1 @c Copyright (C) 1988-2019 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{} 1988-2019 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.
124 Some options take one or more arguments typically separated either
125 by a space or by the equals sign (@samp{=}) from the option name.
126 Unless documented otherwise, an argument can be either numeric or
127 a string. Numeric arguments must typically be small unsigned decimal
128 or hexadecimal integers. Hexadecimal arguments must begin with
129 the @samp{0x} prefix. Arguments to options that specify a size
130 threshold of some sort may be arbitrarily large decimal or hexadecimal
131 integers followed by a byte size suffix designating a multiple of bytes
132 such as @code{kB} and @code{KiB} for kilobyte and kibibyte, respectively,
133 @code{MB} and @code{MiB} for megabyte and mebibyte, @code{GB} and
134 @code{GiB} for gigabyte and gigibyte, and so on. Such arguments are
135 designated by @var{byte-size} in the following text. Refer to the NIST,
136 IEC, and other relevant national and international standards for the full
137 listing and explanation of the binary and decimal byte size prefixes.
141 @xref{Option Index}, for an index to GCC's options.
144 * Option Summary:: Brief list of all options, without explanations.
145 * Overall Options:: Controlling the kind of output:
146 an executable, object files, assembler files,
147 or preprocessed source.
148 * Invoking G++:: Compiling C++ programs.
149 * C Dialect Options:: Controlling the variant of C language compiled.
150 * C++ Dialect Options:: Variations on C++.
151 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
153 * Diagnostic Message Formatting Options:: Controlling how diagnostics should
155 * Warning Options:: How picky should the compiler be?
156 * Debugging Options:: Producing debuggable code.
157 * Optimize Options:: How much optimization?
158 * Instrumentation Options:: Enabling profiling and extra run-time error checking.
159 * Preprocessor Options:: Controlling header files and macro definitions.
160 Also, getting dependency information for Make.
161 * Assembler Options:: Passing options to the assembler.
162 * Link Options:: Specifying libraries and so on.
163 * Directory Options:: Where to find header files and libraries.
164 Where to find the compiler executable files.
165 * Code Gen Options:: Specifying conventions for function calls, data layout
167 * Developer Options:: Printing GCC configuration info, statistics, and
169 * Submodel Options:: Target-specific options, such as compiling for a
170 specific processor variant.
171 * Spec Files:: How to pass switches to sub-processes.
172 * Environment Variables:: Env vars that affect GCC.
173 * Precompiled Headers:: Compiling a header once, and using it many times.
179 @section Option Summary
181 Here is a summary of all the options, grouped by type. Explanations are
182 in the following sections.
185 @item Overall Options
186 @xref{Overall Options,,Options Controlling the Kind of Output}.
187 @gccoptlist{-c -S -E -o @var{file} -x @var{language} @gol
188 -v -### --help@r{[}=@var{class}@r{[},@dots{}@r{]]} --target-help --version @gol
189 -pass-exit-codes -pipe -specs=@var{file} -wrapper @gol
190 @@@var{file} -ffile-prefix-map=@var{old}=@var{new} @gol
191 -fplugin=@var{file} -fplugin-arg-@var{name}=@var{arg} @gol
192 -fdump-ada-spec@r{[}-slim@r{]} -fada-spec-parent=@var{unit} -fdump-go-spec=@var{file}}
194 @item C Language Options
195 @xref{C Dialect Options,,Options Controlling C Dialect}.
196 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
197 -fpermitted-flt-eval-methods=@var{standard} @gol
198 -aux-info @var{filename} -fallow-parameterless-variadic-functions @gol
199 -fno-asm -fno-builtin -fno-builtin-@var{function} -fgimple@gol
200 -fhosted -ffreestanding @gol
201 -fopenacc -fopenacc-dim=@var{geom} @gol
202 -fopenmp -fopenmp-simd @gol
203 -fms-extensions -fplan9-extensions -fsso-struct=@var{endianness} @gol
204 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
205 -fsigned-bitfields -fsigned-char @gol
206 -funsigned-bitfields -funsigned-char}
208 @item C++ Language Options
209 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
210 @gccoptlist{-fabi-version=@var{n} -fno-access-control @gol
211 -faligned-new=@var{n} -fargs-in-order=@var{n} -fchar8_t -fcheck-new @gol
212 -fconstexpr-depth=@var{n} -fconstexpr-cache-depth=@var{n} @gol
213 -fconstexpr-loop-limit=@var{n} -fconstexpr-ops-limit=@var{n} @gol
214 -fno-elide-constructors @gol
215 -fno-enforce-eh-specs @gol
216 -fno-gnu-keywords @gol
217 -fno-implicit-templates @gol
218 -fno-implicit-inline-templates @gol
219 -fno-implement-inlines -fms-extensions @gol
220 -fnew-inheriting-ctors @gol
221 -fnew-ttp-matching @gol
222 -fno-nonansi-builtins -fnothrow-opt -fno-operator-names @gol
223 -fno-optional-diags -fpermissive @gol
224 -fno-pretty-templates @gol
225 -frepo -fno-rtti -fsized-deallocation @gol
226 -ftemplate-backtrace-limit=@var{n} @gol
227 -ftemplate-depth=@var{n} @gol
228 -fno-threadsafe-statics -fuse-cxa-atexit @gol
229 -fno-weak -nostdinc++ @gol
230 -fvisibility-inlines-hidden @gol
231 -fvisibility-ms-compat @gol
232 -fext-numeric-literals @gol
233 -Wabi=@var{n} -Wabi-tag -Wconversion-null -Wctor-dtor-privacy @gol
234 -Wdelete-non-virtual-dtor -Wdeprecated-copy -Wdeprecated-copy-dtor @gol
235 -Wliteral-suffix @gol
236 -Wmultiple-inheritance -Wno-init-list-lifetime @gol
237 -Wnamespaces -Wnarrowing @gol
238 -Wpessimizing-move -Wredundant-move @gol
239 -Wnoexcept -Wnoexcept-type -Wclass-memaccess @gol
240 -Wnon-virtual-dtor -Wreorder -Wregister @gol
241 -Weffc++ -Wstrict-null-sentinel -Wtemplates @gol
242 -Wno-non-template-friend -Wold-style-cast @gol
243 -Woverloaded-virtual -Wno-pmf-conversions @gol
244 -Wno-class-conversion -Wno-terminate @gol
245 -Wsign-promo -Wvirtual-inheritance}
247 @item Objective-C and Objective-C++ Language Options
248 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
249 Objective-C and Objective-C++ Dialects}.
250 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
251 -fgnu-runtime -fnext-runtime @gol
252 -fno-nil-receivers @gol
253 -fobjc-abi-version=@var{n} @gol
254 -fobjc-call-cxx-cdtors @gol
255 -fobjc-direct-dispatch @gol
256 -fobjc-exceptions @gol
259 -fobjc-std=objc1 @gol
260 -fno-local-ivars @gol
261 -fivar-visibility=@r{[}public@r{|}protected@r{|}private@r{|}package@r{]} @gol
262 -freplace-objc-classes @gol
265 -Wassign-intercept @gol
266 -Wno-protocol -Wselector @gol
267 -Wstrict-selector-match @gol
268 -Wundeclared-selector}
270 @item Diagnostic Message Formatting Options
271 @xref{Diagnostic Message Formatting Options,,Options to Control Diagnostic Messages Formatting}.
272 @gccoptlist{-fmessage-length=@var{n} @gol
273 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
274 -fdiagnostics-color=@r{[}auto@r{|}never@r{|}always@r{]} @gol
275 -fdiagnostics-format=@r{[}text@r{|}json@r{]} @gol
276 -fno-diagnostics-show-option -fno-diagnostics-show-caret @gol
277 -fno-diagnostics-show-labels -fno-diagnostics-show-line-numbers @gol
278 -fdiagnostics-minimum-margin-width=@var{width} @gol
279 -fdiagnostics-parseable-fixits -fdiagnostics-generate-patch @gol
280 -fdiagnostics-show-template-tree -fno-elide-type @gol
283 @item Warning Options
284 @xref{Warning Options,,Options to Request or Suppress Warnings}.
285 @gccoptlist{-fsyntax-only -fmax-errors=@var{n} -Wpedantic @gol
286 -pedantic-errors @gol
287 -w -Wextra -Wall -Waddress -Waddress-of-packed-member @gol
288 -Waggregate-return -Waligned-new @gol
289 -Walloc-zero -Walloc-size-larger-than=@var{byte-size} @gol
290 -Walloca -Walloca-larger-than=@var{byte-size} @gol
291 -Wno-aggressive-loop-optimizations -Warray-bounds -Warray-bounds=@var{n} @gol
292 -Wno-attributes -Wattribute-alias=@var{n} @gol
293 -Wbool-compare -Wbool-operation @gol
294 -Wno-builtin-declaration-mismatch @gol
295 -Wno-builtin-macro-redefined -Wc90-c99-compat -Wc99-c11-compat @gol
296 -Wc++-compat -Wc++11-compat -Wc++14-compat -Wc++17-compat @gol
297 -Wcast-align -Wcast-align=strict -Wcast-function-type -Wcast-qual @gol
298 -Wchar-subscripts -Wcatch-value -Wcatch-value=@var{n} @gol
299 -Wclobbered -Wcomment -Wconditionally-supported @gol
300 -Wconversion -Wcoverage-mismatch -Wno-cpp -Wdangling-else -Wdate-time @gol
301 -Wdelete-incomplete @gol
302 -Wno-attribute-warning @gol
303 -Wno-deprecated -Wno-deprecated-declarations -Wno-designated-init @gol
304 -Wdisabled-optimization @gol
305 -Wno-discarded-qualifiers -Wno-discarded-array-qualifiers @gol
306 -Wno-div-by-zero -Wdouble-promotion @gol
307 -Wduplicated-branches -Wduplicated-cond @gol
308 -Wempty-body -Wenum-compare -Wno-endif-labels -Wexpansion-to-defined @gol
309 -Werror -Werror=* -Wextra-semi -Wfatal-errors @gol
310 -Wfloat-equal -Wformat -Wformat=2 @gol
311 -Wno-format-contains-nul -Wno-format-extra-args @gol
312 -Wformat-nonliteral -Wformat-overflow=@var{n} @gol
313 -Wformat-security -Wformat-signedness -Wformat-truncation=@var{n} @gol
314 -Wformat-y2k -Wframe-address @gol
315 -Wframe-larger-than=@var{byte-size} -Wno-free-nonheap-object @gol
316 -Wjump-misses-init @gol
317 -Whsa -Wif-not-aligned @gol
318 -Wignored-qualifiers -Wignored-attributes -Wincompatible-pointer-types @gol
319 -Wimplicit -Wimplicit-fallthrough -Wimplicit-fallthrough=@var{n} @gol
320 -Wimplicit-function-declaration -Wimplicit-int @gol
321 -Winaccessible-base @gol
322 -Winit-self -Winline -Wno-int-conversion -Wint-in-bool-context @gol
323 -Wno-int-to-pointer-cast -Winvalid-memory-model -Wno-invalid-offsetof @gol
324 -Winvalid-pch -Wlarger-than=@var{byte-size} @gol
325 -Wlogical-op -Wlogical-not-parentheses -Wlong-long @gol
326 -Wmain -Wmaybe-uninitialized -Wmemset-elt-size -Wmemset-transposed-args @gol
327 -Wmisleading-indentation -Wmissing-attributes -Wmissing-braces @gol
328 -Wmissing-field-initializers -Wmissing-format-attribute @gol
329 -Wmissing-include-dirs -Wmissing-noreturn -Wmissing-profile @gol
330 -Wno-multichar -Wmultistatement-macros -Wnonnull -Wnonnull-compare @gol
331 -Wnormalized=@r{[}none@r{|}id@r{|}nfc@r{|}nfkc@r{]} @gol
332 -Wnull-dereference -Wodr -Wno-overflow -Wopenmp-simd @gol
333 -Woverride-init-side-effects -Woverlength-strings @gol
334 -Wpacked -Wpacked-bitfield-compat -Wpacked-not-aligned -Wpadded @gol
335 -Wparentheses -Wno-pedantic-ms-format @gol
336 -Wplacement-new -Wplacement-new=@var{n} @gol
337 -Wpointer-arith -Wpointer-compare -Wno-pointer-to-int-cast @gol
338 -Wno-pragmas -Wno-prio-ctor-dtor -Wredundant-decls @gol
339 -Wrestrict -Wno-return-local-addr @gol
340 -Wreturn-type -Wsequence-point -Wshadow -Wno-shadow-ivar @gol
341 -Wshadow=global, -Wshadow=local, -Wshadow=compatible-local @gol
342 -Wshift-overflow -Wshift-overflow=@var{n} @gol
343 -Wshift-count-negative -Wshift-count-overflow -Wshift-negative-value @gol
344 -Wsign-compare -Wsign-conversion -Wfloat-conversion @gol
345 -Wno-scalar-storage-order -Wsizeof-pointer-div @gol
346 -Wsizeof-pointer-memaccess -Wsizeof-array-argument @gol
347 -Wstack-protector -Wstack-usage=@var{byte-size} -Wstrict-aliasing @gol
348 -Wstrict-aliasing=n -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
349 -Wstringop-overflow=@var{n} -Wstringop-truncation -Wsubobject-linkage @gol
350 -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{|}malloc@r{]} @gol
351 -Wsuggest-final-types @gol -Wsuggest-final-methods -Wsuggest-override @gol
352 -Wswitch -Wswitch-bool -Wswitch-default -Wswitch-enum @gol
353 -Wswitch-unreachable -Wsync-nand @gol
354 -Wsystem-headers -Wtautological-compare -Wtrampolines -Wtrigraphs @gol
355 -Wtype-limits -Wundef @gol
356 -Wuninitialized -Wunknown-pragmas @gol
357 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
358 -Wunused-label -Wunused-local-typedefs -Wunused-macros @gol
359 -Wunused-parameter -Wno-unused-result @gol
360 -Wunused-value -Wunused-variable @gol
361 -Wunused-const-variable -Wunused-const-variable=@var{n} @gol
362 -Wunused-but-set-parameter -Wunused-but-set-variable @gol
363 -Wuseless-cast -Wvariadic-macros -Wvector-operation-performance @gol
364 -Wvla -Wvla-larger-than=@var{byte-size} -Wvolatile-register-var @gol
366 -Wzero-as-null-pointer-constant}
368 @item C and Objective-C-only Warning Options
369 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
370 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
371 -Wold-style-declaration -Wold-style-definition @gol
372 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
373 -Wdeclaration-after-statement -Wpointer-sign}
375 @item Debugging Options
376 @xref{Debugging Options,,Options for Debugging Your Program}.
377 @gccoptlist{-g -g@var{level} -gdwarf -gdwarf-@var{version} @gol
378 -ggdb -grecord-gcc-switches -gno-record-gcc-switches @gol
379 -gstabs -gstabs+ -gstrict-dwarf -gno-strict-dwarf @gol
380 -gas-loc-support -gno-as-loc-support @gol
381 -gas-locview-support -gno-as-locview-support @gol
382 -gcolumn-info -gno-column-info @gol
383 -gstatement-frontiers -gno-statement-frontiers @gol
384 -gvariable-location-views -gno-variable-location-views @gol
385 -ginternal-reset-location-views -gno-internal-reset-location-views @gol
386 -ginline-points -gno-inline-points @gol
387 -gvms -gxcoff -gxcoff+ -gz@r{[}=@var{type}@r{]} @gol
388 -gsplit-dwarf -gdescribe-dies -gno-describe-dies @gol
389 -fdebug-prefix-map=@var{old}=@var{new} -fdebug-types-section @gol
390 -fno-eliminate-unused-debug-types @gol
391 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
392 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
393 -fno-eliminate-unused-debug-symbols -femit-class-debug-always @gol
394 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
395 -fvar-tracking -fvar-tracking-assignments}
397 @item Optimization Options
398 @xref{Optimize Options,,Options that Control Optimization}.
399 @gccoptlist{-faggressive-loop-optimizations @gol
400 -falign-functions[=@var{n}[:@var{m}:[@var{n2}[:@var{m2}]]]] @gol
401 -falign-jumps[=@var{n}[:@var{m}:[@var{n2}[:@var{m2}]]]] @gol
402 -falign-labels[=@var{n}[:@var{m}:[@var{n2}[:@var{m2}]]]] @gol
403 -falign-loops[=@var{n}[:@var{m}:[@var{n2}[:@var{m2}]]]] @gol
404 -fassociative-math -fauto-profile -fauto-profile[=@var{path}] @gol
405 -fauto-inc-dec -fbranch-probabilities @gol
406 -fbranch-target-load-optimize -fbranch-target-load-optimize2 @gol
407 -fbtr-bb-exclusive -fcaller-saves @gol
408 -fcombine-stack-adjustments -fconserve-stack @gol
409 -fcompare-elim -fcprop-registers -fcrossjumping @gol
410 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules @gol
411 -fcx-limited-range @gol
412 -fdata-sections -fdce -fdelayed-branch @gol
413 -fdelete-null-pointer-checks -fdevirtualize -fdevirtualize-speculatively @gol
414 -fdevirtualize-at-ltrans -fdse @gol
415 -fearly-inlining -fipa-sra -fexpensive-optimizations -ffat-lto-objects @gol
416 -ffast-math -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
418 -fforward-propagate -ffp-contract=@var{style} -ffunction-sections @gol
419 -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm -fgraphite-identity @gol
420 -fgcse-sm -fhoist-adjacent-loads -fif-conversion @gol
421 -fif-conversion2 -findirect-inlining @gol
422 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
423 -finline-small-functions -fipa-cp -fipa-cp-clone @gol
424 -fipa-bit-cp -fipa-vrp -fipa-pta -fipa-profile -fipa-pure-const @gol
425 -fipa-reference -fipa-reference-addressable @gol
426 -fipa-stack-alignment -fipa-icf -fira-algorithm=@var{algorithm} @gol
427 -flive-patching=@var{level} @gol
428 -fira-region=@var{region} -fira-hoist-pressure @gol
429 -fira-loop-pressure -fno-ira-share-save-slots @gol
430 -fno-ira-share-spill-slots @gol
431 -fisolate-erroneous-paths-dereference -fisolate-erroneous-paths-attribute @gol
432 -fivopts -fkeep-inline-functions -fkeep-static-functions @gol
433 -fkeep-static-consts -flimit-function-alignment -flive-range-shrinkage @gol
434 -floop-block -floop-interchange -floop-strip-mine @gol
435 -floop-unroll-and-jam -floop-nest-optimize @gol
436 -floop-parallelize-all -flra-remat -flto -flto-compression-level @gol
437 -flto-partition=@var{alg} -fmerge-all-constants @gol
438 -fmerge-constants -fmodulo-sched -fmodulo-sched-allow-regmoves @gol
439 -fmove-loop-invariants -fno-branch-count-reg @gol
440 -fno-defer-pop -fno-fp-int-builtin-inexact -fno-function-cse @gol
441 -fno-guess-branch-probability -fno-inline -fno-math-errno -fno-peephole @gol
442 -fno-peephole2 -fno-printf-return-value -fno-sched-interblock @gol
443 -fno-sched-spec -fno-signed-zeros @gol
444 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
445 -fomit-frame-pointer -foptimize-sibling-calls @gol
446 -fpartial-inlining -fpeel-loops -fpredictive-commoning @gol
447 -fprefetch-loop-arrays @gol
448 -fprofile-correction @gol
449 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
450 -fprofile-reorder-functions @gol
451 -freciprocal-math -free -frename-registers -freorder-blocks @gol
452 -freorder-blocks-algorithm=@var{algorithm} @gol
453 -freorder-blocks-and-partition -freorder-functions @gol
454 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
455 -frounding-math -fsave-optimization-record @gol
456 -fsched2-use-superblocks -fsched-pressure @gol
457 -fsched-spec-load -fsched-spec-load-dangerous @gol
458 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
459 -fsched-group-heuristic -fsched-critical-path-heuristic @gol
460 -fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
461 -fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
462 -fschedule-fusion @gol
463 -fschedule-insns -fschedule-insns2 -fsection-anchors @gol
464 -fselective-scheduling -fselective-scheduling2 @gol
465 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
466 -fsemantic-interposition -fshrink-wrap -fshrink-wrap-separate @gol
467 -fsignaling-nans @gol
468 -fsingle-precision-constant -fsplit-ivs-in-unroller -fsplit-loops@gol
470 -fsplit-wide-types -fsplit-wide-types-early -fssa-backprop -fssa-phiopt @gol
471 -fstdarg-opt -fstore-merging -fstrict-aliasing @gol
472 -fthread-jumps -ftracer -ftree-bit-ccp @gol
473 -ftree-builtin-call-dce -ftree-ccp -ftree-ch @gol
474 -ftree-coalesce-vars -ftree-copy-prop -ftree-dce -ftree-dominator-opts @gol
475 -ftree-dse -ftree-forwprop -ftree-fre -fcode-hoisting @gol
476 -ftree-loop-if-convert -ftree-loop-im @gol
477 -ftree-phiprop -ftree-loop-distribution -ftree-loop-distribute-patterns @gol
478 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
479 -ftree-loop-vectorize @gol
480 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-partial-pre -ftree-pta @gol
481 -ftree-reassoc -ftree-scev-cprop -ftree-sink -ftree-slsr -ftree-sra @gol
482 -ftree-switch-conversion -ftree-tail-merge @gol
483 -ftree-ter -ftree-vectorize -ftree-vrp -funconstrained-commons @gol
484 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
485 -funsafe-math-optimizations -funswitch-loops @gol
486 -fipa-ra -fvariable-expansion-in-unroller -fvect-cost-model -fvpt @gol
487 -fweb -fwhole-program -fwpa -fuse-linker-plugin @gol
488 --param @var{name}=@var{value}
489 -O -O0 -O1 -O2 -O3 -Os -Ofast -Og}
491 @item Program Instrumentation Options
492 @xref{Instrumentation Options,,Program Instrumentation Options}.
493 @gccoptlist{-p -pg -fprofile-arcs --coverage -ftest-coverage @gol
494 -fprofile-abs-path @gol
495 -fprofile-dir=@var{path} -fprofile-generate -fprofile-generate=@var{path} @gol
496 -fprofile-note=@var{path} -fprofile-update=@var{method} @gol
497 -fprofile-filter-files=@var{regex} -fprofile-exclude-files=@var{regex} @gol
498 -fsanitize=@var{style} -fsanitize-recover -fsanitize-recover=@var{style} @gol
499 -fasan-shadow-offset=@var{number} -fsanitize-sections=@var{s1},@var{s2},... @gol
500 -fsanitize-undefined-trap-on-error -fbounds-check @gol
501 -fcf-protection=@r{[}full@r{|}branch@r{|}return@r{|}none@r{]} @gol
502 -fstack-protector -fstack-protector-all -fstack-protector-strong @gol
503 -fstack-protector-explicit -fstack-check @gol
504 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
505 -fno-stack-limit -fsplit-stack @gol
506 -fvtable-verify=@r{[}std@r{|}preinit@r{|}none@r{]} @gol
507 -fvtv-counts -fvtv-debug @gol
508 -finstrument-functions @gol
509 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
510 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}}
512 @item Preprocessor Options
513 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
514 @gccoptlist{-A@var{question}=@var{answer} @gol
515 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
516 -C -CC -D@var{macro}@r{[}=@var{defn}@r{]} @gol
517 -dD -dI -dM -dN -dU @gol
518 -fdebug-cpp -fdirectives-only -fdollars-in-identifiers @gol
519 -fexec-charset=@var{charset} -fextended-identifiers @gol
520 -finput-charset=@var{charset} -fmacro-prefix-map=@var{old}=@var{new} @gol
521 -fmax-include-depth=@var{depth} @gol
522 -fno-canonical-system-headers -fpch-deps -fpch-preprocess @gol
523 -fpreprocessed -ftabstop=@var{width} -ftrack-macro-expansion @gol
524 -fwide-exec-charset=@var{charset} -fworking-directory @gol
525 -H -imacros @var{file} -include @var{file} @gol
526 -M -MD -MF -MG -MM -MMD -MP -MQ -MT @gol
527 -no-integrated-cpp -P -pthread -remap @gol
528 -traditional -traditional-cpp -trigraphs @gol
529 -U@var{macro} -undef @gol
530 -Wp,@var{option} -Xpreprocessor @var{option}}
532 @item Assembler Options
533 @xref{Assembler Options,,Passing Options to the Assembler}.
534 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
537 @xref{Link Options,,Options for Linking}.
538 @gccoptlist{@var{object-file-name} -fuse-ld=@var{linker} -l@var{library} @gol
539 -nostartfiles -nodefaultlibs -nolibc -nostdlib @gol
540 -e @var{entry} --entry=@var{entry} @gol
541 -pie -pthread -r -rdynamic @gol
542 -s -static -static-pie -static-libgcc -static-libstdc++ @gol
543 -static-libasan -static-libtsan -static-liblsan -static-libubsan @gol
544 -shared -shared-libgcc -symbolic @gol
545 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
546 -u @var{symbol} -z @var{keyword}}
548 @item Directory Options
549 @xref{Directory Options,,Options for Directory Search}.
550 @gccoptlist{-B@var{prefix} -I@var{dir} -I- @gol
551 -idirafter @var{dir} @gol
552 -imacros @var{file} -imultilib @var{dir} @gol
553 -iplugindir=@var{dir} -iprefix @var{file} @gol
554 -iquote @var{dir} -isysroot @var{dir} -isystem @var{dir} @gol
555 -iwithprefix @var{dir} -iwithprefixbefore @var{dir} @gol
556 -L@var{dir} -no-canonical-prefixes --no-sysroot-suffix @gol
557 -nostdinc -nostdinc++ --sysroot=@var{dir}}
559 @item Code Generation Options
560 @xref{Code Gen Options,,Options for Code Generation Conventions}.
561 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
562 -ffixed-@var{reg} -fexceptions @gol
563 -fnon-call-exceptions -fdelete-dead-exceptions -funwind-tables @gol
564 -fasynchronous-unwind-tables @gol
566 -finhibit-size-directive -fno-common -fno-ident @gol
567 -fpcc-struct-return -fpic -fPIC -fpie -fPIE -fno-plt @gol
568 -fno-jump-tables @gol
569 -frecord-gcc-switches @gol
570 -freg-struct-return -fshort-enums -fshort-wchar @gol
571 -fverbose-asm -fpack-struct[=@var{n}] @gol
572 -fleading-underscore -ftls-model=@var{model} @gol
573 -fstack-reuse=@var{reuse_level} @gol
574 -ftrampolines -ftrapv -fwrapv @gol
575 -fvisibility=@r{[}default@r{|}internal@r{|}hidden@r{|}protected@r{]} @gol
576 -fstrict-volatile-bitfields -fsync-libcalls}
578 @item Developer Options
579 @xref{Developer Options,,GCC Developer Options}.
580 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
581 -dumpfullversion -fchecking -fchecking=@var{n} -fdbg-cnt-list @gol
582 -fdbg-cnt=@var{counter-value-list} @gol
583 -fdisable-ipa-@var{pass_name} @gol
584 -fdisable-rtl-@var{pass_name} @gol
585 -fdisable-rtl-@var{pass-name}=@var{range-list} @gol
586 -fdisable-tree-@var{pass_name} @gol
587 -fdisable-tree-@var{pass-name}=@var{range-list} @gol
588 -fdump-debug -fdump-earlydebug @gol
589 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
590 -fdump-final-insns@r{[}=@var{file}@r{]} @gol
591 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
593 -fdump-lang-@var{switch} @gol
594 -fdump-lang-@var{switch}-@var{options} @gol
595 -fdump-lang-@var{switch}-@var{options}=@var{filename} @gol
597 -fdump-rtl-@var{pass} -fdump-rtl-@var{pass}=@var{filename} @gol
598 -fdump-statistics @gol
600 -fdump-tree-@var{switch} @gol
601 -fdump-tree-@var{switch}-@var{options} @gol
602 -fdump-tree-@var{switch}-@var{options}=@var{filename} @gol
603 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
604 -fenable-@var{kind}-@var{pass} @gol
605 -fenable-@var{kind}-@var{pass}=@var{range-list} @gol
606 -fira-verbose=@var{n} @gol
607 -flto-report -flto-report-wpa -fmem-report-wpa @gol
608 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report @gol
609 -fopt-info -fopt-info-@var{options}@r{[}=@var{file}@r{]} @gol
610 -fprofile-report @gol
611 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
612 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
613 -fstats -fstack-usage -ftime-report -ftime-report-details @gol
614 -fvar-tracking-assignments-toggle -gtoggle @gol
615 -print-file-name=@var{library} -print-libgcc-file-name @gol
616 -print-multi-directory -print-multi-lib -print-multi-os-directory @gol
617 -print-prog-name=@var{program} -print-search-dirs -Q @gol
618 -print-sysroot -print-sysroot-headers-suffix @gol
619 -save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
621 @item Machine-Dependent Options
622 @xref{Submodel Options,,Machine-Dependent Options}.
623 @c This list is ordered alphanumerically by subsection name.
624 @c Try and put the significant identifier (CPU or system) first,
625 @c so users have a clue at guessing where the ones they want will be.
627 @emph{AArch64 Options}
628 @gccoptlist{-mabi=@var{name} -mbig-endian -mlittle-endian @gol
629 -mgeneral-regs-only @gol
630 -mcmodel=tiny -mcmodel=small -mcmodel=large @gol
631 -mstrict-align -mno-strict-align @gol
632 -momit-leaf-frame-pointer @gol
633 -mtls-dialect=desc -mtls-dialect=traditional @gol
634 -mtls-size=@var{size} @gol
635 -mfix-cortex-a53-835769 -mfix-cortex-a53-843419 @gol
636 -mlow-precision-recip-sqrt -mlow-precision-sqrt -mlow-precision-div @gol
637 -mpc-relative-literal-loads @gol
638 -msign-return-address=@var{scope} @gol
639 -mbranch-protection=@var{none}|@var{standard}|@var{pac-ret}[+@var{leaf}
640 +@var{b-key}]|@var{bti} @gol
641 -march=@var{name} -mcpu=@var{name} -mtune=@var{name} @gol
642 -moverride=@var{string} -mverbose-cost-dump @gol
643 -mstack-protector-guard=@var{guard} -mstack-protector-guard-reg=@var{sysreg} @gol
