1 @c Copyright (C) 1988-2017 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{} 1988-2017 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), dbx(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.
75 @xref{Overall Options,,Options Controlling the Kind of Output}.
77 Other options are passed on to one or more stages of processing. Some options
78 control the preprocessor and others the compiler itself. Yet other
79 options control the assembler and linker; most of these are not
80 documented here, since you rarely need to use any of them.
82 @cindex C compilation options
83 Most of the command-line options that you can use with GCC are useful
84 for C programs; when an option is only useful with another language
85 (usually C++), the explanation says so explicitly. If the description
86 for a particular option does not mention a source language, you can use
87 that option with all supported languages.
89 @cindex cross compiling
90 @cindex specifying machine version
91 @cindex specifying compiler version and target machine
92 @cindex compiler version, specifying
93 @cindex target machine, specifying
94 The usual way to run GCC is to run the executable called @command{gcc}, or
95 @command{@var{machine}-gcc} when cross-compiling, or
96 @command{@var{machine}-gcc-@var{version}} to run a specific version of GCC.
97 When you compile C++ programs, you should invoke GCC as @command{g++}
98 instead. @xref{Invoking G++,,Compiling C++ Programs},
99 for information about the differences in behavior between @command{gcc}
100 and @code{g++} when compiling C++ programs.
102 @cindex grouping options
103 @cindex options, grouping
104 The @command{gcc} program accepts options and file names as operands. Many
105 options have multi-letter names; therefore multiple single-letter options
106 may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
109 @cindex order of options
110 @cindex options, order
111 You can mix options and other arguments. For the most part, the order
112 you use doesn't matter. Order does matter when you use several
113 options of the same kind; for example, if you specify @option{-L} more
114 than once, the directories are searched in the order specified. Also,
115 the placement of the @option{-l} option is significant.
117 Many options have long names starting with @samp{-f} or with
118 @samp{-W}---for example,
119 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
120 these have both positive and negative forms; the negative form of
121 @option{-ffoo} is @option{-fno-foo}. This manual documents
122 only one of these two forms, whichever one is not the default.
126 @xref{Option Index}, for an index to GCC's options.
129 * Option Summary:: Brief list of all options, without explanations.
130 * Overall Options:: Controlling the kind of output:
131 an executable, object files, assembler files,
132 or preprocessed source.
133 * Invoking G++:: Compiling C++ programs.
134 * C Dialect Options:: Controlling the variant of C language compiled.
135 * C++ Dialect Options:: Variations on C++.
136 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
138 * Diagnostic Message Formatting Options:: Controlling how diagnostics should
140 * Warning Options:: How picky should the compiler be?
141 * Debugging Options:: Producing debuggable code.
142 * Optimize Options:: How much optimization?
143 * Instrumentation Options:: Enabling profiling and extra run-time error checking.
144 * Preprocessor Options:: Controlling header files and macro definitions.
145 Also, getting dependency information for Make.
146 * Assembler Options:: Passing options to the assembler.
147 * Link Options:: Specifying libraries and so on.
148 * Directory Options:: Where to find header files and libraries.
149 Where to find the compiler executable files.
150 * Code Gen Options:: Specifying conventions for function calls, data layout
152 * Developer Options:: Printing GCC configuration info, statistics, and
154 * Submodel Options:: Target-specific options, such as compiling for a
155 specific processor variant.
156 * Spec Files:: How to pass switches to sub-processes.
157 * Environment Variables:: Env vars that affect GCC.
158 * Precompiled Headers:: Compiling a header once, and using it many times.
164 @section Option Summary
166 Here is a summary of all the options, grouped by type. Explanations are
167 in the following sections.
170 @item Overall Options
171 @xref{Overall Options,,Options Controlling the Kind of Output}.
172 @gccoptlist{-c -S -E -o @var{file} -x @var{language} @gol
173 -v -### --help@r{[}=@var{class}@r{[},@dots{}@r{]]} --target-help --version @gol
174 -pass-exit-codes -pipe -specs=@var{file} -wrapper @gol
175 @@@var{file} -fplugin=@var{file} -fplugin-arg-@var{name}=@var{arg} @gol
176 -fdump-ada-spec@r{[}-slim@r{]} -fada-spec-parent=@var{unit} -fdump-go-spec=@var{file}}
178 @item C Language Options
179 @xref{C Dialect Options,,Options Controlling C Dialect}.
180 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
181 -fpermitted-flt-eval-methods=@var{standard} @gol
182 -aux-info @var{filename} -fallow-parameterless-variadic-functions @gol
183 -fno-asm -fno-builtin -fno-builtin-@var{function} -fgimple@gol
184 -fhosted -ffreestanding -fopenacc -fopenmp -fopenmp-simd @gol
185 -fms-extensions -fplan9-extensions -fsso-struct=@var{endianness} @gol
186 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
187 -fsigned-bitfields -fsigned-char @gol
188 -funsigned-bitfields -funsigned-char}
190 @item C++ Language Options
191 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
192 @gccoptlist{-fabi-version=@var{n} -fno-access-control @gol
193 -faligned-new=@var{n} -fargs-in-order=@var{n} -fcheck-new @gol
194 -fconstexpr-depth=@var{n} -fconstexpr-loop-limit=@var{n} @gol
195 -ffriend-injection @gol
196 -fno-elide-constructors @gol
197 -fno-enforce-eh-specs @gol
198 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
199 -fno-implicit-templates @gol
200 -fno-implicit-inline-templates @gol
201 -fno-implement-inlines -fms-extensions @gol
202 -fnew-inheriting-ctors @gol
203 -fnew-ttp-matching @gol
204 -fno-nonansi-builtins -fnothrow-opt -fno-operator-names @gol
205 -fno-optional-diags -fpermissive @gol
206 -fno-pretty-templates @gol
207 -frepo -fno-rtti -fsized-deallocation @gol
208 -ftemplate-backtrace-limit=@var{n} @gol
209 -ftemplate-depth=@var{n} @gol
210 -fno-threadsafe-statics -fuse-cxa-atexit @gol
211 -fno-weak -nostdinc++ @gol
212 -fvisibility-inlines-hidden @gol
213 -fvisibility-ms-compat @gol
214 -fext-numeric-literals @gol
215 -Wabi=@var{n} -Wabi-tag -Wconversion-null -Wctor-dtor-privacy @gol
216 -Wdelete-non-virtual-dtor -Wliteral-suffix -Wmultiple-inheritance @gol
217 -Wnamespaces -Wnarrowing @gol
218 -Wnoexcept -Wnoexcept-type -Wclass-memaccess @gol
219 -Wnon-virtual-dtor -Wreorder -Wregister @gol
220 -Weffc++ -Wstrict-null-sentinel -Wtemplates @gol
221 -Wno-non-template-friend -Wold-style-cast @gol
222 -Woverloaded-virtual -Wno-pmf-conversions @gol
223 -Wsign-promo -Wvirtual-inheritance}
225 @item Objective-C and Objective-C++ Language Options
226 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
227 Objective-C and Objective-C++ Dialects}.
228 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
229 -fgnu-runtime -fnext-runtime @gol
230 -fno-nil-receivers @gol
231 -fobjc-abi-version=@var{n} @gol
232 -fobjc-call-cxx-cdtors @gol
233 -fobjc-direct-dispatch @gol
234 -fobjc-exceptions @gol
237 -fobjc-std=objc1 @gol
238 -fno-local-ivars @gol
239 -fivar-visibility=@r{[}public@r{|}protected@r{|}private@r{|}package@r{]} @gol
240 -freplace-objc-classes @gol
243 -Wassign-intercept @gol
244 -Wno-protocol -Wselector @gol
245 -Wstrict-selector-match @gol
246 -Wundeclared-selector}
248 @item Diagnostic Message Formatting Options
249 @xref{Diagnostic Message Formatting Options,,Options to Control Diagnostic Messages Formatting}.
250 @gccoptlist{-fmessage-length=@var{n} @gol
251 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
252 -fdiagnostics-color=@r{[}auto@r{|}never@r{|}always@r{]} @gol
253 -fno-diagnostics-show-option -fno-diagnostics-show-caret @gol
254 -fdiagnostics-parseable-fixits -fdiagnostics-generate-patch @gol
255 -fdiagnostics-show-template-tree -fno-elide-type @gol
258 @item Warning Options
259 @xref{Warning Options,,Options to Request or Suppress Warnings}.
260 @gccoptlist{-fsyntax-only -fmax-errors=@var{n} -Wpedantic @gol
261 -pedantic-errors @gol
262 -w -Wextra -Wall -Waddress -Waggregate-return @gol
263 -Walloc-zero -Walloc-size-larger-than=@var{n}
264 -Walloca -Walloca-larger-than=@var{n} @gol
265 -Wno-aggressive-loop-optimizations -Warray-bounds -Warray-bounds=@var{n} @gol
266 -Wno-attributes -Wbool-compare -Wbool-operation @gol
267 -Wno-builtin-declaration-mismatch @gol
268 -Wno-builtin-macro-redefined -Wc90-c99-compat -Wc99-c11-compat @gol
269 -Wc++-compat -Wc++11-compat -Wc++14-compat @gol
270 -Wcast-align -Wcast-align=strict -Wcast-qual @gol
271 -Wchar-subscripts -Wchkp -Wcatch-value -Wcatch-value=@var{n} @gol
272 -Wclobbered -Wcomment -Wconditionally-supported @gol
273 -Wconversion -Wcoverage-mismatch -Wno-cpp -Wdangling-else -Wdate-time @gol
274 -Wdelete-incomplete @gol
275 -Wno-deprecated -Wno-deprecated-declarations -Wno-designated-init @gol
276 -Wdisabled-optimization @gol
277 -Wno-discarded-qualifiers -Wno-discarded-array-qualifiers @gol
278 -Wno-div-by-zero -Wdouble-promotion @gol
279 -Wduplicated-branches -Wduplicated-cond @gol
280 -Wempty-body -Wenum-compare -Wno-endif-labels -Wexpansion-to-defined @gol
281 -Werror -Werror=* -Wextra-semi -Wfatal-errors @gol
282 -Wfloat-equal -Wformat -Wformat=2 @gol
283 -Wno-format-contains-nul -Wno-format-extra-args @gol
284 -Wformat-nonliteral -Wformat-overflow=@var{n} @gol
285 -Wformat-security -Wformat-signedness -Wformat-truncation=@var{n} @gol
286 -Wformat-y2k -Wframe-address @gol
287 -Wframe-larger-than=@var{len} -Wno-free-nonheap-object -Wjump-misses-init @gol
288 -Wif-not-aligned @gol
289 -Wignored-qualifiers -Wignored-attributes -Wincompatible-pointer-types @gol
290 -Wimplicit -Wimplicit-fallthrough -Wimplicit-fallthrough=@var{n} @gol
291 -Wimplicit-function-declaration -Wimplicit-int @gol
292 -Winit-self -Winline -Wno-int-conversion -Wint-in-bool-context @gol
293 -Wno-int-to-pointer-cast -Winvalid-memory-model -Wno-invalid-offsetof @gol
294 -Winvalid-pch -Wlarger-than=@var{len} @gol
295 -Wlogical-op -Wlogical-not-parentheses -Wlong-long @gol
296 -Wmain -Wmaybe-uninitialized -Wmemset-elt-size -Wmemset-transposed-args @gol
297 -Wmisleading-indentation -Wmissing-braces @gol
298 -Wmissing-field-initializers -Wmissing-include-dirs @gol
299 -Wno-multichar -Wmultistatement-macros -Wnonnull -Wnonnull-compare @gol
300 -Wnormalized=@r{[}none@r{|}id@r{|}nfc@r{|}nfkc@r{]} @gol
301 -Wnull-dereference -Wodr -Wno-overflow -Wopenmp-simd @gol
302 -Woverride-init-side-effects -Woverlength-strings @gol
303 -Wpacked -Wpacked-bitfield-compat -Wpacked-not-aligned -Wpadded @gol
304 -Wparentheses -Wno-pedantic-ms-format @gol
305 -Wplacement-new -Wplacement-new=@var{n} @gol
306 -Wpointer-arith -Wpointer-compare -Wno-pointer-to-int-cast @gol
307 -Wno-pragmas -Wredundant-decls -Wrestrict -Wno-return-local-addr @gol
308 -Wreturn-type -Wsequence-point -Wshadow -Wno-shadow-ivar @gol
309 -Wshadow=global, -Wshadow=local, -Wshadow=compatible-local @gol
310 -Wshift-overflow -Wshift-overflow=@var{n} @gol
311 -Wshift-count-negative -Wshift-count-overflow -Wshift-negative-value @gol
312 -Wsign-compare -Wsign-conversion -Wfloat-conversion @gol
313 -Wno-scalar-storage-order -Wsizeof-pointer-div @gol
314 -Wsizeof-pointer-memaccess -Wsizeof-array-argument @gol
315 -Wstack-protector -Wstack-usage=@var{len} -Wstrict-aliasing @gol
316 -Wstrict-aliasing=n -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
317 -Wstringop-overflow=@var{n} -Wstringop-truncation @gol
318 -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{|}malloc@r{]} @gol
319 -Wsuggest-final-types @gol -Wsuggest-final-methods -Wsuggest-override @gol
320 -Wmissing-format-attribute -Wsubobject-linkage @gol
321 -Wswitch -Wswitch-bool -Wswitch-default -Wswitch-enum @gol
322 -Wswitch-unreachable -Wsync-nand @gol
323 -Wsystem-headers -Wtautological-compare -Wtrampolines -Wtrigraphs @gol
324 -Wtype-limits -Wundef @gol
325 -Wuninitialized -Wunknown-pragmas -Wunsafe-loop-optimizations @gol
326 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
327 -Wunused-label -Wunused-local-typedefs -Wunused-macros @gol
328 -Wunused-parameter -Wno-unused-result @gol
329 -Wunused-value -Wunused-variable @gol
330 -Wunused-const-variable -Wunused-const-variable=@var{n} @gol
331 -Wunused-but-set-parameter -Wunused-but-set-variable @gol
332 -Wuseless-cast -Wvariadic-macros -Wvector-operation-performance @gol
333 -Wvla -Wvla-larger-than=@var{n} -Wvolatile-register-var -Wwrite-strings @gol
334 -Wzero-as-null-pointer-constant -Whsa}
336 @item C and Objective-C-only Warning Options
337 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
338 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
339 -Wold-style-declaration -Wold-style-definition @gol
340 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
341 -Wdeclaration-after-statement -Wpointer-sign}
343 @item Debugging Options
344 @xref{Debugging Options,,Options for Debugging Your Program}.
345 @gccoptlist{-g -g@var{level} -gdwarf -gdwarf-@var{version} @gol
346 -ggdb -grecord-gcc-switches -gno-record-gcc-switches @gol
347 -gstabs -gstabs+ -gstrict-dwarf -gno-strict-dwarf @gol
348 -gcolumn-info -gno-column-info @gol
349 -gvms -gxcoff -gxcoff+ -gz@r{[}=@var{type}@r{]} @gol
350 -fdebug-prefix-map=@var{old}=@var{new} -fdebug-types-section @gol
351 -fno-eliminate-unused-debug-types @gol
352 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
353 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
354 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
355 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
356 -fvar-tracking -fvar-tracking-assignments}
358 @item Optimization Options
359 @xref{Optimize Options,,Options that Control Optimization}.
360 @gccoptlist{-faggressive-loop-optimizations -falign-functions[=@var{n}] @gol
361 -falign-jumps[=@var{n}] @gol
362 -falign-labels[=@var{n}] -falign-loops[=@var{n}] @gol
363 -fassociative-math -fauto-profile -fauto-profile[=@var{path}] @gol
364 -fauto-inc-dec -fbranch-probabilities @gol
365 -fbranch-target-load-optimize -fbranch-target-load-optimize2 @gol
366 -fbtr-bb-exclusive -fcaller-saves @gol
367 -fcombine-stack-adjustments -fconserve-stack @gol
368 -fcompare-elim -fcprop-registers -fcrossjumping @gol
369 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules @gol
370 -fcx-limited-range @gol
371 -fdata-sections -fdce -fdelayed-branch @gol
372 -fdelete-null-pointer-checks -fdevirtualize -fdevirtualize-speculatively @gol
373 -fdevirtualize-at-ltrans -fdse @gol
374 -fearly-inlining -fipa-sra -fexpensive-optimizations -ffat-lto-objects @gol
375 -ffast-math -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
376 -fforward-propagate -ffp-contract=@var{style} -ffunction-sections @gol
377 -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm -fgraphite-identity @gol
378 -fgcse-sm -fhoist-adjacent-loads -fif-conversion @gol
379 -fif-conversion2 -findirect-inlining @gol
380 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
381 -finline-small-functions -fipa-cp -fipa-cp-clone @gol
382 -fipa-bit-cp -fipa-vrp @gol
383 -fipa-pta -fipa-profile -fipa-pure-const -fipa-reference -fipa-icf @gol
384 -fira-algorithm=@var{algorithm} @gol
385 -fira-region=@var{region} -fira-hoist-pressure @gol
386 -fira-loop-pressure -fno-ira-share-save-slots @gol
387 -fno-ira-share-spill-slots @gol
388 -fisolate-erroneous-paths-dereference -fisolate-erroneous-paths-attribute @gol
389 -fivopts -fkeep-inline-functions -fkeep-static-functions @gol
390 -fkeep-static-consts -flimit-function-alignment -flive-range-shrinkage @gol
391 -floop-block -floop-interchange -floop-strip-mine @gol
392 -floop-unroll-and-jam -floop-nest-optimize @gol
393 -floop-parallelize-all -flra-remat -flto -flto-compression-level @gol
394 -flto-partition=@var{alg} -fmerge-all-constants @gol
395 -fmerge-constants -fmodulo-sched -fmodulo-sched-allow-regmoves @gol
396 -fmove-loop-invariants -fno-branch-count-reg @gol
397 -fno-defer-pop -fno-fp-int-builtin-inexact -fno-function-cse @gol
398 -fno-guess-branch-probability -fno-inline -fno-math-errno -fno-peephole @gol
399 -fno-peephole2 -fno-printf-return-value -fno-sched-interblock @gol
400 -fno-sched-spec -fno-signed-zeros @gol
401 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
402 -fomit-frame-pointer -foptimize-sibling-calls @gol
403 -fpartial-inlining -fpeel-loops -fpredictive-commoning @gol
404 -fprefetch-loop-arrays @gol
405 -fprofile-correction @gol
406 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
407 -fprofile-reorder-functions @gol
408 -freciprocal-math -free -frename-registers -freorder-blocks @gol
409 -freorder-blocks-algorithm=@var{algorithm} @gol
410 -freorder-blocks-and-partition -freorder-functions @gol
411 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
412 -frounding-math -fsched2-use-superblocks -fsched-pressure @gol
413 -fsched-spec-load -fsched-spec-load-dangerous @gol
414 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
415 -fsched-group-heuristic -fsched-critical-path-heuristic @gol
416 -fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
417 -fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
418 -fschedule-fusion @gol
419 -fschedule-insns -fschedule-insns2 -fsection-anchors @gol
420 -fselective-scheduling -fselective-scheduling2 @gol
421 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
422 -fsemantic-interposition -fshrink-wrap -fshrink-wrap-separate @gol
423 -fsignaling-nans @gol
424 -fsingle-precision-constant -fsplit-ivs-in-unroller -fsplit-loops@gol
426 -fsplit-wide-types -fssa-backprop -fssa-phiopt @gol
427 -fstdarg-opt -fstore-merging -fstrict-aliasing @gol
428 -fthread-jumps -ftracer -ftree-bit-ccp @gol
429 -ftree-builtin-call-dce -ftree-ccp -ftree-ch @gol
430 -ftree-coalesce-vars -ftree-copy-prop -ftree-dce -ftree-dominator-opts @gol
431 -ftree-dse -ftree-forwprop -ftree-fre -fcode-hoisting @gol
432 -ftree-loop-if-convert -ftree-loop-im @gol
433 -ftree-phiprop -ftree-loop-distribution -ftree-loop-distribute-patterns @gol
434 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
435 -ftree-loop-vectorize @gol
436 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-partial-pre -ftree-pta @gol
437 -ftree-reassoc -ftree-sink -ftree-slsr -ftree-sra @gol
438 -ftree-switch-conversion -ftree-tail-merge @gol
439 -ftree-ter -ftree-vectorize -ftree-vrp -funconstrained-commons @gol
440 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
441 -funsafe-math-optimizations -funswitch-loops @gol
442 -fipa-ra -fvariable-expansion-in-unroller -fvect-cost-model -fvpt @gol
443 -fweb -fwhole-program -fwpa -fuse-linker-plugin @gol
444 --param @var{name}=@var{value}
445 -O -O0 -O1 -O2 -O3 -Os -Ofast -Og}
447 @item Program Instrumentation Options
448 @xref{Instrumentation Options,,Program Instrumentation Options}.
449 @gccoptlist{-p -pg -fprofile-arcs --coverage -ftest-coverage @gol
450 -fprofile-abs-path @gol
451 -fprofile-dir=@var{path} -fprofile-generate -fprofile-generate=@var{path} @gol
452 -fsanitize=@var{style} -fsanitize-recover -fsanitize-recover=@var{style} @gol
453 -fasan-shadow-offset=@var{number} -fsanitize-sections=@var{s1},@var{s2},... @gol
454 -fsanitize-undefined-trap-on-error -fbounds-check @gol
455 -fcheck-pointer-bounds -fchkp-check-incomplete-type @gol
456 -fchkp-first-field-has-own-bounds -fchkp-narrow-bounds @gol
457 -fchkp-narrow-to-innermost-array -fchkp-optimize @gol
458 -fchkp-use-fast-string-functions -fchkp-use-nochk-string-functions @gol
459 -fchkp-use-static-bounds -fchkp-use-static-const-bounds @gol
460 -fchkp-treat-zero-dynamic-size-as-infinite -fchkp-check-read @gol
461 -fchkp-check-read -fchkp-check-write -fchkp-store-bounds @gol
462 -fchkp-instrument-calls -fchkp-instrument-marked-only @gol
463 -fchkp-use-wrappers -fchkp-flexible-struct-trailing-arrays@gol
464 -fcf-protection==@r{[}full@r{|}branch@r{|}return@r{|}none@r{]} @gol
465 -fstack-protector -fstack-protector-all -fstack-protector-strong @gol
466 -fstack-protector-explicit -fstack-check @gol
467 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
468 -fno-stack-limit -fsplit-stack @gol
469 -fvtable-verify=@r{[}std@r{|}preinit@r{|}none@r{]} @gol
470 -fvtv-counts -fvtv-debug @gol
471 -finstrument-functions @gol
472 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
473 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}}
475 @item Preprocessor Options
476 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
477 @gccoptlist{-A@var{question}=@var{answer} @gol
478 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
479 -C -CC -D@var{macro}@r{[}=@var{defn}@r{]} @gol
480 -dD -dI -dM -dN -dU @gol
481 -fdebug-cpp -fdirectives-only -fdollars-in-identifiers @gol
482 -fexec-charset=@var{charset} -fextended-identifiers @gol
483 -finput-charset=@var{charset} -fno-canonical-system-headers @gol
484 -fpch-deps -fpch-preprocess -fpreprocessed @gol
485 -ftabstop=@var{width} -ftrack-macro-expansion @gol
486 -fwide-exec-charset=@var{charset} -fworking-directory @gol
487 -H -imacros @var{file} -include @var{file} @gol
488 -M -MD -MF -MG -MM -MMD -MP -MQ -MT @gol
489 -no-integrated-cpp -P -pthread -remap @gol
490 -traditional -traditional-cpp -trigraphs @gol
491 -U@var{macro} -undef @gol
492 -Wp,@var{option} -Xpreprocessor @var{option}}
494 @item Assembler Options
495 @xref{Assembler Options,,Passing Options to the Assembler}.
496 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
499 @xref{Link Options,,Options for Linking}.
500 @gccoptlist{@var{object-file-name} -fuse-ld=@var{linker} -l@var{library} @gol
501 -nostartfiles -nodefaultlibs -nostdlib -pie -pthread -rdynamic @gol
502 -s -static -static-pie -static-libgcc -static-libstdc++ @gol
503 -static-libasan -static-libtsan -static-liblsan -static-libubsan @gol
504 -static-libmpx -static-libmpxwrappers @gol
505 -shared -shared-libgcc -symbolic @gol
506 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
507 -u @var{symbol} -z @var{keyword}}
509 @item Directory Options
510 @xref{Directory Options,,Options for Directory Search}.
511 @gccoptlist{-B@var{prefix} -I@var{dir} -I- @gol
512 -idirafter @var{dir} @gol
513 -imacros @var{file} -imultilib @var{dir} @gol
514 -iplugindir=@var{dir} -iprefix @var{file} @gol
515 -iquote @var{dir} -isysroot @var{dir} -isystem @var{dir} @gol
516 -iwithprefix @var{dir} -iwithprefixbefore @var{dir} @gol
517 -L@var{dir} -no-canonical-prefixes --no-sysroot-suffix @gol
518 -nostdinc -nostdinc++ --sysroot=@var{dir}}
520 @item Code Generation Options
521 @xref{Code Gen Options,,Options for Code Generation Conventions}.
522 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
523 -ffixed-@var{reg} -fexceptions @gol
524 -fnon-call-exceptions -fdelete-dead-exceptions -funwind-tables @gol
525 -fasynchronous-unwind-tables @gol
527 -finhibit-size-directive -fno-common -fno-ident @gol
528 -fpcc-struct-return -fpic -fPIC -fpie -fPIE -fno-plt @gol
529 -fno-jump-tables @gol
530 -frecord-gcc-switches @gol
531 -freg-struct-return -fshort-enums -fshort-wchar @gol
532 -fverbose-asm -fpack-struct[=@var{n}] @gol
533 -fleading-underscore -ftls-model=@var{model} @gol
534 -fstack-reuse=@var{reuse_level} @gol
535 -ftrampolines -ftrapv -fwrapv @gol
536 -fvisibility=@r{[}default@r{|}internal@r{|}hidden@r{|}protected@r{]} @gol
537 -fstrict-volatile-bitfields -fsync-libcalls}
539 @item Developer Options
540 @xref{Developer Options,,GCC Developer Options}.
541 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
542 -dumpfullversion -fchecking -fchecking=@var{n} -fdbg-cnt-list @gol
543 -fdbg-cnt=@var{counter-value-list} @gol
544 -fdisable-ipa-@var{pass_name} @gol
545 -fdisable-rtl-@var{pass_name} @gol
546 -fdisable-rtl-@var{pass-name}=@var{range-list} @gol
547 -fdisable-tree-@var{pass_name} @gol
548 -fdisable-tree-@var{pass-name}=@var{range-list} @gol
549 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
550 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
551 -fdump-final-insns@r{[}=@var{file}@r{]}
552 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
554 -fdump-lang-@var{switch} @gol
555 -fdump-lang-@var{switch}-@var{options} @gol
556 -fdump-lang-@var{switch}-@var{options}=@var{filename} @gol
558 -fdump-rtl-@var{pass} -fdump-rtl-@var{pass}=@var{filename} @gol
559 -fdump-statistics @gol
561 -fdump-tree-@var{switch} @gol
562 -fdump-tree-@var{switch}-@var{options} @gol
563 -fdump-tree-@var{switch}-@var{options}=@var{filename} @gol
564 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
565 -fenable-@var{kind}-@var{pass} @gol
566 -fenable-@var{kind}-@var{pass}=@var{range-list} @gol
567 -fira-verbose=@var{n} @gol
568 -flto-report -flto-report-wpa -fmem-report-wpa @gol
569 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report @gol
570 -fopt-info -fopt-info-@var{options}@r{[}=@var{file}@r{]} @gol
571 -fprofile-report @gol
572 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
573 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
574 -fstats -fstack-usage -ftime-report -ftime-report-details @gol
575 -fvar-tracking-assignments-toggle -gtoggle @gol
576 -print-file-name=@var{library} -print-libgcc-file-name @gol
577 -print-multi-directory -print-multi-lib -print-multi-os-directory @gol
578 -print-prog-name=@var{program} -print-search-dirs -Q @gol
579 -print-sysroot -print-sysroot-headers-suffix @gol
580 -save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
582 @item Machine-Dependent Options
583 @xref{Submodel Options,,Machine-Dependent Options}.
584 @c This list is ordered alphanumerically by subsection name.
585 @c Try and put the significant identifier (CPU or system) first,
586 @c so users have a clue at guessing where the ones they want will be.
588 @emph{AArch64 Options}
589 @gccoptlist{-mabi=@var{name} -mbig-endian -mlittle-endian @gol
590 -mgeneral-regs-only @gol
591 -mcmodel=tiny -mcmodel=small -mcmodel=large @gol
593 -momit-leaf-frame-pointer @gol
594 -mtls-dialect=desc -mtls-dialect=traditional @gol
595 -mtls-size=@var{size} @gol
596 -mfix-cortex-a53-835769 -mfix-cortex-a53-843419 @gol
597 -mlow-precision-recip-sqrt -mlow-precision-sqrt -mlow-precision-div @gol
598 -mpc-relative-literal-loads @gol
599 -msign-return-address=@var{scope} @gol
600 -march=@var{name} -mcpu=@var{name} -mtune=@var{name} -moverride=@var{string}}
602 @emph{Adapteva Epiphany Options}
603 @gccoptlist{-mhalf-reg-file -mprefer-short-insn-regs @gol
604 -mbranch-cost=@var{num} -mcmove -mnops=@var{num} -msoft-cmpsf @gol
605 -msplit-lohi -mpost-inc -mpost-modify -mstack-offset=@var{num} @gol
606 -mround-nearest -mlong-calls -mshort-calls -msmall16 @gol
607 -mfp-mode=@var{mode} -mvect-double -max-vect-align=@var{num} @gol
608 -msplit-vecmove-early -m1reg-@var{reg}}
611 @gccoptlist{-mbarrel-shifter @gol
612 -mcpu=@var{cpu} -mA6 -mARC600 -mA7 -mARC700 @gol
613 -mdpfp -mdpfp-compact -mdpfp-fast -mno-dpfp-lrsr @gol
614 -mea -mno-mpy -mmul32x16 -mmul64 -matomic @gol
615 -mnorm -mspfp -mspfp-compact -mspfp-fast -msimd -msoft-float -mswap @gol
616 -mcrc -mdsp-packa -mdvbf -mlock -mmac-d16 -mmac-24 -mrtsc -mswape @gol
617 -mtelephony -mxy -misize -mannotate-align -marclinux -marclinux_prof @gol
618 -mlong-calls -mmedium-calls -msdata -mirq-ctrl-saved @gol
619 -mrgf-banked-regs -mlpc-width=@var{width} -G @var{num} @gol
620 -mvolatile-cache -mtp-regno=@var{regno} @gol
621 -malign-call -mauto-modify-reg -mbbit-peephole -mno-brcc @gol
622 -mcase-vector-pcrel -mcompact-casesi -mno-cond-exec -mearly-cbranchsi @gol
623 -mexpand-adddi -mindexed-loads -mlra -mlra-priority-none @gol
624 -mlra-priority-compact mlra-priority-noncompact -mno-millicode @gol
625 -mmixed-code -mq-class -mRcq -mRcw -msize-level=@var{level} @gol
626 -mtune=@var{cpu} -mmultcost=@var{num} @gol
627 -munalign-prob-threshold=@var{probability} -mmpy-option=@var{multo} @gol
628 -mdiv-rem -mcode-density -mll64 -mfpu=@var{fpu}}
631 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
632 -mabi=@var{name} @gol
633 -mapcs-stack-check -mno-apcs-stack-check @gol
634 -mapcs-reentrant -mno-apcs-reentrant @gol
635 -msched-prolog -mno-sched-prolog @gol
636 -mlittle-endian -mbig-endian @gol
638 -mfloat-abi=@var{name} @gol
639 -mfp16-format=@var{name}
640 -mthumb-interwork -mno-thumb-interwork @gol
641 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
642 -mtune=@var{name} -mprint-tune-info @gol
643 -mstructure-size-boundary=@var{n} @gol
644 -mabort-on-noreturn @gol
645 -mlong-calls -mno-long-calls @gol
646 -msingle-pic-base -mno-single-pic-base @gol
647 -mpic-register=@var{reg} @gol
648 -mnop-fun-dllimport @gol
649 -mpoke-function-name @gol
651 -mtpcs-frame -mtpcs-leaf-frame @gol
652 -mcaller-super-interworking -mcallee-super-interworking @gol
653 -mtp=@var{name} -mtls-dialect=@var{dialect} @gol
654 -mword-relocations @gol
655 -mfix-cortex-m3-ldrd @gol
656 -munaligned-access @gol
657 -mneon-for-64bits @gol
658 -mslow-flash-data @gol
659 -masm-syntax-unified @gol
665 @gccoptlist{-mmcu=@var{mcu} -mabsdata -maccumulate-args @gol
666 -mbranch-cost=@var{cost} @gol
667 -mcall-prologues -mgas-isr-prologues -mint8 @gol
668 -mn_flash=@var{size} -mno-interrupts @gol
669 -mrelax -mrmw -mstrict-X -mtiny-stack -mfract-convert-truncate @gol
670 -mshort-calls -nodevicelib @gol
671 -Waddr-space-convert -Wmisspelled-isr}
673 @emph{Blackfin Options}
674 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
675 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
676 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
677 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
678 -mno-id-shared-library -mshared-library-id=@var{n} @gol
679 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
680 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
681 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
685 @gccoptlist{-mbig-endian -mlittle-endian -march=@var{cpu} @gol
686 -msim -msdata=@var{sdata-type}}
689 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
690 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
691 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
692 -mstack-align -mdata-align -mconst-align @gol
693 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
694 -melf -maout -melinux -mlinux -sim -sim2 @gol
695 -mmul-bug-workaround -mno-mul-bug-workaround}
698 @gccoptlist{-mmac @gol
699 -mcr16cplus -mcr16c @gol
700 -msim -mint32 -mbit-ops
701 -mdata-model=@var{model}}
703 @emph{Darwin Options}
704 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
705 -arch_only -bind_at_load -bundle -bundle_loader @gol
706 -client_name -compatibility_version -current_version @gol
708 -dependency-file -dylib_file -dylinker_install_name @gol
709 -dynamic -dynamiclib -exported_symbols_list @gol
710 -filelist -flat_namespace -force_cpusubtype_ALL @gol
711 -force_flat_namespace -headerpad_max_install_names @gol
713 -image_base -init -install_name -keep_private_externs @gol
714 -multi_module -multiply_defined -multiply_defined_unused @gol
715 -noall_load -no_dead_strip_inits_and_terms @gol
716 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
717 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
718 -private_bundle -read_only_relocs -sectalign @gol
719 -sectobjectsymbols -whyload -seg1addr @gol
720 -sectcreate -sectobjectsymbols -sectorder @gol
721 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
722 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
723 -segprot -segs_read_only_addr -segs_read_write_addr @gol
724 -single_module -static -sub_library -sub_umbrella @gol
725 -twolevel_namespace -umbrella -undefined @gol
726 -unexported_symbols_list -weak_reference_mismatches @gol
727 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
728 -mkernel -mone-byte-bool}
730 @emph{DEC Alpha Options}
731 @gccoptlist{-mno-fp-regs -msoft-float @gol
732 -mieee -mieee-with-inexact -mieee-conformant @gol
733 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
734 -mtrap-precision=@var{mode} -mbuild-constants @gol
735 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
736 -mbwx -mmax -mfix -mcix @gol
737 -mfloat-vax -mfloat-ieee @gol
738 -mexplicit-relocs -msmall-data -mlarge-data @gol
739 -msmall-text -mlarge-text @gol
740 -mmemory-latency=@var{time}}
743 @gccoptlist{-msmall-model -mno-lsim}
746 @gccoptlist{-msim -mlra -mnodiv -mft32b -mcompress -mnopm}
749 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
750 -mhard-float -msoft-float @gol
751 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
752 -mdouble -mno-double @gol
753 -mmedia -mno-media -mmuladd -mno-muladd @gol
754 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
755 -mlinked-fp -mlong-calls -malign-labels @gol
756 -mlibrary-pic -macc-4 -macc-8 @gol
757 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
758 -moptimize-membar -mno-optimize-membar @gol
759 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
760 -mvliw-branch -mno-vliw-branch @gol
761 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
762 -mno-nested-cond-exec -mtomcat-stats @gol
766 @emph{GNU/Linux Options}
767 @gccoptlist{-mglibc -muclibc -mmusl -mbionic -mandroid @gol
768 -tno-android-cc -tno-android-ld}
770 @emph{H8/300 Options}
771 @gccoptlist{-mrelax -mh -ms -mn -mexr -mno-exr -mint32 -malign-300}
774 @gccoptlist{-march=@var{architecture-type} @gol
775 -mcaller-copies -mdisable-fpregs -mdisable-indexing @gol
776 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
777 -mfixed-range=@var{register-range} @gol
778 -mjump-in-delay -mlinker-opt -mlong-calls @gol
779 -mlong-load-store -mno-disable-fpregs @gol
780 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
781 -mno-jump-in-delay -mno-long-load-store @gol
782 -mno-portable-runtime -mno-soft-float @gol
783 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
784 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
785 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
786 -munix=@var{unix-std} -nolibdld -static -threads}
789 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
790 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
791 -mconstant-gp -mauto-pic -mfused-madd @gol
792 -minline-float-divide-min-latency @gol
793 -minline-float-divide-max-throughput @gol
794 -mno-inline-float-divide @gol
795 -minline-int-divide-min-latency @gol
796 -minline-int-divide-max-throughput @gol
797 -mno-inline-int-divide @gol
798 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
799 -mno-inline-sqrt @gol
800 -mdwarf2-asm -mearly-stop-bits @gol
801 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
802 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
803 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
804 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
805 -msched-spec-ldc -msched-spec-control-ldc @gol
806 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
807 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
808 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
809 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
812 @gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
813 -msign-extend-enabled -muser-enabled}
815 @emph{M32R/D Options}
816 @gccoptlist{-m32r2 -m32rx -m32r @gol
818 -malign-loops -mno-align-loops @gol
819 -missue-rate=@var{number} @gol
820 -mbranch-cost=@var{number} @gol
821 -mmodel=@var{code-size-model-type} @gol
822 -msdata=@var{sdata-type} @gol
823 -mno-flush-func -mflush-func=@var{name} @gol
824 -mno-flush-trap -mflush-trap=@var{number} @gol
828 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
830 @emph{M680x0 Options}
831 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune} @gol
832 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
833 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
834 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
835 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
836 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
837 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
838 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
839 -mxgot -mno-xgot -mlong-jump-table-offsets}
842 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
843 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
844 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
845 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
846 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
849 @gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
850 -mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
851 -mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
852 -mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
855 @emph{MicroBlaze Options}
856 @gccoptlist{-msoft-float -mhard-float -msmall-divides -mcpu=@var{cpu} @gol
857 -mmemcpy -mxl-soft-mul -mxl-soft-div -mxl-barrel-shift @gol
858 -mxl-pattern-compare -mxl-stack-check -mxl-gp-opt -mno-clearbss @gol
859 -mxl-multiply-high -mxl-float-convert -mxl-float-sqrt @gol
860 -mbig-endian -mlittle-endian -mxl-reorder -mxl-mode-@var{app-model}}
863 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
864 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips32r3 -mips32r5 @gol
865 -mips32r6 -mips64 -mips64r2 -mips64r3 -mips64r5 -mips64r6 @gol
866 -mips16 -mno-mips16 -mflip-mips16 @gol
867 -minterlink-compressed -mno-interlink-compressed @gol
868 -minterlink-mips16 -mno-interlink-mips16 @gol
869 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
870 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
871 -mgp32 -mgp64 -mfp32 -mfpxx -mfp64 -mhard-float -msoft-float @gol
872 -mno-float -msingle-float -mdouble-float @gol
873 -modd-spreg -mno-odd-spreg @gol
874 -mabs=@var{mode} -mnan=@var{encoding} @gol
875 -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
878 -mvirt -mno-virt @gol
880 -mmicromips -mno-micromips @gol
882 -mfpu=@var{fpu-type} @gol
883 -msmartmips -mno-smartmips @gol
884 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
885 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
886 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
887 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
888 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
889 -membedded-data -mno-embedded-data @gol
890 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
891 -mcode-readable=@var{setting} @gol
892 -msplit-addresses -mno-split-addresses @gol
893 -mexplicit-relocs -mno-explicit-relocs @gol
894 -mcheck-zero-division -mno-check-zero-division @gol
895 -mdivide-traps -mdivide-breaks @gol
896 -mload-store-pairs -mno-load-store-pairs @gol
897 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
898 -mmad -mno-mad -mimadd -mno-imadd -mfused-madd -mno-fused-madd -nocpp @gol
899 -mfix-24k -mno-fix-24k @gol
900 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
901 -mfix-r10000 -mno-fix-r10000 -mfix-rm7000 -mno-fix-rm7000 @gol
902 -mfix-vr4120 -mno-fix-vr4120 @gol
903 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
904 -mflush-func=@var{func} -mno-flush-func @gol
905 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
906 -mcompact-branches=@var{policy} @gol
907 -mfp-exceptions -mno-fp-exceptions @gol
908 -mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
909 -mlxc1-sxc1 -mno-lxc1-sxc1 -mmadd4 -mno-madd4 @gol
910 -mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address @gol
911 -mframe-header-opt -mno-frame-header-opt}
914 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
915 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
916 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
917 -mno-base-addresses -msingle-exit -mno-single-exit}
919 @emph{MN10300 Options}
920 @gccoptlist{-mmult-bug -mno-mult-bug @gol
921 -mno-am33 -mam33 -mam33-2 -mam34 @gol
922 -mtune=@var{cpu-type} @gol
923 -mreturn-pointer-on-d0 @gol
924 -mno-crt0 -mrelax -mliw -msetlb}
927 @gccoptlist{-meb -mel -mmul.x -mno-crt0}
929 @emph{MSP430 Options}
930 @gccoptlist{-msim -masm-hex -mmcu= -mcpu= -mlarge -msmall -mrelax @gol
932 -mcode-region= -mdata-region= @gol
933 -msilicon-errata= -msilicon-errata-warn= @gol
937 @gccoptlist{-mbig-endian -mlittle-endian @gol
938 -mreduced-regs -mfull-regs @gol
939 -mcmov -mno-cmov @gol
940 -mperf-ext -mno-perf-ext @gol
941 -mv3push -mno-v3push @gol
942 -m16bit -mno-16bit @gol
943 -misr-vector-size=@var{num} @gol
944 -mcache-block-size=@var{num} @gol
945 -march=@var{arch} @gol
946 -mcmodel=@var{code-model} @gol
949 @emph{Nios II Options}
950 @gccoptlist{-G @var{num} -mgpopt=@var{option} -mgpopt -mno-gpopt @gol
951 -mgprel-sec=@var{regexp} -mr0rel-sec=@var{regexp} @gol
953 -mno-bypass-cache -mbypass-cache @gol
954 -mno-cache-volatile -mcache-volatile @gol
955 -mno-fast-sw-div -mfast-sw-div @gol
956 -mhw-mul -mno-hw-mul -mhw-mulx -mno-hw-mulx -mno-hw-div -mhw-div @gol
957 -mcustom-@var{insn}=@var{N} -mno-custom-@var{insn} @gol
958 -mcustom-fpu-cfg=@var{name} @gol
959 -mhal -msmallc -msys-crt0=@var{name} -msys-lib=@var{name} @gol
960 -march=@var{arch} -mbmx -mno-bmx -mcdx -mno-cdx}
962 @emph{Nvidia PTX Options}
963 @gccoptlist{-m32 -m64 -mmainkernel -moptimize}
965 @emph{PDP-11 Options}
966 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
967 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
968 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
969 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
970 -mbranch-expensive -mbranch-cheap @gol
971 -munix-asm -mdec-asm}
973 @emph{picoChip Options}
974 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N} @gol
975 -msymbol-as-address -mno-inefficient-warnings}
977 @emph{PowerPC Options}
978 See RS/6000 and PowerPC Options.
980 @emph{RISC-V Options}
981 @gccoptlist{-mbranch-cost=@var{N-instruction} @gol
982 -mmemcpy -mno-memcpy @gol
984 -mabi=@var{ABI-string} @gol
985 -mfdiv -mno-fdiv @gol
987 -march=@var{ISA-string} @gol
988 -mtune=@var{processor-string} @gol
989 -msmall-data-limit=@var{N-bytes} @gol
990 -msave-restore -mno-save-restore @gol
991 -mstrict-align -mno-strict-align @gol
992 -mcmodel=medlow -mcmodel=medany @gol
993 -mexplicit-relocs -mno-explicit-relocs @gol}
996 @gccoptlist{-msim -mmul=none -mmul=g13 -mmul=g14 -mallregs @gol
997 -mcpu=g10 -mcpu=g13 -mcpu=g14 -mg10 -mg13 -mg14 @gol
998 -m64bit-doubles -m32bit-doubles -msave-mduc-in-interrupts}
1000 @emph{RS/6000 and PowerPC Options}
1001 @gccoptlist{-mcpu=@var{cpu-type} @gol
1002 -mtune=@var{cpu-type} @gol
1003 -mcmodel=@var{code-model} @gol
1005 -maltivec -mno-altivec @gol
1006 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
1007 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
1008 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
1009 -mfprnd -mno-fprnd @gol
1010 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
1011 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
1012 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
1013 -malign-power -malign-natural @gol
1014 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
1015 -msingle-float -mdouble-float -msimple-fpu @gol
1016 -mstring -mno-string -mupdate -mno-update @gol
1017 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
1018 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
1019 -mstrict-align -mno-strict-align -mrelocatable @gol
1020 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
1021 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
1022 -mdynamic-no-pic -maltivec -mswdiv -msingle-pic-base @gol
1023 -mprioritize-restricted-insns=@var{priority} @gol
1024 -msched-costly-dep=@var{dependence_type} @gol
1025 -minsert-sched-nops=@var{scheme} @gol
1026 -mcall-sysv -mcall-netbsd @gol
1027 -maix-struct-return -msvr4-struct-return @gol
1028 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
1029 -mblock-move-inline-limit=@var{num} @gol
1030 -misel -mno-isel @gol
1031 -misel=yes -misel=no @gol
1033 -mspe=yes -mspe=no @gol
1035 -mvrsave -mno-vrsave @gol
1036 -mmulhw -mno-mulhw @gol
1037 -mdlmzb -mno-dlmzb @gol
1038 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
1039 -mprototype -mno-prototype @gol
1040 -msim -mmvme -mads -myellowknife -memb -msdata @gol
1041 -msdata=@var{opt} -mvxworks -G @var{num} @gol
1042 -mrecip -mrecip=@var{opt} -mno-recip -mrecip-precision @gol
1043 -mno-recip-precision @gol
1044 -mveclibabi=@var{type} -mfriz -mno-friz @gol
1045 -mpointers-to-nested-functions -mno-pointers-to-nested-functions @gol
1046 -msave-toc-indirect -mno-save-toc-indirect @gol
1047 -mpower8-fusion -mno-mpower8-fusion -mpower8-vector -mno-power8-vector @gol
1048 -mcrypto -mno-crypto -mhtm -mno-htm -mdirect-move -mno-direct-move @gol
1049 -mquad-memory -mno-quad-memory @gol
1050 -mquad-memory-atomic -mno-quad-memory-atomic @gol
1051 -mcompat-align-parm -mno-compat-align-parm @gol
1052 -mfloat128 -mno-float128 -mfloat128-hardware -mno-float128-hardware @gol
1053 -mgnu-attribute -mno-gnu-attribute @gol
1054 -mstack-protector-guard=@var{guard} -mstack-protector-guard-reg=@var{reg} @gol
1055 -mstack-protector-guard-offset=@var{offset}}
1058 @gccoptlist{-m64bit-doubles -m32bit-doubles -fpu -nofpu@gol
1060 -mbig-endian-data -mlittle-endian-data @gol
1063 -mas100-syntax -mno-as100-syntax@gol
1065 -mmax-constant-size=@gol
1068 -mallow-string-insns -mno-allow-string-insns@gol
1070 -mno-warn-multiple-fast-interrupts@gol
1071 -msave-acc-in-interrupts}
1073 @emph{S/390 and zSeries Options}
1074 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
1075 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
1076 -mlong-double-64 -mlong-double-128 @gol
1077 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
1078 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
1079 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
1080 -mhtm -mvx -mzvector @gol
1081 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
1082 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard @gol
1083 -mhotpatch=@var{halfwords},@var{halfwords}}
1085 @emph{Score Options}
1086 @gccoptlist{-meb -mel @gol
1090 -mscore5 -mscore5u -mscore7 -mscore7d}
1093 @gccoptlist{-m1 -m2 -m2e @gol
1094 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
1096 -m4-nofpu -m4-single-only -m4-single -m4 @gol
1097 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
1098 -mb -ml -mdalign -mrelax @gol
1099 -mbigtable -mfmovd -mrenesas -mno-renesas -mnomacsave @gol
1100 -mieee -mno-ieee -mbitops -misize -minline-ic_invalidate -mpadstruct @gol
1101 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
1102 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
1103 -maccumulate-outgoing-args @gol
1104 -matomic-model=@var{atomic-model} @gol
1105 -mbranch-cost=@var{num} -mzdcbranch -mno-zdcbranch @gol
1106 -mcbranch-force-delay-slot @gol
1107 -mfused-madd -mno-fused-madd -mfsca -mno-fsca -mfsrra -mno-fsrra @gol
1108 -mpretend-cmove -mtas}
1110 @emph{Solaris 2 Options}
1111 @gccoptlist{-mclear-hwcap -mno-clear-hwcap -mimpure-text -mno-impure-text @gol
1114 @emph{SPARC Options}
1115 @gccoptlist{-mcpu=@var{cpu-type} @gol
1116 -mtune=@var{cpu-type} @gol
1117 -mcmodel=@var{code-model} @gol
1118 -mmemory-model=@var{mem-model} @gol
1119 -m32 -m64 -mapp-regs -mno-app-regs @gol
1120 -mfaster-structs -mno-faster-structs -mflat -mno-flat @gol
1121 -mfpu -mno-fpu -mhard-float -msoft-float @gol
1122 -mhard-quad-float -msoft-quad-float @gol
1123 -mstack-bias -mno-stack-bias @gol
1124 -mstd-struct-return -mno-std-struct-return @gol
1125 -munaligned-doubles -mno-unaligned-doubles @gol
1126 -muser-mode -mno-user-mode @gol
1127 -mv8plus -mno-v8plus -mvis -mno-vis @gol
1128 -mvis2 -mno-vis2 -mvis3 -mno-vis3 @gol
1129 -mvis4 -mno-vis4 -mvis4b -mno-vis4b @gol
1130 -mcbcond -mno-cbcond -mfmaf -mno-fmaf -mfsmuld -mno-fsmuld @gol
1131 -mpopc -mno-popc -msubxc -mno-subxc @gol
1132 -mfix-at697f -mfix-ut699 -mfix-ut700 -mfix-gr712rc @gol
1136 @gccoptlist{-mwarn-reloc -merror-reloc @gol
1137 -msafe-dma -munsafe-dma @gol
1139 -msmall-mem -mlarge-mem -mstdmain @gol
1140 -mfixed-range=@var{register-range} @gol
1142 -maddress-space-conversion -mno-address-space-conversion @gol
1143 -mcache-size=@var{cache-size} @gol
1144 -matomic-updates -mno-atomic-updates}
1146 @emph{System V Options}
1147 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
1149 @emph{TILE-Gx Options}
1150 @gccoptlist{-mcpu=CPU -m32 -m64 -mbig-endian -mlittle-endian @gol
1151 -mcmodel=@var{code-model}}
1153 @emph{TILEPro Options}
1154 @gccoptlist{-mcpu=@var{cpu} -m32}
1157 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
1158 -mprolog-function -mno-prolog-function -mspace @gol
1159 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
1160 -mapp-regs -mno-app-regs @gol
1161 -mdisable-callt -mno-disable-callt @gol
1162 -mv850e2v3 -mv850e2 -mv850e1 -mv850es @gol
1163 -mv850e -mv850 -mv850e3v5 @gol
1174 @gccoptlist{-mg -mgnu -munix}
1176 @emph{Visium Options}
1177 @gccoptlist{-mdebug -msim -mfpu -mno-fpu -mhard-float -msoft-float @gol
1178 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} -msv-mode -muser-mode}
1181 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64 @gol
1182 -mpointer-size=@var{size}}
1184 @emph{VxWorks Options}
1185 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
1186 -Xbind-lazy -Xbind-now}
1189 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
1190 -mtune-ctrl=@var{feature-list} -mdump-tune-features -mno-default @gol
1191 -mfpmath=@var{unit} @gol
1192 -masm=@var{dialect} -mno-fancy-math-387 @gol
1193 -mno-fp-ret-in-387 -m80387 -mhard-float -msoft-float @gol
1194 -mno-wide-multiply -mrtd -malign-double @gol
1195 -mpreferred-stack-boundary=@var{num} @gol
1196 -mincoming-stack-boundary=@var{num} @gol
1197 -mcld -mcx16 -msahf -mmovbe -mcrc32 @gol
1198 -mrecip -mrecip=@var{opt} @gol
1199 -mvzeroupper -mprefer-avx128 -mprefer-vector-width=@var{opt} @gol
1200 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
1201 -mavx2 -mavx512f -mavx512pf -mavx512er -mavx512cd -mavx512vl @gol
1202 -mavx512bw -mavx512dq -mavx512ifma -mavx512vbmi -msha -maes @gol
1203 -mpclmul -mfsgsbase -mrdrnd -mf16c -mfma @gol
1204 -mprefetchwt1 -mclflushopt -mxsavec -mxsaves @gol
1205 -msse4a -m3dnow -m3dnowa -mpopcnt -mabm -mbmi -mtbm -mfma4 -mxop @gol
1206 -mlzcnt -mbmi2 -mfxsr -mxsave -mxsaveopt -mrtm -mlwp -mmpx @gol
1207 -mmwaitx -mclzero -mpku -mthreads -mgfni @gol
1208 -mcet -mibt -mshstk -mforce-indirect-call -mavx512vbmi2 @gol
1209 -mms-bitfields -mno-align-stringops -minline-all-stringops @gol
1210 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
1211 -mmemcpy-strategy=@var{strategy} -mmemset-strategy=@var{strategy} @gol
1212 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
1213 -m96bit-long-double -mlong-double-64 -mlong-double-80 -mlong-double-128 @gol
1214 -mregparm=@var{num} -msseregparm @gol
1215 -mveclibabi=@var{type} -mvect8-ret-in-mem @gol
1216 -mpc32 -mpc64 -mpc80 -mstackrealign @gol
1217 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
1218 -mcmodel=@var{code-model} -mabi=@var{name} -maddress-mode=@var{mode} @gol
1219 -m32 -m64 -mx32 -m16 -miamcu -mlarge-data-threshold=@var{num} @gol
1220 -msse2avx -mfentry -mrecord-mcount -mnop-mcount -m8bit-idiv @gol
1221 -mavx256-split-unaligned-load -mavx256-split-unaligned-store @gol
1222 -malign-data=@var{type} -mstack-protector-guard=@var{guard} @gol
1223 -mstack-protector-guard-reg=@var{reg} @gol
1224 -mstack-protector-guard-offset=@var{offset} @gol
1225 -mstack-protector-guard-symbol=@var{symbol} -mmitigate-rop @gol
1226 -mgeneral-regs-only -mcall-ms2sysv-xlogues}
1228 @emph{x86 Windows Options}
1229 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll @gol
1230 -mnop-fun-dllimport -mthread @gol
1231 -municode -mwin32 -mwindows -fno-set-stack-executable}
1233 @emph{Xstormy16 Options}
1236 @emph{Xtensa Options}
1237 @gccoptlist{-mconst16 -mno-const16 @gol
1238 -mfused-madd -mno-fused-madd @gol
1240 -mserialize-volatile -mno-serialize-volatile @gol
1241 -mtext-section-literals -mno-text-section-literals @gol
1242 -mauto-litpools -mno-auto-litpools @gol
1243 -mtarget-align -mno-target-align @gol
1244 -mlongcalls -mno-longcalls}
1246 @emph{zSeries Options}
1247 See S/390 and zSeries Options.
1251 @node Overall Options
1252 @section Options Controlling the Kind of Output
1254 Compilation can involve up to four stages: preprocessing, compilation
1255 proper, assembly and linking, always in that order. GCC is capable of
1256 preprocessing and compiling several files either into several
1257 assembler input files, or into one assembler input file; then each
1258 assembler input file produces an object file, and linking combines all
1259 the object files (those newly compiled, and those specified as input)
1260 into an executable file.
1262 @cindex file name suffix
1263 For any given input file, the file name suffix determines what kind of
1264 compilation is done:
1268 C source code that must be preprocessed.
1271 C source code that should not be preprocessed.
1274 C++ source code that should not be preprocessed.
1277 Objective-C source code. Note that you must link with the @file{libobjc}
1278 library to make an Objective-C program work.
1281 Objective-C source code that should not be preprocessed.
1285 Objective-C++ source code. Note that you must link with the @file{libobjc}
1286 library to make an Objective-C++ program work. Note that @samp{.M} refers
1287 to a literal capital M@.
1289 @item @var{file}.mii
1290 Objective-C++ source code that should not be preprocessed.
1293 C, C++, Objective-C or Objective-C++ header file to be turned into a
1294 precompiled header (default), or C, C++ header file to be turned into an
1295 Ada spec (via the @option{-fdump-ada-spec} switch).
1298 @itemx @var{file}.cp
1299 @itemx @var{file}.cxx
1300 @itemx @var{file}.cpp
1301 @itemx @var{file}.CPP
1302 @itemx @var{file}.c++
1304 C++ source code that must be preprocessed. Note that in @samp{.cxx},
1305 the last two letters must both be literally @samp{x}. Likewise,
1306 @samp{.C} refers to a literal capital C@.
1310 Objective-C++ source code that must be preprocessed.
1312 @item @var{file}.mii
1313 Objective-C++ source code that should not be preprocessed.
1317 @itemx @var{file}.hp
1318 @itemx @var{file}.hxx
1319 @itemx @var{file}.hpp
1320 @itemx @var{file}.HPP
1321 @itemx @var{file}.h++
1322 @itemx @var{file}.tcc
1323 C++ header file to be turned into a precompiled header or Ada spec.
1326 @itemx @var{file}.for
1327 @itemx @var{file}.ftn
1328 Fixed form Fortran source code that should not be preprocessed.
1331 @itemx @var{file}.FOR
1332 @itemx @var{file}.fpp
1333 @itemx @var{file}.FPP
1334 @itemx @var{file}.FTN
1335 Fixed form Fortran source code that must be preprocessed (with the traditional
1338 @item @var{file}.f90
1339 @itemx @var{file}.f95
1340 @itemx @var{file}.f03
1341 @itemx @var{file}.f08
1342 Free form Fortran source code that should not be preprocessed.
1344 @item @var{file}.F90
1345 @itemx @var{file}.F95
1346 @itemx @var{file}.F03
1347 @itemx @var{file}.F08
1348 Free form Fortran source code that must be preprocessed (with the
1349 traditional preprocessor).
1354 @item @var{file}.brig
1355 BRIG files (binary representation of HSAIL).
1357 @item @var{file}.ads
1358 Ada source code file that contains a library unit declaration (a
1359 declaration of a package, subprogram, or generic, or a generic
1360 instantiation), or a library unit renaming declaration (a package,
1361 generic, or subprogram renaming declaration). Such files are also
1364 @item @var{file}.adb
1365 Ada source code file containing a library unit body (a subprogram or
1366 package body). Such files are also called @dfn{bodies}.
1368 @c GCC also knows about some suffixes for languages not yet included:
1379 @itemx @var{file}.sx
1380 Assembler code that must be preprocessed.
1383 An object file to be fed straight into linking.
1384 Any file name with no recognized suffix is treated this way.
1388 You can specify the input language explicitly with the @option{-x} option:
1391 @item -x @var{language}
1392 Specify explicitly the @var{language} for the following input files
1393 (rather than letting the compiler choose a default based on the file
1394 name suffix). This option applies to all following input files until
1395 the next @option{-x} option. Possible values for @var{language} are:
1397 c c-header cpp-output
1398 c++ c++-header c++-cpp-output
1399 objective-c objective-c-header objective-c-cpp-output
1400 objective-c++ objective-c++-header objective-c++-cpp-output
1401 assembler assembler-with-cpp
1403 f77 f77-cpp-input f95 f95-cpp-input
1409 Turn off any specification of a language, so that subsequent files are
1410 handled according to their file name suffixes (as they are if @option{-x}
1411 has not been used at all).
1414 If you only want some of the stages of compilation, you can use
1415 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1416 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1417 @command{gcc} is to stop. Note that some combinations (for example,
1418 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1423 Compile or assemble the source files, but do not link. The linking
1424 stage simply is not done. The ultimate output is in the form of an
1425 object file for each source file.
1427 By default, the object file name for a source file is made by replacing
1428 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1430 Unrecognized input files, not requiring compilation or assembly, are
1435 Stop after the stage of compilation proper; do not assemble. The output
1436 is in the form of an assembler code file for each non-assembler input
1439 By default, the assembler file name for a source file is made by
1440 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1442 Input files that don't require compilation are ignored.
1446 Stop after the preprocessing stage; do not run the compiler proper. The
1447 output is in the form of preprocessed source code, which is sent to the
1450 Input files that don't require preprocessing are ignored.
1452 @cindex output file option
1455 Place output in file @var{file}. This applies to whatever
1456 sort of output is being produced, whether it be an executable file,
1457 an object file, an assembler file or preprocessed C code.
1459 If @option{-o} is not specified, the default is to put an executable
1460 file in @file{a.out}, the object file for
1461 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1462 assembler file in @file{@var{source}.s}, a precompiled header file in
1463 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1468 Print (on standard error output) the commands executed to run the stages
1469 of compilation. Also print the version number of the compiler driver
1470 program and of the preprocessor and the compiler proper.
1474 Like @option{-v} except the commands are not executed and arguments
1475 are quoted unless they contain only alphanumeric characters or @code{./-_}.
1476 This is useful for shell scripts to capture the driver-generated command lines.
1480 Print (on the standard output) a description of the command-line options
1481 understood by @command{gcc}. If the @option{-v} option is also specified
1482 then @option{--help} is also passed on to the various processes
1483 invoked by @command{gcc}, so that they can display the command-line options
1484 they accept. If the @option{-Wextra} option has also been specified
1485 (prior to the @option{--help} option), then command-line options that
1486 have no documentation associated with them are also displayed.
1489 @opindex target-help
1490 Print (on the standard output) a description of target-specific command-line
1491 options for each tool. For some targets extra target-specific
1492 information may also be printed.
1494 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1495 Print (on the standard output) a description of the command-line
1496 options understood by the compiler that fit into all specified classes
1497 and qualifiers. These are the supported classes:
1500 @item @samp{optimizers}
1501 Display all of the optimization options supported by the
1504 @item @samp{warnings}
1505 Display all of the options controlling warning messages
1506 produced by the compiler.
1509 Display target-specific options. Unlike the
1510 @option{--target-help} option however, target-specific options of the
1511 linker and assembler are not displayed. This is because those
1512 tools do not currently support the extended @option{--help=} syntax.
1515 Display the values recognized by the @option{--param}
1518 @item @var{language}
1519 Display the options supported for @var{language}, where
1520 @var{language} is the name of one of the languages supported in this
1524 Display the options that are common to all languages.
1527 These are the supported qualifiers:
1530 @item @samp{undocumented}
1531 Display only those options that are undocumented.
1534 Display options taking an argument that appears after an equal
1535 sign in the same continuous piece of text, such as:
1536 @samp{--help=target}.
1538 @item @samp{separate}
1539 Display options taking an argument that appears as a separate word
1540 following the original option, such as: @samp{-o output-file}.
1543 Thus for example to display all the undocumented target-specific
1544 switches supported by the compiler, use:
1547 --help=target,undocumented
1550 The sense of a qualifier can be inverted by prefixing it with the
1551 @samp{^} character, so for example to display all binary warning
1552 options (i.e., ones that are either on or off and that do not take an
1553 argument) that have a description, use:
1556 --help=warnings,^joined,^undocumented
1559 The argument to @option{--help=} should not consist solely of inverted
1562 Combining several classes is possible, although this usually
1563 restricts the output so much that there is nothing to display. One
1564 case where it does work, however, is when one of the classes is
1565 @var{target}. For example, to display all the target-specific
1566 optimization options, use:
1569 --help=target,optimizers
1572 The @option{--help=} option can be repeated on the command line. Each
1573 successive use displays its requested class of options, skipping
1574 those that have already been displayed.
1576 If the @option{-Q} option appears on the command line before the
1577 @option{--help=} option, then the descriptive text displayed by
1578 @option{--help=} is changed. Instead of describing the displayed
1579 options, an indication is given as to whether the option is enabled,
1580 disabled or set to a specific value (assuming that the compiler
1581 knows this at the point where the @option{--help=} option is used).
1583 Here is a truncated example from the ARM port of @command{gcc}:
1586 % gcc -Q -mabi=2 --help=target -c
1587 The following options are target specific:
1589 -mabort-on-noreturn [disabled]
1593 The output is sensitive to the effects of previous command-line
1594 options, so for example it is possible to find out which optimizations
1595 are enabled at @option{-O2} by using:
1598 -Q -O2 --help=optimizers
1601 Alternatively you can discover which binary optimizations are enabled
1602 by @option{-O3} by using:
1605 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1606 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1607 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1612 Display the version number and copyrights of the invoked GCC@.
1614 @item -pass-exit-codes
1615 @opindex pass-exit-codes
1616 Normally the @command{gcc} program exits with the code of 1 if any
1617 phase of the compiler returns a non-success return code. If you specify
1618 @option{-pass-exit-codes}, the @command{gcc} program instead returns with
1619 the numerically highest error produced by any phase returning an error
1620 indication. The C, C++, and Fortran front ends return 4 if an internal
1621 compiler error is encountered.
1625 Use pipes rather than temporary files for communication between the
1626 various stages of compilation. This fails to work on some systems where
1627 the assembler is unable to read from a pipe; but the GNU assembler has
1630 @item -specs=@var{file}
1632 Process @var{file} after the compiler reads in the standard @file{specs}
1633 file, in order to override the defaults which the @command{gcc} driver
1634 program uses when determining what switches to pass to @command{cc1},
1635 @command{cc1plus}, @command{as}, @command{ld}, etc. More than one
1636 @option{-specs=@var{file}} can be specified on the command line, and they
1637 are processed in order, from left to right. @xref{Spec Files}, for
1638 information about the format of the @var{file}.
1642 Invoke all subcommands under a wrapper program. The name of the
1643 wrapper program and its parameters are passed as a comma separated
1647 gcc -c t.c -wrapper gdb,--args
1651 This invokes all subprograms of @command{gcc} under
1652 @samp{gdb --args}, thus the invocation of @command{cc1} is
1653 @samp{gdb --args cc1 @dots{}}.
1655 @item -fplugin=@var{name}.so
1657 Load the plugin code in file @var{name}.so, assumed to be a
1658 shared object to be dlopen'd by the compiler. The base name of
1659 the shared object file is used to identify the plugin for the
1660 purposes of argument parsing (See
1661 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1662 Each plugin should define the callback functions specified in the
1665 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1666 @opindex fplugin-arg
1667 Define an argument called @var{key} with a value of @var{value}
1668 for the plugin called @var{name}.
1670 @item -fdump-ada-spec@r{[}-slim@r{]}
1671 @opindex fdump-ada-spec
1672 For C and C++ source and include files, generate corresponding Ada specs.
1673 @xref{Generating Ada Bindings for C and C++ headers,,, gnat_ugn,
1674 GNAT User's Guide}, which provides detailed documentation on this feature.
1676 @item -fada-spec-parent=@var{unit}
1677 @opindex fada-spec-parent
1678 In conjunction with @option{-fdump-ada-spec@r{[}-slim@r{]}} above, generate
1679 Ada specs as child units of parent @var{unit}.
1681 @item -fdump-go-spec=@var{file}
1682 @opindex fdump-go-spec
1683 For input files in any language, generate corresponding Go
1684 declarations in @var{file}. This generates Go @code{const},
1685 @code{type}, @code{var}, and @code{func} declarations which may be a
1686 useful way to start writing a Go interface to code written in some
1689 @include @value{srcdir}/../libiberty/at-file.texi
1693 @section Compiling C++ Programs
1695 @cindex suffixes for C++ source
1696 @cindex C++ source file suffixes
1697 C++ source files conventionally use one of the suffixes @samp{.C},
1698 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1699 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1700 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1701 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1702 files with these names and compiles them as C++ programs even if you
1703 call the compiler the same way as for compiling C programs (usually
1704 with the name @command{gcc}).
1708 However, the use of @command{gcc} does not add the C++ library.
1709 @command{g++} is a program that calls GCC and automatically specifies linking
1710 against the C++ library. It treats @samp{.c},
1711 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1712 files unless @option{-x} is used. This program is also useful when
1713 precompiling a C header file with a @samp{.h} extension for use in C++
1714 compilations. On many systems, @command{g++} is also installed with
1715 the name @command{c++}.
1717 @cindex invoking @command{g++}
1718 When you compile C++ programs, you may specify many of the same
1719 command-line options that you use for compiling programs in any
1720 language; or command-line options meaningful for C and related
1721 languages; or options that are meaningful only for C++ programs.
1722 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1723 explanations of options for languages related to C@.
1724 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1725 explanations of options that are meaningful only for C++ programs.
1727 @node C Dialect Options
1728 @section Options Controlling C Dialect
1729 @cindex dialect options
1730 @cindex language dialect options
1731 @cindex options, dialect
1733 The following options control the dialect of C (or languages derived
1734 from C, such as C++, Objective-C and Objective-C++) that the compiler
1738 @cindex ANSI support
1742 In C mode, this is equivalent to @option{-std=c90}. In C++ mode, it is
1743 equivalent to @option{-std=c++98}.
1745 This turns off certain features of GCC that are incompatible with ISO
1746 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1747 such as the @code{asm} and @code{typeof} keywords, and
1748 predefined macros such as @code{unix} and @code{vax} that identify the
1749 type of system you are using. It also enables the undesirable and
1750 rarely used ISO trigraph feature. For the C compiler,
1751 it disables recognition of C++ style @samp{//} comments as well as
1752 the @code{inline} keyword.
1754 The alternate keywords @code{__asm__}, @code{__extension__},
1755 @code{__inline__} and @code{__typeof__} continue to work despite
1756 @option{-ansi}. You would not want to use them in an ISO C program, of
1757 course, but it is useful to put them in header files that might be included
1758 in compilations done with @option{-ansi}. Alternate predefined macros
1759 such as @code{__unix__} and @code{__vax__} are also available, with or
1760 without @option{-ansi}.
1762 The @option{-ansi} option does not cause non-ISO programs to be
1763 rejected gratuitously. For that, @option{-Wpedantic} is required in
1764 addition to @option{-ansi}. @xref{Warning Options}.
1766 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1767 option is used. Some header files may notice this macro and refrain
1768 from declaring certain functions or defining certain macros that the
1769 ISO standard doesn't call for; this is to avoid interfering with any
1770 programs that might use these names for other things.
1772 Functions that are normally built in but do not have semantics
1773 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1774 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1775 built-in functions provided by GCC}, for details of the functions
1780 Determine the language standard. @xref{Standards,,Language Standards
1781 Supported by GCC}, for details of these standard versions. This option
1782 is currently only supported when compiling C or C++.
1784 The compiler can accept several base standards, such as @samp{c90} or
1785 @samp{c++98}, and GNU dialects of those standards, such as
1786 @samp{gnu90} or @samp{gnu++98}. When a base standard is specified, the
1787 compiler accepts all programs following that standard plus those
1788 using GNU extensions that do not contradict it. For example,
1789 @option{-std=c90} turns off certain features of GCC that are
1790 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1791 keywords, but not other GNU extensions that do not have a meaning in
1792 ISO C90, such as omitting the middle term of a @code{?:}
1793 expression. On the other hand, when a GNU dialect of a standard is
1794 specified, all features supported by the compiler are enabled, even when
1795 those features change the meaning of the base standard. As a result, some
1796 strict-conforming programs may be rejected. The particular standard
1797 is used by @option{-Wpedantic} to identify which features are GNU
1798 extensions given that version of the standard. For example
1799 @option{-std=gnu90 -Wpedantic} warns about C++ style @samp{//}
1800 comments, while @option{-std=gnu99 -Wpedantic} does not.
1802 A value for this option must be provided; possible values are
1808 Support all ISO C90 programs (certain GNU extensions that conflict
1809 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1811 @item iso9899:199409
1812 ISO C90 as modified in amendment 1.
1818 ISO C99. This standard is substantially completely supported, modulo
1819 bugs and floating-point issues
1820 (mainly but not entirely relating to optional C99 features from
1821 Annexes F and G). See
1822 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1823 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1828 ISO C11, the 2011 revision of the ISO C standard. This standard is
1829 substantially completely supported, modulo bugs, floating-point issues
1830 (mainly but not entirely relating to optional C11 features from
1831 Annexes F and G) and the optional Annexes K (Bounds-checking
1832 interfaces) and L (Analyzability). The name @samp{c1x} is deprecated.
1838 ISO C17, the 2017 revision of the ISO C standard (expected to be
1839 published in 2018). This standard is
1840 same as C11 except for corrections of defects (all of which are also
1841 applied with @option{-std=c11}) and a new value of
1842 @code{__STDC_VERSION__}, and so is supported to the same extent as C11.
1846 GNU dialect of ISO C90 (including some C99 features).
1850 GNU dialect of ISO C99. The name @samp{gnu9x} is deprecated.
1854 GNU dialect of ISO C11.
1855 The name @samp{gnu1x} is deprecated.
1859 GNU dialect of ISO C17. This is the default for C code.
1863 The 1998 ISO C++ standard plus the 2003 technical corrigendum and some
1864 additional defect reports. Same as @option{-ansi} for C++ code.
1868 GNU dialect of @option{-std=c++98}.
1872 The 2011 ISO C++ standard plus amendments.
1873 The name @samp{c++0x} is deprecated.
1877 GNU dialect of @option{-std=c++11}.
1878 The name @samp{gnu++0x} is deprecated.
1882 The 2014 ISO C++ standard plus amendments.
1883 The name @samp{c++1y} is deprecated.
1887 GNU dialect of @option{-std=c++14}.
1888 This is the default for C++ code.
1889 The name @samp{gnu++1y} is deprecated.
1893 The 2017 ISO C++ standard plus amendments.
1894 The name @samp{c++1z} is deprecated.
1898 GNU dialect of @option{-std=c++17}.
1899 The name @samp{gnu++1z} is deprecated.
1902 The next revision of the ISO C++ standard, tentatively planned for
1903 2020. Support is highly experimental, and will almost certainly
1904 change in incompatible ways in future releases.
1907 GNU dialect of @option{-std=c++2a}. Support is highly experimental,
1908 and will almost certainly change in incompatible ways in future
1912 @item -fgnu89-inline
1913 @opindex fgnu89-inline
1914 The option @option{-fgnu89-inline} tells GCC to use the traditional
1915 GNU semantics for @code{inline} functions when in C99 mode.
1916 @xref{Inline,,An Inline Function is As Fast As a Macro}.
1917 Using this option is roughly equivalent to adding the
1918 @code{gnu_inline} function attribute to all inline functions
1919 (@pxref{Function Attributes}).
1921 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1922 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1923 specifies the default behavior).
1924 This option is not supported in @option{-std=c90} or
1925 @option{-std=gnu90} mode.
1927 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1928 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1929 in effect for @code{inline} functions. @xref{Common Predefined
1930 Macros,,,cpp,The C Preprocessor}.
1932 @item -fpermitted-flt-eval-methods=@var{style}
1933 @opindex fpermitted-flt-eval-methods
1934 @opindex fpermitted-flt-eval-methods=c11
1935 @opindex fpermitted-flt-eval-methods=ts-18661-3
1936 ISO/IEC TS 18661-3 defines new permissible values for
1937 @code{FLT_EVAL_METHOD} that indicate that operations and constants with
1938 a semantic type that is an interchange or extended format should be
1939 evaluated to the precision and range of that type. These new values are
1940 a superset of those permitted under C99/C11, which does not specify the
1941 meaning of other positive values of @code{FLT_EVAL_METHOD}. As such, code
1942 conforming to C11 may not have been written expecting the possibility of
1945 @option{-fpermitted-flt-eval-methods} specifies whether the compiler
1946 should allow only the values of @code{FLT_EVAL_METHOD} specified in C99/C11,
1947 or the extended set of values specified in ISO/IEC TS 18661-3.
1949 @var{style} is either @code{c11} or @code{ts-18661-3} as appropriate.
1951 The default when in a standards compliant mode (@option{-std=c11} or similar)
1952 is @option{-fpermitted-flt-eval-methods=c11}. The default when in a GNU
1953 dialect (@option{-std=gnu11} or similar) is
1954 @option{-fpermitted-flt-eval-methods=ts-18661-3}.
1956 @item -aux-info @var{filename}
1958 Output to the given filename prototyped declarations for all functions
1959 declared and/or defined in a translation unit, including those in header
1960 files. This option is silently ignored in any language other than C@.
1962 Besides declarations, the file indicates, in comments, the origin of
1963 each declaration (source file and line), whether the declaration was
1964 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1965 @samp{O} for old, respectively, in the first character after the line
1966 number and the colon), and whether it came from a declaration or a
1967 definition (@samp{C} or @samp{F}, respectively, in the following
1968 character). In the case of function definitions, a K&R-style list of
1969 arguments followed by their declarations is also provided, inside
1970 comments, after the declaration.
1972 @item -fallow-parameterless-variadic-functions
1973 @opindex fallow-parameterless-variadic-functions
1974 Accept variadic functions without named parameters.
1976 Although it is possible to define such a function, this is not very
1977 useful as it is not possible to read the arguments. This is only
1978 supported for C as this construct is allowed by C++.
1982 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1983 keyword, so that code can use these words as identifiers. You can use
1984 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1985 instead. @option{-ansi} implies @option{-fno-asm}.
1987 In C++, this switch only affects the @code{typeof} keyword, since
1988 @code{asm} and @code{inline} are standard keywords. You may want to
1989 use the @option{-fno-gnu-keywords} flag instead, which has the same
1990 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1991 switch only affects the @code{asm} and @code{typeof} keywords, since
1992 @code{inline} is a standard keyword in ISO C99.
1995 @itemx -fno-builtin-@var{function}
1996 @opindex fno-builtin
1997 @cindex built-in functions
1998 Don't recognize built-in functions that do not begin with
1999 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
2000 functions provided by GCC}, for details of the functions affected,
2001 including those which are not built-in functions when @option{-ansi} or
2002 @option{-std} options for strict ISO C conformance are used because they
2003 do not have an ISO standard meaning.
2005 GCC normally generates special code to handle certain built-in functions
2006 more efficiently; for instance, calls to @code{alloca} may become single
2007 instructions which adjust the stack directly, and calls to @code{memcpy}
2008 may become inline copy loops. The resulting code is often both smaller
2009 and faster, but since the function calls no longer appear as such, you
2010 cannot set a breakpoint on those calls, nor can you change the behavior
2011 of the functions by linking with a different library. In addition,
2012 when a function is recognized as a built-in function, GCC may use
2013 information about that function to warn about problems with calls to
2014 that function, or to generate more efficient code, even if the
2015 resulting code still contains calls to that function. For example,
2016 warnings are given with @option{-Wformat} for bad calls to
2017 @code{printf} when @code{printf} is built in and @code{strlen} is
2018 known not to modify global memory.
2020 With the @option{-fno-builtin-@var{function}} option
2021 only the built-in function @var{function} is
2022 disabled. @var{function} must not begin with @samp{__builtin_}. If a
2023 function is named that is not built-in in this version of GCC, this
2024 option is ignored. There is no corresponding
2025 @option{-fbuiltin-@var{function}} option; if you wish to enable
2026 built-in functions selectively when using @option{-fno-builtin} or
2027 @option{-ffreestanding}, you may define macros such as:
2030 #define abs(n) __builtin_abs ((n))
2031 #define strcpy(d, s) __builtin_strcpy ((d), (s))
2037 Enable parsing of function definitions marked with @code{__GIMPLE}.
2038 This is an experimental feature that allows unit testing of GIMPLE
2043 @cindex hosted environment
2045 Assert that compilation targets a hosted environment. This implies
2046 @option{-fbuiltin}. A hosted environment is one in which the
2047 entire standard library is available, and in which @code{main} has a return
2048 type of @code{int}. Examples are nearly everything except a kernel.
2049 This is equivalent to @option{-fno-freestanding}.
2051 @item -ffreestanding
2052 @opindex ffreestanding
2053 @cindex hosted environment
2055 Assert that compilation targets a freestanding environment. This
2056 implies @option{-fno-builtin}. A freestanding environment
2057 is one in which the standard library may not exist, and program startup may
2058 not necessarily be at @code{main}. The most obvious example is an OS kernel.
2059 This is equivalent to @option{-fno-hosted}.
2061 @xref{Standards,,Language Standards Supported by GCC}, for details of
2062 freestanding and hosted environments.
2066 @cindex OpenACC accelerator programming
2067 Enable handling of OpenACC directives @code{#pragma acc} in C/C++ and
2068 @code{!$acc} in Fortran. When @option{-fopenacc} is specified, the
2069 compiler generates accelerated code according to the OpenACC Application
2070 Programming Interface v2.0 @w{@uref{https://www.openacc.org}}. This option
2071 implies @option{-pthread}, and thus is only supported on targets that
2072 have support for @option{-pthread}.
2074 @item -fopenacc-dim=@var{geom}
2075 @opindex fopenacc-dim
2076 @cindex OpenACC accelerator programming
2077 Specify default compute dimensions for parallel offload regions that do
2078 not explicitly specify. The @var{geom} value is a triple of
2079 ':'-separated sizes, in order 'gang', 'worker' and, 'vector'. A size
2080 can be omitted, to use a target-specific default value.
2084 @cindex OpenMP parallel
2085 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
2086 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
2087 compiler generates parallel code according to the OpenMP Application
2088 Program Interface v4.5 @w{@uref{http://www.openmp.org/}}. This option
2089 implies @option{-pthread}, and thus is only supported on targets that
2090 have support for @option{-pthread}. @option{-fopenmp} implies
2091 @option{-fopenmp-simd}.
2094 @opindex fopenmp-simd
2097 Enable handling of OpenMP's SIMD directives with @code{#pragma omp}
2098 in C/C++ and @code{!$omp} in Fortran. Other OpenMP directives
2103 @cindex Enable Cilk Plus
2104 Enable the usage of Cilk Plus language extension features for C/C++.
2105 When the option @option{-fcilkplus} is specified, enable the usage of
2106 the Cilk Plus Language extension features for C/C++. The present
2107 implementation follows ABI version 1.2. This is an experimental
2108 feature that is only partially complete, and whose interface may
2109 change in future versions of GCC as the official specification
2110 changes. Currently, all features but @code{_Cilk_for} have been
2115 When the option @option{-fgnu-tm} is specified, the compiler
2116 generates code for the Linux variant of Intel's current Transactional
2117 Memory ABI specification document (Revision 1.1, May 6 2009). This is
2118 an experimental feature whose interface may change in future versions
2119 of GCC, as the official specification changes. Please note that not
2120 all architectures are supported for this feature.
2122 For more information on GCC's support for transactional memory,
2123 @xref{Enabling libitm,,The GNU Transactional Memory Library,libitm,GNU
2124 Transactional Memory Library}.
2126 Note that the transactional memory feature is not supported with
2127 non-call exceptions (@option{-fnon-call-exceptions}).
2129 @item -fms-extensions
2130 @opindex fms-extensions
2131 Accept some non-standard constructs used in Microsoft header files.
2133 In C++ code, this allows member names in structures to be similar
2134 to previous types declarations.
2143 Some cases of unnamed fields in structures and unions are only
2144 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
2145 fields within structs/unions}, for details.
2147 Note that this option is off for all targets but x86
2148 targets using ms-abi.
2150 @item -fplan9-extensions
2151 @opindex fplan9-extensions
2152 Accept some non-standard constructs used in Plan 9 code.
2154 This enables @option{-fms-extensions}, permits passing pointers to
2155 structures with anonymous fields to functions that expect pointers to
2156 elements of the type of the field, and permits referring to anonymous
2157 fields declared using a typedef. @xref{Unnamed Fields,,Unnamed
2158 struct/union fields within structs/unions}, for details. This is only
2159 supported for C, not C++.
2161 @item -fcond-mismatch
2162 @opindex fcond-mismatch
2163 Allow conditional expressions with mismatched types in the second and
2164 third arguments. The value of such an expression is void. This option
2165 is not supported for C++.
2167 @item -flax-vector-conversions
2168 @opindex flax-vector-conversions
2169 Allow implicit conversions between vectors with differing numbers of
2170 elements and/or incompatible element types. This option should not be
2173 @item -funsigned-char
2174 @opindex funsigned-char
2175 Let the type @code{char} be unsigned, like @code{unsigned char}.
2177 Each kind of machine has a default for what @code{char} should
2178 be. It is either like @code{unsigned char} by default or like
2179 @code{signed char} by default.
2181 Ideally, a portable program should always use @code{signed char} or
2182 @code{unsigned char} when it depends on the signedness of an object.
2183 But many programs have been written to use plain @code{char} and
2184 expect it to be signed, or expect it to be unsigned, depending on the
2185 machines they were written for. This option, and its inverse, let you
2186 make such a program work with the opposite default.
2188 The type @code{char} is always a distinct type from each of
2189 @code{signed char} or @code{unsigned char}, even though its behavior
2190 is always just like one of those two.
2193 @opindex fsigned-char
2194 Let the type @code{char} be signed, like @code{signed char}.
2196 Note that this is equivalent to @option{-fno-unsigned-char}, which is
2197 the negative form of @option{-funsigned-char}. Likewise, the option
2198 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
2200 @item -fsigned-bitfields
2201 @itemx -funsigned-bitfields
2202 @itemx -fno-signed-bitfields
2203 @itemx -fno-unsigned-bitfields
2204 @opindex fsigned-bitfields
2205 @opindex funsigned-bitfields
2206 @opindex fno-signed-bitfields
2207 @opindex fno-unsigned-bitfields
2208 These options control whether a bit-field is signed or unsigned, when the
2209 declaration does not use either @code{signed} or @code{unsigned}. By
2210 default, such a bit-field is signed, because this is consistent: the
2211 basic integer types such as @code{int} are signed types.
2213 @item -fsso-struct=@var{endianness}
2214 @opindex fsso-struct
2215 Set the default scalar storage order of structures and unions to the
2216 specified endianness. The accepted values are @samp{big-endian},
2217 @samp{little-endian} and @samp{native} for the native endianness of
2218 the target (the default). This option is not supported for C++.
2220 @strong{Warning:} the @option{-fsso-struct} switch causes GCC to generate
2221 code that is not binary compatible with code generated without it if the
2222 specified endianness is not the native endianness of the target.
2225 @node C++ Dialect Options
2226 @section Options Controlling C++ Dialect
2228 @cindex compiler options, C++
2229 @cindex C++ options, command-line
2230 @cindex options, C++
2231 This section describes the command-line options that are only meaningful
2232 for C++ programs. You can also use most of the GNU compiler options
2233 regardless of what language your program is in. For example, you
2234 might compile a file @file{firstClass.C} like this:
2237 g++ -g -fstrict-enums -O -c firstClass.C
2241 In this example, only @option{-fstrict-enums} is an option meant
2242 only for C++ programs; you can use the other options with any
2243 language supported by GCC@.
2245 Some options for compiling C programs, such as @option{-std}, are also
2246 relevant for C++ programs.
2247 @xref{C Dialect Options,,Options Controlling C Dialect}.
2249 Here is a list of options that are @emph{only} for compiling C++ programs:
2253 @item -fabi-version=@var{n}
2254 @opindex fabi-version
2255 Use version @var{n} of the C++ ABI@. The default is version 0.
2257 Version 0 refers to the version conforming most closely to
2258 the C++ ABI specification. Therefore, the ABI obtained using version 0
2259 will change in different versions of G++ as ABI bugs are fixed.
2261 Version 1 is the version of the C++ ABI that first appeared in G++ 3.2.
2263 Version 2 is the version of the C++ ABI that first appeared in G++
2264 3.4, and was the default through G++ 4.9.
2266 Version 3 corrects an error in mangling a constant address as a
2269 Version 4, which first appeared in G++ 4.5, implements a standard
2270 mangling for vector types.
2272 Version 5, which first appeared in G++ 4.6, corrects the mangling of
2273 attribute const/volatile on function pointer types, decltype of a
2274 plain decl, and use of a function parameter in the declaration of
2277 Version 6, which first appeared in G++ 4.7, corrects the promotion
2278 behavior of C++11 scoped enums and the mangling of template argument
2279 packs, const/static_cast, prefix ++ and --, and a class scope function
2280 used as a template argument.
2282 Version 7, which first appeared in G++ 4.8, that treats nullptr_t as a
2283 builtin type and corrects the mangling of lambdas in default argument
2286 Version 8, which first appeared in G++ 4.9, corrects the substitution
2287 behavior of function types with function-cv-qualifiers.
2289 Version 9, which first appeared in G++ 5.2, corrects the alignment of
2292 Version 10, which first appeared in G++ 6.1, adds mangling of
2293 attributes that affect type identity, such as ia32 calling convention
2294 attributes (e.g. @samp{stdcall}).
2296 Version 11, which first appeared in G++ 7, corrects the mangling of
2297 sizeof... expressions and operator names. For multiple entities with
2298 the same name within a function, that are declared in different scopes,
2299 the mangling now changes starting with the twelfth occurrence. It also
2300 implies @option{-fnew-inheriting-ctors}.
2302 See also @option{-Wabi}.
2304 @item -fabi-compat-version=@var{n}
2305 @opindex fabi-compat-version
2306 On targets that support strong aliases, G++
2307 works around mangling changes by creating an alias with the correct
2308 mangled name when defining a symbol with an incorrect mangled name.
2309 This switch specifies which ABI version to use for the alias.
2311 With @option{-fabi-version=0} (the default), this defaults to 8 (GCC 5
2312 compatibility). If another ABI version is explicitly selected, this
2313 defaults to 0. For compatibility with GCC versions 3.2 through 4.9,
2314 use @option{-fabi-compat-version=2}.
2316 If this option is not provided but @option{-Wabi=@var{n}} is, that
2317 version is used for compatibility aliases. If this option is provided
2318 along with @option{-Wabi} (without the version), the version from this
2319 option is used for the warning.
2321 @item -fno-access-control
2322 @opindex fno-access-control
2323 Turn off all access checking. This switch is mainly useful for working
2324 around bugs in the access control code.
2327 @opindex faligned-new
2328 Enable support for C++17 @code{new} of types that require more
2329 alignment than @code{void* ::operator new(std::size_t)} provides. A
2330 numeric argument such as @code{-faligned-new=32} can be used to
2331 specify how much alignment (in bytes) is provided by that function,
2332 but few users will need to override the default of
2333 @code{alignof(std::max_align_t)}.
2335 This flag is enabled by default for @option{-std=c++17}.
2339 Check that the pointer returned by @code{operator new} is non-null
2340 before attempting to modify the storage allocated. This check is
2341 normally unnecessary because the C++ standard specifies that
2342 @code{operator new} only returns @code{0} if it is declared
2343 @code{throw()}, in which case the compiler always checks the
2344 return value even without this option. In all other cases, when
2345 @code{operator new} has a non-empty exception specification, memory
2346 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
2347 @samp{new (nothrow)}.
2351 Enable support for the C++ Extensions for Concepts Technical
2352 Specification, ISO 19217 (2015), which allows code like
2355 template <class T> concept bool Addable = requires (T t) @{ t + t; @};
2356 template <Addable T> T add (T a, T b) @{ return a + b; @}
2359 @item -fconstexpr-depth=@var{n}
2360 @opindex fconstexpr-depth
2361 Set the maximum nested evaluation depth for C++11 constexpr functions
2362 to @var{n}. A limit is needed to detect endless recursion during
2363 constant expression evaluation. The minimum specified by the standard
2366 @item -fconstexpr-loop-limit=@var{n}
2367 @opindex fconstexpr-loop-limit
2368 Set the maximum number of iterations for a loop in C++14 constexpr functions
2369 to @var{n}. A limit is needed to detect infinite loops during
2370 constant expression evaluation. The default is 262144 (1<<18).
2372 @item -fdeduce-init-list
2373 @opindex fdeduce-init-list
2374 Enable deduction of a template type parameter as
2375 @code{std::initializer_list} from a brace-enclosed initializer list, i.e.@:
2378 template <class T> auto forward(T t) -> decltype (realfn (t))
2385 forward(@{1,2@}); // call forward<std::initializer_list<int>>
2389 This deduction was implemented as a possible extension to the
2390 originally proposed semantics for the C++11 standard, but was not part
2391 of the final standard, so it is disabled by default. This option is
2392 deprecated, and may be removed in a future version of G++.
2394 @item -ffriend-injection
2395 @opindex ffriend-injection
2396 Inject friend functions into the enclosing namespace, so that they are
2397 visible outside the scope of the class in which they are declared.
2398 Friend functions were documented to work this way in the old Annotated
2399 C++ Reference Manual.
2400 However, in ISO C++ a friend function that is not declared
2401 in an enclosing scope can only be found using argument dependent
2402 lookup. GCC defaults to the standard behavior.
2404 This option is for compatibility, and may be removed in a future
2407 @item -fno-elide-constructors
2408 @opindex fno-elide-constructors
2409 The C++ standard allows an implementation to omit creating a temporary
2410 that is only used to initialize another object of the same type.
2411 Specifying this option disables that optimization, and forces G++ to
2412 call the copy constructor in all cases. This option also causes G++
2413 to call trivial member functions which otherwise would be expanded inline.
2415 In C++17, the compiler is required to omit these temporaries, but this
2416 option still affects trivial member functions.
2418 @item -fno-enforce-eh-specs
2419 @opindex fno-enforce-eh-specs
2420 Don't generate code to check for violation of exception specifications
2421 at run time. This option violates the C++ standard, but may be useful
2422 for reducing code size in production builds, much like defining
2423 @code{NDEBUG}. This does not give user code permission to throw
2424 exceptions in violation of the exception specifications; the compiler
2425 still optimizes based on the specifications, so throwing an
2426 unexpected exception results in undefined behavior at run time.
2428 @item -fextern-tls-init
2429 @itemx -fno-extern-tls-init
2430 @opindex fextern-tls-init
2431 @opindex fno-extern-tls-init
2432 The C++11 and OpenMP standards allow @code{thread_local} and
2433 @code{threadprivate} variables to have dynamic (runtime)
2434 initialization. To support this, any use of such a variable goes
2435 through a wrapper function that performs any necessary initialization.
2436 When the use and definition of the variable are in the same
2437 translation unit, this overhead can be optimized away, but when the
2438 use is in a different translation unit there is significant overhead
2439 even if the variable doesn't actually need dynamic initialization. If
2440 the programmer can be sure that no use of the variable in a
2441 non-defining TU needs to trigger dynamic initialization (either
2442 because the variable is statically initialized, or a use of the
2443 variable in the defining TU will be executed before any uses in
2444 another TU), they can avoid this overhead with the
2445 @option{-fno-extern-tls-init} option.
2447 On targets that support symbol aliases, the default is
2448 @option{-fextern-tls-init}. On targets that do not support symbol
2449 aliases, the default is @option{-fno-extern-tls-init}.
2452 @itemx -fno-for-scope
2454 @opindex fno-for-scope
2455 If @option{-ffor-scope} is specified, the scope of variables declared in
2456 a @i{for-init-statement} is limited to the @code{for} loop itself,
2457 as specified by the C++ standard.
2458 If @option{-fno-for-scope} is specified, the scope of variables declared in
2459 a @i{for-init-statement} extends to the end of the enclosing scope,
2460 as was the case in old versions of G++, and other (traditional)
2461 implementations of C++.
2463 If neither flag is given, the default is to follow the standard,
2464 but to allow and give a warning for old-style code that would
2465 otherwise be invalid, or have different behavior.
2467 @item -fno-gnu-keywords
2468 @opindex fno-gnu-keywords
2469 Do not recognize @code{typeof} as a keyword, so that code can use this
2470 word as an identifier. You can use the keyword @code{__typeof__} instead.
2471 This option is implied by the strict ISO C++ dialects: @option{-ansi},
2472 @option{-std=c++98}, @option{-std=c++11}, etc.
2474 @item -fno-implicit-templates
2475 @opindex fno-implicit-templates
2476 Never emit code for non-inline templates that are instantiated
2477 implicitly (i.e.@: by use); only emit code for explicit instantiations.
2478 @xref{Template Instantiation}, for more information.
2480 @item -fno-implicit-inline-templates
2481 @opindex fno-implicit-inline-templates
2482 Don't emit code for implicit instantiations of inline templates, either.
2483 The default is to handle inlines differently so that compiles with and
2484 without optimization need the same set of explicit instantiations.
2486 @item -fno-implement-inlines
2487 @opindex fno-implement-inlines
2488 To save space, do not emit out-of-line copies of inline functions
2489 controlled by @code{#pragma implementation}. This causes linker
2490 errors if these functions are not inlined everywhere they are called.
2492 @item -fms-extensions
2493 @opindex fms-extensions
2494 Disable Wpedantic warnings about constructs used in MFC, such as implicit
2495 int and getting a pointer to member function via non-standard syntax.
2497 @item -fnew-inheriting-ctors
2498 @opindex fnew-inheriting-ctors
2499 Enable the P0136 adjustment to the semantics of C++11 constructor
2500 inheritance. This is part of C++17 but also considered to be a Defect
2501 Report against C++11 and C++14. This flag is enabled by default
2502 unless @option{-fabi-version=10} or lower is specified.
2504 @item -fnew-ttp-matching
2505 @opindex fnew-ttp-matching
2506 Enable the P0522 resolution to Core issue 150, template template
2507 parameters and default arguments: this allows a template with default
2508 template arguments as an argument for a template template parameter
2509 with fewer template parameters. This flag is enabled by default for
2510 @option{-std=c++17}.
2512 @item -fno-nonansi-builtins
2513 @opindex fno-nonansi-builtins
2514 Disable built-in declarations of functions that are not mandated by
2515 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
2516 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
2519 @opindex fnothrow-opt
2520 Treat a @code{throw()} exception specification as if it were a
2521 @code{noexcept} specification to reduce or eliminate the text size
2522 overhead relative to a function with no exception specification. If
2523 the function has local variables of types with non-trivial
2524 destructors, the exception specification actually makes the
2525 function smaller because the EH cleanups for those variables can be
2526 optimized away. The semantic effect is that an exception thrown out of
2527 a function with such an exception specification results in a call
2528 to @code{terminate} rather than @code{unexpected}.
2530 @item -fno-operator-names
2531 @opindex fno-operator-names
2532 Do not treat the operator name keywords @code{and}, @code{bitand},
2533 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
2534 synonyms as keywords.
2536 @item -fno-optional-diags
2537 @opindex fno-optional-diags
2538 Disable diagnostics that the standard says a compiler does not need to
2539 issue. Currently, the only such diagnostic issued by G++ is the one for
2540 a name having multiple meanings within a class.
2543 @opindex fpermissive
2544 Downgrade some diagnostics about nonconformant code from errors to
2545 warnings. Thus, using @option{-fpermissive} allows some
2546 nonconforming code to compile.
2548 @item -fno-pretty-templates
2549 @opindex fno-pretty-templates
2550 When an error message refers to a specialization of a function
2551 template, the compiler normally prints the signature of the
2552 template followed by the template arguments and any typedefs or
2553 typenames in the signature (e.g. @code{void f(T) [with T = int]}
2554 rather than @code{void f(int)}) so that it's clear which template is
2555 involved. When an error message refers to a specialization of a class
2556 template, the compiler omits any template arguments that match
2557 the default template arguments for that template. If either of these
2558 behaviors make it harder to understand the error message rather than
2559 easier, you can use @option{-fno-pretty-templates} to disable them.
2563 Enable automatic template instantiation at link time. This option also
2564 implies @option{-fno-implicit-templates}. @xref{Template
2565 Instantiation}, for more information.
2569 Disable generation of information about every class with virtual
2570 functions for use by the C++ run-time type identification features
2571 (@code{dynamic_cast} and @code{typeid}). If you don't use those parts
2572 of the language, you can save some space by using this flag. Note that
2573 exception handling uses the same information, but G++ generates it as
2574 needed. The @code{dynamic_cast} operator can still be used for casts that
2575 do not require run-time type information, i.e.@: casts to @code{void *} or to
2576 unambiguous base classes.
2578 @item -fsized-deallocation
2579 @opindex fsized-deallocation
2580 Enable the built-in global declarations
2582 void operator delete (void *, std::size_t) noexcept;
2583 void operator delete[] (void *, std::size_t) noexcept;
2585 as introduced in C++14. This is useful for user-defined replacement
2586 deallocation functions that, for example, use the size of the object
2587 to make deallocation faster. Enabled by default under
2588 @option{-std=c++14} and above. The flag @option{-Wsized-deallocation}
2589 warns about places that might want to add a definition.
2591 @item -fstrict-enums
2592 @opindex fstrict-enums
2593 Allow the compiler to optimize using the assumption that a value of
2594 enumerated type can only be one of the values of the enumeration (as
2595 defined in the C++ standard; basically, a value that can be
2596 represented in the minimum number of bits needed to represent all the
2597 enumerators). This assumption may not be valid if the program uses a
2598 cast to convert an arbitrary integer value to the enumerated type.
2600 @item -fstrong-eval-order
2601 @opindex fstrong-eval-order
2602 Evaluate member access, array subscripting, and shift expressions in
2603 left-to-right order, and evaluate assignment in right-to-left order,
2604 as adopted for C++17. Enabled by default with @option{-std=c++17}.
2605 @option{-fstrong-eval-order=some} enables just the ordering of member
2606 access and shift expressions, and is the default without
2607 @option{-std=c++17}.
2609 @item -ftemplate-backtrace-limit=@var{n}
2610 @opindex ftemplate-backtrace-limit
2611 Set the maximum number of template instantiation notes for a single
2612 warning or error to @var{n}. The default value is 10.
2614 @item -ftemplate-depth=@var{n}
2615 @opindex ftemplate-depth
2616 Set the maximum instantiation depth for template classes to @var{n}.
2617 A limit on the template instantiation depth is needed to detect
2618 endless recursions during template class instantiation. ANSI/ISO C++
2619 conforming programs must not rely on a maximum depth greater than 17
2620 (changed to 1024 in C++11). The default value is 900, as the compiler
2621 can run out of stack space before hitting 1024 in some situations.
2623 @item -fno-threadsafe-statics
2624 @opindex fno-threadsafe-statics
2625 Do not emit the extra code to use the routines specified in the C++
2626 ABI for thread-safe initialization of local statics. You can use this
2627 option to reduce code size slightly in code that doesn't need to be
2630 @item -fuse-cxa-atexit
2631 @opindex fuse-cxa-atexit
2632 Register destructors for objects with static storage duration with the
2633 @code{__cxa_atexit} function rather than the @code{atexit} function.
2634 This option is required for fully standards-compliant handling of static
2635 destructors, but only works if your C library supports
2636 @code{__cxa_atexit}.
2638 @item -fno-use-cxa-get-exception-ptr
2639 @opindex fno-use-cxa-get-exception-ptr
2640 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2641 causes @code{std::uncaught_exception} to be incorrect, but is necessary
2642 if the runtime routine is not available.
2644 @item -fvisibility-inlines-hidden
2645 @opindex fvisibility-inlines-hidden
2646 This switch declares that the user does not attempt to compare
2647 pointers to inline functions or methods where the addresses of the two functions
2648 are taken in different shared objects.
2650 The effect of this is that GCC may, effectively, mark inline methods with
2651 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2652 appear in the export table of a DSO and do not require a PLT indirection
2653 when used within the DSO@. Enabling this option can have a dramatic effect
2654 on load and link times of a DSO as it massively reduces the size of the
2655 dynamic export table when the library makes heavy use of templates.
2657 The behavior of this switch is not quite the same as marking the
2658 methods as hidden directly, because it does not affect static variables
2659 local to the function or cause the compiler to deduce that
2660 the function is defined in only one shared object.
2662 You may mark a method as having a visibility explicitly to negate the
2663 effect of the switch for that method. For example, if you do want to
2664 compare pointers to a particular inline method, you might mark it as
2665 having default visibility. Marking the enclosing class with explicit
2666 visibility has no effect.
2668 Explicitly instantiated inline methods are unaffected by this option
2669 as their linkage might otherwise cross a shared library boundary.
2670 @xref{Template Instantiation}.
2672 @item -fvisibility-ms-compat
2673 @opindex fvisibility-ms-compat
2674 This flag attempts to use visibility settings to make GCC's C++
2675 linkage model compatible with that of Microsoft Visual Studio.
2677 The flag makes these changes to GCC's linkage model:
2681 It sets the default visibility to @code{hidden}, like
2682 @option{-fvisibility=hidden}.
2685 Types, but not their members, are not hidden by default.
2688 The One Definition Rule is relaxed for types without explicit
2689 visibility specifications that are defined in more than one
2690 shared object: those declarations are permitted if they are
2691 permitted when this option is not used.
2694 In new code it is better to use @option{-fvisibility=hidden} and
2695 export those classes that are intended to be externally visible.
2696 Unfortunately it is possible for code to rely, perhaps accidentally,
2697 on the Visual Studio behavior.
2699 Among the consequences of these changes are that static data members
2700 of the same type with the same name but defined in different shared
2701 objects are different, so changing one does not change the other;
2702 and that pointers to function members defined in different shared
2703 objects may not compare equal. When this flag is given, it is a
2704 violation of the ODR to define types with the same name differently.
2708 Do not use weak symbol support, even if it is provided by the linker.
2709 By default, G++ uses weak symbols if they are available. This
2710 option exists only for testing, and should not be used by end-users;
2711 it results in inferior code and has no benefits. This option may
2712 be removed in a future release of G++.
2716 Do not search for header files in the standard directories specific to
2717 C++, but do still search the other standard directories. (This option
2718 is used when building the C++ library.)
2721 In addition, these optimization, warning, and code generation options
2722 have meanings only for C++ programs:
2725 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2728 Warn when G++ it generates code that is probably not compatible with
2729 the vendor-neutral C++ ABI@. Since G++ now defaults to updating the
2730 ABI with each major release, normally @option{-Wabi} will warn only if
2731 there is a check added later in a release series for an ABI issue
2732 discovered since the initial release. @option{-Wabi} will warn about
2733 more things if an older ABI version is selected (with
2734 @option{-fabi-version=@var{n}}).
2736 @option{-Wabi} can also be used with an explicit version number to
2737 warn about compatibility with a particular @option{-fabi-version}
2738 level, e.g. @option{-Wabi=2} to warn about changes relative to
2739 @option{-fabi-version=2}.
2741 If an explicit version number is provided and
2742 @option{-fabi-compat-version} is not specified, the version number
2743 from this option is used for compatibility aliases. If no explicit
2744 version number is provided with this option, but
2745 @option{-fabi-compat-version} is specified, that version number is
2746 used for ABI warnings.
2748 Although an effort has been made to warn about
2749 all such cases, there are probably some cases that are not warned about,
2750 even though G++ is generating incompatible code. There may also be
2751 cases where warnings are emitted even though the code that is generated
2754 You should rewrite your code to avoid these warnings if you are
2755 concerned about the fact that code generated by G++ may not be binary
2756 compatible with code generated by other compilers.
2758 Known incompatibilities in @option{-fabi-version=2} (which was the
2759 default from GCC 3.4 to 4.9) include:
2764 A template with a non-type template parameter of reference type was
2765 mangled incorrectly:
2768 template <int &> struct S @{@};
2772 This was fixed in @option{-fabi-version=3}.
2775 SIMD vector types declared using @code{__attribute ((vector_size))} were
2776 mangled in a non-standard way that does not allow for overloading of
2777 functions taking vectors of different sizes.
2779 The mangling was changed in @option{-fabi-version=4}.
2782 @code{__attribute ((const))} and @code{noreturn} were mangled as type
2783 qualifiers, and @code{decltype} of a plain declaration was folded away.
2785 These mangling issues were fixed in @option{-fabi-version=5}.
2788 Scoped enumerators passed as arguments to a variadic function are
2789 promoted like unscoped enumerators, causing @code{va_arg} to complain.
2790 On most targets this does not actually affect the parameter passing
2791 ABI, as there is no way to pass an argument smaller than @code{int}.
2793 Also, the ABI changed the mangling of template argument packs,
2794 @code{const_cast}, @code{static_cast}, prefix increment/decrement, and
2795 a class scope function used as a template argument.
2797 These issues were corrected in @option{-fabi-version=6}.
2800 Lambdas in default argument scope were mangled incorrectly, and the
2801 ABI changed the mangling of @code{nullptr_t}.
2803 These issues were corrected in @option{-fabi-version=7}.
2806 When mangling a function type with function-cv-qualifiers, the
2807 un-qualified function type was incorrectly treated as a substitution
2810 This was fixed in @option{-fabi-version=8}, the default for GCC 5.1.
2813 @code{decltype(nullptr)} incorrectly had an alignment of 1, leading to
2814 unaligned accesses. Note that this did not affect the ABI of a
2815 function with a @code{nullptr_t} parameter, as parameters have a
2818 This was fixed in @option{-fabi-version=9}, the default for GCC 5.2.
2821 Target-specific attributes that affect the identity of a type, such as
2822 ia32 calling conventions on a function type (stdcall, regparm, etc.),
2823 did not affect the mangled name, leading to name collisions when
2824 function pointers were used as template arguments.
2826 This was fixed in @option{-fabi-version=10}, the default for GCC 6.1.
2830 It also warns about psABI-related changes. The known psABI changes at this
2836 For SysV/x86-64, unions with @code{long double} members are
2837 passed in memory as specified in psABI. For example:
2847 @code{union U} is always passed in memory.
2851 @item -Wabi-tag @r{(C++ and Objective-C++ only)}
2854 Warn when a type with an ABI tag is used in a context that does not
2855 have that ABI tag. See @ref{C++ Attributes} for more information
2858 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2859 @opindex Wctor-dtor-privacy
2860 @opindex Wno-ctor-dtor-privacy
2861 Warn when a class seems unusable because all the constructors or
2862 destructors in that class are private, and it has neither friends nor
2863 public static member functions. Also warn if there are no non-private
2864 methods, and there's at least one private member function that isn't
2865 a constructor or destructor.
2867 @item -Wdelete-non-virtual-dtor @r{(C++ and Objective-C++ only)}
2868 @opindex Wdelete-non-virtual-dtor
2869 @opindex Wno-delete-non-virtual-dtor
2870 Warn when @code{delete} is used to destroy an instance of a class that
2871 has virtual functions and non-virtual destructor. It is unsafe to delete
2872 an instance of a derived class through a pointer to a base class if the
2873 base class does not have a virtual destructor. This warning is enabled
2876 @item -Wliteral-suffix @r{(C++ and Objective-C++ only)}
2877 @opindex Wliteral-suffix
2878 @opindex Wno-literal-suffix
2879 Warn when a string or character literal is followed by a ud-suffix which does
2880 not begin with an underscore. As a conforming extension, GCC treats such
2881 suffixes as separate preprocessing tokens in order to maintain backwards
2882 compatibility with code that uses formatting macros from @code{<inttypes.h>}.
2886 #define __STDC_FORMAT_MACROS
2887 #include <inttypes.h>
2892 printf("My int64: %" PRId64"\n", i64);
2896 In this case, @code{PRId64} is treated as a separate preprocessing token.
2898 Additionally, warn when a user-defined literal operator is declared with
2899 a literal suffix identifier that doesn't begin with an underscore. Literal
2900 suffix identifiers that don't begin with an underscore are reserved for
2901 future standardization.
2903 This warning is enabled by default.
2905 @item -Wlto-type-mismatch
2906 @opindex Wlto-type-mismatch
2907 @opindex Wno-lto-type-mismatch
2909 During the link-time optimization warn about type mismatches in
2910 global declarations from different compilation units.
2911 Requires @option{-flto} to be enabled. Enabled by default.
2913 @item -Wno-narrowing @r{(C++ and Objective-C++ only)}
2915 @opindex Wno-narrowing
2916 For C++11 and later standards, narrowing conversions are diagnosed by default,
2917 as required by the standard. A narrowing conversion from a constant produces
2918 an error, and a narrowing conversion from a non-constant produces a warning,
2919 but @option{-Wno-narrowing} suppresses the diagnostic.
2920 Note that this does not affect the meaning of well-formed code;
2921 narrowing conversions are still considered ill-formed in SFINAE contexts.
2923 With @option{-Wnarrowing} in C++98, warn when a narrowing
2924 conversion prohibited by C++11 occurs within
2928 int i = @{ 2.2 @}; // error: narrowing from double to int
2931 This flag is included in @option{-Wall} and @option{-Wc++11-compat}.
2933 @item -Wnoexcept @r{(C++ and Objective-C++ only)}
2935 @opindex Wno-noexcept
2936 Warn when a noexcept-expression evaluates to false because of a call
2937 to a function that does not have a non-throwing exception
2938 specification (i.e. @code{throw()} or @code{noexcept}) but is known by
2939 the compiler to never throw an exception.
2941 @item -Wnoexcept-type @r{(C++ and Objective-C++ only)}
2942 @opindex Wnoexcept-type
2943 @opindex Wno-noexcept-type
2944 Warn if the C++17 feature making @code{noexcept} part of a function
2945 type changes the mangled name of a symbol relative to C++14. Enabled
2946 by @option{-Wabi} and @option{-Wc++17-compat}.
2949 template <class T> void f(T t) @{ t(); @};
2951 void h() @{ f(g); @} // in C++14 calls f<void(*)()>, in C++17 calls f<void(*)()noexcept>
2954 @item -Wclass-memaccess @r{(C++ and Objective-C++ only)}
2955 @opindex Wclass-memaccess
2956 Warn when the destination of a call to a raw memory function such as
2957 @code{memset} or @code{memcpy} is an object of class type writing into which
2958 might bypass the class non-trivial or deleted constructor or copy assignment,
2959 violate const-correctness or encapsulation, or corrupt the virtual table.
2960 Modifying the representation of such objects may violate invariants maintained
2961 by member functions of the class. For example, the call to @code{memset}
2962 below is undefined becase it modifies a non-trivial class object and is,
2963 therefore, diagnosed. The safe way to either initialize or clear the storage
2964 of objects of such types is by using the appropriate constructor or assignment
2965 operator, if one is available.
2967 std::string str = "abc";
2968 memset (&str, 0, 3);
2970 The @option{-Wclass-memaccess} option is enabled by @option{-Wall}.
2972 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2973 @opindex Wnon-virtual-dtor
2974 @opindex Wno-non-virtual-dtor
2975 Warn when a class has virtual functions and an accessible non-virtual
2976 destructor itself or in an accessible polymorphic base class, in which
2977 case it is possible but unsafe to delete an instance of a derived
2978 class through a pointer to the class itself or base class. This
2979 warning is automatically enabled if @option{-Weffc++} is specified.
2981 @item -Wregister @r{(C++ and Objective-C++ only)}
2983 @opindex Wno-register
2984 Warn on uses of the @code{register} storage class specifier, except
2985 when it is part of the GNU @ref{Explicit Register Variables} extension.
2986 The use of the @code{register} keyword as storage class specifier has
2987 been deprecated in C++11 and removed in C++17.
2988 Enabled by default with @option{-std=c++17}.
2990 @item -Wreorder @r{(C++ and Objective-C++ only)}
2992 @opindex Wno-reorder
2993 @cindex reordering, warning
2994 @cindex warning for reordering of member initializers
2995 Warn when the order of member initializers given in the code does not
2996 match the order in which they must be executed. For instance:
3002 A(): j (0), i (1) @{ @}
3007 The compiler rearranges the member initializers for @code{i}
3008 and @code{j} to match the declaration order of the members, emitting
3009 a warning to that effect. This warning is enabled by @option{-Wall}.
3011 @item -fext-numeric-literals @r{(C++ and Objective-C++ only)}
3012 @opindex fext-numeric-literals
3013 @opindex fno-ext-numeric-literals
3014 Accept imaginary, fixed-point, or machine-defined
3015 literal number suffixes as GNU extensions.
3016 When this option is turned off these suffixes are treated
3017 as C++11 user-defined literal numeric suffixes.
3018 This is on by default for all pre-C++11 dialects and all GNU dialects:
3019 @option{-std=c++98}, @option{-std=gnu++98}, @option{-std=gnu++11},
3020 @option{-std=gnu++14}.
3021 This option is off by default
3022 for ISO C++11 onwards (@option{-std=c++11}, ...).
3025 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
3028 @item -Weffc++ @r{(C++ and Objective-C++ only)}
3031 Warn about violations of the following style guidelines from Scott Meyers'
3032 @cite{Effective C++} series of books:
3036 Define a copy constructor and an assignment operator for classes
3037 with dynamically-allocated memory.
3040 Prefer initialization to assignment in constructors.
3043 Have @code{operator=} return a reference to @code{*this}.
3046 Don't try to return a reference when you must return an object.
3049 Distinguish between prefix and postfix forms of increment and
3050 decrement operators.
3053 Never overload @code{&&}, @code{||}, or @code{,}.
3057 This option also enables @option{-Wnon-virtual-dtor}, which is also
3058 one of the effective C++ recommendations. However, the check is
3059 extended to warn about the lack of virtual destructor in accessible
3060 non-polymorphic bases classes too.
3062 When selecting this option, be aware that the standard library
3063 headers do not obey all of these guidelines; use @samp{grep -v}
3064 to filter out those warnings.
3066 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
3067 @opindex Wstrict-null-sentinel
3068 @opindex Wno-strict-null-sentinel
3069 Warn about the use of an uncasted @code{NULL} as sentinel. When
3070 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
3071 to @code{__null}. Although it is a null pointer constant rather than a
3072 null pointer, it is guaranteed to be of the same size as a pointer.
3073 But this use is not portable across different compilers.
3075 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
3076 @opindex Wno-non-template-friend
3077 @opindex Wnon-template-friend
3078 Disable warnings when non-template friend functions are declared
3079 within a template. In very old versions of GCC that predate implementation
3080 of the ISO standard, declarations such as
3081 @samp{friend int foo(int)}, where the name of the friend is an unqualified-id,
3082 could be interpreted as a particular specialization of a template
3083 function; the warning exists to diagnose compatibility problems,
3084 and is enabled by default.
3086 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
3087 @opindex Wold-style-cast
3088 @opindex Wno-old-style-cast
3089 Warn if an old-style (C-style) cast to a non-void type is used within
3090 a C++ program. The new-style casts (@code{dynamic_cast},
3091 @code{static_cast}, @code{reinterpret_cast}, and @code{const_cast}) are
3092 less vulnerable to unintended effects and much easier to search for.
3094 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
3095 @opindex Woverloaded-virtual
3096 @opindex Wno-overloaded-virtual
3097 @cindex overloaded virtual function, warning
3098 @cindex warning for overloaded virtual function
3099 Warn when a function declaration hides virtual functions from a
3100 base class. For example, in:
3107 struct B: public A @{
3112 the @code{A} class version of @code{f} is hidden in @code{B}, and code
3123 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
3124 @opindex Wno-pmf-conversions
3125 @opindex Wpmf-conversions
3126 Disable the diagnostic for converting a bound pointer to member function
3129 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
3130 @opindex Wsign-promo
3131 @opindex Wno-sign-promo
3132 Warn when overload resolution chooses a promotion from unsigned or
3133 enumerated type to a signed type, over a conversion to an unsigned type of
3134 the same size. Previous versions of G++ tried to preserve
3135 unsignedness, but the standard mandates the current behavior.
3137 @item -Wtemplates @r{(C++ and Objective-C++ only)}
3139 Warn when a primary template declaration is encountered. Some coding
3140 rules disallow templates, and this may be used to enforce that rule.
3141 The warning is inactive inside a system header file, such as the STL, so
3142 one can still use the STL. One may also instantiate or specialize
3145 @item -Wmultiple-inheritance @r{(C++ and Objective-C++ only)}
3146 @opindex Wmultiple-inheritance
3147 Warn when a class is defined with multiple direct base classes. Some
3148 coding rules disallow multiple inheritance, and this may be used to
3149 enforce that rule. The warning is inactive inside a system header file,
3150 such as the STL, so one can still use the STL. One may also define
3151 classes that indirectly use multiple inheritance.
