
5746 Enable UndefinedBehaviorSanitizer, a fast undefined behavior detector.
5747 Various computations are instrumented to detect undefined behavior
5748 at runtime. Current suboptions are:
5752 @item -fsanitize=shift
5753 @opindex fsanitize=shift
5754 This option enables checking that the result of a shift operation is
5755 not undefined. Note that what exactly is considered undefined differs
5756 slightly between C and C++, as well as between ISO C90 and C99, etc.
5758 @item -fsanitize=integer-divide-by-zero
5759 @opindex fsanitize=integer-divide-by-zero
5760 Detect integer division by zero as well as @code{INT_MIN / -1} division.
5762 @item -fsanitize=unreachable
5763 @opindex fsanitize=unreachable
5764 With this option, the compiler turns the @code{__builtin_unreachable}
5765 call into a diagnostics message call instead. When reaching the
5766 @code{__builtin_unreachable} call, the behavior is undefined.
5768 @item -fsanitize=vla-bound
5769 @opindex fsanitize=vla-bound
5770 This option instructs the compiler to check that the size of a variable
5771 length array is positive.
5773 @item -fsanitize=null
5774 @opindex fsanitize=null
5775 This option enables pointer checking. Particularly, the application
5776 built with this option turned on will issue an error message when it
5777 tries to dereference a NULL pointer, or if a reference (possibly an
5778 rvalue reference) is bound to a NULL pointer, or if a method is invoked
5779 on an object pointed by a NULL pointer.
5781 @item -fsanitize=return
5782 @opindex fsanitize=return
5783 This option enables return statement checking. Programs
5784 built with this option turned on will issue an error message
5785 when the end of a non-void function is reached without actually
5786 returning a value. This option works in C++ only.
5788 @item -fsanitize=signed-integer-overflow
5789 @opindex fsanitize=signed-integer-overflow
5790 This option enables signed integer overflow checking. We check that
5791 the result of @code{+}, @code{*}, and both unary and binary @code{-}
5792 does not overflow in the signed arithmetics. Note, integer promotion
5793 rules must be taken into account. That is, the following is not an
5796 signed char a = SCHAR_MAX;
5800 @item -fsanitize=bounds
5801 @opindex fsanitize=bounds
5802 This option enables instrumentation of array bounds. Various out of bounds
5803 accesses are detected. Flexible array members, flexible array member-like
5804 arrays, and initializers of variables with static storage are not instrumented.
5806 @item -fsanitize=bounds-strict
5807 @opindex fsanitize=bounds-strict
5808 This option enables strict instrumentation of array bounds. Most out of bounds
5809 accesses are detected, including flexible array members and flexible array
5810 member-like arrays. Initializers of variables with static storage are not
5813 @item -fsanitize=alignment
5814 @opindex fsanitize=alignment
5816 This option enables checking of alignment of pointers when they are
5817 dereferenced, or when a reference is bound to insufficiently aligned target,
5818 or when a method or constructor is invoked on insufficiently aligned object.
5820 @item -fsanitize=object-size
5821 @opindex fsanitize=object-size
5822 This option enables instrumentation of memory references using the
5823 @code{__builtin_object_size} function. Various out of bounds pointer
5824 accesses are detected.
5826 @item -fsanitize=float-divide-by-zero
5827 @opindex fsanitize=float-divide-by-zero
5828 Detect floating-point division by zero. Unlike other similar options,
5829 @option{-fsanitize=float-divide-by-zero} is not enabled by
5830 @option{-fsanitize=undefined}, since floating-point division by zero can
5831 be a legitimate way of obtaining infinities and NaNs.
5833 @item -fsanitize=float-cast-overflow
5834 @opindex fsanitize=float-cast-overflow
5835 This option enables floating-point type to integer conversion checking.
5836 We check that the result of the conversion does not overflow.
5837 Unlike other similar options, @option{-fsanitize=float-cast-overflow} is
5838 not enabled by @option{-fsanitize=undefined}.
5839 This option does not work well with @code{FE_INVALID} exceptions enabled.
5841 @item -fsanitize=nonnull-attribute
5842 @opindex fsanitize=nonnull-attribute
5844 This option enables instrumentation of calls, checking whether null values
5845 are not passed to arguments marked as requiring a non-null value by the
5846 @code{nonnull} function attribute.
5848 @item -fsanitize=returns-nonnull-attribute
5849 @opindex fsanitize=returns-nonnull-attribute
5851 This option enables instrumentation of return statements in functions
5852 marked with @code{returns_nonnull} function attribute, to detect returning
5853 of null values from such functions.
5855 @item -fsanitize=bool
5856 @opindex fsanitize=bool
5858 This option enables instrumentation of loads from bool. If a value other
5859 than 0/1 is loaded, a run-time error is issued.
5861 @item -fsanitize=enum
5862 @opindex fsanitize=enum
5864 This option enables instrumentation of loads from an enum type. If
5865 a value outside the range of values for the enum type is loaded,
5866 a run-time error is issued.
5868 @item -fsanitize=vptr
5869 @opindex fsanitize=vptr
5871 This option enables instrumentation of C++ member function calls, member
5872 accesses and some conversions between pointers to base and derived classes,
5873 to verify the referenced object has the correct dynamic type.
5877 While @option{-ftrapv} causes traps for signed overflows to be emitted,
5878 @option{-fsanitize=undefined} gives a diagnostic message.
5879 This currently works only for the C family of languages.
5881 @item -fno-sanitize=all
5882 @opindex fno-sanitize=all
5884 This option disables all previously enabled sanitizers.
5885 @option{-fsanitize=all} is not allowed, as some sanitizers cannot be used
5888 @item -fasan-shadow-offset=@var{number}
5889 @opindex fasan-shadow-offset
5890 This option forces GCC to use custom shadow offset in AddressSanitizer checks.
5891 It is useful for experimenting with different shadow memory layouts in
5892 Kernel AddressSanitizer.
5894 @item -fsanitize-sections=@var{s1},@var{s2},...
5895 @opindex fsanitize-sections
5896 Sanitize global variables in selected user-defined sections. @var{si} may
5899 @item -fsanitize-recover@r{[}=@var{opts}@r{]}
5900 @opindex fsanitize-recover
5901 @opindex fno-sanitize-recover
5902 @option{-fsanitize-recover=} controls error recovery mode for sanitizers
5903 mentioned in comma-separated list of @var{opts}. Enabling this option
5904 for a sanitizer component causes it to attempt to continue
5905 running the program as if no error happened. This means multiple
5906 runtime errors can be reported in a single program run, and the exit
5907 code of the program may indicate success even when errors
5908 have been reported. The @option{-fno-sanitize-recover=} option
5909 can be used to alter
5910 this behavior: only the first detected error is reported
5911 and program then exits with a non-zero exit code.
5913 Currently this feature only works for @option{-fsanitize=undefined} (and its suboptions
5914 except for @option{-fsanitize=unreachable} and @option{-fsanitize=return}),
5915 @option{-fsanitize=float-cast-overflow}, @option{-fsanitize=float-divide-by-zero} and
5916 @option{-fsanitize=kernel-address}. For these sanitizers error recovery is turned on by default.
5917 @option{-fsanitize-recover=all} and @option{-fno-sanitize-recover=all} is also
5918 accepted, the former enables recovery for all sanitizers that support it,
5919 the latter disables recovery for all sanitizers that support it.
5921 Syntax without explicit @var{opts} parameter is deprecated. It is equivalent to
5923 -fsanitize-recover=undefined,float-cast-overflow,float-divide-by-zero
5926 Similarly @option{-fno-sanitize-recover} is equivalent to
5928 -fno-sanitize-recover=undefined,float-cast-overflow,float-divide-by-zero
5931 @item -fsanitize-undefined-trap-on-error
5932 @opindex fsanitize-undefined-trap-on-error
5933 The @option{-fsanitize-undefined-trap-on-error} option instructs the compiler to
5934 report undefined behavior using @code{__builtin_trap} rather than
5935 a @code{libubsan} library routine. The advantage of this is that the
5936 @code{libubsan} library is not needed and is not linked in, so this
5937 is usable even in freestanding environments.
5939 @item -fcheck-pointer-bounds
5940 @opindex fcheck-pointer-bounds
5941 @opindex fno-check-pointer-bounds
5942 @cindex Pointer Bounds Checker options
5943 Enable Pointer Bounds Checker instrumentation. Each memory reference
5944 is instrumented with checks of the pointer used for memory access against
5945 bounds associated with that pointer.
5948 is only an implementation for Intel MPX available, thus x86 target
5949 and @option{-mmpx} are required to enable this feature.
5950 MPX-based instrumentation requires
5951 a runtime library to enable MPX in hardware and handle bounds
5952 violation signals. By default when @option{-fcheck-pointer-bounds}
5953 and @option{-mmpx} options are used to link a program, the GCC driver
5954 links against the @file{libmpx} runtime library and @file{libmpxwrappers}
5955 library. It also passes '-z bndplt' to a linker in case it supports this
5956 option (which is checked on libmpx configuration). Note that old versions
5957 of linker may ignore option. Gold linker doesn't support '-z bndplt'
5958 option. With no '-z bndplt' support in linker all calls to dynamic libraries
5959 lose passed bounds reducing overall protection level. It's highly
5960 recommended to use linker with '-z bndplt' support. In case such linker
5961 is not available it is adviced to always use @option{-static-libmpxwrappers}
5962 for better protection level or use @option{-static} to completely avoid
5963 external calls to dynamic libraries. MPX-based instrumentation
5964 may be used for debugging and also may be included in production code
5965 to increase program security. Depending on usage, you may
5966 have different requirements for the runtime library. The current version
5967 of the MPX runtime library is more oriented for use as a debugging
5968 tool. MPX runtime library usage implies @option{-lpthread}. See
5969 also @option{-static-libmpx}. The runtime library behavior can be
5970 influenced using various @env{CHKP_RT_*} environment variables. See
5971 @uref{https://gcc.gnu.org/wiki/Intel%20MPX%20support%20in%20the%20GCC%20compiler}
5974 Generated instrumentation may be controlled by various
5975 @option{-fchkp-*} options and by the @code{bnd_variable_size}
5976 structure field attribute (@pxref{Type Attributes}) and
5977 @code{bnd_legacy}, and @code{bnd_instrument} function attributes
5978 (@pxref{Function Attributes}). GCC also provides a number of built-in
5979 functions for controlling the Pointer Bounds Checker. @xref{Pointer
5980 Bounds Checker builtins}, for more information.
5982 @item -fchkp-check-incomplete-type
5983 @opindex fchkp-check-incomplete-type
5984 @opindex fno-chkp-check-incomplete-type
5985 Generate pointer bounds checks for variables with incomplete type.
5988 @item -fchkp-narrow-bounds
5989 @opindex fchkp-narrow-bounds
5990 @opindex fno-chkp-narrow-bounds
5991 Controls bounds used by Pointer Bounds Checker for pointers to object
5992 fields. If narrowing is enabled then field bounds are used. Otherwise
5993 object bounds are used. See also @option{-fchkp-narrow-to-innermost-array}
5994 and @option{-fchkp-first-field-has-own-bounds}. Enabled by default.
5996 @item -fchkp-first-field-has-own-bounds
5997 @opindex fchkp-first-field-has-own-bounds
5998 @opindex fno-chkp-first-field-has-own-bounds
5999 Forces Pointer Bounds Checker to use narrowed bounds for the address of the
6000 first field in the structure. By default a pointer to the first field has
6001 the same bounds as a pointer to the whole structure.
6003 @item -fchkp-narrow-to-innermost-array
6004 @opindex fchkp-narrow-to-innermost-array
6005 @opindex fno-chkp-narrow-to-innermost-array
6006 Forces Pointer Bounds Checker to use bounds of the innermost arrays in
6007 case of nested static array access. By default this option is disabled and
6008 bounds of the outermost array are used.
6010 @item -fchkp-optimize
6011 @opindex fchkp-optimize
6012 @opindex fno-chkp-optimize
6013 Enables Pointer Bounds Checker optimizations. Enabled by default at
6014 optimization levels @option{-O}, @option{-O2}, @option{-O3}.
6016 @item -fchkp-use-fast-string-functions
6017 @opindex fchkp-use-fast-string-functions
6018 @opindex fno-chkp-use-fast-string-functions
6019 Enables use of @code{*_nobnd} versions of string functions (not copying bounds)
6020 by Pointer Bounds Checker. Disabled by default.
6022 @item -fchkp-use-nochk-string-functions
6023 @opindex fchkp-use-nochk-string-functions
6024 @opindex fno-chkp-use-nochk-string-functions
6025 Enables use of @code{*_nochk} versions of string functions (not checking bounds)
6026 by Pointer Bounds Checker. Disabled by default.
6028 @item -fchkp-use-static-bounds
6029 @opindex fchkp-use-static-bounds
6030 @opindex fno-chkp-use-static-bounds
6031 Allow Pointer Bounds Checker to generate static bounds holding
6032 bounds of static variables. Enabled by default.
6034 @item -fchkp-use-static-const-bounds
6035 @opindex fchkp-use-static-const-bounds
6036 @opindex fno-chkp-use-static-const-bounds
6037 Use statically-initialized bounds for constant bounds instead of
6038 generating them each time they are required. By default enabled when
6039 @option{-fchkp-use-static-bounds} is enabled.
6041 @item -fchkp-treat-zero-dynamic-size-as-infinite
6042 @opindex fchkp-treat-zero-dynamic-size-as-infinite
6043 @opindex fno-chkp-treat-zero-dynamic-size-as-infinite
6044 With this option, objects with incomplete type whose
6045 dynamically-obtained size is zero are treated as having infinite size
6046 instead by Pointer Bounds
6047 Checker. This option may be helpful if a program is linked with a library
6048 missing size information for some symbols. Disabled by default.
6050 @item -fchkp-check-read
6051 @opindex fchkp-check-read
6052 @opindex fno-chkp-check-read
6053 Instructs Pointer Bounds Checker to generate checks for all read
6054 accesses to memory. Enabled by default.
6056 @item -fchkp-check-write
6057 @opindex fchkp-check-write
6058 @opindex fno-chkp-check-write
6059 Instructs Pointer Bounds Checker to generate checks for all write
6060 accesses to memory. Enabled by default.
6062 @item -fchkp-store-bounds
6063 @opindex fchkp-store-bounds
6064 @opindex fno-chkp-store-bounds
6065 Instructs Pointer Bounds Checker to generate bounds stores for
6066 pointer writes. Enabled by default.
6068 @item -fchkp-instrument-calls
6069 @opindex fchkp-instrument-calls
6070 @opindex fno-chkp-instrument-calls
6071 Instructs Pointer Bounds Checker to pass pointer bounds to calls.
6074 @item -fchkp-instrument-marked-only
6075 @opindex fchkp-instrument-marked-only
6076 @opindex fno-chkp-instrument-marked-only
6077 Instructs Pointer Bounds Checker to instrument only functions
6078 marked with the @code{bnd_instrument} attribute
6079 (@pxref{Function Attributes}). Disabled by default.
6081 @item -fchkp-use-wrappers
6082 @opindex fchkp-use-wrappers
6083 @opindex fno-chkp-use-wrappers
6084 Allows Pointer Bounds Checker to replace calls to built-in functions
6085 with calls to wrapper functions. When @option{-fchkp-use-wrappers}
6086 is used to link a program, the GCC driver automatically links
6087 against @file{libmpxwrappers}. See also @option{-static-libmpxwrappers}.
6090 @item -fdump-final-insns@r{[}=@var{file}@r{]}
6091 @opindex fdump-final-insns
6092 Dump the final internal representation (RTL) to @var{file}. If the
6093 optional argument is omitted (or if @var{file} is @code{.}), the name
6094 of the dump file is determined by appending @code{.gkd} to the
6095 compilation output file name.
6097 @item -fcompare-debug@r{[}=@var{opts}@r{]}
6098 @opindex fcompare-debug
6099 @opindex fno-compare-debug
6100 If no error occurs during compilation, run the compiler a second time,
6101 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
6102 passed to the second compilation. Dump the final internal
6103 representation in both compilations, and print an error if they differ.
6105 If the equal sign is omitted, the default @option{-gtoggle} is used.
6107 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
6108 and nonzero, implicitly enables @option{-fcompare-debug}. If
6109 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
6110 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
6113 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
6114 is equivalent to @option{-fno-compare-debug}, which disables the dumping
6115 of the final representation and the second compilation, preventing even
6116 @env{GCC_COMPARE_DEBUG} from taking effect.
6118 To verify full coverage during @option{-fcompare-debug} testing, set
6119 @env{GCC_COMPARE_DEBUG} to say @option{-fcompare-debug-not-overridden},
6120 which GCC rejects as an invalid option in any actual compilation
6121 (rather than preprocessing, assembly or linking). To get just a
6122 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
6123 not overridden} will do.
6125 @item -fcompare-debug-second
6126 @opindex fcompare-debug-second
6127 This option is implicitly passed to the compiler for the second
6128 compilation requested by @option{-fcompare-debug}, along with options to
6129 silence warnings, and omitting other options that would cause
6130 side-effect compiler outputs to files or to the standard output. Dump
6131 files and preserved temporary files are renamed so as to contain the
6132 @code{.gk} additional extension during the second compilation, to avoid
6133 overwriting those generated by the first.
6135 When this option is passed to the compiler driver, it causes the
6136 @emph{first} compilation to be skipped, which makes it useful for little
6137 other than debugging the compiler proper.
6139 @item -feliminate-dwarf2-dups
6140 @opindex feliminate-dwarf2-dups
6141 Compress DWARF 2 debugging information by eliminating duplicated
6142 information about each symbol. This option only makes sense when
6143 generating DWARF 2 debugging information with @option{-gdwarf-2}.
6145 @item -femit-struct-debug-baseonly
6146 @opindex femit-struct-debug-baseonly
6147 Emit debug information for struct-like types
6148 only when the base name of the compilation source file
6149 matches the base name of file in which the struct is defined.
6151 This option substantially reduces the size of debugging information,
6152 but at significant potential loss in type information to the debugger.
6153 See @option{-femit-struct-debug-reduced} for a less aggressive option.
6154 See @option{-femit-struct-debug-detailed} for more detailed control.
6156 This option works only with DWARF 2.
6158 @item -femit-struct-debug-reduced
6159 @opindex femit-struct-debug-reduced
6160 Emit debug information for struct-like types
6161 only when the base name of the compilation source file
6162 matches the base name of file in which the type is defined,
6163 unless the struct is a template or defined in a system header.
6165 This option significantly reduces the size of debugging information,
6166 with some potential loss in type information to the debugger.
6167 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
6168 See @option{-femit-struct-debug-detailed} for more detailed control.
6170 This option works only with DWARF 2.
6172 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
6173 @opindex femit-struct-debug-detailed
6174 Specify the struct-like types
6175 for which the compiler generates debug information.
6176 The intent is to reduce duplicate struct debug information
6177 between different object files within the same program.
6179 This option is a detailed version of
6180 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
6181 which serves for most needs.
6183 A specification has the syntax@*
6184 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
6186 The optional first word limits the specification to
6187 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
6188 A struct type is used directly when it is the type of a variable, member.
6189 Indirect uses arise through pointers to structs.
6190 That is, when use of an incomplete struct is valid, the use is indirect.
6192 @samp{struct one direct; struct two * indirect;}.
6194 The optional second word limits the specification to
6195 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
6196 Generic structs are a bit complicated to explain.
6197 For C++, these are non-explicit specializations of template classes,
6198 or non-template classes within the above.
6199 Other programming languages have generics,
6200 but @option{-femit-struct-debug-detailed} does not yet implement them.
6202 The third word specifies the source files for those
6203 structs for which the compiler should emit debug information.
6204 The values @samp{none} and @samp{any} have the normal meaning.
6205 The value @samp{base} means that
6206 the base of name of the file in which the type declaration appears
6207 must match the base of the name of the main compilation file.
6208 In practice, this means that when compiling @file{foo.c}, debug information
6209 is generated for types declared in that file and @file{foo.h},
6210 but not other header files.
6211 The value @samp{sys} means those types satisfying @samp{base}
6212 or declared in system or compiler headers.
6214 You may need to experiment to determine the best settings for your application.
6216 The default is @option{-femit-struct-debug-detailed=all}.
6218 This option works only with DWARF 2.
6220 @item -fno-merge-debug-strings
6221 @opindex fmerge-debug-strings
6222 @opindex fno-merge-debug-strings
6223 Direct the linker to not merge together strings in the debugging
6224 information that are identical in different object files. Merging is
6225 not supported by all assemblers or linkers. Merging decreases the size
6226 of the debug information in the output file at the cost of increasing
6227 link processing time. Merging is enabled by default.
6229 @item -fdebug-prefix-map=@var{old}=@var{new}
6230 @opindex fdebug-prefix-map
6231 When compiling files in directory @file{@var{old}}, record debugging
6232 information describing them as in @file{@var{new}} instead.
6234 @item -fno-dwarf2-cfi-asm
6235 @opindex fdwarf2-cfi-asm
6236 @opindex fno-dwarf2-cfi-asm
6237 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
6238 instead of using GAS @code{.cfi_*} directives.
6240 @cindex @command{prof}
6243 Generate extra code to write profile information suitable for the
6244 analysis program @command{prof}. You must use this option when compiling
6245 the source files you want data about, and you must also use it when
6248 @cindex @command{gprof}
6251 Generate extra code to write profile information suitable for the
6252 analysis program @command{gprof}. You must use this option when compiling
6253 the source files you want data about, and you must also use it when
6258 Makes the compiler print out each function name as it is compiled, and
6259 print some statistics about each pass when it finishes.
6262 @opindex ftime-report
6263 Makes the compiler print some statistics about the time consumed by each
6264 pass when it finishes.
6267 @opindex fmem-report
6268 Makes the compiler print some statistics about permanent memory
6269 allocation when it finishes.
6271 @item -fmem-report-wpa
6272 @opindex fmem-report-wpa
6273 Makes the compiler print some statistics about permanent memory
6274 allocation for the WPA phase only.
6276 @item -fpre-ipa-mem-report
6277 @opindex fpre-ipa-mem-report
6278 @item -fpost-ipa-mem-report
6279 @opindex fpost-ipa-mem-report
6280 Makes the compiler print some statistics about permanent memory
6281 allocation before or after interprocedural optimization.
6283 @item -fprofile-report
6284 @opindex fprofile-report
6285 Makes the compiler print some statistics about consistency of the
6286 (estimated) profile and effect of individual passes.
6289 @opindex fstack-usage
6290 Makes the compiler output stack usage information for the program, on a
6291 per-function basis. The filename for the dump is made by appending
6292 @file{.su} to the @var{auxname}. @var{auxname} is generated from the name of
6293 the output file, if explicitly specified and it is not an executable,
6294 otherwise it is the basename of the source file. An entry is made up
6299 The name of the function.
6303 One or more qualifiers: @code{static}, @code{dynamic}, @code{bounded}.
6306 The qualifier @code{static} means that the function manipulates the stack
6307 statically: a fixed number of bytes are allocated for the frame on function
6308 entry and released on function exit; no stack adjustments are otherwise made
6309 in the function. The second field is this fixed number of bytes.
6311 The qualifier @code{dynamic} means that the function manipulates the stack
6312 dynamically: in addition to the static allocation described above, stack
6313 adjustments are made in the body of the function, for example to push/pop
6314 arguments around function calls. If the qualifier @code{bounded} is also
6315 present, the amount of these adjustments is bounded at compile time and
6316 the second field is an upper bound of the total amount of stack used by
6317 the function. If it is not present, the amount of these adjustments is
6318 not bounded at compile time and the second field only represents the
6321 @item -fprofile-arcs
6322 @opindex fprofile-arcs
6323 Add code so that program flow @dfn{arcs} are instrumented. During
6324 execution the program records how many times each branch and call is
6325 executed and how many times it is taken or returns. When the compiled
6326 program exits it saves this data to a file called
6327 @file{@var{auxname}.gcda} for each source file. The data may be used for
6328 profile-directed optimizations (@option{-fbranch-probabilities}), or for
6329 test coverage analysis (@option{-ftest-coverage}). Each object file's
6330 @var{auxname} is generated from the name of the output file, if
6331 explicitly specified and it is not the final executable, otherwise it is
6332 the basename of the source file. In both cases any suffix is removed
6333 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
6334 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
6335 @xref{Cross-profiling}.
6337 @cindex @command{gcov}
6341 This option is used to compile and link code instrumented for coverage
6342 analysis. The option is a synonym for @option{-fprofile-arcs}
6343 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
6344 linking). See the documentation for those options for more details.
6349 Compile the source files with @option{-fprofile-arcs} plus optimization
6350 and code generation options. For test coverage analysis, use the
6351 additional @option{-ftest-coverage} option. You do not need to profile
6352 every source file in a program.
6355 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
6356 (the latter implies the former).
6359 Run the program on a representative workload to generate the arc profile
6360 information. This may be repeated any number of times. You can run
6361 concurrent instances of your program, and provided that the file system
6362 supports locking, the data files will be correctly updated. Also
6363 @code{fork} calls are detected and correctly handled (double counting
6367 For profile-directed optimizations, compile the source files again with
6368 the same optimization and code generation options plus
6369 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
6370 Control Optimization}).
6373 For test coverage analysis, use @command{gcov} to produce human readable
6374 information from the @file{.gcno} and @file{.gcda} files. Refer to the
6375 @command{gcov} documentation for further information.
6379 With @option{-fprofile-arcs}, for each function of your program GCC
6380 creates a program flow graph, then finds a spanning tree for the graph.
6381 Only arcs that are not on the spanning tree have to be instrumented: the
6382 compiler adds code to count the number of times that these arcs are
6383 executed. When an arc is the only exit or only entrance to a block, the
6384 instrumentation code can be added to the block; otherwise, a new basic
6385 block must be created to hold the instrumentation code.
6388 @item -ftest-coverage
6389 @opindex ftest-coverage
6390 Produce a notes file that the @command{gcov} code-coverage utility
6391 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
6392 show program coverage. Each source file's note file is called
6393 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
6394 above for a description of @var{auxname} and instructions on how to
6395 generate test coverage data. Coverage data matches the source files
6396 more closely if you do not optimize.
6398 @item -fdbg-cnt-list
6399 @opindex fdbg-cnt-list
6400 Print the name and the counter upper bound for all debug counters.
6403 @item -fdbg-cnt=@var{counter-value-list}
6405 Set the internal debug counter upper bound. @var{counter-value-list}
6406 is a comma-separated list of @var{name}:@var{value} pairs
6407 which sets the upper bound of each debug counter @var{name} to @var{value}.
6408 All debug counters have the initial upper bound of @code{UINT_MAX};
6409 thus @code{dbg_cnt} returns true always unless the upper bound
6410 is set by this option.
6411 For example, with @option{-fdbg-cnt=dce:10,tail_call:0},
6412 @code{dbg_cnt(dce)} returns true only for first 10 invocations.
6414 @item -fenable-@var{kind}-@var{pass}
6415 @itemx -fdisable-@var{kind}-@var{pass}=@var{range-list}
6419 This is a set of options that are used to explicitly disable/enable
6420 optimization passes. These options are intended for use for debugging GCC.
6421 Compiler users should use regular options for enabling/disabling
6426 @item -fdisable-ipa-@var{pass}
6427 Disable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
6428 statically invoked in the compiler multiple times, the pass name should be
6429 appended with a sequential number starting from 1.
6431 @item -fdisable-rtl-@var{pass}
6432 @itemx -fdisable-rtl-@var{pass}=@var{range-list}
6433 Disable RTL pass @var{pass}. @var{pass} is the pass name. If the same pass is
6434 statically invoked in the compiler multiple times, the pass name should be
6435 appended with a sequential number starting from 1. @var{range-list} is a
6436 comma-separated list of function ranges or assembler names. Each range is a number
6437 pair separated by a colon. The range is inclusive in both ends. If the range
6438 is trivial, the number pair can be simplified as a single number. If the
6439 function's call graph node's @var{uid} falls within one of the specified ranges,
6440 the @var{pass} is disabled for that function. The @var{uid} is shown in the
6441 function header of a dump file, and the pass names can be dumped by using
6442 option @option{-fdump-passes}.
6444 @item -fdisable-tree-@var{pass}
6445 @itemx -fdisable-tree-@var{pass}=@var{range-list}
6446 Disable tree pass @var{pass}. See @option{-fdisable-rtl} for the description of
6449 @item -fenable-ipa-@var{pass}
6450 Enable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
6451 statically invoked in the compiler multiple times, the pass name should be
6452 appended with a sequential number starting from 1.
6454 @item -fenable-rtl-@var{pass}
6455 @itemx -fenable-rtl-@var{pass}=@var{range-list}
6456 Enable RTL pass @var{pass}. See @option{-fdisable-rtl} for option argument
6457 description and examples.
6459 @item -fenable-tree-@var{pass}
6460 @itemx -fenable-tree-@var{pass}=@var{range-list}
6461 Enable tree pass @var{pass}. See @option{-fdisable-rtl} for the description
6462 of option arguments.
6466 Here are some examples showing uses of these options.
6470 # disable ccp1 for all functions
6472 # disable complete unroll for function whose cgraph node uid is 1
6473 -fenable-tree-cunroll=1
6474 # disable gcse2 for functions at the following ranges [1,1],
6475 # [300,400], and [400,1000]
6476 # disable gcse2 for functions foo and foo2
6477 -fdisable-rtl-gcse2=foo,foo2
6478 # disable early inlining
6479 -fdisable-tree-einline
6480 # disable ipa inlining
6481 -fdisable-ipa-inline
6482 # enable tree full unroll
6483 -fenable-tree-unroll
6487 @item -d@var{letters}
6488 @itemx -fdump-rtl-@var{pass}
6489 @itemx -fdump-rtl-@var{pass}=@var{filename}
6491 @opindex fdump-rtl-@var{pass}
6492 Says to make debugging dumps during compilation at times specified by
6493 @var{letters}. This is used for debugging the RTL-based passes of the
6494 compiler. The file names for most of the dumps are made by appending
6495 a pass number and a word to the @var{dumpname}, and the files are
6496 created in the directory of the output file. In case of
6497 @option{=@var{filename}} option, the dump is output on the given file
6498 instead of the pass numbered dump files. Note that the pass number is
6499 computed statically as passes get registered into the pass manager.
6500 Thus the numbering is not related to the dynamic order of execution of
6501 passes. In particular, a pass installed by a plugin could have a
6502 number over 200 even if it executed quite early. @var{dumpname} is
6503 generated from the name of the output file, if explicitly specified
6504 and it is not an executable, otherwise it is the basename of the
6505 source file. These switches may have different effects when
6506 @option{-E} is used for preprocessing.
6508 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
6509 @option{-d} option @var{letters}. Here are the possible
6510 letters for use in @var{pass} and @var{letters}, and their meanings:
6514 @item -fdump-rtl-alignments
6515 @opindex fdump-rtl-alignments
6516 Dump after branch alignments have been computed.
6518 @item -fdump-rtl-asmcons
6519 @opindex fdump-rtl-asmcons
6520 Dump after fixing rtl statements that have unsatisfied in/out constraints.
6522 @item -fdump-rtl-auto_inc_dec
6523 @opindex fdump-rtl-auto_inc_dec
6524 Dump after auto-inc-dec discovery. This pass is only run on
6525 architectures that have auto inc or auto dec instructions.
6527 @item -fdump-rtl-barriers
6528 @opindex fdump-rtl-barriers
6529 Dump after cleaning up the barrier instructions.
6531 @item -fdump-rtl-bbpart
6532 @opindex fdump-rtl-bbpart
6533 Dump after partitioning hot and cold basic blocks.
6535 @item -fdump-rtl-bbro
6536 @opindex fdump-rtl-bbro
6537 Dump after block reordering.
6539 @item -fdump-rtl-btl1
6540 @itemx -fdump-rtl-btl2
6541 @opindex fdump-rtl-btl2
6542 @opindex fdump-rtl-btl2
6543 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
6544 after the two branch
6545 target load optimization passes.
6547 @item -fdump-rtl-bypass
6548 @opindex fdump-rtl-bypass
6549 Dump after jump bypassing and control flow optimizations.
6551 @item -fdump-rtl-combine
6552 @opindex fdump-rtl-combine
6553 Dump after the RTL instruction combination pass.
6555 @item -fdump-rtl-compgotos
6556 @opindex fdump-rtl-compgotos
6557 Dump after duplicating the computed gotos.
6559 @item -fdump-rtl-ce1
6560 @itemx -fdump-rtl-ce2
6561 @itemx -fdump-rtl-ce3
6562 @opindex fdump-rtl-ce1
6563 @opindex fdump-rtl-ce2
6564 @opindex fdump-rtl-ce3
6565 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
6566 @option{-fdump-rtl-ce3} enable dumping after the three
6567 if conversion passes.
6569 @item -fdump-rtl-cprop_hardreg
6570 @opindex fdump-rtl-cprop_hardreg
6571 Dump after hard register copy propagation.
6573 @item -fdump-rtl-csa
6574 @opindex fdump-rtl-csa
6575 Dump after combining stack adjustments.
6577 @item -fdump-rtl-cse1
6578 @itemx -fdump-rtl-cse2
6579 @opindex fdump-rtl-cse1
6580 @opindex fdump-rtl-cse2
6581 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
6582 the two common subexpression elimination passes.
6584 @item -fdump-rtl-dce
6585 @opindex fdump-rtl-dce
6586 Dump after the standalone dead code elimination passes.
6588 @item -fdump-rtl-dbr
6589 @opindex fdump-rtl-dbr
6590 Dump after delayed branch scheduling.
6592 @item -fdump-rtl-dce1
6593 @itemx -fdump-rtl-dce2
6594 @opindex fdump-rtl-dce1
6595 @opindex fdump-rtl-dce2
6596 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
6597 the two dead store elimination passes.
6600 @opindex fdump-rtl-eh
6601 Dump after finalization of EH handling code.
6603 @item -fdump-rtl-eh_ranges
6604 @opindex fdump-rtl-eh_ranges
6605 Dump after conversion of EH handling range regions.
6607 @item -fdump-rtl-expand
6608 @opindex fdump-rtl-expand
6609 Dump after RTL generation.
6611 @item -fdump-rtl-fwprop1
6612 @itemx -fdump-rtl-fwprop2
6613 @opindex fdump-rtl-fwprop1
6614 @opindex fdump-rtl-fwprop2
6615 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
6616 dumping after the two forward propagation passes.
6618 @item -fdump-rtl-gcse1
6619 @itemx -fdump-rtl-gcse2
6620 @opindex fdump-rtl-gcse1
6621 @opindex fdump-rtl-gcse2
6622 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
6623 after global common subexpression elimination.
6625 @item -fdump-rtl-init-regs
6626 @opindex fdump-rtl-init-regs
6627 Dump after the initialization of the registers.
6629 @item -fdump-rtl-initvals
6630 @opindex fdump-rtl-initvals
6631 Dump after the computation of the initial value sets.
6633 @item -fdump-rtl-into_cfglayout
6634 @opindex fdump-rtl-into_cfglayout
6635 Dump after converting to cfglayout mode.
6637 @item -fdump-rtl-ira
6638 @opindex fdump-rtl-ira
6639 Dump after iterated register allocation.
6641 @item -fdump-rtl-jump
6642 @opindex fdump-rtl-jump
6643 Dump after the second jump optimization.
6645 @item -fdump-rtl-loop2
6646 @opindex fdump-rtl-loop2
6647 @option{-fdump-rtl-loop2} enables dumping after the rtl
6648 loop optimization passes.
6650 @item -fdump-rtl-mach
6651 @opindex fdump-rtl-mach
6652 Dump after performing the machine dependent reorganization pass, if that
6655 @item -fdump-rtl-mode_sw
6656 @opindex fdump-rtl-mode_sw
6657 Dump after removing redundant mode switches.
6659 @item -fdump-rtl-rnreg
6660 @opindex fdump-rtl-rnreg
6661 Dump after register renumbering.
6663 @item -fdump-rtl-outof_cfglayout
6664 @opindex fdump-rtl-outof_cfglayout
6665 Dump after converting from cfglayout mode.
6667 @item -fdump-rtl-peephole2
6668 @opindex fdump-rtl-peephole2
6669 Dump after the peephole pass.
6671 @item -fdump-rtl-postreload
6672 @opindex fdump-rtl-postreload
6673 Dump after post-reload optimizations.
6675 @item -fdump-rtl-pro_and_epilogue
6676 @opindex fdump-rtl-pro_and_epilogue
6677 Dump after generating the function prologues and epilogues.
6679 @item -fdump-rtl-sched1
6680 @itemx -fdump-rtl-sched2
6681 @opindex fdump-rtl-sched1
6682 @opindex fdump-rtl-sched2
6683 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
6684 after the basic block scheduling passes.
6686 @item -fdump-rtl-ree
6687 @opindex fdump-rtl-ree
6688 Dump after sign/zero extension elimination.
6690 @item -fdump-rtl-seqabstr
6691 @opindex fdump-rtl-seqabstr
6692 Dump after common sequence discovery.
6694 @item -fdump-rtl-shorten
6695 @opindex fdump-rtl-shorten
6696 Dump after shortening branches.
6698 @item -fdump-rtl-sibling
6699 @opindex fdump-rtl-sibling
6700 Dump after sibling call optimizations.
6702 @item -fdump-rtl-split1
6703 @itemx -fdump-rtl-split2
6704 @itemx -fdump-rtl-split3
6705 @itemx -fdump-rtl-split4
6706 @itemx -fdump-rtl-split5
6707 @opindex fdump-rtl-split1
6708 @opindex fdump-rtl-split2
6709 @opindex fdump-rtl-split3
6710 @opindex fdump-rtl-split4
6711 @opindex fdump-rtl-split5
6712 These options enable dumping after five rounds of
6713 instruction splitting.
6715 @item -fdump-rtl-sms
6716 @opindex fdump-rtl-sms
6717 Dump after modulo scheduling. This pass is only run on some
6720 @item -fdump-rtl-stack
6721 @opindex fdump-rtl-stack
6722 Dump after conversion from GCC's ``flat register file'' registers to the
6723 x87's stack-like registers. This pass is only run on x86 variants.
6725 @item -fdump-rtl-subreg1
6726 @itemx -fdump-rtl-subreg2
6727 @opindex fdump-rtl-subreg1
6728 @opindex fdump-rtl-subreg2
6729 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
6730 the two subreg expansion passes.
6732 @item -fdump-rtl-unshare
6733 @opindex fdump-rtl-unshare
6734 Dump after all rtl has been unshared.
6736 @item -fdump-rtl-vartrack
6737 @opindex fdump-rtl-vartrack
6738 Dump after variable tracking.
6740 @item -fdump-rtl-vregs
6741 @opindex fdump-rtl-vregs
6742 Dump after converting virtual registers to hard registers.
6744 @item -fdump-rtl-web
6745 @opindex fdump-rtl-web
6746 Dump after live range splitting.
6748 @item -fdump-rtl-regclass
6749 @itemx -fdump-rtl-subregs_of_mode_init
6750 @itemx -fdump-rtl-subregs_of_mode_finish
6751 @itemx -fdump-rtl-dfinit
6752 @itemx -fdump-rtl-dfinish
6753 @opindex fdump-rtl-regclass
6754 @opindex fdump-rtl-subregs_of_mode_init
6755 @opindex fdump-rtl-subregs_of_mode_finish
6756 @opindex fdump-rtl-dfinit
6757 @opindex fdump-rtl-dfinish
6758 These dumps are defined but always produce empty files.
6761 @itemx -fdump-rtl-all
6763 @opindex fdump-rtl-all
6764 Produce all the dumps listed above.
6768 Annotate the assembler output with miscellaneous debugging information.
6772 Dump all macro definitions, at the end of preprocessing, in addition to
6777 Produce a core dump whenever an error occurs.
6781 Annotate the assembler output with a comment indicating which
6782 pattern and alternative is used. The length of each instruction is
6787 Dump the RTL in the assembler output as a comment before each instruction.
6788 Also turns on @option{-dp} annotation.
6792 Just generate RTL for a function instead of compiling it. Usually used
6793 with @option{-fdump-rtl-expand}.
6797 @opindex fdump-noaddr
6798 When doing debugging dumps, suppress address output. This makes it more
6799 feasible to use diff on debugging dumps for compiler invocations with
6800 different compiler binaries and/or different
6801 text / bss / data / heap / stack / dso start locations.
6804 @opindex freport-bug
6805 Collect and dump debug information into temporary file if ICE in C/C++
6808 @item -fdump-unnumbered
6809 @opindex fdump-unnumbered
6810 When doing debugging dumps, suppress instruction numbers and address output.
6811 This makes it more feasible to use diff on debugging dumps for compiler
6812 invocations with different options, in particular with and without
6815 @item -fdump-unnumbered-links
6816 @opindex fdump-unnumbered-links
6817 When doing debugging dumps (see @option{-d} option above), suppress
6818 instruction numbers for the links to the previous and next instructions
6821 @item -fdump-translation-unit @r{(C++ only)}
6822 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
6823 @opindex fdump-translation-unit
6824 Dump a representation of the tree structure for the entire translation
6825 unit to a file. The file name is made by appending @file{.tu} to the
6826 source file name, and the file is created in the same directory as the
6827 output file. If the @samp{-@var{options}} form is used, @var{options}
6828 controls the details of the dump as described for the
6829 @option{-fdump-tree} options.
6831 @item -fdump-class-hierarchy @r{(C++ only)}
6832 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
6833 @opindex fdump-class-hierarchy
6834 Dump a representation of each class's hierarchy and virtual function
6835 table layout to a file. The file name is made by appending
6836 @file{.class} to the source file name, and the file is created in the
6837 same directory as the output file. If the @samp{-@var{options}} form
6838 is used, @var{options} controls the details of the dump as described
6839 for the @option{-fdump-tree} options.
6841 @item -fdump-ipa-@var{switch}
6843 Control the dumping at various stages of inter-procedural analysis
6844 language tree to a file. The file name is generated by appending a
6845 switch specific suffix to the source file name, and the file is created
6846 in the same directory as the output file. The following dumps are
6851 Enables all inter-procedural analysis dumps.
6854 Dumps information about call-graph optimization, unused function removal,
6855 and inlining decisions.
6858 Dump after function inlining.
6863 @opindex fdump-passes
6864 Dump the list of optimization passes that are turned on and off by
6865 the current command-line options.
6867 @item -fdump-statistics-@var{option}
6868 @opindex fdump-statistics
6869 Enable and control dumping of pass statistics in a separate file. The
6870 file name is generated by appending a suffix ending in
6871 @samp{.statistics} to the source file name, and the file is created in
6872 the same directory as the output file. If the @samp{-@var{option}}
6873 form is used, @samp{-stats} causes counters to be summed over the
6874 whole compilation unit while @samp{-details} dumps every event as
6875 the passes generate them. The default with no option is to sum
6876 counters for each function compiled.
6878 @item -fdump-tree-@var{switch}
6879 @itemx -fdump-tree-@var{switch}-@var{options}
6880 @itemx -fdump-tree-@var{switch}-@var{options}=@var{filename}
6882 Control the dumping at various stages of processing the intermediate
6883 language tree to a file. The file name is generated by appending a
6884 switch-specific suffix to the source file name, and the file is
6885 created in the same directory as the output file. In case of
6886 @option{=@var{filename}} option, the dump is output on the given file
6887 instead of the auto named dump files. If the @samp{-@var{options}}
6888 form is used, @var{options} is a list of @samp{-} separated options
6889 which control the details of the dump. Not all options are applicable
6890 to all dumps; those that are not meaningful are ignored. The
6891 following options are available
6895 Print the address of each node. Usually this is not meaningful as it
6896 changes according to the environment and source file. Its primary use
6897 is for tying up a dump file with a debug environment.
6899 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
6900 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
6901 use working backward from mangled names in the assembly file.
6903 When dumping front-end intermediate representations, inhibit dumping
6904 of members of a scope or body of a function merely because that scope
6905 has been reached. Only dump such items when they are directly reachable
6908 When dumping pretty-printed trees, this option inhibits dumping the
6909 bodies of control structures.
6911 When dumping RTL, print the RTL in slim (condensed) form instead of
6912 the default LISP-like representation.
6914 Print a raw representation of the tree. By default, trees are
6915 pretty-printed into a C-like representation.
6917 Enable more detailed dumps (not honored by every dump option). Also
6918 include information from the optimization passes.
6920 Enable dumping various statistics about the pass (not honored by every dump
6923 Enable showing basic block boundaries (disabled in raw dumps).
6925 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
6926 dump a representation of the control flow graph suitable for viewing with
6927 GraphViz to @file{@var{file}.@var{passid}.@var{pass}.dot}. Each function in
6928 the file is pretty-printed as a subgraph, so that GraphViz can render them
6929 all in a single plot.
6931 This option currently only works for RTL dumps, and the RTL is always
6932 dumped in slim form.
6934 Enable showing virtual operands for every statement.
6936 Enable showing line numbers for statements.
6938 Enable showing the unique ID (@code{DECL_UID}) for each variable.
6940 Enable showing the tree dump for each statement.
6942 Enable showing the EH region number holding each statement.
6944 Enable showing scalar evolution analysis details.
6946 Enable showing optimization information (only available in certain
6949 Enable showing missed optimization information (only available in certain
6952 Enable other detailed optimization information (only available in
6954 @item =@var{filename}
6955 Instead of an auto named dump file, output into the given file
6956 name. The file names @file{stdout} and @file{stderr} are treated
6957 specially and are considered already open standard streams. For
6961 gcc -O2 -ftree-vectorize -fdump-tree-vect-blocks=foo.dump
6962 -fdump-tree-pre=stderr file.c
6965 outputs vectorizer dump into @file{foo.dump}, while the PRE dump is
6966 output on to @file{stderr}. If two conflicting dump filenames are
6967 given for the same pass, then the latter option overrides the earlier
6971 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
6972 and @option{lineno}.
6975 Turn on all optimization options, i.e., @option{optimized},
6976 @option{missed}, and @option{note}.
6979 The following tree dumps are possible:
6983 @opindex fdump-tree-original
6984 Dump before any tree based optimization, to @file{@var{file}.original}.
6987 @opindex fdump-tree-optimized
6988 Dump after all tree based optimization, to @file{@var{file}.optimized}.
6991 @opindex fdump-tree-gimple
6992 Dump each function before and after the gimplification pass to a file. The
6993 file name is made by appending @file{.gimple} to the source file name.
6996 @opindex fdump-tree-cfg
6997 Dump the control flow graph of each function to a file. The file name is
6998 made by appending @file{.cfg} to the source file name.
7001 @opindex fdump-tree-ch
7002 Dump each function after copying loop headers. The file name is made by
7003 appending @file{.ch} to the source file name.
7006 @opindex fdump-tree-ssa
7007 Dump SSA related information to a file. The file name is made by appending
7008 @file{.ssa} to the source file name.
7011 @opindex fdump-tree-alias
7012 Dump aliasing information for each function. The file name is made by
7013 appending @file{.alias} to the source file name.
7016 @opindex fdump-tree-ccp
7017 Dump each function after CCP@. The file name is made by appending
7018 @file{.ccp} to the source file name.
7021 @opindex fdump-tree-storeccp
7022 Dump each function after STORE-CCP@. The file name is made by appending
7023 @file{.storeccp} to the source file name.
7026 @opindex fdump-tree-pre
7027 Dump trees after partial redundancy elimination. The file name is made
7028 by appending @file{.pre} to the source file name.
7031 @opindex fdump-tree-fre
7032 Dump trees after full redundancy elimination. The file name is made
7033 by appending @file{.fre} to the source file name.
7036 @opindex fdump-tree-copyprop
7037 Dump trees after copy propagation. The file name is made
7038 by appending @file{.copyprop} to the source file name.
7040 @item store_copyprop
7041 @opindex fdump-tree-store_copyprop
7042 Dump trees after store copy-propagation. The file name is made
7043 by appending @file{.store_copyprop} to the source file name.
7046 @opindex fdump-tree-dce
7047 Dump each function after dead code elimination. The file name is made by
7048 appending @file{.dce} to the source file name.
7051 @opindex fdump-tree-sra
7052 Dump each function after performing scalar replacement of aggregates. The
7053 file name is made by appending @file{.sra} to the source file name.
7056 @opindex fdump-tree-sink
7057 Dump each function after performing code sinking. The file name is made
7058 by appending @file{.sink} to the source file name.
7061 @opindex fdump-tree-dom
7062 Dump each function after applying dominator tree optimizations. The file
7063 name is made by appending @file{.dom} to the source file name.
7066 @opindex fdump-tree-dse
7067 Dump each function after applying dead store elimination. The file
7068 name is made by appending @file{.dse} to the source file name.
7071 @opindex fdump-tree-phiopt
7072 Dump each function after optimizing PHI nodes into straightline code. The file
7073 name is made by appending @file{.phiopt} to the source file name.
7076 @opindex fdump-tree-forwprop
7077 Dump each function after forward propagating single use variables. The file
7078 name is made by appending @file{.forwprop} to the source file name.
7081 @opindex fdump-tree-copyrename
7082 Dump each function after applying the copy rename optimization. The file
7083 name is made by appending @file{.copyrename} to the source file name.
7086 @opindex fdump-tree-nrv
7087 Dump each function after applying the named return value optimization on
7088 generic trees. The file name is made by appending @file{.nrv} to the source
7092 @opindex fdump-tree-vect
7093 Dump each function after applying vectorization of loops. The file name is
7094 made by appending @file{.vect} to the source file name.
7097 @opindex fdump-tree-slp
7098 Dump each function after applying vectorization of basic blocks. The file name
7099 is made by appending @file{.slp} to the source file name.
7102 @opindex fdump-tree-vrp
7103 Dump each function after Value Range Propagation (VRP). The file name
7104 is made by appending @file{.vrp} to the source file name.
7107 @opindex fdump-tree-all
7108 Enable all the available tree dumps with the flags provided in this option.
7112 @itemx -fopt-info-@var{options}
7113 @itemx -fopt-info-@var{options}=@var{filename}
7115 Controls optimization dumps from various optimization passes. If the
7116 @samp{-@var{options}} form is used, @var{options} is a list of
7117 @samp{-} separated option keywords to select the dump details and
7120 The @var{options} can be divided into two groups: options describing the
7121 verbosity of the dump, and options describing which optimizations
7122 should be included. The options from both the groups can be freely
7123 mixed as they are non-overlapping. However, in case of any conflicts,
7124 the later options override the earlier options on the command
7127 The following options control the dump verbosity:
7131 Print information when an optimization is successfully applied. It is
7132 up to a pass to decide which information is relevant. For example, the
7133 vectorizer passes print the source location of loops which are
7134 successfully vectorized.
7136 Print information about missed optimizations. Individual passes
7137 control which information to include in the output.
7139 Print verbose information about optimizations, such as certain
7140 transformations, more detailed messages about decisions etc.
7142 Print detailed optimization information. This includes
7143 @samp{optimized}, @samp{missed}, and @samp{note}.
7146 One or more of the following option keywords can be used to describe a
7147 group of optimizations:
7151 Enable dumps from all interprocedural optimizations.
7153 Enable dumps from all loop optimizations.
7155 Enable dumps from all inlining optimizations.
7157 Enable dumps from all vectorization optimizations.
7159 Enable dumps from all optimizations. This is a superset of
7160 the optimization groups listed above.
7164 omitted, it defaults to @samp{optimized-optall}, which means to dump all
7165 info about successful optimizations from all the passes.
7167 If the @var{filename} is provided, then the dumps from all the
7168 applicable optimizations are concatenated into the @var{filename}.
7169 Otherwise the dump is output onto @file{stderr}. Though multiple
7170 @option{-fopt-info} options are accepted, only one of them can include
7171 a @var{filename}. If other filenames are provided then all but the
7172 first such option are ignored.
7174 Note that the output @var{filename} is overwritten
7175 in case of multiple translation units. If a combined output from
7176 multiple translation units is desired, @file{stderr} should be used
7179 In the following example, the optimization info is output to
7188 gcc -O3 -fopt-info-missed=missed.all
7192 outputs missed optimization report from all the passes into
7193 @file{missed.all}, and this one:
7196 gcc -O2 -ftree-vectorize -fopt-info-vec-missed
7200 prints information about missed optimization opportunities from
7201 vectorization passes on @file{stderr}.
7202 Note that @option{-fopt-info-vec-missed} is equivalent to
7203 @option{-fopt-info-missed-vec}.
7207 gcc -O3 -fopt-info-inline-optimized-missed=inline.txt
7211 outputs information about missed optimizations as well as
7212 optimized locations from all the inlining passes into
7218 gcc -fopt-info-vec-missed=vec.miss -fopt-info-loop-optimized=loop.opt
7222 Here the two output filenames @file{vec.miss} and @file{loop.opt} are
7223 in conflict since only one output file is allowed. In this case, only
7224 the first option takes effect and the subsequent options are
7225 ignored. Thus only @file{vec.miss} is produced which contains
7226 dumps from the vectorizer about missed opportunities.
7228 @item -frandom-seed=@var{number}
7229 @opindex frandom-seed
7230 This option provides a seed that GCC uses in place of
7231 random numbers in generating certain symbol names
7232 that have to be different in every compiled file. It is also used to
7233 place unique stamps in coverage data files and the object files that
7234 produce them. You can use the @option{-frandom-seed} option to produce
7235 reproducibly identical object files.
7237 The @var{number} should be different for every file you compile.
7239 @item -fsched-verbose=@var{n}
7240 @opindex fsched-verbose
7241 On targets that use instruction scheduling, this option controls the
7242 amount of debugging output the scheduler prints. This information is
7243 written to standard error, unless @option{-fdump-rtl-sched1} or
7244 @option{-fdump-rtl-sched2} is specified, in which case it is output
7245 to the usual dump listing file, @file{.sched1} or @file{.sched2}
7246 respectively. However for @var{n} greater than nine, the output is
7247 always printed to standard error.
7249 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
7250 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
7251 For @var{n} greater than one, it also output basic block probabilities,
7252 detailed ready list information and unit/insn info. For @var{n} greater
7253 than two, it includes RTL at abort point, control-flow and regions info.
7254 And for @var{n} over four, @option{-fsched-verbose} also includes
7258 @itemx -save-temps=cwd
7260 Store the usual ``temporary'' intermediate files permanently; place them
7261 in the current directory and name them based on the source file. Thus,
7262 compiling @file{foo.c} with @option{-c -save-temps} produces files
7263 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
7264 preprocessed @file{foo.i} output file even though the compiler now
7265 normally uses an integrated preprocessor.
7267 When used in combination with the @option{-x} command-line option,
7268 @option{-save-temps} is sensible enough to avoid over writing an
7269 input source file with the same extension as an intermediate file.
7270 The corresponding intermediate file may be obtained by renaming the
7271 source file before using @option{-save-temps}.
7273 If you invoke GCC in parallel, compiling several different source
7274 files that share a common base name in different subdirectories or the
7275 same source file compiled for multiple output destinations, it is
7276 likely that the different parallel compilers will interfere with each
7277 other, and overwrite the temporary files. For instance:
7280 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
7281 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
7284 may result in @file{foo.i} and @file{foo.o} being written to
7285 simultaneously by both compilers.
7287 @item -save-temps=obj
7288 @opindex save-temps=obj
7289 Store the usual ``temporary'' intermediate files permanently. If the
7290 @option{-o} option is used, the temporary files are based on the
7291 object file. If the @option{-o} option is not used, the
7292 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
7297 gcc -save-temps=obj -c foo.c
7298 gcc -save-temps=obj -c bar.c -o dir/xbar.o
7299 gcc -save-temps=obj foobar.c -o dir2/yfoobar
7303 creates @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
7304 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
7305 @file{dir2/yfoobar.o}.
7307 @item -time@r{[}=@var{file}@r{]}
7309 Report the CPU time taken by each subprocess in the compilation
7310 sequence. For C source files, this is the compiler proper and assembler
7311 (plus the linker if linking is done).
7313 Without the specification of an output file, the output looks like this:
7320 The first number on each line is the ``user time'', that is time spent
7321 executing the program itself. The second number is ``system time'',
7322 time spent executing operating system routines on behalf of the program.
7323 Both numbers are in seconds.
7325 With the specification of an output file, the output is appended to the
7326 named file, and it looks like this:
7329 0.12 0.01 cc1 @var{options}
7330 0.00 0.01 as @var{options}
7333 The ``user time'' and the ``system time'' are moved before the program
7334 name, and the options passed to the program are displayed, so that one
7335 can later tell what file was being compiled, and with which options.
7337 @item -fvar-tracking
7338 @opindex fvar-tracking
7339 Run variable tracking pass. It computes where variables are stored at each
7340 position in code. Better debugging information is then generated
7341 (if the debugging information format supports this information).
7343 It is enabled by default when compiling with optimization (@option{-Os},
7344 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
7345 the debug info format supports it.
7347 @item -fvar-tracking-assignments
7348 @opindex fvar-tracking-assignments
7349 @opindex fno-var-tracking-assignments
7350 Annotate assignments to user variables early in the compilation and
7351 attempt to carry the annotations over throughout the compilation all the
7352 way to the end, in an attempt to improve debug information while
7353 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
7355 It can be enabled even if var-tracking is disabled, in which case
7356 annotations are created and maintained, but discarded at the end.
7357 By default, this flag is enabled together with @option{-fvar-tracking},
7358 except when selective scheduling is enabled.
7360 @item -fvar-tracking-assignments-toggle
7361 @opindex fvar-tracking-assignments-toggle
7362 @opindex fno-var-tracking-assignments-toggle
7363 Toggle @option{-fvar-tracking-assignments}, in the same way that
7364 @option{-gtoggle} toggles @option{-g}.
7366 @item -print-file-name=@var{library}
7367 @opindex print-file-name
7368 Print the full absolute name of the library file @var{library} that
7369 would be used when linking---and don't do anything else. With this
7370 option, GCC does not compile or link anything; it just prints the
7373 @item -print-multi-directory
7374 @opindex print-multi-directory
7375 Print the directory name corresponding to the multilib selected by any
7376 other switches present in the command line. This directory is supposed
7377 to exist in @env{GCC_EXEC_PREFIX}.
7379 @item -print-multi-lib
7380 @opindex print-multi-lib
7381 Print the mapping from multilib directory names to compiler switches
7382 that enable them. The directory name is separated from the switches by
7383 @samp{;}, and each switch starts with an @samp{@@} instead of the
7384 @samp{-}, without spaces between multiple switches. This is supposed to
7385 ease shell processing.
7387 @item -print-multi-os-directory
7388 @opindex print-multi-os-directory
7389 Print the path to OS libraries for the selected
7390 multilib, relative to some @file{lib} subdirectory. If OS libraries are
7391 present in the @file{lib} subdirectory and no multilibs are used, this is
7392 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
7393 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
7394 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
7395 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
7397 @item -print-multiarch
7398 @opindex print-multiarch
7399 Print the path to OS libraries for the selected multiarch,
7400 relative to some @file{lib} subdirectory.
7402 @item -print-prog-name=@var{program}
7403 @opindex print-prog-name
7404 Like @option{-print-file-name}, but searches for a program such as @command{cpp}.
7406 @item -print-libgcc-file-name
7407 @opindex print-libgcc-file-name
7408 Same as @option{-print-file-name=libgcc.a}.
7410 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
7411 but you do want to link with @file{libgcc.a}. You can do:
7414 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
7417 @item -print-search-dirs
7418 @opindex print-search-dirs
7419 Print the name of the configured installation directory and a list of
7420 program and library directories @command{gcc} searches---and don't do anything else.
7422 This is useful when @command{gcc} prints the error message
7423 @samp{installation problem, cannot exec cpp0: No such file or directory}.
7424 To resolve this you either need to put @file{cpp0} and the other compiler
7425 components where @command{gcc} expects to find them, or you can set the environment
7426 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
7427 Don't forget the trailing @samp{/}.
7428 @xref{Environment Variables}.
7430 @item -print-sysroot
7431 @opindex print-sysroot
7432 Print the target sysroot directory that is used during
7433 compilation. This is the target sysroot specified either at configure
7434 time or using the @option{--sysroot} option, possibly with an extra
7435 suffix that depends on compilation options. If no target sysroot is
7436 specified, the option prints nothing.
7438 @item -print-sysroot-headers-suffix
7439 @opindex print-sysroot-headers-suffix
7440 Print the suffix added to the target sysroot when searching for
7441 headers, or give an error if the compiler is not configured with such
7442 a suffix---and don't do anything else.
7445 @opindex dumpmachine
7446 Print the compiler's target machine (for example,
7447 @samp{i686-pc-linux-gnu})---and don't do anything else.
7450 @opindex dumpversion
7451 Print the compiler version (for example, @code{3.0})---and don't do
7456 Print the compiler's built-in specs---and don't do anything else. (This
7457 is used when GCC itself is being built.) @xref{Spec Files}.
7459 @item -fno-eliminate-unused-debug-types
7460 @opindex feliminate-unused-debug-types
7461 @opindex fno-eliminate-unused-debug-types
7462 Normally, when producing DWARF 2 output, GCC avoids producing debug symbol
7463 output for types that are nowhere used in the source file being compiled.
7464 Sometimes it is useful to have GCC emit debugging
7465 information for all types declared in a compilation
7466 unit, regardless of whether or not they are actually used
7467 in that compilation unit, for example
7468 if, in the debugger, you want to cast a value to a type that is
7469 not actually used in your program (but is declared). More often,
7470 however, this results in a significant amount of wasted space.
7473 @node Optimize Options
7474 @section Options That Control Optimization
7475 @cindex optimize options
7476 @cindex options, optimization
7478 These options control various sorts of optimizations.
7480 Without any optimization option, the compiler's goal is to reduce the
7481 cost of compilation and to make debugging produce the expected
7482 results. Statements are independent: if you stop the program with a
7483 breakpoint between statements, you can then assign a new value to any
7484 variable or change the program counter to any other statement in the
7485 function and get exactly the results you expect from the source
7488 Turning on optimization flags makes the compiler attempt to improve
7489 the performance and/or code size at the expense of compilation time
7490 and possibly the ability to debug the program.
7492 The compiler performs optimization based on the knowledge it has of the
7493 program. Compiling multiple files at once to a single output file mode allows
7494 the compiler to use information gained from all of the files when compiling
7497 Not all optimizations are controlled directly by a flag. Only
7498 optimizations that have a flag are listed in this section.
7500 Most optimizations are only enabled if an @option{-O} level is set on
7501 the command line. Otherwise they are disabled, even if individual
7502 optimization flags are specified.
7504 Depending on the target and how GCC was configured, a slightly different
7505 set of optimizations may be enabled at each @option{-O} level than
7506 those listed here. You can invoke GCC with @option{-Q --help=optimizers}
7507 to find out the exact set of optimizations that are enabled at each level.
7508 @xref{Overall Options}, for examples.
7515 Optimize. Optimizing compilation takes somewhat more time, and a lot
7516 more memory for a large function.
7518 With @option{-O}, the compiler tries to reduce code size and execution
7519 time, without performing any optimizations that take a great deal of
7522 @option{-O} turns on the following optimization flags:
7525 -fbranch-count-reg @gol
7526 -fcombine-stack-adjustments @gol
7528 -fcprop-registers @gol
7531 -fdelayed-branch @gol
7533 -fforward-propagate @gol
7534 -fguess-branch-probability @gol
7535 -fif-conversion2 @gol
7536 -fif-conversion @gol
7537 -finline-functions-called-once @gol
7538 -fipa-pure-const @gol
7540 -fipa-reference @gol
7541 -fmerge-constants @gol
7542 -fmove-loop-invariants @gol
7544 -fsplit-wide-types @gol
7549 -ftree-copy-prop @gol
7550 -ftree-copyrename @gol
7552 -ftree-dominator-opts @gol
7554 -ftree-forwprop @gol
7564 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
7565 where doing so does not interfere with debugging.
7569 Optimize even more. GCC performs nearly all supported optimizations
7570 that do not involve a space-speed tradeoff.
7571 As compared to @option{-O}, this option increases both compilation time
7572 and the performance of the generated code.
7574 @option{-O2} turns on all optimization flags specified by @option{-O}. It
7575 also turns on the following optimization flags:
7576 @gccoptlist{-fthread-jumps @gol
7577 -falign-functions -falign-jumps @gol
7578 -falign-loops -falign-labels @gol
7581 -fcse-follow-jumps -fcse-skip-blocks @gol
7582 -fdelete-null-pointer-checks @gol
7583 -fdevirtualize -fdevirtualize-speculatively @gol
7584 -fexpensive-optimizations @gol
7585 -fgcse -fgcse-lm @gol
7586 -fhoist-adjacent-loads @gol
7587 -finline-small-functions @gol
7588 -findirect-inlining @gol
7590 -fipa-cp-alignment @gol
7593 -fisolate-erroneous-paths-dereference @gol
7595 -foptimize-sibling-calls @gol
7596 -foptimize-strlen @gol
7597 -fpartial-inlining @gol
7599 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
7600 -frerun-cse-after-loop @gol
7601 -fsched-interblock -fsched-spec @gol
7602 -fschedule-insns -fschedule-insns2 @gol
7603 -fstrict-aliasing -fstrict-overflow @gol
7604 -ftree-builtin-call-dce @gol
7605 -ftree-switch-conversion -ftree-tail-merge @gol
7610 Please note the warning under @option{-fgcse} about
7611 invoking @option{-O2} on programs that use computed gotos.
7615 Optimize yet more. @option{-O3} turns on all optimizations specified
7616 by @option{-O2} and also turns on the @option{-finline-functions},
7617 @option{-funswitch-loops}, @option{-fpredictive-commoning},
7618 @option{-fgcse-after-reload}, @option{-ftree-loop-vectorize},
7619 @option{-ftree-loop-distribute-patterns},
7620 @option{-ftree-slp-vectorize}, @option{-fvect-cost-model},
7621 @option{-ftree-partial-pre} and @option{-fipa-cp-clone} options.
7625 Reduce compilation time and make debugging produce the expected
7626 results. This is the default.
7630 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
7631 do not typically increase code size. It also performs further
7632 optimizations designed to reduce code size.
7634 @option{-Os} disables the following optimization flags:
7635 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
7636 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
7637 -fprefetch-loop-arrays}
7641 Disregard strict standards compliance. @option{-Ofast} enables all
7642 @option{-O3} optimizations. It also enables optimizations that are not
7643 valid for all standard-compliant programs.
7644 It turns on @option{-ffast-math} and the Fortran-specific
7645 @option{-fno-protect-parens} and @option{-fstack-arrays}.
7649 Optimize debugging experience. @option{-Og} enables optimizations
7650 that do not interfere with debugging. It should be the optimization
7651 level of choice for the standard edit-compile-debug cycle, offering
7652 a reasonable level of optimization while maintaining fast compilation
7653 and a good debugging experience.
7655 If you use multiple @option{-O} options, with or without level numbers,
7656 the last such option is the one that is effective.
7659 Options of the form @option{-f@var{flag}} specify machine-independent
7660 flags. Most flags have both positive and negative forms; the negative
7661 form of @option{-ffoo} is @option{-fno-foo}. In the table
7662 below, only one of the forms is listed---the one you typically
7663 use. You can figure out the other form by either removing @samp{no-}
7666 The following options control specific optimizations. They are either
7667 activated by @option{-O} options or are related to ones that are. You
7668 can use the following flags in the rare cases when ``fine-tuning'' of
7669 optimizations to be performed is desired.
7672 @item -fno-defer-pop
7673 @opindex fno-defer-pop
7674 Always pop the arguments to each function call as soon as that function
7675 returns. For machines that must pop arguments after a function call,
7676 the compiler normally lets arguments accumulate on the stack for several
7677 function calls and pops them all at once.
7679 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7681 @item -fforward-propagate
7682 @opindex fforward-propagate
7683 Perform a forward propagation pass on RTL@. The pass tries to combine two
7684 instructions and checks if the result can be simplified. If loop unrolling
7685 is active, two passes are performed and the second is scheduled after
7688 This option is enabled by default at optimization levels @option{-O},
7689 @option{-O2}, @option{-O3}, @option{-Os}.
7691 @item -ffp-contract=@var{style}
7692 @opindex ffp-contract
7693 @option{-ffp-contract=off} disables floating-point expression contraction.
7694 @option{-ffp-contract=fast} enables floating-point expression contraction
7695 such as forming of fused multiply-add operations if the target has
7696 native support for them.
7697 @option{-ffp-contract=on} enables floating-point expression contraction
7698 if allowed by the language standard. This is currently not implemented
7699 and treated equal to @option{-ffp-contract=off}.
7701 The default is @option{-ffp-contract=fast}.
7703 @item -fomit-frame-pointer
7704 @opindex fomit-frame-pointer
7705 Don't keep the frame pointer in a register for functions that
7706 don't need one. This avoids the instructions to save, set up and
7707 restore frame pointers; it also makes an extra register available
7708 in many functions. @strong{It also makes debugging impossible on
7711 On some machines, such as the VAX, this flag has no effect, because
7712 the standard calling sequence automatically handles the frame pointer
7713 and nothing is saved by pretending it doesn't exist. The
7714 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
7715 whether a target machine supports this flag. @xref{Registers,,Register
7716 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
7718 The default setting (when not optimizing for
7719 size) for 32-bit GNU/Linux x86 and 32-bit Darwin x86 targets is
7720 @option{-fomit-frame-pointer}. You can configure GCC with the
7721 @option{--enable-frame-pointer} configure option to change the default.
7723 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7725 @item -foptimize-sibling-calls
7726 @opindex foptimize-sibling-calls
7727 Optimize sibling and tail recursive calls.
7729 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7731 @item -foptimize-strlen
7732 @opindex foptimize-strlen
7733 Optimize various standard C string functions (e.g. @code{strlen},
7734 @code{strchr} or @code{strcpy}) and
7735 their @code{_FORTIFY_SOURCE} counterparts into faster alternatives.
7737 Enabled at levels @option{-O2}, @option{-O3}.
7741 Do not expand any functions inline apart from those marked with
7742 the @code{always_inline} attribute. This is the default when not
7745 Single functions can be exempted from inlining by marking them
7746 with the @code{noinline} attribute.
7748 @item -finline-small-functions
7749 @opindex finline-small-functions
7750 Integrate functions into their callers when their body is smaller than expected
7751 function call code (so overall size of program gets smaller). The compiler
7752 heuristically decides which functions are simple enough to be worth integrating
7753 in this way. This inlining applies to all functions, even those not declared
7756 Enabled at level @option{-O2}.
7758 @item -findirect-inlining
7759 @opindex findirect-inlining
7760 Inline also indirect calls that are discovered to be known at compile
7761 time thanks to previous inlining. This option has any effect only
7762 when inlining itself is turned on by the @option{-finline-functions}
7763 or @option{-finline-small-functions} options.
7765 Enabled at level @option{-O2}.
7767 @item -finline-functions
7768 @opindex finline-functions
7769 Consider all functions for inlining, even if they are not declared inline.
7770 The compiler heuristically decides which functions are worth integrating
7773 If all calls to a given function are integrated, and the function is
7774 declared @code{static}, then the function is normally not output as
7775 assembler code in its own right.
7777 Enabled at level @option{-O3}.
7779 @item -finline-functions-called-once
7780 @opindex finline-functions-called-once
7781 Consider all @code{static} functions called once for inlining into their
7782 caller even if they are not marked @code{inline}. If a call to a given
7783 function is integrated, then the function is not output as assembler code
7786 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
7788 @item -fearly-inlining
7789 @opindex fearly-inlining
7790 Inline functions marked by @code{always_inline} and functions whose body seems
7791 smaller than the function call overhead early before doing
7792 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
7793 makes profiling significantly cheaper and usually inlining faster on programs
7794 having large chains of nested wrapper functions.
7800 Perform interprocedural scalar replacement of aggregates, removal of
7801 unused parameters and replacement of parameters passed by reference
7802 by parameters passed by value.
7804 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
7806 @item -finline-limit=@var{n}
7807 @opindex finline-limit
7808 By default, GCC limits the size of functions that can be inlined. This flag
7809 allows coarse control of this limit. @var{n} is the size of functions that
7810 can be inlined in number of pseudo instructions.
7812 Inlining is actually controlled by a number of parameters, which may be
7813 specified individually by using @option{--param @var{name}=@var{value}}.
7814 The @option{-finline-limit=@var{n}} option sets some of these parameters
7818 @item max-inline-insns-single
7819 is set to @var{n}/2.
7820 @item max-inline-insns-auto
7821 is set to @var{n}/2.
7824 See below for a documentation of the individual
7825 parameters controlling inlining and for the defaults of these parameters.
7827 @emph{Note:} there may be no value to @option{-finline-limit} that results
7828 in default behavior.
7830 @emph{Note:} pseudo instruction represents, in this particular context, an
7831 abstract measurement of function's size. In no way does it represent a count
7832 of assembly instructions and as such its exact meaning might change from one
7833 release to an another.
7835 @item -fno-keep-inline-dllexport
7836 @opindex fno-keep-inline-dllexport
7837 This is a more fine-grained version of @option{-fkeep-inline-functions},
7838 which applies only to functions that are declared using the @code{dllexport}
7839 attribute or declspec (@xref{Function Attributes,,Declaring Attributes of
7842 @item -fkeep-inline-functions
7843 @opindex fkeep-inline-functions
7844 In C, emit @code{static} functions that are declared @code{inline}
7845 into the object file, even if the function has been inlined into all
7846 of its callers. This switch does not affect functions using the
7847 @code{extern inline} extension in GNU C90@. In C++, emit any and all
7848 inline functions into the object file.
7850 @item -fkeep-static-consts
7851 @opindex fkeep-static-consts
7852 Emit variables declared @code{static const} when optimization isn't turned
7853 on, even if the variables aren't referenced.
7855 GCC enables this option by default. If you want to force the compiler to
7856 check if a variable is referenced, regardless of whether or not
7857 optimization is turned on, use the @option{-fno-keep-static-consts} option.
7859 @item -fmerge-constants
7860 @opindex fmerge-constants
7861 Attempt to merge identical constants (string constants and floating-point
7862 constants) across compilation units.
7864 This option is the default for optimized compilation if the assembler and
7865 linker support it. Use @option{-fno-merge-constants} to inhibit this
7868 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7870 @item -fmerge-all-constants
7871 @opindex fmerge-all-constants
7872 Attempt to merge identical constants and identical variables.
7874 This option implies @option{-fmerge-constants}. In addition to
7875 @option{-fmerge-constants} this considers e.g.@: even constant initialized
7876 arrays or initialized constant variables with integral or floating-point
7877 types. Languages like C or C++ require each variable, including multiple
7878 instances of the same variable in recursive calls, to have distinct locations,
7879 so using this option results in non-conforming
7882 @item -fmodulo-sched
7883 @opindex fmodulo-sched
7884 Perform swing modulo scheduling immediately before the first scheduling
7885 pass. This pass looks at innermost loops and reorders their
7886 instructions by overlapping different iterations.
7888 @item -fmodulo-sched-allow-regmoves
7889 @opindex fmodulo-sched-allow-regmoves
7890 Perform more aggressive SMS-based modulo scheduling with register moves
7891 allowed. By setting this flag certain anti-dependences edges are
7892 deleted, which triggers the generation of reg-moves based on the
7893 life-range analysis. This option is effective only with
7894 @option{-fmodulo-sched} enabled.
7896 @item -fno-branch-count-reg
7897 @opindex fno-branch-count-reg
7898 Do not use ``decrement and branch'' instructions on a count register,
7899 but instead generate a sequence of instructions that decrement a
7900 register, compare it against zero, then branch based upon the result.
7901 This option is only meaningful on architectures that support such
7902 instructions, which include x86, PowerPC, IA-64 and S/390.
7904 Enabled by default at @option{-O1} and higher.
7906 The default is @option{-fbranch-count-reg}.
7908 @item -fno-function-cse
7909 @opindex fno-function-cse
7910 Do not put function addresses in registers; make each instruction that
7911 calls a constant function contain the function's address explicitly.
7913 This option results in less efficient code, but some strange hacks
7914 that alter the assembler output may be confused by the optimizations
7915 performed when this option is not used.
7917 The default is @option{-ffunction-cse}
7919 @item -fno-zero-initialized-in-bss
7920 @opindex fno-zero-initialized-in-bss
7921 If the target supports a BSS section, GCC by default puts variables that
7922 are initialized to zero into BSS@. This can save space in the resulting
7925 This option turns off this behavior because some programs explicitly
7926 rely on variables going to the data section---e.g., so that the
7927 resulting executable can find the beginning of that section and/or make
7928 assumptions based on that.
7930 The default is @option{-fzero-initialized-in-bss}.
7932 @item -fthread-jumps
7933 @opindex fthread-jumps
7934 Perform optimizations that check to see if a jump branches to a
7935 location where another comparison subsumed by the first is found. If
7936 so, the first branch is redirected to either the destination of the
7937 second branch or a point immediately following it, depending on whether
7938 the condition is known to be true or false.
7940 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7942 @item -fsplit-wide-types
7943 @opindex fsplit-wide-types
7944 When using a type that occupies multiple registers, such as @code{long
7945 long} on a 32-bit system, split the registers apart and allocate them
7946 independently. This normally generates better code for those types,
7947 but may make debugging more difficult.
7949 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
7952 @item -fcse-follow-jumps
7953 @opindex fcse-follow-jumps
7954 In common subexpression elimination (CSE), scan through jump instructions
7955 when the target of the jump is not reached by any other path. For
7956 example, when CSE encounters an @code{if} statement with an
7957 @code{else} clause, CSE follows the jump when the condition
7960 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7962 @item -fcse-skip-blocks
7963 @opindex fcse-skip-blocks
7964 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
7965 follow jumps that conditionally skip over blocks. When CSE
7966 encounters a simple @code{if} statement with no else clause,
7967 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
7968 body of the @code{if}.
7970 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7972 @item -frerun-cse-after-loop
7973 @opindex frerun-cse-after-loop
7974 Re-run common subexpression elimination after loop optimizations are
7977 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7981 Perform a global common subexpression elimination pass.
7982 This pass also performs global constant and copy propagation.
7984 @emph{Note:} When compiling a program using computed gotos, a GCC
7985 extension, you may get better run-time performance if you disable
7986 the global common subexpression elimination pass by adding
7987 @option{-fno-gcse} to the command line.
7989 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7993 When @option{-fgcse-lm} is enabled, global common subexpression elimination
7994 attempts to move loads that are only killed by stores into themselves. This
7995 allows a loop containing a load/store sequence to be changed to a load outside
7996 the loop, and a copy/store within the loop.
7998 Enabled by default when @option{-fgcse} is enabled.
8002 When @option{-fgcse-sm} is enabled, a store motion pass is run after
8003 global common subexpression elimination. This pass attempts to move
8004 stores out of loops. When used in conjunction with @option{-fgcse-lm},
8005 loops containing a load/store sequence can be changed to a load before
8006 the loop and a store after the loop.
8008 Not enabled at any optimization level.
8012 When @option{-fgcse-las} is enabled, the global common subexpression
8013 elimination pass eliminates redundant loads that come after stores to the
8014 same memory location (both partial and full redundancies).