644 -mstack-protector-guard-offset=@var{offset} -mtrack-speculation }
646 @emph{Adapteva Epiphany Options}
647 @gccoptlist{-mhalf-reg-file -mprefer-short-insn-regs @gol
648 -mbranch-cost=@var{num} -mcmove -mnops=@var{num} -msoft-cmpsf @gol
649 -msplit-lohi -mpost-inc -mpost-modify -mstack-offset=@var{num} @gol
650 -mround-nearest -mlong-calls -mshort-calls -msmall16 @gol
651 -mfp-mode=@var{mode} -mvect-double -max-vect-align=@var{num} @gol
652 -msplit-vecmove-early -m1reg-@var{reg}}
654 @emph{AMD GCN Options}
655 @gccoptlist{-march=@var{gpu} -mtune=@var{gpu} -mstack-size=@var{bytes}}
658 @gccoptlist{-mbarrel-shifter -mjli-always @gol
659 -mcpu=@var{cpu} -mA6 -mARC600 -mA7 -mARC700 @gol
660 -mdpfp -mdpfp-compact -mdpfp-fast -mno-dpfp-lrsr @gol
661 -mea -mno-mpy -mmul32x16 -mmul64 -matomic @gol
662 -mnorm -mspfp -mspfp-compact -mspfp-fast -msimd -msoft-float -mswap @gol
663 -mcrc -mdsp-packa -mdvbf -mlock -mmac-d16 -mmac-24 -mrtsc -mswape @gol
664 -mtelephony -mxy -misize -mannotate-align -marclinux -marclinux_prof @gol
665 -mlong-calls -mmedium-calls -msdata -mirq-ctrl-saved @gol
666 -mrgf-banked-regs -mlpc-width=@var{width} -G @var{num} @gol
667 -mvolatile-cache -mtp-regno=@var{regno} @gol
668 -malign-call -mauto-modify-reg -mbbit-peephole -mno-brcc @gol
669 -mcase-vector-pcrel -mcompact-casesi -mno-cond-exec -mearly-cbranchsi @gol
670 -mexpand-adddi -mindexed-loads -mlra -mlra-priority-none @gol
671 -mlra-priority-compact mlra-priority-noncompact -mmillicode @gol
672 -mmixed-code -mq-class -mRcq -mRcw -msize-level=@var{level} @gol
673 -mtune=@var{cpu} -mmultcost=@var{num} -mcode-density-frame @gol
674 -munalign-prob-threshold=@var{probability} -mmpy-option=@var{multo} @gol
675 -mdiv-rem -mcode-density -mll64 -mfpu=@var{fpu} -mrf16 -mbranch-index}
678 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
679 -mabi=@var{name} @gol
680 -mapcs-stack-check -mno-apcs-stack-check @gol
681 -mapcs-reentrant -mno-apcs-reentrant @gol
682 -mgeneral-regs-only @gol
683 -msched-prolog -mno-sched-prolog @gol
684 -mlittle-endian -mbig-endian @gol
686 -mfloat-abi=@var{name} @gol
687 -mfp16-format=@var{name}
688 -mthumb-interwork -mno-thumb-interwork @gol
689 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
690 -mtune=@var{name} -mprint-tune-info @gol
691 -mstructure-size-boundary=@var{n} @gol
692 -mabort-on-noreturn @gol
693 -mlong-calls -mno-long-calls @gol
694 -msingle-pic-base -mno-single-pic-base @gol
695 -mpic-register=@var{reg} @gol
696 -mnop-fun-dllimport @gol
697 -mpoke-function-name @gol
698 -mthumb -marm -mflip-thumb @gol
699 -mtpcs-frame -mtpcs-leaf-frame @gol
700 -mcaller-super-interworking -mcallee-super-interworking @gol
701 -mtp=@var{name} -mtls-dialect=@var{dialect} @gol
702 -mword-relocations @gol
703 -mfix-cortex-m3-ldrd @gol
704 -munaligned-access @gol
705 -mneon-for-64bits @gol
706 -mslow-flash-data @gol
707 -masm-syntax-unified @gol
709 -mverbose-cost-dump @gol
714 @gccoptlist{-mmcu=@var{mcu} -mabsdata -maccumulate-args @gol
715 -mbranch-cost=@var{cost} @gol
716 -mcall-prologues -mgas-isr-prologues -mint8 @gol
717 -mn_flash=@var{size} -mno-interrupts @gol
718 -mmain-is-OS_task -mrelax -mrmw -mstrict-X -mtiny-stack @gol
719 -mfract-convert-truncate @gol
720 -mshort-calls -nodevicelib @gol
721 -Waddr-space-convert -Wmisspelled-isr}
723 @emph{Blackfin Options}
724 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
725 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
726 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
727 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
728 -mno-id-shared-library -mshared-library-id=@var{n} @gol
729 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
730 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
731 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
735 @gccoptlist{-mbig-endian -mlittle-endian -march=@var{cpu} @gol
736 -msim -msdata=@var{sdata-type}}
739 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
740 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
741 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
742 -mstack-align -mdata-align -mconst-align @gol
743 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
744 -melf -maout -melinux -mlinux -sim -sim2 @gol
745 -mmul-bug-workaround -mno-mul-bug-workaround}
748 @gccoptlist{-mmac @gol
749 -mcr16cplus -mcr16c @gol
750 -msim -mint32 -mbit-ops
751 -mdata-model=@var{model}}
754 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} @gol
755 -mbig-endian -EB -mlittle-endian -EL @gol
756 -mhard-float -msoft-float -mfpu=@var{fpu} -mdouble-float -mfdivdu @gol
757 -melrw -mistack -mmp -mcp -mcache -msecurity -mtrust @gol
758 -mdsp -medsp -mvdsp @gol
759 -mdiv -msmart -mhigh-registers -manchor @gol
760 -mpushpop -mmultiple-stld -mconstpool -mstack-size -mccrt @gol
761 -mbranch-cost=@var{n} -mcse-cc -msched-prolog}
763 @emph{Darwin Options}
764 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
765 -arch_only -bind_at_load -bundle -bundle_loader @gol
766 -client_name -compatibility_version -current_version @gol
768 -dependency-file -dylib_file -dylinker_install_name @gol
769 -dynamic -dynamiclib -exported_symbols_list @gol
770 -filelist -flat_namespace -force_cpusubtype_ALL @gol
771 -force_flat_namespace -headerpad_max_install_names @gol
773 -image_base -init -install_name -keep_private_externs @gol
774 -multi_module -multiply_defined -multiply_defined_unused @gol
775 -noall_load -no_dead_strip_inits_and_terms @gol
776 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
777 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
778 -private_bundle -read_only_relocs -sectalign @gol
779 -sectobjectsymbols -whyload -seg1addr @gol
780 -sectcreate -sectobjectsymbols -sectorder @gol
781 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
782 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
783 -segprot -segs_read_only_addr -segs_read_write_addr @gol
784 -single_module -static -sub_library -sub_umbrella @gol
785 -twolevel_namespace -umbrella -undefined @gol
786 -unexported_symbols_list -weak_reference_mismatches @gol
787 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
788 -mkernel -mone-byte-bool}
790 @emph{DEC Alpha Options}
791 @gccoptlist{-mno-fp-regs -msoft-float @gol
792 -mieee -mieee-with-inexact -mieee-conformant @gol
793 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
794 -mtrap-precision=@var{mode} -mbuild-constants @gol
795 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
796 -mbwx -mmax -mfix -mcix @gol
797 -mfloat-vax -mfloat-ieee @gol
798 -mexplicit-relocs -msmall-data -mlarge-data @gol
799 -msmall-text -mlarge-text @gol
800 -mmemory-latency=@var{time}}
803 @gccoptlist{-msmall-model -mno-lsim}
806 @gccoptlist{-msim -mlra -mnodiv -mft32b -mcompress -mnopm}
809 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
810 -mhard-float -msoft-float @gol
811 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
812 -mdouble -mno-double @gol
813 -mmedia -mno-media -mmuladd -mno-muladd @gol
814 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
815 -mlinked-fp -mlong-calls -malign-labels @gol
816 -mlibrary-pic -macc-4 -macc-8 @gol
817 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
818 -moptimize-membar -mno-optimize-membar @gol
819 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
820 -mvliw-branch -mno-vliw-branch @gol
821 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
822 -mno-nested-cond-exec -mtomcat-stats @gol
826 @emph{GNU/Linux Options}
827 @gccoptlist{-mglibc -muclibc -mmusl -mbionic -mandroid @gol
828 -tno-android-cc -tno-android-ld}
830 @emph{H8/300 Options}
831 @gccoptlist{-mrelax -mh -ms -mn -mexr -mno-exr -mint32 -malign-300}
834 @gccoptlist{-march=@var{architecture-type} @gol
835 -mcaller-copies -mdisable-fpregs -mdisable-indexing @gol
836 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
837 -mfixed-range=@var{register-range} @gol
838 -mjump-in-delay -mlinker-opt -mlong-calls @gol
839 -mlong-load-store -mno-disable-fpregs @gol
840 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
841 -mno-jump-in-delay -mno-long-load-store @gol
842 -mno-portable-runtime -mno-soft-float @gol
843 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
844 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
845 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
846 -munix=@var{unix-std} -nolibdld -static -threads}
849 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
850 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
851 -mconstant-gp -mauto-pic -mfused-madd @gol
852 -minline-float-divide-min-latency @gol
853 -minline-float-divide-max-throughput @gol
854 -mno-inline-float-divide @gol
855 -minline-int-divide-min-latency @gol
856 -minline-int-divide-max-throughput @gol
857 -mno-inline-int-divide @gol
858 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
859 -mno-inline-sqrt @gol
860 -mdwarf2-asm -mearly-stop-bits @gol
861 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
862 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
863 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
864 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
865 -msched-spec-ldc -msched-spec-control-ldc @gol
866 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
867 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
868 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
869 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
872 @gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
873 -msign-extend-enabled -muser-enabled}
875 @emph{M32R/D Options}
876 @gccoptlist{-m32r2 -m32rx -m32r @gol
878 -malign-loops -mno-align-loops @gol
879 -missue-rate=@var{number} @gol
880 -mbranch-cost=@var{number} @gol
881 -mmodel=@var{code-size-model-type} @gol
882 -msdata=@var{sdata-type} @gol
883 -mno-flush-func -mflush-func=@var{name} @gol
884 -mno-flush-trap -mflush-trap=@var{number} @gol
888 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
890 @emph{M680x0 Options}
891 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune} @gol
892 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
893 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
894 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
895 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
896 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
897 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
898 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
899 -mxgot -mno-xgot -mlong-jump-table-offsets}
902 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
903 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
904 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
905 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
906 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
909 @gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
910 -mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
911 -mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
912 -mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
915 @emph{MicroBlaze Options}
916 @gccoptlist{-msoft-float -mhard-float -msmall-divides -mcpu=@var{cpu} @gol
917 -mmemcpy -mxl-soft-mul -mxl-soft-div -mxl-barrel-shift @gol
918 -mxl-pattern-compare -mxl-stack-check -mxl-gp-opt -mno-clearbss @gol
919 -mxl-multiply-high -mxl-float-convert -mxl-float-sqrt @gol
920 -mbig-endian -mlittle-endian -mxl-reorder -mxl-mode-@var{app-model} @gol
921 -mpic-data-is-text-relative}
924 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
925 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips32r3 -mips32r5 @gol
926 -mips32r6 -mips64 -mips64r2 -mips64r3 -mips64r5 -mips64r6 @gol
927 -mips16 -mno-mips16 -mflip-mips16 @gol
928 -minterlink-compressed -mno-interlink-compressed @gol
929 -minterlink-mips16 -mno-interlink-mips16 @gol
930 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
931 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
932 -mgp32 -mgp64 -mfp32 -mfpxx -mfp64 -mhard-float -msoft-float @gol
933 -mno-float -msingle-float -mdouble-float @gol
934 -modd-spreg -mno-odd-spreg @gol
935 -mabs=@var{mode} -mnan=@var{encoding} @gol
936 -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
939 -mvirt -mno-virt @gol
942 -mginv -mno-ginv @gol
943 -mmicromips -mno-micromips @gol
945 -mloongson-mmi -mno-loongson-mmi @gol
946 -mloongson-ext -mno-loongson-ext @gol
947 -mloongson-ext2 -mno-loongson-ext2 @gol
948 -mfpu=@var{fpu-type} @gol
949 -msmartmips -mno-smartmips @gol
950 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
951 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
952 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
953 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
954 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
955 -membedded-data -mno-embedded-data @gol
956 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
957 -mcode-readable=@var{setting} @gol
958 -msplit-addresses -mno-split-addresses @gol
959 -mexplicit-relocs -mno-explicit-relocs @gol
960 -mcheck-zero-division -mno-check-zero-division @gol
961 -mdivide-traps -mdivide-breaks @gol
962 -mload-store-pairs -mno-load-store-pairs @gol
963 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
964 -mmad -mno-mad -mimadd -mno-imadd -mfused-madd -mno-fused-madd -nocpp @gol
965 -mfix-24k -mno-fix-24k @gol
966 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
967 -mfix-r5900 -mno-fix-r5900 @gol
968 -mfix-r10000 -mno-fix-r10000 -mfix-rm7000 -mno-fix-rm7000 @gol
969 -mfix-vr4120 -mno-fix-vr4120 @gol
970 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
971 -mflush-func=@var{func} -mno-flush-func @gol
972 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
973 -mcompact-branches=@var{policy} @gol
974 -mfp-exceptions -mno-fp-exceptions @gol
975 -mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
976 -mlxc1-sxc1 -mno-lxc1-sxc1 -mmadd4 -mno-madd4 @gol
977 -mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address @gol
978 -mframe-header-opt -mno-frame-header-opt}
981 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
982 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
983 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
984 -mno-base-addresses -msingle-exit -mno-single-exit}
986 @emph{MN10300 Options}
987 @gccoptlist{-mmult-bug -mno-mult-bug @gol
988 -mno-am33 -mam33 -mam33-2 -mam34 @gol
989 -mtune=@var{cpu-type} @gol
990 -mreturn-pointer-on-d0 @gol
991 -mno-crt0 -mrelax -mliw -msetlb}
994 @gccoptlist{-meb -mel -mmul.x -mno-crt0}
996 @emph{MSP430 Options}
997 @gccoptlist{-msim -masm-hex -mmcu= -mcpu= -mlarge -msmall -mrelax @gol
999 -mcode-region= -mdata-region= @gol
1000 -msilicon-errata= -msilicon-errata-warn= @gol
1003 @emph{NDS32 Options}
1004 @gccoptlist{-mbig-endian -mlittle-endian @gol
1005 -mreduced-regs -mfull-regs @gol
1006 -mcmov -mno-cmov @gol
1007 -mext-perf -mno-ext-perf @gol
1008 -mext-perf2 -mno-ext-perf2 @gol
1009 -mext-string -mno-ext-string @gol
1010 -mv3push -mno-v3push @gol
1011 -m16bit -mno-16bit @gol
1012 -misr-vector-size=@var{num} @gol
1013 -mcache-block-size=@var{num} @gol
1014 -march=@var{arch} @gol
1015 -mcmodel=@var{code-model} @gol
1016 -mctor-dtor -mrelax}
1018 @emph{Nios II Options}
1019 @gccoptlist{-G @var{num} -mgpopt=@var{option} -mgpopt -mno-gpopt @gol
1020 -mgprel-sec=@var{regexp} -mr0rel-sec=@var{regexp} @gol
1022 -mno-bypass-cache -mbypass-cache @gol
1023 -mno-cache-volatile -mcache-volatile @gol
1024 -mno-fast-sw-div -mfast-sw-div @gol
1025 -mhw-mul -mno-hw-mul -mhw-mulx -mno-hw-mulx -mno-hw-div -mhw-div @gol
1026 -mcustom-@var{insn}=@var{N} -mno-custom-@var{insn} @gol
1027 -mcustom-fpu-cfg=@var{name} @gol
1028 -mhal -msmallc -msys-crt0=@var{name} -msys-lib=@var{name} @gol
1029 -march=@var{arch} -mbmx -mno-bmx -mcdx -mno-cdx}
1031 @emph{Nvidia PTX Options}
1032 @gccoptlist{-m32 -m64 -mmainkernel -moptimize}
1034 @emph{OpenRISC Options}
1035 @gccoptlist{-mboard=@var{name} -mnewlib -mhard-mul -mhard-div @gol
1036 -msoft-mul -msoft-div @gol
1037 -msoft-float -mhard-float -mdouble-float -munordered-float @gol
1038 -mcmov -mror -mrori -msext -msfimm -mshftimm}
1040 @emph{PDP-11 Options}
1041 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
1042 -mint32 -mno-int16 -mint16 -mno-int32 @gol
1043 -msplit -munix-asm -mdec-asm -mgnu-asm -mlra}
1045 @emph{picoChip Options}
1046 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N} @gol
1047 -msymbol-as-address -mno-inefficient-warnings}
1049 @emph{PowerPC Options}
1050 See RS/6000 and PowerPC Options.
1053 @gccoptlist{-mmcu=@var{mcu} -minrt -mno-relax -mloop @gol
1054 -mabi=@var{variant} @gol}
1056 @emph{RISC-V Options}
1057 @gccoptlist{-mbranch-cost=@var{N-instruction} @gol
1059 -mabi=@var{ABI-string} @gol
1060 -mfdiv -mno-fdiv @gol
1062 -march=@var{ISA-string} @gol
1063 -mtune=@var{processor-string} @gol
1064 -mpreferred-stack-boundary=@var{num} @gol
1065 -msmall-data-limit=@var{N-bytes} @gol
1066 -msave-restore -mno-save-restore @gol
1067 -mstrict-align -mno-strict-align @gol
1068 -mcmodel=medlow -mcmodel=medany @gol
1069 -mexplicit-relocs -mno-explicit-relocs @gol
1070 -mrelax -mno-relax @gol
1071 -mriscv-attribute -mmo-riscv-attribute @gol
1072 -malign-data=@var{type}}
1075 @gccoptlist{-msim -mmul=none -mmul=g13 -mmul=g14 -mallregs @gol
1076 -mcpu=g10 -mcpu=g13 -mcpu=g14 -mg10 -mg13 -mg14 @gol
1077 -m64bit-doubles -m32bit-doubles -msave-mduc-in-interrupts}
1079 @emph{RS/6000 and PowerPC Options}
1080 @gccoptlist{-mcpu=@var{cpu-type} @gol
1081 -mtune=@var{cpu-type} @gol
1082 -mcmodel=@var{code-model} @gol
1084 -maltivec -mno-altivec @gol
1085 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
1086 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
1087 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
1088 -mfprnd -mno-fprnd @gol
1089 -mcmpb -mno-cmpb -mhard-dfp -mno-hard-dfp @gol
1090 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
1091 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
1092 -malign-power -malign-natural @gol
1093 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
1094 -mupdate -mno-update @gol
1095 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
1096 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
1097 -mstrict-align -mno-strict-align -mrelocatable @gol
1098 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
1099 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
1100 -mdynamic-no-pic -mswdiv -msingle-pic-base @gol
1101 -mprioritize-restricted-insns=@var{priority} @gol
1102 -msched-costly-dep=@var{dependence_type} @gol
1103 -minsert-sched-nops=@var{scheme} @gol
1104 -mcall-aixdesc -mcall-eabi -mcall-freebsd @gol
1105 -mcall-linux -mcall-netbsd -mcall-openbsd @gol
1106 -mcall-sysv -mcall-sysv-eabi -mcall-sysv-noeabi @gol
1107 -mtraceback=@var{traceback_type} @gol
1108 -maix-struct-return -msvr4-struct-return @gol
1109 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
1110 -mlongcall -mno-longcall -mpltseq -mno-pltseq @gol
1111 -mblock-move-inline-limit=@var{num} @gol
1112 -mblock-compare-inline-limit=@var{num} @gol
1113 -mblock-compare-inline-loop-limit=@var{num} @gol
1114 -mstring-compare-inline-limit=@var{num} @gol
1115 -misel -mno-isel @gol
1116 -mvrsave -mno-vrsave @gol
1117 -mmulhw -mno-mulhw @gol
1118 -mdlmzb -mno-dlmzb @gol
1119 -mprototype -mno-prototype @gol
1120 -msim -mmvme -mads -myellowknife -memb -msdata @gol
1121 -msdata=@var{opt} -mreadonly-in-sdata -mvxworks -G @var{num} @gol
1122 -mrecip -mrecip=@var{opt} -mno-recip -mrecip-precision @gol
1123 -mno-recip-precision @gol
1124 -mveclibabi=@var{type} -mfriz -mno-friz @gol
1125 -mpointers-to-nested-functions -mno-pointers-to-nested-functions @gol
1126 -msave-toc-indirect -mno-save-toc-indirect @gol
1127 -mpower8-fusion -mno-mpower8-fusion -mpower8-vector -mno-power8-vector @gol
1128 -mcrypto -mno-crypto -mhtm -mno-htm @gol
1129 -mquad-memory -mno-quad-memory @gol
1130 -mquad-memory-atomic -mno-quad-memory-atomic @gol
1131 -mcompat-align-parm -mno-compat-align-parm @gol
1132 -mfloat128 -mno-float128 -mfloat128-hardware -mno-float128-hardware @gol
1133 -mgnu-attribute -mno-gnu-attribute @gol
1134 -mstack-protector-guard=@var{guard} -mstack-protector-guard-reg=@var{reg} @gol
1135 -mstack-protector-guard-offset=@var{offset} -mpcrel -mno-pcrel}
1138 @gccoptlist{-m64bit-doubles -m32bit-doubles -fpu -nofpu@gol
1140 -mbig-endian-data -mlittle-endian-data @gol
1143 -mas100-syntax -mno-as100-syntax@gol
1145 -mmax-constant-size=@gol
1148 -mallow-string-insns -mno-allow-string-insns@gol
1150 -mno-warn-multiple-fast-interrupts@gol
1151 -msave-acc-in-interrupts}
1153 @emph{S/390 and zSeries Options}
1154 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
1155 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
1156 -mlong-double-64 -mlong-double-128 @gol
1157 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
1158 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
1159 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
1160 -mhtm -mvx -mzvector @gol
1161 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
1162 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard @gol
1163 -mhotpatch=@var{halfwords},@var{halfwords}}
1165 @emph{Score Options}
1166 @gccoptlist{-meb -mel @gol
1170 -mscore5 -mscore5u -mscore7 -mscore7d}
1173 @gccoptlist{-m1 -m2 -m2e @gol
1174 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
1176 -m4-nofpu -m4-single-only -m4-single -m4 @gol
1177 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
1178 -mb -ml -mdalign -mrelax @gol
1179 -mbigtable -mfmovd -mrenesas -mno-renesas -mnomacsave @gol
1180 -mieee -mno-ieee -mbitops -misize -minline-ic_invalidate -mpadstruct @gol
1181 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
1182 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
1183 -maccumulate-outgoing-args @gol
1184 -matomic-model=@var{atomic-model} @gol
1185 -mbranch-cost=@var{num} -mzdcbranch -mno-zdcbranch @gol
1186 -mcbranch-force-delay-slot @gol
1187 -mfused-madd -mno-fused-madd -mfsca -mno-fsca -mfsrra -mno-fsrra @gol
1188 -mpretend-cmove -mtas}
1190 @emph{Solaris 2 Options}
1191 @gccoptlist{-mclear-hwcap -mno-clear-hwcap -mimpure-text -mno-impure-text @gol
1194 @emph{SPARC Options}
1195 @gccoptlist{-mcpu=@var{cpu-type} @gol
1196 -mtune=@var{cpu-type} @gol
1197 -mcmodel=@var{code-model} @gol
1198 -mmemory-model=@var{mem-model} @gol
1199 -m32 -m64 -mapp-regs -mno-app-regs @gol
1200 -mfaster-structs -mno-faster-structs -mflat -mno-flat @gol
1201 -mfpu -mno-fpu -mhard-float -msoft-float @gol
1202 -mhard-quad-float -msoft-quad-float @gol
1203 -mstack-bias -mno-stack-bias @gol
1204 -mstd-struct-return -mno-std-struct-return @gol
1205 -munaligned-doubles -mno-unaligned-doubles @gol
1206 -muser-mode -mno-user-mode @gol
1207 -mv8plus -mno-v8plus -mvis -mno-vis @gol
1208 -mvis2 -mno-vis2 -mvis3 -mno-vis3 @gol
1209 -mvis4 -mno-vis4 -mvis4b -mno-vis4b @gol
1210 -mcbcond -mno-cbcond -mfmaf -mno-fmaf -mfsmuld -mno-fsmuld @gol
1211 -mpopc -mno-popc -msubxc -mno-subxc @gol
1212 -mfix-at697f -mfix-ut699 -mfix-ut700 -mfix-gr712rc @gol
1216 @gccoptlist{-mwarn-reloc -merror-reloc @gol
1217 -msafe-dma -munsafe-dma @gol
1219 -msmall-mem -mlarge-mem -mstdmain @gol
1220 -mfixed-range=@var{register-range} @gol
1222 -maddress-space-conversion -mno-address-space-conversion @gol
1223 -mcache-size=@var{cache-size} @gol
1224 -matomic-updates -mno-atomic-updates}
1226 @emph{System V Options}
1227 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
1229 @emph{TILE-Gx Options}
1230 @gccoptlist{-mcpu=CPU -m32 -m64 -mbig-endian -mlittle-endian @gol
1231 -mcmodel=@var{code-model}}
1233 @emph{TILEPro Options}
1234 @gccoptlist{-mcpu=@var{cpu} -m32}
1237 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
1238 -mprolog-function -mno-prolog-function -mspace @gol
1239 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
1240 -mapp-regs -mno-app-regs @gol
1241 -mdisable-callt -mno-disable-callt @gol
1242 -mv850e2v3 -mv850e2 -mv850e1 -mv850es @gol
1243 -mv850e -mv850 -mv850e3v5 @gol
1254 @gccoptlist{-mg -mgnu -munix}
1256 @emph{Visium Options}
1257 @gccoptlist{-mdebug -msim -mfpu -mno-fpu -mhard-float -msoft-float @gol
1258 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} -msv-mode -muser-mode}
1261 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64 @gol
1262 -mpointer-size=@var{size}}
1264 @emph{VxWorks Options}
1265 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
1266 -Xbind-lazy -Xbind-now}
1269 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
1270 -mtune-ctrl=@var{feature-list} -mdump-tune-features -mno-default @gol
1271 -mfpmath=@var{unit} @gol
1272 -masm=@var{dialect} -mno-fancy-math-387 @gol
1273 -mno-fp-ret-in-387 -m80387 -mhard-float -msoft-float @gol
1274 -mno-wide-multiply -mrtd -malign-double @gol
1275 -mpreferred-stack-boundary=@var{num} @gol
1276 -mincoming-stack-boundary=@var{num} @gol
1277 -mcld -mcx16 -msahf -mmovbe -mcrc32 @gol
1278 -mrecip -mrecip=@var{opt} @gol
1279 -mvzeroupper -mprefer-avx128 -mprefer-vector-width=@var{opt} @gol
1280 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
1281 -mavx2 -mavx512f -mavx512pf -mavx512er -mavx512cd -mavx512vl @gol
1282 -mavx512bw -mavx512dq -mavx512ifma -mavx512vbmi -msha -maes @gol
1283 -mpclmul -mfsgsbase -mrdrnd -mf16c -mfma -mpconfig -mwbnoinvd @gol
1284 -mptwrite -mprefetchwt1 -mclflushopt -mclwb -mxsavec -mxsaves @gol
1285 -msse4a -m3dnow -m3dnowa -mpopcnt -mabm -mbmi -mtbm -mfma4 -mxop @gol
1286 -madx -mlzcnt -mbmi2 -mfxsr -mxsave -mxsaveopt -mrtm -mhle -mlwp @gol
1287 -mmwaitx -mclzero -mpku -mthreads -mgfni -mvaes -mwaitpkg @gol
1288 -mshstk -mmanual-endbr -mforce-indirect-call -mavx512vbmi2 -mavx512bf16 -menqcmd @gol
1289 -mvpclmulqdq -mavx512bitalg -mmovdiri -mmovdir64b -mavx512vpopcntdq @gol
1290 -mavx5124fmaps -mavx512vnni -mavx5124vnniw -mprfchw -mrdpid @gol
1291 -mrdseed -msgx -mavx512vp2intersect@gol
1292 -mcldemote -mms-bitfields -mno-align-stringops -minline-all-stringops @gol
1293 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
1294 -mmemcpy-strategy=@var{strategy} -mmemset-strategy=@var{strategy} @gol
1295 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
1296 -m96bit-long-double -mlong-double-64 -mlong-double-80 -mlong-double-128 @gol
1297 -mregparm=@var{num} -msseregparm @gol
1298 -mveclibabi=@var{type} -mvect8-ret-in-mem @gol
1299 -mpc32 -mpc64 -mpc80 -mstackrealign @gol
1300 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
1301 -mcmodel=@var{code-model} -mabi=@var{name} -maddress-mode=@var{mode} @gol
1302 -m32 -m64 -mx32 -m16 -miamcu -mlarge-data-threshold=@var{num} @gol
1303 -msse2avx -mfentry -mrecord-mcount -mnop-mcount -m8bit-idiv @gol
1304 -minstrument-return=@var{type} -mfentry-name=@var{name} -mfentry-section=@var{name} @gol
1305 -mavx256-split-unaligned-load -mavx256-split-unaligned-store @gol
1306 -malign-data=@var{type} -mstack-protector-guard=@var{guard} @gol
1307 -mstack-protector-guard-reg=@var{reg} @gol
1308 -mstack-protector-guard-offset=@var{offset} @gol
1309 -mstack-protector-guard-symbol=@var{symbol} @gol
1310 -mgeneral-regs-only -mcall-ms2sysv-xlogues @gol
1311 -mindirect-branch=@var{choice} -mfunction-return=@var{choice} @gol
1312 -mindirect-branch-register}
1314 @emph{x86 Windows Options}
1315 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll @gol
1316 -mnop-fun-dllimport -mthread @gol
1317 -municode -mwin32 -mwindows -fno-set-stack-executable}
1319 @emph{Xstormy16 Options}
1322 @emph{Xtensa Options}
1323 @gccoptlist{-mconst16 -mno-const16 @gol
1324 -mfused-madd -mno-fused-madd @gol
1326 -mserialize-volatile -mno-serialize-volatile @gol
1327 -mtext-section-literals -mno-text-section-literals @gol
1328 -mauto-litpools -mno-auto-litpools @gol
1329 -mtarget-align -mno-target-align @gol
1330 -mlongcalls -mno-longcalls}
1332 @emph{zSeries Options}
1333 See S/390 and zSeries Options.
1337 @node Overall Options
1338 @section Options Controlling the Kind of Output
1340 Compilation can involve up to four stages: preprocessing, compilation
1341 proper, assembly and linking, always in that order. GCC is capable of
1342 preprocessing and compiling several files either into several
1343 assembler input files, or into one assembler input file; then each
1344 assembler input file produces an object file, and linking combines all
1345 the object files (those newly compiled, and those specified as input)
1346 into an executable file.
1348 @cindex file name suffix
1349 For any given input file, the file name suffix determines what kind of
1350 compilation is done:
1354 C source code that must be preprocessed.
1357 C source code that should not be preprocessed.
1360 C++ source code that should not be preprocessed.
1363 Objective-C source code. Note that you must link with the @file{libobjc}
1364 library to make an Objective-C program work.
1367 Objective-C source code that should not be preprocessed.
1371 Objective-C++ source code. Note that you must link with the @file{libobjc}
1372 library to make an Objective-C++ program work. Note that @samp{.M} refers
1373 to a literal capital M@.
1375 @item @var{file}.mii
1376 Objective-C++ source code that should not be preprocessed.
1379 C, C++, Objective-C or Objective-C++ header file to be turned into a
1380 precompiled header (default), or C, C++ header file to be turned into an
1381 Ada spec (via the @option{-fdump-ada-spec} switch).
1384 @itemx @var{file}.cp
1385 @itemx @var{file}.cxx
1386 @itemx @var{file}.cpp
1387 @itemx @var{file}.CPP
1388 @itemx @var{file}.c++
1390 C++ source code that must be preprocessed. Note that in @samp{.cxx},
1391 the last two letters must both be literally @samp{x}. Likewise,
1392 @samp{.C} refers to a literal capital C@.
1396 Objective-C++ source code that must be preprocessed.
1398 @item @var{file}.mii
1399 Objective-C++ source code that should not be preprocessed.
1403 @itemx @var{file}.hp
1404 @itemx @var{file}.hxx
1405 @itemx @var{file}.hpp
1406 @itemx @var{file}.HPP
1407 @itemx @var{file}.h++
1408 @itemx @var{file}.tcc
1409 C++ header file to be turned into a precompiled header or Ada spec.
1412 @itemx @var{file}.for
1413 @itemx @var{file}.ftn
1414 Fixed form Fortran source code that should not be preprocessed.
1417 @itemx @var{file}.FOR
1418 @itemx @var{file}.fpp
1419 @itemx @var{file}.FPP
1420 @itemx @var{file}.FTN
1421 Fixed form Fortran source code that must be preprocessed (with the traditional
1424 @item @var{file}.f90
1425 @itemx @var{file}.f95
1426 @itemx @var{file}.f03
1427 @itemx @var{file}.f08
1428 Free form Fortran source code that should not be preprocessed.
1430 @item @var{file}.F90
1431 @itemx @var{file}.F95
1432 @itemx @var{file}.F03
1433 @itemx @var{file}.F08
1434 Free form Fortran source code that must be preprocessed (with the
1435 traditional preprocessor).
1440 @item @var{file}.brig
1441 BRIG files (binary representation of HSAIL).
1450 D documentation code (Ddoc).
1452 @item @var{file}.ads
1453 Ada source code file that contains a library unit declaration (a
1454 declaration of a package, subprogram, or generic, or a generic
1455 instantiation), or a library unit renaming declaration (a package,
1456 generic, or subprogram renaming declaration). Such files are also
1459 @item @var{file}.adb
1460 Ada source code file containing a library unit body (a subprogram or
1461 package body). Such files are also called @dfn{bodies}.
1463 @c GCC also knows about some suffixes for languages not yet included:
1471 @itemx @var{file}.sx
1472 Assembler code that must be preprocessed.
1475 An object file to be fed straight into linking.
1476 Any file name with no recognized suffix is treated this way.
1480 You can specify the input language explicitly with the @option{-x} option:
1483 @item -x @var{language}
1484 Specify explicitly the @var{language} for the following input files
1485 (rather than letting the compiler choose a default based on the file
1486 name suffix). This option applies to all following input files until
1487 the next @option{-x} option. Possible values for @var{language} are:
1489 c c-header cpp-output
1490 c++ c++-header c++-cpp-output
1491 objective-c objective-c-header objective-c-cpp-output
1492 objective-c++ objective-c++-header objective-c++-cpp-output
1493 assembler assembler-with-cpp
1496 f77 f77-cpp-input f95 f95-cpp-input
1502 Turn off any specification of a language, so that subsequent files are
1503 handled according to their file name suffixes (as they are if @option{-x}
1504 has not been used at all).
1507 If you only want some of the stages of compilation, you can use
1508 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1509 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1510 @command{gcc} is to stop. Note that some combinations (for example,
1511 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1516 Compile or assemble the source files, but do not link. The linking
1517 stage simply is not done. The ultimate output is in the form of an
1518 object file for each source file.
1520 By default, the object file name for a source file is made by replacing
1521 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1523 Unrecognized input files, not requiring compilation or assembly, are
1528 Stop after the stage of compilation proper; do not assemble. The output
1529 is in the form of an assembler code file for each non-assembler input
1532 By default, the assembler file name for a source file is made by
1533 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1535 Input files that don't require compilation are ignored.
1539 Stop after the preprocessing stage; do not run the compiler proper. The
1540 output is in the form of preprocessed source code, which is sent to the
1543 Input files that don't require preprocessing are ignored.
1545 @cindex output file option
1548 Place output in file @var{file}. This applies to whatever
1549 sort of output is being produced, whether it be an executable file,
1550 an object file, an assembler file or preprocessed C code.
1552 If @option{-o} is not specified, the default is to put an executable
1553 file in @file{a.out}, the object file for
1554 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1555 assembler file in @file{@var{source}.s}, a precompiled header file in
1556 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1561 Print (on standard error output) the commands executed to run the stages
1562 of compilation. Also print the version number of the compiler driver
1563 program and of the preprocessor and the compiler proper.
1567 Like @option{-v} except the commands are not executed and arguments
1568 are quoted unless they contain only alphanumeric characters or @code{./-_}.
1569 This is useful for shell scripts to capture the driver-generated command lines.
1573 Print (on the standard output) a description of the command-line options
1574 understood by @command{gcc}. If the @option{-v} option is also specified
1575 then @option{--help} is also passed on to the various processes
1576 invoked by @command{gcc}, so that they can display the command-line options
1577 they accept. If the @option{-Wextra} option has also been specified
1578 (prior to the @option{--help} option), then command-line options that
1579 have no documentation associated with them are also displayed.
1582 @opindex target-help
1583 Print (on the standard output) a description of target-specific command-line
1584 options for each tool. For some targets extra target-specific
1585 information may also be printed.
1587 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1588 Print (on the standard output) a description of the command-line
1589 options understood by the compiler that fit into all specified classes
1590 and qualifiers. These are the supported classes:
1593 @item @samp{optimizers}
1594 Display all of the optimization options supported by the
1597 @item @samp{warnings}
1598 Display all of the options controlling warning messages
1599 produced by the compiler.
1602 Display target-specific options. Unlike the
1603 @option{--target-help} option however, target-specific options of the
1604 linker and assembler are not displayed. This is because those
1605 tools do not currently support the extended @option{--help=} syntax.
1608 Display the values recognized by the @option{--param}
1611 @item @var{language}
1612 Display the options supported for @var{language}, where
1613 @var{language} is the name of one of the languages supported in this
1617 Display the options that are common to all languages.
1620 These are the supported qualifiers:
1623 @item @samp{undocumented}
1624 Display only those options that are undocumented.
1627 Display options taking an argument that appears after an equal
1628 sign in the same continuous piece of text, such as:
1629 @samp{--help=target}.
1631 @item @samp{separate}
1632 Display options taking an argument that appears as a separate word
1633 following the original option, such as: @samp{-o output-file}.
1636 Thus for example to display all the undocumented target-specific
1637 switches supported by the compiler, use:
1640 --help=target,undocumented
1643 The sense of a qualifier can be inverted by prefixing it with the
1644 @samp{^} character, so for example to display all binary warning
1645 options (i.e., ones that are either on or off and that do not take an
1646 argument) that have a description, use:
1649 --help=warnings,^joined,^undocumented
1652 The argument to @option{--help=} should not consist solely of inverted
1655 Combining several classes is possible, although this usually
1656 restricts the output so much that there is nothing to display. One
1657 case where it does work, however, is when one of the classes is
1658 @var{target}. For example, to display all the target-specific
1659 optimization options, use:
1662 --help=target,optimizers
1665 The @option{--help=} option can be repeated on the command line. Each
1666 successive use displays its requested class of options, skipping
1667 those that have already been displayed. If @option{--help} is also
1668 specified anywhere on the command line then this takes precedence
1669 over any @option{--help=} option.
1671 If the @option{-Q} option appears on the command line before the
1672 @option{--help=} option, then the descriptive text displayed by
1673 @option{--help=} is changed. Instead of describing the displayed
1674 options, an indication is given as to whether the option is enabled,
1675 disabled or set to a specific value (assuming that the compiler
1676 knows this at the point where the @option{--help=} option is used).
1678 Here is a truncated example from the ARM port of @command{gcc}:
1681 % gcc -Q -mabi=2 --help=target -c
1682 The following options are target specific:
1684 -mabort-on-noreturn [disabled]
1688 The output is sensitive to the effects of previous command-line
1689 options, so for example it is possible to find out which optimizations
1690 are enabled at @option{-O2} by using:
1693 -Q -O2 --help=optimizers
1696 Alternatively you can discover which binary optimizations are enabled
1697 by @option{-O3} by using:
1700 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1701 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1702 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1707 Display the version number and copyrights of the invoked GCC@.
1709 @item -pass-exit-codes
1710 @opindex pass-exit-codes
1711 Normally the @command{gcc} program exits with the code of 1 if any
1712 phase of the compiler returns a non-success return code. If you specify
1713 @option{-pass-exit-codes}, the @command{gcc} program instead returns with
1714 the numerically highest error produced by any phase returning an error
1715 indication. The C, C++, and Fortran front ends return 4 if an internal
1716 compiler error is encountered.
1720 Use pipes rather than temporary files for communication between the
1721 various stages of compilation. This fails to work on some systems where
1722 the assembler is unable to read from a pipe; but the GNU assembler has
1725 @item -specs=@var{file}
1727 Process @var{file} after the compiler reads in the standard @file{specs}
1728 file, in order to override the defaults which the @command{gcc} driver
1729 program uses when determining what switches to pass to @command{cc1},
1730 @command{cc1plus}, @command{as}, @command{ld}, etc. More than one
1731 @option{-specs=@var{file}} can be specified on the command line, and they
1732 are processed in order, from left to right. @xref{Spec Files}, for
1733 information about the format of the @var{file}.
1737 Invoke all subcommands under a wrapper program. The name of the
1738 wrapper program and its parameters are passed as a comma separated
1742 gcc -c t.c -wrapper gdb,--args
1746 This invokes all subprograms of @command{gcc} under
1747 @samp{gdb --args}, thus the invocation of @command{cc1} is
1748 @samp{gdb --args cc1 @dots{}}.
1750 @item -ffile-prefix-map=@var{old}=@var{new}
1751 @opindex ffile-prefix-map
1752 When compiling files residing in directory @file{@var{old}}, record
1753 any references to them in the result of the compilation as if the
1754 files resided in directory @file{@var{new}} instead. Specifying this
1755 option is equivalent to specifying all the individual
1756 @option{-f*-prefix-map} options. This can be used to make reproducible
1757 builds that are location independent. See also
1758 @option{-fmacro-prefix-map} and @option{-fdebug-prefix-map}.
1760 @item -fplugin=@var{name}.so
1762 Load the plugin code in file @var{name}.so, assumed to be a
1763 shared object to be dlopen'd by the compiler. The base name of
1764 the shared object file is used to identify the plugin for the
1765 purposes of argument parsing (See
1766 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1767 Each plugin should define the callback functions specified in the
1770 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1771 @opindex fplugin-arg
1772 Define an argument called @var{key} with a value of @var{value}
1773 for the plugin called @var{name}.
1775 @item -fdump-ada-spec@r{[}-slim@r{]}
1776 @opindex fdump-ada-spec
1777 For C and C++ source and include files, generate corresponding Ada specs.
1778 @xref{Generating Ada Bindings for C and C++ headers,,, gnat_ugn,
1779 GNAT User's Guide}, which provides detailed documentation on this feature.
1781 @item -fada-spec-parent=@var{unit}
1782 @opindex fada-spec-parent
1783 In conjunction with @option{-fdump-ada-spec@r{[}-slim@r{]}} above, generate
1784 Ada specs as child units of parent @var{unit}.
1786 @item -fdump-go-spec=@var{file}
1787 @opindex fdump-go-spec
1788 For input files in any language, generate corresponding Go
1789 declarations in @var{file}. This generates Go @code{const},
1790 @code{type}, @code{var}, and @code{func} declarations which may be a
1791 useful way to start writing a Go interface to code written in some
1794 @include @value{srcdir}/../libiberty/at-file.texi
1798 @section Compiling C++ Programs
1800 @cindex suffixes for C++ source
1801 @cindex C++ source file suffixes
1802 C++ source files conventionally use one of the suffixes @samp{.C},
1803 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1804 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1805 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1806 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1807 files with these names and compiles them as C++ programs even if you
1808 call the compiler the same way as for compiling C programs (usually
1809 with the name @command{gcc}).
1813 However, the use of @command{gcc} does not add the C++ library.
1814 @command{g++} is a program that calls GCC and automatically specifies linking
1815 against the C++ library. It treats @samp{.c},
1816 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1817 files unless @option{-x} is used. This program is also useful when
1818 precompiling a C header file with a @samp{.h} extension for use in C++
1819 compilations. On many systems, @command{g++} is also installed with
1820 the name @command{c++}.
1822 @cindex invoking @command{g++}
1823 When you compile C++ programs, you may specify many of the same
1824 command-line options that you use for compiling programs in any
1825 language; or command-line options meaningful for C and related
1826 languages; or options that are meaningful only for C++ programs.
1827 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1828 explanations of options for languages related to C@.
1829 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1830 explanations of options that are meaningful only for C++ programs.
1832 @node C Dialect Options
1833 @section Options Controlling C Dialect
1834 @cindex dialect options
1835 @cindex language dialect options
1836 @cindex options, dialect
1838 The following options control the dialect of C (or languages derived
1839 from C, such as C++, Objective-C and Objective-C++) that the compiler
1843 @cindex ANSI support
1847 In C mode, this is equivalent to @option{-std=c90}. In C++ mode, it is
1848 equivalent to @option{-std=c++98}.
1850 This turns off certain features of GCC that are incompatible with ISO
1851 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1852 such as the @code{asm} and @code{typeof} keywords, and
1853 predefined macros such as @code{unix} and @code{vax} that identify the
1854 type of system you are using. It also enables the undesirable and
1855 rarely used ISO trigraph feature. For the C compiler,
1856 it disables recognition of C++ style @samp{//} comments as well as
1857 the @code{inline} keyword.
1859 The alternate keywords @code{__asm__}, @code{__extension__},
1860 @code{__inline__} and @code{__typeof__} continue to work despite
1861 @option{-ansi}. You would not want to use them in an ISO C program, of
1862 course, but it is useful to put them in header files that might be included
1863 in compilations done with @option{-ansi}. Alternate predefined macros
1864 such as @code{__unix__} and @code{__vax__} are also available, with or
1865 without @option{-ansi}.
1867 The @option{-ansi} option does not cause non-ISO programs to be
1868 rejected gratuitously. For that, @option{-Wpedantic} is required in
1869 addition to @option{-ansi}. @xref{Warning Options}.
1871 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1872 option is used. Some header files may notice this macro and refrain
1873 from declaring certain functions or defining certain macros that the
1874 ISO standard doesn't call for; this is to avoid interfering with any
1875 programs that might use these names for other things.
1877 Functions that are normally built in but do not have semantics
1878 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1879 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1880 built-in functions provided by GCC}, for details of the functions
1885 Determine the language standard. @xref{Standards,,Language Standards
1886 Supported by GCC}, for details of these standard versions. This option
1887 is currently only supported when compiling C or C++.
1889 The compiler can accept several base standards, such as @samp{c90} or
1890 @samp{c++98}, and GNU dialects of those standards, such as
1891 @samp{gnu90} or @samp{gnu++98}. When a base standard is specified, the
1892 compiler accepts all programs following that standard plus those
1893 using GNU extensions that do not contradict it. For example,
1894 @option{-std=c90} turns off certain features of GCC that are
1895 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1896 keywords, but not other GNU extensions that do not have a meaning in
1897 ISO C90, such as omitting the middle term of a @code{?:}
1898 expression. On the other hand, when a GNU dialect of a standard is
1899 specified, all features supported by the compiler are enabled, even when
1900 those features change the meaning of the base standard. As a result, some
1901 strict-conforming programs may be rejected. The particular standard
1902 is used by @option{-Wpedantic} to identify which features are GNU
1903 extensions given that version of the standard. For example
1904 @option{-std=gnu90 -Wpedantic} warns about C++ style @samp{//}
1905 comments, while @option{-std=gnu99 -Wpedantic} does not.
1907 A value for this option must be provided; possible values are
1913 Support all ISO C90 programs (certain GNU extensions that conflict
1914 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1916 @item iso9899:199409
1917 ISO C90 as modified in amendment 1.
1923 ISO C99. This standard is substantially completely supported, modulo
1924 bugs and floating-point issues
1925 (mainly but not entirely relating to optional C99 features from
1926 Annexes F and G). See
1927 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1928 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1933 ISO C11, the 2011 revision of the ISO C standard. This standard is
1934 substantially completely supported, modulo bugs, floating-point issues
1935 (mainly but not entirely relating to optional C11 features from
1936 Annexes F and G) and the optional Annexes K (Bounds-checking
1937 interfaces) and L (Analyzability). The name @samp{c1x} is deprecated.