3153 @item -Wvirtual-inheritance
3154 @opindex Wvirtual-inheritance
3155 Warn when a class is defined with a virtual direct base class. Some
3156 coding rules disallow multiple inheritance, and this may be used to
3157 enforce that rule. The warning is inactive inside a system header file,
3158 such as the STL, so one can still use the STL. One may also define
3159 classes that indirectly use virtual inheritance.
3162 @opindex Wnamespaces
3163 Warn when a namespace definition is opened. Some coding rules disallow
3164 namespaces, and this may be used to enforce that rule. The warning is
3165 inactive inside a system header file, such as the STL, so one can still
3166 use the STL. One may also use using directives and qualified names.
3168 @item -Wno-terminate @r{(C++ and Objective-C++ only)}
3170 @opindex Wno-terminate
3171 Disable the warning about a throw-expression that will immediately
3172 result in a call to @code{terminate}.
3175 @node Objective-C and Objective-C++ Dialect Options
3176 @section Options Controlling Objective-C and Objective-C++ Dialects
3178 @cindex compiler options, Objective-C and Objective-C++
3179 @cindex Objective-C and Objective-C++ options, command-line
3180 @cindex options, Objective-C and Objective-C++
3181 (NOTE: This manual does not describe the Objective-C and Objective-C++
3182 languages themselves. @xref{Standards,,Language Standards
3183 Supported by GCC}, for references.)
3185 This section describes the command-line options that are only meaningful
3186 for Objective-C and Objective-C++ programs. You can also use most of
3187 the language-independent GNU compiler options.
3188 For example, you might compile a file @file{some_class.m} like this:
3191 gcc -g -fgnu-runtime -O -c some_class.m
3195 In this example, @option{-fgnu-runtime} is an option meant only for
3196 Objective-C and Objective-C++ programs; you can use the other options with
3197 any language supported by GCC@.
3199 Note that since Objective-C is an extension of the C language, Objective-C
3200 compilations may also use options specific to the C front-end (e.g.,
3201 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
3202 C++-specific options (e.g., @option{-Wabi}).
3204 Here is a list of options that are @emph{only} for compiling Objective-C
3205 and Objective-C++ programs:
3208 @item -fconstant-string-class=@var{class-name}
3209 @opindex fconstant-string-class
3210 Use @var{class-name} as the name of the class to instantiate for each
3211 literal string specified with the syntax @code{@@"@dots{}"}. The default
3212 class name is @code{NXConstantString} if the GNU runtime is being used, and
3213 @code{NSConstantString} if the NeXT runtime is being used (see below). The
3214 @option{-fconstant-cfstrings} option, if also present, overrides the
3215 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
3216 to be laid out as constant CoreFoundation strings.
3219 @opindex fgnu-runtime
3220 Generate object code compatible with the standard GNU Objective-C
3221 runtime. This is the default for most types of systems.
3223 @item -fnext-runtime
3224 @opindex fnext-runtime
3225 Generate output compatible with the NeXT runtime. This is the default
3226 for NeXT-based systems, including Darwin and Mac OS X@. The macro
3227 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
3230 @item -fno-nil-receivers
3231 @opindex fno-nil-receivers
3232 Assume that all Objective-C message dispatches (@code{[receiver
3233 message:arg]}) in this translation unit ensure that the receiver is
3234 not @code{nil}. This allows for more efficient entry points in the
3235 runtime to be used. This option is only available in conjunction with
3236 the NeXT runtime and ABI version 0 or 1.
3238 @item -fobjc-abi-version=@var{n}
3239 @opindex fobjc-abi-version
3240 Use version @var{n} of the Objective-C ABI for the selected runtime.
3241 This option is currently supported only for the NeXT runtime. In that
3242 case, Version 0 is the traditional (32-bit) ABI without support for
3243 properties and other Objective-C 2.0 additions. Version 1 is the
3244 traditional (32-bit) ABI with support for properties and other
3245 Objective-C 2.0 additions. Version 2 is the modern (64-bit) ABI. If
3246 nothing is specified, the default is Version 0 on 32-bit target
3247 machines, and Version 2 on 64-bit target machines.
3249 @item -fobjc-call-cxx-cdtors
3250 @opindex fobjc-call-cxx-cdtors
3251 For each Objective-C class, check if any of its instance variables is a
3252 C++ object with a non-trivial default constructor. If so, synthesize a
3253 special @code{- (id) .cxx_construct} instance method which runs
3254 non-trivial default constructors on any such instance variables, in order,
3255 and then return @code{self}. Similarly, check if any instance variable
3256 is a C++ object with a non-trivial destructor, and if so, synthesize a
3257 special @code{- (void) .cxx_destruct} method which runs
3258 all such default destructors, in reverse order.
3260 The @code{- (id) .cxx_construct} and @code{- (void) .cxx_destruct}
3261 methods thusly generated only operate on instance variables
3262 declared in the current Objective-C class, and not those inherited
3263 from superclasses. It is the responsibility of the Objective-C
3264 runtime to invoke all such methods in an object's inheritance
3265 hierarchy. The @code{- (id) .cxx_construct} methods are invoked
3266 by the runtime immediately after a new object instance is allocated;
3267 the @code{- (void) .cxx_destruct} methods are invoked immediately
3268 before the runtime deallocates an object instance.
3270 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
3271 support for invoking the @code{- (id) .cxx_construct} and
3272 @code{- (void) .cxx_destruct} methods.
3274 @item -fobjc-direct-dispatch
3275 @opindex fobjc-direct-dispatch
3276 Allow fast jumps to the message dispatcher. On Darwin this is
3277 accomplished via the comm page.
3279 @item -fobjc-exceptions
3280 @opindex fobjc-exceptions
3281 Enable syntactic support for structured exception handling in
3282 Objective-C, similar to what is offered by C++. This option
3283 is required to use the Objective-C keywords @code{@@try},
3284 @code{@@throw}, @code{@@catch}, @code{@@finally} and
3285 @code{@@synchronized}. This option is available with both the GNU
3286 runtime and the NeXT runtime (but not available in conjunction with
3287 the NeXT runtime on Mac OS X 10.2 and earlier).
3291 Enable garbage collection (GC) in Objective-C and Objective-C++
3292 programs. This option is only available with the NeXT runtime; the
3293 GNU runtime has a different garbage collection implementation that
3294 does not require special compiler flags.
3296 @item -fobjc-nilcheck
3297 @opindex fobjc-nilcheck
3298 For the NeXT runtime with version 2 of the ABI, check for a nil
3299 receiver in method invocations before doing the actual method call.
3300 This is the default and can be disabled using
3301 @option{-fno-objc-nilcheck}. Class methods and super calls are never
3302 checked for nil in this way no matter what this flag is set to.
3303 Currently this flag does nothing when the GNU runtime, or an older
3304 version of the NeXT runtime ABI, is used.
3306 @item -fobjc-std=objc1
3308 Conform to the language syntax of Objective-C 1.0, the language
3309 recognized by GCC 4.0. This only affects the Objective-C additions to
3310 the C/C++ language; it does not affect conformance to C/C++ standards,
3311 which is controlled by the separate C/C++ dialect option flags. When
3312 this option is used with the Objective-C or Objective-C++ compiler,
3313 any Objective-C syntax that is not recognized by GCC 4.0 is rejected.
3314 This is useful if you need to make sure that your Objective-C code can
3315 be compiled with older versions of GCC@.
3317 @item -freplace-objc-classes
3318 @opindex freplace-objc-classes
3319 Emit a special marker instructing @command{ld(1)} not to statically link in
3320 the resulting object file, and allow @command{dyld(1)} to load it in at
3321 run time instead. This is used in conjunction with the Fix-and-Continue
3322 debugging mode, where the object file in question may be recompiled and
3323 dynamically reloaded in the course of program execution, without the need
3324 to restart the program itself. Currently, Fix-and-Continue functionality
3325 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
3330 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
3331 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
3332 compile time) with static class references that get initialized at load time,
3333 which improves run-time performance. Specifying the @option{-fzero-link} flag
3334 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
3335 to be retained. This is useful in Zero-Link debugging mode, since it allows
3336 for individual class implementations to be modified during program execution.
3337 The GNU runtime currently always retains calls to @code{objc_get_class("@dots{}")}
3338 regardless of command-line options.
3340 @item -fno-local-ivars
3341 @opindex fno-local-ivars
3342 @opindex flocal-ivars
3343 By default instance variables in Objective-C can be accessed as if
3344 they were local variables from within the methods of the class they're
3345 declared in. This can lead to shadowing between instance variables
3346 and other variables declared either locally inside a class method or
3347 globally with the same name. Specifying the @option{-fno-local-ivars}
3348 flag disables this behavior thus avoiding variable shadowing issues.
3350 @item -fivar-visibility=@r{[}public@r{|}protected@r{|}private@r{|}package@r{]}
3351 @opindex fivar-visibility
3352 Set the default instance variable visibility to the specified option
3353 so that instance variables declared outside the scope of any access
3354 modifier directives default to the specified visibility.
3358 Dump interface declarations for all classes seen in the source file to a
3359 file named @file{@var{sourcename}.decl}.
3361 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
3362 @opindex Wassign-intercept
3363 @opindex Wno-assign-intercept
3364 Warn whenever an Objective-C assignment is being intercepted by the
3367 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
3368 @opindex Wno-protocol
3370 If a class is declared to implement a protocol, a warning is issued for
3371 every method in the protocol that is not implemented by the class. The
3372 default behavior is to issue a warning for every method not explicitly
3373 implemented in the class, even if a method implementation is inherited
3374 from the superclass. If you use the @option{-Wno-protocol} option, then
3375 methods inherited from the superclass are considered to be implemented,
3376 and no warning is issued for them.
3378 @item -Wselector @r{(Objective-C and Objective-C++ only)}
3380 @opindex Wno-selector
3381 Warn if multiple methods of different types for the same selector are
3382 found during compilation. The check is performed on the list of methods
3383 in the final stage of compilation. Additionally, a check is performed
3384 for each selector appearing in a @code{@@selector(@dots{})}
3385 expression, and a corresponding method for that selector has been found
3386 during compilation. Because these checks scan the method table only at
3387 the end of compilation, these warnings are not produced if the final
3388 stage of compilation is not reached, for example because an error is
3389 found during compilation, or because the @option{-fsyntax-only} option is
3392 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
3393 @opindex Wstrict-selector-match
3394 @opindex Wno-strict-selector-match
3395 Warn if multiple methods with differing argument and/or return types are
3396 found for a given selector when attempting to send a message using this
3397 selector to a receiver of type @code{id} or @code{Class}. When this flag
3398 is off (which is the default behavior), the compiler omits such warnings
3399 if any differences found are confined to types that share the same size
3402 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
3403 @opindex Wundeclared-selector
3404 @opindex Wno-undeclared-selector
3405 Warn if a @code{@@selector(@dots{})} expression referring to an
3406 undeclared selector is found. A selector is considered undeclared if no
3407 method with that name has been declared before the
3408 @code{@@selector(@dots{})} expression, either explicitly in an
3409 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
3410 an @code{@@implementation} section. This option always performs its
3411 checks as soon as a @code{@@selector(@dots{})} expression is found,
3412 while @option{-Wselector} only performs its checks in the final stage of
3413 compilation. This also enforces the coding style convention
3414 that methods and selectors must be declared before being used.
3416 @item -print-objc-runtime-info
3417 @opindex print-objc-runtime-info
3418 Generate C header describing the largest structure that is passed by
3423 @node Diagnostic Message Formatting Options
3424 @section Options to Control Diagnostic Messages Formatting
3425 @cindex options to control diagnostics formatting
3426 @cindex diagnostic messages
3427 @cindex message formatting
3429 Traditionally, diagnostic messages have been formatted irrespective of
3430 the output device's aspect (e.g.@: its width, @dots{}). You can use the
3431 options described below
3432 to control the formatting algorithm for diagnostic messages,
3433 e.g.@: how many characters per line, how often source location
3434 information should be reported. Note that some language front ends may not
3435 honor these options.
3438 @item -fmessage-length=@var{n}
3439 @opindex fmessage-length
3440 Try to format error messages so that they fit on lines of about
3441 @var{n} characters. If @var{n} is zero, then no line-wrapping is
3442 done; each error message appears on a single line. This is the
3443 default for all front ends.
3445 @item -fdiagnostics-show-location=once
3446 @opindex fdiagnostics-show-location
3447 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
3448 reporter to emit source location information @emph{once}; that is, in
3449 case the message is too long to fit on a single physical line and has to
3450 be wrapped, the source location won't be emitted (as prefix) again,
3451 over and over, in subsequent continuation lines. This is the default
3454 @item -fdiagnostics-show-location=every-line
3455 Only meaningful in line-wrapping mode. Instructs the diagnostic
3456 messages reporter to emit the same source location information (as
3457 prefix) for physical lines that result from the process of breaking
3458 a message which is too long to fit on a single line.
3460 @item -fdiagnostics-color[=@var{WHEN}]
3461 @itemx -fno-diagnostics-color
3462 @opindex fdiagnostics-color
3463 @cindex highlight, color
3464 @vindex GCC_COLORS @r{environment variable}
3465 Use color in diagnostics. @var{WHEN} is @samp{never}, @samp{always},
3466 or @samp{auto}. The default depends on how the compiler has been configured,
3467 it can be any of the above @var{WHEN} options or also @samp{never}
3468 if @env{GCC_COLORS} environment variable isn't present in the environment,
3469 and @samp{auto} otherwise.
3470 @samp{auto} means to use color only when the standard error is a terminal.
3471 The forms @option{-fdiagnostics-color} and @option{-fno-diagnostics-color} are
3472 aliases for @option{-fdiagnostics-color=always} and
3473 @option{-fdiagnostics-color=never}, respectively.
3475 The colors are defined by the environment variable @env{GCC_COLORS}.
3476 Its value is a colon-separated list of capabilities and Select Graphic
3477 Rendition (SGR) substrings. SGR commands are interpreted by the
3478 terminal or terminal emulator. (See the section in the documentation
3479 of your text terminal for permitted values and their meanings as
3480 character attributes.) These substring values are integers in decimal
3481 representation and can be concatenated with semicolons.
3482 Common values to concatenate include
3484 @samp{4} for underline,
3486 @samp{7} for inverse,
3487 @samp{39} for default foreground color,
3488 @samp{30} to @samp{37} for foreground colors,
3489 @samp{90} to @samp{97} for 16-color mode foreground colors,
3490 @samp{38;5;0} to @samp{38;5;255}
3491 for 88-color and 256-color modes foreground colors,
3492 @samp{49} for default background color,
3493 @samp{40} to @samp{47} for background colors,
3494 @samp{100} to @samp{107} for 16-color mode background colors,
3495 and @samp{48;5;0} to @samp{48;5;255}
3496 for 88-color and 256-color modes background colors.
3498 The default @env{GCC_COLORS} is
3500 error=01;31:warning=01;35:note=01;36:range1=32:range2=34:locus=01:\
3501 quote=01:fixit-insert=32:fixit-delete=31:\
3502 diff-filename=01:diff-hunk=32:diff-delete=31:diff-insert=32:\
3506 where @samp{01;31} is bold red, @samp{01;35} is bold magenta,
3507 @samp{01;36} is bold cyan, @samp{32} is green, @samp{34} is blue,
3508 @samp{01} is bold, and @samp{31} is red.
3509 Setting @env{GCC_COLORS} to the empty string disables colors.
3510 Supported capabilities are as follows.
3514 @vindex error GCC_COLORS @r{capability}
3515 SGR substring for error: markers.
3518 @vindex warning GCC_COLORS @r{capability}
3519 SGR substring for warning: markers.
3522 @vindex note GCC_COLORS @r{capability}
3523 SGR substring for note: markers.
3526 @vindex range1 GCC_COLORS @r{capability}
3527 SGR substring for first additional range.
3530 @vindex range2 GCC_COLORS @r{capability}
3531 SGR substring for second additional range.
3534 @vindex locus GCC_COLORS @r{capability}
3535 SGR substring for location information, @samp{file:line} or
3536 @samp{file:line:column} etc.
3539 @vindex quote GCC_COLORS @r{capability}
3540 SGR substring for information printed within quotes.
3543 @vindex fixit-insert GCC_COLORS @r{capability}
3544 SGR substring for fix-it hints suggesting text to
3545 be inserted or replaced.
3548 @vindex fixit-delete GCC_COLORS @r{capability}
3549 SGR substring for fix-it hints suggesting text to
3552 @item diff-filename=
3553 @vindex diff-filename GCC_COLORS @r{capability}
3554 SGR substring for filename headers within generated patches.
3557 @vindex diff-hunk GCC_COLORS @r{capability}
3558 SGR substring for the starts of hunks within generated patches.
3561 @vindex diff-delete GCC_COLORS @r{capability}
3562 SGR substring for deleted lines within generated patches.
3565 @vindex diff-insert GCC_COLORS @r{capability}
3566 SGR substring for inserted lines within generated patches.
3569 @vindex type-diff GCC_COLORS @r{capability}
3570 SGR substring for highlighting mismatching types within template
3571 arguments in the C++ frontend.
3574 @item -fno-diagnostics-show-option
3575 @opindex fno-diagnostics-show-option
3576 @opindex fdiagnostics-show-option
3577 By default, each diagnostic emitted includes text indicating the
3578 command-line option that directly controls the diagnostic (if such an
3579 option is known to the diagnostic machinery). Specifying the
3580 @option{-fno-diagnostics-show-option} flag suppresses that behavior.
3582 @item -fno-diagnostics-show-caret
3583 @opindex fno-diagnostics-show-caret
3584 @opindex fdiagnostics-show-caret
3585 By default, each diagnostic emitted includes the original source line
3586 and a caret @samp{^} indicating the column. This option suppresses this
3587 information. The source line is truncated to @var{n} characters, if
3588 the @option{-fmessage-length=n} option is given. When the output is done
3589 to the terminal, the width is limited to the width given by the
3590 @env{COLUMNS} environment variable or, if not set, to the terminal width.
3592 @item -fdiagnostics-parseable-fixits
3593 @opindex fdiagnostics-parseable-fixits
3594 Emit fix-it hints in a machine-parseable format, suitable for consumption
3595 by IDEs. For each fix-it, a line will be printed after the relevant
3596 diagnostic, starting with the string ``fix-it:''. For example:
3599 fix-it:"test.c":@{45:3-45:21@}:"gtk_widget_show_all"
3602 The location is expressed as a half-open range, expressed as a count of
3603 bytes, starting at byte 1 for the initial column. In the above example,
3604 bytes 3 through 20 of line 45 of ``test.c'' are to be replaced with the
3608 00000000011111111112222222222
3609 12345678901234567890123456789
3610 gtk_widget_showall (dlg);
3615 The filename and replacement string escape backslash as ``\\", tab as ``\t'',
3616 newline as ``\n'', double quotes as ``\"'', non-printable characters as octal
3617 (e.g. vertical tab as ``\013'').
3619 An empty replacement string indicates that the given range is to be removed.
3620 An empty range (e.g. ``45:3-45:3'') indicates that the string is to
3621 be inserted at the given position.
3623 @item -fdiagnostics-generate-patch
3624 @opindex fdiagnostics-generate-patch
3625 Print fix-it hints to stderr in unified diff format, after any diagnostics
3626 are printed. For example:
3633 void show_cb(GtkDialog *dlg)
3635 - gtk_widget_showall(dlg);
3636 + gtk_widget_show_all(dlg);
3641 The diff may or may not be colorized, following the same rules
3642 as for diagnostics (see @option{-fdiagnostics-color}).
3644 @item -fdiagnostics-show-template-tree
3645 @opindex fdiagnostics-show-template-tree
3647 In the C++ frontend, when printing diagnostics showing mismatching
3648 template types, such as:
3651 could not convert 'std::map<int, std::vector<double> >()'
3652 from 'map<[...],vector<double>>' to 'map<[...],vector<float>>
3655 the @option{-fdiagnostics-show-template-tree} flag enables printing a
3656 tree-like structure showing the common and differing parts of the types,
3666 The parts that differ are highlighted with color (``double'' and
3667 ``float'' in this case).
3669 @item -fno-elide-type
3670 @opindex fno-elide-type
3671 @opindex felide-type
3672 By default when the C++ frontend prints diagnostics showing mismatching
3673 template types, common parts of the types are printed as ``[...]'' to
3674 simplify the error message. For example:
3677 could not convert 'std::map<int, std::vector<double> >()'
3678 from 'map<[...],vector<double>>' to 'map<[...],vector<float>>
3681 Specifying the @option{-fno-elide-type} flag suppresses that behavior.
3682 This flag also affects the output of the
3683 @option{-fdiagnostics-show-template-tree} flag.
3685 @item -fno-show-column
3686 @opindex fno-show-column
3687 Do not print column numbers in diagnostics. This may be necessary if
3688 diagnostics are being scanned by a program that does not understand the
3689 column numbers, such as @command{dejagnu}.
3693 @node Warning Options
3694 @section Options to Request or Suppress Warnings
3695 @cindex options to control warnings
3696 @cindex warning messages
3697 @cindex messages, warning
3698 @cindex suppressing warnings
3700 Warnings are diagnostic messages that report constructions that
3701 are not inherently erroneous but that are risky or suggest there
3702 may have been an error.
3704 The following language-independent options do not enable specific
3705 warnings but control the kinds of diagnostics produced by GCC@.
3708 @cindex syntax checking
3710 @opindex fsyntax-only
3711 Check the code for syntax errors, but don't do anything beyond that.
3713 @item -fmax-errors=@var{n}
3714 @opindex fmax-errors
3715 Limits the maximum number of error messages to @var{n}, at which point
3716 GCC bails out rather than attempting to continue processing the source
3717 code. If @var{n} is 0 (the default), there is no limit on the number
3718 of error messages produced. If @option{-Wfatal-errors} is also
3719 specified, then @option{-Wfatal-errors} takes precedence over this
3724 Inhibit all warning messages.
3729 Make all warnings into errors.
3734 Make the specified warning into an error. The specifier for a warning
3735 is appended; for example @option{-Werror=switch} turns the warnings
3736 controlled by @option{-Wswitch} into errors. This switch takes a
3737 negative form, to be used to negate @option{-Werror} for specific
3738 warnings; for example @option{-Wno-error=switch} makes
3739 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3742 The warning message for each controllable warning includes the
3743 option that controls the warning. That option can then be used with
3744 @option{-Werror=} and @option{-Wno-error=} as described above.
3745 (Printing of the option in the warning message can be disabled using the
3746 @option{-fno-diagnostics-show-option} flag.)
3748 Note that specifying @option{-Werror=}@var{foo} automatically implies
3749 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3752 @item -Wfatal-errors
3753 @opindex Wfatal-errors
3754 @opindex Wno-fatal-errors
3755 This option causes the compiler to abort compilation on the first error
3756 occurred rather than trying to keep going and printing further error
3761 You can request many specific warnings with options beginning with
3762 @samp{-W}, for example @option{-Wimplicit} to request warnings on
3763 implicit declarations. Each of these specific warning options also
3764 has a negative form beginning @samp{-Wno-} to turn off warnings; for
3765 example, @option{-Wno-implicit}. This manual lists only one of the
3766 two forms, whichever is not the default. For further
3767 language-specific options also refer to @ref{C++ Dialect Options} and
3768 @ref{Objective-C and Objective-C++ Dialect Options}.
3770 Some options, such as @option{-Wall} and @option{-Wextra}, turn on other
3771 options, such as @option{-Wunused}, which may turn on further options,
3772 such as @option{-Wunused-value}. The combined effect of positive and
3773 negative forms is that more specific options have priority over less
3774 specific ones, independently of their position in the command-line. For
3775 options of the same specificity, the last one takes effect. Options
3776 enabled or disabled via pragmas (@pxref{Diagnostic Pragmas}) take effect
3777 as if they appeared at the end of the command-line.
3779 When an unrecognized warning option is requested (e.g.,
3780 @option{-Wunknown-warning}), GCC emits a diagnostic stating
3781 that the option is not recognized. However, if the @option{-Wno-} form
3782 is used, the behavior is slightly different: no diagnostic is
3783 produced for @option{-Wno-unknown-warning} unless other diagnostics
3784 are being produced. This allows the use of new @option{-Wno-} options
3785 with old compilers, but if something goes wrong, the compiler
3786 warns that an unrecognized option is present.
3793 Issue all the warnings demanded by strict ISO C and ISO C++;
3794 reject all programs that use forbidden extensions, and some other
3795 programs that do not follow ISO C and ISO C++. For ISO C, follows the
3796 version of the ISO C standard specified by any @option{-std} option used.
3798 Valid ISO C and ISO C++ programs should compile properly with or without
3799 this option (though a rare few require @option{-ansi} or a
3800 @option{-std} option specifying the required version of ISO C)@. However,
3801 without this option, certain GNU extensions and traditional C and C++
3802 features are supported as well. With this option, they are rejected.
3804 @option{-Wpedantic} does not cause warning messages for use of the
3805 alternate keywords whose names begin and end with @samp{__}. Pedantic
3806 warnings are also disabled in the expression that follows
3807 @code{__extension__}. However, only system header files should use
3808 these escape routes; application programs should avoid them.
3809 @xref{Alternate Keywords}.
3811 Some users try to use @option{-Wpedantic} to check programs for strict ISO
3812 C conformance. They soon find that it does not do quite what they want:
3813 it finds some non-ISO practices, but not all---only those for which
3814 ISO C @emph{requires} a diagnostic, and some others for which
3815 diagnostics have been added.
3817 A feature to report any failure to conform to ISO C might be useful in
3818 some instances, but would require considerable additional work and would
3819 be quite different from @option{-Wpedantic}. We don't have plans to
3820 support such a feature in the near future.
3822 Where the standard specified with @option{-std} represents a GNU
3823 extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
3824 corresponding @dfn{base standard}, the version of ISO C on which the GNU
3825 extended dialect is based. Warnings from @option{-Wpedantic} are given
3826 where they are required by the base standard. (It does not make sense
3827 for such warnings to be given only for features not in the specified GNU
3828 C dialect, since by definition the GNU dialects of C include all
3829 features the compiler supports with the given option, and there would be
3830 nothing to warn about.)
3832 @item -pedantic-errors
3833 @opindex pedantic-errors
3834 Give an error whenever the @dfn{base standard} (see @option{-Wpedantic})
3835 requires a diagnostic, in some cases where there is undefined behavior
3836 at compile-time and in some other cases that do not prevent compilation
3837 of programs that are valid according to the standard. This is not
3838 equivalent to @option{-Werror=pedantic}, since there are errors enabled
3839 by this option and not enabled by the latter and vice versa.
3844 This enables all the warnings about constructions that some users
3845 consider questionable, and that are easy to avoid (or modify to
3846 prevent the warning), even in conjunction with macros. This also
3847 enables some language-specific warnings described in @ref{C++ Dialect
3848 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
3850 @option{-Wall} turns on the following warning flags:
3852 @gccoptlist{-Waddress @gol
3853 -Warray-bounds=1 @r{(only with} @option{-O2}@r{)} @gol
3855 -Wbool-operation @gol
3856 -Wc++11-compat -Wc++14-compat @gol
3857 -Wcatch-value @r{(C++ and Objective-C++ only)} @gol
3858 -Wchar-subscripts @gol
3860 -Wduplicate-decl-specifier @r{(C and Objective-C only)} @gol
3861 -Wenum-compare @r{(in C/ObjC; this is on by default in C++)} @gol
3863 -Wint-in-bool-context @gol
3864 -Wimplicit @r{(C and Objective-C only)} @gol
3865 -Wimplicit-int @r{(C and Objective-C only)} @gol
3866 -Wimplicit-function-declaration @r{(C and Objective-C only)} @gol
3867 -Winit-self @r{(only for C++)} @gol
3868 -Wlogical-not-parentheses @gol
3869 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
3870 -Wmaybe-uninitialized @gol
3871 -Wmemset-elt-size @gol
3872 -Wmemset-transposed-args @gol
3873 -Wmisleading-indentation @r{(only for C/C++)} @gol
3874 -Wmissing-braces @r{(only for C/ObjC)} @gol
3875 -Wmultistatement-macros @gol
3876 -Wnarrowing @r{(only for C++)} @gol
3878 -Wnonnull-compare @gol
3884 -Wsequence-point @gol
3885 -Wsign-compare @r{(only in C++)} @gol
3886 -Wsizeof-pointer-div @gol
3887 -Wsizeof-pointer-memaccess @gol
3888 -Wstrict-aliasing @gol
3889 -Wstrict-overflow=1 @gol
3891 -Wtautological-compare @gol
3893 -Wuninitialized @gol
3894 -Wunknown-pragmas @gol
3895 -Wunused-function @gol
3898 -Wunused-variable @gol
3899 -Wvolatile-register-var @gol
3902 Note that some warning flags are not implied by @option{-Wall}. Some of
3903 them warn about constructions that users generally do not consider
3904 questionable, but which occasionally you might wish to check for;
3905 others warn about constructions that are necessary or hard to avoid in
3906 some cases, and there is no simple way to modify the code to suppress
3907 the warning. Some of them are enabled by @option{-Wextra} but many of
3908 them must be enabled individually.
3914 This enables some extra warning flags that are not enabled by
3915 @option{-Wall}. (This option used to be called @option{-W}. The older
3916 name is still supported, but the newer name is more descriptive.)
3918 @gccoptlist{-Wclobbered @gol
3920 -Wignored-qualifiers @gol
3921 -Wimplicit-fallthrough=3 @gol
3922 -Wmissing-field-initializers @gol
3923 -Wmissing-parameter-type @r{(C only)} @gol
3924 -Wold-style-declaration @r{(C only)} @gol
3925 -Woverride-init @gol
3926 -Wsign-compare @r{(C only)} @gol
3928 -Wuninitialized @gol
3929 -Wshift-negative-value @r{(in C++03 and in C99 and newer)} @gol
3930 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
3931 -Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
3934 The option @option{-Wextra} also prints warning messages for the
3940 A pointer is compared against integer zero with @code{<}, @code{<=},
3941 @code{>}, or @code{>=}.
3944 (C++ only) An enumerator and a non-enumerator both appear in a
3945 conditional expression.
3948 (C++ only) Ambiguous virtual bases.
3951 (C++ only) Subscripting an array that has been declared @code{register}.
3954 (C++ only) Taking the address of a variable that has been declared
3958 (C++ only) A base class is not initialized in the copy constructor
3963 @item -Wchar-subscripts
3964 @opindex Wchar-subscripts
3965 @opindex Wno-char-subscripts
3966 Warn if an array subscript has type @code{char}. This is a common cause
3967 of error, as programmers often forget that this type is signed on some
3969 This warning is enabled by @option{-Wall}.
3973 Warn about an invalid memory access that is found by Pointer Bounds Checker
3974 (@option{-fcheck-pointer-bounds}).
3976 @item -Wno-coverage-mismatch
3977 @opindex Wno-coverage-mismatch
3978 Warn if feedback profiles do not match when using the
3979 @option{-fprofile-use} option.
3980 If a source file is changed between compiling with @option{-fprofile-gen} and
3981 with @option{-fprofile-use}, the files with the profile feedback can fail
3982 to match the source file and GCC cannot use the profile feedback
3983 information. By default, this warning is enabled and is treated as an
3984 error. @option{-Wno-coverage-mismatch} can be used to disable the
3985 warning or @option{-Wno-error=coverage-mismatch} can be used to
3986 disable the error. Disabling the error for this warning can result in
3987 poorly optimized code and is useful only in the
3988 case of very minor changes such as bug fixes to an existing code-base.
3989 Completely disabling the warning is not recommended.
3992 @r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
3994 Suppress warning messages emitted by @code{#warning} directives.
3996 @item -Wdouble-promotion @r{(C, C++, Objective-C and Objective-C++ only)}
3997 @opindex Wdouble-promotion
3998 @opindex Wno-double-promotion
3999 Give a warning when a value of type @code{float} is implicitly
4000 promoted to @code{double}. CPUs with a 32-bit ``single-precision''
4001 floating-point unit implement @code{float} in hardware, but emulate
4002 @code{double} in software. On such a machine, doing computations
4003 using @code{double} values is much more expensive because of the
4004 overhead required for software emulation.
4006 It is easy to accidentally do computations with @code{double} because
4007 floating-point literals are implicitly of type @code{double}. For
4011 float area(float radius)
4013 return 3.14159 * radius * radius;
4017 the compiler performs the entire computation with @code{double}
4018 because the floating-point literal is a @code{double}.
4020 @item -Wduplicate-decl-specifier @r{(C and Objective-C only)}
4021 @opindex Wduplicate-decl-specifier
4022 @opindex Wno-duplicate-decl-specifier
4023 Warn if a declaration has duplicate @code{const}, @code{volatile},
4024 @code{restrict} or @code{_Atomic} specifier. This warning is enabled by
4028 @itemx -Wformat=@var{n}
4031 @opindex ffreestanding
4032 @opindex fno-builtin
4034 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
4035 the arguments supplied have types appropriate to the format string
4036 specified, and that the conversions specified in the format string make
4037 sense. This includes standard functions, and others specified by format
4038 attributes (@pxref{Function Attributes}), in the @code{printf},
4039 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
4040 not in the C standard) families (or other target-specific families).
4041 Which functions are checked without format attributes having been
4042 specified depends on the standard version selected, and such checks of
4043 functions without the attribute specified are disabled by
4044 @option{-ffreestanding} or @option{-fno-builtin}.
4046 The formats are checked against the format features supported by GNU
4047 libc version 2.2. These include all ISO C90 and C99 features, as well
4048 as features from the Single Unix Specification and some BSD and GNU
4049 extensions. Other library implementations may not support all these
4050 features; GCC does not support warning about features that go beyond a
4051 particular library's limitations. However, if @option{-Wpedantic} is used
4052 with @option{-Wformat}, warnings are given about format features not
4053 in the selected standard version (but not for @code{strfmon} formats,
4054 since those are not in any version of the C standard). @xref{C Dialect
4055 Options,,Options Controlling C Dialect}.
4062 Option @option{-Wformat} is equivalent to @option{-Wformat=1}, and
4063 @option{-Wno-format} is equivalent to @option{-Wformat=0}. Since
4064 @option{-Wformat} also checks for null format arguments for several
4065 functions, @option{-Wformat} also implies @option{-Wnonnull}. Some
4066 aspects of this level of format checking can be disabled by the
4067 options: @option{-Wno-format-contains-nul},
4068 @option{-Wno-format-extra-args}, and @option{-Wno-format-zero-length}.
4069 @option{-Wformat} is enabled by @option{-Wall}.
4071 @item -Wno-format-contains-nul
4072 @opindex Wno-format-contains-nul
4073 @opindex Wformat-contains-nul
4074 If @option{-Wformat} is specified, do not warn about format strings that
4077 @item -Wno-format-extra-args
4078 @opindex Wno-format-extra-args
4079 @opindex Wformat-extra-args
4080 If @option{-Wformat} is specified, do not warn about excess arguments to a
4081 @code{printf} or @code{scanf} format function. The C standard specifies
4082 that such arguments are ignored.
4084 Where the unused arguments lie between used arguments that are
4085 specified with @samp{$} operand number specifications, normally
4086 warnings are still given, since the implementation could not know what
4087 type to pass to @code{va_arg} to skip the unused arguments. However,
4088 in the case of @code{scanf} formats, this option suppresses the
4089 warning if the unused arguments are all pointers, since the Single
4090 Unix Specification says that such unused arguments are allowed.
4092 @item -Wformat-overflow
4093 @itemx -Wformat-overflow=@var{level}
4094 @opindex Wformat-overflow
4095 @opindex Wno-format-overflow
4096 Warn about calls to formatted input/output functions such as @code{sprintf}
4097 and @code{vsprintf} that might overflow the destination buffer. When the
4098 exact number of bytes written by a format directive cannot be determined
4099 at compile-time it is estimated based on heuristics that depend on the
4100 @var{level} argument and on optimization. While enabling optimization
4101 will in most cases improve the accuracy of the warning, it may also
4102 result in false positives.
4105 @item -Wformat-overflow
4106 @item -Wformat-overflow=1
4107 @opindex Wformat-overflow
4108 @opindex Wno-format-overflow
4109 Level @var{1} of @option{-Wformat-overflow} enabled by @option{-Wformat}
4110 employs a conservative approach that warns only about calls that most
4111 likely overflow the buffer. At this level, numeric arguments to format
4112 directives with unknown values are assumed to have the value of one, and
4113 strings of unknown length to be empty. Numeric arguments that are known
4114 to be bounded to a subrange of their type, or string arguments whose output
4115 is bounded either by their directive's precision or by a finite set of
4116 string literals, are assumed to take on the value within the range that
4117 results in the most bytes on output. For example, the call to @code{sprintf}
4118 below is diagnosed because even with both @var{a} and @var{b} equal to zero,
4119 the terminating NUL character (@code{'\0'}) appended by the function
4120 to the destination buffer will be written past its end. Increasing
4121 the size of the buffer by a single byte is sufficient to avoid the
4122 warning, though it may not be sufficient to avoid the overflow.
4125 void f (int a, int b)
4128 sprintf (buf, "a = %i, b = %i\n", a, b);
4132 @item -Wformat-overflow=2
4133 Level @var{2} warns also about calls that might overflow the destination
4134 buffer given an argument of sufficient length or magnitude. At level
4135 @var{2}, unknown numeric arguments are assumed to have the minimum
4136 representable value for signed types with a precision greater than 1, and
4137 the maximum representable value otherwise. Unknown string arguments whose
4138 length cannot be assumed to be bounded either by the directive's precision,
4139 or by a finite set of string literals they may evaluate to, or the character
4140 array they may point to, are assumed to be 1 character long.
4142 At level @var{2}, the call in the example above is again diagnosed, but
4143 this time because with @var{a} equal to a 32-bit @code{INT_MIN} the first
4144 @code{%i} directive will write some of its digits beyond the end of
4145 the destination buffer. To make the call safe regardless of the values
4146 of the two variables, the size of the destination buffer must be increased
4147 to at least 34 bytes. GCC includes the minimum size of the buffer in
4148 an informational note following the warning.
4150 An alternative to increasing the size of the destination buffer is to
4151 constrain the range of formatted values. The maximum length of string
4152 arguments can be bounded by specifying the precision in the format
4153 directive. When numeric arguments of format directives can be assumed
4154 to be bounded by less than the precision of their type, choosing
4155 an appropriate length modifier to the format specifier will reduce
4156 the required buffer size. For example, if @var{a} and @var{b} in the
4157 example above can be assumed to be within the precision of
4158 the @code{short int} type then using either the @code{%hi} format
4159 directive or casting the argument to @code{short} reduces the maximum
4160 required size of the buffer to 24 bytes.
4163 void f (int a, int b)
4166 sprintf (buf, "a = %hi, b = %i\n", a, (short)b);
4171 @item -Wno-format-zero-length
4172 @opindex Wno-format-zero-length
4173 @opindex Wformat-zero-length
4174 If @option{-Wformat} is specified, do not warn about zero-length formats.
4175 The C standard specifies that zero-length formats are allowed.
4180 Enable @option{-Wformat} plus additional format checks. Currently
4181 equivalent to @option{-Wformat -Wformat-nonliteral -Wformat-security
4184 @item -Wformat-nonliteral
4185 @opindex Wformat-nonliteral
4186 @opindex Wno-format-nonliteral
4187 If @option{-Wformat} is specified, also warn if the format string is not a
4188 string literal and so cannot be checked, unless the format function
4189 takes its format arguments as a @code{va_list}.
4191 @item -Wformat-security
4192 @opindex Wformat-security
4193 @opindex Wno-format-security
4194 If @option{-Wformat} is specified, also warn about uses of format
4195 functions that represent possible security problems. At present, this
4196 warns about calls to @code{printf} and @code{scanf} functions where the
4197 format string is not a string literal and there are no format arguments,
4198 as in @code{printf (foo);}. This may be a security hole if the format
4199 string came from untrusted input and contains @samp{%n}. (This is
4200 currently a subset of what @option{-Wformat-nonliteral} warns about, but
4201 in future warnings may be added to @option{-Wformat-security} that are not
4202 included in @option{-Wformat-nonliteral}.)
4204 @item -Wformat-signedness
4205 @opindex Wformat-signedness
4206 @opindex Wno-format-signedness
4207 If @option{-Wformat} is specified, also warn if the format string
4208 requires an unsigned argument and the argument is signed and vice versa.
4210 @item -Wformat-truncation
4211 @itemx -Wformat-truncation=@var{level}
4212 @opindex Wformat-truncation
4213 @opindex Wno-format-truncation
4214 Warn about calls to formatted input/output functions such as @code{snprintf}
4215 and @code{vsnprintf} that might result in output truncation. When the exact
4216 number of bytes written by a format directive cannot be determined at
4217 compile-time it is estimated based on heuristics that depend on
4218 the @var{level} argument and on optimization. While enabling optimization
4219 will in most cases improve the accuracy of the warning, it may also result
4220 in false positives. Except as noted otherwise, the option uses the same
4221 logic @option{-Wformat-overflow}.
4224 @item -Wformat-truncation
4225 @item -Wformat-truncation=1
4226 @opindex Wformat-truncation
4227 @opindex Wno-format-overflow
4228 Level @var{1} of @option{-Wformat-truncation} enabled by @option{-Wformat}
4229 employs a conservative approach that warns only about calls to bounded
4230 functions whose return value is unused and that will most likely result
4231 in output truncation.
4233 @item -Wformat-truncation=2
4234 Level @var{2} warns also about calls to bounded functions whose return
4235 value is used and that might result in truncation given an argument of
4236 sufficient length or magnitude.
4240 @opindex Wformat-y2k
4241 @opindex Wno-format-y2k
4242 If @option{-Wformat} is specified, also warn about @code{strftime}
4243 formats that may yield only a two-digit year.
4248 @opindex Wno-nonnull
4249 Warn about passing a null pointer for arguments marked as
4250 requiring a non-null value by the @code{nonnull} function attribute.
4252 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
4253 can be disabled with the @option{-Wno-nonnull} option.
4255 @item -Wnonnull-compare
4256 @opindex Wnonnull-compare
4257 @opindex Wno-nonnull-compare
4258 Warn when comparing an argument marked with the @code{nonnull}
4259 function attribute against null inside the function.
4261 @option{-Wnonnull-compare} is included in @option{-Wall}. It
4262 can be disabled with the @option{-Wno-nonnull-compare} option.
4264 @item -Wnull-dereference
4265 @opindex Wnull-dereference
4266 @opindex Wno-null-dereference
4267 Warn if the compiler detects paths that trigger erroneous or
4268 undefined behavior due to dereferencing a null pointer. This option
4269 is only active when @option{-fdelete-null-pointer-checks} is active,
4270 which is enabled by optimizations in most targets. The precision of
4271 the warnings depends on the optimization options used.
4273 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
4275 @opindex Wno-init-self
4276 Warn about uninitialized variables that are initialized with themselves.
4277 Note this option can only be used with the @option{-Wuninitialized} option.
4279 For example, GCC warns about @code{i} being uninitialized in the
4280 following snippet only when @option{-Winit-self} has been specified:
4291 This warning is enabled by @option{-Wall} in C++.
4293 @item -Wimplicit-int @r{(C and Objective-C only)}
4294 @opindex Wimplicit-int
4295 @opindex Wno-implicit-int
4296 Warn when a declaration does not specify a type.
4297 This warning is enabled by @option{-Wall}.
4299 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
4300 @opindex Wimplicit-function-declaration
4301 @opindex Wno-implicit-function-declaration
4302 Give a warning whenever a function is used before being declared. In
4303 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
4304 enabled by default and it is made into an error by
4305 @option{-pedantic-errors}. This warning is also enabled by
4308 @item -Wimplicit @r{(C and Objective-C only)}
4310 @opindex Wno-implicit
4311 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
4312 This warning is enabled by @option{-Wall}.
4314 @item -Wimplicit-fallthrough
4315 @opindex Wimplicit-fallthrough
4316 @opindex Wno-implicit-fallthrough
4317 @option{-Wimplicit-fallthrough} is the same as @option{-Wimplicit-fallthrough=3}
4318 and @option{-Wno-implicit-fallthrough} is the same as
4319 @option{-Wimplicit-fallthrough=0}.
4321 @item -Wimplicit-fallthrough=@var{n}
4322 @opindex Wimplicit-fallthrough=
4323 Warn when a switch case falls through. For example:
4341 This warning does not warn when the last statement of a case cannot
4342 fall through, e.g. when there is a return statement or a call to function
4343 declared with the noreturn attribute. @option{-Wimplicit-fallthrough=}
4344 also takes into account control flow statements, such as ifs, and only
4345 warns when appropriate. E.g.@:
4355 @} else if (i < 1) @{
4365 Since there are occasions where a switch case fall through is desirable,
4366 GCC provides an attribute, @code{__attribute__ ((fallthrough))}, that is
4367 to be used along with a null statement to suppress this warning that
4368 would normally occur:
4376 __attribute__ ((fallthrough));
4383 C++17 provides a standard way to suppress the @option{-Wimplicit-fallthrough}
4384 warning using @code{[[fallthrough]];} instead of the GNU attribute. In C++11
4385 or C++14 users can use @code{[[gnu::fallthrough]];}, which is a GNU extension.
4386 Instead of these attributes, it is also possible to add a fallthrough comment
4387 to silence the warning. The whole body of the C or C++ style comment should
4388 match the given regular expressions listed below. The option argument @var{n}
4389 specifies what kind of comments are accepted:
4393 @item @option{-Wimplicit-fallthrough=0} disables the warning altogether.
4395 @item @option{-Wimplicit-fallthrough=1} matches @code{.*} regular
4396 expression, any comment is used as fallthrough comment.
4398 @item @option{-Wimplicit-fallthrough=2} case insensitively matches
4399 @code{.*falls?[ \t-]*thr(ough|u).*} regular expression.
4401 @item @option{-Wimplicit-fallthrough=3} case sensitively matches one of the
4402 following regular expressions:
4406 @item @code{-fallthrough}
4408 @item @code{@@fallthrough@@}
4410 @item @code{lint -fallthrough[ \t]*}
4412 @item @code{[ \t.!]*(ELSE,? |INTENTIONAL(LY)? )?@*FALL(S | |-)?THR(OUGH|U)[ \t.!]*(-[^\n\r]*)?}
4414 @item @code{[ \t.!]*(Else,? |Intentional(ly)? )?@*Fall((s | |-)[Tt]|t)hr(ough|u)[ \t.!]*(-[^\n\r]*)?}
4416 @item @code{[ \t.!]*([Ee]lse,? |[Ii]ntentional(ly)? )?@*fall(s | |-)?thr(ough|u)[ \t.!]*(-[^\n\r]*)?}
4420 @item @option{-Wimplicit-fallthrough=4} case sensitively matches one of the
4421 following regular expressions:
4425 @item @code{-fallthrough}
4427 @item @code{@@fallthrough@@}
4429 @item @code{lint -fallthrough[ \t]*}
4431 @item @code{[ \t]*FALLTHR(OUGH|U)[ \t]*}
4435 @item @option{-Wimplicit-fallthrough=5} doesn't recognize any comments as
4436 fallthrough comments, only attributes disable the warning.
4440 The comment needs to be followed after optional whitespace and other comments
4441 by @code{case} or @code{default} keywords or by a user label that precedes some
4442 @code{case} or @code{default} label.
4457 The @option{-Wimplicit-fallthrough=3} warning is enabled by @option{-Wextra}.
4459 @item -Wif-not-aligned @r{(C, C++, Objective-C and Objective-C++ only)}
4460 @opindex Wif-not-aligned
4461 @opindex Wno-if-not-aligned
4462 Control if warning triggered by the @code{warn_if_not_aligned} attribute
4463 should be issued. This is is enabled by default.
4464 Use @option{-Wno-if-not-aligned} to disable it.
4466 @item -Wignored-qualifiers @r{(C and C++ only)}
4467 @opindex Wignored-qualifiers
4468 @opindex Wno-ignored-qualifiers
4469 Warn if the return type of a function has a type qualifier
4470 such as @code{const}. For ISO C such a type qualifier has no effect,
4471 since the value returned by a function is not an lvalue.
4472 For C++, the warning is only emitted for scalar types or @code{void}.
4473 ISO C prohibits qualified @code{void} return types on function
4474 definitions, so such return types always receive a warning
4475 even without this option.
4477 This warning is also enabled by @option{-Wextra}.
4479 @item -Wignored-attributes @r{(C and C++ only)}
4480 @opindex Wignored-attributes
4481 @opindex Wno-ignored-attributes
4482 Warn when an attribute is ignored. This is different from the
4483 @option{-Wattributes} option in that it warns whenever the compiler decides
4484 to drop an attribute, not that the attribute is either unknown, used in a
4485 wrong place, etc. This warning is enabled by default.
4490 Warn if the type of @code{main} is suspicious. @code{main} should be
4491 a function with external linkage, returning int, taking either zero
4492 arguments, two, or three arguments of appropriate types. This warning
4493 is enabled by default in C++ and is enabled by either @option{-Wall}
4494 or @option{-Wpedantic}.
4496 @item -Wmisleading-indentation @r{(C and C++ only)}
4497 @opindex Wmisleading-indentation
4498 @opindex Wno-misleading-indentation
4499 Warn when the indentation of the code does not reflect the block structure.
4500 Specifically, a warning is issued for @code{if}, @code{else}, @code{while}, and
4501 @code{for} clauses with a guarded statement that does not use braces,
4502 followed by an unguarded statement with the same indentation.
4504 In the following example, the call to ``bar'' is misleadingly indented as
4505 if it were guarded by the ``if'' conditional.
4508 if (some_condition ())
4510 bar (); /* Gotcha: this is not guarded by the "if". */
4513 In the case of mixed tabs and spaces, the warning uses the
4514 @option{-ftabstop=} option to determine if the statements line up
4517 The warning is not issued for code involving multiline preprocessor logic
4518 such as the following example.
4523 #if SOME_CONDITION_THAT_DOES_NOT_HOLD
4529 The warning is not issued after a @code{#line} directive, since this
4530 typically indicates autogenerated code, and no assumptions can be made
4531 about the layout of the file that the directive references.
4533 This warning is enabled by @option{-Wall} in C and C++.
4535 @item -Wmissing-braces
4536 @opindex Wmissing-braces
4537 @opindex Wno-missing-braces
4538 Warn if an aggregate or union initializer is not fully bracketed. In
4539 the following example, the initializer for @code{a} is not fully
4540 bracketed, but that for @code{b} is fully bracketed. This warning is
4541 enabled by @option{-Wall} in C.
4544 int a[2][2] = @{ 0, 1, 2, 3 @};
4545 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
4548 This warning is enabled by @option{-Wall}.
4550 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
4551 @opindex Wmissing-include-dirs
4552 @opindex Wno-missing-include-dirs
4553 Warn if a user-supplied include directory does not exist.
4555 @item -Wmultistatement-macros
4556 @opindex Wmultistatement-macros
4557 @opindex Wno-multistatement-macros
4558 Warn about unsafe multiple statement macros that appear to be guarded
4559 by a clause such as @code{if}, @code{else}, @code{for}, @code{switch}, or
4560 @code{while}, in which only the first statement is actually guarded after
4561 the macro is expanded.
4566 #define DOIT x++; y++
4571 will increment @code{y} unconditionally, not just when @code{c} holds.
4572 The can usually be fixed by wrapping the macro in a do-while loop:
4574 #define DOIT do @{ x++; y++; @} while (0)
4579 This warning is enabled by @option{-Wall} in C and C++.
4582 @opindex Wparentheses
4583 @opindex Wno-parentheses
4584 Warn if parentheses are omitted in certain contexts, such
4585 as when there is an assignment in a context where a truth value
4586 is expected, or when operators are nested whose precedence people
4587 often get confused about.
4589 Also warn if a comparison like @code{x<=y<=z} appears; this is
4590 equivalent to @code{(x<=y ? 1 : 0) <= z}, which is a different
4591 interpretation from that of ordinary mathematical notation.
4593 Also warn for dangerous uses of the GNU extension to
4594 @code{?:} with omitted middle operand. When the condition
4595 in the @code{?}: operator is a boolean expression, the omitted value is
4596 always 1. Often programmers expect it to be a value computed
4597 inside the conditional expression instead.
4599 For C++ this also warns for some cases of unnecessary parentheses in
4600 declarations, which can indicate an attempt at a function call instead
4604 // Declares a local variable called mymutex.
4605 std::unique_lock<std::mutex> (mymutex);
4606 // User meant std::unique_lock<std::mutex> lock (mymutex);
4610 This warning is enabled by @option{-Wall}.
4612 @item -Wsequence-point
4613 @opindex Wsequence-point
4614 @opindex Wno-sequence-point
4615 Warn about code that may have undefined semantics because of violations
4616 of sequence point rules in the C and C++ standards.
4618 The C and C++ standards define the order in which expressions in a C/C++
4619 program are evaluated in terms of @dfn{sequence points}, which represent
4620 a partial ordering between the execution of parts of the program: those
4621 executed before the sequence point, and those executed after it. These
4622 occur after the evaluation of a full expression (one which is not part
4623 of a larger expression), after the evaluation of the first operand of a
4624 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
4625 function is called (but after the evaluation of its arguments and the
4626 expression denoting the called function), and in certain other places.
4627 Other than as expressed by the sequence point rules, the order of
4628 evaluation of subexpressions of an expression is not specified. All
4629 these rules describe only a partial order rather than a total order,
4630 since, for example, if two functions are called within one expression
4631 with no sequence point between them, the order in which the functions
4632 are called is not specified. However, the standards committee have
4633 ruled that function calls do not overlap.
4635 It is not specified when between sequence points modifications to the
4636 values of objects take effect. Programs whose behavior depends on this
4637 have undefined behavior; the C and C++ standards specify that ``Between
4638 the previous and next sequence point an object shall have its stored
4639 value modified at most once by the evaluation of an expression.
4640 Furthermore, the prior value shall be read only to determine the value
4641 to be stored.''. If a program breaks these rules, the results on any
4642 particular implementation are entirely unpredictable.
4644 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
4645 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
4646 diagnosed by this option, and it may give an occasional false positive
4647 result, but in general it has been found fairly effective at detecting
4648 this sort of problem in programs.
4650 The C++17 standard will define the order of evaluation of operands in
4651 more cases: in particular it requires that the right-hand side of an
4652 assignment be evaluated before the left-hand side, so the above
4653 examples are no longer undefined. But this warning will still warn
4654 about them, to help people avoid writing code that is undefined in C
4655 and earlier revisions of C++.
4657 The standard is worded confusingly, therefore there is some debate
4658 over the precise meaning of the sequence point rules in subtle cases.
4659 Links to discussions of the problem, including proposed formal
4660 definitions, may be found on the GCC readings page, at
4661 @uref{http://gcc.gnu.org/@/readings.html}.
4663 This warning is enabled by @option{-Wall} for C and C++.
4665 @item -Wno-return-local-addr
4666 @opindex Wno-return-local-addr
4667 @opindex Wreturn-local-addr
4668 Do not warn about returning a pointer (or in C++, a reference) to a
4669 variable that goes out of scope after the function returns.
4672 @opindex Wreturn-type
4673 @opindex Wno-return-type
4674 Warn whenever a function is defined with a return type that defaults
4675 to @code{int}. Also warn about any @code{return} statement with no
4676 return value in a function whose return type is not @code{void}
4677 (falling off the end of the function body is considered returning
4680 For C only, warn about a @code{return} statement with an expression in a
4681 function whose return type is @code{void}, unless the expression type is
4682 also @code{void}. As a GNU extension, the latter case is accepted
4683 without a warning unless @option{-Wpedantic} is used.
4685 For C++, a function without return type always produces a diagnostic
4686 message, even when @option{-Wno-return-type} is specified. The only
4687 exceptions are @code{main} and functions defined in system headers.
4689 This warning is enabled by @option{-Wall}.
4691 @item -Wshift-count-negative
4692 @opindex Wshift-count-negative
4693 @opindex Wno-shift-count-negative
4694 Warn if shift count is negative. This warning is enabled by default.
4696 @item -Wshift-count-overflow
4697 @opindex Wshift-count-overflow
4698 @opindex Wno-shift-count-overflow
4699 Warn if shift count >= width of type. This warning is enabled by default.
4701 @item -Wshift-negative-value
4702 @opindex Wshift-negative-value
4703 @opindex Wno-shift-negative-value
4704 Warn if left shifting a negative value. This warning is enabled by
4705 @option{-Wextra} in C99 and C++11 modes (and newer).
4707 @item -Wshift-overflow
4708 @itemx -Wshift-overflow=@var{n}
4709 @opindex Wshift-overflow
4710 @opindex Wno-shift-overflow
4711 Warn about left shift overflows. This warning is enabled by
4712 default in C99 and C++11 modes (and newer).
4715 @item -Wshift-overflow=1
4716 This is the warning level of @option{-Wshift-overflow} and is enabled
4717 by default in C99 and C++11 modes (and newer). This warning level does
4718 not warn about left-shifting 1 into the sign bit. (However, in C, such
4719 an overflow is still rejected in contexts where an integer constant expression
4722 @item -Wshift-overflow=2
4723 This warning level also warns about left-shifting 1 into the sign bit,
4724 unless C++14 mode is active.
4730 Warn whenever a @code{switch} statement has an index of enumerated type
4731 and lacks a @code{case} for one or more of the named codes of that
4732 enumeration. (The presence of a @code{default} label prevents this
4733 warning.) @code{case} labels outside the enumeration range also
4734 provoke warnings when this option is used (even if there is a
4735 @code{default} label).
4736 This warning is enabled by @option{-Wall}.
4738 @item -Wswitch-default
4739 @opindex Wswitch-default
4740 @opindex Wno-switch-default
4741 Warn whenever a @code{switch} statement does not have a @code{default}
4745 @opindex Wswitch-enum
4746 @opindex Wno-switch-enum
4747 Warn whenever a @code{switch} statement has an index of enumerated type
4748 and lacks a @code{case} for one or more of the named codes of that
4749 enumeration. @code{case} labels outside the enumeration range also
4750 provoke warnings when this option is used. The only difference
4751 between @option{-Wswitch} and this option is that this option gives a
4752 warning about an omitted enumeration code even if there is a
4753 @code{default} label.
4756 @opindex Wswitch-bool
4757 @opindex Wno-switch-bool
4758 Warn whenever a @code{switch} statement has an index of boolean type
4759 and the case values are outside the range of a boolean type.
4760 It is possible to suppress this warning by casting the controlling
4761 expression to a type other than @code{bool}. For example:
4764 switch ((int) (a == 4))
4770 This warning is enabled by default for C and C++ programs.
4772 @item -Wswitch-unreachable
4773 @opindex Wswitch-unreachable
4774 @opindex Wno-switch-unreachable
4775 Warn whenever a @code{switch} statement contains statements between the
4776 controlling expression and the first case label, which will never be
4777 executed. For example:
4789 @option{-Wswitch-unreachable} does not warn if the statement between the
4790 controlling expression and the first case label is just a declaration:
4803 This warning is enabled by default for C and C++ programs.
4805 @item -Wsync-nand @r{(C and C++ only)}
4807 @opindex Wno-sync-nand
4808 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
4809 built-in functions are used. These functions changed semantics in GCC 4.4.
4811 @item -Wunused-but-set-parameter
4812 @opindex Wunused-but-set-parameter
4813 @opindex Wno-unused-but-set-parameter
4814 Warn whenever a function parameter is assigned to, but otherwise unused
4815 (aside from its declaration).
4817 To suppress this warning use the @code{unused} attribute
4818 (@pxref{Variable Attributes}).
4820 This warning is also enabled by @option{-Wunused} together with
4823 @item -Wunused-but-set-variable
4824 @opindex Wunused-but-set-variable
4825 @opindex Wno-unused-but-set-variable
4826 Warn whenever a local variable is assigned to, but otherwise unused
4827 (aside from its declaration).
4828 This warning is enabled by @option{-Wall}.
4830 To suppress this warning use the @code{unused} attribute
4831 (@pxref{Variable Attributes}).
4833 This warning is also enabled by @option{-Wunused}, which is enabled
4836 @item -Wunused-function
4837 @opindex Wunused-function
4838 @opindex Wno-unused-function
4839 Warn whenever a static function is declared but not defined or a
4840 non-inline static function is unused.
4841 This warning is enabled by @option{-Wall}.
4843 @item -Wunused-label
4844 @opindex Wunused-label
4845 @opindex Wno-unused-label
4846 Warn whenever a label is declared but not used.
4847 This warning is enabled by @option{-Wall}.
4849 To suppress this warning use the @code{unused} attribute
4850 (@pxref{Variable Attributes}).
4852 @item -Wunused-local-typedefs @r{(C, Objective-C, C++ and Objective-C++ only)}
4853 @opindex Wunused-local-typedefs
4854 Warn when a typedef locally defined in a function is not used.
4855 This warning is enabled by @option{-Wall}.
4857 @item -Wunused-parameter
4858 @opindex Wunused-parameter
4859 @opindex Wno-unused-parameter
4860 Warn whenever a function parameter is unused aside from its declaration.
4862 To suppress this warning use the @code{unused} attribute
4863 (@pxref{Variable Attributes}).
4865 @item -Wno-unused-result
4866 @opindex Wunused-result
4867 @opindex Wno-unused-result
4868 Do not warn if a caller of a function marked with attribute
4869 @code{warn_unused_result} (@pxref{Function Attributes}) does not use
4870 its return value. The default is @option{-Wunused-result}.
4872 @item -Wunused-variable
4873 @opindex Wunused-variable
4874 @opindex Wno-unused-variable
4875 Warn whenever a local or static variable is unused aside from its
4876 declaration. This option implies @option{-Wunused-const-variable=1} for C,
4877 but not for C++. This warning is enabled by @option{-Wall}.
4879 To suppress this warning use the @code{unused} attribute
4880 (@pxref{Variable Attributes}).
4882 @item -Wunused-const-variable
4883 @itemx -Wunused-const-variable=@var{n}
4884 @opindex Wunused-const-variable
4885 @opindex Wno-unused-const-variable
4886 Warn whenever a constant static variable is unused aside from its declaration.
4887 @option{-Wunused-const-variable=1} is enabled by @option{-Wunused-variable}
4888 for C, but not for C++. In C this declares variable storage, but in C++ this
4889 is not an error since const variables take the place of @code{#define}s.
4891 To suppress this warning use the @code{unused} attribute
4892 (@pxref{Variable Attributes}).
4895 @item -Wunused-const-variable=1
4896 This is the warning level that is enabled by @option{-Wunused-variable} for
4897 C. It warns only about unused static const variables defined in the main
4898 compilation unit, but not about static const variables declared in any
4901 @item -Wunused-const-variable=2
4902 This warning level also warns for unused constant static variables in
4903 headers (excluding system headers). This is the warning level of
4904 @option{-Wunused-const-variable} and must be explicitly requested since
4905 in C++ this isn't an error and in C it might be harder to clean up all
4909 @item -Wunused-value
4910 @opindex Wunused-value
4911 @opindex Wno-unused-value
4912 Warn whenever a statement computes a result that is explicitly not
4913 used. To suppress this warning cast the unused expression to
4914 @code{void}. This includes an expression-statement or the left-hand
4915 side of a comma expression that contains no side effects. For example,
4916 an expression such as @code{x[i,j]} causes a warning, while
4917 @code{x[(void)i,j]} does not.
4919 This warning is enabled by @option{-Wall}.
4924 All the above @option{-Wunused} options combined.
4926 In order to get a warning about an unused function parameter, you must
4927 either specify @option{-Wextra -Wunused} (note that @option{-Wall} implies
4928 @option{-Wunused}), or separately specify @option{-Wunused-parameter}.
4930 @item -Wuninitialized
4931 @opindex Wuninitialized
4932 @opindex Wno-uninitialized
4933 Warn if an automatic variable is used without first being initialized
4934 or if a variable may be clobbered by a @code{setjmp} call. In C++,
4935 warn if a non-static reference or non-static @code{const} member
4936 appears in a class without constructors.
4938 If you want to warn about code that uses the uninitialized value of the
4939 variable in its own initializer, use the @option{-Winit-self} option.
4941 These warnings occur for individual uninitialized or clobbered
4942 elements of structure, union or array variables as well as for
4943 variables that are uninitialized or clobbered as a whole. They do
4944 not occur for variables or elements declared @code{volatile}. Because
4945 these warnings depend on optimization, the exact variables or elements
4946 for which there are warnings depends on the precise optimization
4947 options and version of GCC used.
4949 Note that there may be no warning about a variable that is used only
4950 to compute a value that itself is never used, because such
4951 computations may be deleted by data flow analysis before the warnings
4954 @item -Winvalid-memory-model
4955 @opindex Winvalid-memory-model
4956 @opindex Wno-invalid-memory-model
4957 Warn for invocations of @ref{__atomic Builtins}, @ref{__sync Builtins},
4958 and the C11 atomic generic functions with a memory consistency argument
4959 that is either invalid for the operation or outside the range of values
4960 of the @code{memory_order} enumeration. For example, since the
4961 @code{__atomic_store} and @code{__atomic_store_n} built-ins are only
4962 defined for the relaxed, release, and sequentially consistent memory
4963 orders the following code is diagnosed:
4968 __atomic_store_n (i, 0, memory_order_consume);
4972 @option{-Winvalid-memory-model} is enabled by default.
4974 @item -Wmaybe-uninitialized
4975 @opindex Wmaybe-uninitialized
4976 @opindex Wno-maybe-uninitialized
4977 For an automatic (i.e.@ local) variable, if there exists a path from the
4978 function entry to a use of the variable that is initialized, but there exist
4979 some other paths for which the variable is not initialized, the compiler
4980 emits a warning if it cannot prove the uninitialized paths are not
4981 executed at run time.
4983 These warnings are only possible in optimizing compilation, because otherwise
4984 GCC does not keep track of the state of variables.
4986 These warnings are made optional because GCC may not be able to determine when
4987 the code is correct in spite of appearing to have an error. Here is one
4988 example of how this can happen:
5008 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
5009 always initialized, but GCC doesn't know this. To suppress the
5010 warning, you need to provide a default case with assert(0) or
5013 @cindex @code{longjmp} warnings
5014 This option also warns when a non-volatile automatic variable might be
5015 changed by a call to @code{longjmp}.
5016 The compiler sees only the calls to @code{setjmp}. It cannot know
5017 where @code{longjmp} will be called; in fact, a signal handler could
5018 call it at any point in the code. As a result, you may get a warning
5019 even when there is in fact no problem because @code{longjmp} cannot
5020 in fact be called at the place that would cause a problem.
5022 Some spurious warnings can be avoided if you declare all the functions
5023 you use that never return as @code{noreturn}. @xref{Function
5026 This warning is enabled by @option{-Wall} or @option{-Wextra}.
5028 @item -Wunknown-pragmas
5029 @opindex Wunknown-pragmas
5030 @opindex Wno-unknown-pragmas
5031 @cindex warning for unknown pragmas
5032 @cindex unknown pragmas, warning
5033 @cindex pragmas, warning of unknown
5034 Warn when a @code{#pragma} directive is encountered that is not understood by
5035 GCC@. If this command-line option is used, warnings are even issued
5036 for unknown pragmas in system header files. This is not the case if
5037 the warnings are only enabled by the @option{-Wall} command-line option.
5040 @opindex Wno-pragmas
5042 Do not warn about misuses of pragmas, such as incorrect parameters,
5043 invalid syntax, or conflicts between pragmas. See also
5044 @option{-Wunknown-pragmas}.
5046 @item -Wstrict-aliasing
5047 @opindex Wstrict-aliasing
5048 @opindex Wno-strict-aliasing
5049 This option is only active when @option{-fstrict-aliasing} is active.
5050 It warns about code that might break the strict aliasing rules that the
5051 compiler is using for optimization. The warning does not catch all
5052 cases, but does attempt to catch the more common pitfalls. It is
5053 included in @option{-Wall}.
5054 It is equivalent to @option{-Wstrict-aliasing=3}
5056 @item -Wstrict-aliasing=n
5057 @opindex Wstrict-aliasing=n
5058 This option is only active when @option{-fstrict-aliasing} is active.
5059 It warns about code that might break the strict aliasing rules that the
5060 compiler is using for optimization.
5061 Higher levels correspond to higher accuracy (fewer false positives).
5062 Higher levels also correspond to more effort, similar to the way @option{-O}
5064 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=3}.
5066 Level 1: Most aggressive, quick, least accurate.
5067 Possibly useful when higher levels
5068 do not warn but @option{-fstrict-aliasing} still breaks the code, as it has very few
5069 false negatives. However, it has many false positives.
5070 Warns for all pointer conversions between possibly incompatible types,
5071 even if never dereferenced. Runs in the front end only.
5073 Level 2: Aggressive, quick, not too precise.
5074 May still have many false positives (not as many as level 1 though),
5075 and few false negatives (but possibly more than level 1).
5076 Unlike level 1, it only warns when an address is taken. Warns about
5077 incomplete types. Runs in the front end only.
5079 Level 3 (default for @option{-Wstrict-aliasing}):
5080 Should have very few false positives and few false
5081 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
5082 Takes care of the common pun+dereference pattern in the front end:
5083 @code{*(int*)&some_float}.
5084 If optimization is enabled, it also runs in the back end, where it deals
5085 with multiple statement cases using flow-sensitive points-to information.
5086 Only warns when the converted pointer is dereferenced.
5087 Does not warn about incomplete types.
5089 @item -Wstrict-overflow
5090 @itemx -Wstrict-overflow=@var{n}
5091 @opindex Wstrict-overflow
5092 @opindex Wno-strict-overflow
5093 This option is only active when signed overflow is undefined.
5094 It warns about cases where the compiler optimizes based on the
5095 assumption that signed overflow does not occur. Note that it does not
5096 warn about all cases where the code might overflow: it only warns
5097 about cases where the compiler implements some optimization. Thus
5098 this warning depends on the optimization level.
5100 An optimization that assumes that signed overflow does not occur is
5101 perfectly safe if the values of the variables involved are such that
5102 overflow never does, in fact, occur. Therefore this warning can
5103 easily give a false positive: a warning about code that is not
5104 actually a problem. To help focus on important issues, several
5105 warning levels are defined. No warnings are issued for the use of
5106 undefined signed overflow when estimating how many iterations a loop
5107 requires, in particular when determining whether a loop will be
5111 @item -Wstrict-overflow=1
5112 Warn about cases that are both questionable and easy to avoid. For
5113 example the compiler simplifies
5114 @code{x + 1 > x} to @code{1}. This level of
5115 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
5116 are not, and must be explicitly requested.
5118 @item -Wstrict-overflow=2
5119 Also warn about other cases where a comparison is simplified to a
5120 constant. For example: @code{abs (x) >= 0}. This can only be
5121 simplified when signed integer overflow is undefined, because
5122 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
5123 zero. @option{-Wstrict-overflow} (with no level) is the same as
5124 @option{-Wstrict-overflow=2}.
5126 @item -Wstrict-overflow=3
5127 Also warn about other cases where a comparison is simplified. For
5128 example: @code{x + 1 > 1} is simplified to @code{x > 0}.
5130 @item -Wstrict-overflow=4
5131 Also warn about other simplifications not covered by the above cases.
5132 For example: @code{(x * 10) / 5} is simplified to @code{x * 2}.
5134 @item -Wstrict-overflow=5
5135 Also warn about cases where the compiler reduces the magnitude of a
5136 constant involved in a comparison. For example: @code{x + 2 > y} is
5137 simplified to @code{x + 1 >= y}. This is reported only at the
5138 highest warning level because this simplification applies to many
5139 comparisons, so this warning level gives a very large number of
5143 @item -Wstringop-overflow
5144 @itemx -Wstringop-overflow=@var{type}
5145 @opindex Wstringop-overflow
5146 @opindex Wno-stringop-overflow
5147 Warn for calls to string manipulation functions such as @code{memcpy} and
5148 @code{strcpy} that are determined to overflow the destination buffer. The
5149 optional argument is one greater than the type of Object Size Checking to
5150 perform to determine the size of the destination. @xref{Object Size Checking}.
5151 The argument is meaningful only for functions that operate on character arrays
5152 but not for raw memory functions like @code{memcpy} which always make use
5153 of Object Size type-0. The option also warns for calls that specify a size
5154 in excess of the largest possible object or at most @code{SIZE_MAX / 2} bytes.
5155 The option produces the best results with optimization enabled but can detect
5156 a small subset of simple buffer overflows even without optimization in
5157 calls to the GCC built-in functions like @code{__builtin_memcpy} that
5158 correspond to the standard functions. In any case, the option warns about
5159 just a subset of buffer overflows detected by the corresponding overflow
5160 checking built-ins. For example, the option will issue a warning for
5161 the @code{strcpy} call below because it copies at least 5 characters
5162 (the string @code{"blue"} including the terminating NUL) into the buffer
5166 enum Color @{ blue, purple, yellow @};
5167 const char* f (enum Color clr)
5169 static char buf [4];
5173 case blue: str = "blue"; break;
5174 case purple: str = "purple"; break;
5175 case yellow: str = "yellow"; break;
5178 return strcpy (buf, str); // warning here
5182 Option @option{-Wstringop-overflow=2} is enabled by default.
5185 @item -Wstringop-overflow
5186 @item -Wstringop-overflow=1
5187 @opindex Wstringop-overflow
5188 @opindex Wno-stringop-overflow
5189 The @option{-Wstringop-overflow=1} option uses type-zero Object Size Checking
5190 to determine the sizes of destination objects. This is the default setting
5191 of the option. At this setting the option will not warn for writes past
5192 the end of subobjects of larger objects accessed by pointers unless the
5193 size of the largest surrounding object is known. When the destination may
5194 be one of several objects it is assumed to be the largest one of them. On
5195 Linux systems, when optimization is enabled at this setting the option warns
5196 for the same code as when the @code{_FORTIFY_SOURCE} macro is defined to
5199 @item -Wstringop-overflow=2
5200 The @option{-Wstringop-overflow=2} option uses type-one Object Size Checking
5201 to determine the sizes of destination objects. At this setting the option
5202 will warn about overflows when writing to members of the largest complete
5203 objects whose exact size is known. It will, however, not warn for excessive
5204 writes to the same members of unknown objects referenced by pointers since
5205 they may point to arrays containing unknown numbers of elements.
5207 @item -Wstringop-overflow=3
5208 The @option{-Wstringop-overflow=3} option uses type-two Object Size Checking
5209 to determine the sizes of destination objects. At this setting the option
5210 warns about overflowing the smallest object or data member. This is the
5211 most restrictive setting of the option that may result in warnings for safe
5214 @item -Wstringop-overflow=4
5215 The @option{-Wstringop-overflow=4} option uses type-three Object Size Checking
5216 to determine the sizes of destination objects. At this setting the option
5217 will warn about overflowing any data members, and when the destination is
5218 one of several objects it uses the size of the largest of them to decide
5219 whether to issue a warning. Similarly to @option{-Wstringop-overflow=3} this
5220 setting of the option may result in warnings for benign code.
5223 @item -Wstringop-truncation
5224 @opindex Wstringop-truncation
5225 @opindex Wno-stringop-truncation
5226 Warn for calls to bounded string manipulation functions such as @code{strncat},
5227 @code{strncpy}, and @code{stpncpy} that may either truncate the copied string
5228 or leave the destination unchanged.
5230 In the following example, the call to @code{strncat} specifies a bound that
5231 is less than the length of the source string. As a result, the copy of
5232 the source will be truncated and so the call is diagnosed. To avoid the
5233 warning use @code{bufsize - strlen (buf) - 1)} as the bound.
5236 void append (char *buf, size_t bufsize)
5238 strncat (buf, ".txt", 3);
5242 As another example, the following call to @code{strncpy} results in copying
5243 to @code{d} just the characters preceding the terminating NUL, without
5244 appending the NUL to the end. Assuming the result of @code{strncpy} is
5245 necessarily a NUL-terminated string is a common mistake, and so the call
5246 is diagnosed. To avoid the warning when the result is not expected to be
5247 NUL-terminated, call @code{memcpy} instead.
5250 void copy (char *d, const char *s)
5252 strncpy (d, s, strlen (s));
5256 In the following example, the call to @code{strncpy} specifies the size
5257 of the destination buffer as the bound. If the length of the source
5258 string is equal to or greater than this size the result of the copy will
5259 not be NUL-terminated. Therefore, the call is also diagnosed. To avoid
5260 the warning, specify @code{sizeof buf - 1} as the bound and set the last
5261 element of the buffer to @code{NUL}.
5264 void copy (const char *s)
5267 strncpy (buf, s, sizeof buf);
5272 @item -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{|}cold@r{|}malloc@r{]}
5273 @opindex Wsuggest-attribute=
5274 @opindex Wno-suggest-attribute=
5275 Warn for cases where adding an attribute may be beneficial. The
5276 attributes currently supported are listed below.
5279 @item -Wsuggest-attribute=pure
5280 @itemx -Wsuggest-attribute=const
5281 @itemx -Wsuggest-attribute=noreturn
5282 @itemx -Wsuggest-attribute=malloc
5283 @opindex Wsuggest-attribute=pure
5284 @opindex Wno-suggest-attribute=pure
5285 @opindex Wsuggest-attribute=const
5286 @opindex Wno-suggest-attribute=const
5287 @opindex Wsuggest-attribute=noreturn
5288 @opindex Wno-suggest-attribute=noreturn
5289 @opindex Wsuggest-attribute=malloc
5290 @opindex Wno-suggest-attribute=malloc
5292 Warn about functions that might be candidates for attributes
5293 @code{pure}, @code{const} or @code{noreturn} or @code{malloc}. The compiler
5294 only warns for functions visible in other compilation units or (in the case of
5295 @code{pure} and @code{const}) if it cannot prove that the function returns
5296 normally. A function returns normally if it doesn't contain an infinite loop or
5297 return abnormally by throwing, calling @code{abort} or trapping. This analysis
5298 requires option @option{-fipa-pure-const}, which is enabled by default at
5299 @option{-O} and higher. Higher optimization levels improve the accuracy
5302 @item -Wsuggest-attribute=format
5303 @itemx -Wmissing-format-attribute
5304 @opindex Wsuggest-attribute=format
5305 @opindex Wmissing-format-attribute
5306 @opindex Wno-suggest-attribute=format
5307 @opindex Wno-missing-format-attribute
5311 Warn about function pointers that might be candidates for @code{format}
5312 attributes. Note these are only possible candidates, not absolute ones.
5313 GCC guesses that function pointers with @code{format} attributes that
5314 are used in assignment, initialization, parameter passing or return
5315 statements should have a corresponding @code{format} attribute in the
5316 resulting type. I.e.@: the left-hand side of the assignment or
5317 initialization, the type of the parameter variable, or the return type
5318 of the containing function respectively should also have a @code{format}
5319 attribute to avoid the warning.
5321 GCC also warns about function definitions that might be
5322 candidates for @code{format} attributes. Again, these are only
5323 possible candidates. GCC guesses that @code{format} attributes
5324 might be appropriate for any function that calls a function like
5325 @code{vprintf} or @code{vscanf}, but this might not always be the
5326 case, and some functions for which @code{format} attributes are
5327 appropriate may not be detected.
5329 @item -Wsuggest-attribute=cold
5330 @opindex Wsuggest-attribute=cold
5331 @opindex Wno-suggest-attribute=cold
5333 Warn about functions that might be candidates for @code{cold} attribute. This
5334 is based on static detection and generally will only warn about functions which
5335 always leads to a call to another @code{cold} function such as wrappers of
5336 C++ @code{throw} or fatal error reporting functions leading to @code{abort}.
5339 @item -Wsuggest-final-types
5340 @opindex Wno-suggest-final-types
5341 @opindex Wsuggest-final-types
5342 Warn about types with virtual methods where code quality would be improved
5343 if the type were declared with the C++11 @code{final} specifier,
5345 declared in an anonymous namespace. This allows GCC to more aggressively
5346 devirtualize the polymorphic calls. This warning is more effective with link
5347 time optimization, where the information about the class hierarchy graph is
5350 @item -Wsuggest-final-methods
5351 @opindex Wno-suggest-final-methods
5352 @opindex Wsuggest-final-methods
5353 Warn about virtual methods where code quality would be improved if the method
5354 were declared with the C++11 @code{final} specifier,
5355 or, if possible, its type were
5356 declared in an anonymous namespace or with the @code{final} specifier.
5358 more effective with link-time optimization, where the information about the
5359 class hierarchy graph is more complete. It is recommended to first consider
5360 suggestions of @option{-Wsuggest-final-types} and then rebuild with new
5363 @item -Wsuggest-override
5364 Warn about overriding virtual functions that are not marked with the override
5368 @opindex Wno-alloc-zero
5369 @opindex Walloc-zero
5370 Warn about calls to allocation functions decorated with attribute
5371 @code{alloc_size} that specify zero bytes, including those to the built-in
5372 forms of the functions @code{aligned_alloc}, @code{alloca}, @code{calloc},
5373 @code{malloc}, and @code{realloc}. Because the behavior of these functions
5374 when called with a zero size differs among implementations (and in the case
5375 of @code{realloc} has been deprecated) relying on it may result in subtle
5376 portability bugs and should be avoided.
5378 @item -Walloc-size-larger-than=@var{n}
5379 Warn about calls to functions decorated with attribute @code{alloc_size}
5380 that attempt to allocate objects larger than the specified number of bytes,
5381 or where the result of the size computation in an integer type with infinite
5382 precision would exceed @code{SIZE_MAX / 2}. The option argument @var{n}
5383 may end in one of the standard suffixes designating a multiple of bytes
5384 such as @code{kB} and @code{KiB} for kilobyte and kibibyte, respectively,
5385 @code{MB} and @code{MiB} for megabyte and mebibyte, and so on.
5386 @xref{Function Attributes}.
5391 This option warns on all uses of @code{alloca} in the source.
5393 @item -Walloca-larger-than=@var{n}
5394 This option warns on calls to @code{alloca} that are not bounded by a
5395 controlling predicate limiting its argument of integer type to at most
5396 @var{n} bytes, or calls to @code{alloca} where the bound is unknown.
5397 Arguments of non-integer types are considered unbounded even if they
5398 appear to be constrained to the expected range.
5400 For example, a bounded case of @code{alloca} could be:
5403 void func (size_t n)
5414 In the above example, passing @code{-Walloca-larger-than=1000} would not
5415 issue a warning because the call to @code{alloca} is known to be at most
5416 1000 bytes. However, if @code{-Walloca-larger-than=500} were passed,
5417 the compiler would emit a warning.
5419 Unbounded uses, on the other hand, are uses of @code{alloca} with no
5420 controlling predicate constraining its integer argument. For example:
5425 void *p = alloca (n);
5430 If @code{-Walloca-larger-than=500} were passed, the above would trigger
5431 a warning, but this time because of the lack of bounds checking.
5433 Note, that even seemingly correct code involving signed integers could
5437 void func (signed int n)
5447 In the above example, @var{n} could be negative, causing a larger than
5448 expected argument to be implicitly cast into the @code{alloca} call.
5450 This option also warns when @code{alloca} is used in a loop.
5452 This warning is not enabled by @option{-Wall}, and is only active when
5453 @option{-ftree-vrp} is active (default for @option{-O2} and above).
5455 See also @option{-Wvla-larger-than=@var{n}}.
5457 @item -Warray-bounds
5458 @itemx -Warray-bounds=@var{n}
5459 @opindex Wno-array-bounds
5460 @opindex Warray-bounds
5461 This option is only active when @option{-ftree-vrp} is active
5462 (default for @option{-O2} and above). It warns about subscripts to arrays
5463 that are always out of bounds. This warning is enabled by @option{-Wall}.
5466 @item -Warray-bounds=1
5467 This is the warning level of @option{-Warray-bounds} and is enabled
5468 by @option{-Wall}; higher levels are not, and must be explicitly requested.
5470 @item -Warray-bounds=2
5471 This warning level also warns about out of bounds access for
5472 arrays at the end of a struct and for arrays accessed through
5473 pointers. This warning level may give a larger number of
5474 false positives and is deactivated by default.
5477 @item -Wattribute-alias
5478 Warn about declarations using the @code{alias} and similar attributes whose
5479 target is incompatible with the type of the alias. @xref{Function Attributes,
5480 ,Declaring Attributes of Functions}.
5482 @item -Wbool-compare
5483 @opindex Wno-bool-compare
5484 @opindex Wbool-compare
5485 Warn about boolean expression compared with an integer value different from
5486 @code{true}/@code{false}. For instance, the following comparison is
5491 if ((n > 1) == 2) @{ @dots{} @}
5493 This warning is enabled by @option{-Wall}.
5495 @item -Wbool-operation
5496 @opindex Wno-bool-operation
5497 @opindex Wbool-operation
5498 Warn about suspicious operations on expressions of a boolean type. For
5499 instance, bitwise negation of a boolean is very likely a bug in the program.
5500 For C, this warning also warns about incrementing or decrementing a boolean,
5501 which rarely makes sense. (In C++, decrementing a boolean is always invalid.
5502 Incrementing a boolean is invalid in C++17, and deprecated otherwise.)
5504 This warning is enabled by @option{-Wall}.
5506 @item -Wduplicated-branches
5507 @opindex Wno-duplicated-branches
5508 @opindex Wduplicated-branches
5509 Warn when an if-else has identical branches. This warning detects cases like
5516 It doesn't warn when both branches contain just a null statement. This warning
5517 also warn for conditional operators:
5519 int i = x ? *p : *p;
5522 @item -Wduplicated-cond
5523 @opindex Wno-duplicated-cond
5524 @opindex Wduplicated-cond
5525 Warn about duplicated conditions in an if-else-if chain. For instance,
5526 warn for the following code:
5528 if (p->q != NULL) @{ @dots{} @}
5529 else if (p->q != NULL) @{ @dots{} @}
5532 @item -Wframe-address
5533 @opindex Wno-frame-address
5534 @opindex Wframe-address
5535 Warn when the @samp{__builtin_frame_address} or @samp{__builtin_return_address}
5536 is called with an argument greater than 0. Such calls may return indeterminate
5537 values or crash the program. The warning is included in @option{-Wall}.
5539 @item -Wno-discarded-qualifiers @r{(C and Objective-C only)}
5540 @opindex Wno-discarded-qualifiers
5541 @opindex Wdiscarded-qualifiers
5542 Do not warn if type qualifiers on pointers are being discarded.
5543 Typically, the compiler warns if a @code{const char *} variable is
5544 passed to a function that takes a @code{char *} parameter. This option
5545 can be used to suppress such a warning.
5547 @item -Wno-discarded-array-qualifiers @r{(C and Objective-C only)}
5548 @opindex Wno-discarded-array-qualifiers
5549 @opindex Wdiscarded-array-qualifiers
5550 Do not warn if type qualifiers on arrays which are pointer targets
5551 are being discarded. Typically, the compiler warns if a
5552 @code{const int (*)[]} variable is passed to a function that
5553 takes a @code{int (*)[]} parameter. This option can be used to
5554 suppress such a warning.
5556 @item -Wno-incompatible-pointer-types @r{(C and Objective-C only)}
5557 @opindex Wno-incompatible-pointer-types
5558 @opindex Wincompatible-pointer-types
5559 Do not warn when there is a conversion between pointers that have incompatible
5560 types. This warning is for cases not covered by @option{-Wno-pointer-sign},
5561 which warns for pointer argument passing or assignment with different
5564 @item -Wno-int-conversion @r{(C and Objective-C only)}
5565 @opindex Wno-int-conversion
5566 @opindex Wint-conversion
5567 Do not warn about incompatible integer to pointer and pointer to integer
5568 conversions. This warning is about implicit conversions; for explicit
5569 conversions the warnings @option{-Wno-int-to-pointer-cast} and
5570 @option{-Wno-pointer-to-int-cast} may be used.
5572 @item -Wno-div-by-zero
5573 @opindex Wno-div-by-zero
5574 @opindex Wdiv-by-zero
5575 Do not warn about compile-time integer division by zero. Floating-point
5576 division by zero is not warned about, as it can be a legitimate way of
5577 obtaining infinities and NaNs.
5579 @item -Wsystem-headers
5580 @opindex Wsystem-headers
5581 @opindex Wno-system-headers
5582 @cindex warnings from system headers
5583 @cindex system headers, warnings from
5584 Print warning messages for constructs found in system header files.
5585 Warnings from system headers are normally suppressed, on the assumption
5586 that they usually do not indicate real problems and would only make the
5587 compiler output harder to read. Using this command-line option tells
5588 GCC to emit warnings from system headers as if they occurred in user
5589 code. However, note that using @option{-Wall} in conjunction with this
5590 option does @emph{not} warn about unknown pragmas in system
5591 headers---for that, @option{-Wunknown-pragmas} must also be used.
5593 @item -Wtautological-compare
5594 @opindex Wtautological-compare
5595 @opindex Wno-tautological-compare
5596 Warn if a self-comparison always evaluates to true or false. This
5597 warning detects various mistakes such as:
5601 if (i > i) @{ @dots{} @}
5604 This warning also warns about bitwise comparisons that always evaluate
5605 to true or false, for instance:
5607 if ((a & 16) == 10) @{ @dots{} @}
5609 will always be false.
5611 This warning is enabled by @option{-Wall}.
5614 @opindex Wtrampolines
5615 @opindex Wno-trampolines
5616 Warn about trampolines generated for pointers to nested functions.
5617 A trampoline is a small piece of data or code that is created at run
5618 time on the stack when the address of a nested function is taken, and is
5619 used to call the nested function indirectly. For some targets, it is
5620 made up of data only and thus requires no special treatment. But, for
5621 most targets, it is made up of code and thus requires the stack to be
5622 made executable in order for the program to work properly.
5625 @opindex Wfloat-equal
5626 @opindex Wno-float-equal
5627 Warn if floating-point values are used in equality comparisons.
5629 The idea behind this is that sometimes it is convenient (for the
5630 programmer) to consider floating-point values as approximations to
5631 infinitely precise real numbers. If you are doing this, then you need
5632 to compute (by analyzing the code, or in some other way) the maximum or
5633 likely maximum error that the computation introduces, and allow for it
5634 when performing comparisons (and when producing output, but that's a
5635 different problem). In particular, instead of testing for equality, you
5636 should check to see whether the two values have ranges that overlap; and
5637 this is done with the relational operators, so equality comparisons are
5640 @item -Wtraditional @r{(C and Objective-C only)}
5641 @opindex Wtraditional
5642 @opindex Wno-traditional
5643 Warn about certain constructs that behave differently in traditional and
5644 ISO C@. Also warn about ISO C constructs that have no traditional C
5645 equivalent, and/or problematic constructs that should be avoided.
5649 Macro parameters that appear within string literals in the macro body.
5650 In traditional C macro replacement takes place within string literals,
5651 but in ISO C it does not.
5654 In traditional C, some preprocessor directives did not exist.
5655 Traditional preprocessors only considered a line to be a directive
5656 if the @samp{#} appeared in column 1 on the line. Therefore
5657 @option{-Wtraditional} warns about directives that traditional C
5658 understands but ignores because the @samp{#} does not appear as the
5659 first character on the line. It also suggests you hide directives like
5660 @code{#pragma} not understood by traditional C by indenting them. Some
5661 traditional implementations do not recognize @code{#elif}, so this option
5662 suggests avoiding it altogether.
5665 A function-like macro that appears without arguments.
5668 The unary plus operator.
5671 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating-point
5672 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
5673 constants.) Note, these suffixes appear in macros defined in the system
5674 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
5675 Use of these macros in user code might normally lead to spurious
5676 warnings, however GCC's integrated preprocessor has enough context to
5677 avoid warning in these cases.
5680 A function declared external in one block and then used after the end of
5684 A @code{switch} statement has an operand of type @code{long}.
5687 A non-@code{static} function declaration follows a @code{static} one.
5688 This construct is not accepted by some traditional C compilers.
5691 The ISO type of an integer constant has a different width or
5692 signedness from its traditional type. This warning is only issued if
5693 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
5694 typically represent bit patterns, are not warned about.
5697 Usage of ISO string concatenation is detected.
5700 Initialization of automatic aggregates.
5703 Identifier conflicts with labels. Traditional C lacks a separate
5704 namespace for labels.
5707 Initialization of unions. If the initializer is zero, the warning is
5708 omitted. This is done under the assumption that the zero initializer in
5709 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
5710 initializer warnings and relies on default initialization to zero in the
5714 Conversions by prototypes between fixed/floating-point values and vice
5715 versa. The absence of these prototypes when compiling with traditional
5716 C causes serious problems. This is a subset of the possible
5717 conversion warnings; for the full set use @option{-Wtraditional-conversion}.
5720 Use of ISO C style function definitions. This warning intentionally is
5721 @emph{not} issued for prototype declarations or variadic functions
5722 because these ISO C features appear in your code when using
5723 libiberty's traditional C compatibility macros, @code{PARAMS} and
5724 @code{VPARAMS}. This warning is also bypassed for nested functions
5725 because that feature is already a GCC extension and thus not relevant to
5726 traditional C compatibility.
5729 @item -Wtraditional-conversion @r{(C and Objective-C only)}
5730 @opindex Wtraditional-conversion
5731 @opindex Wno-traditional-conversion
5732 Warn if a prototype causes a type conversion that is different from what
5733 would happen to the same argument in the absence of a prototype. This
5734 includes conversions of fixed point to floating and vice versa, and
5735 conversions changing the width or signedness of a fixed-point argument
5736 except when the same as the default promotion.
5738 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
5739 @opindex Wdeclaration-after-statement
5740 @opindex Wno-declaration-after-statement
5741 Warn when a declaration is found after a statement in a block. This
5742 construct, known from C++, was introduced with ISO C99 and is by default
5743 allowed in GCC@. It is not supported by ISO C90. @xref{Mixed Declarations}.
5748 Warn whenever a local variable or type declaration shadows another
5749 variable, parameter, type, class member (in C++), or instance variable
5750 (in Objective-C) or whenever a built-in function is shadowed. Note
5751 that in C++, the compiler warns if a local variable shadows an
5752 explicit typedef, but not if it shadows a struct/class/enum.
5753 Same as @option{-Wshadow=global}.
5755 @item -Wno-shadow-ivar @r{(Objective-C only)}
5756 @opindex Wno-shadow-ivar
5757 @opindex Wshadow-ivar
5758 Do not warn whenever a local variable shadows an instance variable in an
5761 @item -Wshadow=global
5762 @opindex Wshadow=local
5763 The default for @option{-Wshadow}. Warns for any (global) shadowing.
5765 @item -Wshadow=local
5766 @opindex Wshadow=local
5767 Warn when a local variable shadows another local variable or parameter.
5768 This warning is enabled by @option{-Wshadow=global}.
5770 @item -Wshadow=compatible-local
5771 @opindex Wshadow=compatible-local
5772 Warn when a local variable shadows another local variable or parameter
5773 whose type is compatible with that of the shadowing variable. In C++,
5774 type compatibility here means the type of the shadowing variable can be
5775 converted to that of the shadowed variable. The creation of this flag
5776 (in addition to @option{-Wshadow=local}) is based on the idea that when
5777 a local variable shadows another one of incompatible type, it is most
5778 likely intentional, not a bug or typo, as shown in the following example:
5782 for (SomeIterator i = SomeObj.begin(); i != SomeObj.end(); ++i)
5784 for (int i = 0; i < N; ++i)
5793 Since the two variable @code{i} in the example above have incompatible types,
5794 enabling only @option{-Wshadow=compatible-local} will not emit a warning.
5795 Because their types are incompatible, if a programmer accidentally uses one
5796 in place of the other, type checking will catch that and emit an error or
5797 warning. So not warning (about shadowing) in this case will not lead to
5798 undetected bugs. Use of this flag instead of @option{-Wshadow=local} can
5799 possibly reduce the number of warnings triggered by intentional shadowing.
5801 This warning is enabled by @option{-Wshadow=local}.
5803 @item -Wlarger-than=@var{len}
5804 @opindex Wlarger-than=@var{len}
5805 @opindex Wlarger-than-@var{len}
5806 Warn whenever an object of larger than @var{len} bytes is defined.
5808 @item -Wframe-larger-than=@var{len}
5809 @opindex Wframe-larger-than
5810 Warn if the size of a function frame is larger than @var{len} bytes.
5811 The computation done to determine the stack frame size is approximate
5812 and not conservative.
5813 The actual requirements may be somewhat greater than @var{len}
5814 even if you do not get a warning. In addition, any space allocated
5815 via @code{alloca}, variable-length arrays, or related constructs
5816 is not included by the compiler when determining
5817 whether or not to issue a warning.
5819 @item -Wno-free-nonheap-object
5820 @opindex Wno-free-nonheap-object
5821 @opindex Wfree-nonheap-object
5822 Do not warn when attempting to free an object that was not allocated
5825 @item -Wstack-usage=@var{len}
5826 @opindex Wstack-usage
5827 Warn if the stack usage of a function might be larger than @var{len} bytes.
5828 The computation done to determine the stack usage is conservative.
5829 Any space allocated via @code{alloca}, variable-length arrays, or related
5830 constructs is included by the compiler when determining whether or not to
5833 The message is in keeping with the output of @option{-fstack-usage}.
5837 If the stack usage is fully static but exceeds the specified amount, it's:
5840 warning: stack usage is 1120 bytes
5843 If the stack usage is (partly) dynamic but bounded, it's:
5846 warning: stack usage might be 1648 bytes
5849 If the stack usage is (partly) dynamic and not bounded, it's:
5852 warning: stack usage might be unbounded
5856 @item -Wunsafe-loop-optimizations
5857 @opindex Wunsafe-loop-optimizations
5858 @opindex Wno-unsafe-loop-optimizations
5859 Warn if the loop cannot be optimized because the compiler cannot
5860 assume anything on the bounds of the loop indices. With
5861 @option{-funsafe-loop-optimizations} warn if the compiler makes
5864 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
5865 @opindex Wno-pedantic-ms-format
5866 @opindex Wpedantic-ms-format
5867 When used in combination with @option{-Wformat}
5868 and @option{-pedantic} without GNU extensions, this option
5869 disables the warnings about non-ISO @code{printf} / @code{scanf} format
5870 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets,
5871 which depend on the MS runtime.
5874 @opindex Waligned-new
5875 @opindex Wno-aligned-new
5876 Warn about a new-expression of a type that requires greater alignment
5877 than the @code{alignof(std::max_align_t)} but uses an allocation
5878 function without an explicit alignment parameter. This option is
5879 enabled by @option{-Wall}.
5881 Normally this only warns about global allocation functions, but
5882 @option{-Waligned-new=all} also warns about class member allocation
5885 @item -Wplacement-new
5886 @itemx -Wplacement-new=@var{n}
5887 @opindex Wplacement-new
5888 @opindex Wno-placement-new
5889 Warn about placement new expressions with undefined behavior, such as
5890 constructing an object in a buffer that is smaller than the type of
5891 the object. For example, the placement new expression below is diagnosed
5892 because it attempts to construct an array of 64 integers in a buffer only
5898 This warning is enabled by default.
5901 @item -Wplacement-new=1
5902 This is the default warning level of @option{-Wplacement-new}. At this
5903 level the warning is not issued for some strictly undefined constructs that
5904 GCC allows as extensions for compatibility with legacy code. For example,
5905 the following @code{new} expression is not diagnosed at this level even
5906 though it has undefined behavior according to the C++ standard because
5907 it writes past the end of the one-element array.
5909 struct S @{ int n, a[1]; @};
5910 S *s = (S *)malloc (sizeof *s + 31 * sizeof s->a[0]);
5911 new (s->a)int [32]();
5914 @item -Wplacement-new=2
5915 At this level, in addition to diagnosing all the same constructs as at level
5916 1, a diagnostic is also issued for placement new expressions that construct
5917 an object in the last member of structure whose type is an array of a single
5918 element and whose size is less than the size of the object being constructed.
5919 While the previous example would be diagnosed, the following construct makes
5920 use of the flexible member array extension to avoid the warning at level 2.
5922 struct S @{ int n, a[]; @};
5923 S *s = (S *)malloc (sizeof *s + 32 * sizeof s->a[0]);
5924 new (s->a)int [32]();
5929 @item -Wpointer-arith
5930 @opindex Wpointer-arith
5931 @opindex Wno-pointer-arith
5932 Warn about anything that depends on the ``size of'' a function type or
5933 of @code{void}. GNU C assigns these types a size of 1, for
5934 convenience in calculations with @code{void *} pointers and pointers
5935 to functions. In C++, warn also when an arithmetic operation involves
5936 @code{NULL}. This warning is also enabled by @option{-Wpedantic}.
5938 @item -Wpointer-compare
5939 @opindex Wpointer-compare
5940 @opindex Wno-pointer-compare
5941 Warn if a pointer is compared with a zero character constant. This usually
5942 means that the pointer was meant to be dereferenced. For example:
5945 const char *p = foo ();
5950 Note that the code above is invalid in C++11.
5952 This warning is enabled by default.
5955 @opindex Wtype-limits
5956 @opindex Wno-type-limits
5957 Warn if a comparison is always true or always false due to the limited
5958 range of the data type, but do not warn for constant expressions. For
5959 example, warn if an unsigned variable is compared against zero with
5960 @code{<} or @code{>=}. This warning is also enabled by
5963 @include cppwarnopts.texi
5965 @item -Wbad-function-cast @r{(C and Objective-C only)}
5966 @opindex Wbad-function-cast
5967 @opindex Wno-bad-function-cast
5968 Warn when a function call is cast to a non-matching type.
5969 For example, warn if a call to a function returning an integer type
5970 is cast to a pointer type.
5972 @item -Wc90-c99-compat @r{(C and Objective-C only)}
5973 @opindex Wc90-c99-compat
5974 @opindex Wno-c90-c99-compat
5975 Warn about features not present in ISO C90, but present in ISO C99.
5976 For instance, warn about use of variable length arrays, @code{long long}
5977 type, @code{bool} type, compound literals, designated initializers, and so
5978 on. This option is independent of the standards mode. Warnings are disabled
5979 in the expression that follows @code{__extension__}.
5981 @item -Wc99-c11-compat @r{(C and Objective-C only)}
5982 @opindex Wc99-c11-compat
5983 @opindex Wno-c99-c11-compat
5984 Warn about features not present in ISO C99, but present in ISO C11.
5985 For instance, warn about use of anonymous structures and unions,
5986 @code{_Atomic} type qualifier, @code{_Thread_local} storage-class specifier,
5987 @code{_Alignas} specifier, @code{Alignof} operator, @code{_Generic} keyword,
5988 and so on. This option is independent of the standards mode. Warnings are
5989 disabled in the expression that follows @code{__extension__}.
5991 @item -Wc++-compat @r{(C and Objective-C only)}
5992 @opindex Wc++-compat
5993 Warn about ISO C constructs that are outside of the common subset of
5994 ISO C and ISO C++, e.g.@: request for implicit conversion from
5995 @code{void *} to a pointer to non-@code{void} type.
5997 @item -Wc++11-compat @r{(C++ and Objective-C++ only)}
5998 @opindex Wc++11-compat
5999 Warn about C++ constructs whose meaning differs between ISO C++ 1998
6000 and ISO C++ 2011, e.g., identifiers in ISO C++ 1998 that are keywords
6001 in ISO C++ 2011. This warning turns on @option{-Wnarrowing} and is
6002 enabled by @option{-Wall}.
6004 @item -Wc++14-compat @r{(C++ and Objective-C++ only)}
6005 @opindex Wc++14-compat
6006 Warn about C++ constructs whose meaning differs between ISO C++ 2011
6007 and ISO C++ 2014. This warning is enabled by @option{-Wall}.
6009 @item -Wc++17-compat @r{(C++ and Objective-C++ only)}
6010 @opindex Wc++17-compat
6011 Warn about C++ constructs whose meaning differs between ISO C++ 2014
6012 and ISO C++ 2017. This warning is enabled by @option{-Wall}.
6016 @opindex Wno-cast-qual
6017 Warn whenever a pointer is cast so as to remove a type qualifier from
6018 the target type. For example, warn if a @code{const char *} is cast
6019 to an ordinary @code{char *}.
6021 Also warn when making a cast that introduces a type qualifier in an
6022 unsafe way. For example, casting @code{char **} to @code{const char **}
6023 is unsafe, as in this example:
6026 /* p is char ** value. */
6027 const char **q = (const char **) p;
6028 /* Assignment of readonly string to const char * is OK. */
6030 /* Now char** pointer points to read-only memory. */
6035 @opindex Wcast-align
6036 @opindex Wno-cast-align
6037 Warn whenever a pointer is cast such that the required alignment of the
6038 target is increased. For example, warn if a @code{char *} is cast to
6039 an @code{int *} on machines where integers can only be accessed at
6040 two- or four-byte boundaries.
6042 @item -Wcast-align=strict
6043 @opindex Wcast-align=strict
6044 Warn whenever a pointer is cast such that the required alignment of the
6045 target is increased. For example, warn if a @code{char *} is cast to
6046 an @code{int *} regardless of the target machine.
6048 @item -Wwrite-strings
6049 @opindex Wwrite-strings
6050 @opindex Wno-write-strings
6051 When compiling C, give string constants the type @code{const
6052 char[@var{length}]} so that copying the address of one into a
6053 non-@code{const} @code{char *} pointer produces a warning. These
6054 warnings help you find at compile time code that can try to write
6055 into a string constant, but only if you have been very careful about
6056 using @code{const} in declarations and prototypes. Otherwise, it is
6057 just a nuisance. This is why we did not make @option{-Wall} request
6060 When compiling C++, warn about the deprecated conversion from string
6061 literals to @code{char *}. This warning is enabled by default for C++
6065 @itemx -Wcatch-value=@var{n} @r{(C++ and Objective-C++ only)}
6066 @opindex Wcatch-value
6067 @opindex Wno-catch-value
6068 Warn about catch handlers that do not catch via reference.
6069 With @option{-Wcatch-value=1} (or @option{-Wcatch-value} for short)
6070 warn about polymorphic class types that are caught by value.
6071 With @option{-Wcatch-value=2} warn about all class types that are caught
6072 by value. With @option{-Wcatch-value=3} warn about all types that are
6073 not caught by reference. @option{-Wcatch-value} is enabled by @option{-Wall}.
6077 @opindex Wno-clobbered
6078 Warn for variables that might be changed by @code{longjmp} or
6079 @code{vfork}. This warning is also enabled by @option{-Wextra}.
6081 @item -Wconditionally-supported @r{(C++ and Objective-C++ only)}
6082 @opindex Wconditionally-supported
6083 @opindex Wno-conditionally-supported
6084 Warn for conditionally-supported (C++11 [intro.defs]) constructs.
6087 @opindex Wconversion
6088 @opindex Wno-conversion
6089 Warn for implicit conversions that may alter a value. This includes
6090 conversions between real and integer, like @code{abs (x)} when
6091 @code{x} is @code{double}; conversions between signed and unsigned,
6092 like @code{unsigned ui = -1}; and conversions to smaller types, like
6093 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
6094 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
6095 changed by the conversion like in @code{abs (2.0)}. Warnings about
6096 conversions between signed and unsigned integers can be disabled by
6097 using @option{-Wno-sign-conversion}.
6099 For C++, also warn for confusing overload resolution for user-defined
6100 conversions; and conversions that never use a type conversion
6101 operator: conversions to @code{void}, the same type, a base class or a
6102 reference to them. Warnings about conversions between signed and
6103 unsigned integers are disabled by default in C++ unless
6104 @option{-Wsign-conversion} is explicitly enabled.
6106 @item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
6107 @opindex Wconversion-null
6108 @opindex Wno-conversion-null
6109 Do not warn for conversions between @code{NULL} and non-pointer
6110 types. @option{-Wconversion-null} is enabled by default.
6112 @item -Wzero-as-null-pointer-constant @r{(C++ and Objective-C++ only)}
6113 @opindex Wzero-as-null-pointer-constant
6114 @opindex Wno-zero-as-null-pointer-constant
6115 Warn when a literal @samp{0} is used as null pointer constant. This can
6116 be useful to facilitate the conversion to @code{nullptr} in C++11.
6118 @item -Wsubobject-linkage @r{(C++ and Objective-C++ only)}
6119 @opindex Wsubobject-linkage
6120 @opindex Wno-subobject-linkage
6121 Warn if a class type has a base or a field whose type uses the anonymous
6122 namespace or depends on a type with no linkage. If a type A depends on
6123 a type B with no or internal linkage, defining it in multiple
6124 translation units would be an ODR violation because the meaning of B
6125 is different in each translation unit. If A only appears in a single
6126 translation unit, the best way to silence the warning is to give it
6127 internal linkage by putting it in an anonymous namespace as well. The
6128 compiler doesn't give this warning for types defined in the main .C
6129 file, as those are unlikely to have multiple definitions.
6130 @option{-Wsubobject-linkage} is enabled by default.
6132 @item -Wdangling-else
6133 @opindex Wdangling-else
6134 @opindex Wno-dangling-else
6135 Warn about constructions where there may be confusion to which
6136 @code{if} statement an @code{else} branch belongs. Here is an example of
6151 In C/C++, every @code{else} branch belongs to the innermost possible
6152 @code{if} statement, which in this example is @code{if (b)}. This is
6153 often not what the programmer expected, as illustrated in the above
6154 example by indentation the programmer chose. When there is the
6155 potential for this confusion, GCC issues a warning when this flag
6156 is specified. To eliminate the warning, add explicit braces around
6157 the innermost @code{if} statement so there is no way the @code{else}
6158 can belong to the enclosing @code{if}. The resulting code
6175 This warning is enabled by @option{-Wparentheses}.
6179 @opindex Wno-date-time
6180 Warn when macros @code{__TIME__}, @code{__DATE__} or @code{__TIMESTAMP__}
6181 are encountered as they might prevent bit-wise-identical reproducible
6184 @item -Wdelete-incomplete @r{(C++ and Objective-C++ only)}
6185 @opindex Wdelete-incomplete
6186 @opindex Wno-delete-incomplete
6187 Warn when deleting a pointer to incomplete type, which may cause
6188 undefined behavior at runtime. This warning is enabled by default.
6190 @item -Wuseless-cast @r{(C++ and Objective-C++ only)}
6191 @opindex Wuseless-cast
6192 @opindex Wno-useless-cast
6193 Warn when an expression is casted to its own type.
6196 @opindex Wempty-body
6197 @opindex Wno-empty-body
6198 Warn if an empty body occurs in an @code{if}, @code{else} or @code{do
6199 while} statement. This warning is also enabled by @option{-Wextra}.
6201 @item -Wenum-compare
6202 @opindex Wenum-compare
6203 @opindex Wno-enum-compare
6204 Warn about a comparison between values of different enumerated types.
6205 In C++ enumerated type mismatches in conditional expressions are also
6206 diagnosed and the warning is enabled by default. In C this warning is
6207 enabled by @option{-Wall}.
6209 @item -Wextra-semi @r{(C++, Objective-C++ only)}
6210 @opindex Wextra-semi
6211 @opindex Wno-extra-semi
6212 Warn about redundant semicolon after in-class function definition.
6214 @item -Wjump-misses-init @r{(C, Objective-C only)}
6215 @opindex Wjump-misses-init
6216 @opindex Wno-jump-misses-init
6217 Warn if a @code{goto} statement or a @code{switch} statement jumps
6218 forward across the initialization of a variable, or jumps backward to a
6219 label after the variable has been initialized. This only warns about
6220 variables that are initialized when they are declared. This warning is
6221 only supported for C and Objective-C; in C++ this sort of branch is an
6224 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
6225 can be disabled with the @option{-Wno-jump-misses-init} option.
6227 @item -Wsign-compare
6228 @opindex Wsign-compare
6229 @opindex Wno-sign-compare
6230 @cindex warning for comparison of signed and unsigned values
6231 @cindex comparison of signed and unsigned values, warning
6232 @cindex signed and unsigned values, comparison warning
6233 Warn when a comparison between signed and unsigned values could produce
6234 an incorrect result when the signed value is converted to unsigned.
6235 In C++, this warning is also enabled by @option{-Wall}. In C, it is
6236 also enabled by @option{-Wextra}.
6238 @item -Wsign-conversion
6239 @opindex Wsign-conversion
6240 @opindex Wno-sign-conversion
6241 Warn for implicit conversions that may change the sign of an integer
6242 value, like assigning a signed integer expression to an unsigned
6243 integer variable. An explicit cast silences the warning. In C, this
6244 option is enabled also by @option{-Wconversion}.
6246 @item -Wfloat-conversion
6247 @opindex Wfloat-conversion
6248 @opindex Wno-float-conversion
6249 Warn for implicit conversions that reduce the precision of a real value.
6250 This includes conversions from real to integer, and from higher precision
6251 real to lower precision real values. This option is also enabled by
6252 @option{-Wconversion}.
6254 @item -Wno-scalar-storage-order
6255 @opindex -Wno-scalar-storage-order
6256 @opindex -Wscalar-storage-order
6257 Do not warn on suspicious constructs involving reverse scalar storage order.
6259 @item -Wsized-deallocation @r{(C++ and Objective-C++ only)}
6260 @opindex Wsized-deallocation
6261 @opindex Wno-sized-deallocation
6262 Warn about a definition of an unsized deallocation function
6264 void operator delete (void *) noexcept;
6265 void operator delete[] (void *) noexcept;
6267 without a definition of the corresponding sized deallocation function
6269 void operator delete (void *, std::size_t) noexcept;
6270 void operator delete[] (void *, std::size_t) noexcept;
6272 or vice versa. Enabled by @option{-Wextra} along with
6273 @option{-fsized-deallocation}.
6275 @item -Wsizeof-pointer-div
6276 @opindex Wsizeof-pointer-div
6277 @opindex Wno-sizeof-pointer-div
6278 Warn for suspicious divisions of two sizeof expressions that divide
6279 the pointer size by the element size, which is the usual way to compute
6280 the array size but won't work out correctly with pointers. This warning
6281 warns e.g.@: about @code{sizeof (ptr) / sizeof (ptr[0])} if @code{ptr} is
6282 not an array, but a pointer. This warning is enabled by @option{-Wall}.
6284 @item -Wsizeof-pointer-memaccess
6285 @opindex Wsizeof-pointer-memaccess
6286 @opindex Wno-sizeof-pointer-memaccess
6287 Warn for suspicious length parameters to certain string and memory built-in
6288 functions if the argument uses @code{sizeof}. This warning triggers for
6289 example for @code{memset (ptr, 0, sizeof (ptr));} if @code{ptr} is not
6290 an array, but a pointer, and suggests a possible fix, or about
6291 @code{memcpy (&foo, ptr, sizeof (&foo));}. @option{-Wsizeof-pointer-memaccess}
6292 also warns about calls to bounded string copy functions like @code{strncat}
6293 or @code{strncpy} that specify as the bound a @code{sizeof} expression of
6294 the source array. For example, in the following function the call to
6295 @code{strncat} specifies the size of the source string as the bound. That
6296 is almost certainly a mistake and so the call is diagnosed.
6298 void make_file (const char *name)
6300 char path[PATH_MAX];
6301 strncpy (path, name, sizeof path - 1);
6302 strncat (path, ".text", sizeof ".text");
6307 The @option{-Wsizeof-pointer-memaccess} option is enabled by @option{-Wall}.
6309 @item -Wsizeof-array-argument
6310 @opindex Wsizeof-array-argument
6311 @opindex Wno-sizeof-array-argument
6312 Warn when the @code{sizeof} operator is applied to a parameter that is
6313 declared as an array in a function definition. This warning is enabled by
6314 default for C and C++ programs.
6316 @item -Wmemset-elt-size
6317 @opindex Wmemset-elt-size
6318 @opindex Wno-memset-elt-size
6319 Warn for suspicious calls to the @code{memset} built-in function, if the
6320 first argument references an array, and the third argument is a number
6321 equal to the number of elements, but not equal to the size of the array
6322 in memory. This indicates that the user has omitted a multiplication by
6323 the element size. This warning is enabled by @option{-Wall}.
6325 @item -Wmemset-transposed-args
6326 @opindex Wmemset-transposed-args
6327 @opindex Wno-memset-transposed-args
6328 Warn for suspicious calls to the @code{memset} built-in function, if the
6329 second argument is not zero and the third argument is zero. This warns e.g.@
6330 about @code{memset (buf, sizeof buf, 0)} where most probably
6331 @code{memset (buf, 0, sizeof buf)} was meant instead. The diagnostics
6332 is only emitted if the third argument is literal zero. If it is some
6333 expression that is folded to zero, a cast of zero to some type, etc.,
6334 it is far less likely that the user has mistakenly exchanged the arguments
6335 and no warning is emitted. This warning is enabled by @option{-Wall}.
6339 @opindex Wno-address
6340 Warn about suspicious uses of memory addresses. These include using
6341 the address of a function in a conditional expression, such as
6342 @code{void func(void); if (func)}, and comparisons against the memory
6343 address of a string literal, such as @code{if (x == "abc")}. Such
6344 uses typically indicate a programmer error: the address of a function
6345 always evaluates to true, so their use in a conditional usually
6346 indicate that the programmer forgot the parentheses in a function
6347 call; and comparisons against string literals result in unspecified
6348 behavior and are not portable in C, so they usually indicate that the
6349 programmer intended to use @code{strcmp}. This warning is enabled by
6353 @opindex Wlogical-op
6354 @opindex Wno-logical-op
6355 Warn about suspicious uses of logical operators in expressions.
6356 This includes using logical operators in contexts where a
6357 bit-wise operator is likely to be expected. Also warns when
6358 the operands of a logical operator are the same:
6361 if (a < 0 && a < 0) @{ @dots{} @}
6364 @item -Wlogical-not-parentheses
6365 @opindex Wlogical-not-parentheses
6366 @opindex Wno-logical-not-parentheses
6367 Warn about logical not used on the left hand side operand of a comparison.
6368 This option does not warn if the right operand is considered to be a boolean
6369 expression. Its purpose is to detect suspicious code like the following:
6373 if (!a > 1) @{ @dots{} @}
6376 It is possible to suppress the warning by wrapping the LHS into
6379 if ((!a) > 1) @{ @dots{} @}
6382 This warning is enabled by @option{-Wall}.
6384 @item -Waggregate-return
6385 @opindex Waggregate-return
6386 @opindex Wno-aggregate-return
6387 Warn if any functions that return structures or unions are defined or
6388 called. (In languages where you can return an array, this also elicits
6391 @item -Wno-aggressive-loop-optimizations
6392 @opindex Wno-aggressive-loop-optimizations
6393 @opindex Waggressive-loop-optimizations
6394 Warn if in a loop with constant number of iterations the compiler detects
6395 undefined behavior in some statement during one or more of the iterations.
6397 @item -Wno-attributes
6398 @opindex Wno-attributes
6399 @opindex Wattributes
6400 Do not warn if an unexpected @code{__attribute__} is used, such as
6401 unrecognized attributes, function attributes applied to variables,
6402 etc. This does not stop errors for incorrect use of supported
6405 @item -Wno-builtin-declaration-mismatch
6406 @opindex Wno-builtin-declaration-mismatch
6407 @opindex Wbuiltin-declaration-mismatch
6408 Warn if a built-in function is declared with the wrong signature or
6410 This warning is enabled by default.
6412 @item -Wno-builtin-macro-redefined
6413 @opindex Wno-builtin-macro-redefined
6414 @opindex Wbuiltin-macro-redefined
6415 Do not warn if certain built-in macros are redefined. This suppresses
6416 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
6417 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
6419 @item -Wstrict-prototypes @r{(C and Objective-C only)}
6420 @opindex Wstrict-prototypes
6421 @opindex Wno-strict-prototypes
6422 Warn if a function is declared or defined without specifying the
6423 argument types. (An old-style function definition is permitted without
6424 a warning if preceded by a declaration that specifies the argument
6427 @item -Wold-style-declaration @r{(C and Objective-C only)}
6428 @opindex Wold-style-declaration
6429 @opindex Wno-old-style-declaration
6430 Warn for obsolescent usages, according to the C Standard, in a
6431 declaration. For example, warn if storage-class specifiers like
6432 @code{static} are not the first things in a declaration. This warning
6433 is also enabled by @option{-Wextra}.
6435 @item -Wold-style-definition @r{(C and Objective-C only)}
6436 @opindex Wold-style-definition
6437 @opindex Wno-old-style-definition
6438 Warn if an old-style function definition is used. A warning is given
6439 even if there is a previous prototype.