8016 Not enabled at any optimization level.
8018 @item -fgcse-after-reload
8019 @opindex fgcse-after-reload
8020 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
8021 pass is performed after reload. The purpose of this pass is to clean up
8024 @item -faggressive-loop-optimizations
8025 @opindex faggressive-loop-optimizations
8026 This option tells the loop optimizer to use language constraints to
8027 derive bounds for the number of iterations of a loop. This assumes that
8028 loop code does not invoke undefined behavior by for example causing signed
8029 integer overflows or out-of-bound array accesses. The bounds for the
8030 number of iterations of a loop are used to guide loop unrolling and peeling
8031 and loop exit test optimizations.
8032 This option is enabled by default.
8034 @item -funsafe-loop-optimizations
8035 @opindex funsafe-loop-optimizations
8036 This option tells the loop optimizer to assume that loop indices do not
8037 overflow, and that loops with nontrivial exit condition are not
8038 infinite. This enables a wider range of loop optimizations even if
8039 the loop optimizer itself cannot prove that these assumptions are valid.
8040 If you use @option{-Wunsafe-loop-optimizations}, the compiler warns you
8041 if it finds this kind of loop.
8043 @item -fcrossjumping
8044 @opindex fcrossjumping
8045 Perform cross-jumping transformation.
8046 This transformation unifies equivalent code and saves code size. The
8047 resulting code may or may not perform better than without cross-jumping.
8049 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8051 @item -fauto-inc-dec
8052 @opindex fauto-inc-dec
8053 Combine increments or decrements of addresses with memory accesses.
8054 This pass is always skipped on architectures that do not have
8055 instructions to support this. Enabled by default at @option{-O} and
8056 higher on architectures that support this.
8060 Perform dead code elimination (DCE) on RTL@.
8061 Enabled by default at @option{-O} and higher.
8065 Perform dead store elimination (DSE) on RTL@.
8066 Enabled by default at @option{-O} and higher.
8068 @item -fif-conversion
8069 @opindex fif-conversion
8070 Attempt to transform conditional jumps into branch-less equivalents. This
8071 includes use of conditional moves, min, max, set flags and abs instructions, and
8072 some tricks doable by standard arithmetics. The use of conditional execution
8073 on chips where it is available is controlled by @option{-fif-conversion2}.
8075 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8077 @item -fif-conversion2
8078 @opindex fif-conversion2
8079 Use conditional execution (where available) to transform conditional jumps into
8080 branch-less equivalents.
8082 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8084 @item -fdeclone-ctor-dtor
8085 @opindex fdeclone-ctor-dtor
8086 The C++ ABI requires multiple entry points for constructors and
8087 destructors: one for a base subobject, one for a complete object, and
8088 one for a virtual destructor that calls operator delete afterwards.
8089 For a hierarchy with virtual bases, the base and complete variants are
8090 clones, which means two copies of the function. With this option, the
8091 base and complete variants are changed to be thunks that call a common
8094 Enabled by @option{-Os}.
8096 @item -fdelete-null-pointer-checks
8097 @opindex fdelete-null-pointer-checks
8098 Assume that programs cannot safely dereference null pointers, and that
8099 no code or data element resides at address zero.
8100 This option enables simple constant
8101 folding optimizations at all optimization levels. In addition, other
8102 optimization passes in GCC use this flag to control global dataflow
8103 analyses that eliminate useless checks for null pointers; these assume
8104 that a memory access to address zero always results in a trap, so
8105 that if a pointer is checked after it has already been dereferenced,
8108 Note however that in some environments this assumption is not true.
8109 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
8110 for programs that depend on that behavior.
8112 This option is enabled by default on most targets. On Nios II ELF, it
8113 defaults to off. On AVR and CR16, this option is completely disabled.
8115 Passes that use the dataflow information
8116 are enabled independently at different optimization levels.
8118 @item -fdevirtualize
8119 @opindex fdevirtualize
8120 Attempt to convert calls to virtual functions to direct calls. This
8121 is done both within a procedure and interprocedurally as part of
8122 indirect inlining (@option{-findirect-inlining}) and interprocedural constant
8123 propagation (@option{-fipa-cp}).
8124 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8126 @item -fdevirtualize-speculatively
8127 @opindex fdevirtualize-speculatively
8128 Attempt to convert calls to virtual functions to speculative direct calls.
8129 Based on the analysis of the type inheritance graph, determine for a given call
8130 the set of likely targets. If the set is small, preferably of size 1, change
8131 the call into a conditional deciding between direct and indirect calls. The
8132 speculative calls enable more optimizations, such as inlining. When they seem
8133 useless after further optimization, they are converted back into original form.
8135 @item -fdevirtualize-at-ltrans
8136 @opindex fdevirtualize-at-ltrans
8137 Stream extra information needed for aggressive devirtualization when running
8138 the link-time optimizer in local transformation mode.
8139 This option enables more devirtualization but
8140 significantly increases the size of streamed data. For this reason it is
8141 disabled by default.
8143 @item -fexpensive-optimizations
8144 @opindex fexpensive-optimizations
8145 Perform a number of minor optimizations that are relatively expensive.
8147 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8151 Attempt to remove redundant extension instructions. This is especially
8152 helpful for the x86-64 architecture, which implicitly zero-extends in 64-bit
8153 registers after writing to their lower 32-bit half.
8155 Enabled for Alpha, AArch64 and x86 at levels @option{-O2},
8156 @option{-O3}, @option{-Os}.
8158 @item -fno-lifetime-dse
8159 @opindex fno-lifetime-dse
8160 In C++ the value of an object is only affected by changes within its
8161 lifetime: when the constructor begins, the object has an indeterminate
8162 value, and any changes during the lifetime of the object are dead when
8163 the object is destroyed. Normally dead store elimination will take
8164 advantage of this; if your code relies on the value of the object
8165 storage persisting beyond the lifetime of the object, you can use this
8166 flag to disable this optimization.
8168 @item -flive-range-shrinkage
8169 @opindex flive-range-shrinkage
8170 Attempt to decrease register pressure through register live range
8171 shrinkage. This is helpful for fast processors with small or moderate
8174 @item -fira-algorithm=@var{algorithm}
8175 @opindex fira-algorithm
8176 Use the specified coloring algorithm for the integrated register
8177 allocator. The @var{algorithm} argument can be @samp{priority}, which
8178 specifies Chow's priority coloring, or @samp{CB}, which specifies
8179 Chaitin-Briggs coloring. Chaitin-Briggs coloring is not implemented
8180 for all architectures, but for those targets that do support it, it is
8181 the default because it generates better code.
8183 @item -fira-region=@var{region}
8184 @opindex fira-region
8185 Use specified regions for the integrated register allocator. The
8186 @var{region} argument should be one of the following:
8191 Use all loops as register allocation regions.
8192 This can give the best results for machines with a small and/or
8193 irregular register set.
8196 Use all loops except for loops with small register pressure
8197 as the regions. This value usually gives
8198 the best results in most cases and for most architectures,
8199 and is enabled by default when compiling with optimization for speed
8200 (@option{-O}, @option{-O2}, @dots{}).
8203 Use all functions as a single region.
8204 This typically results in the smallest code size, and is enabled by default for
8205 @option{-Os} or @option{-O0}.
8209 @item -fira-hoist-pressure
8210 @opindex fira-hoist-pressure
8211 Use IRA to evaluate register pressure in the code hoisting pass for
8212 decisions to hoist expressions. This option usually results in smaller
8213 code, but it can slow the compiler down.
8215 This option is enabled at level @option{-Os} for all targets.
8217 @item -fira-loop-pressure
8218 @opindex fira-loop-pressure
8219 Use IRA to evaluate register pressure in loops for decisions to move
8220 loop invariants. This option usually results in generation
8221 of faster and smaller code on machines with large register files (>= 32
8222 registers), but it can slow the compiler down.
8224 This option is enabled at level @option{-O3} for some targets.
8226 @item -fno-ira-share-save-slots
8227 @opindex fno-ira-share-save-slots
8228 Disable sharing of stack slots used for saving call-used hard
8229 registers living through a call. Each hard register gets a
8230 separate stack slot, and as a result function stack frames are
8233 @item -fno-ira-share-spill-slots
8234 @opindex fno-ira-share-spill-slots
8235 Disable sharing of stack slots allocated for pseudo-registers. Each
8236 pseudo-register that does not get a hard register gets a separate
8237 stack slot, and as a result function stack frames are larger.
8239 @item -fira-verbose=@var{n}
8240 @opindex fira-verbose
8241 Control the verbosity of the dump file for the integrated register allocator.
8242 The default value is 5. If the value @var{n} is greater or equal to 10,
8243 the dump output is sent to stderr using the same format as @var{n} minus 10.
8247 Enable CFG-sensitive rematerialization in LRA. Instead of loading
8248 values of spilled pseudos, LRA tries to rematerialize (recalculate)
8249 values if it is profitable.
8251 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8253 @item -fdelayed-branch
8254 @opindex fdelayed-branch
8255 If supported for the target machine, attempt to reorder instructions
8256 to exploit instruction slots available after delayed branch
8259 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8261 @item -fschedule-insns
8262 @opindex fschedule-insns
8263 If supported for the target machine, attempt to reorder instructions to
8264 eliminate execution stalls due to required data being unavailable. This
8265 helps machines that have slow floating point or memory load instructions
8266 by allowing other instructions to be issued until the result of the load
8267 or floating-point instruction is required.
8269 Enabled at levels @option{-O2}, @option{-O3}.
8271 @item -fschedule-insns2
8272 @opindex fschedule-insns2
8273 Similar to @option{-fschedule-insns}, but requests an additional pass of
8274 instruction scheduling after register allocation has been done. This is
8275 especially useful on machines with a relatively small number of
8276 registers and where memory load instructions take more than one cycle.
8278 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8280 @item -fno-sched-interblock
8281 @opindex fno-sched-interblock
8282 Don't schedule instructions across basic blocks. This is normally
8283 enabled by default when scheduling before register allocation, i.e.@:
8284 with @option{-fschedule-insns} or at @option{-O2} or higher.
8286 @item -fno-sched-spec
8287 @opindex fno-sched-spec
8288 Don't allow speculative motion of non-load instructions. This is normally
8289 enabled by default when scheduling before register allocation, i.e.@:
8290 with @option{-fschedule-insns} or at @option{-O2} or higher.
8292 @item -fsched-pressure
8293 @opindex fsched-pressure
8294 Enable register pressure sensitive insn scheduling before register
8295 allocation. This only makes sense when scheduling before register
8296 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
8297 @option{-O2} or higher. Usage of this option can improve the
8298 generated code and decrease its size by preventing register pressure
8299 increase above the number of available hard registers and subsequent
8300 spills in register allocation.
8302 @item -fsched-spec-load
8303 @opindex fsched-spec-load
8304 Allow speculative motion of some load instructions. This only makes
8305 sense when scheduling before register allocation, i.e.@: with
8306 @option{-fschedule-insns} or at @option{-O2} or higher.
8308 @item -fsched-spec-load-dangerous
8309 @opindex fsched-spec-load-dangerous
8310 Allow speculative motion of more load instructions. This only makes
8311 sense when scheduling before register allocation, i.e.@: with
8312 @option{-fschedule-insns} or at @option{-O2} or higher.
8314 @item -fsched-stalled-insns
8315 @itemx -fsched-stalled-insns=@var{n}
8316 @opindex fsched-stalled-insns
8317 Define how many insns (if any) can be moved prematurely from the queue
8318 of stalled insns into the ready list during the second scheduling pass.
8319 @option{-fno-sched-stalled-insns} means that no insns are moved
8320 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
8321 on how many queued insns can be moved prematurely.
8322 @option{-fsched-stalled-insns} without a value is equivalent to
8323 @option{-fsched-stalled-insns=1}.
8325 @item -fsched-stalled-insns-dep
8326 @itemx -fsched-stalled-insns-dep=@var{n}
8327 @opindex fsched-stalled-insns-dep
8328 Define how many insn groups (cycles) are examined for a dependency
8329 on a stalled insn that is a candidate for premature removal from the queue
8330 of stalled insns. This has an effect only during the second scheduling pass,
8331 and only if @option{-fsched-stalled-insns} is used.
8332 @option{-fno-sched-stalled-insns-dep} is equivalent to
8333 @option{-fsched-stalled-insns-dep=0}.
8334 @option{-fsched-stalled-insns-dep} without a value is equivalent to
8335 @option{-fsched-stalled-insns-dep=1}.
8337 @item -fsched2-use-superblocks
8338 @opindex fsched2-use-superblocks
8339 When scheduling after register allocation, use superblock scheduling.
8340 This allows motion across basic block boundaries,
8341 resulting in faster schedules. This option is experimental, as not all machine
8342 descriptions used by GCC model the CPU closely enough to avoid unreliable
8343 results from the algorithm.
8345 This only makes sense when scheduling after register allocation, i.e.@: with
8346 @option{-fschedule-insns2} or at @option{-O2} or higher.
8348 @item -fsched-group-heuristic
8349 @opindex fsched-group-heuristic
8350 Enable the group heuristic in the scheduler. This heuristic favors
8351 the instruction that belongs to a schedule group. This is enabled
8352 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
8353 or @option{-fschedule-insns2} or at @option{-O2} or higher.
8355 @item -fsched-critical-path-heuristic
8356 @opindex fsched-critical-path-heuristic
8357 Enable the critical-path heuristic in the scheduler. This heuristic favors
8358 instructions on the critical path. This is enabled by default when
8359 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
8360 or @option{-fschedule-insns2} or at @option{-O2} or higher.
8362 @item -fsched-spec-insn-heuristic
8363 @opindex fsched-spec-insn-heuristic
8364 Enable the speculative instruction heuristic in the scheduler. This
8365 heuristic favors speculative instructions with greater dependency weakness.
8366 This is enabled by default when scheduling is enabled, i.e.@:
8367 with @option{-fschedule-insns} or @option{-fschedule-insns2}
8368 or at @option{-O2} or higher.
8370 @item -fsched-rank-heuristic
8371 @opindex fsched-rank-heuristic
8372 Enable the rank heuristic in the scheduler. This heuristic favors
8373 the instruction belonging to a basic block with greater size or frequency.
8374 This is enabled by default when scheduling is enabled, i.e.@:
8375 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
8376 at @option{-O2} or higher.
8378 @item -fsched-last-insn-heuristic
8379 @opindex fsched-last-insn-heuristic
8380 Enable the last-instruction heuristic in the scheduler. This heuristic
8381 favors the instruction that is less dependent on the last instruction
8382 scheduled. This is enabled by default when scheduling is enabled,
8383 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
8384 at @option{-O2} or higher.
8386 @item -fsched-dep-count-heuristic
8387 @opindex fsched-dep-count-heuristic
8388 Enable the dependent-count heuristic in the scheduler. This heuristic
8389 favors the instruction that has more instructions depending on it.
8390 This is enabled by default when scheduling is enabled, i.e.@:
8391 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
8392 at @option{-O2} or higher.
8394 @item -freschedule-modulo-scheduled-loops
8395 @opindex freschedule-modulo-scheduled-loops
8396 Modulo scheduling is performed before traditional scheduling. If a loop
8397 is modulo scheduled, later scheduling passes may change its schedule.
8398 Use this option to control that behavior.
8400 @item -fselective-scheduling
8401 @opindex fselective-scheduling
8402 Schedule instructions using selective scheduling algorithm. Selective
8403 scheduling runs instead of the first scheduler pass.
8405 @item -fselective-scheduling2
8406 @opindex fselective-scheduling2
8407 Schedule instructions using selective scheduling algorithm. Selective
8408 scheduling runs instead of the second scheduler pass.
8410 @item -fsel-sched-pipelining
8411 @opindex fsel-sched-pipelining
8412 Enable software pipelining of innermost loops during selective scheduling.
8413 This option has no effect unless one of @option{-fselective-scheduling} or
8414 @option{-fselective-scheduling2} is turned on.
8416 @item -fsel-sched-pipelining-outer-loops
8417 @opindex fsel-sched-pipelining-outer-loops
8418 When pipelining loops during selective scheduling, also pipeline outer loops.
8419 This option has no effect unless @option{-fsel-sched-pipelining} is turned on.
8421 @item -fsemantic-interposition
8422 @opindex fsemantic-interposition
8423 Some object formats, like ELF, allow interposing of symbols by the
8425 This means that for symbols exported from the DSO, the compiler cannot perform
8426 interprocedural propagation, inlining and other optimizations in anticipation
8427 that the function or variable in question may change. While this feature is
8428 useful, for example, to rewrite memory allocation functions by a debugging
8429 implementation, it is expensive in the terms of code quality.
8430 With @option{-fno-semantic-interposition} the compiler assumes that
8431 if interposition happens for functions the overwriting function will have
8432 precisely the same semantics (and side effects).
8433 Similarly if interposition happens
8434 for variables, the constructor of the variable will be the same. The flag
8435 has no effect for functions explicitly declared inline
8436 (where it is never allowed for interposition to change semantics)
8437 and for symbols explicitly declared weak.
8440 @opindex fshrink-wrap
8441 Emit function prologues only before parts of the function that need it,
8442 rather than at the top of the function. This flag is enabled by default at
8443 @option{-O} and higher.
8445 @item -fcaller-saves
8446 @opindex fcaller-saves
8447 Enable allocation of values to registers that are clobbered by
8448 function calls, by emitting extra instructions to save and restore the
8449 registers around such calls. Such allocation is done only when it
8450 seems to result in better code.
8452 This option is always enabled by default on certain machines, usually
8453 those which have no call-preserved registers to use instead.
8455 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8457 @item -fcombine-stack-adjustments
8458 @opindex fcombine-stack-adjustments
8459 Tracks stack adjustments (pushes and pops) and stack memory references
8460 and then tries to find ways to combine them.
8462 Enabled by default at @option{-O1} and higher.
8466 Use caller save registers for allocation if those registers are not used by
8467 any called function. In that case it is not necessary to save and restore
8468 them around calls. This is only possible if called functions are part of
8469 same compilation unit as current function and they are compiled before it.
8471 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8473 @item -fconserve-stack
8474 @opindex fconserve-stack
8475 Attempt to minimize stack usage. The compiler attempts to use less
8476 stack space, even if that makes the program slower. This option
8477 implies setting the @option{large-stack-frame} parameter to 100
8478 and the @option{large-stack-frame-growth} parameter to 400.
8480 @item -ftree-reassoc
8481 @opindex ftree-reassoc
8482 Perform reassociation on trees. This flag is enabled by default
8483 at @option{-O} and higher.
8487 Perform partial redundancy elimination (PRE) on trees. This flag is
8488 enabled by default at @option{-O2} and @option{-O3}.
8490 @item -ftree-partial-pre
8491 @opindex ftree-partial-pre
8492 Make partial redundancy elimination (PRE) more aggressive. This flag is
8493 enabled by default at @option{-O3}.
8495 @item -ftree-forwprop
8496 @opindex ftree-forwprop
8497 Perform forward propagation on trees. This flag is enabled by default
8498 at @option{-O} and higher.
8502 Perform full redundancy elimination (FRE) on trees. The difference
8503 between FRE and PRE is that FRE only considers expressions
8504 that are computed on all paths leading to the redundant computation.
8505 This analysis is faster than PRE, though it exposes fewer redundancies.
8506 This flag is enabled by default at @option{-O} and higher.
8508 @item -ftree-phiprop
8509 @opindex ftree-phiprop
8510 Perform hoisting of loads from conditional pointers on trees. This
8511 pass is enabled by default at @option{-O} and higher.
8513 @item -fhoist-adjacent-loads
8514 @opindex fhoist-adjacent-loads
8515 Speculatively hoist loads from both branches of an if-then-else if the
8516 loads are from adjacent locations in the same structure and the target
8517 architecture has a conditional move instruction. This flag is enabled
8518 by default at @option{-O2} and higher.
8520 @item -ftree-copy-prop
8521 @opindex ftree-copy-prop
8522 Perform copy propagation on trees. This pass eliminates unnecessary
8523 copy operations. This flag is enabled by default at @option{-O} and
8526 @item -fipa-pure-const
8527 @opindex fipa-pure-const
8528 Discover which functions are pure or constant.
8529 Enabled by default at @option{-O} and higher.
8531 @item -fipa-reference
8532 @opindex fipa-reference
8533 Discover which static variables do not escape the
8535 Enabled by default at @option{-O} and higher.
8539 Perform interprocedural pointer analysis and interprocedural modification
8540 and reference analysis. This option can cause excessive memory and
8541 compile-time usage on large compilation units. It is not enabled by
8542 default at any optimization level.
8545 @opindex fipa-profile
8546 Perform interprocedural profile propagation. The functions called only from
8547 cold functions are marked as cold. Also functions executed once (such as
8548 @code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
8549 functions and loop less parts of functions executed once are then optimized for
8551 Enabled by default at @option{-O} and higher.
8555 Perform interprocedural constant propagation.
8556 This optimization analyzes the program to determine when values passed
8557 to functions are constants and then optimizes accordingly.
8558 This optimization can substantially increase performance
8559 if the application has constants passed to functions.
8560 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
8562 @item -fipa-cp-clone
8563 @opindex fipa-cp-clone
8564 Perform function cloning to make interprocedural constant propagation stronger.
8565 When enabled, interprocedural constant propagation performs function cloning
8566 when externally visible function can be called with constant arguments.
8567 Because this optimization can create multiple copies of functions,
8568 it may significantly increase code size
8569 (see @option{--param ipcp-unit-growth=@var{value}}).
8570 This flag is enabled by default at @option{-O3}.
8572 @item -fipa-cp-alignment
8573 @opindex -fipa-cp-alignment
8574 When enabled, this optimization propagates alignment of function
8575 parameters to support better vectorization and string operations.
8577 This flag is enabled by default at @option{-O2} and @option{-Os}. It
8578 requires that @option{-fipa-cp} is enabled.
8582 Perform Identical Code Folding for functions and read-only variables.
8583 The optimization reduces code size and may disturb unwind stacks by replacing
8584 a function by equivalent one with a different name. The optimization works
8585 more effectively with link time optimization enabled.
8587 Nevertheless the behavior is similar to Gold Linker ICF optimization, GCC ICF
8588 works on different levels and thus the optimizations are not same - there are
8589 equivalences that are found only by GCC and equivalences found only by Gold.
8591 This flag is enabled by default at @option{-O2} and @option{-Os}.
8593 @item -fisolate-erroneous-paths-dereference
8594 @opindex fisolate-erroneous-paths-dereference
8595 Detect paths that trigger erroneous or undefined behavior due to
8596 dereferencing a null pointer. Isolate those paths from the main control
8597 flow and turn the statement with erroneous or undefined behavior into a trap.
8598 This flag is enabled by default at @option{-O2} and higher and depends on
8599 @option{-fdelete-null-pointer-checks} also being enabled.
8601 @item -fisolate-erroneous-paths-attribute
8602 @opindex fisolate-erroneous-paths-attribute
8603 Detect paths that trigger erroneous or undefined behavior due a null value
8604 being used in a way forbidden by a @code{returns_nonnull} or @code{nonnull}
8605 attribute. Isolate those paths from the main control flow and turn the
8606 statement with erroneous or undefined behavior into a trap. This is not
8607 currently enabled, but may be enabled by @option{-O2} in the future.
8611 Perform forward store motion on trees. This flag is
8612 enabled by default at @option{-O} and higher.
8614 @item -ftree-bit-ccp
8615 @opindex ftree-bit-ccp
8616 Perform sparse conditional bit constant propagation on trees and propagate
8617 pointer alignment information.
8618 This pass only operates on local scalar variables and is enabled by default
8619 at @option{-O} and higher. It requires that @option{-ftree-ccp} is enabled.
8623 Perform sparse conditional constant propagation (CCP) on trees. This
8624 pass only operates on local scalar variables and is enabled by default
8625 at @option{-O} and higher.
8628 @opindex fssa-phiopt
8629 Perform pattern matching on SSA PHI nodes to optimize conditional
8630 code. This pass is enabled by default at @option{-O} and higher.
8632 @item -ftree-switch-conversion
8633 @opindex ftree-switch-conversion
8634 Perform conversion of simple initializations in a switch to
8635 initializations from a scalar array. This flag is enabled by default
8636 at @option{-O2} and higher.
8638 @item -ftree-tail-merge
8639 @opindex ftree-tail-merge
8640 Look for identical code sequences. When found, replace one with a jump to the
8641 other. This optimization is known as tail merging or cross jumping. This flag
8642 is enabled by default at @option{-O2} and higher. The compilation time
8644 be limited using @option{max-tail-merge-comparisons} parameter and
8645 @option{max-tail-merge-iterations} parameter.
8649 Perform dead code elimination (DCE) on trees. This flag is enabled by
8650 default at @option{-O} and higher.
8652 @item -ftree-builtin-call-dce
8653 @opindex ftree-builtin-call-dce
8654 Perform conditional dead code elimination (DCE) for calls to built-in functions
8655 that may set @code{errno} but are otherwise side-effect free. This flag is
8656 enabled by default at @option{-O2} and higher if @option{-Os} is not also
8659 @item -ftree-dominator-opts
8660 @opindex ftree-dominator-opts
8661 Perform a variety of simple scalar cleanups (constant/copy
8662 propagation, redundancy elimination, range propagation and expression
8663 simplification) based on a dominator tree traversal. This also
8664 performs jump threading (to reduce jumps to jumps). This flag is
8665 enabled by default at @option{-O} and higher.
8669 Perform dead store elimination (DSE) on trees. A dead store is a store into
8670 a memory location that is later overwritten by another store without
8671 any intervening loads. In this case the earlier store can be deleted. This
8672 flag is enabled by default at @option{-O} and higher.
8676 Perform loop header copying on trees. This is beneficial since it increases
8677 effectiveness of code motion optimizations. It also saves one jump. This flag
8678 is enabled by default at @option{-O} and higher. It is not enabled
8679 for @option{-Os}, since it usually increases code size.
8681 @item -ftree-loop-optimize
8682 @opindex ftree-loop-optimize
8683 Perform loop optimizations on trees. This flag is enabled by default
8684 at @option{-O} and higher.
8686 @item -ftree-loop-linear
8687 @opindex ftree-loop-linear
8688 Perform loop interchange transformations on tree. Same as
8689 @option{-floop-interchange}. To use this code transformation, GCC has
8690 to be configured with @option{--with-isl} to enable the Graphite loop
8691 transformation infrastructure.
8693 @item -floop-interchange
8694 @opindex floop-interchange
8695 Perform loop interchange transformations on loops. Interchanging two
8696 nested loops switches the inner and outer loops. For example, given a
8701 A(J, I) = A(J, I) * C
8706 loop interchange transforms the loop as if it were written:
8710 A(J, I) = A(J, I) * C
8714 which can be beneficial when @code{N} is larger than the caches,
8715 because in Fortran, the elements of an array are stored in memory
8716 contiguously by column, and the original loop iterates over rows,
8717 potentially creating at each access a cache miss. This optimization
8718 applies to all the languages supported by GCC and is not limited to
8719 Fortran. To use this code transformation, GCC has to be configured
8720 with @option{--with-isl} to enable the Graphite loop transformation
8723 @item -floop-strip-mine
8724 @opindex floop-strip-mine
8725 Perform loop strip mining transformations on loops. Strip mining
8726 splits a loop into two nested loops. The outer loop has strides
8727 equal to the strip size and the inner loop has strides of the
8728 original loop within a strip. The strip length can be changed
8729 using the @option{loop-block-tile-size} parameter. For example,
8737 loop strip mining transforms the loop as if it were written:
8740 DO I = II, min (II + 50, N)
8745 This optimization applies to all the languages supported by GCC and is
8746 not limited to Fortran. To use this code transformation, GCC has to
8747 be configured with @option{--with-isl} to enable the Graphite loop
8748 transformation infrastructure.
8751 @opindex floop-block
8752 Perform loop blocking transformations on loops. Blocking strip mines
8753 each loop in the loop nest such that the memory accesses of the
8754 element loops fit inside caches. The strip length can be changed
8755 using the @option{loop-block-tile-size} parameter. For example, given
8760 A(J, I) = B(I) + C(J)
8765 loop blocking transforms the loop as if it were written:
8769 DO I = II, min (II + 50, N)
8770 DO J = JJ, min (JJ + 50, M)
8771 A(J, I) = B(I) + C(J)
8777 which can be beneficial when @code{M} is larger than the caches,
8778 because the innermost loop iterates over a smaller amount of data
8779 which can be kept in the caches. This optimization applies to all the
8780 languages supported by GCC and is not limited to Fortran. To use this
8781 code transformation, GCC has to be configured with @option{--with-isl}
8782 to enable the Graphite loop transformation infrastructure.
8784 @item -fgraphite-identity
8785 @opindex fgraphite-identity
8786 Enable the identity transformation for graphite. For every SCoP we generate
8787 the polyhedral representation and transform it back to gimple. Using
8788 @option{-fgraphite-identity} we can check the costs or benefits of the
8789 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
8790 are also performed by the code generator ISL, like index splitting and
8791 dead code elimination in loops.
8793 @item -floop-nest-optimize
8794 @opindex floop-nest-optimize
8795 Enable the ISL based loop nest optimizer. This is a generic loop nest
8796 optimizer based on the Pluto optimization algorithms. It calculates a loop
8797 structure optimized for data-locality and parallelism. This option
8800 @item -floop-unroll-and-jam
8801 @opindex floop-unroll-and-jam
8802 Enable unroll and jam for the ISL based loop nest optimizer. The unroll
8803 factor can be changed using the @option{loop-unroll-jam-size} parameter.
8804 The unrolled dimension (counting from the most inner one) can be changed
8805 using the @option{loop-unroll-jam-depth} parameter. .
8807 @item -floop-parallelize-all
8808 @opindex floop-parallelize-all
8809 Use the Graphite data dependence analysis to identify loops that can
8810 be parallelized. Parallelize all the loops that can be analyzed to
8811 not contain loop carried dependences without checking that it is
8812 profitable to parallelize the loops.
8814 @item -fcheck-data-deps
8815 @opindex fcheck-data-deps
8816 Compare the results of several data dependence analyzers. This option
8817 is used for debugging the data dependence analyzers.
8819 @item -ftree-loop-if-convert
8820 @opindex ftree-loop-if-convert
8821 Attempt to transform conditional jumps in the innermost loops to
8822 branch-less equivalents. The intent is to remove control-flow from
8823 the innermost loops in order to improve the ability of the
8824 vectorization pass to handle these loops. This is enabled by default
8825 if vectorization is enabled.
8827 @item -ftree-loop-if-convert-stores
8828 @opindex ftree-loop-if-convert-stores
8829 Attempt to also if-convert conditional jumps containing memory writes.
8830 This transformation can be unsafe for multi-threaded programs as it
8831 transforms conditional memory writes into unconditional memory writes.
8834 for (i = 0; i < N; i++)
8840 for (i = 0; i < N; i++)
8841 A[i] = cond ? expr : A[i];
8843 potentially producing data races.
8845 @item -ftree-loop-distribution
8846 @opindex ftree-loop-distribution
8847 Perform loop distribution. This flag can improve cache performance on
8848 big loop bodies and allow further loop optimizations, like
8849 parallelization or vectorization, to take place. For example, the loop
8866 @item -ftree-loop-distribute-patterns
8867 @opindex ftree-loop-distribute-patterns
8868 Perform loop distribution of patterns that can be code generated with
8869 calls to a library. This flag is enabled by default at @option{-O3}.
8871 This pass distributes the initialization loops and generates a call to
8872 memset zero. For example, the loop
8888 and the initialization loop is transformed into a call to memset zero.
8890 @item -ftree-loop-im
8891 @opindex ftree-loop-im
8892 Perform loop invariant motion on trees. This pass moves only invariants that
8893 are hard to handle at RTL level (function calls, operations that expand to
8894 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
8895 operands of conditions that are invariant out of the loop, so that we can use
8896 just trivial invariantness analysis in loop unswitching. The pass also includes
8899 @item -ftree-loop-ivcanon
8900 @opindex ftree-loop-ivcanon
8901 Create a canonical counter for number of iterations in loops for which
8902 determining number of iterations requires complicated analysis. Later
8903 optimizations then may determine the number easily. Useful especially
8904 in connection with unrolling.
8908 Perform induction variable optimizations (strength reduction, induction
8909 variable merging and induction variable elimination) on trees.
8911 @item -ftree-parallelize-loops=n
8912 @opindex ftree-parallelize-loops
8913 Parallelize loops, i.e., split their iteration space to run in n threads.
8914 This is only possible for loops whose iterations are independent
8915 and can be arbitrarily reordered. The optimization is only
8916 profitable on multiprocessor machines, for loops that are CPU-intensive,
8917 rather than constrained e.g.@: by memory bandwidth. This option
8918 implies @option{-pthread}, and thus is only supported on targets
8919 that have support for @option{-pthread}.
8923 Perform function-local points-to analysis on trees. This flag is
8924 enabled by default at @option{-O} and higher.
8928 Perform scalar replacement of aggregates. This pass replaces structure
8929 references with scalars to prevent committing structures to memory too
8930 early. This flag is enabled by default at @option{-O} and higher.
8932 @item -ftree-copyrename
8933 @opindex ftree-copyrename
8934 Perform copy renaming on trees. This pass attempts to rename compiler
8935 temporaries to other variables at copy locations, usually resulting in
8936 variable names which more closely resemble the original variables. This flag
8937 is enabled by default at @option{-O} and higher.
8939 @item -ftree-coalesce-inlined-vars
8940 @opindex ftree-coalesce-inlined-vars
8941 Tell the copyrename pass (see @option{-ftree-copyrename}) to attempt to
8942 combine small user-defined variables too, but only if they are inlined
8943 from other functions. It is a more limited form of
8944 @option{-ftree-coalesce-vars}. This may harm debug information of such
8945 inlined variables, but it keeps variables of the inlined-into
8946 function apart from each other, such that they are more likely to
8947 contain the expected values in a debugging session.
8949 @item -ftree-coalesce-vars
8950 @opindex ftree-coalesce-vars
8951 Tell the copyrename pass (see @option{-ftree-copyrename}) to attempt to
8952 combine small user-defined variables too, instead of just compiler
8953 temporaries. This may severely limit the ability to debug an optimized
8954 program compiled with @option{-fno-var-tracking-assignments}. In the
8955 negated form, this flag prevents SSA coalescing of user variables,
8956 including inlined ones. This option is enabled by default.
8960 Perform temporary expression replacement during the SSA->normal phase. Single
8961 use/single def temporaries are replaced at their use location with their
8962 defining expression. This results in non-GIMPLE code, but gives the expanders
8963 much more complex trees to work on resulting in better RTL generation. This is
8964 enabled by default at @option{-O} and higher.
8968 Perform straight-line strength reduction on trees. This recognizes related
8969 expressions involving multiplications and replaces them by less expensive
8970 calculations when possible. This is enabled by default at @option{-O} and
8973 @item -ftree-vectorize
8974 @opindex ftree-vectorize
8975 Perform vectorization on trees. This flag enables @option{-ftree-loop-vectorize}
8976 and @option{-ftree-slp-vectorize} if not explicitly specified.
8978 @item -ftree-loop-vectorize
8979 @opindex ftree-loop-vectorize
8980 Perform loop vectorization on trees. This flag is enabled by default at
8981 @option{-O3} and when @option{-ftree-vectorize} is enabled.
8983 @item -ftree-slp-vectorize
8984 @opindex ftree-slp-vectorize
8985 Perform basic block vectorization on trees. This flag is enabled by default at
8986 @option{-O3} and when @option{-ftree-vectorize} is enabled.
8988 @item -fvect-cost-model=@var{model}
8989 @opindex fvect-cost-model
8990 Alter the cost model used for vectorization. The @var{model} argument
8991 should be one of @samp{unlimited}, @samp{dynamic} or @samp{cheap}.
8992 With the @samp{unlimited} model the vectorized code-path is assumed
8993 to be profitable while with the @samp{dynamic} model a runtime check
8994 guards the vectorized code-path to enable it only for iteration
8995 counts that will likely execute faster than when executing the original
8996 scalar loop. The @samp{cheap} model disables vectorization of
8997 loops where doing so would be cost prohibitive for example due to
8998 required runtime checks for data dependence or alignment but otherwise
8999 is equal to the @samp{dynamic} model.
9000 The default cost model depends on other optimization flags and is
9001 either @samp{dynamic} or @samp{cheap}.
9003 @item -fsimd-cost-model=@var{model}
9004 @opindex fsimd-cost-model
9005 Alter the cost model used for vectorization of loops marked with the OpenMP
9006 or Cilk Plus simd directive. The @var{model} argument should be one of
9007 @samp{unlimited}, @samp{dynamic}, @samp{cheap}. All values of @var{model}
9008 have the same meaning as described in @option{-fvect-cost-model} and by
9009 default a cost model defined with @option{-fvect-cost-model} is used.
9013 Perform Value Range Propagation on trees. This is similar to the
9014 constant propagation pass, but instead of values, ranges of values are
9015 propagated. This allows the optimizers to remove unnecessary range
9016 checks like array bound checks and null pointer checks. This is
9017 enabled by default at @option{-O2} and higher. Null pointer check
9018 elimination is only done if @option{-fdelete-null-pointer-checks} is
9021 @item -fsplit-ivs-in-unroller
9022 @opindex fsplit-ivs-in-unroller
9023 Enables expression of values of induction variables in later iterations
9024 of the unrolled loop using the value in the first iteration. This breaks
9025 long dependency chains, thus improving efficiency of the scheduling passes.
9027 A combination of @option{-fweb} and CSE is often sufficient to obtain the
9028 same effect. However, that is not reliable in cases where the loop body
9029 is more complicated than a single basic block. It also does not work at all
9030 on some architectures due to restrictions in the CSE pass.
9032 This optimization is enabled by default.
9034 @item -fvariable-expansion-in-unroller
9035 @opindex fvariable-expansion-in-unroller
9036 With this option, the compiler creates multiple copies of some
9037 local variables when unrolling a loop, which can result in superior code.
9039 @item -fpartial-inlining
9040 @opindex fpartial-inlining
9041 Inline parts of functions. This option has any effect only
9042 when inlining itself is turned on by the @option{-finline-functions}
9043 or @option{-finline-small-functions} options.
9045 Enabled at level @option{-O2}.
9047 @item -fpredictive-commoning
9048 @opindex fpredictive-commoning
9049 Perform predictive commoning optimization, i.e., reusing computations
9050 (especially memory loads and stores) performed in previous
9051 iterations of loops.
9053 This option is enabled at level @option{-O3}.
9055 @item -fprefetch-loop-arrays
9056 @opindex fprefetch-loop-arrays
9057 If supported by the target machine, generate instructions to prefetch
9058 memory to improve the performance of loops that access large arrays.
9060 This option may generate better or worse code; results are highly
9061 dependent on the structure of loops within the source code.
9063 Disabled at level @option{-Os}.
9066 @itemx -fno-peephole2
9067 @opindex fno-peephole
9068 @opindex fno-peephole2
9069 Disable any machine-specific peephole optimizations. The difference
9070 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
9071 are implemented in the compiler; some targets use one, some use the
9072 other, a few use both.
9074 @option{-fpeephole} is enabled by default.
9075 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
9077 @item -fno-guess-branch-probability
9078 @opindex fno-guess-branch-probability
9079 Do not guess branch probabilities using heuristics.
9081 GCC uses heuristics to guess branch probabilities if they are
9082 not provided by profiling feedback (@option{-fprofile-arcs}). These
9083 heuristics are based on the control flow graph. If some branch probabilities
9084 are specified by @code{__builtin_expect}, then the heuristics are
9085 used to guess branch probabilities for the rest of the control flow graph,
9086 taking the @code{__builtin_expect} info into account. The interactions
9087 between the heuristics and @code{__builtin_expect} can be complex, and in
9088 some cases, it may be useful to disable the heuristics so that the effects
9089 of @code{__builtin_expect} are easier to understand.
9091 The default is @option{-fguess-branch-probability} at levels
9092 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
9094 @item -freorder-blocks
9095 @opindex freorder-blocks
9096 Reorder basic blocks in the compiled function in order to reduce number of
9097 taken branches and improve code locality.
9099 Enabled at levels @option{-O2}, @option{-O3}.
9101 @item -freorder-blocks-and-partition
9102 @opindex freorder-blocks-and-partition
9103 In addition to reordering basic blocks in the compiled function, in order
9104 to reduce number of taken branches, partitions hot and cold basic blocks
9105 into separate sections of the assembly and .o files, to improve
9106 paging and cache locality performance.
9108 This optimization is automatically turned off in the presence of
9109 exception handling, for linkonce sections, for functions with a user-defined
9110 section attribute and on any architecture that does not support named
9113 Enabled for x86 at levels @option{-O2}, @option{-O3}.
9115 @item -freorder-functions
9116 @opindex freorder-functions
9117 Reorder functions in the object file in order to
9118 improve code locality. This is implemented by using special
9119 subsections @code{.text.hot} for most frequently executed functions and
9120 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
9121 the linker so object file format must support named sections and linker must
9122 place them in a reasonable way.
9124 Also profile feedback must be available to make this option effective. See
9125 @option{-fprofile-arcs} for details.
9127 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
9129 @item -fstrict-aliasing
9130 @opindex fstrict-aliasing
9131 Allow the compiler to assume the strictest aliasing rules applicable to
9132 the language being compiled. For C (and C++), this activates
9133 optimizations based on the type of expressions. In particular, an
9134 object of one type is assumed never to reside at the same address as an
9135 object of a different type, unless the types are almost the same. For
9136 example, an @code{unsigned int} can alias an @code{int}, but not a
9137 @code{void*} or a @code{double}. A character type may alias any other
9140 @anchor{Type-punning}Pay special attention to code like this:
9153 The practice of reading from a different union member than the one most
9154 recently written to (called ``type-punning'') is common. Even with
9155 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
9156 is accessed through the union type. So, the code above works as
9157 expected. @xref{Structures unions enumerations and bit-fields
9158 implementation}. However, this code might not:
9169 Similarly, access by taking the address, casting the resulting pointer
9170 and dereferencing the result has undefined behavior, even if the cast
9171 uses a union type, e.g.:
9175 return ((union a_union *) &d)->i;
9179 The @option{-fstrict-aliasing} option is enabled at levels
9180 @option{-O2}, @option{-O3}, @option{-Os}.
9182 @item -fstrict-overflow
9183 @opindex fstrict-overflow
9184 Allow the compiler to assume strict signed overflow rules, depending
9185 on the language being compiled. For C (and C++) this means that
9186 overflow when doing arithmetic with signed numbers is undefined, which
9187 means that the compiler may assume that it does not happen. This
9188 permits various optimizations. For example, the compiler assumes
9189 that an expression like @code{i + 10 > i} is always true for
9190 signed @code{i}. This assumption is only valid if signed overflow is
9191 undefined, as the expression is false if @code{i + 10} overflows when
9192 using twos complement arithmetic. When this option is in effect any
9193 attempt to determine whether an operation on signed numbers
9194 overflows must be written carefully to not actually involve overflow.
9196 This option also allows the compiler to assume strict pointer
9197 semantics: given a pointer to an object, if adding an offset to that
9198 pointer does not produce a pointer to the same object, the addition is
9199 undefined. This permits the compiler to conclude that @code{p + u >
9200 p} is always true for a pointer @code{p} and unsigned integer
9201 @code{u}. This assumption is only valid because pointer wraparound is
9202 undefined, as the expression is false if @code{p + u} overflows using
9203 twos complement arithmetic.
9205 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
9206 that integer signed overflow is fully defined: it wraps. When
9207 @option{-fwrapv} is used, there is no difference between
9208 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
9209 integers. With @option{-fwrapv} certain types of overflow are
9210 permitted. For example, if the compiler gets an overflow when doing
9211 arithmetic on constants, the overflowed value can still be used with
9212 @option{-fwrapv}, but not otherwise.
9214 The @option{-fstrict-overflow} option is enabled at levels
9215 @option{-O2}, @option{-O3}, @option{-Os}.
9217 @item -falign-functions
9218 @itemx -falign-functions=@var{n}
9219 @opindex falign-functions
9220 Align the start of functions to the next power-of-two greater than
9221 @var{n}, skipping up to @var{n} bytes. For instance,
9222 @option{-falign-functions=32} aligns functions to the next 32-byte
9223 boundary, but @option{-falign-functions=24} aligns to the next
9224 32-byte boundary only if this can be done by skipping 23 bytes or less.
9226 @option{-fno-align-functions} and @option{-falign-functions=1} are
9227 equivalent and mean that functions are not aligned.
9229 Some assemblers only support this flag when @var{n} is a power of two;
9230 in that case, it is rounded up.
9232 If @var{n} is not specified or is zero, use a machine-dependent default.
9234 Enabled at levels @option{-O2}, @option{-O3}.
9236 @item -falign-labels
9237 @itemx -falign-labels=@var{n}
9238 @opindex falign-labels
9239 Align all branch targets to a power-of-two boundary, skipping up to
9240 @var{n} bytes like @option{-falign-functions}. This option can easily
9241 make code slower, because it must insert dummy operations for when the
9242 branch target is reached in the usual flow of the code.
9244 @option{-fno-align-labels} and @option{-falign-labels=1} are
9245 equivalent and mean that labels are not aligned.
9247 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
9248 are greater than this value, then their values are used instead.
9250 If @var{n} is not specified or is zero, use a machine-dependent default
9251 which is very likely to be @samp{1}, meaning no alignment.
9253 Enabled at levels @option{-O2}, @option{-O3}.
9256 @itemx -falign-loops=@var{n}
9257 @opindex falign-loops
9258 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
9259 like @option{-falign-functions}. If the loops are
9260 executed many times, this makes up for any execution of the dummy
9263 @option{-fno-align-loops} and @option{-falign-loops=1} are
9264 equivalent and mean that loops are not aligned.
9266 If @var{n} is not specified or is zero, use a machine-dependent default.
9268 Enabled at levels @option{-O2}, @option{-O3}.
9271 @itemx -falign-jumps=@var{n}
9272 @opindex falign-jumps
9273 Align branch targets to a power-of-two boundary, for branch targets
9274 where the targets can only be reached by jumping, skipping up to @var{n}
9275 bytes like @option{-falign-functions}. In this case, no dummy operations
9278 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
9279 equivalent and mean that loops are not aligned.
9281 If @var{n} is not specified or is zero, use a machine-dependent default.
9283 Enabled at levels @option{-O2}, @option{-O3}.
9285 @item -funit-at-a-time
9286 @opindex funit-at-a-time
9287 This option is left for compatibility reasons. @option{-funit-at-a-time}
9288 has no effect, while @option{-fno-unit-at-a-time} implies
9289 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
9293 @item -fno-toplevel-reorder
9294 @opindex fno-toplevel-reorder
9295 Do not reorder top-level functions, variables, and @code{asm}
9296 statements. Output them in the same order that they appear in the
9297 input file. When this option is used, unreferenced static variables
9298 are not removed. This option is intended to support existing code
9299 that relies on a particular ordering. For new code, it is better to
9300 use attributes when possible.
9302 Enabled at level @option{-O0}. When disabled explicitly, it also implies
9303 @option{-fno-section-anchors}, which is otherwise enabled at @option{-O0} on some
9308 Constructs webs as commonly used for register allocation purposes and assign
9309 each web individual pseudo register. This allows the register allocation pass
9310 to operate on pseudos directly, but also strengthens several other optimization
9311 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
9312 however, make debugging impossible, since variables no longer stay in a
9315 Enabled by default with @option{-funroll-loops}.
9317 @item -fwhole-program
9318 @opindex fwhole-program
9319 Assume that the current compilation unit represents the whole program being
9320 compiled. All public functions and variables with the exception of @code{main}
9321 and those merged by attribute @code{externally_visible} become static functions
9322 and in effect are optimized more aggressively by interprocedural optimizers.
9324 This option should not be used in combination with @option{-flto}.
9325 Instead relying on a linker plugin should provide safer and more precise
9328 @item -flto[=@var{n}]
9330 This option runs the standard link-time optimizer. When invoked
9331 with source code, it generates GIMPLE (one of GCC's internal
9332 representations) and writes it to special ELF sections in the object
9333 file. When the object files are linked together, all the function
9334 bodies are read from these ELF sections and instantiated as if they
9335 had been part of the same translation unit.
9337 To use the link-time optimizer, @option{-flto} and optimization
9338 options should be specified at compile time and during the final link.
9342 gcc -c -O2 -flto foo.c
9343 gcc -c -O2 -flto bar.c
9344 gcc -o myprog -flto -O2 foo.o bar.o
9347 The first two invocations to GCC save a bytecode representation
9348 of GIMPLE into special ELF sections inside @file{foo.o} and
9349 @file{bar.o}. The final invocation reads the GIMPLE bytecode from
9350 @file{foo.o} and @file{bar.o}, merges the two files into a single
9351 internal image, and compiles the result as usual. Since both
9352 @file{foo.o} and @file{bar.o} are merged into a single image, this
9353 causes all the interprocedural analyses and optimizations in GCC to
9354 work across the two files as if they were a single one. This means,
9355 for example, that the inliner is able to inline functions in
9356 @file{bar.o} into functions in @file{foo.o} and vice-versa.
9358 Another (simpler) way to enable link-time optimization is:
9361 gcc -o myprog -flto -O2 foo.c bar.c
9364 The above generates bytecode for @file{foo.c} and @file{bar.c},
9365 merges them together into a single GIMPLE representation and optimizes
9366 them as usual to produce @file{myprog}.
9368 The only important thing to keep in mind is that to enable link-time
9369 optimizations you need to use the GCC driver to perform the link-step.
9370 GCC then automatically performs link-time optimization if any of the
9371 objects involved were compiled with the @option{-flto} command-line option.
9373 should specify the optimization options to be used for link-time
9374 optimization though GCC tries to be clever at guessing an
9375 optimization level to use from the options used at compile-time
9376 if you fail to specify one at link-time. You can always override
9377 the automatic decision to do link-time optimization at link-time
9378 by passing @option{-fno-lto} to the link command.
9380 To make whole program optimization effective, it is necessary to make
9381 certain whole program assumptions. The compiler needs to know
9382 what functions and variables can be accessed by libraries and runtime
9383 outside of the link-time optimized unit. When supported by the linker,
9384 the linker plugin (see @option{-fuse-linker-plugin}) passes information
9385 to the compiler about used and externally visible symbols. When
9386 the linker plugin is not available, @option{-fwhole-program} should be
9387 used to allow the compiler to make these assumptions, which leads
9388 to more aggressive optimization decisions.
9390 When @option{-fuse-linker-plugin} is not enabled then, when a file is
9391 compiled with @option{-flto}, the generated object file is larger than
9392 a regular object file because it contains GIMPLE bytecodes and the usual
9393 final code (see @option{-ffat-lto-objects}. This means that
9394 object files with LTO information can be linked as normal object
9395 files; if @option{-fno-lto} is passed to the linker, no
9396 interprocedural optimizations are applied. Note that when
9397 @option{-fno-fat-lto-objects} is enabled the compile-stage is faster
9398 but you cannot perform a regular, non-LTO link on them.
9400 Additionally, the optimization flags used to compile individual files
9401 are not necessarily related to those used at link time. For instance,
9404 gcc -c -O0 -ffat-lto-objects -flto foo.c
9405 gcc -c -O0 -ffat-lto-objects -flto bar.c
9406 gcc -o myprog -O3 foo.o bar.o
9409 This produces individual object files with unoptimized assembler
9410 code, but the resulting binary @file{myprog} is optimized at
9411 @option{-O3}. If, instead, the final binary is generated with
9412 @option{-fno-lto}, then @file{myprog} is not optimized.
9414 When producing the final binary, GCC only
9415 applies link-time optimizations to those files that contain bytecode.
9416 Therefore, you can mix and match object files and libraries with
9417 GIMPLE bytecodes and final object code. GCC automatically selects
9418 which files to optimize in LTO mode and which files to link without
9421 There are some code generation flags preserved by GCC when
9422 generating bytecodes, as they need to be used during the final link
9423 stage. Generally options specified at link-time override those
9424 specified at compile-time.
9426 If you do not specify an optimization level option @option{-O} at
9427 link-time then GCC computes one based on the optimization levels
9428 used when compiling the object files. The highest optimization
9431 Currently, the following options and their setting are take from
9432 the first object file that explicitely specified it:
9433 @option{-fPIC}, @option{-fpic}, @option{-fpie}, @option{-fcommon},
9434 @option{-fexceptions}, @option{-fnon-call-exceptions}, @option{-fgnu-tm}
9435 and all the @option{-m} target flags.
9437 Certain ABI changing flags are required to match in all compilation-units
9438 and trying to override this at link-time with a conflicting value
9439 is ignored. This includes options such as @option{-freg-struct-return}
9440 and @option{-fpcc-struct-return}.
9442 Other options such as @option{-ffp-contract}, @option{-fno-strict-overflow},
9443 @option{-fwrapv}, @option{-fno-trapv} or @option{-fno-strict-aliasing}
9444 are passed through to the link stage and merged conservatively for
9445 conflicting translation units. Specifically
9446 @option{-fno-strict-overflow}, @option{-fwrapv} and @option{-fno-trapv} take
9447 precedence and for example @option{-ffp-contract=off} takes precedence
9448 over @option{-ffp-contract=fast}. You can override them at linke-time.
9450 It is recommended that you compile all the files participating in the
9451 same link with the same options and also specify those options at
9454 If LTO encounters objects with C linkage declared with incompatible
9455 types in separate translation units to be linked together (undefined
9456 behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
9457 issued. The behavior is still undefined at run time. Similar
9458 diagnostics may be raised for other languages.
9460 Another feature of LTO is that it is possible to apply interprocedural
9461 optimizations on files written in different languages:
9466 gfortran -c -flto baz.f90
9467 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
9470 Notice that the final link is done with @command{g++} to get the C++
9471 runtime libraries and @option{-lgfortran} is added to get the Fortran
9472 runtime libraries. In general, when mixing languages in LTO mode, you
9473 should use the same link command options as when mixing languages in a
9474 regular (non-LTO) compilation.
9476 If object files containing GIMPLE bytecode are stored in a library archive, say
9477 @file{libfoo.a}, it is possible to extract and use them in an LTO link if you
9478 are using a linker with plugin support. To create static libraries suitable
9479 for LTO, use @command{gcc-ar} and @command{gcc-ranlib} instead of @command{ar}
9480 and @command{ranlib};
9481 to show the symbols of object files with GIMPLE bytecode, use
9482 @command{gcc-nm}. Those commands require that @command{ar}, @command{ranlib}
9483 and @command{nm} have been compiled with plugin support. At link time, use the the
9484 flag @option{-fuse-linker-plugin} to ensure that the library participates in
9485 the LTO optimization process:
9488 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
9491 With the linker plugin enabled, the linker extracts the needed
9492 GIMPLE files from @file{libfoo.a} and passes them on to the running GCC
9493 to make them part of the aggregated GIMPLE image to be optimized.
9495 If you are not using a linker with plugin support and/or do not
9496 enable the linker plugin, then the objects inside @file{libfoo.a}
9497 are extracted and linked as usual, but they do not participate
9498 in the LTO optimization process. In order to make a static library suitable
9499 for both LTO optimization and usual linkage, compile its object files with
9500 @option{-flto} @option{-ffat-lto-objects}.
9502 Link-time optimizations do not require the presence of the whole program to
9503 operate. If the program does not require any symbols to be exported, it is
9504 possible to combine @option{-flto} and @option{-fwhole-program} to allow
9505 the interprocedural optimizers to use more aggressive assumptions which may
9506 lead to improved optimization opportunities.
9507 Use of @option{-fwhole-program} is not needed when linker plugin is
9508 active (see @option{-fuse-linker-plugin}).
9510 The current implementation of LTO makes no
9511 attempt to generate bytecode that is portable between different
9512 types of hosts. The bytecode files are versioned and there is a
9513 strict version check, so bytecode files generated in one version of
9514 GCC do not work with an older or newer version of GCC.
9516 Link-time optimization does not work well with generation of debugging
9517 information. Combining @option{-flto} with
9518 @option{-g} is currently experimental and expected to produce unexpected
9521 If you specify the optional @var{n}, the optimization and code
9522 generation done at link time is executed in parallel using @var{n}
9523 parallel jobs by utilizing an installed @command{make} program. The
9524 environment variable @env{MAKE} may be used to override the program
9525 used. The default value for @var{n} is 1.
9527 You can also specify @option{-flto=jobserver} to use GNU make's
9528 job server mode to determine the number of parallel jobs. This
9529 is useful when the Makefile calling GCC is already executing in parallel.
9530 You must prepend a @samp{+} to the command recipe in the parent Makefile
9531 for this to work. This option likely only works if @env{MAKE} is
9534 @item -flto-partition=@var{alg}
9535 @opindex flto-partition
9536 Specify the partitioning algorithm used by the link-time optimizer.
9537 The value is either @samp{1to1} to specify a partitioning mirroring
9538 the original source files or @samp{balanced} to specify partitioning
9539 into equally sized chunks (whenever possible) or @samp{max} to create
9540 new partition for every symbol where possible. Specifying @samp{none}
9541 as an algorithm disables partitioning and streaming completely.
9542 The default value is @samp{balanced}. While @samp{1to1} can be used
9543 as an workaround for various code ordering issues, the @samp{max}
9544 partitioning is intended for internal testing only.
9545 The value @samp{one} specifies that exactly one partition should be
9546 used while the value @samp{none} bypasses partitioning and executes
9547 the link-time optimization step directly from the WPA phase.
9549 @item -flto-odr-type-merging
9550 @opindex flto-odr-type-merging
9551 Enable streaming of mangled types names of C++ types and their unification
9552 at linktime. This increases size of LTO object files, but enable
9553 diagnostics about One Definition Rule violations.
9555 @item -flto-compression-level=@var{n}
9556 @opindex flto-compression-level
9557 This option specifies the level of compression used for intermediate
9558 language written to LTO object files, and is only meaningful in
9559 conjunction with LTO mode (@option{-flto}). Valid
9560 values are 0 (no compression) to 9 (maximum compression). Values
9561 outside this range are clamped to either 0 or 9. If the option is not
9562 given, a default balanced compression setting is used.
9565 @opindex flto-report
9566 Prints a report with internal details on the workings of the link-time
9567 optimizer. The contents of this report vary from version to version.
9568 It is meant to be useful to GCC developers when processing object
9569 files in LTO mode (via @option{-flto}).
9571 Disabled by default.
9573 @item -flto-report-wpa
9574 @opindex flto-report-wpa
9575 Like @option{-flto-report}, but only print for the WPA phase of Link
9578 @item -fuse-linker-plugin
9579 @opindex fuse-linker-plugin
9580 Enables the use of a linker plugin during link-time optimization. This
9581 option relies on plugin support in the linker, which is available in gold
9582 or in GNU ld 2.21 or newer.
9584 This option enables the extraction of object files with GIMPLE bytecode out
9585 of library archives. This improves the quality of optimization by exposing
9586 more code to the link-time optimizer. This information specifies what
9587 symbols can be accessed externally (by non-LTO object or during dynamic
9588 linking). Resulting code quality improvements on binaries (and shared
9589 libraries that use hidden visibility) are similar to @option{-fwhole-program}.
9590 See @option{-flto} for a description of the effect of this flag and how to
9593 This option is enabled by default when LTO support in GCC is enabled
9594 and GCC was configured for use with
9595 a linker supporting plugins (GNU ld 2.21 or newer or gold).
9597 @item -ffat-lto-objects
9598 @opindex ffat-lto-objects
9599 Fat LTO objects are object files that contain both the intermediate language
9600 and the object code. This makes them usable for both LTO linking and normal
9601 linking. This option is effective only when compiling with @option{-flto}
9602 and is ignored at link time.
9604 @option{-fno-fat-lto-objects} improves compilation time over plain LTO, but
9605 requires the complete toolchain to be aware of LTO. It requires a linker with
9606 linker plugin support for basic functionality. Additionally,
9607 @command{nm}, @command{ar} and @command{ranlib}
9608 need to support linker plugins to allow a full-featured build environment
9609 (capable of building static libraries etc). GCC provides the @command{gcc-ar},
9610 @command{gcc-nm}, @command{gcc-ranlib} wrappers to pass the right options
9611 to these tools. With non fat LTO makefiles need to be modified to use them.
9613 The default is @option{-fno-fat-lto-objects} on targets with linker plugin
9616 @item -fcompare-elim
9617 @opindex fcompare-elim
9618 After register allocation and post-register allocation instruction splitting,
9619 identify arithmetic instructions that compute processor flags similar to a
9620 comparison operation based on that arithmetic. If possible, eliminate the
9621 explicit comparison operation.
9623 This pass only applies to certain targets that cannot explicitly represent
9624 the comparison operation before register allocation is complete.
9626 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
9628 @item -fcprop-registers
9629 @opindex fcprop-registers
9630 After register allocation and post-register allocation instruction splitting,
9631 perform a copy-propagation pass to try to reduce scheduling dependencies
9632 and occasionally eliminate the copy.
9634 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
9636 @item -fprofile-correction
9637 @opindex fprofile-correction
9638 Profiles collected using an instrumented binary for multi-threaded programs may
9639 be inconsistent due to missed counter updates. When this option is specified,
9640 GCC uses heuristics to correct or smooth out such inconsistencies. By
9641 default, GCC emits an error message when an inconsistent profile is detected.
9643 @item -fprofile-dir=@var{path}
9644 @opindex fprofile-dir
9646 Set the directory to search for the profile data files in to @var{path}.
9647 This option affects only the profile data generated by
9648 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
9649 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
9650 and its related options. Both absolute and relative paths can be used.
9651 By default, GCC uses the current directory as @var{path}, thus the
9652 profile data file appears in the same directory as the object file.
9654 @item -fprofile-generate
9655 @itemx -fprofile-generate=@var{path}
9656 @opindex fprofile-generate
9658 Enable options usually used for instrumenting application to produce
9659 profile useful for later recompilation with profile feedback based
9660 optimization. You must use @option{-fprofile-generate} both when
9661 compiling and when linking your program.
9663 The following options are enabled: @option{-fprofile-arcs}, @option{-fprofile-values}, @option{-fvpt}.
9665 If @var{path} is specified, GCC looks at the @var{path} to find
9666 the profile feedback data files. See @option{-fprofile-dir}.
9669 @itemx -fprofile-use=@var{path}
9670 @opindex fprofile-use
9671 Enable profile feedback-directed optimizations,
9672 and the following optimizations
9673 which are generally profitable only with profile feedback available:
9674 @option{-fbranch-probabilities}, @option{-fvpt},
9675 @option{-funroll-loops}, @option{-fpeel-loops}, @option{-ftracer},
9676 @option{-ftree-vectorize}, and @option{ftree-loop-distribute-patterns}.
9678 By default, GCC emits an error message if the feedback profiles do not
9679 match the source code. This error can be turned into a warning by using
9680 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
9683 If @var{path} is specified, GCC looks at the @var{path} to find
9684 the profile feedback data files. See @option{-fprofile-dir}.
9686 @item -fauto-profile
9687 @itemx -fauto-profile=@var{path}
9688 @opindex fauto-profile
9689 Enable sampling-based feedback-directed optimizations,
9690 and the following optimizations
9691 which are generally profitable only with profile feedback available:
9692 @option{-fbranch-probabilities}, @option{-fvpt},
9693 @option{-funroll-loops}, @option{-fpeel-loops}, @option{-ftracer},
9694 @option{-ftree-vectorize},
9695 @option{-finline-functions}, @option{-fipa-cp}, @option{-fipa-cp-clone},
9696 @option{-fpredictive-commoning}, @option{-funswitch-loops},
9697 @option{-fgcse-after-reload}, and @option{-ftree-loop-distribute-patterns}.
9699 @var{path} is the name of a file containing AutoFDO profile information.
9700 If omitted, it defaults to @file{fbdata.afdo} in the current directory.
9702 Producing an AutoFDO profile data file requires running your program
9703 with the @command{perf} utility on a supported GNU/Linux target system.
9704 For more information, see @uref{https://perf.wiki.kernel.org/}.
9708 perf record -e br_inst_retired:near_taken -b -o perf.data \
9712 Then use the @command{create_gcov} tool to convert the raw profile data
9713 to a format that can be used by GCC.@ You must also supply the
9714 unstripped binary for your program to this tool.
9715 See @uref{https://github.com/google/autofdo}.
9719 create_gcov --binary=your_program.unstripped --profile=perf.data \
9724 The following options control compiler behavior regarding floating-point
9725 arithmetic. These options trade off between speed and
9726 correctness. All must be specifically enabled.
9730 @opindex ffloat-store
9731 Do not store floating-point variables in registers, and inhibit other
9732 options that might change whether a floating-point value is taken from a
9735 @cindex floating-point precision
9736 This option prevents undesirable excess precision on machines such as
9737 the 68000 where the floating registers (of the 68881) keep more
9738 precision than a @code{double} is supposed to have. Similarly for the
9739 x86 architecture. For most programs, the excess precision does only
9740 good, but a few programs rely on the precise definition of IEEE floating
9741 point. Use @option{-ffloat-store} for such programs, after modifying
9742 them to store all pertinent intermediate computations into variables.
9744 @item -fexcess-precision=@var{style}
9745 @opindex fexcess-precision
9746 This option allows further control over excess precision on machines
9747 where floating-point registers have more precision than the IEEE
9748 @code{float} and @code{double} types and the processor does not
9749 support operations rounding to those types. By default,
9750 @option{-fexcess-precision=fast} is in effect; this means that
9751 operations are carried out in the precision of the registers and that
9752 it is unpredictable when rounding to the types specified in the source
9753 code takes place. When compiling C, if
9754 @option{-fexcess-precision=standard} is specified then excess
9755 precision follows the rules specified in ISO C99; in particular,
9756 both casts and assignments cause values to be rounded to their
9757 semantic types (whereas @option{-ffloat-store} only affects
9758 assignments). This option is enabled by default for C if a strict
9759 conformance option such as @option{-std=c99} is used.
9762 @option{-fexcess-precision=standard} is not implemented for languages
9763 other than C, and has no effect if
9764 @option{-funsafe-math-optimizations} or @option{-ffast-math} is
9765 specified. On the x86, it also has no effect if @option{-mfpmath=sse}
9766 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
9767 semantics apply without excess precision, and in the latter, rounding
9772 Sets the options @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
9773 @option{-ffinite-math-only}, @option{-fno-rounding-math},
9774 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
9776 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
9778 This option is not turned on by any @option{-O} option besides
9779 @option{-Ofast} since it can result in incorrect output for programs
9780 that depend on an exact implementation of IEEE or ISO rules/specifications
9781 for math functions. It may, however, yield faster code for programs
9782 that do not require the guarantees of these specifications.
9784 @item -fno-math-errno
9785 @opindex fno-math-errno
9786 Do not set @code{errno} after calling math functions that are executed
9787 with a single instruction, e.g., @code{sqrt}. A program that relies on
9788 IEEE exceptions for math error handling may want to use this flag
9789 for speed while maintaining IEEE arithmetic compatibility.
9791 This option is not turned on by any @option{-O} option since
9792 it can result in incorrect output for programs that depend on
9793 an exact implementation of IEEE or ISO rules/specifications for
9794 math functions. It may, however, yield faster code for programs
9795 that do not require the guarantees of these specifications.
9797 The default is @option{-fmath-errno}.
9799 On Darwin systems, the math library never sets @code{errno}. There is
9800 therefore no reason for the compiler to consider the possibility that
9801 it might, and @option{-fno-math-errno} is the default.
9803 @item -funsafe-math-optimizations
9804 @opindex funsafe-math-optimizations
9806 Allow optimizations for floating-point arithmetic that (a) assume
9807 that arguments and results are valid and (b) may violate IEEE or
9808 ANSI standards. When used at link-time, it may include libraries
9809 or startup files that change the default FPU control word or other
9810 similar optimizations.
9812 This option is not turned on by any @option{-O} option since
9813 it can result in incorrect output for programs that depend on
9814 an exact implementation of IEEE or ISO rules/specifications for
9815 math functions. It may, however, yield faster code for programs
9816 that do not require the guarantees of these specifications.
9817 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
9818 @option{-fassociative-math} and @option{-freciprocal-math}.
9820 The default is @option{-fno-unsafe-math-optimizations}.
9822 @item -fassociative-math
9823 @opindex fassociative-math
9825 Allow re-association of operands in series of floating-point operations.
9826 This violates the ISO C and C++ language standard by possibly changing
9827 computation result. NOTE: re-ordering may change the sign of zero as
9828 well as ignore NaNs and inhibit or create underflow or overflow (and
9829 thus cannot be used on code that relies on rounding behavior like
9830 @code{(x + 2**52) - 2**52}. May also reorder floating-point comparisons
9831 and thus may not be used when ordered comparisons are required.
9832 This option requires that both @option{-fno-signed-zeros} and
9833 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
9834 much sense with @option{-frounding-math}. For Fortran the option
9835 is automatically enabled when both @option{-fno-signed-zeros} and
9836 @option{-fno-trapping-math} are in effect.
9838 The default is @option{-fno-associative-math}.
9840 @item -freciprocal-math
9841 @opindex freciprocal-math
9843 Allow the reciprocal of a value to be used instead of dividing by
9844 the value if this enables optimizations. For example @code{x / y}
9845 can be replaced with @code{x * (1/y)}, which is useful if @code{(1/y)}
9846 is subject to common subexpression elimination. Note that this loses
9847 precision and increases the number of flops operating on the value.
9849 The default is @option{-fno-reciprocal-math}.
9851 @item -ffinite-math-only
9852 @opindex ffinite-math-only
9853 Allow optimizations for floating-point arithmetic that assume
9854 that arguments and results are not NaNs or +-Infs.
9856 This option is not turned on by any @option{-O} option since
9857 it can result in incorrect output for programs that depend on
9858 an exact implementation of IEEE or ISO rules/specifications for
9859 math functions. It may, however, yield faster code for programs
9860 that do not require the guarantees of these specifications.
9862 The default is @option{-fno-finite-math-only}.
9864 @item -fno-signed-zeros
9865 @opindex fno-signed-zeros
9866 Allow optimizations for floating-point arithmetic that ignore the
9867 signedness of zero. IEEE arithmetic specifies the behavior of
9868 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
9869 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
9870 This option implies that the sign of a zero result isn't significant.
9872 The default is @option{-fsigned-zeros}.
9874 @item -fno-trapping-math
9875 @opindex fno-trapping-math
9876 Compile code assuming that floating-point operations cannot generate
9877 user-visible traps. These traps include division by zero, overflow,
9878 underflow, inexact result and invalid operation. This option requires
9879 that @option{-fno-signaling-nans} be in effect. Setting this option may
9880 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
9882 This option should never be turned on by any @option{-O} option since
9883 it can result in incorrect output for programs that depend on
9884 an exact implementation of IEEE or ISO rules/specifications for
9887 The default is @option{-ftrapping-math}.
9889 @item -frounding-math
9890 @opindex frounding-math
9891 Disable transformations and optimizations that assume default floating-point
9892 rounding behavior. This is round-to-zero for all floating point
9893 to integer conversions, and round-to-nearest for all other arithmetic
9894 truncations. This option should be specified for programs that change
9895 the FP rounding mode dynamically, or that may be executed with a
9896 non-default rounding mode. This option disables constant folding of
9897 floating-point expressions at compile time (which may be affected by
9898 rounding mode) and arithmetic transformations that are unsafe in the
9899 presence of sign-dependent rounding modes.
9901 The default is @option{-fno-rounding-math}.
9903 This option is experimental and does not currently guarantee to
9904 disable all GCC optimizations that are affected by rounding mode.
9905 Future versions of GCC may provide finer control of this setting
9906 using C99's @code{FENV_ACCESS} pragma. This command-line option
9907 will be used to specify the default state for @code{FENV_ACCESS}.
9909 @item -fsignaling-nans
9910 @opindex fsignaling-nans
9911 Compile code assuming that IEEE signaling NaNs may generate user-visible
9912 traps during floating-point operations. Setting this option disables
9913 optimizations that may change the number of exceptions visible with
9914 signaling NaNs. This option implies @option{-ftrapping-math}.
9916 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
9919 The default is @option{-fno-signaling-nans}.
9921 This option is experimental and does not currently guarantee to
9922 disable all GCC optimizations that affect signaling NaN behavior.
9924 @item -fsingle-precision-constant
9925 @opindex fsingle-precision-constant
9926 Treat floating-point constants as single precision instead of
9927 implicitly converting them to double-precision constants.
9929 @item -fcx-limited-range
9930 @opindex fcx-limited-range
9931 When enabled, this option states that a range reduction step is not
9932 needed when performing complex division. Also, there is no checking
9933 whether the result of a complex multiplication or division is @code{NaN
9934 + I*NaN}, with an attempt to rescue the situation in that case. The
9935 default is @option{-fno-cx-limited-range}, but is enabled by
9936 @option{-ffast-math}.
9938 This option controls the default setting of the ISO C99
9939 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
9942 @item -fcx-fortran-rules
9943 @opindex fcx-fortran-rules
9944 Complex multiplication and division follow Fortran rules. Range
9945 reduction is done as part of complex division, but there is no checking
9946 whether the result of a complex multiplication or division is @code{NaN
9947 + I*NaN}, with an attempt to rescue the situation in that case.
9949 The default is @option{-fno-cx-fortran-rules}.
9953 The following options control optimizations that may improve
9954 performance, but are not enabled by any @option{-O} options. This
9955 section includes experimental options that may produce broken code.
9958 @item -fbranch-probabilities
9959 @opindex fbranch-probabilities
9960 After running a program compiled with @option{-fprofile-arcs}
9961 (@pxref{Debugging Options,, Options for Debugging Your Program or
9962 @command{gcc}}), you can compile it a second time using
9963 @option{-fbranch-probabilities}, to improve optimizations based on
9964 the number of times each branch was taken. When a program
9965 compiled with @option{-fprofile-arcs} exits, it saves arc execution
9966 counts to a file called @file{@var{sourcename}.gcda} for each source
9967 file. The information in this data file is very dependent on the
9968 structure of the generated code, so you must use the same source code
9969 and the same optimization options for both compilations.
9971 With @option{-fbranch-probabilities}, GCC puts a
9972 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
9973 These can be used to improve optimization. Currently, they are only
9974 used in one place: in @file{reorg.c}, instead of guessing which path a
9975 branch is most likely to take, the @samp{REG_BR_PROB} values are used to
9976 exactly determine which path is taken more often.
9978 @item -fprofile-values
9979 @opindex fprofile-values
9980 If combined with @option{-fprofile-arcs}, it adds code so that some
9981 data about values of expressions in the program is gathered.
9983 With @option{-fbranch-probabilities}, it reads back the data gathered
9984 from profiling values of expressions for usage in optimizations.
9986 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
9988 @item -fprofile-reorder-functions
9989 @opindex fprofile-reorder-functions
9990 Function reordering based on profile instrumentation collects
9991 first time of execution of a function and orders these functions
9994 Enabled with @option{-fprofile-use}.
9998 If combined with @option{-fprofile-arcs}, this option instructs the compiler
9999 to add code to gather information about values of expressions.
10001 With @option{-fbranch-probabilities}, it reads back the data gathered
10002 and actually performs the optimizations based on them.
10003 Currently the optimizations include specialization of division operations
10004 using the knowledge about the value of the denominator.
10006 @item -frename-registers
10007 @opindex frename-registers
10008 Attempt to avoid false dependencies in scheduled code by making use
10009 of registers left over after register allocation. This optimization
10010 most benefits processors with lots of registers. Depending on the
10011 debug information format adopted by the target, however, it can
10012 make debugging impossible, since variables no longer stay in
10013 a ``home register''.
10015 Enabled by default with @option{-funroll-loops} and @option{-fpeel-loops}.
10017 @item -fschedule-fusion
10018 @opindex fschedule-fusion
10019 Performs a target dependent pass over the instruction stream to schedule
10020 instructions of same type together because target machine can execute them
10021 more efficiently if they are adjacent to each other in the instruction flow.
10023 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
10027 Perform tail duplication to enlarge superblock size. This transformation
10028 simplifies the control flow of the function allowing other optimizations to do
10031 Enabled with @option{-fprofile-use}.
10033 @item -funroll-loops
10034 @opindex funroll-loops
10035 Unroll loops whose number of iterations can be determined at compile time or
10036 upon entry to the loop. @option{-funroll-loops} implies
10037 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
10038 It also turns on complete loop peeling (i.e.@: complete removal of loops with
10039 a small constant number of iterations). This option makes code larger, and may
10040 or may not make it run faster.
10042 Enabled with @option{-fprofile-use}.
10044 @item -funroll-all-loops
10045 @opindex funroll-all-loops
10046 Unroll all loops, even if their number of iterations is uncertain when
10047 the loop is entered. This usually makes programs run more slowly.
10048 @option{-funroll-all-loops} implies the same options as
10049 @option{-funroll-loops}.
10052 @opindex fpeel-loops
10053 Peels loops for which there is enough information that they do not
10054 roll much (from profile feedback). It also turns on complete loop peeling
10055 (i.e.@: complete removal of loops with small constant number of iterations).
10057 Enabled with @option{-fprofile-use}.
10059 @item -fmove-loop-invariants
10060 @opindex fmove-loop-invariants
10061 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
10062 at level @option{-O1}
10064 @item -funswitch-loops
10065 @opindex funswitch-loops
10066 Move branches with loop invariant conditions out of the loop, with duplicates
10067 of the loop on both branches (modified according to result of the condition).
10069 @item -ffunction-sections
10070 @itemx -fdata-sections
10071 @opindex ffunction-sections
10072 @opindex fdata-sections
10073 Place each function or data item into its own section in the output
10074 file if the target supports arbitrary sections. The name of the
10075 function or the name of the data item determines the section's name
10076 in the output file.
10078 Use these options on systems where the linker can perform optimizations
10079 to improve locality of reference in the instruction space. Most systems
10080 using the ELF object format and SPARC processors running Solaris 2 have
10081 linkers with such optimizations. AIX may have these optimizations in
10084 Only use these options when there are significant benefits from doing
10085 so. When you specify these options, the assembler and linker
10086 create larger object and executable files and are also slower.
10087 You cannot use @command{gprof} on all systems if you
10088 specify this option, and you may have problems with debugging if
10089 you specify both this option and @option{-g}.
10091 @item -fbranch-target-load-optimize
10092 @opindex fbranch-target-load-optimize
10093 Perform branch target register load optimization before prologue / epilogue
10095 The use of target registers can typically be exposed only during reload,
10096 thus hoisting loads out of loops and doing inter-block scheduling needs
10097 a separate optimization pass.
10099 @item -fbranch-target-load-optimize2
10100 @opindex fbranch-target-load-optimize2
10101 Perform branch target register load optimization after prologue / epilogue
10104 @item -fbtr-bb-exclusive
10105 @opindex fbtr-bb-exclusive
10106 When performing branch target register load optimization, don't reuse
10107 branch target registers within any basic block.