1943 ISO C17, the 2017 revision of the ISO C standard
1944 (published in 2018). This standard is
1945 same as C11 except for corrections of defects (all of which are also
1946 applied with @option{-std=c11}) and a new value of
1947 @code{__STDC_VERSION__}, and so is supported to the same extent as C11.
1950 The next version of the ISO C standard, still under development. The
1951 support for this version is experimental and incomplete.
1955 GNU dialect of ISO C90 (including some C99 features).
1959 GNU dialect of ISO C99. The name @samp{gnu9x} is deprecated.
1963 GNU dialect of ISO C11.
1964 The name @samp{gnu1x} is deprecated.
1968 GNU dialect of ISO C17. This is the default for C code.
1971 The next version of the ISO C standard, still under development, plus
1972 GNU extensions. The support for this version is experimental and
1977 The 1998 ISO C++ standard plus the 2003 technical corrigendum and some
1978 additional defect reports. Same as @option{-ansi} for C++ code.
1982 GNU dialect of @option{-std=c++98}.
1986 The 2011 ISO C++ standard plus amendments.
1987 The name @samp{c++0x} is deprecated.
1991 GNU dialect of @option{-std=c++11}.
1992 The name @samp{gnu++0x} is deprecated.
1996 The 2014 ISO C++ standard plus amendments.
1997 The name @samp{c++1y} is deprecated.
2001 GNU dialect of @option{-std=c++14}.
2002 This is the default for C++ code.
2003 The name @samp{gnu++1y} is deprecated.
2007 The 2017 ISO C++ standard plus amendments.
2008 The name @samp{c++1z} is deprecated.
2012 GNU dialect of @option{-std=c++17}.
2013 The name @samp{gnu++1z} is deprecated.
2016 The next revision of the ISO C++ standard, tentatively planned for
2017 2020. Support is highly experimental, and will almost certainly
2018 change in incompatible ways in future releases.
2021 GNU dialect of @option{-std=c++2a}. Support is highly experimental,
2022 and will almost certainly change in incompatible ways in future
2026 @item -fgnu89-inline
2027 @opindex fgnu89-inline
2028 The option @option{-fgnu89-inline} tells GCC to use the traditional
2029 GNU semantics for @code{inline} functions when in C99 mode.
2030 @xref{Inline,,An Inline Function is As Fast As a Macro}.
2031 Using this option is roughly equivalent to adding the
2032 @code{gnu_inline} function attribute to all inline functions
2033 (@pxref{Function Attributes}).
2035 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
2036 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
2037 specifies the default behavior).
2038 This option is not supported in @option{-std=c90} or
2039 @option{-std=gnu90} mode.
2041 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
2042 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
2043 in effect for @code{inline} functions. @xref{Common Predefined
2044 Macros,,,cpp,The C Preprocessor}.
2046 @item -fpermitted-flt-eval-methods=@var{style}
2047 @opindex fpermitted-flt-eval-methods
2048 @opindex fpermitted-flt-eval-methods=c11
2049 @opindex fpermitted-flt-eval-methods=ts-18661-3
2050 ISO/IEC TS 18661-3 defines new permissible values for
2051 @code{FLT_EVAL_METHOD} that indicate that operations and constants with
2052 a semantic type that is an interchange or extended format should be
2053 evaluated to the precision and range of that type. These new values are
2054 a superset of those permitted under C99/C11, which does not specify the
2055 meaning of other positive values of @code{FLT_EVAL_METHOD}. As such, code
2056 conforming to C11 may not have been written expecting the possibility of
2059 @option{-fpermitted-flt-eval-methods} specifies whether the compiler
2060 should allow only the values of @code{FLT_EVAL_METHOD} specified in C99/C11,
2061 or the extended set of values specified in ISO/IEC TS 18661-3.
2063 @var{style} is either @code{c11} or @code{ts-18661-3} as appropriate.
2065 The default when in a standards compliant mode (@option{-std=c11} or similar)
2066 is @option{-fpermitted-flt-eval-methods=c11}. The default when in a GNU
2067 dialect (@option{-std=gnu11} or similar) is
2068 @option{-fpermitted-flt-eval-methods=ts-18661-3}.
2070 @item -aux-info @var{filename}
2072 Output to the given filename prototyped declarations for all functions
2073 declared and/or defined in a translation unit, including those in header
2074 files. This option is silently ignored in any language other than C@.
2076 Besides declarations, the file indicates, in comments, the origin of
2077 each declaration (source file and line), whether the declaration was
2078 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
2079 @samp{O} for old, respectively, in the first character after the line
2080 number and the colon), and whether it came from a declaration or a
2081 definition (@samp{C} or @samp{F}, respectively, in the following
2082 character). In the case of function definitions, a K&R-style list of
2083 arguments followed by their declarations is also provided, inside
2084 comments, after the declaration.
2086 @item -fallow-parameterless-variadic-functions
2087 @opindex fallow-parameterless-variadic-functions
2088 Accept variadic functions without named parameters.
2090 Although it is possible to define such a function, this is not very
2091 useful as it is not possible to read the arguments. This is only
2092 supported for C as this construct is allowed by C++.
2097 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
2098 keyword, so that code can use these words as identifiers. You can use
2099 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
2100 instead. @option{-ansi} implies @option{-fno-asm}.
2102 In C++, this switch only affects the @code{typeof} keyword, since
2103 @code{asm} and @code{inline} are standard keywords. You may want to
2104 use the @option{-fno-gnu-keywords} flag instead, which has the same
2105 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
2106 switch only affects the @code{asm} and @code{typeof} keywords, since
2107 @code{inline} is a standard keyword in ISO C99.
2110 @itemx -fno-builtin-@var{function}
2111 @opindex fno-builtin
2113 @cindex built-in functions
2114 Don't recognize built-in functions that do not begin with
2115 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
2116 functions provided by GCC}, for details of the functions affected,
2117 including those which are not built-in functions when @option{-ansi} or
2118 @option{-std} options for strict ISO C conformance are used because they
2119 do not have an ISO standard meaning.
2121 GCC normally generates special code to handle certain built-in functions
2122 more efficiently; for instance, calls to @code{alloca} may become single
2123 instructions which adjust the stack directly, and calls to @code{memcpy}
2124 may become inline copy loops. The resulting code is often both smaller
2125 and faster, but since the function calls no longer appear as such, you
2126 cannot set a breakpoint on those calls, nor can you change the behavior
2127 of the functions by linking with a different library. In addition,
2128 when a function is recognized as a built-in function, GCC may use
2129 information about that function to warn about problems with calls to
2130 that function, or to generate more efficient code, even if the
2131 resulting code still contains calls to that function. For example,
2132 warnings are given with @option{-Wformat} for bad calls to
2133 @code{printf} when @code{printf} is built in and @code{strlen} is
2134 known not to modify global memory.
2136 With the @option{-fno-builtin-@var{function}} option
2137 only the built-in function @var{function} is
2138 disabled. @var{function} must not begin with @samp{__builtin_}. If a
2139 function is named that is not built-in in this version of GCC, this
2140 option is ignored. There is no corresponding
2141 @option{-fbuiltin-@var{function}} option; if you wish to enable
2142 built-in functions selectively when using @option{-fno-builtin} or
2143 @option{-ffreestanding}, you may define macros such as:
2146 #define abs(n) __builtin_abs ((n))
2147 #define strcpy(d, s) __builtin_strcpy ((d), (s))
2153 Enable parsing of function definitions marked with @code{__GIMPLE}.
2154 This is an experimental feature that allows unit testing of GIMPLE
2159 @cindex hosted environment
2161 Assert that compilation targets a hosted environment. This implies
2162 @option{-fbuiltin}. A hosted environment is one in which the
2163 entire standard library is available, and in which @code{main} has a return
2164 type of @code{int}. Examples are nearly everything except a kernel.
2165 This is equivalent to @option{-fno-freestanding}.
2167 @item -ffreestanding
2168 @opindex ffreestanding
2169 @cindex hosted environment
2171 Assert that compilation targets a freestanding environment. This
2172 implies @option{-fno-builtin}. A freestanding environment
2173 is one in which the standard library may not exist, and program startup may
2174 not necessarily be at @code{main}. The most obvious example is an OS kernel.
2175 This is equivalent to @option{-fno-hosted}.
2177 @xref{Standards,,Language Standards Supported by GCC}, for details of
2178 freestanding and hosted environments.
2182 @cindex OpenACC accelerator programming
2183 Enable handling of OpenACC directives @code{#pragma acc} in C/C++ and
2184 @code{!$acc} in Fortran. When @option{-fopenacc} is specified, the
2185 compiler generates accelerated code according to the OpenACC Application
2186 Programming Interface v2.0 @w{@uref{https://www.openacc.org}}. This option
2187 implies @option{-pthread}, and thus is only supported on targets that
2188 have support for @option{-pthread}.
2190 @item -fopenacc-dim=@var{geom}
2191 @opindex fopenacc-dim
2192 @cindex OpenACC accelerator programming
2193 Specify default compute dimensions for parallel offload regions that do
2194 not explicitly specify. The @var{geom} value is a triple of
2195 ':'-separated sizes, in order 'gang', 'worker' and, 'vector'. A size
2196 can be omitted, to use a target-specific default value.
2200 @cindex OpenMP parallel
2201 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
2202 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
2203 compiler generates parallel code according to the OpenMP Application
2204 Program Interface v4.5 @w{@uref{https://www.openmp.org}}. This option
2205 implies @option{-pthread}, and thus is only supported on targets that
2206 have support for @option{-pthread}. @option{-fopenmp} implies
2207 @option{-fopenmp-simd}.
2210 @opindex fopenmp-simd
2213 Enable handling of OpenMP's SIMD directives with @code{#pragma omp}
2214 in C/C++ and @code{!$omp} in Fortran. Other OpenMP directives
2219 When the option @option{-fgnu-tm} is specified, the compiler
2220 generates code for the Linux variant of Intel's current Transactional
2221 Memory ABI specification document (Revision 1.1, May 6 2009). This is
2222 an experimental feature whose interface may change in future versions
2223 of GCC, as the official specification changes. Please note that not
2224 all architectures are supported for this feature.
2226 For more information on GCC's support for transactional memory,
2227 @xref{Enabling libitm,,The GNU Transactional Memory Library,libitm,GNU
2228 Transactional Memory Library}.
2230 Note that the transactional memory feature is not supported with
2231 non-call exceptions (@option{-fnon-call-exceptions}).
2233 @item -fms-extensions
2234 @opindex fms-extensions
2235 Accept some non-standard constructs used in Microsoft header files.
2237 In C++ code, this allows member names in structures to be similar
2238 to previous types declarations.
2247 Some cases of unnamed fields in structures and unions are only
2248 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
2249 fields within structs/unions}, for details.
2251 Note that this option is off for all targets except for x86
2252 targets using ms-abi.
2254 @item -fplan9-extensions
2255 @opindex fplan9-extensions
2256 Accept some non-standard constructs used in Plan 9 code.
2258 This enables @option{-fms-extensions}, permits passing pointers to
2259 structures with anonymous fields to functions that expect pointers to
2260 elements of the type of the field, and permits referring to anonymous
2261 fields declared using a typedef. @xref{Unnamed Fields,,Unnamed
2262 struct/union fields within structs/unions}, for details. This is only
2263 supported for C, not C++.
2265 @item -fcond-mismatch
2266 @opindex fcond-mismatch
2267 Allow conditional expressions with mismatched types in the second and
2268 third arguments. The value of such an expression is void. This option
2269 is not supported for C++.
2271 @item -flax-vector-conversions
2272 @opindex flax-vector-conversions
2273 Allow implicit conversions between vectors with differing numbers of
2274 elements and/or incompatible element types. This option should not be
2277 @item -funsigned-char
2278 @opindex funsigned-char
2279 Let the type @code{char} be unsigned, like @code{unsigned char}.
2281 Each kind of machine has a default for what @code{char} should
2282 be. It is either like @code{unsigned char} by default or like
2283 @code{signed char} by default.
2285 Ideally, a portable program should always use @code{signed char} or
2286 @code{unsigned char} when it depends on the signedness of an object.
2287 But many programs have been written to use plain @code{char} and
2288 expect it to be signed, or expect it to be unsigned, depending on the
2289 machines they were written for. This option, and its inverse, let you
2290 make such a program work with the opposite default.
2292 The type @code{char} is always a distinct type from each of
2293 @code{signed char} or @code{unsigned char}, even though its behavior
2294 is always just like one of those two.
2297 @opindex fsigned-char
2298 Let the type @code{char} be signed, like @code{signed char}.
2300 Note that this is equivalent to @option{-fno-unsigned-char}, which is
2301 the negative form of @option{-funsigned-char}. Likewise, the option
2302 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
2304 @item -fsigned-bitfields
2305 @itemx -funsigned-bitfields
2306 @itemx -fno-signed-bitfields
2307 @itemx -fno-unsigned-bitfields
2308 @opindex fsigned-bitfields
2309 @opindex funsigned-bitfields
2310 @opindex fno-signed-bitfields
2311 @opindex fno-unsigned-bitfields
2312 These options control whether a bit-field is signed or unsigned, when the
2313 declaration does not use either @code{signed} or @code{unsigned}. By
2314 default, such a bit-field is signed, because this is consistent: the
2315 basic integer types such as @code{int} are signed types.
2317 @item -fsso-struct=@var{endianness}
2318 @opindex fsso-struct
2319 Set the default scalar storage order of structures and unions to the
2320 specified endianness. The accepted values are @samp{big-endian},
2321 @samp{little-endian} and @samp{native} for the native endianness of
2322 the target (the default). This option is not supported for C++.
2324 @strong{Warning:} the @option{-fsso-struct} switch causes GCC to generate
2325 code that is not binary compatible with code generated without it if the
2326 specified endianness is not the native endianness of the target.
2329 @node C++ Dialect Options
2330 @section Options Controlling C++ Dialect
2332 @cindex compiler options, C++
2333 @cindex C++ options, command-line
2334 @cindex options, C++
2335 This section describes the command-line options that are only meaningful
2336 for C++ programs. You can also use most of the GNU compiler options
2337 regardless of what language your program is in. For example, you
2338 might compile a file @file{firstClass.C} like this:
2341 g++ -g -fstrict-enums -O -c firstClass.C
2345 In this example, only @option{-fstrict-enums} is an option meant
2346 only for C++ programs; you can use the other options with any
2347 language supported by GCC@.
2349 Some options for compiling C programs, such as @option{-std}, are also
2350 relevant for C++ programs.
2351 @xref{C Dialect Options,,Options Controlling C Dialect}.
2353 Here is a list of options that are @emph{only} for compiling C++ programs:
2357 @item -fabi-version=@var{n}
2358 @opindex fabi-version
2359 Use version @var{n} of the C++ ABI@. The default is version 0.
2361 Version 0 refers to the version conforming most closely to
2362 the C++ ABI specification. Therefore, the ABI obtained using version 0
2363 will change in different versions of G++ as ABI bugs are fixed.
2365 Version 1 is the version of the C++ ABI that first appeared in G++ 3.2.
2367 Version 2 is the version of the C++ ABI that first appeared in G++
2368 3.4, and was the default through G++ 4.9.
2370 Version 3 corrects an error in mangling a constant address as a
2373 Version 4, which first appeared in G++ 4.5, implements a standard
2374 mangling for vector types.
2376 Version 5, which first appeared in G++ 4.6, corrects the mangling of
2377 attribute const/volatile on function pointer types, decltype of a
2378 plain decl, and use of a function parameter in the declaration of
2381 Version 6, which first appeared in G++ 4.7, corrects the promotion
2382 behavior of C++11 scoped enums and the mangling of template argument
2383 packs, const/static_cast, prefix ++ and --, and a class scope function
2384 used as a template argument.
2386 Version 7, which first appeared in G++ 4.8, that treats nullptr_t as a
2387 builtin type and corrects the mangling of lambdas in default argument
2390 Version 8, which first appeared in G++ 4.9, corrects the substitution
2391 behavior of function types with function-cv-qualifiers.
2393 Version 9, which first appeared in G++ 5.2, corrects the alignment of
2396 Version 10, which first appeared in G++ 6.1, adds mangling of
2397 attributes that affect type identity, such as ia32 calling convention
2398 attributes (e.g.@: @samp{stdcall}).
2400 Version 11, which first appeared in G++ 7, corrects the mangling of
2401 sizeof... expressions and operator names. For multiple entities with
2402 the same name within a function, that are declared in different scopes,
2403 the mangling now changes starting with the twelfth occurrence. It also
2404 implies @option{-fnew-inheriting-ctors}.
2406 Version 12, which first appeared in G++ 8, corrects the calling
2407 conventions for empty classes on the x86_64 target and for classes
2408 with only deleted copy/move constructors. It accidentally changes the
2409 calling convention for classes with a deleted copy constructor and a
2410 trivial move constructor.
2412 Version 13, which first appeared in G++ 8.2, fixes the accidental
2413 change in version 12.
2415 See also @option{-Wabi}.
2417 @item -fabi-compat-version=@var{n}
2418 @opindex fabi-compat-version
2419 On targets that support strong aliases, G++
2420 works around mangling changes by creating an alias with the correct
2421 mangled name when defining a symbol with an incorrect mangled name.
2422 This switch specifies which ABI version to use for the alias.
2424 With @option{-fabi-version=0} (the default), this defaults to 11 (GCC 7
2425 compatibility). If another ABI version is explicitly selected, this
2426 defaults to 0. For compatibility with GCC versions 3.2 through 4.9,
2427 use @option{-fabi-compat-version=2}.
2429 If this option is not provided but @option{-Wabi=@var{n}} is, that
2430 version is used for compatibility aliases. If this option is provided
2431 along with @option{-Wabi} (without the version), the version from this
2432 option is used for the warning.
2434 @item -fno-access-control
2435 @opindex fno-access-control
2436 @opindex faccess-control
2437 Turn off all access checking. This switch is mainly useful for working
2438 around bugs in the access control code.
2441 @opindex faligned-new
2442 Enable support for C++17 @code{new} of types that require more
2443 alignment than @code{void* ::operator new(std::size_t)} provides. A
2444 numeric argument such as @code{-faligned-new=32} can be used to
2445 specify how much alignment (in bytes) is provided by that function,
2446 but few users will need to override the default of
2447 @code{alignof(std::max_align_t)}.
2449 This flag is enabled by default for @option{-std=c++17}.
2454 @opindex fno-char8_t
2455 Enable support for @code{char8_t} as adopted for C++2a. This includes
2456 the addition of a new @code{char8_t} fundamental type, changes to the
2457 types of UTF-8 string and character literals, new signatures for
2458 user-defined literals, associated standard library updates, and new
2459 @code{__cpp_char8_t} and @code{__cpp_lib_char8_t} feature test macros.
2461 This option enables functions to be overloaded for ordinary and UTF-8
2465 int f(const char *); // #1
2466 int f(const char8_t *); // #2
2467 int v1 = f("text"); // Calls #1
2468 int v2 = f(u8"text"); // Calls #2
2472 and introduces new signatures for user-defined literals:
2475 int operator""_udl1(char8_t);
2476 int v3 = u8'x'_udl1;
2477 int operator""_udl2(const char8_t*, std::size_t);
2478 int v4 = u8"text"_udl2;
2479 template<typename T, T...> int operator""_udl3();
2480 int v5 = u8"text"_udl3;
2484 The change to the types of UTF-8 string and character literals introduces
2485 incompatibilities with ISO C++11 and later standards. For example, the
2486 following code is well-formed under ISO C++11, but is ill-formed when
2487 @option{-fchar8_t} is specified.
2490 char ca[] = u8"xx"; // error: char-array initialized from wide
2492 const char *cp = u8"xx";// error: invalid conversion from
2493 // `const char8_t*' to `const char*'
2495 auto v = f(u8"xx"); // error: invalid conversion from
2496 // `const char8_t*' to `const char*'
2497 std::string s@{u8"xx"@}; // error: no matching function for call to
2498 // `std::basic_string<char>::basic_string()'
2499 using namespace std::literals;
2500 s = u8"xx"s; // error: conversion from
2501 // `basic_string<char8_t>' to non-scalar
2502 // type `basic_string<char>' requested
2507 Check that the pointer returned by @code{operator new} is non-null
2508 before attempting to modify the storage allocated. This check is
2509 normally unnecessary because the C++ standard specifies that
2510 @code{operator new} only returns @code{0} if it is declared
2511 @code{throw()}, in which case the compiler always checks the
2512 return value even without this option. In all other cases, when
2513 @code{operator new} has a non-empty exception specification, memory
2514 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
2515 @samp{new (nothrow)}.
2519 Enable support for the C++ Extensions for Concepts Technical
2520 Specification, ISO 19217 (2015), which allows code like
2523 template <class T> concept bool Addable = requires (T t) @{ t + t; @};
2524 template <Addable T> T add (T a, T b) @{ return a + b; @}
2527 @item -fconstexpr-depth=@var{n}
2528 @opindex fconstexpr-depth
2529 Set the maximum nested evaluation depth for C++11 constexpr functions
2530 to @var{n}. A limit is needed to detect endless recursion during
2531 constant expression evaluation. The minimum specified by the standard
2534 @item -fconstexpr-cache-depth=@var{n}
2535 @opindex fconstexpr-cache-depth
2536 Set the maximum level of nested evaluation depth for C++11 constexpr
2537 functions that will be cached to @var{n}. This is a heuristic that
2538 trades off compilation speed (when the cache avoids repeated
2539 calculations) against memory consumption (when the cache grows very
2540 large from highly recursive evaluations). The default is 8. Very few
2541 users are likely to want to adjust it, but if your code does heavy
2542 constexpr calculations you might want to experiment to find which
2543 value works best for you.
2545 @item -fconstexpr-loop-limit=@var{n}
2546 @opindex fconstexpr-loop-limit
2547 Set the maximum number of iterations for a loop in C++14 constexpr functions
2548 to @var{n}. A limit is needed to detect infinite loops during
2549 constant expression evaluation. The default is 262144 (1<<18).
2551 @item -fconstexpr-ops-limit=@var{n}
2552 @opindex fconstexpr-ops-limit
2553 Set the maximum number of operations during a single constexpr evaluation.
2554 Even when number of iterations of a single loop is limited with the above limit,
2555 if there are several nested loops and each of them has many iterations but still
2556 smaller than the above limit, or if in a body of some loop or even outside
2557 of a loop too many expressions need to be evaluated, the resulting constexpr
2558 evaluation might take too long.
2559 The default is 33554432 (1<<25).
2561 @item -fdeduce-init-list
2562 @opindex fdeduce-init-list
2563 Enable deduction of a template type parameter as
2564 @code{std::initializer_list} from a brace-enclosed initializer list, i.e.@:
2567 template <class T> auto forward(T t) -> decltype (realfn (t))
2574 forward(@{1,2@}); // call forward<std::initializer_list<int>>
2578 This deduction was implemented as a possible extension to the
2579 originally proposed semantics for the C++11 standard, but was not part
2580 of the final standard, so it is disabled by default. This option is
2581 deprecated, and may be removed in a future version of G++.
2583 @item -fno-elide-constructors
2584 @opindex fno-elide-constructors
2585 @opindex felide-constructors
2586 The C++ standard allows an implementation to omit creating a temporary
2587 that is only used to initialize another object of the same type.
2588 Specifying this option disables that optimization, and forces G++ to
2589 call the copy constructor in all cases. This option also causes G++
2590 to call trivial member functions which otherwise would be expanded inline.
2592 In C++17, the compiler is required to omit these temporaries, but this
2593 option still affects trivial member functions.
2595 @item -fno-enforce-eh-specs
2596 @opindex fno-enforce-eh-specs
2597 @opindex fenforce-eh-specs
2598 Don't generate code to check for violation of exception specifications
2599 at run time. This option violates the C++ standard, but may be useful
2600 for reducing code size in production builds, much like defining
2601 @code{NDEBUG}. This does not give user code permission to throw
2602 exceptions in violation of the exception specifications; the compiler
2603 still optimizes based on the specifications, so throwing an
2604 unexpected exception results in undefined behavior at run time.
2606 @item -fextern-tls-init
2607 @itemx -fno-extern-tls-init
2608 @opindex fextern-tls-init
2609 @opindex fno-extern-tls-init
2610 The C++11 and OpenMP standards allow @code{thread_local} and
2611 @code{threadprivate} variables to have dynamic (runtime)
2612 initialization. To support this, any use of such a variable goes
2613 through a wrapper function that performs any necessary initialization.
2614 When the use and definition of the variable are in the same
2615 translation unit, this overhead can be optimized away, but when the
2616 use is in a different translation unit there is significant overhead
2617 even if the variable doesn't actually need dynamic initialization. If
2618 the programmer can be sure that no use of the variable in a
2619 non-defining TU needs to trigger dynamic initialization (either
2620 because the variable is statically initialized, or a use of the
2621 variable in the defining TU will be executed before any uses in
2622 another TU), they can avoid this overhead with the
2623 @option{-fno-extern-tls-init} option.
2625 On targets that support symbol aliases, the default is
2626 @option{-fextern-tls-init}. On targets that do not support symbol
2627 aliases, the default is @option{-fno-extern-tls-init}.
2629 @item -fno-gnu-keywords
2630 @opindex fno-gnu-keywords
2631 @opindex fgnu-keywords
2632 Do not recognize @code{typeof} as a keyword, so that code can use this
2633 word as an identifier. You can use the keyword @code{__typeof__} instead.
2634 This option is implied by the strict ISO C++ dialects: @option{-ansi},
2635 @option{-std=c++98}, @option{-std=c++11}, etc.
2637 @item -fno-implicit-templates
2638 @opindex fno-implicit-templates
2639 @opindex fimplicit-templates
2640 Never emit code for non-inline templates that are instantiated
2641 implicitly (i.e.@: by use); only emit code for explicit instantiations.
2642 If you use this option, you must take care to structure your code to
2643 include all the necessary explicit instantiations to avoid getting
2644 undefined symbols at link time.
2645 @xref{Template Instantiation}, for more information.
2647 @item -fno-implicit-inline-templates
2648 @opindex fno-implicit-inline-templates
2649 @opindex fimplicit-inline-templates
2650 Don't emit code for implicit instantiations of inline templates, either.
2651 The default is to handle inlines differently so that compiles with and
2652 without optimization need the same set of explicit instantiations.
2654 @item -fno-implement-inlines
2655 @opindex fno-implement-inlines
2656 @opindex fimplement-inlines
2657 To save space, do not emit out-of-line copies of inline functions
2658 controlled by @code{#pragma implementation}. This causes linker
2659 errors if these functions are not inlined everywhere they are called.
2661 @item -fms-extensions
2662 @opindex fms-extensions
2663 Disable Wpedantic warnings about constructs used in MFC, such as implicit
2664 int and getting a pointer to member function via non-standard syntax.
2666 @item -fnew-inheriting-ctors
2667 @opindex fnew-inheriting-ctors
2668 Enable the P0136 adjustment to the semantics of C++11 constructor
2669 inheritance. This is part of C++17 but also considered to be a Defect
2670 Report against C++11 and C++14. This flag is enabled by default
2671 unless @option{-fabi-version=10} or lower is specified.
2673 @item -fnew-ttp-matching
2674 @opindex fnew-ttp-matching
2675 Enable the P0522 resolution to Core issue 150, template template
2676 parameters and default arguments: this allows a template with default
2677 template arguments as an argument for a template template parameter
2678 with fewer template parameters. This flag is enabled by default for
2679 @option{-std=c++17}.
2681 @item -fno-nonansi-builtins
2682 @opindex fno-nonansi-builtins
2683 @opindex fnonansi-builtins
2684 Disable built-in declarations of functions that are not mandated by
2685 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
2686 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
2689 @opindex fnothrow-opt
2690 Treat a @code{throw()} exception specification as if it were a
2691 @code{noexcept} specification to reduce or eliminate the text size
2692 overhead relative to a function with no exception specification. If
2693 the function has local variables of types with non-trivial
2694 destructors, the exception specification actually makes the
2695 function smaller because the EH cleanups for those variables can be
2696 optimized away. The semantic effect is that an exception thrown out of
2697 a function with such an exception specification results in a call
2698 to @code{terminate} rather than @code{unexpected}.
2700 @item -fno-operator-names
2701 @opindex fno-operator-names
2702 @opindex foperator-names
2703 Do not treat the operator name keywords @code{and}, @code{bitand},
2704 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
2705 synonyms as keywords.
2707 @item -fno-optional-diags
2708 @opindex fno-optional-diags
2709 @opindex foptional-diags
2710 Disable diagnostics that the standard says a compiler does not need to
2711 issue. Currently, the only such diagnostic issued by G++ is the one for
2712 a name having multiple meanings within a class.
2715 @opindex fpermissive
2716 Downgrade some diagnostics about nonconformant code from errors to
2717 warnings. Thus, using @option{-fpermissive} allows some
2718 nonconforming code to compile.
2720 @item -fno-pretty-templates
2721 @opindex fno-pretty-templates
2722 @opindex fpretty-templates
2723 When an error message refers to a specialization of a function
2724 template, the compiler normally prints the signature of the
2725 template followed by the template arguments and any typedefs or
2726 typenames in the signature (e.g.@: @code{void f(T) [with T = int]}
2727 rather than @code{void f(int)}) so that it's clear which template is
2728 involved. When an error message refers to a specialization of a class
2729 template, the compiler omits any template arguments that match
2730 the default template arguments for that template. If either of these
2731 behaviors make it harder to understand the error message rather than
2732 easier, you can use @option{-fno-pretty-templates} to disable them.
2736 Enable automatic template instantiation at link time. This option also
2737 implies @option{-fno-implicit-templates}. @xref{Template
2738 Instantiation}, for more information.
2743 Disable generation of information about every class with virtual
2744 functions for use by the C++ run-time type identification features
2745 (@code{dynamic_cast} and @code{typeid}). If you don't use those parts
2746 of the language, you can save some space by using this flag. Note that
2747 exception handling uses the same information, but G++ generates it as
2748 needed. The @code{dynamic_cast} operator can still be used for casts that
2749 do not require run-time type information, i.e.@: casts to @code{void *} or to
2750 unambiguous base classes.
2752 Mixing code compiled with @option{-frtti} with that compiled with
2753 @option{-fno-rtti} may not work. For example, programs may
2754 fail to link if a class compiled with @option{-fno-rtti} is used as a base
2755 for a class compiled with @option{-frtti}.
2757 @item -fsized-deallocation
2758 @opindex fsized-deallocation
2759 Enable the built-in global declarations
2761 void operator delete (void *, std::size_t) noexcept;
2762 void operator delete[] (void *, std::size_t) noexcept;
2764 as introduced in C++14. This is useful for user-defined replacement
2765 deallocation functions that, for example, use the size of the object
2766 to make deallocation faster. Enabled by default under
2767 @option{-std=c++14} and above. The flag @option{-Wsized-deallocation}
2768 warns about places that might want to add a definition.
2770 @item -fstrict-enums
2771 @opindex fstrict-enums
2772 Allow the compiler to optimize using the assumption that a value of
2773 enumerated type can only be one of the values of the enumeration (as
2774 defined in the C++ standard; basically, a value that can be
2775 represented in the minimum number of bits needed to represent all the
2776 enumerators). This assumption may not be valid if the program uses a
2777 cast to convert an arbitrary integer value to the enumerated type.
2779 @item -fstrong-eval-order
2780 @opindex fstrong-eval-order
2781 Evaluate member access, array subscripting, and shift expressions in
2782 left-to-right order, and evaluate assignment in right-to-left order,
2783 as adopted for C++17. Enabled by default with @option{-std=c++17}.
2784 @option{-fstrong-eval-order=some} enables just the ordering of member
2785 access and shift expressions, and is the default without
2786 @option{-std=c++17}.
2788 @item -ftemplate-backtrace-limit=@var{n}
2789 @opindex ftemplate-backtrace-limit
2790 Set the maximum number of template instantiation notes for a single
2791 warning or error to @var{n}. The default value is 10.
2793 @item -ftemplate-depth=@var{n}
2794 @opindex ftemplate-depth
2795 Set the maximum instantiation depth for template classes to @var{n}.
2796 A limit on the template instantiation depth is needed to detect
2797 endless recursions during template class instantiation. ANSI/ISO C++
2798 conforming programs must not rely on a maximum depth greater than 17
2799 (changed to 1024 in C++11). The default value is 900, as the compiler
2800 can run out of stack space before hitting 1024 in some situations.
2802 @item -fno-threadsafe-statics
2803 @opindex fno-threadsafe-statics
2804 @opindex fthreadsafe-statics
2805 Do not emit the extra code to use the routines specified in the C++
2806 ABI for thread-safe initialization of local statics. You can use this
2807 option to reduce code size slightly in code that doesn't need to be
2810 @item -fuse-cxa-atexit
2811 @opindex fuse-cxa-atexit
2812 Register destructors for objects with static storage duration with the
2813 @code{__cxa_atexit} function rather than the @code{atexit} function.
2814 This option is required for fully standards-compliant handling of static
2815 destructors, but only works if your C library supports
2816 @code{__cxa_atexit}.
2818 @item -fno-use-cxa-get-exception-ptr
2819 @opindex fno-use-cxa-get-exception-ptr
2820 @opindex fuse-cxa-get-exception-ptr
2821 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2822 causes @code{std::uncaught_exception} to be incorrect, but is necessary
2823 if the runtime routine is not available.
2825 @item -fvisibility-inlines-hidden
2826 @opindex fvisibility-inlines-hidden
2827 This switch declares that the user does not attempt to compare
2828 pointers to inline functions or methods where the addresses of the two functions
2829 are taken in different shared objects.
2831 The effect of this is that GCC may, effectively, mark inline methods with
2832 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2833 appear in the export table of a DSO and do not require a PLT indirection
2834 when used within the DSO@. Enabling this option can have a dramatic effect
2835 on load and link times of a DSO as it massively reduces the size of the
2836 dynamic export table when the library makes heavy use of templates.
2838 The behavior of this switch is not quite the same as marking the
2839 methods as hidden directly, because it does not affect static variables
2840 local to the function or cause the compiler to deduce that
2841 the function is defined in only one shared object.
2843 You may mark a method as having a visibility explicitly to negate the
2844 effect of the switch for that method. For example, if you do want to
2845 compare pointers to a particular inline method, you might mark it as
2846 having default visibility. Marking the enclosing class with explicit
2847 visibility has no effect.
2849 Explicitly instantiated inline methods are unaffected by this option
2850 as their linkage might otherwise cross a shared library boundary.
2851 @xref{Template Instantiation}.
2853 @item -fvisibility-ms-compat
2854 @opindex fvisibility-ms-compat
2855 This flag attempts to use visibility settings to make GCC's C++
2856 linkage model compatible with that of Microsoft Visual Studio.
2858 The flag makes these changes to GCC's linkage model:
2862 It sets the default visibility to @code{hidden}, like
2863 @option{-fvisibility=hidden}.
2866 Types, but not their members, are not hidden by default.
2869 The One Definition Rule is relaxed for types without explicit
2870 visibility specifications that are defined in more than one
2871 shared object: those declarations are permitted if they are
2872 permitted when this option is not used.
2875 In new code it is better to use @option{-fvisibility=hidden} and
2876 export those classes that are intended to be externally visible.
2877 Unfortunately it is possible for code to rely, perhaps accidentally,
2878 on the Visual Studio behavior.
2880 Among the consequences of these changes are that static data members
2881 of the same type with the same name but defined in different shared
2882 objects are different, so changing one does not change the other;
2883 and that pointers to function members defined in different shared
2884 objects may not compare equal. When this flag is given, it is a
2885 violation of the ODR to define types with the same name differently.
2890 Do not use weak symbol support, even if it is provided by the linker.
2891 By default, G++ uses weak symbols if they are available. This
2892 option exists only for testing, and should not be used by end-users;
2893 it results in inferior code and has no benefits. This option may
2894 be removed in a future release of G++.
2898 Do not search for header files in the standard directories specific to
2899 C++, but do still search the other standard directories. (This option
2900 is used when building the C++ library.)
2903 In addition, these optimization, warning, and code generation options
2904 have meanings only for C++ programs:
2907 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2910 Warn when G++ it generates code that is probably not compatible with
2911 the vendor-neutral C++ ABI@. Since G++ now defaults to updating the
2912 ABI with each major release, normally @option{-Wabi} will warn only if
2913 there is a check added later in a release series for an ABI issue
2914 discovered since the initial release. @option{-Wabi} will warn about
2915 more things if an older ABI version is selected (with
2916 @option{-fabi-version=@var{n}}).
2918 @option{-Wabi} can also be used with an explicit version number to
2919 warn about compatibility with a particular @option{-fabi-version}
2920 level, e.g.@: @option{-Wabi=2} to warn about changes relative to
2921 @option{-fabi-version=2}.
2923 If an explicit version number is provided and
2924 @option{-fabi-compat-version} is not specified, the version number
2925 from this option is used for compatibility aliases. If no explicit
2926 version number is provided with this option, but
2927 @option{-fabi-compat-version} is specified, that version number is
2928 used for ABI warnings.
2930 Although an effort has been made to warn about
2931 all such cases, there are probably some cases that are not warned about,
2932 even though G++ is generating incompatible code. There may also be
2933 cases where warnings are emitted even though the code that is generated
2936 You should rewrite your code to avoid these warnings if you are
2937 concerned about the fact that code generated by G++ may not be binary
2938 compatible with code generated by other compilers.
2940 Known incompatibilities in @option{-fabi-version=2} (which was the
2941 default from GCC 3.4 to 4.9) include:
2946 A template with a non-type template parameter of reference type was
2947 mangled incorrectly:
2950 template <int &> struct S @{@};
2954 This was fixed in @option{-fabi-version=3}.
2957 SIMD vector types declared using @code{__attribute ((vector_size))} were
2958 mangled in a non-standard way that does not allow for overloading of
2959 functions taking vectors of different sizes.
2961 The mangling was changed in @option{-fabi-version=4}.
2964 @code{__attribute ((const))} and @code{noreturn} were mangled as type
2965 qualifiers, and @code{decltype} of a plain declaration was folded away.
2967 These mangling issues were fixed in @option{-fabi-version=5}.
2970 Scoped enumerators passed as arguments to a variadic function are
2971 promoted like unscoped enumerators, causing @code{va_arg} to complain.
2972 On most targets this does not actually affect the parameter passing
2973 ABI, as there is no way to pass an argument smaller than @code{int}.
2975 Also, the ABI changed the mangling of template argument packs,
2976 @code{const_cast}, @code{static_cast}, prefix increment/decrement, and
2977 a class scope function used as a template argument.
2979 These issues were corrected in @option{-fabi-version=6}.
2982 Lambdas in default argument scope were mangled incorrectly, and the
2983 ABI changed the mangling of @code{nullptr_t}.
2985 These issues were corrected in @option{-fabi-version=7}.
2988 When mangling a function type with function-cv-qualifiers, the
2989 un-qualified function type was incorrectly treated as a substitution
2992 This was fixed in @option{-fabi-version=8}, the default for GCC 5.1.
2995 @code{decltype(nullptr)} incorrectly had an alignment of 1, leading to
2996 unaligned accesses. Note that this did not affect the ABI of a
2997 function with a @code{nullptr_t} parameter, as parameters have a
3000 This was fixed in @option{-fabi-version=9}, the default for GCC 5.2.
3003 Target-specific attributes that affect the identity of a type, such as
3004 ia32 calling conventions on a function type (stdcall, regparm, etc.),
3005 did not affect the mangled name, leading to name collisions when
3006 function pointers were used as template arguments.