6441 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
6442 @opindex Wmissing-parameter-type
6443 @opindex Wno-missing-parameter-type
6444 A function parameter is declared without a type specifier in K&R-style
6451 This warning is also enabled by @option{-Wextra}.
6453 @item -Wmissing-prototypes @r{(C and Objective-C only)}
6454 @opindex Wmissing-prototypes
6455 @opindex Wno-missing-prototypes
6456 Warn if a global function is defined without a previous prototype
6457 declaration. This warning is issued even if the definition itself
6458 provides a prototype. Use this option to detect global functions
6459 that do not have a matching prototype declaration in a header file.
6460 This option is not valid for C++ because all function declarations
6461 provide prototypes and a non-matching declaration declares an
6462 overload rather than conflict with an earlier declaration.
6463 Use @option{-Wmissing-declarations} to detect missing declarations in C++.
6465 @item -Wmissing-declarations
6466 @opindex Wmissing-declarations
6467 @opindex Wno-missing-declarations
6468 Warn if a global function is defined without a previous declaration.
6469 Do so even if the definition itself provides a prototype.
6470 Use this option to detect global functions that are not declared in
6471 header files. In C, no warnings are issued for functions with previous
6472 non-prototype declarations; use @option{-Wmissing-prototypes} to detect
6473 missing prototypes. In C++, no warnings are issued for function templates,
6474 or for inline functions, or for functions in anonymous namespaces.
6476 @item -Wmissing-field-initializers
6477 @opindex Wmissing-field-initializers
6478 @opindex Wno-missing-field-initializers
6482 Warn if a structure's initializer has some fields missing. For
6483 example, the following code causes such a warning, because
6484 @code{x.h} is implicitly zero:
6487 struct s @{ int f, g, h; @};
6488 struct s x = @{ 3, 4 @};
6491 This option does not warn about designated initializers, so the following
6492 modification does not trigger a warning:
6495 struct s @{ int f, g, h; @};
6496 struct s x = @{ .f = 3, .g = 4 @};
6499 In C this option does not warn about the universal zero initializer
6503 struct s @{ int f, g, h; @};
6504 struct s x = @{ 0 @};
6507 Likewise, in C++ this option does not warn about the empty @{ @}
6508 initializer, for example:
6511 struct s @{ int f, g, h; @};
6515 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
6516 warnings without this one, use @option{-Wextra -Wno-missing-field-initializers}.
6518 @item -Wno-multichar
6519 @opindex Wno-multichar
6521 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
6522 Usually they indicate a typo in the user's code, as they have
6523 implementation-defined values, and should not be used in portable code.
6525 @item -Wnormalized=@r{[}none@r{|}id@r{|}nfc@r{|}nfkc@r{]}
6526 @opindex Wnormalized=
6527 @opindex Wnormalized
6528 @opindex Wno-normalized
6531 @cindex character set, input normalization
6532 In ISO C and ISO C++, two identifiers are different if they are
6533 different sequences of characters. However, sometimes when characters
6534 outside the basic ASCII character set are used, you can have two
6535 different character sequences that look the same. To avoid confusion,
6536 the ISO 10646 standard sets out some @dfn{normalization rules} which
6537 when applied ensure that two sequences that look the same are turned into
6538 the same sequence. GCC can warn you if you are using identifiers that
6539 have not been normalized; this option controls that warning.
6541 There are four levels of warning supported by GCC@. The default is
6542 @option{-Wnormalized=nfc}, which warns about any identifier that is
6543 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
6544 recommended form for most uses. It is equivalent to
6545 @option{-Wnormalized}.
6547 Unfortunately, there are some characters allowed in identifiers by
6548 ISO C and ISO C++ that, when turned into NFC, are not allowed in
6549 identifiers. That is, there's no way to use these symbols in portable
6550 ISO C or C++ and have all your identifiers in NFC@.
6551 @option{-Wnormalized=id} suppresses the warning for these characters.
6552 It is hoped that future versions of the standards involved will correct
6553 this, which is why this option is not the default.
6555 You can switch the warning off for all characters by writing
6556 @option{-Wnormalized=none} or @option{-Wno-normalized}. You should
6557 only do this if you are using some other normalization scheme (like
6558 ``D''), because otherwise you can easily create bugs that are
6559 literally impossible to see.
6561 Some characters in ISO 10646 have distinct meanings but look identical
6562 in some fonts or display methodologies, especially once formatting has
6563 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
6564 LETTER N'', displays just like a regular @code{n} that has been
6565 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
6566 normalization scheme to convert all these into a standard form as
6567 well, and GCC warns if your code is not in NFKC if you use
6568 @option{-Wnormalized=nfkc}. This warning is comparable to warning
6569 about every identifier that contains the letter O because it might be
6570 confused with the digit 0, and so is not the default, but may be
6571 useful as a local coding convention if the programming environment
6572 cannot be fixed to display these characters distinctly.
6574 @item -Wno-deprecated
6575 @opindex Wno-deprecated
6576 @opindex Wdeprecated
6577 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
6579 @item -Wno-deprecated-declarations
6580 @opindex Wno-deprecated-declarations
6581 @opindex Wdeprecated-declarations
6582 Do not warn about uses of functions (@pxref{Function Attributes}),
6583 variables (@pxref{Variable Attributes}), and types (@pxref{Type
6584 Attributes}) marked as deprecated by using the @code{deprecated}
6588 @opindex Wno-overflow
6590 Do not warn about compile-time overflow in constant expressions.
6595 Warn about One Definition Rule violations during link-time optimization.
6596 Requires @option{-flto-odr-type-merging} to be enabled. Enabled by default.
6599 @opindex Wopenm-simd
6600 Warn if the vectorizer cost model overrides the OpenMP or the Cilk Plus
6601 simd directive set by user. The @option{-fsimd-cost-model=unlimited}
6602 option can be used to relax the cost model.
6604 @item -Woverride-init @r{(C and Objective-C only)}
6605 @opindex Woverride-init
6606 @opindex Wno-override-init
6610 Warn if an initialized field without side effects is overridden when
6611 using designated initializers (@pxref{Designated Inits, , Designated
6614 This warning is included in @option{-Wextra}. To get other
6615 @option{-Wextra} warnings without this one, use @option{-Wextra
6616 -Wno-override-init}.
6618 @item -Woverride-init-side-effects @r{(C and Objective-C only)}
6619 @opindex Woverride-init-side-effects
6620 @opindex Wno-override-init-side-effects
6621 Warn if an initialized field with side effects is overridden when
6622 using designated initializers (@pxref{Designated Inits, , Designated
6623 Initializers}). This warning is enabled by default.
6628 Warn if a structure is given the packed attribute, but the packed
6629 attribute has no effect on the layout or size of the structure.
6630 Such structures may be mis-aligned for little benefit. For
6631 instance, in this code, the variable @code{f.x} in @code{struct bar}
6632 is misaligned even though @code{struct bar} does not itself
6633 have the packed attribute:
6640 @} __attribute__((packed));
6648 @item -Wpacked-bitfield-compat
6649 @opindex Wpacked-bitfield-compat
6650 @opindex Wno-packed-bitfield-compat
6651 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
6652 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
6653 the change can lead to differences in the structure layout. GCC
6654 informs you when the offset of such a field has changed in GCC 4.4.
6655 For example there is no longer a 4-bit padding between field @code{a}
6656 and @code{b} in this structure:
6663 @} __attribute__ ((packed));
6666 This warning is enabled by default. Use
6667 @option{-Wno-packed-bitfield-compat} to disable this warning.
6669 @item -Wpacked-not-aligned @r{(C, C++, Objective-C and Objective-C++ only)}
6670 @opindex Wpacked-not-aligned
6671 @opindex Wno-packed-not-aligned
6672 Warn if a structure field with explicitly specified alignment in a
6673 packed struct or union is misaligned. For example, a warning will
6674 be issued on @code{struct S}, like, @code{warning: alignment 1 of
6675 'struct S' is less than 8}, in this code:
6679 struct __attribute__ ((aligned (8))) S8 @{ char a[8]; @};
6680 struct __attribute__ ((packed)) S @{
6686 This warning is enabled by @option{-Wall}.
6691 Warn if padding is included in a structure, either to align an element
6692 of the structure or to align the whole structure. Sometimes when this
6693 happens it is possible to rearrange the fields of the structure to
6694 reduce the padding and so make the structure smaller.
6696 @item -Wredundant-decls
6697 @opindex Wredundant-decls
6698 @opindex Wno-redundant-decls
6699 Warn if anything is declared more than once in the same scope, even in
6700 cases where multiple declaration is valid and changes nothing.
6704 @opindex Wno-restrict
6705 Warn when an argument passed to a restrict-qualified parameter
6706 aliases with another argument.
6708 @item -Wnested-externs @r{(C and Objective-C only)}
6709 @opindex Wnested-externs
6710 @opindex Wno-nested-externs
6711 Warn if an @code{extern} declaration is encountered within a function.
6713 @item -Wno-inherited-variadic-ctor
6714 @opindex Winherited-variadic-ctor
6715 @opindex Wno-inherited-variadic-ctor
6716 Suppress warnings about use of C++11 inheriting constructors when the
6717 base class inherited from has a C variadic constructor; the warning is
6718 on by default because the ellipsis is not inherited.
6723 Warn if a function that is declared as inline cannot be inlined.
6724 Even with this option, the compiler does not warn about failures to
6725 inline functions declared in system headers.
6727 The compiler uses a variety of heuristics to determine whether or not
6728 to inline a function. For example, the compiler takes into account
6729 the size of the function being inlined and the amount of inlining
6730 that has already been done in the current function. Therefore,
6731 seemingly insignificant changes in the source program can cause the
6732 warnings produced by @option{-Winline} to appear or disappear.
6734 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
6735 @opindex Wno-invalid-offsetof
6736 @opindex Winvalid-offsetof
6737 Suppress warnings from applying the @code{offsetof} macro to a non-POD
6738 type. According to the 2014 ISO C++ standard, applying @code{offsetof}
6739 to a non-standard-layout type is undefined. In existing C++ implementations,
6740 however, @code{offsetof} typically gives meaningful results.
6741 This flag is for users who are aware that they are
6742 writing nonportable code and who have deliberately chosen to ignore the
6745 The restrictions on @code{offsetof} may be relaxed in a future version
6746 of the C++ standard.
6748 @item -Wint-in-bool-context
6749 @opindex Wint-in-bool-context
6750 @opindex Wno-int-in-bool-context
6751 Warn for suspicious use of integer values where boolean values are expected,
6752 such as conditional expressions (?:) using non-boolean integer constants in
6753 boolean context, like @code{if (a <= b ? 2 : 3)}. Or left shifting of signed
6754 integers in boolean context, like @code{for (a = 0; 1 << a; a++);}. Likewise
6755 for all kinds of multiplications regardless of the data type.
6756 This warning is enabled by @option{-Wall}.
6758 @item -Wno-int-to-pointer-cast
6759 @opindex Wno-int-to-pointer-cast
6760 @opindex Wint-to-pointer-cast
6761 Suppress warnings from casts to pointer type of an integer of a
6762 different size. In C++, casting to a pointer type of smaller size is
6763 an error. @option{Wint-to-pointer-cast} is enabled by default.
6766 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
6767 @opindex Wno-pointer-to-int-cast
6768 @opindex Wpointer-to-int-cast
6769 Suppress warnings from casts from a pointer to an integer type of a
6773 @opindex Winvalid-pch
6774 @opindex Wno-invalid-pch
6775 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
6776 the search path but cannot be used.
6780 @opindex Wno-long-long
6781 Warn if @code{long long} type is used. This is enabled by either
6782 @option{-Wpedantic} or @option{-Wtraditional} in ISO C90 and C++98
6783 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
6785 @item -Wvariadic-macros
6786 @opindex Wvariadic-macros
6787 @opindex Wno-variadic-macros
6788 Warn if variadic macros are used in ISO C90 mode, or if the GNU
6789 alternate syntax is used in ISO C99 mode. This is enabled by either
6790 @option{-Wpedantic} or @option{-Wtraditional}. To inhibit the warning
6791 messages, use @option{-Wno-variadic-macros}.
6795 @opindex Wno-varargs
6796 Warn upon questionable usage of the macros used to handle variable
6797 arguments like @code{va_start}. This is default. To inhibit the
6798 warning messages, use @option{-Wno-varargs}.
6800 @item -Wvector-operation-performance
6801 @opindex Wvector-operation-performance
6802 @opindex Wno-vector-operation-performance
6803 Warn if vector operation is not implemented via SIMD capabilities of the
6804 architecture. Mainly useful for the performance tuning.
6805 Vector operation can be implemented @code{piecewise}, which means that the
6806 scalar operation is performed on every vector element;
6807 @code{in parallel}, which means that the vector operation is implemented
6808 using scalars of wider type, which normally is more performance efficient;
6809 and @code{as a single scalar}, which means that vector fits into a
6812 @item -Wno-virtual-move-assign
6813 @opindex Wvirtual-move-assign
6814 @opindex Wno-virtual-move-assign
6815 Suppress warnings about inheriting from a virtual base with a
6816 non-trivial C++11 move assignment operator. This is dangerous because
6817 if the virtual base is reachable along more than one path, it is
6818 moved multiple times, which can mean both objects end up in the
6819 moved-from state. If the move assignment operator is written to avoid
6820 moving from a moved-from object, this warning can be disabled.
6825 Warn if a variable-length array is used in the code.
6826 @option{-Wno-vla} prevents the @option{-Wpedantic} warning of
6827 the variable-length array.
6829 @item -Wvla-larger-than=@var{n}
6830 If this option is used, the compiler will warn on uses of
6831 variable-length arrays where the size is either unbounded, or bounded
6832 by an argument that can be larger than @var{n} bytes. This is similar
6833 to how @option{-Walloca-larger-than=@var{n}} works, but with
6834 variable-length arrays.
6836 Note that GCC may optimize small variable-length arrays of a known
6837 value into plain arrays, so this warning may not get triggered for
6840 This warning is not enabled by @option{-Wall}, and is only active when
6841 @option{-ftree-vrp} is active (default for @option{-O2} and above).
6843 See also @option{-Walloca-larger-than=@var{n}}.
6845 @item -Wvolatile-register-var
6846 @opindex Wvolatile-register-var
6847 @opindex Wno-volatile-register-var
6848 Warn if a register variable is declared volatile. The volatile
6849 modifier does not inhibit all optimizations that may eliminate reads
6850 and/or writes to register variables. This warning is enabled by
6853 @item -Wdisabled-optimization
6854 @opindex Wdisabled-optimization
6855 @opindex Wno-disabled-optimization
6856 Warn if a requested optimization pass is disabled. This warning does
6857 not generally indicate that there is anything wrong with your code; it
6858 merely indicates that GCC's optimizers are unable to handle the code
6859 effectively. Often, the problem is that your code is too big or too
6860 complex; GCC refuses to optimize programs when the optimization
6861 itself is likely to take inordinate amounts of time.
6863 @item -Wpointer-sign @r{(C and Objective-C only)}
6864 @opindex Wpointer-sign
6865 @opindex Wno-pointer-sign
6866 Warn for pointer argument passing or assignment with different signedness.
6867 This option is only supported for C and Objective-C@. It is implied by
6868 @option{-Wall} and by @option{-Wpedantic}, which can be disabled with
6869 @option{-Wno-pointer-sign}.
6871 @item -Wstack-protector
6872 @opindex Wstack-protector
6873 @opindex Wno-stack-protector
6874 This option is only active when @option{-fstack-protector} is active. It
6875 warns about functions that are not protected against stack smashing.
6877 @item -Woverlength-strings
6878 @opindex Woverlength-strings
6879 @opindex Wno-overlength-strings
6880 Warn about string constants that are longer than the ``minimum
6881 maximum'' length specified in the C standard. Modern compilers
6882 generally allow string constants that are much longer than the
6883 standard's minimum limit, but very portable programs should avoid
6884 using longer strings.
6886 The limit applies @emph{after} string constant concatenation, and does
6887 not count the trailing NUL@. In C90, the limit was 509 characters; in
6888 C99, it was raised to 4095. C++98 does not specify a normative
6889 minimum maximum, so we do not diagnose overlength strings in C++@.
6891 This option is implied by @option{-Wpedantic}, and can be disabled with
6892 @option{-Wno-overlength-strings}.
6894 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
6895 @opindex Wunsuffixed-float-constants
6897 Issue a warning for any floating constant that does not have
6898 a suffix. When used together with @option{-Wsystem-headers} it
6899 warns about such constants in system header files. This can be useful
6900 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
6901 from the decimal floating-point extension to C99.
6903 @item -Wno-designated-init @r{(C and Objective-C only)}
6904 Suppress warnings when a positional initializer is used to initialize
6905 a structure that has been marked with the @code{designated_init}
6909 Issue a warning when HSAIL cannot be emitted for the compiled function or
6914 @node Debugging Options
6915 @section Options for Debugging Your Program
6916 @cindex options, debugging
6917 @cindex debugging information options
6919 To tell GCC to emit extra information for use by a debugger, in almost
6920 all cases you need only to add @option{-g} to your other options.
6922 GCC allows you to use @option{-g} with
6923 @option{-O}. The shortcuts taken by optimized code may occasionally
6924 be surprising: some variables you declared may not exist
6925 at all; flow of control may briefly move where you did not expect it;
6926 some statements may not be executed because they compute constant
6927 results or their values are already at hand; some statements may
6928 execute in different places because they have been moved out of loops.
6929 Nevertheless it is possible to debug optimized output. This makes
6930 it reasonable to use the optimizer for programs that might have bugs.
6932 If you are not using some other optimization option, consider
6933 using @option{-Og} (@pxref{Optimize Options}) with @option{-g}.
6934 With no @option{-O} option at all, some compiler passes that collect
6935 information useful for debugging do not run at all, so that
6936 @option{-Og} may result in a better debugging experience.
6941 Produce debugging information in the operating system's native format
6942 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
6945 On most systems that use stabs format, @option{-g} enables use of extra
6946 debugging information that only GDB can use; this extra information
6947 makes debugging work better in GDB but probably makes other debuggers
6949 refuse to read the program. If you want to control for certain whether
6950 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
6951 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
6955 Produce debugging information for use by GDB@. This means to use the
6956 most expressive format available (DWARF, stabs, or the native format
6957 if neither of those are supported), including GDB extensions if at all
6961 @itemx -gdwarf-@var{version}
6963 Produce debugging information in DWARF format (if that is supported).
6964 The value of @var{version} may be either 2, 3, 4 or 5; the default version
6965 for most targets is 4. DWARF Version 5 is only experimental.
6967 Note that with DWARF Version 2, some ports require and always
6968 use some non-conflicting DWARF 3 extensions in the unwind tables.
6970 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
6971 for maximum benefit.
6973 GCC no longer supports DWARF Version 1, which is substantially
6974 different than Version 2 and later. For historical reasons, some
6975 other DWARF-related options such as
6976 @option{-fno-dwarf2-cfi-asm}) retain a reference to DWARF Version 2
6977 in their names, but apply to all currently-supported versions of DWARF.
6981 Produce debugging information in stabs format (if that is supported),
6982 without GDB extensions. This is the format used by DBX on most BSD
6983 systems. On MIPS, Alpha and System V Release 4 systems this option
6984 produces stabs debugging output that is not understood by DBX@.
6985 On System V Release 4 systems this option requires the GNU assembler.
6989 Produce debugging information in stabs format (if that is supported),
6990 using GNU extensions understood only by the GNU debugger (GDB)@. The
6991 use of these extensions is likely to make other debuggers crash or
6992 refuse to read the program.
6996 Produce debugging information in XCOFF format (if that is supported).
6997 This is the format used by the DBX debugger on IBM RS/6000 systems.
7001 Produce debugging information in XCOFF format (if that is supported),
7002 using GNU extensions understood only by the GNU debugger (GDB)@. The
7003 use of these extensions is likely to make other debuggers crash or
7004 refuse to read the program, and may cause assemblers other than the GNU
7005 assembler (GAS) to fail with an error.
7009 Produce debugging information in Alpha/VMS debug format (if that is
7010 supported). This is the format used by DEBUG on Alpha/VMS systems.
7013 @itemx -ggdb@var{level}
7014 @itemx -gstabs@var{level}
7015 @itemx -gxcoff@var{level}
7016 @itemx -gvms@var{level}
7017 Request debugging information and also use @var{level} to specify how
7018 much information. The default level is 2.
7020 Level 0 produces no debug information at all. Thus, @option{-g0} negates
7023 Level 1 produces minimal information, enough for making backtraces in
7024 parts of the program that you don't plan to debug. This includes
7025 descriptions of functions and external variables, and line number
7026 tables, but no information about local variables.
7028 Level 3 includes extra information, such as all the macro definitions
7029 present in the program. Some debuggers support macro expansion when
7030 you use @option{-g3}.
7032 @option{-gdwarf} does not accept a concatenated debug level, to avoid
7033 confusion with @option{-gdwarf-@var{level}}.
7034 Instead use an additional @option{-g@var{level}} option to change the
7035 debug level for DWARF.
7037 @item -feliminate-unused-debug-symbols
7038 @opindex feliminate-unused-debug-symbols
7039 Produce debugging information in stabs format (if that is supported),
7040 for only symbols that are actually used.
7042 @item -femit-class-debug-always
7043 @opindex femit-class-debug-always
7044 Instead of emitting debugging information for a C++ class in only one
7045 object file, emit it in all object files using the class. This option
7046 should be used only with debuggers that are unable to handle the way GCC
7047 normally emits debugging information for classes because using this
7048 option increases the size of debugging information by as much as a
7051 @item -fno-merge-debug-strings
7052 @opindex fmerge-debug-strings
7053 @opindex fno-merge-debug-strings
7054 Direct the linker to not merge together strings in the debugging
7055 information that are identical in different object files. Merging is
7056 not supported by all assemblers or linkers. Merging decreases the size
7057 of the debug information in the output file at the cost of increasing
7058 link processing time. Merging is enabled by default.
7060 @item -fdebug-prefix-map=@var{old}=@var{new}
7061 @opindex fdebug-prefix-map
7062 When compiling files in directory @file{@var{old}}, record debugging
7063 information describing them as in @file{@var{new}} instead. This can be
7064 used to replace a build-time path with an install-time path in the debug info.
7065 It can also be used to change an absolute path to a relative path by using
7066 @file{.} for @var{new}. This can give more reproducible builds, which are
7067 location independent, but may require an extra command to tell GDB where to
7068 find the source files.
7070 @item -fvar-tracking
7071 @opindex fvar-tracking
7072 Run variable tracking pass. It computes where variables are stored at each
7073 position in code. Better debugging information is then generated
7074 (if the debugging information format supports this information).
7076 It is enabled by default when compiling with optimization (@option{-Os},
7077 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
7078 the debug info format supports it.
7080 @item -fvar-tracking-assignments
7081 @opindex fvar-tracking-assignments
7082 @opindex fno-var-tracking-assignments
7083 Annotate assignments to user variables early in the compilation and
7084 attempt to carry the annotations over throughout the compilation all the
7085 way to the end, in an attempt to improve debug information while
7086 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
7088 It can be enabled even if var-tracking is disabled, in which case
7089 annotations are created and maintained, but discarded at the end.
7090 By default, this flag is enabled together with @option{-fvar-tracking},
7091 except when selective scheduling is enabled.
7094 @opindex gsplit-dwarf
7095 Separate as much DWARF debugging information as possible into a
7096 separate output file with the extension @file{.dwo}. This option allows
7097 the build system to avoid linking files with debug information. To
7098 be useful, this option requires a debugger capable of reading @file{.dwo}
7103 Generate DWARF @code{.debug_pubnames} and @code{.debug_pubtypes} sections.
7105 @item -ggnu-pubnames
7106 @opindex ggnu-pubnames
7107 Generate @code{.debug_pubnames} and @code{.debug_pubtypes} sections in a format
7108 suitable for conversion into a GDB@ index. This option is only useful
7109 with a linker that can produce GDB@ index version 7.
7111 @item -fdebug-types-section
7112 @opindex fdebug-types-section
7113 @opindex fno-debug-types-section
7114 When using DWARF Version 4 or higher, type DIEs can be put into
7115 their own @code{.debug_types} section instead of making them part of the
7116 @code{.debug_info} section. It is more efficient to put them in a separate
7117 comdat sections since the linker can then remove duplicates.
7118 But not all DWARF consumers support @code{.debug_types} sections yet
7119 and on some objects @code{.debug_types} produces larger instead of smaller
7120 debugging information.
7122 @item -grecord-gcc-switches
7123 @item -gno-record-gcc-switches
7124 @opindex grecord-gcc-switches
7125 @opindex gno-record-gcc-switches
7126 This switch causes the command-line options used to invoke the
7127 compiler that may affect code generation to be appended to the
7128 DW_AT_producer attribute in DWARF debugging information. The options
7129 are concatenated with spaces separating them from each other and from
7130 the compiler version.
7131 It is enabled by default.
7132 See also @option{-frecord-gcc-switches} for another
7133 way of storing compiler options into the object file.
7135 @item -gstrict-dwarf
7136 @opindex gstrict-dwarf
7137 Disallow using extensions of later DWARF standard version than selected
7138 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
7139 DWARF extensions from later standard versions is allowed.
7141 @item -gno-strict-dwarf
7142 @opindex gno-strict-dwarf
7143 Allow using extensions of later DWARF standard version than selected with
7144 @option{-gdwarf-@var{version}}.
7147 @item -gno-column-info
7148 @opindex gcolumn-info
7149 @opindex gno-column-info
7150 Emit location column information into DWARF debugging information, rather
7151 than just file and line.
7152 This option is enabled by default.
7154 @item -gz@r{[}=@var{type}@r{]}
7156 Produce compressed debug sections in DWARF format, if that is supported.
7157 If @var{type} is not given, the default type depends on the capabilities
7158 of the assembler and linker used. @var{type} may be one of
7159 @samp{none} (don't compress debug sections), @samp{zlib} (use zlib
7160 compression in ELF gABI format), or @samp{zlib-gnu} (use zlib
7161 compression in traditional GNU format). If the linker doesn't support
7162 writing compressed debug sections, the option is rejected. Otherwise,
7163 if the assembler does not support them, @option{-gz} is silently ignored
7164 when producing object files.
7166 @item -femit-struct-debug-baseonly
7167 @opindex femit-struct-debug-baseonly
7168 Emit debug information for struct-like types
7169 only when the base name of the compilation source file
7170 matches the base name of file in which the struct is defined.
7172 This option substantially reduces the size of debugging information,
7173 but at significant potential loss in type information to the debugger.
7174 See @option{-femit-struct-debug-reduced} for a less aggressive option.
7175 See @option{-femit-struct-debug-detailed} for more detailed control.
7177 This option works only with DWARF debug output.
7179 @item -femit-struct-debug-reduced
7180 @opindex femit-struct-debug-reduced
7181 Emit debug information for struct-like types
7182 only when the base name of the compilation source file
7183 matches the base name of file in which the type is defined,
7184 unless the struct is a template or defined in a system header.
7186 This option significantly reduces the size of debugging information,
7187 with some potential loss in type information to the debugger.
7188 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
7189 See @option{-femit-struct-debug-detailed} for more detailed control.
7191 This option works only with DWARF debug output.
7193 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
7194 @opindex femit-struct-debug-detailed
7195 Specify the struct-like types
7196 for which the compiler generates debug information.
7197 The intent is to reduce duplicate struct debug information
7198 between different object files within the same program.
7200 This option is a detailed version of
7201 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
7202 which serves for most needs.
7204 A specification has the syntax@*
7205 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
7207 The optional first word limits the specification to
7208 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
7209 A struct type is used directly when it is the type of a variable, member.
7210 Indirect uses arise through pointers to structs.
7211 That is, when use of an incomplete struct is valid, the use is indirect.
7213 @samp{struct one direct; struct two * indirect;}.
7215 The optional second word limits the specification to
7216 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
7217 Generic structs are a bit complicated to explain.
7218 For C++, these are non-explicit specializations of template classes,
7219 or non-template classes within the above.
7220 Other programming languages have generics,
7221 but @option{-femit-struct-debug-detailed} does not yet implement them.
7223 The third word specifies the source files for those
7224 structs for which the compiler should emit debug information.
7225 The values @samp{none} and @samp{any} have the normal meaning.
7226 The value @samp{base} means that
7227 the base of name of the file in which the type declaration appears
7228 must match the base of the name of the main compilation file.
7229 In practice, this means that when compiling @file{foo.c}, debug information
7230 is generated for types declared in that file and @file{foo.h},
7231 but not other header files.
7232 The value @samp{sys} means those types satisfying @samp{base}
7233 or declared in system or compiler headers.
7235 You may need to experiment to determine the best settings for your application.
7237 The default is @option{-femit-struct-debug-detailed=all}.
7239 This option works only with DWARF debug output.
7241 @item -fno-dwarf2-cfi-asm
7242 @opindex fdwarf2-cfi-asm
7243 @opindex fno-dwarf2-cfi-asm
7244 Emit DWARF unwind info as compiler generated @code{.eh_frame} section
7245 instead of using GAS @code{.cfi_*} directives.
7247 @item -fno-eliminate-unused-debug-types
7248 @opindex feliminate-unused-debug-types
7249 @opindex fno-eliminate-unused-debug-types
7250 Normally, when producing DWARF output, GCC avoids producing debug symbol
7251 output for types that are nowhere used in the source file being compiled.
7252 Sometimes it is useful to have GCC emit debugging
7253 information for all types declared in a compilation
7254 unit, regardless of whether or not they are actually used
7255 in that compilation unit, for example
7256 if, in the debugger, you want to cast a value to a type that is
7257 not actually used in your program (but is declared). More often,
7258 however, this results in a significant amount of wasted space.
7261 @node Optimize Options
7262 @section Options That Control Optimization
7263 @cindex optimize options
7264 @cindex options, optimization
7266 These options control various sorts of optimizations.
7268 Without any optimization option, the compiler's goal is to reduce the
7269 cost of compilation and to make debugging produce the expected
7270 results. Statements are independent: if you stop the program with a
7271 breakpoint between statements, you can then assign a new value to any
7272 variable or change the program counter to any other statement in the
7273 function and get exactly the results you expect from the source
7276 Turning on optimization flags makes the compiler attempt to improve
7277 the performance and/or code size at the expense of compilation time
7278 and possibly the ability to debug the program.
7280 The compiler performs optimization based on the knowledge it has of the
7281 program. Compiling multiple files at once to a single output file mode allows
7282 the compiler to use information gained from all of the files when compiling
7285 Not all optimizations are controlled directly by a flag. Only
7286 optimizations that have a flag are listed in this section.
7288 Most optimizations are only enabled if an @option{-O} level is set on
7289 the command line. Otherwise they are disabled, even if individual
7290 optimization flags are specified.
7292 Depending on the target and how GCC was configured, a slightly different
7293 set of optimizations may be enabled at each @option{-O} level than
7294 those listed here. You can invoke GCC with @option{-Q --help=optimizers}
7295 to find out the exact set of optimizations that are enabled at each level.
7296 @xref{Overall Options}, for examples.
7303 Optimize. Optimizing compilation takes somewhat more time, and a lot
7304 more memory for a large function.
7306 With @option{-O}, the compiler tries to reduce code size and execution
7307 time, without performing any optimizations that take a great deal of
7310 @option{-O} turns on the following optimization flags:
7313 -fbranch-count-reg @gol
7314 -fcombine-stack-adjustments @gol
7316 -fcprop-registers @gol
7319 -fdelayed-branch @gol
7321 -fforward-propagate @gol
7322 -fguess-branch-probability @gol
7323 -fif-conversion2 @gol
7324 -fif-conversion @gol
7325 -finline-functions-called-once @gol
7326 -fipa-pure-const @gol
7328 -fipa-reference @gol
7329 -fmerge-constants @gol
7330 -fmove-loop-invariants @gol
7331 -fomit-frame-pointer @gol
7332 -freorder-blocks @gol
7334 -fshrink-wrap-separate @gol
7335 -fsplit-wide-types @gol
7341 -ftree-coalesce-vars @gol
7342 -ftree-copy-prop @gol
7344 -ftree-dominator-opts @gol
7346 -ftree-forwprop @gol
7358 Optimize even more. GCC performs nearly all supported optimizations
7359 that do not involve a space-speed tradeoff.
7360 As compared to @option{-O}, this option increases both compilation time
7361 and the performance of the generated code.
7363 @option{-O2} turns on all optimization flags specified by @option{-O}. It
7364 also turns on the following optimization flags:
7365 @gccoptlist{-fthread-jumps @gol
7366 -falign-functions -falign-jumps @gol
7367 -falign-loops -falign-labels @gol
7370 -fcse-follow-jumps -fcse-skip-blocks @gol
7371 -fdelete-null-pointer-checks @gol
7372 -fdevirtualize -fdevirtualize-speculatively @gol
7373 -fexpensive-optimizations @gol
7374 -fgcse -fgcse-lm @gol
7375 -fhoist-adjacent-loads @gol
7376 -finline-small-functions @gol
7377 -findirect-inlining @gol
7383 -fisolate-erroneous-paths-dereference @gol
7385 -foptimize-sibling-calls @gol
7386 -foptimize-strlen @gol
7387 -fpartial-inlining @gol
7389 -freorder-blocks-algorithm=stc @gol
7390 -freorder-blocks-and-partition -freorder-functions @gol
7391 -frerun-cse-after-loop @gol
7392 -fsched-interblock -fsched-spec @gol
7393 -fschedule-insns -fschedule-insns2 @gol
7394 -fstore-merging @gol
7395 -fstrict-aliasing @gol
7396 -ftree-builtin-call-dce @gol
7397 -ftree-switch-conversion -ftree-tail-merge @gol
7398 -fcode-hoisting @gol
7403 Please note the warning under @option{-fgcse} about
7404 invoking @option{-O2} on programs that use computed gotos.
7408 Optimize yet more. @option{-O3} turns on all optimizations specified
7409 by @option{-O2} and also turns on the following optimization flags:
7410 @gccoptlist{-finline-functions @gol
7411 -funswitch-loops @gol
7412 -fpredictive-commoning @gol
7413 -fgcse-after-reload @gol
7414 -ftree-loop-vectorize @gol
7415 -ftree-loop-distribution @gol
7416 -ftree-loop-distribute-patterns @gol
7418 -ftree-slp-vectorize @gol
7419 -fvect-cost-model @gol
7420 -ftree-partial-pre @gol
7426 Reduce compilation time and make debugging produce the expected
7427 results. This is the default.
7431 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
7432 do not typically increase code size. It also performs further
7433 optimizations designed to reduce code size.
7435 @option{-Os} disables the following optimization flags:
7436 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
7437 -falign-labels -freorder-blocks -freorder-blocks-algorithm=stc @gol
7438 -freorder-blocks-and-partition -fprefetch-loop-arrays}
7442 Disregard strict standards compliance. @option{-Ofast} enables all
7443 @option{-O3} optimizations. It also enables optimizations that are not
7444 valid for all standard-compliant programs.
7445 It turns on @option{-ffast-math} and the Fortran-specific
7446 @option{-fstack-arrays}, unless @option{-fmax-stack-var-size} is
7447 specified, and @option{-fno-protect-parens}.
7451 Optimize debugging experience. @option{-Og} enables optimizations
7452 that do not interfere with debugging. It should be the optimization
7453 level of choice for the standard edit-compile-debug cycle, offering
7454 a reasonable level of optimization while maintaining fast compilation
7455 and a good debugging experience.
7458 If you use multiple @option{-O} options, with or without level numbers,
7459 the last such option is the one that is effective.
7461 Options of the form @option{-f@var{flag}} specify machine-independent
7462 flags. Most flags have both positive and negative forms; the negative
7463 form of @option{-ffoo} is @option{-fno-foo}. In the table
7464 below, only one of the forms is listed---the one you typically
7465 use. You can figure out the other form by either removing @samp{no-}
7468 The following options control specific optimizations. They are either
7469 activated by @option{-O} options or are related to ones that are. You
7470 can use the following flags in the rare cases when ``fine-tuning'' of
7471 optimizations to be performed is desired.
7474 @item -fno-defer-pop
7475 @opindex fno-defer-pop
7476 Always pop the arguments to each function call as soon as that function
7477 returns. For machines that must pop arguments after a function call,
7478 the compiler normally lets arguments accumulate on the stack for several
7479 function calls and pops them all at once.
7481 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7483 @item -fforward-propagate
7484 @opindex fforward-propagate
7485 Perform a forward propagation pass on RTL@. The pass tries to combine two
7486 instructions and checks if the result can be simplified. If loop unrolling
7487 is active, two passes are performed and the second is scheduled after
7490 This option is enabled by default at optimization levels @option{-O},
7491 @option{-O2}, @option{-O3}, @option{-Os}.
7493 @item -ffp-contract=@var{style}
7494 @opindex ffp-contract
7495 @option{-ffp-contract=off} disables floating-point expression contraction.
7496 @option{-ffp-contract=fast} enables floating-point expression contraction
7497 such as forming of fused multiply-add operations if the target has
7498 native support for them.
7499 @option{-ffp-contract=on} enables floating-point expression contraction
7500 if allowed by the language standard. This is currently not implemented
7501 and treated equal to @option{-ffp-contract=off}.
7503 The default is @option{-ffp-contract=fast}.
7505 @item -fomit-frame-pointer
7506 @opindex fomit-frame-pointer
7507 Omit the frame pointer in functions that don't need one. This avoids the
7508 instructions to save, set up and restore the frame pointer; on many targets
7509 it also makes an extra register available.
7511 On some targets this flag has no effect because the standard calling sequence
7512 always uses a frame pointer, so it cannot be omitted.
7514 Note that @option{-fno-omit-frame-pointer} doesn't guarantee the frame pointer
7515 is used in all functions. Several targets always omit the frame pointer in
7518 Enabled by default at @option{-O} and higher.
7520 @item -foptimize-sibling-calls
7521 @opindex foptimize-sibling-calls
7522 Optimize sibling and tail recursive calls.
7524 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7526 @item -foptimize-strlen
7527 @opindex foptimize-strlen
7528 Optimize various standard C string functions (e.g. @code{strlen},
7529 @code{strchr} or @code{strcpy}) and
7530 their @code{_FORTIFY_SOURCE} counterparts into faster alternatives.
7532 Enabled at levels @option{-O2}, @option{-O3}.
7536 Do not expand any functions inline apart from those marked with
7537 the @code{always_inline} attribute. This is the default when not
7540 Single functions can be exempted from inlining by marking them
7541 with the @code{noinline} attribute.
7543 @item -finline-small-functions
7544 @opindex finline-small-functions
7545 Integrate functions into their callers when their body is smaller than expected
7546 function call code (so overall size of program gets smaller). The compiler
7547 heuristically decides which functions are simple enough to be worth integrating
7548 in this way. This inlining applies to all functions, even those not declared
7551 Enabled at level @option{-O2}.
7553 @item -findirect-inlining
7554 @opindex findirect-inlining
7555 Inline also indirect calls that are discovered to be known at compile
7556 time thanks to previous inlining. This option has any effect only
7557 when inlining itself is turned on by the @option{-finline-functions}
7558 or @option{-finline-small-functions} options.
7560 Enabled at level @option{-O2}.
7562 @item -finline-functions
7563 @opindex finline-functions
7564 Consider all functions for inlining, even if they are not declared inline.
7565 The compiler heuristically decides which functions are worth integrating
7568 If all calls to a given function are integrated, and the function is
7569 declared @code{static}, then the function is normally not output as
7570 assembler code in its own right.
7572 Enabled at level @option{-O3}.
7574 @item -finline-functions-called-once
7575 @opindex finline-functions-called-once
7576 Consider all @code{static} functions called once for inlining into their
7577 caller even if they are not marked @code{inline}. If a call to a given
7578 function is integrated, then the function is not output as assembler code
7581 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
7583 @item -fearly-inlining
7584 @opindex fearly-inlining
7585 Inline functions marked by @code{always_inline} and functions whose body seems
7586 smaller than the function call overhead early before doing
7587 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
7588 makes profiling significantly cheaper and usually inlining faster on programs
7589 having large chains of nested wrapper functions.
7595 Perform interprocedural scalar replacement of aggregates, removal of
7596 unused parameters and replacement of parameters passed by reference
7597 by parameters passed by value.
7599 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
7601 @item -finline-limit=@var{n}
7602 @opindex finline-limit
7603 By default, GCC limits the size of functions that can be inlined. This flag
7604 allows coarse control of this limit. @var{n} is the size of functions that
7605 can be inlined in number of pseudo instructions.
7607 Inlining is actually controlled by a number of parameters, which may be
7608 specified individually by using @option{--param @var{name}=@var{value}}.
7609 The @option{-finline-limit=@var{n}} option sets some of these parameters
7613 @item max-inline-insns-single
7614 is set to @var{n}/2.
7615 @item max-inline-insns-auto
7616 is set to @var{n}/2.
7619 See below for a documentation of the individual
7620 parameters controlling inlining and for the defaults of these parameters.
7622 @emph{Note:} there may be no value to @option{-finline-limit} that results
7623 in default behavior.
7625 @emph{Note:} pseudo instruction represents, in this particular context, an
7626 abstract measurement of function's size. In no way does it represent a count
7627 of assembly instructions and as such its exact meaning might change from one
7628 release to an another.
7630 @item -fno-keep-inline-dllexport
7631 @opindex fno-keep-inline-dllexport
7632 This is a more fine-grained version of @option{-fkeep-inline-functions},
7633 which applies only to functions that are declared using the @code{dllexport}
7634 attribute or declspec. @xref{Function Attributes,,Declaring Attributes of
7637 @item -fkeep-inline-functions
7638 @opindex fkeep-inline-functions
7639 In C, emit @code{static} functions that are declared @code{inline}
7640 into the object file, even if the function has been inlined into all
7641 of its callers. This switch does not affect functions using the
7642 @code{extern inline} extension in GNU C90@. In C++, emit any and all
7643 inline functions into the object file.
7645 @item -fkeep-static-functions
7646 @opindex fkeep-static-functions
7647 Emit @code{static} functions into the object file, even if the function
7650 @item -fkeep-static-consts
7651 @opindex fkeep-static-consts
7652 Emit variables declared @code{static const} when optimization isn't turned
7653 on, even if the variables aren't referenced.
7655 GCC enables this option by default. If you want to force the compiler to
7656 check if a variable is referenced, regardless of whether or not
7657 optimization is turned on, use the @option{-fno-keep-static-consts} option.
7659 @item -fmerge-constants
7660 @opindex fmerge-constants
7661 Attempt to merge identical constants (string constants and floating-point
7662 constants) across compilation units.
7664 This option is the default for optimized compilation if the assembler and
7665 linker support it. Use @option{-fno-merge-constants} to inhibit this
7668 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7670 @item -fmerge-all-constants
7671 @opindex fmerge-all-constants
7672 Attempt to merge identical constants and identical variables.
7674 This option implies @option{-fmerge-constants}. In addition to
7675 @option{-fmerge-constants} this considers e.g.@: even constant initialized
7676 arrays or initialized constant variables with integral or floating-point
7677 types. Languages like C or C++ require each variable, including multiple
7678 instances of the same variable in recursive calls, to have distinct locations,
7679 so using this option results in non-conforming
7682 @item -fmodulo-sched
7683 @opindex fmodulo-sched
7684 Perform swing modulo scheduling immediately before the first scheduling
7685 pass. This pass looks at innermost loops and reorders their
7686 instructions by overlapping different iterations.
7688 @item -fmodulo-sched-allow-regmoves
7689 @opindex fmodulo-sched-allow-regmoves
7690 Perform more aggressive SMS-based modulo scheduling with register moves
7691 allowed. By setting this flag certain anti-dependences edges are
7692 deleted, which triggers the generation of reg-moves based on the
7693 life-range analysis. This option is effective only with
7694 @option{-fmodulo-sched} enabled.
7696 @item -fno-branch-count-reg
7697 @opindex fno-branch-count-reg
7698 Avoid running a pass scanning for opportunities to use ``decrement and
7699 branch'' instructions on a count register instead of generating sequences
7700 of instructions that decrement a register, compare it against zero, and
7701 then branch based upon the result. This option is only meaningful on
7702 architectures that support such instructions, which include x86, PowerPC,
7703 IA-64 and S/390. Note that the @option{-fno-branch-count-reg} option
7704 doesn't remove the decrement and branch instructions from the generated
7705 instruction stream introduced by other optimization passes.
7707 Enabled by default at @option{-O1} and higher.
7709 The default is @option{-fbranch-count-reg}.
7711 @item -fno-function-cse
7712 @opindex fno-function-cse
7713 Do not put function addresses in registers; make each instruction that
7714 calls a constant function contain the function's address explicitly.
7716 This option results in less efficient code, but some strange hacks
7717 that alter the assembler output may be confused by the optimizations
7718 performed when this option is not used.
7720 The default is @option{-ffunction-cse}
7722 @item -fno-zero-initialized-in-bss
7723 @opindex fno-zero-initialized-in-bss
7724 If the target supports a BSS section, GCC by default puts variables that
7725 are initialized to zero into BSS@. This can save space in the resulting
7728 This option turns off this behavior because some programs explicitly
7729 rely on variables going to the data section---e.g., so that the
7730 resulting executable can find the beginning of that section and/or make
7731 assumptions based on that.
7733 The default is @option{-fzero-initialized-in-bss}.
7735 @item -fthread-jumps
7736 @opindex fthread-jumps
7737 Perform optimizations that check to see if a jump branches to a
7738 location where another comparison subsumed by the first is found. If
7739 so, the first branch is redirected to either the destination of the
7740 second branch or a point immediately following it, depending on whether
7741 the condition is known to be true or false.
7743 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7745 @item -fsplit-wide-types
7746 @opindex fsplit-wide-types
7747 When using a type that occupies multiple registers, such as @code{long
7748 long} on a 32-bit system, split the registers apart and allocate them
7749 independently. This normally generates better code for those types,
7750 but may make debugging more difficult.
7752 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
7755 @item -fcse-follow-jumps
7756 @opindex fcse-follow-jumps
7757 In common subexpression elimination (CSE), scan through jump instructions
7758 when the target of the jump is not reached by any other path. For
7759 example, when CSE encounters an @code{if} statement with an
7760 @code{else} clause, CSE follows the jump when the condition
7763 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7765 @item -fcse-skip-blocks
7766 @opindex fcse-skip-blocks
7767 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
7768 follow jumps that conditionally skip over blocks. When CSE
7769 encounters a simple @code{if} statement with no else clause,
7770 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
7771 body of the @code{if}.
7773 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7775 @item -frerun-cse-after-loop
7776 @opindex frerun-cse-after-loop
7777 Re-run common subexpression elimination after loop optimizations are
7780 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7784 Perform a global common subexpression elimination pass.
7785 This pass also performs global constant and copy propagation.
7787 @emph{Note:} When compiling a program using computed gotos, a GCC
7788 extension, you may get better run-time performance if you disable
7789 the global common subexpression elimination pass by adding
7790 @option{-fno-gcse} to the command line.
7792 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7796 When @option{-fgcse-lm} is enabled, global common subexpression elimination
7797 attempts to move loads that are only killed by stores into themselves. This
7798 allows a loop containing a load/store sequence to be changed to a load outside
7799 the loop, and a copy/store within the loop.
7801 Enabled by default when @option{-fgcse} is enabled.
7805 When @option{-fgcse-sm} is enabled, a store motion pass is run after
7806 global common subexpression elimination. This pass attempts to move
7807 stores out of loops. When used in conjunction with @option{-fgcse-lm},
7808 loops containing a load/store sequence can be changed to a load before
7809 the loop and a store after the loop.
7811 Not enabled at any optimization level.
7815 When @option{-fgcse-las} is enabled, the global common subexpression
7816 elimination pass eliminates redundant loads that come after stores to the
7817 same memory location (both partial and full redundancies).
7819 Not enabled at any optimization level.
7821 @item -fgcse-after-reload
7822 @opindex fgcse-after-reload
7823 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
7824 pass is performed after reload. The purpose of this pass is to clean up
7827 @item -faggressive-loop-optimizations
7828 @opindex faggressive-loop-optimizations
7829 This option tells the loop optimizer to use language constraints to
7830 derive bounds for the number of iterations of a loop. This assumes that
7831 loop code does not invoke undefined behavior by for example causing signed
7832 integer overflows or out-of-bound array accesses. The bounds for the
7833 number of iterations of a loop are used to guide loop unrolling and peeling
7834 and loop exit test optimizations.
7835 This option is enabled by default.
7837 @item -funconstrained-commons
7838 @opindex funconstrained-commons
7839 This option tells the compiler that variables declared in common blocks
7840 (e.g. Fortran) may later be overridden with longer trailing arrays. This
7841 prevents certain optimizations that depend on knowing the array bounds.
7843 @item -fcrossjumping
7844 @opindex fcrossjumping
7845 Perform cross-jumping transformation.
7846 This transformation unifies equivalent code and saves code size. The
7847 resulting code may or may not perform better than without cross-jumping.
7849 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7851 @item -fauto-inc-dec
7852 @opindex fauto-inc-dec
7853 Combine increments or decrements of addresses with memory accesses.
7854 This pass is always skipped on architectures that do not have
7855 instructions to support this. Enabled by default at @option{-O} and
7856 higher on architectures that support this.
7860 Perform dead code elimination (DCE) on RTL@.
7861 Enabled by default at @option{-O} and higher.
7865 Perform dead store elimination (DSE) on RTL@.
7866 Enabled by default at @option{-O} and higher.
7868 @item -fif-conversion
7869 @opindex fif-conversion
7870 Attempt to transform conditional jumps into branch-less equivalents. This
7871 includes use of conditional moves, min, max, set flags and abs instructions, and
7872 some tricks doable by standard arithmetics. The use of conditional execution
7873 on chips where it is available is controlled by @option{-fif-conversion2}.
7875 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7877 @item -fif-conversion2
7878 @opindex fif-conversion2
7879 Use conditional execution (where available) to transform conditional jumps into
7880 branch-less equivalents.
7882 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7884 @item -fdeclone-ctor-dtor
7885 @opindex fdeclone-ctor-dtor
7886 The C++ ABI requires multiple entry points for constructors and
7887 destructors: one for a base subobject, one for a complete object, and
7888 one for a virtual destructor that calls operator delete afterwards.
7889 For a hierarchy with virtual bases, the base and complete variants are
7890 clones, which means two copies of the function. With this option, the
7891 base and complete variants are changed to be thunks that call a common
7894 Enabled by @option{-Os}.
7896 @item -fdelete-null-pointer-checks
7897 @opindex fdelete-null-pointer-checks
7898 Assume that programs cannot safely dereference null pointers, and that
7899 no code or data element resides at address zero.
7900 This option enables simple constant
7901 folding optimizations at all optimization levels. In addition, other
7902 optimization passes in GCC use this flag to control global dataflow
7903 analyses that eliminate useless checks for null pointers; these assume
7904 that a memory access to address zero always results in a trap, so
7905 that if a pointer is checked after it has already been dereferenced,
7908 Note however that in some environments this assumption is not true.
7909 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
7910 for programs that depend on that behavior.
7912 This option is enabled by default on most targets. On Nios II ELF, it
7913 defaults to off. On AVR, CR16, and MSP430, this option is completely disabled.
7915 Passes that use the dataflow information
7916 are enabled independently at different optimization levels.
7918 @item -fdevirtualize
7919 @opindex fdevirtualize
7920 Attempt to convert calls to virtual functions to direct calls. This
7921 is done both within a procedure and interprocedurally as part of
7922 indirect inlining (@option{-findirect-inlining}) and interprocedural constant
7923 propagation (@option{-fipa-cp}).
7924 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7926 @item -fdevirtualize-speculatively
7927 @opindex fdevirtualize-speculatively
7928 Attempt to convert calls to virtual functions to speculative direct calls.
7929 Based on the analysis of the type inheritance graph, determine for a given call
7930 the set of likely targets. If the set is small, preferably of size 1, change
7931 the call into a conditional deciding between direct and indirect calls. The
7932 speculative calls enable more optimizations, such as inlining. When they seem
7933 useless after further optimization, they are converted back into original form.
7935 @item -fdevirtualize-at-ltrans
7936 @opindex fdevirtualize-at-ltrans
7937 Stream extra information needed for aggressive devirtualization when running
7938 the link-time optimizer in local transformation mode.
7939 This option enables more devirtualization but
7940 significantly increases the size of streamed data. For this reason it is
7941 disabled by default.
7943 @item -fexpensive-optimizations
7944 @opindex fexpensive-optimizations
7945 Perform a number of minor optimizations that are relatively expensive.
7947 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7951 Attempt to remove redundant extension instructions. This is especially
7952 helpful for the x86-64 architecture, which implicitly zero-extends in 64-bit
7953 registers after writing to their lower 32-bit half.
7955 Enabled for Alpha, AArch64 and x86 at levels @option{-O2},
7956 @option{-O3}, @option{-Os}.
7958 @item -fno-lifetime-dse
7959 @opindex fno-lifetime-dse
7960 In C++ the value of an object is only affected by changes within its
7961 lifetime: when the constructor begins, the object has an indeterminate
7962 value, and any changes during the lifetime of the object are dead when
7963 the object is destroyed. Normally dead store elimination will take
7964 advantage of this; if your code relies on the value of the object
7965 storage persisting beyond the lifetime of the object, you can use this
7966 flag to disable this optimization. To preserve stores before the
7967 constructor starts (e.g. because your operator new clears the object
7968 storage) but still treat the object as dead after the destructor you,
7969 can use @option{-flifetime-dse=1}. The default behavior can be
7970 explicitly selected with @option{-flifetime-dse=2}.
7971 @option{-flifetime-dse=0} is equivalent to @option{-fno-lifetime-dse}.
7973 @item -flive-range-shrinkage
7974 @opindex flive-range-shrinkage
7975 Attempt to decrease register pressure through register live range
7976 shrinkage. This is helpful for fast processors with small or moderate
7979 @item -fira-algorithm=@var{algorithm}
7980 @opindex fira-algorithm
7981 Use the specified coloring algorithm for the integrated register
7982 allocator. The @var{algorithm} argument can be @samp{priority}, which
7983 specifies Chow's priority coloring, or @samp{CB}, which specifies
7984 Chaitin-Briggs coloring. Chaitin-Briggs coloring is not implemented
7985 for all architectures, but for those targets that do support it, it is
7986 the default because it generates better code.
7988 @item -fira-region=@var{region}
7989 @opindex fira-region
7990 Use specified regions for the integrated register allocator. The
7991 @var{region} argument should be one of the following:
7996 Use all loops as register allocation regions.
7997 This can give the best results for machines with a small and/or
7998 irregular register set.
8001 Use all loops except for loops with small register pressure
8002 as the regions. This value usually gives
8003 the best results in most cases and for most architectures,
8004 and is enabled by default when compiling with optimization for speed
8005 (@option{-O}, @option{-O2}, @dots{}).
8008 Use all functions as a single region.
8009 This typically results in the smallest code size, and is enabled by default for
8010 @option{-Os} or @option{-O0}.
8014 @item -fira-hoist-pressure
8015 @opindex fira-hoist-pressure
8016 Use IRA to evaluate register pressure in the code hoisting pass for
8017 decisions to hoist expressions. This option usually results in smaller
8018 code, but it can slow the compiler down.
8020 This option is enabled at level @option{-Os} for all targets.
8022 @item -fira-loop-pressure
8023 @opindex fira-loop-pressure
8024 Use IRA to evaluate register pressure in loops for decisions to move
8025 loop invariants. This option usually results in generation
8026 of faster and smaller code on machines with large register files (>= 32
8027 registers), but it can slow the compiler down.
8029 This option is enabled at level @option{-O3} for some targets.
8031 @item -fno-ira-share-save-slots
8032 @opindex fno-ira-share-save-slots
8033 Disable sharing of stack slots used for saving call-used hard
8034 registers living through a call. Each hard register gets a
8035 separate stack slot, and as a result function stack frames are
8038 @item -fno-ira-share-spill-slots
8039 @opindex fno-ira-share-spill-slots
8040 Disable sharing of stack slots allocated for pseudo-registers. Each
8041 pseudo-register that does not get a hard register gets a separate
8042 stack slot, and as a result function stack frames are larger.
8046 Enable CFG-sensitive rematerialization in LRA. Instead of loading
8047 values of spilled pseudos, LRA tries to rematerialize (recalculate)
8048 values if it is profitable.
8050 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8052 @item -fdelayed-branch
8053 @opindex fdelayed-branch
8054 If supported for the target machine, attempt to reorder instructions
8055 to exploit instruction slots available after delayed branch
8058 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8060 @item -fschedule-insns
8061 @opindex fschedule-insns
8062 If supported for the target machine, attempt to reorder instructions to
8063 eliminate execution stalls due to required data being unavailable. This
8064 helps machines that have slow floating point or memory load instructions
8065 by allowing other instructions to be issued until the result of the load
8066 or floating-point instruction is required.
8068 Enabled at levels @option{-O2}, @option{-O3}.
8070 @item -fschedule-insns2
8071 @opindex fschedule-insns2
8072 Similar to @option{-fschedule-insns}, but requests an additional pass of
8073 instruction scheduling after register allocation has been done. This is
8074 especially useful on machines with a relatively small number of
8075 registers and where memory load instructions take more than one cycle.
8077 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8079 @item -fno-sched-interblock
8080 @opindex fno-sched-interblock
8081 Don't schedule instructions across basic blocks. This is normally
8082 enabled by default when scheduling before register allocation, i.e.@:
8083 with @option{-fschedule-insns} or at @option{-O2} or higher.
8085 @item -fno-sched-spec
8086 @opindex fno-sched-spec
8087 Don't allow speculative motion of non-load instructions. This is normally
8088 enabled by default when scheduling before register allocation, i.e.@:
8089 with @option{-fschedule-insns} or at @option{-O2} or higher.
8091 @item -fsched-pressure
8092 @opindex fsched-pressure
8093 Enable register pressure sensitive insn scheduling before register
8094 allocation. This only makes sense when scheduling before register
8095 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
8096 @option{-O2} or higher. Usage of this option can improve the
8097 generated code and decrease its size by preventing register pressure
8098 increase above the number of available hard registers and subsequent
8099 spills in register allocation.
8101 @item -fsched-spec-load
8102 @opindex fsched-spec-load
8103 Allow speculative motion of some load instructions. This only makes
8104 sense when scheduling before register allocation, i.e.@: with
8105 @option{-fschedule-insns} or at @option{-O2} or higher.
8107 @item -fsched-spec-load-dangerous
8108 @opindex fsched-spec-load-dangerous
8109 Allow speculative motion of more load instructions. This only makes
8110 sense when scheduling before register allocation, i.e.@: with
8111 @option{-fschedule-insns} or at @option{-O2} or higher.
8113 @item -fsched-stalled-insns
8114 @itemx -fsched-stalled-insns=@var{n}
8115 @opindex fsched-stalled-insns
8116 Define how many insns (if any) can be moved prematurely from the queue
8117 of stalled insns into the ready list during the second scheduling pass.
8118 @option{-fno-sched-stalled-insns} means that no insns are moved
8119 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
8120 on how many queued insns can be moved prematurely.
8121 @option{-fsched-stalled-insns} without a value is equivalent to
8122 @option{-fsched-stalled-insns=1}.
8124 @item -fsched-stalled-insns-dep
8125 @itemx -fsched-stalled-insns-dep=@var{n}
8126 @opindex fsched-stalled-insns-dep
8127 Define how many insn groups (cycles) are examined for a dependency
8128 on a stalled insn that is a candidate for premature removal from the queue
8129 of stalled insns. This has an effect only during the second scheduling pass,
8130 and only if @option{-fsched-stalled-insns} is used.
8131 @option{-fno-sched-stalled-insns-dep} is equivalent to
8132 @option{-fsched-stalled-insns-dep=0}.
8133 @option{-fsched-stalled-insns-dep} without a value is equivalent to
8134 @option{-fsched-stalled-insns-dep=1}.
8136 @item -fsched2-use-superblocks
8137 @opindex fsched2-use-superblocks
8138 When scheduling after register allocation, use superblock scheduling.
8139 This allows motion across basic block boundaries,
8140 resulting in faster schedules. This option is experimental, as not all machine
8141 descriptions used by GCC model the CPU closely enough to avoid unreliable
8142 results from the algorithm.
8144 This only makes sense when scheduling after register allocation, i.e.@: with
8145 @option{-fschedule-insns2} or at @option{-O2} or higher.
8147 @item -fsched-group-heuristic
8148 @opindex fsched-group-heuristic
8149 Enable the group heuristic in the scheduler. This heuristic favors
8150 the instruction that belongs to a schedule group. This is enabled
8151 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
8152 or @option{-fschedule-insns2} or at @option{-O2} or higher.
8154 @item -fsched-critical-path-heuristic
8155 @opindex fsched-critical-path-heuristic
8156 Enable the critical-path heuristic in the scheduler. This heuristic favors
8157 instructions on the critical path. This is enabled by default when
8158 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
8159 or @option{-fschedule-insns2} or at @option{-O2} or higher.
8161 @item -fsched-spec-insn-heuristic
8162 @opindex fsched-spec-insn-heuristic
8163 Enable the speculative instruction heuristic in the scheduler. This
8164 heuristic favors speculative instructions with greater dependency weakness.
8165 This is enabled by default when scheduling is enabled, i.e.@:
8166 with @option{-fschedule-insns} or @option{-fschedule-insns2}
8167 or at @option{-O2} or higher.
8169 @item -fsched-rank-heuristic
8170 @opindex fsched-rank-heuristic
8171 Enable the rank heuristic in the scheduler. This heuristic favors
8172 the instruction belonging to a basic block with greater size or frequency.
8173 This is enabled by default when scheduling is enabled, i.e.@:
8174 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
8175 at @option{-O2} or higher.
8177 @item -fsched-last-insn-heuristic
8178 @opindex fsched-last-insn-heuristic
8179 Enable the last-instruction heuristic in the scheduler. This heuristic
8180 favors the instruction that is less dependent on the last instruction
8181 scheduled. This is enabled by default when scheduling is enabled,
8182 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
8183 at @option{-O2} or higher.
8185 @item -fsched-dep-count-heuristic
8186 @opindex fsched-dep-count-heuristic
8187 Enable the dependent-count heuristic in the scheduler. This heuristic
8188 favors the instruction that has more instructions depending on it.
8189 This is enabled by default when scheduling is enabled, i.e.@:
8190 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
8191 at @option{-O2} or higher.
8193 @item -freschedule-modulo-scheduled-loops
8194 @opindex freschedule-modulo-scheduled-loops
8195 Modulo scheduling is performed before traditional scheduling. If a loop
8196 is modulo scheduled, later scheduling passes may change its schedule.
8197 Use this option to control that behavior.
8199 @item -fselective-scheduling
8200 @opindex fselective-scheduling
8201 Schedule instructions using selective scheduling algorithm. Selective
8202 scheduling runs instead of the first scheduler pass.
8204 @item -fselective-scheduling2
8205 @opindex fselective-scheduling2
8206 Schedule instructions using selective scheduling algorithm. Selective
8207 scheduling runs instead of the second scheduler pass.
8209 @item -fsel-sched-pipelining
8210 @opindex fsel-sched-pipelining
8211 Enable software pipelining of innermost loops during selective scheduling.
8212 This option has no effect unless one of @option{-fselective-scheduling} or
8213 @option{-fselective-scheduling2} is turned on.
8215 @item -fsel-sched-pipelining-outer-loops
8216 @opindex fsel-sched-pipelining-outer-loops
8217 When pipelining loops during selective scheduling, also pipeline outer loops.
8218 This option has no effect unless @option{-fsel-sched-pipelining} is turned on.
8220 @item -fsemantic-interposition
8221 @opindex fsemantic-interposition
8222 Some object formats, like ELF, allow interposing of symbols by the
8224 This means that for symbols exported from the DSO, the compiler cannot perform
8225 interprocedural propagation, inlining and other optimizations in anticipation
8226 that the function or variable in question may change. While this feature is
8227 useful, for example, to rewrite memory allocation functions by a debugging
8228 implementation, it is expensive in the terms of code quality.
8229 With @option{-fno-semantic-interposition} the compiler assumes that
8230 if interposition happens for functions the overwriting function will have
8231 precisely the same semantics (and side effects).
8232 Similarly if interposition happens
8233 for variables, the constructor of the variable will be the same. The flag
8234 has no effect for functions explicitly declared inline
8235 (where it is never allowed for interposition to change semantics)
8236 and for symbols explicitly declared weak.
8239 @opindex fshrink-wrap
8240 Emit function prologues only before parts of the function that need it,
8241 rather than at the top of the function. This flag is enabled by default at
8242 @option{-O} and higher.
8244 @item -fshrink-wrap-separate
8245 @opindex fshrink-wrap-separate
8246 Shrink-wrap separate parts of the prologue and epilogue separately, so that
8247 those parts are only executed when needed.
8248 This option is on by default, but has no effect unless @option{-fshrink-wrap}
8249 is also turned on and the target supports this.
8251 @item -fcaller-saves
8252 @opindex fcaller-saves
8253 Enable allocation of values to registers that are clobbered by
8254 function calls, by emitting extra instructions to save and restore the
8255 registers around such calls. Such allocation is done only when it
8256 seems to result in better code.
8258 This option is always enabled by default on certain machines, usually
8259 those which have no call-preserved registers to use instead.
8261 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8263 @item -fcombine-stack-adjustments
8264 @opindex fcombine-stack-adjustments
8265 Tracks stack adjustments (pushes and pops) and stack memory references
8266 and then tries to find ways to combine them.
8268 Enabled by default at @option{-O1} and higher.
8272 Use caller save registers for allocation if those registers are not used by
8273 any called function. In that case it is not necessary to save and restore
8274 them around calls. This is only possible if called functions are part of
8275 same compilation unit as current function and they are compiled before it.
8277 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}, however the option
8278 is disabled if generated code will be instrumented for profiling
8279 (@option{-p}, or @option{-pg}) or if callee's register usage cannot be known
8280 exactly (this happens on targets that do not expose prologues
8281 and epilogues in RTL).
8283 @item -fconserve-stack
8284 @opindex fconserve-stack
8285 Attempt to minimize stack usage. The compiler attempts to use less
8286 stack space, even if that makes the program slower. This option
8287 implies setting the @option{large-stack-frame} parameter to 100
8288 and the @option{large-stack-frame-growth} parameter to 400.
8290 @item -ftree-reassoc
8291 @opindex ftree-reassoc
8292 Perform reassociation on trees. This flag is enabled by default
8293 at @option{-O} and higher.
8295 @item -fcode-hoisting
8296 @opindex fcode-hoisting
8297 Perform code hoisting. Code hoisting tries to move the
8298 evaluation of expressions executed on all paths to the function exit
8299 as early as possible. This is especially useful as a code size
8300 optimization, but it often helps for code speed as well.
8301 This flag is enabled by default at @option{-O2} and higher.
8305 Perform partial redundancy elimination (PRE) on trees. This flag is
8306 enabled by default at @option{-O2} and @option{-O3}.
8308 @item -ftree-partial-pre
8309 @opindex ftree-partial-pre
8310 Make partial redundancy elimination (PRE) more aggressive. This flag is
8311 enabled by default at @option{-O3}.
8313 @item -ftree-forwprop
8314 @opindex ftree-forwprop
8315 Perform forward propagation on trees. This flag is enabled by default
8316 at @option{-O} and higher.
8320 Perform full redundancy elimination (FRE) on trees. The difference
8321 between FRE and PRE is that FRE only considers expressions
8322 that are computed on all paths leading to the redundant computation.
8323 This analysis is faster than PRE, though it exposes fewer redundancies.
8324 This flag is enabled by default at @option{-O} and higher.
8326 @item -ftree-phiprop
8327 @opindex ftree-phiprop
8328 Perform hoisting of loads from conditional pointers on trees. This
8329 pass is enabled by default at @option{-O} and higher.
8331 @item -fhoist-adjacent-loads
8332 @opindex fhoist-adjacent-loads
8333 Speculatively hoist loads from both branches of an if-then-else if the
8334 loads are from adjacent locations in the same structure and the target
8335 architecture has a conditional move instruction. This flag is enabled
8336 by default at @option{-O2} and higher.
8338 @item -ftree-copy-prop
8339 @opindex ftree-copy-prop
8340 Perform copy propagation on trees. This pass eliminates unnecessary
8341 copy operations. This flag is enabled by default at @option{-O} and
8344 @item -fipa-pure-const
8345 @opindex fipa-pure-const
8346 Discover which functions are pure or constant.
8347 Enabled by default at @option{-O} and higher.
8349 @item -fipa-reference
8350 @opindex fipa-reference
8351 Discover which static variables do not escape the
8353 Enabled by default at @option{-O} and higher.
8357 Perform interprocedural pointer analysis and interprocedural modification
8358 and reference analysis. This option can cause excessive memory and
8359 compile-time usage on large compilation units. It is not enabled by
8360 default at any optimization level.
8363 @opindex fipa-profile
8364 Perform interprocedural profile propagation. The functions called only from
8365 cold functions are marked as cold. Also functions executed once (such as
8366 @code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
8367 functions and loop less parts of functions executed once are then optimized for
8369 Enabled by default at @option{-O} and higher.
8373 Perform interprocedural constant propagation.
8374 This optimization analyzes the program to determine when values passed
8375 to functions are constants and then optimizes accordingly.
8376 This optimization can substantially increase performance
8377 if the application has constants passed to functions.
8378 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
8380 @item -fipa-cp-clone
8381 @opindex fipa-cp-clone
8382 Perform function cloning to make interprocedural constant propagation stronger.
8383 When enabled, interprocedural constant propagation performs function cloning
8384 when externally visible function can be called with constant arguments.
8385 Because this optimization can create multiple copies of functions,
8386 it may significantly increase code size
8387 (see @option{--param ipcp-unit-growth=@var{value}}).
8388 This flag is enabled by default at @option{-O3}.
8391 @opindex -fipa-bit-cp
8392 When enabled, perform interprocedural bitwise constant
8393 propagation. This flag is enabled by default at @option{-O2}. It
8394 requires that @option{-fipa-cp} is enabled.
8398 When enabled, perform interprocedural propagation of value
8399 ranges. This flag is enabled by default at @option{-O2}. It requires
8400 that @option{-fipa-cp} is enabled.
8404 Perform Identical Code Folding for functions and read-only variables.
8405 The optimization reduces code size and may disturb unwind stacks by replacing
8406 a function by equivalent one with a different name. The optimization works
8407 more effectively with link-time optimization enabled.
8409 Nevertheless the behavior is similar to Gold Linker ICF optimization, GCC ICF
8410 works on different levels and thus the optimizations are not same - there are
8411 equivalences that are found only by GCC and equivalences found only by Gold.
8413 This flag is enabled by default at @option{-O2} and @option{-Os}.
8415 @item -fisolate-erroneous-paths-dereference
8416 @opindex fisolate-erroneous-paths-dereference
8417 Detect paths that trigger erroneous or undefined behavior due to
8418 dereferencing a null pointer. Isolate those paths from the main control
8419 flow and turn the statement with erroneous or undefined behavior into a trap.
8420 This flag is enabled by default at @option{-O2} and higher and depends on
8421 @option{-fdelete-null-pointer-checks} also being enabled.
8423 @item -fisolate-erroneous-paths-attribute
8424 @opindex fisolate-erroneous-paths-attribute
8425 Detect paths that trigger erroneous or undefined behavior due a null value
8426 being used in a way forbidden by a @code{returns_nonnull} or @code{nonnull}
8427 attribute. Isolate those paths from the main control flow and turn the
8428 statement with erroneous or undefined behavior into a trap. This is not
8429 currently enabled, but may be enabled by @option{-O2} in the future.
8433 Perform forward store motion on trees. This flag is
8434 enabled by default at @option{-O} and higher.
8436 @item -ftree-bit-ccp
8437 @opindex ftree-bit-ccp
8438 Perform sparse conditional bit constant propagation on trees and propagate
8439 pointer alignment information.
8440 This pass only operates on local scalar variables and is enabled by default
8441 at @option{-O} and higher. It requires that @option{-ftree-ccp} is enabled.
8445 Perform sparse conditional constant propagation (CCP) on trees. This
8446 pass only operates on local scalar variables and is enabled by default
8447 at @option{-O} and higher.
8449 @item -fssa-backprop
8450 @opindex fssa-backprop
8451 Propagate information about uses of a value up the definition chain
8452 in order to simplify the definitions. For example, this pass strips
8453 sign operations if the sign of a value never matters. The flag is
8454 enabled by default at @option{-O} and higher.
8457 @opindex fssa-phiopt
8458 Perform pattern matching on SSA PHI nodes to optimize conditional
8459 code. This pass is enabled by default at @option{-O} and higher.
8461 @item -ftree-switch-conversion
8462 @opindex ftree-switch-conversion
8463 Perform conversion of simple initializations in a switch to
8464 initializations from a scalar array. This flag is enabled by default
8465 at @option{-O2} and higher.
8467 @item -ftree-tail-merge
8468 @opindex ftree-tail-merge
8469 Look for identical code sequences. When found, replace one with a jump to the
8470 other. This optimization is known as tail merging or cross jumping. This flag
8471 is enabled by default at @option{-O2} and higher. The compilation time
8473 be limited using @option{max-tail-merge-comparisons} parameter and
8474 @option{max-tail-merge-iterations} parameter.
8478 Perform dead code elimination (DCE) on trees. This flag is enabled by
8479 default at @option{-O} and higher.
8481 @item -ftree-builtin-call-dce
8482 @opindex ftree-builtin-call-dce
8483 Perform conditional dead code elimination (DCE) for calls to built-in functions
8484 that may set @code{errno} but are otherwise side-effect free. This flag is
8485 enabled by default at @option{-O2} and higher if @option{-Os} is not also
8488 @item -ftree-dominator-opts
8489 @opindex ftree-dominator-opts
8490 Perform a variety of simple scalar cleanups (constant/copy
8491 propagation, redundancy elimination, range propagation and expression
8492 simplification) based on a dominator tree traversal. This also
8493 performs jump threading (to reduce jumps to jumps). This flag is
8494 enabled by default at @option{-O} and higher.
8498 Perform dead store elimination (DSE) on trees. A dead store is a store into
8499 a memory location that is later overwritten by another store without
8500 any intervening loads. In this case the earlier store can be deleted. This
8501 flag is enabled by default at @option{-O} and higher.
8505 Perform loop header copying on trees. This is beneficial since it increases
8506 effectiveness of code motion optimizations. It also saves one jump. This flag
8507 is enabled by default at @option{-O} and higher. It is not enabled
8508 for @option{-Os}, since it usually increases code size.
8510 @item -ftree-loop-optimize
8511 @opindex ftree-loop-optimize
8512 Perform loop optimizations on trees. This flag is enabled by default
8513 at @option{-O} and higher.
8515 @item -ftree-loop-linear
8516 @itemx -floop-interchange
8517 @itemx -floop-strip-mine
8519 @itemx -floop-unroll-and-jam
8520 @opindex ftree-loop-linear
8521 @opindex floop-interchange
8522 @opindex floop-strip-mine
8523 @opindex floop-block
8524 @opindex floop-unroll-and-jam
8525 Perform loop nest optimizations. Same as
8526 @option{-floop-nest-optimize}. To use this code transformation, GCC has
8527 to be configured with @option{--with-isl} to enable the Graphite loop
8528 transformation infrastructure.
8530 @item -fgraphite-identity
8531 @opindex fgraphite-identity
8532 Enable the identity transformation for graphite. For every SCoP we generate
8533 the polyhedral representation and transform it back to gimple. Using
8534 @option{-fgraphite-identity} we can check the costs or benefits of the
8535 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
8536 are also performed by the code generator isl, like index splitting and
8537 dead code elimination in loops.
8539 @item -floop-nest-optimize
8540 @opindex floop-nest-optimize
8541 Enable the isl based loop nest optimizer. This is a generic loop nest
8542 optimizer based on the Pluto optimization algorithms. It calculates a loop
8543 structure optimized for data-locality and parallelism. This option
8546 @item -floop-parallelize-all
8547 @opindex floop-parallelize-all
8548 Use the Graphite data dependence analysis to identify loops that can
8549 be parallelized. Parallelize all the loops that can be analyzed to
8550 not contain loop carried dependences without checking that it is
8551 profitable to parallelize the loops.
8553 @item -ftree-coalesce-vars
8554 @opindex ftree-coalesce-vars
8555 While transforming the program out of the SSA representation, attempt to
8556 reduce copying by coalescing versions of different user-defined
8557 variables, instead of just compiler temporaries. This may severely
8558 limit the ability to debug an optimized program compiled with
8559 @option{-fno-var-tracking-assignments}. In the negated form, this flag
8560 prevents SSA coalescing of user variables. This option is enabled by
8561 default if optimization is enabled, and it does very little otherwise.
8563 @item -ftree-loop-if-convert
8564 @opindex ftree-loop-if-convert
8565 Attempt to transform conditional jumps in the innermost loops to
8566 branch-less equivalents. The intent is to remove control-flow from
8567 the innermost loops in order to improve the ability of the
8568 vectorization pass to handle these loops. This is enabled by default
8569 if vectorization is enabled.
8571 @item -ftree-loop-distribution
8572 @opindex ftree-loop-distribution
8573 Perform loop distribution. This flag can improve cache performance on
8574 big loop bodies and allow further loop optimizations, like
8575 parallelization or vectorization, to take place. For example, the loop
8592 @item -ftree-loop-distribute-patterns
8593 @opindex ftree-loop-distribute-patterns
8594 Perform loop distribution of patterns that can be code generated with
8595 calls to a library. This flag is enabled by default at @option{-O3}.
8597 This pass distributes the initialization loops and generates a call to
8598 memset zero. For example, the loop
8614 and the initialization loop is transformed into a call to memset zero.
8616 @item -ftree-loop-im
8617 @opindex ftree-loop-im
8618 Perform loop invariant motion on trees. This pass moves only invariants that
8619 are hard to handle at RTL level (function calls, operations that expand to
8620 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
8621 operands of conditions that are invariant out of the loop, so that we can use
8622 just trivial invariantness analysis in loop unswitching. The pass also includes
8625 @item -ftree-loop-ivcanon
8626 @opindex ftree-loop-ivcanon
8627 Create a canonical counter for number of iterations in loops for which
8628 determining number of iterations requires complicated analysis. Later
8629 optimizations then may determine the number easily. Useful especially
8630 in connection with unrolling.
8634 Perform induction variable optimizations (strength reduction, induction
8635 variable merging and induction variable elimination) on trees.
8637 @item -ftree-parallelize-loops=n
8638 @opindex ftree-parallelize-loops
8639 Parallelize loops, i.e., split their iteration space to run in n threads.
8640 This is only possible for loops whose iterations are independent
8641 and can be arbitrarily reordered. The optimization is only
8642 profitable on multiprocessor machines, for loops that are CPU-intensive,
8643 rather than constrained e.g.@: by memory bandwidth. This option
8644 implies @option{-pthread}, and thus is only supported on targets
8645 that have support for @option{-pthread}.
8649 Perform function-local points-to analysis on trees. This flag is
8650 enabled by default at @option{-O} and higher.
8654 Perform scalar replacement of aggregates. This pass replaces structure
8655 references with scalars to prevent committing structures to memory too
8656 early. This flag is enabled by default at @option{-O} and higher.
8658 @item -fstore-merging
8659 @opindex fstore-merging
8660 Perform merging of narrow stores to consecutive memory addresses. This pass
8661 merges contiguous stores of immediate values narrower than a word into fewer
8662 wider stores to reduce the number of instructions. This is enabled by default
8663 at @option{-O2} and higher as well as @option{-Os}.
8667 Perform temporary expression replacement during the SSA->normal phase. Single
8668 use/single def temporaries are replaced at their use location with their
8669 defining expression. This results in non-GIMPLE code, but gives the expanders
8670 much more complex trees to work on resulting in better RTL generation. This is
8671 enabled by default at @option{-O} and higher.
8675 Perform straight-line strength reduction on trees. This recognizes related
8676 expressions involving multiplications and replaces them by less expensive
8677 calculations when possible. This is enabled by default at @option{-O} and
8680 @item -ftree-vectorize
8681 @opindex ftree-vectorize
8682 Perform vectorization on trees. This flag enables @option{-ftree-loop-vectorize}
8683 and @option{-ftree-slp-vectorize} if not explicitly specified.
8685 @item -ftree-loop-vectorize
8686 @opindex ftree-loop-vectorize
8687 Perform loop vectorization on trees. This flag is enabled by default at
8688 @option{-O3} and when @option{-ftree-vectorize} is enabled.
8690 @item -ftree-slp-vectorize
8691 @opindex ftree-slp-vectorize
8692 Perform basic block vectorization on trees. This flag is enabled by default at
8693 @option{-O3} and when @option{-ftree-vectorize} is enabled.
8695 @item -fvect-cost-model=@var{model}
8696 @opindex fvect-cost-model
8697 Alter the cost model used for vectorization. The @var{model} argument
8698 should be one of @samp{unlimited}, @samp{dynamic} or @samp{cheap}.
8699 With the @samp{unlimited} model the vectorized code-path is assumed
8700 to be profitable while with the @samp{dynamic} model a runtime check
8701 guards the vectorized code-path to enable it only for iteration
8702 counts that will likely execute faster than when executing the original
8703 scalar loop. The @samp{cheap} model disables vectorization of
8704 loops where doing so would be cost prohibitive for example due to
8705 required runtime checks for data dependence or alignment but otherwise
8706 is equal to the @samp{dynamic} model.
8707 The default cost model depends on other optimization flags and is
8708 either @samp{dynamic} or @samp{cheap}.
8710 @item -fsimd-cost-model=@var{model}
8711 @opindex fsimd-cost-model
8712 Alter the cost model used for vectorization of loops marked with the OpenMP
8713 or Cilk Plus simd directive. The @var{model} argument should be one of
8714 @samp{unlimited}, @samp{dynamic}, @samp{cheap}. All values of @var{model}
8715 have the same meaning as described in @option{-fvect-cost-model} and by
8716 default a cost model defined with @option{-fvect-cost-model} is used.
8720 Perform Value Range Propagation on trees. This is similar to the
8721 constant propagation pass, but instead of values, ranges of values are
8722 propagated. This allows the optimizers to remove unnecessary range
8723 checks like array bound checks and null pointer checks. This is
8724 enabled by default at @option{-O2} and higher. Null pointer check
8725 elimination is only done if @option{-fdelete-null-pointer-checks} is
8729 @opindex fsplit-paths
8730 Split paths leading to loop backedges. This can improve dead code
8731 elimination and common subexpression elimination. This is enabled by
8732 default at @option{-O2} and above.
8734 @item -fsplit-ivs-in-unroller
8735 @opindex fsplit-ivs-in-unroller
8736 Enables expression of values of induction variables in later iterations
8737 of the unrolled loop using the value in the first iteration. This breaks
8738 long dependency chains, thus improving efficiency of the scheduling passes.
8740 A combination of @option{-fweb} and CSE is often sufficient to obtain the
8741 same effect. However, that is not reliable in cases where the loop body
8742 is more complicated than a single basic block. It also does not work at all
8743 on some architectures due to restrictions in the CSE pass.
8745 This optimization is enabled by default.
8747 @item -fvariable-expansion-in-unroller
8748 @opindex fvariable-expansion-in-unroller
8749 With this option, the compiler creates multiple copies of some
8750 local variables when unrolling a loop, which can result in superior code.
8752 @item -fpartial-inlining
8753 @opindex fpartial-inlining
8754 Inline parts of functions. This option has any effect only
8755 when inlining itself is turned on by the @option{-finline-functions}
8756 or @option{-finline-small-functions} options.
8758 Enabled at level @option{-O2}.
8760 @item -fpredictive-commoning
8761 @opindex fpredictive-commoning
8762 Perform predictive commoning optimization, i.e., reusing computations
8763 (especially memory loads and stores) performed in previous
8764 iterations of loops.
8766 This option is enabled at level @option{-O3}.
8768 @item -fprefetch-loop-arrays
8769 @opindex fprefetch-loop-arrays
8770 If supported by the target machine, generate instructions to prefetch
8771 memory to improve the performance of loops that access large arrays.
8773 This option may generate better or worse code; results are highly
8774 dependent on the structure of loops within the source code.
8776 Disabled at level @option{-Os}.
8778 @item -fno-printf-return-value
8779 @opindex fno-printf-return-value
8780 Do not substitute constants for known return value of formatted output
8781 functions such as @code{sprintf}, @code{snprintf}, @code{vsprintf}, and
8782 @code{vsnprintf} (but not @code{printf} of @code{fprintf}). This
8783 transformation allows GCC to optimize or even eliminate branches based
8784 on the known return value of these functions called with arguments that
8785 are either constant, or whose values are known to be in a range that
8786 makes determining the exact return value possible. For example, when
8787 @option{-fprintf-return-value} is in effect, both the branch and the
8788 body of the @code{if} statement (but not the call to @code{snprint})
8789 can be optimized away when @code{i} is a 32-bit or smaller integer
8790 because the return value is guaranteed to be at most 8.
8794 if (snprintf (buf, "%08x", i) >= sizeof buf)
8798 The @option{-fprintf-return-value} option relies on other optimizations
8799 and yields best results with @option{-O2}. It works in tandem with the
8800 @option{-Wformat-overflow} and @option{-Wformat-truncation} options.
8801 The @option{-fprintf-return-value} option is enabled by default.
8804 @itemx -fno-peephole2
8805 @opindex fno-peephole
8806 @opindex fno-peephole2
8807 Disable any machine-specific peephole optimizations. The difference
8808 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
8809 are implemented in the compiler; some targets use one, some use the
8810 other, a few use both.
8812 @option{-fpeephole} is enabled by default.
8813 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8815 @item -fno-guess-branch-probability
8816 @opindex fno-guess-branch-probability
8817 Do not guess branch probabilities using heuristics.
8819 GCC uses heuristics to guess branch probabilities if they are
8820 not provided by profiling feedback (@option{-fprofile-arcs}). These
8821 heuristics are based on the control flow graph. If some branch probabilities
8822 are specified by @code{__builtin_expect}, then the heuristics are
8823 used to guess branch probabilities for the rest of the control flow graph,
8824 taking the @code{__builtin_expect} info into account. The interactions
8825 between the heuristics and @code{__builtin_expect} can be complex, and in
8826 some cases, it may be useful to disable the heuristics so that the effects
8827 of @code{__builtin_expect} are easier to understand.
8829 The default is @option{-fguess-branch-probability} at levels
8830 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8832 @item -freorder-blocks
8833 @opindex freorder-blocks
8834 Reorder basic blocks in the compiled function in order to reduce number of
8835 taken branches and improve code locality.
8837 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8839 @item -freorder-blocks-algorithm=@var{algorithm}
8840 @opindex freorder-blocks-algorithm
8841 Use the specified algorithm for basic block reordering. The
8842 @var{algorithm} argument can be @samp{simple}, which does not increase
8843 code size (except sometimes due to secondary effects like alignment),
8844 or @samp{stc}, the ``software trace cache'' algorithm, which tries to
8845 put all often executed code together, minimizing the number of branches
8846 executed by making extra copies of code.
8848 The default is @samp{simple} at levels @option{-O}, @option{-Os}, and
8849 @samp{stc} at levels @option{-O2}, @option{-O3}.
8851 @item -freorder-blocks-and-partition
8852 @opindex freorder-blocks-and-partition
8853 In addition to reordering basic blocks in the compiled function, in order
8854 to reduce number of taken branches, partitions hot and cold basic blocks
8855 into separate sections of the assembly and @file{.o} files, to improve
8856 paging and cache locality performance.
8858 This optimization is automatically turned off in the presence of
8859 exception handling or unwind tables (on targets using setjump/longjump or target specific scheme), for linkonce sections, for functions with a user-defined
8860 section attribute and on any architecture that does not support named
8861 sections. When @option{-fsplit-stack} is used this option is not
8862 enabled by default (to avoid linker errors), but may be enabled
8863 explicitly (if using a working linker).
8865 Enabled for x86 at levels @option{-O2}, @option{-O3}.
8867 @item -freorder-functions
8868 @opindex freorder-functions
8869 Reorder functions in the object file in order to
8870 improve code locality. This is implemented by using special
8871 subsections @code{.text.hot} for most frequently executed functions and
8872 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
8873 the linker so object file format must support named sections and linker must
8874 place them in a reasonable way.
8876 Also profile feedback must be available to make this option effective. See
8877 @option{-fprofile-arcs} for details.
8879 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8881 @item -fstrict-aliasing
8882 @opindex fstrict-aliasing
8883 Allow the compiler to assume the strictest aliasing rules applicable to
8884 the language being compiled. For C (and C++), this activates
8885 optimizations based on the type of expressions. In particular, an
8886 object of one type is assumed never to reside at the same address as an
8887 object of a different type, unless the types are almost the same. For
8888 example, an @code{unsigned int} can alias an @code{int}, but not a
8889 @code{void*} or a @code{double}. A character type may alias any other
8892 @anchor{Type-punning}Pay special attention to code like this:
8905 The practice of reading from a different union member than the one most
8906 recently written to (called ``type-punning'') is common. Even with
8907 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
8908 is accessed through the union type. So, the code above works as
8909 expected. @xref{Structures unions enumerations and bit-fields
8910 implementation}. However, this code might not:
8921 Similarly, access by taking the address, casting the resulting pointer
8922 and dereferencing the result has undefined behavior, even if the cast
8923 uses a union type, e.g.:
8927 return ((union a_union *) &d)->i;
8931 The @option{-fstrict-aliasing} option is enabled at levels
8932 @option{-O2}, @option{-O3}, @option{-Os}.
8934 @item -falign-functions
8935 @itemx -falign-functions=@var{n}
8936 @opindex falign-functions
8937 Align the start of functions to the next power-of-two greater than
8938 @var{n}, skipping up to @var{n} bytes. For instance,
8939 @option{-falign-functions=32} aligns functions to the next 32-byte
8940 boundary, but @option{-falign-functions=24} aligns to the next
8941 32-byte boundary only if this can be done by skipping 23 bytes or less.
8943 @option{-fno-align-functions} and @option{-falign-functions=1} are
8944 equivalent and mean that functions are not aligned.
8946 Some assemblers only support this flag when @var{n} is a power of two;
8947 in that case, it is rounded up.
8949 If @var{n} is not specified or is zero, use a machine-dependent default.
8951 Enabled at levels @option{-O2}, @option{-O3}.
8953 @item -flimit-function-alignment
8954 If this option is enabled, the compiler tries to avoid unnecessarily
8955 overaligning functions. It attempts to instruct the assembler to align
8956 by the amount specified by @option{-falign-functions}, but not to
8957 skip more bytes than the size of the function.
8959 @item -falign-labels
8960 @itemx -falign-labels=@var{n}
8961 @opindex falign-labels
8962 Align all branch targets to a power-of-two boundary, skipping up to
8963 @var{n} bytes like @option{-falign-functions}. This option can easily
8964 make code slower, because it must insert dummy operations for when the
8965 branch target is reached in the usual flow of the code.
8967 @option{-fno-align-labels} and @option{-falign-labels=1} are
8968 equivalent and mean that labels are not aligned.
8970 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
8971 are greater than this value, then their values are used instead.
8973 If @var{n} is not specified or is zero, use a machine-dependent default
8974 which is very likely to be @samp{1}, meaning no alignment.
8976 Enabled at levels @option{-O2}, @option{-O3}.
8979 @itemx -falign-loops=@var{n}
8980 @opindex falign-loops
8981 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
8982 like @option{-falign-functions}. If the loops are
8983 executed many times, this makes up for any execution of the dummy
8986 @option{-fno-align-loops} and @option{-falign-loops=1} are
8987 equivalent and mean that loops are not aligned.
8989 If @var{n} is not specified or is zero, use a machine-dependent default.
8991 Enabled at levels @option{-O2}, @option{-O3}.
8994 @itemx -falign-jumps=@var{n}
8995 @opindex falign-jumps
8996 Align branch targets to a power-of-two boundary, for branch targets
8997 where the targets can only be reached by jumping, skipping up to @var{n}
8998 bytes like @option{-falign-functions}. In this case, no dummy operations
9001 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
9002 equivalent and mean that loops are not aligned.
9004 If @var{n} is not specified or is zero, use a machine-dependent default.
9006 Enabled at levels @option{-O2}, @option{-O3}.
9008 @item -funit-at-a-time
9009 @opindex funit-at-a-time
9010 This option is left for compatibility reasons. @option{-funit-at-a-time}
9011 has no effect, while @option{-fno-unit-at-a-time} implies
9012 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
9016 @item -fno-toplevel-reorder
9017 @opindex fno-toplevel-reorder
9018 Do not reorder top-level functions, variables, and @code{asm}
9019 statements. Output them in the same order that they appear in the
9020 input file. When this option is used, unreferenced static variables
9021 are not removed. This option is intended to support existing code
9022 that relies on a particular ordering. For new code, it is better to
9023 use attributes when possible.
9025 Enabled at level @option{-O0}. When disabled explicitly, it also implies
9026 @option{-fno-section-anchors}, which is otherwise enabled at @option{-O0} on some
9031 Constructs webs as commonly used for register allocation purposes and assign
9032 each web individual pseudo register. This allows the register allocation pass
9033 to operate on pseudos directly, but also strengthens several other optimization
9034 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
9035 however, make debugging impossible, since variables no longer stay in a
9038 Enabled by default with @option{-funroll-loops}.
9040 @item -fwhole-program
9041 @opindex fwhole-program
9042 Assume that the current compilation unit represents the whole program being
9043 compiled. All public functions and variables with the exception of @code{main}
9044 and those merged by attribute @code{externally_visible} become static functions
9045 and in effect are optimized more aggressively by interprocedural optimizers.
9047 This option should not be used in combination with @option{-flto}.
9048 Instead relying on a linker plugin should provide safer and more precise
9051 @item -flto[=@var{n}]
9053 This option runs the standard link-time optimizer. When invoked
9054 with source code, it generates GIMPLE (one of GCC's internal
9055 representations) and writes it to special ELF sections in the object
9056 file. When the object files are linked together, all the function
9057 bodies are read from these ELF sections and instantiated as if they
9058 had been part of the same translation unit.
9060 To use the link-time optimizer, @option{-flto} and optimization
9061 options should be specified at compile time and during the final link.
9062 It is recommended that you compile all the files participating in the
9063 same link with the same options and also specify those options at
9068 gcc -c -O2 -flto foo.c
9069 gcc -c -O2 -flto bar.c
9070 gcc -o myprog -flto -O2 foo.o bar.o
9073 The first two invocations to GCC save a bytecode representation
9074 of GIMPLE into special ELF sections inside @file{foo.o} and
9075 @file{bar.o}. The final invocation reads the GIMPLE bytecode from
9076 @file{foo.o} and @file{bar.o}, merges the two files into a single
9077 internal image, and compiles the result as usual. Since both
9078 @file{foo.o} and @file{bar.o} are merged into a single image, this
9079 causes all the interprocedural analyses and optimizations in GCC to
9080 work across the two files as if they were a single one. This means,
9081 for example, that the inliner is able to inline functions in
9082 @file{bar.o} into functions in @file{foo.o} and vice-versa.
9084 Another (simpler) way to enable link-time optimization is:
9087 gcc -o myprog -flto -O2 foo.c bar.c
9090 The above generates bytecode for @file{foo.c} and @file{bar.c},
9091 merges them together into a single GIMPLE representation and optimizes
9092 them as usual to produce @file{myprog}.
9094 The only important thing to keep in mind is that to enable link-time
9095 optimizations you need to use the GCC driver to perform the link step.
9096 GCC then automatically performs link-time optimization if any of the
9097 objects involved were compiled with the @option{-flto} command-line option.
9099 should specify the optimization options to be used for link-time
9100 optimization though GCC tries to be clever at guessing an
9101 optimization level to use from the options used at compile time
9102 if you fail to specify one at link time. You can always override
9103 the automatic decision to do link-time optimization
9104 by passing @option{-fno-lto} to the link command.
9106 To make whole program optimization effective, it is necessary to make
9107 certain whole program assumptions. The compiler needs to know
9108 what functions and variables can be accessed by libraries and runtime
9109 outside of the link-time optimized unit. When supported by the linker,
9110 the linker plugin (see @option{-fuse-linker-plugin}) passes information
9111 to the compiler about used and externally visible symbols. When
9112 the linker plugin is not available, @option{-fwhole-program} should be
9113 used to allow the compiler to make these assumptions, which leads
9114 to more aggressive optimization decisions.
9116 When @option{-fuse-linker-plugin} is not enabled, when a file is
9117 compiled with @option{-flto}, the generated object file is larger than
9118 a regular object file because it contains GIMPLE bytecodes and the usual
9119 final code (see @option{-ffat-lto-objects}. This means that
9120 object files with LTO information can be linked as normal object
9121 files; if @option{-fno-lto} is passed to the linker, no
9122 interprocedural optimizations are applied. Note that when
9123 @option{-fno-fat-lto-objects} is enabled the compile stage is faster
9124 but you cannot perform a regular, non-LTO link on them.
9126 Additionally, the optimization flags used to compile individual files
9127 are not necessarily related to those used at link time. For instance,
9130 gcc -c -O0 -ffat-lto-objects -flto foo.c
9131 gcc -c -O0 -ffat-lto-objects -flto bar.c
9132 gcc -o myprog -O3 foo.o bar.o
9135 This produces individual object files with unoptimized assembler
9136 code, but the resulting binary @file{myprog} is optimized at
9137 @option{-O3}. If, instead, the final binary is generated with
9138 @option{-fno-lto}, then @file{myprog} is not optimized.
9140 When producing the final binary, GCC only
9141 applies link-time optimizations to those files that contain bytecode.
9142 Therefore, you can mix and match object files and libraries with
9143 GIMPLE bytecodes and final object code. GCC automatically selects
9144 which files to optimize in LTO mode and which files to link without
9147 There are some code generation flags preserved by GCC when
9148 generating bytecodes, as they need to be used during the final link
9149 stage. Generally options specified at link time override those
9150 specified at compile time.
9152 If you do not specify an optimization level option @option{-O} at
9153 link time, then GCC uses the highest optimization level
9154 used when compiling the object files.
9156 Currently, the following options and their settings are taken from
9157 the first object file that explicitly specifies them:
9158 @option{-fPIC}, @option{-fpic}, @option{-fpie}, @option{-fcommon},
9159 @option{-fexceptions}, @option{-fnon-call-exceptions}, @option{-fgnu-tm}
9160 and all the @option{-m} target flags.
9162 Certain ABI-changing flags are required to match in all compilation units,
9163 and trying to override this at link time with a conflicting value
9164 is ignored. This includes options such as @option{-freg-struct-return}
9165 and @option{-fpcc-struct-return}.
9167 Other options such as @option{-ffp-contract}, @option{-fno-strict-overflow},
9168 @option{-fwrapv}, @option{-fno-trapv} or @option{-fno-strict-aliasing}
9169 are passed through to the link stage and merged conservatively for
9170 conflicting translation units. Specifically
9171 @option{-fno-strict-overflow}, @option{-fwrapv} and @option{-fno-trapv} take
9172 precedence; and for example @option{-ffp-contract=off} takes precedence
9173 over @option{-ffp-contract=fast}. You can override them at link time.
9175 If LTO encounters objects with C linkage declared with incompatible
9176 types in separate translation units to be linked together (undefined
9177 behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
9178 issued. The behavior is still undefined at run time. Similar
9179 diagnostics may be raised for other languages.
9181 Another feature of LTO is that it is possible to apply interprocedural
9182 optimizations on files written in different languages:
9187 gfortran -c -flto baz.f90
9188 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
9191 Notice that the final link is done with @command{g++} to get the C++
9192 runtime libraries and @option{-lgfortran} is added to get the Fortran
9193 runtime libraries. In general, when mixing languages in LTO mode, you
9194 should use the same link command options as when mixing languages in a
9195 regular (non-LTO) compilation.
9197 If object files containing GIMPLE bytecode are stored in a library archive, say
9198 @file{libfoo.a}, it is possible to extract and use them in an LTO link if you
9199 are using a linker with plugin support. To create static libraries suitable
9200 for LTO, use @command{gcc-ar} and @command{gcc-ranlib} instead of @command{ar}
9201 and @command{ranlib};
9202 to show the symbols of object files with GIMPLE bytecode, use
9203 @command{gcc-nm}. Those commands require that @command{ar}, @command{ranlib}
9204 and @command{nm} have been compiled with plugin support. At link time, use the the
9205 flag @option{-fuse-linker-plugin} to ensure that the library participates in
9206 the LTO optimization process:
9209 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
9212 With the linker plugin enabled, the linker extracts the needed
9213 GIMPLE files from @file{libfoo.a} and passes them on to the running GCC
9214 to make them part of the aggregated GIMPLE image to be optimized.
9216 If you are not using a linker with plugin support and/or do not
9217 enable the linker plugin, then the objects inside @file{libfoo.a}
9218 are extracted and linked as usual, but they do not participate
9219 in the LTO optimization process. In order to make a static library suitable
9220 for both LTO optimization and usual linkage, compile its object files with
9221 @option{-flto} @option{-ffat-lto-objects}.
9223 Link-time optimizations do not require the presence of the whole program to
9224 operate. If the program does not require any symbols to be exported, it is
9225 possible to combine @option{-flto} and @option{-fwhole-program} to allow
9226 the interprocedural optimizers to use more aggressive assumptions which may
9227 lead to improved optimization opportunities.
9228 Use of @option{-fwhole-program} is not needed when linker plugin is
9229 active (see @option{-fuse-linker-plugin}).
9231 The current implementation of LTO makes no
9232 attempt to generate bytecode that is portable between different
9233 types of hosts. The bytecode files are versioned and there is a
9234 strict version check, so bytecode files generated in one version of
9235 GCC do not work with an older or newer version of GCC.
9237 Link-time optimization does not work well with generation of debugging
9238 information. Combining @option{-flto} with
9239 @option{-g} is currently experimental and expected to produce unexpected
9242 If you specify the optional @var{n}, the optimization and code
9243 generation done at link time is executed in parallel using @var{n}
9244 parallel jobs by utilizing an installed @command{make} program. The
9245 environment variable @env{MAKE} may be used to override the program
9246 used. The default value for @var{n} is 1.
9248 You can also specify @option{-flto=jobserver} to use GNU make's
9249 job server mode to determine the number of parallel jobs. This
9250 is useful when the Makefile calling GCC is already executing in parallel.
9251 You must prepend a @samp{+} to the command recipe in the parent Makefile
9252 for this to work. This option likely only works if @env{MAKE} is
9255 @item -flto-partition=@var{alg}
9256 @opindex flto-partition
9257 Specify the partitioning algorithm used by the link-time optimizer.
9258 The value is either @samp{1to1} to specify a partitioning mirroring
9259 the original source files or @samp{balanced} to specify partitioning
9260 into equally sized chunks (whenever possible) or @samp{max} to create
9261 new partition for every symbol where possible. Specifying @samp{none}
9262 as an algorithm disables partitioning and streaming completely.
9263 The default value is @samp{balanced}. While @samp{1to1} can be used
9264 as an workaround for various code ordering issues, the @samp{max}
9265 partitioning is intended for internal testing only.
9266 The value @samp{one} specifies that exactly one partition should be
9267 used while the value @samp{none} bypasses partitioning and executes
9268 the link-time optimization step directly from the WPA phase.
9270 @item -flto-odr-type-merging
9271 @opindex flto-odr-type-merging
9272 Enable streaming of mangled types names of C++ types and their unification
9273 at link time. This increases size of LTO object files, but enables
9274 diagnostics about One Definition Rule violations.
9276 @item -flto-compression-level=@var{n}
9277 @opindex flto-compression-level
9278 This option specifies the level of compression used for intermediate
9279 language written to LTO object files, and is only meaningful in
9280 conjunction with LTO mode (@option{-flto}). Valid
9281 values are 0 (no compression) to 9 (maximum compression). Values
9282 outside this range are clamped to either 0 or 9. If the option is not
9283 given, a default balanced compression setting is used.
9285 @item -fuse-linker-plugin
9286 @opindex fuse-linker-plugin
9287 Enables the use of a linker plugin during link-time optimization. This
9288 option relies on plugin support in the linker, which is available in gold
9289 or in GNU ld 2.21 or newer.
9291 This option enables the extraction of object files with GIMPLE bytecode out
9292 of library archives. This improves the quality of optimization by exposing
9293 more code to the link-time optimizer. This information specifies what
9294 symbols can be accessed externally (by non-LTO object or during dynamic
9295 linking). Resulting code quality improvements on binaries (and shared
9296 libraries that use hidden visibility) are similar to @option{-fwhole-program}.
9297 See @option{-flto} for a description of the effect of this flag and how to
9300 This option is enabled by default when LTO support in GCC is enabled
9301 and GCC was configured for use with
9302 a linker supporting plugins (GNU ld 2.21 or newer or gold).
9304 @item -ffat-lto-objects
9305 @opindex ffat-lto-objects
9306 Fat LTO objects are object files that contain both the intermediate language
9307 and the object code. This makes them usable for both LTO linking and normal
9308 linking. This option is effective only when compiling with @option{-flto}
9309 and is ignored at link time.
9311 @option{-fno-fat-lto-objects} improves compilation time over plain LTO, but
9312 requires the complete toolchain to be aware of LTO. It requires a linker with
9313 linker plugin support for basic functionality. Additionally,
9314 @command{nm}, @command{ar} and @command{ranlib}
9315 need to support linker plugins to allow a full-featured build environment
9316 (capable of building static libraries etc). GCC provides the @command{gcc-ar},
9317 @command{gcc-nm}, @command{gcc-ranlib} wrappers to pass the right options
9318 to these tools. With non fat LTO makefiles need to be modified to use them.
9320 The default is @option{-fno-fat-lto-objects} on targets with linker plugin
9323 @item -fcompare-elim
9324 @opindex fcompare-elim
9325 After register allocation and post-register allocation instruction splitting,
9326 identify arithmetic instructions that compute processor flags similar to a
9327 comparison operation based on that arithmetic. If possible, eliminate the
9328 explicit comparison operation.
9330 This pass only applies to certain targets that cannot explicitly represent
9331 the comparison operation before register allocation is complete.
9333 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
9335 @item -fcprop-registers
9336 @opindex fcprop-registers
9337 After register allocation and post-register allocation instruction splitting,
9338 perform a copy-propagation pass to try to reduce scheduling dependencies
9339 and occasionally eliminate the copy.
9341 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
9343 @item -fprofile-correction
9344 @opindex fprofile-correction
9345 Profiles collected using an instrumented binary for multi-threaded programs may
9346 be inconsistent due to missed counter updates. When this option is specified,
9347 GCC uses heuristics to correct or smooth out such inconsistencies. By
9348 default, GCC emits an error message when an inconsistent profile is detected.
9351 @itemx -fprofile-use=@var{path}
9352 @opindex fprofile-use
9353 Enable profile feedback-directed optimizations,
9354 and the following optimizations
9355 which are generally profitable only with profile feedback available:
9356 @option{-fbranch-probabilities}, @option{-fvpt},
9357 @option{-funroll-loops}, @option{-fpeel-loops}, @option{-ftracer},
9358 @option{-ftree-vectorize}, and @option{ftree-loop-distribute-patterns}.
9360 Before you can use this option, you must first generate profiling information.
9361 @xref{Instrumentation Options}, for information about the
9362 @option{-fprofile-generate} option.
9364 By default, GCC emits an error message if the feedback profiles do not
9365 match the source code. This error can be turned into a warning by using
9366 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
9369 If @var{path} is specified, GCC looks at the @var{path} to find
9370 the profile feedback data files. See @option{-fprofile-dir}.
9372 @item -fauto-profile
9373 @itemx -fauto-profile=@var{path}
9374 @opindex fauto-profile
9375 Enable sampling-based feedback-directed optimizations,
9376 and the following optimizations
9377 which are generally profitable only with profile feedback available:
9378 @option{-fbranch-probabilities}, @option{-fvpt},
9379 @option{-funroll-loops}, @option{-fpeel-loops}, @option{-ftracer},
9380 @option{-ftree-vectorize},
9381 @option{-finline-functions}, @option{-fipa-cp}, @option{-fipa-cp-clone},
9382 @option{-fpredictive-commoning}, @option{-funswitch-loops},
9383 @option{-fgcse-after-reload}, and @option{-ftree-loop-distribute-patterns}.
9385 @var{path} is the name of a file containing AutoFDO profile information.
9386 If omitted, it defaults to @file{fbdata.afdo} in the current directory.
9388 Producing an AutoFDO profile data file requires running your program
9389 with the @command{perf} utility on a supported GNU/Linux target system.
9390 For more information, see @uref{https://perf.wiki.kernel.org/}.
9394 perf record -e br_inst_retired:near_taken -b -o perf.data \
9398 Then use the @command{create_gcov} tool to convert the raw profile data
9399 to a format that can be used by GCC.@ You must also supply the
9400 unstripped binary for your program to this tool.
9401 See @uref{https://github.com/google/autofdo}.
9405 create_gcov --binary=your_program.unstripped --profile=perf.data \
9410 The following options control compiler behavior regarding floating-point
9411 arithmetic. These options trade off between speed and
9412 correctness. All must be specifically enabled.
9416 @opindex ffloat-store
9417 Do not store floating-point variables in registers, and inhibit other
9418 options that might change whether a floating-point value is taken from a
9421 @cindex floating-point precision
9422 This option prevents undesirable excess precision on machines such as
9423 the 68000 where the floating registers (of the 68881) keep more
9424 precision than a @code{double} is supposed to have. Similarly for the
9425 x86 architecture. For most programs, the excess precision does only
9426 good, but a few programs rely on the precise definition of IEEE floating
9427 point. Use @option{-ffloat-store} for such programs, after modifying
9428 them to store all pertinent intermediate computations into variables.
9430 @item -fexcess-precision=@var{style}
9431 @opindex fexcess-precision
9432 This option allows further control over excess precision on machines
9433 where floating-point operations occur in a format with more precision or
9434 range than the IEEE standard and interchange floating-point types. By
9435 default, @option{-fexcess-precision=fast} is in effect; this means that
9436 operations may be carried out in a wider precision than the types specified
9437 in the source if that would result in faster code, and it is unpredictable
9438 when rounding to the types specified in the source code takes place.
9439 When compiling C, if @option{-fexcess-precision=standard} is specified then
9440 excess precision follows the rules specified in ISO C99; in particular,
9441 both casts and assignments cause values to be rounded to their
9442 semantic types (whereas @option{-ffloat-store} only affects
9443 assignments). This option is enabled by default for C if a strict
9444 conformance option such as @option{-std=c99} is used.
9445 @option{-ffast-math} enables @option{-fexcess-precision=fast} by default
9446 regardless of whether a strict conformance option is used.
9449 @option{-fexcess-precision=standard} is not implemented for languages
9450 other than C. On the x86, it has no effect if @option{-mfpmath=sse}
9451 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
9452 semantics apply without excess precision, and in the latter, rounding
9457 Sets the options @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
9458 @option{-ffinite-math-only}, @option{-fno-rounding-math},
9459 @option{-fno-signaling-nans}, @option{-fcx-limited-range} and
9460 @option{-fexcess-precision=fast}.
9462 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
9464 This option is not turned on by any @option{-O} option besides
9465 @option{-Ofast} since it can result in incorrect output for programs
9466 that depend on an exact implementation of IEEE or ISO rules/specifications
9467 for math functions. It may, however, yield faster code for programs
9468 that do not require the guarantees of these specifications.
9470 @item -fno-math-errno
9471 @opindex fno-math-errno
9472 Do not set @code{errno} after calling math functions that are executed
9473 with a single instruction, e.g., @code{sqrt}. A program that relies on
9474 IEEE exceptions for math error handling may want to use this flag
9475 for speed while maintaining IEEE arithmetic compatibility.
9477 This option is not turned on by any @option{-O} option since
9478 it can result in incorrect output for programs that depend on
9479 an exact implementation of IEEE or ISO rules/specifications for
9480 math functions. It may, however, yield faster code for programs
9481 that do not require the guarantees of these specifications.
9483 The default is @option{-fmath-errno}.
9485 On Darwin systems, the math library never sets @code{errno}. There is
9486 therefore no reason for the compiler to consider the possibility that
9487 it might, and @option{-fno-math-errno} is the default.
9489 @item -funsafe-math-optimizations
9490 @opindex funsafe-math-optimizations
9492 Allow optimizations for floating-point arithmetic that (a) assume
9493 that arguments and results are valid and (b) may violate IEEE or
9494 ANSI standards. When used at link time, it may include libraries
9495 or startup files that change the default FPU control word or other
9496 similar optimizations.
9498 This option is not turned on by any @option{-O} option since
9499 it can result in incorrect output for programs that depend on
9500 an exact implementation of IEEE or ISO rules/specifications for
9501 math functions. It may, however, yield faster code for programs
9502 that do not require the guarantees of these specifications.
9503 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
9504 @option{-fassociative-math} and @option{-freciprocal-math}.
9506 The default is @option{-fno-unsafe-math-optimizations}.
9508 @item -fassociative-math
9509 @opindex fassociative-math
9511 Allow re-association of operands in series of floating-point operations.
9512 This violates the ISO C and C++ language standard by possibly changing
9513 computation result. NOTE: re-ordering may change the sign of zero as
9514 well as ignore NaNs and inhibit or create underflow or overflow (and
9515 thus cannot be used on code that relies on rounding behavior like
9516 @code{(x + 2**52) - 2**52}. May also reorder floating-point comparisons
9517 and thus may not be used when ordered comparisons are required.
9518 This option requires that both @option{-fno-signed-zeros} and
9519 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
9520 much sense with @option{-frounding-math}. For Fortran the option
9521 is automatically enabled when both @option{-fno-signed-zeros} and
9522 @option{-fno-trapping-math} are in effect.
9524 The default is @option{-fno-associative-math}.
9526 @item -freciprocal-math
9527 @opindex freciprocal-math
9529 Allow the reciprocal of a value to be used instead of dividing by
9530 the value if this enables optimizations. For example @code{x / y}
9531 can be replaced with @code{x * (1/y)}, which is useful if @code{(1/y)}
9532 is subject to common subexpression elimination. Note that this loses
9533 precision and increases the number of flops operating on the value.
9535 The default is @option{-fno-reciprocal-math}.
9537 @item -ffinite-math-only
9538 @opindex ffinite-math-only
9539 Allow optimizations for floating-point arithmetic that assume
9540 that arguments and results are not NaNs or +-Infs.
9542 This option is not turned on by any @option{-O} option since
9543 it can result in incorrect output for programs that depend on
9544 an exact implementation of IEEE or ISO rules/specifications for
9545 math functions. It may, however, yield faster code for programs
9546 that do not require the guarantees of these specifications.
9548 The default is @option{-fno-finite-math-only}.
9550 @item -fno-signed-zeros
9551 @opindex fno-signed-zeros
9552 Allow optimizations for floating-point arithmetic that ignore the
9553 signedness of zero. IEEE arithmetic specifies the behavior of
9554 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
9555 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
9556 This option implies that the sign of a zero result isn't significant.
9558 The default is @option{-fsigned-zeros}.
9560 @item -fno-trapping-math
9561 @opindex fno-trapping-math
9562 Compile code assuming that floating-point operations cannot generate
9563 user-visible traps. These traps include division by zero, overflow,
9564 underflow, inexact result and invalid operation. This option requires
9565 that @option{-fno-signaling-nans} be in effect. Setting this option may
9566 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
9568 This option should never be turned on by any @option{-O} option since
9569 it can result in incorrect output for programs that depend on
9570 an exact implementation of IEEE or ISO rules/specifications for
9573 The default is @option{-ftrapping-math}.
9575 @item -frounding-math
9576 @opindex frounding-math
9577 Disable transformations and optimizations that assume default floating-point
9578 rounding behavior. This is round-to-zero for all floating point
9579 to integer conversions, and round-to-nearest for all other arithmetic
9580 truncations. This option should be specified for programs that change
9581 the FP rounding mode dynamically, or that may be executed with a
9582 non-default rounding mode. This option disables constant folding of
9583 floating-point expressions at compile time (which may be affected by
9584 rounding mode) and arithmetic transformations that are unsafe in the
9585 presence of sign-dependent rounding modes.
9587 The default is @option{-fno-rounding-math}.
9589 This option is experimental and does not currently guarantee to
9590 disable all GCC optimizations that are affected by rounding mode.
9591 Future versions of GCC may provide finer control of this setting
9592 using C99's @code{FENV_ACCESS} pragma. This command-line option
9593 will be used to specify the default state for @code{FENV_ACCESS}.
9595 @item -fsignaling-nans
9596 @opindex fsignaling-nans
9597 Compile code assuming that IEEE signaling NaNs may generate user-visible
9598 traps during floating-point operations. Setting this option disables
9599 optimizations that may change the number of exceptions visible with
9600 signaling NaNs. This option implies @option{-ftrapping-math}.
9602 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
9605 The default is @option{-fno-signaling-nans}.
9607 This option is experimental and does not currently guarantee to
9608 disable all GCC optimizations that affect signaling NaN behavior.
9610 @item -fno-fp-int-builtin-inexact
9611 @opindex fno-fp-int-builtin-inexact
9612 Do not allow the built-in functions @code{ceil}, @code{floor},
9613 @code{round} and @code{trunc}, and their @code{float} and @code{long
9614 double} variants, to generate code that raises the ``inexact''
9615 floating-point exception for noninteger arguments. ISO C99 and C11
9616 allow these functions to raise the ``inexact'' exception, but ISO/IEC
9617 TS 18661-1:2014, the C bindings to IEEE 754-2008, does not allow these
9620 The default is @option{-ffp-int-builtin-inexact}, allowing the
9621 exception to be raised. This option does nothing unless
9622 @option{-ftrapping-math} is in effect.
9624 Even if @option{-fno-fp-int-builtin-inexact} is used, if the functions
9625 generate a call to a library function then the ``inexact'' exception
9626 may be raised if the library implementation does not follow TS 18661.
9628 @item -fsingle-precision-constant
9629 @opindex fsingle-precision-constant
9630 Treat floating-point constants as single precision instead of
9631 implicitly converting them to double-precision constants.
9633 @item -fcx-limited-range
9634 @opindex fcx-limited-range
9635 When enabled, this option states that a range reduction step is not
9636 needed when performing complex division. Also, there is no checking
9637 whether the result of a complex multiplication or division is @code{NaN
9638 + I*NaN}, with an attempt to rescue the situation in that case. The
9639 default is @option{-fno-cx-limited-range}, but is enabled by
9640 @option{-ffast-math}.
9642 This option controls the default setting of the ISO C99
9643 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
9646 @item -fcx-fortran-rules
9647 @opindex fcx-fortran-rules
9648 Complex multiplication and division follow Fortran rules. Range
9649 reduction is done as part of complex division, but there is no checking
9650 whether the result of a complex multiplication or division is @code{NaN
9651 + I*NaN}, with an attempt to rescue the situation in that case.
9653 The default is @option{-fno-cx-fortran-rules}.
9657 The following options control optimizations that may improve
9658 performance, but are not enabled by any @option{-O} options. This
9659 section includes experimental options that may produce broken code.
9662 @item -fbranch-probabilities
9663 @opindex fbranch-probabilities
9664 After running a program compiled with @option{-fprofile-arcs}
9665 (@pxref{Instrumentation Options}),
9666 you can compile it a second time using
9667 @option{-fbranch-probabilities}, to improve optimizations based on
9668 the number of times each branch was taken. When a program
9669 compiled with @option{-fprofile-arcs} exits, it saves arc execution
9670 counts to a file called @file{@var{sourcename}.gcda} for each source
9671 file. The information in this data file is very dependent on the
9672 structure of the generated code, so you must use the same source code
9673 and the same optimization options for both compilations.
9675 With @option{-fbranch-probabilities}, GCC puts a
9676 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
9677 These can be used to improve optimization. Currently, they are only
9678 used in one place: in @file{reorg.c}, instead of guessing which path a
9679 branch is most likely to take, the @samp{REG_BR_PROB} values are used to
9680 exactly determine which path is taken more often.
9682 @item -fprofile-values
9683 @opindex fprofile-values
9684 If combined with @option{-fprofile-arcs}, it adds code so that some
9685 data about values of expressions in the program is gathered.
9687 With @option{-fbranch-probabilities}, it reads back the data gathered
9688 from profiling values of expressions for usage in optimizations.
9690 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
9692 @item -fprofile-reorder-functions
9693 @opindex fprofile-reorder-functions
9694 Function reordering based on profile instrumentation collects
9695 first time of execution of a function and orders these functions
9698 Enabled with @option{-fprofile-use}.
9702 If combined with @option{-fprofile-arcs}, this option instructs the compiler
9703 to add code to gather information about values of expressions.
9705 With @option{-fbranch-probabilities}, it reads back the data gathered
9706 and actually performs the optimizations based on them.