10109 @item -fstack-protector
10110 @opindex fstack-protector
10111 Emit extra code to check for buffer overflows, such as stack smashing
10112 attacks. This is done by adding a guard variable to functions with
10113 vulnerable objects. This includes functions that call @code{alloca}, and
10114 functions with buffers larger than 8 bytes. The guards are initialized
10115 when a function is entered and then checked when the function exits.
10116 If a guard check fails, an error message is printed and the program exits.
10118 @item -fstack-protector-all
10119 @opindex fstack-protector-all
10120 Like @option{-fstack-protector} except that all functions are protected.
10122 @item -fstack-protector-strong
10123 @opindex fstack-protector-strong
10124 Like @option{-fstack-protector} but includes additional functions to
10125 be protected --- those that have local array definitions, or have
10126 references to local frame addresses.
10128 @item -fstack-protector-explicit
10129 @opindex fstack-protector-explicit
10130 Like @option{-fstack-protector} but only protects those functions which
10131 have the @code{stack_protect} attribute
10134 @opindex fstdarg-opt
10135 Optimize the prologue of variadic argument functions with respect to usage of
10138 @item -fsection-anchors
10139 @opindex fsection-anchors
10140 Try to reduce the number of symbolic address calculations by using
10141 shared ``anchor'' symbols to address nearby objects. This transformation
10142 can help to reduce the number of GOT entries and GOT accesses on some
10145 For example, the implementation of the following function @code{foo}:
10148 static int a, b, c;
10149 int foo (void) @{ return a + b + c; @}
10153 usually calculates the addresses of all three variables, but if you
10154 compile it with @option{-fsection-anchors}, it accesses the variables
10155 from a common anchor point instead. The effect is similar to the
10156 following pseudocode (which isn't valid C):
10161 register int *xr = &x;
10162 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
10166 Not all targets support this option.
10168 @item --param @var{name}=@var{value}
10170 In some places, GCC uses various constants to control the amount of
10171 optimization that is done. For example, GCC does not inline functions
10172 that contain more than a certain number of instructions. You can
10173 control some of these constants on the command line using the
10174 @option{--param} option.
10176 The names of specific parameters, and the meaning of the values, are
10177 tied to the internals of the compiler, and are subject to change
10178 without notice in future releases.
10180 In each case, the @var{value} is an integer. The allowable choices for
10184 @item predictable-branch-outcome
10185 When branch is predicted to be taken with probability lower than this threshold
10186 (in percent), then it is considered well predictable. The default is 10.
10188 @item max-crossjump-edges
10189 The maximum number of incoming edges to consider for cross-jumping.
10190 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
10191 the number of edges incoming to each block. Increasing values mean
10192 more aggressive optimization, making the compilation time increase with
10193 probably small improvement in executable size.
10195 @item min-crossjump-insns
10196 The minimum number of instructions that must be matched at the end
10197 of two blocks before cross-jumping is performed on them. This
10198 value is ignored in the case where all instructions in the block being
10199 cross-jumped from are matched. The default value is 5.
10201 @item max-grow-copy-bb-insns
10202 The maximum code size expansion factor when copying basic blocks
10203 instead of jumping. The expansion is relative to a jump instruction.
10204 The default value is 8.
10206 @item max-goto-duplication-insns
10207 The maximum number of instructions to duplicate to a block that jumps
10208 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
10209 passes, GCC factors computed gotos early in the compilation process,
10210 and unfactors them as late as possible. Only computed jumps at the
10211 end of a basic blocks with no more than max-goto-duplication-insns are
10212 unfactored. The default value is 8.
10214 @item max-delay-slot-insn-search
10215 The maximum number of instructions to consider when looking for an
10216 instruction to fill a delay slot. If more than this arbitrary number of
10217 instructions are searched, the time savings from filling the delay slot
10218 are minimal, so stop searching. Increasing values mean more
10219 aggressive optimization, making the compilation time increase with probably
10220 small improvement in execution time.
10222 @item max-delay-slot-live-search
10223 When trying to fill delay slots, the maximum number of instructions to
10224 consider when searching for a block with valid live register
10225 information. Increasing this arbitrarily chosen value means more
10226 aggressive optimization, increasing the compilation time. This parameter
10227 should be removed when the delay slot code is rewritten to maintain the
10228 control-flow graph.
10230 @item max-gcse-memory
10231 The approximate maximum amount of memory that can be allocated in
10232 order to perform the global common subexpression elimination
10233 optimization. If more memory than specified is required, the
10234 optimization is not done.
10236 @item max-gcse-insertion-ratio
10237 If the ratio of expression insertions to deletions is larger than this value
10238 for any expression, then RTL PRE inserts or removes the expression and thus
10239 leaves partially redundant computations in the instruction stream. The default value is 20.
10241 @item max-pending-list-length
10242 The maximum number of pending dependencies scheduling allows
10243 before flushing the current state and starting over. Large functions
10244 with few branches or calls can create excessively large lists which
10245 needlessly consume memory and resources.
10247 @item max-modulo-backtrack-attempts
10248 The maximum number of backtrack attempts the scheduler should make
10249 when modulo scheduling a loop. Larger values can exponentially increase
10252 @item max-inline-insns-single
10253 Several parameters control the tree inliner used in GCC@.
10254 This number sets the maximum number of instructions (counted in GCC's
10255 internal representation) in a single function that the tree inliner
10256 considers for inlining. This only affects functions declared
10257 inline and methods implemented in a class declaration (C++).
10258 The default value is 400.
10260 @item max-inline-insns-auto
10261 When you use @option{-finline-functions} (included in @option{-O3}),
10262 a lot of functions that would otherwise not be considered for inlining
10263 by the compiler are investigated. To those functions, a different
10264 (more restrictive) limit compared to functions declared inline can
10266 The default value is 40.
10268 @item inline-min-speedup
10269 When estimated performance improvement of caller + callee runtime exceeds this
10270 threshold (in precent), the function can be inlined regardless the limit on
10271 @option{--param max-inline-insns-single} and @option{--param
10272 max-inline-insns-auto}.
10274 @item large-function-insns
10275 The limit specifying really large functions. For functions larger than this
10276 limit after inlining, inlining is constrained by
10277 @option{--param large-function-growth}. This parameter is useful primarily
10278 to avoid extreme compilation time caused by non-linear algorithms used by the
10280 The default value is 2700.
10282 @item large-function-growth
10283 Specifies maximal growth of large function caused by inlining in percents.
10284 The default value is 100 which limits large function growth to 2.0 times
10287 @item large-unit-insns
10288 The limit specifying large translation unit. Growth caused by inlining of
10289 units larger than this limit is limited by @option{--param inline-unit-growth}.
10290 For small units this might be too tight.
10291 For example, consider a unit consisting of function A
10292 that is inline and B that just calls A three times. If B is small relative to
10293 A, the growth of unit is 300\% and yet such inlining is very sane. For very
10294 large units consisting of small inlineable functions, however, the overall unit
10295 growth limit is needed to avoid exponential explosion of code size. Thus for
10296 smaller units, the size is increased to @option{--param large-unit-insns}
10297 before applying @option{--param inline-unit-growth}. The default is 10000.
10299 @item inline-unit-growth
10300 Specifies maximal overall growth of the compilation unit caused by inlining.
10301 The default value is 20 which limits unit growth to 1.2 times the original
10302 size. Cold functions (either marked cold via an attribute or by profile
10303 feedback) are not accounted into the unit size.
10305 @item ipcp-unit-growth
10306 Specifies maximal overall growth of the compilation unit caused by
10307 interprocedural constant propagation. The default value is 10 which limits
10308 unit growth to 1.1 times the original size.
10310 @item large-stack-frame
10311 The limit specifying large stack frames. While inlining the algorithm is trying
10312 to not grow past this limit too much. The default value is 256 bytes.
10314 @item large-stack-frame-growth
10315 Specifies maximal growth of large stack frames caused by inlining in percents.
10316 The default value is 1000 which limits large stack frame growth to 11 times
10319 @item max-inline-insns-recursive
10320 @itemx max-inline-insns-recursive-auto
10321 Specifies the maximum number of instructions an out-of-line copy of a
10322 self-recursive inline
10323 function can grow into by performing recursive inlining.
10325 @option{--param max-inline-insns-recursive} applies to functions
10327 For functions not declared inline, recursive inlining
10328 happens only when @option{-finline-functions} (included in @option{-O3}) is
10329 enabled; @option{--param max-inline-insns-recursive-auto} applies instead. The
10330 default value is 450.
10332 @item max-inline-recursive-depth
10333 @itemx max-inline-recursive-depth-auto
10334 Specifies the maximum recursion depth used for recursive inlining.
10336 @option{--param max-inline-recursive-depth} applies to functions
10337 declared inline. For functions not declared inline, recursive inlining
10338 happens only when @option{-finline-functions} (included in @option{-O3}) is
10339 enabled; @option{--param max-inline-recursive-depth-auto} applies instead. The
10340 default value is 8.
10342 @item min-inline-recursive-probability
10343 Recursive inlining is profitable only for function having deep recursion
10344 in average and can hurt for function having little recursion depth by
10345 increasing the prologue size or complexity of function body to other
10348 When profile feedback is available (see @option{-fprofile-generate}) the actual
10349 recursion depth can be guessed from probability that function recurses via a
10350 given call expression. This parameter limits inlining only to call expressions
10351 whose probability exceeds the given threshold (in percents).
10352 The default value is 10.
10354 @item early-inlining-insns
10355 Specify growth that the early inliner can make. In effect it increases
10356 the amount of inlining for code having a large abstraction penalty.
10357 The default value is 14.
10359 @item max-early-inliner-iterations
10360 Limit of iterations of the early inliner. This basically bounds
10361 the number of nested indirect calls the early inliner can resolve.
10362 Deeper chains are still handled by late inlining.
10364 @item comdat-sharing-probability
10365 Probability (in percent) that C++ inline function with comdat visibility
10366 are shared across multiple compilation units. The default value is 20.
10368 @item profile-func-internal-id
10369 A parameter to control whether to use function internal id in profile
10370 database lookup. If the value is 0, the compiler uses an id that
10371 is based on function assembler name and filename, which makes old profile
10372 data more tolerant to source changes such as function reordering etc.
10373 The default value is 0.
10375 @item min-vect-loop-bound
10376 The minimum number of iterations under which loops are not vectorized
10377 when @option{-ftree-vectorize} is used. The number of iterations after
10378 vectorization needs to be greater than the value specified by this option
10379 to allow vectorization. The default value is 0.
10381 @item gcse-cost-distance-ratio
10382 Scaling factor in calculation of maximum distance an expression
10383 can be moved by GCSE optimizations. This is currently supported only in the
10384 code hoisting pass. The bigger the ratio, the more aggressive code hoisting
10385 is with simple expressions, i.e., the expressions that have cost
10386 less than @option{gcse-unrestricted-cost}. Specifying 0 disables
10387 hoisting of simple expressions. The default value is 10.
10389 @item gcse-unrestricted-cost
10390 Cost, roughly measured as the cost of a single typical machine
10391 instruction, at which GCSE optimizations do not constrain
10392 the distance an expression can travel. This is currently
10393 supported only in the code hoisting pass. The lesser the cost,
10394 the more aggressive code hoisting is. Specifying 0
10395 allows all expressions to travel unrestricted distances.
10396 The default value is 3.
10398 @item max-hoist-depth
10399 The depth of search in the dominator tree for expressions to hoist.
10400 This is used to avoid quadratic behavior in hoisting algorithm.
10401 The value of 0 does not limit on the search, but may slow down compilation
10402 of huge functions. The default value is 30.
10404 @item max-tail-merge-comparisons
10405 The maximum amount of similar bbs to compare a bb with. This is used to
10406 avoid quadratic behavior in tree tail merging. The default value is 10.
10408 @item max-tail-merge-iterations
10409 The maximum amount of iterations of the pass over the function. This is used to
10410 limit compilation time in tree tail merging. The default value is 2.
10412 @item max-unrolled-insns
10413 The maximum number of instructions that a loop may have to be unrolled.
10414 If a loop is unrolled, this parameter also determines how many times
10415 the loop code is unrolled.
10417 @item max-average-unrolled-insns
10418 The maximum number of instructions biased by probabilities of their execution
10419 that a loop may have to be unrolled. If a loop is unrolled,
10420 this parameter also determines how many times the loop code is unrolled.
10422 @item max-unroll-times
10423 The maximum number of unrollings of a single loop.
10425 @item max-peeled-insns
10426 The maximum number of instructions that a loop may have to be peeled.
10427 If a loop is peeled, this parameter also determines how many times
10428 the loop code is peeled.
10430 @item max-peel-times
10431 The maximum number of peelings of a single loop.
10433 @item max-peel-branches
10434 The maximum number of branches on the hot path through the peeled sequence.
10436 @item max-completely-peeled-insns
10437 The maximum number of insns of a completely peeled loop.
10439 @item max-completely-peel-times
10440 The maximum number of iterations of a loop to be suitable for complete peeling.
10442 @item max-completely-peel-loop-nest-depth
10443 The maximum depth of a loop nest suitable for complete peeling.
10445 @item max-unswitch-insns
10446 The maximum number of insns of an unswitched loop.
10448 @item max-unswitch-level
10449 The maximum number of branches unswitched in a single loop.
10451 @item lim-expensive
10452 The minimum cost of an expensive expression in the loop invariant motion.
10454 @item iv-consider-all-candidates-bound
10455 Bound on number of candidates for induction variables, below which
10456 all candidates are considered for each use in induction variable
10457 optimizations. If there are more candidates than this,
10458 only the most relevant ones are considered to avoid quadratic time complexity.
10460 @item iv-max-considered-uses
10461 The induction variable optimizations give up on loops that contain more
10462 induction variable uses.
10464 @item iv-always-prune-cand-set-bound
10465 If the number of candidates in the set is smaller than this value,
10466 always try to remove unnecessary ivs from the set
10467 when adding a new one.
10469 @item scev-max-expr-size
10470 Bound on size of expressions used in the scalar evolutions analyzer.
10471 Large expressions slow the analyzer.
10473 @item scev-max-expr-complexity
10474 Bound on the complexity of the expressions in the scalar evolutions analyzer.
10475 Complex expressions slow the analyzer.
10477 @item omega-max-vars
10478 The maximum number of variables in an Omega constraint system.
10479 The default value is 128.
10481 @item omega-max-geqs
10482 The maximum number of inequalities in an Omega constraint system.
10483 The default value is 256.
10485 @item omega-max-eqs
10486 The maximum number of equalities in an Omega constraint system.
10487 The default value is 128.
10489 @item omega-max-wild-cards
10490 The maximum number of wildcard variables that the Omega solver is
10491 able to insert. The default value is 18.
10493 @item omega-hash-table-size
10494 The size of the hash table in the Omega solver. The default value is
10497 @item omega-max-keys
10498 The maximal number of keys used by the Omega solver. The default
10501 @item omega-eliminate-redundant-constraints
10502 When set to 1, use expensive methods to eliminate all redundant
10503 constraints. The default value is 0.
10505 @item vect-max-version-for-alignment-checks
10506 The maximum number of run-time checks that can be performed when
10507 doing loop versioning for alignment in the vectorizer.
10509 @item vect-max-version-for-alias-checks
10510 The maximum number of run-time checks that can be performed when
10511 doing loop versioning for alias in the vectorizer.
10513 @item vect-max-peeling-for-alignment
10514 The maximum number of loop peels to enhance access alignment
10515 for vectorizer. Value -1 means 'no limit'.
10517 @item max-iterations-to-track
10518 The maximum number of iterations of a loop the brute-force algorithm
10519 for analysis of the number of iterations of the loop tries to evaluate.
10521 @item hot-bb-count-ws-permille
10522 A basic block profile count is considered hot if it contributes to
10523 the given permillage (i.e. 0...1000) of the entire profiled execution.
10525 @item hot-bb-frequency-fraction
10526 Select fraction of the entry block frequency of executions of basic block in
10527 function given basic block needs to have to be considered hot.
10529 @item max-predicted-iterations
10530 The maximum number of loop iterations we predict statically. This is useful
10531 in cases where a function contains a single loop with known bound and
10532 another loop with unknown bound.
10533 The known number of iterations is predicted correctly, while
10534 the unknown number of iterations average to roughly 10. This means that the
10535 loop without bounds appears artificially cold relative to the other one.
10537 @item builtin-expect-probability
10538 Control the probability of the expression having the specified value. This
10539 parameter takes a percentage (i.e. 0 ... 100) as input.
10540 The default probability of 90 is obtained empirically.
10542 @item align-threshold
10544 Select fraction of the maximal frequency of executions of a basic block in
10545 a function to align the basic block.
10547 @item align-loop-iterations
10549 A loop expected to iterate at least the selected number of iterations is
10552 @item tracer-dynamic-coverage
10553 @itemx tracer-dynamic-coverage-feedback
10555 This value is used to limit superblock formation once the given percentage of
10556 executed instructions is covered. This limits unnecessary code size
10559 The @option{tracer-dynamic-coverage-feedback} parameter
10560 is used only when profile
10561 feedback is available. The real profiles (as opposed to statically estimated
10562 ones) are much less balanced allowing the threshold to be larger value.
10564 @item tracer-max-code-growth
10565 Stop tail duplication once code growth has reached given percentage. This is
10566 a rather artificial limit, as most of the duplicates are eliminated later in
10567 cross jumping, so it may be set to much higher values than is the desired code
10570 @item tracer-min-branch-ratio
10572 Stop reverse growth when the reverse probability of best edge is less than this
10573 threshold (in percent).
10575 @item tracer-min-branch-ratio
10576 @itemx tracer-min-branch-ratio-feedback
10578 Stop forward growth if the best edge has probability lower than this
10581 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
10582 compilation for profile feedback and one for compilation without. The value
10583 for compilation with profile feedback needs to be more conservative (higher) in
10584 order to make tracer effective.
10586 @item max-cse-path-length
10588 The maximum number of basic blocks on path that CSE considers.
10591 @item max-cse-insns
10592 The maximum number of instructions CSE processes before flushing.
10593 The default is 1000.
10595 @item ggc-min-expand
10597 GCC uses a garbage collector to manage its own memory allocation. This
10598 parameter specifies the minimum percentage by which the garbage
10599 collector's heap should be allowed to expand between collections.
10600 Tuning this may improve compilation speed; it has no effect on code
10603 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
10604 RAM >= 1GB@. If @code{getrlimit} is available, the notion of ``RAM'' is
10605 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
10606 GCC is not able to calculate RAM on a particular platform, the lower
10607 bound of 30% is used. Setting this parameter and
10608 @option{ggc-min-heapsize} to zero causes a full collection to occur at
10609 every opportunity. This is extremely slow, but can be useful for
10612 @item ggc-min-heapsize
10614 Minimum size of the garbage collector's heap before it begins bothering
10615 to collect garbage. The first collection occurs after the heap expands
10616 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
10617 tuning this may improve compilation speed, and has no effect on code
10620 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit that
10621 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
10622 with a lower bound of 4096 (four megabytes) and an upper bound of
10623 131072 (128 megabytes). If GCC is not able to calculate RAM on a
10624 particular platform, the lower bound is used. Setting this parameter
10625 very large effectively disables garbage collection. Setting this
10626 parameter and @option{ggc-min-expand} to zero causes a full collection
10627 to occur at every opportunity.
10629 @item max-reload-search-insns
10630 The maximum number of instruction reload should look backward for equivalent
10631 register. Increasing values mean more aggressive optimization, making the
10632 compilation time increase with probably slightly better performance.
10633 The default value is 100.
10635 @item max-cselib-memory-locations
10636 The maximum number of memory locations cselib should take into account.
10637 Increasing values mean more aggressive optimization, making the compilation time
10638 increase with probably slightly better performance. The default value is 500.
10640 @item reorder-blocks-duplicate
10641 @itemx reorder-blocks-duplicate-feedback
10643 Used by the basic block reordering pass to decide whether to use unconditional
10644 branch or duplicate the code on its destination. Code is duplicated when its
10645 estimated size is smaller than this value multiplied by the estimated size of
10646 unconditional jump in the hot spots of the program.
10648 The @option{reorder-block-duplicate-feedback} parameter
10649 is used only when profile
10650 feedback is available. It may be set to higher values than
10651 @option{reorder-block-duplicate} since information about the hot spots is more
10654 @item max-sched-ready-insns
10655 The maximum number of instructions ready to be issued the scheduler should
10656 consider at any given time during the first scheduling pass. Increasing
10657 values mean more thorough searches, making the compilation time increase
10658 with probably little benefit. The default value is 100.
10660 @item max-sched-region-blocks
10661 The maximum number of blocks in a region to be considered for
10662 interblock scheduling. The default value is 10.
10664 @item max-pipeline-region-blocks
10665 The maximum number of blocks in a region to be considered for
10666 pipelining in the selective scheduler. The default value is 15.
10668 @item max-sched-region-insns
10669 The maximum number of insns in a region to be considered for
10670 interblock scheduling. The default value is 100.
10672 @item max-pipeline-region-insns
10673 The maximum number of insns in a region to be considered for
10674 pipelining in the selective scheduler. The default value is 200.
10676 @item min-spec-prob
10677 The minimum probability (in percents) of reaching a source block
10678 for interblock speculative scheduling. The default value is 40.
10680 @item max-sched-extend-regions-iters
10681 The maximum number of iterations through CFG to extend regions.
10682 A value of 0 (the default) disables region extensions.
10684 @item max-sched-insn-conflict-delay
10685 The maximum conflict delay for an insn to be considered for speculative motion.
10686 The default value is 3.
10688 @item sched-spec-prob-cutoff
10689 The minimal probability of speculation success (in percents), so that
10690 speculative insns are scheduled.
10691 The default value is 40.
10693 @item sched-spec-state-edge-prob-cutoff
10694 The minimum probability an edge must have for the scheduler to save its
10696 The default value is 10.
10698 @item sched-mem-true-dep-cost
10699 Minimal distance (in CPU cycles) between store and load targeting same
10700 memory locations. The default value is 1.
10702 @item selsched-max-lookahead
10703 The maximum size of the lookahead window of selective scheduling. It is a
10704 depth of search for available instructions.
10705 The default value is 50.
10707 @item selsched-max-sched-times
10708 The maximum number of times that an instruction is scheduled during
10709 selective scheduling. This is the limit on the number of iterations
10710 through which the instruction may be pipelined. The default value is 2.
10712 @item selsched-max-insns-to-rename
10713 The maximum number of best instructions in the ready list that are considered
10714 for renaming in the selective scheduler. The default value is 2.
10717 The minimum value of stage count that swing modulo scheduler
10718 generates. The default value is 2.
10720 @item max-last-value-rtl
10721 The maximum size measured as number of RTLs that can be recorded in an expression
10722 in combiner for a pseudo register as last known value of that register. The default
10725 @item max-combine-insns
10726 The maximum number of instructions the RTL combiner tries to combine.
10727 The default value is 2 at @option{-Og} and 4 otherwise.
10729 @item integer-share-limit
10730 Small integer constants can use a shared data structure, reducing the
10731 compiler's memory usage and increasing its speed. This sets the maximum
10732 value of a shared integer constant. The default value is 256.
10734 @item ssp-buffer-size
10735 The minimum size of buffers (i.e.@: arrays) that receive stack smashing
10736 protection when @option{-fstack-protection} is used.
10738 @item min-size-for-stack-sharing
10739 The minimum size of variables taking part in stack slot sharing when not
10740 optimizing. The default value is 32.
10742 @item max-jump-thread-duplication-stmts
10743 Maximum number of statements allowed in a block that needs to be
10744 duplicated when threading jumps.
10746 @item max-fields-for-field-sensitive
10747 Maximum number of fields in a structure treated in
10748 a field sensitive manner during pointer analysis. The default is zero
10749 for @option{-O0} and @option{-O1},
10750 and 100 for @option{-Os}, @option{-O2}, and @option{-O3}.
10752 @item prefetch-latency
10753 Estimate on average number of instructions that are executed before
10754 prefetch finishes. The distance prefetched ahead is proportional
10755 to this constant. Increasing this number may also lead to less
10756 streams being prefetched (see @option{simultaneous-prefetches}).
10758 @item simultaneous-prefetches
10759 Maximum number of prefetches that can run at the same time.
10761 @item l1-cache-line-size
10762 The size of cache line in L1 cache, in bytes.
10764 @item l1-cache-size
10765 The size of L1 cache, in kilobytes.
10767 @item l2-cache-size
10768 The size of L2 cache, in kilobytes.
10770 @item min-insn-to-prefetch-ratio
10771 The minimum ratio between the number of instructions and the
10772 number of prefetches to enable prefetching in a loop.
10774 @item prefetch-min-insn-to-mem-ratio
10775 The minimum ratio between the number of instructions and the
10776 number of memory references to enable prefetching in a loop.
10778 @item use-canonical-types
10779 Whether the compiler should use the ``canonical'' type system. By
10780 default, this should always be 1, which uses a more efficient internal
10781 mechanism for comparing types in C++ and Objective-C++. However, if
10782 bugs in the canonical type system are causing compilation failures,
10783 set this value to 0 to disable canonical types.
10785 @item switch-conversion-max-branch-ratio
10786 Switch initialization conversion refuses to create arrays that are
10787 bigger than @option{switch-conversion-max-branch-ratio} times the number of
10788 branches in the switch.
10790 @item max-partial-antic-length
10791 Maximum length of the partial antic set computed during the tree
10792 partial redundancy elimination optimization (@option{-ftree-pre}) when
10793 optimizing at @option{-O3} and above. For some sorts of source code
10794 the enhanced partial redundancy elimination optimization can run away,
10795 consuming all of the memory available on the host machine. This
10796 parameter sets a limit on the length of the sets that are computed,
10797 which prevents the runaway behavior. Setting a value of 0 for
10798 this parameter allows an unlimited set length.
10800 @item sccvn-max-scc-size
10801 Maximum size of a strongly connected component (SCC) during SCCVN
10802 processing. If this limit is hit, SCCVN processing for the whole
10803 function is not done and optimizations depending on it are
10804 disabled. The default maximum SCC size is 10000.
10806 @item sccvn-max-alias-queries-per-access
10807 Maximum number of alias-oracle queries we perform when looking for
10808 redundancies for loads and stores. If this limit is hit the search
10809 is aborted and the load or store is not considered redundant. The
10810 number of queries is algorithmically limited to the number of
10811 stores on all paths from the load to the function entry.
10812 The default maxmimum number of queries is 1000.
10814 @item ira-max-loops-num
10815 IRA uses regional register allocation by default. If a function
10816 contains more loops than the number given by this parameter, only at most
10817 the given number of the most frequently-executed loops form regions
10818 for regional register allocation. The default value of the
10821 @item ira-max-conflict-table-size
10822 Although IRA uses a sophisticated algorithm to compress the conflict
10823 table, the table can still require excessive amounts of memory for
10824 huge functions. If the conflict table for a function could be more
10825 than the size in MB given by this parameter, the register allocator
10826 instead uses a faster, simpler, and lower-quality
10827 algorithm that does not require building a pseudo-register conflict table.
10828 The default value of the parameter is 2000.
10830 @item ira-loop-reserved-regs
10831 IRA can be used to evaluate more accurate register pressure in loops
10832 for decisions to move loop invariants (see @option{-O3}). The number
10833 of available registers reserved for some other purposes is given
10834 by this parameter. The default value of the parameter is 2, which is
10835 the minimal number of registers needed by typical instructions.
10836 This value is the best found from numerous experiments.
10838 @item lra-inheritance-ebb-probability-cutoff
10839 LRA tries to reuse values reloaded in registers in subsequent insns.
10840 This optimization is called inheritance. EBB is used as a region to
10841 do this optimization. The parameter defines a minimal fall-through
10842 edge probability in percentage used to add BB to inheritance EBB in
10843 LRA. The default value of the parameter is 40. The value was chosen
10844 from numerous runs of SPEC2000 on x86-64.
10846 @item loop-invariant-max-bbs-in-loop
10847 Loop invariant motion can be very expensive, both in compilation time and
10848 in amount of needed compile-time memory, with very large loops. Loops
10849 with more basic blocks than this parameter won't have loop invariant
10850 motion optimization performed on them. The default value of the
10851 parameter is 1000 for @option{-O1} and 10000 for @option{-O2} and above.
10853 @item loop-max-datarefs-for-datadeps
10854 Building data dapendencies is expensive for very large loops. This
10855 parameter limits the number of data references in loops that are
10856 considered for data dependence analysis. These large loops are no
10857 handled by the optimizations using loop data dependencies.
10858 The default value is 1000.
10860 @item max-vartrack-size
10861 Sets a maximum number of hash table slots to use during variable
10862 tracking dataflow analysis of any function. If this limit is exceeded
10863 with variable tracking at assignments enabled, analysis for that
10864 function is retried without it, after removing all debug insns from
10865 the function. If the limit is exceeded even without debug insns, var
10866 tracking analysis is completely disabled for the function. Setting
10867 the parameter to zero makes it unlimited.
10869 @item max-vartrack-expr-depth
10870 Sets a maximum number of recursion levels when attempting to map
10871 variable names or debug temporaries to value expressions. This trades
10872 compilation time for more complete debug information. If this is set too
10873 low, value expressions that are available and could be represented in
10874 debug information may end up not being used; setting this higher may
10875 enable the compiler to find more complex debug expressions, but compile
10876 time and memory use may grow. The default is 12.
10878 @item min-nondebug-insn-uid
10879 Use uids starting at this parameter for nondebug insns. The range below
10880 the parameter is reserved exclusively for debug insns created by
10881 @option{-fvar-tracking-assignments}, but debug insns may get
10882 (non-overlapping) uids above it if the reserved range is exhausted.
10884 @item ipa-sra-ptr-growth-factor
10885 IPA-SRA replaces a pointer to an aggregate with one or more new
10886 parameters only when their cumulative size is less or equal to
10887 @option{ipa-sra-ptr-growth-factor} times the size of the original
10890 @item sra-max-scalarization-size-Ospeed
10891 @item sra-max-scalarization-size-Osize
10892 The two Scalar Reduction of Aggregates passes (SRA and IPA-SRA) aim to
10893 replace scalar parts of aggregates with uses of independent scalar
10894 variables. These parameters control the maximum size, in storage units,
10895 of aggregate which is considered for replacement when compiling for
10897 (@option{sra-max-scalarization-size-Ospeed}) or size
10898 (@option{sra-max-scalarization-size-Osize}) respectively.
10900 @item tm-max-aggregate-size
10901 When making copies of thread-local variables in a transaction, this
10902 parameter specifies the size in bytes after which variables are
10903 saved with the logging functions as opposed to save/restore code
10904 sequence pairs. This option only applies when using
10907 @item graphite-max-nb-scop-params
10908 To avoid exponential effects in the Graphite loop transforms, the
10909 number of parameters in a Static Control Part (SCoP) is bounded. The
10910 default value is 10 parameters. A variable whose value is unknown at
10911 compilation time and defined outside a SCoP is a parameter of the SCoP.
10913 @item graphite-max-bbs-per-function
10914 To avoid exponential effects in the detection of SCoPs, the size of
10915 the functions analyzed by Graphite is bounded. The default value is
10918 @item loop-block-tile-size
10919 Loop blocking or strip mining transforms, enabled with
10920 @option{-floop-block} or @option{-floop-strip-mine}, strip mine each
10921 loop in the loop nest by a given number of iterations. The strip
10922 length can be changed using the @option{loop-block-tile-size}
10923 parameter. The default value is 51 iterations.
10925 @item loop-unroll-jam-size
10926 Specify the unroll factor for the @option{-floop-unroll-and-jam} option. The
10927 default value is 4.
10929 @item loop-unroll-jam-depth
10930 Specify the dimension to be unrolled (counting from the most inner loop)
10931 for the @option{-floop-unroll-and-jam}. The default value is 2.
10933 @item ipa-cp-value-list-size
10934 IPA-CP attempts to track all possible values and types passed to a function's
10935 parameter in order to propagate them and perform devirtualization.
10936 @option{ipa-cp-value-list-size} is the maximum number of values and types it
10937 stores per one formal parameter of a function.
10939 @item ipa-cp-eval-threshold
10940 IPA-CP calculates its own score of cloning profitability heuristics
10941 and performs those cloning opportunities with scores that exceed
10942 @option{ipa-cp-eval-threshold}.
10944 @item ipa-cp-recursion-penalty
10945 Percentage penalty the recursive functions will receive when they
10946 are evaluated for cloning.
10948 @item ipa-cp-single-call-penalty
10949 Percentage penalty functions containg a single call to another
10950 function will receive when they are evaluated for cloning.
10953 @item ipa-max-agg-items
10954 IPA-CP is also capable to propagate a number of scalar values passed
10955 in an aggregate. @option{ipa-max-agg-items} controls the maximum
10956 number of such values per one parameter.
10958 @item ipa-cp-loop-hint-bonus
10959 When IPA-CP determines that a cloning candidate would make the number
10960 of iterations of a loop known, it adds a bonus of
10961 @option{ipa-cp-loop-hint-bonus} to the profitability score of
10964 @item ipa-cp-array-index-hint-bonus
10965 When IPA-CP determines that a cloning candidate would make the index of
10966 an array access known, it adds a bonus of
10967 @option{ipa-cp-array-index-hint-bonus} to the profitability
10968 score of the candidate.
10970 @item ipa-max-aa-steps
10971 During its analysis of function bodies, IPA-CP employs alias analysis
10972 in order to track values pointed to by function parameters. In order
10973 not spend too much time analyzing huge functions, it gives up and
10974 consider all memory clobbered after examining
10975 @option{ipa-max-aa-steps} statements modifying memory.
10977 @item lto-partitions
10978 Specify desired number of partitions produced during WHOPR compilation.
10979 The number of partitions should exceed the number of CPUs used for compilation.
10980 The default value is 32.
10982 @item lto-minpartition
10983 Size of minimal partition for WHOPR (in estimated instructions).
10984 This prevents expenses of splitting very small programs into too many
10987 @item cxx-max-namespaces-for-diagnostic-help
10988 The maximum number of namespaces to consult for suggestions when C++
10989 name lookup fails for an identifier. The default is 1000.
10991 @item sink-frequency-threshold
10992 The maximum relative execution frequency (in percents) of the target block
10993 relative to a statement's original block to allow statement sinking of a
10994 statement. Larger numbers result in more aggressive statement sinking.
10995 The default value is 75. A small positive adjustment is applied for
10996 statements with memory operands as those are even more profitable so sink.
10998 @item max-stores-to-sink
10999 The maximum number of conditional stores paires that can be sunk. Set to 0
11000 if either vectorization (@option{-ftree-vectorize}) or if-conversion
11001 (@option{-ftree-loop-if-convert}) is disabled. The default is 2.
11003 @item allow-store-data-races
11004 Allow optimizers to introduce new data races on stores.
11005 Set to 1 to allow, otherwise to 0. This option is enabled by default
11006 at optimization level @option{-Ofast}.
11008 @item case-values-threshold
11009 The smallest number of different values for which it is best to use a
11010 jump-table instead of a tree of conditional branches. If the value is
11011 0, use the default for the machine. The default is 0.
11013 @item tree-reassoc-width
11014 Set the maximum number of instructions executed in parallel in
11015 reassociated tree. This parameter overrides target dependent
11016 heuristics used by default if has non zero value.
11018 @item sched-pressure-algorithm
11019 Choose between the two available implementations of
11020 @option{-fsched-pressure}. Algorithm 1 is the original implementation
11021 and is the more likely to prevent instructions from being reordered.
11022 Algorithm 2 was designed to be a compromise between the relatively
11023 conservative approach taken by algorithm 1 and the rather aggressive
11024 approach taken by the default scheduler. It relies more heavily on
11025 having a regular register file and accurate register pressure classes.
11026 See @file{haifa-sched.c} in the GCC sources for more details.
11028 The default choice depends on the target.
11030 @item max-slsr-cand-scan
11031 Set the maximum number of existing candidates that are considered when
11032 seeking a basis for a new straight-line strength reduction candidate.
11035 Enable buffer overflow detection for global objects. This kind
11036 of protection is enabled by default if you are using
11037 @option{-fsanitize=address} option.
11038 To disable global objects protection use @option{--param asan-globals=0}.
11041 Enable buffer overflow detection for stack objects. This kind of
11042 protection is enabled by default when using@option{-fsanitize=address}.
11043 To disable stack protection use @option{--param asan-stack=0} option.
11045 @item asan-instrument-reads
11046 Enable buffer overflow detection for memory reads. This kind of
11047 protection is enabled by default when using @option{-fsanitize=address}.
11048 To disable memory reads protection use
11049 @option{--param asan-instrument-reads=0}.
11051 @item asan-instrument-writes
11052 Enable buffer overflow detection for memory writes. This kind of
11053 protection is enabled by default when using @option{-fsanitize=address}.
11054 To disable memory writes protection use
11055 @option{--param asan-instrument-writes=0} option.
11057 @item asan-memintrin
11058 Enable detection for built-in functions. This kind of protection
11059 is enabled by default when using @option{-fsanitize=address}.
11060 To disable built-in functions protection use
11061 @option{--param asan-memintrin=0}.
11063 @item asan-use-after-return
11064 Enable detection of use-after-return. This kind of protection
11065 is enabled by default when using @option{-fsanitize=address} option.
11066 To disable use-after-return detection use
11067 @option{--param asan-use-after-return=0}.
11069 @item asan-instrumentation-with-call-threshold
11070 If number of memory accesses in function being instrumented
11071 is greater or equal to this number, use callbacks instead of inline checks.
11072 E.g. to disable inline code use
11073 @option{--param asan-instrumentation-with-call-threshold=0}.
11075 @item chkp-max-ctor-size
11076 Static constructors generated by Pointer Bounds Checker may become very
11077 large and significantly increase compile time at optimization level
11078 @option{-O1} and higher. This parameter is a maximum nubmer of statements
11079 in a single generated constructor. Default value is 5000.
11081 @item max-fsm-thread-path-insns
11082 Maximum number of instructions to copy when duplicating blocks on a
11083 finite state automaton jump thread path. The default is 100.
11085 @item max-fsm-thread-length
11086 Maximum number of basic blocks on a finite state automaton jump thread
11087 path. The default is 10.
11089 @item max-fsm-thread-paths
11090 Maximum number of new jump thread paths to create for a finite state
11091 automaton. The default is 50.
11096 @node Preprocessor Options
11097 @section Options Controlling the Preprocessor
11098 @cindex preprocessor options
11099 @cindex options, preprocessor
11101 These options control the C preprocessor, which is run on each C source
11102 file before actual compilation.
11104 If you use the @option{-E} option, nothing is done except preprocessing.
11105 Some of these options make sense only together with @option{-E} because
11106 they cause the preprocessor output to be unsuitable for actual
11110 @item -Wp,@var{option}
11112 You can use @option{-Wp,@var{option}} to bypass the compiler driver
11113 and pass @var{option} directly through to the preprocessor. If
11114 @var{option} contains commas, it is split into multiple options at the
11115 commas. However, many options are modified, translated or interpreted
11116 by the compiler driver before being passed to the preprocessor, and
11117 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
11118 interface is undocumented and subject to change, so whenever possible
11119 you should avoid using @option{-Wp} and let the driver handle the
11122 @item -Xpreprocessor @var{option}
11123 @opindex Xpreprocessor
11124 Pass @var{option} as an option to the preprocessor. You can use this to
11125 supply system-specific preprocessor options that GCC does not
11128 If you want to pass an option that takes an argument, you must use
11129 @option{-Xpreprocessor} twice, once for the option and once for the argument.
11131 @item -no-integrated-cpp
11132 @opindex no-integrated-cpp
11133 Perform preprocessing as a separate pass before compilation.
11134 By default, GCC performs preprocessing as an integrated part of
11135 input tokenization and parsing.
11136 If this option is provided, the appropriate language front end
11137 (@command{cc1}, @command{cc1plus}, or @command{cc1obj} for C, C++,
11138 and Objective-C, respectively) is instead invoked twice,
11139 once for preprocessing only and once for actual compilation
11140 of the preprocessed input.
11141 This option may be useful in conjunction with the @option{-B} or
11142 @option{-wrapper} options to specify an alternate preprocessor or
11143 perform additional processing of the program source between
11144 normal preprocessing and compilation.
11147 @include cppopts.texi
11149 @node Assembler Options
11150 @section Passing Options to the Assembler
11152 @c prevent bad page break with this line
11153 You can pass options to the assembler.
11156 @item -Wa,@var{option}
11158 Pass @var{option} as an option to the assembler. If @var{option}
11159 contains commas, it is split into multiple options at the commas.
11161 @item -Xassembler @var{option}
11162 @opindex Xassembler
11163 Pass @var{option} as an option to the assembler. You can use this to
11164 supply system-specific assembler options that GCC does not
11167 If you want to pass an option that takes an argument, you must use
11168 @option{-Xassembler} twice, once for the option and once for the argument.
11173 @section Options for Linking
11174 @cindex link options
11175 @cindex options, linking
11177 These options come into play when the compiler links object files into
11178 an executable output file. They are meaningless if the compiler is
11179 not doing a link step.
11183 @item @var{object-file-name}
11184 A file name that does not end in a special recognized suffix is
11185 considered to name an object file or library. (Object files are
11186 distinguished from libraries by the linker according to the file
11187 contents.) If linking is done, these object files are used as input
11196 If any of these options is used, then the linker is not run, and
11197 object file names should not be used as arguments. @xref{Overall
11201 @opindex fuse-ld=bfd
11202 Use the @command{bfd} linker instead of the default linker.
11204 @item -fuse-ld=gold
11205 @opindex fuse-ld=gold
11206 Use the @command{gold} linker instead of the default linker.
11209 @item -l@var{library}
11210 @itemx -l @var{library}
11212 Search the library named @var{library} when linking. (The second
11213 alternative with the library as a separate argument is only for
11214 POSIX compliance and is not recommended.)
11216 It makes a difference where in the command you write this option; the
11217 linker searches and processes libraries and object files in the order they
11218 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
11219 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
11220 to functions in @samp{z}, those functions may not be loaded.
11222 The linker searches a standard list of directories for the library,
11223 which is actually a file named @file{lib@var{library}.a}. The linker
11224 then uses this file as if it had been specified precisely by name.
11226 The directories searched include several standard system directories
11227 plus any that you specify with @option{-L}.
11229 Normally the files found this way are library files---archive files
11230 whose members are object files. The linker handles an archive file by
11231 scanning through it for members which define symbols that have so far
11232 been referenced but not defined. But if the file that is found is an
11233 ordinary object file, it is linked in the usual fashion. The only
11234 difference between using an @option{-l} option and specifying a file name
11235 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
11236 and searches several directories.
11240 You need this special case of the @option{-l} option in order to
11241 link an Objective-C or Objective-C++ program.
11243 @item -nostartfiles
11244 @opindex nostartfiles
11245 Do not use the standard system startup files when linking.
11246 The standard system libraries are used normally, unless @option{-nostdlib}
11247 or @option{-nodefaultlibs} is used.
11249 @item -nodefaultlibs
11250 @opindex nodefaultlibs
11251 Do not use the standard system libraries when linking.
11252 Only the libraries you specify are passed to the linker, and options
11253 specifying linkage of the system libraries, such as @option{-static-libgcc}
11254 or @option{-shared-libgcc}, are ignored.
11255 The standard startup files are used normally, unless @option{-nostartfiles}
11258 The compiler may generate calls to @code{memcmp},
11259 @code{memset}, @code{memcpy} and @code{memmove}.
11260 These entries are usually resolved by entries in
11261 libc. These entry points should be supplied through some other
11262 mechanism when this option is specified.
11266 Do not use the standard system startup files or libraries when linking.
11267 No startup files and only the libraries you specify are passed to
11268 the linker, and options specifying linkage of the system libraries, such as
11269 @option{-static-libgcc} or @option{-shared-libgcc}, are ignored.
11271 The compiler may generate calls to @code{memcmp}, @code{memset},
11272 @code{memcpy} and @code{memmove}.
11273 These entries are usually resolved by entries in
11274 libc. These entry points should be supplied through some other
11275 mechanism when this option is specified.
11277 @cindex @option{-lgcc}, use with @option{-nostdlib}
11278 @cindex @option{-nostdlib} and unresolved references
11279 @cindex unresolved references and @option{-nostdlib}
11280 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
11281 @cindex @option{-nodefaultlibs} and unresolved references
11282 @cindex unresolved references and @option{-nodefaultlibs}
11283 One of the standard libraries bypassed by @option{-nostdlib} and
11284 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
11285 which GCC uses to overcome shortcomings of particular machines, or special
11286 needs for some languages.
11287 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
11288 Collection (GCC) Internals},
11289 for more discussion of @file{libgcc.a}.)
11290 In most cases, you need @file{libgcc.a} even when you want to avoid
11291 other standard libraries. In other words, when you specify @option{-nostdlib}
11292 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
11293 This ensures that you have no unresolved references to internal GCC
11294 library subroutines.
11295 (An example of such an internal subroutine is @code{__main}, used to ensure C++
11296 constructors are called; @pxref{Collect2,,@code{collect2}, gccint,
11297 GNU Compiler Collection (GCC) Internals}.)
11301 Produce a position independent executable on targets that support it.
11302 For predictable results, you must also specify the same set of options
11303 used for compilation (@option{-fpie}, @option{-fPIE},
11304 or model suboptions) when you specify this linker option.
11308 Don't produce a position independent executable.
11312 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
11313 that support it. This instructs the linker to add all symbols, not
11314 only used ones, to the dynamic symbol table. This option is needed
11315 for some uses of @code{dlopen} or to allow obtaining backtraces
11316 from within a program.
11320 Remove all symbol table and relocation information from the executable.
11324 On systems that support dynamic linking, this prevents linking with the shared
11325 libraries. On other systems, this option has no effect.
11329 Produce a shared object which can then be linked with other objects to
11330 form an executable. Not all systems support this option. For predictable
11331 results, you must also specify the same set of options used for compilation
11332 (@option{-fpic}, @option{-fPIC}, or model suboptions) when
11333 you specify this linker option.@footnote{On some systems, @samp{gcc -shared}
11334 needs to build supplementary stub code for constructors to work. On
11335 multi-libbed systems, @samp{gcc -shared} must select the correct support
11336 libraries to link against. Failing to supply the correct flags may lead
11337 to subtle defects. Supplying them in cases where they are not necessary
11340 @item -shared-libgcc
11341 @itemx -static-libgcc
11342 @opindex shared-libgcc
11343 @opindex static-libgcc
11344 On systems that provide @file{libgcc} as a shared library, these options
11345 force the use of either the shared or static version, respectively.
11346 If no shared version of @file{libgcc} was built when the compiler was
11347 configured, these options have no effect.
11349 There are several situations in which an application should use the
11350 shared @file{libgcc} instead of the static version. The most common
11351 of these is when the application wishes to throw and catch exceptions
11352 across different shared libraries. In that case, each of the libraries
11353 as well as the application itself should use the shared @file{libgcc}.
11355 Therefore, the G++ and GCJ drivers automatically add
11356 @option{-shared-libgcc} whenever you build a shared library or a main
11357 executable, because C++ and Java programs typically use exceptions, so
11358 this is the right thing to do.
11360 If, instead, you use the GCC driver to create shared libraries, you may
11361 find that they are not always linked with the shared @file{libgcc}.
11362 If GCC finds, at its configuration time, that you have a non-GNU linker
11363 or a GNU linker that does not support option @option{--eh-frame-hdr},
11364 it links the shared version of @file{libgcc} into shared libraries
11365 by default. Otherwise, it takes advantage of the linker and optimizes
11366 away the linking with the shared version of @file{libgcc}, linking with
11367 the static version of libgcc by default. This allows exceptions to
11368 propagate through such shared libraries, without incurring relocation
11369 costs at library load time.
11371 However, if a library or main executable is supposed to throw or catch
11372 exceptions, you must link it using the G++ or GCJ driver, as appropriate
11373 for the languages used in the program, or using the option
11374 @option{-shared-libgcc}, such that it is linked with the shared
11377 @item -static-libasan
11378 @opindex static-libasan
11379 When the @option{-fsanitize=address} option is used to link a program,
11380 the GCC driver automatically links against @option{libasan}. If
11381 @file{libasan} is available as a shared library, and the @option{-static}
11382 option is not used, then this links against the shared version of
11383 @file{libasan}. The @option{-static-libasan} option directs the GCC
11384 driver to link @file{libasan} statically, without necessarily linking
11385 other libraries statically.
11387 @item -static-libtsan
11388 @opindex static-libtsan
11389 When the @option{-fsanitize=thread} option is used to link a program,
11390 the GCC driver automatically links against @option{libtsan}. If
11391 @file{libtsan} is available as a shared library, and the @option{-static}
11392 option is not used, then this links against the shared version of
11393 @file{libtsan}. The @option{-static-libtsan} option directs the GCC
11394 driver to link @file{libtsan} statically, without necessarily linking
11395 other libraries statically.
11397 @item -static-liblsan
11398 @opindex static-liblsan
11399 When the @option{-fsanitize=leak} option is used to link a program,
11400 the GCC driver automatically links against @option{liblsan}. If
11401 @file{liblsan} is available as a shared library, and the @option{-static}
11402 option is not used, then this links against the shared version of
11403 @file{liblsan}. The @option{-static-liblsan} option directs the GCC
11404 driver to link @file{liblsan} statically, without necessarily linking
11405 other libraries statically.
11407 @item -static-libubsan
11408 @opindex static-libubsan
11409 When the @option{-fsanitize=undefined} option is used to link a program,
11410 the GCC driver automatically links against @option{libubsan}. If
11411 @file{libubsan} is available as a shared library, and the @option{-static}
11412 option is not used, then this links against the shared version of
11413 @file{libubsan}. The @option{-static-libubsan} option directs the GCC
11414 driver to link @file{libubsan} statically, without necessarily linking
11415 other libraries statically.
11417 @item -static-libmpx
11418 @opindex static-libmpx
11419 When the @option{-fcheck-pointer bounds} and @option{-mmpx} options are
11420 used to link a program, the GCC driver automatically links against
11421 @file{libmpx}. If @file{libmpx} is available as a shared library,
11422 and the @option{-static} option is not used, then this links against
11423 the shared version of @file{libmpx}. The @option{-static-libmpx}
11424 option directs the GCC driver to link @file{libmpx} statically,
11425 without necessarily linking other libraries statically.
11427 @item -static-libmpxwrappers
11428 @opindex static-libmpxwrappers
11429 When the @option{-fcheck-pointer bounds} and @option{-mmpx} options are used
11430 to link a program without also using @option{-fno-chkp-use-wrappers}, the
11431 GCC driver automatically links against @file{libmpxwrappers}. If
11432 @file{libmpxwrappers} is available as a shared library, and the
11433 @option{-static} option is not used, then this links against the shared
11434 version of @file{libmpxwrappers}. The @option{-static-libmpxwrappers}
11435 option directs the GCC driver to link @file{libmpxwrappers} statically,
11436 without necessarily linking other libraries statically.
11438 @item -static-libstdc++
11439 @opindex static-libstdc++
11440 When the @command{g++} program is used to link a C++ program, it
11441 normally automatically links against @option{libstdc++}. If
11442 @file{libstdc++} is available as a shared library, and the
11443 @option{-static} option is not used, then this links against the
11444 shared version of @file{libstdc++}. That is normally fine. However, it
11445 is sometimes useful to freeze the version of @file{libstdc++} used by
11446 the program without going all the way to a fully static link. The
11447 @option{-static-libstdc++} option directs the @command{g++} driver to
11448 link @file{libstdc++} statically, without necessarily linking other
11449 libraries statically.
11453 Bind references to global symbols when building a shared object. Warn
11454 about any unresolved references (unless overridden by the link editor
11455 option @option{-Xlinker -z -Xlinker defs}). Only a few systems support
11458 @item -T @var{script}
11460 @cindex linker script
11461 Use @var{script} as the linker script. This option is supported by most
11462 systems using the GNU linker. On some targets, such as bare-board
11463 targets without an operating system, the @option{-T} option may be required
11464 when linking to avoid references to undefined symbols.
11466 @item -Xlinker @var{option}
11468 Pass @var{option} as an option to the linker. You can use this to
11469 supply system-specific linker options that GCC does not recognize.
11471 If you want to pass an option that takes a separate argument, you must use
11472 @option{-Xlinker} twice, once for the option and once for the argument.
11473 For example, to pass @option{-assert definitions}, you must write
11474 @option{-Xlinker -assert -Xlinker definitions}. It does not work to write
11475 @option{-Xlinker "-assert definitions"}, because this passes the entire
11476 string as a single argument, which is not what the linker expects.
11478 When using the GNU linker, it is usually more convenient to pass
11479 arguments to linker options using the @option{@var{option}=@var{value}}
11480 syntax than as separate arguments. For example, you can specify
11481 @option{-Xlinker -Map=output.map} rather than
11482 @option{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
11483 this syntax for command-line options.
11485 @item -Wl,@var{option}
11487 Pass @var{option} as an option to the linker. If @var{option} contains
11488 commas, it is split into multiple options at the commas. You can use this
11489 syntax to pass an argument to the option.
11490 For example, @option{-Wl,-Map,output.map} passes @option{-Map output.map} to the
11491 linker. When using the GNU linker, you can also get the same effect with
11492 @option{-Wl,-Map=output.map}.
11494 @item -u @var{symbol}
11496 Pretend the symbol @var{symbol} is undefined, to force linking of
11497 library modules to define it. You can use @option{-u} multiple times with
11498 different symbols to force loading of additional library modules.
11500 @item -z @var{keyword}
11502 @option{-z} is passed directly on to the linker along with the keyword
11503 @var{keyword}. See the section in the documentation of your linker for
11504 permitted values and their meanings.
11507 @node Directory Options
11508 @section Options for Directory Search
11509 @cindex directory options
11510 @cindex options, directory search
11511 @cindex search path
11513 These options specify directories to search for header files, for
11514 libraries and for parts of the compiler:
11519 Add the directory @var{dir} to the head of the list of directories to be
11520 searched for header files. This can be used to override a system header
11521 file, substituting your own version, since these directories are
11522 searched before the system header file directories. However, you should
11523 not use this option to add directories that contain vendor-supplied
11524 system header files (use @option{-isystem} for that). If you use more than
11525 one @option{-I} option, the directories are scanned in left-to-right
11526 order; the standard system directories come after.
11528 If a standard system include directory, or a directory specified with
11529 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
11530 option is ignored. The directory is still searched but as a
11531 system directory at its normal position in the system include chain.
11532 This is to ensure that GCC's procedure to fix buggy system headers and
11533 the ordering for the @code{include_next} directive are not inadvertently changed.
11534 If you really need to change the search order for system directories,
11535 use the @option{-nostdinc} and/or @option{-isystem} options.
11537 @item -iplugindir=@var{dir}
11538 @opindex iplugindir=
11539 Set the directory to search for plugins that are passed
11540 by @option{-fplugin=@var{name}} instead of
11541 @option{-fplugin=@var{path}/@var{name}.so}. This option is not meant
11542 to be used by the user, but only passed by the driver.
11544 @item -iquote@var{dir}
11546 Add the directory @var{dir} to the head of the list of directories to
11547 be searched for header files only for the case of @code{#include
11548 "@var{file}"}; they are not searched for @code{#include <@var{file}>},
11549 otherwise just like @option{-I}.
11553 Add directory @var{dir} to the list of directories to be searched
11556 @item -B@var{prefix}
11558 This option specifies where to find the executables, libraries,
11559 include files, and data files of the compiler itself.
11561 The compiler driver program runs one or more of the subprograms
11562 @command{cpp}, @command{cc1}, @command{as} and @command{ld}. It tries
11563 @var{prefix} as a prefix for each program it tries to run, both with and
11564 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
11566 For each subprogram to be run, the compiler driver first tries the
11567 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
11568 is not specified, the driver tries two standard prefixes,
11569 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
11570 those results in a file name that is found, the unmodified program
11571 name is searched for using the directories specified in your
11572 @env{PATH} environment variable.
11574 The compiler checks to see if the path provided by @option{-B}
11575 refers to a directory, and if necessary it adds a directory
11576 separator character at the end of the path.
11578 @option{-B} prefixes that effectively specify directory names also apply
11579 to libraries in the linker, because the compiler translates these
11580 options into @option{-L} options for the linker. They also apply to
11581 include files in the preprocessor, because the compiler translates these
11582 options into @option{-isystem} options for the preprocessor. In this case,
11583 the compiler appends @samp{include} to the prefix.
11585 The runtime support file @file{libgcc.a} can also be searched for using
11586 the @option{-B} prefix, if needed. If it is not found there, the two
11587 standard prefixes above are tried, and that is all. The file is left
11588 out of the link if it is not found by those means.
11590 Another way to specify a prefix much like the @option{-B} prefix is to use
11591 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
11594 As a special kludge, if the path provided by @option{-B} is
11595 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
11596 9, then it is replaced by @file{[dir/]include}. This is to help
11597 with boot-strapping the compiler.
11599 @item -specs=@var{file}
11601 Process @var{file} after the compiler reads in the standard @file{specs}
11602 file, in order to override the defaults which the @command{gcc} driver
11603 program uses when determining what switches to pass to @command{cc1},
11604 @command{cc1plus}, @command{as}, @command{ld}, etc. More than one
11605 @option{-specs=@var{file}} can be specified on the command line, and they
11606 are processed in order, from left to right.
11608 @item --sysroot=@var{dir}
11610 Use @var{dir} as the logical root directory for headers and libraries.
11611 For example, if the compiler normally searches for headers in
11612 @file{/usr/include} and libraries in @file{/usr/lib}, it instead
11613 searches @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
11615 If you use both this option and the @option{-isysroot} option, then
11616 the @option{--sysroot} option applies to libraries, but the
11617 @option{-isysroot} option applies to header files.
11619 The GNU linker (beginning with version 2.16) has the necessary support
11620 for this option. If your linker does not support this option, the
11621 header file aspect of @option{--sysroot} still works, but the
11622 library aspect does not.
11624 @item --no-sysroot-suffix
11625 @opindex no-sysroot-suffix
11626 For some targets, a suffix is added to the root directory specified
11627 with @option{--sysroot}, depending on the other options used, so that
11628 headers may for example be found in
11629 @file{@var{dir}/@var{suffix}/usr/include} instead of
11630 @file{@var{dir}/usr/include}. This option disables the addition of
11635 This option has been deprecated. Please use @option{-iquote} instead for
11636 @option{-I} directories before the @option{-I-} and remove the @option{-I-}
11638 Any directories you specify with @option{-I} options before the @option{-I-}
11639 option are searched only for the case of @code{#include "@var{file}"};
11640 they are not searched for @code{#include <@var{file}>}.
11642 If additional directories are specified with @option{-I} options after
11643 the @option{-I-} option, these directories are searched for all @code{#include}
11644 directives. (Ordinarily @emph{all} @option{-I} directories are used
11647 In addition, the @option{-I-} option inhibits the use of the current
11648 directory (where the current input file came from) as the first search
11649 directory for @code{#include "@var{file}"}. There is no way to
11650 override this effect of @option{-I-}. With @option{-I.} you can specify
11651 searching the directory that is current when the compiler is
11652 invoked. That is not exactly the same as what the preprocessor does
11653 by default, but it is often satisfactory.
11655 @option{-I-} does not inhibit the use of the standard system directories
11656 for header files. Thus, @option{-I-} and @option{-nostdinc} are
11663 @section Specifying Subprocesses and the Switches to Pass to Them
11666 @command{gcc} is a driver program. It performs its job by invoking a
11667 sequence of other programs to do the work of compiling, assembling and
11668 linking. GCC interprets its command-line parameters and uses these to
11669 deduce which programs it should invoke, and which command-line options
11670 it ought to place on their command lines. This behavior is controlled
11671 by @dfn{spec strings}. In most cases there is one spec string for each
11672 program that GCC can invoke, but a few programs have multiple spec
11673 strings to control their behavior. The spec strings built into GCC can
11674 be overridden by using the @option{-specs=} command-line switch to specify
11677 @dfn{Spec files} are plaintext files that are used to construct spec
11678 strings. They consist of a sequence of directives separated by blank
11679 lines. The type of directive is determined by the first non-whitespace
11680 character on the line, which can be one of the following:
11683 @item %@var{command}
11684 Issues a @var{command} to the spec file processor. The commands that can
11688 @item %include <@var{file}>
11689 @cindex @code{%include}
11690 Search for @var{file} and insert its text at the current point in the
11693 @item %include_noerr <@var{file}>
11694 @cindex @code{%include_noerr}
11695 Just like @samp{%include}, but do not generate an error message if the include
11696 file cannot be found.
11698 @item %rename @var{old_name} @var{new_name}
11699 @cindex @code{%rename}
11700 Rename the spec string @var{old_name} to @var{new_name}.
11704 @item *[@var{spec_name}]:
11705 This tells the compiler to create, override or delete the named spec
11706 string. All lines after this directive up to the next directive or
11707 blank line are considered to be the text for the spec string. If this
11708 results in an empty string then the spec is deleted. (Or, if the
11709 spec did not exist, then nothing happens.) Otherwise, if the spec
11710 does not currently exist a new spec is created. If the spec does
11711 exist then its contents are overridden by the text of this
11712 directive, unless the first character of that text is the @samp{+}
11713 character, in which case the text is appended to the spec.
11715 @item [@var{suffix}]:
11716 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
11717 and up to the next directive or blank line are considered to make up the
11718 spec string for the indicated suffix. When the compiler encounters an
11719 input file with the named suffix, it processes the spec string in
11720 order to work out how to compile that file. For example:
11724 z-compile -input %i
11727 This says that any input file whose name ends in @samp{.ZZ} should be
11728 passed to the program @samp{z-compile}, which should be invoked with the
11729 command-line switch @option{-input} and with the result of performing the
11730 @samp{%i} substitution. (See below.)
11732 As an alternative to providing a spec string, the text following a
11733 suffix directive can be one of the following:
11736 @item @@@var{language}
11737 This says that the suffix is an alias for a known @var{language}. This is
11738 similar to using the @option{-x} command-line switch to GCC to specify a
11739 language explicitly. For example:
11746 Says that .ZZ files are, in fact, C++ source files.
11749 This causes an error messages saying:
11752 @var{name} compiler not installed on this system.
11756 GCC already has an extensive list of suffixes built into it.
11757 This directive adds an entry to the end of the list of suffixes, but
11758 since the list is searched from the end backwards, it is effectively
11759 possible to override earlier entries using this technique.
11763 GCC has the following spec strings built into it. Spec files can
11764 override these strings or create their own. Note that individual
11765 targets can also add their own spec strings to this list.
11768 asm Options to pass to the assembler
11769 asm_final Options to pass to the assembler post-processor
11770 cpp Options to pass to the C preprocessor
11771 cc1 Options to pass to the C compiler
11772 cc1plus Options to pass to the C++ compiler
11773 endfile Object files to include at the end of the link
11774 link Options to pass to the linker
11775 lib Libraries to include on the command line to the linker
11776 libgcc Decides which GCC support library to pass to the linker
11777 linker Sets the name of the linker
11778 predefines Defines to be passed to the C preprocessor
11779 signed_char Defines to pass to CPP to say whether @code{char} is signed
11781 startfile Object files to include at the start of the link
11784 Here is a small example of a spec file:
11787 %rename lib old_lib
11790 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
11793 This example renames the spec called @samp{lib} to @samp{old_lib} and
11794 then overrides the previous definition of @samp{lib} with a new one.
11795 The new definition adds in some extra command-line options before
11796 including the text of the old definition.
11798 @dfn{Spec strings} are a list of command-line options to be passed to their
11799 corresponding program. In addition, the spec strings can contain
11800 @samp{%}-prefixed sequences to substitute variable text or to
11801 conditionally insert text into the command line. Using these constructs
11802 it is possible to generate quite complex command lines.
11804 Here is a table of all defined @samp{%}-sequences for spec
11805 strings. Note that spaces are not generated automatically around the
11806 results of expanding these sequences. Therefore you can concatenate them
11807 together or combine them with constant text in a single argument.
11811 Substitute one @samp{%} into the program name or argument.
11814 Substitute the name of the input file being processed.
11817 Substitute the basename of the input file being processed.
11818 This is the substring up to (and not including) the last period
11819 and not including the directory.
11822 This is the same as @samp{%b}, but include the file suffix (text after
11826 Marks the argument containing or following the @samp{%d} as a
11827 temporary file name, so that that file is deleted if GCC exits
11828 successfully. Unlike @samp{%g}, this contributes no text to the
11831 @item %g@var{suffix}
11832 Substitute a file name that has suffix @var{suffix} and is chosen
11833 once per compilation, and mark the argument in the same way as
11834 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
11835 name is now chosen in a way that is hard to predict even when previously
11836 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
11837 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
11838 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
11839 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
11840 was simply substituted with a file name chosen once per compilation,
11841 without regard to any appended suffix (which was therefore treated
11842 just like ordinary text), making such attacks more likely to succeed.
11844 @item %u@var{suffix}
11845 Like @samp{%g}, but generates a new temporary file name
11846 each time it appears instead of once per compilation.
11848 @item %U@var{suffix}
11849 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
11850 new one if there is no such last file name. In the absence of any
11851 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
11852 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
11853 involves the generation of two distinct file names, one
11854 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
11855 simply substituted with a file name chosen for the previous @samp{%u},
11856 without regard to any appended suffix.
11858 @item %j@var{suffix}
11859 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
11860 writable, and if @option{-save-temps} is not used;
11861 otherwise, substitute the name
11862 of a temporary file, just like @samp{%u}. This temporary file is not
11863 meant for communication between processes, but rather as a junk
11864 disposal mechanism.
11866 @item %|@var{suffix}
11867 @itemx %m@var{suffix}
11868 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
11869 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
11870 all. These are the two most common ways to instruct a program that it
11871 should read from standard input or write to standard output. If you
11872 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
11873 construct: see for example @file{f/lang-specs.h}.
11875 @item %.@var{SUFFIX}
11876 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
11877 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
11878 terminated by the next space or %.
11881 Marks the argument containing or following the @samp{%w} as the
11882 designated output file of this compilation. This puts the argument
11883 into the sequence of arguments that @samp{%o} substitutes.
11886 Substitutes the names of all the output files, with spaces
11887 automatically placed around them. You should write spaces
11888 around the @samp{%o} as well or the results are undefined.
11889 @samp{%o} is for use in the specs for running the linker.
11890 Input files whose names have no recognized suffix are not compiled
11891 at all, but they are included among the output files, so they are
11895 Substitutes the suffix for object files. Note that this is
11896 handled specially when it immediately follows @samp{%g, %u, or %U},
11897 because of the need for those to form complete file names. The
11898 handling is such that @samp{%O} is treated exactly as if it had already
11899 been substituted, except that @samp{%g, %u, and %U} do not currently
11900 support additional @var{suffix} characters following @samp{%O} as they do
11901 following, for example, @samp{.o}.
11904 Substitutes the standard macro predefinitions for the
11905 current target machine. Use this when running @command{cpp}.
11908 Like @samp{%p}, but puts @samp{__} before and after the name of each
11909 predefined macro, except for macros that start with @samp{__} or with
11910 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
11914 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
11915 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
11916 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
11917 and @option{-imultilib} as necessary.
11920 Current argument is the name of a library or startup file of some sort.
11921 Search for that file in a standard list of directories and substitute
11922 the full name found. The current working directory is included in the
11923 list of directories scanned.
11926 Current argument is the name of a linker script. Search for that file
11927 in the current list of directories to scan for libraries. If the file
11928 is located insert a @option{--script} option into the command line
11929 followed by the full path name found. If the file is not found then
11930 generate an error message. Note: the current working directory is not
11934 Print @var{str} as an error message. @var{str} is terminated by a newline.
11935 Use this when inconsistent options are detected.
11937 @item %(@var{name})
11938 Substitute the contents of spec string @var{name} at this point.
11940 @item %x@{@var{option}@}
11941 Accumulate an option for @samp{%X}.
11944 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
11948 Output the accumulated assembler options specified by @option{-Wa}.
11951 Output the accumulated preprocessor options specified by @option{-Wp}.
11954 Process the @code{asm} spec. This is used to compute the
11955 switches to be passed to the assembler.
11958 Process the @code{asm_final} spec. This is a spec string for
11959 passing switches to an assembler post-processor, if such a program is
11963 Process the @code{link} spec. This is the spec for computing the
11964 command line passed to the linker. Typically it makes use of the
11965 @samp{%L %G %S %D and %E} sequences.
11968 Dump out a @option{-L} option for each directory that GCC believes might
11969 contain startup files. If the target supports multilibs then the
11970 current multilib directory is prepended to each of these paths.
11973 Process the @code{lib} spec. This is a spec string for deciding which
11974 libraries are included on the command line to the linker.
11977 Process the @code{libgcc} spec. This is a spec string for deciding
11978 which GCC support library is included on the command line to the linker.
11981 Process the @code{startfile} spec. This is a spec for deciding which
11982 object files are the first ones passed to the linker. Typically
11983 this might be a file named @file{crt0.o}.
11986 Process the @code{endfile} spec. This is a spec string that specifies
11987 the last object files that are passed to the linker.
11990 Process the @code{cpp} spec. This is used to construct the arguments
11991 to be passed to the C preprocessor.
11994 Process the @code{cc1} spec. This is used to construct the options to be
11995 passed to the actual C compiler (@command{cc1}).
11998 Process the @code{cc1plus} spec. This is used to construct the options to be
11999 passed to the actual C++ compiler (@command{cc1plus}).
12002 Substitute the variable part of a matched option. See below.
12003 Note that each comma in the substituted string is replaced by
12007 Remove all occurrences of @code{-S} from the command line. Note---this
12008 command is position dependent. @samp{%} commands in the spec string
12009 before this one see @code{-S}, @samp{%} commands in the spec string
12010 after this one do not.
12012 @item %:@var{function}(@var{args})
12013 Call the named function @var{function}, passing it @var{args}.
12014 @var{args} is first processed as a nested spec string, then split
12015 into an argument vector in the usual fashion. The function returns
12016 a string which is processed as if it had appeared literally as part
12017 of the current spec.
12019 The following built-in spec functions are provided:
12022 @item @code{getenv}
12023 The @code{getenv} spec function takes two arguments: an environment
12024 variable name and a string. If the environment variable is not
12025 defined, a fatal error is issued. Otherwise, the return value is the
12026 value of the environment variable concatenated with the string. For
12027 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
12030 %:getenv(TOPDIR /include)
12033 expands to @file{/path/to/top/include}.
12035 @item @code{if-exists}
12036 The @code{if-exists} spec function takes one argument, an absolute
12037 pathname to a file. If the file exists, @code{if-exists} returns the
12038 pathname. Here is a small example of its usage:
12042 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
12045 @item @code{if-exists-else}
12046 The @code{if-exists-else} spec function is similar to the @code{if-exists}
12047 spec function, except that it takes two arguments. The first argument is
12048 an absolute pathname to a file. If the file exists, @code{if-exists-else}
12049 returns the pathname. If it does not exist, it returns the second argument.
12050 This way, @code{if-exists-else} can be used to select one file or another,
12051 based on the existence of the first. Here is a small example of its usage:
12055 crt0%O%s %:if-exists(crti%O%s) \
12056 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
12059 @item @code{replace-outfile}
12060 The @code{replace-outfile} spec function takes two arguments. It looks for the
12061 first argument in the outfiles array and replaces it with the second argument. Here
12062 is a small example of its usage:
12065 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
12068 @item @code{remove-outfile}
12069 The @code{remove-outfile} spec function takes one argument. It looks for the
12070 first argument in the outfiles array and removes it. Here is a small example
12074 %:remove-outfile(-lm)
12077 @item @code{pass-through-libs}
12078 The @code{pass-through-libs} spec function takes any number of arguments. It
12079 finds any @option{-l} options and any non-options ending in @file{.a} (which it
12080 assumes are the names of linker input library archive files) and returns a
12081 result containing all the found arguments each prepended by
12082 @option{-plugin-opt=-pass-through=} and joined by spaces. This list is
12083 intended to be passed to the LTO linker plugin.
12086 %:pass-through-libs(%G %L %G)
12089 @item @code{print-asm-header}
12090 The @code{print-asm-header} function takes no arguments and simply
12091 prints a banner like:
12097 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
12100 It is used to separate compiler options from assembler options
12101 in the @option{--target-help} output.
12104 @item %@{@code{S}@}
12105 Substitutes the @code{-S} switch, if that switch is given to GCC@.
12106 If that switch is not specified, this substitutes nothing. Note that
12107 the leading dash is omitted when specifying this option, and it is
12108 automatically inserted if the substitution is performed. Thus the spec
12109 string @samp{%@{foo@}} matches the command-line option @option{-foo}
12110 and outputs the command-line option @option{-foo}.
12112 @item %W@{@code{S}@}
12113 Like %@{@code{S}@} but mark last argument supplied within as a file to be
12114 deleted on failure.
12116 @item %@{@code{S}*@}
12117 Substitutes all the switches specified to GCC whose names start
12118 with @code{-S}, but which also take an argument. This is used for
12119 switches like @option{-o}, @option{-D}, @option{-I}, etc.
12120 GCC considers @option{-o foo} as being
12121 one switch whose name starts with @samp{o}. %@{o*@} substitutes this
12122 text, including the space. Thus two arguments are generated.
12124 @item %@{@code{S}*&@code{T}*@}
12125 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
12126 (the order of @code{S} and @code{T} in the spec is not significant).
12127 There can be any number of ampersand-separated variables; for each the
12128 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
12130 @item %@{@code{S}:@code{X}@}
12131 Substitutes @code{X}, if the @option{-S} switch is given to GCC@.
12133 @item %@{!@code{S}:@code{X}@}
12134 Substitutes @code{X}, if the @option{-S} switch is @emph{not} given to GCC@.
12136 @item %@{@code{S}*:@code{X}@}
12137 Substitutes @code{X} if one or more switches whose names start with
12138 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
12139 once, no matter how many such switches appeared. However, if @code{%*}
12140 appears somewhere in @code{X}, then @code{X} is substituted once
12141 for each matching switch, with the @code{%*} replaced by the part of
12142 that switch matching the @code{*}.
12144 If @code{%*} appears as the last part of a spec sequence then a space
12145 is added after the end of the last substitution. If there is more
12146 text in the sequence, however, then a space is not generated. This
12147 allows the @code{%*} substitution to be used as part of a larger
12148 string. For example, a spec string like this:
12151 %@{mcu=*:--script=%*/memory.ld@}
12155 when matching an option like @option{-mcu=newchip} produces:
12158 --script=newchip/memory.ld
12161 @item %@{.@code{S}:@code{X}@}
12162 Substitutes @code{X}, if processing a file with suffix @code{S}.
12164 @item %@{!.@code{S}:@code{X}@}
12165 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
12167 @item %@{,@code{S}:@code{X}@}
12168 Substitutes @code{X}, if processing a file for language @code{S}.
12170 @item %@{!,@code{S}:@code{X}@}
12171 Substitutes @code{X}, if not processing a file for language @code{S}.
12173 @item %@{@code{S}|@code{P}:@code{X}@}
12174 Substitutes @code{X} if either @code{-S} or @code{-P} is given to
12175 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
12176 @code{*} sequences as well, although they have a stronger binding than
12177 the @samp{|}. If @code{%*} appears in @code{X}, all of the
12178 alternatives must be starred, and only the first matching alternative
12181 For example, a spec string like this:
12184 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
12188 outputs the following command-line options from the following input
12189 command-line options:
12194 -d fred.c -foo -baz -boggle
12195 -d jim.d -bar -baz -boggle
12198 @item %@{S:X; T:Y; :D@}
12200 If @code{S} is given to GCC, substitutes @code{X}; else if @code{T} is
12201 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
12202 be as many clauses as you need. This may be combined with @code{.},
12203 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
12208 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
12209 construct may contain other nested @samp{%} constructs or spaces, or
12210 even newlines. They are processed as usual, as described above.
12211 Trailing white space in @code{X} is ignored. White space may also
12212 appear anywhere on the left side of the colon in these constructs,
12213 except between @code{.} or @code{*} and the corresponding word.
12215 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
12216 handled specifically in these constructs. If another value of
12217 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
12218 @option{-W} switch is found later in the command line, the earlier
12219 switch value is ignored, except with @{@code{S}*@} where @code{S} is
12220 just one letter, which passes all matching options.
12222 The character @samp{|} at the beginning of the predicate text is used to
12223 indicate that a command should be piped to the following command, but
12224 only if @option{-pipe} is specified.
12226 It is built into GCC which switches take arguments and which do not.
12227 (You might think it would be useful to generalize this to allow each
12228 compiler's spec to say which switches take arguments. But this cannot
12229 be done in a consistent fashion. GCC cannot even decide which input
12230 files have been specified without knowing which switches take arguments,
12231 and it must know which input files to compile in order to tell which
12234 GCC also knows implicitly that arguments starting in @option{-l} are to be
12235 treated as compiler output files, and passed to the linker in their
12236 proper position among the other output files.
12238 @c man begin OPTIONS
12240 @node Target Options
12241 @section Specifying Target Machine and Compiler Version
12242 @cindex target options
12243 @cindex cross compiling
12244 @cindex specifying machine version
12245 @cindex specifying compiler version and target machine
12246 @cindex compiler version, specifying
12247 @cindex target machine, specifying
12249 The usual way to run GCC is to run the executable called @command{gcc}, or
12250 @command{@var{machine}-gcc} when cross-compiling, or
12251 @command{@var{machine}-gcc-@var{version}} to run a version other than the
12252 one that was installed last.
12254 @node Submodel Options
12255 @section Hardware Models and Configurations
12256 @cindex submodel options
12257 @cindex specifying hardware config
12258 @cindex hardware models and configurations, specifying
12259 @cindex machine dependent options
12261 Each target machine types can have its own
12262 special options, starting with @samp{-m}, to choose among various
12263 hardware models or configurations---for example, 68010 vs 68020,
12264 floating coprocessor or none. A single installed version of the
12265 compiler can compile for any model or configuration, according to the
12268 Some configurations of the compiler also support additional special
12269 options, usually for compatibility with other compilers on the same
12272 @c This list is ordered alphanumerically by subsection name.
12273 @c It should be the same order and spelling as these options are listed
12274 @c in Machine Dependent Options
12277 * AArch64 Options::
12278 * Adapteva Epiphany Options::
12282 * Blackfin Options::
12287 * DEC Alpha Options::
12291 * GNU/Linux Options::
12301 * MicroBlaze Options::
12304 * MN10300 Options::
12308 * Nios II Options::
12309 * Nvidia PTX Options::
12311 * picoChip Options::
12312 * PowerPC Options::
12314 * RS/6000 and PowerPC Options::
12316 * S/390 and zSeries Options::
12319 * Solaris 2 Options::
12322 * System V Options::
12323 * TILE-Gx Options::
12324 * TILEPro Options::
12329 * VxWorks Options::
12331 * x86 Windows Options::
12332 * Xstormy16 Options::
12334 * zSeries Options::
12337 @node AArch64 Options
12338 @subsection AArch64 Options
12339 @cindex AArch64 Options
12341 These options are defined for AArch64 implementations:
12345 @item -mabi=@var{name}
12347 Generate code for the specified data model. Permissible values
12348 are @samp{ilp32} for SysV-like data model where int, long int and pointer
12349 are 32-bit, and @samp{lp64} for SysV-like data model where int is 32-bit,
12350 but long int and pointer are 64-bit.