3008 This was fixed in @option{-fabi-version=10}, the default for GCC 6.1.
3012 It also warns about psABI-related changes. The known psABI changes at this
3018 For SysV/x86-64, unions with @code{long double} members are
3019 passed in memory as specified in psABI. For example:
3029 @code{union U} is always passed in memory.
3033 @item -Wabi-tag @r{(C++ and Objective-C++ only)}
3036 Warn when a type with an ABI tag is used in a context that does not
3037 have that ABI tag. See @ref{C++ Attributes} for more information
3040 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
3041 @opindex Wctor-dtor-privacy
3042 @opindex Wno-ctor-dtor-privacy
3043 Warn when a class seems unusable because all the constructors or
3044 destructors in that class are private, and it has neither friends nor
3045 public static member functions. Also warn if there are no non-private
3046 methods, and there's at least one private member function that isn't
3047 a constructor or destructor.
3049 @item -Wdelete-non-virtual-dtor @r{(C++ and Objective-C++ only)}
3050 @opindex Wdelete-non-virtual-dtor
3051 @opindex Wno-delete-non-virtual-dtor
3052 Warn when @code{delete} is used to destroy an instance of a class that
3053 has virtual functions and non-virtual destructor. It is unsafe to delete
3054 an instance of a derived class through a pointer to a base class if the
3055 base class does not have a virtual destructor. This warning is enabled
3058 @item -Wdeprecated-copy @r{(C++ and Objective-C++ only)}
3059 @opindex Wdeprecated-copy
3060 @opindex Wno-deprecated-copy
3061 Warn that the implicit declaration of a copy constructor or copy
3062 assignment operator is deprecated if the class has a user-provided
3063 copy constructor or copy assignment operator, in C++11 and up. This
3064 warning is enabled by @option{-Wextra}. With
3065 @option{-Wdeprecated-copy-dtor}, also deprecate if the class has a
3066 user-provided destructor.
3068 @item -Wno-init-list-lifetime @r{(C++ and Objective-C++ only)}
3069 @opindex Winit-list-lifetime
3070 @opindex Wno-init-list-lifetime
3071 Do not warn about uses of @code{std::initializer_list} that are likely
3072 to result in dangling pointers. Since the underlying array for an
3073 @code{initializer_list} is handled like a normal C++ temporary object,
3074 it is easy to inadvertently keep a pointer to the array past the end
3075 of the array's lifetime. For example:
3079 If a function returns a temporary @code{initializer_list}, or a local
3080 @code{initializer_list} variable, the array's lifetime ends at the end
3081 of the return statement, so the value returned has a dangling pointer.
3084 If a new-expression creates an @code{initializer_list}, the array only
3085 lives until the end of the enclosing full-expression, so the
3086 @code{initializer_list} in the heap has a dangling pointer.
3089 When an @code{initializer_list} variable is assigned from a
3090 brace-enclosed initializer list, the temporary array created for the
3091 right side of the assignment only lives until the end of the
3092 full-expression, so at the next statement the @code{initializer_list}
3093 variable has a dangling pointer.
3096 // li's initial underlying array lives as long as li
3097 std::initializer_list<int> li = @{ 1,2,3 @};
3098 // assignment changes li to point to a temporary array
3100 // now the temporary is gone and li has a dangling pointer
3101 int i = li.begin()[0] // undefined behavior
3105 When a list constructor stores the @code{begin} pointer from the
3106 @code{initializer_list} argument, this doesn't extend the lifetime of
3107 the array, so if a class variable is constructed from a temporary
3108 @code{initializer_list}, the pointer is left dangling by the end of
3109 the variable declaration statement.
3113 @item -Wliteral-suffix @r{(C++ and Objective-C++ only)}
3114 @opindex Wliteral-suffix
3115 @opindex Wno-literal-suffix
3116 Warn when a string or character literal is followed by a ud-suffix which does
3117 not begin with an underscore. As a conforming extension, GCC treats such
3118 suffixes as separate preprocessing tokens in order to maintain backwards
3119 compatibility with code that uses formatting macros from @code{<inttypes.h>}.
3123 #define __STDC_FORMAT_MACROS
3124 #include <inttypes.h>
3129 printf("My int64: %" PRId64"\n", i64);
3133 In this case, @code{PRId64} is treated as a separate preprocessing token.
3135 Additionally, warn when a user-defined literal operator is declared with
3136 a literal suffix identifier that doesn't begin with an underscore. Literal
3137 suffix identifiers that don't begin with an underscore are reserved for
3138 future standardization.
3140 This warning is enabled by default.
3142 @item -Wlto-type-mismatch
3143 @opindex Wlto-type-mismatch
3144 @opindex Wno-lto-type-mismatch
3146 During the link-time optimization warn about type mismatches in
3147 global declarations from different compilation units.
3148 Requires @option{-flto} to be enabled. Enabled by default.
3150 @item -Wno-narrowing @r{(C++ and Objective-C++ only)}
3152 @opindex Wno-narrowing
3153 For C++11 and later standards, narrowing conversions are diagnosed by default,
3154 as required by the standard. A narrowing conversion from a constant produces
3155 an error, and a narrowing conversion from a non-constant produces a warning,
3156 but @option{-Wno-narrowing} suppresses the diagnostic.
3157 Note that this does not affect the meaning of well-formed code;
3158 narrowing conversions are still considered ill-formed in SFINAE contexts.
3160 With @option{-Wnarrowing} in C++98, warn when a narrowing
3161 conversion prohibited by C++11 occurs within
3165 int i = @{ 2.2 @}; // error: narrowing from double to int
3168 This flag is included in @option{-Wall} and @option{-Wc++11-compat}.
3170 @item -Wnoexcept @r{(C++ and Objective-C++ only)}
3172 @opindex Wno-noexcept
3173 Warn when a noexcept-expression evaluates to false because of a call
3174 to a function that does not have a non-throwing exception
3175 specification (i.e. @code{throw()} or @code{noexcept}) but is known by
3176 the compiler to never throw an exception.
3178 @item -Wnoexcept-type @r{(C++ and Objective-C++ only)}
3179 @opindex Wnoexcept-type
3180 @opindex Wno-noexcept-type
3181 Warn if the C++17 feature making @code{noexcept} part of a function
3182 type changes the mangled name of a symbol relative to C++14. Enabled
3183 by @option{-Wabi} and @option{-Wc++17-compat}.
3188 template <class T> void f(T t) @{ t(); @};
3190 void h() @{ f(g); @}
3194 In C++14, @code{f} calls @code{f<void(*)()>}, but in
3195 C++17 it calls @code{f<void(*)()noexcept>}.
3197 @item -Wclass-memaccess @r{(C++ and Objective-C++ only)}
3198 @opindex Wclass-memaccess
3199 @opindex Wno-class-memaccess
3200 Warn when the destination of a call to a raw memory function such as
3201 @code{memset} or @code{memcpy} is an object of class type, and when writing
3202 into such an object might bypass the class non-trivial or deleted constructor
3203 or copy assignment, violate const-correctness or encapsulation, or corrupt
3204 virtual table pointers. Modifying the representation of such objects may
3205 violate invariants maintained by member functions of the class. For example,
3206 the call to @code{memset} below is undefined because it modifies a non-trivial
3207 class object and is, therefore, diagnosed. The safe way to either initialize
3208 or clear the storage of objects of such types is by using the appropriate
3209 constructor or assignment operator, if one is available.
3211 std::string str = "abc";
3212 memset (&str, 0, sizeof str);
3214 The @option{-Wclass-memaccess} option is enabled by @option{-Wall}.
3215 Explicitly casting the pointer to the class object to @code{void *} or
3216 to a type that can be safely accessed by the raw memory function suppresses
3219 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
3220 @opindex Wnon-virtual-dtor
3221 @opindex Wno-non-virtual-dtor
3222 Warn when a class has virtual functions and an accessible non-virtual
3223 destructor itself or in an accessible polymorphic base class, in which
3224 case it is possible but unsafe to delete an instance of a derived
3225 class through a pointer to the class itself or base class. This
3226 warning is automatically enabled if @option{-Weffc++} is specified.
3228 @item -Wregister @r{(C++ and Objective-C++ only)}
3230 @opindex Wno-register
3231 Warn on uses of the @code{register} storage class specifier, except
3232 when it is part of the GNU @ref{Explicit Register Variables} extension.
3233 The use of the @code{register} keyword as storage class specifier has
3234 been deprecated in C++11 and removed in C++17.
3235 Enabled by default with @option{-std=c++17}.
3237 @item -Wreorder @r{(C++ and Objective-C++ only)}
3239 @opindex Wno-reorder
3240 @cindex reordering, warning
3241 @cindex warning for reordering of member initializers
3242 Warn when the order of member initializers given in the code does not
3243 match the order in which they must be executed. For instance:
3249 A(): j (0), i (1) @{ @}
3254 The compiler rearranges the member initializers for @code{i}
3255 and @code{j} to match the declaration order of the members, emitting
3256 a warning to that effect. This warning is enabled by @option{-Wall}.
3258 @item -Wno-pessimizing-move @r{(C++ and Objective-C++ only)}
3259 @opindex Wpessimizing-move
3260 @opindex Wno-pessimizing-move
3261 This warning warns when a call to @code{std::move} prevents copy
3262 elision. A typical scenario when copy elision can occur is when returning in
3263 a function with a class return type, when the expression being returned is the
3264 name of a non-volatile automatic object, and is not a function parameter, and
3265 has the same type as the function return type.
3275 return std::move (t);
3279 But in this example, the @code{std::move} call prevents copy elision.
3281 This warning is enabled by @option{-Wall}.
3283 @item -Wno-redundant-move @r{(C++ and Objective-C++ only)}
3284 @opindex Wredundant-move
3285 @opindex Wno-redundant-move
3286 This warning warns about redundant calls to @code{std::move}; that is, when
3287 a move operation would have been performed even without the @code{std::move}
3288 call. This happens because the compiler is forced to treat the object as if
3289 it were an rvalue in certain situations such as returning a local variable,
3290 where copy elision isn't applicable. Consider:
3299 return std::move (t);
3303 Here, the @code{std::move} call is redundant. Because G++ implements Core
3304 Issue 1579, another example is:
3307 struct T @{ // convertible to U
3317 return std::move (t);
3320 In this example, copy elision isn't applicable because the type of the
3321 expression being returned and the function return type differ, yet G++
3322 treats the return value as if it were designated by an rvalue.
3324 This warning is enabled by @option{-Wextra}.
3326 @item -fext-numeric-literals @r{(C++ and Objective-C++ only)}
3327 @opindex fext-numeric-literals
3328 @opindex fno-ext-numeric-literals
3329 Accept imaginary, fixed-point, or machine-defined
3330 literal number suffixes as GNU extensions.
3331 When this option is turned off these suffixes are treated
3332 as C++11 user-defined literal numeric suffixes.
3333 This is on by default for all pre-C++11 dialects and all GNU dialects:
3334 @option{-std=c++98}, @option{-std=gnu++98}, @option{-std=gnu++11},
3335 @option{-std=gnu++14}.
3336 This option is off by default
3337 for ISO C++11 onwards (@option{-std=c++11}, ...).
3340 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
3343 @item -Weffc++ @r{(C++ and Objective-C++ only)}
3346 Warn about violations of the following style guidelines from Scott Meyers'
3347 @cite{Effective C++} series of books:
3351 Define a copy constructor and an assignment operator for classes
3352 with dynamically-allocated memory.
3355 Prefer initialization to assignment in constructors.
3358 Have @code{operator=} return a reference to @code{*this}.
3361 Don't try to return a reference when you must return an object.
3364 Distinguish between prefix and postfix forms of increment and
3365 decrement operators.
3368 Never overload @code{&&}, @code{||}, or @code{,}.
3372 This option also enables @option{-Wnon-virtual-dtor}, which is also
3373 one of the effective C++ recommendations. However, the check is
3374 extended to warn about the lack of virtual destructor in accessible
3375 non-polymorphic bases classes too.
3377 When selecting this option, be aware that the standard library
3378 headers do not obey all of these guidelines; use @samp{grep -v}
3379 to filter out those warnings.
3381 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
3382 @opindex Wstrict-null-sentinel
3383 @opindex Wno-strict-null-sentinel
3384 Warn about the use of an uncasted @code{NULL} as sentinel. When
3385 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
3386 to @code{__null}. Although it is a null pointer constant rather than a
3387 null pointer, it is guaranteed to be of the same size as a pointer.
3388 But this use is not portable across different compilers.
3390 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
3391 @opindex Wno-non-template-friend
3392 @opindex Wnon-template-friend
3393 Disable warnings when non-template friend functions are declared
3394 within a template. In very old versions of GCC that predate implementation
3395 of the ISO standard, declarations such as
3396 @samp{friend int foo(int)}, where the name of the friend is an unqualified-id,
3397 could be interpreted as a particular specialization of a template
3398 function; the warning exists to diagnose compatibility problems,
3399 and is enabled by default.
3401 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
3402 @opindex Wold-style-cast
3403 @opindex Wno-old-style-cast
3404 Warn if an old-style (C-style) cast to a non-void type is used within
3405 a C++ program. The new-style casts (@code{dynamic_cast},
3406 @code{static_cast}, @code{reinterpret_cast}, and @code{const_cast}) are
3407 less vulnerable to unintended effects and much easier to search for.
3409 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
3410 @opindex Woverloaded-virtual
3411 @opindex Wno-overloaded-virtual
3412 @cindex overloaded virtual function, warning
3413 @cindex warning for overloaded virtual function
3414 Warn when a function declaration hides virtual functions from a
3415 base class. For example, in:
3422 struct B: public A @{
3427 the @code{A} class version of @code{f} is hidden in @code{B}, and code
3438 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
3439 @opindex Wno-pmf-conversions
3440 @opindex Wpmf-conversions
3441 Disable the diagnostic for converting a bound pointer to member function
3444 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
3445 @opindex Wsign-promo
3446 @opindex Wno-sign-promo
3447 Warn when overload resolution chooses a promotion from unsigned or
3448 enumerated type to a signed type, over a conversion to an unsigned type of
3449 the same size. Previous versions of G++ tried to preserve
3450 unsignedness, but the standard mandates the current behavior.
3452 @item -Wtemplates @r{(C++ and Objective-C++ only)}
3454 @opindex Wno-templates
3455 Warn when a primary template declaration is encountered. Some coding
3456 rules disallow templates, and this may be used to enforce that rule.
3457 The warning is inactive inside a system header file, such as the STL, so
3458 one can still use the STL. One may also instantiate or specialize
3461 @item -Wmultiple-inheritance @r{(C++ and Objective-C++ only)}
3462 @opindex Wmultiple-inheritance
3463 @opindex Wno-multiple-inheritance
3464 Warn when a class is defined with multiple direct base classes. Some
3465 coding rules disallow multiple inheritance, and this may be used to
3466 enforce that rule. The warning is inactive inside a system header file,
3467 such as the STL, so one can still use the STL. One may also define
3468 classes that indirectly use multiple inheritance.
3470 @item -Wvirtual-inheritance
3471 @opindex Wvirtual-inheritance
3472 @opindex Wno-virtual-inheritance
3473 Warn when a class is defined with a virtual direct base class. Some
3474 coding rules disallow multiple inheritance, and this may be used to
3475 enforce that rule. The warning is inactive inside a system header file,
3476 such as the STL, so one can still use the STL. One may also define
3477 classes that indirectly use virtual inheritance.
3480 @opindex Wnamespaces
3481 @opindex Wno-namespaces
3482 Warn when a namespace definition is opened. Some coding rules disallow
3483 namespaces, and this may be used to enforce that rule. The warning is
3484 inactive inside a system header file, such as the STL, so one can still
3485 use the STL. One may also use using directives and qualified names.
3487 @item -Wno-terminate @r{(C++ and Objective-C++ only)}
3489 @opindex Wno-terminate
3490 Disable the warning about a throw-expression that will immediately
3491 result in a call to @code{terminate}.
3493 @item -Wno-class-conversion @r{(C++ and Objective-C++ only)}
3494 @opindex Wno-class-conversion
3495 @opindex Wclass-conversion
3496 Disable the warning about the case when a conversion function converts an
3497 object to the same type, to a base class of that type, or to void; such
3498 a conversion function will never be called.
3501 @node Objective-C and Objective-C++ Dialect Options
3502 @section Options Controlling Objective-C and Objective-C++ Dialects
3504 @cindex compiler options, Objective-C and Objective-C++
3505 @cindex Objective-C and Objective-C++ options, command-line
3506 @cindex options, Objective-C and Objective-C++
3507 (NOTE: This manual does not describe the Objective-C and Objective-C++
3508 languages themselves. @xref{Standards,,Language Standards
3509 Supported by GCC}, for references.)
3511 This section describes the command-line options that are only meaningful
3512 for Objective-C and Objective-C++ programs. You can also use most of
3513 the language-independent GNU compiler options.
3514 For example, you might compile a file @file{some_class.m} like this:
3517 gcc -g -fgnu-runtime -O -c some_class.m
3521 In this example, @option{-fgnu-runtime} is an option meant only for
3522 Objective-C and Objective-C++ programs; you can use the other options with
3523 any language supported by GCC@.
3525 Note that since Objective-C is an extension of the C language, Objective-C
3526 compilations may also use options specific to the C front-end (e.g.,
3527 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
3528 C++-specific options (e.g., @option{-Wabi}).
3530 Here is a list of options that are @emph{only} for compiling Objective-C
3531 and Objective-C++ programs:
3534 @item -fconstant-string-class=@var{class-name}
3535 @opindex fconstant-string-class
3536 Use @var{class-name} as the name of the class to instantiate for each
3537 literal string specified with the syntax @code{@@"@dots{}"}. The default
3538 class name is @code{NXConstantString} if the GNU runtime is being used, and
3539 @code{NSConstantString} if the NeXT runtime is being used (see below). The
3540 @option{-fconstant-cfstrings} option, if also present, overrides the
3541 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
3542 to be laid out as constant CoreFoundation strings.
3545 @opindex fgnu-runtime
3546 Generate object code compatible with the standard GNU Objective-C
3547 runtime. This is the default for most types of systems.
3549 @item -fnext-runtime
3550 @opindex fnext-runtime
3551 Generate output compatible with the NeXT runtime. This is the default
3552 for NeXT-based systems, including Darwin and Mac OS X@. The macro
3553 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
3556 @item -fno-nil-receivers
3557 @opindex fno-nil-receivers
3558 @opindex fnil-receivers
3559 Assume that all Objective-C message dispatches (@code{[receiver
3560 message:arg]}) in this translation unit ensure that the receiver is
3561 not @code{nil}. This allows for more efficient entry points in the
3562 runtime to be used. This option is only available in conjunction with
3563 the NeXT runtime and ABI version 0 or 1.
3565 @item -fobjc-abi-version=@var{n}
3566 @opindex fobjc-abi-version
3567 Use version @var{n} of the Objective-C ABI for the selected runtime.
3568 This option is currently supported only for the NeXT runtime. In that
3569 case, Version 0 is the traditional (32-bit) ABI without support for
3570 properties and other Objective-C 2.0 additions. Version 1 is the
3571 traditional (32-bit) ABI with support for properties and other
3572 Objective-C 2.0 additions. Version 2 is the modern (64-bit) ABI. If
3573 nothing is specified, the default is Version 0 on 32-bit target
3574 machines, and Version 2 on 64-bit target machines.
3576 @item -fobjc-call-cxx-cdtors
3577 @opindex fobjc-call-cxx-cdtors
3578 For each Objective-C class, check if any of its instance variables is a
3579 C++ object with a non-trivial default constructor. If so, synthesize a
3580 special @code{- (id) .cxx_construct} instance method which runs
3581 non-trivial default constructors on any such instance variables, in order,
3582 and then return @code{self}. Similarly, check if any instance variable
3583 is a C++ object with a non-trivial destructor, and if so, synthesize a
3584 special @code{- (void) .cxx_destruct} method which runs
3585 all such default destructors, in reverse order.
3587 The @code{- (id) .cxx_construct} and @code{- (void) .cxx_destruct}
3588 methods thusly generated only operate on instance variables
3589 declared in the current Objective-C class, and not those inherited
3590 from superclasses. It is the responsibility of the Objective-C
3591 runtime to invoke all such methods in an object's inheritance
3592 hierarchy. The @code{- (id) .cxx_construct} methods are invoked
3593 by the runtime immediately after a new object instance is allocated;
3594 the @code{- (void) .cxx_destruct} methods are invoked immediately
3595 before the runtime deallocates an object instance.
3597 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
3598 support for invoking the @code{- (id) .cxx_construct} and
3599 @code{- (void) .cxx_destruct} methods.
3601 @item -fobjc-direct-dispatch
3602 @opindex fobjc-direct-dispatch
3603 Allow fast jumps to the message dispatcher. On Darwin this is
3604 accomplished via the comm page.
3606 @item -fobjc-exceptions
3607 @opindex fobjc-exceptions
3608 Enable syntactic support for structured exception handling in
3609 Objective-C, similar to what is offered by C++. This option
3610 is required to use the Objective-C keywords @code{@@try},
3611 @code{@@throw}, @code{@@catch}, @code{@@finally} and
3612 @code{@@synchronized}. This option is available with both the GNU
3613 runtime and the NeXT runtime (but not available in conjunction with
3614 the NeXT runtime on Mac OS X 10.2 and earlier).
3618 Enable garbage collection (GC) in Objective-C and Objective-C++
3619 programs. This option is only available with the NeXT runtime; the
3620 GNU runtime has a different garbage collection implementation that
3621 does not require special compiler flags.
3623 @item -fobjc-nilcheck
3624 @opindex fobjc-nilcheck
3625 For the NeXT runtime with version 2 of the ABI, check for a nil
3626 receiver in method invocations before doing the actual method call.
3627 This is the default and can be disabled using
3628 @option{-fno-objc-nilcheck}. Class methods and super calls are never
3629 checked for nil in this way no matter what this flag is set to.
3630 Currently this flag does nothing when the GNU runtime, or an older
3631 version of the NeXT runtime ABI, is used.
3633 @item -fobjc-std=objc1
3635 Conform to the language syntax of Objective-C 1.0, the language
3636 recognized by GCC 4.0. This only affects the Objective-C additions to
3637 the C/C++ language; it does not affect conformance to C/C++ standards,
3638 which is controlled by the separate C/C++ dialect option flags. When
3639 this option is used with the Objective-C or Objective-C++ compiler,
3640 any Objective-C syntax that is not recognized by GCC 4.0 is rejected.
3641 This is useful if you need to make sure that your Objective-C code can
3642 be compiled with older versions of GCC@.
3644 @item -freplace-objc-classes
3645 @opindex freplace-objc-classes
3646 Emit a special marker instructing @command{ld(1)} not to statically link in
3647 the resulting object file, and allow @command{dyld(1)} to load it in at
3648 run time instead. This is used in conjunction with the Fix-and-Continue
3649 debugging mode, where the object file in question may be recompiled and
3650 dynamically reloaded in the course of program execution, without the need
3651 to restart the program itself. Currently, Fix-and-Continue functionality
3652 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
3657 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
3658 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
3659 compile time) with static class references that get initialized at load time,
3660 which improves run-time performance. Specifying the @option{-fzero-link} flag
3661 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
3662 to be retained. This is useful in Zero-Link debugging mode, since it allows
3663 for individual class implementations to be modified during program execution.
3664 The GNU runtime currently always retains calls to @code{objc_get_class("@dots{}")}
3665 regardless of command-line options.
3667 @item -fno-local-ivars
3668 @opindex fno-local-ivars
3669 @opindex flocal-ivars
3670 By default instance variables in Objective-C can be accessed as if
3671 they were local variables from within the methods of the class they're
3672 declared in. This can lead to shadowing between instance variables
3673 and other variables declared either locally inside a class method or
3674 globally with the same name. Specifying the @option{-fno-local-ivars}
3675 flag disables this behavior thus avoiding variable shadowing issues.
3677 @item -fivar-visibility=@r{[}public@r{|}protected@r{|}private@r{|}package@r{]}
3678 @opindex fivar-visibility
3679 Set the default instance variable visibility to the specified option
3680 so that instance variables declared outside the scope of any access
3681 modifier directives default to the specified visibility.
3685 Dump interface declarations for all classes seen in the source file to a
3686 file named @file{@var{sourcename}.decl}.
3688 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
3689 @opindex Wassign-intercept
3690 @opindex Wno-assign-intercept
3691 Warn whenever an Objective-C assignment is being intercepted by the
3694 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
3695 @opindex Wno-protocol
3697 If a class is declared to implement a protocol, a warning is issued for
3698 every method in the protocol that is not implemented by the class. The
3699 default behavior is to issue a warning for every method not explicitly
3700 implemented in the class, even if a method implementation is inherited
3701 from the superclass. If you use the @option{-Wno-protocol} option, then
3702 methods inherited from the superclass are considered to be implemented,
3703 and no warning is issued for them.
3705 @item -Wselector @r{(Objective-C and Objective-C++ only)}
3707 @opindex Wno-selector
3708 Warn if multiple methods of different types for the same selector are
3709 found during compilation. The check is performed on the list of methods
3710 in the final stage of compilation. Additionally, a check is performed
3711 for each selector appearing in a @code{@@selector(@dots{})}
3712 expression, and a corresponding method for that selector has been found
3713 during compilation. Because these checks scan the method table only at
3714 the end of compilation, these warnings are not produced if the final
3715 stage of compilation is not reached, for example because an error is
3716 found during compilation, or because the @option{-fsyntax-only} option is
3719 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
3720 @opindex Wstrict-selector-match
3721 @opindex Wno-strict-selector-match
3722 Warn if multiple methods with differing argument and/or return types are
3723 found for a given selector when attempting to send a message using this
3724 selector to a receiver of type @code{id} or @code{Class}. When this flag
3725 is off (which is the default behavior), the compiler omits such warnings
3726 if any differences found are confined to types that share the same size
3729 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
3730 @opindex Wundeclared-selector
3731 @opindex Wno-undeclared-selector
3732 Warn if a @code{@@selector(@dots{})} expression referring to an
3733 undeclared selector is found. A selector is considered undeclared if no
3734 method with that name has been declared before the
3735 @code{@@selector(@dots{})} expression, either explicitly in an
3736 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
3737 an @code{@@implementation} section. This option always performs its
3738 checks as soon as a @code{@@selector(@dots{})} expression is found,
3739 while @option{-Wselector} only performs its checks in the final stage of
3740 compilation. This also enforces the coding style convention
3741 that methods and selectors must be declared before being used.
3743 @item -print-objc-runtime-info
3744 @opindex print-objc-runtime-info
3745 Generate C header describing the largest structure that is passed by
3750 @node Diagnostic Message Formatting Options
3751 @section Options to Control Diagnostic Messages Formatting
3752 @cindex options to control diagnostics formatting
3753 @cindex diagnostic messages
3754 @cindex message formatting
3756 Traditionally, diagnostic messages have been formatted irrespective of
3757 the output device's aspect (e.g.@: its width, @dots{}). You can use the
3758 options described below
3759 to control the formatting algorithm for diagnostic messages,
3760 e.g.@: how many characters per line, how often source location
3761 information should be reported. Note that some language front ends may not
3762 honor these options.
3765 @item -fmessage-length=@var{n}
3766 @opindex fmessage-length
3767 Try to format error messages so that they fit on lines of about
3768 @var{n} characters. If @var{n} is zero, then no line-wrapping is
3769 done; each error message appears on a single line. This is the
3770 default for all front ends.
3772 Note - this option also affects the display of the @samp{#error} and
3773 @samp{#warning} pre-processor directives, and the @samp{deprecated}
3774 function/type/variable attribute. It does not however affect the
3775 @samp{pragma GCC warning} and @samp{pragma GCC error} pragmas.
3777 @item -fdiagnostics-show-location=once
3778 @opindex fdiagnostics-show-location
3779 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
3780 reporter to emit source location information @emph{once}; that is, in
3781 case the message is too long to fit on a single physical line and has to
3782 be wrapped, the source location won't be emitted (as prefix) again,
3783 over and over, in subsequent continuation lines. This is the default
3786 @item -fdiagnostics-show-location=every-line
3787 Only meaningful in line-wrapping mode. Instructs the diagnostic
3788 messages reporter to emit the same source location information (as
3789 prefix) for physical lines that result from the process of breaking
3790 a message which is too long to fit on a single line.
3792 @item -fdiagnostics-color[=@var{WHEN}]
3793 @itemx -fno-diagnostics-color
3794 @opindex fdiagnostics-color
3795 @cindex highlight, color
3796 @vindex GCC_COLORS @r{environment variable}
3797 Use color in diagnostics. @var{WHEN} is @samp{never}, @samp{always},
3798 or @samp{auto}. The default depends on how the compiler has been configured,
3799 it can be any of the above @var{WHEN} options or also @samp{never}
3800 if @env{GCC_COLORS} environment variable isn't present in the environment,
3801 and @samp{auto} otherwise.
3802 @samp{auto} means to use color only when the standard error is a terminal.
3803 The forms @option{-fdiagnostics-color} and @option{-fno-diagnostics-color} are
3804 aliases for @option{-fdiagnostics-color=always} and
3805 @option{-fdiagnostics-color=never}, respectively.
3807 The colors are defined by the environment variable @env{GCC_COLORS}.
3808 Its value is a colon-separated list of capabilities and Select Graphic
3809 Rendition (SGR) substrings. SGR commands are interpreted by the
3810 terminal or terminal emulator. (See the section in the documentation
3811 of your text terminal for permitted values and their meanings as
3812 character attributes.) These substring values are integers in decimal
3813 representation and can be concatenated with semicolons.
3814 Common values to concatenate include
3816 @samp{4} for underline,
3818 @samp{7} for inverse,
3819 @samp{39} for default foreground color,
3820 @samp{30} to @samp{37} for foreground colors,
3821 @samp{90} to @samp{97} for 16-color mode foreground colors,
3822 @samp{38;5;0} to @samp{38;5;255}
3823 for 88-color and 256-color modes foreground colors,
3824 @samp{49} for default background color,
3825 @samp{40} to @samp{47} for background colors,
3826 @samp{100} to @samp{107} for 16-color mode background colors,
3827 and @samp{48;5;0} to @samp{48;5;255}
3828 for 88-color and 256-color modes background colors.
3830 The default @env{GCC_COLORS} is
3832 error=01;31:warning=01;35:note=01;36:range1=32:range2=34:locus=01:\
3833 quote=01:fixit-insert=32:fixit-delete=31:\
3834 diff-filename=01:diff-hunk=32:diff-delete=31:diff-insert=32:\
3838 where @samp{01;31} is bold red, @samp{01;35} is bold magenta,
3839 @samp{01;36} is bold cyan, @samp{32} is green, @samp{34} is blue,
3840 @samp{01} is bold, and @samp{31} is red.
3841 Setting @env{GCC_COLORS} to the empty string disables colors.
3842 Supported capabilities are as follows.
3846 @vindex error GCC_COLORS @r{capability}
3847 SGR substring for error: markers.
3850 @vindex warning GCC_COLORS @r{capability}
3851 SGR substring for warning: markers.
3854 @vindex note GCC_COLORS @r{capability}
3855 SGR substring for note: markers.
3858 @vindex range1 GCC_COLORS @r{capability}
3859 SGR substring for first additional range.
3862 @vindex range2 GCC_COLORS @r{capability}
3863 SGR substring for second additional range.
3866 @vindex locus GCC_COLORS @r{capability}
3867 SGR substring for location information, @samp{file:line} or
3868 @samp{file:line:column} etc.
3871 @vindex quote GCC_COLORS @r{capability}
3872 SGR substring for information printed within quotes.
3875 @vindex fixit-insert GCC_COLORS @r{capability}
3876 SGR substring for fix-it hints suggesting text to
3877 be inserted or replaced.
3880 @vindex fixit-delete GCC_COLORS @r{capability}
3881 SGR substring for fix-it hints suggesting text to
3884 @item diff-filename=
3885 @vindex diff-filename GCC_COLORS @r{capability}
3886 SGR substring for filename headers within generated patches.
3889 @vindex diff-hunk GCC_COLORS @r{capability}
3890 SGR substring for the starts of hunks within generated patches.
3893 @vindex diff-delete GCC_COLORS @r{capability}
3894 SGR substring for deleted lines within generated patches.
3897 @vindex diff-insert GCC_COLORS @r{capability}
3898 SGR substring for inserted lines within generated patches.
3901 @vindex type-diff GCC_COLORS @r{capability}
3902 SGR substring for highlighting mismatching types within template
3903 arguments in the C++ frontend.
3906 @item -fno-diagnostics-show-option
3907 @opindex fno-diagnostics-show-option
3908 @opindex fdiagnostics-show-option
3909 By default, each diagnostic emitted includes text indicating the
3910 command-line option that directly controls the diagnostic (if such an
3911 option is known to the diagnostic machinery). Specifying the
3912 @option{-fno-diagnostics-show-option} flag suppresses that behavior.
3914 @item -fno-diagnostics-show-caret
3915 @opindex fno-diagnostics-show-caret
3916 @opindex fdiagnostics-show-caret
3917 By default, each diagnostic emitted includes the original source line
3918 and a caret @samp{^} indicating the column. This option suppresses this
3919 information. The source line is truncated to @var{n} characters, if
3920 the @option{-fmessage-length=n} option is given. When the output is done
3921 to the terminal, the width is limited to the width given by the
3922 @env{COLUMNS} environment variable or, if not set, to the terminal width.
3924 @item -fno-diagnostics-show-labels
3925 @opindex fno-diagnostics-show-labels
3926 @opindex fdiagnostics-show-labels
3927 By default, when printing source code (via @option{-fdiagnostics-show-caret}),
3928 diagnostics can label ranges of source code with pertinent information, such
3929 as the types of expressions:
3932 printf ("foo %s bar", long_i + long_j);
3938 This option suppresses the printing of these labels (in the example above,
3939 the vertical bars and the ``char *'' and ``long int'' text).
3941 @item -fno-diagnostics-show-line-numbers
3942 @opindex fno-diagnostics-show-line-numbers
3943 @opindex fdiagnostics-show-line-numbers
3944 By default, when printing source code (via @option{-fdiagnostics-show-caret}),
3945 a left margin is printed, showing line numbers. This option suppresses this
3948 @item -fdiagnostics-minimum-margin-width=@var{width}
3949 @opindex fdiagnostics-minimum-margin-width
3950 This option controls the minimum width of the left margin printed by
3951 @option{-fdiagnostics-show-line-numbers}. It defaults to 6.
3953 @item -fdiagnostics-parseable-fixits
3954 @opindex fdiagnostics-parseable-fixits
3955 Emit fix-it hints in a machine-parseable format, suitable for consumption
3956 by IDEs. For each fix-it, a line will be printed after the relevant
3957 diagnostic, starting with the string ``fix-it:''. For example:
3960 fix-it:"test.c":@{45:3-45:21@}:"gtk_widget_show_all"
3963 The location is expressed as a half-open range, expressed as a count of
3964 bytes, starting at byte 1 for the initial column. In the above example,
3965 bytes 3 through 20 of line 45 of ``test.c'' are to be replaced with the
3969 00000000011111111112222222222
3970 12345678901234567890123456789
3971 gtk_widget_showall (dlg);
3976 The filename and replacement string escape backslash as ``\\", tab as ``\t'',
3977 newline as ``\n'', double quotes as ``\"'', non-printable characters as octal
3978 (e.g. vertical tab as ``\013'').
3980 An empty replacement string indicates that the given range is to be removed.
3981 An empty range (e.g. ``45:3-45:3'') indicates that the string is to
3982 be inserted at the given position.
3984 @item -fdiagnostics-generate-patch
3985 @opindex fdiagnostics-generate-patch
3986 Print fix-it hints to stderr in unified diff format, after any diagnostics
3987 are printed. For example:
3994 void show_cb(GtkDialog *dlg)
3996 - gtk_widget_showall(dlg);
3997 + gtk_widget_show_all(dlg);
4002 The diff may or may not be colorized, following the same rules
4003 as for diagnostics (see @option{-fdiagnostics-color}).
4005 @item -fdiagnostics-show-template-tree
4006 @opindex fdiagnostics-show-template-tree
4008 In the C++ frontend, when printing diagnostics showing mismatching
4009 template types, such as:
4012 could not convert 'std::map<int, std::vector<double> >()'
4013 from 'map<[...],vector<double>>' to 'map<[...],vector<float>>
4016 the @option{-fdiagnostics-show-template-tree} flag enables printing a
4017 tree-like structure showing the common and differing parts of the types,
4027 The parts that differ are highlighted with color (``double'' and
4028 ``float'' in this case).
4030 @item -fno-elide-type
4031 @opindex fno-elide-type
4032 @opindex felide-type
4033 By default when the C++ frontend prints diagnostics showing mismatching
4034 template types, common parts of the types are printed as ``[...]'' to
4035 simplify the error message. For example:
4038 could not convert 'std::map<int, std::vector<double> >()'
4039 from 'map<[...],vector<double>>' to 'map<[...],vector<float>>
4042 Specifying the @option{-fno-elide-type} flag suppresses that behavior.
4043 This flag also affects the output of the
4044 @option{-fdiagnostics-show-template-tree} flag.
4046 @item -fno-show-column
4047 @opindex fno-show-column
4048 @opindex fshow-column
4049 Do not print column numbers in diagnostics. This may be necessary if
4050 diagnostics are being scanned by a program that does not understand the
4051 column numbers, such as @command{dejagnu}.
4053 @item -fdiagnostics-format=@var{FORMAT}
4054 @opindex fdiagnostics-format
4055 Select a different format for printing diagnostics.
4056 @var{FORMAT} is @samp{text} or @samp{json}.
4057 The default is @samp{text}.
4059 The @samp{json} format consists of a top-level JSON array containing JSON
4060 objects representing the diagnostics.
4062 The JSON is emitted as one line, without formatting; the examples below
4063 have been formatted for clarity.
4065 Diagnostics can have child diagnostics. For example, this error and note:
4068 misleading-indentation.c:15:3: warning: this 'if' clause does not
4069 guard... [-Wmisleading-indentation]
4072 misleading-indentation.c:17:5: note: ...this statement, but the latter
4073 is misleadingly indented as if it were guarded by the 'if'
4079 might be printed in JSON form (after formatting) like this:
4089 "file": "misleading-indentation.c",
4094 "file": "misleading-indentation.c",
4099 "message": "this \u2018if\u2019 clause does not guard...",
4100 "option": "-Wmisleading-indentation",
4108 "file": "misleading-indentation.c",
4113 "message": "...this statement, but the latter is @dots{}"
4122 where the @code{note} is a child of the @code{warning}.