9707 Currently the optimizations include specialization of division operations
9708 using the knowledge about the value of the denominator.
9710 @item -frename-registers
9711 @opindex frename-registers
9712 Attempt to avoid false dependencies in scheduled code by making use
9713 of registers left over after register allocation. This optimization
9714 most benefits processors with lots of registers. Depending on the
9715 debug information format adopted by the target, however, it can
9716 make debugging impossible, since variables no longer stay in
9717 a ``home register''.
9719 Enabled by default with @option{-funroll-loops}.
9721 @item -fschedule-fusion
9722 @opindex fschedule-fusion
9723 Performs a target dependent pass over the instruction stream to schedule
9724 instructions of same type together because target machine can execute them
9725 more efficiently if they are adjacent to each other in the instruction flow.
9727 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
9731 Perform tail duplication to enlarge superblock size. This transformation
9732 simplifies the control flow of the function allowing other optimizations to do
9735 Enabled with @option{-fprofile-use}.
9737 @item -funroll-loops
9738 @opindex funroll-loops
9739 Unroll loops whose number of iterations can be determined at compile time or
9740 upon entry to the loop. @option{-funroll-loops} implies
9741 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
9742 It also turns on complete loop peeling (i.e.@: complete removal of loops with
9743 a small constant number of iterations). This option makes code larger, and may
9744 or may not make it run faster.
9746 Enabled with @option{-fprofile-use}.
9748 @item -funroll-all-loops
9749 @opindex funroll-all-loops
9750 Unroll all loops, even if their number of iterations is uncertain when
9751 the loop is entered. This usually makes programs run more slowly.
9752 @option{-funroll-all-loops} implies the same options as
9753 @option{-funroll-loops}.
9756 @opindex fpeel-loops
9757 Peels loops for which there is enough information that they do not
9758 roll much (from profile feedback or static analysis). It also turns on
9759 complete loop peeling (i.e.@: complete removal of loops with small constant
9760 number of iterations).
9762 Enabled with @option{-O3} and/or @option{-fprofile-use}.
9764 @item -fmove-loop-invariants
9765 @opindex fmove-loop-invariants
9766 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
9767 at level @option{-O1}
9770 @opindex fsplit-loops
9771 Split a loop into two if it contains a condition that's always true
9772 for one side of the iteration space and false for the other.
9774 @item -funswitch-loops
9775 @opindex funswitch-loops
9776 Move branches with loop invariant conditions out of the loop, with duplicates
9777 of the loop on both branches (modified according to result of the condition).
9779 @item -ffunction-sections
9780 @itemx -fdata-sections
9781 @opindex ffunction-sections
9782 @opindex fdata-sections
9783 Place each function or data item into its own section in the output
9784 file if the target supports arbitrary sections. The name of the
9785 function or the name of the data item determines the section's name
9788 Use these options on systems where the linker can perform optimizations to
9789 improve locality of reference in the instruction space. Most systems using the
9790 ELF object format have linkers with such optimizations. On AIX, the linker
9791 rearranges sections (CSECTs) based on the call graph. The performance impact
9794 Together with a linker garbage collection (linker @option{--gc-sections}
9795 option) these options may lead to smaller statically-linked executables (after
9798 On ELF/DWARF systems these options do not degenerate the quality of the debug
9799 information. There could be issues with other object files/debug info formats.
9801 Only use these options when there are significant benefits from doing so. When
9802 you specify these options, the assembler and linker create larger object and
9803 executable files and are also slower. These options affect code generation.
9804 They prevent optimizations by the compiler and assembler using relative
9805 locations inside a translation unit since the locations are unknown until
9806 link time. An example of such an optimization is relaxing calls to short call
9809 @item -fbranch-target-load-optimize
9810 @opindex fbranch-target-load-optimize
9811 Perform branch target register load optimization before prologue / epilogue
9813 The use of target registers can typically be exposed only during reload,
9814 thus hoisting loads out of loops and doing inter-block scheduling needs
9815 a separate optimization pass.
9817 @item -fbranch-target-load-optimize2
9818 @opindex fbranch-target-load-optimize2
9819 Perform branch target register load optimization after prologue / epilogue
9822 @item -fbtr-bb-exclusive
9823 @opindex fbtr-bb-exclusive
9824 When performing branch target register load optimization, don't reuse
9825 branch target registers within any basic block.
9828 @opindex fstdarg-opt
9829 Optimize the prologue of variadic argument functions with respect to usage of
9832 @item -fsection-anchors
9833 @opindex fsection-anchors
9834 Try to reduce the number of symbolic address calculations by using
9835 shared ``anchor'' symbols to address nearby objects. This transformation
9836 can help to reduce the number of GOT entries and GOT accesses on some
9839 For example, the implementation of the following function @code{foo}:
9843 int foo (void) @{ return a + b + c; @}
9847 usually calculates the addresses of all three variables, but if you
9848 compile it with @option{-fsection-anchors}, it accesses the variables
9849 from a common anchor point instead. The effect is similar to the
9850 following pseudocode (which isn't valid C):
9855 register int *xr = &x;
9856 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
9860 Not all targets support this option.
9862 @item --param @var{name}=@var{value}
9864 In some places, GCC uses various constants to control the amount of
9865 optimization that is done. For example, GCC does not inline functions
9866 that contain more than a certain number of instructions. You can
9867 control some of these constants on the command line using the
9868 @option{--param} option.
9870 The names of specific parameters, and the meaning of the values, are
9871 tied to the internals of the compiler, and are subject to change
9872 without notice in future releases.
9874 In each case, the @var{value} is an integer. The allowable choices for
9878 @item predictable-branch-outcome
9879 When branch is predicted to be taken with probability lower than this threshold
9880 (in percent), then it is considered well predictable. The default is 10.
9882 @item max-rtl-if-conversion-insns
9883 RTL if-conversion tries to remove conditional branches around a block and
9884 replace them with conditionally executed instructions. This parameter
9885 gives the maximum number of instructions in a block which should be
9886 considered for if-conversion. The default is 10, though the compiler will
9887 also use other heuristics to decide whether if-conversion is likely to be
9890 @item max-rtl-if-conversion-predictable-cost
9891 @item max-rtl-if-conversion-unpredictable-cost
9892 RTL if-conversion will try to remove conditional branches around a block
9893 and replace them with conditionally executed instructions. These parameters
9894 give the maximum permissible cost for the sequence that would be generated
9895 by if-conversion depending on whether the branch is statically determined
9896 to be predictable or not. The units for this parameter are the same as
9897 those for the GCC internal seq_cost metric. The compiler will try to
9898 provide a reasonable default for this parameter using the BRANCH_COST
9901 @item max-crossjump-edges
9902 The maximum number of incoming edges to consider for cross-jumping.
9903 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
9904 the number of edges incoming to each block. Increasing values mean
9905 more aggressive optimization, making the compilation time increase with
9906 probably small improvement in executable size.
9908 @item min-crossjump-insns
9909 The minimum number of instructions that must be matched at the end
9910 of two blocks before cross-jumping is performed on them. This
9911 value is ignored in the case where all instructions in the block being
9912 cross-jumped from are matched. The default value is 5.
9914 @item max-grow-copy-bb-insns
9915 The maximum code size expansion factor when copying basic blocks
9916 instead of jumping. The expansion is relative to a jump instruction.
9917 The default value is 8.
9919 @item max-goto-duplication-insns
9920 The maximum number of instructions to duplicate to a block that jumps
9921 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
9922 passes, GCC factors computed gotos early in the compilation process,
9923 and unfactors them as late as possible. Only computed jumps at the
9924 end of a basic blocks with no more than max-goto-duplication-insns are
9925 unfactored. The default value is 8.
9927 @item max-delay-slot-insn-search
9928 The maximum number of instructions to consider when looking for an
9929 instruction to fill a delay slot. If more than this arbitrary number of
9930 instructions are searched, the time savings from filling the delay slot
9931 are minimal, so stop searching. Increasing values mean more
9932 aggressive optimization, making the compilation time increase with probably
9933 small improvement in execution time.
9935 @item max-delay-slot-live-search
9936 When trying to fill delay slots, the maximum number of instructions to
9937 consider when searching for a block with valid live register
9938 information. Increasing this arbitrarily chosen value means more
9939 aggressive optimization, increasing the compilation time. This parameter
9940 should be removed when the delay slot code is rewritten to maintain the
9943 @item max-gcse-memory
9944 The approximate maximum amount of memory that can be allocated in
9945 order to perform the global common subexpression elimination
9946 optimization. If more memory than specified is required, the
9947 optimization is not done.
9949 @item max-gcse-insertion-ratio
9950 If the ratio of expression insertions to deletions is larger than this value
9951 for any expression, then RTL PRE inserts or removes the expression and thus
9952 leaves partially redundant computations in the instruction stream. The default value is 20.
9954 @item max-pending-list-length
9955 The maximum number of pending dependencies scheduling allows
9956 before flushing the current state and starting over. Large functions
9957 with few branches or calls can create excessively large lists which
9958 needlessly consume memory and resources.
9960 @item max-modulo-backtrack-attempts
9961 The maximum number of backtrack attempts the scheduler should make
9962 when modulo scheduling a loop. Larger values can exponentially increase
9965 @item max-inline-insns-single
9966 Several parameters control the tree inliner used in GCC@.
9967 This number sets the maximum number of instructions (counted in GCC's
9968 internal representation) in a single function that the tree inliner
9969 considers for inlining. This only affects functions declared
9970 inline and methods implemented in a class declaration (C++).
9971 The default value is 400.
9973 @item max-inline-insns-auto
9974 When you use @option{-finline-functions} (included in @option{-O3}),
9975 a lot of functions that would otherwise not be considered for inlining
9976 by the compiler are investigated. To those functions, a different
9977 (more restrictive) limit compared to functions declared inline can
9979 The default value is 40.
9981 @item inline-min-speedup
9982 When estimated performance improvement of caller + callee runtime exceeds this
9983 threshold (in percent), the function can be inlined regardless of the limit on
9984 @option{--param max-inline-insns-single} and @option{--param
9985 max-inline-insns-auto}.
9987 @item large-function-insns
9988 The limit specifying really large functions. For functions larger than this
9989 limit after inlining, inlining is constrained by
9990 @option{--param large-function-growth}. This parameter is useful primarily
9991 to avoid extreme compilation time caused by non-linear algorithms used by the
9993 The default value is 2700.
9995 @item large-function-growth
9996 Specifies maximal growth of large function caused by inlining in percents.
9997 The default value is 100 which limits large function growth to 2.0 times
10000 @item large-unit-insns
10001 The limit specifying large translation unit. Growth caused by inlining of
10002 units larger than this limit is limited by @option{--param inline-unit-growth}.
10003 For small units this might be too tight.
10004 For example, consider a unit consisting of function A
10005 that is inline and B that just calls A three times. If B is small relative to
10006 A, the growth of unit is 300\% and yet such inlining is very sane. For very
10007 large units consisting of small inlineable functions, however, the overall unit
10008 growth limit is needed to avoid exponential explosion of code size. Thus for
10009 smaller units, the size is increased to @option{--param large-unit-insns}
10010 before applying @option{--param inline-unit-growth}. The default is 10000.
10012 @item inline-unit-growth
10013 Specifies maximal overall growth of the compilation unit caused by inlining.
10014 The default value is 20 which limits unit growth to 1.2 times the original
10015 size. Cold functions (either marked cold via an attribute or by profile
10016 feedback) are not accounted into the unit size.
10018 @item ipcp-unit-growth
10019 Specifies maximal overall growth of the compilation unit caused by
10020 interprocedural constant propagation. The default value is 10 which limits
10021 unit growth to 1.1 times the original size.
10023 @item large-stack-frame
10024 The limit specifying large stack frames. While inlining the algorithm is trying
10025 to not grow past this limit too much. The default value is 256 bytes.
10027 @item large-stack-frame-growth
10028 Specifies maximal growth of large stack frames caused by inlining in percents.
10029 The default value is 1000 which limits large stack frame growth to 11 times
10032 @item max-inline-insns-recursive
10033 @itemx max-inline-insns-recursive-auto
10034 Specifies the maximum number of instructions an out-of-line copy of a
10035 self-recursive inline
10036 function can grow into by performing recursive inlining.
10038 @option{--param max-inline-insns-recursive} applies to functions
10040 For functions not declared inline, recursive inlining
10041 happens only when @option{-finline-functions} (included in @option{-O3}) is
10042 enabled; @option{--param max-inline-insns-recursive-auto} applies instead. The
10043 default value is 450.
10045 @item max-inline-recursive-depth
10046 @itemx max-inline-recursive-depth-auto
10047 Specifies the maximum recursion depth used for recursive inlining.
10049 @option{--param max-inline-recursive-depth} applies to functions
10050 declared inline. For functions not declared inline, recursive inlining
10051 happens only when @option{-finline-functions} (included in @option{-O3}) is
10052 enabled; @option{--param max-inline-recursive-depth-auto} applies instead. The
10053 default value is 8.
10055 @item min-inline-recursive-probability
10056 Recursive inlining is profitable only for function having deep recursion
10057 in average and can hurt for function having little recursion depth by
10058 increasing the prologue size or complexity of function body to other
10061 When profile feedback is available (see @option{-fprofile-generate}) the actual
10062 recursion depth can be guessed from the probability that function recurses
10063 via a given call expression. This parameter limits inlining only to call
10064 expressions whose probability exceeds the given threshold (in percents).
10065 The default value is 10.
10067 @item early-inlining-insns
10068 Specify growth that the early inliner can make. In effect it increases
10069 the amount of inlining for code having a large abstraction penalty.
10070 The default value is 14.
10072 @item max-early-inliner-iterations
10073 Limit of iterations of the early inliner. This basically bounds
10074 the number of nested indirect calls the early inliner can resolve.
10075 Deeper chains are still handled by late inlining.
10077 @item comdat-sharing-probability
10078 Probability (in percent) that C++ inline function with comdat visibility
10079 are shared across multiple compilation units. The default value is 20.
10081 @item profile-func-internal-id
10082 A parameter to control whether to use function internal id in profile
10083 database lookup. If the value is 0, the compiler uses an id that
10084 is based on function assembler name and filename, which makes old profile
10085 data more tolerant to source changes such as function reordering etc.
10086 The default value is 0.
10088 @item min-vect-loop-bound
10089 The minimum number of iterations under which loops are not vectorized
10090 when @option{-ftree-vectorize} is used. The number of iterations after
10091 vectorization needs to be greater than the value specified by this option
10092 to allow vectorization. The default value is 0.
10094 @item gcse-cost-distance-ratio
10095 Scaling factor in calculation of maximum distance an expression
10096 can be moved by GCSE optimizations. This is currently supported only in the
10097 code hoisting pass. The bigger the ratio, the more aggressive code hoisting
10098 is with simple expressions, i.e., the expressions that have cost
10099 less than @option{gcse-unrestricted-cost}. Specifying 0 disables
10100 hoisting of simple expressions. The default value is 10.
10102 @item gcse-unrestricted-cost
10103 Cost, roughly measured as the cost of a single typical machine
10104 instruction, at which GCSE optimizations do not constrain
10105 the distance an expression can travel. This is currently
10106 supported only in the code hoisting pass. The lesser the cost,
10107 the more aggressive code hoisting is. Specifying 0
10108 allows all expressions to travel unrestricted distances.
10109 The default value is 3.
10111 @item max-hoist-depth
10112 The depth of search in the dominator tree for expressions to hoist.
10113 This is used to avoid quadratic behavior in hoisting algorithm.
10114 The value of 0 does not limit on the search, but may slow down compilation
10115 of huge functions. The default value is 30.
10117 @item max-tail-merge-comparisons
10118 The maximum amount of similar bbs to compare a bb with. This is used to
10119 avoid quadratic behavior in tree tail merging. The default value is 10.
10121 @item max-tail-merge-iterations
10122 The maximum amount of iterations of the pass over the function. This is used to
10123 limit compilation time in tree tail merging. The default value is 2.
10125 @item store-merging-allow-unaligned
10126 Allow the store merging pass to introduce unaligned stores if it is legal to
10127 do so. The default value is 1.
10129 @item max-stores-to-merge
10130 The maximum number of stores to attempt to merge into wider stores in the store
10131 merging pass. The minimum value is 2 and the default is 64.
10133 @item max-unrolled-insns
10134 The maximum number of instructions that a loop may have to be unrolled.
10135 If a loop is unrolled, this parameter also determines how many times
10136 the loop code is unrolled.
10138 @item max-average-unrolled-insns
10139 The maximum number of instructions biased by probabilities of their execution
10140 that a loop may have to be unrolled. If a loop is unrolled,
10141 this parameter also determines how many times the loop code is unrolled.
10143 @item max-unroll-times
10144 The maximum number of unrollings of a single loop.
10146 @item max-peeled-insns
10147 The maximum number of instructions that a loop may have to be peeled.
10148 If a loop is peeled, this parameter also determines how many times
10149 the loop code is peeled.
10151 @item max-peel-times
10152 The maximum number of peelings of a single loop.
10154 @item max-peel-branches
10155 The maximum number of branches on the hot path through the peeled sequence.
10157 @item max-completely-peeled-insns
10158 The maximum number of insns of a completely peeled loop.
10160 @item max-completely-peel-times
10161 The maximum number of iterations of a loop to be suitable for complete peeling.
10163 @item max-completely-peel-loop-nest-depth
10164 The maximum depth of a loop nest suitable for complete peeling.
10166 @item max-unswitch-insns
10167 The maximum number of insns of an unswitched loop.
10169 @item max-unswitch-level
10170 The maximum number of branches unswitched in a single loop.
10172 @item max-loop-headers-insns
10173 The maximum number of insns in loop header duplicated by the copy loop headers
10176 @item lim-expensive
10177 The minimum cost of an expensive expression in the loop invariant motion.
10179 @item iv-consider-all-candidates-bound
10180 Bound on number of candidates for induction variables, below which
10181 all candidates are considered for each use in induction variable
10182 optimizations. If there are more candidates than this,
10183 only the most relevant ones are considered to avoid quadratic time complexity.
10185 @item iv-max-considered-uses
10186 The induction variable optimizations give up on loops that contain more
10187 induction variable uses.
10189 @item iv-always-prune-cand-set-bound
10190 If the number of candidates in the set is smaller than this value,
10191 always try to remove unnecessary ivs from the set
10192 when adding a new one.
10194 @item avg-loop-niter
10195 Average number of iterations of a loop.
10197 @item dse-max-object-size
10198 Maximum size (in bytes) of objects tracked bytewise by dead store elimination.
10199 Larger values may result in larger compilation times.
10201 @item scev-max-expr-size
10202 Bound on size of expressions used in the scalar evolutions analyzer.
10203 Large expressions slow the analyzer.
10205 @item scev-max-expr-complexity
10206 Bound on the complexity of the expressions in the scalar evolutions analyzer.
10207 Complex expressions slow the analyzer.
10209 @item max-tree-if-conversion-phi-args
10210 Maximum number of arguments in a PHI supported by TREE if conversion
10211 unless the loop is marked with simd pragma.
10213 @item vect-max-version-for-alignment-checks
10214 The maximum number of run-time checks that can be performed when
10215 doing loop versioning for alignment in the vectorizer.
10217 @item vect-max-version-for-alias-checks
10218 The maximum number of run-time checks that can be performed when
10219 doing loop versioning for alias in the vectorizer.
10221 @item vect-max-peeling-for-alignment
10222 The maximum number of loop peels to enhance access alignment
10223 for vectorizer. Value -1 means no limit.
10225 @item max-iterations-to-track
10226 The maximum number of iterations of a loop the brute-force algorithm
10227 for analysis of the number of iterations of the loop tries to evaluate.
10229 @item hot-bb-count-ws-permille
10230 A basic block profile count is considered hot if it contributes to
10231 the given permillage (i.e. 0...1000) of the entire profiled execution.
10233 @item hot-bb-frequency-fraction
10234 Select fraction of the entry block frequency of executions of basic block in
10235 function given basic block needs to have to be considered hot.
10237 @item max-predicted-iterations
10238 The maximum number of loop iterations we predict statically. This is useful
10239 in cases where a function contains a single loop with known bound and
10240 another loop with unknown bound.
10241 The known number of iterations is predicted correctly, while
10242 the unknown number of iterations average to roughly 10. This means that the
10243 loop without bounds appears artificially cold relative to the other one.
10245 @item builtin-expect-probability
10246 Control the probability of the expression having the specified value. This
10247 parameter takes a percentage (i.e. 0 ... 100) as input.
10248 The default probability of 90 is obtained empirically.
10250 @item align-threshold
10252 Select fraction of the maximal frequency of executions of a basic block in
10253 a function to align the basic block.
10255 @item align-loop-iterations
10257 A loop expected to iterate at least the selected number of iterations is
10260 @item tracer-dynamic-coverage
10261 @itemx tracer-dynamic-coverage-feedback
10263 This value is used to limit superblock formation once the given percentage of
10264 executed instructions is covered. This limits unnecessary code size
10267 The @option{tracer-dynamic-coverage-feedback} parameter
10268 is used only when profile
10269 feedback is available. The real profiles (as opposed to statically estimated
10270 ones) are much less balanced allowing the threshold to be larger value.
10272 @item tracer-max-code-growth
10273 Stop tail duplication once code growth has reached given percentage. This is
10274 a rather artificial limit, as most of the duplicates are eliminated later in
10275 cross jumping, so it may be set to much higher values than is the desired code
10278 @item tracer-min-branch-ratio
10280 Stop reverse growth when the reverse probability of best edge is less than this
10281 threshold (in percent).
10283 @item tracer-min-branch-probability
10284 @itemx tracer-min-branch-probability-feedback
10286 Stop forward growth if the best edge has probability lower than this
10289 Similarly to @option{tracer-dynamic-coverage} two parameters are
10290 provided. @option{tracer-min-branch-probability-feedback} is used for
10291 compilation with profile feedback and @option{tracer-min-branch-probability}
10292 compilation without. The value for compilation with profile feedback
10293 needs to be more conservative (higher) in order to make tracer
10296 @item stack-clash-protection-guard-size
10297 Specify the size of the operating system provided stack guard as
10298 2 raised to @var{num} bytes. The default value is 12 (4096 bytes).
10299 Acceptable values are between 12 and 30. Higher values may reduce the
10300 number of explicit probes, but a value larger than the operating system
10301 provided guard will leave code vulnerable to stack clash style attacks.
10303 @item stack-clash-protection-probe-interval
10304 Stack clash protection involves probing stack space as it is allocated. This
10305 param controls the maximum distance between probes into the stack as 2 raised
10306 to @var{num} bytes. Acceptable values are between 10 and 16 and defaults to
10307 12. Higher values may reduce the number of explicit probes, but a value
10308 larger than the operating system provided guard will leave code vulnerable to
10309 stack clash style attacks.
10311 @item max-cse-path-length
10313 The maximum number of basic blocks on path that CSE considers.
10316 @item max-cse-insns
10317 The maximum number of instructions CSE processes before flushing.
10318 The default is 1000.
10320 @item ggc-min-expand
10322 GCC uses a garbage collector to manage its own memory allocation. This
10323 parameter specifies the minimum percentage by which the garbage
10324 collector's heap should be allowed to expand between collections.
10325 Tuning this may improve compilation speed; it has no effect on code
10328 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
10329 RAM >= 1GB@. If @code{getrlimit} is available, the notion of ``RAM'' is
10330 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
10331 GCC is not able to calculate RAM on a particular platform, the lower
10332 bound of 30% is used. Setting this parameter and
10333 @option{ggc-min-heapsize} to zero causes a full collection to occur at
10334 every opportunity. This is extremely slow, but can be useful for
10337 @item ggc-min-heapsize
10339 Minimum size of the garbage collector's heap before it begins bothering
10340 to collect garbage. The first collection occurs after the heap expands
10341 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
10342 tuning this may improve compilation speed, and has no effect on code
10345 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit that
10346 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
10347 with a lower bound of 4096 (four megabytes) and an upper bound of
10348 131072 (128 megabytes). If GCC is not able to calculate RAM on a
10349 particular platform, the lower bound is used. Setting this parameter
10350 very large effectively disables garbage collection. Setting this
10351 parameter and @option{ggc-min-expand} to zero causes a full collection
10352 to occur at every opportunity.
10354 @item max-reload-search-insns
10355 The maximum number of instruction reload should look backward for equivalent
10356 register. Increasing values mean more aggressive optimization, making the
10357 compilation time increase with probably slightly better performance.
10358 The default value is 100.
10360 @item max-cselib-memory-locations
10361 The maximum number of memory locations cselib should take into account.
10362 Increasing values mean more aggressive optimization, making the compilation time
10363 increase with probably slightly better performance. The default value is 500.
10365 @item max-sched-ready-insns
10366 The maximum number of instructions ready to be issued the scheduler should
10367 consider at any given time during the first scheduling pass. Increasing
10368 values mean more thorough searches, making the compilation time increase
10369 with probably little benefit. The default value is 100.
10371 @item max-sched-region-blocks
10372 The maximum number of blocks in a region to be considered for
10373 interblock scheduling. The default value is 10.
10375 @item max-pipeline-region-blocks
10376 The maximum number of blocks in a region to be considered for
10377 pipelining in the selective scheduler. The default value is 15.
10379 @item max-sched-region-insns
10380 The maximum number of insns in a region to be considered for
10381 interblock scheduling. The default value is 100.
10383 @item max-pipeline-region-insns
10384 The maximum number of insns in a region to be considered for
10385 pipelining in the selective scheduler. The default value is 200.
10387 @item min-spec-prob
10388 The minimum probability (in percents) of reaching a source block
10389 for interblock speculative scheduling. The default value is 40.
10391 @item max-sched-extend-regions-iters
10392 The maximum number of iterations through CFG to extend regions.
10393 A value of 0 (the default) disables region extensions.
10395 @item max-sched-insn-conflict-delay
10396 The maximum conflict delay for an insn to be considered for speculative motion.
10397 The default value is 3.
10399 @item sched-spec-prob-cutoff
10400 The minimal probability of speculation success (in percents), so that
10401 speculative insns are scheduled.
10402 The default value is 40.
10404 @item sched-state-edge-prob-cutoff
10405 The minimum probability an edge must have for the scheduler to save its
10407 The default value is 10.
10409 @item sched-mem-true-dep-cost
10410 Minimal distance (in CPU cycles) between store and load targeting same
10411 memory locations. The default value is 1.
10413 @item selsched-max-lookahead
10414 The maximum size of the lookahead window of selective scheduling. It is a
10415 depth of search for available instructions.
10416 The default value is 50.
10418 @item selsched-max-sched-times
10419 The maximum number of times that an instruction is scheduled during
10420 selective scheduling. This is the limit on the number of iterations
10421 through which the instruction may be pipelined. The default value is 2.
10423 @item selsched-insns-to-rename
10424 The maximum number of best instructions in the ready list that are considered
10425 for renaming in the selective scheduler. The default value is 2.
10428 The minimum value of stage count that swing modulo scheduler
10429 generates. The default value is 2.
10431 @item max-last-value-rtl
10432 The maximum size measured as number of RTLs that can be recorded in an expression
10433 in combiner for a pseudo register as last known value of that register. The default
10436 @item max-combine-insns
10437 The maximum number of instructions the RTL combiner tries to combine.
10438 The default value is 2 at @option{-Og} and 4 otherwise.
10440 @item integer-share-limit
10441 Small integer constants can use a shared data structure, reducing the
10442 compiler's memory usage and increasing its speed. This sets the maximum
10443 value of a shared integer constant. The default value is 256.
10445 @item ssp-buffer-size
10446 The minimum size of buffers (i.e.@: arrays) that receive stack smashing
10447 protection when @option{-fstack-protection} is used.
10449 @item min-size-for-stack-sharing
10450 The minimum size of variables taking part in stack slot sharing when not
10451 optimizing. The default value is 32.
10453 @item max-jump-thread-duplication-stmts
10454 Maximum number of statements allowed in a block that needs to be
10455 duplicated when threading jumps.
10457 @item max-fields-for-field-sensitive
10458 Maximum number of fields in a structure treated in
10459 a field sensitive manner during pointer analysis. The default is zero
10460 for @option{-O0} and @option{-O1},
10461 and 100 for @option{-Os}, @option{-O2}, and @option{-O3}.
10463 @item prefetch-latency
10464 Estimate on average number of instructions that are executed before
10465 prefetch finishes. The distance prefetched ahead is proportional
10466 to this constant. Increasing this number may also lead to less
10467 streams being prefetched (see @option{simultaneous-prefetches}).
10469 @item simultaneous-prefetches
10470 Maximum number of prefetches that can run at the same time.
10472 @item l1-cache-line-size
10473 The size of cache line in L1 cache, in bytes.
10475 @item l1-cache-size
10476 The size of L1 cache, in kilobytes.
10478 @item l2-cache-size
10479 The size of L2 cache, in kilobytes.
10481 @item min-insn-to-prefetch-ratio
10482 The minimum ratio between the number of instructions and the
10483 number of prefetches to enable prefetching in a loop.
10485 @item prefetch-min-insn-to-mem-ratio
10486 The minimum ratio between the number of instructions and the
10487 number of memory references to enable prefetching in a loop.
10489 @item use-canonical-types
10490 Whether the compiler should use the ``canonical'' type system. By
10491 default, this should always be 1, which uses a more efficient internal
10492 mechanism for comparing types in C++ and Objective-C++. However, if
10493 bugs in the canonical type system are causing compilation failures,
10494 set this value to 0 to disable canonical types.
10496 @item switch-conversion-max-branch-ratio
10497 Switch initialization conversion refuses to create arrays that are
10498 bigger than @option{switch-conversion-max-branch-ratio} times the number of
10499 branches in the switch.
10501 @item max-partial-antic-length
10502 Maximum length of the partial antic set computed during the tree
10503 partial redundancy elimination optimization (@option{-ftree-pre}) when
10504 optimizing at @option{-O3} and above. For some sorts of source code
10505 the enhanced partial redundancy elimination optimization can run away,
10506 consuming all of the memory available on the host machine. This
10507 parameter sets a limit on the length of the sets that are computed,
10508 which prevents the runaway behavior. Setting a value of 0 for
10509 this parameter allows an unlimited set length.
10511 @item sccvn-max-scc-size
10512 Maximum size of a strongly connected component (SCC) during SCCVN
10513 processing. If this limit is hit, SCCVN processing for the whole
10514 function is not done and optimizations depending on it are
10515 disabled. The default maximum SCC size is 10000.
10517 @item sccvn-max-alias-queries-per-access
10518 Maximum number of alias-oracle queries we perform when looking for
10519 redundancies for loads and stores. If this limit is hit the search
10520 is aborted and the load or store is not considered redundant. The
10521 number of queries is algorithmically limited to the number of
10522 stores on all paths from the load to the function entry.
10523 The default maximum number of queries is 1000.
10525 @item ira-max-loops-num
10526 IRA uses regional register allocation by default. If a function
10527 contains more loops than the number given by this parameter, only at most
10528 the given number of the most frequently-executed loops form regions
10529 for regional register allocation. The default value of the
10532 @item ira-max-conflict-table-size
10533 Although IRA uses a sophisticated algorithm to compress the conflict
10534 table, the table can still require excessive amounts of memory for
10535 huge functions. If the conflict table for a function could be more
10536 than the size in MB given by this parameter, the register allocator
10537 instead uses a faster, simpler, and lower-quality
10538 algorithm that does not require building a pseudo-register conflict table.
10539 The default value of the parameter is 2000.
10541 @item ira-loop-reserved-regs
10542 IRA can be used to evaluate more accurate register pressure in loops
10543 for decisions to move loop invariants (see @option{-O3}). The number
10544 of available registers reserved for some other purposes is given
10545 by this parameter. The default value of the parameter is 2, which is
10546 the minimal number of registers needed by typical instructions.
10547 This value is the best found from numerous experiments.
10549 @item lra-inheritance-ebb-probability-cutoff
10550 LRA tries to reuse values reloaded in registers in subsequent insns.
10551 This optimization is called inheritance. EBB is used as a region to
10552 do this optimization. The parameter defines a minimal fall-through
10553 edge probability in percentage used to add BB to inheritance EBB in
10554 LRA. The default value of the parameter is 40. The value was chosen
10555 from numerous runs of SPEC2000 on x86-64.
10557 @item loop-invariant-max-bbs-in-loop
10558 Loop invariant motion can be very expensive, both in compilation time and
10559 in amount of needed compile-time memory, with very large loops. Loops
10560 with more basic blocks than this parameter won't have loop invariant
10561 motion optimization performed on them. The default value of the
10562 parameter is 1000 for @option{-O1} and 10000 for @option{-O2} and above.
10564 @item loop-max-datarefs-for-datadeps
10565 Building data dependencies is expensive for very large loops. This
10566 parameter limits the number of data references in loops that are
10567 considered for data dependence analysis. These large loops are no
10568 handled by the optimizations using loop data dependencies.
10569 The default value is 1000.
10571 @item max-vartrack-size
10572 Sets a maximum number of hash table slots to use during variable
10573 tracking dataflow analysis of any function. If this limit is exceeded
10574 with variable tracking at assignments enabled, analysis for that
10575 function is retried without it, after removing all debug insns from
10576 the function. If the limit is exceeded even without debug insns, var
10577 tracking analysis is completely disabled for the function. Setting
10578 the parameter to zero makes it unlimited.
10580 @item max-vartrack-expr-depth
10581 Sets a maximum number of recursion levels when attempting to map
10582 variable names or debug temporaries to value expressions. This trades
10583 compilation time for more complete debug information. If this is set too
10584 low, value expressions that are available and could be represented in
10585 debug information may end up not being used; setting this higher may
10586 enable the compiler to find more complex debug expressions, but compile
10587 time and memory use may grow. The default is 12.
10589 @item min-nondebug-insn-uid
10590 Use uids starting at this parameter for nondebug insns. The range below
10591 the parameter is reserved exclusively for debug insns created by
10592 @option{-fvar-tracking-assignments}, but debug insns may get
10593 (non-overlapping) uids above it if the reserved range is exhausted.
10595 @item ipa-sra-ptr-growth-factor
10596 IPA-SRA replaces a pointer to an aggregate with one or more new
10597 parameters only when their cumulative size is less or equal to
10598 @option{ipa-sra-ptr-growth-factor} times the size of the original
10601 @item sra-max-scalarization-size-Ospeed
10602 @item sra-max-scalarization-size-Osize
10603 The two Scalar Reduction of Aggregates passes (SRA and IPA-SRA) aim to
10604 replace scalar parts of aggregates with uses of independent scalar
10605 variables. These parameters control the maximum size, in storage units,
10606 of aggregate which is considered for replacement when compiling for
10608 (@option{sra-max-scalarization-size-Ospeed}) or size
10609 (@option{sra-max-scalarization-size-Osize}) respectively.
10611 @item tm-max-aggregate-size
10612 When making copies of thread-local variables in a transaction, this
10613 parameter specifies the size in bytes after which variables are
10614 saved with the logging functions as opposed to save/restore code
10615 sequence pairs. This option only applies when using
10618 @item graphite-max-nb-scop-params
10619 To avoid exponential effects in the Graphite loop transforms, the
10620 number of parameters in a Static Control Part (SCoP) is bounded. The
10621 default value is 10 parameters, a value of zero can be used to lift
10622 the bound. A variable whose value is unknown at compilation time and
10623 defined outside a SCoP is a parameter of the SCoP.
10625 @item loop-block-tile-size
10626 Loop blocking or strip mining transforms, enabled with
10627 @option{-floop-block} or @option{-floop-strip-mine}, strip mine each
10628 loop in the loop nest by a given number of iterations. The strip
10629 length can be changed using the @option{loop-block-tile-size}
10630 parameter. The default value is 51 iterations.
10632 @item loop-unroll-jam-size
10633 Specify the unroll factor for the @option{-floop-unroll-and-jam} option. The
10634 default value is 4.
10636 @item loop-unroll-jam-depth
10637 Specify the dimension to be unrolled (counting from the most inner loop)
10638 for the @option{-floop-unroll-and-jam}. The default value is 2.
10640 @item ipa-cp-value-list-size
10641 IPA-CP attempts to track all possible values and types passed to a function's
10642 parameter in order to propagate them and perform devirtualization.
10643 @option{ipa-cp-value-list-size} is the maximum number of values and types it
10644 stores per one formal parameter of a function.
10646 @item ipa-cp-eval-threshold
10647 IPA-CP calculates its own score of cloning profitability heuristics
10648 and performs those cloning opportunities with scores that exceed
10649 @option{ipa-cp-eval-threshold}.
10651 @item ipa-cp-recursion-penalty
10652 Percentage penalty the recursive functions will receive when they
10653 are evaluated for cloning.
10655 @item ipa-cp-single-call-penalty
10656 Percentage penalty functions containing a single call to another
10657 function will receive when they are evaluated for cloning.
10660 @item ipa-max-agg-items
10661 IPA-CP is also capable to propagate a number of scalar values passed
10662 in an aggregate. @option{ipa-max-agg-items} controls the maximum
10663 number of such values per one parameter.
10665 @item ipa-cp-loop-hint-bonus
10666 When IPA-CP determines that a cloning candidate would make the number
10667 of iterations of a loop known, it adds a bonus of
10668 @option{ipa-cp-loop-hint-bonus} to the profitability score of
10671 @item ipa-cp-array-index-hint-bonus
10672 When IPA-CP determines that a cloning candidate would make the index of
10673 an array access known, it adds a bonus of
10674 @option{ipa-cp-array-index-hint-bonus} to the profitability
10675 score of the candidate.
10677 @item ipa-max-aa-steps
10678 During its analysis of function bodies, IPA-CP employs alias analysis
10679 in order to track values pointed to by function parameters. In order
10680 not spend too much time analyzing huge functions, it gives up and
10681 consider all memory clobbered after examining
10682 @option{ipa-max-aa-steps} statements modifying memory.
10684 @item lto-partitions
10685 Specify desired number of partitions produced during WHOPR compilation.
10686 The number of partitions should exceed the number of CPUs used for compilation.
10687 The default value is 32.
10689 @item lto-min-partition
10690 Size of minimal partition for WHOPR (in estimated instructions).
10691 This prevents expenses of splitting very small programs into too many
10694 @item lto-max-partition
10695 Size of max partition for WHOPR (in estimated instructions).
10696 to provide an upper bound for individual size of partition.
10697 Meant to be used only with balanced partitioning.
10699 @item cxx-max-namespaces-for-diagnostic-help
10700 The maximum number of namespaces to consult for suggestions when C++
10701 name lookup fails for an identifier. The default is 1000.
10703 @item sink-frequency-threshold
10704 The maximum relative execution frequency (in percents) of the target block
10705 relative to a statement's original block to allow statement sinking of a
10706 statement. Larger numbers result in more aggressive statement sinking.
10707 The default value is 75. A small positive adjustment is applied for
10708 statements with memory operands as those are even more profitable so sink.
10710 @item max-stores-to-sink
10711 The maximum number of conditional store pairs that can be sunk. Set to 0
10712 if either vectorization (@option{-ftree-vectorize}) or if-conversion
10713 (@option{-ftree-loop-if-convert}) is disabled. The default is 2.
10715 @item allow-store-data-races
10716 Allow optimizers to introduce new data races on stores.
10717 Set to 1 to allow, otherwise to 0. This option is enabled by default
10718 at optimization level @option{-Ofast}.
10720 @item case-values-threshold
10721 The smallest number of different values for which it is best to use a
10722 jump-table instead of a tree of conditional branches. If the value is
10723 0, use the default for the machine. The default is 0.
10725 @item tree-reassoc-width
10726 Set the maximum number of instructions executed in parallel in
10727 reassociated tree. This parameter overrides target dependent
10728 heuristics used by default if has non zero value.
10730 @item sched-pressure-algorithm
10731 Choose between the two available implementations of
10732 @option{-fsched-pressure}. Algorithm 1 is the original implementation
10733 and is the more likely to prevent instructions from being reordered.
10734 Algorithm 2 was designed to be a compromise between the relatively
10735 conservative approach taken by algorithm 1 and the rather aggressive
10736 approach taken by the default scheduler. It relies more heavily on
10737 having a regular register file and accurate register pressure classes.
10738 See @file{haifa-sched.c} in the GCC sources for more details.
10740 The default choice depends on the target.
10742 @item max-slsr-cand-scan
10743 Set the maximum number of existing candidates that are considered when
10744 seeking a basis for a new straight-line strength reduction candidate.
10747 Enable buffer overflow detection for global objects. This kind
10748 of protection is enabled by default if you are using
10749 @option{-fsanitize=address} option.
10750 To disable global objects protection use @option{--param asan-globals=0}.
10753 Enable buffer overflow detection for stack objects. This kind of
10754 protection is enabled by default when using @option{-fsanitize=address}.
10755 To disable stack protection use @option{--param asan-stack=0} option.
10757 @item asan-instrument-reads
10758 Enable buffer overflow detection for memory reads. This kind of
10759 protection is enabled by default when using @option{-fsanitize=address}.
10760 To disable memory reads protection use
10761 @option{--param asan-instrument-reads=0}.
10763 @item asan-instrument-writes
10764 Enable buffer overflow detection for memory writes. This kind of
10765 protection is enabled by default when using @option{-fsanitize=address}.
10766 To disable memory writes protection use
10767 @option{--param asan-instrument-writes=0} option.
10769 @item asan-memintrin
10770 Enable detection for built-in functions. This kind of protection
10771 is enabled by default when using @option{-fsanitize=address}.
10772 To disable built-in functions protection use
10773 @option{--param asan-memintrin=0}.
10775 @item asan-use-after-return
10776 Enable detection of use-after-return. This kind of protection
10777 is enabled by default when using the @option{-fsanitize=address} option.
10778 To disable it use @option{--param asan-use-after-return=0}.
10780 Note: By default the check is disabled at run time. To enable it,
10781 add @code{detect_stack_use_after_return=1} to the environment variable
10782 @env{ASAN_OPTIONS}.
10784 @item asan-instrumentation-with-call-threshold
10785 If number of memory accesses in function being instrumented
10786 is greater or equal to this number, use callbacks instead of inline checks.
10787 E.g. to disable inline code use
10788 @option{--param asan-instrumentation-with-call-threshold=0}.
10790 @item use-after-scope-direct-emission-threshold
10791 If the size of a local variable in bytes is smaller or equal to this
10792 number, directly poison (or unpoison) shadow memory instead of using
10793 run-time callbacks. The default value is 256.
10795 @item chkp-max-ctor-size
10796 Static constructors generated by Pointer Bounds Checker may become very
10797 large and significantly increase compile time at optimization level
10798 @option{-O1} and higher. This parameter is a maximum number of statements
10799 in a single generated constructor. Default value is 5000.
10801 @item max-fsm-thread-path-insns
10802 Maximum number of instructions to copy when duplicating blocks on a
10803 finite state automaton jump thread path. The default is 100.
10805 @item max-fsm-thread-length
10806 Maximum number of basic blocks on a finite state automaton jump thread
10807 path. The default is 10.
10809 @item max-fsm-thread-paths
10810 Maximum number of new jump thread paths to create for a finite state
10811 automaton. The default is 50.
10813 @item parloops-chunk-size
10814 Chunk size of omp schedule for loops parallelized by parloops. The default
10817 @item parloops-schedule
10818 Schedule type of omp schedule for loops parallelized by parloops (static,
10819 dynamic, guided, auto, runtime). The default is static.
10821 @item parloops-min-per-thread
10822 The minimum number of iterations per thread of an innermost parallelized
10823 loop for which the parallelized variant is prefered over the single threaded
10824 one. The default is 100. Note that for a parallelized loop nest the
10825 minimum number of iterations of the outermost loop per thread is two.
10827 @item max-ssa-name-query-depth
10828 Maximum depth of recursion when querying properties of SSA names in things
10829 like fold routines. One level of recursion corresponds to following a
10832 @item hsa-gen-debug-stores
10833 Enable emission of special debug stores within HSA kernels which are
10834 then read and reported by libgomp plugin. Generation of these stores
10835 is disabled by default, use @option{--param hsa-gen-debug-stores=1} to
10838 @item max-speculative-devirt-maydefs
10839 The maximum number of may-defs we analyze when looking for a must-def
10840 specifying the dynamic type of an object that invokes a virtual call
10841 we may be able to devirtualize speculatively.
10843 @item max-vrp-switch-assertions
10844 The maximum number of assertions to add along the default edge of a switch
10845 statement during VRP. The default is 10.
10849 @node Instrumentation Options
10850 @section Program Instrumentation Options
10851 @cindex instrumentation options
10852 @cindex program instrumentation options
10853 @cindex run-time error checking options
10854 @cindex profiling options
10855 @cindex options, program instrumentation
10856 @cindex options, run-time error checking
10857 @cindex options, profiling
10859 GCC supports a number of command-line options that control adding
10860 run-time instrumentation to the code it normally generates.
10861 For example, one purpose of instrumentation is collect profiling
10862 statistics for use in finding program hot spots, code coverage
10863 analysis, or profile-guided optimizations.
10864 Another class of program instrumentation is adding run-time checking
10865 to detect programming errors like invalid pointer
10866 dereferences or out-of-bounds array accesses, as well as deliberately
10867 hostile attacks such as stack smashing or C++ vtable hijacking.
10868 There is also a general hook which can be used to implement other
10869 forms of tracing or function-level instrumentation for debug or
10870 program analysis purposes.
10873 @cindex @command{prof}
10876 Generate extra code to write profile information suitable for the
10877 analysis program @command{prof}. You must use this option when compiling
10878 the source files you want data about, and you must also use it when
10881 @cindex @command{gprof}
10884 Generate extra code to write profile information suitable for the
10885 analysis program @command{gprof}. You must use this option when compiling
10886 the source files you want data about, and you must also use it when
10889 @item -fprofile-arcs
10890 @opindex fprofile-arcs
10891 Add code so that program flow @dfn{arcs} are instrumented. During
10892 execution the program records how many times each branch and call is
10893 executed and how many times it is taken or returns. On targets that support
10894 constructors with priority support, profiling properly handles constructors,
10895 destructors and C++ constructors (and destructors) of classes which are used
10896 as a type of a global variable.
10899 program exits it saves this data to a file called
10900 @file{@var{auxname}.gcda} for each source file. The data may be used for
10901 profile-directed optimizations (@option{-fbranch-probabilities}), or for
10902 test coverage analysis (@option{-ftest-coverage}). Each object file's
10903 @var{auxname} is generated from the name of the output file, if
10904 explicitly specified and it is not the final executable, otherwise it is
10905 the basename of the source file. In both cases any suffix is removed
10906 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
10907 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
10908 @xref{Cross-profiling}.
10910 @cindex @command{gcov}
10914 This option is used to compile and link code instrumented for coverage
10915 analysis. The option is a synonym for @option{-fprofile-arcs}
10916 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
10917 linking). See the documentation for those options for more details.
10922 Compile the source files with @option{-fprofile-arcs} plus optimization
10923 and code generation options. For test coverage analysis, use the
10924 additional @option{-ftest-coverage} option. You do not need to profile
10925 every source file in a program.
10928 Compile the source files additionally with @option{-fprofile-abs-path}
10929 to create absolute path names in the @file{.gcno} files. This allows
10930 @command{gcov} to find the correct sources in projects where compilations
10931 occur with different working directories.
10934 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
10935 (the latter implies the former).
10938 Run the program on a representative workload to generate the arc profile
10939 information. This may be repeated any number of times. You can run
10940 concurrent instances of your program, and provided that the file system
10941 supports locking, the data files will be correctly updated. Unless
10942 a strict ISO C dialect option is in effect, @code{fork} calls are
10943 detected and correctly handled without double counting.
10946 For profile-directed optimizations, compile the source files again with
10947 the same optimization and code generation options plus
10948 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
10949 Control Optimization}).
10952 For test coverage analysis, use @command{gcov} to produce human readable
10953 information from the @file{.gcno} and @file{.gcda} files. Refer to the
10954 @command{gcov} documentation for further information.
10958 With @option{-fprofile-arcs}, for each function of your program GCC
10959 creates a program flow graph, then finds a spanning tree for the graph.
10960 Only arcs that are not on the spanning tree have to be instrumented: the
10961 compiler adds code to count the number of times that these arcs are
10962 executed. When an arc is the only exit or only entrance to a block, the
10963 instrumentation code can be added to the block; otherwise, a new basic
10964 block must be created to hold the instrumentation code.
10967 @item -ftest-coverage
10968 @opindex ftest-coverage
10969 Produce a notes file that the @command{gcov} code-coverage utility
10970 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
10971 show program coverage. Each source file's note file is called
10972 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
10973 above for a description of @var{auxname} and instructions on how to
10974 generate test coverage data. Coverage data matches the source files
10975 more closely if you do not optimize.
10977 @item -fprofile-abs-path
10978 @opindex fprofile-abs-path
10979 Automatically convert relative source file names to absolute path names
10980 in the @file{.gcno} files. This allows @command{gcov} to find the correct
10981 sources in projects where compilations occur with different working
10984 @item -fprofile-dir=@var{path}
10985 @opindex fprofile-dir
10987 Set the directory to search for the profile data files in to @var{path}.
10988 This option affects only the profile data generated by
10989 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
10990 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
10991 and its related options. Both absolute and relative paths can be used.
10992 By default, GCC uses the current directory as @var{path}, thus the
10993 profile data file appears in the same directory as the object file.
10995 @item -fprofile-generate
10996 @itemx -fprofile-generate=@var{path}
10997 @opindex fprofile-generate
10999 Enable options usually used for instrumenting application to produce
11000 profile useful for later recompilation with profile feedback based
11001 optimization. You must use @option{-fprofile-generate} both when
11002 compiling and when linking your program.
11004 The following options are enabled: @option{-fprofile-arcs}, @option{-fprofile-values}, @option{-fvpt}.
11006 If @var{path} is specified, GCC looks at the @var{path} to find
11007 the profile feedback data files. See @option{-fprofile-dir}.
11009 To optimize the program based on the collected profile information, use
11010 @option{-fprofile-use}. @xref{Optimize Options}, for more information.
11012 @item -fprofile-update=@var{method}
11013 @opindex fprofile-update
11015 Alter the update method for an application instrumented for profile
11016 feedback based optimization. The @var{method} argument should be one of
11017 @samp{single}, @samp{atomic} or @samp{prefer-atomic}.
11018 The first one is useful for single-threaded applications,
11019 while the second one prevents profile corruption by emitting thread-safe code.
11021 @strong{Warning:} When an application does not properly join all threads
11022 (or creates an detached thread), a profile file can be still corrupted.
11024 Using @samp{prefer-atomic} would be transformed either to @samp{atomic},
11025 when supported by a target, or to @samp{single} otherwise. The GCC driver
11026 automatically selects @samp{prefer-atomic} when @option{-pthread}
11027 is present in the command line.
11029 @item -fsanitize=address
11030 @opindex fsanitize=address
11031 Enable AddressSanitizer, a fast memory error detector.
11032 Memory access instructions are instrumented to detect
11033 out-of-bounds and use-after-free bugs.
11034 The option enables @option{-fsanitize-address-use-after-scope}.
11035 See @uref{https://github.com/google/sanitizers/wiki/AddressSanitizer} for
11036 more details. The run-time behavior can be influenced using the
11037 @env{ASAN_OPTIONS} environment variable. When set to @code{help=1},
11038 the available options are shown at startup of the instrumented program. See
11039 @url{https://github.com/google/sanitizers/wiki/AddressSanitizerFlags#run-time-flags}
11040 for a list of supported options.
11041 The option cannot be combined with @option{-fsanitize=thread}
11042 and/or @option{-fcheck-pointer-bounds}.
11044 @item -fsanitize=kernel-address
11045 @opindex fsanitize=kernel-address
11046 Enable AddressSanitizer for Linux kernel.
11047 See @uref{https://github.com/google/kasan/wiki} for more details.
11048 The option cannot be combined with @option{-fcheck-pointer-bounds}.
11050 @item -fsanitize=thread
11051 @opindex fsanitize=thread
11052 Enable ThreadSanitizer, a fast data race detector.
11053 Memory access instructions are instrumented to detect
11054 data race bugs. See @uref{https://github.com/google/sanitizers/wiki#threadsanitizer} for more
11055 details. The run-time behavior can be influenced using the @env{TSAN_OPTIONS}
11056 environment variable; see
11057 @url{https://github.com/google/sanitizers/wiki/ThreadSanitizerFlags} for a list of
11059 The option cannot be combined with @option{-fsanitize=address},
11060 @option{-fsanitize=leak} and/or @option{-fcheck-pointer-bounds}.
11062 Note that sanitized atomic builtins cannot throw exceptions when
11063 operating on invalid memory addresses with non-call exceptions
11064 (@option{-fnon-call-exceptions}).
11066 @item -fsanitize=leak
11067 @opindex fsanitize=leak
11068 Enable LeakSanitizer, a memory leak detector.
11069 This option only matters for linking of executables and
11070 the executable is linked against a library that overrides @code{malloc}
11071 and other allocator functions. See
11072 @uref{https://github.com/google/sanitizers/wiki/AddressSanitizerLeakSanitizer} for more
11073 details. The run-time behavior can be influenced using the
11074 @env{LSAN_OPTIONS} environment variable.
11075 The option cannot be combined with @option{-fsanitize=thread}.
11077 @item -fsanitize=undefined
11078 @opindex fsanitize=undefined
11079 Enable UndefinedBehaviorSanitizer, a fast undefined behavior detector.
11080 Various computations are instrumented to detect undefined behavior
11081 at runtime. Current suboptions are:
11085 @item -fsanitize=shift
11086 @opindex fsanitize=shift
11087 This option enables checking that the result of a shift operation is
11088 not undefined. Note that what exactly is considered undefined differs
11089 slightly between C and C++, as well as between ISO C90 and C99, etc.
11090 This option has two suboptions, @option{-fsanitize=shift-base} and
11091 @option{-fsanitize=shift-exponent}.
11093 @item -fsanitize=shift-exponent
11094 @opindex fsanitize=shift-exponent
11095 This option enables checking that the second argument of a shift operation
11096 is not negative and is smaller than the precision of the promoted first
11099 @item -fsanitize=shift-base
11100 @opindex fsanitize=shift-base
11101 If the second argument of a shift operation is within range, check that the
11102 result of a shift operation is not undefined. Note that what exactly is
11103 considered undefined differs slightly between C and C++, as well as between
11104 ISO C90 and C99, etc.
11106 @item -fsanitize=integer-divide-by-zero
11107 @opindex fsanitize=integer-divide-by-zero
11108 Detect integer division by zero as well as @code{INT_MIN / -1} division.
11110 @item -fsanitize=unreachable
11111 @opindex fsanitize=unreachable
11112 With this option, the compiler turns the @code{__builtin_unreachable}
11113 call into a diagnostics message call instead. When reaching the
11114 @code{__builtin_unreachable} call, the behavior is undefined.
11116 @item -fsanitize=vla-bound
11117 @opindex fsanitize=vla-bound
11118 This option instructs the compiler to check that the size of a variable
11119 length array is positive.
11121 @item -fsanitize=null
11122 @opindex fsanitize=null
11123 This option enables pointer checking. Particularly, the application
11124 built with this option turned on will issue an error message when it
11125 tries to dereference a NULL pointer, or if a reference (possibly an
11126 rvalue reference) is bound to a NULL pointer, or if a method is invoked
11127 on an object pointed by a NULL pointer.
11129 @item -fsanitize=return
11130 @opindex fsanitize=return
11131 This option enables return statement checking. Programs
11132 built with this option turned on will issue an error message
11133 when the end of a non-void function is reached without actually
11134 returning a value. This option works in C++ only.
11136 @item -fsanitize=signed-integer-overflow
11137 @opindex fsanitize=signed-integer-overflow
11138 This option enables signed integer overflow checking. We check that
11139 the result of @code{+}, @code{*}, and both unary and binary @code{-}
11140 does not overflow in the signed arithmetics. Note, integer promotion
11141 rules must be taken into account. That is, the following is not an
11144 signed char a = SCHAR_MAX;
11148 @item -fsanitize=bounds
11149 @opindex fsanitize=bounds
11150 This option enables instrumentation of array bounds. Various out of bounds
11151 accesses are detected. Flexible array members, flexible array member-like
11152 arrays, and initializers of variables with static storage are not instrumented.
11153 The option cannot be combined with @option{-fcheck-pointer-bounds}.
11155 @item -fsanitize=bounds-strict
11156 @opindex fsanitize=bounds-strict
11157 This option enables strict instrumentation of array bounds. Most out of bounds
11158 accesses are detected, including flexible array members and flexible array
11159 member-like arrays. Initializers of variables with static storage are not
11160 instrumented. The option cannot be combined
11161 with @option{-fcheck-pointer-bounds}.
11163 @item -fsanitize=alignment
11164 @opindex fsanitize=alignment
11166 This option enables checking of alignment of pointers when they are
11167 dereferenced, or when a reference is bound to insufficiently aligned target,
11168 or when a method or constructor is invoked on insufficiently aligned object.
11170 @item -fsanitize=object-size
11171 @opindex fsanitize=object-size
11172 This option enables instrumentation of memory references using the
11173 @code{__builtin_object_size} function. Various out of bounds pointer
11174 accesses are detected.
11176 @item -fsanitize=float-divide-by-zero
11177 @opindex fsanitize=float-divide-by-zero
11178 Detect floating-point division by zero. Unlike other similar options,
11179 @option{-fsanitize=float-divide-by-zero} is not enabled by
11180 @option{-fsanitize=undefined}, since floating-point division by zero can
11181 be a legitimate way of obtaining infinities and NaNs.
11183 @item -fsanitize=float-cast-overflow
11184 @opindex fsanitize=float-cast-overflow
11185 This option enables floating-point type to integer conversion checking.
11186 We check that the result of the conversion does not overflow.
11187 Unlike other similar options, @option{-fsanitize=float-cast-overflow} is
11188 not enabled by @option{-fsanitize=undefined}.
11189 This option does not work well with @code{FE_INVALID} exceptions enabled.
11191 @item -fsanitize=nonnull-attribute
11192 @opindex fsanitize=nonnull-attribute
11194 This option enables instrumentation of calls, checking whether null values
11195 are not passed to arguments marked as requiring a non-null value by the
11196 @code{nonnull} function attribute.
11198 @item -fsanitize=returns-nonnull-attribute
11199 @opindex fsanitize=returns-nonnull-attribute
11201 This option enables instrumentation of return statements in functions
11202 marked with @code{returns_nonnull} function attribute, to detect returning
11203 of null values from such functions.
11205 @item -fsanitize=bool
11206 @opindex fsanitize=bool
11208 This option enables instrumentation of loads from bool. If a value other
11209 than 0/1 is loaded, a run-time error is issued.
11211 @item -fsanitize=enum
11212 @opindex fsanitize=enum
11214 This option enables instrumentation of loads from an enum type. If
11215 a value outside the range of values for the enum type is loaded,
11216 a run-time error is issued.
11218 @item -fsanitize=vptr
11219 @opindex fsanitize=vptr
11221 This option enables instrumentation of C++ member function calls, member
11222 accesses and some conversions between pointers to base and derived classes,
11223 to verify the referenced object has the correct dynamic type.
11225 @item -fsanitize=pointer-overflow
11226 @opindex fsanitize=pointer-overflow
11228 This option enables instrumentation of pointer arithmetics. If the pointer
11229 arithmetics overflows, a run-time error is issued.
11231 @item -fsanitize=builtin
11232 @opindex fsanitize=builtin
11234 This option enables instrumentation of arguments to selected builtin
11235 functions. If an invalid value is passed to such arguments, a run-time
11236 error is issued. E.g.@ passing 0 as the argument to @code{__builtin_ctz}
11237 or @code{__builtin_clz} invokes undefined behavior and is diagnosed
11242 While @option{-ftrapv} causes traps for signed overflows to be emitted,
11243 @option{-fsanitize=undefined} gives a diagnostic message.
11244 This currently works only for the C family of languages.
11246 @item -fno-sanitize=all
11247 @opindex fno-sanitize=all
11249 This option disables all previously enabled sanitizers.
11250 @option{-fsanitize=all} is not allowed, as some sanitizers cannot be used
11253 @item -fasan-shadow-offset=@var{number}
11254 @opindex fasan-shadow-offset
11255 This option forces GCC to use custom shadow offset in AddressSanitizer checks.
11256 It is useful for experimenting with different shadow memory layouts in
11257 Kernel AddressSanitizer.
11259 @item -fsanitize-sections=@var{s1},@var{s2},...
11260 @opindex fsanitize-sections
11261 Sanitize global variables in selected user-defined sections. @var{si} may
11264 @item -fsanitize-recover@r{[}=@var{opts}@r{]}
11265 @opindex fsanitize-recover
11266 @opindex fno-sanitize-recover
11267 @option{-fsanitize-recover=} controls error recovery mode for sanitizers
11268 mentioned in comma-separated list of @var{opts}. Enabling this option
11269 for a sanitizer component causes it to attempt to continue
11270 running the program as if no error happened. This means multiple
11271 runtime errors can be reported in a single program run, and the exit
11272 code of the program may indicate success even when errors
11273 have been reported. The @option{-fno-sanitize-recover=} option
11274 can be used to alter
11275 this behavior: only the first detected error is reported
11276 and program then exits with a non-zero exit code.
11278 Currently this feature only works for @option{-fsanitize=undefined} (and its suboptions
11279 except for @option{-fsanitize=unreachable} and @option{-fsanitize=return}),
11280 @option{-fsanitize=float-cast-overflow}, @option{-fsanitize=float-divide-by-zero},
11281 @option{-fsanitize=bounds-strict},
11282 @option{-fsanitize=kernel-address} and @option{-fsanitize=address}.
11283 For these sanitizers error recovery is turned on by default,
11284 except @option{-fsanitize=address}, for which this feature is experimental.
11285 @option{-fsanitize-recover=all} and @option{-fno-sanitize-recover=all} is also
11286 accepted, the former enables recovery for all sanitizers that support it,
11287 the latter disables recovery for all sanitizers that support it.
11289 Even if a recovery mode is turned on the compiler side, it needs to be also
11290 enabled on the runtime library side, otherwise the failures are still fatal.
11291 The runtime library defaults to @code{halt_on_error=0} for
11292 ThreadSanitizer and UndefinedBehaviorSanitizer, while default value for
11293 AddressSanitizer is @code{halt_on_error=1}. This can be overridden through
11294 setting the @code{halt_on_error} flag in the corresponding environment variable.
11296 Syntax without an explicit @var{opts} parameter is deprecated. It is
11297 equivalent to specifying an @var{opts} list of:
11300 undefined,float-cast-overflow,float-divide-by-zero,bounds-strict
11303 @item -fsanitize-address-use-after-scope
11304 @opindex fsanitize-address-use-after-scope
11305 Enable sanitization of local variables to detect use-after-scope bugs.
11306 The option sets @option{-fstack-reuse} to @samp{none}.
11308 @item -fsanitize-undefined-trap-on-error
11309 @opindex fsanitize-undefined-trap-on-error
11310 The @option{-fsanitize-undefined-trap-on-error} option instructs the compiler to
11311 report undefined behavior using @code{__builtin_trap} rather than
11312 a @code{libubsan} library routine. The advantage of this is that the
11313 @code{libubsan} library is not needed and is not linked in, so this
11314 is usable even in freestanding environments.
11316 @item -fsanitize-coverage=trace-pc
11317 @opindex fsanitize-coverage=trace-pc
11318 Enable coverage-guided fuzzing code instrumentation.
11319 Inserts a call to @code{__sanitizer_cov_trace_pc} into every basic block.
11321 @item -fsanitize-coverage=trace-cmp
11322 @opindex fsanitize-coverage=trace-cmp
11323 Enable dataflow guided fuzzing code instrumentation.
11324 Inserts a call to @code{__sanitizer_cov_trace_cmp1},
11325 @code{__sanitizer_cov_trace_cmp2}, @code{__sanitizer_cov_trace_cmp4} or
11326 @code{__sanitizer_cov_trace_cmp8} for integral comparison with both operands
11327 variable or @code{__sanitizer_cov_trace_const_cmp1},
11328 @code{__sanitizer_cov_trace_const_cmp2},
11329 @code{__sanitizer_cov_trace_const_cmp4} or
11330 @code{__sanitizer_cov_trace_const_cmp8} for integral comparison with one
11331 operand constant, @code{__sanitizer_cov_trace_cmpf} or
11332 @code{__sanitizer_cov_trace_cmpd} for float or double comparisons and
11333 @code{__sanitizer_cov_trace_switch} for switch statements.
11335 @item -fbounds-check
11336 @opindex fbounds-check
11337 For front ends that support it, generate additional code to check that
11338 indices used to access arrays are within the declared range. This is
11339 currently only supported by the Fortran front end, where this option
11342 @item -fcheck-pointer-bounds
11343 @opindex fcheck-pointer-bounds
11344 @opindex fno-check-pointer-bounds
11345 @cindex Pointer Bounds Checker options
11346 Enable Pointer Bounds Checker instrumentation. Each memory reference
11347 is instrumented with checks of the pointer used for memory access against
11348 bounds associated with that pointer.
11351 is only an implementation for Intel MPX available, thus x86 GNU/Linux target
11352 and @option{-mmpx} are required to enable this feature.
11353 MPX-based instrumentation requires
11354 a runtime library to enable MPX in hardware and handle bounds
11355 violation signals. By default when @option{-fcheck-pointer-bounds}
11356 and @option{-mmpx} options are used to link a program, the GCC driver
11357 links against the @file{libmpx} and @file{libmpxwrappers} libraries.
11358 Bounds checking on calls to dynamic libraries requires a linker
11359 with @option{-z bndplt} support; if GCC was configured with a linker
11360 without support for this option (including the Gold linker and older
11361 versions of ld), a warning is given if you link with @option{-mmpx}
11362 without also specifying @option{-static}, since the overall effectiveness
11363 of the bounds checking protection is reduced.
11364 See also @option{-static-libmpxwrappers}.
11366 MPX-based instrumentation
11367 may be used for debugging and also may be included in production code
11368 to increase program security. Depending on usage, you may
11369 have different requirements for the runtime library. The current version
11370 of the MPX runtime library is more oriented for use as a debugging
11371 tool. MPX runtime library usage implies @option{-lpthread}. See
11372 also @option{-static-libmpx}. The runtime library behavior can be
11373 influenced using various @env{CHKP_RT_*} environment variables. See
11374 @uref{https://gcc.gnu.org/wiki/Intel%20MPX%20support%20in%20the%20GCC%20compiler}
11377 Generated instrumentation may be controlled by various
11378 @option{-fchkp-*} options and by the @code{bnd_variable_size}
11379 structure field attribute (@pxref{Type Attributes}) and
11380 @code{bnd_legacy}, and @code{bnd_instrument} function attributes
11381 (@pxref{Function Attributes}). GCC also provides a number of built-in
11382 functions for controlling the Pointer Bounds Checker. @xref{Pointer
11383 Bounds Checker builtins}, for more information.
11385 @item -fchkp-check-incomplete-type
11386 @opindex fchkp-check-incomplete-type
11387 @opindex fno-chkp-check-incomplete-type
11388 Generate pointer bounds checks for variables with incomplete type.
11389 Enabled by default.
11391 @item -fchkp-narrow-bounds
11392 @opindex fchkp-narrow-bounds
11393 @opindex fno-chkp-narrow-bounds
11394 Controls bounds used by Pointer Bounds Checker for pointers to object
11395 fields. If narrowing is enabled then field bounds are used. Otherwise
11396 object bounds are used. See also @option{-fchkp-narrow-to-innermost-array}
11397 and @option{-fchkp-first-field-has-own-bounds}. Enabled by default.
11399 @item -fchkp-first-field-has-own-bounds
11400 @opindex fchkp-first-field-has-own-bounds
11401 @opindex fno-chkp-first-field-has-own-bounds
11402 Forces Pointer Bounds Checker to use narrowed bounds for the address of the
11403 first field in the structure. By default a pointer to the first field has
11404 the same bounds as a pointer to the whole structure.
11406 @item -fchkp-flexible-struct-trailing-arrays
11407 @opindex fchkp-flexible-struct-trailing-arrays
11408 @opindex fno-chkp-flexible-struct-trailing-arrays
11409 Forces Pointer Bounds Checker to treat all trailing arrays in structures as
11410 possibly flexible. By default only array fields with zero length or that are
11411 marked with attribute bnd_variable_size are treated as flexible.
11413 @item -fchkp-narrow-to-innermost-array
11414 @opindex fchkp-narrow-to-innermost-array
11415 @opindex fno-chkp-narrow-to-innermost-array
11416 Forces Pointer Bounds Checker to use bounds of the innermost arrays in
11417 case of nested static array access. By default this option is disabled and
11418 bounds of the outermost array are used.
11420 @item -fchkp-optimize
11421 @opindex fchkp-optimize
11422 @opindex fno-chkp-optimize
11423 Enables Pointer Bounds Checker optimizations. Enabled by default at
11424 optimization levels @option{-O}, @option{-O2}, @option{-O3}.
11426 @item -fchkp-use-fast-string-functions
11427 @opindex fchkp-use-fast-string-functions
11428 @opindex fno-chkp-use-fast-string-functions
11429 Enables use of @code{*_nobnd} versions of string functions (not copying bounds)
11430 by Pointer Bounds Checker. Disabled by default.
11432 @item -fchkp-use-nochk-string-functions
11433 @opindex fchkp-use-nochk-string-functions
11434 @opindex fno-chkp-use-nochk-string-functions
11435 Enables use of @code{*_nochk} versions of string functions (not checking bounds)
11436 by Pointer Bounds Checker. Disabled by default.
11438 @item -fchkp-use-static-bounds
11439 @opindex fchkp-use-static-bounds
11440 @opindex fno-chkp-use-static-bounds
11441 Allow Pointer Bounds Checker to generate static bounds holding
11442 bounds of static variables. Enabled by default.
11444 @item -fchkp-use-static-const-bounds
11445 @opindex fchkp-use-static-const-bounds
11446 @opindex fno-chkp-use-static-const-bounds
11447 Use statically-initialized bounds for constant bounds instead of
11448 generating them each time they are required. By default enabled when
11449 @option{-fchkp-use-static-bounds} is enabled.
11451 @item -fchkp-treat-zero-dynamic-size-as-infinite
11452 @opindex fchkp-treat-zero-dynamic-size-as-infinite
11453 @opindex fno-chkp-treat-zero-dynamic-size-as-infinite
11454 With this option, objects with incomplete type whose
11455 dynamically-obtained size is zero are treated as having infinite size
11456 instead by Pointer Bounds
11457 Checker. This option may be helpful if a program is linked with a library
11458 missing size information for some symbols. Disabled by default.
11460 @item -fchkp-check-read
11461 @opindex fchkp-check-read
11462 @opindex fno-chkp-check-read
11463 Instructs Pointer Bounds Checker to generate checks for all read
11464 accesses to memory. Enabled by default.
11466 @item -fchkp-check-write
11467 @opindex fchkp-check-write
11468 @opindex fno-chkp-check-write
11469 Instructs Pointer Bounds Checker to generate checks for all write
11470 accesses to memory. Enabled by default.
11472 @item -fchkp-store-bounds
11473 @opindex fchkp-store-bounds
11474 @opindex fno-chkp-store-bounds
11475 Instructs Pointer Bounds Checker to generate bounds stores for
11476 pointer writes. Enabled by default.
11478 @item -fchkp-instrument-calls
11479 @opindex fchkp-instrument-calls
11480 @opindex fno-chkp-instrument-calls
11481 Instructs Pointer Bounds Checker to pass pointer bounds to calls.
11482 Enabled by default.
11484 @item -fchkp-instrument-marked-only
11485 @opindex fchkp-instrument-marked-only
11486 @opindex fno-chkp-instrument-marked-only
11487 Instructs Pointer Bounds Checker to instrument only functions
11488 marked with the @code{bnd_instrument} attribute
11489 (@pxref{Function Attributes}). Disabled by default.
11491 @item -fchkp-use-wrappers
11492 @opindex fchkp-use-wrappers
11493 @opindex fno-chkp-use-wrappers
11494 Allows Pointer Bounds Checker to replace calls to built-in functions
11495 with calls to wrapper functions. When @option{-fchkp-use-wrappers}
11496 is used to link a program, the GCC driver automatically links
11497 against @file{libmpxwrappers}. See also @option{-static-libmpxwrappers}.
11498 Enabled by default.
11500 @item -fcf-protection==@r{[}full@r{|}branch@r{|}return@r{|}none@r{]}
11501 @opindex fcf-protection
11502 Enable code instrumentation of control-flow transfers to increase
11503 program security by checking that target addresses of control-flow
11504 transfer instructions (such as indirect function call, function return,
11505 indirect jump) are valid. This prevents diverting the flow of control
11506 to an unexpected target. This is intended to protect against such
11507 threats as Return-oriented Programming (ROP), and similarly
11508 call/jmp-oriented programming (COP/JOP).
11510 The value @code{branch} tells the compiler to implement checking of
11511 validity of control-flow transfer at the point of indirect branch
11512 instructions, i.e. call/jmp instructions. The value @code{return}
11513 implements checking of validity at the point of returning from a
11514 function. The value @code{full} is an alias for specifying both
11515 @code{branch} and @code{return}. The value @code{none} turns off
11518 You can also use the @code{nocf_check} attribute to identify
11519 which functions and calls should be skipped from instrumentation
11520 (@pxref{Function Attributes}).
11522 Currently the x86 GNU/Linux target provides an implementation based
11523 on Intel Control-flow Enforcement Technology (CET). Instrumentation
11524 for x86 is controlled by target-specific options @option{-mcet},
11525 @option{-mibt} and @option{-mshstk} (@pxref{x86 Options}).
11527 @item -fstack-protector
11528 @opindex fstack-protector
11529 Emit extra code to check for buffer overflows, such as stack smashing
11530 attacks. This is done by adding a guard variable to functions with
11531 vulnerable objects. This includes functions that call @code{alloca}, and
11532 functions with buffers larger than 8 bytes. The guards are initialized
11533 when a function is entered and then checked when the function exits.
11534 If a guard check fails, an error message is printed and the program exits.
11536 @item -fstack-protector-all
11537 @opindex fstack-protector-all
11538 Like @option{-fstack-protector} except that all functions are protected.
11540 @item -fstack-protector-strong
11541 @opindex fstack-protector-strong
11542 Like @option{-fstack-protector} but includes additional functions to
11543 be protected --- those that have local array definitions, or have
11544 references to local frame addresses.
11546 @item -fstack-protector-explicit
11547 @opindex fstack-protector-explicit
11548 Like @option{-fstack-protector} but only protects those functions which
11549 have the @code{stack_protect} attribute.
11551 @item -fstack-check
11552 @opindex fstack-check
11553 Generate code to verify that you do not go beyond the boundary of the
11554 stack. You should specify this flag if you are running in an
11555 environment with multiple threads, but you only rarely need to specify it in
11556 a single-threaded environment since stack overflow is automatically
11557 detected on nearly all systems if there is only one stack.
11559 Note that this switch does not actually cause checking to be done; the
11560 operating system or the language runtime must do that. The switch causes
11561 generation of code to ensure that they see the stack being extended.
11563 You can additionally specify a string parameter: @samp{no} means no
11564 checking, @samp{generic} means force the use of old-style checking,
11565 @samp{specific} means use the best checking method and is equivalent
11566 to bare @option{-fstack-check}.
11568 Old-style checking is a generic mechanism that requires no specific
11569 target support in the compiler but comes with the following drawbacks:
11573 Modified allocation strategy for large objects: they are always
11574 allocated dynamically if their size exceeds a fixed threshold. Note this
11575 may change the semantics of some code.
11578 Fixed limit on the size of the static frame of functions: when it is
11579 topped by a particular function, stack checking is not reliable and
11580 a warning is issued by the compiler.
11583 Inefficiency: because of both the modified allocation strategy and the
11584 generic implementation, code performance is hampered.
11587 Note that old-style stack checking is also the fallback method for
11588 @samp{specific} if no target support has been added in the compiler.
11590 @samp{-fstack-check=} is designed for Ada's needs to detect infinite recursion
11591 and stack overflows. @samp{specific} is an excellent choice when compiling
11592 Ada code. It is not generally sufficient to protect against stack-clash
11593 attacks. To protect against those you want @samp{-fstack-clash-protection}.
11595 @item -fstack-clash-protection
11596 @opindex fstack-clash-protection
11597 Generate code to prevent stack clash style attacks. When this option is
11598 enabled, the compiler will only allocate one page of stack space at a time
11599 and each page is accessed immediately after allocation. Thus, it prevents
11600 allocations from jumping over any stack guard page provided by the
11603 Most targets do not fully support stack clash protection. However, on
11604 those targets @option{-fstack-clash-protection} will protect dynamic stack
11605 allocations. @option{-fstack-clash-protection} may also provide limited
11606 protection for static stack allocations if the target supports
11607 @option{-fstack-check=specific}.
11609 @item -fstack-limit-register=@var{reg}
11610 @itemx -fstack-limit-symbol=@var{sym}
11611 @itemx -fno-stack-limit
11612 @opindex fstack-limit-register
11613 @opindex fstack-limit-symbol
11614 @opindex fno-stack-limit
11615 Generate code to ensure that the stack does not grow beyond a certain value,
11616 either the value of a register or the address of a symbol. If a larger
11617 stack is required, a signal is raised at run time. For most targets,
11618 the signal is raised before the stack overruns the boundary, so
11619 it is possible to catch the signal without taking special precautions.
11621 For instance, if the stack starts at absolute address @samp{0x80000000}
11622 and grows downwards, you can use the flags
11623 @option{-fstack-limit-symbol=__stack_limit} and
11624 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11625 of 128KB@. Note that this may only work with the GNU linker.
11627 You can locally override stack limit checking by using the
11628 @code{no_stack_limit} function attribute (@pxref{Function Attributes}).
11630 @item -fsplit-stack
11631 @opindex fsplit-stack
11632 Generate code to automatically split the stack before it overflows.
11633 The resulting program has a discontiguous stack which can only
11634 overflow if the program is unable to allocate any more memory. This
11635 is most useful when running threaded programs, as it is no longer
11636 necessary to calculate a good stack size to use for each thread. This
11637 is currently only implemented for the x86 targets running
11640 When code compiled with @option{-fsplit-stack} calls code compiled
11641 without @option{-fsplit-stack}, there may not be much stack space
11642 available for the latter code to run. If compiling all code,
11643 including library code, with @option{-fsplit-stack} is not an option,
11644 then the linker can fix up these calls so that the code compiled
11645 without @option{-fsplit-stack} always has a large stack. Support for
11646 this is implemented in the gold linker in GNU binutils release 2.21
11649 @item -fvtable-verify=@r{[}std@r{|}preinit@r{|}none@r{]}
11650 @opindex fvtable-verify
11651 This option is only available when compiling C++ code.
11652 It turns on (or off, if using @option{-fvtable-verify=none}) the security
11653 feature that verifies at run time, for every virtual call, that
11654 the vtable pointer through which the call is made is valid for the type of
11655 the object, and has not been corrupted or overwritten. If an invalid vtable
11656 pointer is detected at run time, an error is reported and execution of the
11657 program is immediately halted.
11659 This option causes run-time data structures to be built at program startup,
11660 which are used for verifying the vtable pointers.
11661 The options @samp{std} and @samp{preinit}
11662 control the timing of when these data structures are built. In both cases the
11663 data structures are built before execution reaches @code{main}. Using
11664 @option{-fvtable-verify=std} causes the data structures to be built after
11665 shared libraries have been loaded and initialized.
11666 @option{-fvtable-verify=preinit} causes them to be built before shared
11667 libraries have been loaded and initialized.
11669 If this option appears multiple times in the command line with different
11670 values specified, @samp{none} takes highest priority over both @samp{std} and
11671 @samp{preinit}; @samp{preinit} takes priority over @samp{std}.
11674 @opindex fvtv-debug
11675 When used in conjunction with @option{-fvtable-verify=std} or
11676 @option{-fvtable-verify=preinit}, causes debug versions of the
11677 runtime functions for the vtable verification feature to be called.
11678 This flag also causes the compiler to log information about which
11679 vtable pointers it finds for each class.
11680 This information is written to a file named @file{vtv_set_ptr_data.log}
11681 in the directory named by the environment variable @env{VTV_LOGS_DIR}
11682 if that is defined or the current working directory otherwise.
11684 Note: This feature @emph{appends} data to the log file. If you want a fresh log
11685 file, be sure to delete any existing one.
11688 @opindex fvtv-counts
11689 This is a debugging flag. When used in conjunction with
11690 @option{-fvtable-verify=std} or @option{-fvtable-verify=preinit}, this
11691 causes the compiler to keep track of the total number of virtual calls
11692 it encounters and the number of verifications it inserts. It also
11693 counts the number of calls to certain run-time library functions
11694 that it inserts and logs this information for each compilation unit.
11695 The compiler writes this information to a file named
11696 @file{vtv_count_data.log} in the directory named by the environment
11697 variable @env{VTV_LOGS_DIR} if that is defined or the current working
11698 directory otherwise. It also counts the size of the vtable pointer sets
11699 for each class, and writes this information to @file{vtv_class_set_sizes.log}
11700 in the same directory.
11702 Note: This feature @emph{appends} data to the log files. To get fresh log
11703 files, be sure to delete any existing ones.
11705 @item -finstrument-functions
11706 @opindex finstrument-functions
11707 Generate instrumentation calls for entry and exit to functions. Just
11708 after function entry and just before function exit, the following
11709 profiling functions are called with the address of the current
11710 function and its call site. (On some platforms,
11711 @code{__builtin_return_address} does not work beyond the current
11712 function, so the call site information may not be available to the
11713 profiling functions otherwise.)
11716 void __cyg_profile_func_enter (void *this_fn,
11718 void __cyg_profile_func_exit (void *this_fn,
11722 The first argument is the address of the start of the current function,
11723 which may be looked up exactly in the symbol table.
11725 This instrumentation is also done for functions expanded inline in other
11726 functions. The profiling calls indicate where, conceptually, the
11727 inline function is entered and exited. This means that addressable
11728 versions of such functions must be available. If all your uses of a
11729 function are expanded inline, this may mean an additional expansion of
11730 code size. If you use @code{extern inline} in your C code, an
11731 addressable version of such functions must be provided. (This is
11732 normally the case anyway, but if you get lucky and the optimizer always
11733 expands the functions inline, you might have gotten away without
11734 providing static copies.)
11736 A function may be given the attribute @code{no_instrument_function}, in
11737 which case this instrumentation is not done. This can be used, for
11738 example, for the profiling functions listed above, high-priority
11739 interrupt routines, and any functions from which the profiling functions
11740 cannot safely be called (perhaps signal handlers, if the profiling
11741 routines generate output or allocate memory).
11743 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
11744 @opindex finstrument-functions-exclude-file-list
11746 Set the list of functions that are excluded from instrumentation (see
11747 the description of @option{-finstrument-functions}). If the file that
11748 contains a function definition matches with one of @var{file}, then
11749 that function is not instrumented. The match is done on substrings:
11750 if the @var{file} parameter is a substring of the file name, it is
11751 considered to be a match.
11756 -finstrument-functions-exclude-file-list=/bits/stl,include/sys
11760 excludes any inline function defined in files whose pathnames
11761 contain @file{/bits/stl} or @file{include/sys}.
11763 If, for some reason, you want to include letter @samp{,} in one of
11764 @var{sym}, write @samp{\,}. For example,
11765 @option{-finstrument-functions-exclude-file-list='\,\,tmp'}
11766 (note the single quote surrounding the option).
11768 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
11769 @opindex finstrument-functions-exclude-function-list
11771 This is similar to @option{-finstrument-functions-exclude-file-list},
11772 but this option sets the list of function names to be excluded from
11773 instrumentation. The function name to be matched is its user-visible
11774 name, such as @code{vector<int> blah(const vector<int> &)}, not the
11775 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
11776 match is done on substrings: if the @var{sym} parameter is a substring
11777 of the function name, it is considered to be a match. For C99 and C++
11778 extended identifiers, the function name must be given in UTF-8, not
11779 using universal character names.
11781 @item -fpatchable-function-entry=@var{N}[,@var{M}]
11782 @opindex fpatchable-function-entry
11783 Generate @var{N} NOPs right at the beginning
11784 of each function, with the function entry point before the @var{M}th NOP.
11785 If @var{M} is omitted, it defaults to @code{0} so the
11786 function entry points to the address just at the first NOP.
11787 The NOP instructions reserve extra space which can be used to patch in
11788 any desired instrumentation at run time, provided that the code segment
11789 is writable. The amount of space is controllable indirectly via
11790 the number of NOPs; the NOP instruction used corresponds to the instruction
11791 emitted by the internal GCC back-end interface @code{gen_nop}. This behavior
11792 is target-specific and may also depend on the architecture variant and/or
11793 other compilation options.
11795 For run-time identification, the starting addresses of these areas,
11796 which correspond to their respective function entries minus @var{M},
11797 are additionally collected in the @code{__patchable_function_entries}
11798 section of the resulting binary.
11800 Note that the value of @code{__attribute__ ((patchable_function_entry
11801 (N,M)))} takes precedence over command-line option
11802 @option{-fpatchable-function-entry=N,M}. This can be used to increase
11803 the area size or to remove it completely on a single function.
11804 If @code{N=0}, no pad location is recorded.
11806 The NOP instructions are inserted at---and maybe before, depending on
11807 @var{M}---the function entry address, even before the prologue.
11812 @node Preprocessor Options
11813 @section Options Controlling the Preprocessor
11814 @cindex preprocessor options
11815 @cindex options, preprocessor
11817 These options control the C preprocessor, which is run on each C source
11818 file before actual compilation.
11820 If you use the @option{-E} option, nothing is done except preprocessing.
11821 Some of these options make sense only together with @option{-E} because
11822 they cause the preprocessor output to be unsuitable for actual
11825 In addition to the options listed here, there are a number of options
11826 to control search paths for include files documented in
11827 @ref{Directory Options}.
11828 Options to control preprocessor diagnostics are listed in
11829 @ref{Warning Options}.
11832 @include cppopts.texi
11834 @item -Wp,@var{option}
11836 You can use @option{-Wp,@var{option}} to bypass the compiler driver
11837 and pass @var{option} directly through to the preprocessor. If
11838 @var{option} contains commas, it is split into multiple options at the
11839 commas. However, many options are modified, translated or interpreted
11840 by the compiler driver before being passed to the preprocessor, and
11841 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
11842 interface is undocumented and subject to change, so whenever possible
11843 you should avoid using @option{-Wp} and let the driver handle the
11846 @item -Xpreprocessor @var{option}
11847 @opindex Xpreprocessor
11848 Pass @var{option} as an option to the preprocessor. You can use this to
11849 supply system-specific preprocessor options that GCC does not
11852 If you want to pass an option that takes an argument, you must use
11853 @option{-Xpreprocessor} twice, once for the option and once for the argument.
11855 @item -no-integrated-cpp
11856 @opindex no-integrated-cpp
11857 Perform preprocessing as a separate pass before compilation.
11858 By default, GCC performs preprocessing as an integrated part of
11859 input tokenization and parsing.
11860 If this option is provided, the appropriate language front end
11861 (@command{cc1}, @command{cc1plus}, or @command{cc1obj} for C, C++,
11862 and Objective-C, respectively) is instead invoked twice,
11863 once for preprocessing only and once for actual compilation
11864 of the preprocessed input.
11865 This option may be useful in conjunction with the @option{-B} or
11866 @option{-wrapper} options to specify an alternate preprocessor or
11867 perform additional processing of the program source between
11868 normal preprocessing and compilation.
11872 @node Assembler Options
11873 @section Passing Options to the Assembler
11875 @c prevent bad page break with this line
11876 You can pass options to the assembler.
11879 @item -Wa,@var{option}
11881 Pass @var{option} as an option to the assembler. If @var{option}
11882 contains commas, it is split into multiple options at the commas.
11884 @item -Xassembler @var{option}
11885 @opindex Xassembler
11886 Pass @var{option} as an option to the assembler. You can use this to
11887 supply system-specific assembler options that GCC does not
11890 If you want to pass an option that takes an argument, you must use
11891 @option{-Xassembler} twice, once for the option and once for the argument.
11896 @section Options for Linking
11897 @cindex link options
11898 @cindex options, linking
11900 These options come into play when the compiler links object files into
11901 an executable output file. They are meaningless if the compiler is
11902 not doing a link step.
11906 @item @var{object-file-name}
11907 A file name that does not end in a special recognized suffix is
11908 considered to name an object file or library. (Object files are
11909 distinguished from libraries by the linker according to the file
11910 contents.) If linking is done, these object files are used as input
11919 If any of these options is used, then the linker is not run, and
11920 object file names should not be used as arguments. @xref{Overall
11924 @opindex fuse-ld=bfd
11925 Use the @command{bfd} linker instead of the default linker.
11927 @item -fuse-ld=gold
11928 @opindex fuse-ld=gold
11929 Use the @command{gold} linker instead of the default linker.
11932 @item -l@var{library}
11933 @itemx -l @var{library}
11935 Search the library named @var{library} when linking. (The second
11936 alternative with the library as a separate argument is only for
11937 POSIX compliance and is not recommended.)
11939 It makes a difference where in the command you write this option; the
11940 linker searches and processes libraries and object files in the order they
11941 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
11942 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
11943 to functions in @samp{z}, those functions may not be loaded.
11945 The linker searches a standard list of directories for the library,
11946 which is actually a file named @file{lib@var{library}.a}. The linker
11947 then uses this file as if it had been specified precisely by name.
11949 The directories searched include several standard system directories
11950 plus any that you specify with @option{-L}.
11952 Normally the files found this way are library files---archive files
11953 whose members are object files. The linker handles an archive file by
11954 scanning through it for members which define symbols that have so far
11955 been referenced but not defined. But if the file that is found is an
11956 ordinary object file, it is linked in the usual fashion. The only
11957 difference between using an @option{-l} option and specifying a file name
11958 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
11959 and searches several directories.
11963 You need this special case of the @option{-l} option in order to
11964 link an Objective-C or Objective-C++ program.
11966 @item -nostartfiles
11967 @opindex nostartfiles
11968 Do not use the standard system startup files when linking.
11969 The standard system libraries are used normally, unless @option{-nostdlib}
11970 or @option{-nodefaultlibs} is used.
11972 @item -nodefaultlibs
11973 @opindex nodefaultlibs
11974 Do not use the standard system libraries when linking.
11975 Only the libraries you specify are passed to the linker, and options
11976 specifying linkage of the system libraries, such as @option{-static-libgcc}
11977 or @option{-shared-libgcc}, are ignored.
11978 The standard startup files are used normally, unless @option{-nostartfiles}
11981 The compiler may generate calls to @code{memcmp},
11982 @code{memset}, @code{memcpy} and @code{memmove}.
11983 These entries are usually resolved by entries in
11984 libc. These entry points should be supplied through some other
11985 mechanism when this option is specified.
11989 Do not use the standard system startup files or libraries when linking.
11990 No startup files and only the libraries you specify are passed to
11991 the linker, and options specifying linkage of the system libraries, such as
11992 @option{-static-libgcc} or @option{-shared-libgcc}, are ignored.
11994 The compiler may generate calls to @code{memcmp}, @code{memset},
11995 @code{memcpy} and @code{memmove}.
11996 These entries are usually resolved by entries in
11997 libc. These entry points should be supplied through some other
11998 mechanism when this option is specified.
12000 @cindex @option{-lgcc}, use with @option{-nostdlib}
12001 @cindex @option{-nostdlib} and unresolved references
12002 @cindex unresolved references and @option{-nostdlib}
12003 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
12004 @cindex @option{-nodefaultlibs} and unresolved references
12005 @cindex unresolved references and @option{-nodefaultlibs}
12006 One of the standard libraries bypassed by @option{-nostdlib} and
12007 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
12008 which GCC uses to overcome shortcomings of particular machines, or special
12009 needs for some languages.
12010 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
12011 Collection (GCC) Internals},
12012 for more discussion of @file{libgcc.a}.)
12013 In most cases, you need @file{libgcc.a} even when you want to avoid
12014 other standard libraries. In other words, when you specify @option{-nostdlib}
12015 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
12016 This ensures that you have no unresolved references to internal GCC
12017 library subroutines.
12018 (An example of such an internal subroutine is @code{__main}, used to ensure C++
12019 constructors are called; @pxref{Collect2,,@code{collect2}, gccint,
12020 GNU Compiler Collection (GCC) Internals}.)
12024 Produce a dynamically linked position independent executable on targets
12025 that support it. For predictable results, you must also specify the same
12026 set of options used for compilation (@option{-fpie}, @option{-fPIE},
12027 or model suboptions) when you specify this linker option.
12031 Don't produce a dynamically linked position independent executable.
12034 @opindex static-pie
12035 Produce a static position independent executable on targets that support
12036 it. A static position independent executable is similar to a static
12037 executable, but can be loaded at any address without a dynamic linker.
12038 For predictable results, you must also specify the same set of options
12039 used for compilation (@option{-fpie}, @option{-fPIE}, or model
12040 suboptions) when you specify this linker option.
12044 Link with the POSIX threads library. This option is supported on
12045 GNU/Linux targets, most other Unix derivatives, and also on
12046 x86 Cygwin and MinGW targets. On some targets this option also sets
12047 flags for the preprocessor, so it should be used consistently for both
12048 compilation and linking.
12052 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
12053 that support it. This instructs the linker to add all symbols, not
12054 only used ones, to the dynamic symbol table. This option is needed
12055 for some uses of @code{dlopen} or to allow obtaining backtraces
12056 from within a program.
12060 Remove all symbol table and relocation information from the executable.
12064 On systems that support dynamic linking, this overrides @option{-pie}
12065 and prevents linking with the shared libraries. On other systems, this
12066 option has no effect.
12070 Produce a shared object which can then be linked with other objects to
12071 form an executable. Not all systems support this option. For predictable
12072 results, you must also specify the same set of options used for compilation
12073 (@option{-fpic}, @option{-fPIC}, or model suboptions) when
12074 you specify this linker option.@footnote{On some systems, @samp{gcc -shared}
12075 needs to build supplementary stub code for constructors to work. On
12076 multi-libbed systems, @samp{gcc -shared} must select the correct support
12077 libraries to link against. Failing to supply the correct flags may lead
12078 to subtle defects. Supplying them in cases where they are not necessary
12081 @item -shared-libgcc
12082 @itemx -static-libgcc
12083 @opindex shared-libgcc
12084 @opindex static-libgcc
12085 On systems that provide @file{libgcc} as a shared library, these options
12086 force the use of either the shared or static version, respectively.
12087 If no shared version of @file{libgcc} was built when the compiler was
12088 configured, these options have no effect.
12090 There are several situations in which an application should use the
12091 shared @file{libgcc} instead of the static version. The most common
12092 of these is when the application wishes to throw and catch exceptions
12093 across different shared libraries. In that case, each of the libraries
12094 as well as the application itself should use the shared @file{libgcc}.
12096 Therefore, the G++ and driver automatically adds @option{-shared-libgcc}
12097 whenever you build a shared library or a main executable, because C++
12098 programs typically use exceptions, so this is the right thing to do.
12100 If, instead, you use the GCC driver to create shared libraries, you may
12101 find that they are not always linked with the shared @file{libgcc}.
12102 If GCC finds, at its configuration time, that you have a non-GNU linker
12103 or a GNU linker that does not support option @option{--eh-frame-hdr},
12104 it links the shared version of @file{libgcc} into shared libraries
12105 by default. Otherwise, it takes advantage of the linker and optimizes
12106 away the linking with the shared version of @file{libgcc}, linking with
12107 the static version of libgcc by default. This allows exceptions to
12108 propagate through such shared libraries, without incurring relocation
12109 costs at library load time.
12111 However, if a library or main executable is supposed to throw or catch
12112 exceptions, you must link it using the G++ driver, as appropriate
12113 for the languages used in the program, or using the option
12114 @option{-shared-libgcc}, such that it is linked with the shared
12117 @item -static-libasan
12118 @opindex static-libasan
12119 When the @option{-fsanitize=address} option is used to link a program,
12120 the GCC driver automatically links against @option{libasan}. If
12121 @file{libasan} is available as a shared library, and the @option{-static}
12122 option is not used, then this links against the shared version of
12123 @file{libasan}. The @option{-static-libasan} option directs the GCC
12124 driver to link @file{libasan} statically, without necessarily linking
12125 other libraries statically.
12127 @item -static-libtsan
12128 @opindex static-libtsan
12129 When the @option{-fsanitize=thread} option is used to link a program,
12130 the GCC driver automatically links against @option{libtsan}. If
12131 @file{libtsan} is available as a shared library, and the @option{-static}
12132 option is not used, then this links against the shared version of
12133 @file{libtsan}. The @option{-static-libtsan} option directs the GCC
12134 driver to link @file{libtsan} statically, without necessarily linking
12135 other libraries statically.
12137 @item -static-liblsan
12138 @opindex static-liblsan
12139 When the @option{-fsanitize=leak} option is used to link a program,
12140 the GCC driver automatically links against @option{liblsan}. If
12141 @file{liblsan} is available as a shared library, and the @option{-static}
12142 option is not used, then this links against the shared version of
12143 @file{liblsan}. The @option{-static-liblsan} option directs the GCC
12144 driver to link @file{liblsan} statically, without necessarily linking
12145 other libraries statically.
12147 @item -static-libubsan
12148 @opindex static-libubsan
12149 When the @option{-fsanitize=undefined} option is used to link a program,
12150 the GCC driver automatically links against @option{libubsan}. If
12151 @file{libubsan} is available as a shared library, and the @option{-static}
12152 option is not used, then this links against the shared version of
12153 @file{libubsan}. The @option{-static-libubsan} option directs the GCC
12154 driver to link @file{libubsan} statically, without necessarily linking
12155 other libraries statically.
12157 @item -static-libmpx
12158 @opindex static-libmpx
12159 When the @option{-fcheck-pointer bounds} and @option{-mmpx} options are
12160 used to link a program, the GCC driver automatically links against
12161 @file{libmpx}. If @file{libmpx} is available as a shared library,
12162 and the @option{-static} option is not used, then this links against
12163 the shared version of @file{libmpx}. The @option{-static-libmpx}
12164 option directs the GCC driver to link @file{libmpx} statically,
12165 without necessarily linking other libraries statically.
12167 @item -static-libmpxwrappers
12168 @opindex static-libmpxwrappers
12169 When the @option{-fcheck-pointer bounds} and @option{-mmpx} options are used
12170 to link a program without also using @option{-fno-chkp-use-wrappers}, the
12171 GCC driver automatically links against @file{libmpxwrappers}. If
12172 @file{libmpxwrappers} is available as a shared library, and the
12173 @option{-static} option is not used, then this links against the shared
12174 version of @file{libmpxwrappers}. The @option{-static-libmpxwrappers}
12175 option directs the GCC driver to link @file{libmpxwrappers} statically,
12176 without necessarily linking other libraries statically.
12178 @item -static-libstdc++
12179 @opindex static-libstdc++
12180 When the @command{g++} program is used to link a C++ program, it
12181 normally automatically links against @option{libstdc++}. If
12182 @file{libstdc++} is available as a shared library, and the
12183 @option{-static} option is not used, then this links against the
12184 shared version of @file{libstdc++}. That is normally fine. However, it
12185 is sometimes useful to freeze the version of @file{libstdc++} used by
12186 the program without going all the way to a fully static link. The
12187 @option{-static-libstdc++} option directs the @command{g++} driver to
12188 link @file{libstdc++} statically, without necessarily linking other
12189 libraries statically.
12193 Bind references to global symbols when building a shared object. Warn
12194 about any unresolved references (unless overridden by the link editor
12195 option @option{-Xlinker -z -Xlinker defs}). Only a few systems support
12198 @item -T @var{script}
12200 @cindex linker script
12201 Use @var{script} as the linker script. This option is supported by most
12202 systems using the GNU linker. On some targets, such as bare-board
12203 targets without an operating system, the @option{-T} option may be required
12204 when linking to avoid references to undefined symbols.
12206 @item -Xlinker @var{option}
12208 Pass @var{option} as an option to the linker. You can use this to
12209 supply system-specific linker options that GCC does not recognize.
12211 If you want to pass an option that takes a separate argument, you must use
12212 @option{-Xlinker} twice, once for the option and once for the argument.
12213 For example, to pass @option{-assert definitions}, you must write
12214 @option{-Xlinker -assert -Xlinker definitions}. It does not work to write
12215 @option{-Xlinker "-assert definitions"}, because this passes the entire
12216 string as a single argument, which is not what the linker expects.
12218 When using the GNU linker, it is usually more convenient to pass
12219 arguments to linker options using the @option{@var{option}=@var{value}}
12220 syntax than as separate arguments. For example, you can specify
12221 @option{-Xlinker -Map=output.map} rather than
12222 @option{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
12223 this syntax for command-line options.
12225 @item -Wl,@var{option}
12227 Pass @var{option} as an option to the linker. If @var{option} contains
12228 commas, it is split into multiple options at the commas. You can use this
12229 syntax to pass an argument to the option.
12230 For example, @option{-Wl,-Map,output.map} passes @option{-Map output.map} to the
12231 linker. When using the GNU linker, you can also get the same effect with
12232 @option{-Wl,-Map=output.map}.
12234 @item -u @var{symbol}
12236 Pretend the symbol @var{symbol} is undefined, to force linking of
12237 library modules to define it. You can use @option{-u} multiple times with
12238 different symbols to force loading of additional library modules.
12240 @item -z @var{keyword}
12242 @option{-z} is passed directly on to the linker along with the keyword
12243 @var{keyword}. See the section in the documentation of your linker for
12244 permitted values and their meanings.
12247 @node Directory Options
12248 @section Options for Directory Search
12249 @cindex directory options
12250 @cindex options, directory search
12251 @cindex search path
12253 These options specify directories to search for header files, for
12254 libraries and for parts of the compiler:
12257 @include cppdiropts.texi
12259 @item -iplugindir=@var{dir}
12260 @opindex iplugindir=
12261 Set the directory to search for plugins that are passed
12262 by @option{-fplugin=@var{name}} instead of
12263 @option{-fplugin=@var{path}/@var{name}.so}. This option is not meant
12264 to be used by the user, but only passed by the driver.
12268 Add directory @var{dir} to the list of directories to be searched
12271 @item -B@var{prefix}
12273 This option specifies where to find the executables, libraries,
12274 include files, and data files of the compiler itself.
12276 The compiler driver program runs one or more of the subprograms
12277 @command{cpp}, @command{cc1}, @command{as} and @command{ld}. It tries
12278 @var{prefix} as a prefix for each program it tries to run, both with and
12279 without @samp{@var{machine}/@var{version}/} for the corresponding target
12280 machine and compiler version.
12282 For each subprogram to be run, the compiler driver first tries the
12283 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
12284 is not specified, the driver tries two standard prefixes,
12285 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
12286 those results in a file name that is found, the unmodified program
12287 name is searched for using the directories specified in your
12288 @env{PATH} environment variable.
12290 The compiler checks to see if the path provided by @option{-B}
12291 refers to a directory, and if necessary it adds a directory
12292 separator character at the end of the path.
12294 @option{-B} prefixes that effectively specify directory names also apply
12295 to libraries in the linker, because the compiler translates these
12296 options into @option{-L} options for the linker. They also apply to
12297 include files in the preprocessor, because the compiler translates these
12298 options into @option{-isystem} options for the preprocessor. In this case,
12299 the compiler appends @samp{include} to the prefix.
12301 The runtime support file @file{libgcc.a} can also be searched for using
12302 the @option{-B} prefix, if needed. If it is not found there, the two
12303 standard prefixes above are tried, and that is all. The file is left
12304 out of the link if it is not found by those means.
12306 Another way to specify a prefix much like the @option{-B} prefix is to use
12307 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
12310 As a special kludge, if the path provided by @option{-B} is
12311 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
12312 9, then it is replaced by @file{[dir/]include}. This is to help
12313 with boot-strapping the compiler.
12315 @item -no-canonical-prefixes
12316 @opindex no-canonical-prefixes
12317 Do not expand any symbolic links, resolve references to @samp{/../}
12318 or @samp{/./}, or make the path absolute when generating a relative
12321 @item --sysroot=@var{dir}
12323 Use @var{dir} as the logical root directory for headers and libraries.
12324 For example, if the compiler normally searches for headers in
12325 @file{/usr/include} and libraries in @file{/usr/lib}, it instead
12326 searches @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
12328 If you use both this option and the @option{-isysroot} option, then
12329 the @option{--sysroot} option applies to libraries, but the
12330 @option{-isysroot} option applies to header files.
12332 The GNU linker (beginning with version 2.16) has the necessary support
12333 for this option. If your linker does not support this option, the
12334 header file aspect of @option{--sysroot} still works, but the
12335 library aspect does not.
12337 @item --no-sysroot-suffix
12338 @opindex no-sysroot-suffix
12339 For some targets, a suffix is added to the root directory specified
12340 with @option{--sysroot}, depending on the other options used, so that
12341 headers may for example be found in
12342 @file{@var{dir}/@var{suffix}/usr/include} instead of
12343 @file{@var{dir}/usr/include}. This option disables the addition of
12348 @node Code Gen Options
12349 @section Options for Code Generation Conventions
12350 @cindex code generation conventions
12351 @cindex options, code generation
12352 @cindex run-time options
12354 These machine-independent options control the interface conventions
12355 used in code generation.
12357 Most of them have both positive and negative forms; the negative form
12358 of @option{-ffoo} is @option{-fno-foo}. In the table below, only
12359 one of the forms is listed---the one that is not the default. You
12360 can figure out the other form by either removing @samp{no-} or adding
12364 @item -fstack-reuse=@var{reuse-level}
12365 @opindex fstack_reuse
12366 This option controls stack space reuse for user declared local/auto variables
12367 and compiler generated temporaries. @var{reuse_level} can be @samp{all},
12368 @samp{named_vars}, or @samp{none}. @samp{all} enables stack reuse for all
12369 local variables and temporaries, @samp{named_vars} enables the reuse only for
12370 user defined local variables with names, and @samp{none} disables stack reuse
12371 completely. The default value is @samp{all}. The option is needed when the
12372 program extends the lifetime of a scoped local variable or a compiler generated
12373 temporary beyond the end point defined by the language. When a lifetime of
12374 a variable ends, and if the variable lives in memory, the optimizing compiler
12375 has the freedom to reuse its stack space with other temporaries or scoped
12376 local variables whose live range does not overlap with it. Legacy code extending
12377 local lifetime is likely to break with the stack reuse optimization.
12396 if (*p == 10) // out of scope use of local1
12407 A(int k) : i(k), j(k) @{ @}
12414 void foo(const A& ar)
12421 foo(A(10)); // temp object's lifetime ends when foo returns
12427 ap->i+= 10; // ap references out of scope temp whose space
12428 // is reused with a. What is the value of ap->i?
12433 The lifetime of a compiler generated temporary is well defined by the C++
12434 standard. When a lifetime of a temporary ends, and if the temporary lives
12435 in memory, the optimizing compiler has the freedom to reuse its stack
12436 space with other temporaries or scoped local variables whose live range
12437 does not overlap with it. However some of the legacy code relies on
12438 the behavior of older compilers in which temporaries' stack space is
12439 not reused, the aggressive stack reuse can lead to runtime errors. This
12440 option is used to control the temporary stack reuse optimization.
12444 This option generates traps for signed overflow on addition, subtraction,
12445 multiplication operations.
12446 The options @option{-ftrapv} and @option{-fwrapv} override each other, so using
12447 @option{-ftrapv} @option{-fwrapv} on the command-line results in
12448 @option{-fwrapv} being effective. Note that only active options override, so
12449 using @option{-ftrapv} @option{-fwrapv} @option{-fno-wrapv} on the command-line
12450 results in @option{-ftrapv} being effective.
12454 This option instructs the compiler to assume that signed arithmetic
12455 overflow of addition, subtraction and multiplication wraps around
12456 using twos-complement representation. This flag enables some optimizations
12457 and disables others.
12458 The options @option{-ftrapv} and @option{-fwrapv} override each other, so using
12459 @option{-ftrapv} @option{-fwrapv} on the command-line results in
12460 @option{-fwrapv} being effective. Note that only active options override, so
12461 using @option{-ftrapv} @option{-fwrapv} @option{-fno-wrapv} on the command-line
12462 results in @option{-ftrapv} being effective.
12465 @opindex fexceptions
12466 Enable exception handling. Generates extra code needed to propagate
12467 exceptions. For some targets, this implies GCC generates frame
12468 unwind information for all functions, which can produce significant data
12469 size overhead, although it does not affect execution. If you do not
12470 specify this option, GCC enables it by default for languages like
12471 C++ that normally require exception handling, and disables it for
12472 languages like C that do not normally require it. However, you may need
12473 to enable this option when compiling C code that needs to interoperate
12474 properly with exception handlers written in C++. You may also wish to
12475 disable this option if you are compiling older C++ programs that don't
12476 use exception handling.
12478 @item -fnon-call-exceptions
12479 @opindex fnon-call-exceptions
12480 Generate code that allows trapping instructions to throw exceptions.
12481 Note that this requires platform-specific runtime support that does
12482 not exist everywhere. Moreover, it only allows @emph{trapping}
12483 instructions to throw exceptions, i.e.@: memory references or floating-point
12484 instructions. It does not allow exceptions to be thrown from
12485 arbitrary signal handlers such as @code{SIGALRM}.
12487 @item -fdelete-dead-exceptions
12488 @opindex fdelete-dead-exceptions
12489 Consider that instructions that may throw exceptions but don't otherwise
12490 contribute to the execution of the program can be optimized away.
12491 This option is enabled by default for the Ada front end, as permitted by
12492 the Ada language specification.
12493 Optimization passes that cause dead exceptions to be removed are enabled independently at different optimization levels.
12495 @item -funwind-tables
12496 @opindex funwind-tables
12497 Similar to @option{-fexceptions}, except that it just generates any needed
12498 static data, but does not affect the generated code in any other way.
12499 You normally do not need to enable this option; instead, a language processor
12500 that needs this handling enables it on your behalf.
12502 @item -fasynchronous-unwind-tables
12503 @opindex fasynchronous-unwind-tables
12504 Generate unwind table in DWARF format, if supported by target machine. The
12505 table is exact at each instruction boundary, so it can be used for stack
12506 unwinding from asynchronous events (such as debugger or garbage collector).
12508 @item -fno-gnu-unique
12509 @opindex fno-gnu-unique
12510 On systems with recent GNU assembler and C library, the C++ compiler
12511 uses the @code{STB_GNU_UNIQUE} binding to make sure that definitions
12512 of template static data members and static local variables in inline
12513 functions are unique even in the presence of @code{RTLD_LOCAL}; this
12514 is necessary to avoid problems with a library used by two different
12515 @code{RTLD_LOCAL} plugins depending on a definition in one of them and
12516 therefore disagreeing with the other one about the binding of the
12517 symbol. But this causes @code{dlclose} to be ignored for affected
12518 DSOs; if your program relies on reinitialization of a DSO via
12519 @code{dlclose} and @code{dlopen}, you can use
12520 @option{-fno-gnu-unique}.
12522 @item -fpcc-struct-return
12523 @opindex fpcc-struct-return
12524 Return ``short'' @code{struct} and @code{union} values in memory like
12525 longer ones, rather than in registers. This convention is less
12526 efficient, but it has the advantage of allowing intercallability between
12527 GCC-compiled files and files compiled with other compilers, particularly
12528 the Portable C Compiler (pcc).
12530 The precise convention for returning structures in memory depends
12531 on the target configuration macros.
12533 Short structures and unions are those whose size and alignment match
12534 that of some integer type.
12536 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
12537 switch is not binary compatible with code compiled with the
12538 @option{-freg-struct-return} switch.
12539 Use it to conform to a non-default application binary interface.
12541 @item -freg-struct-return
12542 @opindex freg-struct-return
12543 Return @code{struct} and @code{union} values in registers when possible.
12544 This is more efficient for small structures than
12545 @option{-fpcc-struct-return}.
12547 If you specify neither @option{-fpcc-struct-return} nor
12548 @option{-freg-struct-return}, GCC defaults to whichever convention is
12549 standard for the target. If there is no standard convention, GCC
12550 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
12551 the principal compiler. In those cases, we can choose the standard, and
12552 we chose the more efficient register return alternative.
12554 @strong{Warning:} code compiled with the @option{-freg-struct-return}
12555 switch is not binary compatible with code compiled with the
12556 @option{-fpcc-struct-return} switch.
12557 Use it to conform to a non-default application binary interface.
12559 @item -fshort-enums
12560 @opindex fshort-enums
12561 Allocate to an @code{enum} type only as many bytes as it needs for the
12562 declared range of possible values. Specifically, the @code{enum} type
12563 is equivalent to the smallest integer type that has enough room.
12565 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
12566 code that is not binary compatible with code generated without that switch.
12567 Use it to conform to a non-default application binary interface.
12569 @item -fshort-wchar
12570 @opindex fshort-wchar
12571 Override the underlying type for @code{wchar_t} to be @code{short
12572 unsigned int} instead of the default for the target. This option is
12573 useful for building programs to run under WINE@.
12575 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
12576 code that is not binary compatible with code generated without that switch.
12577 Use it to conform to a non-default application binary interface.
12580 @opindex fno-common
12581 @cindex tentative definitions
12582 In C code, this option controls the placement of global variables
12583 defined without an initializer, known as @dfn{tentative definitions}
12584 in the C standard. Tentative definitions are distinct from declarations
12585 of a variable with the @code{extern} keyword, which do not allocate storage.
12587 Unix C compilers have traditionally allocated storage for
12588 uninitialized global variables in a common block. This allows the
12589 linker to resolve all tentative definitions of the same variable
12590 in different compilation units to the same object, or to a non-tentative
12592 This is the behavior specified by @option{-fcommon}, and is the default for
12593 GCC on most targets.
12594 On the other hand, this behavior is not required by ISO
12595 C, and on some targets may carry a speed or code size penalty on
12596 variable references.
12598 The @option{-fno-common} option specifies that the compiler should instead
12599 place uninitialized global variables in the data section of the object file.
12600 This inhibits the merging of tentative definitions by the linker so
12601 you get a multiple-definition error if the same
12602 variable is defined in more than one compilation unit.
12603 Compiling with @option{-fno-common} is useful on targets for which
12604 it provides better performance, or if you wish to verify that the
12605 program will work on other systems that always treat uninitialized
12606 variable definitions this way.
12610 Ignore the @code{#ident} directive.
12612 @item -finhibit-size-directive
12613 @opindex finhibit-size-directive
12614 Don't output a @code{.size} assembler directive, or anything else that
12615 would cause trouble if the function is split in the middle, and the
12616 two halves are placed at locations far apart in memory. This option is
12617 used when compiling @file{crtstuff.c}; you should not need to use it
12620 @item -fverbose-asm
12621 @opindex fverbose-asm
12622 Put extra commentary information in the generated assembly code to
12623 make it more readable. This option is generally only of use to those
12624 who actually need to read the generated assembly code (perhaps while
12625 debugging the compiler itself).
12627 @option{-fno-verbose-asm}, the default, causes the
12628 extra information to be omitted and is useful when comparing two assembler
12631 The added comments include:
12636 information on the compiler version and command-line options,
12639 the source code lines associated with the assembly instructions,
12640 in the form FILENAME:LINENUMBER:CONTENT OF LINE,
12643 hints on which high-level expressions correspond to
12644 the various assembly instruction operands.
12648 For example, given this C source file:
12656 for (i = 0; i < n; i++)
12663 compiling to (x86_64) assembly via @option{-S} and emitting the result
12664 direct to stdout via @option{-o} @option{-}
12667 gcc -S test.c -fverbose-asm -Os -o -
12670 gives output similar to this:
12674 # GNU C11 (GCC) version 7.0.0 20160809 (experimental) (x86_64-pc-linux-gnu)
12681 .type test, @@function
12685 # test.c:4: int total = 0;
12686 xorl %eax, %eax # <retval>
12687 # test.c:6: for (i = 0; i < n; i++)
12688 xorl %edx, %edx # i
12690 # test.c:6: for (i = 0; i < n; i++)
12691 cmpl %edi, %edx # n, i
12693 # test.c:7: total += i * i;
12694 movl %edx, %ecx # i, tmp92
12695 imull %edx, %ecx # i, tmp92
12696 # test.c:6: for (i = 0; i < n; i++)
12698 # test.c:7: total += i * i;
12699 addl %ecx, %eax # tmp92, <retval>
12707 .ident "GCC: (GNU) 7.0.0 20160809 (experimental)"
12708 .section .note.GNU-stack,"",@@progbits
12711 The comments are intended for humans rather than machines and hence the
12712 precise format of the comments is subject to change.
12714 @item -frecord-gcc-switches
12715 @opindex frecord-gcc-switches
12716 This switch causes the command line used to invoke the
12717 compiler to be recorded into the object file that is being created.
12718 This switch is only implemented on some targets and the exact format
12719 of the recording is target and binary file format dependent, but it
12720 usually takes the form of a section containing ASCII text. This
12721 switch is related to the @option{-fverbose-asm} switch, but that
12722 switch only records information in the assembler output file as
12723 comments, so it never reaches the object file.
12724 See also @option{-grecord-gcc-switches} for another
12725 way of storing compiler options into the object file.
12729 @cindex global offset table
12731 Generate position-independent code (PIC) suitable for use in a shared
12732 library, if supported for the target machine. Such code accesses all
12733 constant addresses through a global offset table (GOT)@. The dynamic
12734 loader resolves the GOT entries when the program starts (the dynamic
12735 loader is not part of GCC; it is part of the operating system). If
12736 the GOT size for the linked executable exceeds a machine-specific
12737 maximum size, you get an error message from the linker indicating that
12738 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
12739 instead. (These maximums are 8k on the SPARC, 28k on AArch64 and 32k
12740 on the m68k and RS/6000. The x86 has no such limit.)
12742 Position-independent code requires special support, and therefore works
12743 only on certain machines. For the x86, GCC supports PIC for System V
12744 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
12745 position-independent.
12747 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
12752 If supported for the target machine, emit position-independent code,
12753 suitable for dynamic linking and avoiding any limit on the size of the
12754 global offset table. This option makes a difference on AArch64, m68k,
12755 PowerPC and SPARC@.
12757 Position-independent code requires special support, and therefore works
12758 only on certain machines.
12760 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
12767 These options are similar to @option{-fpic} and @option{-fPIC}, but
12768 generated position independent code can be only linked into executables.
12769 Usually these options are used when @option{-pie} GCC option is
12770 used during linking.
12772 @option{-fpie} and @option{-fPIE} both define the macros
12773 @code{__pie__} and @code{__PIE__}. The macros have the value 1
12774 for @option{-fpie} and 2 for @option{-fPIE}.
12778 Do not use the PLT for external function calls in position-independent code.
12779 Instead, load the callee address at call sites from the GOT and branch to it.
12780 This leads to more efficient code by eliminating PLT stubs and exposing
12781 GOT loads to optimizations. On architectures such as 32-bit x86 where
12782 PLT stubs expect the GOT pointer in a specific register, this gives more
12783 register allocation freedom to the compiler.
12784 Lazy binding requires use of the PLT;
12785 with @option{-fno-plt} all external symbols are resolved at load time.
12787 Alternatively, the function attribute @code{noplt} can be used to avoid calls
12788 through the PLT for specific external functions.
12790 In position-dependent code, a few targets also convert calls to
12791 functions that are marked to not use the PLT to use the GOT instead.
12793 @item -fno-jump-tables
12794 @opindex fno-jump-tables
12795 Do not use jump tables for switch statements even where it would be
12796 more efficient than other code generation strategies. This option is
12797 of use in conjunction with @option{-fpic} or @option{-fPIC} for
12798 building code that forms part of a dynamic linker and cannot
12799 reference the address of a jump table. On some targets, jump tables
12800 do not require a GOT and this option is not needed.
12802 @item -ffixed-@var{reg}
12804 Treat the register named @var{reg} as a fixed register; generated code
12805 should never refer to it (except perhaps as a stack pointer, frame
12806 pointer or in some other fixed role).
12808 @var{reg} must be the name of a register. The register names accepted
12809 are machine-specific and are defined in the @code{REGISTER_NAMES}
12810 macro in the machine description macro file.
12812 This flag does not have a negative form, because it specifies a
12815 @item -fcall-used-@var{reg}
12816 @opindex fcall-used
12817 Treat the register named @var{reg} as an allocable register that is
12818 clobbered by function calls. It may be allocated for temporaries or
12819 variables that do not live across a call. Functions compiled this way
12820 do not save and restore the register @var{reg}.