12352 The default depends on the specific target configuration. Note that
12353 the LP64 and ILP32 ABIs are not link-compatible; you must compile your
12354 entire program with the same ABI, and link with a compatible set of libraries.
12357 @opindex mbig-endian
12358 Generate big-endian code. This is the default when GCC is configured for an
12359 @samp{aarch64_be-*-*} target.
12361 @item -mgeneral-regs-only
12362 @opindex mgeneral-regs-only
12363 Generate code which uses only the general-purpose registers. This is equivalent
12364 to feature modifier @option{nofp} of @option{-march} or @option{-mcpu}, except
12365 that @option{-mgeneral-regs-only} takes precedence over any conflicting feature
12366 modifier regardless of sequence.
12368 @item -mlittle-endian
12369 @opindex mlittle-endian
12370 Generate little-endian code. This is the default when GCC is configured for an
12371 @samp{aarch64-*-*} but not an @samp{aarch64_be-*-*} target.
12373 @item -mcmodel=tiny
12374 @opindex mcmodel=tiny
12375 Generate code for the tiny code model. The program and its statically defined
12376 symbols must be within 1GB of each other. Pointers are 64 bits. Programs can
12377 be statically or dynamically linked. This model is not fully implemented and
12378 mostly treated as @samp{small}.
12380 @item -mcmodel=small
12381 @opindex mcmodel=small
12382 Generate code for the small code model. The program and its statically defined
12383 symbols must be within 4GB of each other. Pointers are 64 bits. Programs can
12384 be statically or dynamically linked. This is the default code model.
12386 @item -mcmodel=large
12387 @opindex mcmodel=large
12388 Generate code for the large code model. This makes no assumptions about
12389 addresses and sizes of sections. Pointers are 64 bits. Programs can be
12390 statically linked only.
12392 @item -mstrict-align
12393 @opindex mstrict-align
12394 Do not assume that unaligned memory references are handled by the system.
12396 @item -momit-leaf-frame-pointer
12397 @itemx -mno-omit-leaf-frame-pointer
12398 @opindex momit-leaf-frame-pointer
12399 @opindex mno-omit-leaf-frame-pointer
12400 Omit or keep the frame pointer in leaf functions. The former behaviour is the
12403 @item -mtls-dialect=desc
12404 @opindex mtls-dialect=desc
12405 Use TLS descriptors as the thread-local storage mechanism for dynamic accesses
12406 of TLS variables. This is the default.
12408 @item -mtls-dialect=traditional
12409 @opindex mtls-dialect=traditional
12410 Use traditional TLS as the thread-local storage mechanism for dynamic accesses
12413 @item -mfix-cortex-a53-835769
12414 @itemx -mno-fix-cortex-a53-835769
12415 @opindex mfix-cortex-a53-835769
12416 @opindex mno-fix-cortex-a53-835769
12417 Enable or disable the workaround for the ARM Cortex-A53 erratum number 835769.
12418 This involves inserting a NOP instruction between memory instructions and
12419 64-bit integer multiply-accumulate instructions.
12421 @item -mfix-cortex-a53-843419
12422 @itemx -mno-fix-cortex-a53-843419
12423 @opindex mfix-cortex-a53-843419
12424 @opindex mno-fix-cortex-a53-843419
12425 Enable or disable the workaround for the ARM Cortex-A53 erratum number 843419.
12426 This erratum workaround is made at link time and this will only pass the
12427 corresponding flag to the linker.
12429 @item -march=@var{name}
12431 Specify the name of the target architecture, optionally suffixed by one or
12432 more feature modifiers. This option has the form
12433 @option{-march=@var{arch}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}.
12435 The permissible values for @var{arch} are @samp{armv8-a} or
12438 For the permissible values for @var{feature}, see the sub-section on
12439 @ref{aarch64-feature-modifiers,,@option{-march} and @option{-mcpu}
12440 Feature Modifiers}. Where conflicting feature modifiers are
12441 specified, the right-most feature is used.
12443 Additionally on native AArch64 GNU/Linux systems the value
12444 @samp{native} is available. This option causes the compiler to pick the
12445 architecture of the host system. If the compiler is unable to recognize the
12446 architecture of the host system this option has no effect.
12448 GCC uses @var{name} to determine what kind of instructions it can emit
12449 when generating assembly code. If @option{-march} is specified
12450 without either of @option{-mtune} or @option{-mcpu} also being
12451 specified, the code is tuned to perform well across a range of target
12452 processors implementing the target architecture.
12454 @item -mtune=@var{name}
12456 Specify the name of the target processor for which GCC should tune the
12457 performance of the code. Permissible values for this option are:
12458 @samp{generic}, @samp{cortex-a53}, @samp{cortex-a57}, @samp{cortex-a72},
12459 @samp{exynos-m1}, @samp{thunderx}, @samp{xgene1}.
12461 Additionally, this option can specify that GCC should tune the performance
12462 of the code for a big.LITTLE system. Permissible values for this
12463 option are: @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53}.
12465 Additionally on native AArch64 GNU/Linux systems the value
12466 @samp{native} is available. This option causes the compiler to pick
12467 the architecture of and tune the performance of the code for the
12468 processor of the host system. If the compiler is unable to recognize
12469 the processor of the host system this option has no effect.
12471 Where none of @option{-mtune=}, @option{-mcpu=} or @option{-march=}
12472 are specified, the code is tuned to perform well across a range
12473 of target processors.
12475 This option cannot be suffixed by feature modifiers.
12477 @item -mcpu=@var{name}
12479 Specify the name of the target processor, optionally suffixed by one
12480 or more feature modifiers. This option has the form
12481 @option{-mcpu=@var{cpu}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}, where
12482 the permissible values for @var{cpu} are the same as those available
12483 for @option{-mtune}. The permissible values for @var{feature} are
12484 documented in the sub-section on
12485 @ref{aarch64-feature-modifiers,,@option{-march} and @option{-mcpu}
12486 Feature Modifiers}. Where conflicting feature modifiers are
12487 specified, the right-most feature is used.
12489 Additionally on native AArch64 GNU/Linux systems the value
12490 @samp{native} is available. This option causes the compiler to tune
12491 the performance of the code for the processor of the host system. If
12492 the compiler is unable to recognize the processor of the host system
12493 this option has no effect.
12495 GCC uses @var{name} to determine what kind of instructions it can emit when
12496 generating assembly code (as if by @option{-march}) and to determine
12497 the target processor for which to tune for performance (as if
12498 by @option{-mtune}). Where this option is used in conjunction
12499 with @option{-march} or @option{-mtune}, those options take precedence
12500 over the appropriate part of this option.
12503 @subsubsection @option{-march} and @option{-mcpu} Feature Modifiers
12504 @anchor{aarch64-feature-modifiers}
12505 @cindex @option{-march} feature modifiers
12506 @cindex @option{-mcpu} feature modifiers
12507 Feature modifiers used with @option{-march} and @option{-mcpu} can be any of
12508 the following and their inverses @option{no@var{feature}}:
12512 Enable CRC extension.
12514 Enable Crypto extension. This also enables Advanced SIMD and floating-point
12517 Enable floating-point instructions. This is on by default for all possible
12518 values for options @option{-march} and @option{-mcpu}.
12520 Enable Advanced SIMD instructions. This also enables floating-point
12521 instructions. This is on by default for all possible values for options
12522 @option{-march} and @option{-mcpu}.
12524 Enable Large System Extension instructions.
12526 Enable Privileged Access Never support.
12528 Enable Limited Ordering Regions support.
12530 Enable ARMv8.1 Advanced SIMD instructions. This implies Advanced SIMD
12533 @item -moverride=@var{string}
12535 Override tuning decisions made by the back-end in response to a
12536 @option{-mtune=} switch. The syntax, semantics, and accepted values
12537 for @var{string} in this option are not guaranteed to be consistent
12540 This option is only intended to be useful when developing GCC.
12543 That is, @option{crypto} implies @option{simd} implies @option{fp}.
12544 Conversely, @option{nofp} (or equivalently, @option{-mgeneral-regs-only})
12545 implies @option{nosimd} implies @option{nocrypto}.
12547 @node Adapteva Epiphany Options
12548 @subsection Adapteva Epiphany Options
12550 These @samp{-m} options are defined for Adapteva Epiphany:
12553 @item -mhalf-reg-file
12554 @opindex mhalf-reg-file
12555 Don't allocate any register in the range @code{r32}@dots{}@code{r63}.
12556 That allows code to run on hardware variants that lack these registers.
12558 @item -mprefer-short-insn-regs
12559 @opindex mprefer-short-insn-regs
12560 Preferrentially allocate registers that allow short instruction generation.
12561 This can result in increased instruction count, so this may either reduce or
12562 increase overall code size.
12564 @item -mbranch-cost=@var{num}
12565 @opindex mbranch-cost
12566 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12567 This cost is only a heuristic and is not guaranteed to produce
12568 consistent results across releases.
12572 Enable the generation of conditional moves.
12574 @item -mnops=@var{num}
12576 Emit @var{num} NOPs before every other generated instruction.
12578 @item -mno-soft-cmpsf
12579 @opindex mno-soft-cmpsf
12580 For single-precision floating-point comparisons, emit an @code{fsub} instruction
12581 and test the flags. This is faster than a software comparison, but can
12582 get incorrect results in the presence of NaNs, or when two different small
12583 numbers are compared such that their difference is calculated as zero.
12584 The default is @option{-msoft-cmpsf}, which uses slower, but IEEE-compliant,
12585 software comparisons.
12587 @item -mstack-offset=@var{num}
12588 @opindex mstack-offset
12589 Set the offset between the top of the stack and the stack pointer.
12590 E.g., a value of 8 means that the eight bytes in the range @code{sp+0@dots{}sp+7}
12591 can be used by leaf functions without stack allocation.
12592 Values other than @samp{8} or @samp{16} are untested and unlikely to work.
12593 Note also that this option changes the ABI; compiling a program with a
12594 different stack offset than the libraries have been compiled with
12595 generally does not work.
12596 This option can be useful if you want to evaluate if a different stack
12597 offset would give you better code, but to actually use a different stack
12598 offset to build working programs, it is recommended to configure the
12599 toolchain with the appropriate @option{--with-stack-offset=@var{num}} option.
12601 @item -mno-round-nearest
12602 @opindex mno-round-nearest
12603 Make the scheduler assume that the rounding mode has been set to
12604 truncating. The default is @option{-mround-nearest}.
12607 @opindex mlong-calls
12608 If not otherwise specified by an attribute, assume all calls might be beyond
12609 the offset range of the @code{b} / @code{bl} instructions, and therefore load the
12610 function address into a register before performing a (otherwise direct) call.
12611 This is the default.
12613 @item -mshort-calls
12614 @opindex short-calls
12615 If not otherwise specified by an attribute, assume all direct calls are
12616 in the range of the @code{b} / @code{bl} instructions, so use these instructions
12617 for direct calls. The default is @option{-mlong-calls}.
12621 Assume addresses can be loaded as 16-bit unsigned values. This does not
12622 apply to function addresses for which @option{-mlong-calls} semantics
12625 @item -mfp-mode=@var{mode}
12627 Set the prevailing mode of the floating-point unit.
12628 This determines the floating-point mode that is provided and expected
12629 at function call and return time. Making this mode match the mode you
12630 predominantly need at function start can make your programs smaller and
12631 faster by avoiding unnecessary mode switches.
12633 @var{mode} can be set to one the following values:
12637 Any mode at function entry is valid, and retained or restored when
12638 the function returns, and when it calls other functions.
12639 This mode is useful for compiling libraries or other compilation units
12640 you might want to incorporate into different programs with different
12641 prevailing FPU modes, and the convenience of being able to use a single
12642 object file outweighs the size and speed overhead for any extra
12643 mode switching that might be needed, compared with what would be needed
12644 with a more specific choice of prevailing FPU mode.
12647 This is the mode used for floating-point calculations with
12648 truncating (i.e.@: round towards zero) rounding mode. That includes
12649 conversion from floating point to integer.
12651 @item round-nearest
12652 This is the mode used for floating-point calculations with
12653 round-to-nearest-or-even rounding mode.
12656 This is the mode used to perform integer calculations in the FPU, e.g.@:
12657 integer multiply, or integer multiply-and-accumulate.
12660 The default is @option{-mfp-mode=caller}
12662 @item -mnosplit-lohi
12663 @itemx -mno-postinc
12664 @itemx -mno-postmodify
12665 @opindex mnosplit-lohi
12666 @opindex mno-postinc
12667 @opindex mno-postmodify
12668 Code generation tweaks that disable, respectively, splitting of 32-bit
12669 loads, generation of post-increment addresses, and generation of
12670 post-modify addresses. The defaults are @option{msplit-lohi},
12671 @option{-mpost-inc}, and @option{-mpost-modify}.
12673 @item -mnovect-double
12674 @opindex mno-vect-double
12675 Change the preferred SIMD mode to SImode. The default is
12676 @option{-mvect-double}, which uses DImode as preferred SIMD mode.
12678 @item -max-vect-align=@var{num}
12679 @opindex max-vect-align
12680 The maximum alignment for SIMD vector mode types.
12681 @var{num} may be 4 or 8. The default is 8.
12682 Note that this is an ABI change, even though many library function
12683 interfaces are unaffected if they don't use SIMD vector modes
12684 in places that affect size and/or alignment of relevant types.
12686 @item -msplit-vecmove-early
12687 @opindex msplit-vecmove-early
12688 Split vector moves into single word moves before reload. In theory this
12689 can give better register allocation, but so far the reverse seems to be
12690 generally the case.
12692 @item -m1reg-@var{reg}
12694 Specify a register to hold the constant @minus{}1, which makes loading small negative
12695 constants and certain bitmasks faster.
12696 Allowable values for @var{reg} are @samp{r43} and @samp{r63},
12697 which specify use of that register as a fixed register,
12698 and @samp{none}, which means that no register is used for this
12699 purpose. The default is @option{-m1reg-none}.
12704 @subsection ARC Options
12705 @cindex ARC options
12707 The following options control the architecture variant for which code
12710 @c architecture variants
12713 @item -mbarrel-shifter
12714 @opindex mbarrel-shifter
12715 Generate instructions supported by barrel shifter. This is the default
12716 unless @option{-mcpu=ARC601} is in effect.
12718 @item -mcpu=@var{cpu}
12720 Set architecture type, register usage, and instruction scheduling
12721 parameters for @var{cpu}. There are also shortcut alias options
12722 available for backward compatibility and convenience. Supported
12723 values for @var{cpu} are
12729 Compile for ARC600. Aliases: @option{-mA6}, @option{-mARC600}.
12733 Compile for ARC601. Alias: @option{-mARC601}.
12738 Compile for ARC700. Aliases: @option{-mA7}, @option{-mARC700}.
12739 This is the default when configured with @option{--with-cpu=arc700}@.
12744 @itemx -mdpfp-compact
12745 @opindex mdpfp-compact
12746 FPX: Generate Double Precision FPX instructions, tuned for the compact
12750 @opindex mdpfp-fast
12751 FPX: Generate Double Precision FPX instructions, tuned for the fast
12754 @item -mno-dpfp-lrsr
12755 @opindex mno-dpfp-lrsr
12756 Disable LR and SR instructions from using FPX extension aux registers.
12760 Generate Extended arithmetic instructions. Currently only
12761 @code{divaw}, @code{adds}, @code{subs}, and @code{sat16} are
12762 supported. This is always enabled for @option{-mcpu=ARC700}.
12766 Do not generate mpy instructions for ARC700.
12770 Generate 32x16 bit multiply and mac instructions.
12774 Generate mul64 and mulu64 instructions. Only valid for @option{-mcpu=ARC600}.
12778 Generate norm instruction. This is the default if @option{-mcpu=ARC700}
12783 @itemx -mspfp-compact
12784 @opindex mspfp-compact
12785 FPX: Generate Single Precision FPX instructions, tuned for the compact
12789 @opindex mspfp-fast
12790 FPX: Generate Single Precision FPX instructions, tuned for the fast
12795 Enable generation of ARC SIMD instructions via target-specific
12796 builtins. Only valid for @option{-mcpu=ARC700}.
12799 @opindex msoft-float
12800 This option ignored; it is provided for compatibility purposes only.
12801 Software floating point code is emitted by default, and this default
12802 can overridden by FPX options; @samp{mspfp}, @samp{mspfp-compact}, or
12803 @samp{mspfp-fast} for single precision, and @samp{mdpfp},
12804 @samp{mdpfp-compact}, or @samp{mdpfp-fast} for double precision.
12808 Generate swap instructions.
12812 The following options are passed through to the assembler, and also
12813 define preprocessor macro symbols.
12815 @c Flags used by the assembler, but for which we define preprocessor
12816 @c macro symbols as well.
12819 @opindex mdsp-packa
12820 Passed down to the assembler to enable the DSP Pack A extensions.
12821 Also sets the preprocessor symbol @code{__Xdsp_packa}.
12825 Passed down to the assembler to enable the dual viterbi butterfly
12826 extension. Also sets the preprocessor symbol @code{__Xdvbf}.
12828 @c ARC700 4.10 extension instruction
12831 Passed down to the assembler to enable the Locked Load/Store
12832 Conditional extension. Also sets the preprocessor symbol
12837 Passed down to the assembler. Also sets the preprocessor symbol
12838 @code{__Xxmac_d16}.
12842 Passed down to the assembler. Also sets the preprocessor symbol
12845 @c ARC700 4.10 extension instruction
12848 Passed down to the assembler to enable the 64-bit Time-Stamp Counter
12849 extension instruction. Also sets the preprocessor symbol
12852 @c ARC700 4.10 extension instruction
12855 Passed down to the assembler to enable the swap byte ordering
12856 extension instruction. Also sets the preprocessor symbol
12860 @opindex mtelephony
12861 Passed down to the assembler to enable dual and single operand
12862 instructions for telephony. Also sets the preprocessor symbol
12863 @code{__Xtelephony}.
12867 Passed down to the assembler to enable the XY Memory extension. Also
12868 sets the preprocessor symbol @code{__Xxy}.
12872 The following options control how the assembly code is annotated:
12874 @c Assembly annotation options
12878 Annotate assembler instructions with estimated addresses.
12880 @item -mannotate-align
12881 @opindex mannotate-align
12882 Explain what alignment considerations lead to the decision to make an
12883 instruction short or long.
12887 The following options are passed through to the linker:
12889 @c options passed through to the linker
12893 Passed through to the linker, to specify use of the @code{arclinux} emulation.
12894 This option is enabled by default in tool chains built for
12895 @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets
12896 when profiling is not requested.
12898 @item -marclinux_prof
12899 @opindex marclinux_prof
12900 Passed through to the linker, to specify use of the
12901 @code{arclinux_prof} emulation. This option is enabled by default in
12902 tool chains built for @w{@code{arc-linux-uclibc}} and
12903 @w{@code{arceb-linux-uclibc}} targets when profiling is requested.
12907 The following options control the semantics of generated code:
12909 @c semantically relevant code generation options
12911 @item -mepilogue-cfi
12912 @opindex mepilogue-cfi
12913 Enable generation of call frame information for epilogues.
12915 @item -mno-epilogue-cfi
12916 @opindex mno-epilogue-cfi
12917 Disable generation of call frame information for epilogues.
12920 @opindex mlong-calls
12921 Generate call insns as register indirect calls, thus providing access
12922 to the full 32-bit address range.
12924 @item -mmedium-calls
12925 @opindex mmedium-calls
12926 Don't use less than 25 bit addressing range for calls, which is the
12927 offset available for an unconditional branch-and-link
12928 instruction. Conditional execution of function calls is suppressed, to
12929 allow use of the 25-bit range, rather than the 21-bit range with
12930 conditional branch-and-link. This is the default for tool chains built
12931 for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets.
12935 Do not generate sdata references. This is the default for tool chains
12936 built for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}}
12940 @opindex mucb-mcount
12941 Instrument with mcount calls as used in UCB code. I.e. do the
12942 counting in the callee, not the caller. By default ARC instrumentation
12943 counts in the caller.
12945 @item -mvolatile-cache
12946 @opindex mvolatile-cache
12947 Use ordinarily cached memory accesses for volatile references. This is the
12950 @item -mno-volatile-cache
12951 @opindex mno-volatile-cache
12952 Enable cache bypass for volatile references.
12956 The following options fine tune code generation:
12957 @c code generation tuning options
12960 @opindex malign-call
12961 Do alignment optimizations for call instructions.
12963 @item -mauto-modify-reg
12964 @opindex mauto-modify-reg
12965 Enable the use of pre/post modify with register displacement.
12967 @item -mbbit-peephole
12968 @opindex mbbit-peephole
12969 Enable bbit peephole2.
12973 This option disables a target-specific pass in @file{arc_reorg} to
12974 generate @code{BRcc} instructions. It has no effect on @code{BRcc}
12975 generation driven by the combiner pass.
12977 @item -mcase-vector-pcrel
12978 @opindex mcase-vector-pcrel
12979 Use pc-relative switch case tables - this enables case table shortening.
12980 This is the default for @option{-Os}.
12982 @item -mcompact-casesi
12983 @opindex mcompact-casesi
12984 Enable compact casesi pattern.
12985 This is the default for @option{-Os}.
12987 @item -mno-cond-exec
12988 @opindex mno-cond-exec
12989 Disable ARCompact specific pass to generate conditional execution instructions.
12990 Due to delay slot scheduling and interactions between operand numbers,
12991 literal sizes, instruction lengths, and the support for conditional execution,
12992 the target-independent pass to generate conditional execution is often lacking,
12993 so the ARC port has kept a special pass around that tries to find more
12994 conditional execution generating opportunities after register allocation,
12995 branch shortening, and delay slot scheduling have been done. This pass
12996 generally, but not always, improves performance and code size, at the cost of
12997 extra compilation time, which is why there is an option to switch it off.
12998 If you have a problem with call instructions exceeding their allowable
12999 offset range because they are conditionalized, you should consider using
13000 @option{-mmedium-calls} instead.
13002 @item -mearly-cbranchsi
13003 @opindex mearly-cbranchsi
13004 Enable pre-reload use of the cbranchsi pattern.
13006 @item -mexpand-adddi
13007 @opindex mexpand-adddi
13008 Expand @code{adddi3} and @code{subdi3} at rtl generation time into
13009 @code{add.f}, @code{adc} etc.
13011 @item -mindexed-loads
13012 @opindex mindexed-loads
13013 Enable the use of indexed loads. This can be problematic because some
13014 optimizers then assume that indexed stores exist, which is not
13019 Enable Local Register Allocation. This is still experimental for ARC,
13020 so by default the compiler uses standard reload
13021 (i.e. @option{-mno-lra}).
13023 @item -mlra-priority-none
13024 @opindex mlra-priority-none
13025 Don't indicate any priority for target registers.
13027 @item -mlra-priority-compact
13028 @opindex mlra-priority-compact
13029 Indicate target register priority for r0..r3 / r12..r15.
13031 @item -mlra-priority-noncompact
13032 @opindex mlra-priority-noncompact
13033 Reduce target regsiter priority for r0..r3 / r12..r15.
13035 @item -mno-millicode
13036 @opindex mno-millicode
13037 When optimizing for size (using @option{-Os}), prologues and epilogues
13038 that have to save or restore a large number of registers are often
13039 shortened by using call to a special function in libgcc; this is
13040 referred to as a @emph{millicode} call. As these calls can pose
13041 performance issues, and/or cause linking issues when linking in a
13042 nonstandard way, this option is provided to turn off millicode call
13046 @opindex mmixed-code
13047 Tweak register allocation to help 16-bit instruction generation.
13048 This generally has the effect of decreasing the average instruction size
13049 while increasing the instruction count.
13053 Enable 'q' instruction alternatives.
13054 This is the default for @option{-Os}.
13058 Enable Rcq constraint handling - most short code generation depends on this.
13059 This is the default.
13063 Enable Rcw constraint handling - ccfsm condexec mostly depends on this.
13064 This is the default.
13066 @item -msize-level=@var{level}
13067 @opindex msize-level
13068 Fine-tune size optimization with regards to instruction lengths and alignment.
13069 The recognized values for @var{level} are:
13072 No size optimization. This level is deprecated and treated like @samp{1}.
13075 Short instructions are used opportunistically.
13078 In addition, alignment of loops and of code after barriers are dropped.
13081 In addition, optional data alignment is dropped, and the option @option{Os} is enabled.
13085 This defaults to @samp{3} when @option{-Os} is in effect. Otherwise,
13086 the behavior when this is not set is equivalent to level @samp{1}.
13088 @item -mtune=@var{cpu}
13090 Set instruction scheduling parameters for @var{cpu}, overriding any implied
13091 by @option{-mcpu=}.
13093 Supported values for @var{cpu} are
13097 Tune for ARC600 cpu.
13100 Tune for ARC601 cpu.
13103 Tune for ARC700 cpu with standard multiplier block.
13106 Tune for ARC700 cpu with XMAC block.
13109 Tune for ARC725D cpu.
13112 Tune for ARC750D cpu.
13116 @item -mmultcost=@var{num}
13118 Cost to assume for a multiply instruction, with @samp{4} being equal to a
13119 normal instruction.
13121 @item -munalign-prob-threshold=@var{probability}
13122 @opindex munalign-prob-threshold
13123 Set probability threshold for unaligning branches.
13124 When tuning for @samp{ARC700} and optimizing for speed, branches without
13125 filled delay slot are preferably emitted unaligned and long, unless
13126 profiling indicates that the probability for the branch to be taken
13127 is below @var{probability}. @xref{Cross-profiling}.
13128 The default is (REG_BR_PROB_BASE/2), i.e.@: 5000.
13132 The following options are maintained for backward compatibility, but
13133 are now deprecated and will be removed in a future release:
13135 @c Deprecated options
13143 @opindex mbig-endian
13146 Compile code for big endian targets. Use of these options is now
13147 deprecated. Users wanting big-endian code, should use the
13148 @w{@code{arceb-elf32}} and @w{@code{arceb-linux-uclibc}} targets when
13149 building the tool chain, for which big-endian is the default.
13151 @item -mlittle-endian
13152 @opindex mlittle-endian
13155 Compile code for little endian targets. Use of these options is now
13156 deprecated. Users wanting little-endian code should use the
13157 @w{@code{arc-elf32}} and @w{@code{arc-linux-uclibc}} targets when
13158 building the tool chain, for which little-endian is the default.
13160 @item -mbarrel_shifter
13161 @opindex mbarrel_shifter
13162 Replaced by @option{-mbarrel-shifter}.
13164 @item -mdpfp_compact
13165 @opindex mdpfp_compact
13166 Replaced by @option{-mdpfp-compact}.
13169 @opindex mdpfp_fast
13170 Replaced by @option{-mdpfp-fast}.
13173 @opindex mdsp_packa
13174 Replaced by @option{-mdsp-packa}.
13178 Replaced by @option{-mea}.
13182 Replaced by @option{-mmac-24}.
13186 Replaced by @option{-mmac-d16}.
13188 @item -mspfp_compact
13189 @opindex mspfp_compact
13190 Replaced by @option{-mspfp-compact}.
13193 @opindex mspfp_fast
13194 Replaced by @option{-mspfp-fast}.
13196 @item -mtune=@var{cpu}
13198 Values @samp{arc600}, @samp{arc601}, @samp{arc700} and
13199 @samp{arc700-xmac} for @var{cpu} are replaced by @samp{ARC600},
13200 @samp{ARC601}, @samp{ARC700} and @samp{ARC700-xmac} respectively
13202 @item -multcost=@var{num}
13204 Replaced by @option{-mmultcost}.
13209 @subsection ARM Options
13210 @cindex ARM options
13212 These @samp{-m} options are defined for the ARM port:
13215 @item -mabi=@var{name}
13217 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
13218 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
13221 @opindex mapcs-frame
13222 Generate a stack frame that is compliant with the ARM Procedure Call
13223 Standard for all functions, even if this is not strictly necessary for
13224 correct execution of the code. Specifying @option{-fomit-frame-pointer}
13225 with this option causes the stack frames not to be generated for
13226 leaf functions. The default is @option{-mno-apcs-frame}.
13227 This option is deprecated.
13231 This is a synonym for @option{-mapcs-frame} and is deprecated.
13234 @c not currently implemented
13235 @item -mapcs-stack-check
13236 @opindex mapcs-stack-check
13237 Generate code to check the amount of stack space available upon entry to
13238 every function (that actually uses some stack space). If there is
13239 insufficient space available then either the function
13240 @code{__rt_stkovf_split_small} or @code{__rt_stkovf_split_big} is
13241 called, depending upon the amount of stack space required. The runtime
13242 system is required to provide these functions. The default is
13243 @option{-mno-apcs-stack-check}, since this produces smaller code.
13245 @c not currently implemented
13247 @opindex mapcs-float
13248 Pass floating-point arguments using the floating-point registers. This is
13249 one of the variants of the APCS@. This option is recommended if the
13250 target hardware has a floating-point unit or if a lot of floating-point
13251 arithmetic is going to be performed by the code. The default is
13252 @option{-mno-apcs-float}, since the size of integer-only code is
13253 slightly increased if @option{-mapcs-float} is used.
13255 @c not currently implemented
13256 @item -mapcs-reentrant
13257 @opindex mapcs-reentrant
13258 Generate reentrant, position-independent code. The default is
13259 @option{-mno-apcs-reentrant}.
13262 @item -mthumb-interwork
13263 @opindex mthumb-interwork
13264 Generate code that supports calling between the ARM and Thumb
13265 instruction sets. Without this option, on pre-v5 architectures, the
13266 two instruction sets cannot be reliably used inside one program. The
13267 default is @option{-mno-thumb-interwork}, since slightly larger code
13268 is generated when @option{-mthumb-interwork} is specified. In AAPCS
13269 configurations this option is meaningless.
13271 @item -mno-sched-prolog
13272 @opindex mno-sched-prolog
13273 Prevent the reordering of instructions in the function prologue, or the
13274 merging of those instruction with the instructions in the function's
13275 body. This means that all functions start with a recognizable set
13276 of instructions (or in fact one of a choice from a small set of
13277 different function prologues), and this information can be used to
13278 locate the start of functions inside an executable piece of code. The
13279 default is @option{-msched-prolog}.
13281 @item -mfloat-abi=@var{name}
13282 @opindex mfloat-abi
13283 Specifies which floating-point ABI to use. Permissible values
13284 are: @samp{soft}, @samp{softfp} and @samp{hard}.
13286 Specifying @samp{soft} causes GCC to generate output containing
13287 library calls for floating-point operations.
13288 @samp{softfp} allows the generation of code using hardware floating-point
13289 instructions, but still uses the soft-float calling conventions.
13290 @samp{hard} allows generation of floating-point instructions
13291 and uses FPU-specific calling conventions.
13293 The default depends on the specific target configuration. Note that
13294 the hard-float and soft-float ABIs are not link-compatible; you must
13295 compile your entire program with the same ABI, and link with a
13296 compatible set of libraries.
13298 @item -mlittle-endian
13299 @opindex mlittle-endian
13300 Generate code for a processor running in little-endian mode. This is
13301 the default for all standard configurations.
13304 @opindex mbig-endian
13305 Generate code for a processor running in big-endian mode; the default is
13306 to compile code for a little-endian processor.
13308 @item -march=@var{name}
13310 This specifies the name of the target ARM architecture. GCC uses this
13311 name to determine what kind of instructions it can emit when generating
13312 assembly code. This option can be used in conjunction with or instead
13313 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
13314 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
13315 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
13316 @samp{armv6}, @samp{armv6j},
13317 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
13318 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m}, @samp{armv7e-m},
13319 @samp{armv7ve}, @samp{armv8-a}, @samp{armv8-a+crc},
13320 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
13322 @option{-march=armv7ve} is the armv7-a architecture with virtualization
13325 @option{-march=armv8-a+crc} enables code generation for the ARMv8-A
13326 architecture together with the optional CRC32 extensions.
13328 @option{-march=native} causes the compiler to auto-detect the architecture
13329 of the build computer. At present, this feature is only supported on
13330 GNU/Linux, and not all architectures are recognized. If the auto-detect
13331 is unsuccessful the option has no effect.
13333 @item -mtune=@var{name}
13335 This option specifies the name of the target ARM processor for
13336 which GCC should tune the performance of the code.
13337 For some ARM implementations better performance can be obtained by using
13339 Permissible names are: @samp{arm2}, @samp{arm250},
13340 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
13341 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
13342 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
13343 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
13345 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
13346 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
13347 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
13348 @samp{strongarm1110},
13349 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
13350 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
13351 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
13352 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
13353 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
13354 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
13355 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
13356 @samp{generic-armv7-a}, @samp{cortex-a5}, @samp{cortex-a7}, @samp{cortex-a8},
13357 @samp{cortex-a9}, @samp{cortex-a12}, @samp{cortex-a15}, @samp{cortex-a17},
13358 @samp{cortex-a53}, @samp{cortex-a57}, @samp{cortex-a72},
13360 @samp{cortex-r4f}, @samp{cortex-r5}, @samp{cortex-r7}, @samp{cortex-m7},
13365 @samp{cortex-m0plus},
13366 @samp{cortex-m1.small-multiply},
13367 @samp{cortex-m0.small-multiply},
13368 @samp{cortex-m0plus.small-multiply},
13370 @samp{marvell-pj4},
13371 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312},
13372 @samp{fa526}, @samp{fa626},
13373 @samp{fa606te}, @samp{fa626te}, @samp{fmp626}, @samp{fa726te},
13376 Additionally, this option can specify that GCC should tune the performance
13377 of the code for a big.LITTLE system. Permissible names are:
13378 @samp{cortex-a15.cortex-a7}, @samp{cortex-a17.cortex-a7},
13379 @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53}.
13381 @option{-mtune=generic-@var{arch}} specifies that GCC should tune the
13382 performance for a blend of processors within architecture @var{arch}.
13383 The aim is to generate code that run well on the current most popular
13384 processors, balancing between optimizations that benefit some CPUs in the
13385 range, and avoiding performance pitfalls of other CPUs. The effects of
13386 this option may change in future GCC versions as CPU models come and go.
13388 @option{-mtune=native} causes the compiler to auto-detect the CPU
13389 of the build computer. At present, this feature is only supported on
13390 GNU/Linux, and not all architectures are recognized. If the auto-detect is
13391 unsuccessful the option has no effect.
13393 @item -mcpu=@var{name}
13395 This specifies the name of the target ARM processor. GCC uses this name
13396 to derive the name of the target ARM architecture (as if specified
13397 by @option{-march}) and the ARM processor type for which to tune for
13398 performance (as if specified by @option{-mtune}). Where this option
13399 is used in conjunction with @option{-march} or @option{-mtune},
13400 those options take precedence over the appropriate part of this option.
13402 Permissible names for this option are the same as those for
13405 @option{-mcpu=generic-@var{arch}} is also permissible, and is
13406 equivalent to @option{-march=@var{arch} -mtune=generic-@var{arch}}.
13407 See @option{-mtune} for more information.
13409 @option{-mcpu=native} causes the compiler to auto-detect the CPU
13410 of the build computer. At present, this feature is only supported on
13411 GNU/Linux, and not all architectures are recognized. If the auto-detect
13412 is unsuccessful the option has no effect.
13414 @item -mfpu=@var{name}
13416 This specifies what floating-point hardware (or hardware emulation) is
13417 available on the target. Permissible names are: @samp{vfp}, @samp{vfpv3},
13418 @samp{vfpv3-fp16}, @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd},
13419 @samp{vfpv3xd-fp16}, @samp{neon}, @samp{neon-fp16}, @samp{vfpv4},
13420 @samp{vfpv4-d16}, @samp{fpv4-sp-d16}, @samp{neon-vfpv4},
13421 @samp{fpv5-d16}, @samp{fpv5-sp-d16},
13422 @samp{fp-armv8}, @samp{neon-fp-armv8}, and @samp{crypto-neon-fp-armv8}.
13424 If @option{-msoft-float} is specified this specifies the format of
13425 floating-point values.
13427 If the selected floating-point hardware includes the NEON extension
13428 (e.g. @option{-mfpu}=@samp{neon}), note that floating-point
13429 operations are not generated by GCC's auto-vectorization pass unless
13430 @option{-funsafe-math-optimizations} is also specified. This is
13431 because NEON hardware does not fully implement the IEEE 754 standard for
13432 floating-point arithmetic (in particular denormal values are treated as
13433 zero), so the use of NEON instructions may lead to a loss of precision.
13435 @item -mfp16-format=@var{name}
13436 @opindex mfp16-format
13437 Specify the format of the @code{__fp16} half-precision floating-point type.
13438 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
13439 the default is @samp{none}, in which case the @code{__fp16} type is not
13440 defined. @xref{Half-Precision}, for more information.
13442 @item -mstructure-size-boundary=@var{n}
13443 @opindex mstructure-size-boundary
13444 The sizes of all structures and unions are rounded up to a multiple
13445 of the number of bits set by this option. Permissible values are 8, 32
13446 and 64. The default value varies for different toolchains. For the COFF
13447 targeted toolchain the default value is 8. A value of 64 is only allowed
13448 if the underlying ABI supports it.
13450 Specifying a larger number can produce faster, more efficient code, but
13451 can also increase the size of the program. Different values are potentially
13452 incompatible. Code compiled with one value cannot necessarily expect to
13453 work with code or libraries compiled with another value, if they exchange
13454 information using structures or unions.
13456 @item -mabort-on-noreturn
13457 @opindex mabort-on-noreturn
13458 Generate a call to the function @code{abort} at the end of a
13459 @code{noreturn} function. It is executed if the function tries to
13463 @itemx -mno-long-calls
13464 @opindex mlong-calls
13465 @opindex mno-long-calls
13466 Tells the compiler to perform function calls by first loading the
13467 address of the function into a register and then performing a subroutine
13468 call on this register. This switch is needed if the target function
13469 lies outside of the 64-megabyte addressing range of the offset-based
13470 version of subroutine call instruction.
13472 Even if this switch is enabled, not all function calls are turned
13473 into long calls. The heuristic is that static functions, functions
13474 that have the @code{short_call} attribute, functions that are inside
13475 the scope of a @code{#pragma no_long_calls} directive, and functions whose
13476 definitions have already been compiled within the current compilation
13477 unit are not turned into long calls. The exceptions to this rule are
13478 that weak function definitions, functions with the @code{long_call}
13479 attribute or the @code{section} attribute, and functions that are within
13480 the scope of a @code{#pragma long_calls} directive are always
13481 turned into long calls.
13483 This feature is not enabled by default. Specifying
13484 @option{-mno-long-calls} restores the default behavior, as does
13485 placing the function calls within the scope of a @code{#pragma
13486 long_calls_off} directive. Note these switches have no effect on how
13487 the compiler generates code to handle function calls via function
13490 @item -msingle-pic-base
13491 @opindex msingle-pic-base
13492 Treat the register used for PIC addressing as read-only, rather than
13493 loading it in the prologue for each function. The runtime system is
13494 responsible for initializing this register with an appropriate value
13495 before execution begins.
13497 @item -mpic-register=@var{reg}
13498 @opindex mpic-register
13499 Specify the register to be used for PIC addressing.
13500 For standard PIC base case, the default is any suitable register
13501 determined by compiler. For single PIC base case, the default is
13502 @samp{R9} if target is EABI based or stack-checking is enabled,
13503 otherwise the default is @samp{R10}.
13505 @item -mpic-data-is-text-relative
13506 @opindex mpic-data-is-text-relative
13507 Assume that each data segments are relative to text segment at load time.
13508 Therefore, it permits addressing data using PC-relative operations.
13509 This option is on by default for targets other than VxWorks RTP.
13511 @item -mpoke-function-name
13512 @opindex mpoke-function-name
13513 Write the name of each function into the text section, directly
13514 preceding the function prologue. The generated code is similar to this:
13518 .ascii "arm_poke_function_name", 0
13521 .word 0xff000000 + (t1 - t0)
13522 arm_poke_function_name
13524 stmfd sp!, @{fp, ip, lr, pc@}
13528 When performing a stack backtrace, code can inspect the value of
13529 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
13530 location @code{pc - 12} and the top 8 bits are set, then we know that
13531 there is a function name embedded immediately preceding this location
13532 and has length @code{((pc[-3]) & 0xff000000)}.
13539 Select between generating code that executes in ARM and Thumb
13540 states. The default for most configurations is to generate code
13541 that executes in ARM state, but the default can be changed by
13542 configuring GCC with the @option{--with-mode=}@var{state}
13545 You can also override the ARM and Thumb mode for each function
13546 by using the @code{target("thumb")} and @code{target("arm")} function attributes
13547 (@pxref{ARM Function Attributes}) or pragmas (@pxref{Function Specific Option Pragmas}).
13550 @opindex mtpcs-frame
13551 Generate a stack frame that is compliant with the Thumb Procedure Call
13552 Standard for all non-leaf functions. (A leaf function is one that does
13553 not call any other functions.) The default is @option{-mno-tpcs-frame}.
13555 @item -mtpcs-leaf-frame
13556 @opindex mtpcs-leaf-frame
13557 Generate a stack frame that is compliant with the Thumb Procedure Call
13558 Standard for all leaf functions. (A leaf function is one that does
13559 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
13561 @item -mcallee-super-interworking
13562 @opindex mcallee-super-interworking
13563 Gives all externally visible functions in the file being compiled an ARM
13564 instruction set header which switches to Thumb mode before executing the
13565 rest of the function. This allows these functions to be called from
13566 non-interworking code. This option is not valid in AAPCS configurations
13567 because interworking is enabled by default.
13569 @item -mcaller-super-interworking
13570 @opindex mcaller-super-interworking
13571 Allows calls via function pointers (including virtual functions) to
13572 execute correctly regardless of whether the target code has been
13573 compiled for interworking or not. There is a small overhead in the cost
13574 of executing a function pointer if this option is enabled. This option
13575 is not valid in AAPCS configurations because interworking is enabled
13578 @item -mtp=@var{name}
13580 Specify the access model for the thread local storage pointer. The valid
13581 models are @samp{soft}, which generates calls to @code{__aeabi_read_tp},
13582 @samp{cp15}, which fetches the thread pointer from @code{cp15} directly
13583 (supported in the arm6k architecture), and @samp{auto}, which uses the
13584 best available method for the selected processor. The default setting is
13587 @item -mtls-dialect=@var{dialect}
13588 @opindex mtls-dialect
13589 Specify the dialect to use for accessing thread local storage. Two
13590 @var{dialect}s are supported---@samp{gnu} and @samp{gnu2}. The
13591 @samp{gnu} dialect selects the original GNU scheme for supporting
13592 local and global dynamic TLS models. The @samp{gnu2} dialect
13593 selects the GNU descriptor scheme, which provides better performance
13594 for shared libraries. The GNU descriptor scheme is compatible with
13595 the original scheme, but does require new assembler, linker and
13596 library support. Initial and local exec TLS models are unaffected by
13597 this option and always use the original scheme.
13599 @item -mword-relocations
13600 @opindex mword-relocations
13601 Only generate absolute relocations on word-sized values (i.e. R_ARM_ABS32).
13602 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
13603 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
13606 @item -mfix-cortex-m3-ldrd
13607 @opindex mfix-cortex-m3-ldrd
13608 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
13609 with overlapping destination and base registers are used. This option avoids
13610 generating these instructions. This option is enabled by default when
13611 @option{-mcpu=cortex-m3} is specified.
13613 @item -munaligned-access
13614 @itemx -mno-unaligned-access
13615 @opindex munaligned-access
13616 @opindex mno-unaligned-access
13617 Enables (or disables) reading and writing of 16- and 32- bit values
13618 from addresses that are not 16- or 32- bit aligned. By default
13619 unaligned access is disabled for all pre-ARMv6 and all ARMv6-M
13620 architectures, and enabled for all other architectures. If unaligned
13621 access is not enabled then words in packed data structures are
13622 accessed a byte at a time.
13624 The ARM attribute @code{Tag_CPU_unaligned_access} is set in the
13625 generated object file to either true or false, depending upon the
13626 setting of this option. If unaligned access is enabled then the
13627 preprocessor symbol @code{__ARM_FEATURE_UNALIGNED} is also
13630 @item -mneon-for-64bits
13631 @opindex mneon-for-64bits
13632 Enables using Neon to handle scalar 64-bits operations. This is
13633 disabled by default since the cost of moving data from core registers
13636 @item -mslow-flash-data
13637 @opindex mslow-flash-data
13638 Assume loading data from flash is slower than fetching instruction.
13639 Therefore literal load is minimized for better performance.
13640 This option is only supported when compiling for ARMv7 M-profile and
13643 @item -masm-syntax-unified
13644 @opindex masm-syntax-unified
13645 Assume inline assembler is using unified asm syntax. The default is
13646 currently off which implies divided syntax. Currently this option is
13647 available only for Thumb1 and has no effect on ARM state and Thumb2.
13648 However, this may change in future releases of GCC. Divided syntax
13649 should be considered deprecated.
13651 @item -mrestrict-it
13652 @opindex mrestrict-it
13653 Restricts generation of IT blocks to conform to the rules of ARMv8.
13654 IT blocks can only contain a single 16-bit instruction from a select
13655 set of instructions. This option is on by default for ARMv8 Thumb mode.
13657 @item -mprint-tune-info
13658 @opindex mprint-tune-info
13659 Print CPU tuning information as comment in assembler file. This is
13660 an option used only for regression testing of the compiler and not
13661 intended for ordinary use in compiling code. This option is disabled
13666 @subsection AVR Options
13667 @cindex AVR Options
13669 These options are defined for AVR implementations:
13672 @item -mmcu=@var{mcu}
13674 Specify Atmel AVR instruction set architectures (ISA) or MCU type.
13676 The default for this option is@tie{}@samp{avr2}.
13678 GCC supports the following AVR devices and ISAs:
13680 @include avr-mmcu.texi
13682 @item -maccumulate-args
13683 @opindex maccumulate-args
13684 Accumulate outgoing function arguments and acquire/release the needed
13685 stack space for outgoing function arguments once in function
13686 prologue/epilogue. Without this option, outgoing arguments are pushed
13687 before calling a function and popped afterwards.
13689 Popping the arguments after the function call can be expensive on
13690 AVR so that accumulating the stack space might lead to smaller
13691 executables because arguments need not to be removed from the
13692 stack after such a function call.
13694 This option can lead to reduced code size for functions that perform
13695 several calls to functions that get their arguments on the stack like
13696 calls to printf-like functions.
13698 @item -mbranch-cost=@var{cost}
13699 @opindex mbranch-cost
13700 Set the branch costs for conditional branch instructions to
13701 @var{cost}. Reasonable values for @var{cost} are small, non-negative
13702 integers. The default branch cost is 0.
13704 @item -mcall-prologues
13705 @opindex mcall-prologues
13706 Functions prologues/epilogues are expanded as calls to appropriate
13707 subroutines. Code size is smaller.
13711 Assume @code{int} to be 8-bit integer. This affects the sizes of all types: a
13712 @code{char} is 1 byte, an @code{int} is 1 byte, a @code{long} is 2 bytes,
13713 and @code{long long} is 4 bytes. Please note that this option does not
13714 conform to the C standards, but it results in smaller code
13717 @item -mn-flash=@var{num}
13719 Assume that the flash memory has a size of
13720 @var{num} times 64@tie{}KiB.
13722 @item -mno-interrupts
13723 @opindex mno-interrupts
13724 Generated code is not compatible with hardware interrupts.
13725 Code size is smaller.
13729 Try to replace @code{CALL} resp.@: @code{JMP} instruction by the shorter
13730 @code{RCALL} resp.@: @code{RJMP} instruction if applicable.
13731 Setting @option{-mrelax} just adds the @option{--mlink-relax} option to
13732 the assembler's command line and the @option{--relax} option to the
13733 linker's command line.
13735 Jump relaxing is performed by the linker because jump offsets are not
13736 known before code is located. Therefore, the assembler code generated by the
13737 compiler is the same, but the instructions in the executable may
13738 differ from instructions in the assembler code.
13740 Relaxing must be turned on if linker stubs are needed, see the
13741 section on @code{EIND} and linker stubs below.
13745 Assume that the device supports the Read-Modify-Write
13746 instructions @code{XCH}, @code{LAC}, @code{LAS} and @code{LAT}.
13750 Treat the stack pointer register as an 8-bit register,
13751 i.e.@: assume the high byte of the stack pointer is zero.
13752 In general, you don't need to set this option by hand.
13754 This option is used internally by the compiler to select and
13755 build multilibs for architectures @code{avr2} and @code{avr25}.
13756 These architectures mix devices with and without @code{SPH}.
13757 For any setting other than @option{-mmcu=avr2} or @option{-mmcu=avr25}
13758 the compiler driver adds or removes this option from the compiler
13759 proper's command line, because the compiler then knows if the device
13760 or architecture has an 8-bit stack pointer and thus no @code{SPH}
13765 Use address register @code{X} in a way proposed by the hardware. This means
13766 that @code{X} is only used in indirect, post-increment or
13767 pre-decrement addressing.
13769 Without this option, the @code{X} register may be used in the same way
13770 as @code{Y} or @code{Z} which then is emulated by additional
13772 For example, loading a value with @code{X+const} addressing with a
13773 small non-negative @code{const < 64} to a register @var{Rn} is
13777 adiw r26, const ; X += const
13778 ld @var{Rn}, X ; @var{Rn} = *X
13779 sbiw r26, const ; X -= const
13783 @opindex mtiny-stack
13784 Only change the lower 8@tie{}bits of the stack pointer.
13787 @opindex nodevicelib
13788 Don't link against AVR-LibC's device specific library @code{libdev.a}.
13790 @item -Waddr-space-convert
13791 @opindex Waddr-space-convert
13792 Warn about conversions between address spaces in the case where the
13793 resulting address space is not contained in the incoming address space.
13796 @subsubsection @code{EIND} and Devices with More Than 128 Ki Bytes of Flash
13797 @cindex @code{EIND}
13798 Pointers in the implementation are 16@tie{}bits wide.
13799 The address of a function or label is represented as word address so
13800 that indirect jumps and calls can target any code address in the
13801 range of 64@tie{}Ki words.
13803 In order to facilitate indirect jump on devices with more than 128@tie{}Ki
13804 bytes of program memory space, there is a special function register called
13805 @code{EIND} that serves as most significant part of the target address
13806 when @code{EICALL} or @code{EIJMP} instructions are used.
13808 Indirect jumps and calls on these devices are handled as follows by
13809 the compiler and are subject to some limitations:
13814 The compiler never sets @code{EIND}.
13817 The compiler uses @code{EIND} implicitely in @code{EICALL}/@code{EIJMP}
13818 instructions or might read @code{EIND} directly in order to emulate an
13819 indirect call/jump by means of a @code{RET} instruction.
13822 The compiler assumes that @code{EIND} never changes during the startup
13823 code or during the application. In particular, @code{EIND} is not
13824 saved/restored in function or interrupt service routine
13828 For indirect calls to functions and computed goto, the linker
13829 generates @emph{stubs}. Stubs are jump pads sometimes also called
13830 @emph{trampolines}. Thus, the indirect call/jump jumps to such a stub.
13831 The stub contains a direct jump to the desired address.
13834 Linker relaxation must be turned on so that the linker generates
13835 the stubs correctly in all situations. See the compiler option
13836 @option{-mrelax} and the linker option @option{--relax}.
13837 There are corner cases where the linker is supposed to generate stubs
13838 but aborts without relaxation and without a helpful error message.
13841 The default linker script is arranged for code with @code{EIND = 0}.
13842 If code is supposed to work for a setup with @code{EIND != 0}, a custom
13843 linker script has to be used in order to place the sections whose
13844 name start with @code{.trampolines} into the segment where @code{EIND}
13848 The startup code from libgcc never sets @code{EIND}.
13849 Notice that startup code is a blend of code from libgcc and AVR-LibC.
13850 For the impact of AVR-LibC on @code{EIND}, see the
13851 @w{@uref{http://nongnu.org/avr-libc/user-manual/,AVR-LibC user manual}}.
13854 It is legitimate for user-specific startup code to set up @code{EIND}
13855 early, for example by means of initialization code located in
13856 section @code{.init3}. Such code runs prior to general startup code
13857 that initializes RAM and calls constructors, but after the bit
13858 of startup code from AVR-LibC that sets @code{EIND} to the segment
13859 where the vector table is located.
13861 #include <avr/io.h>
13864 __attribute__((section(".init3"),naked,used,no_instrument_function))
13865 init3_set_eind (void)
13867 __asm volatile ("ldi r24,pm_hh8(__trampolines_start)\n\t"
13868 "out %i0,r24" :: "n" (&EIND) : "r24","memory");
13873 The @code{__trampolines_start} symbol is defined in the linker script.
13876 Stubs are generated automatically by the linker if
13877 the following two conditions are met:
13880 @item The address of a label is taken by means of the @code{gs} modifier
13881 (short for @emph{generate stubs}) like so:
13883 LDI r24, lo8(gs(@var{func}))
13884 LDI r25, hi8(gs(@var{func}))
13886 @item The final location of that label is in a code segment
13887 @emph{outside} the segment where the stubs are located.
13891 The compiler emits such @code{gs} modifiers for code labels in the
13892 following situations:
13894 @item Taking address of a function or code label.
13895 @item Computed goto.
13896 @item If prologue-save function is used, see @option{-mcall-prologues}
13897 command-line option.
13898 @item Switch/case dispatch tables. If you do not want such dispatch
13899 tables you can specify the @option{-fno-jump-tables} command-line option.
13900 @item C and C++ constructors/destructors called during startup/shutdown.
13901 @item If the tools hit a @code{gs()} modifier explained above.
13905 Jumping to non-symbolic addresses like so is @emph{not} supported:
13910 /* Call function at word address 0x2 */
13911 return ((int(*)(void)) 0x2)();
13915 Instead, a stub has to be set up, i.e.@: the function has to be called
13916 through a symbol (@code{func_4} in the example):
13921 extern int func_4 (void);
13923 /* Call function at byte address 0x4 */
13928 and the application be linked with @option{-Wl,--defsym,func_4=0x4}.
13929 Alternatively, @code{func_4} can be defined in the linker script.
13932 @subsubsection Handling of the @code{RAMPD}, @code{RAMPX}, @code{RAMPY} and @code{RAMPZ} Special Function Registers
13933 @cindex @code{RAMPD}
13934 @cindex @code{RAMPX}
13935 @cindex @code{RAMPY}
13936 @cindex @code{RAMPZ}
13937 Some AVR devices support memories larger than the 64@tie{}KiB range
13938 that can be accessed with 16-bit pointers. To access memory locations
13939 outside this 64@tie{}KiB range, the contentent of a @code{RAMP}
13940 register is used as high part of the address:
13941 The @code{X}, @code{Y}, @code{Z} address register is concatenated
13942 with the @code{RAMPX}, @code{RAMPY}, @code{RAMPZ} special function
13943 register, respectively, to get a wide address. Similarly,
13944 @code{RAMPD} is used together with direct addressing.
13948 The startup code initializes the @code{RAMP} special function
13949 registers with zero.
13952 If a @ref{AVR Named Address Spaces,named address space} other than
13953 generic or @code{__flash} is used, then @code{RAMPZ} is set
13954 as needed before the operation.
13957 If the device supports RAM larger than 64@tie{}KiB and the compiler
13958 needs to change @code{RAMPZ} to accomplish an operation, @code{RAMPZ}
13959 is reset to zero after the operation.
13962 If the device comes with a specific @code{RAMP} register, the ISR
13963 prologue/epilogue saves/restores that SFR and initializes it with
13964 zero in case the ISR code might (implicitly) use it.
13967 RAM larger than 64@tie{}KiB is not supported by GCC for AVR targets.
13968 If you use inline assembler to read from locations outside the
13969 16-bit address range and change one of the @code{RAMP} registers,
13970 you must reset it to zero after the access.
13974 @subsubsection AVR Built-in Macros
13976 GCC defines several built-in macros so that the user code can test
13977 for the presence or absence of features. Almost any of the following
13978 built-in macros are deduced from device capabilities and thus
13979 triggered by the @option{-mmcu=} command-line option.
13981 For even more AVR-specific built-in macros see
13982 @ref{AVR Named Address Spaces} and @ref{AVR Built-in Functions}.
13987 Build-in macro that resolves to a decimal number that identifies the
13988 architecture and depends on the @option{-mmcu=@var{mcu}} option.
13989 Possible values are:
13991 @code{2}, @code{25}, @code{3}, @code{31}, @code{35},
13992 @code{4}, @code{5}, @code{51}, @code{6}
13994 for @var{mcu}=@code{avr2}, @code{avr25}, @code{avr3}, @code{avr31},
13995 @code{avr35}, @code{avr4}, @code{avr5}, @code{avr51}, @code{avr6},
13999 @code{100}, @code{102}, @code{104},
14000 @code{105}, @code{106}, @code{107}
14002 for @var{mcu}=@code{avrtiny}, @code{avrxmega2}, @code{avrxmega4},
14003 @code{avrxmega5}, @code{avrxmega6}, @code{avrxmega7}, respectively.
14004 If @var{mcu} specifies a device, this built-in macro is set
14005 accordingly. For example, with @option{-mmcu=atmega8} the macro is
14006 defined to @code{4}.
14008 @item __AVR_@var{Device}__
14009 Setting @option{-mmcu=@var{device}} defines this built-in macro which reflects
14010 the device's name. For example, @option{-mmcu=atmega8} defines the
14011 built-in macro @code{__AVR_ATmega8__}, @option{-mmcu=attiny261a} defines
14012 @code{__AVR_ATtiny261A__}, etc.
14014 The built-in macros' names follow
14015 the scheme @code{__AVR_@var{Device}__} where @var{Device} is
14016 the device name as from the AVR user manual. The difference between
14017 @var{Device} in the built-in macro and @var{device} in
14018 @option{-mmcu=@var{device}} is that the latter is always lowercase.
14020 If @var{device} is not a device but only a core architecture like
14021 @samp{avr51}, this macro is not defined.
14023 @item __AVR_DEVICE_NAME__
14024 Setting @option{-mmcu=@var{device}} defines this built-in macro to
14025 the device's name. For example, with @option{-mmcu=atmega8} the macro
14026 is defined to @code{atmega8}.
14028 If @var{device} is not a device but only a core architecture like
14029 @samp{avr51}, this macro is not defined.
14031 @item __AVR_XMEGA__
14032 The device / architecture belongs to the XMEGA family of devices.
14034 @item __AVR_HAVE_ELPM__
14035 The device has the the @code{ELPM} instruction.
14037 @item __AVR_HAVE_ELPMX__
14038 The device has the @code{ELPM R@var{n},Z} and @code{ELPM
14039 R@var{n},Z+} instructions.
14041 @item __AVR_HAVE_MOVW__
14042 The device has the @code{MOVW} instruction to perform 16-bit
14043 register-register moves.
14045 @item __AVR_HAVE_LPMX__
14046 The device has the @code{LPM R@var{n},Z} and
14047 @code{LPM R@var{n},Z+} instructions.
14049 @item __AVR_HAVE_MUL__
14050 The device has a hardware multiplier.
14052 @item __AVR_HAVE_JMP_CALL__
14053 The device has the @code{JMP} and @code{CALL} instructions.
14054 This is the case for devices with at least 16@tie{}KiB of program
14057 @item __AVR_HAVE_EIJMP_EICALL__
14058 @itemx __AVR_3_BYTE_PC__
14059 The device has the @code{EIJMP} and @code{EICALL} instructions.
14060 This is the case for devices with more than 128@tie{}KiB of program memory.
14061 This also means that the program counter
14062 (PC) is 3@tie{}bytes wide.
14064 @item __AVR_2_BYTE_PC__
14065 The program counter (PC) is 2@tie{}bytes wide. This is the case for devices
14066 with up to 128@tie{}KiB of program memory.
14068 @item __AVR_HAVE_8BIT_SP__
14069 @itemx __AVR_HAVE_16BIT_SP__
14070 The stack pointer (SP) register is treated as 8-bit respectively
14071 16-bit register by the compiler.
14072 The definition of these macros is affected by @option{-mtiny-stack}.
14074 @item __AVR_HAVE_SPH__
14076 The device has the SPH (high part of stack pointer) special function
14077 register or has an 8-bit stack pointer, respectively.
14078 The definition of these macros is affected by @option{-mmcu=} and
14079 in the cases of @option{-mmcu=avr2} and @option{-mmcu=avr25} also
14082 @item __AVR_HAVE_RAMPD__
14083 @itemx __AVR_HAVE_RAMPX__
14084 @itemx __AVR_HAVE_RAMPY__
14085 @itemx __AVR_HAVE_RAMPZ__
14086 The device has the @code{RAMPD}, @code{RAMPX}, @code{RAMPY},
14087 @code{RAMPZ} special function register, respectively.
14089 @item __NO_INTERRUPTS__
14090 This macro reflects the @option{-mno-interrupts} command-line option.
14092 @item __AVR_ERRATA_SKIP__
14093 @itemx __AVR_ERRATA_SKIP_JMP_CALL__
14094 Some AVR devices (AT90S8515, ATmega103) must not skip 32-bit
14095 instructions because of a hardware erratum. Skip instructions are
14096 @code{SBRS}, @code{SBRC}, @code{SBIS}, @code{SBIC} and @code{CPSE}.
14097 The second macro is only defined if @code{__AVR_HAVE_JMP_CALL__} is also
14100 @item __AVR_ISA_RMW__
14101 The device has Read-Modify-Write instructions (XCH, LAC, LAS and LAT).
14103 @item __AVR_SFR_OFFSET__=@var{offset}
14104 Instructions that can address I/O special function registers directly
14105 like @code{IN}, @code{OUT}, @code{SBI}, etc.@: may use a different
14106 address as if addressed by an instruction to access RAM like @code{LD}
14107 or @code{STS}. This offset depends on the device architecture and has
14108 to be subtracted from the RAM address in order to get the
14109 respective I/O@tie{}address.
14111 @item __WITH_AVRLIBC__
14112 The compiler is configured to be used together with AVR-Libc.
14113 See the @option{--with-avrlibc} configure option.
14117 @node Blackfin Options
14118 @subsection Blackfin Options
14119 @cindex Blackfin Options
14122 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
14124 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
14125 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
14126 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
14127 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
14128 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
14129 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
14130 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
14131 @samp{bf561}, @samp{bf592}.
14133 The optional @var{sirevision} specifies the silicon revision of the target
14134 Blackfin processor. Any workarounds available for the targeted silicon revision
14135 are enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
14136 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
14137 are enabled. The @code{__SILICON_REVISION__} macro is defined to two
14138 hexadecimal digits representing the major and minor numbers in the silicon
14139 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
14140 is not defined. If @var{sirevision} is @samp{any}, the
14141 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
14142 If this optional @var{sirevision} is not used, GCC assumes the latest known
14143 silicon revision of the targeted Blackfin processor.
14145 GCC defines a preprocessor macro for the specified @var{cpu}.
14146 For the @samp{bfin-elf} toolchain, this option causes the hardware BSP
14147 provided by libgloss to be linked in if @option{-msim} is not given.
14149 Without this option, @samp{bf532} is used as the processor by default.
14151 Note that support for @samp{bf561} is incomplete. For @samp{bf561},
14152 only the preprocessor macro is defined.
14156 Specifies that the program will be run on the simulator. This causes
14157 the simulator BSP provided by libgloss to be linked in. This option
14158 has effect only for @samp{bfin-elf} toolchain.
14159 Certain other options, such as @option{-mid-shared-library} and
14160 @option{-mfdpic}, imply @option{-msim}.
14162 @item -momit-leaf-frame-pointer
14163 @opindex momit-leaf-frame-pointer
14164 Don't keep the frame pointer in a register for leaf functions. This
14165 avoids the instructions to save, set up and restore frame pointers and
14166 makes an extra register available in leaf functions. The option
14167 @option{-fomit-frame-pointer} removes the frame pointer for all functions,
14168 which might make debugging harder.
14170 @item -mspecld-anomaly
14171 @opindex mspecld-anomaly
14172 When enabled, the compiler ensures that the generated code does not
14173 contain speculative loads after jump instructions. If this option is used,
14174 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
14176 @item -mno-specld-anomaly
14177 @opindex mno-specld-anomaly
14178 Don't generate extra code to prevent speculative loads from occurring.
14180 @item -mcsync-anomaly
14181 @opindex mcsync-anomaly
14182 When enabled, the compiler ensures that the generated code does not
14183 contain CSYNC or SSYNC instructions too soon after conditional branches.
14184 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
14186 @item -mno-csync-anomaly
14187 @opindex mno-csync-anomaly
14188 Don't generate extra code to prevent CSYNC or SSYNC instructions from
14189 occurring too soon after a conditional branch.
14193 When enabled, the compiler is free to take advantage of the knowledge that
14194 the entire program fits into the low 64k of memory.
14197 @opindex mno-low-64k
14198 Assume that the program is arbitrarily large. This is the default.
14200 @item -mstack-check-l1
14201 @opindex mstack-check-l1
14202 Do stack checking using information placed into L1 scratchpad memory by the
14205 @item -mid-shared-library
14206 @opindex mid-shared-library
14207 Generate code that supports shared libraries via the library ID method.
14208 This allows for execute in place and shared libraries in an environment
14209 without virtual memory management. This option implies @option{-fPIC}.
14210 With a @samp{bfin-elf} target, this option implies @option{-msim}.
14212 @item -mno-id-shared-library
14213 @opindex mno-id-shared-library
14214 Generate code that doesn't assume ID-based shared libraries are being used.
14215 This is the default.
14217 @item -mleaf-id-shared-library
14218 @opindex mleaf-id-shared-library
14219 Generate code that supports shared libraries via the library ID method,
14220 but assumes that this library or executable won't link against any other
14221 ID shared libraries. That allows the compiler to use faster code for jumps
14224 @item -mno-leaf-id-shared-library
14225 @opindex mno-leaf-id-shared-library
14226 Do not assume that the code being compiled won't link against any ID shared
14227 libraries. Slower code is generated for jump and call insns.
14229 @item -mshared-library-id=n
14230 @opindex mshared-library-id
14231 Specifies the identification number of the ID-based shared library being
14232 compiled. Specifying a value of 0 generates more compact code; specifying
14233 other values forces the allocation of that number to the current
14234 library but is no more space- or time-efficient than omitting this option.
14238 Generate code that allows the data segment to be located in a different
14239 area of memory from the text segment. This allows for execute in place in
14240 an environment without virtual memory management by eliminating relocations
14241 against the text section.
14243 @item -mno-sep-data
14244 @opindex mno-sep-data
14245 Generate code that assumes that the data segment follows the text segment.
14246 This is the default.
14249 @itemx -mno-long-calls
14250 @opindex mlong-calls
14251 @opindex mno-long-calls
14252 Tells the compiler to perform function calls by first loading the
14253 address of the function into a register and then performing a subroutine
14254 call on this register. This switch is needed if the target function
14255 lies outside of the 24-bit addressing range of the offset-based
14256 version of subroutine call instruction.
14258 This feature is not enabled by default. Specifying
14259 @option{-mno-long-calls} restores the default behavior. Note these
14260 switches have no effect on how the compiler generates code to handle
14261 function calls via function pointers.
14265 Link with the fast floating-point library. This library relaxes some of
14266 the IEEE floating-point standard's rules for checking inputs against
14267 Not-a-Number (NAN), in the interest of performance.
14270 @opindex minline-plt
14271 Enable inlining of PLT entries in function calls to functions that are
14272 not known to bind locally. It has no effect without @option{-mfdpic}.
14275 @opindex mmulticore
14276 Build a standalone application for multicore Blackfin processors.
14277 This option causes proper start files and link scripts supporting
14278 multicore to be used, and defines the macro @code{__BFIN_MULTICORE}.
14279 It can only be used with @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}.
14281 This option can be used with @option{-mcorea} or @option{-mcoreb}, which
14282 selects the one-application-per-core programming model. Without
14283 @option{-mcorea} or @option{-mcoreb}, the single-application/dual-core
14284 programming model is used. In this model, the main function of Core B
14285 should be named as @code{coreb_main}.
14287 If this option is not used, the single-core application programming
14292 Build a standalone application for Core A of BF561 when using
14293 the one-application-per-core programming model. Proper start files
14294 and link scripts are used to support Core A, and the macro
14295 @code{__BFIN_COREA} is defined.
14296 This option can only be used in conjunction with @option{-mmulticore}.
14300 Build a standalone application for Core B of BF561 when using
14301 the one-application-per-core programming model. Proper start files
14302 and link scripts are used to support Core B, and the macro
14303 @code{__BFIN_COREB} is defined. When this option is used, @code{coreb_main}
14304 should be used instead of @code{main}.
14305 This option can only be used in conjunction with @option{-mmulticore}.
14309 Build a standalone application for SDRAM. Proper start files and
14310 link scripts are used to put the application into SDRAM, and the macro
14311 @code{__BFIN_SDRAM} is defined.
14312 The loader should initialize SDRAM before loading the application.
14316 Assume that ICPLBs are enabled at run time. This has an effect on certain
14317 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
14318 are enabled; for standalone applications the default is off.
14322 @subsection C6X Options
14323 @cindex C6X Options
14326 @item -march=@var{name}
14328 This specifies the name of the target architecture. GCC uses this
14329 name to determine what kind of instructions it can emit when generating
14330 assembly code. Permissible names are: @samp{c62x},
14331 @samp{c64x}, @samp{c64x+}, @samp{c67x}, @samp{c67x+}, @samp{c674x}.
14334 @opindex mbig-endian
14335 Generate code for a big-endian target.
14337 @item -mlittle-endian
14338 @opindex mlittle-endian
14339 Generate code for a little-endian target. This is the default.
14343 Choose startup files and linker script suitable for the simulator.
14345 @item -msdata=default
14346 @opindex msdata=default
14347 Put small global and static data in the @code{.neardata} section,
14348 which is pointed to by register @code{B14}. Put small uninitialized
14349 global and static data in the @code{.bss} section, which is adjacent
14350 to the @code{.neardata} section. Put small read-only data into the
14351 @code{.rodata} section. The corresponding sections used for large
14352 pieces of data are @code{.fardata}, @code{.far} and @code{.const}.
14355 @opindex msdata=all
14356 Put all data, not just small objects, into the sections reserved for
14357 small data, and use addressing relative to the @code{B14} register to
14361 @opindex msdata=none
14362 Make no use of the sections reserved for small data, and use absolute
14363 addresses to access all data. Put all initialized global and static
14364 data in the @code{.fardata} section, and all uninitialized data in the
14365 @code{.far} section. Put all constant data into the @code{.const}
14370 @subsection CRIS Options
14371 @cindex CRIS Options
14373 These options are defined specifically for the CRIS ports.
14376 @item -march=@var{architecture-type}
14377 @itemx -mcpu=@var{architecture-type}
14380 Generate code for the specified architecture. The choices for
14381 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
14382 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
14383 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
14386 @item -mtune=@var{architecture-type}
14388 Tune to @var{architecture-type} everything applicable about the generated
14389 code, except for the ABI and the set of available instructions. The
14390 choices for @var{architecture-type} are the same as for
14391 @option{-march=@var{architecture-type}}.
14393 @item -mmax-stack-frame=@var{n}
14394 @opindex mmax-stack-frame
14395 Warn when the stack frame of a function exceeds @var{n} bytes.
14401 The options @option{-metrax4} and @option{-metrax100} are synonyms for
14402 @option{-march=v3} and @option{-march=v8} respectively.
14404 @item -mmul-bug-workaround
14405 @itemx -mno-mul-bug-workaround
14406 @opindex mmul-bug-workaround
14407 @opindex mno-mul-bug-workaround
14408 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
14409 models where it applies. This option is active by default.
14413 Enable CRIS-specific verbose debug-related information in the assembly
14414 code. This option also has the effect of turning off the @samp{#NO_APP}
14415 formatted-code indicator to the assembler at the beginning of the
14420 Do not use condition-code results from previous instruction; always emit
14421 compare and test instructions before use of condition codes.
14423 @item -mno-side-effects
14424 @opindex mno-side-effects
14425 Do not emit instructions with side effects in addressing modes other than
14428 @item -mstack-align
14429 @itemx -mno-stack-align
14430 @itemx -mdata-align
14431 @itemx -mno-data-align
14432 @itemx -mconst-align
14433 @itemx -mno-const-align
14434 @opindex mstack-align
14435 @opindex mno-stack-align
14436 @opindex mdata-align
14437 @opindex mno-data-align
14438 @opindex mconst-align
14439 @opindex mno-const-align
14440 These options (@samp{no-} options) arrange (eliminate arrangements) for the
14441 stack frame, individual data and constants to be aligned for the maximum
14442 single data access size for the chosen CPU model. The default is to
14443 arrange for 32-bit alignment. ABI details such as structure layout are
14444 not affected by these options.
14452 Similar to the stack- data- and const-align options above, these options
14453 arrange for stack frame, writable data and constants to all be 32-bit,
14454 16-bit or 8-bit aligned. The default is 32-bit alignment.
14456 @item -mno-prologue-epilogue
14457 @itemx -mprologue-epilogue
14458 @opindex mno-prologue-epilogue
14459 @opindex mprologue-epilogue
14460 With @option{-mno-prologue-epilogue}, the normal function prologue and
14461 epilogue which set up the stack frame are omitted and no return
14462 instructions or return sequences are generated in the code. Use this
14463 option only together with visual inspection of the compiled code: no
14464 warnings or errors are generated when call-saved registers must be saved,
14465 or storage for local variables needs to be allocated.
14469 @opindex mno-gotplt
14471 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
14472 instruction sequences that load addresses for functions from the PLT part
14473 of the GOT rather than (traditional on other architectures) calls to the
14474 PLT@. The default is @option{-mgotplt}.
14478 Legacy no-op option only recognized with the cris-axis-elf and
14479 cris-axis-linux-gnu targets.
14483 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
14487 This option, recognized for the cris-axis-elf, arranges
14488 to link with input-output functions from a simulator library. Code,
14489 initialized data and zero-initialized data are allocated consecutively.
14493 Like @option{-sim}, but pass linker options to locate initialized data at
14494 0x40000000 and zero-initialized data at 0x80000000.
14498 @subsection CR16 Options
14499 @cindex CR16 Options
14501 These options are defined specifically for the CR16 ports.
14507 Enable the use of multiply-accumulate instructions. Disabled by default.
14511 @opindex mcr16cplus
14513 Generate code for CR16C or CR16C+ architecture. CR16C+ architecture
14518 Links the library libsim.a which is in compatible with simulator. Applicable
14519 to ELF compiler only.
14523 Choose integer type as 32-bit wide.
14527 Generates @code{sbit}/@code{cbit} instructions for bit manipulations.
14529 @item -mdata-model=@var{model}
14530 @opindex mdata-model
14531 Choose a data model. The choices for @var{model} are @samp{near},
14532 @samp{far} or @samp{medium}. @samp{medium} is default.
14533 However, @samp{far} is not valid with @option{-mcr16c}, as the
14534 CR16C architecture does not support the far data model.
14537 @node Darwin Options
14538 @subsection Darwin Options
14539 @cindex Darwin options
14541 These options are defined for all architectures running the Darwin operating
14544 FSF GCC on Darwin does not create ``fat'' object files; it creates
14545 an object file for the single architecture that GCC was built to
14546 target. Apple's GCC on Darwin does create ``fat'' files if multiple
14547 @option{-arch} options are used; it does so by running the compiler or
14548 linker multiple times and joining the results together with
14551 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
14552 @samp{i686}) is determined by the flags that specify the ISA
14553 that GCC is targeting, like @option{-mcpu} or @option{-march}. The
14554 @option{-force_cpusubtype_ALL} option can be used to override this.
14556 The Darwin tools vary in their behavior when presented with an ISA
14557 mismatch. The assembler, @file{as}, only permits instructions to
14558 be used that are valid for the subtype of the file it is generating,
14559 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
14560 The linker for shared libraries, @file{/usr/bin/libtool}, fails
14561 and prints an error if asked to create a shared library with a less
14562 restrictive subtype than its input files (for instance, trying to put
14563 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
14564 for executables, @command{ld}, quietly gives the executable the most
14565 restrictive subtype of any of its input files.
14570 Add the framework directory @var{dir} to the head of the list of
14571 directories to be searched for header files. These directories are
14572 interleaved with those specified by @option{-I} options and are
14573 scanned in a left-to-right order.
14575 A framework directory is a directory with frameworks in it. A
14576 framework is a directory with a @file{Headers} and/or
14577 @file{PrivateHeaders} directory contained directly in it that ends
14578 in @file{.framework}. The name of a framework is the name of this
14579 directory excluding the @file{.framework}. Headers associated with
14580 the framework are found in one of those two directories, with
14581 @file{Headers} being searched first. A subframework is a framework
14582 directory that is in a framework's @file{Frameworks} directory.
14583 Includes of subframework headers can only appear in a header of a
14584 framework that contains the subframework, or in a sibling subframework
14585 header. Two subframeworks are siblings if they occur in the same
14586 framework. A subframework should not have the same name as a
14587 framework; a warning is issued if this is violated. Currently a
14588 subframework cannot have subframeworks; in the future, the mechanism
14589 may be extended to support this. The standard frameworks can be found
14590 in @file{/System/Library/Frameworks} and
14591 @file{/Library/Frameworks}. An example include looks like
14592 @code{#include <Framework/header.h>}, where @file{Framework} denotes
14593 the name of the framework and @file{header.h} is found in the
14594 @file{PrivateHeaders} or @file{Headers} directory.
14596 @item -iframework@var{dir}
14597 @opindex iframework
14598 Like @option{-F} except the directory is a treated as a system
14599 directory. The main difference between this @option{-iframework} and
14600 @option{-F} is that with @option{-iframework} the compiler does not
14601 warn about constructs contained within header files found via
14602 @var{dir}. This option is valid only for the C family of languages.
14606 Emit debugging information for symbols that are used. For stabs
14607 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
14608 This is by default ON@.
14612 Emit debugging information for all symbols and types.
14614 @item -mmacosx-version-min=@var{version}
14615 The earliest version of MacOS X that this executable will run on
14616 is @var{version}. Typical values of @var{version} include @code{10.1},
14617 @code{10.2}, and @code{10.3.9}.
14619 If the compiler was built to use the system's headers by default,
14620 then the default for this option is the system version on which the
14621 compiler is running, otherwise the default is to make choices that
14622 are compatible with as many systems and code bases as possible.
14626 Enable kernel development mode. The @option{-mkernel} option sets
14627 @option{-static}, @option{-fno-common}, @option{-fno-use-cxa-atexit},
14628 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
14629 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
14630 applicable. This mode also sets @option{-mno-altivec},
14631 @option{-msoft-float}, @option{-fno-builtin} and
14632 @option{-mlong-branch} for PowerPC targets.