4124 A diagnostic has a @code{kind}. If this is @code{warning}, then there is
4125 an @code{option} key describing the command-line option controlling the
4128 A diagnostic can contain zero or more locations. Each location has up
4129 to three positions within it: a @code{caret} position and optional
4130 @code{start} and @code{finish} positions. A location can also have
4131 an optional @code{label} string. For example, this error:
4134 bad-binary-ops.c:64:23: error: invalid operands to binary + (have 'S' @{aka
4135 'struct s'@} and 'T' @{aka 'struct t'@})
4136 64 | return callee_4a () + callee_4b ();
4137 | ~~~~~~~~~~~~ ^ ~~~~~~~~~~~~
4139 | | T @{aka struct t@}
4140 | S @{aka struct s@}
4144 has three locations. Its primary location is at the ``+'' token at column
4145 23. It has two secondary locations, describing the left and right-hand sides
4146 of the expression, which have labels. It might be printed in JSON form as:
4155 "column": 23, "file": "bad-binary-ops.c", "line": 64
4160 "column": 10, "file": "bad-binary-ops.c", "line": 64
4163 "column": 21, "file": "bad-binary-ops.c", "line": 64
4165 "label": "S @{aka struct s@}"
4169 "column": 25, "file": "bad-binary-ops.c", "line": 64
4172 "column": 36, "file": "bad-binary-ops.c", "line": 64
4174 "label": "T @{aka struct t@}"
4177 "message": "invalid operands to binary + @dots{}"
4181 If a diagnostic contains fix-it hints, it has a @code{fixits} array,
4182 consisting of half-open intervals, similar to the output of
4183 @option{-fdiagnostics-parseable-fixits}. For example, this diagnostic
4184 with a replacement fix-it hint:
4187 demo.c:8:15: error: 'struct s' has no member named 'colour'; did you
4189 8 | return ptr->colour;
4195 might be printed in JSON form as:
4230 "message": "\u2018struct s\u2019 has no member named @dots{}"
4235 where the fix-it hint suggests replacing the text from @code{start} up
4236 to but not including @code{next} with @code{string}'s value. Deletions
4237 are expressed via an empty value for @code{string}, insertions by
4238 having @code{start} equal @code{next}.
4242 @node Warning Options
4243 @section Options to Request or Suppress Warnings
4244 @cindex options to control warnings
4245 @cindex warning messages
4246 @cindex messages, warning
4247 @cindex suppressing warnings
4249 Warnings are diagnostic messages that report constructions that
4250 are not inherently erroneous but that are risky or suggest there
4251 may have been an error.
4253 The following language-independent options do not enable specific
4254 warnings but control the kinds of diagnostics produced by GCC@.
4257 @cindex syntax checking
4259 @opindex fsyntax-only
4260 Check the code for syntax errors, but don't do anything beyond that.
4262 @item -fmax-errors=@var{n}
4263 @opindex fmax-errors
4264 Limits the maximum number of error messages to @var{n}, at which point
4265 GCC bails out rather than attempting to continue processing the source
4266 code. If @var{n} is 0 (the default), there is no limit on the number
4267 of error messages produced. If @option{-Wfatal-errors} is also
4268 specified, then @option{-Wfatal-errors} takes precedence over this
4273 Inhibit all warning messages.
4278 Make all warnings into errors.
4283 Make the specified warning into an error. The specifier for a warning
4284 is appended; for example @option{-Werror=switch} turns the warnings
4285 controlled by @option{-Wswitch} into errors. This switch takes a
4286 negative form, to be used to negate @option{-Werror} for specific
4287 warnings; for example @option{-Wno-error=switch} makes
4288 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
4291 The warning message for each controllable warning includes the
4292 option that controls the warning. That option can then be used with
4293 @option{-Werror=} and @option{-Wno-error=} as described above.
4294 (Printing of the option in the warning message can be disabled using the
4295 @option{-fno-diagnostics-show-option} flag.)
4297 Note that specifying @option{-Werror=}@var{foo} automatically implies
4298 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
4301 @item -Wfatal-errors
4302 @opindex Wfatal-errors
4303 @opindex Wno-fatal-errors
4304 This option causes the compiler to abort compilation on the first error
4305 occurred rather than trying to keep going and printing further error
4310 You can request many specific warnings with options beginning with
4311 @samp{-W}, for example @option{-Wimplicit} to request warnings on
4312 implicit declarations. Each of these specific warning options also
4313 has a negative form beginning @samp{-Wno-} to turn off warnings; for
4314 example, @option{-Wno-implicit}. This manual lists only one of the
4315 two forms, whichever is not the default. For further
4316 language-specific options also refer to @ref{C++ Dialect Options} and
4317 @ref{Objective-C and Objective-C++ Dialect Options}.
4319 Some options, such as @option{-Wall} and @option{-Wextra}, turn on other
4320 options, such as @option{-Wunused}, which may turn on further options,
4321 such as @option{-Wunused-value}. The combined effect of positive and
4322 negative forms is that more specific options have priority over less
4323 specific ones, independently of their position in the command-line. For
4324 options of the same specificity, the last one takes effect. Options
4325 enabled or disabled via pragmas (@pxref{Diagnostic Pragmas}) take effect
4326 as if they appeared at the end of the command-line.
4328 When an unrecognized warning option is requested (e.g.,
4329 @option{-Wunknown-warning}), GCC emits a diagnostic stating
4330 that the option is not recognized. However, if the @option{-Wno-} form
4331 is used, the behavior is slightly different: no diagnostic is
4332 produced for @option{-Wno-unknown-warning} unless other diagnostics
4333 are being produced. This allows the use of new @option{-Wno-} options
4334 with old compilers, but if something goes wrong, the compiler
4335 warns that an unrecognized option is present.
4342 @opindex Wno-pedantic
4343 Issue all the warnings demanded by strict ISO C and ISO C++;
4344 reject all programs that use forbidden extensions, and some other
4345 programs that do not follow ISO C and ISO C++. For ISO C, follows the
4346 version of the ISO C standard specified by any @option{-std} option used.
4348 Valid ISO C and ISO C++ programs should compile properly with or without
4349 this option (though a rare few require @option{-ansi} or a
4350 @option{-std} option specifying the required version of ISO C)@. However,
4351 without this option, certain GNU extensions and traditional C and C++
4352 features are supported as well. With this option, they are rejected.
4354 @option{-Wpedantic} does not cause warning messages for use of the
4355 alternate keywords whose names begin and end with @samp{__}. This alternate
4356 format can also be used to disable warnings for non-ISO @samp{__intN} types,
4357 i.e. @samp{__intN__}.
4358 Pedantic warnings are also disabled in the expression that follows
4359 @code{__extension__}. However, only system header files should use
4360 these escape routes; application programs should avoid them.
4361 @xref{Alternate Keywords}.
4363 Some users try to use @option{-Wpedantic} to check programs for strict ISO
4364 C conformance. They soon find that it does not do quite what they want:
4365 it finds some non-ISO practices, but not all---only those for which
4366 ISO C @emph{requires} a diagnostic, and some others for which
4367 diagnostics have been added.
4369 A feature to report any failure to conform to ISO C might be useful in
4370 some instances, but would require considerable additional work and would
4371 be quite different from @option{-Wpedantic}. We don't have plans to
4372 support such a feature in the near future.
4374 Where the standard specified with @option{-std} represents a GNU
4375 extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
4376 corresponding @dfn{base standard}, the version of ISO C on which the GNU
4377 extended dialect is based. Warnings from @option{-Wpedantic} are given
4378 where they are required by the base standard. (It does not make sense
4379 for such warnings to be given only for features not in the specified GNU
4380 C dialect, since by definition the GNU dialects of C include all
4381 features the compiler supports with the given option, and there would be
4382 nothing to warn about.)
4384 @item -pedantic-errors
4385 @opindex pedantic-errors
4386 Give an error whenever the @dfn{base standard} (see @option{-Wpedantic})
4387 requires a diagnostic, in some cases where there is undefined behavior
4388 at compile-time and in some other cases that do not prevent compilation
4389 of programs that are valid according to the standard. This is not
4390 equivalent to @option{-Werror=pedantic}, since there are errors enabled
4391 by this option and not enabled by the latter and vice versa.
4396 This enables all the warnings about constructions that some users
4397 consider questionable, and that are easy to avoid (or modify to
4398 prevent the warning), even in conjunction with macros. This also
4399 enables some language-specific warnings described in @ref{C++ Dialect
4400 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
4402 @option{-Wall} turns on the following warning flags:
4404 @gccoptlist{-Waddress @gol
4405 -Warray-bounds=1 @r{(only with} @option{-O2}@r{)} @gol
4407 -Wbool-operation @gol
4408 -Wc++11-compat -Wc++14-compat @gol
4409 -Wcatch-value @r{(C++ and Objective-C++ only)} @gol
4410 -Wchar-subscripts @gol
4412 -Wduplicate-decl-specifier @r{(C and Objective-C only)} @gol
4413 -Wenum-compare @r{(in C/ObjC; this is on by default in C++)} @gol
4415 -Wint-in-bool-context @gol
4416 -Wimplicit @r{(C and Objective-C only)} @gol
4417 -Wimplicit-int @r{(C and Objective-C only)} @gol
4418 -Wimplicit-function-declaration @r{(C and Objective-C only)} @gol
4419 -Winit-self @r{(only for C++)} @gol
4420 -Wlogical-not-parentheses @gol
4421 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
4422 -Wmaybe-uninitialized @gol
4423 -Wmemset-elt-size @gol
4424 -Wmemset-transposed-args @gol
4425 -Wmisleading-indentation @r{(only for C/C++)} @gol
4426 -Wmissing-attributes @gol
4427 -Wmissing-braces @r{(only for C/ObjC)} @gol
4428 -Wmultistatement-macros @gol
4429 -Wnarrowing @r{(only for C++)} @gol
4431 -Wnonnull-compare @gol
4434 -Wpessimizing-move @r{(only for C++)} @gol
4439 -Wsequence-point @gol
4440 -Wsign-compare @r{(only in C++)} @gol
4441 -Wsizeof-pointer-div @gol
4442 -Wsizeof-pointer-memaccess @gol
4443 -Wstrict-aliasing @gol
4444 -Wstrict-overflow=1 @gol
4446 -Wtautological-compare @gol
4448 -Wuninitialized @gol
4449 -Wunknown-pragmas @gol
4450 -Wunused-function @gol
4453 -Wunused-variable @gol
4454 -Wvolatile-register-var}
4456 Note that some warning flags are not implied by @option{-Wall}. Some of
4457 them warn about constructions that users generally do not consider
4458 questionable, but which occasionally you might wish to check for;
4459 others warn about constructions that are necessary or hard to avoid in
4460 some cases, and there is no simple way to modify the code to suppress
4461 the warning. Some of them are enabled by @option{-Wextra} but many of
4462 them must be enabled individually.
4468 This enables some extra warning flags that are not enabled by
4469 @option{-Wall}. (This option used to be called @option{-W}. The older
4470 name is still supported, but the newer name is more descriptive.)
4472 @gccoptlist{-Wclobbered @gol
4473 -Wcast-function-type @gol
4474 -Wdeprecated-copy @r{(C++ only)} @gol
4476 -Wignored-qualifiers @gol
4477 -Wimplicit-fallthrough=3 @gol
4478 -Wmissing-field-initializers @gol
4479 -Wmissing-parameter-type @r{(C only)} @gol
4480 -Wold-style-declaration @r{(C only)} @gol
4481 -Woverride-init @gol
4482 -Wsign-compare @r{(C only)} @gol
4483 -Wredundant-move @r{(only for C++)} @gol
4485 -Wuninitialized @gol
4486 -Wshift-negative-value @r{(in C++03 and in C99 and newer)} @gol
4487 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
4488 -Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)}}
4491 The option @option{-Wextra} also prints warning messages for the
4497 A pointer is compared against integer zero with @code{<}, @code{<=},
4498 @code{>}, or @code{>=}.
4501 (C++ only) An enumerator and a non-enumerator both appear in a
4502 conditional expression.
4505 (C++ only) Ambiguous virtual bases.
4508 (C++ only) Subscripting an array that has been declared @code{register}.
4511 (C++ only) Taking the address of a variable that has been declared
4515 (C++ only) A base class is not initialized in the copy constructor
4520 @item -Wchar-subscripts
4521 @opindex Wchar-subscripts
4522 @opindex Wno-char-subscripts
4523 Warn if an array subscript has type @code{char}. This is a common cause
4524 of error, as programmers often forget that this type is signed on some
4526 This warning is enabled by @option{-Wall}.
4528 @item -Wno-coverage-mismatch
4529 @opindex Wno-coverage-mismatch
4530 @opindex Wcoverage-mismatch
4531 Warn if feedback profiles do not match when using the
4532 @option{-fprofile-use} option.
4533 If a source file is changed between compiling with @option{-fprofile-generate}
4534 and with @option{-fprofile-use}, the files with the profile feedback can fail
4535 to match the source file and GCC cannot use the profile feedback
4536 information. By default, this warning is enabled and is treated as an
4537 error. @option{-Wno-coverage-mismatch} can be used to disable the
4538 warning or @option{-Wno-error=coverage-mismatch} can be used to
4539 disable the error. Disabling the error for this warning can result in
4540 poorly optimized code and is useful only in the
4541 case of very minor changes such as bug fixes to an existing code-base.
4542 Completely disabling the warning is not recommended.
4545 @r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
4547 Suppress warning messages emitted by @code{#warning} directives.
4549 @item -Wdouble-promotion @r{(C, C++, Objective-C and Objective-C++ only)}
4550 @opindex Wdouble-promotion
4551 @opindex Wno-double-promotion
4552 Give a warning when a value of type @code{float} is implicitly
4553 promoted to @code{double}. CPUs with a 32-bit ``single-precision''
4554 floating-point unit implement @code{float} in hardware, but emulate
4555 @code{double} in software. On such a machine, doing computations
4556 using @code{double} values is much more expensive because of the
4557 overhead required for software emulation.
4559 It is easy to accidentally do computations with @code{double} because
4560 floating-point literals are implicitly of type @code{double}. For
4564 float area(float radius)
4566 return 3.14159 * radius * radius;
4570 the compiler performs the entire computation with @code{double}
4571 because the floating-point literal is a @code{double}.
4573 @item -Wduplicate-decl-specifier @r{(C and Objective-C only)}
4574 @opindex Wduplicate-decl-specifier
4575 @opindex Wno-duplicate-decl-specifier
4576 Warn if a declaration has duplicate @code{const}, @code{volatile},
4577 @code{restrict} or @code{_Atomic} specifier. This warning is enabled by
4581 @itemx -Wformat=@var{n}
4584 @opindex ffreestanding
4585 @opindex fno-builtin
4587 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
4588 the arguments supplied have types appropriate to the format string
4589 specified, and that the conversions specified in the format string make
4590 sense. This includes standard functions, and others specified by format
4591 attributes (@pxref{Function Attributes}), in the @code{printf},
4592 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
4593 not in the C standard) families (or other target-specific families).
4594 Which functions are checked without format attributes having been
4595 specified depends on the standard version selected, and such checks of
4596 functions without the attribute specified are disabled by
4597 @option{-ffreestanding} or @option{-fno-builtin}.
4599 The formats are checked against the format features supported by GNU
4600 libc version 2.2. These include all ISO C90 and C99 features, as well
4601 as features from the Single Unix Specification and some BSD and GNU
4602 extensions. Other library implementations may not support all these
4603 features; GCC does not support warning about features that go beyond a
4604 particular library's limitations. However, if @option{-Wpedantic} is used
4605 with @option{-Wformat}, warnings are given about format features not
4606 in the selected standard version (but not for @code{strfmon} formats,
4607 since those are not in any version of the C standard). @xref{C Dialect
4608 Options,,Options Controlling C Dialect}.
4615 Option @option{-Wformat} is equivalent to @option{-Wformat=1}, and
4616 @option{-Wno-format} is equivalent to @option{-Wformat=0}. Since
4617 @option{-Wformat} also checks for null format arguments for several
4618 functions, @option{-Wformat} also implies @option{-Wnonnull}. Some
4619 aspects of this level of format checking can be disabled by the
4620 options: @option{-Wno-format-contains-nul},
4621 @option{-Wno-format-extra-args}, and @option{-Wno-format-zero-length}.
4622 @option{-Wformat} is enabled by @option{-Wall}.
4624 @item -Wno-format-contains-nul
4625 @opindex Wno-format-contains-nul
4626 @opindex Wformat-contains-nul
4627 If @option{-Wformat} is specified, do not warn about format strings that
4630 @item -Wno-format-extra-args
4631 @opindex Wno-format-extra-args
4632 @opindex Wformat-extra-args
4633 If @option{-Wformat} is specified, do not warn about excess arguments to a
4634 @code{printf} or @code{scanf} format function. The C standard specifies
4635 that such arguments are ignored.
4637 Where the unused arguments lie between used arguments that are
4638 specified with @samp{$} operand number specifications, normally
4639 warnings are still given, since the implementation could not know what
4640 type to pass to @code{va_arg} to skip the unused arguments. However,
4641 in the case of @code{scanf} formats, this option suppresses the
4642 warning if the unused arguments are all pointers, since the Single
4643 Unix Specification says that such unused arguments are allowed.
4645 @item -Wformat-overflow
4646 @itemx -Wformat-overflow=@var{level}
4647 @opindex Wformat-overflow
4648 @opindex Wno-format-overflow
4649 Warn about calls to formatted input/output functions such as @code{sprintf}
4650 and @code{vsprintf} that might overflow the destination buffer. When the
4651 exact number of bytes written by a format directive cannot be determined
4652 at compile-time it is estimated based on heuristics that depend on the
4653 @var{level} argument and on optimization. While enabling optimization
4654 will in most cases improve the accuracy of the warning, it may also
4655 result in false positives.
4658 @item -Wformat-overflow
4659 @itemx -Wformat-overflow=1
4660 @opindex Wformat-overflow
4661 @opindex Wno-format-overflow
4662 Level @var{1} of @option{-Wformat-overflow} enabled by @option{-Wformat}
4663 employs a conservative approach that warns only about calls that most
4664 likely overflow the buffer. At this level, numeric arguments to format
4665 directives with unknown values are assumed to have the value of one, and
4666 strings of unknown length to be empty. Numeric arguments that are known
4667 to be bounded to a subrange of their type, or string arguments whose output
4668 is bounded either by their directive's precision or by a finite set of
4669 string literals, are assumed to take on the value within the range that
4670 results in the most bytes on output. For example, the call to @code{sprintf}
4671 below is diagnosed because even with both @var{a} and @var{b} equal to zero,
4672 the terminating NUL character (@code{'\0'}) appended by the function
4673 to the destination buffer will be written past its end. Increasing
4674 the size of the buffer by a single byte is sufficient to avoid the
4675 warning, though it may not be sufficient to avoid the overflow.
4678 void f (int a, int b)
4681 sprintf (buf, "a = %i, b = %i\n", a, b);
4685 @item -Wformat-overflow=2
4686 Level @var{2} warns also about calls that might overflow the destination
4687 buffer given an argument of sufficient length or magnitude. At level
4688 @var{2}, unknown numeric arguments are assumed to have the minimum
4689 representable value for signed types with a precision greater than 1, and
4690 the maximum representable value otherwise. Unknown string arguments whose
4691 length cannot be assumed to be bounded either by the directive's precision,
4692 or by a finite set of string literals they may evaluate to, or the character
4693 array they may point to, are assumed to be 1 character long.
4695 At level @var{2}, the call in the example above is again diagnosed, but
4696 this time because with @var{a} equal to a 32-bit @code{INT_MIN} the first
4697 @code{%i} directive will write some of its digits beyond the end of
4698 the destination buffer. To make the call safe regardless of the values
4699 of the two variables, the size of the destination buffer must be increased
4700 to at least 34 bytes. GCC includes the minimum size of the buffer in
4701 an informational note following the warning.
4703 An alternative to increasing the size of the destination buffer is to
4704 constrain the range of formatted values. The maximum length of string
4705 arguments can be bounded by specifying the precision in the format
4706 directive. When numeric arguments of format directives can be assumed
4707 to be bounded by less than the precision of their type, choosing
4708 an appropriate length modifier to the format specifier will reduce
4709 the required buffer size. For example, if @var{a} and @var{b} in the
4710 example above can be assumed to be within the precision of
4711 the @code{short int} type then using either the @code{%hi} format
4712 directive or casting the argument to @code{short} reduces the maximum
4713 required size of the buffer to 24 bytes.
4716 void f (int a, int b)
4719 sprintf (buf, "a = %hi, b = %i\n", a, (short)b);
4724 @item -Wno-format-zero-length
4725 @opindex Wno-format-zero-length
4726 @opindex Wformat-zero-length
4727 If @option{-Wformat} is specified, do not warn about zero-length formats.
4728 The C standard specifies that zero-length formats are allowed.
4733 Enable @option{-Wformat} plus additional format checks. Currently
4734 equivalent to @option{-Wformat -Wformat-nonliteral -Wformat-security
4737 @item -Wformat-nonliteral
4738 @opindex Wformat-nonliteral
4739 @opindex Wno-format-nonliteral
4740 If @option{-Wformat} is specified, also warn if the format string is not a
4741 string literal and so cannot be checked, unless the format function
4742 takes its format arguments as a @code{va_list}.
4744 @item -Wformat-security
4745 @opindex Wformat-security
4746 @opindex Wno-format-security
4747 If @option{-Wformat} is specified, also warn about uses of format
4748 functions that represent possible security problems. At present, this
4749 warns about calls to @code{printf} and @code{scanf} functions where the
4750 format string is not a string literal and there are no format arguments,
4751 as in @code{printf (foo);}. This may be a security hole if the format
4752 string came from untrusted input and contains @samp{%n}. (This is
4753 currently a subset of what @option{-Wformat-nonliteral} warns about, but
4754 in future warnings may be added to @option{-Wformat-security} that are not
4755 included in @option{-Wformat-nonliteral}.)
4757 @item -Wformat-signedness
4758 @opindex Wformat-signedness
4759 @opindex Wno-format-signedness
4760 If @option{-Wformat} is specified, also warn if the format string
4761 requires an unsigned argument and the argument is signed and vice versa.
4763 @item -Wformat-truncation
4764 @itemx -Wformat-truncation=@var{level}
4765 @opindex Wformat-truncation
4766 @opindex Wno-format-truncation
4767 Warn about calls to formatted input/output functions such as @code{snprintf}
4768 and @code{vsnprintf} that might result in output truncation. When the exact
4769 number of bytes written by a format directive cannot be determined at
4770 compile-time it is estimated based on heuristics that depend on
4771 the @var{level} argument and on optimization. While enabling optimization
4772 will in most cases improve the accuracy of the warning, it may also result
4773 in false positives. Except as noted otherwise, the option uses the same
4774 logic @option{-Wformat-overflow}.
4777 @item -Wformat-truncation
4778 @itemx -Wformat-truncation=1
4779 @opindex Wformat-truncation
4780 @opindex Wno-format-truncation
4781 Level @var{1} of @option{-Wformat-truncation} enabled by @option{-Wformat}
4782 employs a conservative approach that warns only about calls to bounded
4783 functions whose return value is unused and that will most likely result
4784 in output truncation.
4786 @item -Wformat-truncation=2
4787 Level @var{2} warns also about calls to bounded functions whose return
4788 value is used and that might result in truncation given an argument of
4789 sufficient length or magnitude.
4793 @opindex Wformat-y2k
4794 @opindex Wno-format-y2k
4795 If @option{-Wformat} is specified, also warn about @code{strftime}
4796 formats that may yield only a two-digit year.
4801 @opindex Wno-nonnull
4802 Warn about passing a null pointer for arguments marked as
4803 requiring a non-null value by the @code{nonnull} function attribute.
4805 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
4806 can be disabled with the @option{-Wno-nonnull} option.
4808 @item -Wnonnull-compare
4809 @opindex Wnonnull-compare
4810 @opindex Wno-nonnull-compare
4811 Warn when comparing an argument marked with the @code{nonnull}
4812 function attribute against null inside the function.
4814 @option{-Wnonnull-compare} is included in @option{-Wall}. It
4815 can be disabled with the @option{-Wno-nonnull-compare} option.
4817 @item -Wnull-dereference
4818 @opindex Wnull-dereference
4819 @opindex Wno-null-dereference
4820 Warn if the compiler detects paths that trigger erroneous or
4821 undefined behavior due to dereferencing a null pointer. This option
4822 is only active when @option{-fdelete-null-pointer-checks} is active,
4823 which is enabled by optimizations in most targets. The precision of
4824 the warnings depends on the optimization options used.
4826 @item -Winaccessible-base @r{(C++, Objective-C++ only)}
4827 @opindex Winaccessible-base
4828 @opindex Wno-inaccessible-base
4829 Warn when a base class is inaccessible in a class derived from it due to
4830 ambiguity. The warning is enabled by default. Note the warning for virtual
4831 bases is enabled by the @option{-Wextra} option.
4834 struct A @{ int a; @};
4838 struct C : B, A @{ @};
4842 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
4844 @opindex Wno-init-self
4845 Warn about uninitialized variables that are initialized with themselves.
4846 Note this option can only be used with the @option{-Wuninitialized} option.
4848 For example, GCC warns about @code{i} being uninitialized in the
4849 following snippet only when @option{-Winit-self} has been specified:
4860 This warning is enabled by @option{-Wall} in C++.
4862 @item -Wimplicit-int @r{(C and Objective-C only)}
4863 @opindex Wimplicit-int
4864 @opindex Wno-implicit-int
4865 Warn when a declaration does not specify a type.
4866 This warning is enabled by @option{-Wall}.
4868 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
4869 @opindex Wimplicit-function-declaration
4870 @opindex Wno-implicit-function-declaration
4871 Give a warning whenever a function is used before being declared. In
4872 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
4873 enabled by default and it is made into an error by
4874 @option{-pedantic-errors}. This warning is also enabled by
4877 @item -Wimplicit @r{(C and Objective-C only)}
4879 @opindex Wno-implicit
4880 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
4881 This warning is enabled by @option{-Wall}.
4883 @item -Wimplicit-fallthrough
4884 @opindex Wimplicit-fallthrough
4885 @opindex Wno-implicit-fallthrough
4886 @option{-Wimplicit-fallthrough} is the same as @option{-Wimplicit-fallthrough=3}
4887 and @option{-Wno-implicit-fallthrough} is the same as
4888 @option{-Wimplicit-fallthrough=0}.
4890 @item -Wimplicit-fallthrough=@var{n}
4891 @opindex Wimplicit-fallthrough=
4892 Warn when a switch case falls through. For example:
4910 This warning does not warn when the last statement of a case cannot
4911 fall through, e.g. when there is a return statement or a call to function
4912 declared with the noreturn attribute. @option{-Wimplicit-fallthrough=}
4913 also takes into account control flow statements, such as ifs, and only
4914 warns when appropriate. E.g.@:
4924 @} else if (i < 1) @{
4934 Since there are occasions where a switch case fall through is desirable,
4935 GCC provides an attribute, @code{__attribute__ ((fallthrough))}, that is
4936 to be used along with a null statement to suppress this warning that
4937 would normally occur:
4945 __attribute__ ((fallthrough));
4952 C++17 provides a standard way to suppress the @option{-Wimplicit-fallthrough}
4953 warning using @code{[[fallthrough]];} instead of the GNU attribute. In C++11
4954 or C++14 users can use @code{[[gnu::fallthrough]];}, which is a GNU extension.
4955 Instead of these attributes, it is also possible to add a fallthrough comment
4956 to silence the warning. The whole body of the C or C++ style comment should
4957 match the given regular expressions listed below. The option argument @var{n}
4958 specifies what kind of comments are accepted:
4962 @item @option{-Wimplicit-fallthrough=0} disables the warning altogether.
4964 @item @option{-Wimplicit-fallthrough=1} matches @code{.*} regular
4965 expression, any comment is used as fallthrough comment.
4967 @item @option{-Wimplicit-fallthrough=2} case insensitively matches
4968 @code{.*falls?[ \t-]*thr(ough|u).*} regular expression.
4970 @item @option{-Wimplicit-fallthrough=3} case sensitively matches one of the
4971 following regular expressions:
4975 @item @code{-fallthrough}
4977 @item @code{@@fallthrough@@}
4979 @item @code{lint -fallthrough[ \t]*}
4981 @item @code{[ \t.!]*(ELSE,? |INTENTIONAL(LY)? )?@*FALL(S | |-)?THR(OUGH|U)[ \t.!]*(-[^\n\r]*)?}
4983 @item @code{[ \t.!]*(Else,? |Intentional(ly)? )?@*Fall((s | |-)[Tt]|t)hr(ough|u)[ \t.!]*(-[^\n\r]*)?}
4985 @item @code{[ \t.!]*([Ee]lse,? |[Ii]ntentional(ly)? )?@*fall(s | |-)?thr(ough|u)[ \t.!]*(-[^\n\r]*)?}
4989 @item @option{-Wimplicit-fallthrough=4} case sensitively matches one of the
4990 following regular expressions:
4994 @item @code{-fallthrough}
4996 @item @code{@@fallthrough@@}
4998 @item @code{lint -fallthrough[ \t]*}
5000 @item @code{[ \t]*FALLTHR(OUGH|U)[ \t]*}
5004 @item @option{-Wimplicit-fallthrough=5} doesn't recognize any comments as
5005 fallthrough comments, only attributes disable the warning.
5009 The comment needs to be followed after optional whitespace and other comments
5010 by @code{case} or @code{default} keywords or by a user label that precedes some
5011 @code{case} or @code{default} label.
5026 The @option{-Wimplicit-fallthrough=3} warning is enabled by @option{-Wextra}.
5028 @item -Wif-not-aligned @r{(C, C++, Objective-C and Objective-C++ only)}
5029 @opindex Wif-not-aligned
5030 @opindex Wno-if-not-aligned
5031 Control if warning triggered by the @code{warn_if_not_aligned} attribute
5032 should be issued. This is enabled by default.
5033 Use @option{-Wno-if-not-aligned} to disable it.
5035 @item -Wignored-qualifiers @r{(C and C++ only)}
5036 @opindex Wignored-qualifiers
5037 @opindex Wno-ignored-qualifiers
5038 Warn if the return type of a function has a type qualifier
5039 such as @code{const}. For ISO C such a type qualifier has no effect,
5040 since the value returned by a function is not an lvalue.
5041 For C++, the warning is only emitted for scalar types or @code{void}.
5042 ISO C prohibits qualified @code{void} return types on function
5043 definitions, so such return types always receive a warning
5044 even without this option.
5046 This warning is also enabled by @option{-Wextra}.
5048 @item -Wignored-attributes @r{(C and C++ only)}
5049 @opindex Wignored-attributes
5050 @opindex Wno-ignored-attributes
5051 Warn when an attribute is ignored. This is different from the
5052 @option{-Wattributes} option in that it warns whenever the compiler decides
5053 to drop an attribute, not that the attribute is either unknown, used in a
5054 wrong place, etc. This warning is enabled by default.
5059 Warn if the type of @code{main} is suspicious. @code{main} should be
5060 a function with external linkage, returning int, taking either zero
5061 arguments, two, or three arguments of appropriate types. This warning
5062 is enabled by default in C++ and is enabled by either @option{-Wall}
5063 or @option{-Wpedantic}.
5065 @item -Wmisleading-indentation @r{(C and C++ only)}
5066 @opindex Wmisleading-indentation
5067 @opindex Wno-misleading-indentation
5068 Warn when the indentation of the code does not reflect the block structure.
5069 Specifically, a warning is issued for @code{if}, @code{else}, @code{while}, and
5070 @code{for} clauses with a guarded statement that does not use braces,
5071 followed by an unguarded statement with the same indentation.
5073 In the following example, the call to ``bar'' is misleadingly indented as
5074 if it were guarded by the ``if'' conditional.
5077 if (some_condition ())
5079 bar (); /* Gotcha: this is not guarded by the "if". */
5082 In the case of mixed tabs and spaces, the warning uses the
5083 @option{-ftabstop=} option to determine if the statements line up
5086 The warning is not issued for code involving multiline preprocessor logic
5087 such as the following example.
5092 #if SOME_CONDITION_THAT_DOES_NOT_HOLD
5098 The warning is not issued after a @code{#line} directive, since this
5099 typically indicates autogenerated code, and no assumptions can be made
5100 about the layout of the file that the directive references.
5102 This warning is enabled by @option{-Wall} in C and C++.
5104 @item -Wmissing-attributes
5105 @opindex Wmissing-attributes
5106 @opindex Wno-missing-attributes
5107 Warn when a declaration of a function is missing one or more attributes
5108 that a related function is declared with and whose absence may adversely
5109 affect the correctness or efficiency of generated code. For example,
5110 the warning is issued for declarations of aliases that use attributes
5111 to specify less restrictive requirements than those of their targets.
5112 This typically represents a potential optimization opportunity.
5113 By contrast, the @option{-Wattribute-alias=2} option controls warnings
5114 issued when the alias is more restrictive than the target, which could
5115 lead to incorrect code generation.
5116 Attributes considered include @code{alloc_align}, @code{alloc_size},
5117 @code{cold}, @code{const}, @code{hot}, @code{leaf}, @code{malloc},
5118 @code{nonnull}, @code{noreturn}, @code{nothrow}, @code{pure},
5119 @code{returns_nonnull}, and @code{returns_twice}.
5121 In C++, the warning is issued when an explicit specialization of a primary
5122 template declared with attribute @code{alloc_align}, @code{alloc_size},
5123 @code{assume_aligned}, @code{format}, @code{format_arg}, @code{malloc},
5124 or @code{nonnull} is declared without it. Attributes @code{deprecated},
5125 @code{error}, and @code{warning} suppress the warning.
5126 (@pxref{Function Attributes}).
5128 You can use the @code{copy} attribute to apply the same
5129 set of attributes to a declaration as that on another declaration without
5130 explicitly enumerating the attributes. This attribute can be applied
5131 to declarations of functions (@pxref{Common Function Attributes}),
5132 variables (@pxref{Common Variable Attributes}), or types
5133 (@pxref{Common Type Attributes}).
5135 @option{-Wmissing-attributes} is enabled by @option{-Wall}.
5137 For example, since the declaration of the primary function template
5138 below makes use of both attribute @code{malloc} and @code{alloc_size}
5139 the declaration of the explicit specialization of the template is
5140 diagnosed because it is missing one of the attributes.
5144 T* __attribute__ ((malloc, alloc_size (1)))
5148 void* __attribute__ ((malloc)) // missing alloc_size
5149 allocate<void> (size_t);
5152 @item -Wmissing-braces
5153 @opindex Wmissing-braces
5154 @opindex Wno-missing-braces
5155 Warn if an aggregate or union initializer is not fully bracketed. In
5156 the following example, the initializer for @code{a} is not fully
5157 bracketed, but that for @code{b} is fully bracketed. This warning is
5158 enabled by @option{-Wall} in C.
5161 int a[2][2] = @{ 0, 1, 2, 3 @};
5162 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
5165 This warning is enabled by @option{-Wall}.
5167 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
5168 @opindex Wmissing-include-dirs
5169 @opindex Wno-missing-include-dirs
5170 Warn if a user-supplied include directory does not exist.
5172 @item -Wmissing-profile
5173 @opindex Wmissing-profile
5174 @opindex Wno-missing-profile
5175 Warn if feedback profiles are missing when using the
5176 @option{-fprofile-use} option.
5177 This option diagnoses those cases where a new function or a new file is added
5178 to the user code between compiling with @option{-fprofile-generate} and with
5179 @option{-fprofile-use}, without regenerating the profiles. In these cases, the
5180 profile feedback data files do not contain any profile feedback information for
5181 the newly added function or file respectively. Also, in the case when profile
5182 count data (.gcda) files are removed, GCC cannot use any profile feedback
5183 information. In all these cases, warnings are issued to inform the user that a
5184 profile generation step is due. @option{-Wno-missing-profile} can be used to
5185 disable the warning. Ignoring the warning can result in poorly optimized code.
5186 Completely disabling the warning is not recommended and should be done only
5187 when non-existent profile data is justified.
5189 @item -Wmultistatement-macros
5190 @opindex Wmultistatement-macros
5191 @opindex Wno-multistatement-macros
5192 Warn about unsafe multiple statement macros that appear to be guarded
5193 by a clause such as @code{if}, @code{else}, @code{for}, @code{switch}, or
5194 @code{while}, in which only the first statement is actually guarded after
5195 the macro is expanded.
5200 #define DOIT x++; y++
5205 will increment @code{y} unconditionally, not just when @code{c} holds.
5206 The can usually be fixed by wrapping the macro in a do-while loop:
5208 #define DOIT do @{ x++; y++; @} while (0)
5213 This warning is enabled by @option{-Wall} in C and C++.
5216 @opindex Wparentheses
5217 @opindex Wno-parentheses
5218 Warn if parentheses are omitted in certain contexts, such
5219 as when there is an assignment in a context where a truth value
5220 is expected, or when operators are nested whose precedence people
5221 often get confused about.
5223 Also warn if a comparison like @code{x<=y<=z} appears; this is
5224 equivalent to @code{(x<=y ? 1 : 0) <= z}, which is a different
5225 interpretation from that of ordinary mathematical notation.
5227 Also warn for dangerous uses of the GNU extension to
5228 @code{?:} with omitted middle operand. When the condition
5229 in the @code{?}: operator is a boolean expression, the omitted value is
5230 always 1. Often programmers expect it to be a value computed
5231 inside the conditional expression instead.
5233 For C++ this also warns for some cases of unnecessary parentheses in
5234 declarations, which can indicate an attempt at a function call instead
5238 // Declares a local variable called mymutex.
5239 std::unique_lock<std::mutex> (mymutex);
5240 // User meant std::unique_lock<std::mutex> lock (mymutex);
5244 This warning is enabled by @option{-Wall}.
5246 @item -Wsequence-point
5247 @opindex Wsequence-point
5248 @opindex Wno-sequence-point
5249 Warn about code that may have undefined semantics because of violations
5250 of sequence point rules in the C and C++ standards.
5252 The C and C++ standards define the order in which expressions in a C/C++
5253 program are evaluated in terms of @dfn{sequence points}, which represent
5254 a partial ordering between the execution of parts of the program: those
5255 executed before the sequence point, and those executed after it. These
5256 occur after the evaluation of a full expression (one which is not part
5257 of a larger expression), after the evaluation of the first operand of a
5258 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
5259 function is called (but after the evaluation of its arguments and the
5260 expression denoting the called function), and in certain other places.
5261 Other than as expressed by the sequence point rules, the order of
5262 evaluation of subexpressions of an expression is not specified. All
5263 these rules describe only a partial order rather than a total order,
5264 since, for example, if two functions are called within one expression
5265 with no sequence point between them, the order in which the functions
5266 are called is not specified. However, the standards committee have
5267 ruled that function calls do not overlap.
5269 It is not specified when between sequence points modifications to the
5270 values of objects take effect. Programs whose behavior depends on this
5271 have undefined behavior; the C and C++ standards specify that ``Between
5272 the previous and next sequence point an object shall have its stored
5273 value modified at most once by the evaluation of an expression.
5274 Furthermore, the prior value shall be read only to determine the value
5275 to be stored.''. If a program breaks these rules, the results on any
5276 particular implementation are entirely unpredictable.
5278 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
5279 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
5280 diagnosed by this option, and it may give an occasional false positive
5281 result, but in general it has been found fairly effective at detecting
5282 this sort of problem in programs.
5284 The C++17 standard will define the order of evaluation of operands in
5285 more cases: in particular it requires that the right-hand side of an
5286 assignment be evaluated before the left-hand side, so the above
5287 examples are no longer undefined. But this warning will still warn
5288 about them, to help people avoid writing code that is undefined in C
5289 and earlier revisions of C++.
5291 The standard is worded confusingly, therefore there is some debate
5292 over the precise meaning of the sequence point rules in subtle cases.
5293 Links to discussions of the problem, including proposed formal
5294 definitions, may be found on the GCC readings page, at
5295 @uref{http://gcc.gnu.org/@/readings.html}.