12822 It is an error to use this flag with the frame pointer or stack pointer.
12823 Use of this flag for other registers that have fixed pervasive roles in
12824 the machine's execution model produces disastrous results.
12826 This flag does not have a negative form, because it specifies a
12829 @item -fcall-saved-@var{reg}
12830 @opindex fcall-saved
12831 Treat the register named @var{reg} as an allocable register saved by
12832 functions. It may be allocated even for temporaries or variables that
12833 live across a call. Functions compiled this way save and restore
12834 the register @var{reg} if they use it.
12836 It is an error to use this flag with the frame pointer or stack pointer.
12837 Use of this flag for other registers that have fixed pervasive roles in
12838 the machine's execution model produces disastrous results.
12840 A different sort of disaster results from the use of this flag for
12841 a register in which function values may be returned.
12843 This flag does not have a negative form, because it specifies a
12846 @item -fpack-struct[=@var{n}]
12847 @opindex fpack-struct
12848 Without a value specified, pack all structure members together without
12849 holes. When a value is specified (which must be a small power of two), pack
12850 structure members according to this value, representing the maximum
12851 alignment (that is, objects with default alignment requirements larger than
12852 this are output potentially unaligned at the next fitting location.
12854 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
12855 code that is not binary compatible with code generated without that switch.
12856 Additionally, it makes the code suboptimal.
12857 Use it to conform to a non-default application binary interface.
12859 @item -fleading-underscore
12860 @opindex fleading-underscore
12861 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
12862 change the way C symbols are represented in the object file. One use
12863 is to help link with legacy assembly code.
12865 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
12866 generate code that is not binary compatible with code generated without that
12867 switch. Use it to conform to a non-default application binary interface.
12868 Not all targets provide complete support for this switch.
12870 @item -ftls-model=@var{model}
12871 @opindex ftls-model
12872 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
12873 The @var{model} argument should be one of @samp{global-dynamic},
12874 @samp{local-dynamic}, @samp{initial-exec} or @samp{local-exec}.
12875 Note that the choice is subject to optimization: the compiler may use
12876 a more efficient model for symbols not visible outside of the translation
12877 unit, or if @option{-fpic} is not given on the command line.
12879 The default without @option{-fpic} is @samp{initial-exec}; with
12880 @option{-fpic} the default is @samp{global-dynamic}.
12882 @item -ftrampolines
12883 @opindex ftrampolines
12884 For targets that normally need trampolines for nested functions, always
12885 generate them instead of using descriptors. Otherwise, for targets that
12886 do not need them, like for example HP-PA or IA-64, do nothing.
12888 A trampoline is a small piece of code that is created at run time on the
12889 stack when the address of a nested function is taken, and is used to call
12890 the nested function indirectly. Therefore, it requires the stack to be
12891 made executable in order for the program to work properly.
12893 @option{-fno-trampolines} is enabled by default on a language by language
12894 basis to let the compiler avoid generating them, if it computes that this
12895 is safe, and replace them with descriptors. Descriptors are made up of data
12896 only, but the generated code must be prepared to deal with them. As of this
12897 writing, @option{-fno-trampolines} is enabled by default only for Ada.
12899 Moreover, code compiled with @option{-ftrampolines} and code compiled with
12900 @option{-fno-trampolines} are not binary compatible if nested functions are
12901 present. This option must therefore be used on a program-wide basis and be
12902 manipulated with extreme care.
12904 @item -fvisibility=@r{[}default@r{|}internal@r{|}hidden@r{|}protected@r{]}
12905 @opindex fvisibility
12906 Set the default ELF image symbol visibility to the specified option---all
12907 symbols are marked with this unless overridden within the code.
12908 Using this feature can very substantially improve linking and
12909 load times of shared object libraries, produce more optimized
12910 code, provide near-perfect API export and prevent symbol clashes.
12911 It is @strong{strongly} recommended that you use this in any shared objects
12914 Despite the nomenclature, @samp{default} always means public; i.e.,
12915 available to be linked against from outside the shared object.
12916 @samp{protected} and @samp{internal} are pretty useless in real-world
12917 usage so the only other commonly used option is @samp{hidden}.
12918 The default if @option{-fvisibility} isn't specified is
12919 @samp{default}, i.e., make every symbol public.
12921 A good explanation of the benefits offered by ensuring ELF
12922 symbols have the correct visibility is given by ``How To Write
12923 Shared Libraries'' by Ulrich Drepper (which can be found at
12924 @w{@uref{https://www.akkadia.org/drepper/}})---however a superior
12925 solution made possible by this option to marking things hidden when
12926 the default is public is to make the default hidden and mark things
12927 public. This is the norm with DLLs on Windows and with @option{-fvisibility=hidden}
12928 and @code{__attribute__ ((visibility("default")))} instead of
12929 @code{__declspec(dllexport)} you get almost identical semantics with
12930 identical syntax. This is a great boon to those working with
12931 cross-platform projects.
12933 For those adding visibility support to existing code, you may find
12934 @code{#pragma GCC visibility} of use. This works by you enclosing
12935 the declarations you wish to set visibility for with (for example)
12936 @code{#pragma GCC visibility push(hidden)} and
12937 @code{#pragma GCC visibility pop}.
12938 Bear in mind that symbol visibility should be viewed @strong{as
12939 part of the API interface contract} and thus all new code should
12940 always specify visibility when it is not the default; i.e., declarations
12941 only for use within the local DSO should @strong{always} be marked explicitly
12942 as hidden as so to avoid PLT indirection overheads---making this
12943 abundantly clear also aids readability and self-documentation of the code.
12944 Note that due to ISO C++ specification requirements, @code{operator new} and
12945 @code{operator delete} must always be of default visibility.
12947 Be aware that headers from outside your project, in particular system
12948 headers and headers from any other library you use, may not be
12949 expecting to be compiled with visibility other than the default. You
12950 may need to explicitly say @code{#pragma GCC visibility push(default)}
12951 before including any such headers.
12953 @code{extern} declarations are not affected by @option{-fvisibility}, so
12954 a lot of code can be recompiled with @option{-fvisibility=hidden} with
12955 no modifications. However, this means that calls to @code{extern}
12956 functions with no explicit visibility use the PLT, so it is more
12957 effective to use @code{__attribute ((visibility))} and/or
12958 @code{#pragma GCC visibility} to tell the compiler which @code{extern}
12959 declarations should be treated as hidden.
12961 Note that @option{-fvisibility} does affect C++ vague linkage
12962 entities. This means that, for instance, an exception class that is
12963 be thrown between DSOs must be explicitly marked with default
12964 visibility so that the @samp{type_info} nodes are unified between
12967 An overview of these techniques, their benefits and how to use them
12968 is at @uref{http://gcc.gnu.org/@/wiki/@/Visibility}.
12970 @item -fstrict-volatile-bitfields
12971 @opindex fstrict-volatile-bitfields
12972 This option should be used if accesses to volatile bit-fields (or other
12973 structure fields, although the compiler usually honors those types
12974 anyway) should use a single access of the width of the
12975 field's type, aligned to a natural alignment if possible. For
12976 example, targets with memory-mapped peripheral registers might require
12977 all such accesses to be 16 bits wide; with this flag you can
12978 declare all peripheral bit-fields as @code{unsigned short} (assuming short
12979 is 16 bits on these targets) to force GCC to use 16-bit accesses
12980 instead of, perhaps, a more efficient 32-bit access.
12982 If this option is disabled, the compiler uses the most efficient
12983 instruction. In the previous example, that might be a 32-bit load
12984 instruction, even though that accesses bytes that do not contain
12985 any portion of the bit-field, or memory-mapped registers unrelated to
12986 the one being updated.
12988 In some cases, such as when the @code{packed} attribute is applied to a
12989 structure field, it may not be possible to access the field with a single
12990 read or write that is correctly aligned for the target machine. In this
12991 case GCC falls back to generating multiple accesses rather than code that
12992 will fault or truncate the result at run time.
12994 Note: Due to restrictions of the C/C++11 memory model, write accesses are
12995 not allowed to touch non bit-field members. It is therefore recommended
12996 to define all bits of the field's type as bit-field members.
12998 The default value of this option is determined by the application binary
12999 interface for the target processor.
13001 @item -fsync-libcalls
13002 @opindex fsync-libcalls
13003 This option controls whether any out-of-line instance of the @code{__sync}
13004 family of functions may be used to implement the C++11 @code{__atomic}
13005 family of functions.
13007 The default value of this option is enabled, thus the only useful form
13008 of the option is @option{-fno-sync-libcalls}. This option is used in
13009 the implementation of the @file{libatomic} runtime library.
13013 @node Developer Options
13014 @section GCC Developer Options
13015 @cindex developer options
13016 @cindex debugging GCC
13017 @cindex debug dump options
13018 @cindex dump options
13019 @cindex compilation statistics
13021 This section describes command-line options that are primarily of
13022 interest to GCC developers, including options to support compiler
13023 testing and investigation of compiler bugs and compile-time
13024 performance problems. This includes options that produce debug dumps
13025 at various points in the compilation; that print statistics such as
13026 memory use and execution time; and that print information about GCC's
13027 configuration, such as where it searches for libraries. You should
13028 rarely need to use any of these options for ordinary compilation and
13033 @item -d@var{letters}
13034 @itemx -fdump-rtl-@var{pass}
13035 @itemx -fdump-rtl-@var{pass}=@var{filename}
13037 @opindex fdump-rtl-@var{pass}
13038 Says to make debugging dumps during compilation at times specified by
13039 @var{letters}. This is used for debugging the RTL-based passes of the
13040 compiler. The file names for most of the dumps are made by appending
13041 a pass number and a word to the @var{dumpname}, and the files are
13042 created in the directory of the output file. In case of
13043 @option{=@var{filename}} option, the dump is output on the given file
13044 instead of the pass numbered dump files. Note that the pass number is
13045 assigned as passes are registered into the pass manager. Most passes
13046 are registered in the order that they will execute and for these passes
13047 the number corresponds to the pass execution order. However, passes
13048 registered by plugins, passes specific to compilation targets, or
13049 passes that are otherwise registered after all the other passes are
13050 numbered higher than a pass named "final", even if they are executed
13051 earlier. @var{dumpname} is generated from the name of the output
13052 file if explicitly specified and not an executable, otherwise it is
13053 the basename of the source file.
13055 Some @option{-d@var{letters}} switches have different meaning when
13056 @option{-E} is used for preprocessing. @xref{Preprocessor Options},
13057 for information about preprocessor-specific dump options.
13059 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
13060 @option{-d} option @var{letters}. Here are the possible
13061 letters for use in @var{pass} and @var{letters}, and their meanings:
13065 @item -fdump-rtl-alignments
13066 @opindex fdump-rtl-alignments
13067 Dump after branch alignments have been computed.
13069 @item -fdump-rtl-asmcons
13070 @opindex fdump-rtl-asmcons
13071 Dump after fixing rtl statements that have unsatisfied in/out constraints.
13073 @item -fdump-rtl-auto_inc_dec
13074 @opindex fdump-rtl-auto_inc_dec
13075 Dump after auto-inc-dec discovery. This pass is only run on
13076 architectures that have auto inc or auto dec instructions.
13078 @item -fdump-rtl-barriers
13079 @opindex fdump-rtl-barriers
13080 Dump after cleaning up the barrier instructions.
13082 @item -fdump-rtl-bbpart
13083 @opindex fdump-rtl-bbpart
13084 Dump after partitioning hot and cold basic blocks.
13086 @item -fdump-rtl-bbro
13087 @opindex fdump-rtl-bbro
13088 Dump after block reordering.
13090 @item -fdump-rtl-btl1
13091 @itemx -fdump-rtl-btl2
13092 @opindex fdump-rtl-btl2
13093 @opindex fdump-rtl-btl2
13094 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
13095 after the two branch
13096 target load optimization passes.
13098 @item -fdump-rtl-bypass
13099 @opindex fdump-rtl-bypass
13100 Dump after jump bypassing and control flow optimizations.
13102 @item -fdump-rtl-combine
13103 @opindex fdump-rtl-combine
13104 Dump after the RTL instruction combination pass.
13106 @item -fdump-rtl-compgotos
13107 @opindex fdump-rtl-compgotos
13108 Dump after duplicating the computed gotos.
13110 @item -fdump-rtl-ce1
13111 @itemx -fdump-rtl-ce2
13112 @itemx -fdump-rtl-ce3
13113 @opindex fdump-rtl-ce1
13114 @opindex fdump-rtl-ce2
13115 @opindex fdump-rtl-ce3
13116 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
13117 @option{-fdump-rtl-ce3} enable dumping after the three
13118 if conversion passes.
13120 @item -fdump-rtl-cprop_hardreg
13121 @opindex fdump-rtl-cprop_hardreg
13122 Dump after hard register copy propagation.
13124 @item -fdump-rtl-csa
13125 @opindex fdump-rtl-csa
13126 Dump after combining stack adjustments.
13128 @item -fdump-rtl-cse1
13129 @itemx -fdump-rtl-cse2
13130 @opindex fdump-rtl-cse1
13131 @opindex fdump-rtl-cse2
13132 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
13133 the two common subexpression elimination passes.
13135 @item -fdump-rtl-dce
13136 @opindex fdump-rtl-dce
13137 Dump after the standalone dead code elimination passes.
13139 @item -fdump-rtl-dbr
13140 @opindex fdump-rtl-dbr
13141 Dump after delayed branch scheduling.
13143 @item -fdump-rtl-dce1
13144 @itemx -fdump-rtl-dce2
13145 @opindex fdump-rtl-dce1
13146 @opindex fdump-rtl-dce2
13147 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
13148 the two dead store elimination passes.
13150 @item -fdump-rtl-eh
13151 @opindex fdump-rtl-eh
13152 Dump after finalization of EH handling code.
13154 @item -fdump-rtl-eh_ranges
13155 @opindex fdump-rtl-eh_ranges
13156 Dump after conversion of EH handling range regions.
13158 @item -fdump-rtl-expand
13159 @opindex fdump-rtl-expand
13160 Dump after RTL generation.
13162 @item -fdump-rtl-fwprop1
13163 @itemx -fdump-rtl-fwprop2
13164 @opindex fdump-rtl-fwprop1
13165 @opindex fdump-rtl-fwprop2
13166 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
13167 dumping after the two forward propagation passes.
13169 @item -fdump-rtl-gcse1
13170 @itemx -fdump-rtl-gcse2
13171 @opindex fdump-rtl-gcse1
13172 @opindex fdump-rtl-gcse2
13173 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
13174 after global common subexpression elimination.
13176 @item -fdump-rtl-init-regs
13177 @opindex fdump-rtl-init-regs
13178 Dump after the initialization of the registers.
13180 @item -fdump-rtl-initvals
13181 @opindex fdump-rtl-initvals
13182 Dump after the computation of the initial value sets.
13184 @item -fdump-rtl-into_cfglayout
13185 @opindex fdump-rtl-into_cfglayout
13186 Dump after converting to cfglayout mode.
13188 @item -fdump-rtl-ira
13189 @opindex fdump-rtl-ira
13190 Dump after iterated register allocation.
13192 @item -fdump-rtl-jump
13193 @opindex fdump-rtl-jump
13194 Dump after the second jump optimization.
13196 @item -fdump-rtl-loop2
13197 @opindex fdump-rtl-loop2
13198 @option{-fdump-rtl-loop2} enables dumping after the rtl
13199 loop optimization passes.
13201 @item -fdump-rtl-mach
13202 @opindex fdump-rtl-mach
13203 Dump after performing the machine dependent reorganization pass, if that
13206 @item -fdump-rtl-mode_sw
13207 @opindex fdump-rtl-mode_sw
13208 Dump after removing redundant mode switches.
13210 @item -fdump-rtl-rnreg
13211 @opindex fdump-rtl-rnreg
13212 Dump after register renumbering.
13214 @item -fdump-rtl-outof_cfglayout
13215 @opindex fdump-rtl-outof_cfglayout
13216 Dump after converting from cfglayout mode.
13218 @item -fdump-rtl-peephole2
13219 @opindex fdump-rtl-peephole2
13220 Dump after the peephole pass.
13222 @item -fdump-rtl-postreload
13223 @opindex fdump-rtl-postreload
13224 Dump after post-reload optimizations.
13226 @item -fdump-rtl-pro_and_epilogue
13227 @opindex fdump-rtl-pro_and_epilogue
13228 Dump after generating the function prologues and epilogues.
13230 @item -fdump-rtl-sched1
13231 @itemx -fdump-rtl-sched2
13232 @opindex fdump-rtl-sched1
13233 @opindex fdump-rtl-sched2
13234 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
13235 after the basic block scheduling passes.
13237 @item -fdump-rtl-ree
13238 @opindex fdump-rtl-ree
13239 Dump after sign/zero extension elimination.
13241 @item -fdump-rtl-seqabstr
13242 @opindex fdump-rtl-seqabstr
13243 Dump after common sequence discovery.
13245 @item -fdump-rtl-shorten
13246 @opindex fdump-rtl-shorten
13247 Dump after shortening branches.
13249 @item -fdump-rtl-sibling
13250 @opindex fdump-rtl-sibling
13251 Dump after sibling call optimizations.
13253 @item -fdump-rtl-split1
13254 @itemx -fdump-rtl-split2
13255 @itemx -fdump-rtl-split3
13256 @itemx -fdump-rtl-split4
13257 @itemx -fdump-rtl-split5
13258 @opindex fdump-rtl-split1
13259 @opindex fdump-rtl-split2
13260 @opindex fdump-rtl-split3
13261 @opindex fdump-rtl-split4
13262 @opindex fdump-rtl-split5
13263 These options enable dumping after five rounds of
13264 instruction splitting.
13266 @item -fdump-rtl-sms
13267 @opindex fdump-rtl-sms
13268 Dump after modulo scheduling. This pass is only run on some
13271 @item -fdump-rtl-stack
13272 @opindex fdump-rtl-stack
13273 Dump after conversion from GCC's ``flat register file'' registers to the
13274 x87's stack-like registers. This pass is only run on x86 variants.
13276 @item -fdump-rtl-subreg1
13277 @itemx -fdump-rtl-subreg2
13278 @opindex fdump-rtl-subreg1
13279 @opindex fdump-rtl-subreg2
13280 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
13281 the two subreg expansion passes.
13283 @item -fdump-rtl-unshare
13284 @opindex fdump-rtl-unshare
13285 Dump after all rtl has been unshared.
13287 @item -fdump-rtl-vartrack
13288 @opindex fdump-rtl-vartrack
13289 Dump after variable tracking.
13291 @item -fdump-rtl-vregs
13292 @opindex fdump-rtl-vregs
13293 Dump after converting virtual registers to hard registers.
13295 @item -fdump-rtl-web
13296 @opindex fdump-rtl-web
13297 Dump after live range splitting.
13299 @item -fdump-rtl-regclass
13300 @itemx -fdump-rtl-subregs_of_mode_init
13301 @itemx -fdump-rtl-subregs_of_mode_finish
13302 @itemx -fdump-rtl-dfinit
13303 @itemx -fdump-rtl-dfinish
13304 @opindex fdump-rtl-regclass
13305 @opindex fdump-rtl-subregs_of_mode_init
13306 @opindex fdump-rtl-subregs_of_mode_finish
13307 @opindex fdump-rtl-dfinit
13308 @opindex fdump-rtl-dfinish
13309 These dumps are defined but always produce empty files.
13312 @itemx -fdump-rtl-all
13314 @opindex fdump-rtl-all
13315 Produce all the dumps listed above.
13319 Annotate the assembler output with miscellaneous debugging information.
13323 Dump all macro definitions, at the end of preprocessing, in addition to
13328 Produce a core dump whenever an error occurs.
13332 Annotate the assembler output with a comment indicating which
13333 pattern and alternative is used. The length of each instruction is
13338 Dump the RTL in the assembler output as a comment before each instruction.
13339 Also turns on @option{-dp} annotation.
13343 Just generate RTL for a function instead of compiling it. Usually used
13344 with @option{-fdump-rtl-expand}.
13347 @item -fdump-noaddr
13348 @opindex fdump-noaddr
13349 When doing debugging dumps, suppress address output. This makes it more
13350 feasible to use diff on debugging dumps for compiler invocations with
13351 different compiler binaries and/or different
13352 text / bss / data / heap / stack / dso start locations.
13355 @opindex freport-bug
13356 Collect and dump debug information into a temporary file if an
13357 internal compiler error (ICE) occurs.
13359 @item -fdump-unnumbered
13360 @opindex fdump-unnumbered
13361 When doing debugging dumps, suppress instruction numbers and address output.
13362 This makes it more feasible to use diff on debugging dumps for compiler
13363 invocations with different options, in particular with and without
13366 @item -fdump-unnumbered-links
13367 @opindex fdump-unnumbered-links
13368 When doing debugging dumps (see @option{-d} option above), suppress
13369 instruction numbers for the links to the previous and next instructions
13372 @item -fdump-ipa-@var{switch}
13374 Control the dumping at various stages of inter-procedural analysis
13375 language tree to a file. The file name is generated by appending a
13376 switch specific suffix to the source file name, and the file is created
13377 in the same directory as the output file. The following dumps are
13382 Enables all inter-procedural analysis dumps.
13385 Dumps information about call-graph optimization, unused function removal,
13386 and inlining decisions.
13389 Dump after function inlining.
13393 @item -fdump-lang-all
13394 @itemx -fdump-lang-@var{switch}
13395 @itemx -fdump-lang-@var{switch}-@var{options}
13396 @itemx -fdump-lang-@var{switch}-@var{options}=@var{filename}
13397 @opindex fdump-lang-all
13398 @opindex fdump-lang
13399 Control the dumping of language-specific information. The @var{options}
13400 and @var{filename} portions behave as described in the
13401 @option{-fdump-tree} option. The following @var{switch} values are
13407 Enable all language-specific dumps.
13410 Dump class hierarchy information. Virtual table information is emitted
13411 unless '@option{slim}' is specified. This option is applicable to C++ only.
13414 Dump the raw internal tree data. This option is applicable to C++ only.
13418 @item -fdump-passes
13419 @opindex fdump-passes
13420 Print on @file{stderr} the list of optimization passes that are turned
13421 on and off by the current command-line options.
13423 @item -fdump-statistics-@var{option}
13424 @opindex fdump-statistics
13425 Enable and control dumping of pass statistics in a separate file. The
13426 file name is generated by appending a suffix ending in
13427 @samp{.statistics} to the source file name, and the file is created in
13428 the same directory as the output file. If the @samp{-@var{option}}
13429 form is used, @samp{-stats} causes counters to be summed over the
13430 whole compilation unit while @samp{-details} dumps every event as
13431 the passes generate them. The default with no option is to sum
13432 counters for each function compiled.
13434 @item -fdump-tree-all
13435 @itemx -fdump-tree-@var{switch}
13436 @itemx -fdump-tree-@var{switch}-@var{options}
13437 @itemx -fdump-tree-@var{switch}-@var{options}=@var{filename}
13438 @opindex fdump-tree-all
13439 @opindex fdump-tree
13440 Control the dumping at various stages of processing the intermediate
13441 language tree to a file. The file name is generated by appending a
13442 switch-specific suffix to the source file name, and the file is
13443 created in the same directory as the output file. In case of
13444 @option{=@var{filename}} option, the dump is output on the given file
13445 instead of the auto named dump files. If the @samp{-@var{options}}
13446 form is used, @var{options} is a list of @samp{-} separated options
13447 which control the details of the dump. Not all options are applicable
13448 to all dumps; those that are not meaningful are ignored. The
13449 following options are available
13453 Print the address of each node. Usually this is not meaningful as it
13454 changes according to the environment and source file. Its primary use
13455 is for tying up a dump file with a debug environment.
13457 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
13458 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
13459 use working backward from mangled names in the assembly file.
13461 When dumping front-end intermediate representations, inhibit dumping
13462 of members of a scope or body of a function merely because that scope
13463 has been reached. Only dump such items when they are directly reachable
13464 by some other path.
13466 When dumping pretty-printed trees, this option inhibits dumping the
13467 bodies of control structures.
13469 When dumping RTL, print the RTL in slim (condensed) form instead of
13470 the default LISP-like representation.
13472 Print a raw representation of the tree. By default, trees are
13473 pretty-printed into a C-like representation.
13475 Enable more detailed dumps (not honored by every dump option). Also
13476 include information from the optimization passes.
13478 Enable dumping various statistics about the pass (not honored by every dump
13481 Enable showing basic block boundaries (disabled in raw dumps).
13483 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
13484 dump a representation of the control flow graph suitable for viewing with
13485 GraphViz to @file{@var{file}.@var{passid}.@var{pass}.dot}. Each function in
13486 the file is pretty-printed as a subgraph, so that GraphViz can render them
13487 all in a single plot.
13489 This option currently only works for RTL dumps, and the RTL is always
13490 dumped in slim form.
13492 Enable showing virtual operands for every statement.
13494 Enable showing line numbers for statements.
13496 Enable showing the unique ID (@code{DECL_UID}) for each variable.
13498 Enable showing the tree dump for each statement.
13500 Enable showing the EH region number holding each statement.
13502 Enable showing scalar evolution analysis details.
13504 Enable showing optimization information (only available in certain
13507 Enable showing missed optimization information (only available in certain
13510 Enable other detailed optimization information (only available in
13512 @item =@var{filename}
13513 Instead of an auto named dump file, output into the given file
13514 name. The file names @file{stdout} and @file{stderr} are treated
13515 specially and are considered already open standard streams. For
13519 gcc -O2 -ftree-vectorize -fdump-tree-vect-blocks=foo.dump
13520 -fdump-tree-pre=/dev/stderr file.c
13523 outputs vectorizer dump into @file{foo.dump}, while the PRE dump is
13524 output on to @file{stderr}. If two conflicting dump filenames are
13525 given for the same pass, then the latter option overrides the earlier
13529 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
13530 and @option{lineno}.
13533 Turn on all optimization options, i.e., @option{optimized},
13534 @option{missed}, and @option{note}.
13537 To determine what tree dumps are available or find the dump for a pass
13538 of interest follow the steps below.
13542 Invoke GCC with @option{-fdump-passes} and in the @file{stderr} output
13543 look for a code that corresponds to the pass you are interested in.
13544 For example, the codes @code{tree-evrp}, @code{tree-vrp1}, and
13545 @code{tree-vrp2} correspond to the three Value Range Propagation passes.
13546 The number at the end distinguishes distinct invocations of the same pass.
13548 To enable the creation of the dump file, append the pass code to
13549 the @option{-fdump-} option prefix and invoke GCC with it. For example,
13550 to enable the dump from the Early Value Range Propagation pass, invoke
13551 GCC with the @option{-fdump-tree-evrp} option. Optionally, you may
13552 specify the name of the dump file. If you don't specify one, GCC
13553 creates as described below.
13555 Find the pass dump in a file whose name is composed of three components
13556 separated by a period: the name of the source file GCC was invoked to
13557 compile, a numeric suffix indicating the pass number followed by the
13558 letter @samp{t} for tree passes (and the letter @samp{r} for RTL passes),
13559 and finally the pass code. For example, the Early VRP pass dump might
13560 be in a file named @file{myfile.c.038t.evrp} in the current working
13561 directory. Note that the numeric codes are not stable and may change
13562 from one version of GCC to another.
13566 @itemx -fopt-info-@var{options}
13567 @itemx -fopt-info-@var{options}=@var{filename}
13569 Controls optimization dumps from various optimization passes. If the
13570 @samp{-@var{options}} form is used, @var{options} is a list of
13571 @samp{-} separated option keywords to select the dump details and
13574 The @var{options} can be divided into two groups: options describing the
13575 verbosity of the dump, and options describing which optimizations
13576 should be included. The options from both the groups can be freely
13577 mixed as they are non-overlapping. However, in case of any conflicts,
13578 the later options override the earlier options on the command
13581 The following options control the dump verbosity:
13585 Print information when an optimization is successfully applied. It is
13586 up to a pass to decide which information is relevant. For example, the
13587 vectorizer passes print the source location of loops which are
13588 successfully vectorized.
13590 Print information about missed optimizations. Individual passes
13591 control which information to include in the output.
13593 Print verbose information about optimizations, such as certain
13594 transformations, more detailed messages about decisions etc.
13596 Print detailed optimization information. This includes
13597 @samp{optimized}, @samp{missed}, and @samp{note}.
13600 One or more of the following option keywords can be used to describe a
13601 group of optimizations:
13605 Enable dumps from all interprocedural optimizations.
13607 Enable dumps from all loop optimizations.
13609 Enable dumps from all inlining optimizations.
13611 Enable dumps from all OMP (Offloading and Multi Processing) optimizations.
13613 Enable dumps from all vectorization optimizations.
13615 Enable dumps from all optimizations. This is a superset of
13616 the optimization groups listed above.
13619 If @var{options} is
13620 omitted, it defaults to @samp{optimized-optall}, which means to dump all
13621 info about successful optimizations from all the passes.
13623 If the @var{filename} is provided, then the dumps from all the
13624 applicable optimizations are concatenated into the @var{filename}.
13625 Otherwise the dump is output onto @file{stderr}. Though multiple
13626 @option{-fopt-info} options are accepted, only one of them can include
13627 a @var{filename}. If other filenames are provided then all but the
13628 first such option are ignored.
13630 Note that the output @var{filename} is overwritten
13631 in case of multiple translation units. If a combined output from
13632 multiple translation units is desired, @file{stderr} should be used
13635 In the following example, the optimization info is output to
13644 gcc -O3 -fopt-info-missed=missed.all
13648 outputs missed optimization report from all the passes into
13649 @file{missed.all}, and this one:
13652 gcc -O2 -ftree-vectorize -fopt-info-vec-missed
13656 prints information about missed optimization opportunities from
13657 vectorization passes on @file{stderr}.
13658 Note that @option{-fopt-info-vec-missed} is equivalent to
13659 @option{-fopt-info-missed-vec}. The order of the optimization group
13660 names and message types listed after @option{-fopt-info} does not matter.
13662 As another example,
13664 gcc -O3 -fopt-info-inline-optimized-missed=inline.txt
13668 outputs information about missed optimizations as well as
13669 optimized locations from all the inlining passes into
13675 gcc -fopt-info-vec-missed=vec.miss -fopt-info-loop-optimized=loop.opt
13679 Here the two output filenames @file{vec.miss} and @file{loop.opt} are
13680 in conflict since only one output file is allowed. In this case, only
13681 the first option takes effect and the subsequent options are
13682 ignored. Thus only @file{vec.miss} is produced which contains
13683 dumps from the vectorizer about missed opportunities.
13685 @item -fsched-verbose=@var{n}
13686 @opindex fsched-verbose
13687 On targets that use instruction scheduling, this option controls the
13688 amount of debugging output the scheduler prints to the dump files.
13690 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
13691 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
13692 For @var{n} greater than one, it also output basic block probabilities,
13693 detailed ready list information and unit/insn info. For @var{n} greater
13694 than two, it includes RTL at abort point, control-flow and regions info.
13695 And for @var{n} over four, @option{-fsched-verbose} also includes
13700 @item -fenable-@var{kind}-@var{pass}
13701 @itemx -fdisable-@var{kind}-@var{pass}=@var{range-list}
13705 This is a set of options that are used to explicitly disable/enable
13706 optimization passes. These options are intended for use for debugging GCC.
13707 Compiler users should use regular options for enabling/disabling
13712 @item -fdisable-ipa-@var{pass}
13713 Disable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
13714 statically invoked in the compiler multiple times, the pass name should be
13715 appended with a sequential number starting from 1.
13717 @item -fdisable-rtl-@var{pass}
13718 @itemx -fdisable-rtl-@var{pass}=@var{range-list}
13719 Disable RTL pass @var{pass}. @var{pass} is the pass name. If the same pass is
13720 statically invoked in the compiler multiple times, the pass name should be
13721 appended with a sequential number starting from 1. @var{range-list} is a
13722 comma-separated list of function ranges or assembler names. Each range is a number
13723 pair separated by a colon. The range is inclusive in both ends. If the range
13724 is trivial, the number pair can be simplified as a single number. If the
13725 function's call graph node's @var{uid} falls within one of the specified ranges,
13726 the @var{pass} is disabled for that function. The @var{uid} is shown in the
13727 function header of a dump file, and the pass names can be dumped by using
13728 option @option{-fdump-passes}.
13730 @item -fdisable-tree-@var{pass}
13731 @itemx -fdisable-tree-@var{pass}=@var{range-list}
13732 Disable tree pass @var{pass}. See @option{-fdisable-rtl} for the description of
13735 @item -fenable-ipa-@var{pass}
13736 Enable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
13737 statically invoked in the compiler multiple times, the pass name should be
13738 appended with a sequential number starting from 1.
13740 @item -fenable-rtl-@var{pass}
13741 @itemx -fenable-rtl-@var{pass}=@var{range-list}
13742 Enable RTL pass @var{pass}. See @option{-fdisable-rtl} for option argument
13743 description and examples.
13745 @item -fenable-tree-@var{pass}
13746 @itemx -fenable-tree-@var{pass}=@var{range-list}
13747 Enable tree pass @var{pass}. See @option{-fdisable-rtl} for the description
13748 of option arguments.
13752 Here are some examples showing uses of these options.
13756 # disable ccp1 for all functions
13757 -fdisable-tree-ccp1
13758 # disable complete unroll for function whose cgraph node uid is 1
13759 -fenable-tree-cunroll=1
13760 # disable gcse2 for functions at the following ranges [1,1],
13761 # [300,400], and [400,1000]
13762 # disable gcse2 for functions foo and foo2
13763 -fdisable-rtl-gcse2=foo,foo2
13764 # disable early inlining
13765 -fdisable-tree-einline
13766 # disable ipa inlining
13767 -fdisable-ipa-inline
13768 # enable tree full unroll
13769 -fenable-tree-unroll
13774 @itemx -fchecking=@var{n}
13776 @opindex fno-checking
13777 Enable internal consistency checking. The default depends on
13778 the compiler configuration. @option{-fchecking=2} enables further
13779 internal consistency checking that might affect code generation.
13781 @item -frandom-seed=@var{string}
13782 @opindex frandom-seed
13783 This option provides a seed that GCC uses in place of
13784 random numbers in generating certain symbol names
13785 that have to be different in every compiled file. It is also used to
13786 place unique stamps in coverage data files and the object files that
13787 produce them. You can use the @option{-frandom-seed} option to produce
13788 reproducibly identical object files.
13790 The @var{string} can either be a number (decimal, octal or hex) or an
13791 arbitrary string (in which case it's converted to a number by
13794 The @var{string} should be different for every file you compile.
13797 @itemx -save-temps=cwd
13798 @opindex save-temps
13799 Store the usual ``temporary'' intermediate files permanently; place them
13800 in the current directory and name them based on the source file. Thus,
13801 compiling @file{foo.c} with @option{-c -save-temps} produces files
13802 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
13803 preprocessed @file{foo.i} output file even though the compiler now
13804 normally uses an integrated preprocessor.
13806 When used in combination with the @option{-x} command-line option,
13807 @option{-save-temps} is sensible enough to avoid over writing an
13808 input source file with the same extension as an intermediate file.
13809 The corresponding intermediate file may be obtained by renaming the
13810 source file before using @option{-save-temps}.
13812 If you invoke GCC in parallel, compiling several different source
13813 files that share a common base name in different subdirectories or the
13814 same source file compiled for multiple output destinations, it is
13815 likely that the different parallel compilers will interfere with each
13816 other, and overwrite the temporary files. For instance:
13819 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
13820 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
13823 may result in @file{foo.i} and @file{foo.o} being written to
13824 simultaneously by both compilers.
13826 @item -save-temps=obj
13827 @opindex save-temps=obj
13828 Store the usual ``temporary'' intermediate files permanently. If the
13829 @option{-o} option is used, the temporary files are based on the
13830 object file. If the @option{-o} option is not used, the
13831 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
13836 gcc -save-temps=obj -c foo.c
13837 gcc -save-temps=obj -c bar.c -o dir/xbar.o
13838 gcc -save-temps=obj foobar.c -o dir2/yfoobar
13842 creates @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
13843 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
13844 @file{dir2/yfoobar.o}.
13846 @item -time@r{[}=@var{file}@r{]}
13848 Report the CPU time taken by each subprocess in the compilation
13849 sequence. For C source files, this is the compiler proper and assembler
13850 (plus the linker if linking is done).
13852 Without the specification of an output file, the output looks like this:
13859 The first number on each line is the ``user time'', that is time spent
13860 executing the program itself. The second number is ``system time'',
13861 time spent executing operating system routines on behalf of the program.
13862 Both numbers are in seconds.
13864 With the specification of an output file, the output is appended to the
13865 named file, and it looks like this:
13868 0.12 0.01 cc1 @var{options}
13869 0.00 0.01 as @var{options}
13872 The ``user time'' and the ``system time'' are moved before the program
13873 name, and the options passed to the program are displayed, so that one
13874 can later tell what file was being compiled, and with which options.
13876 @item -fdump-final-insns@r{[}=@var{file}@r{]}
13877 @opindex fdump-final-insns
13878 Dump the final internal representation (RTL) to @var{file}. If the
13879 optional argument is omitted (or if @var{file} is @code{.}), the name
13880 of the dump file is determined by appending @code{.gkd} to the
13881 compilation output file name.
13883 @item -fcompare-debug@r{[}=@var{opts}@r{]}
13884 @opindex fcompare-debug
13885 @opindex fno-compare-debug
13886 If no error occurs during compilation, run the compiler a second time,
13887 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
13888 passed to the second compilation. Dump the final internal
13889 representation in both compilations, and print an error if they differ.
13891 If the equal sign is omitted, the default @option{-gtoggle} is used.
13893 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
13894 and nonzero, implicitly enables @option{-fcompare-debug}. If
13895 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
13896 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
13899 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
13900 is equivalent to @option{-fno-compare-debug}, which disables the dumping
13901 of the final representation and the second compilation, preventing even
13902 @env{GCC_COMPARE_DEBUG} from taking effect.
13904 To verify full coverage during @option{-fcompare-debug} testing, set
13905 @env{GCC_COMPARE_DEBUG} to say @option{-fcompare-debug-not-overridden},
13906 which GCC rejects as an invalid option in any actual compilation
13907 (rather than preprocessing, assembly or linking). To get just a
13908 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
13909 not overridden} will do.
13911 @item -fcompare-debug-second
13912 @opindex fcompare-debug-second
13913 This option is implicitly passed to the compiler for the second
13914 compilation requested by @option{-fcompare-debug}, along with options to
13915 silence warnings, and omitting other options that would cause
13916 side-effect compiler outputs to files or to the standard output. Dump
13917 files and preserved temporary files are renamed so as to contain the
13918 @code{.gk} additional extension during the second compilation, to avoid
13919 overwriting those generated by the first.
13921 When this option is passed to the compiler driver, it causes the
13922 @emph{first} compilation to be skipped, which makes it useful for little
13923 other than debugging the compiler proper.
13927 Turn off generation of debug info, if leaving out this option
13928 generates it, or turn it on at level 2 otherwise. The position of this
13929 argument in the command line does not matter; it takes effect after all
13930 other options are processed, and it does so only once, no matter how
13931 many times it is given. This is mainly intended to be used with
13932 @option{-fcompare-debug}.
13934 @item -fvar-tracking-assignments-toggle
13935 @opindex fvar-tracking-assignments-toggle
13936 @opindex fno-var-tracking-assignments-toggle
13937 Toggle @option{-fvar-tracking-assignments}, in the same way that
13938 @option{-gtoggle} toggles @option{-g}.
13942 Makes the compiler print out each function name as it is compiled, and
13943 print some statistics about each pass when it finishes.
13945 @item -ftime-report
13946 @opindex ftime-report
13947 Makes the compiler print some statistics about the time consumed by each
13948 pass when it finishes.
13950 @item -ftime-report-details
13951 @opindex ftime-report-details
13952 Record the time consumed by infrastructure parts separately for each pass.
13954 @item -fira-verbose=@var{n}
13955 @opindex fira-verbose
13956 Control the verbosity of the dump file for the integrated register allocator.
13957 The default value is 5. If the value @var{n} is greater or equal to 10,
13958 the dump output is sent to stderr using the same format as @var{n} minus 10.
13961 @opindex flto-report
13962 Prints a report with internal details on the workings of the link-time
13963 optimizer. The contents of this report vary from version to version.
13964 It is meant to be useful to GCC developers when processing object
13965 files in LTO mode (via @option{-flto}).
13967 Disabled by default.
13969 @item -flto-report-wpa
13970 @opindex flto-report-wpa
13971 Like @option{-flto-report}, but only print for the WPA phase of Link
13975 @opindex fmem-report
13976 Makes the compiler print some statistics about permanent memory
13977 allocation when it finishes.
13979 @item -fmem-report-wpa
13980 @opindex fmem-report-wpa
13981 Makes the compiler print some statistics about permanent memory
13982 allocation for the WPA phase only.
13984 @item -fpre-ipa-mem-report
13985 @opindex fpre-ipa-mem-report
13986 @item -fpost-ipa-mem-report
13987 @opindex fpost-ipa-mem-report
13988 Makes the compiler print some statistics about permanent memory
13989 allocation before or after interprocedural optimization.
13991 @item -fprofile-report
13992 @opindex fprofile-report
13993 Makes the compiler print some statistics about consistency of the
13994 (estimated) profile and effect of individual passes.
13996 @item -fstack-usage
13997 @opindex fstack-usage
13998 Makes the compiler output stack usage information for the program, on a
13999 per-function basis. The filename for the dump is made by appending
14000 @file{.su} to the @var{auxname}. @var{auxname} is generated from the name of
14001 the output file, if explicitly specified and it is not an executable,
14002 otherwise it is the basename of the source file. An entry is made up
14007 The name of the function.
14011 One or more qualifiers: @code{static}, @code{dynamic}, @code{bounded}.
14014 The qualifier @code{static} means that the function manipulates the stack
14015 statically: a fixed number of bytes are allocated for the frame on function
14016 entry and released on function exit; no stack adjustments are otherwise made
14017 in the function. The second field is this fixed number of bytes.
14019 The qualifier @code{dynamic} means that the function manipulates the stack
14020 dynamically: in addition to the static allocation described above, stack
14021 adjustments are made in the body of the function, for example to push/pop
14022 arguments around function calls. If the qualifier @code{bounded} is also
14023 present, the amount of these adjustments is bounded at compile time and
14024 the second field is an upper bound of the total amount of stack used by
14025 the function. If it is not present, the amount of these adjustments is
14026 not bounded at compile time and the second field only represents the
14031 Emit statistics about front-end processing at the end of the compilation.
14032 This option is supported only by the C++ front end, and
14033 the information is generally only useful to the G++ development team.
14035 @item -fdbg-cnt-list
14036 @opindex fdbg-cnt-list
14037 Print the name and the counter upper bound for all debug counters.
14040 @item -fdbg-cnt=@var{counter-value-list}
14042 Set the internal debug counter upper bound. @var{counter-value-list}
14043 is a comma-separated list of @var{name}:@var{value} pairs
14044 which sets the upper bound of each debug counter @var{name} to @var{value}.
14045 All debug counters have the initial upper bound of @code{UINT_MAX};
14046 thus @code{dbg_cnt} returns true always unless the upper bound
14047 is set by this option.
14048 For example, with @option{-fdbg-cnt=dce:10,tail_call:0},
14049 @code{dbg_cnt(dce)} returns true only for first 10 invocations.
14051 @item -print-file-name=@var{library}
14052 @opindex print-file-name
14053 Print the full absolute name of the library file @var{library} that
14054 would be used when linking---and don't do anything else. With this
14055 option, GCC does not compile or link anything; it just prints the
14058 @item -print-multi-directory
14059 @opindex print-multi-directory
14060 Print the directory name corresponding to the multilib selected by any
14061 other switches present in the command line. This directory is supposed
14062 to exist in @env{GCC_EXEC_PREFIX}.
14064 @item -print-multi-lib
14065 @opindex print-multi-lib
14066 Print the mapping from multilib directory names to compiler switches
14067 that enable them. The directory name is separated from the switches by
14068 @samp{;}, and each switch starts with an @samp{@@} instead of the
14069 @samp{-}, without spaces between multiple switches. This is supposed to
14070 ease shell processing.
14072 @item -print-multi-os-directory
14073 @opindex print-multi-os-directory
14074 Print the path to OS libraries for the selected
14075 multilib, relative to some @file{lib} subdirectory. If OS libraries are
14076 present in the @file{lib} subdirectory and no multilibs are used, this is
14077 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
14078 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
14079 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
14080 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
14082 @item -print-multiarch
14083 @opindex print-multiarch
14084 Print the path to OS libraries for the selected multiarch,
14085 relative to some @file{lib} subdirectory.
14087 @item -print-prog-name=@var{program}
14088 @opindex print-prog-name
14089 Like @option{-print-file-name}, but searches for a program such as @command{cpp}.
14091 @item -print-libgcc-file-name
14092 @opindex print-libgcc-file-name
14093 Same as @option{-print-file-name=libgcc.a}.
14095 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
14096 but you do want to link with @file{libgcc.a}. You can do:
14099 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
14102 @item -print-search-dirs
14103 @opindex print-search-dirs
14104 Print the name of the configured installation directory and a list of
14105 program and library directories @command{gcc} searches---and don't do anything else.
14107 This is useful when @command{gcc} prints the error message
14108 @samp{installation problem, cannot exec cpp0: No such file or directory}.
14109 To resolve this you either need to put @file{cpp0} and the other compiler
14110 components where @command{gcc} expects to find them, or you can set the environment
14111 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
14112 Don't forget the trailing @samp{/}.
14113 @xref{Environment Variables}.
14115 @item -print-sysroot
14116 @opindex print-sysroot
14117 Print the target sysroot directory that is used during
14118 compilation. This is the target sysroot specified either at configure
14119 time or using the @option{--sysroot} option, possibly with an extra
14120 suffix that depends on compilation options. If no target sysroot is
14121 specified, the option prints nothing.
14123 @item -print-sysroot-headers-suffix
14124 @opindex print-sysroot-headers-suffix
14125 Print the suffix added to the target sysroot when searching for
14126 headers, or give an error if the compiler is not configured with such
14127 a suffix---and don't do anything else.
14130 @opindex dumpmachine
14131 Print the compiler's target machine (for example,
14132 @samp{i686-pc-linux-gnu})---and don't do anything else.
14135 @opindex dumpversion
14136 Print the compiler version (for example, @code{3.0}, @code{6.3.0} or @code{7})---and don't do
14137 anything else. This is the compiler version used in filesystem paths,
14138 specs, can be depending on how the compiler has been configured just
14139 a single number (major version), two numbers separated by dot (major and
14140 minor version) or three numbers separated by dots (major, minor and patchlevel
14143 @item -dumpfullversion
14144 @opindex dumpfullversion
14145 Print the full compiler version, always 3 numbers separated by dots,
14146 major, minor and patchlevel version.
14150 Print the compiler's built-in specs---and don't do anything else. (This
14151 is used when GCC itself is being built.) @xref{Spec Files}.
14154 @node Submodel Options
14155 @section Machine-Dependent Options
14156 @cindex submodel options
14157 @cindex specifying hardware config
14158 @cindex hardware models and configurations, specifying
14159 @cindex target-dependent options
14160 @cindex machine-dependent options
14162 Each target machine supported by GCC can have its own options---for
14163 example, to allow you to compile for a particular processor variant or
14164 ABI, or to control optimizations specific to that machine. By
14165 convention, the names of machine-specific options start with
14168 Some configurations of the compiler also support additional target-specific
14169 options, usually for compatibility with other compilers on the same
14172 @c This list is ordered alphanumerically by subsection name.
14173 @c It should be the same order and spelling as these options are listed
14174 @c in Machine Dependent Options
14177 * AArch64 Options::
14178 * Adapteva Epiphany Options::
14182 * Blackfin Options::
14187 * DEC Alpha Options::
14191 * GNU/Linux Options::
14201 * MicroBlaze Options::
14204 * MN10300 Options::
14208 * Nios II Options::
14209 * Nvidia PTX Options::
14211 * picoChip Options::
14212 * PowerPC Options::
14215 * RS/6000 and PowerPC Options::
14217 * S/390 and zSeries Options::
14220 * Solaris 2 Options::
14223 * System V Options::
14224 * TILE-Gx Options::
14225 * TILEPro Options::
14230 * VxWorks Options::
14232 * x86 Windows Options::
14233 * Xstormy16 Options::
14235 * zSeries Options::
14238 @node AArch64 Options
14239 @subsection AArch64 Options
14240 @cindex AArch64 Options
14242 These options are defined for AArch64 implementations:
14246 @item -mabi=@var{name}
14248 Generate code for the specified data model. Permissible values
14249 are @samp{ilp32} for SysV-like data model where int, long int and pointers
14250 are 32 bits, and @samp{lp64} for SysV-like data model where int is 32 bits,
14251 but long int and pointers are 64 bits.
14253 The default depends on the specific target configuration. Note that
14254 the LP64 and ILP32 ABIs are not link-compatible; you must compile your
14255 entire program with the same ABI, and link with a compatible set of libraries.
14258 @opindex mbig-endian
14259 Generate big-endian code. This is the default when GCC is configured for an
14260 @samp{aarch64_be-*-*} target.
14262 @item -mgeneral-regs-only
14263 @opindex mgeneral-regs-only
14264 Generate code which uses only the general-purpose registers. This will prevent
14265 the compiler from using floating-point and Advanced SIMD registers but will not
14266 impose any restrictions on the assembler.
14268 @item -mlittle-endian
14269 @opindex mlittle-endian
14270 Generate little-endian code. This is the default when GCC is configured for an
14271 @samp{aarch64-*-*} but not an @samp{aarch64_be-*-*} target.
14273 @item -mcmodel=tiny
14274 @opindex mcmodel=tiny
14275 Generate code for the tiny code model. The program and its statically defined
14276 symbols must be within 1MB of each other. Programs can be statically or
14277 dynamically linked.
14279 @item -mcmodel=small
14280 @opindex mcmodel=small
14281 Generate code for the small code model. The program and its statically defined
14282 symbols must be within 4GB of each other. Programs can be statically or
14283 dynamically linked. This is the default code model.
14285 @item -mcmodel=large
14286 @opindex mcmodel=large
14287 Generate code for the large code model. This makes no assumptions about
14288 addresses and sizes of sections. Programs can be statically linked only.
14290 @item -mstrict-align
14291 @opindex mstrict-align
14292 Avoid generating memory accesses that may not be aligned on a natural object
14293 boundary as described in the architecture specification.
14295 @item -momit-leaf-frame-pointer
14296 @itemx -mno-omit-leaf-frame-pointer
14297 @opindex momit-leaf-frame-pointer
14298 @opindex mno-omit-leaf-frame-pointer
14299 Omit or keep the frame pointer in leaf functions. The former behavior is the
14302 @item -mtls-dialect=desc
14303 @opindex mtls-dialect=desc
14304 Use TLS descriptors as the thread-local storage mechanism for dynamic accesses
14305 of TLS variables. This is the default.
14307 @item -mtls-dialect=traditional
14308 @opindex mtls-dialect=traditional
14309 Use traditional TLS as the thread-local storage mechanism for dynamic accesses
14312 @item -mtls-size=@var{size}
14314 Specify bit size of immediate TLS offsets. Valid values are 12, 24, 32, 48.
14315 This option requires binutils 2.26 or newer.
14317 @item -mfix-cortex-a53-835769
14318 @itemx -mno-fix-cortex-a53-835769
14319 @opindex mfix-cortex-a53-835769
14320 @opindex mno-fix-cortex-a53-835769
14321 Enable or disable the workaround for the ARM Cortex-A53 erratum number 835769.
14322 This involves inserting a NOP instruction between memory instructions and
14323 64-bit integer multiply-accumulate instructions.
14325 @item -mfix-cortex-a53-843419
14326 @itemx -mno-fix-cortex-a53-843419
14327 @opindex mfix-cortex-a53-843419
14328 @opindex mno-fix-cortex-a53-843419
14329 Enable or disable the workaround for the ARM Cortex-A53 erratum number 843419.
14330 This erratum workaround is made at link time and this will only pass the
14331 corresponding flag to the linker.
14333 @item -mlow-precision-recip-sqrt
14334 @item -mno-low-precision-recip-sqrt
14335 @opindex mlow-precision-recip-sqrt
14336 @opindex mno-low-precision-recip-sqrt
14337 Enable or disable the reciprocal square root approximation.
14338 This option only has an effect if @option{-ffast-math} or
14339 @option{-funsafe-math-optimizations} is used as well. Enabling this reduces
14340 precision of reciprocal square root results to about 16 bits for
14341 single precision and to 32 bits for double precision.
14343 @item -mlow-precision-sqrt
14344 @item -mno-low-precision-sqrt
14345 @opindex -mlow-precision-sqrt
14346 @opindex -mno-low-precision-sqrt
14347 Enable or disable the square root approximation.
14348 This option only has an effect if @option{-ffast-math} or
14349 @option{-funsafe-math-optimizations} is used as well. Enabling this reduces
14350 precision of square root results to about 16 bits for
14351 single precision and to 32 bits for double precision.
14352 If enabled, it implies @option{-mlow-precision-recip-sqrt}.
14354 @item -mlow-precision-div
14355 @item -mno-low-precision-div
14356 @opindex -mlow-precision-div
14357 @opindex -mno-low-precision-div
14358 Enable or disable the division approximation.
14359 This option only has an effect if @option{-ffast-math} or
14360 @option{-funsafe-math-optimizations} is used as well. Enabling this reduces
14361 precision of division results to about 16 bits for
14362 single precision and to 32 bits for double precision.
14364 @item -march=@var{name}
14366 Specify the name of the target architecture and, optionally, one or
14367 more feature modifiers. This option has the form
14368 @option{-march=@var{arch}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}.
14370 The permissible values for @var{arch} are @samp{armv8-a},
14371 @samp{armv8.1-a}, @samp{armv8.2-a}, @samp{armv8.3-a} or @var{native}.
14373 The value @samp{armv8.3-a} implies @samp{armv8.2-a} and enables compiler
14374 support for the ARMv8.3-A architecture extensions.
14376 The value @samp{armv8.2-a} implies @samp{armv8.1-a} and enables compiler
14377 support for the ARMv8.2-A architecture extensions.
14379 The value @samp{armv8.1-a} implies @samp{armv8-a} and enables compiler
14380 support for the ARMv8.1-A architecture extension. In particular, it
14381 enables the @samp{+crc}, @samp{+lse}, and @samp{+rdma} features.
14383 The value @samp{native} is available on native AArch64 GNU/Linux and
14384 causes the compiler to pick the architecture of the host system. This
14385 option has no effect if the compiler is unable to recognize the
14386 architecture of the host system,
14388 The permissible values for @var{feature} are listed in the sub-section
14389 on @ref{aarch64-feature-modifiers,,@option{-march} and @option{-mcpu}
14390 Feature Modifiers}. Where conflicting feature modifiers are
14391 specified, the right-most feature is used.
14393 GCC uses @var{name} to determine what kind of instructions it can emit
14394 when generating assembly code. If @option{-march} is specified
14395 without either of @option{-mtune} or @option{-mcpu} also being
14396 specified, the code is tuned to perform well across a range of target
14397 processors implementing the target architecture.
14399 @item -mtune=@var{name}
14401 Specify the name of the target processor for which GCC should tune the
14402 performance of the code. Permissible values for this option are:
14403 @samp{generic}, @samp{cortex-a35}, @samp{cortex-a53}, @samp{cortex-a55},
14404 @samp{cortex-a57}, @samp{cortex-a72}, @samp{cortex-a73}, @samp{cortex-a75},
14405 @samp{exynos-m1}, @samp{falkor}, @samp{qdf24xx}, @samp{saphira},
14406 @samp{xgene1}, @samp{vulcan}, @samp{thunderx},
14407 @samp{thunderxt88}, @samp{thunderxt88p1}, @samp{thunderxt81},
14408 @samp{thunderxt83}, @samp{thunderx2t99}, @samp{cortex-a57.cortex-a53},
14409 @samp{cortex-a72.cortex-a53}, @samp{cortex-a73.cortex-a35},
14410 @samp{cortex-a73.cortex-a53}, @samp{cortex-a75.cortex-a55},
14413 The values @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53},
14414 @samp{cortex-a73.cortex-a35}, @samp{cortex-a73.cortex-a53},
14415 @samp{cortex-a75.cortex-a55} specify that GCC should tune for a
14418 Additionally on native AArch64 GNU/Linux systems the value
14419 @samp{native} tunes performance to the host system. This option has no effect
14420 if the compiler is unable to recognize the processor of the host system.
14422 Where none of @option{-mtune=}, @option{-mcpu=} or @option{-march=}
14423 are specified, the code is tuned to perform well across a range
14424 of target processors.
14426 This option cannot be suffixed by feature modifiers.
14428 @item -mcpu=@var{name}
14430 Specify the name of the target processor, optionally suffixed by one
14431 or more feature modifiers. This option has the form
14432 @option{-mcpu=@var{cpu}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}, where
14433 the permissible values for @var{cpu} are the same as those available
14434 for @option{-mtune}. The permissible values for @var{feature} are
14435 documented in the sub-section on
14436 @ref{aarch64-feature-modifiers,,@option{-march} and @option{-mcpu}
14437 Feature Modifiers}. Where conflicting feature modifiers are
14438 specified, the right-most feature is used.
14440 GCC uses @var{name} to determine what kind of instructions it can emit when
14441 generating assembly code (as if by @option{-march}) and to determine
14442 the target processor for which to tune for performance (as if
14443 by @option{-mtune}). Where this option is used in conjunction
14444 with @option{-march} or @option{-mtune}, those options take precedence
14445 over the appropriate part of this option.
14447 @item -moverride=@var{string}
14449 Override tuning decisions made by the back-end in response to a
14450 @option{-mtune=} switch. The syntax, semantics, and accepted values
14451 for @var{string} in this option are not guaranteed to be consistent
14454 This option is only intended to be useful when developing GCC.
14456 @item -mpc-relative-literal-loads
14457 @itemx -mno-pc-relative-literal-loads
14458 @opindex mpc-relative-literal-loads
14459 @opindex mno-pc-relative-literal-loads
14460 Enable or disable PC-relative literal loads. With this option literal pools are
14461 accessed using a single instruction and emitted after each function. This
14462 limits the maximum size of functions to 1MB. This is enabled by default for
14463 @option{-mcmodel=tiny}.
14465 @item -msign-return-address=@var{scope}
14466 @opindex msign-return-address
14467 Select the function scope on which return address signing will be applied.
14468 Permissible values are @samp{none}, which disables return address signing,
14469 @samp{non-leaf}, which enables pointer signing for functions which are not leaf
14470 functions, and @samp{all}, which enables pointer signing for all functions. The
14471 default value is @samp{none}.
14475 @subsubsection @option{-march} and @option{-mcpu} Feature Modifiers
14476 @anchor{aarch64-feature-modifiers}
14477 @cindex @option{-march} feature modifiers
14478 @cindex @option{-mcpu} feature modifiers
14479 Feature modifiers used with @option{-march} and @option{-mcpu} can be any of
14480 the following and their inverses @option{no@var{feature}}:
14484 Enable CRC extension. This is on by default for
14485 @option{-march=armv8.1-a}.
14487 Enable Crypto extension. This also enables Advanced SIMD and floating-point
14490 Enable floating-point instructions. This is on by default for all possible
14491 values for options @option{-march} and @option{-mcpu}.
14493 Enable Advanced SIMD instructions. This also enables floating-point
14494 instructions. This is on by default for all possible values for options
14495 @option{-march} and @option{-mcpu}.
14497 Enable Large System Extension instructions. This is on by default for
14498 @option{-march=armv8.1-a}.
14500 Enable Round Double Multiply Accumulate instructions. This is on by default
14501 for @option{-march=armv8.1-a}.
14503 Enable FP16 extension. This also enables floating-point instructions.
14505 Enable the RcPc extension. This does not change code generation from GCC,
14506 but is passed on to the assembler, enabling inline asm statements to use
14507 instructions from the RcPc extension.
14509 Enable the Dot Product extension. This also enables Advanced SIMD instructions.
14513 Feature @option{crypto} implies @option{simd}, which implies @option{fp}.
14514 Conversely, @option{nofp} implies @option{nosimd}, which implies
14517 @node Adapteva Epiphany Options
14518 @subsection Adapteva Epiphany Options
14520 These @samp{-m} options are defined for Adapteva Epiphany:
14523 @item -mhalf-reg-file
14524 @opindex mhalf-reg-file
14525 Don't allocate any register in the range @code{r32}@dots{}@code{r63}.
14526 That allows code to run on hardware variants that lack these registers.
14528 @item -mprefer-short-insn-regs
14529 @opindex mprefer-short-insn-regs
14530 Preferentially allocate registers that allow short instruction generation.
14531 This can result in increased instruction count, so this may either reduce or
14532 increase overall code size.
14534 @item -mbranch-cost=@var{num}
14535 @opindex mbranch-cost
14536 Set the cost of branches to roughly @var{num} ``simple'' instructions.
14537 This cost is only a heuristic and is not guaranteed to produce
14538 consistent results across releases.
14542 Enable the generation of conditional moves.
14544 @item -mnops=@var{num}
14546 Emit @var{num} NOPs before every other generated instruction.
14548 @item -mno-soft-cmpsf
14549 @opindex mno-soft-cmpsf
14550 For single-precision floating-point comparisons, emit an @code{fsub} instruction
14551 and test the flags. This is faster than a software comparison, but can
14552 get incorrect results in the presence of NaNs, or when two different small
14553 numbers are compared such that their difference is calculated as zero.
14554 The default is @option{-msoft-cmpsf}, which uses slower, but IEEE-compliant,
14555 software comparisons.
14557 @item -mstack-offset=@var{num}
14558 @opindex mstack-offset
14559 Set the offset between the top of the stack and the stack pointer.
14560 E.g., a value of 8 means that the eight bytes in the range @code{sp+0@dots{}sp+7}
14561 can be used by leaf functions without stack allocation.
14562 Values other than @samp{8} or @samp{16} are untested and unlikely to work.
14563 Note also that this option changes the ABI; compiling a program with a
14564 different stack offset than the libraries have been compiled with
14565 generally does not work.
14566 This option can be useful if you want to evaluate if a different stack
14567 offset would give you better code, but to actually use a different stack
14568 offset to build working programs, it is recommended to configure the
14569 toolchain with the appropriate @option{--with-stack-offset=@var{num}} option.
14571 @item -mno-round-nearest
14572 @opindex mno-round-nearest
14573 Make the scheduler assume that the rounding mode has been set to
14574 truncating. The default is @option{-mround-nearest}.
14577 @opindex mlong-calls
14578 If not otherwise specified by an attribute, assume all calls might be beyond
14579 the offset range of the @code{b} / @code{bl} instructions, and therefore load the
14580 function address into a register before performing a (otherwise direct) call.
14581 This is the default.
14583 @item -mshort-calls
14584 @opindex short-calls
14585 If not otherwise specified by an attribute, assume all direct calls are
14586 in the range of the @code{b} / @code{bl} instructions, so use these instructions
14587 for direct calls. The default is @option{-mlong-calls}.
14591 Assume addresses can be loaded as 16-bit unsigned values. This does not
14592 apply to function addresses for which @option{-mlong-calls} semantics
14595 @item -mfp-mode=@var{mode}
14597 Set the prevailing mode of the floating-point unit.
14598 This determines the floating-point mode that is provided and expected
14599 at function call and return time. Making this mode match the mode you
14600 predominantly need at function start can make your programs smaller and
14601 faster by avoiding unnecessary mode switches.
14603 @var{mode} can be set to one the following values:
14607 Any mode at function entry is valid, and retained or restored when
14608 the function returns, and when it calls other functions.
14609 This mode is useful for compiling libraries or other compilation units
14610 you might want to incorporate into different programs with different
14611 prevailing FPU modes, and the convenience of being able to use a single
14612 object file outweighs the size and speed overhead for any extra
14613 mode switching that might be needed, compared with what would be needed
14614 with a more specific choice of prevailing FPU mode.
14617 This is the mode used for floating-point calculations with
14618 truncating (i.e.@: round towards zero) rounding mode. That includes
14619 conversion from floating point to integer.
14621 @item round-nearest
14622 This is the mode used for floating-point calculations with
14623 round-to-nearest-or-even rounding mode.
14626 This is the mode used to perform integer calculations in the FPU, e.g.@:
14627 integer multiply, or integer multiply-and-accumulate.
14630 The default is @option{-mfp-mode=caller}
14632 @item -mnosplit-lohi
14633 @itemx -mno-postinc
14634 @itemx -mno-postmodify
14635 @opindex mnosplit-lohi
14636 @opindex mno-postinc
14637 @opindex mno-postmodify
14638 Code generation tweaks that disable, respectively, splitting of 32-bit
14639 loads, generation of post-increment addresses, and generation of
14640 post-modify addresses. The defaults are @option{msplit-lohi},
14641 @option{-mpost-inc}, and @option{-mpost-modify}.
14643 @item -mnovect-double
14644 @opindex mno-vect-double
14645 Change the preferred SIMD mode to SImode. The default is
14646 @option{-mvect-double}, which uses DImode as preferred SIMD mode.
14648 @item -max-vect-align=@var{num}
14649 @opindex max-vect-align
14650 The maximum alignment for SIMD vector mode types.
14651 @var{num} may be 4 or 8. The default is 8.
14652 Note that this is an ABI change, even though many library function
14653 interfaces are unaffected if they don't use SIMD vector modes
14654 in places that affect size and/or alignment of relevant types.
14656 @item -msplit-vecmove-early
14657 @opindex msplit-vecmove-early
14658 Split vector moves into single word moves before reload. In theory this
14659 can give better register allocation, but so far the reverse seems to be
14660 generally the case.
14662 @item -m1reg-@var{reg}
14664 Specify a register to hold the constant @minus{}1, which makes loading small negative
14665 constants and certain bitmasks faster.
14666 Allowable values for @var{reg} are @samp{r43} and @samp{r63},
14667 which specify use of that register as a fixed register,
14668 and @samp{none}, which means that no register is used for this
14669 purpose. The default is @option{-m1reg-none}.
14674 @subsection ARC Options
14675 @cindex ARC options
14677 The following options control the architecture variant for which code
14680 @c architecture variants
14683 @item -mbarrel-shifter
14684 @opindex mbarrel-shifter
14685 Generate instructions supported by barrel shifter. This is the default
14686 unless @option{-mcpu=ARC601} or @samp{-mcpu=ARCEM} is in effect.
14688 @item -mcpu=@var{cpu}
14690 Set architecture type, register usage, and instruction scheduling
14691 parameters for @var{cpu}. There are also shortcut alias options
14692 available for backward compatibility and convenience. Supported
14693 values for @var{cpu} are
14699 Compile for ARC600. Aliases: @option{-mA6}, @option{-mARC600}.
14703 Compile for ARC601. Alias: @option{-mARC601}.
14708 Compile for ARC700. Aliases: @option{-mA7}, @option{-mARC700}.
14709 This is the default when configured with @option{--with-cpu=arc700}@.
14712 Compile for ARC EM.
14715 Compile for ARC HS.
14718 Compile for ARC EM CPU with no hardware extensions.
14721 Compile for ARC EM4 CPU.
14724 Compile for ARC EM4 DMIPS CPU.
14727 Compile for ARC EM4 DMIPS CPU with the single-precision floating-point
14731 Compile for ARC EM4 DMIPS CPU with single-precision floating-point and
14732 double assist instructions.
14735 Compile for ARC HS CPU with no hardware extensions except the atomic
14739 Compile for ARC HS34 CPU.
14742 Compile for ARC HS38 CPU.
14745 Compile for ARC HS38 CPU with all hardware extensions on.
14748 Compile for ARC 600 CPU with @code{norm} instructions enabled.
14750 @item arc600_mul32x16
14751 Compile for ARC 600 CPU with @code{norm} and 32x16-bit multiply
14752 instructions enabled.
14755 Compile for ARC 600 CPU with @code{norm} and @code{mul64}-family
14756 instructions enabled.
14759 Compile for ARC 601 CPU with @code{norm} instructions enabled.
14761 @item arc601_mul32x16
14762 Compile for ARC 601 CPU with @code{norm} and 32x16-bit multiply
14763 instructions enabled.
14766 Compile for ARC 601 CPU with @code{norm} and @code{mul64}-family
14767 instructions enabled.
14770 Compile for ARC 700 on NPS400 chip.
14776 @itemx -mdpfp-compact
14777 @opindex mdpfp-compact
14778 Generate double-precision FPX instructions, tuned for the compact
14782 @opindex mdpfp-fast
14783 Generate double-precision FPX instructions, tuned for the fast
14786 @item -mno-dpfp-lrsr
14787 @opindex mno-dpfp-lrsr
14788 Disable @code{lr} and @code{sr} instructions from using FPX extension
14793 Generate extended arithmetic instructions. Currently only
14794 @code{divaw}, @code{adds}, @code{subs}, and @code{sat16} are
14795 supported. This is always enabled for @option{-mcpu=ARC700}.
14799 Do not generate @code{mpy}-family instructions for ARC700. This option is
14804 Generate 32x16-bit multiply and multiply-accumulate instructions.
14808 Generate @code{mul64} and @code{mulu64} instructions.
14809 Only valid for @option{-mcpu=ARC600}.
14813 Generate @code{norm} instructions. This is the default if @option{-mcpu=ARC700}
14818 @itemx -mspfp-compact
14819 @opindex mspfp-compact
14820 Generate single-precision FPX instructions, tuned for the compact
14824 @opindex mspfp-fast
14825 Generate single-precision FPX instructions, tuned for the fast
14830 Enable generation of ARC SIMD instructions via target-specific
14831 builtins. Only valid for @option{-mcpu=ARC700}.
14834 @opindex msoft-float
14835 This option ignored; it is provided for compatibility purposes only.
14836 Software floating-point code is emitted by default, and this default
14837 can overridden by FPX options; @option{-mspfp}, @option{-mspfp-compact}, or
14838 @option{-mspfp-fast} for single precision, and @option{-mdpfp},
14839 @option{-mdpfp-compact}, or @option{-mdpfp-fast} for double precision.
14843 Generate @code{swap} instructions.
14847 This enables use of the locked load/store conditional extension to implement
14848 atomic memory built-in functions. Not available for ARC 6xx or ARC
14853 Enable @code{div} and @code{rem} instructions for ARCv2 cores.
14855 @item -mcode-density
14856 @opindex mcode-density
14857 Enable code density instructions for ARC EM.
14858 This option is on by default for ARC HS.
14862 Enable double load/store operations for ARC HS cores.
14864 @item -mtp-regno=@var{regno}
14866 Specify thread pointer register number.
14868 @item -mmpy-option=@var{multo}
14869 @opindex mmpy-option
14870 Compile ARCv2 code with a multiplier design option. You can specify
14871 the option using either a string or numeric value for @var{multo}.
14872 @samp{wlh1} is the default value. The recognized values are:
14877 No multiplier available.
14881 16x16 multiplier, fully pipelined.
14882 The following instructions are enabled: @code{mpyw} and @code{mpyuw}.
14886 32x32 multiplier, fully
14887 pipelined (1 stage). The following instructions are additionally
14888 enabled: @code{mpy}, @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
14892 32x32 multiplier, fully pipelined
14893 (2 stages). The following instructions are additionally enabled: @code{mpy},
14894 @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
14898 Two 16x16 multipliers, blocking,
14899 sequential. The following instructions are additionally enabled: @code{mpy},
14900 @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
14904 One 16x16 multiplier, blocking,
14905 sequential. The following instructions are additionally enabled: @code{mpy},
14906 @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
14910 One 32x4 multiplier, blocking,
14911 sequential. The following instructions are additionally enabled: @code{mpy},
14912 @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
14916 ARC HS SIMD support.
14920 ARC HS SIMD support.
14924 ARC HS SIMD support.
14928 This option is only available for ARCv2 cores@.
14930 @item -mfpu=@var{fpu}
14932 Enables support for specific floating-point hardware extensions for ARCv2
14933 cores. Supported values for @var{fpu} are:
14938 Enables support for single-precision floating-point hardware
14942 Enables support for double-precision floating-point hardware
14943 extensions. The single-precision floating-point extension is also
14944 enabled. Not available for ARC EM@.
14947 Enables support for double-precision floating-point hardware
14948 extensions using double-precision assist instructions. The single-precision
14949 floating-point extension is also enabled. This option is
14950 only available for ARC EM@.
14953 Enables support for double-precision floating-point hardware
14954 extensions using double-precision assist instructions.
14955 The single-precision floating-point, square-root, and divide
14956 extensions are also enabled. This option is
14957 only available for ARC EM@.
14960 Enables support for double-precision floating-point hardware
14961 extensions using double-precision assist instructions.
14962 The single-precision floating-point and fused multiply and add
14963 hardware extensions are also enabled. This option is
14964 only available for ARC EM@.
14967 Enables support for double-precision floating-point hardware
14968 extensions using double-precision assist instructions.
14969 All single-precision floating-point hardware extensions are also
14970 enabled. This option is only available for ARC EM@.
14973 Enables support for single-precision floating-point, square-root and divide
14974 hardware extensions@.
14977 Enables support for double-precision floating-point, square-root and divide
14978 hardware extensions. This option
14979 includes option @samp{fpus_div}. Not available for ARC EM@.
14982 Enables support for single-precision floating-point and
14983 fused multiply and add hardware extensions@.
14986 Enables support for double-precision floating-point and
14987 fused multiply and add hardware extensions. This option
14988 includes option @samp{fpus_fma}. Not available for ARC EM@.
14991 Enables support for all single-precision floating-point hardware
14995 Enables support for all single- and double-precision floating-point
14996 hardware extensions. Not available for ARC EM@.
15000 @item -mirq-ctrl-saved=@var{register-range}, @var{blink}, @var{lp_count}
15001 @opindex mirq-ctrl-saved
15002 Specifies general-purposes registers that the processor automatically
15003 saves/restores on interrupt entry and exit. @var{register-range} is
15004 specified as two registers separated by a dash. The register range
15005 always starts with @code{r0}, the upper limit is @code{fp} register.
15006 @var{blink} and @var{lp_count} are optional. This option is only
15007 valid for ARC EM and ARC HS cores.
15009 @item -mrgf-banked-regs=@var{number}
15010 @opindex mrgf-banked-regs
15011 Specifies the number of registers replicated in second register bank
15012 on entry to fast interrupt. Fast interrupts are interrupts with the
15013 highest priority level P0. These interrupts save only PC and STATUS32
15014 registers to avoid memory transactions during interrupt entry and exit
15015 sequences. Use this option when you are using fast interrupts in an
15016 ARC V2 family processor. Permitted values are 4, 8, 16, and 32.
15018 @item -mlpc-width=@var{width}
15019 @opindex mlpc-width
15020 Specify the width of the @code{lp_count} register. Valid values for
15021 @var{width} are 8, 16, 20, 24, 28 and 32 bits. The default width is
15022 fixed to 32 bits. If the width is less than 32, the compiler does not
15023 attempt to transform loops in your program to use the zero-delay loop
15024 mechanism unless it is known that the @code{lp_count} register can
15025 hold the required loop-counter value. Depending on the width
15026 specified, the compiler and run-time library might continue to use the
15027 loop mechanism for various needs. This option defines macro
15028 @code{__ARC_LPC_WIDTH__} with the value of @var{width}.
15032 The following options are passed through to the assembler, and also
15033 define preprocessor macro symbols.
15035 @c Flags used by the assembler, but for which we define preprocessor
15036 @c macro symbols as well.
15039 @opindex mdsp-packa
15040 Passed down to the assembler to enable the DSP Pack A extensions.
15041 Also sets the preprocessor symbol @code{__Xdsp_packa}. This option is
15046 Passed down to the assembler to enable the dual Viterbi butterfly
15047 extension. Also sets the preprocessor symbol @code{__Xdvbf}. This
15048 option is deprecated.
15050 @c ARC700 4.10 extension instruction
15053 Passed down to the assembler to enable the locked load/store
15054 conditional extension. Also sets the preprocessor symbol
15059 Passed down to the assembler. Also sets the preprocessor symbol
15060 @code{__Xxmac_d16}. This option is deprecated.
15064 Passed down to the assembler. Also sets the preprocessor symbol
15065 @code{__Xxmac_24}. This option is deprecated.
15067 @c ARC700 4.10 extension instruction
15070 Passed down to the assembler to enable the 64-bit time-stamp counter
15071 extension instruction. Also sets the preprocessor symbol
15072 @code{__Xrtsc}. This option is deprecated.
15074 @c ARC700 4.10 extension instruction
15077 Passed down to the assembler to enable the swap byte ordering
15078 extension instruction. Also sets the preprocessor symbol
15082 @opindex mtelephony
15083 Passed down to the assembler to enable dual- and single-operand
15084 instructions for telephony. Also sets the preprocessor symbol
15085 @code{__Xtelephony}. This option is deprecated.
15089 Passed down to the assembler to enable the XY memory extension. Also
15090 sets the preprocessor symbol @code{__Xxy}.
15094 The following options control how the assembly code is annotated:
15096 @c Assembly annotation options
15100 Annotate assembler instructions with estimated addresses.
15102 @item -mannotate-align
15103 @opindex mannotate-align
15104 Explain what alignment considerations lead to the decision to make an
15105 instruction short or long.
15109 The following options are passed through to the linker:
15111 @c options passed through to the linker
15115 Passed through to the linker, to specify use of the @code{arclinux} emulation.
15116 This option is enabled by default in tool chains built for
15117 @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets
15118 when profiling is not requested.
15120 @item -marclinux_prof
15121 @opindex marclinux_prof
15122 Passed through to the linker, to specify use of the
15123 @code{arclinux_prof} emulation. This option is enabled by default in
15124 tool chains built for @w{@code{arc-linux-uclibc}} and
15125 @w{@code{arceb-linux-uclibc}} targets when profiling is requested.
15129 The following options control the semantics of generated code:
15131 @c semantically relevant code generation options
15134 @opindex mlong-calls
15135 Generate calls as register indirect calls, thus providing access
15136 to the full 32-bit address range.
15138 @item -mmedium-calls
15139 @opindex mmedium-calls
15140 Don't use less than 25-bit addressing range for calls, which is the
15141 offset available for an unconditional branch-and-link
15142 instruction. Conditional execution of function calls is suppressed, to
15143 allow use of the 25-bit range, rather than the 21-bit range with
15144 conditional branch-and-link. This is the default for tool chains built
15145 for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets.
15149 Put definitions of externally-visible data in a small data section if
15150 that data is no bigger than @var{num} bytes. The default value of
15151 @var{num} is 4 for any ARC configuration, or 8 when we have double
15152 load/store operations.
15156 Do not generate sdata references. This is the default for tool chains
15157 built for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}}
15160 @item -mvolatile-cache
15161 @opindex mvolatile-cache
15162 Use ordinarily cached memory accesses for volatile references. This is the
15165 @item -mno-volatile-cache
15166 @opindex mno-volatile-cache
15167 Enable cache bypass for volatile references.
15171 The following options fine tune code generation:
15172 @c code generation tuning options
15175 @opindex malign-call
15176 Do alignment optimizations for call instructions.
15178 @item -mauto-modify-reg
15179 @opindex mauto-modify-reg
15180 Enable the use of pre/post modify with register displacement.
15182 @item -mbbit-peephole
15183 @opindex mbbit-peephole
15184 Enable bbit peephole2.
15188 This option disables a target-specific pass in @file{arc_reorg} to
15189 generate compare-and-branch (@code{br@var{cc}}) instructions.
15190 It has no effect on
15191 generation of these instructions driven by the combiner pass.
15193 @item -mcase-vector-pcrel
15194 @opindex mcase-vector-pcrel
15195 Use PC-relative switch case tables to enable case table shortening.
15196 This is the default for @option{-Os}.
15198 @item -mcompact-casesi
15199 @opindex mcompact-casesi
15200 Enable compact @code{casesi} pattern. This is the default for @option{-Os},
15201 and only available for ARCv1 cores.
15203 @item -mno-cond-exec
15204 @opindex mno-cond-exec
15205 Disable the ARCompact-specific pass to generate conditional
15206 execution instructions.
15208 Due to delay slot scheduling and interactions between operand numbers,
15209 literal sizes, instruction lengths, and the support for conditional execution,
15210 the target-independent pass to generate conditional execution is often lacking,
15211 so the ARC port has kept a special pass around that tries to find more
15212 conditional execution generation opportunities after register allocation,
15213 branch shortening, and delay slot scheduling have been done. This pass
15214 generally, but not always, improves performance and code size, at the cost of
15215 extra compilation time, which is why there is an option to switch it off.
15216 If you have a problem with call instructions exceeding their allowable
15217 offset range because they are conditionalized, you should consider using
15218 @option{-mmedium-calls} instead.
15220 @item -mearly-cbranchsi
15221 @opindex mearly-cbranchsi
15222 Enable pre-reload use of the @code{cbranchsi} pattern.
15224 @item -mexpand-adddi
15225 @opindex mexpand-adddi
15226 Expand @code{adddi3} and @code{subdi3} at RTL generation time into
15227 @code{add.f}, @code{adc} etc. This option is deprecated.
15229 @item -mindexed-loads
15230 @opindex mindexed-loads
15231 Enable the use of indexed loads. This can be problematic because some
15232 optimizers then assume that indexed stores exist, which is not
15236 Enable Local Register Allocation. This is still experimental for ARC,
15237 so by default the compiler uses standard reload
15238 (i.e. @option{-mno-lra}).
15240 @item -mlra-priority-none
15241 @opindex mlra-priority-none
15242 Don't indicate any priority for target registers.
15244 @item -mlra-priority-compact
15245 @opindex mlra-priority-compact
15246 Indicate target register priority for r0..r3 / r12..r15.
15248 @item -mlra-priority-noncompact
15249 @opindex mlra-priority-noncompact
15250 Reduce target register priority for r0..r3 / r12..r15.
15252 @item -mno-millicode
15253 @opindex mno-millicode
15254 When optimizing for size (using @option{-Os}), prologues and epilogues
15255 that have to save or restore a large number of registers are often
15256 shortened by using call to a special function in libgcc; this is
15257 referred to as a @emph{millicode} call. As these calls can pose
15258 performance issues, and/or cause linking issues when linking in a
15259 nonstandard way, this option is provided to turn off millicode call
15263 @opindex mmixed-code
15264 Tweak register allocation to help 16-bit instruction generation.
15265 This generally has the effect of decreasing the average instruction size
15266 while increasing the instruction count.
15270 Enable @samp{q} instruction alternatives.
15271 This is the default for @option{-Os}.
15275 Enable @samp{Rcq} constraint handling.
15276 Most short code generation depends on this.
15277 This is the default.
15281 Enable @samp{Rcw} constraint handling.
15282 Most ccfsm condexec mostly depends on this.
15283 This is the default.
15285 @item -msize-level=@var{level}
15286 @opindex msize-level
15287 Fine-tune size optimization with regards to instruction lengths and alignment.
15288 The recognized values for @var{level} are:
15291 No size optimization. This level is deprecated and treated like @samp{1}.
15294 Short instructions are used opportunistically.
15297 In addition, alignment of loops and of code after barriers are dropped.
15300 In addition, optional data alignment is dropped, and the option @option{Os} is enabled.
15304 This defaults to @samp{3} when @option{-Os} is in effect. Otherwise,
15305 the behavior when this is not set is equivalent to level @samp{1}.
15307 @item -mtune=@var{cpu}
15309 Set instruction scheduling parameters for @var{cpu}, overriding any implied
15310 by @option{-mcpu=}.
15312 Supported values for @var{cpu} are
15316 Tune for ARC600 CPU.
15319 Tune for ARC601 CPU.
15322 Tune for ARC700 CPU with standard multiplier block.
15325 Tune for ARC700 CPU with XMAC block.
15328 Tune for ARC725D CPU.
15331 Tune for ARC750D CPU.
15335 @item -mmultcost=@var{num}
15337 Cost to assume for a multiply instruction, with @samp{4} being equal to a
15338 normal instruction.
15340 @item -munalign-prob-threshold=@var{probability}
15341 @opindex munalign-prob-threshold
15342 Set probability threshold for unaligning branches.
15343 When tuning for @samp{ARC700} and optimizing for speed, branches without
15344 filled delay slot are preferably emitted unaligned and long, unless
15345 profiling indicates that the probability for the branch to be taken
15346 is below @var{probability}. @xref{Cross-profiling}.
15347 The default is (REG_BR_PROB_BASE/2), i.e.@: 5000.
15351 The following options are maintained for backward compatibility, but
15352 are now deprecated and will be removed in a future release:
15354 @c Deprecated options
15362 @opindex mbig-endian
15365 Compile code for big-endian targets. Use of these options is now
15366 deprecated. Big-endian code is supported by configuring GCC to build
15367 @w{@code{arceb-elf32}} and @w{@code{arceb-linux-uclibc}} targets,
15368 for which big endian is the default.
15370 @item -mlittle-endian
15371 @opindex mlittle-endian
15374 Compile code for little-endian targets. Use of these options is now
15375 deprecated. Little-endian code is supported by configuring GCC to build
15376 @w{@code{arc-elf32}} and @w{@code{arc-linux-uclibc}} targets,
15377 for which little endian is the default.
15379 @item -mbarrel_shifter
15380 @opindex mbarrel_shifter
15381 Replaced by @option{-mbarrel-shifter}.
15383 @item -mdpfp_compact
15384 @opindex mdpfp_compact
15385 Replaced by @option{-mdpfp-compact}.
15388 @opindex mdpfp_fast
15389 Replaced by @option{-mdpfp-fast}.
15392 @opindex mdsp_packa
15393 Replaced by @option{-mdsp-packa}.
15397 Replaced by @option{-mea}.
15401 Replaced by @option{-mmac-24}.
15405 Replaced by @option{-mmac-d16}.
15407 @item -mspfp_compact
15408 @opindex mspfp_compact
15409 Replaced by @option{-mspfp-compact}.
15412 @opindex mspfp_fast
15413 Replaced by @option{-mspfp-fast}.
15415 @item -mtune=@var{cpu}
15417 Values @samp{arc600}, @samp{arc601}, @samp{arc700} and
15418 @samp{arc700-xmac} for @var{cpu} are replaced by @samp{ARC600},
15419 @samp{ARC601}, @samp{ARC700} and @samp{ARC700-xmac} respectively.
15421 @item -multcost=@var{num}
15423 Replaced by @option{-mmultcost}.
15428 @subsection ARM Options
15429 @cindex ARM options
15431 These @samp{-m} options are defined for the ARM port:
15434 @item -mabi=@var{name}
15436 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
15437 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
15440 @opindex mapcs-frame
15441 Generate a stack frame that is compliant with the ARM Procedure Call
15442 Standard for all functions, even if this is not strictly necessary for
15443 correct execution of the code. Specifying @option{-fomit-frame-pointer}
15444 with this option causes the stack frames not to be generated for
15445 leaf functions. The default is @option{-mno-apcs-frame}.
15446 This option is deprecated.
15450 This is a synonym for @option{-mapcs-frame} and is deprecated.
15453 @c not currently implemented
15454 @item -mapcs-stack-check
15455 @opindex mapcs-stack-check
15456 Generate code to check the amount of stack space available upon entry to
15457 every function (that actually uses some stack space). If there is
15458 insufficient space available then either the function
15459 @code{__rt_stkovf_split_small} or @code{__rt_stkovf_split_big} is
15460 called, depending upon the amount of stack space required. The runtime
15461 system is required to provide these functions. The default is
15462 @option{-mno-apcs-stack-check}, since this produces smaller code.
15464 @c not currently implemented
15465 @item -mapcs-reentrant
15466 @opindex mapcs-reentrant
15467 Generate reentrant, position-independent code. The default is
15468 @option{-mno-apcs-reentrant}.
15471 @item -mthumb-interwork
15472 @opindex mthumb-interwork
15473 Generate code that supports calling between the ARM and Thumb
15474 instruction sets. Without this option, on pre-v5 architectures, the
15475 two instruction sets cannot be reliably used inside one program. The
15476 default is @option{-mno-thumb-interwork}, since slightly larger code
15477 is generated when @option{-mthumb-interwork} is specified. In AAPCS
15478 configurations this option is meaningless.
15480 @item -mno-sched-prolog
15481 @opindex mno-sched-prolog
15482 Prevent the reordering of instructions in the function prologue, or the
15483 merging of those instruction with the instructions in the function's
15484 body. This means that all functions start with a recognizable set
15485 of instructions (or in fact one of a choice from a small set of
15486 different function prologues), and this information can be used to
15487 locate the start of functions inside an executable piece of code. The
15488 default is @option{-msched-prolog}.
15490 @item -mfloat-abi=@var{name}
15491 @opindex mfloat-abi
15492 Specifies which floating-point ABI to use. Permissible values
15493 are: @samp{soft}, @samp{softfp} and @samp{hard}.
15495 Specifying @samp{soft} causes GCC to generate output containing
15496 library calls for floating-point operations.
15497 @samp{softfp} allows the generation of code using hardware floating-point
15498 instructions, but still uses the soft-float calling conventions.
15499 @samp{hard} allows generation of floating-point instructions
15500 and uses FPU-specific calling conventions.
15502 The default depends on the specific target configuration. Note that
15503 the hard-float and soft-float ABIs are not link-compatible; you must
15504 compile your entire program with the same ABI, and link with a
15505 compatible set of libraries.
15507 @item -mlittle-endian
15508 @opindex mlittle-endian
15509 Generate code for a processor running in little-endian mode. This is
15510 the default for all standard configurations.
15513 @opindex mbig-endian
15514 Generate code for a processor running in big-endian mode; the default is
15515 to compile code for a little-endian processor.
15520 When linking a big-endian image select between BE8 and BE32 formats.
15521 The option has no effect for little-endian images and is ignored. The
15522 default is dependent on the selected target architecture. For ARMv6
15523 and later architectures the default is BE8, for older architectures
15524 the default is BE32. BE32 format has been deprecated by ARM.
15526 @item -march=@var{name}@r{[}+extension@dots{}@r{]}
15528 This specifies the name of the target ARM architecture. GCC uses this
15529 name to determine what kind of instructions it can emit when generating
15530 assembly code. This option can be used in conjunction with or instead
15531 of the @option{-mcpu=} option.
15533 Permissible names are:
15535 @samp{armv5t}, @samp{armv5te},
15536 @samp{armv6}, @samp{armv6j}, @samp{armv6k}, @samp{armv6kz}, @samp{armv6t2},
15537 @samp{armv6z}, @samp{armv6zk},
15538 @samp{armv7}, @samp{armv7-a}, @samp{armv7ve},
15539 @samp{armv8-a}, @samp{armv8.1-a}, @samp{armv8.2-a}, @samp{armv8.3-a},
15542 @samp{armv6-m}, @samp{armv6s-m},
15543 @samp{armv7-m}, @samp{armv7e-m},
15544 @samp{armv8-m.base}, @samp{armv8-m.main},
15545 @samp{iwmmxt} and @samp{iwmmxt2}.
15547 Additionally, the following architectures, which lack support for the
15548 Thumb exection state, are recognized but support is deprecated:
15549 @samp{armv2}, @samp{armv2a}, @samp{armv3}, @samp{armv3m},
15550 @samp{armv4}, @samp{armv5} and @samp{armv5e}.
15552 Many of the architectures support extensions. These can be added by
15553 appending @samp{+@var{extension}} to the architecture name. Extension
15554 options are processed in order and capabilities accumulate. An extension
15555 will also enable any necessary base extensions
15556 upon which it depends. For example, the @samp{+crypto} extension
15557 will always enable the @samp{+simd} extension. The exception to the
15558 additive construction is for extensions that are prefixed with
15559 @samp{+no@dots{}}: these extensions disable the specified option and
15560 any other extensions that may depend on the presence of that
15563 For example, @samp{-march=armv7-a+simd+nofp+vfpv4} is equivalent to
15564 writing @samp{-march=armv7-a+vfpv4} since the @samp{+simd} option is
15565 entirely disabled by the @samp{+nofp} option that follows it.
15567 Most extension names are generically named, but have an effect that is
15568 dependent upon the architecture to which it is applied. For example,
15569 the @samp{+simd} option can be applied to both @samp{armv7-a} and
15570 @samp{armv8-a} architectures, but will enable the original ARMv7
15571 Advanced SIMD (Neon) extensions for @samp{armv7-a} and the ARMv8-a
15572 variant for @samp{armv8-a}.
15574 The table below lists the supported extensions for each architecture.
15575 Architectures not mentioned do not support any extensions.
15589 The VFPv2 floating-point instructions. The extension @samp{+vfpv2} can be
15590 used as an alias for this extension.
15593 Disable the floating-point instructions.
15597 The common subset of the ARMv7-A, ARMv7-R and ARMv7-M architectures.
15600 The VFPv3 floating-point instructions, with 16 double-precision
15601 registers. The extension @samp{+vfpv3-d16} can be used as an alias
15602 for this extension. Note that floating-point is not supported by the
15603 base ARMv7-M architecture, but is compatible with both the ARMv7-A and
15604 ARMv7-R architectures.
15607 Disable the floating-point instructions.
15613 The VFPv3 floating-point instructions, with 16 double-precision
15614 registers. The extension @samp{+vfpv3-d16} can be used as an alias
15615 for this extension.
15618 The Advanced SIMD (Neon) v1 and the VFPv3 floating-point instructions.
15619 The extensions @samp{+neon} and @samp{+neon-vfpv3} can be used as aliases
15620 for this extension.
15623 The VFPv3 floating-point instructions, with 32 double-precision
15626 @item +vfpv3-d16-fp16
15627 The VFPv3 floating-point instructions, with 16 double-precision
15628 registers and the half-precision floating-point conversion operations.
15631 The VFPv3 floating-point instructions, with 32 double-precision
15632 registers and the half-precision floating-point conversion operations.
15635 The VFPv4 floating-point instructions, with 16 double-precision
15639 The VFPv4 floating-point instructions, with 32 double-precision
15643 The Advanced SIMD (Neon) v1 and the VFPv3 floating-point instructions, with
15644 the half-precision floating-point conversion operations.
15647 The Advanced SIMD (Neon) v2 and the VFPv4 floating-point instructions.
15650 Disable the Advanced SIMD instructions (does not disable floating point).
15653 Disable the floating-point and Advanced SIMD instructions.
15657 The extended version of the ARMv7-A architecture with support for
15661 The VFPv4 floating-point instructions, with 16 double-precision registers.
15662 The extension @samp{+vfpv4-d16} can be used as an alias for this extension.
15665 The Advanced SIMD (Neon) v2 and the VFPv4 floating-point instructions. The
15666 extension @samp{+neon-vfpv4} can be used as an alias for this extension.
15669 The VFPv3 floating-point instructions, with 16 double-precision
15673 The VFPv3 floating-point instructions, with 32 double-precision
15676 @item +vfpv3-d16-fp16
15677 The VFPv3 floating-point instructions, with 16 double-precision
15678 registers and the half-precision floating-point conversion operations.
15681 The VFPv3 floating-point instructions, with 32 double-precision
15682 registers and the half-precision floating-point conversion operations.
15685 The VFPv4 floating-point instructions, with 16 double-precision
15689 The VFPv4 floating-point instructions, with 32 double-precision
15693 The Advanced SIMD (Neon) v1 and the VFPv3 floating-point instructions.
15694 The extension @samp{+neon-vfpv3} can be used as an alias for this extension.
15697 The Advanced SIMD (Neon) v1 and the VFPv3 floating-point instructions, with
15698 the half-precision floating-point conversion operations.
15701 Disable the Advanced SIMD instructions (does not disable floating point).
15704 Disable the floating-point and Advanced SIMD instructions.
15710 The Cyclic Redundancy Check (CRC) instructions.
15712 The ARMv8 Advanced SIMD and floating-point instructions.
15714 The cryptographic instructions.
15716 Disable the cryptographic isntructions.
15718 Disable the floating-point, Advanced SIMD and cryptographic instructions.
15724 The ARMv8.1 Advanced SIMD and floating-point instructions.
15727 The cryptographic instructions. This also enables the Advanced SIMD and
15728 floating-point instructions.
15731 Disable the cryptographic isntructions.
15734 Disable the floating-point, Advanced SIMD and cryptographic instructions.
15740 The half-precision floating-point data processing instructions.
15741 This also enables the Advanced SIMD and floating-point instructions.
15744 The ARMv8.1 Advanced SIMD and floating-point instructions.
15747 The cryptographic instructions. This also enables the Advanced SIMD and
15748 floating-point instructions.
15751 Enable the Dot Product extension. This also enables Advanced SIMD instructions.
15754 Disable the cryptographic extension.
15757 Disable the floating-point, Advanced SIMD and cryptographic instructions.
15763 The single-precision VFPv3 floating-point instructions. The extension
15764 @samp{+vfpv3xd} can be used as an alias for this extension.
15767 The VFPv3 floating-point instructions with 16 double-precision registers.
15768 The extension +vfpv3-d16 can be used as an alias for this extension.
15771 Disable the floating-point extension.
15774 The ARM-state integer division instructions.
15777 Disable the ARM-state integer division extension.
15783 The single-precision VFPv4 floating-point instructions.
15786 The single-precision FPv5 floating-point instructions.
15789 The single- and double-precision FPv5 floating-point instructions.
15792 Disable the floating-point extensions.
15798 The DSP instructions.
15801 Disable the DSP extension.
15804 The single-precision floating-point instructions.
15807 The single- and double-precision floating-point instructions.
15810 Disable the floating-point extension.
15816 The Cyclic Redundancy Check (CRC) instructions.
15818 The single-precision FPv5 floating-point instructions.
15820 The ARMv8 Advanced SIMD and floating-point instructions.
15822 The cryptographic instructions.
15824 Disable the cryptographic isntructions.
15826 Disable the floating-point, Advanced SIMD and cryptographic instructions.
15831 @option{-march=native} causes the compiler to auto-detect the architecture
15832 of the build computer. At present, this feature is only supported on
15833 GNU/Linux, and not all architectures are recognized. If the auto-detect
15834 is unsuccessful the option has no effect.
15836 @item -mtune=@var{name}
15838 This option specifies the name of the target ARM processor for
15839 which GCC should tune the performance of the code.
15840 For some ARM implementations better performance can be obtained by using
15842 Permissible names are: @samp{arm2}, @samp{arm250},
15843 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
15844 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
15845 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
15846 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
15848 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
15849 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
15850 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
15851 @samp{strongarm1110},
15852 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
15853 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
15854 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
15855 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
15856 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
15857 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
15858 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
15859 @samp{generic-armv7-a}, @samp{cortex-a5}, @samp{cortex-a7}, @samp{cortex-a8},
15860 @samp{cortex-a9}, @samp{cortex-a12}, @samp{cortex-a15}, @samp{cortex-a17},
15861 @samp{cortex-a32}, @samp{cortex-a35}, @samp{cortex-a53}, @samp{cortex-a55},
15862 @samp{cortex-a57}, @samp{cortex-a72}, @samp{cortex-a73}, @samp{cortex-a75},
15863 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-r5}, @samp{cortex-r7},
15864 @samp{cortex-r8}, @samp{cortex-r52},
15872 @samp{cortex-m0plus},
15873 @samp{cortex-m1.small-multiply},
15874 @samp{cortex-m0.small-multiply},
15875 @samp{cortex-m0plus.small-multiply},
15877 @samp{marvell-pj4},
15878 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312},
15879 @samp{fa526}, @samp{fa626},
15880 @samp{fa606te}, @samp{fa626te}, @samp{fmp626}, @samp{fa726te},
15883 Additionally, this option can specify that GCC should tune the performance
15884 of the code for a big.LITTLE system. Permissible names are:
15885 @samp{cortex-a15.cortex-a7}, @samp{cortex-a17.cortex-a7},
15886 @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53},
15887 @samp{cortex-a72.cortex-a35}, @samp{cortex-a73.cortex-a53},
15888 @samp{cortex-a75.cortex-a55}.
15890 @option{-mtune=generic-@var{arch}} specifies that GCC should tune the
15891 performance for a blend of processors within architecture @var{arch}.
15892 The aim is to generate code that run well on the current most popular
15893 processors, balancing between optimizations that benefit some CPUs in the
15894 range, and avoiding performance pitfalls of other CPUs. The effects of
15895 this option may change in future GCC versions as CPU models come and go.
15897 @option{-mtune} permits the same extension options as @option{-mcpu}, but
15898 the extension options do not affect the tuning of the generated code.
15900 @option{-mtune=native} causes the compiler to auto-detect the CPU
15901 of the build computer. At present, this feature is only supported on
15902 GNU/Linux, and not all architectures are recognized. If the auto-detect is
15903 unsuccessful the option has no effect.
15905 @item -mcpu=@var{name}@r{[}+extension@dots{}@r{]}
15907 This specifies the name of the target ARM processor. GCC uses this name
15908 to derive the name of the target ARM architecture (as if specified
15909 by @option{-march}) and the ARM processor type for which to tune for
15910 performance (as if specified by @option{-mtune}). Where this option
15911 is used in conjunction with @option{-march} or @option{-mtune},
15912 those options take precedence over the appropriate part of this option.
15914 Many of the supported CPUs implement optional architectural
15915 extensions. Where this is so the architectural extensions are
15916 normally enabled by default. If implementations that lack the
15917 extension exist, then the extension syntax can be used to disable
15918 those extensions that have been omitted. For floating-point and
15919 Advanced SIMD (Neon) instructions, the settings of the options
15920 @option{-mfloat-abi} and @option{-mfpu} must also be considered:
15921 floating-point and Advanced SIMD instructions will only be used if
15922 @option{-mfloat-abi} is not set to @samp{soft}; and any setting of
15923 @option{-mfpu} other than @samp{auto} will override the available
15924 floating-point and SIMD extension instructions.
15926 For example, @samp{cortex-a9} can be found in three major
15927 configurations: integer only, with just a floating-point unit or with
15928 floating-point and Advanced SIMD. The default is to enable all the
15929 instructions, but the extensions @samp{+nosimd} and @samp{+nofp} can
15930 be used to disable just the SIMD or both the SIMD and floating-point
15931 instructions respectively.
15933 Permissible names for this option are the same as those for
15936 The following extension options are common to the listed CPUs:
15940 Disable the DSP instructions on @samp{cortex-m33}.
15943 Disables the floating-point instructions on @samp{arm9e},
15944 @samp{arm946e-s}, @samp{arm966e-s}, @samp{arm968e-s}, @samp{arm10e},
15945 @samp{arm1020e}, @samp{arm1022e}, @samp{arm926ej-s},
15946 @samp{arm1026ej-s}, @samp{cortex-r5}, @samp{cortex-r7}, @samp{cortex-r8},
15947 @samp{cortex-m4}, @samp{cortex-m7} and @samp{cortex-m33}.
15948 Disables the floating-point and SIMD instructions on
15949 @samp{generic-armv7-a}, @samp{cortex-a5}, @samp{cortex-a7},
15950 @samp{cortex-a8}, @samp{cortex-a9}, @samp{cortex-a12},
15951 @samp{cortex-a15}, @samp{cortex-a17}, @samp{cortex-a15.cortex-a7},
15952 @samp{cortex-a17.cortex-a7}, @samp{cortex-a32}, @samp{cortex-a35},
15953 @samp{cortex-a53} and @samp{cortex-a55}.
15956 Disables the double-precision component of the floating-point instructions
15957 on @samp{cortex-r5}, @samp{cortex-r52} and @samp{cortex-m7}.
15960 Disables the SIMD (but not floating-point) instructions on
15961 @samp{generic-armv7-a}, @samp{cortex-a5}, @samp{cortex-a7}
15962 and @samp{cortex-a9}.
15965 Enables the cryptographic instructions on @samp{cortex-a32},
15966 @samp{cortex-a35}, @samp{cortex-a53}, @samp{cortex-a55}, @samp{cortex-a57},
15967 @samp{cortex-a72}, @samp{cortex-a73}, @samp{cortex-a75}, @samp{exynos-m1},
15968 @samp{xgene1}, @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53},
15969 @samp{cortex-a73.cortex-a35}, @samp{cortex-a73.cortex-a53} and
15970 @samp{cortex-a75.cortex-a55}.
15973 Additionally the @samp{generic-armv7-a} pseudo target defaults to
15974 VFPv3 with 16 double-precision registers. It supports the following
15975 extension options: @samp{vfpv3-d16}, @samp{vfpv3},
15976 @samp{vfpv3-d16-fp16}, @samp{vfpv3-fp16}, @samp{vfpv4-d16},
15977 @samp{vfpv4}, @samp{neon}, @samp{neon-vfpv3}, @samp{neon-fp16},
15978 @samp{neon-vfpv4}. The meanings are the same as for the extensions to
15979 @option{-march=armv7-a}.
15981 @option{-mcpu=generic-@var{arch}} is also permissible, and is
15982 equivalent to @option{-march=@var{arch} -mtune=generic-@var{arch}}.
15983 See @option{-mtune} for more information.
15985 @option{-mcpu=native} causes the compiler to auto-detect the CPU
15986 of the build computer. At present, this feature is only supported on
15987 GNU/Linux, and not all architectures are recognized. If the auto-detect
15988 is unsuccessful the option has no effect.
15990 @item -mfpu=@var{name}
15992 This specifies what floating-point hardware (or hardware emulation) is
15993 available on the target. Permissible names are: @samp{auto}, @samp{vfpv2},
15995 @samp{vfpv3-fp16}, @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd},
15996 @samp{vfpv3xd-fp16}, @samp{neon-vfpv3}, @samp{neon-fp16}, @samp{vfpv4},
15997 @samp{vfpv4-d16}, @samp{fpv4-sp-d16}, @samp{neon-vfpv4},
15998 @samp{fpv5-d16}, @samp{fpv5-sp-d16},
15999 @samp{fp-armv8}, @samp{neon-fp-armv8} and @samp{crypto-neon-fp-armv8}.
16000 Note that @samp{neon} is an alias for @samp{neon-vfpv3} and @samp{vfp}
16001 is an alias for @samp{vfpv2}.
16003 The setting @samp{auto} is the default and is special. It causes the
16004 compiler to select the floating-point and Advanced SIMD instructions
16005 based on the settings of @option{-mcpu} and @option{-march}.
16007 If the selected floating-point hardware includes the NEON extension
16008 (e.g. @option{-mfpu=neon}), note that floating-point
16009 operations are not generated by GCC's auto-vectorization pass unless
16010 @option{-funsafe-math-optimizations} is also specified. This is
16011 because NEON hardware does not fully implement the IEEE 754 standard for
16012 floating-point arithmetic (in particular denormal values are treated as
16013 zero), so the use of NEON instructions may lead to a loss of precision.
16015 You can also set the fpu name at function level by using the @code{target("fpu=")} function attributes (@pxref{ARM Function Attributes}) or pragmas (@pxref{Function Specific Option Pragmas}).
16017 @item -mfp16-format=@var{name}
16018 @opindex mfp16-format
16019 Specify the format of the @code{__fp16} half-precision floating-point type.
16020 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
16021 the default is @samp{none}, in which case the @code{__fp16} type is not
16022 defined. @xref{Half-Precision}, for more information.
16024 @item -mstructure-size-boundary=@var{n}
16025 @opindex mstructure-size-boundary
16026 The sizes of all structures and unions are rounded up to a multiple
16027 of the number of bits set by this option. Permissible values are 8, 32
16028 and 64. The default value varies for different toolchains. For the COFF
16029 targeted toolchain the default value is 8. A value of 64 is only allowed
16030 if the underlying ABI supports it.
16032 Specifying a larger number can produce faster, more efficient code, but
16033 can also increase the size of the program. Different values are potentially
16034 incompatible. Code compiled with one value cannot necessarily expect to
16035 work with code or libraries compiled with another value, if they exchange
16036 information using structures or unions.
16038 This option is deprecated.
16040 @item -mabort-on-noreturn
16041 @opindex mabort-on-noreturn
16042 Generate a call to the function @code{abort} at the end of a
16043 @code{noreturn} function. It is executed if the function tries to
16047 @itemx -mno-long-calls
16048 @opindex mlong-calls
16049 @opindex mno-long-calls
16050 Tells the compiler to perform function calls by first loading the
16051 address of the function into a register and then performing a subroutine
16052 call on this register. This switch is needed if the target function
16053 lies outside of the 64-megabyte addressing range of the offset-based
16054 version of subroutine call instruction.
16056 Even if this switch is enabled, not all function calls are turned
16057 into long calls. The heuristic is that static functions, functions
16058 that have the @code{short_call} attribute, functions that are inside
16059 the scope of a @code{#pragma no_long_calls} directive, and functions whose
16060 definitions have already been compiled within the current compilation
16061 unit are not turned into long calls. The exceptions to this rule are
16062 that weak function definitions, functions with the @code{long_call}
16063 attribute or the @code{section} attribute, and functions that are within
16064 the scope of a @code{#pragma long_calls} directive are always
16065 turned into long calls.
16067 This feature is not enabled by default. Specifying
16068 @option{-mno-long-calls} restores the default behavior, as does
16069 placing the function calls within the scope of a @code{#pragma
16070 long_calls_off} directive. Note these switches have no effect on how
16071 the compiler generates code to handle function calls via function
16074 @item -msingle-pic-base
16075 @opindex msingle-pic-base
16076 Treat the register used for PIC addressing as read-only, rather than
16077 loading it in the prologue for each function. The runtime system is
16078 responsible for initializing this register with an appropriate value
16079 before execution begins.
16081 @item -mpic-register=@var{reg}
16082 @opindex mpic-register
16083 Specify the register to be used for PIC addressing.
16084 For standard PIC base case, the default is any suitable register
16085 determined by compiler. For single PIC base case, the default is
16086 @samp{R9} if target is EABI based or stack-checking is enabled,
16087 otherwise the default is @samp{R10}.
16089 @item -mpic-data-is-text-relative
16090 @opindex mpic-data-is-text-relative
16091 Assume that the displacement between the text and data segments is fixed
16092 at static link time. This permits using PC-relative addressing
16093 operations to access data known to be in the data segment. For
16094 non-VxWorks RTP targets, this option is enabled by default. When
16095 disabled on such targets, it will enable @option{-msingle-pic-base} by
16098 @item -mpoke-function-name
16099 @opindex mpoke-function-name
16100 Write the name of each function into the text section, directly
16101 preceding the function prologue. The generated code is similar to this:
16105 .ascii "arm_poke_function_name", 0
16108 .word 0xff000000 + (t1 - t0)
16109 arm_poke_function_name
16111 stmfd sp!, @{fp, ip, lr, pc@}
16115 When performing a stack backtrace, code can inspect the value of
16116 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
16117 location @code{pc - 12} and the top 8 bits are set, then we know that
16118 there is a function name embedded immediately preceding this location
16119 and has length @code{((pc[-3]) & 0xff000000)}.
16126 Select between generating code that executes in ARM and Thumb
16127 states. The default for most configurations is to generate code
16128 that executes in ARM state, but the default can be changed by
16129 configuring GCC with the @option{--with-mode=}@var{state}
16132 You can also override the ARM and Thumb mode for each function
16133 by using the @code{target("thumb")} and @code{target("arm")} function attributes
16134 (@pxref{ARM Function Attributes}) or pragmas (@pxref{Function Specific Option Pragmas}).
16137 @opindex mtpcs-frame
16138 Generate a stack frame that is compliant with the Thumb Procedure Call
16139 Standard for all non-leaf functions. (A leaf function is one that does
16140 not call any other functions.) The default is @option{-mno-tpcs-frame}.
16142 @item -mtpcs-leaf-frame
16143 @opindex mtpcs-leaf-frame
16144 Generate a stack frame that is compliant with the Thumb Procedure Call
16145 Standard for all leaf functions. (A leaf function is one that does
16146 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
16148 @item -mcallee-super-interworking
16149 @opindex mcallee-super-interworking
16150 Gives all externally visible functions in the file being compiled an ARM
16151 instruction set header which switches to Thumb mode before executing the
16152 rest of the function. This allows these functions to be called from
16153 non-interworking code. This option is not valid in AAPCS configurations
16154 because interworking is enabled by default.
16156 @item -mcaller-super-interworking
16157 @opindex mcaller-super-interworking
16158 Allows calls via function pointers (including virtual functions) to
16159 execute correctly regardless of whether the target code has been
16160 compiled for interworking or not. There is a small overhead in the cost
16161 of executing a function pointer if this option is enabled. This option
16162 is not valid in AAPCS configurations because interworking is enabled
16165 @item -mtp=@var{name}
16167 Specify the access model for the thread local storage pointer. The valid
16168 models are @samp{soft}, which generates calls to @code{__aeabi_read_tp},
16169 @samp{cp15}, which fetches the thread pointer from @code{cp15} directly
16170 (supported in the arm6k architecture), and @samp{auto}, which uses the
16171 best available method for the selected processor. The default setting is
16174 @item -mtls-dialect=@var{dialect}
16175 @opindex mtls-dialect
16176 Specify the dialect to use for accessing thread local storage. Two
16177 @var{dialect}s are supported---@samp{gnu} and @samp{gnu2}. The
16178 @samp{gnu} dialect selects the original GNU scheme for supporting
16179 local and global dynamic TLS models. The @samp{gnu2} dialect
16180 selects the GNU descriptor scheme, which provides better performance
16181 for shared libraries. The GNU descriptor scheme is compatible with
16182 the original scheme, but does require new assembler, linker and
16183 library support. Initial and local exec TLS models are unaffected by
16184 this option and always use the original scheme.
16186 @item -mword-relocations
16187 @opindex mword-relocations
16188 Only generate absolute relocations on word-sized values (i.e. R_ARM_ABS32).
16189 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
16190 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
16193 @item -mfix-cortex-m3-ldrd
16194 @opindex mfix-cortex-m3-ldrd
16195 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
16196 with overlapping destination and base registers are used. This option avoids
16197 generating these instructions. This option is enabled by default when
16198 @option{-mcpu=cortex-m3} is specified.
16200 @item -munaligned-access
16201 @itemx -mno-unaligned-access
16202 @opindex munaligned-access
16203 @opindex mno-unaligned-access
16204 Enables (or disables) reading and writing of 16- and 32- bit values
16205 from addresses that are not 16- or 32- bit aligned. By default
16206 unaligned access is disabled for all pre-ARMv6, all ARMv6-M and for
16207 ARMv8-M Baseline architectures, and enabled for all other
16208 architectures. If unaligned access is not enabled then words in packed
16209 data structures are accessed a byte at a time.
16211 The ARM attribute @code{Tag_CPU_unaligned_access} is set in the
16212 generated object file to either true or false, depending upon the
16213 setting of this option. If unaligned access is enabled then the
16214 preprocessor symbol @code{__ARM_FEATURE_UNALIGNED} is also
16217 @item -mneon-for-64bits
16218 @opindex mneon-for-64bits
16219 Enables using Neon to handle scalar 64-bits operations. This is
16220 disabled by default since the cost of moving data from core registers
16223 @item -mslow-flash-data
16224 @opindex mslow-flash-data
16225 Assume loading data from flash is slower than fetching instruction.
16226 Therefore literal load is minimized for better performance.
16227 This option is only supported when compiling for ARMv7 M-profile and
16230 @item -masm-syntax-unified
16231 @opindex masm-syntax-unified
16232 Assume inline assembler is using unified asm syntax. The default is
16233 currently off which implies divided syntax. This option has no impact
16234 on Thumb2. However, this may change in future releases of GCC.
16235 Divided syntax should be considered deprecated.
16237 @item -mrestrict-it
16238 @opindex mrestrict-it
16239 Restricts generation of IT blocks to conform to the rules of ARMv8.
16240 IT blocks can only contain a single 16-bit instruction from a select
16241 set of instructions. This option is on by default for ARMv8 Thumb mode.
16243 @item -mprint-tune-info
16244 @opindex mprint-tune-info
16245 Print CPU tuning information as comment in assembler file. This is
16246 an option used only for regression testing of the compiler and not
16247 intended for ordinary use in compiling code. This option is disabled
16251 @opindex mpure-code
16252 Do not allow constant data to be placed in code sections.
16253 Additionally, when compiling for ELF object format give all text sections the
16254 ELF processor-specific section attribute @code{SHF_ARM_PURECODE}. This option
16255 is only available when generating non-pic code for M-profile targets with the
16260 Generate secure code as per the "ARMv8-M Security Extensions: Requirements on
16261 Development Tools Engineering Specification", which can be found on
16262 @url{http://infocenter.arm.com/help/topic/com.arm.doc.ecm0359818/ECM0359818_armv8m_security_extensions_reqs_on_dev_tools_1_0.pdf}.
16266 @subsection AVR Options
16267 @cindex AVR Options
16269 These options are defined for AVR implementations:
16272 @item -mmcu=@var{mcu}
16274 Specify Atmel AVR instruction set architectures (ISA) or MCU type.
16276 The default for this option is@tie{}@samp{avr2}.
16278 GCC supports the following AVR devices and ISAs:
16280 @include avr-mmcu.texi
16285 Assume that all data in static storage can be accessed by LDS / STS
16286 instructions. This option has only an effect on reduced Tiny devices like
16287 ATtiny40. See also the @code{absdata}
16288 @ref{AVR Variable Attributes,variable attribute}.
16290 @item -maccumulate-args
16291 @opindex maccumulate-args
16292 Accumulate outgoing function arguments and acquire/release the needed
16293 stack space for outgoing function arguments once in function
16294 prologue/epilogue. Without this option, outgoing arguments are pushed
16295 before calling a function and popped afterwards.
16297 Popping the arguments after the function call can be expensive on
16298 AVR so that accumulating the stack space might lead to smaller
16299 executables because arguments need not be removed from the
16300 stack after such a function call.
16302 This option can lead to reduced code size for functions that perform
16303 several calls to functions that get their arguments on the stack like
16304 calls to printf-like functions.
16306 @item -mbranch-cost=@var{cost}
16307 @opindex mbranch-cost
16308 Set the branch costs for conditional branch instructions to
16309 @var{cost}. Reasonable values for @var{cost} are small, non-negative
16310 integers. The default branch cost is 0.
16312 @item -mcall-prologues
16313 @opindex mcall-prologues
16314 Functions prologues/epilogues are expanded as calls to appropriate
16315 subroutines. Code size is smaller.
16317 @item -mgas-isr-prologues
16318 @opindex mgas-isr-prologues
16319 Interrupt service routines (ISRs) may use the @code{__gcc_isr} pseudo
16320 instruction supported by GNU Binutils.
16321 If this option is on, the feature can still be disabled for individual
16322 ISRs by means of the @ref{AVR Function Attributes,,@code{no_gccisr}}
16323 function attribute. This feature is activated per default
16324 if optimization is on (but not with @option{-Og}, @pxref{Optimize Options}),
16325 and if GNU Binutils support @w{@uref{https://sourceware.org/PR21683,PR21683}}.
16329 Assume @code{int} to be 8-bit integer. This affects the sizes of all types: a
16330 @code{char} is 1 byte, an @code{int} is 1 byte, a @code{long} is 2 bytes,
16331 and @code{long long} is 4 bytes. Please note that this option does not
16332 conform to the C standards, but it results in smaller code
16335 @item -mn-flash=@var{num}
16337 Assume that the flash memory has a size of
16338 @var{num} times 64@tie{}KiB.
16340 @item -mno-interrupts
16341 @opindex mno-interrupts
16342 Generated code is not compatible with hardware interrupts.
16343 Code size is smaller.
16347 Try to replace @code{CALL} resp.@: @code{JMP} instruction by the shorter
16348 @code{RCALL} resp.@: @code{RJMP} instruction if applicable.
16349 Setting @option{-mrelax} just adds the @option{--mlink-relax} option to
16350 the assembler's command line and the @option{--relax} option to the
16351 linker's command line.
16353 Jump relaxing is performed by the linker because jump offsets are not
16354 known before code is located. Therefore, the assembler code generated by the
16355 compiler is the same, but the instructions in the executable may
16356 differ from instructions in the assembler code.
16358 Relaxing must be turned on if linker stubs are needed, see the
16359 section on @code{EIND} and linker stubs below.
16363 Assume that the device supports the Read-Modify-Write
16364 instructions @code{XCH}, @code{LAC}, @code{LAS} and @code{LAT}.
16366 @item -mshort-calls
16367 @opindex mshort-calls
16369 Assume that @code{RJMP} and @code{RCALL} can target the whole
16372 This option is used internally for multilib selection. It is
16373 not an optimization option, and you don't need to set it by hand.
16377 Treat the stack pointer register as an 8-bit register,
16378 i.e.@: assume the high byte of the stack pointer is zero.