14634 @item -mone-byte-bool
14635 @opindex mone-byte-bool
14636 Override the defaults for @code{bool} so that @code{sizeof(bool)==1}.
14637 By default @code{sizeof(bool)} is @code{4} when compiling for
14638 Darwin/PowerPC and @code{1} when compiling for Darwin/x86, so this
14639 option has no effect on x86.
14641 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
14642 to generate code that is not binary compatible with code generated
14643 without that switch. Using this switch may require recompiling all
14644 other modules in a program, including system libraries. Use this
14645 switch to conform to a non-default data model.
14647 @item -mfix-and-continue
14648 @itemx -ffix-and-continue
14649 @itemx -findirect-data
14650 @opindex mfix-and-continue
14651 @opindex ffix-and-continue
14652 @opindex findirect-data
14653 Generate code suitable for fast turnaround development, such as to
14654 allow GDB to dynamically load @file{.o} files into already-running
14655 programs. @option{-findirect-data} and @option{-ffix-and-continue}
14656 are provided for backwards compatibility.
14660 Loads all members of static archive libraries.
14661 See man ld(1) for more information.
14663 @item -arch_errors_fatal
14664 @opindex arch_errors_fatal
14665 Cause the errors having to do with files that have the wrong architecture
14668 @item -bind_at_load
14669 @opindex bind_at_load
14670 Causes the output file to be marked such that the dynamic linker will
14671 bind all undefined references when the file is loaded or launched.
14675 Produce a Mach-o bundle format file.
14676 See man ld(1) for more information.
14678 @item -bundle_loader @var{executable}
14679 @opindex bundle_loader
14680 This option specifies the @var{executable} that will load the build
14681 output file being linked. See man ld(1) for more information.
14684 @opindex dynamiclib
14685 When passed this option, GCC produces a dynamic library instead of
14686 an executable when linking, using the Darwin @file{libtool} command.
14688 @item -force_cpusubtype_ALL
14689 @opindex force_cpusubtype_ALL
14690 This causes GCC's output file to have the @samp{ALL} subtype, instead of
14691 one controlled by the @option{-mcpu} or @option{-march} option.
14693 @item -allowable_client @var{client_name}
14694 @itemx -client_name
14695 @itemx -compatibility_version
14696 @itemx -current_version
14698 @itemx -dependency-file
14700 @itemx -dylinker_install_name
14702 @itemx -exported_symbols_list
14705 @itemx -flat_namespace
14706 @itemx -force_flat_namespace
14707 @itemx -headerpad_max_install_names
14710 @itemx -install_name
14711 @itemx -keep_private_externs
14712 @itemx -multi_module
14713 @itemx -multiply_defined
14714 @itemx -multiply_defined_unused
14717 @itemx -no_dead_strip_inits_and_terms
14718 @itemx -nofixprebinding
14719 @itemx -nomultidefs
14721 @itemx -noseglinkedit
14722 @itemx -pagezero_size
14724 @itemx -prebind_all_twolevel_modules
14725 @itemx -private_bundle
14727 @itemx -read_only_relocs
14729 @itemx -sectobjectsymbols
14733 @itemx -sectobjectsymbols
14736 @itemx -segs_read_only_addr
14738 @itemx -segs_read_write_addr
14739 @itemx -seg_addr_table
14740 @itemx -seg_addr_table_filename
14741 @itemx -seglinkedit
14743 @itemx -segs_read_only_addr
14744 @itemx -segs_read_write_addr
14745 @itemx -single_module
14747 @itemx -sub_library
14749 @itemx -sub_umbrella
14750 @itemx -twolevel_namespace
14753 @itemx -unexported_symbols_list
14754 @itemx -weak_reference_mismatches
14755 @itemx -whatsloaded
14756 @opindex allowable_client
14757 @opindex client_name
14758 @opindex compatibility_version
14759 @opindex current_version
14760 @opindex dead_strip
14761 @opindex dependency-file
14762 @opindex dylib_file
14763 @opindex dylinker_install_name
14765 @opindex exported_symbols_list
14767 @opindex flat_namespace
14768 @opindex force_flat_namespace
14769 @opindex headerpad_max_install_names
14770 @opindex image_base
14772 @opindex install_name
14773 @opindex keep_private_externs
14774 @opindex multi_module
14775 @opindex multiply_defined
14776 @opindex multiply_defined_unused
14777 @opindex noall_load
14778 @opindex no_dead_strip_inits_and_terms
14779 @opindex nofixprebinding
14780 @opindex nomultidefs
14782 @opindex noseglinkedit
14783 @opindex pagezero_size
14785 @opindex prebind_all_twolevel_modules
14786 @opindex private_bundle
14787 @opindex read_only_relocs
14789 @opindex sectobjectsymbols
14792 @opindex sectcreate
14793 @opindex sectobjectsymbols
14796 @opindex segs_read_only_addr
14797 @opindex segs_read_write_addr
14798 @opindex seg_addr_table
14799 @opindex seg_addr_table_filename
14800 @opindex seglinkedit
14802 @opindex segs_read_only_addr
14803 @opindex segs_read_write_addr
14804 @opindex single_module
14806 @opindex sub_library
14807 @opindex sub_umbrella
14808 @opindex twolevel_namespace
14811 @opindex unexported_symbols_list
14812 @opindex weak_reference_mismatches
14813 @opindex whatsloaded
14814 These options are passed to the Darwin linker. The Darwin linker man page
14815 describes them in detail.
14818 @node DEC Alpha Options
14819 @subsection DEC Alpha Options
14821 These @samp{-m} options are defined for the DEC Alpha implementations:
14824 @item -mno-soft-float
14825 @itemx -msoft-float
14826 @opindex mno-soft-float
14827 @opindex msoft-float
14828 Use (do not use) the hardware floating-point instructions for
14829 floating-point operations. When @option{-msoft-float} is specified,
14830 functions in @file{libgcc.a} are used to perform floating-point
14831 operations. Unless they are replaced by routines that emulate the
14832 floating-point operations, or compiled in such a way as to call such
14833 emulations routines, these routines issue floating-point
14834 operations. If you are compiling for an Alpha without floating-point
14835 operations, you must ensure that the library is built so as not to call
14838 Note that Alpha implementations without floating-point operations are
14839 required to have floating-point registers.
14842 @itemx -mno-fp-regs
14844 @opindex mno-fp-regs
14845 Generate code that uses (does not use) the floating-point register set.
14846 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
14847 register set is not used, floating-point operands are passed in integer
14848 registers as if they were integers and floating-point results are passed
14849 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
14850 so any function with a floating-point argument or return value called by code
14851 compiled with @option{-mno-fp-regs} must also be compiled with that
14854 A typical use of this option is building a kernel that does not use,
14855 and hence need not save and restore, any floating-point registers.
14859 The Alpha architecture implements floating-point hardware optimized for
14860 maximum performance. It is mostly compliant with the IEEE floating-point
14861 standard. However, for full compliance, software assistance is
14862 required. This option generates code fully IEEE-compliant code
14863 @emph{except} that the @var{inexact-flag} is not maintained (see below).
14864 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
14865 defined during compilation. The resulting code is less efficient but is
14866 able to correctly support denormalized numbers and exceptional IEEE
14867 values such as not-a-number and plus/minus infinity. Other Alpha
14868 compilers call this option @option{-ieee_with_no_inexact}.
14870 @item -mieee-with-inexact
14871 @opindex mieee-with-inexact
14872 This is like @option{-mieee} except the generated code also maintains
14873 the IEEE @var{inexact-flag}. Turning on this option causes the
14874 generated code to implement fully-compliant IEEE math. In addition to
14875 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
14876 macro. On some Alpha implementations the resulting code may execute
14877 significantly slower than the code generated by default. Since there is
14878 very little code that depends on the @var{inexact-flag}, you should
14879 normally not specify this option. Other Alpha compilers call this
14880 option @option{-ieee_with_inexact}.
14882 @item -mfp-trap-mode=@var{trap-mode}
14883 @opindex mfp-trap-mode
14884 This option controls what floating-point related traps are enabled.
14885 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
14886 The trap mode can be set to one of four values:
14890 This is the default (normal) setting. The only traps that are enabled
14891 are the ones that cannot be disabled in software (e.g., division by zero
14895 In addition to the traps enabled by @samp{n}, underflow traps are enabled
14899 Like @samp{u}, but the instructions are marked to be safe for software
14900 completion (see Alpha architecture manual for details).
14903 Like @samp{su}, but inexact traps are enabled as well.
14906 @item -mfp-rounding-mode=@var{rounding-mode}
14907 @opindex mfp-rounding-mode
14908 Selects the IEEE rounding mode. Other Alpha compilers call this option
14909 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
14914 Normal IEEE rounding mode. Floating-point numbers are rounded towards
14915 the nearest machine number or towards the even machine number in case
14919 Round towards minus infinity.
14922 Chopped rounding mode. Floating-point numbers are rounded towards zero.
14925 Dynamic rounding mode. A field in the floating-point control register
14926 (@var{fpcr}, see Alpha architecture reference manual) controls the
14927 rounding mode in effect. The C library initializes this register for
14928 rounding towards plus infinity. Thus, unless your program modifies the
14929 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
14932 @item -mtrap-precision=@var{trap-precision}
14933 @opindex mtrap-precision
14934 In the Alpha architecture, floating-point traps are imprecise. This
14935 means without software assistance it is impossible to recover from a
14936 floating trap and program execution normally needs to be terminated.
14937 GCC can generate code that can assist operating system trap handlers
14938 in determining the exact location that caused a floating-point trap.
14939 Depending on the requirements of an application, different levels of
14940 precisions can be selected:
14944 Program precision. This option is the default and means a trap handler
14945 can only identify which program caused a floating-point exception.
14948 Function precision. The trap handler can determine the function that
14949 caused a floating-point exception.
14952 Instruction precision. The trap handler can determine the exact
14953 instruction that caused a floating-point exception.
14956 Other Alpha compilers provide the equivalent options called
14957 @option{-scope_safe} and @option{-resumption_safe}.
14959 @item -mieee-conformant
14960 @opindex mieee-conformant
14961 This option marks the generated code as IEEE conformant. You must not
14962 use this option unless you also specify @option{-mtrap-precision=i} and either
14963 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
14964 is to emit the line @samp{.eflag 48} in the function prologue of the
14965 generated assembly file.
14967 @item -mbuild-constants
14968 @opindex mbuild-constants
14969 Normally GCC examines a 32- or 64-bit integer constant to
14970 see if it can construct it from smaller constants in two or three
14971 instructions. If it cannot, it outputs the constant as a literal and
14972 generates code to load it from the data segment at run time.
14974 Use this option to require GCC to construct @emph{all} integer constants
14975 using code, even if it takes more instructions (the maximum is six).
14977 You typically use this option to build a shared library dynamic
14978 loader. Itself a shared library, it must relocate itself in memory
14979 before it can find the variables and constants in its own data segment.
14997 Indicate whether GCC should generate code to use the optional BWX,
14998 CIX, FIX and MAX instruction sets. The default is to use the instruction
14999 sets supported by the CPU type specified via @option{-mcpu=} option or that
15000 of the CPU on which GCC was built if none is specified.
15003 @itemx -mfloat-ieee
15004 @opindex mfloat-vax
15005 @opindex mfloat-ieee
15006 Generate code that uses (does not use) VAX F and G floating-point
15007 arithmetic instead of IEEE single and double precision.
15009 @item -mexplicit-relocs
15010 @itemx -mno-explicit-relocs
15011 @opindex mexplicit-relocs
15012 @opindex mno-explicit-relocs
15013 Older Alpha assemblers provided no way to generate symbol relocations
15014 except via assembler macros. Use of these macros does not allow
15015 optimal instruction scheduling. GNU binutils as of version 2.12
15016 supports a new syntax that allows the compiler to explicitly mark
15017 which relocations should apply to which instructions. This option
15018 is mostly useful for debugging, as GCC detects the capabilities of
15019 the assembler when it is built and sets the default accordingly.
15022 @itemx -mlarge-data
15023 @opindex msmall-data
15024 @opindex mlarge-data
15025 When @option{-mexplicit-relocs} is in effect, static data is
15026 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
15027 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
15028 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
15029 16-bit relocations off of the @code{$gp} register. This limits the
15030 size of the small data area to 64KB, but allows the variables to be
15031 directly accessed via a single instruction.
15033 The default is @option{-mlarge-data}. With this option the data area
15034 is limited to just below 2GB@. Programs that require more than 2GB of
15035 data must use @code{malloc} or @code{mmap} to allocate the data in the
15036 heap instead of in the program's data segment.
15038 When generating code for shared libraries, @option{-fpic} implies
15039 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
15042 @itemx -mlarge-text
15043 @opindex msmall-text
15044 @opindex mlarge-text
15045 When @option{-msmall-text} is used, the compiler assumes that the
15046 code of the entire program (or shared library) fits in 4MB, and is
15047 thus reachable with a branch instruction. When @option{-msmall-data}
15048 is used, the compiler can assume that all local symbols share the
15049 same @code{$gp} value, and thus reduce the number of instructions
15050 required for a function call from 4 to 1.
15052 The default is @option{-mlarge-text}.
15054 @item -mcpu=@var{cpu_type}
15056 Set the instruction set and instruction scheduling parameters for
15057 machine type @var{cpu_type}. You can specify either the @samp{EV}
15058 style name or the corresponding chip number. GCC supports scheduling
15059 parameters for the EV4, EV5 and EV6 family of processors and
15060 chooses the default values for the instruction set from the processor
15061 you specify. If you do not specify a processor type, GCC defaults
15062 to the processor on which the compiler was built.
15064 Supported values for @var{cpu_type} are
15070 Schedules as an EV4 and has no instruction set extensions.
15074 Schedules as an EV5 and has no instruction set extensions.
15078 Schedules as an EV5 and supports the BWX extension.
15083 Schedules as an EV5 and supports the BWX and MAX extensions.
15087 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
15091 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
15094 Native toolchains also support the value @samp{native},
15095 which selects the best architecture option for the host processor.
15096 @option{-mcpu=native} has no effect if GCC does not recognize
15099 @item -mtune=@var{cpu_type}
15101 Set only the instruction scheduling parameters for machine type
15102 @var{cpu_type}. The instruction set is not changed.
15104 Native toolchains also support the value @samp{native},
15105 which selects the best architecture option for the host processor.
15106 @option{-mtune=native} has no effect if GCC does not recognize
15109 @item -mmemory-latency=@var{time}
15110 @opindex mmemory-latency
15111 Sets the latency the scheduler should assume for typical memory
15112 references as seen by the application. This number is highly
15113 dependent on the memory access patterns used by the application
15114 and the size of the external cache on the machine.
15116 Valid options for @var{time} are
15120 A decimal number representing clock cycles.
15126 The compiler contains estimates of the number of clock cycles for
15127 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
15128 (also called Dcache, Scache, and Bcache), as well as to main memory.
15129 Note that L3 is only valid for EV5.
15135 @subsection FR30 Options
15136 @cindex FR30 Options
15138 These options are defined specifically for the FR30 port.
15142 @item -msmall-model
15143 @opindex msmall-model
15144 Use the small address space model. This can produce smaller code, but
15145 it does assume that all symbolic values and addresses fit into a
15150 Assume that runtime support has been provided and so there is no need
15151 to include the simulator library (@file{libsim.a}) on the linker
15157 @subsection FT32 Options
15158 @cindex FT32 Options
15160 These options are defined specifically for the FT32 port.
15166 Specifies that the program will be run on the simulator. This causes
15167 an alternate runtime startup and library to be linked.
15168 You must not use this option when generating programs that will run on
15169 real hardware; you must provide your own runtime library for whatever
15170 I/O functions are needed.
15174 Enable Local Register Allocation. This is still experimental for FT32,
15175 so by default the compiler uses standard reload.
15180 @subsection FRV Options
15181 @cindex FRV Options
15187 Only use the first 32 general-purpose registers.
15192 Use all 64 general-purpose registers.
15197 Use only the first 32 floating-point registers.
15202 Use all 64 floating-point registers.
15205 @opindex mhard-float
15207 Use hardware instructions for floating-point operations.
15210 @opindex msoft-float
15212 Use library routines for floating-point operations.
15217 Dynamically allocate condition code registers.
15222 Do not try to dynamically allocate condition code registers, only
15223 use @code{icc0} and @code{fcc0}.
15228 Change ABI to use double word insns.
15233 Do not use double word instructions.
15238 Use floating-point double instructions.
15241 @opindex mno-double
15243 Do not use floating-point double instructions.
15248 Use media instructions.
15253 Do not use media instructions.
15258 Use multiply and add/subtract instructions.
15261 @opindex mno-muladd
15263 Do not use multiply and add/subtract instructions.
15268 Select the FDPIC ABI, which uses function descriptors to represent
15269 pointers to functions. Without any PIC/PIE-related options, it
15270 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
15271 assumes GOT entries and small data are within a 12-bit range from the
15272 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
15273 are computed with 32 bits.
15274 With a @samp{bfin-elf} target, this option implies @option{-msim}.
15277 @opindex minline-plt
15279 Enable inlining of PLT entries in function calls to functions that are
15280 not known to bind locally. It has no effect without @option{-mfdpic}.
15281 It's enabled by default if optimizing for speed and compiling for
15282 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
15283 optimization option such as @option{-O3} or above is present in the
15289 Assume a large TLS segment when generating thread-local code.
15294 Do not assume a large TLS segment when generating thread-local code.
15299 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
15300 that is known to be in read-only sections. It's enabled by default,
15301 except for @option{-fpic} or @option{-fpie}: even though it may help
15302 make the global offset table smaller, it trades 1 instruction for 4.
15303 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
15304 one of which may be shared by multiple symbols, and it avoids the need
15305 for a GOT entry for the referenced symbol, so it's more likely to be a
15306 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
15308 @item -multilib-library-pic
15309 @opindex multilib-library-pic
15311 Link with the (library, not FD) pic libraries. It's implied by
15312 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
15313 @option{-fpic} without @option{-mfdpic}. You should never have to use
15317 @opindex mlinked-fp
15319 Follow the EABI requirement of always creating a frame pointer whenever
15320 a stack frame is allocated. This option is enabled by default and can
15321 be disabled with @option{-mno-linked-fp}.
15324 @opindex mlong-calls
15326 Use indirect addressing to call functions outside the current
15327 compilation unit. This allows the functions to be placed anywhere
15328 within the 32-bit address space.
15330 @item -malign-labels
15331 @opindex malign-labels
15333 Try to align labels to an 8-byte boundary by inserting NOPs into the
15334 previous packet. This option only has an effect when VLIW packing
15335 is enabled. It doesn't create new packets; it merely adds NOPs to
15338 @item -mlibrary-pic
15339 @opindex mlibrary-pic
15341 Generate position-independent EABI code.
15346 Use only the first four media accumulator registers.
15351 Use all eight media accumulator registers.
15356 Pack VLIW instructions.
15361 Do not pack VLIW instructions.
15364 @opindex mno-eflags
15366 Do not mark ABI switches in e_flags.
15369 @opindex mcond-move
15371 Enable the use of conditional-move instructions (default).
15373 This switch is mainly for debugging the compiler and will likely be removed
15374 in a future version.
15376 @item -mno-cond-move
15377 @opindex mno-cond-move
15379 Disable the use of conditional-move instructions.
15381 This switch is mainly for debugging the compiler and will likely be removed
15382 in a future version.
15387 Enable the use of conditional set instructions (default).
15389 This switch is mainly for debugging the compiler and will likely be removed
15390 in a future version.
15395 Disable the use of conditional set instructions.
15397 This switch is mainly for debugging the compiler and will likely be removed
15398 in a future version.
15401 @opindex mcond-exec
15403 Enable the use of conditional execution (default).
15405 This switch is mainly for debugging the compiler and will likely be removed
15406 in a future version.
15408 @item -mno-cond-exec
15409 @opindex mno-cond-exec
15411 Disable the use of conditional execution.
15413 This switch is mainly for debugging the compiler and will likely be removed
15414 in a future version.
15416 @item -mvliw-branch
15417 @opindex mvliw-branch
15419 Run a pass to pack branches into VLIW instructions (default).
15421 This switch is mainly for debugging the compiler and will likely be removed
15422 in a future version.
15424 @item -mno-vliw-branch
15425 @opindex mno-vliw-branch
15427 Do not run a pass to pack branches into VLIW instructions.
15429 This switch is mainly for debugging the compiler and will likely be removed
15430 in a future version.
15432 @item -mmulti-cond-exec
15433 @opindex mmulti-cond-exec
15435 Enable optimization of @code{&&} and @code{||} in conditional execution
15438 This switch is mainly for debugging the compiler and will likely be removed
15439 in a future version.
15441 @item -mno-multi-cond-exec
15442 @opindex mno-multi-cond-exec
15444 Disable optimization of @code{&&} and @code{||} in conditional execution.
15446 This switch is mainly for debugging the compiler and will likely be removed
15447 in a future version.
15449 @item -mnested-cond-exec
15450 @opindex mnested-cond-exec
15452 Enable nested conditional execution optimizations (default).
15454 This switch is mainly for debugging the compiler and will likely be removed
15455 in a future version.
15457 @item -mno-nested-cond-exec
15458 @opindex mno-nested-cond-exec
15460 Disable nested conditional execution optimizations.
15462 This switch is mainly for debugging the compiler and will likely be removed
15463 in a future version.
15465 @item -moptimize-membar
15466 @opindex moptimize-membar
15468 This switch removes redundant @code{membar} instructions from the
15469 compiler-generated code. It is enabled by default.
15471 @item -mno-optimize-membar
15472 @opindex mno-optimize-membar
15474 This switch disables the automatic removal of redundant @code{membar}
15475 instructions from the generated code.
15477 @item -mtomcat-stats
15478 @opindex mtomcat-stats
15480 Cause gas to print out tomcat statistics.
15482 @item -mcpu=@var{cpu}
15485 Select the processor type for which to generate code. Possible values are
15486 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
15487 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
15491 @node GNU/Linux Options
15492 @subsection GNU/Linux Options
15494 These @samp{-m} options are defined for GNU/Linux targets:
15499 Use the GNU C library. This is the default except
15500 on @samp{*-*-linux-*uclibc*}, @samp{*-*-linux-*musl*} and
15501 @samp{*-*-linux-*android*} targets.
15505 Use uClibc C library. This is the default on
15506 @samp{*-*-linux-*uclibc*} targets.
15510 Use the musl C library. This is the default on
15511 @samp{*-*-linux-*musl*} targets.
15515 Use Bionic C library. This is the default on
15516 @samp{*-*-linux-*android*} targets.
15520 Compile code compatible with Android platform. This is the default on
15521 @samp{*-*-linux-*android*} targets.
15523 When compiling, this option enables @option{-mbionic}, @option{-fPIC},
15524 @option{-fno-exceptions} and @option{-fno-rtti} by default. When linking,
15525 this option makes the GCC driver pass Android-specific options to the linker.
15526 Finally, this option causes the preprocessor macro @code{__ANDROID__}
15529 @item -tno-android-cc
15530 @opindex tno-android-cc
15531 Disable compilation effects of @option{-mandroid}, i.e., do not enable
15532 @option{-mbionic}, @option{-fPIC}, @option{-fno-exceptions} and
15533 @option{-fno-rtti} by default.
15535 @item -tno-android-ld
15536 @opindex tno-android-ld
15537 Disable linking effects of @option{-mandroid}, i.e., pass standard Linux
15538 linking options to the linker.
15542 @node H8/300 Options
15543 @subsection H8/300 Options
15545 These @samp{-m} options are defined for the H8/300 implementations:
15550 Shorten some address references at link time, when possible; uses the
15551 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
15552 ld, Using ld}, for a fuller description.
15556 Generate code for the H8/300H@.
15560 Generate code for the H8S@.
15564 Generate code for the H8S and H8/300H in the normal mode. This switch
15565 must be used either with @option{-mh} or @option{-ms}.
15569 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
15573 Extended registers are stored on stack before execution of function
15574 with monitor attribute. Default option is @option{-mexr}.
15575 This option is valid only for H8S targets.
15579 Extended registers are not stored on stack before execution of function
15580 with monitor attribute. Default option is @option{-mno-exr}.
15581 This option is valid only for H8S targets.
15585 Make @code{int} data 32 bits by default.
15588 @opindex malign-300
15589 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
15590 The default for the H8/300H and H8S is to align longs and floats on
15592 @option{-malign-300} causes them to be aligned on 2-byte boundaries.
15593 This option has no effect on the H8/300.
15597 @subsection HPPA Options
15598 @cindex HPPA Options
15600 These @samp{-m} options are defined for the HPPA family of computers:
15603 @item -march=@var{architecture-type}
15605 Generate code for the specified architecture. The choices for
15606 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
15607 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
15608 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
15609 architecture option for your machine. Code compiled for lower numbered
15610 architectures runs on higher numbered architectures, but not the
15613 @item -mpa-risc-1-0
15614 @itemx -mpa-risc-1-1
15615 @itemx -mpa-risc-2-0
15616 @opindex mpa-risc-1-0
15617 @opindex mpa-risc-1-1
15618 @opindex mpa-risc-2-0
15619 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
15621 @item -mjump-in-delay
15622 @opindex mjump-in-delay
15623 This option is ignored and provided for compatibility purposes only.
15625 @item -mdisable-fpregs
15626 @opindex mdisable-fpregs
15627 Prevent floating-point registers from being used in any manner. This is
15628 necessary for compiling kernels that perform lazy context switching of
15629 floating-point registers. If you use this option and attempt to perform
15630 floating-point operations, the compiler aborts.
15632 @item -mdisable-indexing
15633 @opindex mdisable-indexing
15634 Prevent the compiler from using indexing address modes. This avoids some
15635 rather obscure problems when compiling MIG generated code under MACH@.
15637 @item -mno-space-regs
15638 @opindex mno-space-regs
15639 Generate code that assumes the target has no space registers. This allows
15640 GCC to generate faster indirect calls and use unscaled index address modes.
15642 Such code is suitable for level 0 PA systems and kernels.
15644 @item -mfast-indirect-calls
15645 @opindex mfast-indirect-calls
15646 Generate code that assumes calls never cross space boundaries. This
15647 allows GCC to emit code that performs faster indirect calls.
15649 This option does not work in the presence of shared libraries or nested
15652 @item -mfixed-range=@var{register-range}
15653 @opindex mfixed-range
15654 Generate code treating the given register range as fixed registers.
15655 A fixed register is one that the register allocator cannot use. This is
15656 useful when compiling kernel code. A register range is specified as
15657 two registers separated by a dash. Multiple register ranges can be
15658 specified separated by a comma.
15660 @item -mlong-load-store
15661 @opindex mlong-load-store
15662 Generate 3-instruction load and store sequences as sometimes required by
15663 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
15666 @item -mportable-runtime
15667 @opindex mportable-runtime
15668 Use the portable calling conventions proposed by HP for ELF systems.
15672 Enable the use of assembler directives only GAS understands.
15674 @item -mschedule=@var{cpu-type}
15676 Schedule code according to the constraints for the machine type
15677 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
15678 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
15679 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
15680 proper scheduling option for your machine. The default scheduling is
15684 @opindex mlinker-opt
15685 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
15686 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
15687 linkers in which they give bogus error messages when linking some programs.
15690 @opindex msoft-float
15691 Generate output containing library calls for floating point.
15692 @strong{Warning:} the requisite libraries are not available for all HPPA
15693 targets. Normally the facilities of the machine's usual C compiler are
15694 used, but this cannot be done directly in cross-compilation. You must make
15695 your own arrangements to provide suitable library functions for
15698 @option{-msoft-float} changes the calling convention in the output file;
15699 therefore, it is only useful if you compile @emph{all} of a program with
15700 this option. In particular, you need to compile @file{libgcc.a}, the
15701 library that comes with GCC, with @option{-msoft-float} in order for
15706 Generate the predefine, @code{_SIO}, for server IO@. The default is
15707 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
15708 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
15709 options are available under HP-UX and HI-UX@.
15713 Use options specific to GNU @command{ld}.
15714 This passes @option{-shared} to @command{ld} when
15715 building a shared library. It is the default when GCC is configured,
15716 explicitly or implicitly, with the GNU linker. This option does not
15717 affect which @command{ld} is called; it only changes what parameters
15718 are passed to that @command{ld}.
15719 The @command{ld} that is called is determined by the
15720 @option{--with-ld} configure option, GCC's program search path, and
15721 finally by the user's @env{PATH}. The linker used by GCC can be printed
15722 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
15723 on the 64-bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
15727 Use options specific to HP @command{ld}.
15728 This passes @option{-b} to @command{ld} when building
15729 a shared library and passes @option{+Accept TypeMismatch} to @command{ld} on all
15730 links. It is the default when GCC is configured, explicitly or
15731 implicitly, with the HP linker. This option does not affect
15732 which @command{ld} is called; it only changes what parameters are passed to that
15734 The @command{ld} that is called is determined by the @option{--with-ld}
15735 configure option, GCC's program search path, and finally by the user's
15736 @env{PATH}. The linker used by GCC can be printed using @samp{which
15737 `gcc -print-prog-name=ld`}. This option is only available on the 64-bit
15738 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
15741 @opindex mno-long-calls
15742 Generate code that uses long call sequences. This ensures that a call
15743 is always able to reach linker generated stubs. The default is to generate
15744 long calls only when the distance from the call site to the beginning
15745 of the function or translation unit, as the case may be, exceeds a
15746 predefined limit set by the branch type being used. The limits for
15747 normal calls are 7,600,000 and 240,000 bytes, respectively for the
15748 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
15751 Distances are measured from the beginning of functions when using the
15752 @option{-ffunction-sections} option, or when using the @option{-mgas}
15753 and @option{-mno-portable-runtime} options together under HP-UX with
15756 It is normally not desirable to use this option as it degrades
15757 performance. However, it may be useful in large applications,
15758 particularly when partial linking is used to build the application.
15760 The types of long calls used depends on the capabilities of the
15761 assembler and linker, and the type of code being generated. The
15762 impact on systems that support long absolute calls, and long pic
15763 symbol-difference or pc-relative calls should be relatively small.
15764 However, an indirect call is used on 32-bit ELF systems in pic code
15765 and it is quite long.
15767 @item -munix=@var{unix-std}
15769 Generate compiler predefines and select a startfile for the specified
15770 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
15771 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
15772 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
15773 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
15774 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
15777 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
15778 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
15779 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
15780 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
15781 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
15782 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
15784 It is @emph{important} to note that this option changes the interfaces
15785 for various library routines. It also affects the operational behavior
15786 of the C library. Thus, @emph{extreme} care is needed in using this
15789 Library code that is intended to operate with more than one UNIX
15790 standard must test, set and restore the variable @code{__xpg4_extended_mask}
15791 as appropriate. Most GNU software doesn't provide this capability.
15795 Suppress the generation of link options to search libdld.sl when the
15796 @option{-static} option is specified on HP-UX 10 and later.
15800 The HP-UX implementation of setlocale in libc has a dependency on
15801 libdld.sl. There isn't an archive version of libdld.sl. Thus,
15802 when the @option{-static} option is specified, special link options
15803 are needed to resolve this dependency.
15805 On HP-UX 10 and later, the GCC driver adds the necessary options to
15806 link with libdld.sl when the @option{-static} option is specified.
15807 This causes the resulting binary to be dynamic. On the 64-bit port,
15808 the linkers generate dynamic binaries by default in any case. The
15809 @option{-nolibdld} option can be used to prevent the GCC driver from
15810 adding these link options.
15814 Add support for multithreading with the @dfn{dce thread} library
15815 under HP-UX@. This option sets flags for both the preprocessor and
15819 @node IA-64 Options
15820 @subsection IA-64 Options
15821 @cindex IA-64 Options
15823 These are the @samp{-m} options defined for the Intel IA-64 architecture.
15827 @opindex mbig-endian
15828 Generate code for a big-endian target. This is the default for HP-UX@.
15830 @item -mlittle-endian
15831 @opindex mlittle-endian
15832 Generate code for a little-endian target. This is the default for AIX5
15838 @opindex mno-gnu-as
15839 Generate (or don't) code for the GNU assembler. This is the default.
15840 @c Also, this is the default if the configure option @option{--with-gnu-as}
15846 @opindex mno-gnu-ld
15847 Generate (or don't) code for the GNU linker. This is the default.
15848 @c Also, this is the default if the configure option @option{--with-gnu-ld}
15853 Generate code that does not use a global pointer register. The result
15854 is not position independent code, and violates the IA-64 ABI@.
15856 @item -mvolatile-asm-stop
15857 @itemx -mno-volatile-asm-stop
15858 @opindex mvolatile-asm-stop
15859 @opindex mno-volatile-asm-stop
15860 Generate (or don't) a stop bit immediately before and after volatile asm
15863 @item -mregister-names
15864 @itemx -mno-register-names
15865 @opindex mregister-names
15866 @opindex mno-register-names
15867 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
15868 the stacked registers. This may make assembler output more readable.
15874 Disable (or enable) optimizations that use the small data section. This may
15875 be useful for working around optimizer bugs.
15877 @item -mconstant-gp
15878 @opindex mconstant-gp
15879 Generate code that uses a single constant global pointer value. This is
15880 useful when compiling kernel code.
15884 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
15885 This is useful when compiling firmware code.
15887 @item -minline-float-divide-min-latency
15888 @opindex minline-float-divide-min-latency
15889 Generate code for inline divides of floating-point values
15890 using the minimum latency algorithm.
15892 @item -minline-float-divide-max-throughput
15893 @opindex minline-float-divide-max-throughput
15894 Generate code for inline divides of floating-point values
15895 using the maximum throughput algorithm.
15897 @item -mno-inline-float-divide
15898 @opindex mno-inline-float-divide
15899 Do not generate inline code for divides of floating-point values.
15901 @item -minline-int-divide-min-latency
15902 @opindex minline-int-divide-min-latency
15903 Generate code for inline divides of integer values
15904 using the minimum latency algorithm.
15906 @item -minline-int-divide-max-throughput
15907 @opindex minline-int-divide-max-throughput
15908 Generate code for inline divides of integer values
15909 using the maximum throughput algorithm.
15911 @item -mno-inline-int-divide
15912 @opindex mno-inline-int-divide
15913 Do not generate inline code for divides of integer values.
15915 @item -minline-sqrt-min-latency
15916 @opindex minline-sqrt-min-latency
15917 Generate code for inline square roots
15918 using the minimum latency algorithm.
15920 @item -minline-sqrt-max-throughput
15921 @opindex minline-sqrt-max-throughput
15922 Generate code for inline square roots
15923 using the maximum throughput algorithm.
15925 @item -mno-inline-sqrt
15926 @opindex mno-inline-sqrt
15927 Do not generate inline code for @code{sqrt}.
15930 @itemx -mno-fused-madd
15931 @opindex mfused-madd
15932 @opindex mno-fused-madd
15933 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
15934 instructions. The default is to use these instructions.
15936 @item -mno-dwarf2-asm
15937 @itemx -mdwarf2-asm
15938 @opindex mno-dwarf2-asm
15939 @opindex mdwarf2-asm
15940 Don't (or do) generate assembler code for the DWARF 2 line number debugging
15941 info. This may be useful when not using the GNU assembler.
15943 @item -mearly-stop-bits
15944 @itemx -mno-early-stop-bits
15945 @opindex mearly-stop-bits
15946 @opindex mno-early-stop-bits
15947 Allow stop bits to be placed earlier than immediately preceding the
15948 instruction that triggered the stop bit. This can improve instruction
15949 scheduling, but does not always do so.
15951 @item -mfixed-range=@var{register-range}
15952 @opindex mfixed-range
15953 Generate code treating the given register range as fixed registers.
15954 A fixed register is one that the register allocator cannot use. This is
15955 useful when compiling kernel code. A register range is specified as
15956 two registers separated by a dash. Multiple register ranges can be
15957 specified separated by a comma.
15959 @item -mtls-size=@var{tls-size}
15961 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
15964 @item -mtune=@var{cpu-type}
15966 Tune the instruction scheduling for a particular CPU, Valid values are
15967 @samp{itanium}, @samp{itanium1}, @samp{merced}, @samp{itanium2},
15968 and @samp{mckinley}.
15974 Generate code for a 32-bit or 64-bit environment.
15975 The 32-bit environment sets int, long and pointer to 32 bits.
15976 The 64-bit environment sets int to 32 bits and long and pointer
15977 to 64 bits. These are HP-UX specific flags.
15979 @item -mno-sched-br-data-spec
15980 @itemx -msched-br-data-spec
15981 @opindex mno-sched-br-data-spec
15982 @opindex msched-br-data-spec
15983 (Dis/En)able data speculative scheduling before reload.
15984 This results in generation of @code{ld.a} instructions and
15985 the corresponding check instructions (@code{ld.c} / @code{chk.a}).
15986 The default is 'disable'.
15988 @item -msched-ar-data-spec
15989 @itemx -mno-sched-ar-data-spec
15990 @opindex msched-ar-data-spec
15991 @opindex mno-sched-ar-data-spec
15992 (En/Dis)able data speculative scheduling after reload.
15993 This results in generation of @code{ld.a} instructions and
15994 the corresponding check instructions (@code{ld.c} / @code{chk.a}).
15995 The default is 'enable'.
15997 @item -mno-sched-control-spec
15998 @itemx -msched-control-spec
15999 @opindex mno-sched-control-spec
16000 @opindex msched-control-spec
16001 (Dis/En)able control speculative scheduling. This feature is
16002 available only during region scheduling (i.e.@: before reload).
16003 This results in generation of the @code{ld.s} instructions and
16004 the corresponding check instructions @code{chk.s}.
16005 The default is 'disable'.
16007 @item -msched-br-in-data-spec
16008 @itemx -mno-sched-br-in-data-spec
16009 @opindex msched-br-in-data-spec
16010 @opindex mno-sched-br-in-data-spec
16011 (En/Dis)able speculative scheduling of the instructions that
16012 are dependent on the data speculative loads before reload.
16013 This is effective only with @option{-msched-br-data-spec} enabled.
16014 The default is 'enable'.
16016 @item -msched-ar-in-data-spec
16017 @itemx -mno-sched-ar-in-data-spec
16018 @opindex msched-ar-in-data-spec
16019 @opindex mno-sched-ar-in-data-spec
16020 (En/Dis)able speculative scheduling of the instructions that
16021 are dependent on the data speculative loads after reload.
16022 This is effective only with @option{-msched-ar-data-spec} enabled.
16023 The default is 'enable'.
16025 @item -msched-in-control-spec
16026 @itemx -mno-sched-in-control-spec
16027 @opindex msched-in-control-spec
16028 @opindex mno-sched-in-control-spec
16029 (En/Dis)able speculative scheduling of the instructions that
16030 are dependent on the control speculative loads.
16031 This is effective only with @option{-msched-control-spec} enabled.
16032 The default is 'enable'.
16034 @item -mno-sched-prefer-non-data-spec-insns
16035 @itemx -msched-prefer-non-data-spec-insns
16036 @opindex mno-sched-prefer-non-data-spec-insns
16037 @opindex msched-prefer-non-data-spec-insns
16038 If enabled, data-speculative instructions are chosen for schedule
16039 only if there are no other choices at the moment. This makes
16040 the use of the data speculation much more conservative.
16041 The default is 'disable'.
16043 @item -mno-sched-prefer-non-control-spec-insns
16044 @itemx -msched-prefer-non-control-spec-insns
16045 @opindex mno-sched-prefer-non-control-spec-insns
16046 @opindex msched-prefer-non-control-spec-insns
16047 If enabled, control-speculative instructions are chosen for schedule
16048 only if there are no other choices at the moment. This makes
16049 the use of the control speculation much more conservative.
16050 The default is 'disable'.
16052 @item -mno-sched-count-spec-in-critical-path
16053 @itemx -msched-count-spec-in-critical-path
16054 @opindex mno-sched-count-spec-in-critical-path
16055 @opindex msched-count-spec-in-critical-path
16056 If enabled, speculative dependencies are considered during
16057 computation of the instructions priorities. This makes the use of the
16058 speculation a bit more conservative.
16059 The default is 'disable'.
16061 @item -msched-spec-ldc
16062 @opindex msched-spec-ldc
16063 Use a simple data speculation check. This option is on by default.
16065 @item -msched-control-spec-ldc
16066 @opindex msched-spec-ldc
16067 Use a simple check for control speculation. This option is on by default.
16069 @item -msched-stop-bits-after-every-cycle
16070 @opindex msched-stop-bits-after-every-cycle
16071 Place a stop bit after every cycle when scheduling. This option is on
16074 @item -msched-fp-mem-deps-zero-cost
16075 @opindex msched-fp-mem-deps-zero-cost
16076 Assume that floating-point stores and loads are not likely to cause a conflict
16077 when placed into the same instruction group. This option is disabled by
16080 @item -msel-sched-dont-check-control-spec
16081 @opindex msel-sched-dont-check-control-spec
16082 Generate checks for control speculation in selective scheduling.
16083 This flag is disabled by default.
16085 @item -msched-max-memory-insns=@var{max-insns}
16086 @opindex msched-max-memory-insns
16087 Limit on the number of memory insns per instruction group, giving lower
16088 priority to subsequent memory insns attempting to schedule in the same
16089 instruction group. Frequently useful to prevent cache bank conflicts.
16090 The default value is 1.
16092 @item -msched-max-memory-insns-hard-limit
16093 @opindex msched-max-memory-insns-hard-limit
16094 Makes the limit specified by @option{msched-max-memory-insns} a hard limit,
16095 disallowing more than that number in an instruction group.
16096 Otherwise, the limit is ``soft'', meaning that non-memory operations
16097 are preferred when the limit is reached, but memory operations may still
16103 @subsection LM32 Options
16104 @cindex LM32 options
16106 These @option{-m} options are defined for the LatticeMico32 architecture:
16109 @item -mbarrel-shift-enabled
16110 @opindex mbarrel-shift-enabled
16111 Enable barrel-shift instructions.
16113 @item -mdivide-enabled
16114 @opindex mdivide-enabled
16115 Enable divide and modulus instructions.
16117 @item -mmultiply-enabled
16118 @opindex multiply-enabled
16119 Enable multiply instructions.
16121 @item -msign-extend-enabled
16122 @opindex msign-extend-enabled
16123 Enable sign extend instructions.
16125 @item -muser-enabled
16126 @opindex muser-enabled
16127 Enable user-defined instructions.
16132 @subsection M32C Options
16133 @cindex M32C options
16136 @item -mcpu=@var{name}
16138 Select the CPU for which code is generated. @var{name} may be one of
16139 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
16140 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
16141 the M32C/80 series.
16145 Specifies that the program will be run on the simulator. This causes
16146 an alternate runtime library to be linked in which supports, for
16147 example, file I/O@. You must not use this option when generating
16148 programs that will run on real hardware; you must provide your own
16149 runtime library for whatever I/O functions are needed.
16151 @item -memregs=@var{number}
16153 Specifies the number of memory-based pseudo-registers GCC uses
16154 during code generation. These pseudo-registers are used like real
16155 registers, so there is a tradeoff between GCC's ability to fit the
16156 code into available registers, and the performance penalty of using
16157 memory instead of registers. Note that all modules in a program must
16158 be compiled with the same value for this option. Because of that, you
16159 must not use this option with GCC's default runtime libraries.
16163 @node M32R/D Options
16164 @subsection M32R/D Options
16165 @cindex M32R/D options
16167 These @option{-m} options are defined for Renesas M32R/D architectures:
16172 Generate code for the M32R/2@.
16176 Generate code for the M32R/X@.
16180 Generate code for the M32R@. This is the default.
16182 @item -mmodel=small
16183 @opindex mmodel=small
16184 Assume all objects live in the lower 16MB of memory (so that their addresses
16185 can be loaded with the @code{ld24} instruction), and assume all subroutines
16186 are reachable with the @code{bl} instruction.
16187 This is the default.
16189 The addressability of a particular object can be set with the
16190 @code{model} attribute.
16192 @item -mmodel=medium
16193 @opindex mmodel=medium
16194 Assume objects may be anywhere in the 32-bit address space (the compiler
16195 generates @code{seth/add3} instructions to load their addresses), and
16196 assume all subroutines are reachable with the @code{bl} instruction.
16198 @item -mmodel=large
16199 @opindex mmodel=large
16200 Assume objects may be anywhere in the 32-bit address space (the compiler
16201 generates @code{seth/add3} instructions to load their addresses), and
16202 assume subroutines may not be reachable with the @code{bl} instruction
16203 (the compiler generates the much slower @code{seth/add3/jl}
16204 instruction sequence).
16207 @opindex msdata=none
16208 Disable use of the small data area. Variables are put into
16209 one of @code{.data}, @code{.bss}, or @code{.rodata} (unless the
16210 @code{section} attribute has been specified).
16211 This is the default.
16213 The small data area consists of sections @code{.sdata} and @code{.sbss}.
16214 Objects may be explicitly put in the small data area with the
16215 @code{section} attribute using one of these sections.
16217 @item -msdata=sdata
16218 @opindex msdata=sdata
16219 Put small global and static data in the small data area, but do not
16220 generate special code to reference them.
16223 @opindex msdata=use
16224 Put small global and static data in the small data area, and generate
16225 special instructions to reference them.
16229 @cindex smaller data references
16230 Put global and static objects less than or equal to @var{num} bytes
16231 into the small data or BSS sections instead of the normal data or BSS
16232 sections. The default value of @var{num} is 8.
16233 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
16234 for this option to have any effect.
16236 All modules should be compiled with the same @option{-G @var{num}} value.
16237 Compiling with different values of @var{num} may or may not work; if it
16238 doesn't the linker gives an error message---incorrect code is not
16243 Makes the M32R-specific code in the compiler display some statistics
16244 that might help in debugging programs.
16246 @item -malign-loops
16247 @opindex malign-loops
16248 Align all loops to a 32-byte boundary.
16250 @item -mno-align-loops
16251 @opindex mno-align-loops
16252 Do not enforce a 32-byte alignment for loops. This is the default.
16254 @item -missue-rate=@var{number}
16255 @opindex missue-rate=@var{number}
16256 Issue @var{number} instructions per cycle. @var{number} can only be 1
16259 @item -mbranch-cost=@var{number}
16260 @opindex mbranch-cost=@var{number}
16261 @var{number} can only be 1 or 2. If it is 1 then branches are
16262 preferred over conditional code, if it is 2, then the opposite applies.
16264 @item -mflush-trap=@var{number}
16265 @opindex mflush-trap=@var{number}
16266 Specifies the trap number to use to flush the cache. The default is
16267 12. Valid numbers are between 0 and 15 inclusive.
16269 @item -mno-flush-trap
16270 @opindex mno-flush-trap
16271 Specifies that the cache cannot be flushed by using a trap.
16273 @item -mflush-func=@var{name}
16274 @opindex mflush-func=@var{name}
16275 Specifies the name of the operating system function to call to flush
16276 the cache. The default is @samp{_flush_cache}, but a function call
16277 is only used if a trap is not available.
16279 @item -mno-flush-func
16280 @opindex mno-flush-func
16281 Indicates that there is no OS function for flushing the cache.
16285 @node M680x0 Options
16286 @subsection M680x0 Options
16287 @cindex M680x0 options
16289 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
16290 The default settings depend on which architecture was selected when
16291 the compiler was configured; the defaults for the most common choices
16295 @item -march=@var{arch}
16297 Generate code for a specific M680x0 or ColdFire instruction set
16298 architecture. Permissible values of @var{arch} for M680x0
16299 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
16300 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
16301 architectures are selected according to Freescale's ISA classification
16302 and the permissible values are: @samp{isaa}, @samp{isaaplus},
16303 @samp{isab} and @samp{isac}.
16305 GCC defines a macro @code{__mcf@var{arch}__} whenever it is generating
16306 code for a ColdFire target. The @var{arch} in this macro is one of the
16307 @option{-march} arguments given above.
16309 When used together, @option{-march} and @option{-mtune} select code
16310 that runs on a family of similar processors but that is optimized
16311 for a particular microarchitecture.
16313 @item -mcpu=@var{cpu}
16315 Generate code for a specific M680x0 or ColdFire processor.
16316 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
16317 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
16318 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
16319 below, which also classifies the CPUs into families:
16321 @multitable @columnfractions 0.20 0.80
16322 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
16323 @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}
16324 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
16325 @item @samp{5206e} @tab @samp{5206e}
16326 @item @samp{5208} @tab @samp{5207} @samp{5208}
16327 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
16328 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
16329 @item @samp{5216} @tab @samp{5214} @samp{5216}
16330 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
16331 @item @samp{5225} @tab @samp{5224} @samp{5225}
16332 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
16333 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
16334 @item @samp{5249} @tab @samp{5249}
16335 @item @samp{5250} @tab @samp{5250}
16336 @item @samp{5271} @tab @samp{5270} @samp{5271}
16337 @item @samp{5272} @tab @samp{5272}
16338 @item @samp{5275} @tab @samp{5274} @samp{5275}
16339 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
16340 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
16341 @item @samp{5307} @tab @samp{5307}
16342 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
16343 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
16344 @item @samp{5407} @tab @samp{5407}
16345 @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}
16348 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
16349 @var{arch} is compatible with @var{cpu}. Other combinations of
16350 @option{-mcpu} and @option{-march} are rejected.
16352 GCC defines the macro @code{__mcf_cpu_@var{cpu}} when ColdFire target
16353 @var{cpu} is selected. It also defines @code{__mcf_family_@var{family}},
16354 where the value of @var{family} is given by the table above.
16356 @item -mtune=@var{tune}
16358 Tune the code for a particular microarchitecture within the
16359 constraints set by @option{-march} and @option{-mcpu}.
16360 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
16361 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
16362 and @samp{cpu32}. The ColdFire microarchitectures
16363 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
16365 You can also use @option{-mtune=68020-40} for code that needs
16366 to run relatively well on 68020, 68030 and 68040 targets.
16367 @option{-mtune=68020-60} is similar but includes 68060 targets
16368 as well. These two options select the same tuning decisions as
16369 @option{-m68020-40} and @option{-m68020-60} respectively.
16371 GCC defines the macros @code{__mc@var{arch}} and @code{__mc@var{arch}__}
16372 when tuning for 680x0 architecture @var{arch}. It also defines
16373 @code{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
16374 option is used. If GCC is tuning for a range of architectures,
16375 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
16376 it defines the macros for every architecture in the range.
16378 GCC also defines the macro @code{__m@var{uarch}__} when tuning for
16379 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
16380 of the arguments given above.
16386 Generate output for a 68000. This is the default
16387 when the compiler is configured for 68000-based systems.
16388 It is equivalent to @option{-march=68000}.
16390 Use this option for microcontrollers with a 68000 or EC000 core,
16391 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
16395 Generate output for a 68010. This is the default
16396 when the compiler is configured for 68010-based systems.
16397 It is equivalent to @option{-march=68010}.
16403 Generate output for a 68020. This is the default
16404 when the compiler is configured for 68020-based systems.
16405 It is equivalent to @option{-march=68020}.
16409 Generate output for a 68030. This is the default when the compiler is
16410 configured for 68030-based systems. It is equivalent to
16411 @option{-march=68030}.
16415 Generate output for a 68040. This is the default when the compiler is
16416 configured for 68040-based systems. It is equivalent to
16417 @option{-march=68040}.
16419 This option inhibits the use of 68881/68882 instructions that have to be
16420 emulated by software on the 68040. Use this option if your 68040 does not
16421 have code to emulate those instructions.
16425 Generate output for a 68060. This is the default when the compiler is
16426 configured for 68060-based systems. It is equivalent to
16427 @option{-march=68060}.
16429 This option inhibits the use of 68020 and 68881/68882 instructions that
16430 have to be emulated by software on the 68060. Use this option if your 68060
16431 does not have code to emulate those instructions.
16435 Generate output for a CPU32. This is the default
16436 when the compiler is configured for CPU32-based systems.
16437 It is equivalent to @option{-march=cpu32}.
16439 Use this option for microcontrollers with a
16440 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
16441 68336, 68340, 68341, 68349 and 68360.
16445 Generate output for a 520X ColdFire CPU@. This is the default
16446 when the compiler is configured for 520X-based systems.
16447 It is equivalent to @option{-mcpu=5206}, and is now deprecated
16448 in favor of that option.
16450 Use this option for microcontroller with a 5200 core, including
16451 the MCF5202, MCF5203, MCF5204 and MCF5206.
16455 Generate output for a 5206e ColdFire CPU@. The option is now
16456 deprecated in favor of the equivalent @option{-mcpu=5206e}.
16460 Generate output for a member of the ColdFire 528X family.
16461 The option is now deprecated in favor of the equivalent
16462 @option{-mcpu=528x}.
16466 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
16467 in favor of the equivalent @option{-mcpu=5307}.
16471 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
16472 in favor of the equivalent @option{-mcpu=5407}.
16476 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
16477 This includes use of hardware floating-point instructions.
16478 The option is equivalent to @option{-mcpu=547x}, and is now
16479 deprecated in favor of that option.
16483 Generate output for a 68040, without using any of the new instructions.
16484 This results in code that can run relatively efficiently on either a
16485 68020/68881 or a 68030 or a 68040. The generated code does use the
16486 68881 instructions that are emulated on the 68040.
16488 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
16492 Generate output for a 68060, without using any of the new instructions.
16493 This results in code that can run relatively efficiently on either a
16494 68020/68881 or a 68030 or a 68040. The generated code does use the
16495 68881 instructions that are emulated on the 68060.
16497 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
16501 @opindex mhard-float
16503 Generate floating-point instructions. This is the default for 68020
16504 and above, and for ColdFire devices that have an FPU@. It defines the
16505 macro @code{__HAVE_68881__} on M680x0 targets and @code{__mcffpu__}
16506 on ColdFire targets.
16509 @opindex msoft-float
16510 Do not generate floating-point instructions; use library calls instead.
16511 This is the default for 68000, 68010, and 68832 targets. It is also
16512 the default for ColdFire devices that have no FPU.
16518 Generate (do not generate) ColdFire hardware divide and remainder
16519 instructions. If @option{-march} is used without @option{-mcpu},
16520 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
16521 architectures. Otherwise, the default is taken from the target CPU
16522 (either the default CPU, or the one specified by @option{-mcpu}). For
16523 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
16524 @option{-mcpu=5206e}.
16526 GCC defines the macro @code{__mcfhwdiv__} when this option is enabled.
16530 Consider type @code{int} to be 16 bits wide, like @code{short int}.
16531 Additionally, parameters passed on the stack are also aligned to a
16532 16-bit boundary even on targets whose API mandates promotion to 32-bit.
16536 Do not consider type @code{int} to be 16 bits wide. This is the default.
16539 @itemx -mno-bitfield
16540 @opindex mnobitfield
16541 @opindex mno-bitfield
16542 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
16543 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
16547 Do use the bit-field instructions. The @option{-m68020} option implies
16548 @option{-mbitfield}. This is the default if you use a configuration
16549 designed for a 68020.
16553 Use a different function-calling convention, in which functions
16554 that take a fixed number of arguments return with the @code{rtd}
16555 instruction, which pops their arguments while returning. This
16556 saves one instruction in the caller since there is no need to pop
16557 the arguments there.
16559 This calling convention is incompatible with the one normally
16560 used on Unix, so you cannot use it if you need to call libraries
16561 compiled with the Unix compiler.
16563 Also, you must provide function prototypes for all functions that
16564 take variable numbers of arguments (including @code{printf});
16565 otherwise incorrect code is generated for calls to those
16568 In addition, seriously incorrect code results if you call a
16569 function with too many arguments. (Normally, extra arguments are
16570 harmlessly ignored.)
16572 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
16573 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
16577 Do not use the calling conventions selected by @option{-mrtd}.
16578 This is the default.
16581 @itemx -mno-align-int
16582 @opindex malign-int
16583 @opindex mno-align-int
16584 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
16585 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
16586 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
16587 Aligning variables on 32-bit boundaries produces code that runs somewhat
16588 faster on processors with 32-bit busses at the expense of more memory.
16590 @strong{Warning:} if you use the @option{-malign-int} switch, GCC
16591 aligns structures containing the above types differently than
16592 most published application binary interface specifications for the m68k.
16596 Use the pc-relative addressing mode of the 68000 directly, instead of
16597 using a global offset table. At present, this option implies @option{-fpic},
16598 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
16599 not presently supported with @option{-mpcrel}, though this could be supported for
16600 68020 and higher processors.
16602 @item -mno-strict-align
16603 @itemx -mstrict-align
16604 @opindex mno-strict-align
16605 @opindex mstrict-align
16606 Do not (do) assume that unaligned memory references are handled by
16610 Generate code that allows the data segment to be located in a different
16611 area of memory from the text segment. This allows for execute-in-place in
16612 an environment without virtual memory management. This option implies
16615 @item -mno-sep-data
16616 Generate code that assumes that the data segment follows the text segment.
16617 This is the default.
16619 @item -mid-shared-library
16620 Generate code that supports shared libraries via the library ID method.
16621 This allows for execute-in-place and shared libraries in an environment
16622 without virtual memory management. This option implies @option{-fPIC}.
16624 @item -mno-id-shared-library
16625 Generate code that doesn't assume ID-based shared libraries are being used.
16626 This is the default.
16628 @item -mshared-library-id=n
16629 Specifies the identification number of the ID-based shared library being
16630 compiled. Specifying a value of 0 generates more compact code; specifying
16631 other values forces the allocation of that number to the current
16632 library, but is no more space- or time-efficient than omitting this option.
16638 When generating position-independent code for ColdFire, generate code
16639 that works if the GOT has more than 8192 entries. This code is
16640 larger and slower than code generated without this option. On M680x0
16641 processors, this option is not needed; @option{-fPIC} suffices.
16643 GCC normally uses a single instruction to load values from the GOT@.
16644 While this is relatively efficient, it only works if the GOT
16645 is smaller than about 64k. Anything larger causes the linker
16646 to report an error such as:
16648 @cindex relocation truncated to fit (ColdFire)
16650 relocation truncated to fit: R_68K_GOT16O foobar
16653 If this happens, you should recompile your code with @option{-mxgot}.
16654 It should then work with very large GOTs. However, code generated with
16655 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
16656 the value of a global symbol.
16658 Note that some linkers, including newer versions of the GNU linker,
16659 can create multiple GOTs and sort GOT entries. If you have such a linker,
16660 you should only need to use @option{-mxgot} when compiling a single
16661 object file that accesses more than 8192 GOT entries. Very few do.
16663 These options have no effect unless GCC is generating
16664 position-independent code.
16668 @node MCore Options
16669 @subsection MCore Options
16670 @cindex MCore options
16672 These are the @samp{-m} options defined for the Motorola M*Core
16678 @itemx -mno-hardlit
16680 @opindex mno-hardlit
16681 Inline constants into the code stream if it can be done in two
16682 instructions or less.
16688 Use the divide instruction. (Enabled by default).
16690 @item -mrelax-immediate
16691 @itemx -mno-relax-immediate
16692 @opindex mrelax-immediate
16693 @opindex mno-relax-immediate
16694 Allow arbitrary-sized immediates in bit operations.
16696 @item -mwide-bitfields
16697 @itemx -mno-wide-bitfields
16698 @opindex mwide-bitfields
16699 @opindex mno-wide-bitfields
16700 Always treat bit-fields as @code{int}-sized.
16702 @item -m4byte-functions
16703 @itemx -mno-4byte-functions
16704 @opindex m4byte-functions
16705 @opindex mno-4byte-functions
16706 Force all functions to be aligned to a 4-byte boundary.
16708 @item -mcallgraph-data
16709 @itemx -mno-callgraph-data
16710 @opindex mcallgraph-data
16711 @opindex mno-callgraph-data
16712 Emit callgraph information.
16715 @itemx -mno-slow-bytes
16716 @opindex mslow-bytes
16717 @opindex mno-slow-bytes
16718 Prefer word access when reading byte quantities.
16720 @item -mlittle-endian
16721 @itemx -mbig-endian
16722 @opindex mlittle-endian
16723 @opindex mbig-endian
16724 Generate code for a little-endian target.
16730 Generate code for the 210 processor.
16734 Assume that runtime support has been provided and so omit the
16735 simulator library (@file{libsim.a)} from the linker command line.
16737 @item -mstack-increment=@var{size}
16738 @opindex mstack-increment
16739 Set the maximum amount for a single stack increment operation. Large
16740 values can increase the speed of programs that contain functions
16741 that need a large amount of stack space, but they can also trigger a
16742 segmentation fault if the stack is extended too much. The default
16748 @subsection MeP Options
16749 @cindex MeP options
16755 Enables the @code{abs} instruction, which is the absolute difference
16756 between two registers.
16760 Enables all the optional instructions---average, multiply, divide, bit
16761 operations, leading zero, absolute difference, min/max, clip, and
16767 Enables the @code{ave} instruction, which computes the average of two
16770 @item -mbased=@var{n}
16772 Variables of size @var{n} bytes or smaller are placed in the
16773 @code{.based} section by default. Based variables use the @code{$tp}
16774 register as a base register, and there is a 128-byte limit to the
16775 @code{.based} section.
16779 Enables the bit operation instructions---bit test (@code{btstm}), set
16780 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
16781 test-and-set (@code{tas}).
16783 @item -mc=@var{name}
16785 Selects which section constant data is placed in. @var{name} may
16786 be @samp{tiny}, @samp{near}, or @samp{far}.
16790 Enables the @code{clip} instruction. Note that @option{-mclip} is not
16791 useful unless you also provide @option{-mminmax}.
16793 @item -mconfig=@var{name}
16795 Selects one of the built-in core configurations. Each MeP chip has
16796 one or more modules in it; each module has a core CPU and a variety of
16797 coprocessors, optional instructions, and peripherals. The
16798 @code{MeP-Integrator} tool, not part of GCC, provides these
16799 configurations through this option; using this option is the same as
16800 using all the corresponding command-line options. The default
16801 configuration is @samp{default}.
16805 Enables the coprocessor instructions. By default, this is a 32-bit
16806 coprocessor. Note that the coprocessor is normally enabled via the
16807 @option{-mconfig=} option.
16811 Enables the 32-bit coprocessor's instructions.
16815 Enables the 64-bit coprocessor's instructions.
16819 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
16823 Causes constant variables to be placed in the @code{.near} section.
16827 Enables the @code{div} and @code{divu} instructions.
16831 Generate big-endian code.
16835 Generate little-endian code.
16837 @item -mio-volatile
16838 @opindex mio-volatile
16839 Tells the compiler that any variable marked with the @code{io}
16840 attribute is to be considered volatile.
16844 Causes variables to be assigned to the @code{.far} section by default.
16848 Enables the @code{leadz} (leading zero) instruction.
16852 Causes variables to be assigned to the @code{.near} section by default.
16856 Enables the @code{min} and @code{max} instructions.
16860 Enables the multiplication and multiply-accumulate instructions.
16864 Disables all the optional instructions enabled by @option{-mall-opts}.
16868 Enables the @code{repeat} and @code{erepeat} instructions, used for
16869 low-overhead looping.
16873 Causes all variables to default to the @code{.tiny} section. Note
16874 that there is a 65536-byte limit to this section. Accesses to these
16875 variables use the @code{%gp} base register.
16879 Enables the saturation instructions. Note that the compiler does not
16880 currently generate these itself, but this option is included for
16881 compatibility with other tools, like @code{as}.
16885 Link the SDRAM-based runtime instead of the default ROM-based runtime.
16889 Link the simulator run-time libraries.
16893 Link the simulator runtime libraries, excluding built-in support
16894 for reset and exception vectors and tables.
16898 Causes all functions to default to the @code{.far} section. Without
16899 this option, functions default to the @code{.near} section.
16901 @item -mtiny=@var{n}
16903 Variables that are @var{n} bytes or smaller are allocated to the
16904 @code{.tiny} section. These variables use the @code{$gp} base
16905 register. The default for this option is 4, but note that there's a
16906 65536-byte limit to the @code{.tiny} section.
16910 @node MicroBlaze Options
16911 @subsection MicroBlaze Options
16912 @cindex MicroBlaze Options
16917 @opindex msoft-float
16918 Use software emulation for floating point (default).
16921 @opindex mhard-float
16922 Use hardware floating-point instructions.
16926 Do not optimize block moves, use @code{memcpy}.
16928 @item -mno-clearbss
16929 @opindex mno-clearbss
16930 This option is deprecated. Use @option{-fno-zero-initialized-in-bss} instead.
16932 @item -mcpu=@var{cpu-type}
16934 Use features of, and schedule code for, the given CPU.
16935 Supported values are in the format @samp{v@var{X}.@var{YY}.@var{Z}},
16936 where @var{X} is a major version, @var{YY} is the minor version, and
16937 @var{Z} is compatibility code. Example values are @samp{v3.00.a},
16938 @samp{v4.00.b}, @samp{v5.00.a}, @samp{v5.00.b}, @samp{v5.00.b}, @samp{v6.00.a}.
16940 @item -mxl-soft-mul
16941 @opindex mxl-soft-mul
16942 Use software multiply emulation (default).
16944 @item -mxl-soft-div
16945 @opindex mxl-soft-div
16946 Use software emulation for divides (default).
16948 @item -mxl-barrel-shift
16949 @opindex mxl-barrel-shift
16950 Use the hardware barrel shifter.
16952 @item -mxl-pattern-compare
16953 @opindex mxl-pattern-compare
16954 Use pattern compare instructions.
16956 @item -msmall-divides
16957 @opindex msmall-divides
16958 Use table lookup optimization for small signed integer divisions.
16960 @item -mxl-stack-check
16961 @opindex mxl-stack-check
16962 This option is deprecated. Use @option{-fstack-check} instead.
16965 @opindex mxl-gp-opt
16966 Use GP-relative @code{.sdata}/@code{.sbss} sections.
16968 @item -mxl-multiply-high
16969 @opindex mxl-multiply-high
16970 Use multiply high instructions for high part of 32x32 multiply.
16972 @item -mxl-float-convert
16973 @opindex mxl-float-convert
16974 Use hardware floating-point conversion instructions.
16976 @item -mxl-float-sqrt
16977 @opindex mxl-float-sqrt
16978 Use hardware floating-point square root instruction.
16981 @opindex mbig-endian
16982 Generate code for a big-endian target.
16984 @item -mlittle-endian
16985 @opindex mlittle-endian
16986 Generate code for a little-endian target.
16989 @opindex mxl-reorder
16990 Use reorder instructions (swap and byte reversed load/store).
16992 @item -mxl-mode-@var{app-model}
16993 Select application model @var{app-model}. Valid models are
16996 normal executable (default), uses startup code @file{crt0.o}.
16999 for use with Xilinx Microprocessor Debugger (XMD) based
17000 software intrusive debug agent called xmdstub. This uses startup file
17001 @file{crt1.o} and sets the start address of the program to 0x800.
17004 for applications that are loaded using a bootloader.
17005 This model uses startup file @file{crt2.o} which does not contain a processor
17006 reset vector handler. This is suitable for transferring control on a
17007 processor reset to the bootloader rather than the application.
17010 for applications that do not require any of the
17011 MicroBlaze vectors. This option may be useful for applications running
17012 within a monitoring application. This model uses @file{crt3.o} as a startup file.
17015 Option @option{-xl-mode-@var{app-model}} is a deprecated alias for
17016 @option{-mxl-mode-@var{app-model}}.
17021 @subsection MIPS Options
17022 @cindex MIPS options
17028 Generate big-endian code.
17032 Generate little-endian code. This is the default for @samp{mips*el-*-*}
17035 @item -march=@var{arch}
17037 Generate code that runs on @var{arch}, which can be the name of a
17038 generic MIPS ISA, or the name of a particular processor.
17040 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
17041 @samp{mips32}, @samp{mips32r2}, @samp{mips32r3}, @samp{mips32r5},
17042 @samp{mips32r6}, @samp{mips64}, @samp{mips64r2}, @samp{mips64r3},
17043 @samp{mips64r5} and @samp{mips64r6}.
17044 The processor names are:
17045 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
17046 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
17047 @samp{5kc}, @samp{5kf},
17049 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
17050 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
17051 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1}, @samp{34kn},
17052 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
17053 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
17054 @samp{loongson2e}, @samp{loongson2f}, @samp{loongson3a},
17056 @samp{m14k}, @samp{m14kc}, @samp{m14ke}, @samp{m14kec},
17057 @samp{octeon}, @samp{octeon+}, @samp{octeon2}, @samp{octeon3},
17060 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
17061 @samp{r4600}, @samp{r4650}, @samp{r4700}, @samp{r6000}, @samp{r8000},
17062 @samp{rm7000}, @samp{rm9000},
17063 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
17066 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
17067 @samp{vr5000}, @samp{vr5400}, @samp{vr5500},
17068 @samp{xlr} and @samp{xlp}.
17069 The special value @samp{from-abi} selects the
17070 most compatible architecture for the selected ABI (that is,
17071 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
17073 The native Linux/GNU toolchain also supports the value @samp{native},
17074 which selects the best architecture option for the host processor.
17075 @option{-march=native} has no effect if GCC does not recognize
17078 In processor names, a final @samp{000} can be abbreviated as @samp{k}
17079 (for example, @option{-march=r2k}). Prefixes are optional, and
17080 @samp{vr} may be written @samp{r}.
17082 Names of the form @samp{@var{n}f2_1} refer to processors with
17083 FPUs clocked at half the rate of the core, names of the form
17084 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
17085 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
17086 processors with FPUs clocked a ratio of 3:2 with respect to the core.
17087 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
17088 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
17089 accepted as synonyms for @samp{@var{n}f1_1}.
17091 GCC defines two macros based on the value of this option. The first
17092 is @code{_MIPS_ARCH}, which gives the name of target architecture, as
17093 a string. The second has the form @code{_MIPS_ARCH_@var{foo}},
17094 where @var{foo} is the capitalized value of @code{_MIPS_ARCH}@.
17095 For example, @option{-march=r2000} sets @code{_MIPS_ARCH}
17096 to @code{"r2000"} and defines the macro @code{_MIPS_ARCH_R2000}.
17098 Note that the @code{_MIPS_ARCH} macro uses the processor names given
17099 above. In other words, it has the full prefix and does not
17100 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
17101 the macro names the resolved architecture (either @code{"mips1"} or
17102 @code{"mips3"}). It names the default architecture when no
17103 @option{-march} option is given.
17105 @item -mtune=@var{arch}
17107 Optimize for @var{arch}. Among other things, this option controls
17108 the way instructions are scheduled, and the perceived cost of arithmetic
17109 operations. The list of @var{arch} values is the same as for
17112 When this option is not used, GCC optimizes for the processor
17113 specified by @option{-march}. By using @option{-march} and
17114 @option{-mtune} together, it is possible to generate code that
17115 runs on a family of processors, but optimize the code for one
17116 particular member of that family.
17118 @option{-mtune} defines the macros @code{_MIPS_TUNE} and
17119 @code{_MIPS_TUNE_@var{foo}}, which work in the same way as the
17120 @option{-march} ones described above.
17124 Equivalent to @option{-march=mips1}.
17128 Equivalent to @option{-march=mips2}.
17132 Equivalent to @option{-march=mips3}.
17136 Equivalent to @option{-march=mips4}.
17140 Equivalent to @option{-march=mips32}.
17144 Equivalent to @option{-march=mips32r3}.
17148 Equivalent to @option{-march=mips32r5}.
17152 Equivalent to @option{-march=mips32r6}.
17156 Equivalent to @option{-march=mips64}.
17160 Equivalent to @option{-march=mips64r2}.
17164 Equivalent to @option{-march=mips64r3}.
17168 Equivalent to @option{-march=mips64r5}.
17172 Equivalent to @option{-march=mips64r6}.
17177 @opindex mno-mips16
17178 Generate (do not generate) MIPS16 code. If GCC is targeting a
17179 MIPS32 or MIPS64 architecture, it makes use of the MIPS16e ASE@.
17181 MIPS16 code generation can also be controlled on a per-function basis
17182 by means of @code{mips16} and @code{nomips16} attributes.
17183 @xref{Function Attributes}, for more information.
17185 @item -mflip-mips16
17186 @opindex mflip-mips16
17187 Generate MIPS16 code on alternating functions. This option is provided
17188 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
17189 not intended for ordinary use in compiling user code.
17191 @item -minterlink-compressed
17192 @item -mno-interlink-compressed
17193 @opindex minterlink-compressed
17194 @opindex mno-interlink-compressed
17195 Require (do not require) that code using the standard (uncompressed) MIPS ISA
17196 be link-compatible with MIPS16 and microMIPS code, and vice versa.
17198 For example, code using the standard ISA encoding cannot jump directly
17199 to MIPS16 or microMIPS code; it must either use a call or an indirect jump.
17200 @option{-minterlink-compressed} therefore disables direct jumps unless GCC
17201 knows that the target of the jump is not compressed.
17203 @item -minterlink-mips16
17204 @itemx -mno-interlink-mips16
17205 @opindex minterlink-mips16
17206 @opindex mno-interlink-mips16
17207 Aliases of @option{-minterlink-compressed} and
17208 @option{-mno-interlink-compressed}. These options predate the microMIPS ASE
17209 and are retained for backwards compatibility.
17221 Generate code for the given ABI@.
17223 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
17224 generates 64-bit code when you select a 64-bit architecture, but you
17225 can use @option{-mgp32} to get 32-bit code instead.
17227 For information about the O64 ABI, see
17228 @uref{http://gcc.gnu.org/@/projects/@/mipso64-abi.html}.
17230 GCC supports a variant of the o32 ABI in which floating-point registers
17231 are 64 rather than 32 bits wide. You can select this combination with
17232 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @code{mthc1}
17233 and @code{mfhc1} instructions and is therefore only supported for
17234 MIPS32R2, MIPS32R3 and MIPS32R5 processors.
17236 The register assignments for arguments and return values remain the
17237 same, but each scalar value is passed in a single 64-bit register
17238 rather than a pair of 32-bit registers. For example, scalar
17239 floating-point values are returned in @samp{$f0} only, not a
17240 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
17241 remains the same in that the even-numbered double-precision registers
17244 Two additional variants of the o32 ABI are supported to enable
17245 a transition from 32-bit to 64-bit registers. These are FPXX
17246 (@option{-mfpxx}) and FP64A (@option{-mfp64} @option{-mno-odd-spreg}).
17247 The FPXX extension mandates that all code must execute correctly
17248 when run using 32-bit or 64-bit registers. The code can be interlinked
17249 with either FP32 or FP64, but not both.
17250 The FP64A extension is similar to the FP64 extension but forbids the
17251 use of odd-numbered single-precision registers. This can be used
17252 in conjunction with the @code{FRE} mode of FPUs in MIPS32R5
17253 processors and allows both FP32 and FP64A code to interlink and
17254 run in the same process without changing FPU modes.
17257 @itemx -mno-abicalls
17259 @opindex mno-abicalls
17260 Generate (do not generate) code that is suitable for SVR4-style
17261 dynamic objects. @option{-mabicalls} is the default for SVR4-based
17266 Generate (do not generate) code that is fully position-independent,
17267 and that can therefore be linked into shared libraries. This option
17268 only affects @option{-mabicalls}.
17270 All @option{-mabicalls} code has traditionally been position-independent,
17271 regardless of options like @option{-fPIC} and @option{-fpic}. However,
17272 as an extension, the GNU toolchain allows executables to use absolute
17273 accesses for locally-binding symbols. It can also use shorter GP
17274 initialization sequences and generate direct calls to locally-defined
17275 functions. This mode is selected by @option{-mno-shared}.
17277 @option{-mno-shared} depends on binutils 2.16 or higher and generates
17278 objects that can only be linked by the GNU linker. However, the option
17279 does not affect the ABI of the final executable; it only affects the ABI
17280 of relocatable objects. Using @option{-mno-shared} generally makes
17281 executables both smaller and quicker.
17283 @option{-mshared} is the default.
17289 Assume (do not assume) that the static and dynamic linkers
17290 support PLTs and copy relocations. This option only affects
17291 @option{-mno-shared -mabicalls}. For the n64 ABI, this option
17292 has no effect without @option{-msym32}.
17294 You can make @option{-mplt} the default by configuring
17295 GCC with @option{--with-mips-plt}. The default is
17296 @option{-mno-plt} otherwise.
17302 Lift (do not lift) the usual restrictions on the size of the global
17305 GCC normally uses a single instruction to load values from the GOT@.
17306 While this is relatively efficient, it only works if the GOT
17307 is smaller than about 64k. Anything larger causes the linker
17308 to report an error such as:
17310 @cindex relocation truncated to fit (MIPS)
17312 relocation truncated to fit: R_MIPS_GOT16 foobar
17315 If this happens, you should recompile your code with @option{-mxgot}.
17316 This works with very large GOTs, although the code is also
17317 less efficient, since it takes three instructions to fetch the
17318 value of a global symbol.
17320 Note that some linkers can create multiple GOTs. If you have such a
17321 linker, you should only need to use @option{-mxgot} when a single object
17322 file accesses more than 64k's worth of GOT entries. Very few do.
17324 These options have no effect unless GCC is generating position
17329 Assume that general-purpose registers are 32 bits wide.
17333 Assume that general-purpose registers are 64 bits wide.
17337 Assume that floating-point registers are 32 bits wide.
17341 Assume that floating-point registers are 64 bits wide.
17345 Do not assume the width of floating-point registers.
17348 @opindex mhard-float
17349 Use floating-point coprocessor instructions.
17352 @opindex msoft-float
17353 Do not use floating-point coprocessor instructions. Implement
17354 floating-point calculations using library calls instead.
17358 Equivalent to @option{-msoft-float}, but additionally asserts that the
17359 program being compiled does not perform any floating-point operations.
17360 This option is presently supported only by some bare-metal MIPS
17361 configurations, where it may select a special set of libraries
17362 that lack all floating-point support (including, for example, the
17363 floating-point @code{printf} formats).
17364 If code compiled with @option{-mno-float} accidentally contains
17365 floating-point operations, it is likely to suffer a link-time
17366 or run-time failure.
17368 @item -msingle-float
17369 @opindex msingle-float
17370 Assume that the floating-point coprocessor only supports single-precision
17373 @item -mdouble-float
17374 @opindex mdouble-float
17375 Assume that the floating-point coprocessor supports double-precision
17376 operations. This is the default.
17379 @itemx -mno-odd-spreg
17380 @opindex modd-spreg
17381 @opindex mno-odd-spreg
17382 Enable the use of odd-numbered single-precision floating-point registers
17383 for the o32 ABI. This is the default for processors that are known to
17384 support these registers. When using the o32 FPXX ABI, @option{-mno-odd-spreg}
17388 @itemx -mabs=legacy
17390 @opindex mabs=legacy
17391 These options control the treatment of the special not-a-number (NaN)
17392 IEEE 754 floating-point data with the @code{abs.@i{fmt}} and
17393 @code{neg.@i{fmt}} machine instructions.