5297 This warning is enabled by @option{-Wall} for C and C++.
5299 @item -Wno-return-local-addr
5300 @opindex Wno-return-local-addr
5301 @opindex Wreturn-local-addr
5302 Do not warn about returning a pointer (or in C++, a reference) to a
5303 variable that goes out of scope after the function returns.
5306 @opindex Wreturn-type
5307 @opindex Wno-return-type
5308 Warn whenever a function is defined with a return type that defaults
5309 to @code{int}. Also warn about any @code{return} statement with no
5310 return value in a function whose return type is not @code{void}
5311 (falling off the end of the function body is considered returning
5314 For C only, warn about a @code{return} statement with an expression in a
5315 function whose return type is @code{void}, unless the expression type is
5316 also @code{void}. As a GNU extension, the latter case is accepted
5317 without a warning unless @option{-Wpedantic} is used. Attempting
5318 to use the return value of a non-@code{void} function other than @code{main}
5319 that flows off the end by reaching the closing curly brace that terminates
5320 the function is undefined.
5322 Unlike in C, in C++, flowing off the end of a non-@code{void} function other
5323 than @code{main} results in undefined behavior even when the value of
5324 the function is not used.
5326 This warning is enabled by default in C++ and by @option{-Wall} otherwise.
5328 @item -Wshift-count-negative
5329 @opindex Wshift-count-negative
5330 @opindex Wno-shift-count-negative
5331 Warn if shift count is negative. This warning is enabled by default.
5333 @item -Wshift-count-overflow
5334 @opindex Wshift-count-overflow
5335 @opindex Wno-shift-count-overflow
5336 Warn if shift count >= width of type. This warning is enabled by default.
5338 @item -Wshift-negative-value
5339 @opindex Wshift-negative-value
5340 @opindex Wno-shift-negative-value
5341 Warn if left shifting a negative value. This warning is enabled by
5342 @option{-Wextra} in C99 and C++11 modes (and newer).
5344 @item -Wshift-overflow
5345 @itemx -Wshift-overflow=@var{n}
5346 @opindex Wshift-overflow
5347 @opindex Wno-shift-overflow
5348 Warn about left shift overflows. This warning is enabled by
5349 default in C99 and C++11 modes (and newer).
5352 @item -Wshift-overflow=1
5353 This is the warning level of @option{-Wshift-overflow} and is enabled
5354 by default in C99 and C++11 modes (and newer). This warning level does
5355 not warn about left-shifting 1 into the sign bit. (However, in C, such
5356 an overflow is still rejected in contexts where an integer constant expression
5357 is required.) No warning is emitted in C++2A mode (and newer), as signed left
5360 @item -Wshift-overflow=2
5361 This warning level also warns about left-shifting 1 into the sign bit,
5362 unless C++14 mode (or newer) is active.
5368 Warn whenever a @code{switch} statement has an index of enumerated type
5369 and lacks a @code{case} for one or more of the named codes of that
5370 enumeration. (The presence of a @code{default} label prevents this
5371 warning.) @code{case} labels outside the enumeration range also
5372 provoke warnings when this option is used (even if there is a
5373 @code{default} label).
5374 This warning is enabled by @option{-Wall}.
5376 @item -Wswitch-default
5377 @opindex Wswitch-default
5378 @opindex Wno-switch-default
5379 Warn whenever a @code{switch} statement does not have a @code{default}
5383 @opindex Wswitch-enum
5384 @opindex Wno-switch-enum
5385 Warn whenever a @code{switch} statement has an index of enumerated type
5386 and lacks a @code{case} for one or more of the named codes of that
5387 enumeration. @code{case} labels outside the enumeration range also
5388 provoke warnings when this option is used. The only difference
5389 between @option{-Wswitch} and this option is that this option gives a
5390 warning about an omitted enumeration code even if there is a
5391 @code{default} label.
5394 @opindex Wswitch-bool
5395 @opindex Wno-switch-bool
5396 Warn whenever a @code{switch} statement has an index of boolean type
5397 and the case values are outside the range of a boolean type.
5398 It is possible to suppress this warning by casting the controlling
5399 expression to a type other than @code{bool}. For example:
5402 switch ((int) (a == 4))
5408 This warning is enabled by default for C and C++ programs.
5410 @item -Wswitch-outside-range
5411 @opindex Wswitch-outside-range
5412 @opindex Wno-switch-outside-range
5413 Warn whenever a @code{switch} case has a value that is outside of its
5414 respective type range. This warning is enabled by default for
5417 @item -Wswitch-unreachable
5418 @opindex Wswitch-unreachable
5419 @opindex Wno-switch-unreachable
5420 Warn whenever a @code{switch} statement contains statements between the
5421 controlling expression and the first case label, which will never be
5422 executed. For example:
5434 @option{-Wswitch-unreachable} does not warn if the statement between the
5435 controlling expression and the first case label is just a declaration:
5448 This warning is enabled by default for C and C++ programs.
5450 @item -Wsync-nand @r{(C and C++ only)}
5452 @opindex Wno-sync-nand
5453 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
5454 built-in functions are used. These functions changed semantics in GCC 4.4.
5456 @item -Wunused-but-set-parameter
5457 @opindex Wunused-but-set-parameter
5458 @opindex Wno-unused-but-set-parameter
5459 Warn whenever a function parameter is assigned to, but otherwise unused
5460 (aside from its declaration).
5462 To suppress this warning use the @code{unused} attribute
5463 (@pxref{Variable Attributes}).
5465 This warning is also enabled by @option{-Wunused} together with
5468 @item -Wunused-but-set-variable
5469 @opindex Wunused-but-set-variable
5470 @opindex Wno-unused-but-set-variable
5471 Warn whenever a local variable is assigned to, but otherwise unused
5472 (aside from its declaration).
5473 This warning is enabled by @option{-Wall}.
5475 To suppress this warning use the @code{unused} attribute
5476 (@pxref{Variable Attributes}).
5478 This warning is also enabled by @option{-Wunused}, which is enabled
5481 @item -Wunused-function
5482 @opindex Wunused-function
5483 @opindex Wno-unused-function
5484 Warn whenever a static function is declared but not defined or a
5485 non-inline static function is unused.
5486 This warning is enabled by @option{-Wall}.
5488 @item -Wunused-label
5489 @opindex Wunused-label
5490 @opindex Wno-unused-label
5491 Warn whenever a label is declared but not used.
5492 This warning is enabled by @option{-Wall}.
5494 To suppress this warning use the @code{unused} attribute
5495 (@pxref{Variable Attributes}).
5497 @item -Wunused-local-typedefs @r{(C, Objective-C, C++ and Objective-C++ only)}
5498 @opindex Wunused-local-typedefs
5499 @opindex Wno-unused-local-typedefs
5500 Warn when a typedef locally defined in a function is not used.
5501 This warning is enabled by @option{-Wall}.
5503 @item -Wunused-parameter
5504 @opindex Wunused-parameter
5505 @opindex Wno-unused-parameter
5506 Warn whenever a function parameter is unused aside from its declaration.
5508 To suppress this warning use the @code{unused} attribute
5509 (@pxref{Variable Attributes}).
5511 @item -Wno-unused-result
5512 @opindex Wunused-result
5513 @opindex Wno-unused-result
5514 Do not warn if a caller of a function marked with attribute
5515 @code{warn_unused_result} (@pxref{Function Attributes}) does not use
5516 its return value. The default is @option{-Wunused-result}.
5518 @item -Wunused-variable
5519 @opindex Wunused-variable
5520 @opindex Wno-unused-variable
5521 Warn whenever a local or static variable is unused aside from its
5522 declaration. This option implies @option{-Wunused-const-variable=1} for C,
5523 but not for C++. This warning is enabled by @option{-Wall}.
5525 To suppress this warning use the @code{unused} attribute
5526 (@pxref{Variable Attributes}).
5528 @item -Wunused-const-variable
5529 @itemx -Wunused-const-variable=@var{n}
5530 @opindex Wunused-const-variable
5531 @opindex Wno-unused-const-variable
5532 Warn whenever a constant static variable is unused aside from its declaration.
5533 @option{-Wunused-const-variable=1} is enabled by @option{-Wunused-variable}
5534 for C, but not for C++. In C this declares variable storage, but in C++ this
5535 is not an error since const variables take the place of @code{#define}s.
5537 To suppress this warning use the @code{unused} attribute
5538 (@pxref{Variable Attributes}).
5541 @item -Wunused-const-variable=1
5542 This is the warning level that is enabled by @option{-Wunused-variable} for
5543 C. It warns only about unused static const variables defined in the main
5544 compilation unit, but not about static const variables declared in any
5547 @item -Wunused-const-variable=2
5548 This warning level also warns for unused constant static variables in
5549 headers (excluding system headers). This is the warning level of
5550 @option{-Wunused-const-variable} and must be explicitly requested since
5551 in C++ this isn't an error and in C it might be harder to clean up all
5555 @item -Wunused-value
5556 @opindex Wunused-value
5557 @opindex Wno-unused-value
5558 Warn whenever a statement computes a result that is explicitly not
5559 used. To suppress this warning cast the unused expression to
5560 @code{void}. This includes an expression-statement or the left-hand
5561 side of a comma expression that contains no side effects. For example,
5562 an expression such as @code{x[i,j]} causes a warning, while
5563 @code{x[(void)i,j]} does not.
5565 This warning is enabled by @option{-Wall}.
5570 All the above @option{-Wunused} options combined.
5572 In order to get a warning about an unused function parameter, you must
5573 either specify @option{-Wextra -Wunused} (note that @option{-Wall} implies
5574 @option{-Wunused}), or separately specify @option{-Wunused-parameter}.
5576 @item -Wuninitialized
5577 @opindex Wuninitialized
5578 @opindex Wno-uninitialized
5579 Warn if an automatic variable is used without first being initialized
5580 or if a variable may be clobbered by a @code{setjmp} call. In C++,
5581 warn if a non-static reference or non-static @code{const} member
5582 appears in a class without constructors.
5584 If you want to warn about code that uses the uninitialized value of the
5585 variable in its own initializer, use the @option{-Winit-self} option.
5587 These warnings occur for individual uninitialized or clobbered
5588 elements of structure, union or array variables as well as for
5589 variables that are uninitialized or clobbered as a whole. They do
5590 not occur for variables or elements declared @code{volatile}. Because
5591 these warnings depend on optimization, the exact variables or elements
5592 for which there are warnings depends on the precise optimization
5593 options and version of GCC used.
5595 Note that there may be no warning about a variable that is used only
5596 to compute a value that itself is never used, because such
5597 computations may be deleted by data flow analysis before the warnings
5600 @item -Winvalid-memory-model
5601 @opindex Winvalid-memory-model
5602 @opindex Wno-invalid-memory-model
5603 Warn for invocations of @ref{__atomic Builtins}, @ref{__sync Builtins},
5604 and the C11 atomic generic functions with a memory consistency argument
5605 that is either invalid for the operation or outside the range of values
5606 of the @code{memory_order} enumeration. For example, since the
5607 @code{__atomic_store} and @code{__atomic_store_n} built-ins are only
5608 defined for the relaxed, release, and sequentially consistent memory
5609 orders the following code is diagnosed:
5614 __atomic_store_n (i, 0, memory_order_consume);
5618 @option{-Winvalid-memory-model} is enabled by default.
5620 @item -Wmaybe-uninitialized
5621 @opindex Wmaybe-uninitialized
5622 @opindex Wno-maybe-uninitialized
5623 For an automatic (i.e.@: local) variable, if there exists a path from the
5624 function entry to a use of the variable that is initialized, but there exist
5625 some other paths for which the variable is not initialized, the compiler
5626 emits a warning if it cannot prove the uninitialized paths are not
5627 executed at run time.
5629 These warnings are only possible in optimizing compilation, because otherwise
5630 GCC does not keep track of the state of variables.
5632 These warnings are made optional because GCC may not be able to determine when
5633 the code is correct in spite of appearing to have an error. Here is one
5634 example of how this can happen:
5654 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
5655 always initialized, but GCC doesn't know this. To suppress the
5656 warning, you need to provide a default case with assert(0) or
5659 @cindex @code{longjmp} warnings
5660 This option also warns when a non-volatile automatic variable might be
5661 changed by a call to @code{longjmp}.
5662 The compiler sees only the calls to @code{setjmp}. It cannot know
5663 where @code{longjmp} will be called; in fact, a signal handler could
5664 call it at any point in the code. As a result, you may get a warning
5665 even when there is in fact no problem because @code{longjmp} cannot
5666 in fact be called at the place that would cause a problem.
5668 Some spurious warnings can be avoided if you declare all the functions
5669 you use that never return as @code{noreturn}. @xref{Function
5672 This warning is enabled by @option{-Wall} or @option{-Wextra}.
5674 @item -Wunknown-pragmas
5675 @opindex Wunknown-pragmas
5676 @opindex Wno-unknown-pragmas
5677 @cindex warning for unknown pragmas
5678 @cindex unknown pragmas, warning
5679 @cindex pragmas, warning of unknown
5680 Warn when a @code{#pragma} directive is encountered that is not understood by
5681 GCC@. If this command-line option is used, warnings are even issued
5682 for unknown pragmas in system header files. This is not the case if
5683 the warnings are only enabled by the @option{-Wall} command-line option.
5686 @opindex Wno-pragmas
5688 Do not warn about misuses of pragmas, such as incorrect parameters,
5689 invalid syntax, or conflicts between pragmas. See also
5690 @option{-Wunknown-pragmas}.
5692 @item -Wno-prio-ctor-dtor
5693 @opindex Wno-prio-ctor-dtor
5694 @opindex Wprio-ctor-dtor
5695 Do not warn if a priority from 0 to 100 is used for constructor or destructor.
5696 The use of constructor and destructor attributes allow you to assign a
5697 priority to the constructor/destructor to control its order of execution
5698 before @code{main} is called or after it returns. The priority values must be
5699 greater than 100 as the compiler reserves priority values between 0--100 for
5702 @item -Wstrict-aliasing
5703 @opindex Wstrict-aliasing
5704 @opindex Wno-strict-aliasing
5705 This option is only active when @option{-fstrict-aliasing} is active.
5706 It warns about code that might break the strict aliasing rules that the
5707 compiler is using for optimization. The warning does not catch all
5708 cases, but does attempt to catch the more common pitfalls. It is
5709 included in @option{-Wall}.
5710 It is equivalent to @option{-Wstrict-aliasing=3}
5712 @item -Wstrict-aliasing=n
5713 @opindex Wstrict-aliasing=n
5714 This option is only active when @option{-fstrict-aliasing} is active.
5715 It warns about code that might break the strict aliasing rules that the
5716 compiler is using for optimization.
5717 Higher levels correspond to higher accuracy (fewer false positives).
5718 Higher levels also correspond to more effort, similar to the way @option{-O}
5720 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=3}.
5722 Level 1: Most aggressive, quick, least accurate.
5723 Possibly useful when higher levels
5724 do not warn but @option{-fstrict-aliasing} still breaks the code, as it has very few
5725 false negatives. However, it has many false positives.
5726 Warns for all pointer conversions between possibly incompatible types,
5727 even if never dereferenced. Runs in the front end only.
5729 Level 2: Aggressive, quick, not too precise.
5730 May still have many false positives (not as many as level 1 though),
5731 and few false negatives (but possibly more than level 1).
5732 Unlike level 1, it only warns when an address is taken. Warns about
5733 incomplete types. Runs in the front end only.
5735 Level 3 (default for @option{-Wstrict-aliasing}):
5736 Should have very few false positives and few false
5737 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
5738 Takes care of the common pun+dereference pattern in the front end:
5739 @code{*(int*)&some_float}.
5740 If optimization is enabled, it also runs in the back end, where it deals
5741 with multiple statement cases using flow-sensitive points-to information.
5742 Only warns when the converted pointer is dereferenced.
5743 Does not warn about incomplete types.
5745 @item -Wstrict-overflow
5746 @itemx -Wstrict-overflow=@var{n}
5747 @opindex Wstrict-overflow
5748 @opindex Wno-strict-overflow
5749 This option is only active when signed overflow is undefined.
5750 It warns about cases where the compiler optimizes based on the
5751 assumption that signed overflow does not occur. Note that it does not
5752 warn about all cases where the code might overflow: it only warns
5753 about cases where the compiler implements some optimization. Thus
5754 this warning depends on the optimization level.
5756 An optimization that assumes that signed overflow does not occur is
5757 perfectly safe if the values of the variables involved are such that
5758 overflow never does, in fact, occur. Therefore this warning can
5759 easily give a false positive: a warning about code that is not
5760 actually a problem. To help focus on important issues, several
5761 warning levels are defined. No warnings are issued for the use of
5762 undefined signed overflow when estimating how many iterations a loop
5763 requires, in particular when determining whether a loop will be
5767 @item -Wstrict-overflow=1
5768 Warn about cases that are both questionable and easy to avoid. For
5769 example the compiler simplifies
5770 @code{x + 1 > x} to @code{1}. This level of
5771 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
5772 are not, and must be explicitly requested.
5774 @item -Wstrict-overflow=2
5775 Also warn about other cases where a comparison is simplified to a
5776 constant. For example: @code{abs (x) >= 0}. This can only be
5777 simplified when signed integer overflow is undefined, because
5778 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
5779 zero. @option{-Wstrict-overflow} (with no level) is the same as
5780 @option{-Wstrict-overflow=2}.
5782 @item -Wstrict-overflow=3
5783 Also warn about other cases where a comparison is simplified. For
5784 example: @code{x + 1 > 1} is simplified to @code{x > 0}.
5786 @item -Wstrict-overflow=4
5787 Also warn about other simplifications not covered by the above cases.
5788 For example: @code{(x * 10) / 5} is simplified to @code{x * 2}.
5790 @item -Wstrict-overflow=5
5791 Also warn about cases where the compiler reduces the magnitude of a
5792 constant involved in a comparison. For example: @code{x + 2 > y} is
5793 simplified to @code{x + 1 >= y}. This is reported only at the
5794 highest warning level because this simplification applies to many
5795 comparisons, so this warning level gives a very large number of
5799 @item -Wstringop-overflow
5800 @itemx -Wstringop-overflow=@var{type}
5801 @opindex Wstringop-overflow
5802 @opindex Wno-stringop-overflow
5803 Warn for calls to string manipulation functions such as @code{memcpy} and
5804 @code{strcpy} that are determined to overflow the destination buffer. The
5805 optional argument is one greater than the type of Object Size Checking to
5806 perform to determine the size of the destination. @xref{Object Size Checking}.
5807 The argument is meaningful only for functions that operate on character arrays
5808 but not for raw memory functions like @code{memcpy} which always make use
5809 of Object Size type-0. The option also warns for calls that specify a size
5810 in excess of the largest possible object or at most @code{SIZE_MAX / 2} bytes.
5811 The option produces the best results with optimization enabled but can detect
5812 a small subset of simple buffer overflows even without optimization in
5813 calls to the GCC built-in functions like @code{__builtin_memcpy} that
5814 correspond to the standard functions. In any case, the option warns about
5815 just a subset of buffer overflows detected by the corresponding overflow
5816 checking built-ins. For example, the option will issue a warning for
5817 the @code{strcpy} call below because it copies at least 5 characters
5818 (the string @code{"blue"} including the terminating NUL) into the buffer
5822 enum Color @{ blue, purple, yellow @};
5823 const char* f (enum Color clr)
5825 static char buf [4];
5829 case blue: str = "blue"; break;
5830 case purple: str = "purple"; break;
5831 case yellow: str = "yellow"; break;
5834 return strcpy (buf, str); // warning here
5838 Option @option{-Wstringop-overflow=2} is enabled by default.
5841 @item -Wstringop-overflow
5842 @itemx -Wstringop-overflow=1
5843 @opindex Wstringop-overflow
5844 @opindex Wno-stringop-overflow
5845 The @option{-Wstringop-overflow=1} option uses type-zero Object Size Checking
5846 to determine the sizes of destination objects. This is the default setting
5847 of the option. At this setting the option will not warn for writes past
5848 the end of subobjects of larger objects accessed by pointers unless the
5849 size of the largest surrounding object is known. When the destination may
5850 be one of several objects it is assumed to be the largest one of them. On
5851 Linux systems, when optimization is enabled at this setting the option warns
5852 for the same code as when the @code{_FORTIFY_SOURCE} macro is defined to
5855 @item -Wstringop-overflow=2
5856 The @option{-Wstringop-overflow=2} option uses type-one Object Size Checking
5857 to determine the sizes of destination objects. At this setting the option
5858 will warn about overflows when writing to members of the largest complete
5859 objects whose exact size is known. It will, however, not warn for excessive
5860 writes to the same members of unknown objects referenced by pointers since
5861 they may point to arrays containing unknown numbers of elements.
5863 @item -Wstringop-overflow=3
5864 The @option{-Wstringop-overflow=3} option uses type-two Object Size Checking
5865 to determine the sizes of destination objects. At this setting the option
5866 warns about overflowing the smallest object or data member. This is the
5867 most restrictive setting of the option that may result in warnings for safe
5870 @item -Wstringop-overflow=4
5871 The @option{-Wstringop-overflow=4} option uses type-three Object Size Checking
5872 to determine the sizes of destination objects. At this setting the option
5873 will warn about overflowing any data members, and when the destination is
5874 one of several objects it uses the size of the largest of them to decide
5875 whether to issue a warning. Similarly to @option{-Wstringop-overflow=3} this
5876 setting of the option may result in warnings for benign code.
5879 @item -Wstringop-truncation
5880 @opindex Wstringop-truncation
5881 @opindex Wno-stringop-truncation
5882 Warn for calls to bounded string manipulation functions such as @code{strncat},
5883 @code{strncpy}, and @code{stpncpy} that may either truncate the copied string
5884 or leave the destination unchanged.
5886 In the following example, the call to @code{strncat} specifies a bound that
5887 is less than the length of the source string. As a result, the copy of
5888 the source will be truncated and so the call is diagnosed. To avoid the
5889 warning use @code{bufsize - strlen (buf) - 1)} as the bound.
5892 void append (char *buf, size_t bufsize)
5894 strncat (buf, ".txt", 3);
5898 As another example, the following call to @code{strncpy} results in copying
5899 to @code{d} just the characters preceding the terminating NUL, without
5900 appending the NUL to the end. Assuming the result of @code{strncpy} is
5901 necessarily a NUL-terminated string is a common mistake, and so the call
5902 is diagnosed. To avoid the warning when the result is not expected to be
5903 NUL-terminated, call @code{memcpy} instead.
5906 void copy (char *d, const char *s)
5908 strncpy (d, s, strlen (s));
5912 In the following example, the call to @code{strncpy} specifies the size
5913 of the destination buffer as the bound. If the length of the source
5914 string is equal to or greater than this size the result of the copy will
5915 not be NUL-terminated. Therefore, the call is also diagnosed. To avoid
5916 the warning, specify @code{sizeof buf - 1} as the bound and set the last
5917 element of the buffer to @code{NUL}.
5920 void copy (const char *s)
5923 strncpy (buf, s, sizeof buf);
5928 In situations where a character array is intended to store a sequence
5929 of bytes with no terminating @code{NUL} such an array may be annotated
5930 with attribute @code{nonstring} to avoid this warning. Such arrays,
5931 however, are not suitable arguments to functions that expect
5932 @code{NUL}-terminated strings. To help detect accidental misuses of
5933 such arrays GCC issues warnings unless it can prove that the use is
5934 safe. @xref{Common Variable Attributes}.
5936 @item -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{|}cold@r{|}malloc@r{]}
5937 @opindex Wsuggest-attribute=
5938 @opindex Wno-suggest-attribute=
5939 Warn for cases where adding an attribute may be beneficial. The
5940 attributes currently supported are listed below.
5943 @item -Wsuggest-attribute=pure
5944 @itemx -Wsuggest-attribute=const
5945 @itemx -Wsuggest-attribute=noreturn
5946 @itemx -Wmissing-noreturn
5947 @itemx -Wsuggest-attribute=malloc
5948 @opindex Wsuggest-attribute=pure
5949 @opindex Wno-suggest-attribute=pure
5950 @opindex Wsuggest-attribute=const
5951 @opindex Wno-suggest-attribute=const
5952 @opindex Wsuggest-attribute=noreturn
5953 @opindex Wno-suggest-attribute=noreturn
5954 @opindex Wmissing-noreturn
5955 @opindex Wno-missing-noreturn
5956 @opindex Wsuggest-attribute=malloc
5957 @opindex Wno-suggest-attribute=malloc
5959 Warn about functions that might be candidates for attributes
5960 @code{pure}, @code{const} or @code{noreturn} or @code{malloc}. The compiler
5961 only warns for functions visible in other compilation units or (in the case of
5962 @code{pure} and @code{const}) if it cannot prove that the function returns
5963 normally. A function returns normally if it doesn't contain an infinite loop or
5964 return abnormally by throwing, calling @code{abort} or trapping. This analysis
5965 requires option @option{-fipa-pure-const}, which is enabled by default at
5966 @option{-O} and higher. Higher optimization levels improve the accuracy
5969 @item -Wsuggest-attribute=format
5970 @itemx -Wmissing-format-attribute
5971 @opindex Wsuggest-attribute=format
5972 @opindex Wmissing-format-attribute
5973 @opindex Wno-suggest-attribute=format
5974 @opindex Wno-missing-format-attribute
5978 Warn about function pointers that might be candidates for @code{format}
5979 attributes. Note these are only possible candidates, not absolute ones.
5980 GCC guesses that function pointers with @code{format} attributes that
5981 are used in assignment, initialization, parameter passing or return
5982 statements should have a corresponding @code{format} attribute in the
5983 resulting type. I.e.@: the left-hand side of the assignment or
5984 initialization, the type of the parameter variable, or the return type
5985 of the containing function respectively should also have a @code{format}
5986 attribute to avoid the warning.
5988 GCC also warns about function definitions that might be
5989 candidates for @code{format} attributes. Again, these are only
5990 possible candidates. GCC guesses that @code{format} attributes
5991 might be appropriate for any function that calls a function like
5992 @code{vprintf} or @code{vscanf}, but this might not always be the
5993 case, and some functions for which @code{format} attributes are
5994 appropriate may not be detected.
5996 @item -Wsuggest-attribute=cold
5997 @opindex Wsuggest-attribute=cold
5998 @opindex Wno-suggest-attribute=cold
6000 Warn about functions that might be candidates for @code{cold} attribute. This
6001 is based on static detection and generally will only warn about functions which
6002 always leads to a call to another @code{cold} function such as wrappers of
6003 C++ @code{throw} or fatal error reporting functions leading to @code{abort}.
6006 @item -Wsuggest-final-types
6007 @opindex Wno-suggest-final-types
6008 @opindex Wsuggest-final-types
6009 Warn about types with virtual methods where code quality would be improved
6010 if the type were declared with the C++11 @code{final} specifier,
6012 declared in an anonymous namespace. This allows GCC to more aggressively
6013 devirtualize the polymorphic calls. This warning is more effective with
6014 link-time optimization,
6015 where the information about the class hierarchy graph is
6018 @item -Wsuggest-final-methods
6019 @opindex Wno-suggest-final-methods
6020 @opindex Wsuggest-final-methods
6021 Warn about virtual methods where code quality would be improved if the method
6022 were declared with the C++11 @code{final} specifier,
6023 or, if possible, its type were
6024 declared in an anonymous namespace or with the @code{final} specifier.
6026 more effective with link-time optimization, where the information about the
6027 class hierarchy graph is more complete. It is recommended to first consider
6028 suggestions of @option{-Wsuggest-final-types} and then rebuild with new
6031 @item -Wsuggest-override
6032 Warn about overriding virtual functions that are not marked with the override
6036 @opindex Wno-alloc-zero
6037 @opindex Walloc-zero
6038 Warn about calls to allocation functions decorated with attribute
6039 @code{alloc_size} that specify zero bytes, including those to the built-in
6040 forms of the functions @code{aligned_alloc}, @code{alloca}, @code{calloc},
6041 @code{malloc}, and @code{realloc}. Because the behavior of these functions
6042 when called with a zero size differs among implementations (and in the case
6043 of @code{realloc} has been deprecated) relying on it may result in subtle
6044 portability bugs and should be avoided.
6046 @item -Walloc-size-larger-than=@var{byte-size}
6047 @opindex Walloc-size-larger-than=
6048 @opindex Wno-alloc-size-larger-than
6049 Warn about calls to functions decorated with attribute @code{alloc_size}
6050 that attempt to allocate objects larger than the specified number of bytes,
6051 or where the result of the size computation in an integer type with infinite
6052 precision would exceed the value of @samp{PTRDIFF_MAX} on the target.
6053 @option{-Walloc-size-larger-than=}@samp{PTRDIFF_MAX} is enabled by default.
6054 Warnings controlled by the option can be disabled either by specifying
6055 @var{byte-size} of @samp{SIZE_MAX} or more or by
6056 @option{-Wno-alloc-size-larger-than}.
6057 @xref{Function Attributes}.
6059 @item -Wno-alloc-size-larger-than
6060 @opindex Wno-alloc-size-larger-than
6061 Disable @option{-Walloc-size-larger-than=} warnings. The option is
6062 equivalent to @option{-Walloc-size-larger-than=}@samp{SIZE_MAX} or
6068 This option warns on all uses of @code{alloca} in the source.
6070 @item -Walloca-larger-than=@var{byte-size}
6071 @opindex Walloca-larger-than=
6072 @opindex Wno-alloca-larger-than
6073 This option warns on calls to @code{alloca} with an integer argument whose
6074 value is either zero, or that is not bounded by a controlling predicate
6075 that limits its value to at most @var{byte-size}. It also warns for calls
6076 to @code{alloca} where the bound value is unknown. Arguments of non-integer
6077 types are considered unbounded even if they appear to be constrained to
6080 For example, a bounded case of @code{alloca} could be:
6083 void func (size_t n)
6094 In the above example, passing @code{-Walloca-larger-than=1000} would not
6095 issue a warning because the call to @code{alloca} is known to be at most
6096 1000 bytes. However, if @code{-Walloca-larger-than=500} were passed,
6097 the compiler would emit a warning.
6099 Unbounded uses, on the other hand, are uses of @code{alloca} with no
6100 controlling predicate constraining its integer argument. For example:
6105 void *p = alloca (n);
6110 If @code{-Walloca-larger-than=500} were passed, the above would trigger
6111 a warning, but this time because of the lack of bounds checking.
6113 Note, that even seemingly correct code involving signed integers could
6117 void func (signed int n)
6127 In the above example, @var{n} could be negative, causing a larger than
6128 expected argument to be implicitly cast into the @code{alloca} call.
6130 This option also warns when @code{alloca} is used in a loop.
6132 @option{-Walloca-larger-than=}@samp{PTRDIFF_MAX} is enabled by default
6133 but is usually only effective when @option{-ftree-vrp} is active (default
6134 for @option{-O2} and above).
6136 See also @option{-Wvla-larger-than=}@samp{byte-size}.
6138 @item -Wno-alloca-larger-than
6139 @opindex Wno-alloca-larger-than
6140 Disable @option{-Walloca-larger-than=} warnings. The option is
6141 equivalent to @option{-Walloca-larger-than=}@samp{SIZE_MAX} or larger.
6143 @item -Warray-bounds
6144 @itemx -Warray-bounds=@var{n}
6145 @opindex Wno-array-bounds
6146 @opindex Warray-bounds
6147 This option is only active when @option{-ftree-vrp} is active
6148 (default for @option{-O2} and above). It warns about subscripts to arrays
6149 that are always out of bounds. This warning is enabled by @option{-Wall}.
6152 @item -Warray-bounds=1
6153 This is the warning level of @option{-Warray-bounds} and is enabled
6154 by @option{-Wall}; higher levels are not, and must be explicitly requested.
6156 @item -Warray-bounds=2
6157 This warning level also warns about out of bounds access for
6158 arrays at the end of a struct and for arrays accessed through
6159 pointers. This warning level may give a larger number of
6160 false positives and is deactivated by default.
6163 @item -Wattribute-alias=@var{n}
6164 @itemx -Wno-attribute-alias
6165 @opindex Wattribute-alias
6166 @opindex Wno-attribute-alias
6167 Warn about declarations using the @code{alias} and similar attributes whose
6168 target is incompatible with the type of the alias.
6169 @xref{Function Attributes,,Declaring Attributes of Functions}.
6172 @item -Wattribute-alias=1
6173 The default warning level of the @option{-Wattribute-alias} option diagnoses
6174 incompatibilities between the type of the alias declaration and that of its
6175 target. Such incompatibilities are typically indicative of bugs.
6177 @item -Wattribute-alias=2
6179 At this level @option{-Wattribute-alias} also diagnoses cases where
6180 the attributes of the alias declaration are more restrictive than the
6181 attributes applied to its target. These mismatches can potentially
6182 result in incorrect code generation. In other cases they may be
6183 benign and could be resolved simply by adding the missing attribute to
6184 the target. For comparison, see the @option{-Wmissing-attributes}
6185 option, which controls diagnostics when the alias declaration is less
6186 restrictive than the target, rather than more restrictive.
6188 Attributes considered include @code{alloc_align}, @code{alloc_size},
6189 @code{cold}, @code{const}, @code{hot}, @code{leaf}, @code{malloc},
6190 @code{nonnull}, @code{noreturn}, @code{nothrow}, @code{pure},
6191 @code{returns_nonnull}, and @code{returns_twice}.
6194 @option{-Wattribute-alias} is equivalent to @option{-Wattribute-alias=1}.
6195 This is the default. You can disable these warnings with either
6196 @option{-Wno-attribute-alias} or @option{-Wattribute-alias=0}.
6198 @item -Wbool-compare
6199 @opindex Wno-bool-compare
6200 @opindex Wbool-compare
6201 Warn about boolean expression compared with an integer value different from
6202 @code{true}/@code{false}. For instance, the following comparison is
6207 if ((n > 1) == 2) @{ @dots{} @}
6209 This warning is enabled by @option{-Wall}.
6211 @item -Wbool-operation
6212 @opindex Wno-bool-operation
6213 @opindex Wbool-operation
6214 Warn about suspicious operations on expressions of a boolean type. For
6215 instance, bitwise negation of a boolean is very likely a bug in the program.
6216 For C, this warning also warns about incrementing or decrementing a boolean,
6217 which rarely makes sense. (In C++, decrementing a boolean is always invalid.
6218 Incrementing a boolean is invalid in C++17, and deprecated otherwise.)
6220 This warning is enabled by @option{-Wall}.
6222 @item -Wduplicated-branches
6223 @opindex Wno-duplicated-branches
6224 @opindex Wduplicated-branches
6225 Warn when an if-else has identical branches. This warning detects cases like
6232 It doesn't warn when both branches contain just a null statement. This warning
6233 also warn for conditional operators:
6235 int i = x ? *p : *p;
6238 @item -Wduplicated-cond
6239 @opindex Wno-duplicated-cond
6240 @opindex Wduplicated-cond
6241 Warn about duplicated conditions in an if-else-if chain. For instance,
6242 warn for the following code:
6244 if (p->q != NULL) @{ @dots{} @}
6245 else if (p->q != NULL) @{ @dots{} @}
6248 @item -Wframe-address
6249 @opindex Wno-frame-address
6250 @opindex Wframe-address
6251 Warn when the @samp{__builtin_frame_address} or @samp{__builtin_return_address}
6252 is called with an argument greater than 0. Such calls may return indeterminate
6253 values or crash the program. The warning is included in @option{-Wall}.
6255 @item -Wno-discarded-qualifiers @r{(C and Objective-C only)}
6256 @opindex Wno-discarded-qualifiers
6257 @opindex Wdiscarded-qualifiers
6258 Do not warn if type qualifiers on pointers are being discarded.
6259 Typically, the compiler warns if a @code{const char *} variable is
6260 passed to a function that takes a @code{char *} parameter. This option
6261 can be used to suppress such a warning.
6263 @item -Wno-discarded-array-qualifiers @r{(C and Objective-C only)}
6264 @opindex Wno-discarded-array-qualifiers
6265 @opindex Wdiscarded-array-qualifiers
6266 Do not warn if type qualifiers on arrays which are pointer targets
6267 are being discarded. Typically, the compiler warns if a
6268 @code{const int (*)[]} variable is passed to a function that
6269 takes a @code{int (*)[]} parameter. This option can be used to
6270 suppress such a warning.
6272 @item -Wno-incompatible-pointer-types @r{(C and Objective-C only)}
6273 @opindex Wno-incompatible-pointer-types
6274 @opindex Wincompatible-pointer-types
6275 Do not warn when there is a conversion between pointers that have incompatible
6276 types. This warning is for cases not covered by @option{-Wno-pointer-sign},
6277 which warns for pointer argument passing or assignment with different
6280 @item -Wno-int-conversion @r{(C and Objective-C only)}
6281 @opindex Wno-int-conversion
6282 @opindex Wint-conversion
6283 Do not warn about incompatible integer to pointer and pointer to integer
6284 conversions. This warning is about implicit conversions; for explicit
6285 conversions the warnings @option{-Wno-int-to-pointer-cast} and
6286 @option{-Wno-pointer-to-int-cast} may be used.
6288 @item -Wno-div-by-zero
6289 @opindex Wno-div-by-zero
6290 @opindex Wdiv-by-zero
6291 Do not warn about compile-time integer division by zero. Floating-point
6292 division by zero is not warned about, as it can be a legitimate way of
6293 obtaining infinities and NaNs.
6295 @item -Wsystem-headers
6296 @opindex Wsystem-headers
6297 @opindex Wno-system-headers
6298 @cindex warnings from system headers
6299 @cindex system headers, warnings from
6300 Print warning messages for constructs found in system header files.
6301 Warnings from system headers are normally suppressed, on the assumption
6302 that they usually do not indicate real problems and would only make the
6303 compiler output harder to read. Using this command-line option tells
6304 GCC to emit warnings from system headers as if they occurred in user
6305 code. However, note that using @option{-Wall} in conjunction with this
6306 option does @emph{not} warn about unknown pragmas in system
6307 headers---for that, @option{-Wunknown-pragmas} must also be used.
6309 @item -Wtautological-compare
6310 @opindex Wtautological-compare
6311 @opindex Wno-tautological-compare
6312 Warn if a self-comparison always evaluates to true or false. This
6313 warning detects various mistakes such as:
6317 if (i > i) @{ @dots{} @}
6320 This warning also warns about bitwise comparisons that always evaluate
6321 to true or false, for instance:
6323 if ((a & 16) == 10) @{ @dots{} @}
6325 will always be false.
6327 This warning is enabled by @option{-Wall}.
6330 @opindex Wtrampolines
6331 @opindex Wno-trampolines
6332 Warn about trampolines generated for pointers to nested functions.
6333 A trampoline is a small piece of data or code that is created at run
6334 time on the stack when the address of a nested function is taken, and is
6335 used to call the nested function indirectly. For some targets, it is
6336 made up of data only and thus requires no special treatment. But, for
6337 most targets, it is made up of code and thus requires the stack to be
6338 made executable in order for the program to work properly.