16379 In general, you don't need to set this option by hand.
16381 This option is used internally by the compiler to select and
16382 build multilibs for architectures @code{avr2} and @code{avr25}.
16383 These architectures mix devices with and without @code{SPH}.
16384 For any setting other than @option{-mmcu=avr2} or @option{-mmcu=avr25}
16385 the compiler driver adds or removes this option from the compiler
16386 proper's command line, because the compiler then knows if the device
16387 or architecture has an 8-bit stack pointer and thus no @code{SPH}
16392 Use address register @code{X} in a way proposed by the hardware. This means
16393 that @code{X} is only used in indirect, post-increment or
16394 pre-decrement addressing.
16396 Without this option, the @code{X} register may be used in the same way
16397 as @code{Y} or @code{Z} which then is emulated by additional
16399 For example, loading a value with @code{X+const} addressing with a
16400 small non-negative @code{const < 64} to a register @var{Rn} is
16404 adiw r26, const ; X += const
16405 ld @var{Rn}, X ; @var{Rn} = *X
16406 sbiw r26, const ; X -= const
16410 @opindex mtiny-stack
16411 Only change the lower 8@tie{}bits of the stack pointer.
16413 @item -mfract-convert-truncate
16414 @opindex mfract-convert-truncate
16415 Allow to use truncation instead of rounding towards zero for fractional fixed-point types.
16418 @opindex nodevicelib
16419 Don't link against AVR-LibC's device specific library @code{lib<mcu>.a}.
16421 @item -Waddr-space-convert
16422 @opindex Waddr-space-convert
16423 Warn about conversions between address spaces in the case where the
16424 resulting address space is not contained in the incoming address space.
16426 @item -Wmisspelled-isr
16427 @opindex Wmisspelled-isr
16428 Warn if the ISR is misspelled, i.e. without __vector prefix.
16429 Enabled by default.
16432 @subsubsection @code{EIND} and Devices with More Than 128 Ki Bytes of Flash
16433 @cindex @code{EIND}
16434 Pointers in the implementation are 16@tie{}bits wide.
16435 The address of a function or label is represented as word address so
16436 that indirect jumps and calls can target any code address in the
16437 range of 64@tie{}Ki words.
16439 In order to facilitate indirect jump on devices with more than 128@tie{}Ki
16440 bytes of program memory space, there is a special function register called
16441 @code{EIND} that serves as most significant part of the target address
16442 when @code{EICALL} or @code{EIJMP} instructions are used.
16444 Indirect jumps and calls on these devices are handled as follows by
16445 the compiler and are subject to some limitations:
16450 The compiler never sets @code{EIND}.
16453 The compiler uses @code{EIND} implicitly in @code{EICALL}/@code{EIJMP}
16454 instructions or might read @code{EIND} directly in order to emulate an
16455 indirect call/jump by means of a @code{RET} instruction.
16458 The compiler assumes that @code{EIND} never changes during the startup
16459 code or during the application. In particular, @code{EIND} is not
16460 saved/restored in function or interrupt service routine
16464 For indirect calls to functions and computed goto, the linker
16465 generates @emph{stubs}. Stubs are jump pads sometimes also called
16466 @emph{trampolines}. Thus, the indirect call/jump jumps to such a stub.
16467 The stub contains a direct jump to the desired address.
16470 Linker relaxation must be turned on so that the linker generates
16471 the stubs correctly in all situations. See the compiler option
16472 @option{-mrelax} and the linker option @option{--relax}.
16473 There are corner cases where the linker is supposed to generate stubs
16474 but aborts without relaxation and without a helpful error message.
16477 The default linker script is arranged for code with @code{EIND = 0}.
16478 If code is supposed to work for a setup with @code{EIND != 0}, a custom
16479 linker script has to be used in order to place the sections whose
16480 name start with @code{.trampolines} into the segment where @code{EIND}
16484 The startup code from libgcc never sets @code{EIND}.
16485 Notice that startup code is a blend of code from libgcc and AVR-LibC.
16486 For the impact of AVR-LibC on @code{EIND}, see the
16487 @w{@uref{http://nongnu.org/avr-libc/user-manual/,AVR-LibC user manual}}.
16490 It is legitimate for user-specific startup code to set up @code{EIND}
16491 early, for example by means of initialization code located in
16492 section @code{.init3}. Such code runs prior to general startup code
16493 that initializes RAM and calls constructors, but after the bit
16494 of startup code from AVR-LibC that sets @code{EIND} to the segment
16495 where the vector table is located.
16497 #include <avr/io.h>
16500 __attribute__((section(".init3"),naked,used,no_instrument_function))
16501 init3_set_eind (void)
16503 __asm volatile ("ldi r24,pm_hh8(__trampolines_start)\n\t"
16504 "out %i0,r24" :: "n" (&EIND) : "r24","memory");
16509 The @code{__trampolines_start} symbol is defined in the linker script.
16512 Stubs are generated automatically by the linker if
16513 the following two conditions are met:
16516 @item The address of a label is taken by means of the @code{gs} modifier
16517 (short for @emph{generate stubs}) like so:
16519 LDI r24, lo8(gs(@var{func}))
16520 LDI r25, hi8(gs(@var{func}))
16522 @item The final location of that label is in a code segment
16523 @emph{outside} the segment where the stubs are located.
16527 The compiler emits such @code{gs} modifiers for code labels in the
16528 following situations:
16530 @item Taking address of a function or code label.
16531 @item Computed goto.
16532 @item If prologue-save function is used, see @option{-mcall-prologues}
16533 command-line option.
16534 @item Switch/case dispatch tables. If you do not want such dispatch
16535 tables you can specify the @option{-fno-jump-tables} command-line option.
16536 @item C and C++ constructors/destructors called during startup/shutdown.
16537 @item If the tools hit a @code{gs()} modifier explained above.
16541 Jumping to non-symbolic addresses like so is @emph{not} supported:
16546 /* Call function at word address 0x2 */
16547 return ((int(*)(void)) 0x2)();
16551 Instead, a stub has to be set up, i.e.@: the function has to be called
16552 through a symbol (@code{func_4} in the example):
16557 extern int func_4 (void);
16559 /* Call function at byte address 0x4 */
16564 and the application be linked with @option{-Wl,--defsym,func_4=0x4}.
16565 Alternatively, @code{func_4} can be defined in the linker script.
16568 @subsubsection Handling of the @code{RAMPD}, @code{RAMPX}, @code{RAMPY} and @code{RAMPZ} Special Function Registers
16569 @cindex @code{RAMPD}
16570 @cindex @code{RAMPX}
16571 @cindex @code{RAMPY}
16572 @cindex @code{RAMPZ}
16573 Some AVR devices support memories larger than the 64@tie{}KiB range
16574 that can be accessed with 16-bit pointers. To access memory locations
16575 outside this 64@tie{}KiB range, the content of a @code{RAMP}
16576 register is used as high part of the address:
16577 The @code{X}, @code{Y}, @code{Z} address register is concatenated
16578 with the @code{RAMPX}, @code{RAMPY}, @code{RAMPZ} special function
16579 register, respectively, to get a wide address. Similarly,
16580 @code{RAMPD} is used together with direct addressing.
16584 The startup code initializes the @code{RAMP} special function
16585 registers with zero.
16588 If a @ref{AVR Named Address Spaces,named address space} other than
16589 generic or @code{__flash} is used, then @code{RAMPZ} is set
16590 as needed before the operation.
16593 If the device supports RAM larger than 64@tie{}KiB and the compiler
16594 needs to change @code{RAMPZ} to accomplish an operation, @code{RAMPZ}
16595 is reset to zero after the operation.
16598 If the device comes with a specific @code{RAMP} register, the ISR
16599 prologue/epilogue saves/restores that SFR and initializes it with
16600 zero in case the ISR code might (implicitly) use it.
16603 RAM larger than 64@tie{}KiB is not supported by GCC for AVR targets.
16604 If you use inline assembler to read from locations outside the
16605 16-bit address range and change one of the @code{RAMP} registers,
16606 you must reset it to zero after the access.
16610 @subsubsection AVR Built-in Macros
16612 GCC defines several built-in macros so that the user code can test
16613 for the presence or absence of features. Almost any of the following
16614 built-in macros are deduced from device capabilities and thus
16615 triggered by the @option{-mmcu=} command-line option.
16617 For even more AVR-specific built-in macros see
16618 @ref{AVR Named Address Spaces} and @ref{AVR Built-in Functions}.
16623 Build-in macro that resolves to a decimal number that identifies the
16624 architecture and depends on the @option{-mmcu=@var{mcu}} option.
16625 Possible values are:
16627 @code{2}, @code{25}, @code{3}, @code{31}, @code{35},
16628 @code{4}, @code{5}, @code{51}, @code{6}
16630 for @var{mcu}=@code{avr2}, @code{avr25}, @code{avr3}, @code{avr31},
16631 @code{avr35}, @code{avr4}, @code{avr5}, @code{avr51}, @code{avr6},
16636 @code{102}, @code{103}, @code{104},
16637 @code{105}, @code{106}, @code{107}
16639 for @var{mcu}=@code{avrtiny},
16640 @code{avrxmega2}, @code{avrxmega3}, @code{avrxmega4},
16641 @code{avrxmega5}, @code{avrxmega6}, @code{avrxmega7}, respectively.
16642 If @var{mcu} specifies a device, this built-in macro is set
16643 accordingly. For example, with @option{-mmcu=atmega8} the macro is
16644 defined to @code{4}.
16646 @item __AVR_@var{Device}__
16647 Setting @option{-mmcu=@var{device}} defines this built-in macro which reflects
16648 the device's name. For example, @option{-mmcu=atmega8} defines the
16649 built-in macro @code{__AVR_ATmega8__}, @option{-mmcu=attiny261a} defines
16650 @code{__AVR_ATtiny261A__}, etc.
16652 The built-in macros' names follow
16653 the scheme @code{__AVR_@var{Device}__} where @var{Device} is
16654 the device name as from the AVR user manual. The difference between
16655 @var{Device} in the built-in macro and @var{device} in
16656 @option{-mmcu=@var{device}} is that the latter is always lowercase.
16658 If @var{device} is not a device but only a core architecture like
16659 @samp{avr51}, this macro is not defined.
16661 @item __AVR_DEVICE_NAME__
16662 Setting @option{-mmcu=@var{device}} defines this built-in macro to
16663 the device's name. For example, with @option{-mmcu=atmega8} the macro
16664 is defined to @code{atmega8}.
16666 If @var{device} is not a device but only a core architecture like
16667 @samp{avr51}, this macro is not defined.
16669 @item __AVR_XMEGA__
16670 The device / architecture belongs to the XMEGA family of devices.
16672 @item __AVR_HAVE_ELPM__
16673 The device has the @code{ELPM} instruction.
16675 @item __AVR_HAVE_ELPMX__
16676 The device has the @code{ELPM R@var{n},Z} and @code{ELPM
16677 R@var{n},Z+} instructions.
16679 @item __AVR_HAVE_MOVW__
16680 The device has the @code{MOVW} instruction to perform 16-bit
16681 register-register moves.
16683 @item __AVR_HAVE_LPMX__
16684 The device has the @code{LPM R@var{n},Z} and
16685 @code{LPM R@var{n},Z+} instructions.
16687 @item __AVR_HAVE_MUL__
16688 The device has a hardware multiplier.
16690 @item __AVR_HAVE_JMP_CALL__
16691 The device has the @code{JMP} and @code{CALL} instructions.
16692 This is the case for devices with more than 8@tie{}KiB of program
16695 @item __AVR_HAVE_EIJMP_EICALL__
16696 @itemx __AVR_3_BYTE_PC__
16697 The device has the @code{EIJMP} and @code{EICALL} instructions.
16698 This is the case for devices with more than 128@tie{}KiB of program memory.
16699 This also means that the program counter
16700 (PC) is 3@tie{}bytes wide.
16702 @item __AVR_2_BYTE_PC__
16703 The program counter (PC) is 2@tie{}bytes wide. This is the case for devices
16704 with up to 128@tie{}KiB of program memory.
16706 @item __AVR_HAVE_8BIT_SP__
16707 @itemx __AVR_HAVE_16BIT_SP__
16708 The stack pointer (SP) register is treated as 8-bit respectively
16709 16-bit register by the compiler.
16710 The definition of these macros is affected by @option{-mtiny-stack}.
16712 @item __AVR_HAVE_SPH__
16714 The device has the SPH (high part of stack pointer) special function
16715 register or has an 8-bit stack pointer, respectively.
16716 The definition of these macros is affected by @option{-mmcu=} and
16717 in the cases of @option{-mmcu=avr2} and @option{-mmcu=avr25} also
16720 @item __AVR_HAVE_RAMPD__
16721 @itemx __AVR_HAVE_RAMPX__
16722 @itemx __AVR_HAVE_RAMPY__
16723 @itemx __AVR_HAVE_RAMPZ__
16724 The device has the @code{RAMPD}, @code{RAMPX}, @code{RAMPY},
16725 @code{RAMPZ} special function register, respectively.
16727 @item __NO_INTERRUPTS__
16728 This macro reflects the @option{-mno-interrupts} command-line option.
16730 @item __AVR_ERRATA_SKIP__
16731 @itemx __AVR_ERRATA_SKIP_JMP_CALL__
16732 Some AVR devices (AT90S8515, ATmega103) must not skip 32-bit
16733 instructions because of a hardware erratum. Skip instructions are
16734 @code{SBRS}, @code{SBRC}, @code{SBIS}, @code{SBIC} and @code{CPSE}.
16735 The second macro is only defined if @code{__AVR_HAVE_JMP_CALL__} is also
16738 @item __AVR_ISA_RMW__
16739 The device has Read-Modify-Write instructions (XCH, LAC, LAS and LAT).
16741 @item __AVR_SFR_OFFSET__=@var{offset}
16742 Instructions that can address I/O special function registers directly
16743 like @code{IN}, @code{OUT}, @code{SBI}, etc.@: may use a different
16744 address as if addressed by an instruction to access RAM like @code{LD}
16745 or @code{STS}. This offset depends on the device architecture and has
16746 to be subtracted from the RAM address in order to get the
16747 respective I/O@tie{}address.
16749 @item __AVR_SHORT_CALLS__
16750 The @option{-mshort-calls} command line option is set.
16752 @item __AVR_PM_BASE_ADDRESS__=@var{addr}
16753 Some devices support reading from flash memory by means of @code{LD*}
16754 instructions. The flash memory is seen in the data address space
16755 at an offset of @code{__AVR_PM_BASE_ADDRESS__}. If this macro
16756 is not defined, this feature is not available. If defined,
16757 the address space is linear and there is no need to put
16758 @code{.rodata} into RAM. This is handled by the default linker
16759 description file, and is currently available for
16760 @code{avrtiny} and @code{avrxmega3}. Even more convenient,
16761 there is no need to use address spaces like @code{__flash} or
16762 features like attribute @code{progmem} and @code{pgm_read_*}.
16764 @item __WITH_AVRLIBC__
16765 The compiler is configured to be used together with AVR-Libc.
16766 See the @option{--with-avrlibc} configure option.
16770 @node Blackfin Options
16771 @subsection Blackfin Options
16772 @cindex Blackfin Options
16775 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
16777 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
16778 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
16779 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
16780 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
16781 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
16782 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
16783 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
16784 @samp{bf561}, @samp{bf592}.
16786 The optional @var{sirevision} specifies the silicon revision of the target
16787 Blackfin processor. Any workarounds available for the targeted silicon revision
16788 are enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
16789 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
16790 are enabled. The @code{__SILICON_REVISION__} macro is defined to two
16791 hexadecimal digits representing the major and minor numbers in the silicon
16792 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
16793 is not defined. If @var{sirevision} is @samp{any}, the
16794 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
16795 If this optional @var{sirevision} is not used, GCC assumes the latest known
16796 silicon revision of the targeted Blackfin processor.
16798 GCC defines a preprocessor macro for the specified @var{cpu}.
16799 For the @samp{bfin-elf} toolchain, this option causes the hardware BSP
16800 provided by libgloss to be linked in if @option{-msim} is not given.
16802 Without this option, @samp{bf532} is used as the processor by default.
16804 Note that support for @samp{bf561} is incomplete. For @samp{bf561},
16805 only the preprocessor macro is defined.
16809 Specifies that the program will be run on the simulator. This causes
16810 the simulator BSP provided by libgloss to be linked in. This option
16811 has effect only for @samp{bfin-elf} toolchain.
16812 Certain other options, such as @option{-mid-shared-library} and
16813 @option{-mfdpic}, imply @option{-msim}.
16815 @item -momit-leaf-frame-pointer
16816 @opindex momit-leaf-frame-pointer
16817 Don't keep the frame pointer in a register for leaf functions. This
16818 avoids the instructions to save, set up and restore frame pointers and
16819 makes an extra register available in leaf functions.
16821 @item -mspecld-anomaly
16822 @opindex mspecld-anomaly
16823 When enabled, the compiler ensures that the generated code does not
16824 contain speculative loads after jump instructions. If this option is used,
16825 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
16827 @item -mno-specld-anomaly
16828 @opindex mno-specld-anomaly
16829 Don't generate extra code to prevent speculative loads from occurring.
16831 @item -mcsync-anomaly
16832 @opindex mcsync-anomaly
16833 When enabled, the compiler ensures that the generated code does not
16834 contain CSYNC or SSYNC instructions too soon after conditional branches.
16835 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
16837 @item -mno-csync-anomaly
16838 @opindex mno-csync-anomaly
16839 Don't generate extra code to prevent CSYNC or SSYNC instructions from
16840 occurring too soon after a conditional branch.
16844 When enabled, the compiler is free to take advantage of the knowledge that
16845 the entire program fits into the low 64k of memory.
16848 @opindex mno-low-64k
16849 Assume that the program is arbitrarily large. This is the default.
16851 @item -mstack-check-l1
16852 @opindex mstack-check-l1
16853 Do stack checking using information placed into L1 scratchpad memory by the
16856 @item -mid-shared-library
16857 @opindex mid-shared-library
16858 Generate code that supports shared libraries via the library ID method.
16859 This allows for execute in place and shared libraries in an environment
16860 without virtual memory management. This option implies @option{-fPIC}.
16861 With a @samp{bfin-elf} target, this option implies @option{-msim}.
16863 @item -mno-id-shared-library
16864 @opindex mno-id-shared-library
16865 Generate code that doesn't assume ID-based shared libraries are being used.
16866 This is the default.
16868 @item -mleaf-id-shared-library
16869 @opindex mleaf-id-shared-library
16870 Generate code that supports shared libraries via the library ID method,
16871 but assumes that this library or executable won't link against any other
16872 ID shared libraries. That allows the compiler to use faster code for jumps
16875 @item -mno-leaf-id-shared-library
16876 @opindex mno-leaf-id-shared-library
16877 Do not assume that the code being compiled won't link against any ID shared
16878 libraries. Slower code is generated for jump and call insns.
16880 @item -mshared-library-id=n
16881 @opindex mshared-library-id
16882 Specifies the identification number of the ID-based shared library being
16883 compiled. Specifying a value of 0 generates more compact code; specifying
16884 other values forces the allocation of that number to the current
16885 library but is no more space- or time-efficient than omitting this option.
16889 Generate code that allows the data segment to be located in a different
16890 area of memory from the text segment. This allows for execute in place in
16891 an environment without virtual memory management by eliminating relocations
16892 against the text section.
16894 @item -mno-sep-data
16895 @opindex mno-sep-data
16896 Generate code that assumes that the data segment follows the text segment.
16897 This is the default.
16900 @itemx -mno-long-calls
16901 @opindex mlong-calls
16902 @opindex mno-long-calls
16903 Tells the compiler to perform function calls by first loading the
16904 address of the function into a register and then performing a subroutine
16905 call on this register. This switch is needed if the target function
16906 lies outside of the 24-bit addressing range of the offset-based
16907 version of subroutine call instruction.
16909 This feature is not enabled by default. Specifying
16910 @option{-mno-long-calls} restores the default behavior. Note these
16911 switches have no effect on how the compiler generates code to handle
16912 function calls via function pointers.
16916 Link with the fast floating-point library. This library relaxes some of
16917 the IEEE floating-point standard's rules for checking inputs against
16918 Not-a-Number (NAN), in the interest of performance.
16921 @opindex minline-plt
16922 Enable inlining of PLT entries in function calls to functions that are
16923 not known to bind locally. It has no effect without @option{-mfdpic}.
16926 @opindex mmulticore
16927 Build a standalone application for multicore Blackfin processors.
16928 This option causes proper start files and link scripts supporting
16929 multicore to be used, and defines the macro @code{__BFIN_MULTICORE}.
16930 It can only be used with @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}.
16932 This option can be used with @option{-mcorea} or @option{-mcoreb}, which
16933 selects the one-application-per-core programming model. Without
16934 @option{-mcorea} or @option{-mcoreb}, the single-application/dual-core
16935 programming model is used. In this model, the main function of Core B
16936 should be named as @code{coreb_main}.
16938 If this option is not used, the single-core application programming
16943 Build a standalone application for Core A of BF561 when using
16944 the one-application-per-core programming model. Proper start files
16945 and link scripts are used to support Core A, and the macro
16946 @code{__BFIN_COREA} is defined.
16947 This option can only be used in conjunction with @option{-mmulticore}.
16951 Build a standalone application for Core B of BF561 when using
16952 the one-application-per-core programming model. Proper start files
16953 and link scripts are used to support Core B, and the macro
16954 @code{__BFIN_COREB} is defined. When this option is used, @code{coreb_main}
16955 should be used instead of @code{main}.
16956 This option can only be used in conjunction with @option{-mmulticore}.
16960 Build a standalone application for SDRAM. Proper start files and
16961 link scripts are used to put the application into SDRAM, and the macro
16962 @code{__BFIN_SDRAM} is defined.
16963 The loader should initialize SDRAM before loading the application.
16967 Assume that ICPLBs are enabled at run time. This has an effect on certain
16968 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
16969 are enabled; for standalone applications the default is off.
16973 @subsection C6X Options
16974 @cindex C6X Options
16977 @item -march=@var{name}
16979 This specifies the name of the target architecture. GCC uses this
16980 name to determine what kind of instructions it can emit when generating
16981 assembly code. Permissible names are: @samp{c62x},
16982 @samp{c64x}, @samp{c64x+}, @samp{c67x}, @samp{c67x+}, @samp{c674x}.
16985 @opindex mbig-endian
16986 Generate code for a big-endian target.
16988 @item -mlittle-endian
16989 @opindex mlittle-endian
16990 Generate code for a little-endian target. This is the default.
16994 Choose startup files and linker script suitable for the simulator.
16996 @item -msdata=default
16997 @opindex msdata=default
16998 Put small global and static data in the @code{.neardata} section,
16999 which is pointed to by register @code{B14}. Put small uninitialized
17000 global and static data in the @code{.bss} section, which is adjacent
17001 to the @code{.neardata} section. Put small read-only data into the
17002 @code{.rodata} section. The corresponding sections used for large
17003 pieces of data are @code{.fardata}, @code{.far} and @code{.const}.
17006 @opindex msdata=all
17007 Put all data, not just small objects, into the sections reserved for
17008 small data, and use addressing relative to the @code{B14} register to
17012 @opindex msdata=none
17013 Make no use of the sections reserved for small data, and use absolute
17014 addresses to access all data. Put all initialized global and static
17015 data in the @code{.fardata} section, and all uninitialized data in the
17016 @code{.far} section. Put all constant data into the @code{.const}
17021 @subsection CRIS Options
17022 @cindex CRIS Options
17024 These options are defined specifically for the CRIS ports.
17027 @item -march=@var{architecture-type}
17028 @itemx -mcpu=@var{architecture-type}
17031 Generate code for the specified architecture. The choices for
17032 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
17033 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
17034 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
17037 @item -mtune=@var{architecture-type}
17039 Tune to @var{architecture-type} everything applicable about the generated
17040 code, except for the ABI and the set of available instructions. The
17041 choices for @var{architecture-type} are the same as for
17042 @option{-march=@var{architecture-type}}.
17044 @item -mmax-stack-frame=@var{n}
17045 @opindex mmax-stack-frame
17046 Warn when the stack frame of a function exceeds @var{n} bytes.
17052 The options @option{-metrax4} and @option{-metrax100} are synonyms for
17053 @option{-march=v3} and @option{-march=v8} respectively.
17055 @item -mmul-bug-workaround
17056 @itemx -mno-mul-bug-workaround
17057 @opindex mmul-bug-workaround
17058 @opindex mno-mul-bug-workaround
17059 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
17060 models where it applies. This option is active by default.
17064 Enable CRIS-specific verbose debug-related information in the assembly
17065 code. This option also has the effect of turning off the @samp{#NO_APP}
17066 formatted-code indicator to the assembler at the beginning of the
17071 Do not use condition-code results from previous instruction; always emit
17072 compare and test instructions before use of condition codes.
17074 @item -mno-side-effects
17075 @opindex mno-side-effects
17076 Do not emit instructions with side effects in addressing modes other than
17079 @item -mstack-align
17080 @itemx -mno-stack-align
17081 @itemx -mdata-align
17082 @itemx -mno-data-align
17083 @itemx -mconst-align
17084 @itemx -mno-const-align
17085 @opindex mstack-align
17086 @opindex mno-stack-align
17087 @opindex mdata-align
17088 @opindex mno-data-align
17089 @opindex mconst-align
17090 @opindex mno-const-align
17091 These options (@samp{no-} options) arrange (eliminate arrangements) for the
17092 stack frame, individual data and constants to be aligned for the maximum
17093 single data access size for the chosen CPU model. The default is to
17094 arrange for 32-bit alignment. ABI details such as structure layout are
17095 not affected by these options.
17103 Similar to the stack- data- and const-align options above, these options
17104 arrange for stack frame, writable data and constants to all be 32-bit,
17105 16-bit or 8-bit aligned. The default is 32-bit alignment.
17107 @item -mno-prologue-epilogue
17108 @itemx -mprologue-epilogue
17109 @opindex mno-prologue-epilogue
17110 @opindex mprologue-epilogue
17111 With @option{-mno-prologue-epilogue}, the normal function prologue and
17112 epilogue which set up the stack frame are omitted and no return
17113 instructions or return sequences are generated in the code. Use this
17114 option only together with visual inspection of the compiled code: no
17115 warnings or errors are generated when call-saved registers must be saved,
17116 or storage for local variables needs to be allocated.
17120 @opindex mno-gotplt
17122 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
17123 instruction sequences that load addresses for functions from the PLT part
17124 of the GOT rather than (traditional on other architectures) calls to the
17125 PLT@. The default is @option{-mgotplt}.
17129 Legacy no-op option only recognized with the cris-axis-elf and
17130 cris-axis-linux-gnu targets.
17134 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
17138 This option, recognized for the cris-axis-elf, arranges
17139 to link with input-output functions from a simulator library. Code,
17140 initialized data and zero-initialized data are allocated consecutively.
17144 Like @option{-sim}, but pass linker options to locate initialized data at
17145 0x40000000 and zero-initialized data at 0x80000000.
17149 @subsection CR16 Options
17150 @cindex CR16 Options
17152 These options are defined specifically for the CR16 ports.
17158 Enable the use of multiply-accumulate instructions. Disabled by default.
17162 @opindex mcr16cplus
17164 Generate code for CR16C or CR16C+ architecture. CR16C+ architecture
17169 Links the library libsim.a which is in compatible with simulator. Applicable
17170 to ELF compiler only.
17174 Choose integer type as 32-bit wide.
17178 Generates @code{sbit}/@code{cbit} instructions for bit manipulations.
17180 @item -mdata-model=@var{model}
17181 @opindex mdata-model
17182 Choose a data model. The choices for @var{model} are @samp{near},
17183 @samp{far} or @samp{medium}. @samp{medium} is default.
17184 However, @samp{far} is not valid with @option{-mcr16c}, as the
17185 CR16C architecture does not support the far data model.
17188 @node Darwin Options
17189 @subsection Darwin Options
17190 @cindex Darwin options
17192 These options are defined for all architectures running the Darwin operating
17195 FSF GCC on Darwin does not create ``fat'' object files; it creates
17196 an object file for the single architecture that GCC was built to
17197 target. Apple's GCC on Darwin does create ``fat'' files if multiple
17198 @option{-arch} options are used; it does so by running the compiler or
17199 linker multiple times and joining the results together with
17202 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
17203 @samp{i686}) is determined by the flags that specify the ISA
17204 that GCC is targeting, like @option{-mcpu} or @option{-march}. The
17205 @option{-force_cpusubtype_ALL} option can be used to override this.
17207 The Darwin tools vary in their behavior when presented with an ISA
17208 mismatch. The assembler, @file{as}, only permits instructions to
17209 be used that are valid for the subtype of the file it is generating,
17210 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
17211 The linker for shared libraries, @file{/usr/bin/libtool}, fails
17212 and prints an error if asked to create a shared library with a less
17213 restrictive subtype than its input files (for instance, trying to put
17214 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
17215 for executables, @command{ld}, quietly gives the executable the most
17216 restrictive subtype of any of its input files.
17221 Add the framework directory @var{dir} to the head of the list of
17222 directories to be searched for header files. These directories are
17223 interleaved with those specified by @option{-I} options and are
17224 scanned in a left-to-right order.
17226 A framework directory is a directory with frameworks in it. A
17227 framework is a directory with a @file{Headers} and/or
17228 @file{PrivateHeaders} directory contained directly in it that ends
17229 in @file{.framework}. The name of a framework is the name of this
17230 directory excluding the @file{.framework}. Headers associated with
17231 the framework are found in one of those two directories, with
17232 @file{Headers} being searched first. A subframework is a framework
17233 directory that is in a framework's @file{Frameworks} directory.
17234 Includes of subframework headers can only appear in a header of a
17235 framework that contains the subframework, or in a sibling subframework
17236 header. Two subframeworks are siblings if they occur in the same
17237 framework. A subframework should not have the same name as a
17238 framework; a warning is issued if this is violated. Currently a
17239 subframework cannot have subframeworks; in the future, the mechanism
17240 may be extended to support this. The standard frameworks can be found
17241 in @file{/System/Library/Frameworks} and
17242 @file{/Library/Frameworks}. An example include looks like
17243 @code{#include <Framework/header.h>}, where @file{Framework} denotes
17244 the name of the framework and @file{header.h} is found in the
17245 @file{PrivateHeaders} or @file{Headers} directory.
17247 @item -iframework@var{dir}
17248 @opindex iframework
17249 Like @option{-F} except the directory is a treated as a system
17250 directory. The main difference between this @option{-iframework} and
17251 @option{-F} is that with @option{-iframework} the compiler does not
17252 warn about constructs contained within header files found via
17253 @var{dir}. This option is valid only for the C family of languages.
17257 Emit debugging information for symbols that are used. For stabs
17258 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
17259 This is by default ON@.
17263 Emit debugging information for all symbols and types.
17265 @item -mmacosx-version-min=@var{version}
17266 The earliest version of MacOS X that this executable will run on
17267 is @var{version}. Typical values of @var{version} include @code{10.1},
17268 @code{10.2}, and @code{10.3.9}.
17270 If the compiler was built to use the system's headers by default,
17271 then the default for this option is the system version on which the
17272 compiler is running, otherwise the default is to make choices that
17273 are compatible with as many systems and code bases as possible.
17277 Enable kernel development mode. The @option{-mkernel} option sets
17278 @option{-static}, @option{-fno-common}, @option{-fno-use-cxa-atexit},
17279 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
17280 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
17281 applicable. This mode also sets @option{-mno-altivec},
17282 @option{-msoft-float}, @option{-fno-builtin} and
17283 @option{-mlong-branch} for PowerPC targets.
17285 @item -mone-byte-bool
17286 @opindex mone-byte-bool
17287 Override the defaults for @code{bool} so that @code{sizeof(bool)==1}.
17288 By default @code{sizeof(bool)} is @code{4} when compiling for
17289 Darwin/PowerPC and @code{1} when compiling for Darwin/x86, so this
17290 option has no effect on x86.
17292 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
17293 to generate code that is not binary compatible with code generated
17294 without that switch. Using this switch may require recompiling all
17295 other modules in a program, including system libraries. Use this
17296 switch to conform to a non-default data model.
17298 @item -mfix-and-continue
17299 @itemx -ffix-and-continue
17300 @itemx -findirect-data
17301 @opindex mfix-and-continue
17302 @opindex ffix-and-continue
17303 @opindex findirect-data
17304 Generate code suitable for fast turnaround development, such as to
17305 allow GDB to dynamically load @file{.o} files into already-running
17306 programs. @option{-findirect-data} and @option{-ffix-and-continue}
17307 are provided for backwards compatibility.
17311 Loads all members of static archive libraries.
17312 See man ld(1) for more information.
17314 @item -arch_errors_fatal
17315 @opindex arch_errors_fatal
17316 Cause the errors having to do with files that have the wrong architecture
17319 @item -bind_at_load
17320 @opindex bind_at_load
17321 Causes the output file to be marked such that the dynamic linker will
17322 bind all undefined references when the file is loaded or launched.
17326 Produce a Mach-o bundle format file.
17327 See man ld(1) for more information.
17329 @item -bundle_loader @var{executable}
17330 @opindex bundle_loader
17331 This option specifies the @var{executable} that will load the build
17332 output file being linked. See man ld(1) for more information.
17335 @opindex dynamiclib
17336 When passed this option, GCC produces a dynamic library instead of
17337 an executable when linking, using the Darwin @file{libtool} command.
17339 @item -force_cpusubtype_ALL
17340 @opindex force_cpusubtype_ALL
17341 This causes GCC's output file to have the @samp{ALL} subtype, instead of
17342 one controlled by the @option{-mcpu} or @option{-march} option.
17344 @item -allowable_client @var{client_name}
17345 @itemx -client_name
17346 @itemx -compatibility_version
17347 @itemx -current_version
17349 @itemx -dependency-file
17351 @itemx -dylinker_install_name
17353 @itemx -exported_symbols_list
17356 @itemx -flat_namespace
17357 @itemx -force_flat_namespace
17358 @itemx -headerpad_max_install_names
17361 @itemx -install_name
17362 @itemx -keep_private_externs
17363 @itemx -multi_module
17364 @itemx -multiply_defined
17365 @itemx -multiply_defined_unused
17368 @itemx -no_dead_strip_inits_and_terms
17369 @itemx -nofixprebinding
17370 @itemx -nomultidefs
17372 @itemx -noseglinkedit
17373 @itemx -pagezero_size
17375 @itemx -prebind_all_twolevel_modules
17376 @itemx -private_bundle
17378 @itemx -read_only_relocs
17380 @itemx -sectobjectsymbols
17384 @itemx -sectobjectsymbols
17387 @itemx -segs_read_only_addr
17389 @itemx -segs_read_write_addr
17390 @itemx -seg_addr_table
17391 @itemx -seg_addr_table_filename
17392 @itemx -seglinkedit
17394 @itemx -segs_read_only_addr
17395 @itemx -segs_read_write_addr
17396 @itemx -single_module
17398 @itemx -sub_library
17400 @itemx -sub_umbrella
17401 @itemx -twolevel_namespace
17404 @itemx -unexported_symbols_list
17405 @itemx -weak_reference_mismatches
17406 @itemx -whatsloaded
17407 @opindex allowable_client
17408 @opindex client_name
17409 @opindex compatibility_version
17410 @opindex current_version
17411 @opindex dead_strip
17412 @opindex dependency-file
17413 @opindex dylib_file
17414 @opindex dylinker_install_name
17416 @opindex exported_symbols_list
17418 @opindex flat_namespace
17419 @opindex force_flat_namespace
17420 @opindex headerpad_max_install_names
17421 @opindex image_base
17423 @opindex install_name
17424 @opindex keep_private_externs
17425 @opindex multi_module
17426 @opindex multiply_defined
17427 @opindex multiply_defined_unused
17428 @opindex noall_load
17429 @opindex no_dead_strip_inits_and_terms
17430 @opindex nofixprebinding
17431 @opindex nomultidefs
17433 @opindex noseglinkedit
17434 @opindex pagezero_size
17436 @opindex prebind_all_twolevel_modules
17437 @opindex private_bundle
17438 @opindex read_only_relocs
17440 @opindex sectobjectsymbols
17443 @opindex sectcreate
17444 @opindex sectobjectsymbols
17447 @opindex segs_read_only_addr
17448 @opindex segs_read_write_addr
17449 @opindex seg_addr_table
17450 @opindex seg_addr_table_filename
17451 @opindex seglinkedit
17453 @opindex segs_read_only_addr
17454 @opindex segs_read_write_addr
17455 @opindex single_module
17457 @opindex sub_library
17458 @opindex sub_umbrella
17459 @opindex twolevel_namespace
17462 @opindex unexported_symbols_list
17463 @opindex weak_reference_mismatches
17464 @opindex whatsloaded
17465 These options are passed to the Darwin linker. The Darwin linker man page
17466 describes them in detail.
17469 @node DEC Alpha Options
17470 @subsection DEC Alpha Options
17472 These @samp{-m} options are defined for the DEC Alpha implementations:
17475 @item -mno-soft-float
17476 @itemx -msoft-float
17477 @opindex mno-soft-float
17478 @opindex msoft-float
17479 Use (do not use) the hardware floating-point instructions for
17480 floating-point operations. When @option{-msoft-float} is specified,
17481 functions in @file{libgcc.a} are used to perform floating-point
17482 operations. Unless they are replaced by routines that emulate the
17483 floating-point operations, or compiled in such a way as to call such
17484 emulations routines, these routines issue floating-point
17485 operations. If you are compiling for an Alpha without floating-point
17486 operations, you must ensure that the library is built so as not to call
17489 Note that Alpha implementations without floating-point operations are
17490 required to have floating-point registers.
17493 @itemx -mno-fp-regs
17495 @opindex mno-fp-regs
17496 Generate code that uses (does not use) the floating-point register set.
17497 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
17498 register set is not used, floating-point operands are passed in integer
17499 registers as if they were integers and floating-point results are passed
17500 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
17501 so any function with a floating-point argument or return value called by code
17502 compiled with @option{-mno-fp-regs} must also be compiled with that
17505 A typical use of this option is building a kernel that does not use,
17506 and hence need not save and restore, any floating-point registers.
17510 The Alpha architecture implements floating-point hardware optimized for
17511 maximum performance. It is mostly compliant with the IEEE floating-point
17512 standard. However, for full compliance, software assistance is
17513 required. This option generates code fully IEEE-compliant code
17514 @emph{except} that the @var{inexact-flag} is not maintained (see below).
17515 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
17516 defined during compilation. The resulting code is less efficient but is
17517 able to correctly support denormalized numbers and exceptional IEEE
17518 values such as not-a-number and plus/minus infinity. Other Alpha
17519 compilers call this option @option{-ieee_with_no_inexact}.
17521 @item -mieee-with-inexact
17522 @opindex mieee-with-inexact
17523 This is like @option{-mieee} except the generated code also maintains
17524 the IEEE @var{inexact-flag}. Turning on this option causes the
17525 generated code to implement fully-compliant IEEE math. In addition to
17526 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
17527 macro. On some Alpha implementations the resulting code may execute
17528 significantly slower than the code generated by default. Since there is
17529 very little code that depends on the @var{inexact-flag}, you should
17530 normally not specify this option. Other Alpha compilers call this
17531 option @option{-ieee_with_inexact}.
17533 @item -mfp-trap-mode=@var{trap-mode}
17534 @opindex mfp-trap-mode
17535 This option controls what floating-point related traps are enabled.
17536 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
17537 The trap mode can be set to one of four values:
17541 This is the default (normal) setting. The only traps that are enabled
17542 are the ones that cannot be disabled in software (e.g., division by zero
17546 In addition to the traps enabled by @samp{n}, underflow traps are enabled
17550 Like @samp{u}, but the instructions are marked to be safe for software
17551 completion (see Alpha architecture manual for details).
17554 Like @samp{su}, but inexact traps are enabled as well.
17557 @item -mfp-rounding-mode=@var{rounding-mode}
17558 @opindex mfp-rounding-mode
17559 Selects the IEEE rounding mode. Other Alpha compilers call this option
17560 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
17565 Normal IEEE rounding mode. Floating-point numbers are rounded towards
17566 the nearest machine number or towards the even machine number in case
17570 Round towards minus infinity.
17573 Chopped rounding mode. Floating-point numbers are rounded towards zero.
17576 Dynamic rounding mode. A field in the floating-point control register
17577 (@var{fpcr}, see Alpha architecture reference manual) controls the
17578 rounding mode in effect. The C library initializes this register for
17579 rounding towards plus infinity. Thus, unless your program modifies the
17580 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
17583 @item -mtrap-precision=@var{trap-precision}
17584 @opindex mtrap-precision
17585 In the Alpha architecture, floating-point traps are imprecise. This
17586 means without software assistance it is impossible to recover from a
17587 floating trap and program execution normally needs to be terminated.
17588 GCC can generate code that can assist operating system trap handlers
17589 in determining the exact location that caused a floating-point trap.
17590 Depending on the requirements of an application, different levels of
17591 precisions can be selected:
17595 Program precision. This option is the default and means a trap handler
17596 can only identify which program caused a floating-point exception.
17599 Function precision. The trap handler can determine the function that
17600 caused a floating-point exception.
17603 Instruction precision. The trap handler can determine the exact
17604 instruction that caused a floating-point exception.
17607 Other Alpha compilers provide the equivalent options called
17608 @option{-scope_safe} and @option{-resumption_safe}.
17610 @item -mieee-conformant
17611 @opindex mieee-conformant
17612 This option marks the generated code as IEEE conformant. You must not
17613 use this option unless you also specify @option{-mtrap-precision=i} and either
17614 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
17615 is to emit the line @samp{.eflag 48} in the function prologue of the
17616 generated assembly file.
17618 @item -mbuild-constants
17619 @opindex mbuild-constants
17620 Normally GCC examines a 32- or 64-bit integer constant to
17621 see if it can construct it from smaller constants in two or three
17622 instructions. If it cannot, it outputs the constant as a literal and
17623 generates code to load it from the data segment at run time.
17625 Use this option to require GCC to construct @emph{all} integer constants
17626 using code, even if it takes more instructions (the maximum is six).
17628 You typically use this option to build a shared library dynamic
17629 loader. Itself a shared library, it must relocate itself in memory
17630 before it can find the variables and constants in its own data segment.
17648 Indicate whether GCC should generate code to use the optional BWX,
17649 CIX, FIX and MAX instruction sets. The default is to use the instruction
17650 sets supported by the CPU type specified via @option{-mcpu=} option or that
17651 of the CPU on which GCC was built if none is specified.
17654 @itemx -mfloat-ieee
17655 @opindex mfloat-vax
17656 @opindex mfloat-ieee
17657 Generate code that uses (does not use) VAX F and G floating-point
17658 arithmetic instead of IEEE single and double precision.
17660 @item -mexplicit-relocs
17661 @itemx -mno-explicit-relocs
17662 @opindex mexplicit-relocs
17663 @opindex mno-explicit-relocs
17664 Older Alpha assemblers provided no way to generate symbol relocations
17665 except via assembler macros. Use of these macros does not allow
17666 optimal instruction scheduling. GNU binutils as of version 2.12
17667 supports a new syntax that allows the compiler to explicitly mark
17668 which relocations should apply to which instructions. This option
17669 is mostly useful for debugging, as GCC detects the capabilities of
17670 the assembler when it is built and sets the default accordingly.
17673 @itemx -mlarge-data
17674 @opindex msmall-data
17675 @opindex mlarge-data
17676 When @option{-mexplicit-relocs} is in effect, static data is
17677 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
17678 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
17679 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
17680 16-bit relocations off of the @code{$gp} register. This limits the
17681 size of the small data area to 64KB, but allows the variables to be
17682 directly accessed via a single instruction.
17684 The default is @option{-mlarge-data}. With this option the data area
17685 is limited to just below 2GB@. Programs that require more than 2GB of
17686 data must use @code{malloc} or @code{mmap} to allocate the data in the
17687 heap instead of in the program's data segment.
17689 When generating code for shared libraries, @option{-fpic} implies
17690 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
17693 @itemx -mlarge-text
17694 @opindex msmall-text
17695 @opindex mlarge-text
17696 When @option{-msmall-text} is used, the compiler assumes that the
17697 code of the entire program (or shared library) fits in 4MB, and is
17698 thus reachable with a branch instruction. When @option{-msmall-data}
17699 is used, the compiler can assume that all local symbols share the
17700 same @code{$gp} value, and thus reduce the number of instructions
17701 required for a function call from 4 to 1.
17703 The default is @option{-mlarge-text}.
17705 @item -mcpu=@var{cpu_type}
17707 Set the instruction set and instruction scheduling parameters for
17708 machine type @var{cpu_type}. You can specify either the @samp{EV}
17709 style name or the corresponding chip number. GCC supports scheduling
17710 parameters for the EV4, EV5 and EV6 family of processors and
17711 chooses the default values for the instruction set from the processor
17712 you specify. If you do not specify a processor type, GCC defaults
17713 to the processor on which the compiler was built.
17715 Supported values for @var{cpu_type} are
17721 Schedules as an EV4 and has no instruction set extensions.
17725 Schedules as an EV5 and has no instruction set extensions.
17729 Schedules as an EV5 and supports the BWX extension.
17734 Schedules as an EV5 and supports the BWX and MAX extensions.
17738 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
17742 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
17745 Native toolchains also support the value @samp{native},
17746 which selects the best architecture option for the host processor.
17747 @option{-mcpu=native} has no effect if GCC does not recognize
17750 @item -mtune=@var{cpu_type}
17752 Set only the instruction scheduling parameters for machine type
17753 @var{cpu_type}. The instruction set is not changed.
17755 Native toolchains also support the value @samp{native},
17756 which selects the best architecture option for the host processor.
17757 @option{-mtune=native} has no effect if GCC does not recognize
17760 @item -mmemory-latency=@var{time}
17761 @opindex mmemory-latency
17762 Sets the latency the scheduler should assume for typical memory
17763 references as seen by the application. This number is highly
17764 dependent on the memory access patterns used by the application
17765 and the size of the external cache on the machine.
17767 Valid options for @var{time} are
17771 A decimal number representing clock cycles.
17777 The compiler contains estimates of the number of clock cycles for
17778 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
17779 (also called Dcache, Scache, and Bcache), as well as to main memory.
17780 Note that L3 is only valid for EV5.
17786 @subsection FR30 Options
17787 @cindex FR30 Options
17789 These options are defined specifically for the FR30 port.
17793 @item -msmall-model
17794 @opindex msmall-model
17795 Use the small address space model. This can produce smaller code, but
17796 it does assume that all symbolic values and addresses fit into a
17801 Assume that runtime support has been provided and so there is no need
17802 to include the simulator library (@file{libsim.a}) on the linker
17808 @subsection FT32 Options
17809 @cindex FT32 Options
17811 These options are defined specifically for the FT32 port.
17817 Specifies that the program will be run on the simulator. This causes
17818 an alternate runtime startup and library to be linked.
17819 You must not use this option when generating programs that will run on
17820 real hardware; you must provide your own runtime library for whatever
17821 I/O functions are needed.
17825 Enable Local Register Allocation. This is still experimental for FT32,
17826 so by default the compiler uses standard reload.
17830 Do not use div and mod instructions.
17834 Enable use of the extended instructions of the FT32B processor.
17838 Compress all code using the Ft32B code compression scheme.
17842 Do not generate code that reads program memory.
17847 @subsection FRV Options
17848 @cindex FRV Options
17854 Only use the first 32 general-purpose registers.
17859 Use all 64 general-purpose registers.
17864 Use only the first 32 floating-point registers.
17869 Use all 64 floating-point registers.
17872 @opindex mhard-float
17874 Use hardware instructions for floating-point operations.
17877 @opindex msoft-float
17879 Use library routines for floating-point operations.
17884 Dynamically allocate condition code registers.
17889 Do not try to dynamically allocate condition code registers, only
17890 use @code{icc0} and @code{fcc0}.
17895 Change ABI to use double word insns.
17900 Do not use double word instructions.
17905 Use floating-point double instructions.
17908 @opindex mno-double
17910 Do not use floating-point double instructions.
17915 Use media instructions.
17920 Do not use media instructions.
17925 Use multiply and add/subtract instructions.
17928 @opindex mno-muladd
17930 Do not use multiply and add/subtract instructions.
17935 Select the FDPIC ABI, which uses function descriptors to represent
17936 pointers to functions. Without any PIC/PIE-related options, it
17937 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
17938 assumes GOT entries and small data are within a 12-bit range from the
17939 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
17940 are computed with 32 bits.
17941 With a @samp{bfin-elf} target, this option implies @option{-msim}.
17944 @opindex minline-plt
17946 Enable inlining of PLT entries in function calls to functions that are
17947 not known to bind locally. It has no effect without @option{-mfdpic}.
17948 It's enabled by default if optimizing for speed and compiling for
17949 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
17950 optimization option such as @option{-O3} or above is present in the
17956 Assume a large TLS segment when generating thread-local code.
17961 Do not assume a large TLS segment when generating thread-local code.
17966 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
17967 that is known to be in read-only sections. It's enabled by default,
17968 except for @option{-fpic} or @option{-fpie}: even though it may help
17969 make the global offset table smaller, it trades 1 instruction for 4.
17970 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
17971 one of which may be shared by multiple symbols, and it avoids the need
17972 for a GOT entry for the referenced symbol, so it's more likely to be a
17973 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
17975 @item -multilib-library-pic
17976 @opindex multilib-library-pic
17978 Link with the (library, not FD) pic libraries. It's implied by
17979 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
17980 @option{-fpic} without @option{-mfdpic}. You should never have to use
17984 @opindex mlinked-fp
17986 Follow the EABI requirement of always creating a frame pointer whenever
17987 a stack frame is allocated. This option is enabled by default and can
17988 be disabled with @option{-mno-linked-fp}.
17991 @opindex mlong-calls
17993 Use indirect addressing to call functions outside the current
17994 compilation unit. This allows the functions to be placed anywhere
17995 within the 32-bit address space.
17997 @item -malign-labels
17998 @opindex malign-labels
18000 Try to align labels to an 8-byte boundary by inserting NOPs into the
18001 previous packet. This option only has an effect when VLIW packing
18002 is enabled. It doesn't create new packets; it merely adds NOPs to
18005 @item -mlibrary-pic
18006 @opindex mlibrary-pic
18008 Generate position-independent EABI code.
18013 Use only the first four media accumulator registers.
18018 Use all eight media accumulator registers.
18023 Pack VLIW instructions.
18028 Do not pack VLIW instructions.
18031 @opindex mno-eflags
18033 Do not mark ABI switches in e_flags.
18036 @opindex mcond-move
18038 Enable the use of conditional-move instructions (default).
18040 This switch is mainly for debugging the compiler and will likely be removed
18041 in a future version.
18043 @item -mno-cond-move
18044 @opindex mno-cond-move
18046 Disable the use of conditional-move instructions.
18048 This switch is mainly for debugging the compiler and will likely be removed
18049 in a future version.
18054 Enable the use of conditional set instructions (default).
18056 This switch is mainly for debugging the compiler and will likely be removed
18057 in a future version.
18062 Disable the use of conditional set instructions.
18064 This switch is mainly for debugging the compiler and will likely be removed
18065 in a future version.
18068 @opindex mcond-exec
18070 Enable the use of conditional execution (default).
18072 This switch is mainly for debugging the compiler and will likely be removed
18073 in a future version.
18075 @item -mno-cond-exec
18076 @opindex mno-cond-exec
18078 Disable the use of conditional execution.
18080 This switch is mainly for debugging the compiler and will likely be removed
18081 in a future version.
18083 @item -mvliw-branch
18084 @opindex mvliw-branch
18086 Run a pass to pack branches into VLIW instructions (default).
18088 This switch is mainly for debugging the compiler and will likely be removed
18089 in a future version.
18091 @item -mno-vliw-branch
18092 @opindex mno-vliw-branch
18094 Do not run a pass to pack branches into VLIW instructions.
18096 This switch is mainly for debugging the compiler and will likely be removed
18097 in a future version.
18099 @item -mmulti-cond-exec
18100 @opindex mmulti-cond-exec
18102 Enable optimization of @code{&&} and @code{||} in conditional execution
18105 This switch is mainly for debugging the compiler and will likely be removed
18106 in a future version.
18108 @item -mno-multi-cond-exec
18109 @opindex mno-multi-cond-exec
18111 Disable optimization of @code{&&} and @code{||} in conditional execution.
18113 This switch is mainly for debugging the compiler and will likely be removed
18114 in a future version.
18116 @item -mnested-cond-exec
18117 @opindex mnested-cond-exec
18119 Enable nested conditional execution optimizations (default).
18121 This switch is mainly for debugging the compiler and will likely be removed
18122 in a future version.
18124 @item -mno-nested-cond-exec
18125 @opindex mno-nested-cond-exec
18127 Disable nested conditional execution optimizations.
18129 This switch is mainly for debugging the compiler and will likely be removed
18130 in a future version.
18132 @item -moptimize-membar
18133 @opindex moptimize-membar
18135 This switch removes redundant @code{membar} instructions from the
18136 compiler-generated code. It is enabled by default.
18138 @item -mno-optimize-membar
18139 @opindex mno-optimize-membar
18141 This switch disables the automatic removal of redundant @code{membar}
18142 instructions from the generated code.
18144 @item -mtomcat-stats
18145 @opindex mtomcat-stats
18147 Cause gas to print out tomcat statistics.
18149 @item -mcpu=@var{cpu}
18152 Select the processor type for which to generate code. Possible values are
18153 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
18154 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
18158 @node GNU/Linux Options
18159 @subsection GNU/Linux Options
18161 These @samp{-m} options are defined for GNU/Linux targets:
18166 Use the GNU C library. This is the default except
18167 on @samp{*-*-linux-*uclibc*}, @samp{*-*-linux-*musl*} and
18168 @samp{*-*-linux-*android*} targets.
18172 Use uClibc C library. This is the default on
18173 @samp{*-*-linux-*uclibc*} targets.
18177 Use the musl C library. This is the default on
18178 @samp{*-*-linux-*musl*} targets.
18182 Use Bionic C library. This is the default on
18183 @samp{*-*-linux-*android*} targets.
18187 Compile code compatible with Android platform. This is the default on
18188 @samp{*-*-linux-*android*} targets.
18190 When compiling, this option enables @option{-mbionic}, @option{-fPIC},
18191 @option{-fno-exceptions} and @option{-fno-rtti} by default. When linking,
18192 this option makes the GCC driver pass Android-specific options to the linker.
18193 Finally, this option causes the preprocessor macro @code{__ANDROID__}
18196 @item -tno-android-cc
18197 @opindex tno-android-cc
18198 Disable compilation effects of @option{-mandroid}, i.e., do not enable
18199 @option{-mbionic}, @option{-fPIC}, @option{-fno-exceptions} and
18200 @option{-fno-rtti} by default.
18202 @item -tno-android-ld
18203 @opindex tno-android-ld
18204 Disable linking effects of @option{-mandroid}, i.e., pass standard Linux
18205 linking options to the linker.
18209 @node H8/300 Options
18210 @subsection H8/300 Options
18212 These @samp{-m} options are defined for the H8/300 implementations:
18217 Shorten some address references at link time, when possible; uses the
18218 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
18219 ld, Using ld}, for a fuller description.
18223 Generate code for the H8/300H@.
18227 Generate code for the H8S@.
18231 Generate code for the H8S and H8/300H in the normal mode. This switch
18232 must be used either with @option{-mh} or @option{-ms}.
18236 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
18240 Extended registers are stored on stack before execution of function
18241 with monitor attribute. Default option is @option{-mexr}.
18242 This option is valid only for H8S targets.
18246 Extended registers are not stored on stack before execution of function
18247 with monitor attribute. Default option is @option{-mno-exr}.
18248 This option is valid only for H8S targets.
18252 Make @code{int} data 32 bits by default.
18255 @opindex malign-300
18256 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
18257 The default for the H8/300H and H8S is to align longs and floats on
18259 @option{-malign-300} causes them to be aligned on 2-byte boundaries.
18260 This option has no effect on the H8/300.
18264 @subsection HPPA Options
18265 @cindex HPPA Options
18267 These @samp{-m} options are defined for the HPPA family of computers:
18270 @item -march=@var{architecture-type}
18272 Generate code for the specified architecture. The choices for
18273 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
18274 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
18275 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
18276 architecture option for your machine. Code compiled for lower numbered
18277 architectures runs on higher numbered architectures, but not the
18280 @item -mpa-risc-1-0
18281 @itemx -mpa-risc-1-1
18282 @itemx -mpa-risc-2-0
18283 @opindex mpa-risc-1-0
18284 @opindex mpa-risc-1-1
18285 @opindex mpa-risc-2-0
18286 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
18288 @item -mcaller-copies
18289 @opindex mcaller-copies
18290 The caller copies function arguments passed by hidden reference. This
18291 option should be used with care as it is not compatible with the default
18292 32-bit runtime. However, only aggregates larger than eight bytes are
18293 passed by hidden reference and the option provides better compatibility
18296 @item -mjump-in-delay
18297 @opindex mjump-in-delay
18298 This option is ignored and provided for compatibility purposes only.
18300 @item -mdisable-fpregs
18301 @opindex mdisable-fpregs
18302 Prevent floating-point registers from being used in any manner. This is
18303 necessary for compiling kernels that perform lazy context switching of
18304 floating-point registers. If you use this option and attempt to perform
18305 floating-point operations, the compiler aborts.
18307 @item -mdisable-indexing
18308 @opindex mdisable-indexing
18309 Prevent the compiler from using indexing address modes. This avoids some
18310 rather obscure problems when compiling MIG generated code under MACH@.
18312 @item -mno-space-regs
18313 @opindex mno-space-regs
18314 Generate code that assumes the target has no space registers. This allows
18315 GCC to generate faster indirect calls and use unscaled index address modes.
18317 Such code is suitable for level 0 PA systems and kernels.
18319 @item -mfast-indirect-calls
18320 @opindex mfast-indirect-calls
18321 Generate code that assumes calls never cross space boundaries. This
18322 allows GCC to emit code that performs faster indirect calls.
18324 This option does not work in the presence of shared libraries or nested
18327 @item -mfixed-range=@var{register-range}
18328 @opindex mfixed-range
18329 Generate code treating the given register range as fixed registers.
18330 A fixed register is one that the register allocator cannot use. This is
18331 useful when compiling kernel code. A register range is specified as
18332 two registers separated by a dash. Multiple register ranges can be
18333 specified separated by a comma.
18335 @item -mlong-load-store
18336 @opindex mlong-load-store
18337 Generate 3-instruction load and store sequences as sometimes required by
18338 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
18341 @item -mportable-runtime
18342 @opindex mportable-runtime
18343 Use the portable calling conventions proposed by HP for ELF systems.
18347 Enable the use of assembler directives only GAS understands.
18349 @item -mschedule=@var{cpu-type}
18351 Schedule code according to the constraints for the machine type
18352 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
18353 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
18354 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
18355 proper scheduling option for your machine. The default scheduling is
18359 @opindex mlinker-opt
18360 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
18361 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
18362 linkers in which they give bogus error messages when linking some programs.
18365 @opindex msoft-float
18366 Generate output containing library calls for floating point.
18367 @strong{Warning:} the requisite libraries are not available for all HPPA
18368 targets. Normally the facilities of the machine's usual C compiler are
18369 used, but this cannot be done directly in cross-compilation. You must make
18370 your own arrangements to provide suitable library functions for
18373 @option{-msoft-float} changes the calling convention in the output file;
18374 therefore, it is only useful if you compile @emph{all} of a program with
18375 this option. In particular, you need to compile @file{libgcc.a}, the
18376 library that comes with GCC, with @option{-msoft-float} in order for
18381 Generate the predefine, @code{_SIO}, for server IO@. The default is
18382 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
18383 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
18384 options are available under HP-UX and HI-UX@.
18388 Use options specific to GNU @command{ld}.
18389 This passes @option{-shared} to @command{ld} when
18390 building a shared library. It is the default when GCC is configured,
18391 explicitly or implicitly, with the GNU linker. This option does not
18392 affect which @command{ld} is called; it only changes what parameters
18393 are passed to that @command{ld}.
18394 The @command{ld} that is called is determined by the
18395 @option{--with-ld} configure option, GCC's program search path, and
18396 finally by the user's @env{PATH}. The linker used by GCC can be printed
18397 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
18398 on the 64-bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
18402 Use options specific to HP @command{ld}.
18403 This passes @option{-b} to @command{ld} when building
18404 a shared library and passes @option{+Accept TypeMismatch} to @command{ld} on all
18405 links. It is the default when GCC is configured, explicitly or
18406 implicitly, with the HP linker. This option does not affect
18407 which @command{ld} is called; it only changes what parameters are passed to that
18409 The @command{ld} that is called is determined by the @option{--with-ld}
18410 configure option, GCC's program search path, and finally by the user's
18411 @env{PATH}. The linker used by GCC can be printed using @samp{which
18412 `gcc -print-prog-name=ld`}. This option is only available on the 64-bit
18413 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
18416 @opindex mno-long-calls
18417 Generate code that uses long call sequences. This ensures that a call
18418 is always able to reach linker generated stubs. The default is to generate
18419 long calls only when the distance from the call site to the beginning
18420 of the function or translation unit, as the case may be, exceeds a
18421 predefined limit set by the branch type being used. The limits for
18422 normal calls are 7,600,000 and 240,000 bytes, respectively for the
18423 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
18426 Distances are measured from the beginning of functions when using the
18427 @option{-ffunction-sections} option, or when using the @option{-mgas}
18428 and @option{-mno-portable-runtime} options together under HP-UX with
18431 It is normally not desirable to use this option as it degrades
18432 performance. However, it may be useful in large applications,
18433 particularly when partial linking is used to build the application.
18435 The types of long calls used depends on the capabilities of the
18436 assembler and linker, and the type of code being generated. The
18437 impact on systems that support long absolute calls, and long pic
18438 symbol-difference or pc-relative calls should be relatively small.
18439 However, an indirect call is used on 32-bit ELF systems in pic code
18440 and it is quite long.
18442 @item -munix=@var{unix-std}
18444 Generate compiler predefines and select a startfile for the specified
18445 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
18446 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
18447 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
18448 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
18449 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
18452 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
18453 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
18454 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
18455 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
18456 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
18457 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
18459 It is @emph{important} to note that this option changes the interfaces
18460 for various library routines. It also affects the operational behavior
18461 of the C library. Thus, @emph{extreme} care is needed in using this
18464 Library code that is intended to operate with more than one UNIX
18465 standard must test, set and restore the variable @code{__xpg4_extended_mask}
18466 as appropriate. Most GNU software doesn't provide this capability.
18470 Suppress the generation of link options to search libdld.sl when the
18471 @option{-static} option is specified on HP-UX 10 and later.
18475 The HP-UX implementation of setlocale in libc has a dependency on
18476 libdld.sl. There isn't an archive version of libdld.sl. Thus,
18477 when the @option{-static} option is specified, special link options
18478 are needed to resolve this dependency.
18480 On HP-UX 10 and later, the GCC driver adds the necessary options to
18481 link with libdld.sl when the @option{-static} option is specified.
18482 This causes the resulting binary to be dynamic. On the 64-bit port,
18483 the linkers generate dynamic binaries by default in any case. The
18484 @option{-nolibdld} option can be used to prevent the GCC driver from
18485 adding these link options.
18489 Add support for multithreading with the @dfn{dce thread} library
18490 under HP-UX@. This option sets flags for both the preprocessor and
18494 @node IA-64 Options
18495 @subsection IA-64 Options
18496 @cindex IA-64 Options
18498 These are the @samp{-m} options defined for the Intel IA-64 architecture.
18502 @opindex mbig-endian
18503 Generate code for a big-endian target. This is the default for HP-UX@.
18505 @item -mlittle-endian
18506 @opindex mlittle-endian
18507 Generate code for a little-endian target. This is the default for AIX5
18513 @opindex mno-gnu-as
18514 Generate (or don't) code for the GNU assembler. This is the default.
18515 @c Also, this is the default if the configure option @option{--with-gnu-as}
18521 @opindex mno-gnu-ld
18522 Generate (or don't) code for the GNU linker. This is the default.
18523 @c Also, this is the default if the configure option @option{--with-gnu-ld}
18528 Generate code that does not use a global pointer register. The result
18529 is not position independent code, and violates the IA-64 ABI@.
18531 @item -mvolatile-asm-stop
18532 @itemx -mno-volatile-asm-stop
18533 @opindex mvolatile-asm-stop
18534 @opindex mno-volatile-asm-stop
18535 Generate (or don't) a stop bit immediately before and after volatile asm
18538 @item -mregister-names
18539 @itemx -mno-register-names
18540 @opindex mregister-names
18541 @opindex mno-register-names
18542 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
18543 the stacked registers. This may make assembler output more readable.
18549 Disable (or enable) optimizations that use the small data section. This may
18550 be useful for working around optimizer bugs.
18552 @item -mconstant-gp
18553 @opindex mconstant-gp
18554 Generate code that uses a single constant global pointer value. This is
18555 useful when compiling kernel code.
18559 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
18560 This is useful when compiling firmware code.
18562 @item -minline-float-divide-min-latency
18563 @opindex minline-float-divide-min-latency
18564 Generate code for inline divides of floating-point values
18565 using the minimum latency algorithm.
18567 @item -minline-float-divide-max-throughput
18568 @opindex minline-float-divide-max-throughput
18569 Generate code for inline divides of floating-point values
18570 using the maximum throughput algorithm.
18572 @item -mno-inline-float-divide
18573 @opindex mno-inline-float-divide
18574 Do not generate inline code for divides of floating-point values.
18576 @item -minline-int-divide-min-latency
18577 @opindex minline-int-divide-min-latency
18578 Generate code for inline divides of integer values
18579 using the minimum latency algorithm.
18581 @item -minline-int-divide-max-throughput
18582 @opindex minline-int-divide-max-throughput
18583 Generate code for inline divides of integer values
18584 using the maximum throughput algorithm.
18586 @item -mno-inline-int-divide
18587 @opindex mno-inline-int-divide
18588 Do not generate inline code for divides of integer values.
18590 @item -minline-sqrt-min-latency
18591 @opindex minline-sqrt-min-latency
18592 Generate code for inline square roots
18593 using the minimum latency algorithm.
18595 @item -minline-sqrt-max-throughput
18596 @opindex minline-sqrt-max-throughput
18597 Generate code for inline square roots
18598 using the maximum throughput algorithm.
18600 @item -mno-inline-sqrt
18601 @opindex mno-inline-sqrt
18602 Do not generate inline code for @code{sqrt}.
18605 @itemx -mno-fused-madd
18606 @opindex mfused-madd
18607 @opindex mno-fused-madd
18608 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
18609 instructions. The default is to use these instructions.
18611 @item -mno-dwarf2-asm
18612 @itemx -mdwarf2-asm
18613 @opindex mno-dwarf2-asm
18614 @opindex mdwarf2-asm
18615 Don't (or do) generate assembler code for the DWARF line number debugging
18616 info. This may be useful when not using the GNU assembler.
18618 @item -mearly-stop-bits
18619 @itemx -mno-early-stop-bits
18620 @opindex mearly-stop-bits
18621 @opindex mno-early-stop-bits
18622 Allow stop bits to be placed earlier than immediately preceding the
18623 instruction that triggered the stop bit. This can improve instruction
18624 scheduling, but does not always do so.
18626 @item -mfixed-range=@var{register-range}
18627 @opindex mfixed-range
18628 Generate code treating the given register range as fixed registers.
18629 A fixed register is one that the register allocator cannot use. This is
18630 useful when compiling kernel code. A register range is specified as
18631 two registers separated by a dash. Multiple register ranges can be
18632 specified separated by a comma.
18634 @item -mtls-size=@var{tls-size}
18636 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
18639 @item -mtune=@var{cpu-type}
18641 Tune the instruction scheduling for a particular CPU, Valid values are
18642 @samp{itanium}, @samp{itanium1}, @samp{merced}, @samp{itanium2},
18643 and @samp{mckinley}.
18649 Generate code for a 32-bit or 64-bit environment.
18650 The 32-bit environment sets int, long and pointer to 32 bits.
18651 The 64-bit environment sets int to 32 bits and long and pointer
18652 to 64 bits. These are HP-UX specific flags.
18654 @item -mno-sched-br-data-spec
18655 @itemx -msched-br-data-spec
18656 @opindex mno-sched-br-data-spec
18657 @opindex msched-br-data-spec
18658 (Dis/En)able data speculative scheduling before reload.
18659 This results in generation of @code{ld.a} instructions and
18660 the corresponding check instructions (@code{ld.c} / @code{chk.a}).
18661 The default setting is disabled.
18663 @item -msched-ar-data-spec
18664 @itemx -mno-sched-ar-data-spec
18665 @opindex msched-ar-data-spec
18666 @opindex mno-sched-ar-data-spec
18667 (En/Dis)able data speculative scheduling after reload.
18668 This results in generation of @code{ld.a} instructions and
18669 the corresponding check instructions (@code{ld.c} / @code{chk.a}).
18670 The default setting is enabled.
18672 @item -mno-sched-control-spec
18673 @itemx -msched-control-spec
18674 @opindex mno-sched-control-spec
18675 @opindex msched-control-spec
18676 (Dis/En)able control speculative scheduling. This feature is
18677 available only during region scheduling (i.e.@: before reload).
18678 This results in generation of the @code{ld.s} instructions and
18679 the corresponding check instructions @code{chk.s}.
18680 The default setting is disabled.
18682 @item -msched-br-in-data-spec
18683 @itemx -mno-sched-br-in-data-spec
18684 @opindex msched-br-in-data-spec
18685 @opindex mno-sched-br-in-data-spec
18686 (En/Dis)able speculative scheduling of the instructions that
18687 are dependent on the data speculative loads before reload.
18688 This is effective only with @option{-msched-br-data-spec} enabled.
18689 The default setting is enabled.
18691 @item -msched-ar-in-data-spec
18692 @itemx -mno-sched-ar-in-data-spec
18693 @opindex msched-ar-in-data-spec
18694 @opindex mno-sched-ar-in-data-spec
18695 (En/Dis)able speculative scheduling of the instructions that
18696 are dependent on the data speculative loads after reload.
18697 This is effective only with @option{-msched-ar-data-spec} enabled.
18698 The default setting is enabled.
18700 @item -msched-in-control-spec
18701 @itemx -mno-sched-in-control-spec
18702 @opindex msched-in-control-spec
18703 @opindex mno-sched-in-control-spec
18704 (En/Dis)able speculative scheduling of the instructions that
18705 are dependent on the control speculative loads.
18706 This is effective only with @option{-msched-control-spec} enabled.
18707 The default setting is enabled.
18709 @item -mno-sched-prefer-non-data-spec-insns
18710 @itemx -msched-prefer-non-data-spec-insns
18711 @opindex mno-sched-prefer-non-data-spec-insns
18712 @opindex msched-prefer-non-data-spec-insns
18713 If enabled, data-speculative instructions are chosen for schedule
18714 only if there are no other choices at the moment. This makes
18715 the use of the data speculation much more conservative.
18716 The default setting is disabled.
18718 @item -mno-sched-prefer-non-control-spec-insns
18719 @itemx -msched-prefer-non-control-spec-insns
18720 @opindex mno-sched-prefer-non-control-spec-insns
18721 @opindex msched-prefer-non-control-spec-insns
18722 If enabled, control-speculative instructions are chosen for schedule
18723 only if there are no other choices at the moment. This makes
18724 the use of the control speculation much more conservative.
18725 The default setting is disabled.
18727 @item -mno-sched-count-spec-in-critical-path
18728 @itemx -msched-count-spec-in-critical-path
18729 @opindex mno-sched-count-spec-in-critical-path
18730 @opindex msched-count-spec-in-critical-path
18731 If enabled, speculative dependencies are considered during
18732 computation of the instructions priorities. This makes the use of the
18733 speculation a bit more conservative.
18734 The default setting is disabled.
18736 @item -msched-spec-ldc
18737 @opindex msched-spec-ldc
18738 Use a simple data speculation check. This option is on by default.
18740 @item -msched-control-spec-ldc
18741 @opindex msched-spec-ldc
18742 Use a simple check for control speculation. This option is on by default.
18744 @item -msched-stop-bits-after-every-cycle
18745 @opindex msched-stop-bits-after-every-cycle
18746 Place a stop bit after every cycle when scheduling. This option is on
18749 @item -msched-fp-mem-deps-zero-cost
18750 @opindex msched-fp-mem-deps-zero-cost
18751 Assume that floating-point stores and loads are not likely to cause a conflict
18752 when placed into the same instruction group. This option is disabled by
18755 @item -msel-sched-dont-check-control-spec
18756 @opindex msel-sched-dont-check-control-spec
18757 Generate checks for control speculation in selective scheduling.
18758 This flag is disabled by default.
18760 @item -msched-max-memory-insns=@var{max-insns}
18761 @opindex msched-max-memory-insns
18762 Limit on the number of memory insns per instruction group, giving lower
18763 priority to subsequent memory insns attempting to schedule in the same
18764 instruction group. Frequently useful to prevent cache bank conflicts.
18765 The default value is 1.
18767 @item -msched-max-memory-insns-hard-limit
18768 @opindex msched-max-memory-insns-hard-limit
18769 Makes the limit specified by @option{msched-max-memory-insns} a hard limit,
18770 disallowing more than that number in an instruction group.
18771 Otherwise, the limit is ``soft'', meaning that non-memory operations
18772 are preferred when the limit is reached, but memory operations may still
18778 @subsection LM32 Options
18779 @cindex LM32 options
18781 These @option{-m} options are defined for the LatticeMico32 architecture:
18784 @item -mbarrel-shift-enabled
18785 @opindex mbarrel-shift-enabled
18786 Enable barrel-shift instructions.
18788 @item -mdivide-enabled
18789 @opindex mdivide-enabled
18790 Enable divide and modulus instructions.
18792 @item -mmultiply-enabled
18793 @opindex multiply-enabled
18794 Enable multiply instructions.
18796 @item -msign-extend-enabled
18797 @opindex msign-extend-enabled
18798 Enable sign extend instructions.
18800 @item -muser-enabled
18801 @opindex muser-enabled
18802 Enable user-defined instructions.
18807 @subsection M32C Options
18808 @cindex M32C options
18811 @item -mcpu=@var{name}
18813 Select the CPU for which code is generated. @var{name} may be one of
18814 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
18815 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
18816 the M32C/80 series.
18820 Specifies that the program will be run on the simulator. This causes
18821 an alternate runtime library to be linked in which supports, for
18822 example, file I/O@. You must not use this option when generating
18823 programs that will run on real hardware; you must provide your own
18824 runtime library for whatever I/O functions are needed.
18826 @item -memregs=@var{number}
18828 Specifies the number of memory-based pseudo-registers GCC uses
18829 during code generation. These pseudo-registers are used like real
18830 registers, so there is a tradeoff between GCC's ability to fit the
18831 code into available registers, and the performance penalty of using
18832 memory instead of registers. Note that all modules in a program must
18833 be compiled with the same value for this option. Because of that, you
18834 must not use this option with GCC's default runtime libraries.
18838 @node M32R/D Options
18839 @subsection M32R/D Options
18840 @cindex M32R/D options
18842 These @option{-m} options are defined for Renesas M32R/D architectures:
18847 Generate code for the M32R/2@.
18851 Generate code for the M32R/X@.
18855 Generate code for the M32R@. This is the default.
18857 @item -mmodel=small
18858 @opindex mmodel=small
18859 Assume all objects live in the lower 16MB of memory (so that their addresses
18860 can be loaded with the @code{ld24} instruction), and assume all subroutines
18861 are reachable with the @code{bl} instruction.
18862 This is the default.
18864 The addressability of a particular object can be set with the
18865 @code{model} attribute.
18867 @item -mmodel=medium
18868 @opindex mmodel=medium
18869 Assume objects may be anywhere in the 32-bit address space (the compiler
18870 generates @code{seth/add3} instructions to load their addresses), and
18871 assume all subroutines are reachable with the @code{bl} instruction.
18873 @item -mmodel=large
18874 @opindex mmodel=large
18875 Assume objects may be anywhere in the 32-bit address space (the compiler
18876 generates @code{seth/add3} instructions to load their addresses), and
18877 assume subroutines may not be reachable with the @code{bl} instruction
18878 (the compiler generates the much slower @code{seth/add3/jl}
18879 instruction sequence).
18882 @opindex msdata=none
18883 Disable use of the small data area. Variables are put into
18884 one of @code{.data}, @code{.bss}, or @code{.rodata} (unless the
18885 @code{section} attribute has been specified).
18886 This is the default.
18888 The small data area consists of sections @code{.sdata} and @code{.sbss}.
18889 Objects may be explicitly put in the small data area with the
18890 @code{section} attribute using one of these sections.
18892 @item -msdata=sdata
18893 @opindex msdata=sdata
18894 Put small global and static data in the small data area, but do not
18895 generate special code to reference them.
18898 @opindex msdata=use
18899 Put small global and static data in the small data area, and generate
18900 special instructions to reference them.
18904 @cindex smaller data references
18905 Put global and static objects less than or equal to @var{num} bytes
18906 into the small data or BSS sections instead of the normal data or BSS
18907 sections. The default value of @var{num} is 8.
18908 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
18909 for this option to have any effect.
18911 All modules should be compiled with the same @option{-G @var{num}} value.
18912 Compiling with different values of @var{num} may or may not work; if it
18913 doesn't the linker gives an error message---incorrect code is not
18918 Makes the M32R-specific code in the compiler display some statistics
18919 that might help in debugging programs.
18921 @item -malign-loops
18922 @opindex malign-loops
18923 Align all loops to a 32-byte boundary.
18925 @item -mno-align-loops
18926 @opindex mno-align-loops
18927 Do not enforce a 32-byte alignment for loops. This is the default.
18929 @item -missue-rate=@var{number}
18930 @opindex missue-rate=@var{number}
18931 Issue @var{number} instructions per cycle. @var{number} can only be 1
18934 @item -mbranch-cost=@var{number}
18935 @opindex mbranch-cost=@var{number}
18936 @var{number} can only be 1 or 2. If it is 1 then branches are
18937 preferred over conditional code, if it is 2, then the opposite applies.
18939 @item -mflush-trap=@var{number}
18940 @opindex mflush-trap=@var{number}
18941 Specifies the trap number to use to flush the cache. The default is
18942 12. Valid numbers are between 0 and 15 inclusive.
18944 @item -mno-flush-trap
18945 @opindex mno-flush-trap
18946 Specifies that the cache cannot be flushed by using a trap.
18948 @item -mflush-func=@var{name}
18949 @opindex mflush-func=@var{name}
18950 Specifies the name of the operating system function to call to flush
18951 the cache. The default is @samp{_flush_cache}, but a function call
18952 is only used if a trap is not available.
18954 @item -mno-flush-func
18955 @opindex mno-flush-func
18956 Indicates that there is no OS function for flushing the cache.
18960 @node M680x0 Options
18961 @subsection M680x0 Options
18962 @cindex M680x0 options
18964 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
18965 The default settings depend on which architecture was selected when
18966 the compiler was configured; the defaults for the most common choices
18970 @item -march=@var{arch}
18972 Generate code for a specific M680x0 or ColdFire instruction set
18973 architecture. Permissible values of @var{arch} for M680x0
18974 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
18975 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
18976 architectures are selected according to Freescale's ISA classification
18977 and the permissible values are: @samp{isaa}, @samp{isaaplus},
18978 @samp{isab} and @samp{isac}.
18980 GCC defines a macro @code{__mcf@var{arch}__} whenever it is generating
18981 code for a ColdFire target. The @var{arch} in this macro is one of the
18982 @option{-march} arguments given above.
18984 When used together, @option{-march} and @option{-mtune} select code
18985 that runs on a family of similar processors but that is optimized
18986 for a particular microarchitecture.
18988 @item -mcpu=@var{cpu}
18990 Generate code for a specific M680x0 or ColdFire processor.
18991 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
18992 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
18993 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
18994 below, which also classifies the CPUs into families:
18996 @multitable @columnfractions 0.20 0.80
18997 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
18998 @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}
18999 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
19000 @item @samp{5206e} @tab @samp{5206e}
19001 @item @samp{5208} @tab @samp{5207} @samp{5208}
19002 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
19003 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
19004 @item @samp{5216} @tab @samp{5214} @samp{5216}
19005 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
19006 @item @samp{5225} @tab @samp{5224} @samp{5225}
19007 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
19008 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
19009 @item @samp{5249} @tab @samp{5249}
19010 @item @samp{5250} @tab @samp{5250}
19011 @item @samp{5271} @tab @samp{5270} @samp{5271}
19012 @item @samp{5272} @tab @samp{5272}
19013 @item @samp{5275} @tab @samp{5274} @samp{5275}
19014 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
19015 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
19016 @item @samp{5307} @tab @samp{5307}
19017 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
19018 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
19019 @item @samp{5407} @tab @samp{5407}
19020 @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}
19023 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
19024 @var{arch} is compatible with @var{cpu}. Other combinations of
19025 @option{-mcpu} and @option{-march} are rejected.
19027 GCC defines the macro @code{__mcf_cpu_@var{cpu}} when ColdFire target
19028 @var{cpu} is selected. It also defines @code{__mcf_family_@var{family}},
19029 where the value of @var{family} is given by the table above.
19031 @item -mtune=@var{tune}
19033 Tune the code for a particular microarchitecture within the
19034 constraints set by @option{-march} and @option{-mcpu}.
19035 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
19036 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
19037 and @samp{cpu32}. The ColdFire microarchitectures
19038 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
19040 You can also use @option{-mtune=68020-40} for code that needs
19041 to run relatively well on 68020, 68030 and 68040 targets.
19042 @option{-mtune=68020-60} is similar but includes 68060 targets
19043 as well. These two options select the same tuning decisions as
19044 @option{-m68020-40} and @option{-m68020-60} respectively.
19046 GCC defines the macros @code{__mc@var{arch}} and @code{__mc@var{arch}__}
19047 when tuning for 680x0 architecture @var{arch}. It also defines
19048 @code{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
19049 option is used. If GCC is tuning for a range of architectures,
19050 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
19051 it defines the macros for every architecture in the range.
19053 GCC also defines the macro @code{__m@var{uarch}__} when tuning for
19054 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
19055 of the arguments given above.
19061 Generate output for a 68000. This is the default
19062 when the compiler is configured for 68000-based systems.
19063 It is equivalent to @option{-march=68000}.
19065 Use this option for microcontrollers with a 68000 or EC000 core,
19066 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
19070 Generate output for a 68010. This is the default
19071 when the compiler is configured for 68010-based systems.
19072 It is equivalent to @option{-march=68010}.
19078 Generate output for a 68020. This is the default
19079 when the compiler is configured for 68020-based systems.
19080 It is equivalent to @option{-march=68020}.
19084 Generate output for a 68030. This is the default when the compiler is
19085 configured for 68030-based systems. It is equivalent to
19086 @option{-march=68030}.
19090 Generate output for a 68040. This is the default when the compiler is
19091 configured for 68040-based systems. It is equivalent to
19092 @option{-march=68040}.
19094 This option inhibits the use of 68881/68882 instructions that have to be
19095 emulated by software on the 68040. Use this option if your 68040 does not
19096 have code to emulate those instructions.
19100 Generate output for a 68060. This is the default when the compiler is
19101 configured for 68060-based systems. It is equivalent to
19102 @option{-march=68060}.
19104 This option inhibits the use of 68020 and 68881/68882 instructions that
19105 have to be emulated by software on the 68060. Use this option if your 68060
19106 does not have code to emulate those instructions.
19110 Generate output for a CPU32. This is the default
19111 when the compiler is configured for CPU32-based systems.
19112 It is equivalent to @option{-march=cpu32}.
19114 Use this option for microcontrollers with a
19115 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
19116 68336, 68340, 68341, 68349 and 68360.
19120 Generate output for a 520X ColdFire CPU@. This is the default
19121 when the compiler is configured for 520X-based systems.
19122 It is equivalent to @option{-mcpu=5206}, and is now deprecated
19123 in favor of that option.
19125 Use this option for microcontroller with a 5200 core, including
19126 the MCF5202, MCF5203, MCF5204 and MCF5206.
19130 Generate output for a 5206e ColdFire CPU@. The option is now
19131 deprecated in favor of the equivalent @option{-mcpu=5206e}.
19135 Generate output for a member of the ColdFire 528X family.
19136 The option is now deprecated in favor of the equivalent
19137 @option{-mcpu=528x}.
19141 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
19142 in favor of the equivalent @option{-mcpu=5307}.
19146 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
19147 in favor of the equivalent @option{-mcpu=5407}.
19151 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
19152 This includes use of hardware floating-point instructions.
19153 The option is equivalent to @option{-mcpu=547x}, and is now
19154 deprecated in favor of that option.
19158 Generate output for a 68040, without using any of the new instructions.
19159 This results in code that can run relatively efficiently on either a
19160 68020/68881 or a 68030 or a 68040. The generated code does use the
19161 68881 instructions that are emulated on the 68040.
19163 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
19167 Generate output for a 68060, without using any of the new instructions.
19168 This results in code that can run relatively efficiently on either a
19169 68020/68881 or a 68030 or a 68040. The generated code does use the
19170 68881 instructions that are emulated on the 68060.
19172 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
19176 @opindex mhard-float
19178 Generate floating-point instructions. This is the default for 68020
19179 and above, and for ColdFire devices that have an FPU@. It defines the
19180 macro @code{__HAVE_68881__} on M680x0 targets and @code{__mcffpu__}
19181 on ColdFire targets.
19184 @opindex msoft-float
19185 Do not generate floating-point instructions; use library calls instead.
19186 This is the default for 68000, 68010, and 68832 targets. It is also
19187 the default for ColdFire devices that have no FPU.
19193 Generate (do not generate) ColdFire hardware divide and remainder
19194 instructions. If @option{-march} is used without @option{-mcpu},
19195 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
19196 architectures. Otherwise, the default is taken from the target CPU
19197 (either the default CPU, or the one specified by @option{-mcpu}). For
19198 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
19199 @option{-mcpu=5206e}.
19201 GCC defines the macro @code{__mcfhwdiv__} when this option is enabled.
19205 Consider type @code{int} to be 16 bits wide, like @code{short int}.
19206 Additionally, parameters passed on the stack are also aligned to a
19207 16-bit boundary even on targets whose API mandates promotion to 32-bit.
19211 Do not consider type @code{int} to be 16 bits wide. This is the default.
19214 @itemx -mno-bitfield
19215 @opindex mnobitfield
19216 @opindex mno-bitfield
19217 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
19218 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
19222 Do use the bit-field instructions. The @option{-m68020} option implies
19223 @option{-mbitfield}. This is the default if you use a configuration
19224 designed for a 68020.
19228 Use a different function-calling convention, in which functions
19229 that take a fixed number of arguments return with the @code{rtd}
19230 instruction, which pops their arguments while returning. This
19231 saves one instruction in the caller since there is no need to pop
19232 the arguments there.
19234 This calling convention is incompatible with the one normally
19235 used on Unix, so you cannot use it if you need to call libraries
19236 compiled with the Unix compiler.
19238 Also, you must provide function prototypes for all functions that
19239 take variable numbers of arguments (including @code{printf});
19240 otherwise incorrect code is generated for calls to those
19243 In addition, seriously incorrect code results if you call a
19244 function with too many arguments. (Normally, extra arguments are
19245 harmlessly ignored.)
19247 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
19248 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
19252 Do not use the calling conventions selected by @option{-mrtd}.
19253 This is the default.
19256 @itemx -mno-align-int
19257 @opindex malign-int
19258 @opindex mno-align-int
19259 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
19260 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
19261 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
19262 Aligning variables on 32-bit boundaries produces code that runs somewhat
19263 faster on processors with 32-bit busses at the expense of more memory.
19265 @strong{Warning:} if you use the @option{-malign-int} switch, GCC
19266 aligns structures containing the above types differently than
19267 most published application binary interface specifications for the m68k.
19271 Use the pc-relative addressing mode of the 68000 directly, instead of
19272 using a global offset table. At present, this option implies @option{-fpic},
19273 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
19274 not presently supported with @option{-mpcrel}, though this could be supported for
19275 68020 and higher processors.
19277 @item -mno-strict-align
19278 @itemx -mstrict-align
19279 @opindex mno-strict-align
19280 @opindex mstrict-align
19281 Do not (do) assume that unaligned memory references are handled by
19285 Generate code that allows the data segment to be located in a different
19286 area of memory from the text segment. This allows for execute-in-place in
19287 an environment without virtual memory management. This option implies
19290 @item -mno-sep-data
19291 Generate code that assumes that the data segment follows the text segment.
19292 This is the default.
19294 @item -mid-shared-library
19295 Generate code that supports shared libraries via the library ID method.
19296 This allows for execute-in-place and shared libraries in an environment
19297 without virtual memory management. This option implies @option{-fPIC}.
19299 @item -mno-id-shared-library
19300 Generate code that doesn't assume ID-based shared libraries are being used.
19301 This is the default.
19303 @item -mshared-library-id=n
19304 Specifies the identification number of the ID-based shared library being
19305 compiled. Specifying a value of 0 generates more compact code; specifying
19306 other values forces the allocation of that number to the current
19307 library, but is no more space- or time-efficient than omitting this option.
19313 When generating position-independent code for ColdFire, generate code
19314 that works if the GOT has more than 8192 entries. This code is
19315 larger and slower than code generated without this option. On M680x0
19316 processors, this option is not needed; @option{-fPIC} suffices.
19318 GCC normally uses a single instruction to load values from the GOT@.
19319 While this is relatively efficient, it only works if the GOT
19320 is smaller than about 64k. Anything larger causes the linker
19321 to report an error such as:
19323 @cindex relocation truncated to fit (ColdFire)
19325 relocation truncated to fit: R_68K_GOT16O foobar
19328 If this happens, you should recompile your code with @option{-mxgot}.
19329 It should then work with very large GOTs. However, code generated with
19330 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
19331 the value of a global symbol.
19333 Note that some linkers, including newer versions of the GNU linker,
19334 can create multiple GOTs and sort GOT entries. If you have such a linker,
19335 you should only need to use @option{-mxgot} when compiling a single
19336 object file that accesses more than 8192 GOT entries. Very few do.
19338 These options have no effect unless GCC is generating
19339 position-independent code.
19341 @item -mlong-jump-table-offsets
19342 @opindex mlong-jump-table-offsets
19343 Use 32-bit offsets in @code{switch} tables. The default is to use
19348 @node MCore Options
19349 @subsection MCore Options
19350 @cindex MCore options
19352 These are the @samp{-m} options defined for the Motorola M*Core
19358 @itemx -mno-hardlit
19360 @opindex mno-hardlit
19361 Inline constants into the code stream if it can be done in two
19362 instructions or less.
19368 Use the divide instruction. (Enabled by default).
19370 @item -mrelax-immediate
19371 @itemx -mno-relax-immediate
19372 @opindex mrelax-immediate
19373 @opindex mno-relax-immediate
19374 Allow arbitrary-sized immediates in bit operations.
19376 @item -mwide-bitfields
19377 @itemx -mno-wide-bitfields
19378 @opindex mwide-bitfields
19379 @opindex mno-wide-bitfields
19380 Always treat bit-fields as @code{int}-sized.
19382 @item -m4byte-functions
19383 @itemx -mno-4byte-functions
19384 @opindex m4byte-functions
19385 @opindex mno-4byte-functions
19386 Force all functions to be aligned to a 4-byte boundary.
19388 @item -mcallgraph-data
19389 @itemx -mno-callgraph-data
19390 @opindex mcallgraph-data
19391 @opindex mno-callgraph-data
19392 Emit callgraph information.
19395 @itemx -mno-slow-bytes
19396 @opindex mslow-bytes
19397 @opindex mno-slow-bytes
19398 Prefer word access when reading byte quantities.
19400 @item -mlittle-endian
19401 @itemx -mbig-endian
19402 @opindex mlittle-endian
19403 @opindex mbig-endian
19404 Generate code for a little-endian target.
19410 Generate code for the 210 processor.
19414 Assume that runtime support has been provided and so omit the
19415 simulator library (@file{libsim.a)} from the linker command line.
19417 @item -mstack-increment=@var{size}
19418 @opindex mstack-increment
19419 Set the maximum amount for a single stack increment operation. Large
19420 values can increase the speed of programs that contain functions
19421 that need a large amount of stack space, but they can also trigger a
19422 segmentation fault if the stack is extended too much. The default
19428 @subsection MeP Options
19429 @cindex MeP options
19435 Enables the @code{abs} instruction, which is the absolute difference
19436 between two registers.
19440 Enables all the optional instructions---average, multiply, divide, bit
19441 operations, leading zero, absolute difference, min/max, clip, and
19447 Enables the @code{ave} instruction, which computes the average of two
19450 @item -mbased=@var{n}
19452 Variables of size @var{n} bytes or smaller are placed in the
19453 @code{.based} section by default. Based variables use the @code{$tp}
19454 register as a base register, and there is a 128-byte limit to the
19455 @code{.based} section.
19459 Enables the bit operation instructions---bit test (@code{btstm}), set
19460 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
19461 test-and-set (@code{tas}).
19463 @item -mc=@var{name}
19465 Selects which section constant data is placed in. @var{name} may
19466 be @samp{tiny}, @samp{near}, or @samp{far}.
19470 Enables the @code{clip} instruction. Note that @option{-mclip} is not
19471 useful unless you also provide @option{-mminmax}.
19473 @item -mconfig=@var{name}
19475 Selects one of the built-in core configurations. Each MeP chip has
19476 one or more modules in it; each module has a core CPU and a variety of
19477 coprocessors, optional instructions, and peripherals. The
19478 @code{MeP-Integrator} tool, not part of GCC, provides these
19479 configurations through this option; using this option is the same as
19480 using all the corresponding command-line options. The default
19481 configuration is @samp{default}.
19485 Enables the coprocessor instructions. By default, this is a 32-bit
19486 coprocessor. Note that the coprocessor is normally enabled via the
19487 @option{-mconfig=} option.
19491 Enables the 32-bit coprocessor's instructions.
19495 Enables the 64-bit coprocessor's instructions.
19499 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
19503 Causes constant variables to be placed in the @code{.near} section.
19507 Enables the @code{div} and @code{divu} instructions.
19511 Generate big-endian code.
19515 Generate little-endian code.
19517 @item -mio-volatile
19518 @opindex mio-volatile
19519 Tells the compiler that any variable marked with the @code{io}
19520 attribute is to be considered volatile.
19524 Causes variables to be assigned to the @code{.far} section by default.
19528 Enables the @code{leadz} (leading zero) instruction.
19532 Causes variables to be assigned to the @code{.near} section by default.
19536 Enables the @code{min} and @code{max} instructions.
19540 Enables the multiplication and multiply-accumulate instructions.
19544 Disables all the optional instructions enabled by @option{-mall-opts}.
19548 Enables the @code{repeat} and @code{erepeat} instructions, used for
19549 low-overhead looping.
19553 Causes all variables to default to the @code{.tiny} section. Note
19554 that there is a 65536-byte limit to this section. Accesses to these
19555 variables use the @code{%gp} base register.
19559 Enables the saturation instructions. Note that the compiler does not
19560 currently generate these itself, but this option is included for
19561 compatibility with other tools, like @code{as}.
19565 Link the SDRAM-based runtime instead of the default ROM-based runtime.
19569 Link the simulator run-time libraries.
19573 Link the simulator runtime libraries, excluding built-in support
19574 for reset and exception vectors and tables.
19578 Causes all functions to default to the @code{.far} section. Without
19579 this option, functions default to the @code{.near} section.
19581 @item -mtiny=@var{n}
19583 Variables that are @var{n} bytes or smaller are allocated to the
19584 @code{.tiny} section. These variables use the @code{$gp} base
19585 register. The default for this option is 4, but note that there's a
19586 65536-byte limit to the @code{.tiny} section.
19590 @node MicroBlaze Options
19591 @subsection MicroBlaze Options
19592 @cindex MicroBlaze Options
19597 @opindex msoft-float
19598 Use software emulation for floating point (default).
19601 @opindex mhard-float
19602 Use hardware floating-point instructions.
19606 Do not optimize block moves, use @code{memcpy}.
19608 @item -mno-clearbss
19609 @opindex mno-clearbss
19610 This option is deprecated. Use @option{-fno-zero-initialized-in-bss} instead.
19612 @item -mcpu=@var{cpu-type}
19614 Use features of, and schedule code for, the given CPU.
19615 Supported values are in the format @samp{v@var{X}.@var{YY}.@var{Z}},
19616 where @var{X} is a major version, @var{YY} is the minor version, and
19617 @var{Z} is compatibility code. Example values are @samp{v3.00.a},
19618 @samp{v4.00.b}, @samp{v5.00.a}, @samp{v5.00.b}, @samp{v5.00.b}, @samp{v6.00.a}.
19620 @item -mxl-soft-mul
19621 @opindex mxl-soft-mul
19622 Use software multiply emulation (default).
19624 @item -mxl-soft-div
19625 @opindex mxl-soft-div
19626 Use software emulation for divides (default).
19628 @item -mxl-barrel-shift
19629 @opindex mxl-barrel-shift
19630 Use the hardware barrel shifter.
19632 @item -mxl-pattern-compare
19633 @opindex mxl-pattern-compare
19634 Use pattern compare instructions.
19636 @item -msmall-divides
19637 @opindex msmall-divides
19638 Use table lookup optimization for small signed integer divisions.
19640 @item -mxl-stack-check
19641 @opindex mxl-stack-check
19642 This option is deprecated. Use @option{-fstack-check} instead.
19645 @opindex mxl-gp-opt
19646 Use GP-relative @code{.sdata}/@code{.sbss} sections.
19648 @item -mxl-multiply-high
19649 @opindex mxl-multiply-high
19650 Use multiply high instructions for high part of 32x32 multiply.
19652 @item -mxl-float-convert
19653 @opindex mxl-float-convert
19654 Use hardware floating-point conversion instructions.
19656 @item -mxl-float-sqrt
19657 @opindex mxl-float-sqrt
19658 Use hardware floating-point square root instruction.
19661 @opindex mbig-endian
19662 Generate code for a big-endian target.
19664 @item -mlittle-endian
19665 @opindex mlittle-endian
19666 Generate code for a little-endian target.
19669 @opindex mxl-reorder
19670 Use reorder instructions (swap and byte reversed load/store).
19672 @item -mxl-mode-@var{app-model}
19673 Select application model @var{app-model}. Valid models are
19676 normal executable (default), uses startup code @file{crt0.o}.
19679 for use with Xilinx Microprocessor Debugger (XMD) based
19680 software intrusive debug agent called xmdstub. This uses startup file
19681 @file{crt1.o} and sets the start address of the program to 0x800.
19684 for applications that are loaded using a bootloader.
19685 This model uses startup file @file{crt2.o} which does not contain a processor
19686 reset vector handler. This is suitable for transferring control on a
19687 processor reset to the bootloader rather than the application.
19690 for applications that do not require any of the
19691 MicroBlaze vectors. This option may be useful for applications running
19692 within a monitoring application. This model uses @file{crt3.o} as a startup file.
19695 Option @option{-xl-mode-@var{app-model}} is a deprecated alias for
19696 @option{-mxl-mode-@var{app-model}}.
19701 @subsection MIPS Options
19702 @cindex MIPS options
19708 Generate big-endian code.
19712 Generate little-endian code. This is the default for @samp{mips*el-*-*}
19715 @item -march=@var{arch}
19717 Generate code that runs on @var{arch}, which can be the name of a
19718 generic MIPS ISA, or the name of a particular processor.
19720 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
19721 @samp{mips32}, @samp{mips32r2}, @samp{mips32r3}, @samp{mips32r5},
19722 @samp{mips32r6}, @samp{mips64}, @samp{mips64r2}, @samp{mips64r3},
19723 @samp{mips64r5} and @samp{mips64r6}.
19724 The processor names are:
19725 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
19726 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
19727 @samp{5kc}, @samp{5kf},
19729 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
19730 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
19731 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1}, @samp{34kn},
19732 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
19733 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
19736 @samp{loongson2e}, @samp{loongson2f}, @samp{loongson3a},
19738 @samp{m14k}, @samp{m14kc}, @samp{m14ke}, @samp{m14kec},
19739 @samp{m5100}, @samp{m5101},
19740 @samp{octeon}, @samp{octeon+}, @samp{octeon2}, @samp{octeon3},
19743 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
19744 @samp{r4600}, @samp{r4650}, @samp{r4700}, @samp{r6000}, @samp{r8000},
19745 @samp{rm7000}, @samp{rm9000},
19746 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
19749 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
19750 @samp{vr5000}, @samp{vr5400}, @samp{vr5500},
19751 @samp{xlr} and @samp{xlp}.
19752 The special value @samp{from-abi} selects the
19753 most compatible architecture for the selected ABI (that is,
19754 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
19756 The native Linux/GNU toolchain also supports the value @samp{native},
19757 which selects the best architecture option for the host processor.
19758 @option{-march=native} has no effect if GCC does not recognize
19761 In processor names, a final @samp{000} can be abbreviated as @samp{k}
19762 (for example, @option{-march=r2k}). Prefixes are optional, and
19763 @samp{vr} may be written @samp{r}.
19765 Names of the form @samp{@var{n}f2_1} refer to processors with
19766 FPUs clocked at half the rate of the core, names of the form
19767 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
19768 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
19769 processors with FPUs clocked a ratio of 3:2 with respect to the core.
19770 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
19771 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
19772 accepted as synonyms for @samp{@var{n}f1_1}.
19774 GCC defines two macros based on the value of this option. The first
19775 is @code{_MIPS_ARCH}, which gives the name of target architecture, as
19776 a string. The second has the form @code{_MIPS_ARCH_@var{foo}},
19777 where @var{foo} is the capitalized value of @code{_MIPS_ARCH}@.
19778 For example, @option{-march=r2000} sets @code{_MIPS_ARCH}
19779 to @code{"r2000"} and defines the macro @code{_MIPS_ARCH_R2000}.
19781 Note that the @code{_MIPS_ARCH} macro uses the processor names given
19782 above. In other words, it has the full prefix and does not
19783 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
19784 the macro names the resolved architecture (either @code{"mips1"} or
19785 @code{"mips3"}). It names the default architecture when no
19786 @option{-march} option is given.
19788 @item -mtune=@var{arch}
19790 Optimize for @var{arch}. Among other things, this option controls
19791 the way instructions are scheduled, and the perceived cost of arithmetic
19792 operations. The list of @var{arch} values is the same as for
19795 When this option is not used, GCC optimizes for the processor
19796 specified by @option{-march}. By using @option{-march} and
19797 @option{-mtune} together, it is possible to generate code that
19798 runs on a family of processors, but optimize the code for one
19799 particular member of that family.
19801 @option{-mtune} defines the macros @code{_MIPS_TUNE} and
19802 @code{_MIPS_TUNE_@var{foo}}, which work in the same way as the
19803 @option{-march} ones described above.
19807 Equivalent to @option{-march=mips1}.
19811 Equivalent to @option{-march=mips2}.
19815 Equivalent to @option{-march=mips3}.
19819 Equivalent to @option{-march=mips4}.
19823 Equivalent to @option{-march=mips32}.
19827 Equivalent to @option{-march=mips32r3}.
19831 Equivalent to @option{-march=mips32r5}.
19835 Equivalent to @option{-march=mips32r6}.
19839 Equivalent to @option{-march=mips64}.
19843 Equivalent to @option{-march=mips64r2}.
19847 Equivalent to @option{-march=mips64r3}.
19851 Equivalent to @option{-march=mips64r5}.
19855 Equivalent to @option{-march=mips64r6}.
19860 @opindex mno-mips16
19861 Generate (do not generate) MIPS16 code. If GCC is targeting a
19862 MIPS32 or MIPS64 architecture, it makes use of the MIPS16e ASE@.
19864 MIPS16 code generation can also be controlled on a per-function basis
19865 by means of @code{mips16} and @code{nomips16} attributes.
19866 @xref{Function Attributes}, for more information.
19868 @item -mflip-mips16
19869 @opindex mflip-mips16
19870 Generate MIPS16 code on alternating functions. This option is provided
19871 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
19872 not intended for ordinary use in compiling user code.
19874 @item -minterlink-compressed
19875 @item -mno-interlink-compressed
19876 @opindex minterlink-compressed
19877 @opindex mno-interlink-compressed
19878 Require (do not require) that code using the standard (uncompressed) MIPS ISA
19879 be link-compatible with MIPS16 and microMIPS code, and vice versa.
19881 For example, code using the standard ISA encoding cannot jump directly
19882 to MIPS16 or microMIPS code; it must either use a call or an indirect jump.
19883 @option{-minterlink-compressed} therefore disables direct jumps unless GCC
19884 knows that the target of the jump is not compressed.
19886 @item -minterlink-mips16
19887 @itemx -mno-interlink-mips16
19888 @opindex minterlink-mips16
19889 @opindex mno-interlink-mips16
19890 Aliases of @option{-minterlink-compressed} and
19891 @option{-mno-interlink-compressed}. These options predate the microMIPS ASE
19892 and are retained for backwards compatibility.
19904 Generate code for the given ABI@.
19906 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
19907 generates 64-bit code when you select a 64-bit architecture, but you
19908 can use @option{-mgp32} to get 32-bit code instead.
19910 For information about the O64 ABI, see
19911 @uref{http://gcc.gnu.org/@/projects/@/mipso64-abi.html}.
19913 GCC supports a variant of the o32 ABI in which floating-point registers
19914 are 64 rather than 32 bits wide. You can select this combination with
19915 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @code{mthc1}
19916 and @code{mfhc1} instructions and is therefore only supported for
19917 MIPS32R2, MIPS32R3 and MIPS32R5 processors.
19919 The register assignments for arguments and return values remain the
19920 same, but each scalar value is passed in a single 64-bit register
19921 rather than a pair of 32-bit registers. For example, scalar
19922 floating-point values are returned in @samp{$f0} only, not a
19923 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
19924 remains the same in that the even-numbered double-precision registers
19927 Two additional variants of the o32 ABI are supported to enable
19928 a transition from 32-bit to 64-bit registers. These are FPXX
19929 (@option{-mfpxx}) and FP64A (@option{-mfp64} @option{-mno-odd-spreg}).
19930 The FPXX extension mandates that all code must execute correctly
19931 when run using 32-bit or 64-bit registers. The code can be interlinked
19932 with either FP32 or FP64, but not both.
19933 The FP64A extension is similar to the FP64 extension but forbids the
19934 use of odd-numbered single-precision registers. This can be used
19935 in conjunction with the @code{FRE} mode of FPUs in MIPS32R5
19936 processors and allows both FP32 and FP64A code to interlink and
19937 run in the same process without changing FPU modes.
19940 @itemx -mno-abicalls
19942 @opindex mno-abicalls
19943 Generate (do not generate) code that is suitable for SVR4-style
19944 dynamic objects. @option{-mabicalls} is the default for SVR4-based
19949 Generate (do not generate) code that is fully position-independent,
19950 and that can therefore be linked into shared libraries. This option
19951 only affects @option{-mabicalls}.
19953 All @option{-mabicalls} code has traditionally been position-independent,
19954 regardless of options like @option{-fPIC} and @option{-fpic}. However,
19955 as an extension, the GNU toolchain allows executables to use absolute
19956 accesses for locally-binding symbols. It can also use shorter GP
19957 initialization sequences and generate direct calls to locally-defined
19958 functions. This mode is selected by @option{-mno-shared}.
19960 @option{-mno-shared} depends on binutils 2.16 or higher and generates
19961 objects that can only be linked by the GNU linker. However, the option
19962 does not affect the ABI of the final executable; it only affects the ABI
19963 of relocatable objects. Using @option{-mno-shared} generally makes
19964 executables both smaller and quicker.
19966 @option{-mshared} is the default.
19972 Assume (do not assume) that the static and dynamic linkers
19973 support PLTs and copy relocations. This option only affects
19974 @option{-mno-shared -mabicalls}. For the n64 ABI, this option
19975 has no effect without @option{-msym32}.
19977 You can make @option{-mplt} the default by configuring
19978 GCC with @option{--with-mips-plt}. The default is
19979 @option{-mno-plt} otherwise.
19985 Lift (do not lift) the usual restrictions on the size of the global
19988 GCC normally uses a single instruction to load values from the GOT@.
19989 While this is relatively efficient, it only works if the GOT
19990 is smaller than about 64k. Anything larger causes the linker
19991 to report an error such as:
19993 @cindex relocation truncated to fit (MIPS)
19995 relocation truncated to fit: R_MIPS_GOT16 foobar
19998 If this happens, you should recompile your code with @option{-mxgot}.
19999 This works with very large GOTs, although the code is also
20000 less efficient, since it takes three instructions to fetch the
20001 value of a global symbol.
20003 Note that some linkers can create multiple GOTs. If you have such a
20004 linker, you should only need to use @option{-mxgot} when a single object
20005 file accesses more than 64k's worth of GOT entries. Very few do.
20007 These options have no effect unless GCC is generating position
20012 Assume that general-purpose registers are 32 bits wide.
20016 Assume that general-purpose registers are 64 bits wide.
20020 Assume that floating-point registers are 32 bits wide.
20024 Assume that floating-point registers are 64 bits wide.
20028 Do not assume the width of floating-point registers.
20031 @opindex mhard-float
20032 Use floating-point coprocessor instructions.
20035 @opindex msoft-float
20036 Do not use floating-point coprocessor instructions. Implement
20037 floating-point calculations using library calls instead.
20041 Equivalent to @option{-msoft-float}, but additionally asserts that the
20042 program being compiled does not perform any floating-point operations.
20043 This option is presently supported only by some bare-metal MIPS
20044 configurations, where it may select a special set of libraries
20045 that lack all floating-point support (including, for example, the
20046 floating-point @code{printf} formats).
20047 If code compiled with @option{-mno-float} accidentally contains
20048 floating-point operations, it is likely to suffer a link-time
20049 or run-time failure.
20051 @item -msingle-float
20052 @opindex msingle-float
20053 Assume that the floating-point coprocessor only supports single-precision
20056 @item -mdouble-float
20057 @opindex mdouble-float
20058 Assume that the floating-point coprocessor supports double-precision
20059 operations. This is the default.
20062 @itemx -mno-odd-spreg
20063 @opindex modd-spreg
20064 @opindex mno-odd-spreg
20065 Enable the use of odd-numbered single-precision floating-point registers
20066 for the o32 ABI. This is the default for processors that are known to
20067 support these registers. When using the o32 FPXX ABI, @option{-mno-odd-spreg}
20071 @itemx -mabs=legacy
20073 @opindex mabs=legacy
20074 These options control the treatment of the special not-a-number (NaN)
20075 IEEE 754 floating-point data with the @code{abs.@i{fmt}} and
20076 @code{neg.@i{fmt}} machine instructions.
20078 By default or when @option{-mabs=legacy} is used the legacy
20079 treatment is selected. In this case these instructions are considered
20080 arithmetic and avoided where correct operation is required and the
20081 input operand might be a NaN. A longer sequence of instructions that
20082 manipulate the sign bit of floating-point datum manually is used
20083 instead unless the @option{-ffinite-math-only} option has also been
20086 The @option{-mabs=2008} option selects the IEEE 754-2008 treatment. In
20087 this case these instructions are considered non-arithmetic and therefore
20088 operating correctly in all cases, including in particular where the
20089 input operand is a NaN. These instructions are therefore always used
20090 for the respective operations.
20093 @itemx -mnan=legacy
20095 @opindex mnan=legacy
20096 These options control the encoding of the special not-a-number (NaN)
20097 IEEE 754 floating-point data.
20099 The @option{-mnan=legacy} option selects the legacy encoding. In this
20100 case quiet NaNs (qNaNs) are denoted by the first bit of their trailing
20101 significand field being 0, whereas signaling NaNs (sNaNs) are denoted
20102 by the first bit of their trailing significand field being 1.
20104 The @option{-mnan=2008} option selects the IEEE 754-2008 encoding. In
20105 this case qNaNs are denoted by the first bit of their trailing
20106 significand field being 1, whereas sNaNs are denoted by the first bit of
20107 their trailing significand field being 0.
20109 The default is @option{-mnan=legacy} unless GCC has been configured with
20110 @option{--with-nan=2008}.
20116 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
20117 implement atomic memory built-in functions. When neither option is
20118 specified, GCC uses the instructions if the target architecture
20121 @option{-mllsc} is useful if the runtime environment can emulate the
20122 instructions and @option{-mno-llsc} can be useful when compiling for
20123 nonstandard ISAs. You can make either option the default by
20124 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
20125 respectively. @option{--with-llsc} is the default for some
20126 configurations; see the installation documentation for details.
20132 Use (do not use) revision 1 of the MIPS DSP ASE@.
20133 @xref{MIPS DSP Built-in Functions}. This option defines the
20134 preprocessor macro @code{__mips_dsp}. It also defines
20135 @code{__mips_dsp_rev} to 1.
20141 Use (do not use) revision 2 of the MIPS DSP ASE@.
20142 @xref{MIPS DSP Built-in Functions}. This option defines the
20143 preprocessor macros @code{__mips_dsp} and @code{__mips_dspr2}.
20144 It also defines @code{__mips_dsp_rev} to 2.
20147 @itemx -mno-smartmips
20148 @opindex msmartmips
20149 @opindex mno-smartmips
20150 Use (do not use) the MIPS SmartMIPS ASE.
20152 @item -mpaired-single
20153 @itemx -mno-paired-single
20154 @opindex mpaired-single
20155 @opindex mno-paired-single
20156 Use (do not use) paired-single floating-point instructions.
20157 @xref{MIPS Paired-Single Support}. This option requires
20158 hardware floating-point support to be enabled.
20164 Use (do not use) MIPS Digital Media Extension instructions.
20165 This option can only be used when generating 64-bit code and requires
20166 hardware floating-point support to be enabled.
20171 @opindex mno-mips3d
20172 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
20173 The option @option{-mips3d} implies @option{-mpaired-single}.
20176 @itemx -mno-micromips
20177 @opindex mmicromips
20178 @opindex mno-mmicromips
20179 Generate (do not generate) microMIPS code.
20181 MicroMIPS code generation can also be controlled on a per-function basis
20182 by means of @code{micromips} and @code{nomicromips} attributes.
20183 @xref{Function Attributes}, for more information.
20189 Use (do not use) MT Multithreading instructions.
20195 Use (do not use) the MIPS MCU ASE instructions.
20201 Use (do not use) the MIPS Enhanced Virtual Addressing instructions.
20207 Use (do not use) the MIPS Virtualization (VZ) instructions.
20213 Use (do not use) the MIPS eXtended Physical Address (XPA) instructions.
20217 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
20218 an explanation of the default and the way that the pointer size is
20223 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
20225 The default size of @code{int}s, @code{long}s and pointers depends on
20226 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
20227 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
20228 32-bit @code{long}s. Pointers are the same size as @code{long}s,
20229 or the same size as integer registers, whichever is smaller.
20235 Assume (do not assume) that all symbols have 32-bit values, regardless
20236 of the selected ABI@. This option is useful in combination with
20237 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
20238 to generate shorter and faster references to symbolic addresses.
20242 Put definitions of externally-visible data in a small data section
20243 if that data is no bigger than @var{num} bytes. GCC can then generate
20244 more efficient accesses to the data; see @option{-mgpopt} for details.
20246 The default @option{-G} option depends on the configuration.
20248 @item -mlocal-sdata
20249 @itemx -mno-local-sdata
20250 @opindex mlocal-sdata
20251 @opindex mno-local-sdata
20252 Extend (do not extend) the @option{-G} behavior to local data too,
20253 such as to static variables in C@. @option{-mlocal-sdata} is the
20254 default for all configurations.
20256 If the linker complains that an application is using too much small data,
20257 you might want to try rebuilding the less performance-critical parts with
20258 @option{-mno-local-sdata}. You might also want to build large
20259 libraries with @option{-mno-local-sdata}, so that the libraries leave
20260 more room for the main program.
20262 @item -mextern-sdata
20263 @itemx -mno-extern-sdata
20264 @opindex mextern-sdata
20265 @opindex mno-extern-sdata
20266 Assume (do not assume) that externally-defined data is in
20267 a small data section if the size of that data is within the @option{-G} limit.
20268 @option{-mextern-sdata} is the default for all configurations.
20270 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
20271 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
20272 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
20273 is placed in a small data section. If @var{Var} is defined by another
20274 module, you must either compile that module with a high-enough
20275 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
20276 definition. If @var{Var} is common, you must link the application
20277 with a high-enough @option{-G} setting.
20279 The easiest way of satisfying these restrictions is to compile
20280 and link every module with the same @option{-G} option. However,
20281 you may wish to build a library that supports several different
20282 small data limits. You can do this by compiling the library with
20283 the highest supported @option{-G} setting and additionally using
20284 @option{-mno-extern-sdata} to stop the library from making assumptions
20285 about externally-defined data.
20291 Use (do not use) GP-relative accesses for symbols that are known to be
20292 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
20293 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
20296 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
20297 might not hold the value of @code{_gp}. For example, if the code is
20298 part of a library that might be used in a boot monitor, programs that
20299 call boot monitor routines pass an unknown value in @code{$gp}.
20300 (In such situations, the boot monitor itself is usually compiled
20301 with @option{-G0}.)
20303 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
20304 @option{-mno-extern-sdata}.
20306 @item -membedded-data
20307 @itemx -mno-embedded-data
20308 @opindex membedded-data
20309 @opindex mno-embedded-data
20310 Allocate variables to the read-only data section first if possible, then
20311 next in the small data section if possible, otherwise in data. This gives
20312 slightly slower code than the default, but reduces the amount of RAM required
20313 when executing, and thus may be preferred for some embedded systems.
20315 @item -muninit-const-in-rodata
20316 @itemx -mno-uninit-const-in-rodata
20317 @opindex muninit-const-in-rodata
20318 @opindex mno-uninit-const-in-rodata
20319 Put uninitialized @code{const} variables in the read-only data section.
20320 This option is only meaningful in conjunction with @option{-membedded-data}.
20322 @item -mcode-readable=@var{setting}
20323 @opindex mcode-readable
20324 Specify whether GCC may generate code that reads from executable sections.
20325 There are three possible settings:
20328 @item -mcode-readable=yes
20329 Instructions may freely access executable sections. This is the
20332 @item -mcode-readable=pcrel
20333 MIPS16 PC-relative load instructions can access executable sections,
20334 but other instructions must not do so. This option is useful on 4KSc
20335 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
20336 It is also useful on processors that can be configured to have a dual
20337 instruction/data SRAM interface and that, like the M4K, automatically
20338 redirect PC-relative loads to the instruction RAM.
20340 @item -mcode-readable=no
20341 Instructions must not access executable sections. This option can be
20342 useful on targets that are configured to have a dual instruction/data
20343 SRAM interface but that (unlike the M4K) do not automatically redirect
20344 PC-relative loads to the instruction RAM.
20347 @item -msplit-addresses
20348 @itemx -mno-split-addresses
20349 @opindex msplit-addresses
20350 @opindex mno-split-addresses
20351 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
20352 relocation operators. This option has been superseded by
20353 @option{-mexplicit-relocs} but is retained for backwards compatibility.
20355 @item -mexplicit-relocs
20356 @itemx -mno-explicit-relocs
20357 @opindex mexplicit-relocs
20358 @opindex mno-explicit-relocs
20359 Use (do not use) assembler relocation operators when dealing with symbolic
20360 addresses. The alternative, selected by @option{-mno-explicit-relocs},
20361 is to use assembler macros instead.
20363 @option{-mexplicit-relocs} is the default if GCC was configured
20364 to use an assembler that supports relocation operators.
20366 @item -mcheck-zero-division
20367 @itemx -mno-check-zero-division
20368 @opindex mcheck-zero-division
20369 @opindex mno-check-zero-division
20370 Trap (do not trap) on integer division by zero.
20372 The default is @option{-mcheck-zero-division}.
20374 @item -mdivide-traps
20375 @itemx -mdivide-breaks
20376 @opindex mdivide-traps
20377 @opindex mdivide-breaks
20378 MIPS systems check for division by zero by generating either a
20379 conditional trap or a break instruction. Using traps results in
20380 smaller code, but is only supported on MIPS II and later. Also, some
20381 versions of the Linux kernel have a bug that prevents trap from
20382 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
20383 allow conditional traps on architectures that support them and
20384 @option{-mdivide-breaks} to force the use of breaks.
20386 The default is usually @option{-mdivide-traps}, but this can be
20387 overridden at configure time using @option{--with-divide=breaks}.