17395 By default or when @option{-mabs=legacy} is used the legacy
17396 treatment is selected. In this case these instructions are considered
17397 arithmetic and avoided where correct operation is required and the
17398 input operand might be a NaN. A longer sequence of instructions that
17399 manipulate the sign bit of floating-point datum manually is used
17400 instead unless the @option{-ffinite-math-only} option has also been
17403 The @option{-mabs=2008} option selects the IEEE 754-2008 treatment. In
17404 this case these instructions are considered non-arithmetic and therefore
17405 operating correctly in all cases, including in particular where the
17406 input operand is a NaN. These instructions are therefore always used
17407 for the respective operations.
17410 @itemx -mnan=legacy
17412 @opindex mnan=legacy
17413 These options control the encoding of the special not-a-number (NaN)
17414 IEEE 754 floating-point data.
17416 The @option{-mnan=legacy} option selects the legacy encoding. In this
17417 case quiet NaNs (qNaNs) are denoted by the first bit of their trailing
17418 significand field being 0, whereas signalling NaNs (sNaNs) are denoted
17419 by the first bit of their trailing significand field being 1.
17421 The @option{-mnan=2008} option selects the IEEE 754-2008 encoding. In
17422 this case qNaNs are denoted by the first bit of their trailing
17423 significand field being 1, whereas sNaNs are denoted by the first bit of
17424 their trailing significand field being 0.
17426 The default is @option{-mnan=legacy} unless GCC has been configured with
17427 @option{--with-nan=2008}.
17433 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
17434 implement atomic memory built-in functions. When neither option is
17435 specified, GCC uses the instructions if the target architecture
17438 @option{-mllsc} is useful if the runtime environment can emulate the
17439 instructions and @option{-mno-llsc} can be useful when compiling for
17440 nonstandard ISAs. You can make either option the default by
17441 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
17442 respectively. @option{--with-llsc} is the default for some
17443 configurations; see the installation documentation for details.
17449 Use (do not use) revision 1 of the MIPS DSP ASE@.
17450 @xref{MIPS DSP Built-in Functions}. This option defines the
17451 preprocessor macro @code{__mips_dsp}. It also defines
17452 @code{__mips_dsp_rev} to 1.
17458 Use (do not use) revision 2 of the MIPS DSP ASE@.
17459 @xref{MIPS DSP Built-in Functions}. This option defines the
17460 preprocessor macros @code{__mips_dsp} and @code{__mips_dspr2}.
17461 It also defines @code{__mips_dsp_rev} to 2.
17464 @itemx -mno-smartmips
17465 @opindex msmartmips
17466 @opindex mno-smartmips
17467 Use (do not use) the MIPS SmartMIPS ASE.
17469 @item -mpaired-single
17470 @itemx -mno-paired-single
17471 @opindex mpaired-single
17472 @opindex mno-paired-single
17473 Use (do not use) paired-single floating-point instructions.
17474 @xref{MIPS Paired-Single Support}. This option requires
17475 hardware floating-point support to be enabled.
17481 Use (do not use) MIPS Digital Media Extension instructions.
17482 This option can only be used when generating 64-bit code and requires
17483 hardware floating-point support to be enabled.
17488 @opindex mno-mips3d
17489 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
17490 The option @option{-mips3d} implies @option{-mpaired-single}.
17493 @itemx -mno-micromips
17494 @opindex mmicromips
17495 @opindex mno-mmicromips
17496 Generate (do not generate) microMIPS code.
17498 MicroMIPS code generation can also be controlled on a per-function basis
17499 by means of @code{micromips} and @code{nomicromips} attributes.
17500 @xref{Function Attributes}, for more information.
17506 Use (do not use) MT Multithreading instructions.
17512 Use (do not use) the MIPS MCU ASE instructions.
17518 Use (do not use) the MIPS Enhanced Virtual Addressing instructions.
17524 Use (do not use) the MIPS Virtualization Application Specific instructions.
17530 Use (do not use) the MIPS eXtended Physical Address (XPA) instructions.
17534 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
17535 an explanation of the default and the way that the pointer size is
17540 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
17542 The default size of @code{int}s, @code{long}s and pointers depends on
17543 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
17544 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
17545 32-bit @code{long}s. Pointers are the same size as @code{long}s,
17546 or the same size as integer registers, whichever is smaller.
17552 Assume (do not assume) that all symbols have 32-bit values, regardless
17553 of the selected ABI@. This option is useful in combination with
17554 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
17555 to generate shorter and faster references to symbolic addresses.
17559 Put definitions of externally-visible data in a small data section
17560 if that data is no bigger than @var{num} bytes. GCC can then generate
17561 more efficient accesses to the data; see @option{-mgpopt} for details.
17563 The default @option{-G} option depends on the configuration.
17565 @item -mlocal-sdata
17566 @itemx -mno-local-sdata
17567 @opindex mlocal-sdata
17568 @opindex mno-local-sdata
17569 Extend (do not extend) the @option{-G} behavior to local data too,
17570 such as to static variables in C@. @option{-mlocal-sdata} is the
17571 default for all configurations.
17573 If the linker complains that an application is using too much small data,
17574 you might want to try rebuilding the less performance-critical parts with
17575 @option{-mno-local-sdata}. You might also want to build large
17576 libraries with @option{-mno-local-sdata}, so that the libraries leave
17577 more room for the main program.
17579 @item -mextern-sdata
17580 @itemx -mno-extern-sdata
17581 @opindex mextern-sdata
17582 @opindex mno-extern-sdata
17583 Assume (do not assume) that externally-defined data is in
17584 a small data section if the size of that data is within the @option{-G} limit.
17585 @option{-mextern-sdata} is the default for all configurations.
17587 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
17588 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
17589 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
17590 is placed in a small data section. If @var{Var} is defined by another
17591 module, you must either compile that module with a high-enough
17592 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
17593 definition. If @var{Var} is common, you must link the application
17594 with a high-enough @option{-G} setting.
17596 The easiest way of satisfying these restrictions is to compile
17597 and link every module with the same @option{-G} option. However,
17598 you may wish to build a library that supports several different
17599 small data limits. You can do this by compiling the library with
17600 the highest supported @option{-G} setting and additionally using
17601 @option{-mno-extern-sdata} to stop the library from making assumptions
17602 about externally-defined data.
17608 Use (do not use) GP-relative accesses for symbols that are known to be
17609 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
17610 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
17613 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
17614 might not hold the value of @code{_gp}. For example, if the code is
17615 part of a library that might be used in a boot monitor, programs that
17616 call boot monitor routines pass an unknown value in @code{$gp}.
17617 (In such situations, the boot monitor itself is usually compiled
17618 with @option{-G0}.)
17620 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
17621 @option{-mno-extern-sdata}.
17623 @item -membedded-data
17624 @itemx -mno-embedded-data
17625 @opindex membedded-data
17626 @opindex mno-embedded-data
17627 Allocate variables to the read-only data section first if possible, then
17628 next in the small data section if possible, otherwise in data. This gives
17629 slightly slower code than the default, but reduces the amount of RAM required
17630 when executing, and thus may be preferred for some embedded systems.
17632 @item -muninit-const-in-rodata
17633 @itemx -mno-uninit-const-in-rodata
17634 @opindex muninit-const-in-rodata
17635 @opindex mno-uninit-const-in-rodata
17636 Put uninitialized @code{const} variables in the read-only data section.
17637 This option is only meaningful in conjunction with @option{-membedded-data}.
17639 @item -mcode-readable=@var{setting}
17640 @opindex mcode-readable
17641 Specify whether GCC may generate code that reads from executable sections.
17642 There are three possible settings:
17645 @item -mcode-readable=yes
17646 Instructions may freely access executable sections. This is the
17649 @item -mcode-readable=pcrel
17650 MIPS16 PC-relative load instructions can access executable sections,
17651 but other instructions must not do so. This option is useful on 4KSc
17652 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
17653 It is also useful on processors that can be configured to have a dual
17654 instruction/data SRAM interface and that, like the M4K, automatically
17655 redirect PC-relative loads to the instruction RAM.
17657 @item -mcode-readable=no
17658 Instructions must not access executable sections. This option can be
17659 useful on targets that are configured to have a dual instruction/data
17660 SRAM interface but that (unlike the M4K) do not automatically redirect
17661 PC-relative loads to the instruction RAM.
17664 @item -msplit-addresses
17665 @itemx -mno-split-addresses
17666 @opindex msplit-addresses
17667 @opindex mno-split-addresses
17668 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
17669 relocation operators. This option has been superseded by
17670 @option{-mexplicit-relocs} but is retained for backwards compatibility.
17672 @item -mexplicit-relocs
17673 @itemx -mno-explicit-relocs
17674 @opindex mexplicit-relocs
17675 @opindex mno-explicit-relocs
17676 Use (do not use) assembler relocation operators when dealing with symbolic
17677 addresses. The alternative, selected by @option{-mno-explicit-relocs},
17678 is to use assembler macros instead.
17680 @option{-mexplicit-relocs} is the default if GCC was configured
17681 to use an assembler that supports relocation operators.
17683 @item -mcheck-zero-division
17684 @itemx -mno-check-zero-division
17685 @opindex mcheck-zero-division
17686 @opindex mno-check-zero-division
17687 Trap (do not trap) on integer division by zero.
17689 The default is @option{-mcheck-zero-division}.
17691 @item -mdivide-traps
17692 @itemx -mdivide-breaks
17693 @opindex mdivide-traps
17694 @opindex mdivide-breaks
17695 MIPS systems check for division by zero by generating either a
17696 conditional trap or a break instruction. Using traps results in
17697 smaller code, but is only supported on MIPS II and later. Also, some
17698 versions of the Linux kernel have a bug that prevents trap from
17699 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
17700 allow conditional traps on architectures that support them and
17701 @option{-mdivide-breaks} to force the use of breaks.
17703 The default is usually @option{-mdivide-traps}, but this can be
17704 overridden at configure time using @option{--with-divide=breaks}.
17705 Divide-by-zero checks can be completely disabled using
17706 @option{-mno-check-zero-division}.
17711 @opindex mno-memcpy
17712 Force (do not force) the use of @code{memcpy} for non-trivial block
17713 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
17714 most constant-sized copies.
17717 @itemx -mno-long-calls
17718 @opindex mlong-calls
17719 @opindex mno-long-calls
17720 Disable (do not disable) use of the @code{jal} instruction. Calling
17721 functions using @code{jal} is more efficient but requires the caller
17722 and callee to be in the same 256 megabyte segment.
17724 This option has no effect on abicalls code. The default is
17725 @option{-mno-long-calls}.
17731 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
17732 instructions, as provided by the R4650 ISA@.
17738 Enable (disable) use of the @code{madd} and @code{msub} integer
17739 instructions. The default is @option{-mimadd} on architectures
17740 that support @code{madd} and @code{msub} except for the 74k
17741 architecture where it was found to generate slower code.
17744 @itemx -mno-fused-madd
17745 @opindex mfused-madd
17746 @opindex mno-fused-madd
17747 Enable (disable) use of the floating-point multiply-accumulate
17748 instructions, when they are available. The default is
17749 @option{-mfused-madd}.
17751 On the R8000 CPU when multiply-accumulate instructions are used,
17752 the intermediate product is calculated to infinite precision
17753 and is not subject to the FCSR Flush to Zero bit. This may be
17754 undesirable in some circumstances. On other processors the result
17755 is numerically identical to the equivalent computation using
17756 separate multiply, add, subtract and negate instructions.
17760 Tell the MIPS assembler to not run its preprocessor over user
17761 assembler files (with a @samp{.s} suffix) when assembling them.
17766 @opindex mno-fix-24k
17767 Work around the 24K E48 (lost data on stores during refill) errata.
17768 The workarounds are implemented by the assembler rather than by GCC@.
17771 @itemx -mno-fix-r4000
17772 @opindex mfix-r4000
17773 @opindex mno-fix-r4000
17774 Work around certain R4000 CPU errata:
17777 A double-word or a variable shift may give an incorrect result if executed
17778 immediately after starting an integer division.
17780 A double-word or a variable shift may give an incorrect result if executed
17781 while an integer multiplication is in progress.
17783 An integer division may give an incorrect result if started in a delay slot
17784 of a taken branch or a jump.
17788 @itemx -mno-fix-r4400
17789 @opindex mfix-r4400
17790 @opindex mno-fix-r4400
17791 Work around certain R4400 CPU errata:
17794 A double-word or a variable shift may give an incorrect result if executed
17795 immediately after starting an integer division.
17799 @itemx -mno-fix-r10000
17800 @opindex mfix-r10000
17801 @opindex mno-fix-r10000
17802 Work around certain R10000 errata:
17805 @code{ll}/@code{sc} sequences may not behave atomically on revisions
17806 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
17809 This option can only be used if the target architecture supports
17810 branch-likely instructions. @option{-mfix-r10000} is the default when
17811 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
17815 @itemx -mno-fix-rm7000
17816 @opindex mfix-rm7000
17817 Work around the RM7000 @code{dmult}/@code{dmultu} errata. The
17818 workarounds are implemented by the assembler rather than by GCC@.
17821 @itemx -mno-fix-vr4120
17822 @opindex mfix-vr4120
17823 Work around certain VR4120 errata:
17826 @code{dmultu} does not always produce the correct result.
17828 @code{div} and @code{ddiv} do not always produce the correct result if one
17829 of the operands is negative.
17831 The workarounds for the division errata rely on special functions in
17832 @file{libgcc.a}. At present, these functions are only provided by
17833 the @code{mips64vr*-elf} configurations.
17835 Other VR4120 errata require a NOP to be inserted between certain pairs of
17836 instructions. These errata are handled by the assembler, not by GCC itself.
17839 @opindex mfix-vr4130
17840 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
17841 workarounds are implemented by the assembler rather than by GCC,
17842 although GCC avoids using @code{mflo} and @code{mfhi} if the
17843 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
17844 instructions are available instead.
17847 @itemx -mno-fix-sb1
17849 Work around certain SB-1 CPU core errata.
17850 (This flag currently works around the SB-1 revision 2
17851 ``F1'' and ``F2'' floating-point errata.)
17853 @item -mr10k-cache-barrier=@var{setting}
17854 @opindex mr10k-cache-barrier
17855 Specify whether GCC should insert cache barriers to avoid the
17856 side-effects of speculation on R10K processors.
17858 In common with many processors, the R10K tries to predict the outcome
17859 of a conditional branch and speculatively executes instructions from
17860 the ``taken'' branch. It later aborts these instructions if the
17861 predicted outcome is wrong. However, on the R10K, even aborted
17862 instructions can have side effects.
17864 This problem only affects kernel stores and, depending on the system,
17865 kernel loads. As an example, a speculatively-executed store may load
17866 the target memory into cache and mark the cache line as dirty, even if
17867 the store itself is later aborted. If a DMA operation writes to the
17868 same area of memory before the ``dirty'' line is flushed, the cached
17869 data overwrites the DMA-ed data. See the R10K processor manual
17870 for a full description, including other potential problems.
17872 One workaround is to insert cache barrier instructions before every memory
17873 access that might be speculatively executed and that might have side
17874 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
17875 controls GCC's implementation of this workaround. It assumes that
17876 aborted accesses to any byte in the following regions does not have
17881 the memory occupied by the current function's stack frame;
17884 the memory occupied by an incoming stack argument;
17887 the memory occupied by an object with a link-time-constant address.
17890 It is the kernel's responsibility to ensure that speculative
17891 accesses to these regions are indeed safe.
17893 If the input program contains a function declaration such as:
17899 then the implementation of @code{foo} must allow @code{j foo} and
17900 @code{jal foo} to be executed speculatively. GCC honors this
17901 restriction for functions it compiles itself. It expects non-GCC
17902 functions (such as hand-written assembly code) to do the same.
17904 The option has three forms:
17907 @item -mr10k-cache-barrier=load-store
17908 Insert a cache barrier before a load or store that might be
17909 speculatively executed and that might have side effects even
17912 @item -mr10k-cache-barrier=store
17913 Insert a cache barrier before a store that might be speculatively
17914 executed and that might have side effects even if aborted.
17916 @item -mr10k-cache-barrier=none
17917 Disable the insertion of cache barriers. This is the default setting.
17920 @item -mflush-func=@var{func}
17921 @itemx -mno-flush-func
17922 @opindex mflush-func
17923 Specifies the function to call to flush the I and D caches, or to not
17924 call any such function. If called, the function must take the same
17925 arguments as the common @code{_flush_func}, that is, the address of the
17926 memory range for which the cache is being flushed, the size of the
17927 memory range, and the number 3 (to flush both caches). The default
17928 depends on the target GCC was configured for, but commonly is either
17929 @code{_flush_func} or @code{__cpu_flush}.
17931 @item mbranch-cost=@var{num}
17932 @opindex mbranch-cost
17933 Set the cost of branches to roughly @var{num} ``simple'' instructions.
17934 This cost is only a heuristic and is not guaranteed to produce
17935 consistent results across releases. A zero cost redundantly selects
17936 the default, which is based on the @option{-mtune} setting.
17938 @item -mbranch-likely
17939 @itemx -mno-branch-likely
17940 @opindex mbranch-likely
17941 @opindex mno-branch-likely
17942 Enable or disable use of Branch Likely instructions, regardless of the
17943 default for the selected architecture. By default, Branch Likely
17944 instructions may be generated if they are supported by the selected
17945 architecture. An exception is for the MIPS32 and MIPS64 architectures
17946 and processors that implement those architectures; for those, Branch
17947 Likely instructions are not be generated by default because the MIPS32
17948 and MIPS64 architectures specifically deprecate their use.
17950 @item -mfp-exceptions
17951 @itemx -mno-fp-exceptions
17952 @opindex mfp-exceptions
17953 Specifies whether FP exceptions are enabled. This affects how
17954 FP instructions are scheduled for some processors.
17955 The default is that FP exceptions are
17958 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
17959 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
17962 @item -mvr4130-align
17963 @itemx -mno-vr4130-align
17964 @opindex mvr4130-align
17965 The VR4130 pipeline is two-way superscalar, but can only issue two
17966 instructions together if the first one is 8-byte aligned. When this
17967 option is enabled, GCC aligns pairs of instructions that it
17968 thinks should execute in parallel.
17970 This option only has an effect when optimizing for the VR4130.
17971 It normally makes code faster, but at the expense of making it bigger.
17972 It is enabled by default at optimization level @option{-O3}.
17977 Enable (disable) generation of @code{synci} instructions on
17978 architectures that support it. The @code{synci} instructions (if
17979 enabled) are generated when @code{__builtin___clear_cache} is
17982 This option defaults to @option{-mno-synci}, but the default can be
17983 overridden by configuring GCC with @option{--with-synci}.
17985 When compiling code for single processor systems, it is generally safe
17986 to use @code{synci}. However, on many multi-core (SMP) systems, it
17987 does not invalidate the instruction caches on all cores and may lead
17988 to undefined behavior.
17990 @item -mrelax-pic-calls
17991 @itemx -mno-relax-pic-calls
17992 @opindex mrelax-pic-calls
17993 Try to turn PIC calls that are normally dispatched via register
17994 @code{$25} into direct calls. This is only possible if the linker can
17995 resolve the destination at link-time and if the destination is within
17996 range for a direct call.
17998 @option{-mrelax-pic-calls} is the default if GCC was configured to use
17999 an assembler and a linker that support the @code{.reloc} assembly
18000 directive and @option{-mexplicit-relocs} is in effect. With
18001 @option{-mno-explicit-relocs}, this optimization can be performed by the
18002 assembler and the linker alone without help from the compiler.
18004 @item -mmcount-ra-address
18005 @itemx -mno-mcount-ra-address
18006 @opindex mmcount-ra-address
18007 @opindex mno-mcount-ra-address
18008 Emit (do not emit) code that allows @code{_mcount} to modify the
18009 calling function's return address. When enabled, this option extends
18010 the usual @code{_mcount} interface with a new @var{ra-address}
18011 parameter, which has type @code{intptr_t *} and is passed in register
18012 @code{$12}. @code{_mcount} can then modify the return address by
18013 doing both of the following:
18016 Returning the new address in register @code{$31}.
18018 Storing the new address in @code{*@var{ra-address}},
18019 if @var{ra-address} is nonnull.
18022 The default is @option{-mno-mcount-ra-address}.
18027 @subsection MMIX Options
18028 @cindex MMIX Options
18030 These options are defined for the MMIX:
18034 @itemx -mno-libfuncs
18036 @opindex mno-libfuncs
18037 Specify that intrinsic library functions are being compiled, passing all
18038 values in registers, no matter the size.
18041 @itemx -mno-epsilon
18043 @opindex mno-epsilon
18044 Generate floating-point comparison instructions that compare with respect
18045 to the @code{rE} epsilon register.
18047 @item -mabi=mmixware
18049 @opindex mabi=mmixware
18051 Generate code that passes function parameters and return values that (in
18052 the called function) are seen as registers @code{$0} and up, as opposed to
18053 the GNU ABI which uses global registers @code{$231} and up.
18055 @item -mzero-extend
18056 @itemx -mno-zero-extend
18057 @opindex mzero-extend
18058 @opindex mno-zero-extend
18059 When reading data from memory in sizes shorter than 64 bits, use (do not
18060 use) zero-extending load instructions by default, rather than
18061 sign-extending ones.
18064 @itemx -mno-knuthdiv
18066 @opindex mno-knuthdiv
18067 Make the result of a division yielding a remainder have the same sign as
18068 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
18069 remainder follows the sign of the dividend. Both methods are
18070 arithmetically valid, the latter being almost exclusively used.
18072 @item -mtoplevel-symbols
18073 @itemx -mno-toplevel-symbols
18074 @opindex mtoplevel-symbols
18075 @opindex mno-toplevel-symbols
18076 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
18077 code can be used with the @code{PREFIX} assembly directive.
18081 Generate an executable in the ELF format, rather than the default
18082 @samp{mmo} format used by the @command{mmix} simulator.
18084 @item -mbranch-predict
18085 @itemx -mno-branch-predict
18086 @opindex mbranch-predict
18087 @opindex mno-branch-predict
18088 Use (do not use) the probable-branch instructions, when static branch
18089 prediction indicates a probable branch.
18091 @item -mbase-addresses
18092 @itemx -mno-base-addresses
18093 @opindex mbase-addresses
18094 @opindex mno-base-addresses
18095 Generate (do not generate) code that uses @emph{base addresses}. Using a
18096 base address automatically generates a request (handled by the assembler
18097 and the linker) for a constant to be set up in a global register. The
18098 register is used for one or more base address requests within the range 0
18099 to 255 from the value held in the register. The generally leads to short
18100 and fast code, but the number of different data items that can be
18101 addressed is limited. This means that a program that uses lots of static
18102 data may require @option{-mno-base-addresses}.
18104 @item -msingle-exit
18105 @itemx -mno-single-exit
18106 @opindex msingle-exit
18107 @opindex mno-single-exit
18108 Force (do not force) generated code to have a single exit point in each
18112 @node MN10300 Options
18113 @subsection MN10300 Options
18114 @cindex MN10300 options
18116 These @option{-m} options are defined for Matsushita MN10300 architectures:
18121 Generate code to avoid bugs in the multiply instructions for the MN10300
18122 processors. This is the default.
18124 @item -mno-mult-bug
18125 @opindex mno-mult-bug
18126 Do not generate code to avoid bugs in the multiply instructions for the
18127 MN10300 processors.
18131 Generate code using features specific to the AM33 processor.
18135 Do not generate code using features specific to the AM33 processor. This
18140 Generate code using features specific to the AM33/2.0 processor.
18144 Generate code using features specific to the AM34 processor.
18146 @item -mtune=@var{cpu-type}
18148 Use the timing characteristics of the indicated CPU type when
18149 scheduling instructions. This does not change the targeted processor
18150 type. The CPU type must be one of @samp{mn10300}, @samp{am33},
18151 @samp{am33-2} or @samp{am34}.
18153 @item -mreturn-pointer-on-d0
18154 @opindex mreturn-pointer-on-d0
18155 When generating a function that returns a pointer, return the pointer
18156 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
18157 only in @code{a0}, and attempts to call such functions without a prototype
18158 result in errors. Note that this option is on by default; use
18159 @option{-mno-return-pointer-on-d0} to disable it.
18163 Do not link in the C run-time initialization object file.
18167 Indicate to the linker that it should perform a relaxation optimization pass
18168 to shorten branches, calls and absolute memory addresses. This option only
18169 has an effect when used on the command line for the final link step.
18171 This option makes symbolic debugging impossible.
18175 Allow the compiler to generate @emph{Long Instruction Word}
18176 instructions if the target is the @samp{AM33} or later. This is the
18177 default. This option defines the preprocessor macro @code{__LIW__}.
18181 Do not allow the compiler to generate @emph{Long Instruction Word}
18182 instructions. This option defines the preprocessor macro
18187 Allow the compiler to generate the @emph{SETLB} and @emph{Lcc}
18188 instructions if the target is the @samp{AM33} or later. This is the
18189 default. This option defines the preprocessor macro @code{__SETLB__}.
18193 Do not allow the compiler to generate @emph{SETLB} or @emph{Lcc}
18194 instructions. This option defines the preprocessor macro
18195 @code{__NO_SETLB__}.
18199 @node Moxie Options
18200 @subsection Moxie Options
18201 @cindex Moxie Options
18207 Generate big-endian code. This is the default for @samp{moxie-*-*}
18212 Generate little-endian code.
18216 Generate mul.x and umul.x instructions. This is the default for
18217 @samp{moxiebox-*-*} configurations.
18221 Do not link in the C run-time initialization object file.
18225 @node MSP430 Options
18226 @subsection MSP430 Options
18227 @cindex MSP430 Options
18229 These options are defined for the MSP430:
18235 Force assembly output to always use hex constants. Normally such
18236 constants are signed decimals, but this option is available for
18237 testsuite and/or aesthetic purposes.
18241 Select the MCU to target. This is used to create a C preprocessor
18242 symbol based upon the MCU name, converted to upper case and pre- and
18243 post-fixed with @samp{__}. This in turn is used by the
18244 @file{msp430.h} header file to select an MCU-specific supplementary
18247 The option also sets the ISA to use. If the MCU name is one that is
18248 known to only support the 430 ISA then that is selected, otherwise the
18249 430X ISA is selected. A generic MCU name of @samp{msp430} can also be
18250 used to select the 430 ISA. Similarly the generic @samp{msp430x} MCU
18251 name selects the 430X ISA.
18253 In addition an MCU-specific linker script is added to the linker
18254 command line. The script's name is the name of the MCU with
18255 @file{.ld} appended. Thus specifying @option{-mmcu=xxx} on the @command{gcc}
18256 command line defines the C preprocessor symbol @code{__XXX__} and
18257 cause the linker to search for a script called @file{xxx.ld}.
18259 This option is also passed on to the assembler.
18263 Specifies the ISA to use. Accepted values are @samp{msp430},
18264 @samp{msp430x} and @samp{msp430xv2}. This option is deprecated. The
18265 @option{-mmcu=} option should be used to select the ISA.
18269 Link to the simulator runtime libraries and linker script. Overrides
18270 any scripts that would be selected by the @option{-mmcu=} option.
18274 Use large-model addressing (20-bit pointers, 32-bit @code{size_t}).
18278 Use small-model addressing (16-bit pointers, 16-bit @code{size_t}).
18282 This option is passed to the assembler and linker, and allows the
18283 linker to perform certain optimizations that cannot be done until
18288 Describes the type of hardware multiply supported by the target.
18289 Accepted values are @samp{none} for no hardware multiply, @samp{16bit}
18290 for the original 16-bit-only multiply supported by early MCUs.
18291 @samp{32bit} for the 16/32-bit multiply supported by later MCUs and
18292 @samp{f5series} for the 16/32-bit multiply supported by F5-series MCUs.
18293 A value of @samp{auto} can also be given. This tells GCC to deduce
18294 the hardware multiply support based upon the MCU name provided by the
18295 @option{-mmcu} option. If no @option{-mmcu} option is specified then
18296 @samp{32bit} hardware multiply support is assumed. @samp{auto} is the
18299 Hardware multiplies are normally performed by calling a library
18300 routine. This saves space in the generated code. When compiling at
18301 @option{-O3} or higher however the hardware multiplier is invoked
18302 inline. This makes for bigger, but faster code.
18304 The hardware multiply routines disable interrupts whilst running and
18305 restore the previous interrupt state when they finish. This makes
18306 them safe to use inside interrupt handlers as well as in normal code.
18310 Enable the use of a minimum runtime environment - no static
18311 initializers or constructors. This is intended for memory-constrained
18312 devices. The compiler includes special symbols in some objects
18313 that tell the linker and runtime which code fragments are required.
18315 @item -mcode-region=
18316 @itemx -mdata-region=
18317 @opindex mcode-region
18318 @opindex mdata-region
18319 These options tell the compiler where to place functions and data that
18320 do not have one of the @code{lower}, @code{upper}, @code{either} or
18321 @code{section} attributes. Possible values are @code{lower},
18322 @code{upper}, @code{either} or @code{any}. The first three behave
18323 like the corresponding attribute. The fourth possible value -
18324 @code{any} - is the default. It leaves placement entirely up to the
18325 linker script and how it assigns the standard sections (.text, .data
18326 etc) to the memory regions.
18330 @node NDS32 Options
18331 @subsection NDS32 Options
18332 @cindex NDS32 Options
18334 These options are defined for NDS32 implementations:
18339 @opindex mbig-endian
18340 Generate code in big-endian mode.
18342 @item -mlittle-endian
18343 @opindex mlittle-endian
18344 Generate code in little-endian mode.
18346 @item -mreduced-regs
18347 @opindex mreduced-regs
18348 Use reduced-set registers for register allocation.
18351 @opindex mfull-regs
18352 Use full-set registers for register allocation.
18356 Generate conditional move instructions.
18360 Do not generate conditional move instructions.
18364 Generate performance extension instructions.
18366 @item -mno-perf-ext
18367 @opindex mno-perf-ext
18368 Do not generate performance extension instructions.
18372 Generate v3 push25/pop25 instructions.
18375 @opindex mno-v3push
18376 Do not generate v3 push25/pop25 instructions.
18380 Generate 16-bit instructions.
18383 @opindex mno-16-bit
18384 Do not generate 16-bit instructions.
18386 @item -misr-vector-size=@var{num}
18387 @opindex misr-vector-size
18388 Specify the size of each interrupt vector, which must be 4 or 16.
18390 @item -mcache-block-size=@var{num}
18391 @opindex mcache-block-size
18392 Specify the size of each cache block,
18393 which must be a power of 2 between 4 and 512.
18395 @item -march=@var{arch}
18397 Specify the name of the target architecture.
18399 @item -mcmodel=@var{code-model}
18401 Set the code model to one of
18404 All the data and read-only data segments must be within 512KB addressing space.
18405 The text segment must be within 16MB addressing space.
18406 @item @samp{medium}
18407 The data segment must be within 512KB while the read-only data segment can be
18408 within 4GB addressing space. The text segment should be still within 16MB
18411 All the text and data segments can be within 4GB addressing space.
18415 @opindex mctor-dtor
18416 Enable constructor/destructor feature.
18420 Guide linker to relax instructions.
18424 @node Nios II Options
18425 @subsection Nios II Options
18426 @cindex Nios II options
18427 @cindex Altera Nios II options
18429 These are the options defined for the Altera Nios II processor.
18435 @cindex smaller data references
18436 Put global and static objects less than or equal to @var{num} bytes
18437 into the small data or BSS sections instead of the normal data or BSS
18438 sections. The default value of @var{num} is 8.
18440 @item -mgpopt=@var{option}
18445 Generate (do not generate) GP-relative accesses. The following
18446 @var{option} names are recognized:
18451 Do not generate GP-relative accesses.
18454 Generate GP-relative accesses for small data objects that are not
18455 external or weak. Also use GP-relative addressing for objects that
18456 have been explicitly placed in a small data section via a @code{section}
18460 As for @samp{local}, but also generate GP-relative accesses for
18461 small data objects that are external or weak. If you use this option,
18462 you must ensure that all parts of your program (including libraries) are
18463 compiled with the same @option{-G} setting.
18466 Generate GP-relative accesses for all data objects in the program. If you
18467 use this option, the entire data and BSS segments
18468 of your program must fit in 64K of memory and you must use an appropriate
18469 linker script to allocate them within the addressible range of the
18473 Generate GP-relative addresses for function pointers as well as data
18474 pointers. If you use this option, the entire text, data, and BSS segments
18475 of your program must fit in 64K of memory and you must use an appropriate
18476 linker script to allocate them within the addressible range of the
18481 @option{-mgpopt} is equivalent to @option{-mgpopt=local}, and
18482 @option{-mno-gpopt} is equivalent to @option{-mgpopt=none}.
18484 The default is @option{-mgpopt} except when @option{-fpic} or
18485 @option{-fPIC} is specified to generate position-independent code.
18486 Note that the Nios II ABI does not permit GP-relative accesses from
18489 You may need to specify @option{-mno-gpopt} explicitly when building
18490 programs that include large amounts of small data, including large
18491 GOT data sections. In this case, the 16-bit offset for GP-relative
18492 addressing may not be large enough to allow access to the entire
18493 small data section.
18499 Generate little-endian (default) or big-endian (experimental) code,
18502 @item -mbypass-cache
18503 @itemx -mno-bypass-cache
18504 @opindex mno-bypass-cache
18505 @opindex mbypass-cache
18506 Force all load and store instructions to always bypass cache by
18507 using I/O variants of the instructions. The default is not to
18510 @item -mno-cache-volatile
18511 @itemx -mcache-volatile
18512 @opindex mcache-volatile
18513 @opindex mno-cache-volatile
18514 Volatile memory access bypass the cache using the I/O variants of
18515 the load and store instructions. The default is not to bypass the cache.
18517 @item -mno-fast-sw-div
18518 @itemx -mfast-sw-div
18519 @opindex mno-fast-sw-div
18520 @opindex mfast-sw-div
18521 Do not use table-based fast divide for small numbers. The default
18522 is to use the fast divide at @option{-O3} and above.
18526 @itemx -mno-hw-mulx
18530 @opindex mno-hw-mul
18532 @opindex mno-hw-mulx
18534 @opindex mno-hw-div
18536 Enable or disable emitting @code{mul}, @code{mulx} and @code{div} family of
18537 instructions by the compiler. The default is to emit @code{mul}
18538 and not emit @code{div} and @code{mulx}.
18540 @item -mcustom-@var{insn}=@var{N}
18541 @itemx -mno-custom-@var{insn}
18542 @opindex mcustom-@var{insn}
18543 @opindex mno-custom-@var{insn}
18544 Each @option{-mcustom-@var{insn}=@var{N}} option enables use of a
18545 custom instruction with encoding @var{N} when generating code that uses
18546 @var{insn}. For example, @option{-mcustom-fadds=253} generates custom
18547 instruction 253 for single-precision floating-point add operations instead
18548 of the default behavior of using a library call.
18550 The following values of @var{insn} are supported. Except as otherwise
18551 noted, floating-point operations are expected to be implemented with
18552 normal IEEE 754 semantics and correspond directly to the C operators or the
18553 equivalent GCC built-in functions (@pxref{Other Builtins}).
18555 Single-precision floating point:
18558 @item @samp{fadds}, @samp{fsubs}, @samp{fdivs}, @samp{fmuls}
18559 Binary arithmetic operations.
18565 Unary absolute value.
18567 @item @samp{fcmpeqs}, @samp{fcmpges}, @samp{fcmpgts}, @samp{fcmples}, @samp{fcmplts}, @samp{fcmpnes}
18568 Comparison operations.
18570 @item @samp{fmins}, @samp{fmaxs}
18571 Floating-point minimum and maximum. These instructions are only
18572 generated if @option{-ffinite-math-only} is specified.
18574 @item @samp{fsqrts}
18575 Unary square root operation.
18577 @item @samp{fcoss}, @samp{fsins}, @samp{ftans}, @samp{fatans}, @samp{fexps}, @samp{flogs}
18578 Floating-point trigonometric and exponential functions. These instructions
18579 are only generated if @option{-funsafe-math-optimizations} is also specified.
18583 Double-precision floating point:
18586 @item @samp{faddd}, @samp{fsubd}, @samp{fdivd}, @samp{fmuld}
18587 Binary arithmetic operations.
18593 Unary absolute value.
18595 @item @samp{fcmpeqd}, @samp{fcmpged}, @samp{fcmpgtd}, @samp{fcmpled}, @samp{fcmpltd}, @samp{fcmpned}
18596 Comparison operations.
18598 @item @samp{fmind}, @samp{fmaxd}
18599 Double-precision minimum and maximum. These instructions are only
18600 generated if @option{-ffinite-math-only} is specified.
18602 @item @samp{fsqrtd}
18603 Unary square root operation.
18605 @item @samp{fcosd}, @samp{fsind}, @samp{ftand}, @samp{fatand}, @samp{fexpd}, @samp{flogd}
18606 Double-precision trigonometric and exponential functions. These instructions
18607 are only generated if @option{-funsafe-math-optimizations} is also specified.
18613 @item @samp{fextsd}
18614 Conversion from single precision to double precision.
18616 @item @samp{ftruncds}
18617 Conversion from double precision to single precision.
18619 @item @samp{fixsi}, @samp{fixsu}, @samp{fixdi}, @samp{fixdu}
18620 Conversion from floating point to signed or unsigned integer types, with
18621 truncation towards zero.
18624 Conversion from single-precision floating point to signed integer,
18625 rounding to the nearest integer and ties away from zero.
18626 This corresponds to the @code{__builtin_lroundf} function when
18627 @option{-fno-math-errno} is used.
18629 @item @samp{floatis}, @samp{floatus}, @samp{floatid}, @samp{floatud}
18630 Conversion from signed or unsigned integer types to floating-point types.
18634 In addition, all of the following transfer instructions for internal
18635 registers X and Y must be provided to use any of the double-precision
18636 floating-point instructions. Custom instructions taking two
18637 double-precision source operands expect the first operand in the
18638 64-bit register X. The other operand (or only operand of a unary
18639 operation) is given to the custom arithmetic instruction with the
18640 least significant half in source register @var{src1} and the most
18641 significant half in @var{src2}. A custom instruction that returns a
18642 double-precision result returns the most significant 32 bits in the
18643 destination register and the other half in 32-bit register Y.
18644 GCC automatically generates the necessary code sequences to write
18645 register X and/or read register Y when double-precision floating-point
18646 instructions are used.
18651 Write @var{src1} into the least significant half of X and @var{src2} into
18652 the most significant half of X.
18655 Write @var{src1} into Y.
18657 @item @samp{frdxhi}, @samp{frdxlo}
18658 Read the most or least (respectively) significant half of X and store it in
18662 Read the value of Y and store it into @var{dest}.
18665 Note that you can gain more local control over generation of Nios II custom
18666 instructions by using the @code{target("custom-@var{insn}=@var{N}")}
18667 and @code{target("no-custom-@var{insn}")} function attributes
18668 (@pxref{Function Attributes})
18669 or pragmas (@pxref{Function Specific Option Pragmas}).
18671 @item -mcustom-fpu-cfg=@var{name}
18672 @opindex mcustom-fpu-cfg
18674 This option enables a predefined, named set of custom instruction encodings
18675 (see @option{-mcustom-@var{insn}} above).
18676 Currently, the following sets are defined:
18678 @option{-mcustom-fpu-cfg=60-1} is equivalent to:
18679 @gccoptlist{-mcustom-fmuls=252 @gol
18680 -mcustom-fadds=253 @gol
18681 -mcustom-fsubs=254 @gol
18682 -fsingle-precision-constant}
18684 @option{-mcustom-fpu-cfg=60-2} is equivalent to:
18685 @gccoptlist{-mcustom-fmuls=252 @gol
18686 -mcustom-fadds=253 @gol
18687 -mcustom-fsubs=254 @gol
18688 -mcustom-fdivs=255 @gol
18689 -fsingle-precision-constant}
18691 @option{-mcustom-fpu-cfg=72-3} is equivalent to:
18692 @gccoptlist{-mcustom-floatus=243 @gol
18693 -mcustom-fixsi=244 @gol
18694 -mcustom-floatis=245 @gol
18695 -mcustom-fcmpgts=246 @gol
18696 -mcustom-fcmples=249 @gol
18697 -mcustom-fcmpeqs=250 @gol
18698 -mcustom-fcmpnes=251 @gol
18699 -mcustom-fmuls=252 @gol
18700 -mcustom-fadds=253 @gol
18701 -mcustom-fsubs=254 @gol
18702 -mcustom-fdivs=255 @gol
18703 -fsingle-precision-constant}
18705 Custom instruction assignments given by individual
18706 @option{-mcustom-@var{insn}=} options override those given by
18707 @option{-mcustom-fpu-cfg=}, regardless of the
18708 order of the options on the command line.
18710 Note that you can gain more local control over selection of a FPU
18711 configuration by using the @code{target("custom-fpu-cfg=@var{name}")}
18712 function attribute (@pxref{Function Attributes})
18713 or pragma (@pxref{Function Specific Option Pragmas}).
18717 These additional @samp{-m} options are available for the Altera Nios II
18718 ELF (bare-metal) target:
18724 Link with HAL BSP. This suppresses linking with the GCC-provided C runtime
18725 startup and termination code, and is typically used in conjunction with
18726 @option{-msys-crt0=} to specify the location of the alternate startup code
18727 provided by the HAL BSP.
18731 Link with a limited version of the C library, @option{-lsmallc}, rather than
18734 @item -msys-crt0=@var{startfile}
18736 @var{startfile} is the file name of the startfile (crt0) to use
18737 when linking. This option is only useful in conjunction with @option{-mhal}.
18739 @item -msys-lib=@var{systemlib}
18741 @var{systemlib} is the library name of the library that provides
18742 low-level system calls required by the C library,
18743 e.g. @code{read} and @code{write}.
18744 This option is typically used to link with a library provided by a HAL BSP.
18748 @node Nvidia PTX Options
18749 @subsection Nvidia PTX Options
18750 @cindex Nvidia PTX options
18751 @cindex nvptx options
18753 These options are defined for Nvidia PTX:
18761 Generate code for 32-bit or 64-bit ABI.
18764 @opindex mmainkernel
18765 Link in code for a __main kernel. This is for stand-alone instead of
18766 offloading execution.
18770 @node PDP-11 Options
18771 @subsection PDP-11 Options
18772 @cindex PDP-11 Options
18774 These options are defined for the PDP-11:
18779 Use hardware FPP floating point. This is the default. (FIS floating
18780 point on the PDP-11/40 is not supported.)
18783 @opindex msoft-float
18784 Do not use hardware floating point.
18788 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
18792 Return floating-point results in memory. This is the default.
18796 Generate code for a PDP-11/40.
18800 Generate code for a PDP-11/45. This is the default.
18804 Generate code for a PDP-11/10.
18806 @item -mbcopy-builtin
18807 @opindex mbcopy-builtin
18808 Use inline @code{movmemhi} patterns for copying memory. This is the
18813 Do not use inline @code{movmemhi} patterns for copying memory.
18819 Use 16-bit @code{int}. This is the default.
18825 Use 32-bit @code{int}.
18828 @itemx -mno-float32
18830 @opindex mno-float32
18831 Use 64-bit @code{float}. This is the default.
18834 @itemx -mno-float64
18836 @opindex mno-float64
18837 Use 32-bit @code{float}.
18841 Use @code{abshi2} pattern. This is the default.
18845 Do not use @code{abshi2} pattern.
18847 @item -mbranch-expensive
18848 @opindex mbranch-expensive
18849 Pretend that branches are expensive. This is for experimenting with
18850 code generation only.
18852 @item -mbranch-cheap
18853 @opindex mbranch-cheap
18854 Do not pretend that branches are expensive. This is the default.
18858 Use Unix assembler syntax. This is the default when configured for
18859 @samp{pdp11-*-bsd}.
18863 Use DEC assembler syntax. This is the default when configured for any
18864 PDP-11 target other than @samp{pdp11-*-bsd}.
18867 @node picoChip Options
18868 @subsection picoChip Options
18869 @cindex picoChip options
18871 These @samp{-m} options are defined for picoChip implementations:
18875 @item -mae=@var{ae_type}
18877 Set the instruction set, register set, and instruction scheduling
18878 parameters for array element type @var{ae_type}. Supported values
18879 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
18881 @option{-mae=ANY} selects a completely generic AE type. Code
18882 generated with this option runs on any of the other AE types. The
18883 code is not as efficient as it would be if compiled for a specific
18884 AE type, and some types of operation (e.g., multiplication) do not
18885 work properly on all types of AE.
18887 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
18888 for compiled code, and is the default.
18890 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
18891 option may suffer from poor performance of byte (char) manipulation,
18892 since the DSP AE does not provide hardware support for byte load/stores.
18894 @item -msymbol-as-address
18895 Enable the compiler to directly use a symbol name as an address in a
18896 load/store instruction, without first loading it into a
18897 register. Typically, the use of this option generates larger
18898 programs, which run faster than when the option isn't used. However, the
18899 results vary from program to program, so it is left as a user option,
18900 rather than being permanently enabled.
18902 @item -mno-inefficient-warnings
18903 Disables warnings about the generation of inefficient code. These
18904 warnings can be generated, for example, when compiling code that
18905 performs byte-level memory operations on the MAC AE type. The MAC AE has
18906 no hardware support for byte-level memory operations, so all byte
18907 load/stores must be synthesized from word load/store operations. This is
18908 inefficient and a warning is generated to indicate
18909 that you should rewrite the code to avoid byte operations, or to target
18910 an AE type that has the necessary hardware support. This option disables
18915 @node PowerPC Options
18916 @subsection PowerPC Options
18917 @cindex PowerPC options
18919 These are listed under @xref{RS/6000 and PowerPC Options}.
18922 @subsection RL78 Options
18923 @cindex RL78 Options
18929 Links in additional target libraries to support operation within a
18938 Specifies the type of hardware multiplication and division support to
18939 be used. The simplest is @code{none}, which uses software for both
18940 multiplication and division. This is the default. The @code{g13}
18941 value is for the hardware multiply/divide peripheral found on the
18942 RL78/G13 (S2 core) targets. The @code{g14} value selects the use of
18943 the multiplication and division instructions supported by the RL78/G14
18944 (S3 core) parts. The value @code{rl78} is an alias for @code{g14} and
18945 the value @code{mg10} is an alias for @code{none}.
18947 In addition a C preprocessor macro is defined, based upon the setting
18948 of this option. Possible values are: @code{__RL78_MUL_NONE__},
18949 @code{__RL78_MUL_G13__} or @code{__RL78_MUL_G14__}.
18956 Specifies the RL78 core to target. The default is the G14 core, also
18957 known as an S3 core or just RL78. The G13 or S2 core does not have
18958 multiply or divide instructions, instead it uses a hardware peripheral
18959 for these operations. The G10 or S1 core does not have register
18960 banks, so it uses a different calling convention.
18962 If this option is set it also selects the type of hardware multiply
18963 support to use, unless this is overridden by an explicit
18964 @option{-mmul=none} option on the command line. Thus specifying
18965 @option{-mcpu=g13} enables the use of the G13 hardware multiply
18966 peripheral and specifying @option{-mcpu=g10} disables the use of
18967 hardware multipications altogether.
18969 Note, although the RL78/G14 core is the default target, specifying
18970 @option{-mcpu=g14} or @option{-mcpu=rl78} on the command line does
18971 change the behaviour of the toolchain since it also enables G14
18972 hardware multiply support. If these options are not specified on the
18973 command line then software multiplication routines will be used even
18974 though the code targets the RL78 core. This is for backwards
18975 compatibility with older toolchains which did not have hardware
18976 multiply and divide support.
18978 In addition a C preprocessor macro is defined, based upon the setting
18979 of this option. Possible values are: @code{__RL78_G10__},
18980 @code{__RL78_G13__} or @code{__RL78_G14__}.
18990 These are aliases for the corresponding @option{-mcpu=} option. They
18991 are provided for backwards compatibility.
18995 Allow the compiler to use all of the available registers. By default
18996 registers @code{r24..r31} are reserved for use in interrupt handlers.
18997 With this option enabled these registers can be used in ordinary
19000 @item -m64bit-doubles
19001 @itemx -m32bit-doubles
19002 @opindex m64bit-doubles
19003 @opindex m32bit-doubles
19004 Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
19005 or 32 bits (@option{-m32bit-doubles}) in size. The default is
19006 @option{-m32bit-doubles}.
19010 @node RS/6000 and PowerPC Options
19011 @subsection IBM RS/6000 and PowerPC Options
19012 @cindex RS/6000 and PowerPC Options
19013 @cindex IBM RS/6000 and PowerPC Options
19015 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
19017 @item -mpowerpc-gpopt
19018 @itemx -mno-powerpc-gpopt
19019 @itemx -mpowerpc-gfxopt
19020 @itemx -mno-powerpc-gfxopt
19023 @itemx -mno-powerpc64
19027 @itemx -mno-popcntb
19029 @itemx -mno-popcntd
19038 @itemx -mno-hard-dfp
19039 @opindex mpowerpc-gpopt
19040 @opindex mno-powerpc-gpopt
19041 @opindex mpowerpc-gfxopt
19042 @opindex mno-powerpc-gfxopt
19043 @opindex mpowerpc64
19044 @opindex mno-powerpc64
19048 @opindex mno-popcntb
19050 @opindex mno-popcntd
19056 @opindex mno-mfpgpr
19058 @opindex mno-hard-dfp
19059 You use these options to specify which instructions are available on the
19060 processor you are using. The default value of these options is
19061 determined when configuring GCC@. Specifying the
19062 @option{-mcpu=@var{cpu_type}} overrides the specification of these
19063 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
19064 rather than the options listed above.
19066 Specifying @option{-mpowerpc-gpopt} allows
19067 GCC to use the optional PowerPC architecture instructions in the
19068 General Purpose group, including floating-point square root. Specifying
19069 @option{-mpowerpc-gfxopt} allows GCC to
19070 use the optional PowerPC architecture instructions in the Graphics
19071 group, including floating-point select.
19073 The @option{-mmfcrf} option allows GCC to generate the move from
19074 condition register field instruction implemented on the POWER4
19075 processor and other processors that support the PowerPC V2.01
19077 The @option{-mpopcntb} option allows GCC to generate the popcount and
19078 double-precision FP reciprocal estimate instruction implemented on the
19079 POWER5 processor and other processors that support the PowerPC V2.02
19081 The @option{-mpopcntd} option allows GCC to generate the popcount
19082 instruction implemented on the POWER7 processor and other processors
19083 that support the PowerPC V2.06 architecture.
19084 The @option{-mfprnd} option allows GCC to generate the FP round to
19085 integer instructions implemented on the POWER5+ processor and other
19086 processors that support the PowerPC V2.03 architecture.
19087 The @option{-mcmpb} option allows GCC to generate the compare bytes
19088 instruction implemented on the POWER6 processor and other processors
19089 that support the PowerPC V2.05 architecture.
19090 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
19091 general-purpose register instructions implemented on the POWER6X
19092 processor and other processors that support the extended PowerPC V2.05
19094 The @option{-mhard-dfp} option allows GCC to generate the decimal
19095 floating-point instructions implemented on some POWER processors.
19097 The @option{-mpowerpc64} option allows GCC to generate the additional
19098 64-bit instructions that are found in the full PowerPC64 architecture
19099 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
19100 @option{-mno-powerpc64}.
19102 @item -mcpu=@var{cpu_type}
19104 Set architecture type, register usage, and
19105 instruction scheduling parameters for machine type @var{cpu_type}.
19106 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
19107 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
19108 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
19109 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
19110 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
19111 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
19112 @samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{e5500},
19113 @samp{e6500}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
19114 @samp{titan}, @samp{power3}, @samp{power4}, @samp{power5}, @samp{power5+},
19115 @samp{power6}, @samp{power6x}, @samp{power7}, @samp{power8}, @samp{powerpc},
19116 @samp{powerpc64}, @samp{powerpc64le}, and @samp{rs64}.
19118 @option{-mcpu=powerpc}, @option{-mcpu=powerpc64}, and
19119 @option{-mcpu=powerpc64le} specify pure 32-bit PowerPC (either
19120 endian), 64-bit big endian PowerPC and 64-bit little endian PowerPC
19121 architecture machine types, with an appropriate, generic processor
19122 model assumed for scheduling purposes.
19124 The other options specify a specific processor. Code generated under
19125 those options runs best on that processor, and may not run at all on
19128 The @option{-mcpu} options automatically enable or disable the
19131 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
19132 -mpopcntb -mpopcntd -mpowerpc64 @gol
19133 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
19134 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx @gol
19135 -mcrypto -mdirect-move -mpower8-fusion -mpower8-vector @gol
19136 -mquad-memory -mquad-memory-atomic}
19138 The particular options set for any particular CPU varies between
19139 compiler versions, depending on what setting seems to produce optimal
19140 code for that CPU; it doesn't necessarily reflect the actual hardware's
19141 capabilities. If you wish to set an individual option to a particular
19142 value, you may specify it after the @option{-mcpu} option, like
19143 @option{-mcpu=970 -mno-altivec}.
19145 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
19146 not enabled or disabled by the @option{-mcpu} option at present because
19147 AIX does not have full support for these options. You may still
19148 enable or disable them individually if you're sure it'll work in your
19151 @item -mtune=@var{cpu_type}
19153 Set the instruction scheduling parameters for machine type
19154 @var{cpu_type}, but do not set the architecture type or register usage,
19155 as @option{-mcpu=@var{cpu_type}} does. The same
19156 values for @var{cpu_type} are used for @option{-mtune} as for
19157 @option{-mcpu}. If both are specified, the code generated uses the
19158 architecture and registers set by @option{-mcpu}, but the
19159 scheduling parameters set by @option{-mtune}.
19161 @item -mcmodel=small
19162 @opindex mcmodel=small
19163 Generate PowerPC64 code for the small model: The TOC is limited to
19166 @item -mcmodel=medium
19167 @opindex mcmodel=medium
19168 Generate PowerPC64 code for the medium model: The TOC and other static
19169 data may be up to a total of 4G in size.
19171 @item -mcmodel=large
19172 @opindex mcmodel=large
19173 Generate PowerPC64 code for the large model: The TOC may be up to 4G
19174 in size. Other data and code is only limited by the 64-bit address
19178 @itemx -mno-altivec
19180 @opindex mno-altivec
19181 Generate code that uses (does not use) AltiVec instructions, and also
19182 enable the use of built-in functions that allow more direct access to
19183 the AltiVec instruction set. You may also need to set
19184 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
19187 When @option{-maltivec} is used, rather than @option{-maltivec=le} or
19188 @option{-maltivec=be}, the element order for Altivec intrinsics such
19189 as @code{vec_splat}, @code{vec_extract}, and @code{vec_insert}
19190 match array element order corresponding to the endianness of the
19191 target. That is, element zero identifies the leftmost element in a
19192 vector register when targeting a big-endian platform, and identifies
19193 the rightmost element in a vector register when targeting a
19194 little-endian platform.
19197 @opindex maltivec=be
19198 Generate Altivec instructions using big-endian element order,
19199 regardless of whether the target is big- or little-endian. This is
19200 the default when targeting a big-endian platform.
19202 The element order is used to interpret element numbers in Altivec
19203 intrinsics such as @code{vec_splat}, @code{vec_extract}, and
19204 @code{vec_insert}. By default, these match array element order
19205 corresponding to the endianness for the target.
19208 @opindex maltivec=le
19209 Generate Altivec instructions using little-endian element order,
19210 regardless of whether the target is big- or little-endian. This is
19211 the default when targeting a little-endian platform. This option is
19212 currently ignored when targeting a big-endian platform.
19214 The element order is used to interpret element numbers in Altivec
19215 intrinsics such as @code{vec_splat}, @code{vec_extract}, and
19216 @code{vec_insert}. By default, these match array element order
19217 corresponding to the endianness for the target.
19222 @opindex mno-vrsave
19223 Generate VRSAVE instructions when generating AltiVec code.
19225 @item -mgen-cell-microcode
19226 @opindex mgen-cell-microcode
19227 Generate Cell microcode instructions.
19229 @item -mwarn-cell-microcode
19230 @opindex mwarn-cell-microcode
19231 Warn when a Cell microcode instruction is emitted. An example
19232 of a Cell microcode instruction is a variable shift.
19235 @opindex msecure-plt
19236 Generate code that allows @command{ld} and @command{ld.so}
19237 to build executables and shared
19238 libraries with non-executable @code{.plt} and @code{.got} sections.
19240 32-bit SYSV ABI option.
19244 Generate code that uses a BSS @code{.plt} section that @command{ld.so}
19246 requires @code{.plt} and @code{.got}
19247 sections that are both writable and executable.
19248 This is a PowerPC 32-bit SYSV ABI option.
19254 This switch enables or disables the generation of ISEL instructions.
19256 @item -misel=@var{yes/no}
19257 This switch has been deprecated. Use @option{-misel} and
19258 @option{-mno-isel} instead.
19264 This switch enables or disables the generation of SPE simd
19270 @opindex mno-paired
19271 This switch enables or disables the generation of PAIRED simd
19274 @item -mspe=@var{yes/no}
19275 This option has been deprecated. Use @option{-mspe} and
19276 @option{-mno-spe} instead.
19282 Generate code that uses (does not use) vector/scalar (VSX)
19283 instructions, and also enable the use of built-in functions that allow
19284 more direct access to the VSX instruction set.
19289 @opindex mno-crypto
19290 Enable the use (disable) of the built-in functions that allow direct
19291 access to the cryptographic instructions that were added in version
19292 2.07 of the PowerPC ISA.
19294 @item -mdirect-move
19295 @itemx -mno-direct-move
19296 @opindex mdirect-move
19297 @opindex mno-direct-move
19298 Generate code that uses (does not use) the instructions to move data
19299 between the general purpose registers and the vector/scalar (VSX)
19300 registers that were added in version 2.07 of the PowerPC ISA.
19302 @item -mpower8-fusion
19303 @itemx -mno-power8-fusion
19304 @opindex mpower8-fusion
19305 @opindex mno-power8-fusion
19306 Generate code that keeps (does not keeps) some integer operations
19307 adjacent so that the instructions can be fused together on power8 and
19310 @item -mpower8-vector
19311 @itemx -mno-power8-vector
19312 @opindex mpower8-vector
19313 @opindex mno-power8-vector
19314 Generate code that uses (does not use) the vector and scalar
19315 instructions that were added in version 2.07 of the PowerPC ISA. Also
19316 enable the use of built-in functions that allow more direct access to
19317 the vector instructions.
19319 @item -mquad-memory
19320 @itemx -mno-quad-memory
19321 @opindex mquad-memory
19322 @opindex mno-quad-memory
19323 Generate code that uses (does not use) the non-atomic quad word memory
19324 instructions. The @option{-mquad-memory} option requires use of
19327 @item -mquad-memory-atomic
19328 @itemx -mno-quad-memory-atomic
19329 @opindex mquad-memory-atomic
19330 @opindex mno-quad-memory-atomic
19331 Generate code that uses (does not use) the atomic quad word memory
19332 instructions. The @option{-mquad-memory-atomic} option requires use of
19335 @item -mupper-regs-df
19336 @itemx -mno-upper-regs-df
19337 @opindex mupper-regs-df
19338 @opindex mno-upper-regs-df
19339 Generate code that uses (does not use) the scalar double precision
19340 instructions that target all 64 registers in the vector/scalar
19341 floating point register set that were added in version 2.06 of the
19342 PowerPC ISA. @option{-mupper-regs-df} is turned on by default if you
19343 use any of the @option{-mcpu=power7}, @option{-mcpu=power8}, or
19344 @option{-mvsx} options.
19346 @item -mupper-regs-sf
19347 @itemx -mno-upper-regs-sf
19348 @opindex mupper-regs-sf
19349 @opindex mno-upper-regs-sf
19350 Generate code that uses (does not use) the scalar single precision
19351 instructions that target all 64 registers in the vector/scalar
19352 floating point register set that were added in version 2.07 of the
19353 PowerPC ISA. @option{-mupper-regs-sf} is turned on by default if you
19354 use either of the @option{-mcpu=power8} or @option{-mpower8-vector}
19358 @itemx -mno-upper-regs
19359 @opindex mupper-regs
19360 @opindex mno-upper-regs
19361 Generate code that uses (does not use) the scalar
19362 instructions that target all 64 registers in the vector/scalar
19363 floating point register set, depending on the model of the machine.
19365 If the @option{-mno-upper-regs} option is used, it turns off both
19366 @option{-mupper-regs-sf} and @option{-mupper-regs-df} options.
19368 @item -mfloat-gprs=@var{yes/single/double/no}
19369 @itemx -mfloat-gprs
19370 @opindex mfloat-gprs
19371 This switch enables or disables the generation of floating-point
19372 operations on the general-purpose registers for architectures that
19375 The argument @samp{yes} or @samp{single} enables the use of
19376 single-precision floating-point operations.
19378 The argument @samp{double} enables the use of single and
19379 double-precision floating-point operations.
19381 The argument @samp{no} disables floating-point operations on the
19382 general-purpose registers.
19384 This option is currently only available on the MPC854x.
19390 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
19391 targets (including GNU/Linux). The 32-bit environment sets int, long
19392 and pointer to 32 bits and generates code that runs on any PowerPC
19393 variant. The 64-bit environment sets int to 32 bits and long and
19394 pointer to 64 bits, and generates code for PowerPC64, as for
19395 @option{-mpowerpc64}.
19398 @itemx -mno-fp-in-toc
19399 @itemx -mno-sum-in-toc
19400 @itemx -mminimal-toc
19402 @opindex mno-fp-in-toc
19403 @opindex mno-sum-in-toc
19404 @opindex mminimal-toc
19405 Modify generation of the TOC (Table Of Contents), which is created for
19406 every executable file. The @option{-mfull-toc} option is selected by
19407 default. In that case, GCC allocates at least one TOC entry for
19408 each unique non-automatic variable reference in your program. GCC
19409 also places floating-point constants in the TOC@. However, only
19410 16,384 entries are available in the TOC@.
19412 If you receive a linker error message that saying you have overflowed
19413 the available TOC space, you can reduce the amount of TOC space used
19414 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
19415 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
19416 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
19417 generate code to calculate the sum of an address and a constant at
19418 run time instead of putting that sum into the TOC@. You may specify one
19419 or both of these options. Each causes GCC to produce very slightly
19420 slower and larger code at the expense of conserving TOC space.
19422 If you still run out of space in the TOC even when you specify both of
19423 these options, specify @option{-mminimal-toc} instead. This option causes
19424 GCC to make only one TOC entry for every file. When you specify this
19425 option, GCC produces code that is slower and larger but which
19426 uses extremely little TOC space. You may wish to use this option
19427 only on files that contain less frequently-executed code.
19433 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
19434 @code{long} type, and the infrastructure needed to support them.
19435 Specifying @option{-maix64} implies @option{-mpowerpc64},
19436 while @option{-maix32} disables the 64-bit ABI and
19437 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
19440 @itemx -mno-xl-compat
19441 @opindex mxl-compat
19442 @opindex mno-xl-compat
19443 Produce code that conforms more closely to IBM XL compiler semantics
19444 when using AIX-compatible ABI@. Pass floating-point arguments to
19445 prototyped functions beyond the register save area (RSA) on the stack
19446 in addition to argument FPRs. Do not assume that most significant
19447 double in 128-bit long double value is properly rounded when comparing
19448 values and converting to double. Use XL symbol names for long double
19451 The AIX calling convention was extended but not initially documented to
19452 handle an obscure K&R C case of calling a function that takes the
19453 address of its arguments with fewer arguments than declared. IBM XL
19454 compilers access floating-point arguments that do not fit in the
19455 RSA from the stack when a subroutine is compiled without
19456 optimization. Because always storing floating-point arguments on the
19457 stack is inefficient and rarely needed, this option is not enabled by
19458 default and only is necessary when calling subroutines compiled by IBM
19459 XL compilers without optimization.
19463 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
19464 application written to use message passing with special startup code to
19465 enable the application to run. The system must have PE installed in the
19466 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
19467 must be overridden with the @option{-specs=} option to specify the
19468 appropriate directory location. The Parallel Environment does not
19469 support threads, so the @option{-mpe} option and the @option{-pthread}
19470 option are incompatible.
19472 @item -malign-natural
19473 @itemx -malign-power
19474 @opindex malign-natural
19475 @opindex malign-power
19476 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
19477 @option{-malign-natural} overrides the ABI-defined alignment of larger
19478 types, such as floating-point doubles, on their natural size-based boundary.
19479 The option @option{-malign-power} instructs GCC to follow the ABI-specified
19480 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
19482 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
19486 @itemx -mhard-float
19487 @opindex msoft-float
19488 @opindex mhard-float
19489 Generate code that does not use (uses) the floating-point register set.
19490 Software floating-point emulation is provided if you use the
19491 @option{-msoft-float} option, and pass the option to GCC when linking.
19493 @item -msingle-float
19494 @itemx -mdouble-float
19495 @opindex msingle-float
19496 @opindex mdouble-float
19497 Generate code for single- or double-precision floating-point operations.
19498 @option{-mdouble-float} implies @option{-msingle-float}.
19501 @opindex msimple-fpu
19502 Do not generate @code{sqrt} and @code{div} instructions for hardware
19503 floating-point unit.
19505 @item -mfpu=@var{name}
19507 Specify type of floating-point unit. Valid values for @var{name} are
19508 @samp{sp_lite} (equivalent to @option{-msingle-float -msimple-fpu}),
19509 @samp{dp_lite} (equivalent to @option{-mdouble-float -msimple-fpu}),
19510 @samp{sp_full} (equivalent to @option{-msingle-float}),
19511 and @samp{dp_full} (equivalent to @option{-mdouble-float}).
19514 @opindex mxilinx-fpu
19515 Perform optimizations for the floating-point unit on Xilinx PPC 405/440.
19518 @itemx -mno-multiple
19520 @opindex mno-multiple
19521 Generate code that uses (does not use) the load multiple word
19522 instructions and the store multiple word instructions. These
19523 instructions are generated by default on POWER systems, and not
19524 generated on PowerPC systems. Do not use @option{-mmultiple} on little-endian
19525 PowerPC systems, since those instructions do not work when the
19526 processor is in little-endian mode. The exceptions are PPC740 and
19527 PPC750 which permit these instructions in little-endian mode.
19532 @opindex mno-string
19533 Generate code that uses (does not use) the load string instructions
19534 and the store string word instructions to save multiple registers and
19535 do small block moves. These instructions are generated by default on
19536 POWER systems, and not generated on PowerPC systems. Do not use
19537 @option{-mstring} on little-endian PowerPC systems, since those
19538 instructions do not work when the processor is in little-endian mode.
19539 The exceptions are PPC740 and PPC750 which permit these instructions
19540 in little-endian mode.
19545 @opindex mno-update
19546 Generate code that uses (does not use) the load or store instructions
19547 that update the base register to the address of the calculated memory
19548 location. These instructions are generated by default. If you use
19549 @option{-mno-update}, there is a small window between the time that the
19550 stack pointer is updated and the address of the previous frame is
19551 stored, which means code that walks the stack frame across interrupts or
19552 signals may get corrupted data.
19554 @item -mavoid-indexed-addresses
19555 @itemx -mno-avoid-indexed-addresses
19556 @opindex mavoid-indexed-addresses
19557 @opindex mno-avoid-indexed-addresses
19558 Generate code that tries to avoid (not avoid) the use of indexed load
19559 or store instructions. These instructions can incur a performance
19560 penalty on Power6 processors in certain situations, such as when
19561 stepping through large arrays that cross a 16M boundary. This option
19562 is enabled by default when targeting Power6 and disabled otherwise.
19565 @itemx -mno-fused-madd
19566 @opindex mfused-madd
19567 @opindex mno-fused-madd
19568 Generate code that uses (does not use) the floating-point multiply and
19569 accumulate instructions. These instructions are generated by default
19570 if hardware floating point is used. The machine-dependent
19571 @option{-mfused-madd} option is now mapped to the machine-independent
19572 @option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
19573 mapped to @option{-ffp-contract=off}.
19579 Generate code that uses (does not use) the half-word multiply and
19580 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
19581 These instructions are generated by default when targeting those
19588 Generate code that uses (does not use) the string-search @samp{dlmzb}
19589 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
19590 generated by default when targeting those processors.
19592 @item -mno-bit-align
19594 @opindex mno-bit-align
19595 @opindex mbit-align
19596 On System V.4 and embedded PowerPC systems do not (do) force structures
19597 and unions that contain bit-fields to be aligned to the base type of the
19600 For example, by default a structure containing nothing but 8
19601 @code{unsigned} bit-fields of length 1 is aligned to a 4-byte
19602 boundary and has a size of 4 bytes. By using @option{-mno-bit-align},
19603 the structure is aligned to a 1-byte boundary and is 1 byte in
19606 @item -mno-strict-align
19607 @itemx -mstrict-align
19608 @opindex mno-strict-align
19609 @opindex mstrict-align
19610 On System V.4 and embedded PowerPC systems do not (do) assume that
19611 unaligned memory references are handled by the system.
19613 @item -mrelocatable
19614 @itemx -mno-relocatable
19615 @opindex mrelocatable
19616 @opindex mno-relocatable
19617 Generate code that allows (does not allow) a static executable to be
19618 relocated to a different address at run time. A simple embedded
19619 PowerPC system loader should relocate the entire contents of
19620 @code{.got2} and 4-byte locations listed in the @code{.fixup} section,
19621 a table of 32-bit addresses generated by this option. For this to
19622 work, all objects linked together must be compiled with
19623 @option{-mrelocatable} or @option{-mrelocatable-lib}.
19624 @option{-mrelocatable} code aligns the stack to an 8-byte boundary.
19626 @item -mrelocatable-lib
19627 @itemx -mno-relocatable-lib
19628 @opindex mrelocatable-lib
19629 @opindex mno-relocatable-lib
19630 Like @option{-mrelocatable}, @option{-mrelocatable-lib} generates a
19631 @code{.fixup} section to allow static executables to be relocated at
19632 run time, but @option{-mrelocatable-lib} does not use the smaller stack
19633 alignment of @option{-mrelocatable}. Objects compiled with
19634 @option{-mrelocatable-lib} may be linked with objects compiled with
19635 any combination of the @option{-mrelocatable} options.
19641 On System V.4 and embedded PowerPC systems do not (do) assume that
19642 register 2 contains a pointer to a global area pointing to the addresses
19643 used in the program.
19646 @itemx -mlittle-endian
19648 @opindex mlittle-endian
19649 On System V.4 and embedded PowerPC systems compile code for the
19650 processor in little-endian mode. The @option{-mlittle-endian} option is
19651 the same as @option{-mlittle}.
19654 @itemx -mbig-endian
19656 @opindex mbig-endian
19657 On System V.4 and embedded PowerPC systems compile code for the
19658 processor in big-endian mode. The @option{-mbig-endian} option is
19659 the same as @option{-mbig}.
19661 @item -mdynamic-no-pic
19662 @opindex mdynamic-no-pic
19663 On Darwin and Mac OS X systems, compile code so that it is not
19664 relocatable, but that its external references are relocatable. The
19665 resulting code is suitable for applications, but not shared
19668 @item -msingle-pic-base
19669 @opindex msingle-pic-base
19670 Treat the register used for PIC addressing as read-only, rather than
19671 loading it in the prologue for each function. The runtime system is
19672 responsible for initializing this register with an appropriate value
19673 before execution begins.
19675 @item -mprioritize-restricted-insns=@var{priority}
19676 @opindex mprioritize-restricted-insns
19677 This option controls the priority that is assigned to
19678 dispatch-slot restricted instructions during the second scheduling
19679 pass. The argument @var{priority} takes the value @samp{0}, @samp{1},
19680 or @samp{2} to assign no, highest, or second-highest (respectively)
19681 priority to dispatch-slot restricted
19684 @item -msched-costly-dep=@var{dependence_type}
19685 @opindex msched-costly-dep
19686 This option controls which dependences are considered costly
19687 by the target during instruction scheduling. The argument
19688 @var{dependence_type} takes one of the following values:
19692 No dependence is costly.
19695 All dependences are costly.
19697 @item @samp{true_store_to_load}
19698 A true dependence from store to load is costly.
19700 @item @samp{store_to_load}
19701 Any dependence from store to load is costly.
19704 Any dependence for which the latency is greater than or equal to
19705 @var{number} is costly.
19708 @item -minsert-sched-nops=@var{scheme}
19709 @opindex minsert-sched-nops
19710 This option controls which NOP insertion scheme is used during
19711 the second scheduling pass. The argument @var{scheme} takes one of the
19719 Pad with NOPs any dispatch group that has vacant issue slots,
19720 according to the scheduler's grouping.
19722 @item @samp{regroup_exact}
19723 Insert NOPs to force costly dependent insns into
19724 separate groups. Insert exactly as many NOPs as needed to force an insn
19725 to a new group, according to the estimated processor grouping.
19728 Insert NOPs to force costly dependent insns into
19729 separate groups. Insert @var{number} NOPs to force an insn to a new group.
19733 @opindex mcall-sysv
19734 On System V.4 and embedded PowerPC systems compile code using calling
19735 conventions that adhere to the March 1995 draft of the System V
19736 Application Binary Interface, PowerPC processor supplement. This is the
19737 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
19739 @item -mcall-sysv-eabi
19741 @opindex mcall-sysv-eabi
19742 @opindex mcall-eabi
19743 Specify both @option{-mcall-sysv} and @option{-meabi} options.
19745 @item -mcall-sysv-noeabi
19746 @opindex mcall-sysv-noeabi
19747 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
19749 @item -mcall-aixdesc
19751 On System V.4 and embedded PowerPC systems compile code for the AIX
19755 @opindex mcall-linux
19756 On System V.4 and embedded PowerPC systems compile code for the
19757 Linux-based GNU system.
19759 @item -mcall-freebsd
19760 @opindex mcall-freebsd
19761 On System V.4 and embedded PowerPC systems compile code for the
19762 FreeBSD operating system.
19764 @item -mcall-netbsd
19765 @opindex mcall-netbsd
19766 On System V.4 and embedded PowerPC systems compile code for the
19767 NetBSD operating system.
19769 @item -mcall-openbsd
19770 @opindex mcall-netbsd
19771 On System V.4 and embedded PowerPC systems compile code for the
19772 OpenBSD operating system.
19774 @item -maix-struct-return
19775 @opindex maix-struct-return
19776 Return all structures in memory (as specified by the AIX ABI)@.
19778 @item -msvr4-struct-return
19779 @opindex msvr4-struct-return
19780 Return structures smaller than 8 bytes in registers (as specified by the
19783 @item -mabi=@var{abi-type}
19785 Extend the current ABI with a particular extension, or remove such extension.
19786 Valid values are @samp{altivec}, @samp{no-altivec}, @samp{spe},
19787 @samp{no-spe}, @samp{ibmlongdouble}, @samp{ieeelongdouble},
19788 @samp{elfv1}, @samp{elfv2}@.
19792 Extend the current ABI with SPE ABI extensions. This does not change
19793 the default ABI, instead it adds the SPE ABI extensions to the current
19797 @opindex mabi=no-spe
19798 Disable Book-E SPE ABI extensions for the current ABI@.
19800 @item -mabi=ibmlongdouble
19801 @opindex mabi=ibmlongdouble
19802 Change the current ABI to use IBM extended-precision long double.
19803 This is a PowerPC 32-bit SYSV ABI option.
19805 @item -mabi=ieeelongdouble
19806 @opindex mabi=ieeelongdouble
19807 Change the current ABI to use IEEE extended-precision long double.
19808 This is a PowerPC 32-bit Linux ABI option.
19811 @opindex mabi=elfv1
19812 Change the current ABI to use the ELFv1 ABI.
19813 This is the default ABI for big-endian PowerPC 64-bit Linux.
19814 Overriding the default ABI requires special system support and is
19815 likely to fail in spectacular ways.
19818 @opindex mabi=elfv2
19819 Change the current ABI to use the ELFv2 ABI.
19820 This is the default ABI for little-endian PowerPC 64-bit Linux.
19821 Overriding the default ABI requires special system support and is
19822 likely to fail in spectacular ways.
19825 @itemx -mno-prototype
19826 @opindex mprototype
19827 @opindex mno-prototype
19828 On System V.4 and embedded PowerPC systems assume that all calls to
19829 variable argument functions are properly prototyped. Otherwise, the
19830 compiler must insert an instruction before every non-prototyped call to
19831 set or clear bit 6 of the condition code register (@code{CR}) to
19832 indicate whether floating-point values are passed in the floating-point
19833 registers in case the function takes variable arguments. With
19834 @option{-mprototype}, only calls to prototyped variable argument functions
19835 set or clear the bit.
19839 On embedded PowerPC systems, assume that the startup module is called
19840 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
19841 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
19846 On embedded PowerPC systems, assume that the startup module is called
19847 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
19852 On embedded PowerPC systems, assume that the startup module is called
19853 @file{crt0.o} and the standard C libraries are @file{libads.a} and
19856 @item -myellowknife
19857 @opindex myellowknife
19858 On embedded PowerPC systems, assume that the startup module is called
19859 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
19864 On System V.4 and embedded PowerPC systems, specify that you are
19865 compiling for a VxWorks system.
19869 On embedded PowerPC systems, set the @code{PPC_EMB} bit in the ELF flags
19870 header to indicate that @samp{eabi} extended relocations are used.
19876 On System V.4 and embedded PowerPC systems do (do not) adhere to the
19877 Embedded Applications Binary Interface (EABI), which is a set of
19878 modifications to the System V.4 specifications. Selecting @option{-meabi}
19879 means that the stack is aligned to an 8-byte boundary, a function
19880 @code{__eabi} is called from @code{main} to set up the EABI
19881 environment, and the @option{-msdata} option can use both @code{r2} and
19882 @code{r13} to point to two separate small data areas. Selecting
19883 @option{-mno-eabi} means that the stack is aligned to a 16-byte boundary,
19884 no EABI initialization function is called from @code{main}, and the
19885 @option{-msdata} option only uses @code{r13} to point to a single
19886 small data area. The @option{-meabi} option is on by default if you
19887 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
19890 @opindex msdata=eabi
19891 On System V.4 and embedded PowerPC systems, put small initialized
19892 @code{const} global and static data in the @code{.sdata2} section, which
19893 is pointed to by register @code{r2}. Put small initialized
19894 non-@code{const} global and static data in the @code{.sdata} section,
19895 which is pointed to by register @code{r13}. Put small uninitialized
19896 global and static data in the @code{.sbss} section, which is adjacent to
19897 the @code{.sdata} section. The @option{-msdata=eabi} option is
19898 incompatible with the @option{-mrelocatable} option. The
19899 @option{-msdata=eabi} option also sets the @option{-memb} option.
19902 @opindex msdata=sysv
19903 On System V.4 and embedded PowerPC systems, put small global and static
19904 data in the @code{.sdata} section, which is pointed to by register
19905 @code{r13}. Put small uninitialized global and static data in the
19906 @code{.sbss} section, which is adjacent to the @code{.sdata} section.
19907 The @option{-msdata=sysv} option is incompatible with the
19908 @option{-mrelocatable} option.
19910 @item -msdata=default
19912 @opindex msdata=default
19914 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
19915 compile code the same as @option{-msdata=eabi}, otherwise compile code the
19916 same as @option{-msdata=sysv}.