6341 @opindex Wfloat-equal
6342 @opindex Wno-float-equal
6343 Warn if floating-point values are used in equality comparisons.
6345 The idea behind this is that sometimes it is convenient (for the
6346 programmer) to consider floating-point values as approximations to
6347 infinitely precise real numbers. If you are doing this, then you need
6348 to compute (by analyzing the code, or in some other way) the maximum or
6349 likely maximum error that the computation introduces, and allow for it
6350 when performing comparisons (and when producing output, but that's a
6351 different problem). In particular, instead of testing for equality, you
6352 should check to see whether the two values have ranges that overlap; and
6353 this is done with the relational operators, so equality comparisons are
6356 @item -Wtraditional @r{(C and Objective-C only)}
6357 @opindex Wtraditional
6358 @opindex Wno-traditional
6359 Warn about certain constructs that behave differently in traditional and
6360 ISO C@. Also warn about ISO C constructs that have no traditional C
6361 equivalent, and/or problematic constructs that should be avoided.
6365 Macro parameters that appear within string literals in the macro body.
6366 In traditional C macro replacement takes place within string literals,
6367 but in ISO C it does not.
6370 In traditional C, some preprocessor directives did not exist.
6371 Traditional preprocessors only considered a line to be a directive
6372 if the @samp{#} appeared in column 1 on the line. Therefore
6373 @option{-Wtraditional} warns about directives that traditional C
6374 understands but ignores because the @samp{#} does not appear as the
6375 first character on the line. It also suggests you hide directives like
6376 @code{#pragma} not understood by traditional C by indenting them. Some
6377 traditional implementations do not recognize @code{#elif}, so this option
6378 suggests avoiding it altogether.
6381 A function-like macro that appears without arguments.
6384 The unary plus operator.
6387 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating-point
6388 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
6389 constants.) Note, these suffixes appear in macros defined in the system
6390 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
6391 Use of these macros in user code might normally lead to spurious
6392 warnings, however GCC's integrated preprocessor has enough context to
6393 avoid warning in these cases.
6396 A function declared external in one block and then used after the end of
6400 A @code{switch} statement has an operand of type @code{long}.
6403 A non-@code{static} function declaration follows a @code{static} one.
6404 This construct is not accepted by some traditional C compilers.
6407 The ISO type of an integer constant has a different width or
6408 signedness from its traditional type. This warning is only issued if
6409 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
6410 typically represent bit patterns, are not warned about.
6413 Usage of ISO string concatenation is detected.
6416 Initialization of automatic aggregates.
6419 Identifier conflicts with labels. Traditional C lacks a separate
6420 namespace for labels.
6423 Initialization of unions. If the initializer is zero, the warning is
6424 omitted. This is done under the assumption that the zero initializer in
6425 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
6426 initializer warnings and relies on default initialization to zero in the
6430 Conversions by prototypes between fixed/floating-point values and vice
6431 versa. The absence of these prototypes when compiling with traditional
6432 C causes serious problems. This is a subset of the possible
6433 conversion warnings; for the full set use @option{-Wtraditional-conversion}.
6436 Use of ISO C style function definitions. This warning intentionally is
6437 @emph{not} issued for prototype declarations or variadic functions
6438 because these ISO C features appear in your code when using
6439 libiberty's traditional C compatibility macros, @code{PARAMS} and
6440 @code{VPARAMS}. This warning is also bypassed for nested functions
6441 because that feature is already a GCC extension and thus not relevant to
6442 traditional C compatibility.
6445 @item -Wtraditional-conversion @r{(C and Objective-C only)}
6446 @opindex Wtraditional-conversion
6447 @opindex Wno-traditional-conversion
6448 Warn if a prototype causes a type conversion that is different from what
6449 would happen to the same argument in the absence of a prototype. This
6450 includes conversions of fixed point to floating and vice versa, and
6451 conversions changing the width or signedness of a fixed-point argument
6452 except when the same as the default promotion.
6454 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
6455 @opindex Wdeclaration-after-statement
6456 @opindex Wno-declaration-after-statement
6457 Warn when a declaration is found after a statement in a block. This
6458 construct, known from C++, was introduced with ISO C99 and is by default
6459 allowed in GCC@. It is not supported by ISO C90. @xref{Mixed Declarations}.
6464 Warn whenever a local variable or type declaration shadows another
6465 variable, parameter, type, class member (in C++), or instance variable
6466 (in Objective-C) or whenever a built-in function is shadowed. Note
6467 that in C++, the compiler warns if a local variable shadows an
6468 explicit typedef, but not if it shadows a struct/class/enum.
6469 Same as @option{-Wshadow=global}.
6471 @item -Wno-shadow-ivar @r{(Objective-C only)}
6472 @opindex Wno-shadow-ivar
6473 @opindex Wshadow-ivar
6474 Do not warn whenever a local variable shadows an instance variable in an
6477 @item -Wshadow=global
6478 @opindex Wshadow=local
6479 The default for @option{-Wshadow}. Warns for any (global) shadowing.
6481 @item -Wshadow=local
6482 @opindex Wshadow=local
6483 Warn when a local variable shadows another local variable or parameter.
6484 This warning is enabled by @option{-Wshadow=global}.
6486 @item -Wshadow=compatible-local
6487 @opindex Wshadow=compatible-local
6488 Warn when a local variable shadows another local variable or parameter
6489 whose type is compatible with that of the shadowing variable. In C++,
6490 type compatibility here means the type of the shadowing variable can be
6491 converted to that of the shadowed variable. The creation of this flag
6492 (in addition to @option{-Wshadow=local}) is based on the idea that when
6493 a local variable shadows another one of incompatible type, it is most
6494 likely intentional, not a bug or typo, as shown in the following example:
6498 for (SomeIterator i = SomeObj.begin(); i != SomeObj.end(); ++i)
6500 for (int i = 0; i < N; ++i)
6509 Since the two variable @code{i} in the example above have incompatible types,
6510 enabling only @option{-Wshadow=compatible-local} will not emit a warning.
6511 Because their types are incompatible, if a programmer accidentally uses one
6512 in place of the other, type checking will catch that and emit an error or
6513 warning. So not warning (about shadowing) in this case will not lead to
6514 undetected bugs. Use of this flag instead of @option{-Wshadow=local} can
6515 possibly reduce the number of warnings triggered by intentional shadowing.
6517 This warning is enabled by @option{-Wshadow=local}.
6519 @item -Wlarger-than=@var{byte-size}
6520 @opindex Wlarger-than=
6521 @opindex Wlarger-than-@var{byte-size}
6522 Warn whenever an object is defined whose size exceeds @var{byte-size}.
6523 @option{-Wlarger-than=}@samp{PTRDIFF_MAX} is enabled by default.
6524 Warnings controlled by the option can be disabled either by specifying
6525 @var{byte-size} of @samp{SIZE_MAX} or more or by
6526 @option{-Wno-larger-than}.
6528 @item -Wno-larger-than
6529 @opindex Wno-larger-than
6530 Disable @option{-Wlarger-than=} warnings. The option is equivalent
6531 to @option{-Wlarger-than=}@samp{SIZE_MAX} or larger.
6533 @item -Wframe-larger-than=@var{byte-size}
6534 @opindex Wframe-larger-than=
6535 @opindex Wno-frame-larger-than
6536 Warn if the size of a function frame exceeds @var{byte-size}.
6537 The computation done to determine the stack frame size is approximate
6538 and not conservative.
6539 The actual requirements may be somewhat greater than @var{byte-size}
6540 even if you do not get a warning. In addition, any space allocated
6541 via @code{alloca}, variable-length arrays, or related constructs
6542 is not included by the compiler when determining
6543 whether or not to issue a warning.
6544 @option{-Wframe-larger-than=}@samp{PTRDIFF_MAX} is enabled by default.
6545 Warnings controlled by the option can be disabled either by specifying
6546 @var{byte-size} of @samp{SIZE_MAX} or more or by
6547 @option{-Wno-frame-larger-than}.
6549 @item -Wno-frame-larger-than
6550 @opindex Wno-frame-larger-than
6551 Disable @option{-Wframe-larger-than=} warnings. The option is equivalent
6552 to @option{-Wframe-larger-than=}@samp{SIZE_MAX} or larger.
6554 @item -Wno-free-nonheap-object
6555 @opindex Wno-free-nonheap-object
6556 @opindex Wfree-nonheap-object
6557 Do not warn when attempting to free an object that was not allocated
6560 @item -Wstack-usage=@var{byte-size}
6561 @opindex Wstack-usage
6562 @opindex Wno-stack-usage
6563 Warn if the stack usage of a function might exceed @var{byte-size}.
6564 The computation done to determine the stack usage is conservative.
6565 Any space allocated via @code{alloca}, variable-length arrays, or related
6566 constructs is included by the compiler when determining whether or not to
6569 The message is in keeping with the output of @option{-fstack-usage}.
6573 If the stack usage is fully static but exceeds the specified amount, it's:
6576 warning: stack usage is 1120 bytes
6579 If the stack usage is (partly) dynamic but bounded, it's:
6582 warning: stack usage might be 1648 bytes
6585 If the stack usage is (partly) dynamic and not bounded, it's:
6588 warning: stack usage might be unbounded
6592 @option{-Wstack-usage=}@samp{PTRDIFF_MAX} is enabled by default.
6593 Warnings controlled by the option can be disabled either by specifying
6594 @var{byte-size} of @samp{SIZE_MAX} or more or by
6595 @option{-Wno-stack-usage}.
6597 @item -Wno-stack-usage
6598 @opindex Wno-stack-usage
6599 Disable @option{-Wstack-usage=} warnings. The option is equivalent
6600 to @option{-Wstack-usage=}@samp{SIZE_MAX} or larger.
6602 @item -Wunsafe-loop-optimizations
6603 @opindex Wunsafe-loop-optimizations
6604 @opindex Wno-unsafe-loop-optimizations
6605 Warn if the loop cannot be optimized because the compiler cannot
6606 assume anything on the bounds of the loop indices. With
6607 @option{-funsafe-loop-optimizations} warn if the compiler makes
6610 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
6611 @opindex Wno-pedantic-ms-format
6612 @opindex Wpedantic-ms-format
6613 When used in combination with @option{-Wformat}
6614 and @option{-pedantic} without GNU extensions, this option
6615 disables the warnings about non-ISO @code{printf} / @code{scanf} format
6616 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets,
6617 which depend on the MS runtime.
6620 @opindex Waligned-new
6621 @opindex Wno-aligned-new
6622 Warn about a new-expression of a type that requires greater alignment
6623 than the @code{alignof(std::max_align_t)} but uses an allocation
6624 function without an explicit alignment parameter. This option is
6625 enabled by @option{-Wall}.
6627 Normally this only warns about global allocation functions, but
6628 @option{-Waligned-new=all} also warns about class member allocation
6631 @item -Wplacement-new
6632 @itemx -Wplacement-new=@var{n}
6633 @opindex Wplacement-new
6634 @opindex Wno-placement-new
6635 Warn about placement new expressions with undefined behavior, such as
6636 constructing an object in a buffer that is smaller than the type of
6637 the object. For example, the placement new expression below is diagnosed
6638 because it attempts to construct an array of 64 integers in a buffer only
6644 This warning is enabled by default.
6647 @item -Wplacement-new=1
6648 This is the default warning level of @option{-Wplacement-new}. At this
6649 level the warning is not issued for some strictly undefined constructs that
6650 GCC allows as extensions for compatibility with legacy code. For example,
6651 the following @code{new} expression is not diagnosed at this level even
6652 though it has undefined behavior according to the C++ standard because
6653 it writes past the end of the one-element array.
6655 struct S @{ int n, a[1]; @};
6656 S *s = (S *)malloc (sizeof *s + 31 * sizeof s->a[0]);
6657 new (s->a)int [32]();
6660 @item -Wplacement-new=2
6661 At this level, in addition to diagnosing all the same constructs as at level
6662 1, a diagnostic is also issued for placement new expressions that construct
6663 an object in the last member of structure whose type is an array of a single
6664 element and whose size is less than the size of the object being constructed.
6665 While the previous example would be diagnosed, the following construct makes
6666 use of the flexible member array extension to avoid the warning at level 2.
6668 struct S @{ int n, a[]; @};
6669 S *s = (S *)malloc (sizeof *s + 32 * sizeof s->a[0]);
6670 new (s->a)int [32]();
6675 @item -Wpointer-arith
6676 @opindex Wpointer-arith
6677 @opindex Wno-pointer-arith
6678 Warn about anything that depends on the ``size of'' a function type or
6679 of @code{void}. GNU C assigns these types a size of 1, for
6680 convenience in calculations with @code{void *} pointers and pointers
6681 to functions. In C++, warn also when an arithmetic operation involves
6682 @code{NULL}. This warning is also enabled by @option{-Wpedantic}.
6684 @item -Wpointer-compare
6685 @opindex Wpointer-compare
6686 @opindex Wno-pointer-compare
6687 Warn if a pointer is compared with a zero character constant. This usually
6688 means that the pointer was meant to be dereferenced. For example:
6691 const char *p = foo ();
6696 Note that the code above is invalid in C++11.
6698 This warning is enabled by default.
6701 @opindex Wtype-limits
6702 @opindex Wno-type-limits
6703 Warn if a comparison is always true or always false due to the limited
6704 range of the data type, but do not warn for constant expressions. For
6705 example, warn if an unsigned variable is compared against zero with
6706 @code{<} or @code{>=}. This warning is also enabled by
6709 @item -Wabsolute-value @r{(C and Objective-C only)}
6710 @opindex Wabsolute-value
6711 @opindex Wno-absolute-value
6712 Warn for calls to standard functions that compute the absolute value
6713 of an argument when a more appropriate standard function is available.
6714 For example, calling @code{abs(3.14)} triggers the warning because the
6715 appropriate function to call to compute the absolute value of a double
6716 argument is @code{fabs}. The option also triggers warnings when the
6717 argument in a call to such a function has an unsigned type. This
6718 warning can be suppressed with an explicit type cast and it is also
6719 enabled by @option{-Wextra}.
6721 @include cppwarnopts.texi
6723 @item -Wbad-function-cast @r{(C and Objective-C only)}
6724 @opindex Wbad-function-cast
6725 @opindex Wno-bad-function-cast
6726 Warn when a function call is cast to a non-matching type.
6727 For example, warn if a call to a function returning an integer type
6728 is cast to a pointer type.
6730 @item -Wc90-c99-compat @r{(C and Objective-C only)}
6731 @opindex Wc90-c99-compat
6732 @opindex Wno-c90-c99-compat
6733 Warn about features not present in ISO C90, but present in ISO C99.
6734 For instance, warn about use of variable length arrays, @code{long long}
6735 type, @code{bool} type, compound literals, designated initializers, and so
6736 on. This option is independent of the standards mode. Warnings are disabled
6737 in the expression that follows @code{__extension__}.
6739 @item -Wc99-c11-compat @r{(C and Objective-C only)}
6740 @opindex Wc99-c11-compat
6741 @opindex Wno-c99-c11-compat
6742 Warn about features not present in ISO C99, but present in ISO C11.
6743 For instance, warn about use of anonymous structures and unions,
6744 @code{_Atomic} type qualifier, @code{_Thread_local} storage-class specifier,
6745 @code{_Alignas} specifier, @code{Alignof} operator, @code{_Generic} keyword,
6746 and so on. This option is independent of the standards mode. Warnings are
6747 disabled in the expression that follows @code{__extension__}.
6749 @item -Wc++-compat @r{(C and Objective-C only)}
6750 @opindex Wc++-compat
6751 @opindex Wno-c++-compat
6752 Warn about ISO C constructs that are outside of the common subset of
6753 ISO C and ISO C++, e.g.@: request for implicit conversion from
6754 @code{void *} to a pointer to non-@code{void} type.
6756 @item -Wc++11-compat @r{(C++ and Objective-C++ only)}
6757 @opindex Wc++11-compat
6758 @opindex Wno-c++11-compat
6759 Warn about C++ constructs whose meaning differs between ISO C++ 1998
6760 and ISO C++ 2011, e.g., identifiers in ISO C++ 1998 that are keywords
6761 in ISO C++ 2011. This warning turns on @option{-Wnarrowing} and is
6762 enabled by @option{-Wall}.
6764 @item -Wc++14-compat @r{(C++ and Objective-C++ only)}
6765 @opindex Wc++14-compat
6766 @opindex Wno-c++14-compat
6767 Warn about C++ constructs whose meaning differs between ISO C++ 2011
6768 and ISO C++ 2014. This warning is enabled by @option{-Wall}.
6770 @item -Wc++17-compat @r{(C++ and Objective-C++ only)}
6771 @opindex Wc++17-compat
6772 @opindex Wno-c++17-compat
6773 Warn about C++ constructs whose meaning differs between ISO C++ 2014
6774 and ISO C++ 2017. This warning is enabled by @option{-Wall}.
6778 @opindex Wno-cast-qual
6779 Warn whenever a pointer is cast so as to remove a type qualifier from
6780 the target type. For example, warn if a @code{const char *} is cast
6781 to an ordinary @code{char *}.
6783 Also warn when making a cast that introduces a type qualifier in an
6784 unsafe way. For example, casting @code{char **} to @code{const char **}
6785 is unsafe, as in this example:
6788 /* p is char ** value. */
6789 const char **q = (const char **) p;
6790 /* Assignment of readonly string to const char * is OK. */
6792 /* Now char** pointer points to read-only memory. */
6797 @opindex Wcast-align
6798 @opindex Wno-cast-align
6799 Warn whenever a pointer is cast such that the required alignment of the
6800 target is increased. For example, warn if a @code{char *} is cast to
6801 an @code{int *} on machines where integers can only be accessed at
6802 two- or four-byte boundaries.
6804 @item -Wcast-align=strict
6805 @opindex Wcast-align=strict
6806 Warn whenever a pointer is cast such that the required alignment of the
6807 target is increased. For example, warn if a @code{char *} is cast to
6808 an @code{int *} regardless of the target machine.
6810 @item -Wcast-function-type
6811 @opindex Wcast-function-type
6812 @opindex Wno-cast-function-type
6813 Warn when a function pointer is cast to an incompatible function pointer.
6814 In a cast involving function types with a variable argument list only
6815 the types of initial arguments that are provided are considered.
6816 Any parameter of pointer-type matches any other pointer-type. Any benign
6817 differences in integral types are ignored, like @code{int} vs.@: @code{long}
6818 on ILP32 targets. Likewise type qualifiers are ignored. The function
6819 type @code{void (*) (void)} is special and matches everything, which can
6820 be used to suppress this warning.
6821 In a cast involving pointer to member types this warning warns whenever
6822 the type cast is changing the pointer to member type.
6823 This warning is enabled by @option{-Wextra}.
6825 @item -Wwrite-strings
6826 @opindex Wwrite-strings
6827 @opindex Wno-write-strings
6828 When compiling C, give string constants the type @code{const
6829 char[@var{length}]} so that copying the address of one into a
6830 non-@code{const} @code{char *} pointer produces a warning. These
6831 warnings help you find at compile time code that can try to write
6832 into a string constant, but only if you have been very careful about
6833 using @code{const} in declarations and prototypes. Otherwise, it is
6834 just a nuisance. This is why we did not make @option{-Wall} request
6837 When compiling C++, warn about the deprecated conversion from string
6838 literals to @code{char *}. This warning is enabled by default for C++
6842 @itemx -Wcatch-value=@var{n} @r{(C++ and Objective-C++ only)}
6843 @opindex Wcatch-value
6844 @opindex Wno-catch-value
6845 Warn about catch handlers that do not catch via reference.
6846 With @option{-Wcatch-value=1} (or @option{-Wcatch-value} for short)
6847 warn about polymorphic class types that are caught by value.
6848 With @option{-Wcatch-value=2} warn about all class types that are caught
6849 by value. With @option{-Wcatch-value=3} warn about all types that are
6850 not caught by reference. @option{-Wcatch-value} is enabled by @option{-Wall}.
6854 @opindex Wno-clobbered
6855 Warn for variables that might be changed by @code{longjmp} or
6856 @code{vfork}. This warning is also enabled by @option{-Wextra}.
6858 @item -Wconditionally-supported @r{(C++ and Objective-C++ only)}
6859 @opindex Wconditionally-supported
6860 @opindex Wno-conditionally-supported
6861 Warn for conditionally-supported (C++11 [intro.defs]) constructs.
6864 @opindex Wconversion
6865 @opindex Wno-conversion
6866 Warn for implicit conversions that may alter a value. This includes
6867 conversions between real and integer, like @code{abs (x)} when
6868 @code{x} is @code{double}; conversions between signed and unsigned,
6869 like @code{unsigned ui = -1}; and conversions to smaller types, like
6870 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
6871 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
6872 changed by the conversion like in @code{abs (2.0)}. Warnings about
6873 conversions between signed and unsigned integers can be disabled by
6874 using @option{-Wno-sign-conversion}.
6876 For C++, also warn for confusing overload resolution for user-defined
6877 conversions; and conversions that never use a type conversion
6878 operator: conversions to @code{void}, the same type, a base class or a
6879 reference to them. Warnings about conversions between signed and
6880 unsigned integers are disabled by default in C++ unless
6881 @option{-Wsign-conversion} is explicitly enabled.
6883 @item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
6884 @opindex Wconversion-null
6885 @opindex Wno-conversion-null
6886 Do not warn for conversions between @code{NULL} and non-pointer
6887 types. @option{-Wconversion-null} is enabled by default.
6889 @item -Wzero-as-null-pointer-constant @r{(C++ and Objective-C++ only)}
6890 @opindex Wzero-as-null-pointer-constant
6891 @opindex Wno-zero-as-null-pointer-constant
6892 Warn when a literal @samp{0} is used as null pointer constant. This can
6893 be useful to facilitate the conversion to @code{nullptr} in C++11.
6895 @item -Wsubobject-linkage @r{(C++ and Objective-C++ only)}
6896 @opindex Wsubobject-linkage
6897 @opindex Wno-subobject-linkage
6898 Warn if a class type has a base or a field whose type uses the anonymous
6899 namespace or depends on a type with no linkage. If a type A depends on
6900 a type B with no or internal linkage, defining it in multiple
6901 translation units would be an ODR violation because the meaning of B
6902 is different in each translation unit. If A only appears in a single
6903 translation unit, the best way to silence the warning is to give it
6904 internal linkage by putting it in an anonymous namespace as well. The
6905 compiler doesn't give this warning for types defined in the main .C
6906 file, as those are unlikely to have multiple definitions.
6907 @option{-Wsubobject-linkage} is enabled by default.
6909 @item -Wdangling-else
6910 @opindex Wdangling-else
6911 @opindex Wno-dangling-else
6912 Warn about constructions where there may be confusion to which
6913 @code{if} statement an @code{else} branch belongs. Here is an example of
6928 In C/C++, every @code{else} branch belongs to the innermost possible
6929 @code{if} statement, which in this example is @code{if (b)}. This is
6930 often not what the programmer expected, as illustrated in the above
6931 example by indentation the programmer chose. When there is the
6932 potential for this confusion, GCC issues a warning when this flag
6933 is specified. To eliminate the warning, add explicit braces around
6934 the innermost @code{if} statement so there is no way the @code{else}
6935 can belong to the enclosing @code{if}. The resulting code
6952 This warning is enabled by @option{-Wparentheses}.
6956 @opindex Wno-date-time
6957 Warn when macros @code{__TIME__}, @code{__DATE__} or @code{__TIMESTAMP__}
6958 are encountered as they might prevent bit-wise-identical reproducible
6961 @item -Wdelete-incomplete @r{(C++ and Objective-C++ only)}
6962 @opindex Wdelete-incomplete
6963 @opindex Wno-delete-incomplete
6964 Warn when deleting a pointer to incomplete type, which may cause
6965 undefined behavior at runtime. This warning is enabled by default.
6967 @item -Wuseless-cast @r{(C++ and Objective-C++ only)}
6968 @opindex Wuseless-cast
6969 @opindex Wno-useless-cast
6970 Warn when an expression is casted to its own type.
6973 @opindex Wempty-body
6974 @opindex Wno-empty-body
6975 Warn if an empty body occurs in an @code{if}, @code{else} or @code{do
6976 while} statement. This warning is also enabled by @option{-Wextra}.
6978 @item -Wenum-compare
6979 @opindex Wenum-compare
6980 @opindex Wno-enum-compare
6981 Warn about a comparison between values of different enumerated types.
6982 In C++ enumerated type mismatches in conditional expressions are also
6983 diagnosed and the warning is enabled by default. In C this warning is
6984 enabled by @option{-Wall}.
6986 @item -Wextra-semi @r{(C++, Objective-C++ only)}
6987 @opindex Wextra-semi
6988 @opindex Wno-extra-semi
6989 Warn about redundant semicolon after in-class function definition.
6991 @item -Wjump-misses-init @r{(C, Objective-C only)}
6992 @opindex Wjump-misses-init
6993 @opindex Wno-jump-misses-init
6994 Warn if a @code{goto} statement or a @code{switch} statement jumps
6995 forward across the initialization of a variable, or jumps backward to a
6996 label after the variable has been initialized. This only warns about
6997 variables that are initialized when they are declared. This warning is
6998 only supported for C and Objective-C; in C++ this sort of branch is an
7001 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
7002 can be disabled with the @option{-Wno-jump-misses-init} option.
7004 @item -Wsign-compare
7005 @opindex Wsign-compare
7006 @opindex Wno-sign-compare
7007 @cindex warning for comparison of signed and unsigned values
7008 @cindex comparison of signed and unsigned values, warning
7009 @cindex signed and unsigned values, comparison warning
7010 Warn when a comparison between signed and unsigned values could produce
7011 an incorrect result when the signed value is converted to unsigned.
7012 In C++, this warning is also enabled by @option{-Wall}. In C, it is
7013 also enabled by @option{-Wextra}.
7015 @item -Wsign-conversion
7016 @opindex Wsign-conversion
7017 @opindex Wno-sign-conversion
7018 Warn for implicit conversions that may change the sign of an integer
7019 value, like assigning a signed integer expression to an unsigned
7020 integer variable. An explicit cast silences the warning. In C, this
7021 option is enabled also by @option{-Wconversion}.
7023 @item -Wfloat-conversion
7024 @opindex Wfloat-conversion
7025 @opindex Wno-float-conversion
7026 Warn for implicit conversions that reduce the precision of a real value.
7027 This includes conversions from real to integer, and from higher precision
7028 real to lower precision real values. This option is also enabled by
7029 @option{-Wconversion}.
7031 @item -Wno-scalar-storage-order
7032 @opindex Wno-scalar-storage-order
7033 @opindex Wscalar-storage-order
7034 Do not warn on suspicious constructs involving reverse scalar storage order.
7036 @item -Wsized-deallocation @r{(C++ and Objective-C++ only)}
7037 @opindex Wsized-deallocation
7038 @opindex Wno-sized-deallocation
7039 Warn about a definition of an unsized deallocation function
7041 void operator delete (void *) noexcept;
7042 void operator delete[] (void *) noexcept;
7044 without a definition of the corresponding sized deallocation function
7046 void operator delete (void *, std::size_t) noexcept;
7047 void operator delete[] (void *, std::size_t) noexcept;
7049 or vice versa. Enabled by @option{-Wextra} along with
7050 @option{-fsized-deallocation}.
7052 @item -Wsizeof-pointer-div
7053 @opindex Wsizeof-pointer-div
7054 @opindex Wno-sizeof-pointer-div
7055 Warn for suspicious divisions of two sizeof expressions that divide
7056 the pointer size by the element size, which is the usual way to compute
7057 the array size but won't work out correctly with pointers. This warning
7058 warns e.g.@: about @code{sizeof (ptr) / sizeof (ptr[0])} if @code{ptr} is
7059 not an array, but a pointer. This warning is enabled by @option{-Wall}.
7061 @item -Wsizeof-pointer-memaccess
7062 @opindex Wsizeof-pointer-memaccess
7063 @opindex Wno-sizeof-pointer-memaccess
7064 Warn for suspicious length parameters to certain string and memory built-in
7065 functions if the argument uses @code{sizeof}. This warning triggers for
7066 example for @code{memset (ptr, 0, sizeof (ptr));} if @code{ptr} is not
7067 an array, but a pointer, and suggests a possible fix, or about
7068 @code{memcpy (&foo, ptr, sizeof (&foo));}. @option{-Wsizeof-pointer-memaccess}
7069 also warns about calls to bounded string copy functions like @code{strncat}
7070 or @code{strncpy} that specify as the bound a @code{sizeof} expression of
7071 the source array. For example, in the following function the call to
7072 @code{strncat} specifies the size of the source string as the bound. That
7073 is almost certainly a mistake and so the call is diagnosed.
7075 void make_file (const char *name)
7077 char path[PATH_MAX];
7078 strncpy (path, name, sizeof path - 1);
7079 strncat (path, ".text", sizeof ".text");
7084 The @option{-Wsizeof-pointer-memaccess} option is enabled by @option{-Wall}.
7086 @item -Wsizeof-array-argument
7087 @opindex Wsizeof-array-argument
7088 @opindex Wno-sizeof-array-argument
7089 Warn when the @code{sizeof} operator is applied to a parameter that is
7090 declared as an array in a function definition. This warning is enabled by
7091 default for C and C++ programs.
7093 @item -Wmemset-elt-size
7094 @opindex Wmemset-elt-size
7095 @opindex Wno-memset-elt-size
7096 Warn for suspicious calls to the @code{memset} built-in function, if the
7097 first argument references an array, and the third argument is a number
7098 equal to the number of elements, but not equal to the size of the array
7099 in memory. This indicates that the user has omitted a multiplication by
7100 the element size. This warning is enabled by @option{-Wall}.
7102 @item -Wmemset-transposed-args
7103 @opindex Wmemset-transposed-args
7104 @opindex Wno-memset-transposed-args
7105 Warn for suspicious calls to the @code{memset} built-in function where
7106 the second argument is not zero and the third argument is zero. For
7107 example, the call @code{memset (buf, sizeof buf, 0)} is diagnosed because
7108 @code{memset (buf, 0, sizeof buf)} was meant instead. The diagnostic
7109 is only emitted if the third argument is a literal zero. Otherwise, if
7110 it is an expression that is folded to zero, or a cast of zero to some
7111 type, it is far less likely that the arguments have been mistakenly
7112 transposed and no warning is emitted. This warning is enabled
7117 @opindex Wno-address
7118 Warn about suspicious uses of memory addresses. These include using
7119 the address of a function in a conditional expression, such as
7120 @code{void func(void); if (func)}, and comparisons against the memory
7121 address of a string literal, such as @code{if (x == "abc")}. Such
7122 uses typically indicate a programmer error: the address of a function
7123 always evaluates to true, so their use in a conditional usually
7124 indicate that the programmer forgot the parentheses in a function
7125 call; and comparisons against string literals result in unspecified
7126 behavior and are not portable in C, so they usually indicate that the
7127 programmer intended to use @code{strcmp}. This warning is enabled by
7130 @item -Waddress-of-packed-member
7131 @opindex Waddress-of-packed-member
7132 @opindex Wno-address-of-packed-member
7133 Warn when the address of packed member of struct or union is taken,
7134 which usually results in an unaligned pointer value. This is
7138 @opindex Wlogical-op
7139 @opindex Wno-logical-op
7140 Warn about suspicious uses of logical operators in expressions.
7141 This includes using logical operators in contexts where a
7142 bit-wise operator is likely to be expected. Also warns when
7143 the operands of a logical operator are the same:
7146 if (a < 0 && a < 0) @{ @dots{} @}
7149 @item -Wlogical-not-parentheses
7150 @opindex Wlogical-not-parentheses
7151 @opindex Wno-logical-not-parentheses
7152 Warn about logical not used on the left hand side operand of a comparison.
7153 This option does not warn if the right operand is considered to be a boolean
7154 expression. Its purpose is to detect suspicious code like the following:
7158 if (!a > 1) @{ @dots{} @}
7161 It is possible to suppress the warning by wrapping the LHS into
7164 if ((!a) > 1) @{ @dots{} @}
7167 This warning is enabled by @option{-Wall}.
7169 @item -Waggregate-return
7170 @opindex Waggregate-return
7171 @opindex Wno-aggregate-return
7172 Warn if any functions that return structures or unions are defined or
7173 called. (In languages where you can return an array, this also elicits
7176 @item -Wno-aggressive-loop-optimizations
7177 @opindex Wno-aggressive-loop-optimizations
7178 @opindex Waggressive-loop-optimizations
7179 Warn if in a loop with constant number of iterations the compiler detects
7180 undefined behavior in some statement during one or more of the iterations.
7182 @item -Wno-attributes
7183 @opindex Wno-attributes
7184 @opindex Wattributes
7185 Do not warn if an unexpected @code{__attribute__} is used, such as
7186 unrecognized attributes, function attributes applied to variables,
7187 etc. This does not stop errors for incorrect use of supported
7190 @item -Wno-builtin-declaration-mismatch
7191 @opindex Wno-builtin-declaration-mismatch
7192 @opindex Wbuiltin-declaration-mismatch
7193 Warn if a built-in function is declared with an incompatible signature
7194 or as a non-function, or when a built-in function declared with a type
7195 that does not include a prototype is called with arguments whose promoted
7196 types do not match those expected by the function. When @option{-Wextra}
7197 is specified, also warn when a built-in function that takes arguments is
7198 declared without a prototype. The @option{-Wno-builtin-declaration-mismatch}
7199 warning is enabled by default. To avoid the warning include the appropriate
7200 header to bring the prototypes of built-in functions into scope.
7202 For example, the call to @code{memset} below is diagnosed by the warning
7203 because the function expects a value of type @code{size_t} as its argument
7204 but the type of @code{32} is @code{int}. With @option{-Wextra},
7205 the declaration of the function is diagnosed as well.
7207 extern void* memset ();
7210 memset (d, '\0', 32);
7214 @item -Wno-builtin-macro-redefined
7215 @opindex Wno-builtin-macro-redefined
7216 @opindex Wbuiltin-macro-redefined
7217 Do not warn if certain built-in macros are redefined. This suppresses
7218 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
7219 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
7221 @item -Wstrict-prototypes @r{(C and Objective-C only)}
7222 @opindex Wstrict-prototypes
7223 @opindex Wno-strict-prototypes
7224 Warn if a function is declared or defined without specifying the
7225 argument types. (An old-style function definition is permitted without
7226 a warning if preceded by a declaration that specifies the argument
7229 @item -Wold-style-declaration @r{(C and Objective-C only)}
7230 @opindex Wold-style-declaration
7231 @opindex Wno-old-style-declaration
7232 Warn for obsolescent usages, according to the C Standard, in a
7233 declaration. For example, warn if storage-class specifiers like
7234 @code{static} are not the first things in a declaration. This warning
7235 is also enabled by @option{-Wextra}.
7237 @item -Wold-style-definition @r{(C and Objective-C only)}
7238 @opindex Wold-style-definition
7239 @opindex Wno-old-style-definition
7240 Warn if an old-style function definition is used. A warning is given
7241 even if there is a previous prototype.
7243 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
7244 @opindex Wmissing-parameter-type
7245 @opindex Wno-missing-parameter-type
7246 A function parameter is declared without a type specifier in K&R-style
7253 This warning is also enabled by @option{-Wextra}.
7255 @item -Wmissing-prototypes @r{(C and Objective-C only)}
7256 @opindex Wmissing-prototypes
7257 @opindex Wno-missing-prototypes
7258 Warn if a global function is defined without a previous prototype
7259 declaration. This warning is issued even if the definition itself
7260 provides a prototype. Use this option to detect global functions
7261 that do not have a matching prototype declaration in a header file.
7262 This option is not valid for C++ because all function declarations
7263 provide prototypes and a non-matching declaration declares an
7264 overload rather than conflict with an earlier declaration.
7265 Use @option{-Wmissing-declarations} to detect missing declarations in C++.
7267 @item -Wmissing-declarations
7268 @opindex Wmissing-declarations
7269 @opindex Wno-missing-declarations
7270 Warn if a global function is defined without a previous declaration.
7271 Do so even if the definition itself provides a prototype.
7272 Use this option to detect global functions that are not declared in
7273 header files. In C, no warnings are issued for functions with previous
7274 non-prototype declarations; use @option{-Wmissing-prototypes} to detect
7275 missing prototypes. In C++, no warnings are issued for function templates,
7276 or for inline functions, or for functions in anonymous namespaces.
7278 @item -Wmissing-field-initializers
7279 @opindex Wmissing-field-initializers
7280 @opindex Wno-missing-field-initializers
7284 Warn if a structure's initializer has some fields missing. For
7285 example, the following code causes such a warning, because
7286 @code{x.h} is implicitly zero:
7289 struct s @{ int f, g, h; @};
7290 struct s x = @{ 3, 4 @};
7293 This option does not warn about designated initializers, so the following
7294 modification does not trigger a warning:
7297 struct s @{ int f, g, h; @};
7298 struct s x = @{ .f = 3, .g = 4 @};
7301 In C this option does not warn about the universal zero initializer
7305 struct s @{ int f, g, h; @};
7306 struct s x = @{ 0 @};
7309 Likewise, in C++ this option does not warn about the empty @{ @}
7310 initializer, for example:
7313 struct s @{ int f, g, h; @};
7317 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
7318 warnings without this one, use @option{-Wextra -Wno-missing-field-initializers}.
7320 @item -Wno-multichar
7321 @opindex Wno-multichar
7323 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
7324 Usually they indicate a typo in the user's code, as they have
7325 implementation-defined values, and should not be used in portable code.
7327 @item -Wnormalized=@r{[}none@r{|}id@r{|}nfc@r{|}nfkc@r{]}
7328 @opindex Wnormalized=
7329 @opindex Wnormalized
7330 @opindex Wno-normalized
7333 @cindex character set, input normalization
7334 In ISO C and ISO C++, two identifiers are different if they are
7335 different sequences of characters. However, sometimes when characters
7336 outside the basic ASCII character set are used, you can have two
7337 different character sequences that look the same. To avoid confusion,
7338 the ISO 10646 standard sets out some @dfn{normalization rules} which
7339 when applied ensure that two sequences that look the same are turned into
7340 the same sequence. GCC can warn you if you are using identifiers that
7341 have not been normalized; this option controls that warning.
7343 There are four levels of warning supported by GCC@. The default is
7344 @option{-Wnormalized=nfc}, which warns about any identifier that is
7345 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
7346 recommended form for most uses. It is equivalent to
7347 @option{-Wnormalized}.
7349 Unfortunately, there are some characters allowed in identifiers by
7350 ISO C and ISO C++ that, when turned into NFC, are not allowed in
7351 identifiers. That is, there's no way to use these symbols in portable
7352 ISO C or C++ and have all your identifiers in NFC@.
7353 @option{-Wnormalized=id} suppresses the warning for these characters.
7354 It is hoped that future versions of the standards involved will correct
7355 this, which is why this option is not the default.
7357 You can switch the warning off for all characters by writing
7358 @option{-Wnormalized=none} or @option{-Wno-normalized}. You should
7359 only do this if you are using some other normalization scheme (like
7360 ``D''), because otherwise you can easily create bugs that are
7361 literally impossible to see.
7363 Some characters in ISO 10646 have distinct meanings but look identical
7364 in some fonts or display methodologies, especially once formatting has
7365 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
7366 LETTER N'', displays just like a regular @code{n} that has been
7367 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
7368 normalization scheme to convert all these into a standard form as
7369 well, and GCC warns if your code is not in NFKC if you use
7370 @option{-Wnormalized=nfkc}. This warning is comparable to warning
7371 about every identifier that contains the letter O because it might be
7372 confused with the digit 0, and so is not the default, but may be
7373 useful as a local coding convention if the programming environment
7374 cannot be fixed to display these characters distinctly.