20388 Divide-by-zero checks can be completely disabled using
20389 @option{-mno-check-zero-division}.
20391 @item -mload-store-pairs
20392 @itemx -mno-load-store-pairs
20393 @opindex mload-store-pairs
20394 @opindex mno-load-store-pairs
20395 Enable (disable) an optimization that pairs consecutive load or store
20396 instructions to enable load/store bonding. This option is enabled by
20397 default but only takes effect when the selected architecture is known
20398 to support bonding.
20403 @opindex mno-memcpy
20404 Force (do not force) the use of @code{memcpy} for non-trivial block
20405 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
20406 most constant-sized copies.
20409 @itemx -mno-long-calls
20410 @opindex mlong-calls
20411 @opindex mno-long-calls
20412 Disable (do not disable) use of the @code{jal} instruction. Calling
20413 functions using @code{jal} is more efficient but requires the caller
20414 and callee to be in the same 256 megabyte segment.
20416 This option has no effect on abicalls code. The default is
20417 @option{-mno-long-calls}.
20423 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
20424 instructions, as provided by the R4650 ISA@.
20430 Enable (disable) use of the @code{madd} and @code{msub} integer
20431 instructions. The default is @option{-mimadd} on architectures
20432 that support @code{madd} and @code{msub} except for the 74k
20433 architecture where it was found to generate slower code.
20436 @itemx -mno-fused-madd
20437 @opindex mfused-madd
20438 @opindex mno-fused-madd
20439 Enable (disable) use of the floating-point multiply-accumulate
20440 instructions, when they are available. The default is
20441 @option{-mfused-madd}.
20443 On the R8000 CPU when multiply-accumulate instructions are used,
20444 the intermediate product is calculated to infinite precision
20445 and is not subject to the FCSR Flush to Zero bit. This may be
20446 undesirable in some circumstances. On other processors the result
20447 is numerically identical to the equivalent computation using
20448 separate multiply, add, subtract and negate instructions.
20452 Tell the MIPS assembler to not run its preprocessor over user
20453 assembler files (with a @samp{.s} suffix) when assembling them.
20458 @opindex mno-fix-24k
20459 Work around the 24K E48 (lost data on stores during refill) errata.
20460 The workarounds are implemented by the assembler rather than by GCC@.
20463 @itemx -mno-fix-r4000
20464 @opindex mfix-r4000
20465 @opindex mno-fix-r4000
20466 Work around certain R4000 CPU errata:
20469 A double-word or a variable shift may give an incorrect result if executed
20470 immediately after starting an integer division.
20472 A double-word or a variable shift may give an incorrect result if executed
20473 while an integer multiplication is in progress.
20475 An integer division may give an incorrect result if started in a delay slot
20476 of a taken branch or a jump.
20480 @itemx -mno-fix-r4400
20481 @opindex mfix-r4400
20482 @opindex mno-fix-r4400
20483 Work around certain R4400 CPU errata:
20486 A double-word or a variable shift may give an incorrect result if executed
20487 immediately after starting an integer division.
20491 @itemx -mno-fix-r10000
20492 @opindex mfix-r10000
20493 @opindex mno-fix-r10000
20494 Work around certain R10000 errata:
20497 @code{ll}/@code{sc} sequences may not behave atomically on revisions
20498 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
20501 This option can only be used if the target architecture supports
20502 branch-likely instructions. @option{-mfix-r10000} is the default when
20503 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
20507 @itemx -mno-fix-rm7000
20508 @opindex mfix-rm7000
20509 Work around the RM7000 @code{dmult}/@code{dmultu} errata. The
20510 workarounds are implemented by the assembler rather than by GCC@.
20513 @itemx -mno-fix-vr4120
20514 @opindex mfix-vr4120
20515 Work around certain VR4120 errata:
20518 @code{dmultu} does not always produce the correct result.
20520 @code{div} and @code{ddiv} do not always produce the correct result if one
20521 of the operands is negative.
20523 The workarounds for the division errata rely on special functions in
20524 @file{libgcc.a}. At present, these functions are only provided by
20525 the @code{mips64vr*-elf} configurations.
20527 Other VR4120 errata require a NOP to be inserted between certain pairs of
20528 instructions. These errata are handled by the assembler, not by GCC itself.
20531 @opindex mfix-vr4130
20532 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
20533 workarounds are implemented by the assembler rather than by GCC,
20534 although GCC avoids using @code{mflo} and @code{mfhi} if the
20535 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
20536 instructions are available instead.
20539 @itemx -mno-fix-sb1
20541 Work around certain SB-1 CPU core errata.
20542 (This flag currently works around the SB-1 revision 2
20543 ``F1'' and ``F2'' floating-point errata.)
20545 @item -mr10k-cache-barrier=@var{setting}
20546 @opindex mr10k-cache-barrier
20547 Specify whether GCC should insert cache barriers to avoid the
20548 side-effects of speculation on R10K processors.
20550 In common with many processors, the R10K tries to predict the outcome
20551 of a conditional branch and speculatively executes instructions from
20552 the ``taken'' branch. It later aborts these instructions if the
20553 predicted outcome is wrong. However, on the R10K, even aborted
20554 instructions can have side effects.
20556 This problem only affects kernel stores and, depending on the system,
20557 kernel loads. As an example, a speculatively-executed store may load
20558 the target memory into cache and mark the cache line as dirty, even if
20559 the store itself is later aborted. If a DMA operation writes to the
20560 same area of memory before the ``dirty'' line is flushed, the cached
20561 data overwrites the DMA-ed data. See the R10K processor manual
20562 for a full description, including other potential problems.
20564 One workaround is to insert cache barrier instructions before every memory
20565 access that might be speculatively executed and that might have side
20566 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
20567 controls GCC's implementation of this workaround. It assumes that
20568 aborted accesses to any byte in the following regions does not have
20573 the memory occupied by the current function's stack frame;
20576 the memory occupied by an incoming stack argument;
20579 the memory occupied by an object with a link-time-constant address.
20582 It is the kernel's responsibility to ensure that speculative
20583 accesses to these regions are indeed safe.
20585 If the input program contains a function declaration such as:
20591 then the implementation of @code{foo} must allow @code{j foo} and
20592 @code{jal foo} to be executed speculatively. GCC honors this
20593 restriction for functions it compiles itself. It expects non-GCC
20594 functions (such as hand-written assembly code) to do the same.
20596 The option has three forms:
20599 @item -mr10k-cache-barrier=load-store
20600 Insert a cache barrier before a load or store that might be
20601 speculatively executed and that might have side effects even
20604 @item -mr10k-cache-barrier=store
20605 Insert a cache barrier before a store that might be speculatively
20606 executed and that might have side effects even if aborted.
20608 @item -mr10k-cache-barrier=none
20609 Disable the insertion of cache barriers. This is the default setting.
20612 @item -mflush-func=@var{func}
20613 @itemx -mno-flush-func
20614 @opindex mflush-func
20615 Specifies the function to call to flush the I and D caches, or to not
20616 call any such function. If called, the function must take the same
20617 arguments as the common @code{_flush_func}, that is, the address of the
20618 memory range for which the cache is being flushed, the size of the
20619 memory range, and the number 3 (to flush both caches). The default
20620 depends on the target GCC was configured for, but commonly is either
20621 @code{_flush_func} or @code{__cpu_flush}.
20623 @item mbranch-cost=@var{num}
20624 @opindex mbranch-cost
20625 Set the cost of branches to roughly @var{num} ``simple'' instructions.
20626 This cost is only a heuristic and is not guaranteed to produce
20627 consistent results across releases. A zero cost redundantly selects
20628 the default, which is based on the @option{-mtune} setting.
20630 @item -mbranch-likely
20631 @itemx -mno-branch-likely
20632 @opindex mbranch-likely
20633 @opindex mno-branch-likely
20634 Enable or disable use of Branch Likely instructions, regardless of the
20635 default for the selected architecture. By default, Branch Likely
20636 instructions may be generated if they are supported by the selected
20637 architecture. An exception is for the MIPS32 and MIPS64 architectures
20638 and processors that implement those architectures; for those, Branch
20639 Likely instructions are not be generated by default because the MIPS32
20640 and MIPS64 architectures specifically deprecate their use.
20642 @item -mcompact-branches=never
20643 @itemx -mcompact-branches=optimal
20644 @itemx -mcompact-branches=always
20645 @opindex mcompact-branches=never
20646 @opindex mcompact-branches=optimal
20647 @opindex mcompact-branches=always
20648 These options control which form of branches will be generated. The
20649 default is @option{-mcompact-branches=optimal}.
20651 The @option{-mcompact-branches=never} option ensures that compact branch
20652 instructions will never be generated.
20654 The @option{-mcompact-branches=always} option ensures that a compact
20655 branch instruction will be generated if available. If a compact branch
20656 instruction is not available, a delay slot form of the branch will be
20659 This option is supported from MIPS Release 6 onwards.
20661 The @option{-mcompact-branches=optimal} option will cause a delay slot
20662 branch to be used if one is available in the current ISA and the delay
20663 slot is successfully filled. If the delay slot is not filled, a compact
20664 branch will be chosen if one is available.
20666 @item -mfp-exceptions
20667 @itemx -mno-fp-exceptions
20668 @opindex mfp-exceptions
20669 Specifies whether FP exceptions are enabled. This affects how
20670 FP instructions are scheduled for some processors.
20671 The default is that FP exceptions are
20674 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
20675 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
20678 @item -mvr4130-align
20679 @itemx -mno-vr4130-align
20680 @opindex mvr4130-align
20681 The VR4130 pipeline is two-way superscalar, but can only issue two
20682 instructions together if the first one is 8-byte aligned. When this
20683 option is enabled, GCC aligns pairs of instructions that it
20684 thinks should execute in parallel.
20686 This option only has an effect when optimizing for the VR4130.
20687 It normally makes code faster, but at the expense of making it bigger.
20688 It is enabled by default at optimization level @option{-O3}.
20693 Enable (disable) generation of @code{synci} instructions on
20694 architectures that support it. The @code{synci} instructions (if
20695 enabled) are generated when @code{__builtin___clear_cache} is
20698 This option defaults to @option{-mno-synci}, but the default can be
20699 overridden by configuring GCC with @option{--with-synci}.
20701 When compiling code for single processor systems, it is generally safe
20702 to use @code{synci}. However, on many multi-core (SMP) systems, it
20703 does not invalidate the instruction caches on all cores and may lead
20704 to undefined behavior.
20706 @item -mrelax-pic-calls
20707 @itemx -mno-relax-pic-calls
20708 @opindex mrelax-pic-calls
20709 Try to turn PIC calls that are normally dispatched via register
20710 @code{$25} into direct calls. This is only possible if the linker can
20711 resolve the destination at link time and if the destination is within
20712 range for a direct call.
20714 @option{-mrelax-pic-calls} is the default if GCC was configured to use
20715 an assembler and a linker that support the @code{.reloc} assembly
20716 directive and @option{-mexplicit-relocs} is in effect. With
20717 @option{-mno-explicit-relocs}, this optimization can be performed by the
20718 assembler and the linker alone without help from the compiler.
20720 @item -mmcount-ra-address
20721 @itemx -mno-mcount-ra-address
20722 @opindex mmcount-ra-address
20723 @opindex mno-mcount-ra-address
20724 Emit (do not emit) code that allows @code{_mcount} to modify the
20725 calling function's return address. When enabled, this option extends
20726 the usual @code{_mcount} interface with a new @var{ra-address}
20727 parameter, which has type @code{intptr_t *} and is passed in register
20728 @code{$12}. @code{_mcount} can then modify the return address by
20729 doing both of the following:
20732 Returning the new address in register @code{$31}.
20734 Storing the new address in @code{*@var{ra-address}},
20735 if @var{ra-address} is nonnull.
20738 The default is @option{-mno-mcount-ra-address}.
20740 @item -mframe-header-opt
20741 @itemx -mno-frame-header-opt
20742 @opindex mframe-header-opt
20743 Enable (disable) frame header optimization in the o32 ABI. When using the
20744 o32 ABI, calling functions will allocate 16 bytes on the stack for the called
20745 function to write out register arguments. When enabled, this optimization
20746 will suppress the allocation of the frame header if it can be determined that
20749 This optimization is off by default at all optimization levels.
20752 @itemx -mno-lxc1-sxc1
20753 @opindex mlxc1-sxc1
20754 When applicable, enable (disable) the generation of @code{lwxc1},
20755 @code{swxc1}, @code{ldxc1}, @code{sdxc1} instructions. Enabled by default.
20760 When applicable, enable (disable) the generation of 4-operand @code{madd.s},
20761 @code{madd.d} and related instructions. Enabled by default.
20766 @subsection MMIX Options
20767 @cindex MMIX Options
20769 These options are defined for the MMIX:
20773 @itemx -mno-libfuncs
20775 @opindex mno-libfuncs
20776 Specify that intrinsic library functions are being compiled, passing all
20777 values in registers, no matter the size.
20780 @itemx -mno-epsilon
20782 @opindex mno-epsilon
20783 Generate floating-point comparison instructions that compare with respect
20784 to the @code{rE} epsilon register.
20786 @item -mabi=mmixware
20788 @opindex mabi=mmixware
20790 Generate code that passes function parameters and return values that (in
20791 the called function) are seen as registers @code{$0} and up, as opposed to
20792 the GNU ABI which uses global registers @code{$231} and up.
20794 @item -mzero-extend
20795 @itemx -mno-zero-extend
20796 @opindex mzero-extend
20797 @opindex mno-zero-extend
20798 When reading data from memory in sizes shorter than 64 bits, use (do not
20799 use) zero-extending load instructions by default, rather than
20800 sign-extending ones.
20803 @itemx -mno-knuthdiv
20805 @opindex mno-knuthdiv
20806 Make the result of a division yielding a remainder have the same sign as
20807 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
20808 remainder follows the sign of the dividend. Both methods are
20809 arithmetically valid, the latter being almost exclusively used.
20811 @item -mtoplevel-symbols
20812 @itemx -mno-toplevel-symbols
20813 @opindex mtoplevel-symbols
20814 @opindex mno-toplevel-symbols
20815 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
20816 code can be used with the @code{PREFIX} assembly directive.
20820 Generate an executable in the ELF format, rather than the default
20821 @samp{mmo} format used by the @command{mmix} simulator.
20823 @item -mbranch-predict
20824 @itemx -mno-branch-predict
20825 @opindex mbranch-predict
20826 @opindex mno-branch-predict
20827 Use (do not use) the probable-branch instructions, when static branch
20828 prediction indicates a probable branch.
20830 @item -mbase-addresses
20831 @itemx -mno-base-addresses
20832 @opindex mbase-addresses
20833 @opindex mno-base-addresses
20834 Generate (do not generate) code that uses @emph{base addresses}. Using a
20835 base address automatically generates a request (handled by the assembler
20836 and the linker) for a constant to be set up in a global register. The
20837 register is used for one or more base address requests within the range 0
20838 to 255 from the value held in the register. The generally leads to short
20839 and fast code, but the number of different data items that can be
20840 addressed is limited. This means that a program that uses lots of static
20841 data may require @option{-mno-base-addresses}.
20843 @item -msingle-exit
20844 @itemx -mno-single-exit
20845 @opindex msingle-exit
20846 @opindex mno-single-exit
20847 Force (do not force) generated code to have a single exit point in each
20851 @node MN10300 Options
20852 @subsection MN10300 Options
20853 @cindex MN10300 options
20855 These @option{-m} options are defined for Matsushita MN10300 architectures:
20860 Generate code to avoid bugs in the multiply instructions for the MN10300
20861 processors. This is the default.
20863 @item -mno-mult-bug
20864 @opindex mno-mult-bug
20865 Do not generate code to avoid bugs in the multiply instructions for the
20866 MN10300 processors.
20870 Generate code using features specific to the AM33 processor.
20874 Do not generate code using features specific to the AM33 processor. This
20879 Generate code using features specific to the AM33/2.0 processor.
20883 Generate code using features specific to the AM34 processor.
20885 @item -mtune=@var{cpu-type}
20887 Use the timing characteristics of the indicated CPU type when
20888 scheduling instructions. This does not change the targeted processor
20889 type. The CPU type must be one of @samp{mn10300}, @samp{am33},
20890 @samp{am33-2} or @samp{am34}.
20892 @item -mreturn-pointer-on-d0
20893 @opindex mreturn-pointer-on-d0
20894 When generating a function that returns a pointer, return the pointer
20895 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
20896 only in @code{a0}, and attempts to call such functions without a prototype
20897 result in errors. Note that this option is on by default; use
20898 @option{-mno-return-pointer-on-d0} to disable it.
20902 Do not link in the C run-time initialization object file.
20906 Indicate to the linker that it should perform a relaxation optimization pass
20907 to shorten branches, calls and absolute memory addresses. This option only
20908 has an effect when used on the command line for the final link step.
20910 This option makes symbolic debugging impossible.
20914 Allow the compiler to generate @emph{Long Instruction Word}
20915 instructions if the target is the @samp{AM33} or later. This is the
20916 default. This option defines the preprocessor macro @code{__LIW__}.
20920 Do not allow the compiler to generate @emph{Long Instruction Word}
20921 instructions. This option defines the preprocessor macro
20926 Allow the compiler to generate the @emph{SETLB} and @emph{Lcc}
20927 instructions if the target is the @samp{AM33} or later. This is the
20928 default. This option defines the preprocessor macro @code{__SETLB__}.
20932 Do not allow the compiler to generate @emph{SETLB} or @emph{Lcc}
20933 instructions. This option defines the preprocessor macro
20934 @code{__NO_SETLB__}.
20938 @node Moxie Options
20939 @subsection Moxie Options
20940 @cindex Moxie Options
20946 Generate big-endian code. This is the default for @samp{moxie-*-*}
20951 Generate little-endian code.
20955 Generate mul.x and umul.x instructions. This is the default for
20956 @samp{moxiebox-*-*} configurations.
20960 Do not link in the C run-time initialization object file.
20964 @node MSP430 Options
20965 @subsection MSP430 Options
20966 @cindex MSP430 Options
20968 These options are defined for the MSP430:
20974 Force assembly output to always use hex constants. Normally such
20975 constants are signed decimals, but this option is available for
20976 testsuite and/or aesthetic purposes.
20980 Select the MCU to target. This is used to create a C preprocessor
20981 symbol based upon the MCU name, converted to upper case and pre- and
20982 post-fixed with @samp{__}. This in turn is used by the
20983 @file{msp430.h} header file to select an MCU-specific supplementary
20986 The option also sets the ISA to use. If the MCU name is one that is
20987 known to only support the 430 ISA then that is selected, otherwise the
20988 430X ISA is selected. A generic MCU name of @samp{msp430} can also be
20989 used to select the 430 ISA. Similarly the generic @samp{msp430x} MCU
20990 name selects the 430X ISA.
20992 In addition an MCU-specific linker script is added to the linker
20993 command line. The script's name is the name of the MCU with
20994 @file{.ld} appended. Thus specifying @option{-mmcu=xxx} on the @command{gcc}
20995 command line defines the C preprocessor symbol @code{__XXX__} and
20996 cause the linker to search for a script called @file{xxx.ld}.
20998 This option is also passed on to the assembler.
21001 @itemx -mno-warn-mcu
21003 @opindex mno-warn-mcu
21004 This option enables or disables warnings about conflicts between the
21005 MCU name specified by the @option{-mmcu} option and the ISA set by the
21006 @option{-mcpu} option and/or the hardware multiply support set by the
21007 @option{-mhwmult} option. It also toggles warnings about unrecognized
21008 MCU names. This option is on by default.
21012 Specifies the ISA to use. Accepted values are @samp{msp430},
21013 @samp{msp430x} and @samp{msp430xv2}. This option is deprecated. The
21014 @option{-mmcu=} option should be used to select the ISA.
21018 Link to the simulator runtime libraries and linker script. Overrides
21019 any scripts that would be selected by the @option{-mmcu=} option.
21023 Use large-model addressing (20-bit pointers, 32-bit @code{size_t}).
21027 Use small-model addressing (16-bit pointers, 16-bit @code{size_t}).
21031 This option is passed to the assembler and linker, and allows the
21032 linker to perform certain optimizations that cannot be done until
21037 Describes the type of hardware multiply supported by the target.
21038 Accepted values are @samp{none} for no hardware multiply, @samp{16bit}
21039 for the original 16-bit-only multiply supported by early MCUs.
21040 @samp{32bit} for the 16/32-bit multiply supported by later MCUs and
21041 @samp{f5series} for the 16/32-bit multiply supported by F5-series MCUs.
21042 A value of @samp{auto} can also be given. This tells GCC to deduce
21043 the hardware multiply support based upon the MCU name provided by the
21044 @option{-mmcu} option. If no @option{-mmcu} option is specified or if
21045 the MCU name is not recognized then no hardware multiply support is
21046 assumed. @code{auto} is the default setting.
21048 Hardware multiplies are normally performed by calling a library
21049 routine. This saves space in the generated code. When compiling at
21050 @option{-O3} or higher however the hardware multiplier is invoked
21051 inline. This makes for bigger, but faster code.
21053 The hardware multiply routines disable interrupts whilst running and
21054 restore the previous interrupt state when they finish. This makes
21055 them safe to use inside interrupt handlers as well as in normal code.
21059 Enable the use of a minimum runtime environment - no static
21060 initializers or constructors. This is intended for memory-constrained
21061 devices. The compiler includes special symbols in some objects
21062 that tell the linker and runtime which code fragments are required.
21064 @item -mcode-region=
21065 @itemx -mdata-region=
21066 @opindex mcode-region
21067 @opindex mdata-region
21068 These options tell the compiler where to place functions and data that
21069 do not have one of the @code{lower}, @code{upper}, @code{either} or
21070 @code{section} attributes. Possible values are @code{lower},
21071 @code{upper}, @code{either} or @code{any}. The first three behave
21072 like the corresponding attribute. The fourth possible value -
21073 @code{any} - is the default. It leaves placement entirely up to the
21074 linker script and how it assigns the standard sections
21075 (@code{.text}, @code{.data}, etc) to the memory regions.
21077 @item -msilicon-errata=
21078 @opindex msilicon-errata
21079 This option passes on a request to assembler to enable the fixes for
21080 the named silicon errata.
21082 @item -msilicon-errata-warn=
21083 @opindex msilicon-errata-warn
21084 This option passes on a request to the assembler to enable warning
21085 messages when a silicon errata might need to be applied.
21089 @node NDS32 Options
21090 @subsection NDS32 Options
21091 @cindex NDS32 Options
21093 These options are defined for NDS32 implementations:
21098 @opindex mbig-endian
21099 Generate code in big-endian mode.
21101 @item -mlittle-endian
21102 @opindex mlittle-endian
21103 Generate code in little-endian mode.
21105 @item -mreduced-regs
21106 @opindex mreduced-regs
21107 Use reduced-set registers for register allocation.
21110 @opindex mfull-regs
21111 Use full-set registers for register allocation.
21115 Generate conditional move instructions.
21119 Do not generate conditional move instructions.
21123 Generate performance extension instructions.
21125 @item -mno-perf-ext
21126 @opindex mno-perf-ext
21127 Do not generate performance extension instructions.
21131 Generate v3 push25/pop25 instructions.
21134 @opindex mno-v3push
21135 Do not generate v3 push25/pop25 instructions.
21139 Generate 16-bit instructions.
21142 @opindex mno-16-bit
21143 Do not generate 16-bit instructions.
21145 @item -misr-vector-size=@var{num}
21146 @opindex misr-vector-size
21147 Specify the size of each interrupt vector, which must be 4 or 16.
21149 @item -mcache-block-size=@var{num}
21150 @opindex mcache-block-size
21151 Specify the size of each cache block,
21152 which must be a power of 2 between 4 and 512.
21154 @item -march=@var{arch}
21156 Specify the name of the target architecture.
21158 @item -mcmodel=@var{code-model}
21160 Set the code model to one of
21163 All the data and read-only data segments must be within 512KB addressing space.
21164 The text segment must be within 16MB addressing space.
21165 @item @samp{medium}
21166 The data segment must be within 512KB while the read-only data segment can be
21167 within 4GB addressing space. The text segment should be still within 16MB
21170 All the text and data segments can be within 4GB addressing space.
21174 @opindex mctor-dtor
21175 Enable constructor/destructor feature.
21179 Guide linker to relax instructions.
21183 @node Nios II Options
21184 @subsection Nios II Options
21185 @cindex Nios II options
21186 @cindex Altera Nios II options
21188 These are the options defined for the Altera Nios II processor.
21194 @cindex smaller data references
21195 Put global and static objects less than or equal to @var{num} bytes
21196 into the small data or BSS sections instead of the normal data or BSS
21197 sections. The default value of @var{num} is 8.
21199 @item -mgpopt=@var{option}
21204 Generate (do not generate) GP-relative accesses. The following
21205 @var{option} names are recognized:
21210 Do not generate GP-relative accesses.
21213 Generate GP-relative accesses for small data objects that are not
21214 external, weak, or uninitialized common symbols.
21215 Also use GP-relative addressing for objects that
21216 have been explicitly placed in a small data section via a @code{section}
21220 As for @samp{local}, but also generate GP-relative accesses for
21221 small data objects that are external, weak, or common. If you use this option,
21222 you must ensure that all parts of your program (including libraries) are
21223 compiled with the same @option{-G} setting.
21226 Generate GP-relative accesses for all data objects in the program. If you
21227 use this option, the entire data and BSS segments
21228 of your program must fit in 64K of memory and you must use an appropriate
21229 linker script to allocate them within the addressable range of the
21233 Generate GP-relative addresses for function pointers as well as data
21234 pointers. If you use this option, the entire text, data, and BSS segments
21235 of your program must fit in 64K of memory and you must use an appropriate
21236 linker script to allocate them within the addressable range of the
21241 @option{-mgpopt} is equivalent to @option{-mgpopt=local}, and
21242 @option{-mno-gpopt} is equivalent to @option{-mgpopt=none}.
21244 The default is @option{-mgpopt} except when @option{-fpic} or
21245 @option{-fPIC} is specified to generate position-independent code.
21246 Note that the Nios II ABI does not permit GP-relative accesses from
21249 You may need to specify @option{-mno-gpopt} explicitly when building
21250 programs that include large amounts of small data, including large
21251 GOT data sections. In this case, the 16-bit offset for GP-relative
21252 addressing may not be large enough to allow access to the entire
21253 small data section.
21255 @item -mgprel-sec=@var{regexp}
21256 @opindex mgprel-sec
21257 This option specifies additional section names that can be accessed via
21258 GP-relative addressing. It is most useful in conjunction with
21259 @code{section} attributes on variable declarations
21260 (@pxref{Common Variable Attributes}) and a custom linker script.
21261 The @var{regexp} is a POSIX Extended Regular Expression.
21263 This option does not affect the behavior of the @option{-G} option, and
21264 and the specified sections are in addition to the standard @code{.sdata}
21265 and @code{.sbss} small-data sections that are recognized by @option{-mgpopt}.
21267 @item -mr0rel-sec=@var{regexp}
21268 @opindex mr0rel-sec
21269 This option specifies names of sections that can be accessed via a
21270 16-bit offset from @code{r0}; that is, in the low 32K or high 32K
21271 of the 32-bit address space. It is most useful in conjunction with
21272 @code{section} attributes on variable declarations
21273 (@pxref{Common Variable Attributes}) and a custom linker script.
21274 The @var{regexp} is a POSIX Extended Regular Expression.
21276 In contrast to the use of GP-relative addressing for small data,
21277 zero-based addressing is never generated by default and there are no
21278 conventional section names used in standard linker scripts for sections
21279 in the low or high areas of memory.
21285 Generate little-endian (default) or big-endian (experimental) code,
21288 @item -march=@var{arch}
21290 This specifies the name of the target Nios II architecture. GCC uses this
21291 name to determine what kind of instructions it can emit when generating
21292 assembly code. Permissible names are: @samp{r1}, @samp{r2}.
21294 The preprocessor macro @code{__nios2_arch__} is available to programs,
21295 with value 1 or 2, indicating the targeted ISA level.
21297 @item -mbypass-cache
21298 @itemx -mno-bypass-cache
21299 @opindex mno-bypass-cache
21300 @opindex mbypass-cache
21301 Force all load and store instructions to always bypass cache by
21302 using I/O variants of the instructions. The default is not to
21305 @item -mno-cache-volatile
21306 @itemx -mcache-volatile
21307 @opindex mcache-volatile
21308 @opindex mno-cache-volatile
21309 Volatile memory access bypass the cache using the I/O variants of
21310 the load and store instructions. The default is not to bypass the cache.
21312 @item -mno-fast-sw-div
21313 @itemx -mfast-sw-div
21314 @opindex mno-fast-sw-div
21315 @opindex mfast-sw-div
21316 Do not use table-based fast divide for small numbers. The default
21317 is to use the fast divide at @option{-O3} and above.
21321 @itemx -mno-hw-mulx
21325 @opindex mno-hw-mul
21327 @opindex mno-hw-mulx
21329 @opindex mno-hw-div
21331 Enable or disable emitting @code{mul}, @code{mulx} and @code{div} family of
21332 instructions by the compiler. The default is to emit @code{mul}
21333 and not emit @code{div} and @code{mulx}.
21339 Enable or disable generation of Nios II R2 BMX (bit manipulation) and
21340 CDX (code density) instructions. Enabling these instructions also
21341 requires @option{-march=r2}. Since these instructions are optional
21342 extensions to the R2 architecture, the default is not to emit them.
21344 @item -mcustom-@var{insn}=@var{N}
21345 @itemx -mno-custom-@var{insn}
21346 @opindex mcustom-@var{insn}
21347 @opindex mno-custom-@var{insn}
21348 Each @option{-mcustom-@var{insn}=@var{N}} option enables use of a
21349 custom instruction with encoding @var{N} when generating code that uses
21350 @var{insn}. For example, @option{-mcustom-fadds=253} generates custom
21351 instruction 253 for single-precision floating-point add operations instead
21352 of the default behavior of using a library call.
21354 The following values of @var{insn} are supported. Except as otherwise
21355 noted, floating-point operations are expected to be implemented with
21356 normal IEEE 754 semantics and correspond directly to the C operators or the
21357 equivalent GCC built-in functions (@pxref{Other Builtins}).
21359 Single-precision floating point:
21362 @item @samp{fadds}, @samp{fsubs}, @samp{fdivs}, @samp{fmuls}
21363 Binary arithmetic operations.
21369 Unary absolute value.
21371 @item @samp{fcmpeqs}, @samp{fcmpges}, @samp{fcmpgts}, @samp{fcmples}, @samp{fcmplts}, @samp{fcmpnes}
21372 Comparison operations.
21374 @item @samp{fmins}, @samp{fmaxs}
21375 Floating-point minimum and maximum. These instructions are only
21376 generated if @option{-ffinite-math-only} is specified.
21378 @item @samp{fsqrts}
21379 Unary square root operation.
21381 @item @samp{fcoss}, @samp{fsins}, @samp{ftans}, @samp{fatans}, @samp{fexps}, @samp{flogs}
21382 Floating-point trigonometric and exponential functions. These instructions
21383 are only generated if @option{-funsafe-math-optimizations} is also specified.
21387 Double-precision floating point:
21390 @item @samp{faddd}, @samp{fsubd}, @samp{fdivd}, @samp{fmuld}
21391 Binary arithmetic operations.
21397 Unary absolute value.
21399 @item @samp{fcmpeqd}, @samp{fcmpged}, @samp{fcmpgtd}, @samp{fcmpled}, @samp{fcmpltd}, @samp{fcmpned}
21400 Comparison operations.
21402 @item @samp{fmind}, @samp{fmaxd}
21403 Double-precision minimum and maximum. These instructions are only
21404 generated if @option{-ffinite-math-only} is specified.
21406 @item @samp{fsqrtd}
21407 Unary square root operation.
21409 @item @samp{fcosd}, @samp{fsind}, @samp{ftand}, @samp{fatand}, @samp{fexpd}, @samp{flogd}
21410 Double-precision trigonometric and exponential functions. These instructions
21411 are only generated if @option{-funsafe-math-optimizations} is also specified.
21417 @item @samp{fextsd}
21418 Conversion from single precision to double precision.
21420 @item @samp{ftruncds}
21421 Conversion from double precision to single precision.
21423 @item @samp{fixsi}, @samp{fixsu}, @samp{fixdi}, @samp{fixdu}
21424 Conversion from floating point to signed or unsigned integer types, with
21425 truncation towards zero.
21428 Conversion from single-precision floating point to signed integer,
21429 rounding to the nearest integer and ties away from zero.
21430 This corresponds to the @code{__builtin_lroundf} function when
21431 @option{-fno-math-errno} is used.
21433 @item @samp{floatis}, @samp{floatus}, @samp{floatid}, @samp{floatud}
21434 Conversion from signed or unsigned integer types to floating-point types.
21438 In addition, all of the following transfer instructions for internal
21439 registers X and Y must be provided to use any of the double-precision
21440 floating-point instructions. Custom instructions taking two
21441 double-precision source operands expect the first operand in the
21442 64-bit register X. The other operand (or only operand of a unary
21443 operation) is given to the custom arithmetic instruction with the
21444 least significant half in source register @var{src1} and the most
21445 significant half in @var{src2}. A custom instruction that returns a
21446 double-precision result returns the most significant 32 bits in the
21447 destination register and the other half in 32-bit register Y.
21448 GCC automatically generates the necessary code sequences to write
21449 register X and/or read register Y when double-precision floating-point
21450 instructions are used.
21455 Write @var{src1} into the least significant half of X and @var{src2} into
21456 the most significant half of X.
21459 Write @var{src1} into Y.
21461 @item @samp{frdxhi}, @samp{frdxlo}
21462 Read the most or least (respectively) significant half of X and store it in
21466 Read the value of Y and store it into @var{dest}.
21469 Note that you can gain more local control over generation of Nios II custom
21470 instructions by using the @code{target("custom-@var{insn}=@var{N}")}
21471 and @code{target("no-custom-@var{insn}")} function attributes
21472 (@pxref{Function Attributes})
21473 or pragmas (@pxref{Function Specific Option Pragmas}).
21475 @item -mcustom-fpu-cfg=@var{name}
21476 @opindex mcustom-fpu-cfg
21478 This option enables a predefined, named set of custom instruction encodings
21479 (see @option{-mcustom-@var{insn}} above).
21480 Currently, the following sets are defined:
21482 @option{-mcustom-fpu-cfg=60-1} is equivalent to:
21483 @gccoptlist{-mcustom-fmuls=252 @gol
21484 -mcustom-fadds=253 @gol
21485 -mcustom-fsubs=254 @gol
21486 -fsingle-precision-constant}
21488 @option{-mcustom-fpu-cfg=60-2} is equivalent to:
21489 @gccoptlist{-mcustom-fmuls=252 @gol
21490 -mcustom-fadds=253 @gol
21491 -mcustom-fsubs=254 @gol
21492 -mcustom-fdivs=255 @gol
21493 -fsingle-precision-constant}
21495 @option{-mcustom-fpu-cfg=72-3} is equivalent to:
21496 @gccoptlist{-mcustom-floatus=243 @gol
21497 -mcustom-fixsi=244 @gol
21498 -mcustom-floatis=245 @gol
21499 -mcustom-fcmpgts=246 @gol
21500 -mcustom-fcmples=249 @gol
21501 -mcustom-fcmpeqs=250 @gol
21502 -mcustom-fcmpnes=251 @gol
21503 -mcustom-fmuls=252 @gol
21504 -mcustom-fadds=253 @gol
21505 -mcustom-fsubs=254 @gol
21506 -mcustom-fdivs=255 @gol
21507 -fsingle-precision-constant}
21509 Custom instruction assignments given by individual
21510 @option{-mcustom-@var{insn}=} options override those given by
21511 @option{-mcustom-fpu-cfg=}, regardless of the
21512 order of the options on the command line.
21514 Note that you can gain more local control over selection of a FPU
21515 configuration by using the @code{target("custom-fpu-cfg=@var{name}")}
21516 function attribute (@pxref{Function Attributes})
21517 or pragma (@pxref{Function Specific Option Pragmas}).
21521 These additional @samp{-m} options are available for the Altera Nios II
21522 ELF (bare-metal) target:
21528 Link with HAL BSP. This suppresses linking with the GCC-provided C runtime
21529 startup and termination code, and is typically used in conjunction with
21530 @option{-msys-crt0=} to specify the location of the alternate startup code
21531 provided by the HAL BSP.
21535 Link with a limited version of the C library, @option{-lsmallc}, rather than
21538 @item -msys-crt0=@var{startfile}
21540 @var{startfile} is the file name of the startfile (crt0) to use
21541 when linking. This option is only useful in conjunction with @option{-mhal}.
21543 @item -msys-lib=@var{systemlib}
21545 @var{systemlib} is the library name of the library that provides
21546 low-level system calls required by the C library,
21547 e.g. @code{read} and @code{write}.
21548 This option is typically used to link with a library provided by a HAL BSP.
21552 @node Nvidia PTX Options
21553 @subsection Nvidia PTX Options
21554 @cindex Nvidia PTX options
21555 @cindex nvptx options
21557 These options are defined for Nvidia PTX:
21565 Generate code for 32-bit or 64-bit ABI.
21568 @opindex mmainkernel
21569 Link in code for a __main kernel. This is for stand-alone instead of
21570 offloading execution.
21574 Apply partitioned execution optimizations. This is the default when any
21575 level of optimization is selected.
21578 @opindex msoft-stack
21579 Generate code that does not use @code{.local} memory
21580 directly for stack storage. Instead, a per-warp stack pointer is
21581 maintained explicitly. This enables variable-length stack allocation (with
21582 variable-length arrays or @code{alloca}), and when global memory is used for
21583 underlying storage, makes it possible to access automatic variables from other
21584 threads, or with atomic instructions. This code generation variant is used
21585 for OpenMP offloading, but the option is exposed on its own for the purpose
21586 of testing the compiler; to generate code suitable for linking into programs
21587 using OpenMP offloading, use option @option{-mgomp}.
21589 @item -muniform-simt
21590 @opindex muniform-simt
21591 Switch to code generation variant that allows to execute all threads in each
21592 warp, while maintaining memory state and side effects as if only one thread
21593 in each warp was active outside of OpenMP SIMD regions. All atomic operations
21594 and calls to runtime (malloc, free, vprintf) are conditionally executed (iff
21595 current lane index equals the master lane index), and the register being
21596 assigned is copied via a shuffle instruction from the master lane. Outside of
21597 SIMD regions lane 0 is the master; inside, each thread sees itself as the
21598 master. Shared memory array @code{int __nvptx_uni[]} stores all-zeros or
21599 all-ones bitmasks for each warp, indicating current mode (0 outside of SIMD
21600 regions). Each thread can bitwise-and the bitmask at position @code{tid.y}
21601 with current lane index to compute the master lane index.
21605 Generate code for use in OpenMP offloading: enables @option{-msoft-stack} and
21606 @option{-muniform-simt} options, and selects corresponding multilib variant.
21610 @node PDP-11 Options
21611 @subsection PDP-11 Options
21612 @cindex PDP-11 Options
21614 These options are defined for the PDP-11:
21619 Use hardware FPP floating point. This is the default. (FIS floating
21620 point on the PDP-11/40 is not supported.)
21623 @opindex msoft-float
21624 Do not use hardware floating point.
21628 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
21632 Return floating-point results in memory. This is the default.
21636 Generate code for a PDP-11/40.
21640 Generate code for a PDP-11/45. This is the default.
21644 Generate code for a PDP-11/10.
21646 @item -mbcopy-builtin
21647 @opindex mbcopy-builtin
21648 Use inline @code{movmemhi} patterns for copying memory. This is the
21653 Do not use inline @code{movmemhi} patterns for copying memory.
21659 Use 16-bit @code{int}. This is the default.
21665 Use 32-bit @code{int}.
21668 @itemx -mno-float32
21670 @opindex mno-float32
21671 Use 64-bit @code{float}. This is the default.
21674 @itemx -mno-float64
21676 @opindex mno-float64
21677 Use 32-bit @code{float}.
21681 Use @code{abshi2} pattern. This is the default.
21685 Do not use @code{abshi2} pattern.
21687 @item -mbranch-expensive
21688 @opindex mbranch-expensive
21689 Pretend that branches are expensive. This is for experimenting with
21690 code generation only.
21692 @item -mbranch-cheap
21693 @opindex mbranch-cheap
21694 Do not pretend that branches are expensive. This is the default.
21698 Use Unix assembler syntax. This is the default when configured for
21699 @samp{pdp11-*-bsd}.
21703 Use DEC assembler syntax. This is the default when configured for any
21704 PDP-11 target other than @samp{pdp11-*-bsd}.
21707 @node picoChip Options
21708 @subsection picoChip Options
21709 @cindex picoChip options
21711 These @samp{-m} options are defined for picoChip implementations:
21715 @item -mae=@var{ae_type}
21717 Set the instruction set, register set, and instruction scheduling
21718 parameters for array element type @var{ae_type}. Supported values
21719 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
21721 @option{-mae=ANY} selects a completely generic AE type. Code
21722 generated with this option runs on any of the other AE types. The
21723 code is not as efficient as it would be if compiled for a specific
21724 AE type, and some types of operation (e.g., multiplication) do not
21725 work properly on all types of AE.
21727 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
21728 for compiled code, and is the default.
21730 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
21731 option may suffer from poor performance of byte (char) manipulation,
21732 since the DSP AE does not provide hardware support for byte load/stores.
21734 @item -msymbol-as-address
21735 Enable the compiler to directly use a symbol name as an address in a
21736 load/store instruction, without first loading it into a
21737 register. Typically, the use of this option generates larger
21738 programs, which run faster than when the option isn't used. However, the
21739 results vary from program to program, so it is left as a user option,
21740 rather than being permanently enabled.
21742 @item -mno-inefficient-warnings
21743 Disables warnings about the generation of inefficient code. These
21744 warnings can be generated, for example, when compiling code that
21745 performs byte-level memory operations on the MAC AE type. The MAC AE has
21746 no hardware support for byte-level memory operations, so all byte
21747 load/stores must be synthesized from word load/store operations. This is
21748 inefficient and a warning is generated to indicate
21749 that you should rewrite the code to avoid byte operations, or to target
21750 an AE type that has the necessary hardware support. This option disables
21755 @node PowerPC Options
21756 @subsection PowerPC Options
21757 @cindex PowerPC options
21759 These are listed under @xref{RS/6000 and PowerPC Options}.
21761 @node RISC-V Options
21762 @subsection RISC-V Options
21763 @cindex RISC-V Options
21765 These command-line options are defined for RISC-V targets:
21768 @item -mbranch-cost=@var{n}
21769 @opindex mbranch-cost
21770 Set the cost of branches to roughly @var{n} instructions.
21775 Don't optimize block moves.
21780 When generating PIC code, allow the use of PLTs. Ignored for non-PIC.
21782 @item -mabi=@var{ABI-string}
21784 @item -mabi=@var{ABI-string}
21786 Specify integer and floating-point calling convention. @var{ABI-string}
21787 contains two parts: the size of integer types and the registers used for
21788 floating-point types. For example @samp{-march=rv64ifd -mabi=lp64d} means that
21789 @samp{long} and pointers are 64-bit (implicitly defining @samp{int} to be
21790 32-bit), and that floating-point values up to 64 bits wide are passed in F
21791 registers. Contrast this with @samp{-march=rv64ifd -mabi=lp64f}, which still
21792 allows the compiler to generate code that uses the F and D extensions but only
21793 allows floating-point values up to 32 bits long to be passed in registers; or
21794 @samp{-march=rv64ifd -mabi=lp64}, in which no floating-point arguments will be
21795 passed in registers.
21797 The default for this argument is system dependent, users who want a specific
21798 calling convention should specify one explicitly. The valid calling
21799 conventions are: @samp{ilp32}, @samp{ilp32f}, @samp{ilp32d}, @samp{lp64},
21800 @samp{lp64f}, and @samp{lp64d}. Some calling conventions are impossible to
21801 implement on some ISAs: for example, @samp{-march=rv32if -mabi=ilp32d} is
21802 invalid because the ABI requires 64-bit values be passed in F registers, but F
21803 registers are only 32 bits wide.
21808 Use hardware floating-point divide and square root instructions. This requires
21809 the F or D extensions for floating-point registers.
21814 Use hardware instructions for integer division. This requires the M extension.
21816 @item -march=@var{ISA-string}
21818 Generate code for given RISC-V ISA (e.g.@ @samp{rv64im}). ISA strings must be
21819 lower-case. Examples include @samp{rv64i}, @samp{rv32g}, and @samp{rv32imaf}.
21821 @item -mtune=@var{processor-string}
21823 Optimize the output for the given processor, specified by microarchitecture
21826 @item -msmall-data-limit=@var{n}
21827 @opindex msmall-data-limit
21828 Put global and static data smaller than @var{n} bytes into a special section
21831 @item -msave-restore
21832 @itemx -mno-save-restore
21833 @opindex msave-restore
21834 Use smaller but slower prologue and epilogue code.
21836 @item -mstrict-align
21837 @itemx -mno-strict-align
21838 @opindex mstrict-align
21839 Do not generate unaligned memory accesses.
21841 @item -mcmodel=medlow
21842 @opindex mcmodel=medlow
21843 Generate code for the medium-low code model. The program and its statically
21844 defined symbols must lie within a single 2 GiB address range and must lie
21845 between absolute addresses @minus{}2 GiB and +2 GiB. Programs can be
21846 statically or dynamically linked. This is the default code model.
21848 @item -mcmodel=medany
21849 @opindex mcmodel=medany
21850 Generate code for the medium-any code model. The program and its statically
21851 defined symbols must be within any single 2 GiB address range. Programs can be
21852 statically or dynamically linked.
21857 @subsection RL78 Options
21858 @cindex RL78 Options
21864 Links in additional target libraries to support operation within a
21873 Specifies the type of hardware multiplication and division support to
21874 be used. The simplest is @code{none}, which uses software for both
21875 multiplication and division. This is the default. The @code{g13}
21876 value is for the hardware multiply/divide peripheral found on the
21877 RL78/G13 (S2 core) targets. The @code{g14} value selects the use of
21878 the multiplication and division instructions supported by the RL78/G14
21879 (S3 core) parts. The value @code{rl78} is an alias for @code{g14} and
21880 the value @code{mg10} is an alias for @code{none}.
21882 In addition a C preprocessor macro is defined, based upon the setting
21883 of this option. Possible values are: @code{__RL78_MUL_NONE__},
21884 @code{__RL78_MUL_G13__} or @code{__RL78_MUL_G14__}.
21891 Specifies the RL78 core to target. The default is the G14 core, also
21892 known as an S3 core or just RL78. The G13 or S2 core does not have
21893 multiply or divide instructions, instead it uses a hardware peripheral
21894 for these operations. The G10 or S1 core does not have register
21895 banks, so it uses a different calling convention.
21897 If this option is set it also selects the type of hardware multiply
21898 support to use, unless this is overridden by an explicit
21899 @option{-mmul=none} option on the command line. Thus specifying
21900 @option{-mcpu=g13} enables the use of the G13 hardware multiply
21901 peripheral and specifying @option{-mcpu=g10} disables the use of
21902 hardware multiplications altogether.
21904 Note, although the RL78/G14 core is the default target, specifying
21905 @option{-mcpu=g14} or @option{-mcpu=rl78} on the command line does
21906 change the behavior of the toolchain since it also enables G14
21907 hardware multiply support. If these options are not specified on the
21908 command line then software multiplication routines will be used even
21909 though the code targets the RL78 core. This is for backwards
21910 compatibility with older toolchains which did not have hardware
21911 multiply and divide support.
21913 In addition a C preprocessor macro is defined, based upon the setting
21914 of this option. Possible values are: @code{__RL78_G10__},
21915 @code{__RL78_G13__} or @code{__RL78_G14__}.
21925 These are aliases for the corresponding @option{-mcpu=} option. They
21926 are provided for backwards compatibility.
21930 Allow the compiler to use all of the available registers. By default
21931 registers @code{r24..r31} are reserved for use in interrupt handlers.
21932 With this option enabled these registers can be used in ordinary
21935 @item -m64bit-doubles
21936 @itemx -m32bit-doubles
21937 @opindex m64bit-doubles
21938 @opindex m32bit-doubles
21939 Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
21940 or 32 bits (@option{-m32bit-doubles}) in size. The default is
21941 @option{-m32bit-doubles}.
21943 @item -msave-mduc-in-interrupts
21944 @item -mno-save-mduc-in-interrupts
21945 @opindex msave-mduc-in-interrupts
21946 @opindex mno-save-mduc-in-interrupts
21947 Specifies that interrupt handler functions should preserve the
21948 MDUC registers. This is only necessary if normal code might use
21949 the MDUC registers, for example because it performs multiplication
21950 and division operations. The default is to ignore the MDUC registers
21951 as this makes the interrupt handlers faster. The target option -mg13
21952 needs to be passed for this to work as this feature is only available
21953 on the G13 target (S2 core). The MDUC registers will only be saved
21954 if the interrupt handler performs a multiplication or division
21955 operation or it calls another function.
21959 @node RS/6000 and PowerPC Options
21960 @subsection IBM RS/6000 and PowerPC Options
21961 @cindex RS/6000 and PowerPC Options
21962 @cindex IBM RS/6000 and PowerPC Options
21964 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
21966 @item -mpowerpc-gpopt
21967 @itemx -mno-powerpc-gpopt
21968 @itemx -mpowerpc-gfxopt
21969 @itemx -mno-powerpc-gfxopt
21972 @itemx -mno-powerpc64
21976 @itemx -mno-popcntb
21978 @itemx -mno-popcntd
21987 @itemx -mno-hard-dfp
21988 @opindex mpowerpc-gpopt
21989 @opindex mno-powerpc-gpopt
21990 @opindex mpowerpc-gfxopt
21991 @opindex mno-powerpc-gfxopt
21992 @opindex mpowerpc64
21993 @opindex mno-powerpc64
21997 @opindex mno-popcntb
21999 @opindex mno-popcntd
22005 @opindex mno-mfpgpr
22007 @opindex mno-hard-dfp
22008 You use these options to specify which instructions are available on the
22009 processor you are using. The default value of these options is
22010 determined when configuring GCC@. Specifying the
22011 @option{-mcpu=@var{cpu_type}} overrides the specification of these
22012 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
22013 rather than the options listed above.
22015 Specifying @option{-mpowerpc-gpopt} allows
22016 GCC to use the optional PowerPC architecture instructions in the
22017 General Purpose group, including floating-point square root. Specifying
22018 @option{-mpowerpc-gfxopt} allows GCC to
22019 use the optional PowerPC architecture instructions in the Graphics
22020 group, including floating-point select.
22022 The @option{-mmfcrf} option allows GCC to generate the move from
22023 condition register field instruction implemented on the POWER4
22024 processor and other processors that support the PowerPC V2.01
22026 The @option{-mpopcntb} option allows GCC to generate the popcount and
22027 double-precision FP reciprocal estimate instruction implemented on the
22028 POWER5 processor and other processors that support the PowerPC V2.02
22030 The @option{-mpopcntd} option allows GCC to generate the popcount
22031 instruction implemented on the POWER7 processor and other processors
22032 that support the PowerPC V2.06 architecture.
22033 The @option{-mfprnd} option allows GCC to generate the FP round to
22034 integer instructions implemented on the POWER5+ processor and other
22035 processors that support the PowerPC V2.03 architecture.
22036 The @option{-mcmpb} option allows GCC to generate the compare bytes
22037 instruction implemented on the POWER6 processor and other processors
22038 that support the PowerPC V2.05 architecture.
22039 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
22040 general-purpose register instructions implemented on the POWER6X
22041 processor and other processors that support the extended PowerPC V2.05
22043 The @option{-mhard-dfp} option allows GCC to generate the decimal
22044 floating-point instructions implemented on some POWER processors.
22046 The @option{-mpowerpc64} option allows GCC to generate the additional
22047 64-bit instructions that are found in the full PowerPC64 architecture
22048 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
22049 @option{-mno-powerpc64}.
22051 @item -mcpu=@var{cpu_type}
22053 Set architecture type, register usage, and
22054 instruction scheduling parameters for machine type @var{cpu_type}.
22055 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
22056 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
22057 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
22058 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
22059 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
22060 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
22061 @samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{e5500},
22062 @samp{e6500}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
22063 @samp{titan}, @samp{power3}, @samp{power4}, @samp{power5}, @samp{power5+},
22064 @samp{power6}, @samp{power6x}, @samp{power7}, @samp{power8},
22065 @samp{power9}, @samp{powerpc}, @samp{powerpc64}, @samp{powerpc64le},
22068 @option{-mcpu=powerpc}, @option{-mcpu=powerpc64}, and
22069 @option{-mcpu=powerpc64le} specify pure 32-bit PowerPC (either
22070 endian), 64-bit big endian PowerPC and 64-bit little endian PowerPC
22071 architecture machine types, with an appropriate, generic processor
22072 model assumed for scheduling purposes.
22074 The other options specify a specific processor. Code generated under
22075 those options runs best on that processor, and may not run at all on
22078 The @option{-mcpu} options automatically enable or disable the
22081 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
22082 -mpopcntb -mpopcntd -mpowerpc64 @gol
22083 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
22084 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx @gol
22085 -mcrypto -mdirect-move -mhtm -mpower8-fusion -mpower8-vector @gol
22086 -mquad-memory -mquad-memory-atomic -mfloat128 -mfloat128-hardware}
22088 The particular options set for any particular CPU varies between
22089 compiler versions, depending on what setting seems to produce optimal
22090 code for that CPU; it doesn't necessarily reflect the actual hardware's
22091 capabilities. If you wish to set an individual option to a particular
22092 value, you may specify it after the @option{-mcpu} option, like
22093 @option{-mcpu=970 -mno-altivec}.
22095 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
22096 not enabled or disabled by the @option{-mcpu} option at present because
22097 AIX does not have full support for these options. You may still
22098 enable or disable them individually if you're sure it'll work in your
22101 @item -mtune=@var{cpu_type}
22103 Set the instruction scheduling parameters for machine type
22104 @var{cpu_type}, but do not set the architecture type or register usage,
22105 as @option{-mcpu=@var{cpu_type}} does. The same
22106 values for @var{cpu_type} are used for @option{-mtune} as for
22107 @option{-mcpu}. If both are specified, the code generated uses the
22108 architecture and registers set by @option{-mcpu}, but the
22109 scheduling parameters set by @option{-mtune}.
22111 @item -mcmodel=small
22112 @opindex mcmodel=small
22113 Generate PowerPC64 code for the small model: The TOC is limited to
22116 @item -mcmodel=medium
22117 @opindex mcmodel=medium
22118 Generate PowerPC64 code for the medium model: The TOC and other static
22119 data may be up to a total of 4G in size. This is the default for 64-bit
22122 @item -mcmodel=large
22123 @opindex mcmodel=large
22124 Generate PowerPC64 code for the large model: The TOC may be up to 4G
22125 in size. Other data and code is only limited by the 64-bit address
22129 @itemx -mno-altivec
22131 @opindex mno-altivec
22132 Generate code that uses (does not use) AltiVec instructions, and also
22133 enable the use of built-in functions that allow more direct access to
22134 the AltiVec instruction set. You may also need to set
22135 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
22138 When @option{-maltivec} is used, rather than @option{-maltivec=le} or
22139 @option{-maltivec=be}, the element order for AltiVec intrinsics such
22140 as @code{vec_splat}, @code{vec_extract}, and @code{vec_insert}
22141 match array element order corresponding to the endianness of the
22142 target. That is, element zero identifies the leftmost element in a
22143 vector register when targeting a big-endian platform, and identifies
22144 the rightmost element in a vector register when targeting a
22145 little-endian platform.
22148 @opindex maltivec=be
22149 Generate AltiVec instructions using big-endian element order,
22150 regardless of whether the target is big- or little-endian. This is
22151 the default when targeting a big-endian platform.
22153 The element order is used to interpret element numbers in AltiVec
22154 intrinsics such as @code{vec_splat}, @code{vec_extract}, and
22155 @code{vec_insert}. By default, these match array element order
22156 corresponding to the endianness for the target.
22159 @opindex maltivec=le
22160 Generate AltiVec instructions using little-endian element order,
22161 regardless of whether the target is big- or little-endian. This is
22162 the default when targeting a little-endian platform. This option is
22163 currently ignored when targeting a big-endian platform.
22165 The element order is used to interpret element numbers in AltiVec
22166 intrinsics such as @code{vec_splat}, @code{vec_extract}, and
22167 @code{vec_insert}. By default, these match array element order
22168 corresponding to the endianness for the target.
22173 @opindex mno-vrsave
22174 Generate VRSAVE instructions when generating AltiVec code.
22177 @opindex msecure-plt
22178 Generate code that allows @command{ld} and @command{ld.so}
22179 to build executables and shared
22180 libraries with non-executable @code{.plt} and @code{.got} sections.
22182 32-bit SYSV ABI option.
22186 Generate code that uses a BSS @code{.plt} section that @command{ld.so}
22188 requires @code{.plt} and @code{.got}
22189 sections that are both writable and executable.
22190 This is a PowerPC 32-bit SYSV ABI option.
22196 This switch enables or disables the generation of ISEL instructions.
22198 @item -misel=@var{yes/no}
22199 This switch has been deprecated. Use @option{-misel} and
22200 @option{-mno-isel} instead.
22206 This switch enables or disables the generation of SPE simd
22212 @opindex mno-paired
22213 This switch enables or disables the generation of PAIRED simd
22216 @item -mspe=@var{yes/no}
22217 This option has been deprecated. Use @option{-mspe} and
22218 @option{-mno-spe} instead.
22224 Generate code that uses (does not use) vector/scalar (VSX)
22225 instructions, and also enable the use of built-in functions that allow
22226 more direct access to the VSX instruction set.
22231 @opindex mno-crypto
22232 Enable the use (disable) of the built-in functions that allow direct
22233 access to the cryptographic instructions that were added in version
22234 2.07 of the PowerPC ISA.
22236 @item -mdirect-move
22237 @itemx -mno-direct-move
22238 @opindex mdirect-move
22239 @opindex mno-direct-move
22240 Generate code that uses (does not use) the instructions to move data
22241 between the general purpose registers and the vector/scalar (VSX)
22242 registers that were added in version 2.07 of the PowerPC ISA.
22248 Enable (disable) the use of the built-in functions that allow direct
22249 access to the Hardware Transactional Memory (HTM) instructions that
22250 were added in version 2.07 of the PowerPC ISA.
22252 @item -mpower8-fusion
22253 @itemx -mno-power8-fusion
22254 @opindex mpower8-fusion
22255 @opindex mno-power8-fusion
22256 Generate code that keeps (does not keeps) some integer operations
22257 adjacent so that the instructions can be fused together on power8 and
22260 @item -mpower8-vector
22261 @itemx -mno-power8-vector
22262 @opindex mpower8-vector
22263 @opindex mno-power8-vector
22264 Generate code that uses (does not use) the vector and scalar
22265 instructions that were added in version 2.07 of the PowerPC ISA. Also
22266 enable the use of built-in functions that allow more direct access to
22267 the vector instructions.
22269 @item -mquad-memory
22270 @itemx -mno-quad-memory
22271 @opindex mquad-memory
22272 @opindex mno-quad-memory
22273 Generate code that uses (does not use) the non-atomic quad word memory
22274 instructions. The @option{-mquad-memory} option requires use of
22277 @item -mquad-memory-atomic
22278 @itemx -mno-quad-memory-atomic
22279 @opindex mquad-memory-atomic
22280 @opindex mno-quad-memory-atomic
22281 Generate code that uses (does not use) the atomic quad word memory
22282 instructions. The @option{-mquad-memory-atomic} option requires use of
22286 @itemx -mno-float128
22288 @opindex mno-float128
22289 Enable/disable the @var{__float128} keyword for IEEE 128-bit floating point
22290 and use either software emulation for IEEE 128-bit floating point or
22291 hardware instructions.
22293 The VSX instruction set (@option{-mvsx}, @option{-mcpu=power7},
22294 @option{-mcpu=power8}), or @option{-mcpu=power9} must be enabled to
22295 use the IEEE 128-bit floating point support. The IEEE 128-bit
22296 floating point support only works on PowerPC Linux systems.
22298 The default for @option{-mfloat128} is enabled on PowerPC Linux
22299 systems using the VSX instruction set, and disabled on other systems.
22301 If you use the ISA 3.0 instruction set (@option{-mpower9-vector} or
22302 @option{-mcpu=power9}) on a 64-bit system, the IEEE 128-bit floating
22303 point support will also enable the generation of ISA 3.0 IEEE 128-bit
22304 floating point instructions. Otherwise, if you do not specify to
22305 generate ISA 3.0 instructions or you are targeting a 32-bit big endian
22306 system, IEEE 128-bit floating point will be done with software
22309 @item -mfloat128-hardware
22310 @itemx -mno-float128-hardware
22311 @opindex mfloat128-hardware
22312 @opindex mno-float128-hardware
22313 Enable/disable using ISA 3.0 hardware instructions to support the
22314 @var{__float128} data type.
22316 The default for @option{-mfloat128-hardware} is enabled on PowerPC
22317 Linux systems using the ISA 3.0 instruction set, and disabled on other
22320 @item -mfloat-gprs=@var{yes/single/double/no}
22321 @itemx -mfloat-gprs
22322 @opindex mfloat-gprs
22323 This switch enables or disables the generation of floating-point
22324 operations on the general-purpose registers for architectures that
22327 The argument @samp{yes} or @samp{single} enables the use of
22328 single-precision floating-point operations.
22330 The argument @samp{double} enables the use of single and
22331 double-precision floating-point operations.
22333 The argument @samp{no} disables floating-point operations on the
22334 general-purpose registers.
22336 This option is currently only available on the MPC854x.
22342 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
22343 targets (including GNU/Linux). The 32-bit environment sets int, long
22344 and pointer to 32 bits and generates code that runs on any PowerPC
22345 variant. The 64-bit environment sets int to 32 bits and long and
22346 pointer to 64 bits, and generates code for PowerPC64, as for
22347 @option{-mpowerpc64}.
22350 @itemx -mno-fp-in-toc
22351 @itemx -mno-sum-in-toc
22352 @itemx -mminimal-toc
22354 @opindex mno-fp-in-toc
22355 @opindex mno-sum-in-toc
22356 @opindex mminimal-toc
22357 Modify generation of the TOC (Table Of Contents), which is created for
22358 every executable file. The @option{-mfull-toc} option is selected by
22359 default. In that case, GCC allocates at least one TOC entry for
22360 each unique non-automatic variable reference in your program. GCC
22361 also places floating-point constants in the TOC@. However, only
22362 16,384 entries are available in the TOC@.
22364 If you receive a linker error message that saying you have overflowed
22365 the available TOC space, you can reduce the amount of TOC space used
22366 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
22367 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
22368 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
22369 generate code to calculate the sum of an address and a constant at
22370 run time instead of putting that sum into the TOC@. You may specify one
22371 or both of these options. Each causes GCC to produce very slightly
22372 slower and larger code at the expense of conserving TOC space.
22374 If you still run out of space in the TOC even when you specify both of
22375 these options, specify @option{-mminimal-toc} instead. This option causes
22376 GCC to make only one TOC entry for every file. When you specify this
22377 option, GCC produces code that is slower and larger but which
22378 uses extremely little TOC space. You may wish to use this option
22379 only on files that contain less frequently-executed code.
22385 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
22386 @code{long} type, and the infrastructure needed to support them.
22387 Specifying @option{-maix64} implies @option{-mpowerpc64},
22388 while @option{-maix32} disables the 64-bit ABI and
22389 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
22392 @itemx -mno-xl-compat
22393 @opindex mxl-compat
22394 @opindex mno-xl-compat
22395 Produce code that conforms more closely to IBM XL compiler semantics
22396 when using AIX-compatible ABI@. Pass floating-point arguments to
22397 prototyped functions beyond the register save area (RSA) on the stack
22398 in addition to argument FPRs. Do not assume that most significant
22399 double in 128-bit long double value is properly rounded when comparing
22400 values and converting to double. Use XL symbol names for long double
22403 The AIX calling convention was extended but not initially documented to
22404 handle an obscure K&R C case of calling a function that takes the
22405 address of its arguments with fewer arguments than declared. IBM XL
22406 compilers access floating-point arguments that do not fit in the
22407 RSA from the stack when a subroutine is compiled without
22408 optimization. Because always storing floating-point arguments on the
22409 stack is inefficient and rarely needed, this option is not enabled by
22410 default and only is necessary when calling subroutines compiled by IBM
22411 XL compilers without optimization.
22415 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
22416 application written to use message passing with special startup code to
22417 enable the application to run. The system must have PE installed in the
22418 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
22419 must be overridden with the @option{-specs=} option to specify the
22420 appropriate directory location. The Parallel Environment does not
22421 support threads, so the @option{-mpe} option and the @option{-pthread}
22422 option are incompatible.
22424 @item -malign-natural
22425 @itemx -malign-power
22426 @opindex malign-natural
22427 @opindex malign-power
22428 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
22429 @option{-malign-natural} overrides the ABI-defined alignment of larger
22430 types, such as floating-point doubles, on their natural size-based boundary.
22431 The option @option{-malign-power} instructs GCC to follow the ABI-specified
22432 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
22434 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
22438 @itemx -mhard-float
22439 @opindex msoft-float
22440 @opindex mhard-float
22441 Generate code that does not use (uses) the floating-point register set.
22442 Software floating-point emulation is provided if you use the
22443 @option{-msoft-float} option, and pass the option to GCC when linking.
22445 @item -msingle-float
22446 @itemx -mdouble-float
22447 @opindex msingle-float
22448 @opindex mdouble-float
22449 Generate code for single- or double-precision floating-point operations.
22450 @option{-mdouble-float} implies @option{-msingle-float}.
22453 @opindex msimple-fpu
22454 Do not generate @code{sqrt} and @code{div} instructions for hardware
22455 floating-point unit.
22457 @item -mfpu=@var{name}
22459 Specify type of floating-point unit. Valid values for @var{name} are
22460 @samp{sp_lite} (equivalent to @option{-msingle-float -msimple-fpu}),
22461 @samp{dp_lite} (equivalent to @option{-mdouble-float -msimple-fpu}),
22462 @samp{sp_full} (equivalent to @option{-msingle-float}),
22463 and @samp{dp_full} (equivalent to @option{-mdouble-float}).
22466 @opindex mxilinx-fpu
22467 Perform optimizations for the floating-point unit on Xilinx PPC 405/440.
22470 @itemx -mno-multiple
22472 @opindex mno-multiple
22473 Generate code that uses (does not use) the load multiple word
22474 instructions and the store multiple word instructions. These
22475 instructions are generated by default on POWER systems, and not
22476 generated on PowerPC systems. Do not use @option{-mmultiple} on little-endian
22477 PowerPC systems, since those instructions do not work when the
22478 processor is in little-endian mode. The exceptions are PPC740 and
22479 PPC750 which permit these instructions in little-endian mode.
22484 @opindex mno-string
22485 Generate code that uses (does not use) the load string instructions
22486 and the store string word instructions to save multiple registers and
22487 do small block moves. These instructions are generated by default on
22488 POWER systems, and not generated on PowerPC systems. Do not use
22489 @option{-mstring} on little-endian PowerPC systems, since those
22490 instructions do not work when the processor is in little-endian mode.
22491 The exceptions are PPC740 and PPC750 which permit these instructions
22492 in little-endian mode.
22497 @opindex mno-update
22498 Generate code that uses (does not use) the load or store instructions
22499 that update the base register to the address of the calculated memory
22500 location. These instructions are generated by default. If you use
22501 @option{-mno-update}, there is a small window between the time that the
22502 stack pointer is updated and the address of the previous frame is
22503 stored, which means code that walks the stack frame across interrupts or
22504 signals may get corrupted data.
22506 @item -mavoid-indexed-addresses
22507 @itemx -mno-avoid-indexed-addresses
22508 @opindex mavoid-indexed-addresses
22509 @opindex mno-avoid-indexed-addresses
22510 Generate code that tries to avoid (not avoid) the use of indexed load
22511 or store instructions. These instructions can incur a performance
22512 penalty on Power6 processors in certain situations, such as when
22513 stepping through large arrays that cross a 16M boundary. This option
22514 is enabled by default when targeting Power6 and disabled otherwise.
22517 @itemx -mno-fused-madd
22518 @opindex mfused-madd
22519 @opindex mno-fused-madd
22520 Generate code that uses (does not use) the floating-point multiply and
22521 accumulate instructions. These instructions are generated by default
22522 if hardware floating point is used. The machine-dependent
22523 @option{-mfused-madd} option is now mapped to the machine-independent
22524 @option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
22525 mapped to @option{-ffp-contract=off}.
22531 Generate code that uses (does not use) the half-word multiply and
22532 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
22533 These instructions are generated by default when targeting those
22540 Generate code that uses (does not use) the string-search @samp{dlmzb}
22541 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
22542 generated by default when targeting those processors.
22544 @item -mno-bit-align
22546 @opindex mno-bit-align
22547 @opindex mbit-align
22548 On System V.4 and embedded PowerPC systems do not (do) force structures
22549 and unions that contain bit-fields to be aligned to the base type of the
22552 For example, by default a structure containing nothing but 8
22553 @code{unsigned} bit-fields of length 1 is aligned to a 4-byte
22554 boundary and has a size of 4 bytes. By using @option{-mno-bit-align},
22555 the structure is aligned to a 1-byte boundary and is 1 byte in
22558 @item -mno-strict-align
22559 @itemx -mstrict-align
22560 @opindex mno-strict-align
22561 @opindex mstrict-align
22562 On System V.4 and embedded PowerPC systems do not (do) assume that
22563 unaligned memory references are handled by the system.
22565 @item -mrelocatable
22566 @itemx -mno-relocatable
22567 @opindex mrelocatable
22568 @opindex mno-relocatable
22569 Generate code that allows (does not allow) a static executable to be
22570 relocated to a different address at run time. A simple embedded
22571 PowerPC system loader should relocate the entire contents of
22572 @code{.got2} and 4-byte locations listed in the @code{.fixup} section,
22573 a table of 32-bit addresses generated by this option. For this to
22574 work, all objects linked together must be compiled with
22575 @option{-mrelocatable} or @option{-mrelocatable-lib}.
22576 @option{-mrelocatable} code aligns the stack to an 8-byte boundary.
22578 @item -mrelocatable-lib
22579 @itemx -mno-relocatable-lib
22580 @opindex mrelocatable-lib
22581 @opindex mno-relocatable-lib
22582 Like @option{-mrelocatable}, @option{-mrelocatable-lib} generates a
22583 @code{.fixup} section to allow static executables to be relocated at
22584 run time, but @option{-mrelocatable-lib} does not use the smaller stack
22585 alignment of @option{-mrelocatable}. Objects compiled with
22586 @option{-mrelocatable-lib} may be linked with objects compiled with
22587 any combination of the @option{-mrelocatable} options.
22593 On System V.4 and embedded PowerPC systems do not (do) assume that
22594 register 2 contains a pointer to a global area pointing to the addresses
22595 used in the program.
22598 @itemx -mlittle-endian
22600 @opindex mlittle-endian
22601 On System V.4 and embedded PowerPC systems compile code for the
22602 processor in little-endian mode. The @option{-mlittle-endian} option is
22603 the same as @option{-mlittle}.
22606 @itemx -mbig-endian
22608 @opindex mbig-endian
22609 On System V.4 and embedded PowerPC systems compile code for the
22610 processor in big-endian mode. The @option{-mbig-endian} option is
22611 the same as @option{-mbig}.
22613 @item -mdynamic-no-pic
22614 @opindex mdynamic-no-pic
22615 On Darwin and Mac OS X systems, compile code so that it is not
22616 relocatable, but that its external references are relocatable. The
22617 resulting code is suitable for applications, but not shared
22620 @item -msingle-pic-base
22621 @opindex msingle-pic-base
22622 Treat the register used for PIC addressing as read-only, rather than
22623 loading it in the prologue for each function. The runtime system is
22624 responsible for initializing this register with an appropriate value
22625 before execution begins.
22627 @item -mprioritize-restricted-insns=@var{priority}
22628 @opindex mprioritize-restricted-insns
22629 This option controls the priority that is assigned to
22630 dispatch-slot restricted instructions during the second scheduling
22631 pass. The argument @var{priority} takes the value @samp{0}, @samp{1},
22632 or @samp{2} to assign no, highest, or second-highest (respectively)
22633 priority to dispatch-slot restricted
22636 @item -msched-costly-dep=@var{dependence_type}
22637 @opindex msched-costly-dep
22638 This option controls which dependences are considered costly
22639 by the target during instruction scheduling. The argument
22640 @var{dependence_type} takes one of the following values:
22644 No dependence is costly.
22647 All dependences are costly.
22649 @item @samp{true_store_to_load}
22650 A true dependence from store to load is costly.
22652 @item @samp{store_to_load}
22653 Any dependence from store to load is costly.
22656 Any dependence for which the latency is greater than or equal to
22657 @var{number} is costly.
22660 @item -minsert-sched-nops=@var{scheme}
22661 @opindex minsert-sched-nops
22662 This option controls which NOP insertion scheme is used during
22663 the second scheduling pass. The argument @var{scheme} takes one of the
22671 Pad with NOPs any dispatch group that has vacant issue slots,
22672 according to the scheduler's grouping.
22674 @item @samp{regroup_exact}
22675 Insert NOPs to force costly dependent insns into
22676 separate groups. Insert exactly as many NOPs as needed to force an insn
22677 to a new group, according to the estimated processor grouping.
22680 Insert NOPs to force costly dependent insns into
22681 separate groups. Insert @var{number} NOPs to force an insn to a new group.
22685 @opindex mcall-sysv
22686 On System V.4 and embedded PowerPC systems compile code using calling
22687 conventions that adhere to the March 1995 draft of the System V
22688 Application Binary Interface, PowerPC processor supplement. This is the
22689 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
22691 @item -mcall-sysv-eabi
22693 @opindex mcall-sysv-eabi
22694 @opindex mcall-eabi
22695 Specify both @option{-mcall-sysv} and @option{-meabi} options.
22697 @item -mcall-sysv-noeabi
22698 @opindex mcall-sysv-noeabi
22699 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
22701 @item -mcall-aixdesc
22703 On System V.4 and embedded PowerPC systems compile code for the AIX
22707 @opindex mcall-linux
22708 On System V.4 and embedded PowerPC systems compile code for the
22709 Linux-based GNU system.
22711 @item -mcall-freebsd
22712 @opindex mcall-freebsd
22713 On System V.4 and embedded PowerPC systems compile code for the
22714 FreeBSD operating system.
22716 @item -mcall-netbsd
22717 @opindex mcall-netbsd
22718 On System V.4 and embedded PowerPC systems compile code for the
22719 NetBSD operating system.
22721 @item -mcall-openbsd
22722 @opindex mcall-netbsd
22723 On System V.4 and embedded PowerPC systems compile code for the
22724 OpenBSD operating system.
22726 @item -maix-struct-return
22727 @opindex maix-struct-return
22728 Return all structures in memory (as specified by the AIX ABI)@.
22730 @item -msvr4-struct-return
22731 @opindex msvr4-struct-return
22732 Return structures smaller than 8 bytes in registers (as specified by the
22735 @item -mabi=@var{abi-type}
22737 Extend the current ABI with a particular extension, or remove such extension.
22738 Valid values are @samp{altivec}, @samp{no-altivec}, @samp{spe},
22739 @samp{no-spe}, @samp{ibmlongdouble}, @samp{ieeelongdouble},
22740 @samp{elfv1}, @samp{elfv2}@.
22744 Extend the current ABI with SPE ABI extensions. This does not change
22745 the default ABI, instead it adds the SPE ABI extensions to the current
22749 @opindex mabi=no-spe
22750 Disable Book-E SPE ABI extensions for the current ABI@.
22752 @item -mabi=ibmlongdouble
22753 @opindex mabi=ibmlongdouble
22754 Change the current ABI to use IBM extended-precision long double.
22755 This is not likely to work if your system defaults to using IEEE
22756 extended-precision long double. If you change the long double type
22757 from IEEE extended-precision, the compiler will issue a warning unless
22758 you use the @option{-Wno-psabi} option.
22760 @item -mabi=ieeelongdouble
22761 @opindex mabi=ieeelongdouble
22762 Change the current ABI to use IEEE extended-precision long double.
22763 This is not likely to work if your system defaults to using IBM
22764 extended-precision long double. If you change the long double type
22765 from IBM extended-precision, the compiler will issue a warning unless
22766 you use the @option{-Wno-psabi} option.
22769 @opindex mabi=elfv1
22770 Change the current ABI to use the ELFv1 ABI.
22771 This is the default ABI for big-endian PowerPC 64-bit Linux.
22772 Overriding the default ABI requires special system support and is
22773 likely to fail in spectacular ways.
22776 @opindex mabi=elfv2
22777 Change the current ABI to use the ELFv2 ABI.
22778 This is the default ABI for little-endian PowerPC 64-bit Linux.
22779 Overriding the default ABI requires special system support and is
22780 likely to fail in spectacular ways.
22782 @item -mgnu-attribute
22783 @itemx -mno-gnu-attribute
22784 @opindex mgnu-attribute
22785 @opindex mno-gnu-attribute
22786 Emit .gnu_attribute assembly directives to set tag/value pairs in a
22787 .gnu.attributes section that specify ABI variations in function
22788 parameters or return values.
22791 @itemx -mno-prototype
22792 @opindex mprototype
22793 @opindex mno-prototype
22794 On System V.4 and embedded PowerPC systems assume that all calls to
22795 variable argument functions are properly prototyped. Otherwise, the
22796 compiler must insert an instruction before every non-prototyped call to
22797 set or clear bit 6 of the condition code register (@code{CR}) to
22798 indicate whether floating-point values are passed in the floating-point
22799 registers in case the function takes variable arguments. With
22800 @option{-mprototype}, only calls to prototyped variable argument functions
22801 set or clear the bit.
22805 On embedded PowerPC systems, assume that the startup module is called
22806 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
22807 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
22812 On embedded PowerPC systems, assume that the startup module is called
22813 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
22818 On embedded PowerPC systems, assume that the startup module is called
22819 @file{crt0.o} and the standard C libraries are @file{libads.a} and
22822 @item -myellowknife
22823 @opindex myellowknife
22824 On embedded PowerPC systems, assume that the startup module is called
22825 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
22830 On System V.4 and embedded PowerPC systems, specify that you are
22831 compiling for a VxWorks system.
22835 On embedded PowerPC systems, set the @code{PPC_EMB} bit in the ELF flags
22836 header to indicate that @samp{eabi} extended relocations are used.
22842 On System V.4 and embedded PowerPC systems do (do not) adhere to the
22843 Embedded Applications Binary Interface (EABI), which is a set of
22844 modifications to the System V.4 specifications. Selecting @option{-meabi}
22845 means that the stack is aligned to an 8-byte boundary, a function
22846 @code{__eabi} is called from @code{main} to set up the EABI
22847 environment, and the @option{-msdata} option can use both @code{r2} and
22848 @code{r13} to point to two separate small data areas. Selecting
22849 @option{-mno-eabi} means that the stack is aligned to a 16-byte boundary,
22850 no EABI initialization function is called from @code{main}, and the
22851 @option{-msdata} option only uses @code{r13} to point to a single
22852 small data area. The @option{-meabi} option is on by default if you
22853 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
22856 @opindex msdata=eabi
22857 On System V.4 and embedded PowerPC systems, put small initialized
22858 @code{const} global and static data in the @code{.sdata2} section, which
22859 is pointed to by register @code{r2}. Put small initialized
22860 non-@code{const} global and static data in the @code{.sdata} section,
22861 which is pointed to by register @code{r13}. Put small uninitialized
22862 global and static data in the @code{.sbss} section, which is adjacent to
22863 the @code{.sdata} section. The @option{-msdata=eabi} option is
22864 incompatible with the @option{-mrelocatable} option. The
22865 @option{-msdata=eabi} option also sets the @option{-memb} option.
22868 @opindex msdata=sysv
22869 On System V.4 and embedded PowerPC systems, put small global and static
22870 data in the @code{.sdata} section, which is pointed to by register
22871 @code{r13}. Put small uninitialized global and static data in the
22872 @code{.sbss} section, which is adjacent to the @code{.sdata} section.
22873 The @option{-msdata=sysv} option is incompatible with the
22874 @option{-mrelocatable} option.
22876 @item -msdata=default
22878 @opindex msdata=default
22880 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
22881 compile code the same as @option{-msdata=eabi}, otherwise compile code the
22882 same as @option{-msdata=sysv}.
22885 @opindex msdata=data
22886 On System V.4 and embedded PowerPC systems, put small global
22887 data in the @code{.sdata} section. Put small uninitialized global
22888 data in the @code{.sbss} section. Do not use register @code{r13}
22889 to address small data however. This is the default behavior unless
22890 other @option{-msdata} options are used.
22894 @opindex msdata=none
22896 On embedded PowerPC systems, put all initialized global and static data
22897 in the @code{.data} section, and all uninitialized data in the
22898 @code{.bss} section.
22900 @item -mblock-move-inline-limit=@var{num}
22901 @opindex mblock-move-inline-limit
22902 Inline all block moves (such as calls to @code{memcpy} or structure
22903 copies) less than or equal to @var{num} bytes. The minimum value for
22904 @var{num} is 32 bytes on 32-bit targets and 64 bytes on 64-bit
22905 targets. The default value is target-specific.
22909 @cindex smaller data references (PowerPC)
22910 @cindex .sdata/.sdata2 references (PowerPC)
22911 On embedded PowerPC systems, put global and static items less than or
22912 equal to @var{num} bytes into the small data or BSS sections instead of
22913 the normal data or BSS section. By default, @var{num} is 8. The
22914 @option{-G @var{num}} switch is also passed to the linker.
22915 All modules should be compiled with the same @option{-G @var{num}} value.
22918 @itemx -mno-regnames
22920 @opindex mno-regnames
22921 On System V.4 and embedded PowerPC systems do (do not) emit register
22922 names in the assembly language output using symbolic forms.
22925 @itemx -mno-longcall
22927 @opindex mno-longcall
22928 By default assume that all calls are far away so that a longer and more
22929 expensive calling sequence is required. This is required for calls
22930 farther than 32 megabytes (33,554,432 bytes) from the current location.
22931 A short call is generated if the compiler knows
22932 the call cannot be that far away. This setting can be overridden by
22933 the @code{shortcall} function attribute, or by @code{#pragma
22936 Some linkers are capable of detecting out-of-range calls and generating
22937 glue code on the fly. On these systems, long calls are unnecessary and
22938 generate slower code. As of this writing, the AIX linker can do this,
22939 as can the GNU linker for PowerPC/64. It is planned to add this feature
22940 to the GNU linker for 32-bit PowerPC systems as well.
22942 On Darwin/PPC systems, @code{#pragma longcall} generates @code{jbsr
22943 callee, L42}, plus a @dfn{branch island} (glue code). The two target
22944 addresses represent the callee and the branch island. The
22945 Darwin/PPC linker prefers the first address and generates a @code{bl
22946 callee} if the PPC @code{bl} instruction reaches the callee directly;
22947 otherwise, the linker generates @code{bl L42} to call the branch
22948 island. The branch island is appended to the body of the
22949 calling function; it computes the full 32-bit address of the callee
22952 On Mach-O (Darwin) systems, this option directs the compiler emit to
22953 the glue for every direct call, and the Darwin linker decides whether
22954 to use or discard it.
22956 In the future, GCC may ignore all longcall specifications
22957 when the linker is known to generate glue.
22959 @item -mtls-markers
22960 @itemx -mno-tls-markers
22961 @opindex mtls-markers
22962 @opindex mno-tls-markers
22963 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
22964 specifying the function argument. The relocation allows the linker to
22965 reliably associate function call with argument setup instructions for
22966 TLS optimization, which in turn allows GCC to better schedule the
22972 This option enables use of the reciprocal estimate and
22973 reciprocal square root estimate instructions with additional
22974 Newton-Raphson steps to increase precision instead of doing a divide or
22975 square root and divide for floating-point arguments. You should use
22976 the @option{-ffast-math} option when using @option{-mrecip} (or at
22977 least @option{-funsafe-math-optimizations},
22978 @option{-ffinite-math-only}, @option{-freciprocal-math} and
22979 @option{-fno-trapping-math}). Note that while the throughput of the
22980 sequence is generally higher than the throughput of the non-reciprocal
22981 instruction, the precision of the sequence can be decreased by up to 2
22982 ulp (i.e.@: the inverse of 1.0 equals 0.99999994) for reciprocal square
22985 @item -mrecip=@var{opt}
22986 @opindex mrecip=opt
22987 This option controls which reciprocal estimate instructions
22988 may be used. @var{opt} is a comma-separated list of options, which may
22989 be preceded by a @code{!} to invert the option:
22994 Enable all estimate instructions.
22997 Enable the default instructions, equivalent to @option{-mrecip}.
23000 Disable all estimate instructions, equivalent to @option{-mno-recip}.
23003 Enable the reciprocal approximation instructions for both
23004 single and double precision.
23007 Enable the single-precision reciprocal approximation instructions.
23010 Enable the double-precision reciprocal approximation instructions.
23013 Enable the reciprocal square root approximation instructions for both
23014 single and double precision.
23017 Enable the single-precision reciprocal square root approximation instructions.
23020 Enable the double-precision reciprocal square root approximation instructions.
23024 So, for example, @option{-mrecip=all,!rsqrtd} enables
23025 all of the reciprocal estimate instructions, except for the
23026 @code{FRSQRTE}, @code{XSRSQRTEDP}, and @code{XVRSQRTEDP} instructions
23027 which handle the double-precision reciprocal square root calculations.
23029 @item -mrecip-precision
23030 @itemx -mno-recip-precision
23031 @opindex mrecip-precision
23032 Assume (do not assume) that the reciprocal estimate instructions
23033 provide higher-precision estimates than is mandated by the PowerPC
23034 ABI. Selecting @option{-mcpu=power6}, @option{-mcpu=power7} or
23035 @option{-mcpu=power8} automatically selects @option{-mrecip-precision}.
23036 The double-precision square root estimate instructions are not generated by
23037 default on low-precision machines, since they do not provide an
23038 estimate that converges after three steps.
23040 @item -mveclibabi=@var{type}
23041 @opindex mveclibabi
23042 Specifies the ABI type to use for vectorizing intrinsics using an
23043 external library. The only type supported at present is @samp{mass},
23044 which specifies to use IBM's Mathematical Acceleration Subsystem
23045 (MASS) libraries for vectorizing intrinsics using external libraries.
23046 GCC currently emits calls to @code{acosd2}, @code{acosf4},
23047 @code{acoshd2}, @code{acoshf4}, @code{asind2}, @code{asinf4},
23048 @code{asinhd2}, @code{asinhf4}, @code{atan2d2}, @code{atan2f4},
23049 @code{atand2}, @code{atanf4}, @code{atanhd2}, @code{atanhf4},
23050 @code{cbrtd2}, @code{cbrtf4}, @code{cosd2}, @code{cosf4},
23051 @code{coshd2}, @code{coshf4}, @code{erfcd2}, @code{erfcf4},
23052 @code{erfd2}, @code{erff4}, @code{exp2d2}, @code{exp2f4},
23053 @code{expd2}, @code{expf4}, @code{expm1d2}, @code{expm1f4},
23054 @code{hypotd2}, @code{hypotf4}, @code{lgammad2}, @code{lgammaf4},
23055 @code{log10d2}, @code{log10f4}, @code{log1pd2}, @code{log1pf4},
23056 @code{log2d2}, @code{log2f4}, @code{logd2}, @code{logf4},
23057 @code{powd2}, @code{powf4}, @code{sind2}, @code{sinf4}, @code{sinhd2},
23058 @code{sinhf4}, @code{sqrtd2}, @code{sqrtf4}, @code{tand2},
23059 @code{tanf4}, @code{tanhd2}, and @code{tanhf4} when generating code
23060 for power7. Both @option{-ftree-vectorize} and
23061 @option{-funsafe-math-optimizations} must also be enabled. The MASS
23062 libraries must be specified at link time.
23067 Generate (do not generate) the @code{friz} instruction when the
23068 @option{-funsafe-math-optimizations} option is used to optimize
23069 rounding of floating-point values to 64-bit integer and back to floating
23070 point. The @code{friz} instruction does not return the same value if
23071 the floating-point number is too large to fit in an integer.
23073 @item -mpointers-to-nested-functions
23074 @itemx -mno-pointers-to-nested-functions
23075 @opindex mpointers-to-nested-functions
23076 Generate (do not generate) code to load up the static chain register
23077 (@code{r11}) when calling through a pointer on AIX and 64-bit Linux
23078 systems where a function pointer points to a 3-word descriptor giving
23079 the function address, TOC value to be loaded in register @code{r2}, and
23080 static chain value to be loaded in register @code{r11}. The
23081 @option{-mpointers-to-nested-functions} is on by default. You cannot
23082 call through pointers to nested functions or pointers
23083 to functions compiled in other languages that use the static chain if
23084 you use @option{-mno-pointers-to-nested-functions}.
23086 @item -msave-toc-indirect
23087 @itemx -mno-save-toc-indirect
23088 @opindex msave-toc-indirect
23089 Generate (do not generate) code to save the TOC value in the reserved
23090 stack location in the function prologue if the function calls through
23091 a pointer on AIX and 64-bit Linux systems. If the TOC value is not
23092 saved in the prologue, it is saved just before the call through the
23093 pointer. The @option{-mno-save-toc-indirect} option is the default.
23095 @item -mcompat-align-parm
23096 @itemx -mno-compat-align-parm
23097 @opindex mcompat-align-parm
23098 Generate (do not generate) code to pass structure parameters with a
23099 maximum alignment of 64 bits, for compatibility with older versions
23102 Older versions of GCC (prior to 4.9.0) incorrectly did not align a
23103 structure parameter on a 128-bit boundary when that structure contained
23104 a member requiring 128-bit alignment. This is corrected in more
23105 recent versions of GCC. This option may be used to generate code
23106 that is compatible with functions compiled with older versions of
23109 The @option{-mno-compat-align-parm} option is the default.
23111 @item -mstack-protector-guard=@var{guard}
23112 @itemx -mstack-protector-guard-reg=@var{reg}
23113 @itemx -mstack-protector-guard-offset=@var{offset}
23114 @itemx -mstack-protector-guard-symbol=@var{symbol}
23115 @opindex mstack-protector-guard
23116 @opindex mstack-protector-guard-reg
23117 @opindex mstack-protector-guard-offset
23118 @opindex mstack-protector-guard-symbol
23119 Generate stack protection code using canary at @var{guard}. Supported
23120 locations are @samp{global} for global canary or @samp{tls} for per-thread
23121 canary in the TLS block (the default with GNU libc version 2.4 or later).
23123 With the latter choice the options
23124 @option{-mstack-protector-guard-reg=@var{reg}} and
23125 @option{-mstack-protector-guard-offset=@var{offset}} furthermore specify
23126 which register to use as base register for reading the canary, and from what
23127 offset from that base register. The default for those is as specified in the
23128 relevant ABI. @option{-mstack-protector-guard-symbol=@var{symbol}} overrides
23129 the offset with a symbol reference to a canary in the TLS block.
23133 @subsection RX Options
23136 These command-line options are defined for RX targets:
23139 @item -m64bit-doubles
23140 @itemx -m32bit-doubles
23141 @opindex m64bit-doubles
23142 @opindex m32bit-doubles
23143 Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
23144 or 32 bits (@option{-m32bit-doubles}) in size. The default is
23145 @option{-m32bit-doubles}. @emph{Note} RX floating-point hardware only
23146 works on 32-bit values, which is why the default is
23147 @option{-m32bit-doubles}.
23153 Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
23154 floating-point hardware. The default is enabled for the RX600
23155 series and disabled for the RX200 series.
23157 Floating-point instructions are only generated for 32-bit floating-point
23158 values, however, so the FPU hardware is not used for doubles if the
23159 @option{-m64bit-doubles} option is used.
23161 @emph{Note} If the @option{-fpu} option is enabled then
23162 @option{-funsafe-math-optimizations} is also enabled automatically.
23163 This is because the RX FPU instructions are themselves unsafe.
23165 @item -mcpu=@var{name}
23167 Selects the type of RX CPU to be targeted. Currently three types are
23168 supported, the generic @samp{RX600} and @samp{RX200} series hardware and
23169 the specific @samp{RX610} CPU. The default is @samp{RX600}.
23171 The only difference between @samp{RX600} and @samp{RX610} is that the
23172 @samp{RX610} does not support the @code{MVTIPL} instruction.
23174 The @samp{RX200} series does not have a hardware floating-point unit
23175 and so @option{-nofpu} is enabled by default when this type is
23178 @item -mbig-endian-data
23179 @itemx -mlittle-endian-data
23180 @opindex mbig-endian-data
23181 @opindex mlittle-endian-data
23182 Store data (but not code) in the big-endian format. The default is
23183 @option{-mlittle-endian-data}, i.e.@: to store data in the little-endian
23186 @item -msmall-data-limit=@var{N}
23187 @opindex msmall-data-limit
23188 Specifies the maximum size in bytes of global and static variables
23189 which can be placed into the small data area. Using the small data
23190 area can lead to smaller and faster code, but the size of area is
23191 limited and it is up to the programmer to ensure that the area does
23192 not overflow. Also when the small data area is used one of the RX's
23193 registers (usually @code{r13}) is reserved for use pointing to this
23194 area, so it is no longer available for use by the compiler. This
23195 could result in slower and/or larger code if variables are pushed onto
23196 the stack instead of being held in this register.
23198 Note, common variables (variables that have not been initialized) and
23199 constants are not placed into the small data area as they are assigned
23200 to other sections in the output executable.
23202 The default value is zero, which disables this feature. Note, this
23203 feature is not enabled by default with higher optimization levels
23204 (@option{-O2} etc) because of the potentially detrimental effects of
23205 reserving a register. It is up to the programmer to experiment and
23206 discover whether this feature is of benefit to their program. See the
23207 description of the @option{-mpid} option for a description of how the
23208 actual register to hold the small data area pointer is chosen.
23214 Use the simulator runtime. The default is to use the libgloss
23215 board-specific runtime.
23217 @item -mas100-syntax
23218 @itemx -mno-as100-syntax
23219 @opindex mas100-syntax
23220 @opindex mno-as100-syntax
23221 When generating assembler output use a syntax that is compatible with
23222 Renesas's AS100 assembler. This syntax can also be handled by the GAS
23223 assembler, but it has some restrictions so it is not generated by default.
23225 @item -mmax-constant-size=@var{N}
23226 @opindex mmax-constant-size
23227 Specifies the maximum size, in bytes, of a constant that can be used as
23228 an operand in a RX instruction. Although the RX instruction set does
23229 allow constants of up to 4 bytes in length to be used in instructions,
23230 a longer value equates to a longer instruction. Thus in some
23231 circumstances it can be beneficial to restrict the size of constants
23232 that are used in instructions. Constants that are too big are instead
23233 placed into a constant pool and referenced via register indirection.
23235 The value @var{N} can be between 0 and 4. A value of 0 (the default)
23236 or 4 means that constants of any size are allowed.
23240 Enable linker relaxation. Linker relaxation is a process whereby the
23241 linker attempts to reduce the size of a program by finding shorter
23242 versions of various instructions. Disabled by default.
23244 @item -mint-register=@var{N}
23245 @opindex mint-register
23246 Specify the number of registers to reserve for fast interrupt handler
23247 functions. The value @var{N} can be between 0 and 4. A value of 1
23248 means that register @code{r13} is reserved for the exclusive use
23249 of fast interrupt handlers. A value of 2 reserves @code{r13} and
23250 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
23251 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
23252 A value of 0, the default, does not reserve any registers.
23254 @item -msave-acc-in-interrupts
23255 @opindex msave-acc-in-interrupts
23256 Specifies that interrupt handler functions should preserve the
23257 accumulator register. This is only necessary if normal code might use
23258 the accumulator register, for example because it performs 64-bit
23259 multiplications. The default is to ignore the accumulator as this
23260 makes the interrupt handlers faster.
23266 Enables the generation of position independent data. When enabled any
23267 access to constant data is done via an offset from a base address
23268 held in a register. This allows the location of constant data to be
23269 determined at run time without requiring the executable to be
23270 relocated, which is a benefit to embedded applications with tight
23271 memory constraints. Data that can be modified is not affected by this
23274 Note, using this feature reserves a register, usually @code{r13}, for
23275 the constant data base address. This can result in slower and/or
23276 larger code, especially in complicated functions.
23278 The actual register chosen to hold the constant data base address
23279 depends upon whether the @option{-msmall-data-limit} and/or the
23280 @option{-mint-register} command-line options are enabled. Starting
23281 with register @code{r13} and proceeding downwards, registers are
23282 allocated first to satisfy the requirements of @option{-mint-register},
23283 then @option{-mpid} and finally @option{-msmall-data-limit}. Thus it
23284 is possible for the small data area register to be @code{r8} if both
23285 @option{-mint-register=4} and @option{-mpid} are specified on the
23288 By default this feature is not enabled. The default can be restored
23289 via the @option{-mno-pid} command-line option.
23291 @item -mno-warn-multiple-fast-interrupts
23292 @itemx -mwarn-multiple-fast-interrupts
23293 @opindex mno-warn-multiple-fast-interrupts
23294 @opindex mwarn-multiple-fast-interrupts
23295 Prevents GCC from issuing a warning message if it finds more than one
23296 fast interrupt handler when it is compiling a file. The default is to
23297 issue a warning for each extra fast interrupt handler found, as the RX
23298 only supports one such interrupt.
23300 @item -mallow-string-insns
23301 @itemx -mno-allow-string-insns
23302 @opindex mallow-string-insns
23303 @opindex mno-allow-string-insns
23304 Enables or disables the use of the string manipulation instructions
23305 @code{SMOVF}, @code{SCMPU}, @code{SMOVB}, @code{SMOVU}, @code{SUNTIL}
23306 @code{SWHILE} and also the @code{RMPA} instruction. These
23307 instructions may prefetch data, which is not safe to do if accessing
23308 an I/O register. (See section 12.2.7 of the RX62N Group User's Manual
23309 for more information).
23311 The default is to allow these instructions, but it is not possible for
23312 GCC to reliably detect all circumstances where a string instruction
23313 might be used to access an I/O register, so their use cannot be
23314 disabled automatically. Instead it is reliant upon the programmer to
23315 use the @option{-mno-allow-string-insns} option if their program
23316 accesses I/O space.
23318 When the instructions are enabled GCC defines the C preprocessor
23319 symbol @code{__RX_ALLOW_STRING_INSNS__}, otherwise it defines the
23320 symbol @code{__RX_DISALLOW_STRING_INSNS__}.
23326 Use only (or not only) @code{JSR} instructions to access functions.
23327 This option can be used when code size exceeds the range of @code{BSR}
23328 instructions. Note that @option{-mno-jsr} does not mean to not use
23329 @code{JSR} but instead means that any type of branch may be used.
23332 @emph{Note:} The generic GCC command-line option @option{-ffixed-@var{reg}}
23333 has special significance to the RX port when used with the
23334 @code{interrupt} function attribute. This attribute indicates a
23335 function intended to process fast interrupts. GCC ensures
23336 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
23337 and/or @code{r13} and only provided that the normal use of the
23338 corresponding registers have been restricted via the
23339 @option{-ffixed-@var{reg}} or @option{-mint-register} command-line
23342 @node S/390 and zSeries Options
23343 @subsection S/390 and zSeries Options
23344 @cindex S/390 and zSeries Options
23346 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
23350 @itemx -msoft-float
23351 @opindex mhard-float
23352 @opindex msoft-float
23353 Use (do not use) the hardware floating-point instructions and registers
23354 for floating-point operations. When @option{-msoft-float} is specified,
23355 functions in @file{libgcc.a} are used to perform floating-point
23356 operations. When @option{-mhard-float} is specified, the compiler
23357 generates IEEE floating-point instructions. This is the default.
23360 @itemx -mno-hard-dfp
23362 @opindex mno-hard-dfp
23363 Use (do not use) the hardware decimal-floating-point instructions for
23364 decimal-floating-point operations. When @option{-mno-hard-dfp} is
23365 specified, functions in @file{libgcc.a} are used to perform
23366 decimal-floating-point operations. When @option{-mhard-dfp} is
23367 specified, the compiler generates decimal-floating-point hardware
23368 instructions. This is the default for @option{-march=z9-ec} or higher.
23370 @item -mlong-double-64
23371 @itemx -mlong-double-128
23372 @opindex mlong-double-64
23373 @opindex mlong-double-128
23374 These switches control the size of @code{long double} type. A size
23375 of 64 bits makes the @code{long double} type equivalent to the @code{double}
23376 type. This is the default.
23379 @itemx -mno-backchain
23380 @opindex mbackchain
23381 @opindex mno-backchain
23382 Store (do not store) the address of the caller's frame as backchain pointer
23383 into the callee's stack frame.
23384 A backchain may be needed to allow debugging using tools that do not understand
23385 DWARF call frame information.
23386 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
23387 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
23388 the backchain is placed into the topmost word of the 96/160 byte register
23391 In general, code compiled with @option{-mbackchain} is call-compatible with
23392 code compiled with @option{-mmo-backchain}; however, use of the backchain
23393 for debugging purposes usually requires that the whole binary is built with
23394 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
23395 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
23396 to build a linux kernel use @option{-msoft-float}.
23398 The default is to not maintain the backchain.
23400 @item -mpacked-stack
23401 @itemx -mno-packed-stack
23402 @opindex mpacked-stack
23403 @opindex mno-packed-stack
23404 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
23405 specified, the compiler uses the all fields of the 96/160 byte register save
23406 area only for their default purpose; unused fields still take up stack space.
23407 When @option{-mpacked-stack} is specified, register save slots are densely
23408 packed at the top of the register save area; unused space is reused for other
23409 purposes, allowing for more efficient use of the available stack space.
23410 However, when @option{-mbackchain} is also in effect, the topmost word of
23411 the save area is always used to store the backchain, and the return address
23412 register is always saved two words below the backchain.
23414 As long as the stack frame backchain is not used, code generated with
23415 @option{-mpacked-stack} is call-compatible with code generated with
23416 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
23417 S/390 or zSeries generated code that uses the stack frame backchain at run
23418 time, not just for debugging purposes. Such code is not call-compatible
23419 with code compiled with @option{-mpacked-stack}. Also, note that the
23420 combination of @option{-mbackchain},
23421 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
23422 to build a linux kernel use @option{-msoft-float}.
23424 The default is to not use the packed stack layout.
23427 @itemx -mno-small-exec
23428 @opindex msmall-exec
23429 @opindex mno-small-exec
23430 Generate (or do not generate) code using the @code{bras} instruction
23431 to do subroutine calls.
23432 This only works reliably if the total executable size does not
23433 exceed 64k. The default is to use the @code{basr} instruction instead,
23434 which does not have this limitation.
23440 When @option{-m31} is specified, generate code compliant to the
23441 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
23442 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
23443 particular to generate 64-bit instructions. For the @samp{s390}
23444 targets, the default is @option{-m31}, while the @samp{s390x}
23445 targets default to @option{-m64}.
23451 When @option{-mzarch} is specified, generate code using the
23452 instructions available on z/Architecture.
23453 When @option{-mesa} is specified, generate code using the
23454 instructions available on ESA/390. Note that @option{-mesa} is
23455 not possible with @option{-m64}.
23456 When generating code compliant to the GNU/Linux for S/390 ABI,
23457 the default is @option{-mesa}. When generating code compliant
23458 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
23464 The @option{-mhtm} option enables a set of builtins making use of
23465 instructions available with the transactional execution facility
23466 introduced with the IBM zEnterprise EC12 machine generation
23467 @ref{S/390 System z Built-in Functions}.
23468 @option{-mhtm} is enabled by default when using @option{-march=zEC12}.
23474 When @option{-mvx} is specified, generate code using the instructions
23475 available with the vector extension facility introduced with the IBM
23476 z13 machine generation.
23477 This option changes the ABI for some vector type values with regard to
23478 alignment and calling conventions. In case vector type values are
23479 being used in an ABI-relevant context a GAS @samp{.gnu_attribute}
23480 command will be added to mark the resulting binary with the ABI used.
23481 @option{-mvx} is enabled by default when using @option{-march=z13}.
23484 @itemx -mno-zvector
23486 @opindex mno-zvector
23487 The @option{-mzvector} option enables vector language extensions and
23488 builtins using instructions available with the vector extension
23489 facility introduced with the IBM z13 machine generation.
23490 This option adds support for @samp{vector} to be used as a keyword to
23491 define vector type variables and arguments. @samp{vector} is only
23492 available when GNU extensions are enabled. It will not be expanded
23493 when requesting strict standard compliance e.g. with @option{-std=c99}.
23494 In addition to the GCC low-level builtins @option{-mzvector} enables
23495 a set of builtins added for compatibility with AltiVec-style
23496 implementations like Power and Cell. In order to make use of these
23497 builtins the header file @file{vecintrin.h} needs to be included.
23498 @option{-mzvector} is disabled by default.
23504 Generate (or do not generate) code using the @code{mvcle} instruction
23505 to perform block moves. When @option{-mno-mvcle} is specified,
23506 use a @code{mvc} loop instead. This is the default unless optimizing for
23513 Print (or do not print) additional debug information when compiling.
23514 The default is to not print debug information.
23516 @item -march=@var{cpu-type}
23518 Generate code that runs on @var{cpu-type}, which is the name of a
23519 system representing a certain processor type. Possible values for
23520 @var{cpu-type} are @samp{z900}/@samp{arch5}, @samp{z990}/@samp{arch6},
23521 @samp{z9-109}, @samp{z9-ec}/@samp{arch7}, @samp{z10}/@samp{arch8},
23522 @samp{z196}/@samp{arch9}, @samp{zEC12}, @samp{z13}/@samp{arch11}, and
23525 The default is @option{-march=z900}. @samp{g5}/@samp{arch3} and
23526 @samp{g6} are deprecated and will be removed with future releases.
23528 Specifying @samp{native} as cpu type can be used to select the best
23529 architecture option for the host processor.
23530 @option{-march=native} has no effect if GCC does not recognize the
23533 @item -mtune=@var{cpu-type}
23535 Tune to @var{cpu-type} everything applicable about the generated code,
23536 except for the ABI and the set of available instructions.
23537 The list of @var{cpu-type} values is the same as for @option{-march}.
23538 The default is the value used for @option{-march}.
23541 @itemx -mno-tpf-trace
23542 @opindex mtpf-trace
23543 @opindex mno-tpf-trace
23544 Generate code that adds (does not add) in TPF OS specific branches to trace
23545 routines in the operating system. This option is off by default, even
23546 when compiling for the TPF OS@.
23549 @itemx -mno-fused-madd
23550 @opindex mfused-madd
23551 @opindex mno-fused-madd
23552 Generate code that uses (does not use) the floating-point multiply and
23553 accumulate instructions. These instructions are generated by default if
23554 hardware floating point is used.
23556 @item -mwarn-framesize=@var{framesize}
23557 @opindex mwarn-framesize
23558 Emit a warning if the current function exceeds the given frame size. Because
23559 this is a compile-time check it doesn't need to be a real problem when the program
23560 runs. It is intended to identify functions that most probably cause
23561 a stack overflow. It is useful to be used in an environment with limited stack
23562 size e.g.@: the linux kernel.
23564 @item -mwarn-dynamicstack
23565 @opindex mwarn-dynamicstack
23566 Emit a warning if the function calls @code{alloca} or uses dynamically-sized
23567 arrays. This is generally a bad idea with a limited stack size.
23569 @item -mstack-guard=@var{stack-guard}
23570 @itemx -mstack-size=@var{stack-size}
23571 @opindex mstack-guard
23572 @opindex mstack-size
23573 If these options are provided the S/390 back end emits additional instructions in
23574 the function prologue that trigger a trap if the stack size is @var{stack-guard}
23575 bytes above the @var{stack-size} (remember that the stack on S/390 grows downward).
23576 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
23577 the frame size of the compiled function is chosen.
23578 These options are intended to be used to help debugging stack overflow problems.
23579 The additionally emitted code causes only little overhead and hence can also be
23580 used in production-like systems without greater performance degradation. The given
23581 values have to be exact powers of 2 and @var{stack-size} has to be greater than
23582 @var{stack-guard} without exceeding 64k.
23583 In order to be efficient the extra code makes the assumption that the stack starts
23584 at an address aligned to the value given by @var{stack-size}.
23585 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
23587 @item -mhotpatch=@var{pre-halfwords},@var{post-halfwords}
23589 If the hotpatch option is enabled, a ``hot-patching'' function
23590 prologue is generated for all functions in the compilation unit.
23591 The funtion label is prepended with the given number of two-byte
23592 NOP instructions (@var{pre-halfwords}, maximum 1000000). After
23593 the label, 2 * @var{post-halfwords} bytes are appended, using the
23594 largest NOP like instructions the architecture allows (maximum
23597 If both arguments are zero, hotpatching is disabled.
23599 This option can be overridden for individual functions with the
23600 @code{hotpatch} attribute.
23603 @node Score Options
23604 @subsection Score Options
23605 @cindex Score Options
23607 These options are defined for Score implementations:
23612 Compile code for big-endian mode. This is the default.
23616 Compile code for little-endian mode.
23620 Disable generation of @code{bcnz} instructions.
23624 Enable generation of unaligned load and store instructions.
23628 Enable the use of multiply-accumulate instructions. Disabled by default.
23632 Specify the SCORE5 as the target architecture.
23636 Specify the SCORE5U of the target architecture.
23640 Specify the SCORE7 as the target architecture. This is the default.
23644 Specify the SCORE7D as the target architecture.
23648 @subsection SH Options
23650 These @samp{-m} options are defined for the SH implementations:
23655 Generate code for the SH1.
23659 Generate code for the SH2.
23662 Generate code for the SH2e.
23666 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
23667 that the floating-point unit is not used.
23669 @item -m2a-single-only
23670 @opindex m2a-single-only
23671 Generate code for the SH2a-FPU, in such a way that no double-precision
23672 floating-point operations are used.
23675 @opindex m2a-single
23676 Generate code for the SH2a-FPU assuming the floating-point unit is in
23677 single-precision mode by default.
23681 Generate code for the SH2a-FPU assuming the floating-point unit is in
23682 double-precision mode by default.
23686 Generate code for the SH3.
23690 Generate code for the SH3e.
23694 Generate code for the SH4 without a floating-point unit.
23696 @item -m4-single-only
23697 @opindex m4-single-only
23698 Generate code for the SH4 with a floating-point unit that only
23699 supports single-precision arithmetic.
23703 Generate code for the SH4 assuming the floating-point unit is in
23704 single-precision mode by default.
23708 Generate code for the SH4.
23712 Generate code for SH4-100.
23714 @item -m4-100-nofpu
23715 @opindex m4-100-nofpu
23716 Generate code for SH4-100 in such a way that the
23717 floating-point unit is not used.
23719 @item -m4-100-single
23720 @opindex m4-100-single
23721 Generate code for SH4-100 assuming the floating-point unit is in
23722 single-precision mode by default.
23724 @item -m4-100-single-only
23725 @opindex m4-100-single-only
23726 Generate code for SH4-100 in such a way that no double-precision
23727 floating-point operations are used.
23731 Generate code for SH4-200.
23733 @item -m4-200-nofpu
23734 @opindex m4-200-nofpu
23735 Generate code for SH4-200 without in such a way that the
23736 floating-point unit is not used.
23738 @item -m4-200-single
23739 @opindex m4-200-single
23740 Generate code for SH4-200 assuming the floating-point unit is in
23741 single-precision mode by default.
23743 @item -m4-200-single-only
23744 @opindex m4-200-single-only
23745 Generate code for SH4-200 in such a way that no double-precision
23746 floating-point operations are used.
23750 Generate code for SH4-300.
23752 @item -m4-300-nofpu
23753 @opindex m4-300-nofpu
23754 Generate code for SH4-300 without in such a way that the
23755 floating-point unit is not used.
23757 @item -m4-300-single
23758 @opindex m4-300-single
23759 Generate code for SH4-300 in such a way that no double-precision
23760 floating-point operations are used.
23762 @item -m4-300-single-only
23763 @opindex m4-300-single-only
23764 Generate code for SH4-300 in such a way that no double-precision
23765 floating-point operations are used.
23769 Generate code for SH4-340 (no MMU, no FPU).
23773 Generate code for SH4-500 (no FPU). Passes @option{-isa=sh4-nofpu} to the
23778 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
23779 floating-point unit is not used.
23781 @item -m4a-single-only
23782 @opindex m4a-single-only
23783 Generate code for the SH4a, in such a way that no double-precision
23784 floating-point operations are used.
23787 @opindex m4a-single
23788 Generate code for the SH4a assuming the floating-point unit is in
23789 single-precision mode by default.
23793 Generate code for the SH4a.
23797 Same as @option{-m4a-nofpu}, except that it implicitly passes
23798 @option{-dsp} to the assembler. GCC doesn't generate any DSP
23799 instructions at the moment.
23803 Compile code for the processor in big-endian mode.
23807 Compile code for the processor in little-endian mode.
23811 Align doubles at 64-bit boundaries. Note that this changes the calling
23812 conventions, and thus some functions from the standard C library do
23813 not work unless you recompile it first with @option{-mdalign}.
23817 Shorten some address references at link time, when possible; uses the
23818 linker option @option{-relax}.
23822 Use 32-bit offsets in @code{switch} tables. The default is to use
23827 Enable the use of bit manipulation instructions on SH2A.
23831 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
23832 alignment constraints.
23836 Comply with the calling conventions defined by Renesas.
23839 @opindex mno-renesas
23840 Comply with the calling conventions defined for GCC before the Renesas
23841 conventions were available. This option is the default for all
23842 targets of the SH toolchain.
23845 @opindex mnomacsave
23846 Mark the @code{MAC} register as call-clobbered, even if
23847 @option{-mrenesas} is given.
23853 Control the IEEE compliance of floating-point comparisons, which affects the
23854 handling of cases where the result of a comparison is unordered. By default
23855 @option{-mieee} is implicitly enabled. If @option{-ffinite-math-only} is
23856 enabled @option{-mno-ieee} is implicitly set, which results in faster
23857 floating-point greater-equal and less-equal comparisons. The implicit settings
23858 can be overridden by specifying either @option{-mieee} or @option{-mno-ieee}.
23860 @item -minline-ic_invalidate
23861 @opindex minline-ic_invalidate
23862 Inline code to invalidate instruction cache entries after setting up
23863 nested function trampolines.
23864 This option has no effect if @option{-musermode} is in effect and the selected
23865 code generation option (e.g. @option{-m4}) does not allow the use of the @code{icbi}
23867 If the selected code generation option does not allow the use of the @code{icbi}
23868 instruction, and @option{-musermode} is not in effect, the inlined code
23869 manipulates the instruction cache address array directly with an associative
23870 write. This not only requires privileged mode at run time, but it also
23871 fails if the cache line had been mapped via the TLB and has become unmapped.
23875 Dump instruction size and location in the assembly code.
23878 @opindex mpadstruct
23879 This option is deprecated. It pads structures to multiple of 4 bytes,
23880 which is incompatible with the SH ABI@.
23882 @item -matomic-model=@var{model}
23883 @opindex matomic-model=@var{model}
23884 Sets the model of atomic operations and additional parameters as a comma
23885 separated list. For details on the atomic built-in functions see
23886 @ref{__atomic Builtins}. The following models and parameters are supported:
23891 Disable compiler generated atomic sequences and emit library calls for atomic
23892 operations. This is the default if the target is not @code{sh*-*-linux*}.
23895 Generate GNU/Linux compatible gUSA software atomic sequences for the atomic
23896 built-in functions. The generated atomic sequences require additional support
23897 from the interrupt/exception handling code of the system and are only suitable
23898 for SH3* and SH4* single-core systems. This option is enabled by default when
23899 the target is @code{sh*-*-linux*} and SH3* or SH4*. When the target is SH4A,
23900 this option also partially utilizes the hardware atomic instructions
23901 @code{movli.l} and @code{movco.l} to create more efficient code, unless
23902 @samp{strict} is specified.
23905 Generate software atomic sequences that use a variable in the thread control
23906 block. This is a variation of the gUSA sequences which can also be used on
23907 SH1* and SH2* targets. The generated atomic sequences require additional
23908 support from the interrupt/exception handling code of the system and are only
23909 suitable for single-core systems. When using this model, the @samp{gbr-offset=}
23910 parameter has to be specified as well.
23913 Generate software atomic sequences that temporarily disable interrupts by
23914 setting @code{SR.IMASK = 1111}. This model works only when the program runs
23915 in privileged mode and is only suitable for single-core systems. Additional
23916 support from the interrupt/exception handling code of the system is not
23917 required. This model is enabled by default when the target is
23918 @code{sh*-*-linux*} and SH1* or SH2*.
23921 Generate hardware atomic sequences using the @code{movli.l} and @code{movco.l}
23922 instructions only. This is only available on SH4A and is suitable for
23923 multi-core systems. Since the hardware instructions support only 32 bit atomic
23924 variables access to 8 or 16 bit variables is emulated with 32 bit accesses.
23925 Code compiled with this option is also compatible with other software
23926 atomic model interrupt/exception handling systems if executed on an SH4A
23927 system. Additional support from the interrupt/exception handling code of the
23928 system is not required for this model.
23931 This parameter specifies the offset in bytes of the variable in the thread
23932 control block structure that should be used by the generated atomic sequences
23933 when the @samp{soft-tcb} model has been selected. For other models this
23934 parameter is ignored. The specified value must be an integer multiple of four
23935 and in the range 0-1020.
23938 This parameter prevents mixed usage of multiple atomic models, even if they
23939 are compatible, and makes the compiler generate atomic sequences of the
23940 specified model only.
23946 Generate the @code{tas.b} opcode for @code{__atomic_test_and_set}.
23947 Notice that depending on the particular hardware and software configuration
23948 this can degrade overall performance due to the operand cache line flushes
23949 that are implied by the @code{tas.b} instruction. On multi-core SH4A
23950 processors the @code{tas.b} instruction must be used with caution since it
23951 can result in data corruption for certain cache configurations.
23954 @opindex mprefergot
23955 When generating position-independent code, emit function calls using
23956 the Global Offset Table instead of the Procedure Linkage Table.
23959 @itemx -mno-usermode
23961 @opindex mno-usermode
23962 Don't allow (allow) the compiler generating privileged mode code. Specifying
23963 @option{-musermode} also implies @option{-mno-inline-ic_invalidate} if the
23964 inlined code would not work in user mode. @option{-musermode} is the default
23965 when the target is @code{sh*-*-linux*}. If the target is SH1* or SH2*
23966 @option{-musermode} has no effect, since there is no user mode.
23968 @item -multcost=@var{number}
23969 @opindex multcost=@var{number}
23970 Set the cost to assume for a multiply insn.
23972 @item -mdiv=@var{strategy}
23973 @opindex mdiv=@var{strategy}
23974 Set the division strategy to be used for integer division operations.
23975 @var{strategy} can be one of:
23980 Calls a library function that uses the single-step division instruction
23981 @code{div1} to perform the operation. Division by zero calculates an
23982 unspecified result and does not trap. This is the default except for SH4,
23983 SH2A and SHcompact.
23986 Calls a library function that performs the operation in double precision
23987 floating point. Division by zero causes a floating-point exception. This is
23988 the default for SHcompact with FPU. Specifying this for targets that do not
23989 have a double precision FPU defaults to @code{call-div1}.
23992 Calls a library function that uses a lookup table for small divisors and
23993 the @code{div1} instruction with case distinction for larger divisors. Division
23994 by zero calculates an unspecified result and does not trap. This is the default
23995 for SH4. Specifying this for targets that do not have dynamic shift
23996 instructions defaults to @code{call-div1}.
24000 When a division strategy has not been specified the default strategy is
24001 selected based on the current target. For SH2A the default strategy is to
24002 use the @code{divs} and @code{divu} instructions instead of library function
24005 @item -maccumulate-outgoing-args
24006 @opindex maccumulate-outgoing-args
24007 Reserve space once for outgoing arguments in the function prologue rather
24008 than around each call. Generally beneficial for performance and size. Also
24009 needed for unwinding to avoid changing the stack frame around conditional code.
24011 @item -mdivsi3_libfunc=@var{name}
24012 @opindex mdivsi3_libfunc=@var{name}
24013 Set the name of the library function used for 32-bit signed division to
24015 This only affects the name used in the @samp{call} division strategies, and
24016 the compiler still expects the same sets of input/output/clobbered registers as
24017 if this option were not present.