19919 @opindex msdata=data
19920 On System V.4 and embedded PowerPC systems, put small global
19921 data in the @code{.sdata} section. Put small uninitialized global
19922 data in the @code{.sbss} section. Do not use register @code{r13}
19923 to address small data however. This is the default behavior unless
19924 other @option{-msdata} options are used.
19928 @opindex msdata=none
19930 On embedded PowerPC systems, put all initialized global and static data
19931 in the @code{.data} section, and all uninitialized data in the
19932 @code{.bss} section.
19934 @item -mblock-move-inline-limit=@var{num}
19935 @opindex mblock-move-inline-limit
19936 Inline all block moves (such as calls to @code{memcpy} or structure
19937 copies) less than or equal to @var{num} bytes. The minimum value for
19938 @var{num} is 32 bytes on 32-bit targets and 64 bytes on 64-bit
19939 targets. The default value is target-specific.
19943 @cindex smaller data references (PowerPC)
19944 @cindex .sdata/.sdata2 references (PowerPC)
19945 On embedded PowerPC systems, put global and static items less than or
19946 equal to @var{num} bytes into the small data or BSS sections instead of
19947 the normal data or BSS section. By default, @var{num} is 8. The
19948 @option{-G @var{num}} switch is also passed to the linker.
19949 All modules should be compiled with the same @option{-G @var{num}} value.
19952 @itemx -mno-regnames
19954 @opindex mno-regnames
19955 On System V.4 and embedded PowerPC systems do (do not) emit register
19956 names in the assembly language output using symbolic forms.
19959 @itemx -mno-longcall
19961 @opindex mno-longcall
19962 By default assume that all calls are far away so that a longer and more
19963 expensive calling sequence is required. This is required for calls
19964 farther than 32 megabytes (33,554,432 bytes) from the current location.
19965 A short call is generated if the compiler knows
19966 the call cannot be that far away. This setting can be overridden by
19967 the @code{shortcall} function attribute, or by @code{#pragma
19970 Some linkers are capable of detecting out-of-range calls and generating
19971 glue code on the fly. On these systems, long calls are unnecessary and
19972 generate slower code. As of this writing, the AIX linker can do this,
19973 as can the GNU linker for PowerPC/64. It is planned to add this feature
19974 to the GNU linker for 32-bit PowerPC systems as well.
19976 On Darwin/PPC systems, @code{#pragma longcall} generates @code{jbsr
19977 callee, L42}, plus a @dfn{branch island} (glue code). The two target
19978 addresses represent the callee and the branch island. The
19979 Darwin/PPC linker prefers the first address and generates a @code{bl
19980 callee} if the PPC @code{bl} instruction reaches the callee directly;
19981 otherwise, the linker generates @code{bl L42} to call the branch
19982 island. The branch island is appended to the body of the
19983 calling function; it computes the full 32-bit address of the callee
19986 On Mach-O (Darwin) systems, this option directs the compiler emit to
19987 the glue for every direct call, and the Darwin linker decides whether
19988 to use or discard it.
19990 In the future, GCC may ignore all longcall specifications
19991 when the linker is known to generate glue.
19993 @item -mtls-markers
19994 @itemx -mno-tls-markers
19995 @opindex mtls-markers
19996 @opindex mno-tls-markers
19997 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
19998 specifying the function argument. The relocation allows the linker to
19999 reliably associate function call with argument setup instructions for
20000 TLS optimization, which in turn allows GCC to better schedule the
20005 Adds support for multithreading with the @dfn{pthreads} library.
20006 This option sets flags for both the preprocessor and linker.
20011 This option enables use of the reciprocal estimate and
20012 reciprocal square root estimate instructions with additional
20013 Newton-Raphson steps to increase precision instead of doing a divide or
20014 square root and divide for floating-point arguments. You should use
20015 the @option{-ffast-math} option when using @option{-mrecip} (or at
20016 least @option{-funsafe-math-optimizations},
20017 @option{-finite-math-only}, @option{-freciprocal-math} and
20018 @option{-fno-trapping-math}). Note that while the throughput of the
20019 sequence is generally higher than the throughput of the non-reciprocal
20020 instruction, the precision of the sequence can be decreased by up to 2
20021 ulp (i.e.@: the inverse of 1.0 equals 0.99999994) for reciprocal square
20024 @item -mrecip=@var{opt}
20025 @opindex mrecip=opt
20026 This option controls which reciprocal estimate instructions
20027 may be used. @var{opt} is a comma-separated list of options, which may
20028 be preceded by a @code{!} to invert the option:
20033 Enable all estimate instructions.
20036 Enable the default instructions, equivalent to @option{-mrecip}.
20039 Disable all estimate instructions, equivalent to @option{-mno-recip}.
20042 Enable the reciprocal approximation instructions for both
20043 single and double precision.
20046 Enable the single-precision reciprocal approximation instructions.
20049 Enable the double-precision reciprocal approximation instructions.
20052 Enable the reciprocal square root approximation instructions for both
20053 single and double precision.
20056 Enable the single-precision reciprocal square root approximation instructions.
20059 Enable the double-precision reciprocal square root approximation instructions.
20063 So, for example, @option{-mrecip=all,!rsqrtd} enables
20064 all of the reciprocal estimate instructions, except for the
20065 @code{FRSQRTE}, @code{XSRSQRTEDP}, and @code{XVRSQRTEDP} instructions
20066 which handle the double-precision reciprocal square root calculations.
20068 @item -mrecip-precision
20069 @itemx -mno-recip-precision
20070 @opindex mrecip-precision
20071 Assume (do not assume) that the reciprocal estimate instructions
20072 provide higher-precision estimates than is mandated by the PowerPC
20073 ABI. Selecting @option{-mcpu=power6}, @option{-mcpu=power7} or
20074 @option{-mcpu=power8} automatically selects @option{-mrecip-precision}.
20075 The double-precision square root estimate instructions are not generated by
20076 default on low-precision machines, since they do not provide an
20077 estimate that converges after three steps.
20079 @item -mveclibabi=@var{type}
20080 @opindex mveclibabi
20081 Specifies the ABI type to use for vectorizing intrinsics using an
20082 external library. The only type supported at present is @samp{mass},
20083 which specifies to use IBM's Mathematical Acceleration Subsystem
20084 (MASS) libraries for vectorizing intrinsics using external libraries.
20085 GCC currently emits calls to @code{acosd2}, @code{acosf4},
20086 @code{acoshd2}, @code{acoshf4}, @code{asind2}, @code{asinf4},
20087 @code{asinhd2}, @code{asinhf4}, @code{atan2d2}, @code{atan2f4},
20088 @code{atand2}, @code{atanf4}, @code{atanhd2}, @code{atanhf4},
20089 @code{cbrtd2}, @code{cbrtf4}, @code{cosd2}, @code{cosf4},
20090 @code{coshd2}, @code{coshf4}, @code{erfcd2}, @code{erfcf4},
20091 @code{erfd2}, @code{erff4}, @code{exp2d2}, @code{exp2f4},
20092 @code{expd2}, @code{expf4}, @code{expm1d2}, @code{expm1f4},
20093 @code{hypotd2}, @code{hypotf4}, @code{lgammad2}, @code{lgammaf4},
20094 @code{log10d2}, @code{log10f4}, @code{log1pd2}, @code{log1pf4},
20095 @code{log2d2}, @code{log2f4}, @code{logd2}, @code{logf4},
20096 @code{powd2}, @code{powf4}, @code{sind2}, @code{sinf4}, @code{sinhd2},
20097 @code{sinhf4}, @code{sqrtd2}, @code{sqrtf4}, @code{tand2},
20098 @code{tanf4}, @code{tanhd2}, and @code{tanhf4} when generating code
20099 for power7. Both @option{-ftree-vectorize} and
20100 @option{-funsafe-math-optimizations} must also be enabled. The MASS
20101 libraries must be specified at link time.
20106 Generate (do not generate) the @code{friz} instruction when the
20107 @option{-funsafe-math-optimizations} option is used to optimize
20108 rounding of floating-point values to 64-bit integer and back to floating
20109 point. The @code{friz} instruction does not return the same value if
20110 the floating-point number is too large to fit in an integer.
20112 @item -mpointers-to-nested-functions
20113 @itemx -mno-pointers-to-nested-functions
20114 @opindex mpointers-to-nested-functions
20115 Generate (do not generate) code to load up the static chain register
20116 (@code{r11}) when calling through a pointer on AIX and 64-bit Linux
20117 systems where a function pointer points to a 3-word descriptor giving
20118 the function address, TOC value to be loaded in register @code{r2}, and
20119 static chain value to be loaded in register @code{r11}. The
20120 @option{-mpointers-to-nested-functions} is on by default. You cannot
20121 call through pointers to nested functions or pointers
20122 to functions compiled in other languages that use the static chain if
20123 you use @option{-mno-pointers-to-nested-functions}.
20125 @item -msave-toc-indirect
20126 @itemx -mno-save-toc-indirect
20127 @opindex msave-toc-indirect
20128 Generate (do not generate) code to save the TOC value in the reserved
20129 stack location in the function prologue if the function calls through
20130 a pointer on AIX and 64-bit Linux systems. If the TOC value is not
20131 saved in the prologue, it is saved just before the call through the
20132 pointer. The @option{-mno-save-toc-indirect} option is the default.
20134 @item -mcompat-align-parm
20135 @itemx -mno-compat-align-parm
20136 @opindex mcompat-align-parm
20137 Generate (do not generate) code to pass structure parameters with a
20138 maximum alignment of 64 bits, for compatibility with older versions
20141 Older versions of GCC (prior to 4.9.0) incorrectly did not align a
20142 structure parameter on a 128-bit boundary when that structure contained
20143 a member requiring 128-bit alignment. This is corrected in more
20144 recent versions of GCC. This option may be used to generate code
20145 that is compatible with functions compiled with older versions of
20148 The @option{-mno-compat-align-parm} option is the default.
20152 @subsection RX Options
20155 These command-line options are defined for RX targets:
20158 @item -m64bit-doubles
20159 @itemx -m32bit-doubles
20160 @opindex m64bit-doubles
20161 @opindex m32bit-doubles
20162 Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
20163 or 32 bits (@option{-m32bit-doubles}) in size. The default is
20164 @option{-m32bit-doubles}. @emph{Note} RX floating-point hardware only
20165 works on 32-bit values, which is why the default is
20166 @option{-m32bit-doubles}.
20172 Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
20173 floating-point hardware. The default is enabled for the RX600
20174 series and disabled for the RX200 series.
20176 Floating-point instructions are only generated for 32-bit floating-point
20177 values, however, so the FPU hardware is not used for doubles if the
20178 @option{-m64bit-doubles} option is used.
20180 @emph{Note} If the @option{-fpu} option is enabled then
20181 @option{-funsafe-math-optimizations} is also enabled automatically.
20182 This is because the RX FPU instructions are themselves unsafe.
20184 @item -mcpu=@var{name}
20186 Selects the type of RX CPU to be targeted. Currently three types are
20187 supported, the generic @samp{RX600} and @samp{RX200} series hardware and
20188 the specific @samp{RX610} CPU. The default is @samp{RX600}.
20190 The only difference between @samp{RX600} and @samp{RX610} is that the
20191 @samp{RX610} does not support the @code{MVTIPL} instruction.
20193 The @samp{RX200} series does not have a hardware floating-point unit
20194 and so @option{-nofpu} is enabled by default when this type is
20197 @item -mbig-endian-data
20198 @itemx -mlittle-endian-data
20199 @opindex mbig-endian-data
20200 @opindex mlittle-endian-data
20201 Store data (but not code) in the big-endian format. The default is
20202 @option{-mlittle-endian-data}, i.e.@: to store data in the little-endian
20205 @item -msmall-data-limit=@var{N}
20206 @opindex msmall-data-limit
20207 Specifies the maximum size in bytes of global and static variables
20208 which can be placed into the small data area. Using the small data
20209 area can lead to smaller and faster code, but the size of area is
20210 limited and it is up to the programmer to ensure that the area does
20211 not overflow. Also when the small data area is used one of the RX's
20212 registers (usually @code{r13}) is reserved for use pointing to this
20213 area, so it is no longer available for use by the compiler. This
20214 could result in slower and/or larger code if variables are pushed onto
20215 the stack instead of being held in this register.
20217 Note, common variables (variables that have not been initialized) and
20218 constants are not placed into the small data area as they are assigned
20219 to other sections in the output executable.
20221 The default value is zero, which disables this feature. Note, this
20222 feature is not enabled by default with higher optimization levels
20223 (@option{-O2} etc) because of the potentially detrimental effects of
20224 reserving a register. It is up to the programmer to experiment and
20225 discover whether this feature is of benefit to their program. See the
20226 description of the @option{-mpid} option for a description of how the
20227 actual register to hold the small data area pointer is chosen.
20233 Use the simulator runtime. The default is to use the libgloss
20234 board-specific runtime.
20236 @item -mas100-syntax
20237 @itemx -mno-as100-syntax
20238 @opindex mas100-syntax
20239 @opindex mno-as100-syntax
20240 When generating assembler output use a syntax that is compatible with
20241 Renesas's AS100 assembler. This syntax can also be handled by the GAS
20242 assembler, but it has some restrictions so it is not generated by default.
20244 @item -mmax-constant-size=@var{N}
20245 @opindex mmax-constant-size
20246 Specifies the maximum size, in bytes, of a constant that can be used as
20247 an operand in a RX instruction. Although the RX instruction set does
20248 allow constants of up to 4 bytes in length to be used in instructions,
20249 a longer value equates to a longer instruction. Thus in some
20250 circumstances it can be beneficial to restrict the size of constants
20251 that are used in instructions. Constants that are too big are instead
20252 placed into a constant pool and referenced via register indirection.
20254 The value @var{N} can be between 0 and 4. A value of 0 (the default)
20255 or 4 means that constants of any size are allowed.
20259 Enable linker relaxation. Linker relaxation is a process whereby the
20260 linker attempts to reduce the size of a program by finding shorter
20261 versions of various instructions. Disabled by default.
20263 @item -mint-register=@var{N}
20264 @opindex mint-register
20265 Specify the number of registers to reserve for fast interrupt handler
20266 functions. The value @var{N} can be between 0 and 4. A value of 1
20267 means that register @code{r13} is reserved for the exclusive use
20268 of fast interrupt handlers. A value of 2 reserves @code{r13} and
20269 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
20270 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
20271 A value of 0, the default, does not reserve any registers.
20273 @item -msave-acc-in-interrupts
20274 @opindex msave-acc-in-interrupts
20275 Specifies that interrupt handler functions should preserve the
20276 accumulator register. This is only necessary if normal code might use
20277 the accumulator register, for example because it performs 64-bit
20278 multiplications. The default is to ignore the accumulator as this
20279 makes the interrupt handlers faster.
20285 Enables the generation of position independent data. When enabled any
20286 access to constant data is done via an offset from a base address
20287 held in a register. This allows the location of constant data to be
20288 determined at run time without requiring the executable to be
20289 relocated, which is a benefit to embedded applications with tight
20290 memory constraints. Data that can be modified is not affected by this
20293 Note, using this feature reserves a register, usually @code{r13}, for
20294 the constant data base address. This can result in slower and/or
20295 larger code, especially in complicated functions.
20297 The actual register chosen to hold the constant data base address
20298 depends upon whether the @option{-msmall-data-limit} and/or the
20299 @option{-mint-register} command-line options are enabled. Starting
20300 with register @code{r13} and proceeding downwards, registers are
20301 allocated first to satisfy the requirements of @option{-mint-register},
20302 then @option{-mpid} and finally @option{-msmall-data-limit}. Thus it
20303 is possible for the small data area register to be @code{r8} if both
20304 @option{-mint-register=4} and @option{-mpid} are specified on the
20307 By default this feature is not enabled. The default can be restored
20308 via the @option{-mno-pid} command-line option.
20310 @item -mno-warn-multiple-fast-interrupts
20311 @itemx -mwarn-multiple-fast-interrupts
20312 @opindex mno-warn-multiple-fast-interrupts
20313 @opindex mwarn-multiple-fast-interrupts
20314 Prevents GCC from issuing a warning message if it finds more than one
20315 fast interrupt handler when it is compiling a file. The default is to
20316 issue a warning for each extra fast interrupt handler found, as the RX
20317 only supports one such interrupt.
20319 @item -mallow-string-insns
20320 @itemx -mno-allow-string-insns
20321 @opindex mallow-string-insns
20322 @opindex mno-allow-string-insns
20323 Enables or disables the use of the string manipulation instructions
20324 @code{SMOVF}, @code{SCMPU}, @code{SMOVB}, @code{SMOVU}, @code{SUNTIL}
20325 @code{SWHILE} and also the @code{RMPA} instruction. These
20326 instructions may prefetch data, which is not safe to do if accessing
20327 an I/O register. (See section 12.2.7 of the RX62N Group User's Manual
20328 for more information).
20330 The default is to allow these instructions, but it is not possible for
20331 GCC to reliably detect all circumstances where a string instruction
20332 might be used to access an I/O register, so their use cannot be
20333 disabled automatically. Instead it is reliant upon the programmer to
20334 use the @option{-mno-allow-string-insns} option if their program
20335 accesses I/O space.
20337 When the instructions are enabled GCC defines the C preprocessor
20338 symbol @code{__RX_ALLOW_STRING_INSNS__}, otherwise it defines the
20339 symbol @code{__RX_DISALLOW_STRING_INSNS__}.
20342 @emph{Note:} The generic GCC command-line option @option{-ffixed-@var{reg}}
20343 has special significance to the RX port when used with the
20344 @code{interrupt} function attribute. This attribute indicates a
20345 function intended to process fast interrupts. GCC ensures
20346 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
20347 and/or @code{r13} and only provided that the normal use of the
20348 corresponding registers have been restricted via the
20349 @option{-ffixed-@var{reg}} or @option{-mint-register} command-line
20352 @node S/390 and zSeries Options
20353 @subsection S/390 and zSeries Options
20354 @cindex S/390 and zSeries Options
20356 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
20360 @itemx -msoft-float
20361 @opindex mhard-float
20362 @opindex msoft-float
20363 Use (do not use) the hardware floating-point instructions and registers
20364 for floating-point operations. When @option{-msoft-float} is specified,
20365 functions in @file{libgcc.a} are used to perform floating-point
20366 operations. When @option{-mhard-float} is specified, the compiler
20367 generates IEEE floating-point instructions. This is the default.
20370 @itemx -mno-hard-dfp
20372 @opindex mno-hard-dfp
20373 Use (do not use) the hardware decimal-floating-point instructions for
20374 decimal-floating-point operations. When @option{-mno-hard-dfp} is
20375 specified, functions in @file{libgcc.a} are used to perform
20376 decimal-floating-point operations. When @option{-mhard-dfp} is
20377 specified, the compiler generates decimal-floating-point hardware
20378 instructions. This is the default for @option{-march=z9-ec} or higher.
20380 @item -mlong-double-64
20381 @itemx -mlong-double-128
20382 @opindex mlong-double-64
20383 @opindex mlong-double-128
20384 These switches control the size of @code{long double} type. A size
20385 of 64 bits makes the @code{long double} type equivalent to the @code{double}
20386 type. This is the default.
20389 @itemx -mno-backchain
20390 @opindex mbackchain
20391 @opindex mno-backchain
20392 Store (do not store) the address of the caller's frame as backchain pointer
20393 into the callee's stack frame.
20394 A backchain may be needed to allow debugging using tools that do not understand
20395 DWARF 2 call frame information.
20396 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
20397 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
20398 the backchain is placed into the topmost word of the 96/160 byte register
20401 In general, code compiled with @option{-mbackchain} is call-compatible with
20402 code compiled with @option{-mmo-backchain}; however, use of the backchain
20403 for debugging purposes usually requires that the whole binary is built with
20404 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
20405 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
20406 to build a linux kernel use @option{-msoft-float}.
20408 The default is to not maintain the backchain.
20410 @item -mpacked-stack
20411 @itemx -mno-packed-stack
20412 @opindex mpacked-stack
20413 @opindex mno-packed-stack
20414 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
20415 specified, the compiler uses the all fields of the 96/160 byte register save
20416 area only for their default purpose; unused fields still take up stack space.
20417 When @option{-mpacked-stack} is specified, register save slots are densely
20418 packed at the top of the register save area; unused space is reused for other
20419 purposes, allowing for more efficient use of the available stack space.
20420 However, when @option{-mbackchain} is also in effect, the topmost word of
20421 the save area is always used to store the backchain, and the return address
20422 register is always saved two words below the backchain.
20424 As long as the stack frame backchain is not used, code generated with
20425 @option{-mpacked-stack} is call-compatible with code generated with
20426 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
20427 S/390 or zSeries generated code that uses the stack frame backchain at run
20428 time, not just for debugging purposes. Such code is not call-compatible
20429 with code compiled with @option{-mpacked-stack}. Also, note that the
20430 combination of @option{-mbackchain},
20431 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
20432 to build a linux kernel use @option{-msoft-float}.
20434 The default is to not use the packed stack layout.
20437 @itemx -mno-small-exec
20438 @opindex msmall-exec
20439 @opindex mno-small-exec
20440 Generate (or do not generate) code using the @code{bras} instruction
20441 to do subroutine calls.
20442 This only works reliably if the total executable size does not
20443 exceed 64k. The default is to use the @code{basr} instruction instead,
20444 which does not have this limitation.
20450 When @option{-m31} is specified, generate code compliant to the
20451 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
20452 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
20453 particular to generate 64-bit instructions. For the @samp{s390}
20454 targets, the default is @option{-m31}, while the @samp{s390x}
20455 targets default to @option{-m64}.
20461 When @option{-mzarch} is specified, generate code using the
20462 instructions available on z/Architecture.
20463 When @option{-mesa} is specified, generate code using the
20464 instructions available on ESA/390. Note that @option{-mesa} is
20465 not possible with @option{-m64}.
20466 When generating code compliant to the GNU/Linux for S/390 ABI,
20467 the default is @option{-mesa}. When generating code compliant
20468 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
20474 Generate (or do not generate) code using the @code{mvcle} instruction
20475 to perform block moves. When @option{-mno-mvcle} is specified,
20476 use a @code{mvc} loop instead. This is the default unless optimizing for
20483 Print (or do not print) additional debug information when compiling.
20484 The default is to not print debug information.
20486 @item -march=@var{cpu-type}
20488 Generate code that runs on @var{cpu-type}, which is the name of a system
20489 representing a certain processor type. Possible values for
20490 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
20491 @samp{z9-109}, @samp{z9-ec}, @samp{z10}, @samp{z196}, @samp{zEC12},
20493 When generating code using the instructions available on z/Architecture,
20494 the default is @option{-march=z900}. Otherwise, the default is
20495 @option{-march=g5}.
20497 @item -mtune=@var{cpu-type}
20499 Tune to @var{cpu-type} everything applicable about the generated code,
20500 except for the ABI and the set of available instructions.
20501 The list of @var{cpu-type} values is the same as for @option{-march}.
20502 The default is the value used for @option{-march}.
20505 @itemx -mno-tpf-trace
20506 @opindex mtpf-trace
20507 @opindex mno-tpf-trace
20508 Generate code that adds (does not add) in TPF OS specific branches to trace
20509 routines in the operating system. This option is off by default, even
20510 when compiling for the TPF OS@.
20513 @itemx -mno-fused-madd
20514 @opindex mfused-madd
20515 @opindex mno-fused-madd
20516 Generate code that uses (does not use) the floating-point multiply and
20517 accumulate instructions. These instructions are generated by default if
20518 hardware floating point is used.
20520 @item -mwarn-framesize=@var{framesize}
20521 @opindex mwarn-framesize
20522 Emit a warning if the current function exceeds the given frame size. Because
20523 this is a compile-time check it doesn't need to be a real problem when the program
20524 runs. It is intended to identify functions that most probably cause
20525 a stack overflow. It is useful to be used in an environment with limited stack
20526 size e.g.@: the linux kernel.
20528 @item -mwarn-dynamicstack
20529 @opindex mwarn-dynamicstack
20530 Emit a warning if the function calls @code{alloca} or uses dynamically-sized
20531 arrays. This is generally a bad idea with a limited stack size.
20533 @item -mstack-guard=@var{stack-guard}
20534 @itemx -mstack-size=@var{stack-size}
20535 @opindex mstack-guard
20536 @opindex mstack-size
20537 If these options are provided the S/390 back end emits additional instructions in
20538 the function prologue that trigger a trap if the stack size is @var{stack-guard}
20539 bytes above the @var{stack-size} (remember that the stack on S/390 grows downward).
20540 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
20541 the frame size of the compiled function is chosen.
20542 These options are intended to be used to help debugging stack overflow problems.
20543 The additionally emitted code causes only little overhead and hence can also be
20544 used in production-like systems without greater performance degradation. The given
20545 values have to be exact powers of 2 and @var{stack-size} has to be greater than
20546 @var{stack-guard} without exceeding 64k.
20547 In order to be efficient the extra code makes the assumption that the stack starts
20548 at an address aligned to the value given by @var{stack-size}.
20549 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
20551 @item -mhotpatch=@var{pre-halfwords},@var{post-halfwords}
20553 If the hotpatch option is enabled, a ``hot-patching'' function
20554 prologue is generated for all functions in the compilation unit.
20555 The funtion label is prepended with the given number of two-byte
20556 NOP instructions (@var{pre-halfwords}, maximum 1000000). After
20557 the label, 2 * @var{post-halfwords} bytes are appended, using the
20558 largest NOP like instructions the architecture allows (maximum
20561 If both arguments are zero, hotpatching is disabled.
20563 This option can be overridden for individual functions with the
20564 @code{hotpatch} attribute.
20567 @node Score Options
20568 @subsection Score Options
20569 @cindex Score Options
20571 These options are defined for Score implementations:
20576 Compile code for big-endian mode. This is the default.
20580 Compile code for little-endian mode.
20584 Disable generation of @code{bcnz} instructions.
20588 Enable generation of unaligned load and store instructions.
20592 Enable the use of multiply-accumulate instructions. Disabled by default.
20596 Specify the SCORE5 as the target architecture.
20600 Specify the SCORE5U of the target architecture.
20604 Specify the SCORE7 as the target architecture. This is the default.
20608 Specify the SCORE7D as the target architecture.
20612 @subsection SH Options
20614 These @samp{-m} options are defined for the SH implementations:
20619 Generate code for the SH1.
20623 Generate code for the SH2.
20626 Generate code for the SH2e.
20630 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
20631 that the floating-point unit is not used.
20633 @item -m2a-single-only
20634 @opindex m2a-single-only
20635 Generate code for the SH2a-FPU, in such a way that no double-precision
20636 floating-point operations are used.
20639 @opindex m2a-single
20640 Generate code for the SH2a-FPU assuming the floating-point unit is in
20641 single-precision mode by default.
20645 Generate code for the SH2a-FPU assuming the floating-point unit is in
20646 double-precision mode by default.
20650 Generate code for the SH3.
20654 Generate code for the SH3e.
20658 Generate code for the SH4 without a floating-point unit.
20660 @item -m4-single-only
20661 @opindex m4-single-only
20662 Generate code for the SH4 with a floating-point unit that only
20663 supports single-precision arithmetic.
20667 Generate code for the SH4 assuming the floating-point unit is in
20668 single-precision mode by default.
20672 Generate code for the SH4.
20676 Generate code for SH4-100.
20678 @item -m4-100-nofpu
20679 @opindex m4-100-nofpu
20680 Generate code for SH4-100 in such a way that the
20681 floating-point unit is not used.
20683 @item -m4-100-single
20684 @opindex m4-100-single
20685 Generate code for SH4-100 assuming the floating-point unit is in
20686 single-precision mode by default.
20688 @item -m4-100-single-only
20689 @opindex m4-100-single-only
20690 Generate code for SH4-100 in such a way that no double-precision
20691 floating-point operations are used.
20695 Generate code for SH4-200.
20697 @item -m4-200-nofpu
20698 @opindex m4-200-nofpu
20699 Generate code for SH4-200 without in such a way that the
20700 floating-point unit is not used.
20702 @item -m4-200-single
20703 @opindex m4-200-single
20704 Generate code for SH4-200 assuming the floating-point unit is in
20705 single-precision mode by default.
20707 @item -m4-200-single-only
20708 @opindex m4-200-single-only
20709 Generate code for SH4-200 in such a way that no double-precision
20710 floating-point operations are used.
20714 Generate code for SH4-300.
20716 @item -m4-300-nofpu
20717 @opindex m4-300-nofpu
20718 Generate code for SH4-300 without in such a way that the
20719 floating-point unit is not used.
20721 @item -m4-300-single
20722 @opindex m4-300-single
20723 Generate code for SH4-300 in such a way that no double-precision
20724 floating-point operations are used.
20726 @item -m4-300-single-only
20727 @opindex m4-300-single-only
20728 Generate code for SH4-300 in such a way that no double-precision
20729 floating-point operations are used.
20733 Generate code for SH4-340 (no MMU, no FPU).
20737 Generate code for SH4-500 (no FPU). Passes @option{-isa=sh4-nofpu} to the
20742 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
20743 floating-point unit is not used.
20745 @item -m4a-single-only
20746 @opindex m4a-single-only
20747 Generate code for the SH4a, in such a way that no double-precision
20748 floating-point operations are used.
20751 @opindex m4a-single
20752 Generate code for the SH4a assuming the floating-point unit is in
20753 single-precision mode by default.
20757 Generate code for the SH4a.
20761 Same as @option{-m4a-nofpu}, except that it implicitly passes
20762 @option{-dsp} to the assembler. GCC doesn't generate any DSP
20763 instructions at the moment.
20766 @opindex m5-32media
20767 Generate 32-bit code for SHmedia.
20769 @item -m5-32media-nofpu
20770 @opindex m5-32media-nofpu
20771 Generate 32-bit code for SHmedia in such a way that the
20772 floating-point unit is not used.
20775 @opindex m5-64media
20776 Generate 64-bit code for SHmedia.
20778 @item -m5-64media-nofpu
20779 @opindex m5-64media-nofpu
20780 Generate 64-bit code for SHmedia in such a way that the
20781 floating-point unit is not used.
20784 @opindex m5-compact
20785 Generate code for SHcompact.
20787 @item -m5-compact-nofpu
20788 @opindex m5-compact-nofpu
20789 Generate code for SHcompact in such a way that the
20790 floating-point unit is not used.
20794 Compile code for the processor in big-endian mode.
20798 Compile code for the processor in little-endian mode.
20802 Align doubles at 64-bit boundaries. Note that this changes the calling
20803 conventions, and thus some functions from the standard C library do
20804 not work unless you recompile it first with @option{-mdalign}.
20808 Shorten some address references at link time, when possible; uses the
20809 linker option @option{-relax}.
20813 Use 32-bit offsets in @code{switch} tables. The default is to use
20818 Enable the use of bit manipulation instructions on SH2A.
20822 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
20823 alignment constraints.
20827 Comply with the calling conventions defined by Renesas.
20830 @opindex mno-renesas
20831 Comply with the calling conventions defined for GCC before the Renesas
20832 conventions were available. This option is the default for all
20833 targets of the SH toolchain.
20836 @opindex mnomacsave
20837 Mark the @code{MAC} register as call-clobbered, even if
20838 @option{-mrenesas} is given.
20844 Control the IEEE compliance of floating-point comparisons, which affects the
20845 handling of cases where the result of a comparison is unordered. By default
20846 @option{-mieee} is implicitly enabled. If @option{-ffinite-math-only} is
20847 enabled @option{-mno-ieee} is implicitly set, which results in faster
20848 floating-point greater-equal and less-equal comparisons. The implcit settings
20849 can be overridden by specifying either @option{-mieee} or @option{-mno-ieee}.
20851 @item -minline-ic_invalidate
20852 @opindex minline-ic_invalidate
20853 Inline code to invalidate instruction cache entries after setting up
20854 nested function trampolines.
20855 This option has no effect if @option{-musermode} is in effect and the selected
20856 code generation option (e.g. @option{-m4}) does not allow the use of the @code{icbi}
20858 If the selected code generation option does not allow the use of the @code{icbi}
20859 instruction, and @option{-musermode} is not in effect, the inlined code
20860 manipulates the instruction cache address array directly with an associative
20861 write. This not only requires privileged mode at run time, but it also
20862 fails if the cache line had been mapped via the TLB and has become unmapped.
20866 Dump instruction size and location in the assembly code.
20869 @opindex mpadstruct
20870 This option is deprecated. It pads structures to multiple of 4 bytes,
20871 which is incompatible with the SH ABI@.
20873 @item -matomic-model=@var{model}
20874 @opindex matomic-model=@var{model}
20875 Sets the model of atomic operations and additional parameters as a comma
20876 separated list. For details on the atomic built-in functions see
20877 @ref{__atomic Builtins}. The following models and parameters are supported:
20882 Disable compiler generated atomic sequences and emit library calls for atomic
20883 operations. This is the default if the target is not @code{sh*-*-linux*}.
20886 Generate GNU/Linux compatible gUSA software atomic sequences for the atomic
20887 built-in functions. The generated atomic sequences require additional support
20888 from the interrupt/exception handling code of the system and are only suitable
20889 for SH3* and SH4* single-core systems. This option is enabled by default when
20890 the target is @code{sh*-*-linux*} and SH3* or SH4*. When the target is SH4A,
20891 this option also partially utilizes the hardware atomic instructions
20892 @code{movli.l} and @code{movco.l} to create more efficient code, unless
20893 @samp{strict} is specified.
20896 Generate software atomic sequences that use a variable in the thread control
20897 block. This is a variation of the gUSA sequences which can also be used on
20898 SH1* and SH2* targets. The generated atomic sequences require additional
20899 support from the interrupt/exception handling code of the system and are only
20900 suitable for single-core systems. When using this model, the @samp{gbr-offset=}
20901 parameter has to be specified as well.
20904 Generate software atomic sequences that temporarily disable interrupts by
20905 setting @code{SR.IMASK = 1111}. This model works only when the program runs
20906 in privileged mode and is only suitable for single-core systems. Additional
20907 support from the interrupt/exception handling code of the system is not
20908 required. This model is enabled by default when the target is
20909 @code{sh*-*-linux*} and SH1* or SH2*.
20912 Generate hardware atomic sequences using the @code{movli.l} and @code{movco.l}
20913 instructions only. This is only available on SH4A and is suitable for
20914 multi-core systems. Since the hardware instructions support only 32 bit atomic
20915 variables access to 8 or 16 bit variables is emulated with 32 bit accesses.
20916 Code compiled with this option is also compatible with other software
20917 atomic model interrupt/exception handling systems if executed on an SH4A
20918 system. Additional support from the interrupt/exception handling code of the
20919 system is not required for this model.
20922 This parameter specifies the offset in bytes of the variable in the thread
20923 control block structure that should be used by the generated atomic sequences
20924 when the @samp{soft-tcb} model has been selected. For other models this
20925 parameter is ignored. The specified value must be an integer multiple of four
20926 and in the range 0-1020.
20929 This parameter prevents mixed usage of multiple atomic models, even if they
20930 are compatible, and makes the compiler generate atomic sequences of the
20931 specified model only.
20937 Generate the @code{tas.b} opcode for @code{__atomic_test_and_set}.
20938 Notice that depending on the particular hardware and software configuration
20939 this can degrade overall performance due to the operand cache line flushes
20940 that are implied by the @code{tas.b} instruction. On multi-core SH4A
20941 processors the @code{tas.b} instruction must be used with caution since it
20942 can result in data corruption for certain cache configurations.
20945 @opindex mprefergot
20946 When generating position-independent code, emit function calls using
20947 the Global Offset Table instead of the Procedure Linkage Table.
20950 @itemx -mno-usermode
20952 @opindex mno-usermode
20953 Don't allow (allow) the compiler generating privileged mode code. Specifying
20954 @option{-musermode} also implies @option{-mno-inline-ic_invalidate} if the
20955 inlined code would not work in user mode. @option{-musermode} is the default
20956 when the target is @code{sh*-*-linux*}. If the target is SH1* or SH2*
20957 @option{-musermode} has no effect, since there is no user mode.
20959 @item -multcost=@var{number}
20960 @opindex multcost=@var{number}
20961 Set the cost to assume for a multiply insn.
20963 @item -mdiv=@var{strategy}
20964 @opindex mdiv=@var{strategy}
20965 Set the division strategy to be used for integer division operations.
20966 For SHmedia @var{strategy} can be one of:
20971 Performs the operation in floating point. This has a very high latency,
20972 but needs only a few instructions, so it might be a good choice if
20973 your code has enough easily-exploitable ILP to allow the compiler to
20974 schedule the floating-point instructions together with other instructions.
20975 Division by zero causes a floating-point exception.
20978 Uses integer operations to calculate the inverse of the divisor,
20979 and then multiplies the dividend with the inverse. This strategy allows
20980 CSE and hoisting of the inverse calculation. Division by zero calculates
20981 an unspecified result, but does not trap.
20984 A variant of @samp{inv} where, if no CSE or hoisting opportunities
20985 have been found, or if the entire operation has been hoisted to the same
20986 place, the last stages of the inverse calculation are intertwined with the
20987 final multiply to reduce the overall latency, at the expense of using a few
20988 more instructions, and thus offering fewer scheduling opportunities with
20992 Calls a library function that usually implements the @samp{inv:minlat}
20994 This gives high code density for @code{m5-*media-nofpu} compilations.
20997 Uses a different entry point of the same library function, where it
20998 assumes that a pointer to a lookup table has already been set up, which
20999 exposes the pointer load to CSE and code hoisting optimizations.
21004 Use the @samp{inv} algorithm for initial
21005 code generation, but if the code stays unoptimized, revert to the @samp{call},
21006 @samp{call2}, or @samp{fp} strategies, respectively. Note that the
21007 potentially-trapping side effect of division by zero is carried by a
21008 separate instruction, so it is possible that all the integer instructions
21009 are hoisted out, but the marker for the side effect stays where it is.
21010 A recombination to floating-point operations or a call is not possible
21015 Variants of the @samp{inv:minlat} strategy. In the case
21016 that the inverse calculation is not separated from the multiply, they speed
21017 up division where the dividend fits into 20 bits (plus sign where applicable)
21018 by inserting a test to skip a number of operations in this case; this test
21019 slows down the case of larger dividends. @samp{inv20u} assumes the case of a such
21020 a small dividend to be unlikely, and @samp{inv20l} assumes it to be likely.
21024 For targets other than SHmedia @var{strategy} can be one of:
21029 Calls a library function that uses the single-step division instruction
21030 @code{div1} to perform the operation. Division by zero calculates an
21031 unspecified result and does not trap. This is the default except for SH4,
21032 SH2A and SHcompact.
21035 Calls a library function that performs the operation in double precision
21036 floating point. Division by zero causes a floating-point exception. This is
21037 the default for SHcompact with FPU. Specifying this for targets that do not
21038 have a double precision FPU defaults to @code{call-div1}.
21041 Calls a library function that uses a lookup table for small divisors and
21042 the @code{div1} instruction with case distinction for larger divisors. Division
21043 by zero calculates an unspecified result and does not trap. This is the default
21044 for SH4. Specifying this for targets that do not have dynamic shift
21045 instructions defaults to @code{call-div1}.
21049 When a division strategy has not been specified the default strategy is
21050 selected based on the current target. For SH2A the default strategy is to
21051 use the @code{divs} and @code{divu} instructions instead of library function
21054 @item -maccumulate-outgoing-args
21055 @opindex maccumulate-outgoing-args
21056 Reserve space once for outgoing arguments in the function prologue rather
21057 than around each call. Generally beneficial for performance and size. Also
21058 needed for unwinding to avoid changing the stack frame around conditional code.
21060 @item -mdivsi3_libfunc=@var{name}
21061 @opindex mdivsi3_libfunc=@var{name}
21062 Set the name of the library function used for 32-bit signed division to
21064 This only affects the name used in the @samp{call} and @samp{inv:call}
21065 division strategies, and the compiler still expects the same
21066 sets of input/output/clobbered registers as if this option were not present.
21068 @item -mfixed-range=@var{register-range}
21069 @opindex mfixed-range
21070 Generate code treating the given register range as fixed registers.
21071 A fixed register is one that the register allocator can not use. This is
21072 useful when compiling kernel code. A register range is specified as
21073 two registers separated by a dash. Multiple register ranges can be
21074 specified separated by a comma.
21076 @item -mindexed-addressing
21077 @opindex mindexed-addressing
21078 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
21079 This is only safe if the hardware and/or OS implement 32-bit wrap-around
21080 semantics for the indexed addressing mode. The architecture allows the
21081 implementation of processors with 64-bit MMU, which the OS could use to
21082 get 32-bit addressing, but since no current hardware implementation supports
21083 this or any other way to make the indexed addressing mode safe to use in
21084 the 32-bit ABI, the default is @option{-mno-indexed-addressing}.
21086 @item -mgettrcost=@var{number}
21087 @opindex mgettrcost=@var{number}
21088 Set the cost assumed for the @code{gettr} instruction to @var{number}.
21089 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
21093 Assume @code{pt*} instructions won't trap. This generally generates
21094 better-scheduled code, but is unsafe on current hardware.
21095 The current architecture
21096 definition says that @code{ptabs} and @code{ptrel} trap when the target
21098 This has the unintentional effect of making it unsafe to schedule these
21099 instructions before a branch, or hoist them out of a loop. For example,
21100 @code{__do_global_ctors}, a part of @file{libgcc}
21101 that runs constructors at program
21102 startup, calls functions in a list which is delimited by @minus{}1. With the
21103 @option{-mpt-fixed} option, the @code{ptabs} is done before testing against @minus{}1.
21104 That means that all the constructors run a bit more quickly, but when
21105 the loop comes to the end of the list, the program crashes because @code{ptabs}
21106 loads @minus{}1 into a target register.
21108 Since this option is unsafe for any
21109 hardware implementing the current architecture specification, the default
21110 is @option{-mno-pt-fixed}. Unless specified explicitly with
21111 @option{-mgettrcost}, @option{-mno-pt-fixed} also implies @option{-mgettrcost=100};
21112 this deters register allocation from using target registers for storing
21115 @item -minvalid-symbols
21116 @opindex minvalid-symbols
21117 Assume symbols might be invalid. Ordinary function symbols generated by
21118 the compiler are always valid to load with
21119 @code{movi}/@code{shori}/@code{ptabs} or
21120 @code{movi}/@code{shori}/@code{ptrel},
21121 but with assembler and/or linker tricks it is possible
21122 to generate symbols that cause @code{ptabs} or @code{ptrel} to trap.
21123 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
21124 It prevents cross-basic-block CSE, hoisting and most scheduling
21125 of symbol loads. The default is @option{-mno-invalid-symbols}.
21127 @item -mbranch-cost=@var{num}
21128 @opindex mbranch-cost=@var{num}
21129 Assume @var{num} to be the cost for a branch instruction. Higher numbers
21130 make the compiler try to generate more branch-free code if possible.
21131 If not specified the value is selected depending on the processor type that
21132 is being compiled for.
21135 @itemx -mno-zdcbranch
21136 @opindex mzdcbranch
21137 @opindex mno-zdcbranch
21138 Assume (do not assume) that zero displacement conditional branch instructions
21139 @code{bt} and @code{bf} are fast. If @option{-mzdcbranch} is specified, the
21140 compiler prefers zero displacement branch code sequences. This is
21141 enabled by default when generating code for SH4 and SH4A. It can be explicitly
21142 disabled by specifying @option{-mno-zdcbranch}.
21144 @item -mcbranch-force-delay-slot
21145 @opindex mcbranch-force-delay-slot
21146 Force the usage of delay slots for conditional branches, which stuffs the delay
21147 slot with a @code{nop} if a suitable instruction can't be found. By default
21148 this option is disabled. It can be enabled to work around hardware bugs as
21149 found in the original SH7055.
21152 @itemx -mno-fused-madd
21153 @opindex mfused-madd
21154 @opindex mno-fused-madd
21155 Generate code that uses (does not use) the floating-point multiply and
21156 accumulate instructions. These instructions are generated by default
21157 if hardware floating point is used. The machine-dependent
21158 @option{-mfused-madd} option is now mapped to the machine-independent
21159 @option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
21160 mapped to @option{-ffp-contract=off}.
21166 Allow or disallow the compiler to emit the @code{fsca} instruction for sine
21167 and cosine approximations. The option @option{-mfsca} must be used in
21168 combination with @option{-funsafe-math-optimizations}. It is enabled by default
21169 when generating code for SH4A. Using @option{-mno-fsca} disables sine and cosine
21170 approximations even if @option{-funsafe-math-optimizations} is in effect.
21176 Allow or disallow the compiler to emit the @code{fsrra} instruction for
21177 reciprocal square root approximations. The option @option{-mfsrra} must be used
21178 in combination with @option{-funsafe-math-optimizations} and
21179 @option{-ffinite-math-only}. It is enabled by default when generating code for
21180 SH4A. Using @option{-mno-fsrra} disables reciprocal square root approximations
21181 even if @option{-funsafe-math-optimizations} and @option{-ffinite-math-only} are
21184 @item -mpretend-cmove
21185 @opindex mpretend-cmove
21186 Prefer zero-displacement conditional branches for conditional move instruction
21187 patterns. This can result in faster code on the SH4 processor.
21191 @node Solaris 2 Options
21192 @subsection Solaris 2 Options
21193 @cindex Solaris 2 options
21195 These @samp{-m} options are supported on Solaris 2:
21198 @item -mclear-hwcap
21199 @opindex mclear-hwcap
21200 @option{-mclear-hwcap} tells the compiler to remove the hardware
21201 capabilities generated by the Solaris assembler. This is only necessary
21202 when object files use ISA extensions not supported by the current
21203 machine, but check at runtime whether or not to use them.
21205 @item -mimpure-text
21206 @opindex mimpure-text
21207 @option{-mimpure-text}, used in addition to @option{-shared}, tells
21208 the compiler to not pass @option{-z text} to the linker when linking a
21209 shared object. Using this option, you can link position-dependent
21210 code into a shared object.
21212 @option{-mimpure-text} suppresses the ``relocations remain against
21213 allocatable but non-writable sections'' linker error message.
21214 However, the necessary relocations trigger copy-on-write, and the
21215 shared object is not actually shared across processes. Instead of
21216 using @option{-mimpure-text}, you should compile all source code with
21217 @option{-fpic} or @option{-fPIC}.
21221 These switches are supported in addition to the above on Solaris 2:
21226 Add support for multithreading using the POSIX threads library. This
21227 option sets flags for both the preprocessor and linker. This option does
21228 not affect the thread safety of object code produced by the compiler or
21229 that of libraries supplied with it.
21233 This is a synonym for @option{-pthreads}.
21236 @node SPARC Options
21237 @subsection SPARC Options
21238 @cindex SPARC options
21240 These @samp{-m} options are supported on the SPARC:
21243 @item -mno-app-regs
21245 @opindex mno-app-regs
21247 Specify @option{-mapp-regs} to generate output using the global registers
21248 2 through 4, which the SPARC SVR4 ABI reserves for applications. Like the
21249 global register 1, each global register 2 through 4 is then treated as an
21250 allocable register that is clobbered by function calls. This is the default.
21252 To be fully SVR4 ABI-compliant at the cost of some performance loss,
21253 specify @option{-mno-app-regs}. You should compile libraries and system
21254 software with this option.
21260 With @option{-mflat}, the compiler does not generate save/restore instructions
21261 and uses a ``flat'' or single register window model. This model is compatible
21262 with the regular register window model. The local registers and the input
21263 registers (0--5) are still treated as ``call-saved'' registers and are
21264 saved on the stack as needed.
21266 With @option{-mno-flat} (the default), the compiler generates save/restore
21267 instructions (except for leaf functions). This is the normal operating mode.
21270 @itemx -mhard-float
21272 @opindex mhard-float
21273 Generate output containing floating-point instructions. This is the
21277 @itemx -msoft-float
21279 @opindex msoft-float
21280 Generate output containing library calls for floating point.
21281 @strong{Warning:} the requisite libraries are not available for all SPARC
21282 targets. Normally the facilities of the machine's usual C compiler are
21283 used, but this cannot be done directly in cross-compilation. You must make
21284 your own arrangements to provide suitable library functions for
21285 cross-compilation. The embedded targets @samp{sparc-*-aout} and
21286 @samp{sparclite-*-*} do provide software floating-point support.
21288 @option{-msoft-float} changes the calling convention in the output file;
21289 therefore, it is only useful if you compile @emph{all} of a program with
21290 this option. In particular, you need to compile @file{libgcc.a}, the
21291 library that comes with GCC, with @option{-msoft-float} in order for
21294 @item -mhard-quad-float
21295 @opindex mhard-quad-float
21296 Generate output containing quad-word (long double) floating-point
21299 @item -msoft-quad-float
21300 @opindex msoft-quad-float
21301 Generate output containing library calls for quad-word (long double)
21302 floating-point instructions. The functions called are those specified
21303 in the SPARC ABI@. This is the default.
21305 As of this writing, there are no SPARC implementations that have hardware
21306 support for the quad-word floating-point instructions. They all invoke
21307 a trap handler for one of these instructions, and then the trap handler
21308 emulates the effect of the instruction. Because of the trap handler overhead,
21309 this is much slower than calling the ABI library routines. Thus the
21310 @option{-msoft-quad-float} option is the default.
21312 @item -mno-unaligned-doubles
21313 @itemx -munaligned-doubles
21314 @opindex mno-unaligned-doubles
21315 @opindex munaligned-doubles
21316 Assume that doubles have 8-byte alignment. This is the default.
21318 With @option{-munaligned-doubles}, GCC assumes that doubles have 8-byte
21319 alignment only if they are contained in another type, or if they have an
21320 absolute address. Otherwise, it assumes they have 4-byte alignment.
21321 Specifying this option avoids some rare compatibility problems with code
21322 generated by other compilers. It is not the default because it results
21323 in a performance loss, especially for floating-point code.
21326 @itemx -mno-user-mode
21327 @opindex muser-mode
21328 @opindex mno-user-mode
21329 Do not generate code that can only run in supervisor mode. This is relevant
21330 only for the @code{casa} instruction emitted for the LEON3 processor. The
21331 default is @option{-mno-user-mode}.
21333 @item -mno-faster-structs
21334 @itemx -mfaster-structs
21335 @opindex mno-faster-structs
21336 @opindex mfaster-structs
21337 With @option{-mfaster-structs}, the compiler assumes that structures
21338 should have 8-byte alignment. This enables the use of pairs of
21339 @code{ldd} and @code{std} instructions for copies in structure
21340 assignment, in place of twice as many @code{ld} and @code{st} pairs.
21341 However, the use of this changed alignment directly violates the SPARC
21342 ABI@. Thus, it's intended only for use on targets where the developer
21343 acknowledges that their resulting code is not directly in line with
21344 the rules of the ABI@.
21346 @item -mcpu=@var{cpu_type}
21348 Set the instruction set, register set, and instruction scheduling parameters
21349 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
21350 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{hypersparc},
21351 @samp{leon}, @samp{leon3}, @samp{leon3v7}, @samp{sparclite}, @samp{f930},
21352 @samp{f934}, @samp{sparclite86x}, @samp{sparclet}, @samp{tsc701}, @samp{v9},
21353 @samp{ultrasparc}, @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2},
21354 @samp{niagara3} and @samp{niagara4}.
21356 Native Solaris and GNU/Linux toolchains also support the value @samp{native},
21357 which selects the best architecture option for the host processor.
21358 @option{-mcpu=native} has no effect if GCC does not recognize
21361 Default instruction scheduling parameters are used for values that select
21362 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
21363 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
21365 Here is a list of each supported architecture and their supported
21373 supersparc, hypersparc, leon, leon3
21376 f930, f934, sparclite86x
21382 ultrasparc, ultrasparc3, niagara, niagara2, niagara3, niagara4
21385 By default (unless configured otherwise), GCC generates code for the V7
21386 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
21387 additionally optimizes it for the Cypress CY7C602 chip, as used in the
21388 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
21389 SPARCStation 1, 2, IPX etc.
21391 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
21392 architecture. The only difference from V7 code is that the compiler emits
21393 the integer multiply and integer divide instructions which exist in SPARC-V8
21394 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
21395 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
21398 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
21399 the SPARC architecture. This adds the integer multiply, integer divide step
21400 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
21401 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
21402 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
21403 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
21404 MB86934 chip, which is the more recent SPARClite with FPU@.
21406 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
21407 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
21408 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
21409 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
21410 optimizes it for the TEMIC SPARClet chip.
21412 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
21413 architecture. This adds 64-bit integer and floating-point move instructions,
21414 3 additional floating-point condition code registers and conditional move
21415 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
21416 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
21417 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
21418 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
21419 @option{-mcpu=niagara}, the compiler additionally optimizes it for
21420 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
21421 additionally optimizes it for Sun UltraSPARC T2 chips. With
21422 @option{-mcpu=niagara3}, the compiler additionally optimizes it for Sun
21423 UltraSPARC T3 chips. With @option{-mcpu=niagara4}, the compiler
21424 additionally optimizes it for Sun UltraSPARC T4 chips.
21426 @item -mtune=@var{cpu_type}
21428 Set the instruction scheduling parameters for machine type
21429 @var{cpu_type}, but do not set the instruction set or register set that the
21430 option @option{-mcpu=@var{cpu_type}} does.
21432 The same values for @option{-mcpu=@var{cpu_type}} can be used for
21433 @option{-mtune=@var{cpu_type}}, but the only useful values are those
21434 that select a particular CPU implementation. Those are @samp{cypress},
21435 @samp{supersparc}, @samp{hypersparc}, @samp{leon}, @samp{leon3},
21436 @samp{leon3v7}, @samp{f930}, @samp{f934}, @samp{sparclite86x}, @samp{tsc701},
21437 @samp{ultrasparc}, @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2},
21438 @samp{niagara3} and @samp{niagara4}. With native Solaris and GNU/Linux
21439 toolchains, @samp{native} can also be used.
21444 @opindex mno-v8plus
21445 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
21446 difference from the V8 ABI is that the global and out registers are
21447 considered 64 bits wide. This is enabled by default on Solaris in 32-bit
21448 mode for all SPARC-V9 processors.
21454 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
21455 Visual Instruction Set extensions. The default is @option{-mno-vis}.
21461 With @option{-mvis2}, GCC generates code that takes advantage of
21462 version 2.0 of the UltraSPARC Visual Instruction Set extensions. The
21463 default is @option{-mvis2} when targeting a cpu that supports such
21464 instructions, such as UltraSPARC-III and later. Setting @option{-mvis2}
21465 also sets @option{-mvis}.
21471 With @option{-mvis3}, GCC generates code that takes advantage of
21472 version 3.0 of the UltraSPARC Visual Instruction Set extensions. The
21473 default is @option{-mvis3} when targeting a cpu that supports such
21474 instructions, such as niagara-3 and later. Setting @option{-mvis3}
21475 also sets @option{-mvis2} and @option{-mvis}.
21480 @opindex mno-cbcond
21481 With @option{-mcbcond}, GCC generates code that takes advantage of
21482 compare-and-branch instructions, as defined in the Sparc Architecture 2011.
21483 The default is @option{-mcbcond} when targeting a cpu that supports such
21484 instructions, such as niagara-4 and later.
21490 With @option{-mpopc}, GCC generates code that takes advantage of the UltraSPARC
21491 population count instruction. The default is @option{-mpopc}
21492 when targeting a cpu that supports such instructions, such as Niagara-2 and
21499 With @option{-mfmaf}, GCC generates code that takes advantage of the UltraSPARC
21500 Fused Multiply-Add Floating-point extensions. The default is @option{-mfmaf}
21501 when targeting a cpu that supports such instructions, such as Niagara-3 and
21505 @opindex mfix-at697f
21506 Enable the documented workaround for the single erratum of the Atmel AT697F
21507 processor (which corresponds to erratum #13 of the AT697E processor).
21510 @opindex mfix-ut699
21511 Enable the documented workarounds for the floating-point errata and the data
21512 cache nullify errata of the UT699 processor.
21515 These @samp{-m} options are supported in addition to the above
21516 on SPARC-V9 processors in 64-bit environments:
21523 Generate code for a 32-bit or 64-bit environment.
21524 The 32-bit environment sets int, long and pointer to 32 bits.
21525 The 64-bit environment sets int to 32 bits and long and pointer
21528 @item -mcmodel=@var{which}
21530 Set the code model to one of
21534 The Medium/Low code model: 64-bit addresses, programs
21535 must be linked in the low 32 bits of memory. Programs can be statically
21536 or dynamically linked.
21539 The Medium/Middle code model: 64-bit addresses, programs
21540 must be linked in the low 44 bits of memory, the text and data segments must
21541 be less than 2GB in size and the data segment must be located within 2GB of
21545 The Medium/Anywhere code model: 64-bit addresses, programs
21546 may be linked anywhere in memory, the text and data segments must be less
21547 than 2GB in size and the data segment must be located within 2GB of the
21551 The Medium/Anywhere code model for embedded systems:
21552 64-bit addresses, the text and data segments must be less than 2GB in
21553 size, both starting anywhere in memory (determined at link time). The
21554 global register %g4 points to the base of the data segment. Programs
21555 are statically linked and PIC is not supported.
21558 @item -mmemory-model=@var{mem-model}
21559 @opindex mmemory-model
21560 Set the memory model in force on the processor to one of
21564 The default memory model for the processor and operating system.
21567 Relaxed Memory Order
21570 Partial Store Order
21576 Sequential Consistency
21579 These memory models are formally defined in Appendix D of the Sparc V9
21580 architecture manual, as set in the processor's @code{PSTATE.MM} field.
21583 @itemx -mno-stack-bias
21584 @opindex mstack-bias
21585 @opindex mno-stack-bias
21586 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
21587 frame pointer if present, are offset by @minus{}2047 which must be added back
21588 when making stack frame references. This is the default in 64-bit mode.
21589 Otherwise, assume no such offset is present.
21593 @subsection SPU Options
21594 @cindex SPU options
21596 These @samp{-m} options are supported on the SPU:
21600 @itemx -merror-reloc
21601 @opindex mwarn-reloc
21602 @opindex merror-reloc
21604 The loader for SPU does not handle dynamic relocations. By default, GCC
21605 gives an error when it generates code that requires a dynamic
21606 relocation. @option{-mno-error-reloc} disables the error,
21607 @option{-mwarn-reloc} generates a warning instead.
21610 @itemx -munsafe-dma
21612 @opindex munsafe-dma
21614 Instructions that initiate or test completion of DMA must not be
21615 reordered with respect to loads and stores of the memory that is being
21617 With @option{-munsafe-dma} you must use the @code{volatile} keyword to protect
21618 memory accesses, but that can lead to inefficient code in places where the
21619 memory is known to not change. Rather than mark the memory as volatile,
21620 you can use @option{-msafe-dma} to tell the compiler to treat
21621 the DMA instructions as potentially affecting all memory.
21623 @item -mbranch-hints
21624 @opindex mbranch-hints
21626 By default, GCC generates a branch hint instruction to avoid
21627 pipeline stalls for always-taken or probably-taken branches. A hint
21628 is not generated closer than 8 instructions away from its branch.
21629 There is little reason to disable them, except for debugging purposes,
21630 or to make an object a little bit smaller.
21634 @opindex msmall-mem
21635 @opindex mlarge-mem
21637 By default, GCC generates code assuming that addresses are never larger
21638 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
21639 a full 32-bit address.
21644 By default, GCC links against startup code that assumes the SPU-style
21645 main function interface (which has an unconventional parameter list).
21646 With @option{-mstdmain}, GCC links your program against startup
21647 code that assumes a C99-style interface to @code{main}, including a
21648 local copy of @code{argv} strings.
21650 @item -mfixed-range=@var{register-range}
21651 @opindex mfixed-range
21652 Generate code treating the given register range as fixed registers.
21653 A fixed register is one that the register allocator cannot use. This is
21654 useful when compiling kernel code. A register range is specified as
21655 two registers separated by a dash. Multiple register ranges can be
21656 specified separated by a comma.
21662 Compile code assuming that pointers to the PPU address space accessed
21663 via the @code{__ea} named address space qualifier are either 32 or 64
21664 bits wide. The default is 32 bits. As this is an ABI-changing option,
21665 all object code in an executable must be compiled with the same setting.
21667 @item -maddress-space-conversion
21668 @itemx -mno-address-space-conversion
21669 @opindex maddress-space-conversion
21670 @opindex mno-address-space-conversion
21671 Allow/disallow treating the @code{__ea} address space as superset
21672 of the generic address space. This enables explicit type casts
21673 between @code{__ea} and generic pointer as well as implicit
21674 conversions of generic pointers to @code{__ea} pointers. The
21675 default is to allow address space pointer conversions.
21677 @item -mcache-size=@var{cache-size}
21678 @opindex mcache-size
21679 This option controls the version of libgcc that the compiler links to an
21680 executable and selects a software-managed cache for accessing variables
21681 in the @code{__ea} address space with a particular cache size. Possible
21682 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
21683 and @samp{128}. The default cache size is 64KB.
21685 @item -matomic-updates
21686 @itemx -mno-atomic-updates
21687 @opindex matomic-updates
21688 @opindex mno-atomic-updates
21689 This option controls the version of libgcc that the compiler links to an
21690 executable and selects whether atomic updates to the software-managed
21691 cache of PPU-side variables are used. If you use atomic updates, changes
21692 to a PPU variable from SPU code using the @code{__ea} named address space
21693 qualifier do not interfere with changes to other PPU variables residing
21694 in the same cache line from PPU code. If you do not use atomic updates,
21695 such interference may occur; however, writing back cache lines is
21696 more efficient. The default behavior is to use atomic updates.
21699 @itemx -mdual-nops=@var{n}
21700 @opindex mdual-nops
21701 By default, GCC inserts nops to increase dual issue when it expects
21702 it to increase performance. @var{n} can be a value from 0 to 10. A
21703 smaller @var{n} inserts fewer nops. 10 is the default, 0 is the
21704 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
21706 @item -mhint-max-nops=@var{n}
21707 @opindex mhint-max-nops
21708 Maximum number of nops to insert for a branch hint. A branch hint must
21709 be at least 8 instructions away from the branch it is affecting. GCC
21710 inserts up to @var{n} nops to enforce this, otherwise it does not
21711 generate the branch hint.
21713 @item -mhint-max-distance=@var{n}
21714 @opindex mhint-max-distance
21715 The encoding of the branch hint instruction limits the hint to be within
21716 256 instructions of the branch it is affecting. By default, GCC makes
21717 sure it is within 125.
21720 @opindex msafe-hints
21721 Work around a hardware bug that causes the SPU to stall indefinitely.
21722 By default, GCC inserts the @code{hbrp} instruction to make sure
21723 this stall won't happen.
21727 @node System V Options
21728 @subsection Options for System V
21730 These additional options are available on System V Release 4 for
21731 compatibility with other compilers on those systems:
21736 Create a shared object.
21737 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
21741 Identify the versions of each tool used by the compiler, in a
21742 @code{.ident} assembler directive in the output.
21746 Refrain from adding @code{.ident} directives to the output file (this is
21749 @item -YP,@var{dirs}
21751 Search the directories @var{dirs}, and no others, for libraries
21752 specified with @option{-l}.
21754 @item -Ym,@var{dir}
21756 Look in the directory @var{dir} to find the M4 preprocessor.
21757 The assembler uses this option.
21758 @c This is supposed to go with a -Yd for predefined M4 macro files, but
21759 @c the generic assembler that comes with Solaris takes just -Ym.
21762 @node TILE-Gx Options
21763 @subsection TILE-Gx Options
21764 @cindex TILE-Gx options
21766 These @samp{-m} options are supported on the TILE-Gx:
21769 @item -mcmodel=small
21770 @opindex mcmodel=small
21771 Generate code for the small model. The distance for direct calls is
21772 limited to 500M in either direction. PC-relative addresses are 32
21773 bits. Absolute addresses support the full address range.
21775 @item -mcmodel=large
21776 @opindex mcmodel=large
21777 Generate code for the large model. There is no limitation on call
21778 distance, pc-relative addresses, or absolute addresses.
21780 @item -mcpu=@var{name}
21782 Selects the type of CPU to be targeted. Currently the only supported
21783 type is @samp{tilegx}.
21789 Generate code for a 32-bit or 64-bit environment. The 32-bit
21790 environment sets int, long, and pointer to 32 bits. The 64-bit
21791 environment sets int to 32 bits and long and pointer to 64 bits.
21794 @itemx -mlittle-endian
21795 @opindex mbig-endian
21796 @opindex mlittle-endian
21797 Generate code in big/little endian mode, respectively.
21800 @node TILEPro Options
21801 @subsection TILEPro Options
21802 @cindex TILEPro options
21804 These @samp{-m} options are supported on the TILEPro:
21807 @item -mcpu=@var{name}
21809 Selects the type of CPU to be targeted. Currently the only supported
21810 type is @samp{tilepro}.
21814 Generate code for a 32-bit environment, which sets int, long, and
21815 pointer to 32 bits. This is the only supported behavior so the flag
21816 is essentially ignored.
21820 @subsection V850 Options
21821 @cindex V850 Options
21823 These @samp{-m} options are defined for V850 implementations:
21827 @itemx -mno-long-calls
21828 @opindex mlong-calls
21829 @opindex mno-long-calls
21830 Treat all calls as being far away (near). If calls are assumed to be
21831 far away, the compiler always loads the function's address into a
21832 register, and calls indirect through the pointer.
21838 Do not optimize (do optimize) basic blocks that use the same index
21839 pointer 4 or more times to copy pointer into the @code{ep} register, and
21840 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
21841 option is on by default if you optimize.
21843 @item -mno-prolog-function
21844 @itemx -mprolog-function
21845 @opindex mno-prolog-function
21846 @opindex mprolog-function
21847 Do not use (do use) external functions to save and restore registers
21848 at the prologue and epilogue of a function. The external functions
21849 are slower, but use less code space if more than one function saves
21850 the same number of registers. The @option{-mprolog-function} option
21851 is on by default if you optimize.
21855 Try to make the code as small as possible. At present, this just turns
21856 on the @option{-mep} and @option{-mprolog-function} options.
21858 @item -mtda=@var{n}
21860 Put static or global variables whose size is @var{n} bytes or less into
21861 the tiny data area that register @code{ep} points to. The tiny data
21862 area can hold up to 256 bytes in total (128 bytes for byte references).
21864 @item -msda=@var{n}
21866 Put static or global variables whose size is @var{n} bytes or less into
21867 the small data area that register @code{gp} points to. The small data
21868 area can hold up to 64 kilobytes.
21870 @item -mzda=@var{n}
21872 Put static or global variables whose size is @var{n} bytes or less into
21873 the first 32 kilobytes of memory.
21877 Specify that the target processor is the V850.
21881 Specify that the target processor is the V850E3V5. The preprocessor
21882 constant @code{__v850e3v5__} is defined if this option is used.
21886 Specify that the target processor is the V850E3V5. This is an alias for
21887 the @option{-mv850e3v5} option.
21891 Specify that the target processor is the V850E2V3. The preprocessor
21892 constant @code{__v850e2v3__} is defined if this option is used.
21896 Specify that the target processor is the V850E2. The preprocessor
21897 constant @code{__v850e2__} is defined if this option is used.
21901 Specify that the target processor is the V850E1. The preprocessor
21902 constants @code{__v850e1__} and @code{__v850e__} are defined if
21903 this option is used.
21907 Specify that the target processor is the V850ES. This is an alias for
21908 the @option{-mv850e1} option.
21912 Specify that the target processor is the V850E@. The preprocessor
21913 constant @code{__v850e__} is defined if this option is used.
21915 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
21916 nor @option{-mv850e2} nor @option{-mv850e2v3} nor @option{-mv850e3v5}
21917 are defined then a default target processor is chosen and the
21918 relevant @samp{__v850*__} preprocessor constant is defined.
21920 The preprocessor constants @code{__v850} and @code{__v851__} are always
21921 defined, regardless of which processor variant is the target.
21923 @item -mdisable-callt
21924 @itemx -mno-disable-callt
21925 @opindex mdisable-callt
21926 @opindex mno-disable-callt
21927 This option suppresses generation of the @code{CALLT} instruction for the
21928 v850e, v850e1, v850e2, v850e2v3 and v850e3v5 flavors of the v850
21931 This option is enabled by default when the RH850 ABI is
21932 in use (see @option{-mrh850-abi}), and disabled by default when the
21933 GCC ABI is in use. If @code{CALLT} instructions are being generated
21934 then the C preprocessor symbol @code{__V850_CALLT__} is defined.
21940 Pass on (or do not pass on) the @option{-mrelax} command-line option
21944 @itemx -mno-long-jumps
21945 @opindex mlong-jumps
21946 @opindex mno-long-jumps
21947 Disable (or re-enable) the generation of PC-relative jump instructions.
21950 @itemx -mhard-float
21951 @opindex msoft-float
21952 @opindex mhard-float
21953 Disable (or re-enable) the generation of hardware floating point
21954 instructions. This option is only significant when the target
21955 architecture is @samp{V850E2V3} or higher. If hardware floating point
21956 instructions are being generated then the C preprocessor symbol
21957 @code{__FPU_OK__} is defined, otherwise the symbol
21958 @code{__NO_FPU__} is defined.
21962 Enables the use of the e3v5 LOOP instruction. The use of this
21963 instruction is not enabled by default when the e3v5 architecture is
21964 selected because its use is still experimental.
21968 @opindex mrh850-abi
21970 Enables support for the RH850 version of the V850 ABI. This is the
21971 default. With this version of the ABI the following rules apply:
21975 Integer sized structures and unions are returned via a memory pointer
21976 rather than a register.
21979 Large structures and unions (more than 8 bytes in size) are passed by
21983 Functions are aligned to 16-bit boundaries.
21986 The @option{-m8byte-align} command-line option is supported.
21989 The @option{-mdisable-callt} command-line option is enabled by
21990 default. The @option{-mno-disable-callt} command-line option is not
21994 When this version of the ABI is enabled the C preprocessor symbol
21995 @code{__V850_RH850_ABI__} is defined.
21999 Enables support for the old GCC version of the V850 ABI. With this
22000 version of the ABI the following rules apply:
22004 Integer sized structures and unions are returned in register @code{r10}.
22007 Large structures and unions (more than 8 bytes in size) are passed by
22011 Functions are aligned to 32-bit boundaries, unless optimizing for
22015 The @option{-m8byte-align} command-line option is not supported.
22018 The @option{-mdisable-callt} command-line option is supported but not
22019 enabled by default.
22022 When this version of the ABI is enabled the C preprocessor symbol
22023 @code{__V850_GCC_ABI__} is defined.
22025 @item -m8byte-align
22026 @itemx -mno-8byte-align
22027 @opindex m8byte-align
22028 @opindex mno-8byte-align
22029 Enables support for @code{double} and @code{long long} types to be
22030 aligned on 8-byte boundaries. The default is to restrict the
22031 alignment of all objects to at most 4-bytes. When
22032 @option{-m8byte-align} is in effect the C preprocessor symbol
22033 @code{__V850_8BYTE_ALIGN__} is defined.
22036 @opindex mbig-switch
22037 Generate code suitable for big switch tables. Use this option only if
22038 the assembler/linker complain about out of range branches within a switch
22043 This option causes r2 and r5 to be used in the code generated by
22044 the compiler. This setting is the default.
22046 @item -mno-app-regs
22047 @opindex mno-app-regs
22048 This option causes r2 and r5 to be treated as fixed registers.
22053 @subsection VAX Options
22054 @cindex VAX options
22056 These @samp{-m} options are defined for the VAX:
22061 Do not output certain jump instructions (@code{aobleq} and so on)
22062 that the Unix assembler for the VAX cannot handle across long
22067 Do output those jump instructions, on the assumption that the
22068 GNU assembler is being used.
22072 Output code for G-format floating-point numbers instead of D-format.
22075 @node Visium Options
22076 @subsection Visium Options
22077 @cindex Visium options
22083 A program which performs file I/O and is destined to run on an MCM target
22084 should be linked with this option. It causes the libraries libc.a and
22085 libdebug.a to be linked. The program should be run on the target under
22086 the control of the GDB remote debugging stub.
22090 A program which performs file I/O and is destined to run on the simulator
22091 should be linked with option. This causes libraries libc.a and libsim.a to
22095 @itemx -mhard-float
22097 @opindex mhard-float
22098 Generate code containing floating-point instructions. This is the
22102 @itemx -msoft-float
22104 @opindex msoft-float
22105 Generate code containing library calls for floating-point.
22107 @option{-msoft-float} changes the calling convention in the output file;
22108 therefore, it is only useful if you compile @emph{all} of a program with
22109 this option. In particular, you need to compile @file{libgcc.a}, the
22110 library that comes with GCC, with @option{-msoft-float} in order for
22113 @item -mcpu=@var{cpu_type}
22115 Set the instruction set, register set, and instruction scheduling parameters
22116 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
22117 @samp{mcm}, @samp{gr5} and @samp{gr6}.
22119 @samp{mcm} is a synonym of @samp{gr5} present for backward compatibility.
22121 By default (unless configured otherwise), GCC generates code for the GR5
22122 variant of the Visium architecture.
22124 With @option{-mcpu=gr6}, GCC generates code for the GR6 variant of the Visium
22125 architecture. The only difference from GR5 code is that the compiler will
22126 generate block move instructions.
22128 @item -mtune=@var{cpu_type}
22130 Set the instruction scheduling parameters for machine type @var{cpu_type},
22131 but do not set the instruction set or register set that the option
22132 @option{-mcpu=@var{cpu_type}} would.
22136 Generate code for the supervisor mode, where there are no restrictions on
22137 the access to general registers. This is the default.
22140 @opindex muser-mode
22141 Generate code for the user mode, where the access to some general registers
22142 is forbidden: on the GR5, registers r24 to r31 cannot be accessed in this
22143 mode; on the GR6, only registers r29 to r31 are affected.
22147 @subsection VMS Options
22149 These @samp{-m} options are defined for the VMS implementations:
22152 @item -mvms-return-codes
22153 @opindex mvms-return-codes
22154 Return VMS condition codes from @code{main}. The default is to return POSIX-style
22155 condition (e.g.@ error) codes.
22157 @item -mdebug-main=@var{prefix}
22158 @opindex mdebug-main=@var{prefix}
22159 Flag the first routine whose name starts with @var{prefix} as the main
22160 routine for the debugger.
22164 Default to 64-bit memory allocation routines.
22166 @item -mpointer-size=@var{size}
22167 @opindex mpointer-size=@var{size}
22168 Set the default size of pointers. Possible options for @var{size} are
22169 @samp{32} or @samp{short} for 32 bit pointers, @samp{64} or @samp{long}
22170 for 64 bit pointers, and @samp{no} for supporting only 32 bit pointers.
22171 The later option disables @code{pragma pointer_size}.
22174 @node VxWorks Options
22175 @subsection VxWorks Options
22176 @cindex VxWorks Options
22178 The options in this section are defined for all VxWorks targets.
22179 Options specific to the target hardware are listed with the other
22180 options for that target.
22185 GCC can generate code for both VxWorks kernels and real time processes
22186 (RTPs). This option switches from the former to the latter. It also
22187 defines the preprocessor macro @code{__RTP__}.
22190 @opindex non-static
22191 Link an RTP executable against shared libraries rather than static
22192 libraries. The options @option{-static} and @option{-shared} can
22193 also be used for RTPs (@pxref{Link Options}); @option{-static}
22200 These options are passed down to the linker. They are defined for
22201 compatibility with Diab.
22204 @opindex Xbind-lazy
22205 Enable lazy binding of function calls. This option is equivalent to
22206 @option{-Wl,-z,now} and is defined for compatibility with Diab.
22210 Disable lazy binding of function calls. This option is the default and
22211 is defined for compatibility with Diab.
22215 @subsection x86 Options
22216 @cindex x86 Options
22218 These @samp{-m} options are defined for the x86 family of computers.
22222 @item -march=@var{cpu-type}
22224 Generate instructions for the machine type @var{cpu-type}. In contrast to
22225 @option{-mtune=@var{cpu-type}}, which merely tunes the generated code
22226 for the specified @var{cpu-type}, @option{-march=@var{cpu-type}} allows GCC
22227 to generate code that may not run at all on processors other than the one
22228 indicated. Specifying @option{-march=@var{cpu-type}} implies
22229 @option{-mtune=@var{cpu-type}}.
22231 The choices for @var{cpu-type} are:
22235 This selects the CPU to generate code for at compilation time by determining
22236 the processor type of the compiling machine. Using @option{-march=native}
22237 enables all instruction subsets supported by the local machine (hence
22238 the result might not run on different machines). Using @option{-mtune=native}
22239 produces code optimized for the local machine under the constraints
22240 of the selected instruction set.
22243 Original Intel i386 CPU@.
22246 Intel i486 CPU@. (No scheduling is implemented for this chip.)
22250 Intel Pentium CPU with no MMX support.
22253 Intel Pentium MMX CPU, based on Pentium core with MMX instruction set support.
22256 Intel Pentium Pro CPU@.
22259 When used with @option{-march}, the Pentium Pro
22260 instruction set is used, so the code runs on all i686 family chips.
22261 When used with @option{-mtune}, it has the same meaning as @samp{generic}.
22264 Intel Pentium II CPU, based on Pentium Pro core with MMX instruction set
22269 Intel Pentium III CPU, based on Pentium Pro core with MMX and SSE instruction
22273 Intel Pentium M; low-power version of Intel Pentium III CPU
22274 with MMX, SSE and SSE2 instruction set support. Used by Centrino notebooks.
22278 Intel Pentium 4 CPU with MMX, SSE and SSE2 instruction set support.
22281 Improved version of Intel Pentium 4 CPU with MMX, SSE, SSE2 and SSE3 instruction
22285 Improved version of Intel Pentium 4 CPU with 64-bit extensions, MMX, SSE,
22286 SSE2 and SSE3 instruction set support.
22289 Intel Core 2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
22290 instruction set support.
22293 Intel Nehalem CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
22294 SSE4.1, SSE4.2 and POPCNT instruction set support.
22297 Intel Westmere CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
22298 SSE4.1, SSE4.2, POPCNT, AES and PCLMUL instruction set support.
22301 Intel Sandy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
22302 SSE4.1, SSE4.2, POPCNT, AVX, AES and PCLMUL instruction set support.
22305 Intel Ivy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
22306 SSE4.1, SSE4.2, POPCNT, AVX, AES, PCLMUL, FSGSBASE, RDRND and F16C
22307 instruction set support.
22310 Intel Haswell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
22311 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
22312 BMI, BMI2 and F16C instruction set support.
22315 Intel Broadwell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
22316 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
22317 BMI, BMI2, F16C, RDSEED, ADCX and PREFETCHW instruction set support.
22320 Intel Bonnell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3 and SSSE3
22321 instruction set support.