24019 @item -mfixed-range=@var{register-range}
24020 @opindex mfixed-range
24021 Generate code treating the given register range as fixed registers.
24022 A fixed register is one that the register allocator can not use. This is
24023 useful when compiling kernel code. A register range is specified as
24024 two registers separated by a dash. Multiple register ranges can be
24025 specified separated by a comma.
24027 @item -mbranch-cost=@var{num}
24028 @opindex mbranch-cost=@var{num}
24029 Assume @var{num} to be the cost for a branch instruction. Higher numbers
24030 make the compiler try to generate more branch-free code if possible.
24031 If not specified the value is selected depending on the processor type that
24032 is being compiled for.
24035 @itemx -mno-zdcbranch
24036 @opindex mzdcbranch
24037 @opindex mno-zdcbranch
24038 Assume (do not assume) that zero displacement conditional branch instructions
24039 @code{bt} and @code{bf} are fast. If @option{-mzdcbranch} is specified, the
24040 compiler prefers zero displacement branch code sequences. This is
24041 enabled by default when generating code for SH4 and SH4A. It can be explicitly
24042 disabled by specifying @option{-mno-zdcbranch}.
24044 @item -mcbranch-force-delay-slot
24045 @opindex mcbranch-force-delay-slot
24046 Force the usage of delay slots for conditional branches, which stuffs the delay
24047 slot with a @code{nop} if a suitable instruction cannot be found. By default
24048 this option is disabled. It can be enabled to work around hardware bugs as
24049 found in the original SH7055.
24052 @itemx -mno-fused-madd
24053 @opindex mfused-madd
24054 @opindex mno-fused-madd
24055 Generate code that uses (does not use) the floating-point multiply and
24056 accumulate instructions. These instructions are generated by default
24057 if hardware floating point is used. The machine-dependent
24058 @option{-mfused-madd} option is now mapped to the machine-independent
24059 @option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
24060 mapped to @option{-ffp-contract=off}.
24066 Allow or disallow the compiler to emit the @code{fsca} instruction for sine
24067 and cosine approximations. The option @option{-mfsca} must be used in
24068 combination with @option{-funsafe-math-optimizations}. It is enabled by default
24069 when generating code for SH4A. Using @option{-mno-fsca} disables sine and cosine
24070 approximations even if @option{-funsafe-math-optimizations} is in effect.
24076 Allow or disallow the compiler to emit the @code{fsrra} instruction for
24077 reciprocal square root approximations. The option @option{-mfsrra} must be used
24078 in combination with @option{-funsafe-math-optimizations} and
24079 @option{-ffinite-math-only}. It is enabled by default when generating code for
24080 SH4A. Using @option{-mno-fsrra} disables reciprocal square root approximations
24081 even if @option{-funsafe-math-optimizations} and @option{-ffinite-math-only} are
24084 @item -mpretend-cmove
24085 @opindex mpretend-cmove
24086 Prefer zero-displacement conditional branches for conditional move instruction
24087 patterns. This can result in faster code on the SH4 processor.
24091 Generate code using the FDPIC ABI.
24095 @node Solaris 2 Options
24096 @subsection Solaris 2 Options
24097 @cindex Solaris 2 options
24099 These @samp{-m} options are supported on Solaris 2:
24102 @item -mclear-hwcap
24103 @opindex mclear-hwcap
24104 @option{-mclear-hwcap} tells the compiler to remove the hardware
24105 capabilities generated by the Solaris assembler. This is only necessary
24106 when object files use ISA extensions not supported by the current
24107 machine, but check at runtime whether or not to use them.
24109 @item -mimpure-text
24110 @opindex mimpure-text
24111 @option{-mimpure-text}, used in addition to @option{-shared}, tells
24112 the compiler to not pass @option{-z text} to the linker when linking a
24113 shared object. Using this option, you can link position-dependent
24114 code into a shared object.
24116 @option{-mimpure-text} suppresses the ``relocations remain against
24117 allocatable but non-writable sections'' linker error message.
24118 However, the necessary relocations trigger copy-on-write, and the
24119 shared object is not actually shared across processes. Instead of
24120 using @option{-mimpure-text}, you should compile all source code with
24121 @option{-fpic} or @option{-fPIC}.
24125 These switches are supported in addition to the above on Solaris 2:
24130 This is a synonym for @option{-pthread}.
24133 @node SPARC Options
24134 @subsection SPARC Options
24135 @cindex SPARC options
24137 These @samp{-m} options are supported on the SPARC:
24140 @item -mno-app-regs
24142 @opindex mno-app-regs
24144 Specify @option{-mapp-regs} to generate output using the global registers
24145 2 through 4, which the SPARC SVR4 ABI reserves for applications. Like the
24146 global register 1, each global register 2 through 4 is then treated as an
24147 allocable register that is clobbered by function calls. This is the default.
24149 To be fully SVR4 ABI-compliant at the cost of some performance loss,
24150 specify @option{-mno-app-regs}. You should compile libraries and system
24151 software with this option.
24157 With @option{-mflat}, the compiler does not generate save/restore instructions
24158 and uses a ``flat'' or single register window model. This model is compatible
24159 with the regular register window model. The local registers and the input
24160 registers (0--5) are still treated as ``call-saved'' registers and are
24161 saved on the stack as needed.
24163 With @option{-mno-flat} (the default), the compiler generates save/restore
24164 instructions (except for leaf functions). This is the normal operating mode.
24167 @itemx -mhard-float
24169 @opindex mhard-float
24170 Generate output containing floating-point instructions. This is the
24174 @itemx -msoft-float
24176 @opindex msoft-float
24177 Generate output containing library calls for floating point.
24178 @strong{Warning:} the requisite libraries are not available for all SPARC
24179 targets. Normally the facilities of the machine's usual C compiler are
24180 used, but this cannot be done directly in cross-compilation. You must make
24181 your own arrangements to provide suitable library functions for
24182 cross-compilation. The embedded targets @samp{sparc-*-aout} and
24183 @samp{sparclite-*-*} do provide software floating-point support.
24185 @option{-msoft-float} changes the calling convention in the output file;
24186 therefore, it is only useful if you compile @emph{all} of a program with
24187 this option. In particular, you need to compile @file{libgcc.a}, the
24188 library that comes with GCC, with @option{-msoft-float} in order for
24191 @item -mhard-quad-float
24192 @opindex mhard-quad-float
24193 Generate output containing quad-word (long double) floating-point
24196 @item -msoft-quad-float
24197 @opindex msoft-quad-float
24198 Generate output containing library calls for quad-word (long double)
24199 floating-point instructions. The functions called are those specified
24200 in the SPARC ABI@. This is the default.
24202 As of this writing, there are no SPARC implementations that have hardware
24203 support for the quad-word floating-point instructions. They all invoke
24204 a trap handler for one of these instructions, and then the trap handler
24205 emulates the effect of the instruction. Because of the trap handler overhead,
24206 this is much slower than calling the ABI library routines. Thus the
24207 @option{-msoft-quad-float} option is the default.
24209 @item -mno-unaligned-doubles
24210 @itemx -munaligned-doubles
24211 @opindex mno-unaligned-doubles
24212 @opindex munaligned-doubles
24213 Assume that doubles have 8-byte alignment. This is the default.
24215 With @option{-munaligned-doubles}, GCC assumes that doubles have 8-byte
24216 alignment only if they are contained in another type, or if they have an
24217 absolute address. Otherwise, it assumes they have 4-byte alignment.
24218 Specifying this option avoids some rare compatibility problems with code
24219 generated by other compilers. It is not the default because it results
24220 in a performance loss, especially for floating-point code.
24223 @itemx -mno-user-mode
24224 @opindex muser-mode
24225 @opindex mno-user-mode
24226 Do not generate code that can only run in supervisor mode. This is relevant
24227 only for the @code{casa} instruction emitted for the LEON3 processor. This
24230 @item -mfaster-structs
24231 @itemx -mno-faster-structs
24232 @opindex mfaster-structs
24233 @opindex mno-faster-structs
24234 With @option{-mfaster-structs}, the compiler assumes that structures
24235 should have 8-byte alignment. This enables the use of pairs of
24236 @code{ldd} and @code{std} instructions for copies in structure
24237 assignment, in place of twice as many @code{ld} and @code{st} pairs.
24238 However, the use of this changed alignment directly violates the SPARC
24239 ABI@. Thus, it's intended only for use on targets where the developer
24240 acknowledges that their resulting code is not directly in line with
24241 the rules of the ABI@.
24243 @item -mstd-struct-return
24244 @itemx -mno-std-struct-return
24245 @opindex mstd-struct-return
24246 @opindex mno-std-struct-return
24247 With @option{-mstd-struct-return}, the compiler generates checking code
24248 in functions returning structures or unions to detect size mismatches
24249 between the two sides of function calls, as per the 32-bit ABI@.
24251 The default is @option{-mno-std-struct-return}. This option has no effect
24258 Enable Local Register Allocation. This is the default for SPARC since GCC 7
24259 so @option{-mno-lra} needs to be passed to get old Reload.
24261 @item -mcpu=@var{cpu_type}
24263 Set the instruction set, register set, and instruction scheduling parameters
24264 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
24265 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{hypersparc},
24266 @samp{leon}, @samp{leon3}, @samp{leon3v7}, @samp{sparclite}, @samp{f930},
24267 @samp{f934}, @samp{sparclite86x}, @samp{sparclet}, @samp{tsc701}, @samp{v9},
24268 @samp{ultrasparc}, @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2},
24269 @samp{niagara3}, @samp{niagara4}, @samp{niagara7} and @samp{m8}.
24271 Native Solaris and GNU/Linux toolchains also support the value @samp{native},
24272 which selects the best architecture option for the host processor.
24273 @option{-mcpu=native} has no effect if GCC does not recognize
24276 Default instruction scheduling parameters are used for values that select
24277 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
24278 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
24280 Here is a list of each supported architecture and their supported
24288 supersparc, hypersparc, leon, leon3
24291 f930, f934, sparclite86x
24297 ultrasparc, ultrasparc3, niagara, niagara2, niagara3, niagara4,
24301 By default (unless configured otherwise), GCC generates code for the V7
24302 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
24303 additionally optimizes it for the Cypress CY7C602 chip, as used in the
24304 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
24305 SPARCStation 1, 2, IPX etc.
24307 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
24308 architecture. The only difference from V7 code is that the compiler emits
24309 the integer multiply and integer divide instructions which exist in SPARC-V8
24310 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
24311 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
24314 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
24315 the SPARC architecture. This adds the integer multiply, integer divide step
24316 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
24317 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
24318 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
24319 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
24320 MB86934 chip, which is the more recent SPARClite with FPU@.
24322 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
24323 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
24324 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
24325 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
24326 optimizes it for the TEMIC SPARClet chip.
24328 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
24329 architecture. This adds 64-bit integer and floating-point move instructions,
24330 3 additional floating-point condition code registers and conditional move
24331 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
24332 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
24333 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
24334 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
24335 @option{-mcpu=niagara}, the compiler additionally optimizes it for
24336 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
24337 additionally optimizes it for Sun UltraSPARC T2 chips. With
24338 @option{-mcpu=niagara3}, the compiler additionally optimizes it for Sun
24339 UltraSPARC T3 chips. With @option{-mcpu=niagara4}, the compiler
24340 additionally optimizes it for Sun UltraSPARC T4 chips. With
24341 @option{-mcpu=niagara7}, the compiler additionally optimizes it for
24342 Oracle SPARC M7 chips. With @option{-mcpu=m8}, the compiler
24343 additionally optimizes it for Oracle M8 chips.
24345 @item -mtune=@var{cpu_type}
24347 Set the instruction scheduling parameters for machine type
24348 @var{cpu_type}, but do not set the instruction set or register set that the
24349 option @option{-mcpu=@var{cpu_type}} does.
24351 The same values for @option{-mcpu=@var{cpu_type}} can be used for
24352 @option{-mtune=@var{cpu_type}}, but the only useful values are those
24353 that select a particular CPU implementation. Those are
24354 @samp{cypress}, @samp{supersparc}, @samp{hypersparc}, @samp{leon},
24355 @samp{leon3}, @samp{leon3v7}, @samp{f930}, @samp{f934},
24356 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
24357 @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2}, @samp{niagara3},
24358 @samp{niagara4}, @samp{niagara7} and @samp{m8}. With native Solaris
24359 and GNU/Linux toolchains, @samp{native} can also be used.
24364 @opindex mno-v8plus
24365 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
24366 difference from the V8 ABI is that the global and out registers are
24367 considered 64 bits wide. This is enabled by default on Solaris in 32-bit
24368 mode for all SPARC-V9 processors.
24374 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
24375 Visual Instruction Set extensions. The default is @option{-mno-vis}.
24381 With @option{-mvis2}, GCC generates code that takes advantage of
24382 version 2.0 of the UltraSPARC Visual Instruction Set extensions. The
24383 default is @option{-mvis2} when targeting a cpu that supports such
24384 instructions, such as UltraSPARC-III and later. Setting @option{-mvis2}
24385 also sets @option{-mvis}.
24391 With @option{-mvis3}, GCC generates code that takes advantage of
24392 version 3.0 of the UltraSPARC Visual Instruction Set extensions. The
24393 default is @option{-mvis3} when targeting a cpu that supports such
24394 instructions, such as niagara-3 and later. Setting @option{-mvis3}
24395 also sets @option{-mvis2} and @option{-mvis}.
24401 With @option{-mvis4}, GCC generates code that takes advantage of
24402 version 4.0 of the UltraSPARC Visual Instruction Set extensions. The
24403 default is @option{-mvis4} when targeting a cpu that supports such
24404 instructions, such as niagara-7 and later. Setting @option{-mvis4}
24405 also sets @option{-mvis3}, @option{-mvis2} and @option{-mvis}.
24411 With @option{-mvis4b}, GCC generates code that takes advantage of
24412 version 4.0 of the UltraSPARC Visual Instruction Set extensions, plus
24413 the additional VIS instructions introduced in the Oracle SPARC
24414 Architecture 2017. The default is @option{-mvis4b} when targeting a
24415 cpu that supports such instructions, such as m8 and later. Setting
24416 @option{-mvis4b} also sets @option{-mvis4}, @option{-mvis3},
24417 @option{-mvis2} and @option{-mvis}.
24422 @opindex mno-cbcond
24423 With @option{-mcbcond}, GCC generates code that takes advantage of the UltraSPARC
24424 Compare-and-Branch-on-Condition instructions. The default is @option{-mcbcond}
24425 when targeting a CPU that supports such instructions, such as Niagara-4 and
24432 With @option{-mfmaf}, GCC generates code that takes advantage of the UltraSPARC
24433 Fused Multiply-Add Floating-point instructions. The default is @option{-mfmaf}
24434 when targeting a CPU that supports such instructions, such as Niagara-3 and
24440 @opindex mno-fsmuld
24441 With @option{-mfsmuld}, GCC generates code that takes advantage of the
24442 Floating-point Multiply Single to Double (FsMULd) instruction. The default is
24443 @option{-mfsmuld} when targeting a CPU supporting the architecture versions V8
24444 or V9 with FPU except @option{-mcpu=leon}.
24450 With @option{-mpopc}, GCC generates code that takes advantage of the UltraSPARC
24451 Population Count instruction. The default is @option{-mpopc}
24452 when targeting a CPU that supports such an instruction, such as Niagara-2 and
24459 With @option{-msubxc}, GCC generates code that takes advantage of the UltraSPARC
24460 Subtract-Extended-with-Carry instruction. The default is @option{-msubxc}
24461 when targeting a CPU that supports such an instruction, such as Niagara-7 and
24465 @opindex mfix-at697f
24466 Enable the documented workaround for the single erratum of the Atmel AT697F
24467 processor (which corresponds to erratum #13 of the AT697E processor).
24470 @opindex mfix-ut699
24471 Enable the documented workarounds for the floating-point errata and the data
24472 cache nullify errata of the UT699 processor.
24475 @opindex mfix-ut700
24476 Enable the documented workaround for the back-to-back store errata of
24477 the UT699E/UT700 processor.
24479 @item -mfix-gr712rc
24480 @opindex mfix-gr712rc
24481 Enable the documented workaround for the back-to-back store errata of
24482 the GR712RC processor.
24485 These @samp{-m} options are supported in addition to the above
24486 on SPARC-V9 processors in 64-bit environments:
24493 Generate code for a 32-bit or 64-bit environment.
24494 The 32-bit environment sets int, long and pointer to 32 bits.
24495 The 64-bit environment sets int to 32 bits and long and pointer
24498 @item -mcmodel=@var{which}
24500 Set the code model to one of
24504 The Medium/Low code model: 64-bit addresses, programs
24505 must be linked in the low 32 bits of memory. Programs can be statically
24506 or dynamically linked.
24509 The Medium/Middle code model: 64-bit addresses, programs
24510 must be linked in the low 44 bits of memory, the text and data segments must
24511 be less than 2GB in size and the data segment must be located within 2GB of
24515 The Medium/Anywhere code model: 64-bit addresses, programs
24516 may be linked anywhere in memory, the text and data segments must be less
24517 than 2GB in size and the data segment must be located within 2GB of the
24521 The Medium/Anywhere code model for embedded systems:
24522 64-bit addresses, the text and data segments must be less than 2GB in
24523 size, both starting anywhere in memory (determined at link time). The
24524 global register %g4 points to the base of the data segment. Programs
24525 are statically linked and PIC is not supported.
24528 @item -mmemory-model=@var{mem-model}
24529 @opindex mmemory-model
24530 Set the memory model in force on the processor to one of
24534 The default memory model for the processor and operating system.
24537 Relaxed Memory Order
24540 Partial Store Order
24546 Sequential Consistency
24549 These memory models are formally defined in Appendix D of the SPARC-V9
24550 architecture manual, as set in the processor's @code{PSTATE.MM} field.
24553 @itemx -mno-stack-bias
24554 @opindex mstack-bias
24555 @opindex mno-stack-bias
24556 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
24557 frame pointer if present, are offset by @minus{}2047 which must be added back
24558 when making stack frame references. This is the default in 64-bit mode.
24559 Otherwise, assume no such offset is present.
24563 @subsection SPU Options
24564 @cindex SPU options
24566 These @samp{-m} options are supported on the SPU:
24570 @itemx -merror-reloc
24571 @opindex mwarn-reloc
24572 @opindex merror-reloc
24574 The loader for SPU does not handle dynamic relocations. By default, GCC
24575 gives an error when it generates code that requires a dynamic
24576 relocation. @option{-mno-error-reloc} disables the error,
24577 @option{-mwarn-reloc} generates a warning instead.
24580 @itemx -munsafe-dma
24582 @opindex munsafe-dma
24584 Instructions that initiate or test completion of DMA must not be
24585 reordered with respect to loads and stores of the memory that is being
24587 With @option{-munsafe-dma} you must use the @code{volatile} keyword to protect
24588 memory accesses, but that can lead to inefficient code in places where the
24589 memory is known to not change. Rather than mark the memory as volatile,
24590 you can use @option{-msafe-dma} to tell the compiler to treat
24591 the DMA instructions as potentially affecting all memory.
24593 @item -mbranch-hints
24594 @opindex mbranch-hints
24596 By default, GCC generates a branch hint instruction to avoid
24597 pipeline stalls for always-taken or probably-taken branches. A hint
24598 is not generated closer than 8 instructions away from its branch.
24599 There is little reason to disable them, except for debugging purposes,
24600 or to make an object a little bit smaller.
24604 @opindex msmall-mem
24605 @opindex mlarge-mem
24607 By default, GCC generates code assuming that addresses are never larger
24608 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
24609 a full 32-bit address.
24614 By default, GCC links against startup code that assumes the SPU-style
24615 main function interface (which has an unconventional parameter list).
24616 With @option{-mstdmain}, GCC links your program against startup
24617 code that assumes a C99-style interface to @code{main}, including a
24618 local copy of @code{argv} strings.
24620 @item -mfixed-range=@var{register-range}
24621 @opindex mfixed-range
24622 Generate code treating the given register range as fixed registers.
24623 A fixed register is one that the register allocator cannot use. This is
24624 useful when compiling kernel code. A register range is specified as
24625 two registers separated by a dash. Multiple register ranges can be
24626 specified separated by a comma.
24632 Compile code assuming that pointers to the PPU address space accessed
24633 via the @code{__ea} named address space qualifier are either 32 or 64
24634 bits wide. The default is 32 bits. As this is an ABI-changing option,
24635 all object code in an executable must be compiled with the same setting.
24637 @item -maddress-space-conversion
24638 @itemx -mno-address-space-conversion
24639 @opindex maddress-space-conversion
24640 @opindex mno-address-space-conversion
24641 Allow/disallow treating the @code{__ea} address space as superset
24642 of the generic address space. This enables explicit type casts
24643 between @code{__ea} and generic pointer as well as implicit
24644 conversions of generic pointers to @code{__ea} pointers. The
24645 default is to allow address space pointer conversions.
24647 @item -mcache-size=@var{cache-size}
24648 @opindex mcache-size
24649 This option controls the version of libgcc that the compiler links to an
24650 executable and selects a software-managed cache for accessing variables
24651 in the @code{__ea} address space with a particular cache size. Possible
24652 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
24653 and @samp{128}. The default cache size is 64KB.
24655 @item -matomic-updates
24656 @itemx -mno-atomic-updates
24657 @opindex matomic-updates
24658 @opindex mno-atomic-updates
24659 This option controls the version of libgcc that the compiler links to an
24660 executable and selects whether atomic updates to the software-managed
24661 cache of PPU-side variables are used. If you use atomic updates, changes
24662 to a PPU variable from SPU code using the @code{__ea} named address space
24663 qualifier do not interfere with changes to other PPU variables residing
24664 in the same cache line from PPU code. If you do not use atomic updates,
24665 such interference may occur; however, writing back cache lines is
24666 more efficient. The default behavior is to use atomic updates.
24669 @itemx -mdual-nops=@var{n}
24670 @opindex mdual-nops
24671 By default, GCC inserts NOPs to increase dual issue when it expects
24672 it to increase performance. @var{n} can be a value from 0 to 10. A
24673 smaller @var{n} inserts fewer NOPs. 10 is the default, 0 is the
24674 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
24676 @item -mhint-max-nops=@var{n}
24677 @opindex mhint-max-nops
24678 Maximum number of NOPs to insert for a branch hint. A branch hint must
24679 be at least 8 instructions away from the branch it is affecting. GCC
24680 inserts up to @var{n} NOPs to enforce this, otherwise it does not
24681 generate the branch hint.
24683 @item -mhint-max-distance=@var{n}
24684 @opindex mhint-max-distance
24685 The encoding of the branch hint instruction limits the hint to be within
24686 256 instructions of the branch it is affecting. By default, GCC makes
24687 sure it is within 125.
24690 @opindex msafe-hints
24691 Work around a hardware bug that causes the SPU to stall indefinitely.
24692 By default, GCC inserts the @code{hbrp} instruction to make sure
24693 this stall won't happen.
24697 @node System V Options
24698 @subsection Options for System V
24700 These additional options are available on System V Release 4 for
24701 compatibility with other compilers on those systems:
24706 Create a shared object.
24707 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
24711 Identify the versions of each tool used by the compiler, in a
24712 @code{.ident} assembler directive in the output.
24716 Refrain from adding @code{.ident} directives to the output file (this is
24719 @item -YP,@var{dirs}
24721 Search the directories @var{dirs}, and no others, for libraries
24722 specified with @option{-l}.
24724 @item -Ym,@var{dir}
24726 Look in the directory @var{dir} to find the M4 preprocessor.
24727 The assembler uses this option.
24728 @c This is supposed to go with a -Yd for predefined M4 macro files, but
24729 @c the generic assembler that comes with Solaris takes just -Ym.
24732 @node TILE-Gx Options
24733 @subsection TILE-Gx Options
24734 @cindex TILE-Gx options
24736 These @samp{-m} options are supported on the TILE-Gx:
24739 @item -mcmodel=small
24740 @opindex mcmodel=small
24741 Generate code for the small model. The distance for direct calls is
24742 limited to 500M in either direction. PC-relative addresses are 32
24743 bits. Absolute addresses support the full address range.
24745 @item -mcmodel=large
24746 @opindex mcmodel=large
24747 Generate code for the large model. There is no limitation on call
24748 distance, pc-relative addresses, or absolute addresses.
24750 @item -mcpu=@var{name}
24752 Selects the type of CPU to be targeted. Currently the only supported
24753 type is @samp{tilegx}.
24759 Generate code for a 32-bit or 64-bit environment. The 32-bit
24760 environment sets int, long, and pointer to 32 bits. The 64-bit
24761 environment sets int to 32 bits and long and pointer to 64 bits.
24764 @itemx -mlittle-endian
24765 @opindex mbig-endian
24766 @opindex mlittle-endian
24767 Generate code in big/little endian mode, respectively.
24770 @node TILEPro Options
24771 @subsection TILEPro Options
24772 @cindex TILEPro options
24774 These @samp{-m} options are supported on the TILEPro:
24777 @item -mcpu=@var{name}
24779 Selects the type of CPU to be targeted. Currently the only supported
24780 type is @samp{tilepro}.
24784 Generate code for a 32-bit environment, which sets int, long, and
24785 pointer to 32 bits. This is the only supported behavior so the flag
24786 is essentially ignored.
24790 @subsection V850 Options
24791 @cindex V850 Options
24793 These @samp{-m} options are defined for V850 implementations:
24797 @itemx -mno-long-calls
24798 @opindex mlong-calls
24799 @opindex mno-long-calls
24800 Treat all calls as being far away (near). If calls are assumed to be
24801 far away, the compiler always loads the function's address into a
24802 register, and calls indirect through the pointer.
24808 Do not optimize (do optimize) basic blocks that use the same index
24809 pointer 4 or more times to copy pointer into the @code{ep} register, and
24810 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
24811 option is on by default if you optimize.
24813 @item -mno-prolog-function
24814 @itemx -mprolog-function
24815 @opindex mno-prolog-function
24816 @opindex mprolog-function
24817 Do not use (do use) external functions to save and restore registers
24818 at the prologue and epilogue of a function. The external functions
24819 are slower, but use less code space if more than one function saves
24820 the same number of registers. The @option{-mprolog-function} option
24821 is on by default if you optimize.
24825 Try to make the code as small as possible. At present, this just turns
24826 on the @option{-mep} and @option{-mprolog-function} options.
24828 @item -mtda=@var{n}
24830 Put static or global variables whose size is @var{n} bytes or less into
24831 the tiny data area that register @code{ep} points to. The tiny data
24832 area can hold up to 256 bytes in total (128 bytes for byte references).
24834 @item -msda=@var{n}
24836 Put static or global variables whose size is @var{n} bytes or less into
24837 the small data area that register @code{gp} points to. The small data
24838 area can hold up to 64 kilobytes.
24840 @item -mzda=@var{n}
24842 Put static or global variables whose size is @var{n} bytes or less into
24843 the first 32 kilobytes of memory.
24847 Specify that the target processor is the V850.
24851 Specify that the target processor is the V850E3V5. The preprocessor
24852 constant @code{__v850e3v5__} is defined if this option is used.
24856 Specify that the target processor is the V850E3V5. This is an alias for
24857 the @option{-mv850e3v5} option.
24861 Specify that the target processor is the V850E2V3. The preprocessor
24862 constant @code{__v850e2v3__} is defined if this option is used.
24866 Specify that the target processor is the V850E2. The preprocessor
24867 constant @code{__v850e2__} is defined if this option is used.
24871 Specify that the target processor is the V850E1. The preprocessor
24872 constants @code{__v850e1__} and @code{__v850e__} are defined if
24873 this option is used.
24877 Specify that the target processor is the V850ES. This is an alias for
24878 the @option{-mv850e1} option.
24882 Specify that the target processor is the V850E@. The preprocessor
24883 constant @code{__v850e__} is defined if this option is used.
24885 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
24886 nor @option{-mv850e2} nor @option{-mv850e2v3} nor @option{-mv850e3v5}
24887 are defined then a default target processor is chosen and the
24888 relevant @samp{__v850*__} preprocessor constant is defined.
24890 The preprocessor constants @code{__v850} and @code{__v851__} are always
24891 defined, regardless of which processor variant is the target.
24893 @item -mdisable-callt
24894 @itemx -mno-disable-callt
24895 @opindex mdisable-callt
24896 @opindex mno-disable-callt
24897 This option suppresses generation of the @code{CALLT} instruction for the
24898 v850e, v850e1, v850e2, v850e2v3 and v850e3v5 flavors of the v850
24901 This option is enabled by default when the RH850 ABI is
24902 in use (see @option{-mrh850-abi}), and disabled by default when the
24903 GCC ABI is in use. If @code{CALLT} instructions are being generated
24904 then the C preprocessor symbol @code{__V850_CALLT__} is defined.
24910 Pass on (or do not pass on) the @option{-mrelax} command-line option
24914 @itemx -mno-long-jumps
24915 @opindex mlong-jumps
24916 @opindex mno-long-jumps
24917 Disable (or re-enable) the generation of PC-relative jump instructions.
24920 @itemx -mhard-float
24921 @opindex msoft-float
24922 @opindex mhard-float
24923 Disable (or re-enable) the generation of hardware floating point
24924 instructions. This option is only significant when the target
24925 architecture is @samp{V850E2V3} or higher. If hardware floating point
24926 instructions are being generated then the C preprocessor symbol
24927 @code{__FPU_OK__} is defined, otherwise the symbol
24928 @code{__NO_FPU__} is defined.
24932 Enables the use of the e3v5 LOOP instruction. The use of this
24933 instruction is not enabled by default when the e3v5 architecture is
24934 selected because its use is still experimental.
24938 @opindex mrh850-abi
24940 Enables support for the RH850 version of the V850 ABI. This is the
24941 default. With this version of the ABI the following rules apply:
24945 Integer sized structures and unions are returned via a memory pointer
24946 rather than a register.
24949 Large structures and unions (more than 8 bytes in size) are passed by
24953 Functions are aligned to 16-bit boundaries.
24956 The @option{-m8byte-align} command-line option is supported.
24959 The @option{-mdisable-callt} command-line option is enabled by
24960 default. The @option{-mno-disable-callt} command-line option is not
24964 When this version of the ABI is enabled the C preprocessor symbol
24965 @code{__V850_RH850_ABI__} is defined.
24969 Enables support for the old GCC version of the V850 ABI. With this
24970 version of the ABI the following rules apply:
24974 Integer sized structures and unions are returned in register @code{r10}.
24977 Large structures and unions (more than 8 bytes in size) are passed by
24981 Functions are aligned to 32-bit boundaries, unless optimizing for
24985 The @option{-m8byte-align} command-line option is not supported.
24988 The @option{-mdisable-callt} command-line option is supported but not
24989 enabled by default.
24992 When this version of the ABI is enabled the C preprocessor symbol
24993 @code{__V850_GCC_ABI__} is defined.
24995 @item -m8byte-align
24996 @itemx -mno-8byte-align
24997 @opindex m8byte-align
24998 @opindex mno-8byte-align
24999 Enables support for @code{double} and @code{long long} types to be
25000 aligned on 8-byte boundaries. The default is to restrict the
25001 alignment of all objects to at most 4-bytes. When
25002 @option{-m8byte-align} is in effect the C preprocessor symbol
25003 @code{__V850_8BYTE_ALIGN__} is defined.
25006 @opindex mbig-switch
25007 Generate code suitable for big switch tables. Use this option only if
25008 the assembler/linker complain about out of range branches within a switch
25013 This option causes r2 and r5 to be used in the code generated by
25014 the compiler. This setting is the default.
25016 @item -mno-app-regs
25017 @opindex mno-app-regs
25018 This option causes r2 and r5 to be treated as fixed registers.
25023 @subsection VAX Options
25024 @cindex VAX options
25026 These @samp{-m} options are defined for the VAX:
25031 Do not output certain jump instructions (@code{aobleq} and so on)
25032 that the Unix assembler for the VAX cannot handle across long
25037 Do output those jump instructions, on the assumption that the
25038 GNU assembler is being used.
25042 Output code for G-format floating-point numbers instead of D-format.
25045 @node Visium Options
25046 @subsection Visium Options
25047 @cindex Visium options
25053 A program which performs file I/O and is destined to run on an MCM target
25054 should be linked with this option. It causes the libraries libc.a and
25055 libdebug.a to be linked. The program should be run on the target under
25056 the control of the GDB remote debugging stub.
25060 A program which performs file I/O and is destined to run on the simulator
25061 should be linked with option. This causes libraries libc.a and libsim.a to
25065 @itemx -mhard-float
25067 @opindex mhard-float
25068 Generate code containing floating-point instructions. This is the
25072 @itemx -msoft-float
25074 @opindex msoft-float
25075 Generate code containing library calls for floating-point.
25077 @option{-msoft-float} changes the calling convention in the output file;
25078 therefore, it is only useful if you compile @emph{all} of a program with
25079 this option. In particular, you need to compile @file{libgcc.a}, the
25080 library that comes with GCC, with @option{-msoft-float} in order for
25083 @item -mcpu=@var{cpu_type}
25085 Set the instruction set, register set, and instruction scheduling parameters
25086 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
25087 @samp{mcm}, @samp{gr5} and @samp{gr6}.
25089 @samp{mcm} is a synonym of @samp{gr5} present for backward compatibility.
25091 By default (unless configured otherwise), GCC generates code for the GR5
25092 variant of the Visium architecture.
25094 With @option{-mcpu=gr6}, GCC generates code for the GR6 variant of the Visium
25095 architecture. The only difference from GR5 code is that the compiler will
25096 generate block move instructions.
25098 @item -mtune=@var{cpu_type}
25100 Set the instruction scheduling parameters for machine type @var{cpu_type},
25101 but do not set the instruction set or register set that the option
25102 @option{-mcpu=@var{cpu_type}} would.
25106 Generate code for the supervisor mode, where there are no restrictions on
25107 the access to general registers. This is the default.
25110 @opindex muser-mode
25111 Generate code for the user mode, where the access to some general registers
25112 is forbidden: on the GR5, registers r24 to r31 cannot be accessed in this
25113 mode; on the GR6, only registers r29 to r31 are affected.
25117 @subsection VMS Options
25119 These @samp{-m} options are defined for the VMS implementations:
25122 @item -mvms-return-codes
25123 @opindex mvms-return-codes
25124 Return VMS condition codes from @code{main}. The default is to return POSIX-style
25125 condition (e.g.@ error) codes.
25127 @item -mdebug-main=@var{prefix}
25128 @opindex mdebug-main=@var{prefix}
25129 Flag the first routine whose name starts with @var{prefix} as the main
25130 routine for the debugger.
25134 Default to 64-bit memory allocation routines.
25136 @item -mpointer-size=@var{size}
25137 @opindex mpointer-size=@var{size}
25138 Set the default size of pointers. Possible options for @var{size} are
25139 @samp{32} or @samp{short} for 32 bit pointers, @samp{64} or @samp{long}
25140 for 64 bit pointers, and @samp{no} for supporting only 32 bit pointers.
25141 The later option disables @code{pragma pointer_size}.
25144 @node VxWorks Options
25145 @subsection VxWorks Options
25146 @cindex VxWorks Options
25148 The options in this section are defined for all VxWorks targets.
25149 Options specific to the target hardware are listed with the other
25150 options for that target.
25155 GCC can generate code for both VxWorks kernels and real time processes
25156 (RTPs). This option switches from the former to the latter. It also
25157 defines the preprocessor macro @code{__RTP__}.
25160 @opindex non-static
25161 Link an RTP executable against shared libraries rather than static
25162 libraries. The options @option{-static} and @option{-shared} can
25163 also be used for RTPs (@pxref{Link Options}); @option{-static}
25170 These options are passed down to the linker. They are defined for
25171 compatibility with Diab.
25174 @opindex Xbind-lazy
25175 Enable lazy binding of function calls. This option is equivalent to
25176 @option{-Wl,-z,now} and is defined for compatibility with Diab.
25180 Disable lazy binding of function calls. This option is the default and
25181 is defined for compatibility with Diab.
25185 @subsection x86 Options
25186 @cindex x86 Options
25188 These @samp{-m} options are defined for the x86 family of computers.
25192 @item -march=@var{cpu-type}
25194 Generate instructions for the machine type @var{cpu-type}. In contrast to
25195 @option{-mtune=@var{cpu-type}}, which merely tunes the generated code
25196 for the specified @var{cpu-type}, @option{-march=@var{cpu-type}} allows GCC
25197 to generate code that may not run at all on processors other than the one
25198 indicated. Specifying @option{-march=@var{cpu-type}} implies
25199 @option{-mtune=@var{cpu-type}}.
25201 The choices for @var{cpu-type} are:
25205 This selects the CPU to generate code for at compilation time by determining
25206 the processor type of the compiling machine. Using @option{-march=native}
25207 enables all instruction subsets supported by the local machine (hence
25208 the result might not run on different machines). Using @option{-mtune=native}
25209 produces code optimized for the local machine under the constraints
25210 of the selected instruction set.
25213 Original Intel i386 CPU@.
25216 Intel i486 CPU@. (No scheduling is implemented for this chip.)
25220 Intel Pentium CPU with no MMX support.
25223 Intel Lakemont MCU, based on Intel Pentium CPU.
25226 Intel Pentium MMX CPU, based on Pentium core with MMX instruction set support.
25229 Intel Pentium Pro CPU@.
25232 When used with @option{-march}, the Pentium Pro
25233 instruction set is used, so the code runs on all i686 family chips.
25234 When used with @option{-mtune}, it has the same meaning as @samp{generic}.
25237 Intel Pentium II CPU, based on Pentium Pro core with MMX instruction set
25242 Intel Pentium III CPU, based on Pentium Pro core with MMX and SSE instruction
25246 Intel Pentium M; low-power version of Intel Pentium III CPU
25247 with MMX, SSE and SSE2 instruction set support. Used by Centrino notebooks.
25251 Intel Pentium 4 CPU with MMX, SSE and SSE2 instruction set support.
25254 Improved version of Intel Pentium 4 CPU with MMX, SSE, SSE2 and SSE3 instruction
25258 Improved version of Intel Pentium 4 CPU with 64-bit extensions, MMX, SSE,
25259 SSE2 and SSE3 instruction set support.
25262 Intel Core 2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
25263 instruction set support.
25266 Intel Nehalem CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
25267 SSE4.1, SSE4.2 and POPCNT instruction set support.
25270 Intel Westmere CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
25271 SSE4.1, SSE4.2, POPCNT, AES and PCLMUL instruction set support.
25274 Intel Sandy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
25275 SSE4.1, SSE4.2, POPCNT, AVX, AES and PCLMUL instruction set support.
25278 Intel Ivy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
25279 SSE4.1, SSE4.2, POPCNT, AVX, AES, PCLMUL, FSGSBASE, RDRND and F16C
25280 instruction set support.
25283 Intel Haswell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
25284 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
25285 BMI, BMI2 and F16C instruction set support.
25288 Intel Broadwell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
25289 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
25290 BMI, BMI2, F16C, RDSEED, ADCX and PREFETCHW instruction set support.
25293 Intel Skylake CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
25294 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
25295 BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC and
25296 XSAVES instruction set support.
25299 Intel Bonnell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3 and SSSE3
25300 instruction set support.
25303 Intel Silvermont CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
25304 SSE4.1, SSE4.2, POPCNT, AES, PCLMUL and RDRND instruction set support.
25307 Intel Knight's Landing CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
25308 SSSE3, SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
25309 BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, AVX512F, AVX512PF, AVX512ER and
25310 AVX512CD instruction set support.
25313 Intel Knights Mill CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
25314 SSSE3, SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
25315 BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, AVX512F, AVX512PF, AVX512ER, AVX512CD,
25316 AVX5124VNNIW, AVX5124FMAPS and AVX512VPOPCNTDQ instruction set support.
25318 @item skylake-avx512
25319 Intel Skylake Server CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
25320 SSSE3, SSE4.1, SSE4.2, POPCNT, PKU, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
25321 BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC, XSAVES, AVX512F,
25322 AVX512VL, AVX512BW, AVX512DQ and AVX512CD instruction set support.
25325 AMD K6 CPU with MMX instruction set support.
25329 Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
25332 @itemx athlon-tbird
25333 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
25339 Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
25340 instruction set support.
25346 Processors based on the AMD K8 core with x86-64 instruction set support,
25347 including the AMD Opteron, Athlon 64, and Athlon 64 FX processors.
25348 (This supersets MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit
25349 instruction set extensions.)
25352 @itemx opteron-sse3
25353 @itemx athlon64-sse3
25354 Improved versions of AMD K8 cores with SSE3 instruction set support.
25358 CPUs based on AMD Family 10h cores with x86-64 instruction set support. (This
25359 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
25360 instruction set extensions.)
25363 CPUs based on AMD Family 15h cores with x86-64 instruction set support. (This
25364 supersets FMA4, AVX, XOP, LWP, AES, PCL_MUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A,
25365 SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set extensions.)
25367 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
25368 supersets BMI, TBM, F16C, FMA, FMA4, AVX, XOP, LWP, AES, PCL_MUL, CX16, MMX,
25369 SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set
25372 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
25373 supersets BMI, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, XOP, LWP, AES,
25374 PCL_MUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and
25375 64-bit instruction set extensions.
25377 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
25378 supersets BMI, BMI2, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, AVX2, XOP, LWP,
25379 AES, PCL_MUL, CX16, MOVBE, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1,
25380 SSE4.2, ABM and 64-bit instruction set extensions.
25383 AMD Family 17h core based CPUs with x86-64 instruction set support. (This
25384 supersets BMI, BMI2, F16C, FMA, FSGSBASE, AVX, AVX2, ADCX, RDSEED, MWAITX,
25385 SHA, CLZERO, AES, PCL_MUL, CX16, MOVBE, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3,
25386 SSE4.1, SSE4.2, ABM, XSAVEC, XSAVES, CLFLUSHOPT, POPCNT, and 64-bit
25387 instruction set extensions.
25390 CPUs based on AMD Family 14h cores with x86-64 instruction set support. (This
25391 supersets MMX, SSE, SSE2, SSE3, SSSE3, SSE4A, CX16, ABM and 64-bit
25392 instruction set extensions.)
25395 CPUs based on AMD Family 16h cores with x86-64 instruction set support. This
25396 includes MOVBE, F16C, BMI, AVX, PCL_MUL, AES, SSE4.2, SSE4.1, CX16, ABM,
25397 SSE4A, SSSE3, SSE3, SSE2, SSE, MMX and 64-bit instruction set extensions.
25400 IDT WinChip C6 CPU, dealt in same way as i486 with additional MMX instruction
25404 IDT WinChip 2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
25405 instruction set support.
25408 VIA C3 CPU with MMX and 3DNow!@: instruction set support.
25409 (No scheduling is implemented for this chip.)
25412 VIA C3-2 (Nehemiah/C5XL) CPU with MMX and SSE instruction set support.
25413 (No scheduling is implemented for this chip.)
25416 VIA C7 (Esther) CPU with MMX, SSE, SSE2 and SSE3 instruction set support.
25417 (No scheduling is implemented for this chip.)
25420 VIA Eden Samuel 2 CPU with MMX and 3DNow!@: instruction set support.
25421 (No scheduling is implemented for this chip.)
25424 VIA Eden Nehemiah CPU with MMX and SSE instruction set support.
25425 (No scheduling is implemented for this chip.)
25428 VIA Eden Esther CPU with MMX, SSE, SSE2 and SSE3 instruction set support.
25429 (No scheduling is implemented for this chip.)
25432 VIA Eden X2 CPU with x86-64, MMX, SSE, SSE2 and SSE3 instruction set support.
25433 (No scheduling is implemented for this chip.)
25436 VIA Eden X4 CPU with x86-64, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2,
25437 AVX and AVX2 instruction set support.
25438 (No scheduling is implemented for this chip.)
25441 Generic VIA Nano CPU with x86-64, MMX, SSE, SSE2, SSE3 and SSSE3
25442 instruction set support.
25443 (No scheduling is implemented for this chip.)
25446 VIA Nano 1xxx CPU with x86-64, MMX, SSE, SSE2, SSE3 and SSSE3
25447 instruction set support.
25448 (No scheduling is implemented for this chip.)
25451 VIA Nano 2xxx CPU with x86-64, MMX, SSE, SSE2, SSE3 and SSSE3
25452 instruction set support.
25453 (No scheduling is implemented for this chip.)
25456 VIA Nano 3xxx CPU with x86-64, MMX, SSE, SSE2, SSE3, SSSE3 and SSE4.1
25457 instruction set support.
25458 (No scheduling is implemented for this chip.)
25461 VIA Nano Dual Core CPU with x86-64, MMX, SSE, SSE2, SSE3, SSSE3 and SSE4.1
25462 instruction set support.
25463 (No scheduling is implemented for this chip.)
25466 VIA Nano Quad Core CPU with x86-64, MMX, SSE, SSE2, SSE3, SSSE3 and SSE4.1
25467 instruction set support.
25468 (No scheduling is implemented for this chip.)
25471 AMD Geode embedded processor with MMX and 3DNow!@: instruction set support.
25474 @item -mtune=@var{cpu-type}
25476 Tune to @var{cpu-type} everything applicable about the generated code, except
25477 for the ABI and the set of available instructions.
25478 While picking a specific @var{cpu-type} schedules things appropriately
25479 for that particular chip, the compiler does not generate any code that
25480 cannot run on the default machine type unless you use a
25481 @option{-march=@var{cpu-type}} option.
25482 For example, if GCC is configured for i686-pc-linux-gnu
25483 then @option{-mtune=pentium4} generates code that is tuned for Pentium 4
25484 but still runs on i686 machines.
25486 The choices for @var{cpu-type} are the same as for @option{-march}.
25487 In addition, @option{-mtune} supports 2 extra choices for @var{cpu-type}:
25491 Produce code optimized for the most common IA32/@/AMD64/@/EM64T processors.
25492 If you know the CPU on which your code will run, then you should use
25493 the corresponding @option{-mtune} or @option{-march} option instead of
25494 @option{-mtune=generic}. But, if you do not know exactly what CPU users
25495 of your application will have, then you should use this option.
25497 As new processors are deployed in the marketplace, the behavior of this
25498 option will change. Therefore, if you upgrade to a newer version of
25499 GCC, code generation controlled by this option will change to reflect
25501 that are most common at the time that version of GCC is released.
25503 There is no @option{-march=generic} option because @option{-march}
25504 indicates the instruction set the compiler can use, and there is no
25505 generic instruction set applicable to all processors. In contrast,
25506 @option{-mtune} indicates the processor (or, in this case, collection of
25507 processors) for which the code is optimized.
25510 Produce code optimized for the most current Intel processors, which are
25511 Haswell and Silvermont for this version of GCC. If you know the CPU
25512 on which your code will run, then you should use the corresponding
25513 @option{-mtune} or @option{-march} option instead of @option{-mtune=intel}.
25514 But, if you want your application performs better on both Haswell and
25515 Silvermont, then you should use this option.
25517 As new Intel processors are deployed in the marketplace, the behavior of
25518 this option will change. Therefore, if you upgrade to a newer version of
25519 GCC, code generation controlled by this option will change to reflect
25520 the most current Intel processors at the time that version of GCC is
25523 There is no @option{-march=intel} option because @option{-march} indicates
25524 the instruction set the compiler can use, and there is no common
25525 instruction set applicable to all processors. In contrast,
25526 @option{-mtune} indicates the processor (or, in this case, collection of
25527 processors) for which the code is optimized.
25530 @item -mcpu=@var{cpu-type}
25532 A deprecated synonym for @option{-mtune}.
25534 @item -mfpmath=@var{unit}
25536 Generate floating-point arithmetic for selected unit @var{unit}. The choices
25537 for @var{unit} are:
25541 Use the standard 387 floating-point coprocessor present on the majority of chips and
25542 emulated otherwise. Code compiled with this option runs almost everywhere.
25543 The temporary results are computed in 80-bit precision instead of the precision
25544 specified by the type, resulting in slightly different results compared to most
25545 of other chips. See @option{-ffloat-store} for more detailed description.
25547 This is the default choice for non-Darwin x86-32 targets.
25550 Use scalar floating-point instructions present in the SSE instruction set.
25551 This instruction set is supported by Pentium III and newer chips,
25552 and in the AMD line
25553 by Athlon-4, Athlon XP and Athlon MP chips. The earlier version of the SSE
25554 instruction set supports only single-precision arithmetic, thus the double and
25555 extended-precision arithmetic are still done using 387. A later version, present
25556 only in Pentium 4 and AMD x86-64 chips, supports double-precision
25559 For the x86-32 compiler, you must use @option{-march=@var{cpu-type}}, @option{-msse}
25560 or @option{-msse2} switches to enable SSE extensions and make this option
25561 effective. For the x86-64 compiler, these extensions are enabled by default.
25563 The resulting code should be considerably faster in the majority of cases and avoid
25564 the numerical instability problems of 387 code, but may break some existing
25565 code that expects temporaries to be 80 bits.
25567 This is the default choice for the x86-64 compiler, Darwin x86-32 targets,
25568 and the default choice for x86-32 targets with the SSE2 instruction set
25569 when @option{-ffast-math} is enabled.
25574 Attempt to utilize both instruction sets at once. This effectively doubles the
25575 amount of available registers, and on chips with separate execution units for
25576 387 and SSE the execution resources too. Use this option with care, as it is
25577 still experimental, because the GCC register allocator does not model separate
25578 functional units well, resulting in unstable performance.
25581 @item -masm=@var{dialect}
25582 @opindex masm=@var{dialect}
25583 Output assembly instructions using selected @var{dialect}. Also affects
25584 which dialect is used for basic @code{asm} (@pxref{Basic Asm}) and
25585 extended @code{asm} (@pxref{Extended Asm}). Supported choices (in dialect
25586 order) are @samp{att} or @samp{intel}. The default is @samp{att}. Darwin does
25587 not support @samp{intel}.
25590 @itemx -mno-ieee-fp
25592 @opindex mno-ieee-fp
25593 Control whether or not the compiler uses IEEE floating-point
25594 comparisons. These correctly handle the case where the result of a
25595 comparison is unordered.
25600 @opindex mhard-float
25601 Generate output containing 80387 instructions for floating point.
25606 @opindex msoft-float
25607 Generate output containing library calls for floating point.
25609 @strong{Warning:} the requisite libraries are not part of GCC@.
25610 Normally the facilities of the machine's usual C compiler are used, but
25611 this cannot be done directly in cross-compilation. You must make your
25612 own arrangements to provide suitable library functions for
25615 On machines where a function returns floating-point results in the 80387
25616 register stack, some floating-point opcodes may be emitted even if
25617 @option{-msoft-float} is used.
25619 @item -mno-fp-ret-in-387
25620 @opindex mno-fp-ret-in-387
25621 Do not use the FPU registers for return values of functions.
25623 The usual calling convention has functions return values of types
25624 @code{float} and @code{double} in an FPU register, even if there
25625 is no FPU@. The idea is that the operating system should emulate
25628 The option @option{-mno-fp-ret-in-387} causes such values to be returned
25629 in ordinary CPU registers instead.
25631 @item -mno-fancy-math-387
25632 @opindex mno-fancy-math-387
25633 Some 387 emulators do not support the @code{sin}, @code{cos} and
25634 @code{sqrt} instructions for the 387. Specify this option to avoid
25635 generating those instructions. This option is the default on
25636 OpenBSD and NetBSD@. This option is overridden when @option{-march}
25637 indicates that the target CPU always has an FPU and so the
25638 instruction does not need emulation. These
25639 instructions are not generated unless you also use the
25640 @option{-funsafe-math-optimizations} switch.
25642 @item -malign-double
25643 @itemx -mno-align-double
25644 @opindex malign-double
25645 @opindex mno-align-double
25646 Control whether GCC aligns @code{double}, @code{long double}, and
25647 @code{long long} variables on a two-word boundary or a one-word
25648 boundary. Aligning @code{double} variables on a two-word boundary
25649 produces code that runs somewhat faster on a Pentium at the
25650 expense of more memory.
25652 On x86-64, @option{-malign-double} is enabled by default.
25654 @strong{Warning:} if you use the @option{-malign-double} switch,
25655 structures containing the above types are aligned differently than
25656 the published application binary interface specifications for the x86-32
25657 and are not binary compatible with structures in code compiled
25658 without that switch.
25660 @item -m96bit-long-double
25661 @itemx -m128bit-long-double
25662 @opindex m96bit-long-double
25663 @opindex m128bit-long-double
25664 These switches control the size of @code{long double} type. The x86-32
25665 application binary interface specifies the size to be 96 bits,
25666 so @option{-m96bit-long-double} is the default in 32-bit mode.
25668 Modern architectures (Pentium and newer) prefer @code{long double}
25669 to be aligned to an 8- or 16-byte boundary. In arrays or structures
25670 conforming to the ABI, this is not possible. So specifying
25671 @option{-m128bit-long-double} aligns @code{long double}
25672 to a 16-byte boundary by padding the @code{long double} with an additional
25675 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
25676 its ABI specifies that @code{long double} is aligned on 16-byte boundary.
25678 Notice that neither of these options enable any extra precision over the x87
25679 standard of 80 bits for a @code{long double}.
25681 @strong{Warning:} if you override the default value for your target ABI, this
25682 changes the size of
25683 structures and arrays containing @code{long double} variables,
25684 as well as modifying the function calling convention for functions taking
25685 @code{long double}. Hence they are not binary-compatible
25686 with code compiled without that switch.
25688 @item -mlong-double-64
25689 @itemx -mlong-double-80
25690 @itemx -mlong-double-128
25691 @opindex mlong-double-64
25692 @opindex mlong-double-80
25693 @opindex mlong-double-128
25694 These switches control the size of @code{long double} type. A size
25695 of 64 bits makes the @code{long double} type equivalent to the @code{double}
25696 type. This is the default for 32-bit Bionic C library. A size
25697 of 128 bits makes the @code{long double} type equivalent to the
25698 @code{__float128} type. This is the default for 64-bit Bionic C library.
25700 @strong{Warning:} if you override the default value for your target ABI, this
25701 changes the size of
25702 structures and arrays containing @code{long double} variables,
25703 as well as modifying the function calling convention for functions taking
25704 @code{long double}. Hence they are not binary-compatible
25705 with code compiled without that switch.
25707 @item -malign-data=@var{type}
25708 @opindex malign-data
25709 Control how GCC aligns variables. Supported values for @var{type} are
25710 @samp{compat} uses increased alignment value compatible uses GCC 4.8
25711 and earlier, @samp{abi} uses alignment value as specified by the
25712 psABI, and @samp{cacheline} uses increased alignment value to match
25713 the cache line size. @samp{compat} is the default.
25715 @item -mlarge-data-threshold=@var{threshold}
25716 @opindex mlarge-data-threshold
25717 When @option{-mcmodel=medium} is specified, data objects larger than
25718 @var{threshold} are placed in the large data section. This value must be the
25719 same across all objects linked into the binary, and defaults to 65535.
25723 Use a different function-calling convention, in which functions that
25724 take a fixed number of arguments return with the @code{ret @var{num}}
25725 instruction, which pops their arguments while returning. This saves one
25726 instruction in the caller since there is no need to pop the arguments
25729 You can specify that an individual function is called with this calling
25730 sequence with the function attribute @code{stdcall}. You can also
25731 override the @option{-mrtd} option by using the function attribute
25732 @code{cdecl}. @xref{Function Attributes}.
25734 @strong{Warning:} this calling convention is incompatible with the one
25735 normally used on Unix, so you cannot use it if you need to call
25736 libraries compiled with the Unix compiler.
25738 Also, you must provide function prototypes for all functions that
25739 take variable numbers of arguments (including @code{printf});
25740 otherwise incorrect code is generated for calls to those
25743 In addition, seriously incorrect code results if you call a
25744 function with too many arguments. (Normally, extra arguments are
25745 harmlessly ignored.)
25747 @item -mregparm=@var{num}
25749 Control how many registers are used to pass integer arguments. By
25750 default, no registers are used to pass arguments, and at most 3
25751 registers can be used. You can control this behavior for a specific
25752 function by using the function attribute @code{regparm}.
25753 @xref{Function Attributes}.
25755 @strong{Warning:} if you use this switch, and
25756 @var{num} is nonzero, then you must build all modules with the same
25757 value, including any libraries. This includes the system libraries and
25761 @opindex msseregparm
25762 Use SSE register passing conventions for float and double arguments
25763 and return values. You can control this behavior for a specific
25764 function by using the function attribute @code{sseregparm}.
25765 @xref{Function Attributes}.
25767 @strong{Warning:} if you use this switch then you must build all
25768 modules with the same value, including any libraries. This includes
25769 the system libraries and startup modules.
25771 @item -mvect8-ret-in-mem
25772 @opindex mvect8-ret-in-mem
25773 Return 8-byte vectors in memory instead of MMX registers. This is the
25774 default on Solaris@tie{}8 and 9 and VxWorks to match the ABI of the Sun
25775 Studio compilers until version 12. Later compiler versions (starting
25776 with Studio 12 Update@tie{}1) follow the ABI used by other x86 targets, which
25777 is the default on Solaris@tie{}10 and later. @emph{Only} use this option if
25778 you need to remain compatible with existing code produced by those
25779 previous compiler versions or older versions of GCC@.
25788 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
25789 is specified, the significands of results of floating-point operations are
25790 rounded to 24 bits (single precision); @option{-mpc64} rounds the
25791 significands of results of floating-point operations to 53 bits (double
25792 precision) and @option{-mpc80} rounds the significands of results of
25793 floating-point operations to 64 bits (extended double precision), which is
25794 the default. When this option is used, floating-point operations in higher
25795 precisions are not available to the programmer without setting the FPU
25796 control word explicitly.
25798 Setting the rounding of floating-point operations to less than the default
25799 80 bits can speed some programs by 2% or more. Note that some mathematical
25800 libraries assume that extended-precision (80-bit) floating-point operations
25801 are enabled by default; routines in such libraries could suffer significant
25802 loss of accuracy, typically through so-called ``catastrophic cancellation'',
25803 when this option is used to set the precision to less than extended precision.
25805 @item -mstackrealign
25806 @opindex mstackrealign
25807 Realign the stack at entry. On the x86, the @option{-mstackrealign}
25808 option generates an alternate prologue and epilogue that realigns the
25809 run-time stack if necessary. This supports mixing legacy codes that keep
25810 4-byte stack alignment with modern codes that keep 16-byte stack alignment for
25811 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
25812 applicable to individual functions.
25814 @item -mpreferred-stack-boundary=@var{num}
25815 @opindex mpreferred-stack-boundary
25816 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
25817 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
25818 the default is 4 (16 bytes or 128 bits).
25820 @strong{Warning:} When generating code for the x86-64 architecture with
25821 SSE extensions disabled, @option{-mpreferred-stack-boundary=3} can be
25822 used to keep the stack boundary aligned to 8 byte boundary. Since
25823 x86-64 ABI require 16 byte stack alignment, this is ABI incompatible and
25824 intended to be used in controlled environment where stack space is
25825 important limitation. This option leads to wrong code when functions
25826 compiled with 16 byte stack alignment (such as functions from a standard
25827 library) are called with misaligned stack. In this case, SSE
25828 instructions may lead to misaligned memory access traps. In addition,
25829 variable arguments are handled incorrectly for 16 byte aligned
25830 objects (including x87 long double and __int128), leading to wrong
25831 results. You must build all modules with
25832 @option{-mpreferred-stack-boundary=3}, including any libraries. This
25833 includes the system libraries and startup modules.
25835 @item -mincoming-stack-boundary=@var{num}
25836 @opindex mincoming-stack-boundary
25837 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
25838 boundary. If @option{-mincoming-stack-boundary} is not specified,
25839 the one specified by @option{-mpreferred-stack-boundary} is used.
25841 On Pentium and Pentium Pro, @code{double} and @code{long double} values
25842 should be aligned to an 8-byte boundary (see @option{-malign-double}) or
25843 suffer significant run time performance penalties. On Pentium III, the
25844 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
25845 properly if it is not 16-byte aligned.
25847 To ensure proper alignment of this values on the stack, the stack boundary
25848 must be as aligned as that required by any value stored on the stack.
25849 Further, every function must be generated such that it keeps the stack
25850 aligned. Thus calling a function compiled with a higher preferred
25851 stack boundary from a function compiled with a lower preferred stack
25852 boundary most likely misaligns the stack. It is recommended that
25853 libraries that use callbacks always use the default setting.
25855 This extra alignment does consume extra stack space, and generally
25856 increases code size. Code that is sensitive to stack space usage, such
25857 as embedded systems and operating system kernels, may want to reduce the
25858 preferred alignment to @option{-mpreferred-stack-boundary=2}.
25915 @itemx -mavx512ifma
25916 @opindex mavx512ifma
25918 @itemx -mavx512vbmi
25919 @opindex mavx512vbmi
25931 @opindex mclfushopt
25948 @itemx -mprefetchwt1
25949 @opindex mprefetchwt1
26013 @itemx -mavx512vbmi2
26014 @opindex mavx512vbmi2
26018 These switches enable the use of instructions in the MMX, SSE,
26019 SSE2, SSE3, SSSE3, SSE4.1, AVX, AVX2, AVX512F, AVX512PF, AVX512ER, AVX512CD,
26020 SHA, AES, PCLMUL, FSGSBASE, RDRND, F16C, FMA, SSE4A, FMA4, XOP, LWP, ABM,
26021 AVX512VL, AVX512BW, AVX512DQ, AVX512IFMA, AVX512VBMI, BMI, BMI2,
26022 FXSR, XSAVE, XSAVEOPT, LZCNT, RTM, MPX, MWAITX, PKU, IBT, SHSTK, AVX512VBMI2,
26023 GFNI, 3DNow!@: or enhanced 3DNow!@: extended instruction sets. Each has a
26024 corresponding @option{-mno-} option to disable use of these instructions.
26026 These extensions are also available as built-in functions: see
26027 @ref{x86 Built-in Functions}, for details of the functions enabled and
26028 disabled by these switches.
26030 To generate SSE/SSE2 instructions automatically from floating-point
26031 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
26033 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
26034 generates new AVX instructions or AVX equivalence for all SSEx instructions
26037 These options enable GCC to use these extended instructions in
26038 generated code, even without @option{-mfpmath=sse}. Applications that
26039 perform run-time CPU detection must compile separate files for each
26040 supported architecture, using the appropriate flags. In particular,
26041 the file containing the CPU detection code should be compiled without
26044 The @option{-mcet} option turns on the @option{-mibt} and @option{-mshstk}
26045 options. The @option{-mibt} option enables indirect branch tracking support
26046 and the @option{-mshstk} option enables shadow stack support from
26047 Intel Control-flow Enforcement Technology (CET). The compiler also provides
26048 a number of built-in functions for fine-grained control in a CET-based
26049 application. See @xref{x86 Built-in Functions}, for more information.
26051 @item -mdump-tune-features
26052 @opindex mdump-tune-features
26053 This option instructs GCC to dump the names of the x86 performance
26054 tuning features and default settings. The names can be used in
26055 @option{-mtune-ctrl=@var{feature-list}}.
26057 @item -mtune-ctrl=@var{feature-list}
26058 @opindex mtune-ctrl=@var{feature-list}
26059 This option is used to do fine grain control of x86 code generation features.
26060 @var{feature-list} is a comma separated list of @var{feature} names. See also
26061 @option{-mdump-tune-features}. When specified, the @var{feature} is turned
26062 on if it is not preceded with @samp{^}, otherwise, it is turned off.
26063 @option{-mtune-ctrl=@var{feature-list}} is intended to be used by GCC
26064 developers. Using it may lead to code paths not covered by testing and can
26065 potentially result in compiler ICEs or runtime errors.
26068 @opindex mno-default
26069 This option instructs GCC to turn off all tunable features. See also
26070 @option{-mtune-ctrl=@var{feature-list}} and @option{-mdump-tune-features}.
26074 This option instructs GCC to emit a @code{cld} instruction in the prologue
26075 of functions that use string instructions. String instructions depend on
26076 the DF flag to select between autoincrement or autodecrement mode. While the
26077 ABI specifies the DF flag to be cleared on function entry, some operating
26078 systems violate this specification by not clearing the DF flag in their
26079 exception dispatchers. The exception handler can be invoked with the DF flag
26080 set, which leads to wrong direction mode when string instructions are used.
26081 This option can be enabled by default on 32-bit x86 targets by configuring
26082 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
26083 instructions can be suppressed with the @option{-mno-cld} compiler option
26087 @opindex mvzeroupper
26088 This option instructs GCC to emit a @code{vzeroupper} instruction
26089 before a transfer of control flow out of the function to minimize
26090 the AVX to SSE transition penalty as well as remove unnecessary @code{zeroupper}
26093 @item -mprefer-avx128
26094 @opindex mprefer-avx128
26095 This option instructs GCC to use 128-bit AVX instructions instead of
26096 256-bit AVX instructions in the auto-vectorizer.
26098 @item -mprefer-vector-width=@var{opt}
26099 @opindex mprefer-vector-width
26100 This option instructs GCC to use @var{opt}-bit vector width in instructions
26101 instead of default on the selected platform.
26105 No extra limitations applied to GCC other than defined by the selected platform.
26108 Prefer 128-bit vector width for instructions.
26111 Prefer 256-bit vector width for instructions.
26114 Prefer 512-bit vector width for instructions.
26119 This option enables GCC to generate @code{CMPXCHG16B} instructions in 64-bit
26120 code to implement compare-and-exchange operations on 16-byte aligned 128-bit
26121 objects. This is useful for atomic updates of data structures exceeding one
26122 machine word in size. The compiler uses this instruction to implement
26123 @ref{__sync Builtins}. However, for @ref{__atomic Builtins} operating on
26124 128-bit integers, a library call is always used.
26128 This option enables generation of @code{SAHF} instructions in 64-bit code.
26129 Early Intel Pentium 4 CPUs with Intel 64 support,
26130 prior to the introduction of Pentium 4 G1 step in December 2005,
26131 lacked the @code{LAHF} and @code{SAHF} instructions
26132 which are supported by AMD64.
26133 These are load and store instructions, respectively, for certain status flags.
26134 In 64-bit mode, the @code{SAHF} instruction is used to optimize @code{fmod},
26135 @code{drem}, and @code{remainder} built-in functions;
26136 see @ref{Other Builtins} for details.
26140 This option enables use of the @code{movbe} instruction to implement
26141 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
26145 This option tells the compiler to use indirect branch tracking support
26146 (for indirect calls and jumps) from x86 Control-flow Enforcement
26147 Technology (CET). The option has effect only if the
26148 @option{-fcf-protection=full} or @option{-fcf-protection=branch} option
26149 is specified. The option @option{-mibt} is on by default when the
26150 @code{-mcet} option is specified.
26154 This option tells the compiler to use shadow stack support (return
26155 address tracking) from x86 Control-flow Enforcement Technology (CET).
26156 The option has effect only if the @option{-fcf-protection=full} or
26157 @option{-fcf-protection=return} option is specified. The option
26158 @option{-mshstk} is on by default when the @option{-mcet} option is
26163 This option enables built-in functions @code{__builtin_ia32_crc32qi},
26164 @code{__builtin_ia32_crc32hi}, @code{__builtin_ia32_crc32si} and
26165 @code{__builtin_ia32_crc32di} to generate the @code{crc32} machine instruction.
26169 This option enables use of @code{RCPSS} and @code{RSQRTSS} instructions
26170 (and their vectorized variants @code{RCPPS} and @code{RSQRTPS})
26171 with an additional Newton-Raphson step
26172 to increase precision instead of @code{DIVSS} and @code{SQRTSS}
26173 (and their vectorized
26174 variants) for single-precision floating-point arguments. These instructions
26175 are generated only when @option{-funsafe-math-optimizations} is enabled
26176 together with @option{-ffinite-math-only} and @option{-fno-trapping-math}.
26177 Note that while the throughput of the sequence is higher than the throughput
26178 of the non-reciprocal instruction, the precision of the sequence can be
26179 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
26181 Note that GCC implements @code{1.0f/sqrtf(@var{x})} in terms of @code{RSQRTSS}
26182 (or @code{RSQRTPS}) already with @option{-ffast-math} (or the above option
26183 combination), and doesn't need @option{-mrecip}.
26185 Also note that GCC emits the above sequence with additional Newton-Raphson step
26186 for vectorized single-float division and vectorized @code{sqrtf(@var{x})}
26187 already with @option{-ffast-math} (or the above option combination), and
26188 doesn't need @option{-mrecip}.
26190 @item -mrecip=@var{opt}
26191 @opindex mrecip=opt
26192 This option controls which reciprocal estimate instructions
26193 may be used. @var{opt} is a comma-separated list of options, which may
26194 be preceded by a @samp{!} to invert the option:
26198 Enable all estimate instructions.
26201 Enable the default instructions, equivalent to @option{-mrecip}.
26204 Disable all estimate instructions, equivalent to @option{-mno-recip}.
26207 Enable the approximation for scalar division.
26210 Enable the approximation for vectorized division.
26213 Enable the approximation for scalar square root.
26216 Enable the approximation for vectorized square root.
26219 So, for example, @option{-mrecip=all,!sqrt} enables
26220 all of the reciprocal approximations, except for square root.
26222 @item -mveclibabi=@var{type}
26223 @opindex mveclibabi
26224 Specifies the ABI type to use for vectorizing intrinsics using an
26225 external library. Supported values for @var{type} are @samp{svml}
26226 for the Intel short
26227 vector math library and @samp{acml} for the AMD math core library.
26228 To use this option, both @option{-ftree-vectorize} and
26229 @option{-funsafe-math-optimizations} have to be enabled, and an SVML or ACML
26230 ABI-compatible library must be specified at link time.
26232 GCC currently emits calls to @code{vmldExp2},
26233 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
26234 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
26235 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
26236 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
26237 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
26238 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
26239 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
26240 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
26241 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
26242 function type when @option{-mveclibabi=svml} is used, and @code{__vrd2_sin},
26243 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
26244 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
26245 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
26246 @code{__vrs4_log10f} and @code{__vrs4_powf} for the corresponding function type
26247 when @option{-mveclibabi=acml} is used.
26249 @item -mabi=@var{name}
26251 Generate code for the specified calling convention. Permissible values
26252 are @samp{sysv} for the ABI used on GNU/Linux and other systems, and
26253 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
26254 ABI when targeting Microsoft Windows and the SysV ABI on all other systems.
26255 You can control this behavior for specific functions by
26256 using the function attributes @code{ms_abi} and @code{sysv_abi}.
26257 @xref{Function Attributes}.
26259 @item -mforce-indirect-call
26260 @opindex mforce-indirect-call
26261 Force all calls to functions to be indirect. This is useful
26262 when using Intel Processor Trace where it generates more precise timing
26263 information for function calls.
26265 @item -mcall-ms2sysv-xlogues
26266 @opindex mcall-ms2sysv-xlogues
26267 @opindex mno-call-ms2sysv-xlogues
26268 Due to differences in 64-bit ABIs, any Microsoft ABI function that calls a
26269 System V ABI function must consider RSI, RDI and XMM6-15 as clobbered. By
26270 default, the code for saving and restoring these registers is emitted inline,
26271 resulting in fairly lengthy prologues and epilogues. Using
26272 @option{-mcall-ms2sysv-xlogues} emits prologues and epilogues that
26273 use stubs in the static portion of libgcc to perform these saves and restores,
26274 thus reducing function size at the cost of a few extra instructions.
26276 @item -mtls-dialect=@var{type}
26277 @opindex mtls-dialect
26278 Generate code to access thread-local storage using the @samp{gnu} or
26279 @samp{gnu2} conventions. @samp{gnu} is the conservative default;
26280 @samp{gnu2} is more efficient, but it may add compile- and run-time
26281 requirements that cannot be satisfied on all systems.
26284 @itemx -mno-push-args
26285 @opindex mpush-args
26286 @opindex mno-push-args
26287 Use PUSH operations to store outgoing parameters. This method is shorter
26288 and usually equally fast as method using SUB/MOV operations and is enabled
26289 by default. In some cases disabling it may improve performance because of
26290 improved scheduling and reduced dependencies.
26292 @item -maccumulate-outgoing-args
26293 @opindex maccumulate-outgoing-args
26294 If enabled, the maximum amount of space required for outgoing arguments is
26295 computed in the function prologue. This is faster on most modern CPUs
26296 because of reduced dependencies, improved scheduling and reduced stack usage
26297 when the preferred stack boundary is not equal to 2. The drawback is a notable
26298 increase in code size. This switch implies @option{-mno-push-args}.
26302 Support thread-safe exception handling on MinGW. Programs that rely
26303 on thread-safe exception handling must compile and link all code with the
26304 @option{-mthreads} option. When compiling, @option{-mthreads} defines
26305 @option{-D_MT}; when linking, it links in a special thread helper library
26306 @option{-lmingwthrd} which cleans up per-thread exception-handling data.
26308 @item -mms-bitfields
26309 @itemx -mno-ms-bitfields
26310 @opindex mms-bitfields
26311 @opindex mno-ms-bitfields
26313 Enable/disable bit-field layout compatible with the native Microsoft
26316 If @code{packed} is used on a structure, or if bit-fields are used,
26317 it may be that the Microsoft ABI lays out the structure differently
26318 than the way GCC normally does. Particularly when moving packed
26319 data between functions compiled with GCC and the native Microsoft compiler
26320 (either via function call or as data in a file), it may be necessary to access
26323 This option is enabled by default for Microsoft Windows
26324 targets. This behavior can also be controlled locally by use of variable
26325 or type attributes. For more information, see @ref{x86 Variable Attributes}
26326 and @ref{x86 Type Attributes}.
26328 The Microsoft structure layout algorithm is fairly simple with the exception
26329 of the bit-field packing.
26330 The padding and alignment of members of structures and whether a bit-field
26331 can straddle a storage-unit boundary are determine by these rules:
26334 @item Structure members are stored sequentially in the order in which they are
26335 declared: the first member has the lowest memory address and the last member
26338 @item Every data object has an alignment requirement. The alignment requirement
26339 for all data except structures, unions, and arrays is either the size of the
26340 object or the current packing size (specified with either the
26341 @code{aligned} attribute or the @code{pack} pragma),
26342 whichever is less. For structures, unions, and arrays,
26343 the alignment requirement is the largest alignment requirement of its members.
26344 Every object is allocated an offset so that:
26347 offset % alignment_requirement == 0
26350 @item Adjacent bit-fields are packed into the same 1-, 2-, or 4-byte allocation
26351 unit if the integral types are the same size and if the next bit-field fits
26352 into the current allocation unit without crossing the boundary imposed by the
26353 common alignment requirements of the bit-fields.
26356 MSVC interprets zero-length bit-fields in the following ways:
26359 @item If a zero-length bit-field is inserted between two bit-fields that
26360 are normally coalesced, the bit-fields are not coalesced.
26367 unsigned long bf_1 : 12;
26369 unsigned long bf_2 : 12;
26374 The size of @code{t1} is 8 bytes with the zero-length bit-field. If the
26375 zero-length bit-field were removed, @code{t1}'s size would be 4 bytes.
26377 @item If a zero-length bit-field is inserted after a bit-field, @code{foo}, and the
26378 alignment of the zero-length bit-field is greater than the member that follows it,
26379 @code{bar}, @code{bar} is aligned as the type of the zero-length bit-field.
26400 For @code{t2}, @code{bar} is placed at offset 2, rather than offset 1.
26401 Accordingly, the size of @code{t2} is 4. For @code{t3}, the zero-length
26402 bit-field does not affect the alignment of @code{bar} or, as a result, the size
26405 Taking this into account, it is important to note the following:
26408 @item If a zero-length bit-field follows a normal bit-field, the type of the
26409 zero-length bit-field may affect the alignment of the structure as whole. For
26410 example, @code{t2} has a size of 4 bytes, since the zero-length bit-field follows a
26411 normal bit-field, and is of type short.
26413 @item Even if a zero-length bit-field is not followed by a normal bit-field, it may
26414 still affect the alignment of the structure:
26425 Here, @code{t4} takes up 4 bytes.
26428 @item Zero-length bit-fields following non-bit-field members are ignored:
26440 Here, @code{t5} takes up 2 bytes.
26444 @item -mno-align-stringops
26445 @opindex mno-align-stringops
26446 Do not align the destination of inlined string operations. This switch reduces
26447 code size and improves performance in case the destination is already aligned,
26448 but GCC doesn't know about it.
26450 @item -minline-all-stringops
26451 @opindex minline-all-stringops
26452 By default GCC inlines string operations only when the destination is
26453 known to be aligned to least a 4-byte boundary.
26454 This enables more inlining and increases code
26455 size, but may improve performance of code that depends on fast
26456 @code{memcpy}, @code{strlen},
26457 and @code{memset} for short lengths.
26459 @item -minline-stringops-dynamically
26460 @opindex minline-stringops-dynamically
26461 For string operations of unknown size, use run-time checks with
26462 inline code for small blocks and a library call for large blocks.
26464 @item -mstringop-strategy=@var{alg}
26465 @opindex mstringop-strategy=@var{alg}
26466 Override the internal decision heuristic for the particular algorithm to use
26467 for inlining string operations. The allowed values for @var{alg} are:
26473 Expand using i386 @code{rep} prefix of the specified size.
26477 @itemx unrolled_loop
26478 Expand into an inline loop.
26481 Always use a library call.
26484 @item -mmemcpy-strategy=@var{strategy}
26485 @opindex mmemcpy-strategy=@var{strategy}
26486 Override the internal decision heuristic to decide if @code{__builtin_memcpy}
26487 should be inlined and what inline algorithm to use when the expected size
26488 of the copy operation is known. @var{strategy}
26489 is a comma-separated list of @var{alg}:@var{max_size}:@var{dest_align} triplets.
26490 @var{alg} is specified in @option{-mstringop-strategy}, @var{max_size} specifies
26491 the max byte size with which inline algorithm @var{alg} is allowed. For the last
26492 triplet, the @var{max_size} must be @code{-1}. The @var{max_size} of the triplets
26493 in the list must be specified in increasing order. The minimal byte size for
26494 @var{alg} is @code{0} for the first triplet and @code{@var{max_size} + 1} of the
26497 @item -mmemset-strategy=@var{strategy}
26498 @opindex mmemset-strategy=@var{strategy}
26499 The option is similar to @option{-mmemcpy-strategy=} except that it is to control
26500 @code{__builtin_memset} expansion.
26502 @item -momit-leaf-frame-pointer
26503 @opindex momit-leaf-frame-pointer
26504 Don't keep the frame pointer in a register for leaf functions. This
26505 avoids the instructions to save, set up, and restore frame pointers and
26506 makes an extra register available in leaf functions. The option
26507 @option{-fomit-leaf-frame-pointer} removes the frame pointer for leaf functions,
26508 which might make debugging harder.
26510 @item -mtls-direct-seg-refs
26511 @itemx -mno-tls-direct-seg-refs
26512 @opindex mtls-direct-seg-refs
26513 Controls whether TLS variables may be accessed with offsets from the
26514 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
26515 or whether the thread base pointer must be added. Whether or not this
26516 is valid depends on the operating system, and whether it maps the
26517 segment to cover the entire TLS area.
26519 For systems that use the GNU C Library, the default is on.
26522 @itemx -mno-sse2avx
26524 Specify that the assembler should encode SSE instructions with VEX
26525 prefix. The option @option{-mavx} turns this on by default.
26530 If profiling is active (@option{-pg}), put the profiling
26531 counter call before the prologue.
26532 Note: On x86 architectures the attribute @code{ms_hook_prologue}
26533 isn't possible at the moment for @option{-mfentry} and @option{-pg}.
26535 @item -mrecord-mcount
26536 @itemx -mno-record-mcount
26537 @opindex mrecord-mcount
26538 If profiling is active (@option{-pg}), generate a __mcount_loc section
26539 that contains pointers to each profiling call. This is useful for
26540 automatically patching and out calls.
26543 @itemx -mno-nop-mcount
26544 @opindex mnop-mcount
26545 If profiling is active (@option{-pg}), generate the calls to
26546 the profiling functions as NOPs. This is useful when they
26547 should be patched in later dynamically. This is likely only
26548 useful together with @option{-mrecord-mcount}.
26550 @item -mskip-rax-setup
26551 @itemx -mno-skip-rax-setup
26552 @opindex mskip-rax-setup
26553 When generating code for the x86-64 architecture with SSE extensions
26554 disabled, @option{-mskip-rax-setup} can be used to skip setting up RAX
26555 register when there are no variable arguments passed in vector registers.
26557 @strong{Warning:} Since RAX register is used to avoid unnecessarily
26558 saving vector registers on stack when passing variable arguments, the
26559 impacts of this option are callees may waste some stack space,
26560 misbehave or jump to a random location. GCC 4.4 or newer don't have
26561 those issues, regardless the RAX register value.
26564 @itemx -mno-8bit-idiv
26565 @opindex m8bit-idiv
26566 On some processors, like Intel Atom, 8-bit unsigned integer divide is
26567 much faster than 32-bit/64-bit integer divide. This option generates a
26568 run-time check. If both dividend and divisor are within range of 0
26569 to 255, 8-bit unsigned integer divide is used instead of
26570 32-bit/64-bit integer divide.
26572 @item -mavx256-split-unaligned-load
26573 @itemx -mavx256-split-unaligned-store
26574 @opindex mavx256-split-unaligned-load
26575 @opindex mavx256-split-unaligned-store
26576 Split 32-byte AVX unaligned load and store.
26578 @item -mstack-protector-guard=@var{guard}
26579 @itemx -mstack-protector-guard-reg=@var{reg}
26580 @itemx -mstack-protector-guard-offset=@var{offset}
26581 @opindex mstack-protector-guard
26582 @opindex mstack-protector-guard-reg
26583 @opindex mstack-protector-guard-offset
26584 Generate stack protection code using canary at @var{guard}. Supported
26585 locations are @samp{global} for global canary or @samp{tls} for per-thread
26586 canary in the TLS block (the default). This option has effect only when
26587 @option{-fstack-protector} or @option{-fstack-protector-all} is specified.
26589 With the latter choice the options
26590 @option{-mstack-protector-guard-reg=@var{reg}} and
26591 @option{-mstack-protector-guard-offset=@var{offset}} furthermore specify
26592 which segment register (@code{%fs} or @code{%gs}) to use as base register
26593 for reading the canary, and from what offset from that base register.
26594 The default for those is as specified in the relevant ABI.
26596 @item -mmitigate-rop
26597 @opindex mmitigate-rop
26598 Try to avoid generating code sequences that contain unintended return
26599 opcodes, to mitigate against certain forms of attack. At the moment,
26600 this option is limited in what it can do and should not be relied
26601 on to provide serious protection.
26603 @item -mgeneral-regs-only
26604 @opindex mgeneral-regs-only
26605 Generate code that uses only the general-purpose registers. This
26606 prevents the compiler from using floating-point, vector, mask and bound
26611 These @samp{-m} switches are supported in addition to the above
26612 on x86-64 processors in 64-bit environments.
26625 Generate code for a 16-bit, 32-bit or 64-bit environment.
26626 The @option{-m32} option sets @code{int}, @code{long}, and pointer types
26628 generates code that runs on any i386 system.
26630 The @option{-m64} option sets @code{int} to 32 bits and @code{long} and pointer
26631 types to 64 bits, and generates code for the x86-64 architecture.
26632 For Darwin only the @option{-m64} option also turns off the @option{-fno-pic}
26633 and @option{-mdynamic-no-pic} options.
26635 The @option{-mx32} option sets @code{int}, @code{long}, and pointer types
26637 generates code for the x86-64 architecture.
26639 The @option{-m16} option is the same as @option{-m32}, except for that
26640 it outputs the @code{.code16gcc} assembly directive at the beginning of
26641 the assembly output so that the binary can run in 16-bit mode.
26643 The @option{-miamcu} option generates code which conforms to Intel MCU
26644 psABI. It requires the @option{-m32} option to be turned on.
26646 @item -mno-red-zone
26647 @opindex mno-red-zone
26648 Do not use a so-called ``red zone'' for x86-64 code. The red zone is mandated
26649 by the x86-64 ABI; it is a 128-byte area beyond the location of the
26650 stack pointer that is not modified by signal or interrupt handlers
26651 and therefore can be used for temporary data without adjusting the stack
26652 pointer. The flag @option{-mno-red-zone} disables this red zone.
26654 @item -mcmodel=small
26655 @opindex mcmodel=small
26656 Generate code for the small code model: the program and its symbols must
26657 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
26658 Programs can be statically or dynamically linked. This is the default
26661 @item -mcmodel=kernel
26662 @opindex mcmodel=kernel
26663 Generate code for the kernel code model. The kernel runs in the
26664 negative 2 GB of the address space.
26665 This model has to be used for Linux kernel code.
26667 @item -mcmodel=medium
26668 @opindex mcmodel=medium
26669 Generate code for the medium model: the program is linked in the lower 2
26670 GB of the address space. Small symbols are also placed there. Symbols
26671 with sizes larger than @option{-mlarge-data-threshold} are put into
26672 large data or BSS sections and can be located above 2GB. Programs can
26673 be statically or dynamically linked.
26675 @item -mcmodel=large
26676 @opindex mcmodel=large
26677 Generate code for the large model. This model makes no assumptions
26678 about addresses and sizes of sections.
26680 @item -maddress-mode=long
26681 @opindex maddress-mode=long
26682 Generate code for long address mode. This is only supported for 64-bit
26683 and x32 environments. It is the default address mode for 64-bit
26686 @item -maddress-mode=short
26687 @opindex maddress-mode=short
26688 Generate code for short address mode. This is only supported for 32-bit
26689 and x32 environments. It is the default address mode for 32-bit and
26693 @node x86 Windows Options
26694 @subsection x86 Windows Options
26695 @cindex x86 Windows Options
26696 @cindex Windows Options for x86
26698 These additional options are available for Microsoft Windows targets:
26704 specifies that a console application is to be generated, by
26705 instructing the linker to set the PE header subsystem type
26706 required for console applications.
26707 This option is available for Cygwin and MinGW targets and is
26708 enabled by default on those targets.
26712 This option is available for Cygwin and MinGW targets. It
26713 specifies that a DLL---a dynamic link library---is to be
26714 generated, enabling the selection of the required runtime
26715 startup object and entry point.
26717 @item -mnop-fun-dllimport
26718 @opindex mnop-fun-dllimport
26719 This option is available for Cygwin and MinGW targets. It
26720 specifies that the @code{dllimport} attribute should be ignored.
26724 This option is available for MinGW targets. It specifies
26725 that MinGW-specific thread support is to be used.
26729 This option is available for MinGW-w64 targets. It causes
26730 the @code{UNICODE} preprocessor macro to be predefined, and
26731 chooses Unicode-capable runtime startup code.
26735 This option is available for Cygwin and MinGW targets. It
26736 specifies that the typical Microsoft Windows predefined macros are to
26737 be set in the pre-processor, but does not influence the choice
26738 of runtime library/startup code.
26742 This option is available for Cygwin and MinGW targets. It
26743 specifies that a GUI application is to be generated by
26744 instructing the linker to set the PE header subsystem type
26747 @item -fno-set-stack-executable
26748 @opindex fno-set-stack-executable
26749 This option is available for MinGW targets. It specifies that
26750 the executable flag for the stack used by nested functions isn't
26751 set. This is necessary for binaries running in kernel mode of
26752 Microsoft Windows, as there the User32 API, which is used to set executable
26753 privileges, isn't available.
26755 @item -fwritable-relocated-rdata
26756 @opindex fno-writable-relocated-rdata
26757 This option is available for MinGW and Cygwin targets. It specifies
26758 that relocated-data in read-only section is put into the @code{.data}
26759 section. This is a necessary for older runtimes not supporting
26760 modification of @code{.rdata} sections for pseudo-relocation.
26762 @item -mpe-aligned-commons
26763 @opindex mpe-aligned-commons
26764 This option is available for Cygwin and MinGW targets. It
26765 specifies that the GNU extension to the PE file format that
26766 permits the correct alignment of COMMON variables should be
26767 used when generating code. It is enabled by default if
26768 GCC detects that the target assembler found during configuration
26769 supports the feature.
26772 See also under @ref{x86 Options} for standard options.
26774 @node Xstormy16 Options
26775 @subsection Xstormy16 Options
26776 @cindex Xstormy16 Options
26778 These options are defined for Xstormy16:
26783 Choose startup files and linker script suitable for the simulator.
26786 @node Xtensa Options
26787 @subsection Xtensa Options
26788 @cindex Xtensa Options
26790 These options are supported for Xtensa targets:
26794 @itemx -mno-const16
26796 @opindex mno-const16
26797 Enable or disable use of @code{CONST16} instructions for loading
26798 constant values. The @code{CONST16} instruction is currently not a
26799 standard option from Tensilica. When enabled, @code{CONST16}
26800 instructions are always used in place of the standard @code{L32R}
26801 instructions. The use of @code{CONST16} is enabled by default only if
26802 the @code{L32R} instruction is not available.
26805 @itemx -mno-fused-madd
26806 @opindex mfused-madd
26807 @opindex mno-fused-madd
26808 Enable or disable use of fused multiply/add and multiply/subtract
26809 instructions in the floating-point option. This has no effect if the
26810 floating-point option is not also enabled. Disabling fused multiply/add
26811 and multiply/subtract instructions forces the compiler to use separate
26812 instructions for the multiply and add/subtract operations. This may be
26813 desirable in some cases where strict IEEE 754-compliant results are
26814 required: the fused multiply add/subtract instructions do not round the
26815 intermediate result, thereby producing results with @emph{more} bits of
26816 precision than specified by the IEEE standard. Disabling fused multiply
26817 add/subtract instructions also ensures that the program output is not
26818 sensitive to the compiler's ability to combine multiply and add/subtract
26821 @item -mserialize-volatile
26822 @itemx -mno-serialize-volatile
26823 @opindex mserialize-volatile
26824 @opindex mno-serialize-volatile
26825 When this option is enabled, GCC inserts @code{MEMW} instructions before
26826 @code{volatile} memory references to guarantee sequential consistency.
26827 The default is @option{-mserialize-volatile}. Use
26828 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
26830 @item -mforce-no-pic
26831 @opindex mforce-no-pic
26832 For targets, like GNU/Linux, where all user-mode Xtensa code must be
26833 position-independent code (PIC), this option disables PIC for compiling
26836 @item -mtext-section-literals
26837 @itemx -mno-text-section-literals
26838 @opindex mtext-section-literals
26839 @opindex mno-text-section-literals
26840 These options control the treatment of literal pools. The default is
26841 @option{-mno-text-section-literals}, which places literals in a separate
26842 section in the output file. This allows the literal pool to be placed
26843 in a data RAM/ROM, and it also allows the linker to combine literal
26844 pools from separate object files to remove redundant literals and
26845 improve code size. With @option{-mtext-section-literals}, the literals
26846 are interspersed in the text section in order to keep them as close as
26847 possible to their references. This may be necessary for large assembly
26848 files. Literals for each function are placed right before that function.
26850 @item -mauto-litpools
26851 @itemx -mno-auto-litpools
26852 @opindex mauto-litpools
26853 @opindex mno-auto-litpools
26854 These options control the treatment of literal pools. The default is
26855 @option{-mno-auto-litpools}, which places literals in a separate
26856 section in the output file unless @option{-mtext-section-literals} is
26857 used. With @option{-mauto-litpools} the literals are interspersed in
26858 the text section by the assembler. Compiler does not produce explicit
26859 @code{.literal} directives and loads literals into registers with
26860 @code{MOVI} instructions instead of @code{L32R} to let the assembler
26861 do relaxation and place literals as necessary. This option allows
26862 assembler to create several literal pools per function and assemble
26863 very big functions, which may not be possible with
26864 @option{-mtext-section-literals}.
26866 @item -mtarget-align
26867 @itemx -mno-target-align
26868 @opindex mtarget-align
26869 @opindex mno-target-align
26870 When this option is enabled, GCC instructs the assembler to
26871 automatically align instructions to reduce branch penalties at the
26872 expense of some code density. The assembler attempts to widen density
26873 instructions to align branch targets and the instructions following call
26874 instructions. If there are not enough preceding safe density
26875 instructions to align a target, no widening is performed. The
26876 default is @option{-mtarget-align}. These options do not affect the
26877 treatment of auto-aligned instructions like @code{LOOP}, which the
26878 assembler always aligns, either by widening density instructions or
26879 by inserting NOP instructions.
26882 @itemx -mno-longcalls
26883 @opindex mlongcalls
26884 @opindex mno-longcalls
26885 When this option is enabled, GCC instructs the assembler to translate
26886 direct calls to indirect calls unless it can determine that the target
26887 of a direct call is in the range allowed by the call instruction. This
26888 translation typically occurs for calls to functions in other source
26889 files. Specifically, the assembler translates a direct @code{CALL}
26890 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
26891 The default is @option{-mno-longcalls}. This option should be used in
26892 programs where the call target can potentially be out of range. This
26893 option is implemented in the assembler, not the compiler, so the
26894 assembly code generated by GCC still shows direct call
26895 instructions---look at the disassembled object code to see the actual
26896 instructions. Note that the assembler uses an indirect call for
26897 every cross-file call, not just those that really are out of range.
26900 @node zSeries Options
26901 @subsection zSeries Options
26902 @cindex zSeries options
26904 These are listed under @xref{S/390 and zSeries Options}.
26910 @section Specifying Subprocesses and the Switches to Pass to Them
26913 @command{gcc} is a driver program. It performs its job by invoking a
26914 sequence of other programs to do the work of compiling, assembling and
26915 linking. GCC interprets its command-line parameters and uses these to
26916 deduce which programs it should invoke, and which command-line options
26917 it ought to place on their command lines. This behavior is controlled
26918 by @dfn{spec strings}. In most cases there is one spec string for each
26919 program that GCC can invoke, but a few programs have multiple spec
26920 strings to control their behavior. The spec strings built into GCC can
26921 be overridden by using the @option{-specs=} command-line switch to specify
26924 @dfn{Spec files} are plain-text files that are used to construct spec
26925 strings. They consist of a sequence of directives separated by blank
26926 lines. The type of directive is determined by the first non-whitespace
26927 character on the line, which can be one of the following:
26930 @item %@var{command}
26931 Issues a @var{command} to the spec file processor. The commands that can
26935 @item %include <@var{file}>
26936 @cindex @code{%include}
26937 Search for @var{file} and insert its text at the current point in the
26940 @item %include_noerr <@var{file}>
26941 @cindex @code{%include_noerr}
26942 Just like @samp{%include}, but do not generate an error message if the include
26943 file cannot be found.
26945 @item %rename @var{old_name} @var{new_name}
26946 @cindex @code{%rename}
26947 Rename the spec string @var{old_name} to @var{new_name}.
26951 @item *[@var{spec_name}]:
26952 This tells the compiler to create, override or delete the named spec
26953 string. All lines after this directive up to the next directive or
26954 blank line are considered to be the text for the spec string. If this
26955 results in an empty string then the spec is deleted. (Or, if the
26956 spec did not exist, then nothing happens.) Otherwise, if the spec
26957 does not currently exist a new spec is created. If the spec does
26958 exist then its contents are overridden by the text of this
26959 directive, unless the first character of that text is the @samp{+}
26960 character, in which case the text is appended to the spec.
26962 @item [@var{suffix}]:
26963 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
26964 and up to the next directive or blank line are considered to make up the
26965 spec string for the indicated suffix. When the compiler encounters an
26966 input file with the named suffix, it processes the spec string in
26967 order to work out how to compile that file. For example:
26971 z-compile -input %i
26974 This says that any input file whose name ends in @samp{.ZZ} should be
26975 passed to the program @samp{z-compile}, which should be invoked with the
26976 command-line switch @option{-input} and with the result of performing the
26977 @samp{%i} substitution. (See below.)
26979 As an alternative to providing a spec string, the text following a
26980 suffix directive can be one of the following:
26983 @item @@@var{language}
26984 This says that the suffix is an alias for a known @var{language}. This is
26985 similar to using the @option{-x} command-line switch to GCC to specify a
26986 language explicitly. For example:
26993 Says that .ZZ files are, in fact, C++ source files.
26996 This causes an error messages saying:
26999 @var{name} compiler not installed on this system.
27003 GCC already has an extensive list of suffixes built into it.
27004 This directive adds an entry to the end of the list of suffixes, but
27005 since the list is searched from the end backwards, it is effectively
27006 possible to override earlier entries using this technique.
27010 GCC has the following spec strings built into it. Spec files can
27011 override these strings or create their own. Note that individual
27012 targets can also add their own spec strings to this list.
27015 asm Options to pass to the assembler
27016 asm_final Options to pass to the assembler post-processor
27017 cpp Options to pass to the C preprocessor
27018 cc1 Options to pass to the C compiler
27019 cc1plus Options to pass to the C++ compiler
27020 endfile Object files to include at the end of the link
27021 link Options to pass to the linker
27022 lib Libraries to include on the command line to the linker
27023 libgcc Decides which GCC support library to pass to the linker
27024 linker Sets the name of the linker
27025 predefines Defines to be passed to the C preprocessor
27026 signed_char Defines to pass to CPP to say whether @code{char} is signed
27028 startfile Object files to include at the start of the link
27031 Here is a small example of a spec file:
27034 %rename lib old_lib
27037 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
27040 This example renames the spec called @samp{lib} to @samp{old_lib} and
27041 then overrides the previous definition of @samp{lib} with a new one.
27042 The new definition adds in some extra command-line options before
27043 including the text of the old definition.
27045 @dfn{Spec strings} are a list of command-line options to be passed to their
27046 corresponding program. In addition, the spec strings can contain
27047 @samp{%}-prefixed sequences to substitute variable text or to
27048 conditionally insert text into the command line. Using these constructs
27049 it is possible to generate quite complex command lines.
27051 Here is a table of all defined @samp{%}-sequences for spec
27052 strings. Note that spaces are not generated automatically around the
27053 results of expanding these sequences. Therefore you can concatenate them
27054 together or combine them with constant text in a single argument.
27058 Substitute one @samp{%} into the program name or argument.
27061 Substitute the name of the input file being processed.
27064 Substitute the basename of the input file being processed.
27065 This is the substring up to (and not including) the last period
27066 and not including the directory.
27069 This is the same as @samp{%b}, but include the file suffix (text after
27073 Marks the argument containing or following the @samp{%d} as a
27074 temporary file name, so that that file is deleted if GCC exits
27075 successfully. Unlike @samp{%g}, this contributes no text to the
27078 @item %g@var{suffix}
27079 Substitute a file name that has suffix @var{suffix} and is chosen
27080 once per compilation, and mark the argument in the same way as
27081 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
27082 name is now chosen in a way that is hard to predict even when previously
27083 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
27084 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
27085 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
27086 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
27087 was simply substituted with a file name chosen once per compilation,
27088 without regard to any appended suffix (which was therefore treated
27089 just like ordinary text), making such attacks more likely to succeed.
27091 @item %u@var{suffix}
27092 Like @samp{%g}, but generates a new temporary file name
27093 each time it appears instead of once per compilation.
27095 @item %U@var{suffix}
27096 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
27097 new one if there is no such last file name. In the absence of any
27098 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
27099 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
27100 involves the generation of two distinct file names, one
27101 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
27102 simply substituted with a file name chosen for the previous @samp{%u},
27103 without regard to any appended suffix.
27105 @item %j@var{suffix}
27106 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
27107 writable, and if @option{-save-temps} is not used;
27108 otherwise, substitute the name
27109 of a temporary file, just like @samp{%u}. This temporary file is not
27110 meant for communication between processes, but rather as a junk
27111 disposal mechanism.
27113 @item %|@var{suffix}
27114 @itemx %m@var{suffix}
27115 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
27116 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
27117 all. These are the two most common ways to instruct a program that it
27118 should read from standard input or write to standard output. If you
27119 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
27120 construct: see for example @file{f/lang-specs.h}.
27122 @item %.@var{SUFFIX}
27123 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
27124 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
27125 terminated by the next space or %.
27128 Marks the argument containing or following the @samp{%w} as the
27129 designated output file of this compilation. This puts the argument
27130 into the sequence of arguments that @samp{%o} substitutes.
27133 Substitutes the names of all the output files, with spaces
27134 automatically placed around them. You should write spaces
27135 around the @samp{%o} as well or the results are undefined.
27136 @samp{%o} is for use in the specs for running the linker.
27137 Input files whose names have no recognized suffix are not compiled
27138 at all, but they are included among the output files, so they are
27142 Substitutes the suffix for object files. Note that this is
27143 handled specially when it immediately follows @samp{%g, %u, or %U},
27144 because of the need for those to form complete file names. The
27145 handling is such that @samp{%O} is treated exactly as if it had already
27146 been substituted, except that @samp{%g, %u, and %U} do not currently
27147 support additional @var{suffix} characters following @samp{%O} as they do
27148 following, for example, @samp{.o}.
27151 Substitutes the standard macro predefinitions for the
27152 current target machine. Use this when running @command{cpp}.
27155 Like @samp{%p}, but puts @samp{__} before and after the name of each
27156 predefined macro, except for macros that start with @samp{__} or with
27157 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
27161 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
27162 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
27163 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
27164 and @option{-imultilib} as necessary.
27167 Current argument is the name of a library or startup file of some sort.
27168 Search for that file in a standard list of directories and substitute
27169 the full name found. The current working directory is included in the
27170 list of directories scanned.
27173 Current argument is the name of a linker script. Search for that file
27174 in the current list of directories to scan for libraries. If the file
27175 is located insert a @option{--script} option into the command line
27176 followed by the full path name found. If the file is not found then
27177 generate an error message. Note: the current working directory is not
27181 Print @var{str} as an error message. @var{str} is terminated by a newline.
27182 Use this when inconsistent options are detected.
27184 @item %(@var{name})
27185 Substitute the contents of spec string @var{name} at this point.
27187 @item %x@{@var{option}@}
27188 Accumulate an option for @samp{%X}.
27191 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
27195 Output the accumulated assembler options specified by @option{-Wa}.
27198 Output the accumulated preprocessor options specified by @option{-Wp}.
27201 Process the @code{asm} spec. This is used to compute the
27202 switches to be passed to the assembler.
27205 Process the @code{asm_final} spec. This is a spec string for
27206 passing switches to an assembler post-processor, if such a program is
27210 Process the @code{link} spec. This is the spec for computing the
27211 command line passed to the linker. Typically it makes use of the
27212 @samp{%L %G %S %D and %E} sequences.
27215 Dump out a @option{-L} option for each directory that GCC believes might
27216 contain startup files. If the target supports multilibs then the
27217 current multilib directory is prepended to each of these paths.
27220 Process the @code{lib} spec. This is a spec string for deciding which
27221 libraries are included on the command line to the linker.
27224 Process the @code{libgcc} spec. This is a spec string for deciding
27225 which GCC support library is included on the command line to the linker.
27228 Process the @code{startfile} spec. This is a spec for deciding which
27229 object files are the first ones passed to the linker. Typically
27230 this might be a file named @file{crt0.o}.
27233 Process the @code{endfile} spec. This is a spec string that specifies
27234 the last object files that are passed to the linker.
27237 Process the @code{cpp} spec. This is used to construct the arguments
27238 to be passed to the C preprocessor.
27241 Process the @code{cc1} spec. This is used to construct the options to be
27242 passed to the actual C compiler (@command{cc1}).
27245 Process the @code{cc1plus} spec. This is used to construct the options to be
27246 passed to the actual C++ compiler (@command{cc1plus}).
27249 Substitute the variable part of a matched option. See below.
27250 Note that each comma in the substituted string is replaced by
27254 Remove all occurrences of @code{-S} from the command line. Note---this
27255 command is position dependent. @samp{%} commands in the spec string
27256 before this one see @code{-S}, @samp{%} commands in the spec string
27257 after this one do not.
27259 @item %:@var{function}(@var{args})
27260 Call the named function @var{function}, passing it @var{args}.
27261 @var{args} is first processed as a nested spec string, then split
27262 into an argument vector in the usual fashion. The function returns
27263 a string which is processed as if it had appeared literally as part
27264 of the current spec.
27266 The following built-in spec functions are provided:
27269 @item @code{getenv}
27270 The @code{getenv} spec function takes two arguments: an environment
27271 variable name and a string. If the environment variable is not
27272 defined, a fatal error is issued. Otherwise, the return value is the
27273 value of the environment variable concatenated with the string. For
27274 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
27277 %:getenv(TOPDIR /include)
27280 expands to @file{/path/to/top/include}.
27282 @item @code{if-exists}
27283 The @code{if-exists} spec function takes one argument, an absolute
27284 pathname to a file. If the file exists, @code{if-exists} returns the
27285 pathname. Here is a small example of its usage:
27289 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
27292 @item @code{if-exists-else}
27293 The @code{if-exists-else} spec function is similar to the @code{if-exists}
27294 spec function, except that it takes two arguments. The first argument is
27295 an absolute pathname to a file. If the file exists, @code{if-exists-else}
27296 returns the pathname. If it does not exist, it returns the second argument.
27297 This way, @code{if-exists-else} can be used to select one file or another,
27298 based on the existence of the first. Here is a small example of its usage:
27302 crt0%O%s %:if-exists(crti%O%s) \
27303 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
27306 @item @code{replace-outfile}
27307 The @code{replace-outfile} spec function takes two arguments. It looks for the
27308 first argument in the outfiles array and replaces it with the second argument. Here
27309 is a small example of its usage:
27312 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
27315 @item @code{remove-outfile}
27316 The @code{remove-outfile} spec function takes one argument. It looks for the
27317 first argument in the outfiles array and removes it. Here is a small example
27321 %:remove-outfile(-lm)
27324 @item @code{pass-through-libs}
27325 The @code{pass-through-libs} spec function takes any number of arguments. It
27326 finds any @option{-l} options and any non-options ending in @file{.a} (which it
27327 assumes are the names of linker input library archive files) and returns a
27328 result containing all the found arguments each prepended by
27329 @option{-plugin-opt=-pass-through=} and joined by spaces. This list is
27330 intended to be passed to the LTO linker plugin.
27333 %:pass-through-libs(%G %L %G)
27336 @item @code{print-asm-header}
27337 The @code{print-asm-header} function takes no arguments and simply
27338 prints a banner like:
27344 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
27347 It is used to separate compiler options from assembler options
27348 in the @option{--target-help} output.
27352 Substitutes the @code{-S} switch, if that switch is given to GCC@.
27353 If that switch is not specified, this substitutes nothing. Note that
27354 the leading dash is omitted when specifying this option, and it is
27355 automatically inserted if the substitution is performed. Thus the spec
27356 string @samp{%@{foo@}} matches the command-line option @option{-foo}
27357 and outputs the command-line option @option{-foo}.
27360 Like %@{@code{S}@} but mark last argument supplied within as a file to be
27361 deleted on failure.
27364 Substitutes all the switches specified to GCC whose names start
27365 with @code{-S}, but which also take an argument. This is used for
27366 switches like @option{-o}, @option{-D}, @option{-I}, etc.
27367 GCC considers @option{-o foo} as being
27368 one switch whose name starts with @samp{o}. %@{o*@} substitutes this
27369 text, including the space. Thus two arguments are generated.
27372 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
27373 (the order of @code{S} and @code{T} in the spec is not significant).
27374 There can be any number of ampersand-separated variables; for each the
27375 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
27378 Substitutes @code{X}, if the @option{-S} switch is given to GCC@.
27381 Substitutes @code{X}, if the @option{-S} switch is @emph{not} given to GCC@.
27384 Substitutes @code{X} if one or more switches whose names start with
27385 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
27386 once, no matter how many such switches appeared. However, if @code{%*}
27387 appears somewhere in @code{X}, then @code{X} is substituted once
27388 for each matching switch, with the @code{%*} replaced by the part of
27389 that switch matching the @code{*}.
27391 If @code{%*} appears as the last part of a spec sequence then a space
27392 is added after the end of the last substitution. If there is more
27393 text in the sequence, however, then a space is not generated. This
27394 allows the @code{%*} substitution to be used as part of a larger
27395 string. For example, a spec string like this:
27398 %@{mcu=*:--script=%*/memory.ld@}
27402 when matching an option like @option{-mcu=newchip} produces:
27405 --script=newchip/memory.ld
27409 Substitutes @code{X}, if processing a file with suffix @code{S}.
27412 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
27415 Substitutes @code{X}, if processing a file for language @code{S}.
27418 Substitutes @code{X}, if not processing a file for language @code{S}.
27421 Substitutes @code{X} if either @code{-S} or @code{-P} is given to
27422 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
27423 @code{*} sequences as well, although they have a stronger binding than
27424 the @samp{|}. If @code{%*} appears in @code{X}, all of the
27425 alternatives must be starred, and only the first matching alternative
27428 For example, a spec string like this:
27431 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
27435 outputs the following command-line options from the following input
27436 command-line options:
27441 -d fred.c -foo -baz -boggle
27442 -d jim.d -bar -baz -boggle
27445 @item %@{S:X; T:Y; :D@}
27447 If @code{S} is given to GCC, substitutes @code{X}; else if @code{T} is
27448 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
27449 be as many clauses as you need. This may be combined with @code{.},
27450 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
27455 The switch matching text @code{S} in a @samp{%@{S@}}, @samp{%@{S:X@}}
27456 or similar construct can use a backslash to ignore the special meaning
27457 of the character following it, thus allowing literal matching of a
27458 character that is otherwise specially treated. For example,
27459 @samp{%@{std=iso9899\:1999:X@}} substitutes @code{X} if the
27460 @option{-std=iso9899:1999} option is given.
27462 The conditional text @code{X} in a @samp{%@{S:X@}} or similar
27463 construct may contain other nested @samp{%} constructs or spaces, or
27464 even newlines. They are processed as usual, as described above.
27465 Trailing white space in @code{X} is ignored. White space may also
27466 appear anywhere on the left side of the colon in these constructs,
27467 except between @code{.} or @code{*} and the corresponding word.
27469 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
27470 handled specifically in these constructs. If another value of
27471 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
27472 @option{-W} switch is found later in the command line, the earlier
27473 switch value is ignored, except with @{@code{S}*@} where @code{S} is
27474 just one letter, which passes all matching options.
27476 The character @samp{|} at the beginning of the predicate text is used to
27477 indicate that a command should be piped to the following command, but
27478 only if @option{-pipe} is specified.
27480 It is built into GCC which switches take arguments and which do not.
27481 (You might think it would be useful to generalize this to allow each
27482 compiler's spec to say which switches take arguments. But this cannot
27483 be done in a consistent fashion. GCC cannot even decide which input
27484 files have been specified without knowing which switches take arguments,
27485 and it must know which input files to compile in order to tell which
27488 GCC also knows implicitly that arguments starting in @option{-l} are to be
27489 treated as compiler output files, and passed to the linker in their
27490 proper position among the other output files.
27492 @node Environment Variables
27493 @section Environment Variables Affecting GCC
27494 @cindex environment variables
27496 @c man begin ENVIRONMENT
27497 This section describes several environment variables that affect how GCC
27498 operates. Some of them work by specifying directories or prefixes to use
27499 when searching for various kinds of files. Some are used to specify other
27500 aspects of the compilation environment.
27502 Note that you can also specify places to search using options such as
27503 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
27504 take precedence over places specified using environment variables, which
27505 in turn take precedence over those specified by the configuration of GCC@.
27506 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
27507 GNU Compiler Collection (GCC) Internals}.
27512 @c @itemx LC_COLLATE
27514 @c @itemx LC_MONETARY
27515 @c @itemx LC_NUMERIC
27520 @c @findex LC_COLLATE
27521 @findex LC_MESSAGES
27522 @c @findex LC_MONETARY
27523 @c @findex LC_NUMERIC
27527 These environment variables control the way that GCC uses
27528 localization information which allows GCC to work with different
27529 national conventions. GCC inspects the locale categories
27530 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
27531 so. These locale categories can be set to any value supported by your
27532 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
27533 Kingdom encoded in UTF-8.
27535 The @env{LC_CTYPE} environment variable specifies character
27536 classification. GCC uses it to determine the character boundaries in
27537 a string; this is needed for some multibyte encodings that contain quote
27538 and escape characters that are otherwise interpreted as a string
27541 The @env{LC_MESSAGES} environment variable specifies the language to
27542 use in diagnostic messages.
27544 If the @env{LC_ALL} environment variable is set, it overrides the value
27545 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
27546 and @env{LC_MESSAGES} default to the value of the @env{LANG}
27547 environment variable. If none of these variables are set, GCC
27548 defaults to traditional C English behavior.
27552 If @env{TMPDIR} is set, it specifies the directory to use for temporary
27553 files. GCC uses temporary files to hold the output of one stage of
27554 compilation which is to be used as input to the next stage: for example,
27555 the output of the preprocessor, which is the input to the compiler
27558 @item GCC_COMPARE_DEBUG
27559 @findex GCC_COMPARE_DEBUG
27560 Setting @env{GCC_COMPARE_DEBUG} is nearly equivalent to passing
27561 @option{-fcompare-debug} to the compiler driver. See the documentation
27562 of this option for more details.
27564 @item GCC_EXEC_PREFIX
27565 @findex GCC_EXEC_PREFIX
27566 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
27567 names of the subprograms executed by the compiler. No slash is added
27568 when this prefix is combined with the name of a subprogram, but you can
27569 specify a prefix that ends with a slash if you wish.
27571 If @env{GCC_EXEC_PREFIX} is not set, GCC attempts to figure out
27572 an appropriate prefix to use based on the pathname it is invoked with.
27574 If GCC cannot find the subprogram using the specified prefix, it
27575 tries looking in the usual places for the subprogram.
27577 The default value of @env{GCC_EXEC_PREFIX} is
27578 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
27579 the installed compiler. In many cases @var{prefix} is the value
27580 of @code{prefix} when you ran the @file{configure} script.
27582 Other prefixes specified with @option{-B} take precedence over this prefix.
27584 This prefix is also used for finding files such as @file{crt0.o} that are
27587 In addition, the prefix is used in an unusual way in finding the
27588 directories to search for header files. For each of the standard
27589 directories whose name normally begins with @samp{/usr/local/lib/gcc}
27590 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
27591 replacing that beginning with the specified prefix to produce an
27592 alternate directory name. Thus, with @option{-Bfoo/}, GCC searches
27593 @file{foo/bar} just before it searches the standard directory
27594 @file{/usr/local/lib/bar}.
27595 If a standard directory begins with the configured
27596 @var{prefix} then the value of @var{prefix} is replaced by
27597 @env{GCC_EXEC_PREFIX} when looking for header files.
27599 @item COMPILER_PATH
27600 @findex COMPILER_PATH
27601 The value of @env{COMPILER_PATH} is a colon-separated list of
27602 directories, much like @env{PATH}. GCC tries the directories thus
27603 specified when searching for subprograms, if it cannot find the
27604 subprograms using @env{GCC_EXEC_PREFIX}.
27607 @findex LIBRARY_PATH
27608 The value of @env{LIBRARY_PATH} is a colon-separated list of
27609 directories, much like @env{PATH}. When configured as a native compiler,
27610 GCC tries the directories thus specified when searching for special
27611 linker files, if it cannot find them using @env{GCC_EXEC_PREFIX}. Linking
27612 using GCC also uses these directories when searching for ordinary
27613 libraries for the @option{-l} option (but directories specified with
27614 @option{-L} come first).
27618 @cindex locale definition
27619 This variable is used to pass locale information to the compiler. One way in
27620 which this information is used is to determine the character set to be used
27621 when character literals, string literals and comments are parsed in C and C++.
27622 When the compiler is configured to allow multibyte characters,
27623 the following values for @env{LANG} are recognized:
27627 Recognize JIS characters.
27629 Recognize SJIS characters.
27631 Recognize EUCJP characters.
27634 If @env{LANG} is not defined, or if it has some other value, then the
27635 compiler uses @code{mblen} and @code{mbtowc} as defined by the default locale to
27636 recognize and translate multibyte characters.
27640 Some additional environment variables affect the behavior of the
27643 @include cppenv.texi
27647 @node Precompiled Headers
27648 @section Using Precompiled Headers
27649 @cindex precompiled headers
27650 @cindex speed of compilation
27652 Often large projects have many header files that are included in every
27653 source file. The time the compiler takes to process these header files
27654 over and over again can account for nearly all of the time required to
27655 build the project. To make builds faster, GCC allows you to
27656 @dfn{precompile} a header file.
27658 To create a precompiled header file, simply compile it as you would any
27659 other file, if necessary using the @option{-x} option to make the driver
27660 treat it as a C or C++ header file. You may want to use a
27661 tool like @command{make} to keep the precompiled header up-to-date when
27662 the headers it contains change.
27664 A precompiled header file is searched for when @code{#include} is
27665 seen in the compilation. As it searches for the included file
27666 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
27667 compiler looks for a precompiled header in each directory just before it
27668 looks for the include file in that directory. The name searched for is
27669 the name specified in the @code{#include} with @samp{.gch} appended. If
27670 the precompiled header file cannot be used, it is ignored.
27672 For instance, if you have @code{#include "all.h"}, and you have
27673 @file{all.h.gch} in the same directory as @file{all.h}, then the
27674 precompiled header file is used if possible, and the original
27675 header is used otherwise.
27677 Alternatively, you might decide to put the precompiled header file in a
27678 directory and use @option{-I} to ensure that directory is searched
27679 before (or instead of) the directory containing the original header.
27680 Then, if you want to check that the precompiled header file is always
27681 used, you can put a file of the same name as the original header in this
27682 directory containing an @code{#error} command.
27684 This also works with @option{-include}. So yet another way to use
27685 precompiled headers, good for projects not designed with precompiled
27686 header files in mind, is to simply take most of the header files used by
27687 a project, include them from another header file, precompile that header
27688 file, and @option{-include} the precompiled header. If the header files
27689 have guards against multiple inclusion, they are skipped because
27690 they've already been included (in the precompiled header).
27692 If you need to precompile the same header file for different
27693 languages, targets, or compiler options, you can instead make a
27694 @emph{directory} named like @file{all.h.gch}, and put each precompiled
27695 header in the directory, perhaps using @option{-o}. It doesn't matter
27696 what you call the files in the directory; every precompiled header in
27697 the directory is considered. The first precompiled header
27698 encountered in the directory that is valid for this compilation is
27699 used; they're searched in no particular order.
27701 There are many other possibilities, limited only by your imagination,
27702 good sense, and the constraints of your build system.
27704 A precompiled header file can be used only when these conditions apply:
27708 Only one precompiled header can be used in a particular compilation.
27711 A precompiled header cannot be used once the first C token is seen. You
27712 can have preprocessor directives before a precompiled header; you cannot
27713 include a precompiled header from inside another header.
27716 The precompiled header file must be produced for the same language as
27717 the current compilation. You cannot use a C precompiled header for a C++
27721 The precompiled header file must have been produced by the same compiler
27722 binary as the current compilation is using.
27725 Any macros defined before the precompiled header is included must
27726 either be defined in the same way as when the precompiled header was
27727 generated, or must not affect the precompiled header, which usually
27728 means that they don't appear in the precompiled header at all.
27730 The @option{-D} option is one way to define a macro before a
27731 precompiled header is included; using a @code{#define} can also do it.
27732 There are also some options that define macros implicitly, like
27733 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
27736 @item If debugging information is output when using the precompiled
27737 header, using @option{-g} or similar, the same kind of debugging information
27738 must have been output when building the precompiled header. However,
27739 a precompiled header built using @option{-g} can be used in a compilation
27740 when no debugging information is being output.
27742 @item The same @option{-m} options must generally be used when building
27743 and using the precompiled header. @xref{Submodel Options},
27744 for any cases where this rule is relaxed.
27746 @item Each of the following options must be the same when building and using
27747 the precompiled header:
27749 @gccoptlist{-fexceptions}
27752 Some other command-line options starting with @option{-f},
27753 @option{-p}, or @option{-O} must be defined in the same way as when
27754 the precompiled header was generated. At present, it's not clear
27755 which options are safe to change and which are not; the safest choice
27756 is to use exactly the same options when generating and using the
27757 precompiled header. The following are known to be safe:
27759 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
27760 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
27761 -fsched-verbose=@var{number} -fschedule-insns -fvisibility= @gol
27766 For all of these except the last, the compiler automatically
27767 ignores the precompiled header if the conditions aren't met. If you
27768 find an option combination that doesn't work and doesn't cause the
27769 precompiled header to be ignored, please consider filing a bug report,
27772 If you do use differing options when generating and using the
27773 precompiled header, the actual behavior is a mixture of the
27774 behavior for the options. For instance, if you use @option{-g} to
27775 generate the precompiled header but not when using it, you may or may
27776 not get debugging information for routines in the precompiled header.