22324 Intel Silvermont CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
22325 SSE4.1, SSE4.2, POPCNT, AES, PCLMUL and RDRND instruction set support.
22328 Intel Knight's Landing CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
22329 SSSE3, SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
22330 BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, AVX512F, AVX512PF, AVX512ER and
22331 AVX512CD instruction set support.
22334 AMD K6 CPU with MMX instruction set support.
22338 Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
22341 @itemx athlon-tbird
22342 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
22348 Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
22349 instruction set support.
22355 Processors based on the AMD K8 core with x86-64 instruction set support,
22356 including the AMD Opteron, Athlon 64, and Athlon 64 FX processors.
22357 (This supersets MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit
22358 instruction set extensions.)
22361 @itemx opteron-sse3
22362 @itemx athlon64-sse3
22363 Improved versions of AMD K8 cores with SSE3 instruction set support.
22367 CPUs based on AMD Family 10h cores with x86-64 instruction set support. (This
22368 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
22369 instruction set extensions.)
22372 CPUs based on AMD Family 15h cores with x86-64 instruction set support. (This
22373 supersets FMA4, AVX, XOP, LWP, AES, PCL_MUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A,
22374 SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set extensions.)
22376 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
22377 supersets BMI, TBM, F16C, FMA, FMA4, AVX, XOP, LWP, AES, PCL_MUL, CX16, MMX,
22378 SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set
22381 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
22382 supersets BMI, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, XOP, LWP, AES,
22383 PCL_MUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and
22384 64-bit instruction set extensions.
22386 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
22387 supersets BMI, BMI2, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, AVX2, XOP, LWP,
22388 AES, PCL_MUL, CX16, MOVBE, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1,
22389 SSE4.2, ABM and 64-bit instruction set extensions.
22392 CPUs based on AMD Family 14h cores with x86-64 instruction set support. (This
22393 supersets MMX, SSE, SSE2, SSE3, SSSE3, SSE4A, CX16, ABM and 64-bit
22394 instruction set extensions.)
22397 CPUs based on AMD Family 16h cores with x86-64 instruction set support. This
22398 includes MOVBE, F16C, BMI, AVX, PCL_MUL, AES, SSE4.2, SSE4.1, CX16, ABM,
22399 SSE4A, SSSE3, SSE3, SSE2, SSE, MMX and 64-bit instruction set extensions.
22402 IDT WinChip C6 CPU, dealt in same way as i486 with additional MMX instruction
22406 IDT WinChip 2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
22407 instruction set support.
22410 VIA C3 CPU with MMX and 3DNow!@: instruction set support. (No scheduling is
22411 implemented for this chip.)
22414 VIA C3-2 (Nehemiah/C5XL) CPU with MMX and SSE instruction set support.
22416 implemented for this chip.)
22419 AMD Geode embedded processor with MMX and 3DNow!@: instruction set support.
22422 @item -mtune=@var{cpu-type}
22424 Tune to @var{cpu-type} everything applicable about the generated code, except
22425 for the ABI and the set of available instructions.
22426 While picking a specific @var{cpu-type} schedules things appropriately
22427 for that particular chip, the compiler does not generate any code that
22428 cannot run on the default machine type unless you use a
22429 @option{-march=@var{cpu-type}} option.
22430 For example, if GCC is configured for i686-pc-linux-gnu
22431 then @option{-mtune=pentium4} generates code that is tuned for Pentium 4
22432 but still runs on i686 machines.
22434 The choices for @var{cpu-type} are the same as for @option{-march}.
22435 In addition, @option{-mtune} supports 2 extra choices for @var{cpu-type}:
22439 Produce code optimized for the most common IA32/@/AMD64/@/EM64T processors.
22440 If you know the CPU on which your code will run, then you should use
22441 the corresponding @option{-mtune} or @option{-march} option instead of
22442 @option{-mtune=generic}. But, if you do not know exactly what CPU users
22443 of your application will have, then you should use this option.
22445 As new processors are deployed in the marketplace, the behavior of this
22446 option will change. Therefore, if you upgrade to a newer version of
22447 GCC, code generation controlled by this option will change to reflect
22449 that are most common at the time that version of GCC is released.
22451 There is no @option{-march=generic} option because @option{-march}
22452 indicates the instruction set the compiler can use, and there is no
22453 generic instruction set applicable to all processors. In contrast,
22454 @option{-mtune} indicates the processor (or, in this case, collection of
22455 processors) for which the code is optimized.
22458 Produce code optimized for the most current Intel processors, which are
22459 Haswell and Silvermont for this version of GCC. If you know the CPU
22460 on which your code will run, then you should use the corresponding
22461 @option{-mtune} or @option{-march} option instead of @option{-mtune=intel}.
22462 But, if you want your application performs better on both Haswell and
22463 Silvermont, then you should use this option.
22465 As new Intel processors are deployed in the marketplace, the behavior of
22466 this option will change. Therefore, if you upgrade to a newer version of
22467 GCC, code generation controlled by this option will change to reflect
22468 the most current Intel processors at the time that version of GCC is
22471 There is no @option{-march=intel} option because @option{-march} indicates
22472 the instruction set the compiler can use, and there is no common
22473 instruction set applicable to all processors. In contrast,
22474 @option{-mtune} indicates the processor (or, in this case, collection of
22475 processors) for which the code is optimized.
22478 @item -mcpu=@var{cpu-type}
22480 A deprecated synonym for @option{-mtune}.
22482 @item -mfpmath=@var{unit}
22484 Generate floating-point arithmetic for selected unit @var{unit}. The choices
22485 for @var{unit} are:
22489 Use the standard 387 floating-point coprocessor present on the majority of chips and
22490 emulated otherwise. Code compiled with this option runs almost everywhere.
22491 The temporary results are computed in 80-bit precision instead of the precision
22492 specified by the type, resulting in slightly different results compared to most
22493 of other chips. See @option{-ffloat-store} for more detailed description.
22495 This is the default choice for x86-32 targets.
22498 Use scalar floating-point instructions present in the SSE instruction set.
22499 This instruction set is supported by Pentium III and newer chips,
22500 and in the AMD line
22501 by Athlon-4, Athlon XP and Athlon MP chips. The earlier version of the SSE
22502 instruction set supports only single-precision arithmetic, thus the double and
22503 extended-precision arithmetic are still done using 387. A later version, present
22504 only in Pentium 4 and AMD x86-64 chips, supports double-precision
22507 For the x86-32 compiler, you must use @option{-march=@var{cpu-type}}, @option{-msse}
22508 or @option{-msse2} switches to enable SSE extensions and make this option
22509 effective. For the x86-64 compiler, these extensions are enabled by default.
22511 The resulting code should be considerably faster in the majority of cases and avoid
22512 the numerical instability problems of 387 code, but may break some existing
22513 code that expects temporaries to be 80 bits.
22515 This is the default choice for the x86-64 compiler.
22520 Attempt to utilize both instruction sets at once. This effectively doubles the
22521 amount of available registers, and on chips with separate execution units for
22522 387 and SSE the execution resources too. Use this option with care, as it is
22523 still experimental, because the GCC register allocator does not model separate
22524 functional units well, resulting in unstable performance.
22527 @item -masm=@var{dialect}
22528 @opindex masm=@var{dialect}
22529 Output assembly instructions using selected @var{dialect}. Also affects
22530 which dialect is used for basic @code{asm} (@pxref{Basic Asm}) and
22531 extended @code{asm} (@pxref{Extended Asm}). Supported choices (in dialect
22532 order) are @samp{att} or @samp{intel}. The default is @samp{att}. Darwin does
22533 not support @samp{intel}.
22536 @itemx -mno-ieee-fp
22538 @opindex mno-ieee-fp
22539 Control whether or not the compiler uses IEEE floating-point
22540 comparisons. These correctly handle the case where the result of a
22541 comparison is unordered.
22544 @opindex msoft-float
22545 Generate output containing library calls for floating point.
22547 @strong{Warning:} the requisite libraries are not part of GCC@.
22548 Normally the facilities of the machine's usual C compiler are used, but
22549 this can't be done directly in cross-compilation. You must make your
22550 own arrangements to provide suitable library functions for
22553 On machines where a function returns floating-point results in the 80387
22554 register stack, some floating-point opcodes may be emitted even if
22555 @option{-msoft-float} is used.
22557 @item -mno-fp-ret-in-387
22558 @opindex mno-fp-ret-in-387
22559 Do not use the FPU registers for return values of functions.
22561 The usual calling convention has functions return values of types
22562 @code{float} and @code{double} in an FPU register, even if there
22563 is no FPU@. The idea is that the operating system should emulate
22566 The option @option{-mno-fp-ret-in-387} causes such values to be returned
22567 in ordinary CPU registers instead.
22569 @item -mno-fancy-math-387
22570 @opindex mno-fancy-math-387
22571 Some 387 emulators do not support the @code{sin}, @code{cos} and
22572 @code{sqrt} instructions for the 387. Specify this option to avoid
22573 generating those instructions. This option is the default on FreeBSD,
22574 OpenBSD and NetBSD@. This option is overridden when @option{-march}
22575 indicates that the target CPU always has an FPU and so the
22576 instruction does not need emulation. These
22577 instructions are not generated unless you also use the
22578 @option{-funsafe-math-optimizations} switch.
22580 @item -malign-double
22581 @itemx -mno-align-double
22582 @opindex malign-double
22583 @opindex mno-align-double
22584 Control whether GCC aligns @code{double}, @code{long double}, and
22585 @code{long long} variables on a two-word boundary or a one-word
22586 boundary. Aligning @code{double} variables on a two-word boundary
22587 produces code that runs somewhat faster on a Pentium at the
22588 expense of more memory.
22590 On x86-64, @option{-malign-double} is enabled by default.
22592 @strong{Warning:} if you use the @option{-malign-double} switch,
22593 structures containing the above types are aligned differently than
22594 the published application binary interface specifications for the x86-32
22595 and are not binary compatible with structures in code compiled
22596 without that switch.
22598 @item -m96bit-long-double
22599 @itemx -m128bit-long-double
22600 @opindex m96bit-long-double
22601 @opindex m128bit-long-double
22602 These switches control the size of @code{long double} type. The x86-32
22603 application binary interface specifies the size to be 96 bits,
22604 so @option{-m96bit-long-double} is the default in 32-bit mode.
22606 Modern architectures (Pentium and newer) prefer @code{long double}
22607 to be aligned to an 8- or 16-byte boundary. In arrays or structures
22608 conforming to the ABI, this is not possible. So specifying
22609 @option{-m128bit-long-double} aligns @code{long double}
22610 to a 16-byte boundary by padding the @code{long double} with an additional
22613 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
22614 its ABI specifies that @code{long double} is aligned on 16-byte boundary.
22616 Notice that neither of these options enable any extra precision over the x87
22617 standard of 80 bits for a @code{long double}.
22619 @strong{Warning:} if you override the default value for your target ABI, this
22620 changes the size of
22621 structures and arrays containing @code{long double} variables,
22622 as well as modifying the function calling convention for functions taking
22623 @code{long double}. Hence they are not binary-compatible
22624 with code compiled without that switch.
22626 @item -mlong-double-64
22627 @itemx -mlong-double-80
22628 @itemx -mlong-double-128
22629 @opindex mlong-double-64
22630 @opindex mlong-double-80
22631 @opindex mlong-double-128
22632 These switches control the size of @code{long double} type. A size
22633 of 64 bits makes the @code{long double} type equivalent to the @code{double}
22634 type. This is the default for 32-bit Bionic C library. A size
22635 of 128 bits makes the @code{long double} type equivalent to the
22636 @code{__float128} type. This is the default for 64-bit Bionic C library.
22638 @strong{Warning:} if you override the default value for your target ABI, this
22639 changes the size of
22640 structures and arrays containing @code{long double} variables,
22641 as well as modifying the function calling convention for functions taking
22642 @code{long double}. Hence they are not binary-compatible
22643 with code compiled without that switch.
22645 @item -malign-data=@var{type}
22646 @opindex malign-data
22647 Control how GCC aligns variables. Supported values for @var{type} are
22648 @samp{compat} uses increased alignment value compatible uses GCC 4.8
22649 and earlier, @samp{abi} uses alignment value as specified by the
22650 psABI, and @samp{cacheline} uses increased alignment value to match
22651 the cache line size. @samp{compat} is the default.
22653 @item -mlarge-data-threshold=@var{threshold}
22654 @opindex mlarge-data-threshold
22655 When @option{-mcmodel=medium} is specified, data objects larger than
22656 @var{threshold} are placed in the large data section. This value must be the
22657 same across all objects linked into the binary, and defaults to 65535.
22661 Use a different function-calling convention, in which functions that
22662 take a fixed number of arguments return with the @code{ret @var{num}}
22663 instruction, which pops their arguments while returning. This saves one
22664 instruction in the caller since there is no need to pop the arguments
22667 You can specify that an individual function is called with this calling
22668 sequence with the function attribute @code{stdcall}. You can also
22669 override the @option{-mrtd} option by using the function attribute
22670 @code{cdecl}. @xref{Function Attributes}.
22672 @strong{Warning:} this calling convention is incompatible with the one
22673 normally used on Unix, so you cannot use it if you need to call
22674 libraries compiled with the Unix compiler.
22676 Also, you must provide function prototypes for all functions that
22677 take variable numbers of arguments (including @code{printf});
22678 otherwise incorrect code is generated for calls to those
22681 In addition, seriously incorrect code results if you call a
22682 function with too many arguments. (Normally, extra arguments are
22683 harmlessly ignored.)
22685 @item -mregparm=@var{num}
22687 Control how many registers are used to pass integer arguments. By
22688 default, no registers are used to pass arguments, and at most 3
22689 registers can be used. You can control this behavior for a specific
22690 function by using the function attribute @code{regparm}.
22691 @xref{Function Attributes}.
22693 @strong{Warning:} if you use this switch, and
22694 @var{num} is nonzero, then you must build all modules with the same
22695 value, including any libraries. This includes the system libraries and
22699 @opindex msseregparm
22700 Use SSE register passing conventions for float and double arguments
22701 and return values. You can control this behavior for a specific
22702 function by using the function attribute @code{sseregparm}.
22703 @xref{Function Attributes}.
22705 @strong{Warning:} if you use this switch then you must build all
22706 modules with the same value, including any libraries. This includes
22707 the system libraries and startup modules.
22709 @item -mvect8-ret-in-mem
22710 @opindex mvect8-ret-in-mem
22711 Return 8-byte vectors in memory instead of MMX registers. This is the
22712 default on Solaris@tie{}8 and 9 and VxWorks to match the ABI of the Sun
22713 Studio compilers until version 12. Later compiler versions (starting
22714 with Studio 12 Update@tie{}1) follow the ABI used by other x86 targets, which
22715 is the default on Solaris@tie{}10 and later. @emph{Only} use this option if
22716 you need to remain compatible with existing code produced by those
22717 previous compiler versions or older versions of GCC@.
22726 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
22727 is specified, the significands of results of floating-point operations are
22728 rounded to 24 bits (single precision); @option{-mpc64} rounds the
22729 significands of results of floating-point operations to 53 bits (double
22730 precision) and @option{-mpc80} rounds the significands of results of
22731 floating-point operations to 64 bits (extended double precision), which is
22732 the default. When this option is used, floating-point operations in higher
22733 precisions are not available to the programmer without setting the FPU
22734 control word explicitly.
22736 Setting the rounding of floating-point operations to less than the default
22737 80 bits can speed some programs by 2% or more. Note that some mathematical
22738 libraries assume that extended-precision (80-bit) floating-point operations
22739 are enabled by default; routines in such libraries could suffer significant
22740 loss of accuracy, typically through so-called ``catastrophic cancellation'',
22741 when this option is used to set the precision to less than extended precision.
22743 @item -mstackrealign
22744 @opindex mstackrealign
22745 Realign the stack at entry. On the x86, the @option{-mstackrealign}
22746 option generates an alternate prologue and epilogue that realigns the
22747 run-time stack if necessary. This supports mixing legacy codes that keep
22748 4-byte stack alignment with modern codes that keep 16-byte stack alignment for
22749 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
22750 applicable to individual functions.
22752 @item -mpreferred-stack-boundary=@var{num}
22753 @opindex mpreferred-stack-boundary
22754 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
22755 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
22756 the default is 4 (16 bytes or 128 bits).
22758 @strong{Warning:} When generating code for the x86-64 architecture with
22759 SSE extensions disabled, @option{-mpreferred-stack-boundary=3} can be
22760 used to keep the stack boundary aligned to 8 byte boundary. Since
22761 x86-64 ABI require 16 byte stack alignment, this is ABI incompatible and
22762 intended to be used in controlled environment where stack space is
22763 important limitation. This option leads to wrong code when functions
22764 compiled with 16 byte stack alignment (such as functions from a standard
22765 library) are called with misaligned stack. In this case, SSE
22766 instructions may lead to misaligned memory access traps. In addition,
22767 variable arguments are handled incorrectly for 16 byte aligned
22768 objects (including x87 long double and __int128), leading to wrong
22769 results. You must build all modules with
22770 @option{-mpreferred-stack-boundary=3}, including any libraries. This
22771 includes the system libraries and startup modules.
22773 @item -mincoming-stack-boundary=@var{num}
22774 @opindex mincoming-stack-boundary
22775 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
22776 boundary. If @option{-mincoming-stack-boundary} is not specified,
22777 the one specified by @option{-mpreferred-stack-boundary} is used.
22779 On Pentium and Pentium Pro, @code{double} and @code{long double} values
22780 should be aligned to an 8-byte boundary (see @option{-malign-double}) or
22781 suffer significant run time performance penalties. On Pentium III, the
22782 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
22783 properly if it is not 16-byte aligned.
22785 To ensure proper alignment of this values on the stack, the stack boundary
22786 must be as aligned as that required by any value stored on the stack.
22787 Further, every function must be generated such that it keeps the stack
22788 aligned. Thus calling a function compiled with a higher preferred
22789 stack boundary from a function compiled with a lower preferred stack
22790 boundary most likely misaligns the stack. It is recommended that
22791 libraries that use callbacks always use the default setting.
22793 This extra alignment does consume extra stack space, and generally
22794 increases code size. Code that is sensitive to stack space usage, such
22795 as embedded systems and operating system kernels, may want to reduce the
22796 preferred alignment to @option{-mpreferred-stack-boundary=2}.
22842 @opindex mclfushopt
22860 @itemx -mprefetchwt1
22861 @opindex mprefetchwt1
22912 These switches enable the use of instructions in the MMX, SSE,
22913 SSE2, SSE3, SSSE3, SSE4.1, AVX, AVX2, AVX512F, AVX512PF, AVX512ER, AVX512CD,
22914 SHA, AES, PCLMUL, FSGSBASE, RDRND, F16C, FMA, SSE4A, FMA4, XOP, LWP, ABM,
22915 BMI, BMI2, FXSR, XSAVE, XSAVEOPT, LZCNT, RTM, MPX, MWAITX or 3DNow!@:
22916 extended instruction sets. Each has a corresponding @option{-mno-} option
22917 to disable use of these instructions.
22919 These extensions are also available as built-in functions: see
22920 @ref{x86 Built-in Functions}, for details of the functions enabled and
22921 disabled by these switches.
22923 To generate SSE/SSE2 instructions automatically from floating-point
22924 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
22926 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
22927 generates new AVX instructions or AVX equivalence for all SSEx instructions
22930 These options enable GCC to use these extended instructions in
22931 generated code, even without @option{-mfpmath=sse}. Applications that
22932 perform run-time CPU detection must compile separate files for each
22933 supported architecture, using the appropriate flags. In particular,
22934 the file containing the CPU detection code should be compiled without
22937 @item -mdump-tune-features
22938 @opindex mdump-tune-features
22939 This option instructs GCC to dump the names of the x86 performance
22940 tuning features and default settings. The names can be used in
22941 @option{-mtune-ctrl=@var{feature-list}}.
22943 @item -mtune-ctrl=@var{feature-list}
22944 @opindex mtune-ctrl=@var{feature-list}
22945 This option is used to do fine grain control of x86 code generation features.
22946 @var{feature-list} is a comma separated list of @var{feature} names. See also
22947 @option{-mdump-tune-features}. When specified, the @var{feature} is turned
22948 on if it is not preceded with @samp{^}, otherwise, it is turned off.
22949 @option{-mtune-ctrl=@var{feature-list}} is intended to be used by GCC
22950 developers. Using it may lead to code paths not covered by testing and can
22951 potentially result in compiler ICEs or runtime errors.
22954 @opindex mno-default
22955 This option instructs GCC to turn off all tunable features. See also
22956 @option{-mtune-ctrl=@var{feature-list}} and @option{-mdump-tune-features}.
22960 This option instructs GCC to emit a @code{cld} instruction in the prologue
22961 of functions that use string instructions. String instructions depend on
22962 the DF flag to select between autoincrement or autodecrement mode. While the
22963 ABI specifies the DF flag to be cleared on function entry, some operating
22964 systems violate this specification by not clearing the DF flag in their
22965 exception dispatchers. The exception handler can be invoked with the DF flag
22966 set, which leads to wrong direction mode when string instructions are used.
22967 This option can be enabled by default on 32-bit x86 targets by configuring
22968 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
22969 instructions can be suppressed with the @option{-mno-cld} compiler option
22973 @opindex mvzeroupper
22974 This option instructs GCC to emit a @code{vzeroupper} instruction
22975 before a transfer of control flow out of the function to minimize
22976 the AVX to SSE transition penalty as well as remove unnecessary @code{zeroupper}
22979 @item -mprefer-avx128
22980 @opindex mprefer-avx128
22981 This option instructs GCC to use 128-bit AVX instructions instead of
22982 256-bit AVX instructions in the auto-vectorizer.
22986 This option enables GCC to generate @code{CMPXCHG16B} instructions.
22987 @code{CMPXCHG16B} allows for atomic operations on 128-bit double quadword
22988 (or oword) data types.
22989 This is useful for high-resolution counters that can be updated
22990 by multiple processors (or cores). This instruction is generated as part of
22991 atomic built-in functions: see @ref{__sync Builtins} or
22992 @ref{__atomic Builtins} for details.
22996 This option enables generation of @code{SAHF} instructions in 64-bit code.
22997 Early Intel Pentium 4 CPUs with Intel 64 support,
22998 prior to the introduction of Pentium 4 G1 step in December 2005,
22999 lacked the @code{LAHF} and @code{SAHF} instructions
23000 which are supported by AMD64.
23001 These are load and store instructions, respectively, for certain status flags.
23002 In 64-bit mode, the @code{SAHF} instruction is used to optimize @code{fmod},
23003 @code{drem}, and @code{remainder} built-in functions;
23004 see @ref{Other Builtins} for details.
23008 This option enables use of the @code{movbe} instruction to implement
23009 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
23013 This option enables built-in functions @code{__builtin_ia32_crc32qi},
23014 @code{__builtin_ia32_crc32hi}, @code{__builtin_ia32_crc32si} and
23015 @code{__builtin_ia32_crc32di} to generate the @code{crc32} machine instruction.
23019 This option enables use of @code{RCPSS} and @code{RSQRTSS} instructions
23020 (and their vectorized variants @code{RCPPS} and @code{RSQRTPS})
23021 with an additional Newton-Raphson step
23022 to increase precision instead of @code{DIVSS} and @code{SQRTSS}
23023 (and their vectorized
23024 variants) for single-precision floating-point arguments. These instructions
23025 are generated only when @option{-funsafe-math-optimizations} is enabled
23026 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
23027 Note that while the throughput of the sequence is higher than the throughput
23028 of the non-reciprocal instruction, the precision of the sequence can be
23029 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
23031 Note that GCC implements @code{1.0f/sqrtf(@var{x})} in terms of @code{RSQRTSS}
23032 (or @code{RSQRTPS}) already with @option{-ffast-math} (or the above option
23033 combination), and doesn't need @option{-mrecip}.
23035 Also note that GCC emits the above sequence with additional Newton-Raphson step
23036 for vectorized single-float division and vectorized @code{sqrtf(@var{x})}
23037 already with @option{-ffast-math} (or the above option combination), and
23038 doesn't need @option{-mrecip}.
23040 @item -mrecip=@var{opt}
23041 @opindex mrecip=opt
23042 This option controls which reciprocal estimate instructions
23043 may be used. @var{opt} is a comma-separated list of options, which may
23044 be preceded by a @samp{!} to invert the option:
23048 Enable all estimate instructions.
23051 Enable the default instructions, equivalent to @option{-mrecip}.
23054 Disable all estimate instructions, equivalent to @option{-mno-recip}.
23057 Enable the approximation for scalar division.
23060 Enable the approximation for vectorized division.
23063 Enable the approximation for scalar square root.
23066 Enable the approximation for vectorized square root.
23069 So, for example, @option{-mrecip=all,!sqrt} enables
23070 all of the reciprocal approximations, except for square root.
23072 @item -mveclibabi=@var{type}
23073 @opindex mveclibabi
23074 Specifies the ABI type to use for vectorizing intrinsics using an
23075 external library. Supported values for @var{type} are @samp{svml}
23076 for the Intel short
23077 vector math library and @samp{acml} for the AMD math core library.
23078 To use this option, both @option{-ftree-vectorize} and
23079 @option{-funsafe-math-optimizations} have to be enabled, and an SVML or ACML
23080 ABI-compatible library must be specified at link time.
23082 GCC currently emits calls to @code{vmldExp2},
23083 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
23084 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
23085 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
23086 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
23087 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
23088 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
23089 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
23090 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
23091 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
23092 function type when @option{-mveclibabi=svml} is used, and @code{__vrd2_sin},
23093 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
23094 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
23095 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
23096 @code{__vrs4_log10f} and @code{__vrs4_powf} for the corresponding function type
23097 when @option{-mveclibabi=acml} is used.
23099 @item -mabi=@var{name}
23101 Generate code for the specified calling convention. Permissible values
23102 are @samp{sysv} for the ABI used on GNU/Linux and other systems, and
23103 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
23104 ABI when targeting Microsoft Windows and the SysV ABI on all other systems.
23105 You can control this behavior for specific functions by
23106 using the function attributes @code{ms_abi} and @code{sysv_abi}.
23107 @xref{Function Attributes}.
23109 @item -mtls-dialect=@var{type}
23110 @opindex mtls-dialect
23111 Generate code to access thread-local storage using the @samp{gnu} or
23112 @samp{gnu2} conventions. @samp{gnu} is the conservative default;
23113 @samp{gnu2} is more efficient, but it may add compile- and run-time
23114 requirements that cannot be satisfied on all systems.
23117 @itemx -mno-push-args
23118 @opindex mpush-args
23119 @opindex mno-push-args
23120 Use PUSH operations to store outgoing parameters. This method is shorter
23121 and usually equally fast as method using SUB/MOV operations and is enabled
23122 by default. In some cases disabling it may improve performance because of
23123 improved scheduling and reduced dependencies.
23125 @item -maccumulate-outgoing-args
23126 @opindex maccumulate-outgoing-args
23127 If enabled, the maximum amount of space required for outgoing arguments is
23128 computed in the function prologue. This is faster on most modern CPUs
23129 because of reduced dependencies, improved scheduling and reduced stack usage
23130 when the preferred stack boundary is not equal to 2. The drawback is a notable
23131 increase in code size. This switch implies @option{-mno-push-args}.
23135 Support thread-safe exception handling on MinGW. Programs that rely
23136 on thread-safe exception handling must compile and link all code with the
23137 @option{-mthreads} option. When compiling, @option{-mthreads} defines
23138 @option{-D_MT}; when linking, it links in a special thread helper library
23139 @option{-lmingwthrd} which cleans up per-thread exception-handling data.
23141 @item -mno-align-stringops
23142 @opindex mno-align-stringops
23143 Do not align the destination of inlined string operations. This switch reduces
23144 code size and improves performance in case the destination is already aligned,
23145 but GCC doesn't know about it.
23147 @item -minline-all-stringops
23148 @opindex minline-all-stringops
23149 By default GCC inlines string operations only when the destination is
23150 known to be aligned to least a 4-byte boundary.
23151 This enables more inlining and increases code
23152 size, but may improve performance of code that depends on fast
23153 @code{memcpy}, @code{strlen},
23154 and @code{memset} for short lengths.
23156 @item -minline-stringops-dynamically
23157 @opindex minline-stringops-dynamically
23158 For string operations of unknown size, use run-time checks with
23159 inline code for small blocks and a library call for large blocks.
23161 @item -mstringop-strategy=@var{alg}
23162 @opindex mstringop-strategy=@var{alg}
23163 Override the internal decision heuristic for the particular algorithm to use
23164 for inlining string operations. The allowed values for @var{alg} are:
23170 Expand using i386 @code{rep} prefix of the specified size.
23174 @itemx unrolled_loop
23175 Expand into an inline loop.
23178 Always use a library call.
23181 @item -mmemcpy-strategy=@var{strategy}
23182 @opindex mmemcpy-strategy=@var{strategy}
23183 Override the internal decision heuristic to decide if @code{__builtin_memcpy}
23184 should be inlined and what inline algorithm to use when the expected size
23185 of the copy operation is known. @var{strategy}
23186 is a comma-separated list of @var{alg}:@var{max_size}:@var{dest_align} triplets.
23187 @var{alg} is specified in @option{-mstringop-strategy}, @var{max_size} specifies
23188 the max byte size with which inline algorithm @var{alg} is allowed. For the last
23189 triplet, the @var{max_size} must be @code{-1}. The @var{max_size} of the triplets
23190 in the list must be specified in increasing order. The minimal byte size for
23191 @var{alg} is @code{0} for the first triplet and @code{@var{max_size} + 1} of the
23194 @item -mmemset-strategy=@var{strategy}
23195 @opindex mmemset-strategy=@var{strategy}
23196 The option is similar to @option{-mmemcpy-strategy=} except that it is to control
23197 @code{__builtin_memset} expansion.
23199 @item -momit-leaf-frame-pointer
23200 @opindex momit-leaf-frame-pointer
23201 Don't keep the frame pointer in a register for leaf functions. This
23202 avoids the instructions to save, set up, and restore frame pointers and
23203 makes an extra register available in leaf functions. The option
23204 @option{-fomit-leaf-frame-pointer} removes the frame pointer for leaf functions,
23205 which might make debugging harder.
23207 @item -mtls-direct-seg-refs
23208 @itemx -mno-tls-direct-seg-refs
23209 @opindex mtls-direct-seg-refs
23210 Controls whether TLS variables may be accessed with offsets from the
23211 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
23212 or whether the thread base pointer must be added. Whether or not this
23213 is valid depends on the operating system, and whether it maps the
23214 segment to cover the entire TLS area.
23216 For systems that use the GNU C Library, the default is on.
23219 @itemx -mno-sse2avx
23221 Specify that the assembler should encode SSE instructions with VEX
23222 prefix. The option @option{-mavx} turns this on by default.
23227 If profiling is active (@option{-pg}), put the profiling
23228 counter call before the prologue.
23229 Note: On x86 architectures the attribute @code{ms_hook_prologue}
23230 isn't possible at the moment for @option{-mfentry} and @option{-pg}.
23232 @item -mrecord-mcount
23233 @itemx -mno-record-mcount
23234 @opindex mrecord-mcount
23235 If profiling is active (@option{-pg}), generate a __mcount_loc section
23236 that contains pointers to each profiling call. This is useful for
23237 automatically patching and out calls.
23240 @itemx -mno-nop-mcount
23241 @opindex mnop-mcount
23242 If profiling is active (@option{-pg}), generate the calls to
23243 the profiling functions as nops. This is useful when they
23244 should be patched in later dynamically. This is likely only
23245 useful together with @option{-mrecord-mcount}.
23247 @item -mskip-rax-setup
23248 @itemx -mno-skip-rax-setup
23249 @opindex mskip-rax-setup
23250 When generating code for the x86-64 architecture with SSE extensions
23251 disabled, @option{-skip-rax-setup} can be used to skip setting up RAX
23252 register when there are no variable arguments passed in vector registers.
23254 @strong{Warning:} Since RAX register is used to avoid unnecessarily
23255 saving vector registers on stack when passing variable arguments, the
23256 impacts of this option are callees may waste some stack space,
23257 misbehave or jump to a random location. GCC 4.4 or newer don't have
23258 those issues, regardless the RAX register value.
23261 @itemx -mno-8bit-idiv
23262 @opindex m8bit-idiv
23263 On some processors, like Intel Atom, 8-bit unsigned integer divide is
23264 much faster than 32-bit/64-bit integer divide. This option generates a
23265 run-time check. If both dividend and divisor are within range of 0
23266 to 255, 8-bit unsigned integer divide is used instead of
23267 32-bit/64-bit integer divide.
23269 @item -mavx256-split-unaligned-load
23270 @itemx -mavx256-split-unaligned-store
23271 @opindex mavx256-split-unaligned-load
23272 @opindex mavx256-split-unaligned-store
23273 Split 32-byte AVX unaligned load and store.
23275 @item -mstack-protector-guard=@var{guard}
23276 @opindex mstack-protector-guard=@var{guard}
23277 Generate stack protection code using canary at @var{guard}. Supported
23278 locations are @samp{global} for global canary or @samp{tls} for per-thread
23279 canary in the TLS block (the default). This option has effect only when
23280 @option{-fstack-protector} or @option{-fstack-protector-all} is specified.
23284 These @samp{-m} switches are supported in addition to the above
23285 on x86-64 processors in 64-bit environments.
23298 Generate code for a 16-bit, 32-bit or 64-bit environment.
23299 The @option{-m32} option sets @code{int}, @code{long}, and pointer types
23301 generates code that runs on any i386 system.
23303 The @option{-m64} option sets @code{int} to 32 bits and @code{long} and pointer
23304 types to 64 bits, and generates code for the x86-64 architecture.
23305 For Darwin only the @option{-m64} option also turns off the @option{-fno-pic}
23306 and @option{-mdynamic-no-pic} options.
23308 The @option{-mx32} option sets @code{int}, @code{long}, and pointer types
23310 generates code for the x86-64 architecture.
23312 The @option{-m16} option is the same as @option{-m32}, except for that
23313 it outputs the @code{.code16gcc} assembly directive at the beginning of
23314 the assembly output so that the binary can run in 16-bit mode.
23316 The @option{-miamcu} option generates code which conforms to Intel MCU
23317 psABI. It requires the @option{-m32} option to be turned on.
23319 @item -mno-red-zone
23320 @opindex mno-red-zone
23321 Do not use a so-called ``red zone'' for x86-64 code. The red zone is mandated
23322 by the x86-64 ABI; it is a 128-byte area beyond the location of the
23323 stack pointer that is not modified by signal or interrupt handlers
23324 and therefore can be used for temporary data without adjusting the stack
23325 pointer. The flag @option{-mno-red-zone} disables this red zone.
23327 @item -mcmodel=small
23328 @opindex mcmodel=small
23329 Generate code for the small code model: the program and its symbols must
23330 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
23331 Programs can be statically or dynamically linked. This is the default
23334 @item -mcmodel=kernel
23335 @opindex mcmodel=kernel
23336 Generate code for the kernel code model. The kernel runs in the
23337 negative 2 GB of the address space.
23338 This model has to be used for Linux kernel code.
23340 @item -mcmodel=medium
23341 @opindex mcmodel=medium
23342 Generate code for the medium model: the program is linked in the lower 2
23343 GB of the address space. Small symbols are also placed there. Symbols
23344 with sizes larger than @option{-mlarge-data-threshold} are put into
23345 large data or BSS sections and can be located above 2GB. Programs can
23346 be statically or dynamically linked.
23348 @item -mcmodel=large
23349 @opindex mcmodel=large
23350 Generate code for the large model. This model makes no assumptions
23351 about addresses and sizes of sections.
23353 @item -maddress-mode=long
23354 @opindex maddress-mode=long
23355 Generate code for long address mode. This is only supported for 64-bit
23356 and x32 environments. It is the default address mode for 64-bit
23359 @item -maddress-mode=short
23360 @opindex maddress-mode=short
23361 Generate code for short address mode. This is only supported for 32-bit
23362 and x32 environments. It is the default address mode for 32-bit and
23366 @node x86 Windows Options
23367 @subsection x86 Windows Options
23368 @cindex x86 Windows Options
23369 @cindex Windows Options for x86
23371 These additional options are available for Microsoft Windows targets:
23377 specifies that a console application is to be generated, by
23378 instructing the linker to set the PE header subsystem type
23379 required for console applications.
23380 This option is available for Cygwin and MinGW targets and is
23381 enabled by default on those targets.
23385 This option is available for Cygwin and MinGW targets. It
23386 specifies that a DLL---a dynamic link library---is to be
23387 generated, enabling the selection of the required runtime
23388 startup object and entry point.
23390 @item -mnop-fun-dllimport
23391 @opindex mnop-fun-dllimport
23392 This option is available for Cygwin and MinGW targets. It
23393 specifies that the @code{dllimport} attribute should be ignored.
23397 This option is available for MinGW targets. It specifies
23398 that MinGW-specific thread support is to be used.
23402 This option is available for MinGW-w64 targets. It causes
23403 the @code{UNICODE} preprocessor macro to be predefined, and
23404 chooses Unicode-capable runtime startup code.
23408 This option is available for Cygwin and MinGW targets. It
23409 specifies that the typical Microsoft Windows predefined macros are to
23410 be set in the pre-processor, but does not influence the choice
23411 of runtime library/startup code.
23415 This option is available for Cygwin and MinGW targets. It
23416 specifies that a GUI application is to be generated by
23417 instructing the linker to set the PE header subsystem type
23420 @item -fno-set-stack-executable
23421 @opindex fno-set-stack-executable
23422 This option is available for MinGW targets. It specifies that
23423 the executable flag for the stack used by nested functions isn't
23424 set. This is necessary for binaries running in kernel mode of
23425 Microsoft Windows, as there the User32 API, which is used to set executable
23426 privileges, isn't available.
23428 @item -fwritable-relocated-rdata
23429 @opindex fno-writable-relocated-rdata
23430 This option is available for MinGW and Cygwin targets. It specifies
23431 that relocated-data in read-only section is put into .data
23432 section. This is a necessary for older runtimes not supporting
23433 modification of .rdata sections for pseudo-relocation.
23435 @item -mpe-aligned-commons
23436 @opindex mpe-aligned-commons
23437 This option is available for Cygwin and MinGW targets. It
23438 specifies that the GNU extension to the PE file format that
23439 permits the correct alignment of COMMON variables should be
23440 used when generating code. It is enabled by default if
23441 GCC detects that the target assembler found during configuration
23442 supports the feature.
23445 See also under @ref{x86 Options} for standard options.
23447 @node Xstormy16 Options
23448 @subsection Xstormy16 Options
23449 @cindex Xstormy16 Options
23451 These options are defined for Xstormy16:
23456 Choose startup files and linker script suitable for the simulator.
23459 @node Xtensa Options
23460 @subsection Xtensa Options
23461 @cindex Xtensa Options
23463 These options are supported for Xtensa targets:
23467 @itemx -mno-const16
23469 @opindex mno-const16
23470 Enable or disable use of @code{CONST16} instructions for loading
23471 constant values. The @code{CONST16} instruction is currently not a
23472 standard option from Tensilica. When enabled, @code{CONST16}
23473 instructions are always used in place of the standard @code{L32R}
23474 instructions. The use of @code{CONST16} is enabled by default only if
23475 the @code{L32R} instruction is not available.
23478 @itemx -mno-fused-madd
23479 @opindex mfused-madd
23480 @opindex mno-fused-madd
23481 Enable or disable use of fused multiply/add and multiply/subtract
23482 instructions in the floating-point option. This has no effect if the
23483 floating-point option is not also enabled. Disabling fused multiply/add
23484 and multiply/subtract instructions forces the compiler to use separate
23485 instructions for the multiply and add/subtract operations. This may be
23486 desirable in some cases where strict IEEE 754-compliant results are
23487 required: the fused multiply add/subtract instructions do not round the
23488 intermediate result, thereby producing results with @emph{more} bits of
23489 precision than specified by the IEEE standard. Disabling fused multiply
23490 add/subtract instructions also ensures that the program output is not
23491 sensitive to the compiler's ability to combine multiply and add/subtract
23494 @item -mserialize-volatile
23495 @itemx -mno-serialize-volatile
23496 @opindex mserialize-volatile
23497 @opindex mno-serialize-volatile
23498 When this option is enabled, GCC inserts @code{MEMW} instructions before
23499 @code{volatile} memory references to guarantee sequential consistency.
23500 The default is @option{-mserialize-volatile}. Use
23501 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
23503 @item -mforce-no-pic
23504 @opindex mforce-no-pic
23505 For targets, like GNU/Linux, where all user-mode Xtensa code must be
23506 position-independent code (PIC), this option disables PIC for compiling
23509 @item -mtext-section-literals
23510 @itemx -mno-text-section-literals
23511 @opindex mtext-section-literals
23512 @opindex mno-text-section-literals
23513 These options control the treatment of literal pools. The default is
23514 @option{-mno-text-section-literals}, which places literals in a separate
23515 section in the output file. This allows the literal pool to be placed
23516 in a data RAM/ROM, and it also allows the linker to combine literal
23517 pools from separate object files to remove redundant literals and
23518 improve code size. With @option{-mtext-section-literals}, the literals
23519 are interspersed in the text section in order to keep them as close as
23520 possible to their references. This may be necessary for large assembly
23523 @item -mtarget-align
23524 @itemx -mno-target-align
23525 @opindex mtarget-align
23526 @opindex mno-target-align
23527 When this option is enabled, GCC instructs the assembler to
23528 automatically align instructions to reduce branch penalties at the
23529 expense of some code density. The assembler attempts to widen density
23530 instructions to align branch targets and the instructions following call
23531 instructions. If there are not enough preceding safe density
23532 instructions to align a target, no widening is performed. The
23533 default is @option{-mtarget-align}. These options do not affect the
23534 treatment of auto-aligned instructions like @code{LOOP}, which the
23535 assembler always aligns, either by widening density instructions or
23536 by inserting NOP instructions.
23539 @itemx -mno-longcalls
23540 @opindex mlongcalls
23541 @opindex mno-longcalls
23542 When this option is enabled, GCC instructs the assembler to translate
23543 direct calls to indirect calls unless it can determine that the target
23544 of a direct call is in the range allowed by the call instruction. This
23545 translation typically occurs for calls to functions in other source
23546 files. Specifically, the assembler translates a direct @code{CALL}
23547 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
23548 The default is @option{-mno-longcalls}. This option should be used in
23549 programs where the call target can potentially be out of range. This
23550 option is implemented in the assembler, not the compiler, so the
23551 assembly code generated by GCC still shows direct call
23552 instructions---look at the disassembled object code to see the actual
23553 instructions. Note that the assembler uses an indirect call for
23554 every cross-file call, not just those that really are out of range.
23557 @node zSeries Options
23558 @subsection zSeries Options
23559 @cindex zSeries options
23561 These are listed under @xref{S/390 and zSeries Options}.
23563 @node Code Gen Options
23564 @section Options for Code Generation Conventions
23565 @cindex code generation conventions
23566 @cindex options, code generation
23567 @cindex run-time options
23569 These machine-independent options control the interface conventions
23570 used in code generation.
23572 Most of them have both positive and negative forms; the negative form
23573 of @option{-ffoo} is @option{-fno-foo}. In the table below, only
23574 one of the forms is listed---the one that is not the default. You
23575 can figure out the other form by either removing @samp{no-} or adding
23579 @item -fbounds-check
23580 @opindex fbounds-check
23581 For front ends that support it, generate additional code to check that
23582 indices used to access arrays are within the declared range. This is
23583 currently only supported by the Java and Fortran front ends, where
23584 this option defaults to true and false respectively.
23586 @item -fstack-reuse=@var{reuse-level}
23587 @opindex fstack_reuse
23588 This option controls stack space reuse for user declared local/auto variables
23589 and compiler generated temporaries. @var{reuse_level} can be @samp{all},
23590 @samp{named_vars}, or @samp{none}. @samp{all} enables stack reuse for all
23591 local variables and temporaries, @samp{named_vars} enables the reuse only for
23592 user defined local variables with names, and @samp{none} disables stack reuse
23593 completely. The default value is @samp{all}. The option is needed when the
23594 program extends the lifetime of a scoped local variable or a compiler generated
23595 temporary beyond the end point defined by the language. When a lifetime of
23596 a variable ends, and if the variable lives in memory, the optimizing compiler
23597 has the freedom to reuse its stack space with other temporaries or scoped
23598 local variables whose live range does not overlap with it. Legacy code extending
23599 local lifetime is likely to break with the stack reuse optimization.
23618 if (*p == 10) // out of scope use of local1
23629 A(int k) : i(k), j(k) @{ @}
23636 void foo(const A& ar)
23643 foo(A(10)); // temp object's lifetime ends when foo returns
23649 ap->i+= 10; // ap references out of scope temp whose space
23650 // is reused with a. What is the value of ap->i?
23655 The lifetime of a compiler generated temporary is well defined by the C++
23656 standard. When a lifetime of a temporary ends, and if the temporary lives
23657 in memory, the optimizing compiler has the freedom to reuse its stack
23658 space with other temporaries or scoped local variables whose live range
23659 does not overlap with it. However some of the legacy code relies on
23660 the behavior of older compilers in which temporaries' stack space is
23661 not reused, the aggressive stack reuse can lead to runtime errors. This
23662 option is used to control the temporary stack reuse optimization.
23666 This option generates traps for signed overflow on addition, subtraction,
23667 multiplication operations.
23671 This option instructs the compiler to assume that signed arithmetic
23672 overflow of addition, subtraction and multiplication wraps around
23673 using twos-complement representation. This flag enables some optimizations
23674 and disables others. This option is enabled by default for the Java
23675 front end, as required by the Java language specification.
23678 @opindex fexceptions
23679 Enable exception handling. Generates extra code needed to propagate
23680 exceptions. For some targets, this implies GCC generates frame
23681 unwind information for all functions, which can produce significant data
23682 size overhead, although it does not affect execution. If you do not
23683 specify this option, GCC enables it by default for languages like
23684 C++ that normally require exception handling, and disables it for
23685 languages like C that do not normally require it. However, you may need
23686 to enable this option when compiling C code that needs to interoperate
23687 properly with exception handlers written in C++. You may also wish to
23688 disable this option if you are compiling older C++ programs that don't
23689 use exception handling.
23691 @item -fnon-call-exceptions
23692 @opindex fnon-call-exceptions
23693 Generate code that allows trapping instructions to throw exceptions.
23694 Note that this requires platform-specific runtime support that does
23695 not exist everywhere. Moreover, it only allows @emph{trapping}
23696 instructions to throw exceptions, i.e.@: memory references or floating-point
23697 instructions. It does not allow exceptions to be thrown from
23698 arbitrary signal handlers such as @code{SIGALRM}.
23700 @item -fdelete-dead-exceptions
23701 @opindex fdelete-dead-exceptions
23702 Consider that instructions that may throw exceptions but don't otherwise
23703 contribute to the execution of the program can be optimized away.
23704 This option is enabled by default for the Ada front end, as permitted by
23705 the Ada language specification.
23706 Optimization passes that cause dead exceptions to be removed are enabled independently at different optimization levels.
23708 @item -funwind-tables
23709 @opindex funwind-tables
23710 Similar to @option{-fexceptions}, except that it just generates any needed
23711 static data, but does not affect the generated code in any other way.
23712 You normally do not need to enable this option; instead, a language processor
23713 that needs this handling enables it on your behalf.
23715 @item -fasynchronous-unwind-tables
23716 @opindex fasynchronous-unwind-tables
23717 Generate unwind table in DWARF 2 format, if supported by target machine. The
23718 table is exact at each instruction boundary, so it can be used for stack
23719 unwinding from asynchronous events (such as debugger or garbage collector).
23721 @item -fno-gnu-unique
23722 @opindex fno-gnu-unique
23723 On systems with recent GNU assembler and C library, the C++ compiler
23724 uses the @code{STB_GNU_UNIQUE} binding to make sure that definitions
23725 of template static data members and static local variables in inline
23726 functions are unique even in the presence of @code{RTLD_LOCAL}; this
23727 is necessary to avoid problems with a library used by two different
23728 @code{RTLD_LOCAL} plugins depending on a definition in one of them and
23729 therefore disagreeing with the other one about the binding of the
23730 symbol. But this causes @code{dlclose} to be ignored for affected
23731 DSOs; if your program relies on reinitialization of a DSO via
23732 @code{dlclose} and @code{dlopen}, you can use
23733 @option{-fno-gnu-unique}.
23735 @item -fpcc-struct-return
23736 @opindex fpcc-struct-return
23737 Return ``short'' @code{struct} and @code{union} values in memory like
23738 longer ones, rather than in registers. This convention is less
23739 efficient, but it has the advantage of allowing intercallability between
23740 GCC-compiled files and files compiled with other compilers, particularly
23741 the Portable C Compiler (pcc).
23743 The precise convention for returning structures in memory depends
23744 on the target configuration macros.
23746 Short structures and unions are those whose size and alignment match
23747 that of some integer type.
23749 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
23750 switch is not binary compatible with code compiled with the
23751 @option{-freg-struct-return} switch.
23752 Use it to conform to a non-default application binary interface.
23754 @item -freg-struct-return
23755 @opindex freg-struct-return
23756 Return @code{struct} and @code{union} values in registers when possible.
23757 This is more efficient for small structures than
23758 @option{-fpcc-struct-return}.
23760 If you specify neither @option{-fpcc-struct-return} nor
23761 @option{-freg-struct-return}, GCC defaults to whichever convention is
23762 standard for the target. If there is no standard convention, GCC
23763 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
23764 the principal compiler. In those cases, we can choose the standard, and
23765 we chose the more efficient register return alternative.
23767 @strong{Warning:} code compiled with the @option{-freg-struct-return}
23768 switch is not binary compatible with code compiled with the
23769 @option{-fpcc-struct-return} switch.
23770 Use it to conform to a non-default application binary interface.
23772 @item -fshort-enums
23773 @opindex fshort-enums
23774 Allocate to an @code{enum} type only as many bytes as it needs for the
23775 declared range of possible values. Specifically, the @code{enum} type
23776 is equivalent to the smallest integer type that has enough room.
23778 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
23779 code that is not binary compatible with code generated without that switch.
23780 Use it to conform to a non-default application binary interface.
23782 @item -fshort-double
23783 @opindex fshort-double
23784 Use the same size for @code{double} as for @code{float}.
23786 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
23787 code that is not binary compatible with code generated without that switch.
23788 Use it to conform to a non-default application binary interface.
23790 @item -fshort-wchar
23791 @opindex fshort-wchar
23792 Override the underlying type for @code{wchar_t} to be @code{short
23793 unsigned int} instead of the default for the target. This option is
23794 useful for building programs to run under WINE@.
23796 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
23797 code that is not binary compatible with code generated without that switch.
23798 Use it to conform to a non-default application binary interface.
23801 @opindex fno-common
23802 In C code, controls the placement of uninitialized global variables.
23803 Unix C compilers have traditionally permitted multiple definitions of
23804 such variables in different compilation units by placing the variables
23806 This is the behavior specified by @option{-fcommon}, and is the default
23807 for GCC on most targets.
23808 On the other hand, this behavior is not required by ISO C, and on some
23809 targets may carry a speed or code size penalty on variable references.
23810 The @option{-fno-common} option specifies that the compiler should place
23811 uninitialized global variables in the data section of the object file,
23812 rather than generating them as common blocks.
23813 This has the effect that if the same variable is declared
23814 (without @code{extern}) in two different compilations,
23815 you get a multiple-definition error when you link them.
23816 In this case, you must compile with @option{-fcommon} instead.
23817 Compiling with @option{-fno-common} is useful on targets for which
23818 it provides better performance, or if you wish to verify that the
23819 program will work on other systems that always treat uninitialized
23820 variable declarations this way.
23824 Ignore the @code{#ident} directive.
23826 @item -finhibit-size-directive
23827 @opindex finhibit-size-directive
23828 Don't output a @code{.size} assembler directive, or anything else that
23829 would cause trouble if the function is split in the middle, and the
23830 two halves are placed at locations far apart in memory. This option is
23831 used when compiling @file{crtstuff.c}; you should not need to use it
23834 @item -fverbose-asm
23835 @opindex fverbose-asm
23836 Put extra commentary information in the generated assembly code to
23837 make it more readable. This option is generally only of use to those
23838 who actually need to read the generated assembly code (perhaps while
23839 debugging the compiler itself).
23841 @option{-fno-verbose-asm}, the default, causes the
23842 extra information to be omitted and is useful when comparing two assembler
23845 @item -frecord-gcc-switches
23846 @opindex frecord-gcc-switches
23847 This switch causes the command line used to invoke the
23848 compiler to be recorded into the object file that is being created.
23849 This switch is only implemented on some targets and the exact format
23850 of the recording is target and binary file format dependent, but it
23851 usually takes the form of a section containing ASCII text. This
23852 switch is related to the @option{-fverbose-asm} switch, but that
23853 switch only records information in the assembler output file as
23854 comments, so it never reaches the object file.
23855 See also @option{-grecord-gcc-switches} for another
23856 way of storing compiler options into the object file.
23860 @cindex global offset table
23862 Generate position-independent code (PIC) suitable for use in a shared
23863 library, if supported for the target machine. Such code accesses all
23864 constant addresses through a global offset table (GOT)@. The dynamic
23865 loader resolves the GOT entries when the program starts (the dynamic
23866 loader is not part of GCC; it is part of the operating system). If
23867 the GOT size for the linked executable exceeds a machine-specific
23868 maximum size, you get an error message from the linker indicating that
23869 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
23870 instead. (These maximums are 8k on the SPARC and 32k
23871 on the m68k and RS/6000. The x86 has no such limit.)
23873 Position-independent code requires special support, and therefore works
23874 only on certain machines. For the x86, GCC supports PIC for System V
23875 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
23876 position-independent.
23878 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
23883 If supported for the target machine, emit position-independent code,
23884 suitable for dynamic linking and avoiding any limit on the size of the
23885 global offset table. This option makes a difference on the m68k,
23886 PowerPC and SPARC@.
23888 Position-independent code requires special support, and therefore works
23889 only on certain machines.
23891 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
23898 These options are similar to @option{-fpic} and @option{-fPIC}, but
23899 generated position independent code can be only linked into executables.
23900 Usually these options are used when @option{-pie} GCC option is
23901 used during linking.
23903 @option{-fpie} and @option{-fPIE} both define the macros
23904 @code{__pie__} and @code{__PIE__}. The macros have the value 1
23905 for @option{-fpie} and 2 for @option{-fPIE}.
23909 Do not use PLT for external function calls in position-independent code.
23910 Instead, load callee address at call site from GOT and branch to it.
23911 This leads to more efficient code by eliminating PLT stubs and exposing
23912 GOT load to optimizations. On architectures such as 32-bit x86 where
23913 PLT stubs expect GOT pointer in a specific register, this gives more
23914 register allocation freedom to the compiler. Lazy binding requires PLT:
23915 with @option{-fno-plt} all external symbols are resolved at load time.
23917 Alternatively, function attribute @code{noplt} can be used to avoid PLT
23918 for calls to specific external functions by marking those functions with
23921 Additionally, a few targets also convert calls to those functions that are
23922 marked to not use the PLT to use the GOT instead for non-position independent
23925 @item -fno-jump-tables
23926 @opindex fno-jump-tables
23927 Do not use jump tables for switch statements even where it would be
23928 more efficient than other code generation strategies. This option is
23929 of use in conjunction with @option{-fpic} or @option{-fPIC} for
23930 building code that forms part of a dynamic linker and cannot
23931 reference the address of a jump table. On some targets, jump tables
23932 do not require a GOT and this option is not needed.
23934 @item -ffixed-@var{reg}
23936 Treat the register named @var{reg} as a fixed register; generated code
23937 should never refer to it (except perhaps as a stack pointer, frame
23938 pointer or in some other fixed role).
23940 @var{reg} must be the name of a register. The register names accepted
23941 are machine-specific and are defined in the @code{REGISTER_NAMES}
23942 macro in the machine description macro file.
23944 This flag does not have a negative form, because it specifies a
23947 @item -fcall-used-@var{reg}
23948 @opindex fcall-used
23949 Treat the register named @var{reg} as an allocable register that is
23950 clobbered by function calls. It may be allocated for temporaries or
23951 variables that do not live across a call. Functions compiled this way
23952 do not save and restore the register @var{reg}.
23954 It is an error to use this flag with the frame pointer or stack pointer.
23955 Use of this flag for other registers that have fixed pervasive roles in
23956 the machine's execution model produces disastrous results.
23958 This flag does not have a negative form, because it specifies a
23961 @item -fcall-saved-@var{reg}
23962 @opindex fcall-saved
23963 Treat the register named @var{reg} as an allocable register saved by
23964 functions. It may be allocated even for temporaries or variables that
23965 live across a call. Functions compiled this way save and restore
23966 the register @var{reg} if they use it.
23968 It is an error to use this flag with the frame pointer or stack pointer.
23969 Use of this flag for other registers that have fixed pervasive roles in
23970 the machine's execution model produces disastrous results.
23972 A different sort of disaster results from the use of this flag for
23973 a register in which function values may be returned.
23975 This flag does not have a negative form, because it specifies a
23978 @item -fpack-struct[=@var{n}]
23979 @opindex fpack-struct
23980 Without a value specified, pack all structure members together without
23981 holes. When a value is specified (which must be a small power of two), pack
23982 structure members according to this value, representing the maximum
23983 alignment (that is, objects with default alignment requirements larger than
23984 this are output potentially unaligned at the next fitting location.
23986 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
23987 code that is not binary compatible with code generated without that switch.
23988 Additionally, it makes the code suboptimal.
23989 Use it to conform to a non-default application binary interface.
23991 @item -finstrument-functions
23992 @opindex finstrument-functions
23993 Generate instrumentation calls for entry and exit to functions. Just
23994 after function entry and just before function exit, the following
23995 profiling functions are called with the address of the current
23996 function and its call site. (On some platforms,
23997 @code{__builtin_return_address} does not work beyond the current
23998 function, so the call site information may not be available to the
23999 profiling functions otherwise.)
24002 void __cyg_profile_func_enter (void *this_fn,
24004 void __cyg_profile_func_exit (void *this_fn,
24008 The first argument is the address of the start of the current function,
24009 which may be looked up exactly in the symbol table.
24011 This instrumentation is also done for functions expanded inline in other
24012 functions. The profiling calls indicate where, conceptually, the
24013 inline function is entered and exited. This means that addressable
24014 versions of such functions must be available. If all your uses of a
24015 function are expanded inline, this may mean an additional expansion of
24016 code size. If you use @code{extern inline} in your C code, an
24017 addressable version of such functions must be provided. (This is
24018 normally the case anyway, but if you get lucky and the optimizer always
24019 expands the functions inline, you might have gotten away without
24020 providing static copies.)
24022 A function may be given the attribute @code{no_instrument_function}, in
24023 which case this instrumentation is not done. This can be used, for
24024 example, for the profiling functions listed above, high-priority
24025 interrupt routines, and any functions from which the profiling functions
24026 cannot safely be called (perhaps signal handlers, if the profiling
24027 routines generate output or allocate memory).
24029 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
24030 @opindex finstrument-functions-exclude-file-list
24032 Set the list of functions that are excluded from instrumentation (see
24033 the description of @option{-finstrument-functions}). If the file that
24034 contains a function definition matches with one of @var{file}, then
24035 that function is not instrumented. The match is done on substrings:
24036 if the @var{file} parameter is a substring of the file name, it is
24037 considered to be a match.
24042 -finstrument-functions-exclude-file-list=/bits/stl,include/sys
24046 excludes any inline function defined in files whose pathnames
24047 contain @file{/bits/stl} or @file{include/sys}.
24049 If, for some reason, you want to include letter @samp{,} in one of
24050 @var{sym}, write @samp{\,}. For example,
24051 @option{-finstrument-functions-exclude-file-list='\,\,tmp'}
24052 (note the single quote surrounding the option).
24054 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
24055 @opindex finstrument-functions-exclude-function-list
24057 This is similar to @option{-finstrument-functions-exclude-file-list},
24058 but this option sets the list of function names to be excluded from
24059 instrumentation. The function name to be matched is its user-visible
24060 name, such as @code{vector<int> blah(const vector<int> &)}, not the
24061 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
24062 match is done on substrings: if the @var{sym} parameter is a substring
24063 of the function name, it is considered to be a match. For C99 and C++
24064 extended identifiers, the function name must be given in UTF-8, not
24065 using universal character names.
24067 @item -fstack-check
24068 @opindex fstack-check
24069 Generate code to verify that you do not go beyond the boundary of the
24070 stack. You should specify this flag if you are running in an
24071 environment with multiple threads, but you only rarely need to specify it in
24072 a single-threaded environment since stack overflow is automatically
24073 detected on nearly all systems if there is only one stack.
24075 Note that this switch does not actually cause checking to be done; the
24076 operating system or the language runtime must do that. The switch causes
24077 generation of code to ensure that they see the stack being extended.
24079 You can additionally specify a string parameter: @samp{no} means no
24080 checking, @samp{generic} means force the use of old-style checking,
24081 @samp{specific} means use the best checking method and is equivalent
24082 to bare @option{-fstack-check}.
24084 Old-style checking is a generic mechanism that requires no specific
24085 target support in the compiler but comes with the following drawbacks:
24089 Modified allocation strategy for large objects: they are always
24090 allocated dynamically if their size exceeds a fixed threshold.
24093 Fixed limit on the size of the static frame of functions: when it is
24094 topped by a particular function, stack checking is not reliable and
24095 a warning is issued by the compiler.
24098 Inefficiency: because of both the modified allocation strategy and the
24099 generic implementation, code performance is hampered.
24102 Note that old-style stack checking is also the fallback method for
24103 @samp{specific} if no target support has been added in the compiler.
24105 @item -fstack-limit-register=@var{reg}
24106 @itemx -fstack-limit-symbol=@var{sym}
24107 @itemx -fno-stack-limit
24108 @opindex fstack-limit-register
24109 @opindex fstack-limit-symbol
24110 @opindex fno-stack-limit
24111 Generate code to ensure that the stack does not grow beyond a certain value,
24112 either the value of a register or the address of a symbol. If a larger
24113 stack is required, a signal is raised at run time. For most targets,
24114 the signal is raised before the stack overruns the boundary, so
24115 it is possible to catch the signal without taking special precautions.
24117 For instance, if the stack starts at absolute address @samp{0x80000000}
24118 and grows downwards, you can use the flags
24119 @option{-fstack-limit-symbol=__stack_limit} and
24120 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
24121 of 128KB@. Note that this may only work with the GNU linker.
24123 @item -fsplit-stack
24124 @opindex fsplit-stack
24125 Generate code to automatically split the stack before it overflows.
24126 The resulting program has a discontiguous stack which can only
24127 overflow if the program is unable to allocate any more memory. This
24128 is most useful when running threaded programs, as it is no longer
24129 necessary to calculate a good stack size to use for each thread. This
24130 is currently only implemented for the x86 targets running
24133 When code compiled with @option{-fsplit-stack} calls code compiled
24134 without @option{-fsplit-stack}, there may not be much stack space
24135 available for the latter code to run. If compiling all code,
24136 including library code, with @option{-fsplit-stack} is not an option,
24137 then the linker can fix up these calls so that the code compiled
24138 without @option{-fsplit-stack} always has a large stack. Support for
24139 this is implemented in the gold linker in GNU binutils release 2.21
24142 @item -fleading-underscore
24143 @opindex fleading-underscore
24144 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
24145 change the way C symbols are represented in the object file. One use
24146 is to help link with legacy assembly code.
24148 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
24149 generate code that is not binary compatible with code generated without that
24150 switch. Use it to conform to a non-default application binary interface.
24151 Not all targets provide complete support for this switch.
24153 @item -ftls-model=@var{model}
24154 @opindex ftls-model
24155 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
24156 The @var{model} argument should be one of @samp{global-dynamic},
24157 @samp{local-dynamic}, @samp{initial-exec} or @samp{local-exec}.
24158 Note that the choice is subject to optimization: the compiler may use
24159 a more efficient model for symbols not visible outside of the translation
24160 unit, or if @option{-fpic} is not given on the command line.
24162 The default without @option{-fpic} is @samp{initial-exec}; with
24163 @option{-fpic} the default is @samp{global-dynamic}.
24165 @item -fvisibility=@r{[}default@r{|}internal@r{|}hidden@r{|}protected@r{]}
24166 @opindex fvisibility
24167 Set the default ELF image symbol visibility to the specified option---all
24168 symbols are marked with this unless overridden within the code.
24169 Using this feature can very substantially improve linking and
24170 load times of shared object libraries, produce more optimized
24171 code, provide near-perfect API export and prevent symbol clashes.
24172 It is @strong{strongly} recommended that you use this in any shared objects
24175 Despite the nomenclature, @samp{default} always means public; i.e.,
24176 available to be linked against from outside the shared object.
24177 @samp{protected} and @samp{internal} are pretty useless in real-world
24178 usage so the only other commonly used option is @samp{hidden}.
24179 The default if @option{-fvisibility} isn't specified is
24180 @samp{default}, i.e., make every symbol public.
24182 A good explanation of the benefits offered by ensuring ELF
24183 symbols have the correct visibility is given by ``How To Write
24184 Shared Libraries'' by Ulrich Drepper (which can be found at
24185 @w{@uref{http://www.akkadia.org/drepper/}})---however a superior
24186 solution made possible by this option to marking things hidden when
24187 the default is public is to make the default hidden and mark things
24188 public. This is the norm with DLLs on Windows and with @option{-fvisibility=hidden}
24189 and @code{__attribute__ ((visibility("default")))} instead of
24190 @code{__declspec(dllexport)} you get almost identical semantics with
24191 identical syntax. This is a great boon to those working with
24192 cross-platform projects.
24194 For those adding visibility support to existing code, you may find
24195 @code{#pragma GCC visibility} of use. This works by you enclosing
24196 the declarations you wish to set visibility for with (for example)
24197 @code{#pragma GCC visibility push(hidden)} and
24198 @code{#pragma GCC visibility pop}.
24199 Bear in mind that symbol visibility should be viewed @strong{as
24200 part of the API interface contract} and thus all new code should
24201 always specify visibility when it is not the default; i.e., declarations
24202 only for use within the local DSO should @strong{always} be marked explicitly
24203 as hidden as so to avoid PLT indirection overheads---making this
24204 abundantly clear also aids readability and self-documentation of the code.
24205 Note that due to ISO C++ specification requirements, @code{operator new} and
24206 @code{operator delete} must always be of default visibility.
24208 Be aware that headers from outside your project, in particular system
24209 headers and headers from any other library you use, may not be
24210 expecting to be compiled with visibility other than the default. You
24211 may need to explicitly say @code{#pragma GCC visibility push(default)}
24212 before including any such headers.
24214 @code{extern} declarations are not affected by @option{-fvisibility}, so
24215 a lot of code can be recompiled with @option{-fvisibility=hidden} with
24216 no modifications. However, this means that calls to @code{extern}
24217 functions with no explicit visibility use the PLT, so it is more
24218 effective to use @code{__attribute ((visibility))} and/or
24219 @code{#pragma GCC visibility} to tell the compiler which @code{extern}
24220 declarations should be treated as hidden.
24222 Note that @option{-fvisibility} does affect C++ vague linkage
24223 entities. This means that, for instance, an exception class that is
24224 be thrown between DSOs must be explicitly marked with default
24225 visibility so that the @samp{type_info} nodes are unified between
24228 An overview of these techniques, their benefits and how to use them
24229 is at @uref{http://gcc.gnu.org/@/wiki/@/Visibility}.
24231 @item -fstrict-volatile-bitfields
24232 @opindex fstrict-volatile-bitfields
24233 This option should be used if accesses to volatile bit-fields (or other
24234 structure fields, although the compiler usually honors those types
24235 anyway) should use a single access of the width of the
24236 field's type, aligned to a natural alignment if possible. For
24237 example, targets with memory-mapped peripheral registers might require
24238 all such accesses to be 16 bits wide; with this flag you can
24239 declare all peripheral bit-fields as @code{unsigned short} (assuming short
24240 is 16 bits on these targets) to force GCC to use 16-bit accesses
24241 instead of, perhaps, a more efficient 32-bit access.
24243 If this option is disabled, the compiler uses the most efficient
24244 instruction. In the previous example, that might be a 32-bit load
24245 instruction, even though that accesses bytes that do not contain
24246 any portion of the bit-field, or memory-mapped registers unrelated to
24247 the one being updated.
24249 In some cases, such as when the @code{packed} attribute is applied to a
24250 structure field, it may not be possible to access the field with a single
24251 read or write that is correctly aligned for the target machine. In this
24252 case GCC falls back to generating multiple accesses rather than code that
24253 will fault or truncate the result at run time.
24255 Note: Due to restrictions of the C/C++11 memory model, write accesses are
24256 not allowed to touch non bit-field members. It is therefore recommended
24257 to define all bits of the field's type as bit-field members.
24259 The default value of this option is determined by the application binary
24260 interface for the target processor.
24262 @item -fsync-libcalls
24263 @opindex fsync-libcalls
24264 This option controls whether any out-of-line instance of the @code{__sync}
24265 family of functions may be used to implement the C++11 @code{__atomic}
24266 family of functions.
24268 The default value of this option is enabled, thus the only useful form
24269 of the option is @option{-fno-sync-libcalls}. This option is used in
24270 the implementation of the @file{libatomic} runtime library.
24276 @node Environment Variables
24277 @section Environment Variables Affecting GCC
24278 @cindex environment variables
24280 @c man begin ENVIRONMENT
24281 This section describes several environment variables that affect how GCC
24282 operates. Some of them work by specifying directories or prefixes to use
24283 when searching for various kinds of files. Some are used to specify other
24284 aspects of the compilation environment.
24286 Note that you can also specify places to search using options such as
24287 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
24288 take precedence over places specified using environment variables, which
24289 in turn take precedence over those specified by the configuration of GCC@.
24290 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
24291 GNU Compiler Collection (GCC) Internals}.
24296 @c @itemx LC_COLLATE
24298 @c @itemx LC_MONETARY
24299 @c @itemx LC_NUMERIC
24304 @c @findex LC_COLLATE
24305 @findex LC_MESSAGES
24306 @c @findex LC_MONETARY
24307 @c @findex LC_NUMERIC
24311 These environment variables control the way that GCC uses
24312 localization information which allows GCC to work with different
24313 national conventions. GCC inspects the locale categories
24314 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
24315 so. These locale categories can be set to any value supported by your
24316 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
24317 Kingdom encoded in UTF-8.
24319 The @env{LC_CTYPE} environment variable specifies character
24320 classification. GCC uses it to determine the character boundaries in
24321 a string; this is needed for some multibyte encodings that contain quote
24322 and escape characters that are otherwise interpreted as a string
24325 The @env{LC_MESSAGES} environment variable specifies the language to
24326 use in diagnostic messages.
24328 If the @env{LC_ALL} environment variable is set, it overrides the value
24329 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
24330 and @env{LC_MESSAGES} default to the value of the @env{LANG}
24331 environment variable. If none of these variables are set, GCC
24332 defaults to traditional C English behavior.
24336 If @env{TMPDIR} is set, it specifies the directory to use for temporary
24337 files. GCC uses temporary files to hold the output of one stage of
24338 compilation which is to be used as input to the next stage: for example,
24339 the output of the preprocessor, which is the input to the compiler
24342 @item GCC_COMPARE_DEBUG
24343 @findex GCC_COMPARE_DEBUG
24344 Setting @env{GCC_COMPARE_DEBUG} is nearly equivalent to passing
24345 @option{-fcompare-debug} to the compiler driver. See the documentation
24346 of this option for more details.
24348 @item GCC_EXEC_PREFIX
24349 @findex GCC_EXEC_PREFIX
24350 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
24351 names of the subprograms executed by the compiler. No slash is added
24352 when this prefix is combined with the name of a subprogram, but you can
24353 specify a prefix that ends with a slash if you wish.
24355 If @env{GCC_EXEC_PREFIX} is not set, GCC attempts to figure out
24356 an appropriate prefix to use based on the pathname it is invoked with.
24358 If GCC cannot find the subprogram using the specified prefix, it
24359 tries looking in the usual places for the subprogram.
24361 The default value of @env{GCC_EXEC_PREFIX} is
24362 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
24363 the installed compiler. In many cases @var{prefix} is the value
24364 of @code{prefix} when you ran the @file{configure} script.
24366 Other prefixes specified with @option{-B} take precedence over this prefix.
24368 This prefix is also used for finding files such as @file{crt0.o} that are
24371 In addition, the prefix is used in an unusual way in finding the
24372 directories to search for header files. For each of the standard
24373 directories whose name normally begins with @samp{/usr/local/lib/gcc}
24374 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
24375 replacing that beginning with the specified prefix to produce an
24376 alternate directory name. Thus, with @option{-Bfoo/}, GCC searches
24377 @file{foo/bar} just before it searches the standard directory
24378 @file{/usr/local/lib/bar}.
24379 If a standard directory begins with the configured
24380 @var{prefix} then the value of @var{prefix} is replaced by
24381 @env{GCC_EXEC_PREFIX} when looking for header files.
24383 @item COMPILER_PATH
24384 @findex COMPILER_PATH
24385 The value of @env{COMPILER_PATH} is a colon-separated list of
24386 directories, much like @env{PATH}. GCC tries the directories thus
24387 specified when searching for subprograms, if it can't find the
24388 subprograms using @env{GCC_EXEC_PREFIX}.
24391 @findex LIBRARY_PATH
24392 The value of @env{LIBRARY_PATH} is a colon-separated list of
24393 directories, much like @env{PATH}. When configured as a native compiler,
24394 GCC tries the directories thus specified when searching for special
24395 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
24396 using GCC also uses these directories when searching for ordinary
24397 libraries for the @option{-l} option (but directories specified with
24398 @option{-L} come first).
24402 @cindex locale definition
24403 This variable is used to pass locale information to the compiler. One way in
24404 which this information is used is to determine the character set to be used
24405 when character literals, string literals and comments are parsed in C and C++.
24406 When the compiler is configured to allow multibyte characters,
24407 the following values for @env{LANG} are recognized:
24411 Recognize JIS characters.
24413 Recognize SJIS characters.
24415 Recognize EUCJP characters.
24418 If @env{LANG} is not defined, or if it has some other value, then the
24419 compiler uses @code{mblen} and @code{mbtowc} as defined by the default locale to
24420 recognize and translate multibyte characters.
24424 Some additional environment variables affect the behavior of the
24427 @include cppenv.texi
24431 @node Precompiled Headers
24432 @section Using Precompiled Headers
24433 @cindex precompiled headers
24434 @cindex speed of compilation
24436 Often large projects have many header files that are included in every
24437 source file. The time the compiler takes to process these header files
24438 over and over again can account for nearly all of the time required to
24439 build the project. To make builds faster, GCC allows you to
24440 @dfn{precompile} a header file.
24442 To create a precompiled header file, simply compile it as you would any
24443 other file, if necessary using the @option{-x} option to make the driver
24444 treat it as a C or C++ header file. You may want to use a
24445 tool like @command{make} to keep the precompiled header up-to-date when
24446 the headers it contains change.
24448 A precompiled header file is searched for when @code{#include} is
24449 seen in the compilation. As it searches for the included file
24450 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
24451 compiler looks for a precompiled header in each directory just before it
24452 looks for the include file in that directory. The name searched for is
24453 the name specified in the @code{#include} with @samp{.gch} appended. If
24454 the precompiled header file can't be used, it is ignored.
24456 For instance, if you have @code{#include "all.h"}, and you have
24457 @file{all.h.gch} in the same directory as @file{all.h}, then the
24458 precompiled header file is used if possible, and the original
24459 header is used otherwise.
24461 Alternatively, you might decide to put the precompiled header file in a
24462 directory and use @option{-I} to ensure that directory is searched
24463 before (or instead of) the directory containing the original header.
24464 Then, if you want to check that the precompiled header file is always
24465 used, you can put a file of the same name as the original header in this
24466 directory containing an @code{#error} command.
24468 This also works with @option{-include}. So yet another way to use
24469 precompiled headers, good for projects not designed with precompiled
24470 header files in mind, is to simply take most of the header files used by
24471 a project, include them from another header file, precompile that header
24472 file, and @option{-include} the precompiled header. If the header files
24473 have guards against multiple inclusion, they are skipped because
24474 they've already been included (in the precompiled header).
24476 If you need to precompile the same header file for different
24477 languages, targets, or compiler options, you can instead make a
24478 @emph{directory} named like @file{all.h.gch}, and put each precompiled
24479 header in the directory, perhaps using @option{-o}. It doesn't matter
24480 what you call the files in the directory; every precompiled header in
24481 the directory is considered. The first precompiled header
24482 encountered in the directory that is valid for this compilation is
24483 used; they're searched in no particular order.
24485 There are many other possibilities, limited only by your imagination,
24486 good sense, and the constraints of your build system.
24488 A precompiled header file can be used only when these conditions apply:
24492 Only one precompiled header can be used in a particular compilation.
24495 A precompiled header can't be used once the first C token is seen. You
24496 can have preprocessor directives before a precompiled header; you cannot
24497 include a precompiled header from inside another header.
24500 The precompiled header file must be produced for the same language as
24501 the current compilation. You can't use a C precompiled header for a C++
24505 The precompiled header file must have been produced by the same compiler
24506 binary as the current compilation is using.
24509 Any macros defined before the precompiled header is included must
24510 either be defined in the same way as when the precompiled header was
24511 generated, or must not affect the precompiled header, which usually
24512 means that they don't appear in the precompiled header at all.
24514 The @option{-D} option is one way to define a macro before a
24515 precompiled header is included; using a @code{#define} can also do it.
24516 There are also some options that define macros implicitly, like
24517 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
24520 @item If debugging information is output when using the precompiled
24521 header, using @option{-g} or similar, the same kind of debugging information
24522 must have been output when building the precompiled header. However,
24523 a precompiled header built using @option{-g} can be used in a compilation
24524 when no debugging information is being output.
24526 @item The same @option{-m} options must generally be used when building
24527 and using the precompiled header. @xref{Submodel Options},
24528 for any cases where this rule is relaxed.
24530 @item Each of the following options must be the same when building and using
24531 the precompiled header:
24533 @gccoptlist{-fexceptions}
24536 Some other command-line options starting with @option{-f},
24537 @option{-p}, or @option{-O} must be defined in the same way as when
24538 the precompiled header was generated. At present, it's not clear
24539 which options are safe to change and which are not; the safest choice
24540 is to use exactly the same options when generating and using the
24541 precompiled header. The following are known to be safe:
24543 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
24544 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
24545 -fsched-verbose=@var{number} -fschedule-insns -fvisibility= @gol
24550 For all of these except the last, the compiler automatically
24551 ignores the precompiled header if the conditions aren't met. If you
24552 find an option combination that doesn't work and doesn't cause the
24553 precompiled header to be ignored, please consider filing a bug report,
24556 If you do use differing options when generating and using the
24557 precompiled header, the actual behavior is a mixture of the
24558 behavior for the options. For instance, if you use @option{-g} to
24559 generate the precompiled header but not when using it, you may or may
24560 not get debugging information for routines in the precompiled header.