7376 @item -Wno-attribute-warning
7377 @opindex Wno-attribute-warning
7378 @opindex Wattribute-warning
7379 Do not warn about usage of functions (@pxref{Function Attributes})
7380 declared with @code{warning} attribute. By default, this warning is
7381 enabled. @option{-Wno-attribute-warning} can be used to disable the
7382 warning or @option{-Wno-error=attribute-warning} can be used to
7383 disable the error when compiled with @option{-Werror} flag.
7385 @item -Wno-deprecated
7386 @opindex Wno-deprecated
7387 @opindex Wdeprecated
7388 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
7390 @item -Wno-deprecated-declarations
7391 @opindex Wno-deprecated-declarations
7392 @opindex Wdeprecated-declarations
7393 Do not warn about uses of functions (@pxref{Function Attributes}),
7394 variables (@pxref{Variable Attributes}), and types (@pxref{Type
7395 Attributes}) marked as deprecated by using the @code{deprecated}
7399 @opindex Wno-overflow
7401 Do not warn about compile-time overflow in constant expressions.
7406 Warn about One Definition Rule violations during link-time optimization.
7410 @opindex Wopenmp-simd
7411 @opindex Wno-openmp-simd
7412 Warn if the vectorizer cost model overrides the OpenMP
7413 simd directive set by user. The @option{-fsimd-cost-model=unlimited}
7414 option can be used to relax the cost model.
7416 @item -Woverride-init @r{(C and Objective-C only)}
7417 @opindex Woverride-init
7418 @opindex Wno-override-init
7422 Warn if an initialized field without side effects is overridden when
7423 using designated initializers (@pxref{Designated Inits, , Designated
7426 This warning is included in @option{-Wextra}. To get other
7427 @option{-Wextra} warnings without this one, use @option{-Wextra
7428 -Wno-override-init}.
7430 @item -Woverride-init-side-effects @r{(C and Objective-C only)}
7431 @opindex Woverride-init-side-effects
7432 @opindex Wno-override-init-side-effects
7433 Warn if an initialized field with side effects is overridden when
7434 using designated initializers (@pxref{Designated Inits, , Designated
7435 Initializers}). This warning is enabled by default.
7440 Warn if a structure is given the packed attribute, but the packed
7441 attribute has no effect on the layout or size of the structure.
7442 Such structures may be mis-aligned for little benefit. For
7443 instance, in this code, the variable @code{f.x} in @code{struct bar}
7444 is misaligned even though @code{struct bar} does not itself
7445 have the packed attribute:
7452 @} __attribute__((packed));
7460 @item -Wpacked-bitfield-compat
7461 @opindex Wpacked-bitfield-compat
7462 @opindex Wno-packed-bitfield-compat
7463 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
7464 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
7465 the change can lead to differences in the structure layout. GCC
7466 informs you when the offset of such a field has changed in GCC 4.4.
7467 For example there is no longer a 4-bit padding between field @code{a}
7468 and @code{b} in this structure:
7475 @} __attribute__ ((packed));
7478 This warning is enabled by default. Use
7479 @option{-Wno-packed-bitfield-compat} to disable this warning.
7481 @item -Wpacked-not-aligned @r{(C, C++, Objective-C and Objective-C++ only)}
7482 @opindex Wpacked-not-aligned
7483 @opindex Wno-packed-not-aligned
7484 Warn if a structure field with explicitly specified alignment in a
7485 packed struct or union is misaligned. For example, a warning will
7486 be issued on @code{struct S}, like, @code{warning: alignment 1 of
7487 'struct S' is less than 8}, in this code:
7491 struct __attribute__ ((aligned (8))) S8 @{ char a[8]; @};
7492 struct __attribute__ ((packed)) S @{
7498 This warning is enabled by @option{-Wall}.
7503 Warn if padding is included in a structure, either to align an element
7504 of the structure or to align the whole structure. Sometimes when this
7505 happens it is possible to rearrange the fields of the structure to
7506 reduce the padding and so make the structure smaller.
7508 @item -Wredundant-decls
7509 @opindex Wredundant-decls
7510 @opindex Wno-redundant-decls
7511 Warn if anything is declared more than once in the same scope, even in
7512 cases where multiple declaration is valid and changes nothing.
7516 @opindex Wno-restrict
7517 Warn when an object referenced by a @code{restrict}-qualified parameter
7518 (or, in C++, a @code{__restrict}-qualified parameter) is aliased by another
7519 argument, or when copies between such objects overlap. For example,
7520 the call to the @code{strcpy} function below attempts to truncate the string
7521 by replacing its initial characters with the last four. However, because
7522 the call writes the terminating NUL into @code{a[4]}, the copies overlap and
7523 the call is diagnosed.
7528 char a[] = "abcd1234";
7533 The @option{-Wrestrict} option detects some instances of simple overlap
7534 even without optimization but works best at @option{-O2} and above. It
7535 is included in @option{-Wall}.
7537 @item -Wnested-externs @r{(C and Objective-C only)}
7538 @opindex Wnested-externs
7539 @opindex Wno-nested-externs
7540 Warn if an @code{extern} declaration is encountered within a function.
7542 @item -Wno-inherited-variadic-ctor
7543 @opindex Winherited-variadic-ctor
7544 @opindex Wno-inherited-variadic-ctor
7545 Suppress warnings about use of C++11 inheriting constructors when the
7546 base class inherited from has a C variadic constructor; the warning is
7547 on by default because the ellipsis is not inherited.
7552 Warn if a function that is declared as inline cannot be inlined.
7553 Even with this option, the compiler does not warn about failures to
7554 inline functions declared in system headers.
7556 The compiler uses a variety of heuristics to determine whether or not
7557 to inline a function. For example, the compiler takes into account
7558 the size of the function being inlined and the amount of inlining
7559 that has already been done in the current function. Therefore,
7560 seemingly insignificant changes in the source program can cause the
7561 warnings produced by @option{-Winline} to appear or disappear.
7563 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
7564 @opindex Wno-invalid-offsetof
7565 @opindex Winvalid-offsetof
7566 Suppress warnings from applying the @code{offsetof} macro to a non-POD
7567 type. According to the 2014 ISO C++ standard, applying @code{offsetof}
7568 to a non-standard-layout type is undefined. In existing C++ implementations,
7569 however, @code{offsetof} typically gives meaningful results.
7570 This flag is for users who are aware that they are
7571 writing nonportable code and who have deliberately chosen to ignore the
7574 The restrictions on @code{offsetof} may be relaxed in a future version
7575 of the C++ standard.
7577 @item -Wint-in-bool-context
7578 @opindex Wint-in-bool-context
7579 @opindex Wno-int-in-bool-context
7580 Warn for suspicious use of integer values where boolean values are expected,
7581 such as conditional expressions (?:) using non-boolean integer constants in
7582 boolean context, like @code{if (a <= b ? 2 : 3)}. Or left shifting of signed
7583 integers in boolean context, like @code{for (a = 0; 1 << a; a++);}. Likewise
7584 for all kinds of multiplications regardless of the data type.
7585 This warning is enabled by @option{-Wall}.
7587 @item -Wno-int-to-pointer-cast
7588 @opindex Wno-int-to-pointer-cast
7589 @opindex Wint-to-pointer-cast
7590 Suppress warnings from casts to pointer type of an integer of a
7591 different size. In C++, casting to a pointer type of smaller size is
7592 an error. @option{Wint-to-pointer-cast} is enabled by default.
7595 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
7596 @opindex Wno-pointer-to-int-cast
7597 @opindex Wpointer-to-int-cast
7598 Suppress warnings from casts from a pointer to an integer type of a
7602 @opindex Winvalid-pch
7603 @opindex Wno-invalid-pch
7604 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
7605 the search path but cannot be used.
7609 @opindex Wno-long-long
7610 Warn if @code{long long} type is used. This is enabled by either
7611 @option{-Wpedantic} or @option{-Wtraditional} in ISO C90 and C++98
7612 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
7614 @item -Wvariadic-macros
7615 @opindex Wvariadic-macros
7616 @opindex Wno-variadic-macros
7617 Warn if variadic macros are used in ISO C90 mode, or if the GNU
7618 alternate syntax is used in ISO C99 mode. This is enabled by either
7619 @option{-Wpedantic} or @option{-Wtraditional}. To inhibit the warning
7620 messages, use @option{-Wno-variadic-macros}.
7624 @opindex Wno-varargs
7625 Warn upon questionable usage of the macros used to handle variable
7626 arguments like @code{va_start}. This is default. To inhibit the
7627 warning messages, use @option{-Wno-varargs}.
7629 @item -Wvector-operation-performance
7630 @opindex Wvector-operation-performance
7631 @opindex Wno-vector-operation-performance
7632 Warn if vector operation is not implemented via SIMD capabilities of the
7633 architecture. Mainly useful for the performance tuning.
7634 Vector operation can be implemented @code{piecewise}, which means that the
7635 scalar operation is performed on every vector element;
7636 @code{in parallel}, which means that the vector operation is implemented
7637 using scalars of wider type, which normally is more performance efficient;
7638 and @code{as a single scalar}, which means that vector fits into a
7641 @item -Wno-virtual-move-assign
7642 @opindex Wvirtual-move-assign
7643 @opindex Wno-virtual-move-assign
7644 Suppress warnings about inheriting from a virtual base with a
7645 non-trivial C++11 move assignment operator. This is dangerous because
7646 if the virtual base is reachable along more than one path, it is
7647 moved multiple times, which can mean both objects end up in the
7648 moved-from state. If the move assignment operator is written to avoid
7649 moving from a moved-from object, this warning can be disabled.
7654 Warn if a variable-length array is used in the code.
7655 @option{-Wno-vla} prevents the @option{-Wpedantic} warning of
7656 the variable-length array.
7658 @item -Wvla-larger-than=@var{byte-size}
7659 @opindex Wvla-larger-than=
7660 @opindex Wno-vla-larger-than
7661 If this option is used, the compiler will warn for declarations of
7662 variable-length arrays whose size is either unbounded, or bounded
7663 by an argument that allows the array size to exceed @var{byte-size}
7664 bytes. This is similar to how @option{-Walloca-larger-than=}@var{byte-size}
7665 works, but with variable-length arrays.
7667 Note that GCC may optimize small variable-length arrays of a known
7668 value into plain arrays, so this warning may not get triggered for
7671 @option{-Wvla-larger-than=}@samp{PTRDIFF_MAX} is enabled by default but
7672 is typically only effective when @option{-ftree-vrp} is active (default
7673 for @option{-O2} and above).
7675 See also @option{-Walloca-larger-than=@var{byte-size}}.
7677 @item -Wno-vla-larger-than
7678 @opindex Wno-vla-larger-than
7679 Disable @option{-Wvla-larger-than=} warnings. The option is equivalent
7680 to @option{-Wvla-larger-than=}@samp{SIZE_MAX} or larger.
7682 @item -Wvolatile-register-var
7683 @opindex Wvolatile-register-var
7684 @opindex Wno-volatile-register-var
7685 Warn if a register variable is declared volatile. The volatile
7686 modifier does not inhibit all optimizations that may eliminate reads
7687 and/or writes to register variables. This warning is enabled by
7690 @item -Wdisabled-optimization
7691 @opindex Wdisabled-optimization
7692 @opindex Wno-disabled-optimization
7693 Warn if a requested optimization pass is disabled. This warning does
7694 not generally indicate that there is anything wrong with your code; it
7695 merely indicates that GCC's optimizers are unable to handle the code
7696 effectively. Often, the problem is that your code is too big or too
7697 complex; GCC refuses to optimize programs when the optimization
7698 itself is likely to take inordinate amounts of time.
7700 @item -Wpointer-sign @r{(C and Objective-C only)}
7701 @opindex Wpointer-sign
7702 @opindex Wno-pointer-sign
7703 Warn for pointer argument passing or assignment with different signedness.
7704 This option is only supported for C and Objective-C@. It is implied by
7705 @option{-Wall} and by @option{-Wpedantic}, which can be disabled with
7706 @option{-Wno-pointer-sign}.
7708 @item -Wstack-protector
7709 @opindex Wstack-protector
7710 @opindex Wno-stack-protector
7711 This option is only active when @option{-fstack-protector} is active. It
7712 warns about functions that are not protected against stack smashing.
7714 @item -Woverlength-strings
7715 @opindex Woverlength-strings
7716 @opindex Wno-overlength-strings
7717 Warn about string constants that are longer than the ``minimum
7718 maximum'' length specified in the C standard. Modern compilers
7719 generally allow string constants that are much longer than the
7720 standard's minimum limit, but very portable programs should avoid
7721 using longer strings.
7723 The limit applies @emph{after} string constant concatenation, and does
7724 not count the trailing NUL@. In C90, the limit was 509 characters; in
7725 C99, it was raised to 4095. C++98 does not specify a normative
7726 minimum maximum, so we do not diagnose overlength strings in C++@.
7728 This option is implied by @option{-Wpedantic}, and can be disabled with
7729 @option{-Wno-overlength-strings}.
7731 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
7732 @opindex Wunsuffixed-float-constants
7733 @opindex Wno-unsuffixed-float-constants
7735 Issue a warning for any floating constant that does not have
7736 a suffix. When used together with @option{-Wsystem-headers} it
7737 warns about such constants in system header files. This can be useful
7738 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
7739 from the decimal floating-point extension to C99.
7741 @item -Wno-designated-init @r{(C and Objective-C only)}
7742 Suppress warnings when a positional initializer is used to initialize
7743 a structure that has been marked with the @code{designated_init}
7747 Issue a warning when HSAIL cannot be emitted for the compiled function or
7752 @node Debugging Options
7753 @section Options for Debugging Your Program
7754 @cindex options, debugging
7755 @cindex debugging information options
7757 To tell GCC to emit extra information for use by a debugger, in almost
7758 all cases you need only to add @option{-g} to your other options.
7760 GCC allows you to use @option{-g} with
7761 @option{-O}. The shortcuts taken by optimized code may occasionally
7762 be surprising: some variables you declared may not exist
7763 at all; flow of control may briefly move where you did not expect it;
7764 some statements may not be executed because they compute constant
7765 results or their values are already at hand; some statements may
7766 execute in different places because they have been moved out of loops.
7767 Nevertheless it is possible to debug optimized output. This makes
7768 it reasonable to use the optimizer for programs that might have bugs.
7770 If you are not using some other optimization option, consider
7771 using @option{-Og} (@pxref{Optimize Options}) with @option{-g}.
7772 With no @option{-O} option at all, some compiler passes that collect
7773 information useful for debugging do not run at all, so that
7774 @option{-Og} may result in a better debugging experience.
7779 Produce debugging information in the operating system's native format
7780 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
7783 On most systems that use stabs format, @option{-g} enables use of extra
7784 debugging information that only GDB can use; this extra information
7785 makes debugging work better in GDB but probably makes other debuggers
7787 refuse to read the program. If you want to control for certain whether
7788 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
7789 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
7793 Produce debugging information for use by GDB@. This means to use the
7794 most expressive format available (DWARF, stabs, or the native format
7795 if neither of those are supported), including GDB extensions if at all
7799 @itemx -gdwarf-@var{version}
7801 Produce debugging information in DWARF format (if that is supported).
7802 The value of @var{version} may be either 2, 3, 4 or 5; the default version
7803 for most targets is 4. DWARF Version 5 is only experimental.
7805 Note that with DWARF Version 2, some ports require and always
7806 use some non-conflicting DWARF 3 extensions in the unwind tables.
7808 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
7809 for maximum benefit.
7811 GCC no longer supports DWARF Version 1, which is substantially
7812 different than Version 2 and later. For historical reasons, some
7813 other DWARF-related options such as
7814 @option{-fno-dwarf2-cfi-asm}) retain a reference to DWARF Version 2
7815 in their names, but apply to all currently-supported versions of DWARF.
7819 Produce debugging information in stabs format (if that is supported),
7820 without GDB extensions. This is the format used by DBX on most BSD
7821 systems. On MIPS, Alpha and System V Release 4 systems this option
7822 produces stabs debugging output that is not understood by DBX@.
7823 On System V Release 4 systems this option requires the GNU assembler.
7827 Produce debugging information in stabs format (if that is supported),
7828 using GNU extensions understood only by the GNU debugger (GDB)@. The
7829 use of these extensions is likely to make other debuggers crash or
7830 refuse to read the program.
7834 Produce debugging information in XCOFF format (if that is supported).
7835 This is the format used by the DBX debugger on IBM RS/6000 systems.
7839 Produce debugging information in XCOFF format (if that is supported),
7840 using GNU extensions understood only by the GNU debugger (GDB)@. The
7841 use of these extensions is likely to make other debuggers crash or
7842 refuse to read the program, and may cause assemblers other than the GNU
7843 assembler (GAS) to fail with an error.
7847 Produce debugging information in Alpha/VMS debug format (if that is
7848 supported). This is the format used by DEBUG on Alpha/VMS systems.
7851 @itemx -ggdb@var{level}
7852 @itemx -gstabs@var{level}
7853 @itemx -gxcoff@var{level}
7854 @itemx -gvms@var{level}
7855 Request debugging information and also use @var{level} to specify how
7856 much information. The default level is 2.
7858 Level 0 produces no debug information at all. Thus, @option{-g0} negates
7861 Level 1 produces minimal information, enough for making backtraces in
7862 parts of the program that you don't plan to debug. This includes
7863 descriptions of functions and external variables, and line number
7864 tables, but no information about local variables.
7866 Level 3 includes extra information, such as all the macro definitions
7867 present in the program. Some debuggers support macro expansion when
7868 you use @option{-g3}.
7870 If you use multiple @option{-g} options, with or without level numbers,
7871 the last such option is the one that is effective.
7873 @option{-gdwarf} does not accept a concatenated debug level, to avoid
7874 confusion with @option{-gdwarf-@var{level}}.
7875 Instead use an additional @option{-g@var{level}} option to change the
7876 debug level for DWARF.
7878 @item -fno-eliminate-unused-debug-symbols
7879 @opindex feliminate-unused-debug-symbols
7880 @opindex fno-eliminate-unused-debug-symbols
7881 By default, no debug information is produced for symbols that are not actually
7882 used. Use this option if you want debug information for all symbols.
7884 @item -femit-class-debug-always
7885 @opindex femit-class-debug-always
7886 Instead of emitting debugging information for a C++ class in only one
7887 object file, emit it in all object files using the class. This option
7888 should be used only with debuggers that are unable to handle the way GCC
7889 normally emits debugging information for classes because using this
7890 option increases the size of debugging information by as much as a
7893 @item -fno-merge-debug-strings
7894 @opindex fmerge-debug-strings
7895 @opindex fno-merge-debug-strings
7896 Direct the linker to not merge together strings in the debugging
7897 information that are identical in different object files. Merging is
7898 not supported by all assemblers or linkers. Merging decreases the size
7899 of the debug information in the output file at the cost of increasing
7900 link processing time. Merging is enabled by default.
7902 @item -fdebug-prefix-map=@var{old}=@var{new}
7903 @opindex fdebug-prefix-map
7904 When compiling files residing in directory @file{@var{old}}, record
7905 debugging information describing them as if the files resided in
7906 directory @file{@var{new}} instead. This can be used to replace a
7907 build-time path with an install-time path in the debug info. It can
7908 also be used to change an absolute path to a relative path by using
7909 @file{.} for @var{new}. This can give more reproducible builds, which
7910 are location independent, but may require an extra command to tell GDB
7911 where to find the source files. See also @option{-ffile-prefix-map}.
7913 @item -fvar-tracking
7914 @opindex fvar-tracking
7915 Run variable tracking pass. It computes where variables are stored at each
7916 position in code. Better debugging information is then generated
7917 (if the debugging information format supports this information).
7919 It is enabled by default when compiling with optimization (@option{-Os},
7920 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
7921 the debug info format supports it.
7923 @item -fvar-tracking-assignments
7924 @opindex fvar-tracking-assignments
7925 @opindex fno-var-tracking-assignments
7926 Annotate assignments to user variables early in the compilation and
7927 attempt to carry the annotations over throughout the compilation all the
7928 way to the end, in an attempt to improve debug information while
7929 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
7931 It can be enabled even if var-tracking is disabled, in which case
7932 annotations are created and maintained, but discarded at the end.
7933 By default, this flag is enabled together with @option{-fvar-tracking},
7934 except when selective scheduling is enabled.
7937 @opindex gsplit-dwarf
7938 Separate as much DWARF debugging information as possible into a
7939 separate output file with the extension @file{.dwo}. This option allows
7940 the build system to avoid linking files with debug information. To
7941 be useful, this option requires a debugger capable of reading @file{.dwo}
7944 @item -gdescribe-dies
7945 @opindex gdescribe-dies
7946 Add description attributes to some DWARF DIEs that have no name attribute,
7947 such as artificial variables, external references and call site
7952 Generate DWARF @code{.debug_pubnames} and @code{.debug_pubtypes} sections.
7954 @item -ggnu-pubnames
7955 @opindex ggnu-pubnames
7956 Generate @code{.debug_pubnames} and @code{.debug_pubtypes} sections in a format
7957 suitable for conversion into a GDB@ index. This option is only useful
7958 with a linker that can produce GDB@ index version 7.
7960 @item -fdebug-types-section
7961 @opindex fdebug-types-section
7962 @opindex fno-debug-types-section
7963 When using DWARF Version 4 or higher, type DIEs can be put into
7964 their own @code{.debug_types} section instead of making them part of the
7965 @code{.debug_info} section. It is more efficient to put them in a separate
7966 comdat section since the linker can then remove duplicates.
7967 But not all DWARF consumers support @code{.debug_types} sections yet
7968 and on some objects @code{.debug_types} produces larger instead of smaller
7969 debugging information.
7971 @item -grecord-gcc-switches
7972 @itemx -gno-record-gcc-switches
7973 @opindex grecord-gcc-switches
7974 @opindex gno-record-gcc-switches
7975 This switch causes the command-line options used to invoke the
7976 compiler that may affect code generation to be appended to the
7977 DW_AT_producer attribute in DWARF debugging information. The options
7978 are concatenated with spaces separating them from each other and from
7979 the compiler version.
7980 It is enabled by default.
7981 See also @option{-frecord-gcc-switches} for another
7982 way of storing compiler options into the object file.
7984 @item -gstrict-dwarf
7985 @opindex gstrict-dwarf
7986 Disallow using extensions of later DWARF standard version than selected
7987 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
7988 DWARF extensions from later standard versions is allowed.
7990 @item -gno-strict-dwarf
7991 @opindex gno-strict-dwarf
7992 Allow using extensions of later DWARF standard version than selected with
7993 @option{-gdwarf-@var{version}}.
7995 @item -gas-loc-support
7996 @opindex gas-loc-support
7997 Inform the compiler that the assembler supports @code{.loc} directives.
7998 It may then use them for the assembler to generate DWARF2+ line number
8001 This is generally desirable, because assembler-generated line-number
8002 tables are a lot more compact than those the compiler can generate
8005 This option will be enabled by default if, at GCC configure time, the
8006 assembler was found to support such directives.
8008 @item -gno-as-loc-support
8009 @opindex gno-as-loc-support
8010 Force GCC to generate DWARF2+ line number tables internally, if DWARF2+
8011 line number tables are to be generated.
8013 @item -gas-locview-support
8014 @opindex gas-locview-support
8015 Inform the compiler that the assembler supports @code{view} assignment
8016 and reset assertion checking in @code{.loc} directives.
8018 This option will be enabled by default if, at GCC configure time, the
8019 assembler was found to support them.
8021 @item -gno-as-locview-support
8022 Force GCC to assign view numbers internally, if
8023 @option{-gvariable-location-views} are explicitly requested.
8026 @itemx -gno-column-info
8027 @opindex gcolumn-info
8028 @opindex gno-column-info
8029 Emit location column information into DWARF debugging information, rather
8030 than just file and line.
8031 This option is enabled by default.
8033 @item -gstatement-frontiers
8034 @itemx -gno-statement-frontiers
8035 @opindex gstatement-frontiers
8036 @opindex gno-statement-frontiers
8037 This option causes GCC to create markers in the internal representation
8038 at the beginning of statements, and to keep them roughly in place
8039 throughout compilation, using them to guide the output of @code{is_stmt}
8040 markers in the line number table. This is enabled by default when
8041 compiling with optimization (@option{-Os}, @option{-O}, @option{-O2},
8042 @dots{}), and outputting DWARF 2 debug information at the normal level.
8044 @item -gvariable-location-views
8045 @itemx -gvariable-location-views=incompat5
8046 @itemx -gno-variable-location-views
8047 @opindex gvariable-location-views
8048 @opindex gvariable-location-views=incompat5
8049 @opindex gno-variable-location-views
8050 Augment variable location lists with progressive view numbers implied
8051 from the line number table. This enables debug information consumers to
8052 inspect state at certain points of the program, even if no instructions
8053 associated with the corresponding source locations are present at that
8054 point. If the assembler lacks support for view numbers in line number
8055 tables, this will cause the compiler to emit the line number table,
8056 which generally makes them somewhat less compact. The augmented line
8057 number tables and location lists are fully backward-compatible, so they
8058 can be consumed by debug information consumers that are not aware of
8059 these augmentations, but they won't derive any benefit from them either.
8061 This is enabled by default when outputting DWARF 2 debug information at
8062 the normal level, as long as there is assembler support,
8063 @option{-fvar-tracking-assignments} is enabled and
8064 @option{-gstrict-dwarf} is not. When assembler support is not
8065 available, this may still be enabled, but it will force GCC to output
8066 internal line number tables, and if
8067 @option{-ginternal-reset-location-views} is not enabled, that will most
8068 certainly lead to silently mismatching location views.
8070 There is a proposed representation for view numbers that is not backward
8071 compatible with the location list format introduced in DWARF 5, that can
8072 be enabled with @option{-gvariable-location-views=incompat5}. This
8073 option may be removed in the future, is only provided as a reference
8074 implementation of the proposed representation. Debug information
8075 consumers are not expected to support this extended format, and they
8076 would be rendered unable to decode location lists using it.
8078 @item -ginternal-reset-location-views
8079 @itemx -gnointernal-reset-location-views
8080 @opindex ginternal-reset-location-views
8081 @opindex gno-internal-reset-location-views
8082 Attempt to determine location views that can be omitted from location
8083 view lists. This requires the compiler to have very accurate insn
8084 length estimates, which isn't always the case, and it may cause
8085 incorrect view lists to be generated silently when using an assembler
8086 that does not support location view lists. The GNU assembler will flag
8087 any such error as a @code{view number mismatch}. This is only enabled
8088 on ports that define a reliable estimation function.
8090 @item -ginline-points
8091 @itemx -gno-inline-points
8092 @opindex ginline-points
8093 @opindex gno-inline-points
8094 Generate extended debug information for inlined functions. Location
8095 view tracking markers are inserted at inlined entry points, so that
8096 address and view numbers can be computed and output in debug
8097 information. This can be enabled independently of location views, in
8098 which case the view numbers won't be output, but it can only be enabled
8099 along with statement frontiers, and it is only enabled by default if
8100 location views are enabled.
8102 @item -gz@r{[}=@var{type}@r{]}
8104 Produce compressed debug sections in DWARF format, if that is supported.
8105 If @var{type} is not given, the default type depends on the capabilities
8106 of the assembler and linker used. @var{type} may be one of
8107 @samp{none} (don't compress debug sections), @samp{zlib} (use zlib
8108 compression in ELF gABI format), or @samp{zlib-gnu} (use zlib
8109 compression in traditional GNU format). If the linker doesn't support
8110 writing compressed debug sections, the option is rejected. Otherwise,
8111 if the assembler does not support them, @option{-gz} is silently ignored
8112 when producing object files.
8114 @item -femit-struct-debug-baseonly
8115 @opindex femit-struct-debug-baseonly
8116 Emit debug information for struct-like types
8117 only when the base name of the compilation source file
8118 matches the base name of file in which the struct is defined.
8120 This option substantially reduces the size of debugging information,
8121 but at significant potential loss in type information to the debugger.
8122 See @option{-femit-struct-debug-reduced} for a less aggressive option.
8123 See @option{-femit-struct-debug-detailed} for more detailed control.
8125 This option works only with DWARF debug output.
8127 @item -femit-struct-debug-reduced
8128 @opindex femit-struct-debug-reduced
8129 Emit debug information for struct-like types
8130 only when the base name of the compilation source file
8131 matches the base name of file in which the type is defined,
8132 unless the struct is a template or defined in a system header.
8134 This option significantly reduces the size of debugging information,
8135 with some potential loss in type information to the debugger.
8136 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
8137 See @option{-femit-struct-debug-detailed} for more detailed control.
8139 This option works only with DWARF debug output.
8141 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
8142 @opindex femit-struct-debug-detailed
8143 Specify the struct-like types
8144 for which the compiler generates debug information.
8145 The intent is to reduce duplicate struct debug information
8146 between different object files within the same program.
8148 This option is a detailed version of
8149 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
8150 which serves for most needs.
8152 A specification has the syntax@*
8153 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
8155 The optional first word limits the specification to
8156 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
8157 A struct type is used directly when it is the type of a variable, member.
8158 Indirect uses arise through pointers to structs.
8159 That is, when use of an incomplete struct is valid, the use is indirect.
8161 @samp{struct one direct; struct two * indirect;}.
8163 The optional second word limits the specification to
8164 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
8165 Generic structs are a bit complicated to explain.
8166 For C++, these are non-explicit specializations of template classes,
8167 or non-template classes within the above.
8168 Other programming languages have generics,
8169 but @option{-femit-struct-debug-detailed} does not yet implement them.
8171 The third word specifies the source files for those
8172 structs for which the compiler should emit debug information.
8173 The values @samp{none} and @samp{any} have the normal meaning.
8174 The value @samp{base} means that
8175 the base of name of the file in which the type declaration appears
8176 must match the base of the name of the main compilation file.
8177 In practice, this means that when compiling @file{foo.c}, debug information
8178 is generated for types declared in that file and @file{foo.h},
8179 but not other header files.
8180 The value @samp{sys} means those types satisfying @samp{base}
8181 or declared in system or compiler headers.
8183 You may need to experiment to determine the best settings for your application.
8185 The default is @option{-femit-struct-debug-detailed=all}.
8187 This option works only with DWARF debug output.
8189 @item -fno-dwarf2-cfi-asm
8190 @opindex fdwarf2-cfi-asm
8191 @opindex fno-dwarf2-cfi-asm
8192 Emit DWARF unwind info as compiler generated @code{.eh_frame} section
8193 instead of using GAS @code{.cfi_*} directives.
8195 @item -fno-eliminate-unused-debug-types
8196 @opindex feliminate-unused-debug-types
8197 @opindex fno-eliminate-unused-debug-types
8198 Normally, when producing DWARF output, GCC avoids producing debug symbol
8199 output for types that are nowhere used in the source file being compiled.
8200 Sometimes it is useful to have GCC emit debugging
8201 information for all types declared in a compilation
8202 unit, regardless of whether or not they are actually used
8203 in that compilation unit, for example
8204 if, in the debugger, you want to cast a value to a type that is
8205 not actually used in your program (but is declared). More often,
8206 however, this results in a significant amount of wasted space.
8209 @node Optimize Options
8210 @section Options That Control Optimization
8211 @cindex optimize options
8212 @cindex options, optimization
8214 These options control various sorts of optimizations.
8216 Without any optimization option, the compiler's goal is to reduce the
8217 cost of compilation and to make debugging produce the expected
8218 results. Statements are independent: if you stop the program with a
8219 breakpoint between statements, you can then assign a new value to any
8220 variable or change the program counter to any other statement in the
8221 function and get exactly the results you expect from the source
8224 Turning on optimization flags makes the compiler attempt to improve
8225 the performance and/or code size at the expense of compilation time
8226 and possibly the ability to debug the program.
8228 The compiler performs optimization based on the knowledge it has of the
8229 program. Compiling multiple files at once to a single output file mode allows
8230 the compiler to use information gained from all of the files when compiling
8233 Not all optimizations are controlled directly by a flag. Only
8234 optimizations that have a flag are listed in this section.
8236 Most optimizations are completely disabled at @option{-O0} or if an
8237 @option{-O} level is not set on the command line, even if individual
8238 optimization flags are specified. Similarly, @option{-Og} suppresses
8239 many optimization passes.
8241 Depending on the target and how GCC was configured, a slightly different
8242 set of optimizations may be enabled at each @option{-O} level than
8243 those listed here. You can invoke GCC with @option{-Q --help=optimizers}
8244 to find out the exact set of optimizations that are enabled at each level.
8245 @xref{Overall Options}, for examples.
8252 Optimize. Optimizing compilation takes somewhat more time, and a lot
8253 more memory for a large function.
8255 With @option{-O}, the compiler tries to reduce code size and execution
8256 time, without performing any optimizations that take a great deal of
8259 @c Note that in addition to the default_options_table list in opts.c,
8260 @c several optimization flags default to true but control optimization
8261 @c passes that are explicitly disabled at -O0.
8263 @option{-O} turns on the following optimization flags:
8265 @c Please keep the following list alphabetized.
8266 @gccoptlist{-fauto-inc-dec @gol
8267 -fbranch-count-reg @gol
8268 -fcombine-stack-adjustments @gol
8270 -fcprop-registers @gol
8273 -fdelayed-branch @gol
8275 -fforward-propagate @gol
8276 -fguess-branch-probability @gol
8277 -fif-conversion @gol
8278 -fif-conversion2 @gol
8279 -finline-functions-called-once @gol
8281 -fipa-pure-const @gol
8282 -fipa-reference @gol
8283 -fipa-reference-addressable @gol
8284 -fmerge-constants @gol
8285 -fmove-loop-invariants @gol
8286 -fomit-frame-pointer @gol
8287 -freorder-blocks @gol
8289 -fshrink-wrap-separate @gol
8290 -fsplit-wide-types @gol
8296 -ftree-coalesce-vars @gol
8297 -ftree-copy-prop @gol
8299 -ftree-dominator-opts @gol
8301 -ftree-forwprop @gol
8305 -ftree-scev-cprop @gol
8314 Optimize even more. GCC performs nearly all supported optimizations
8315 that do not involve a space-speed tradeoff.
8316 As compared to @option{-O}, this option increases both compilation time
8317 and the performance of the generated code.
8319 @option{-O2} turns on all optimization flags specified by @option{-O}. It
8320 also turns on the following optimization flags:
8322 @c Please keep the following list alphabetized!
8323 @gccoptlist{-falign-functions -falign-jumps @gol
8324 -falign-labels -falign-loops @gol
8326 -fcode-hoisting @gol
8328 -fcse-follow-jumps -fcse-skip-blocks @gol
8329 -fdelete-null-pointer-checks @gol
8330 -fdevirtualize -fdevirtualize-speculatively @gol
8331 -fexpensive-optimizations @gol
8333 -fgcse -fgcse-lm @gol
8334 -fhoist-adjacent-loads @gol
8335 -finline-small-functions @gol
8336 -findirect-inlining @gol
8337 -fipa-bit-cp -fipa-cp -fipa-icf @gol
8338 -fipa-ra -fipa-sra -fipa-vrp @gol
8339 -fisolate-erroneous-paths-dereference @gol
8341 -foptimize-sibling-calls @gol
8342 -foptimize-strlen @gol
8343 -fpartial-inlining @gol
8345 -freorder-blocks-algorithm=stc @gol
8346 -freorder-blocks-and-partition -freorder-functions @gol
8347 -frerun-cse-after-loop @gol
8348 -fschedule-insns -fschedule-insns2 @gol
8349 -fsched-interblock -fsched-spec @gol
8350 -fstore-merging @gol
8351 -fstrict-aliasing @gol
8353 -ftree-builtin-call-dce @gol
8355 -ftree-switch-conversion -ftree-tail-merge @gol
8358 Please note the warning under @option{-fgcse} about
8359 invoking @option{-O2} on programs that use computed gotos.
8363 Optimize yet more. @option{-O3} turns on all optimizations specified
8364 by @option{-O2} and also turns on the following optimization flags:
8366 @c Please keep the following list alphabetized!
8367 @gccoptlist{-fgcse-after-reload @gol
8368 -finline-functions @gol
8370 -floop-interchange @gol
8371 -floop-unroll-and-jam @gol
8373 -fpredictive-commoning @gol
8375 -ftree-loop-distribute-patterns @gol
8376 -ftree-loop-distribution @gol
8377 -ftree-loop-vectorize @gol
8378 -ftree-partial-pre @gol
8379 -ftree-slp-vectorize @gol
8380 -funswitch-loops @gol
8381 -fvect-cost-model @gol
8382 -fversion-loops-for-strides}
8386 Reduce compilation time and make debugging produce the expected
8387 results. This is the default.
8391 Optimize for size. @option{-Os} enables all @option{-O2} optimizations
8392 except those that often increase code size:
8394 @gccoptlist{-falign-functions -falign-jumps @gol
8395 -falign-labels -falign-loops @gol
8396 -fprefetch-loop-arrays -freorder-blocks-algorithm=stc}
8398 It also enables @option{-finline-functions}, causes the compiler to tune for
8399 code size rather than execution speed, and performs further optimizations
8400 designed to reduce code size.
8404 Disregard strict standards compliance. @option{-Ofast} enables all
8405 @option{-O3} optimizations. It also enables optimizations that are not
8406 valid for all standard-compliant programs.
8407 It turns on @option{-ffast-math} and the Fortran-specific
8408 @option{-fstack-arrays}, unless @option{-fmax-stack-var-size} is
8409 specified, and @option{-fno-protect-parens}.
8413 Optimize debugging experience. @option{-Og} should be the optimization
8414 level of choice for the standard edit-compile-debug cycle, offering
8415 a reasonable level of optimization while maintaining fast compilation
8416 and a good debugging experience. It is a better choice than @option{-O0}
8417 for producing debuggable code because some compiler passes
8418 that collect debug information are disabled at @option{-O0}.
8420 Like @option{-O0}, @option{-Og} completely disables a number of
8421 optimization passes so that individual options controlling them have
8422 no effect. Otherwise @option{-Og} enables all @option{-O1}
8423 optimization flags except for those that may interfere with debugging:
8425 @gccoptlist{-fbranch-count-reg -fdelayed-branch @gol
8426 -fdse -fif-conversion -fif-conversion2 @gol
8427 -finline-functions-called-once @gol
8428 -fmove-loop-invariants -fssa-phiopt @gol
8429 -ftree-bit-ccp -ftree-dse -ftree-pta -ftree-sra}
8433 If you use multiple @option{-O} options, with or without level numbers,
8434 the last such option is the one that is effective.
8436 Options of the form @option{-f@var{flag}} specify machine-independent
8437 flags. Most flags have both positive and negative forms; the negative
8438 form of @option{-ffoo} is @option{-fno-foo}. In the table
8439 below, only one of the forms is listed---the one you typically
8440 use. You can figure out the other form by either removing @samp{no-}
8443 The following options control specific optimizations. They are either
8444 activated by @option{-O} options or are related to ones that are. You
8445 can use the following flags in the rare cases when ``fine-tuning'' of
8446 optimizations to be performed is desired.
8449 @item -fno-defer-pop
8450 @opindex fno-defer-pop
8452 For machines that must pop arguments after a function call, always pop
8453 the arguments as soon as each function returns.
8454 At levels @option{-O1} and higher, @option{-fdefer-pop} is the default;
8455 this allows the compiler to let arguments accumulate on the stack for several
8456 function calls and pop them all at once.
8458 @item -fforward-propagate
8459 @opindex fforward-propagate
8460 Perform a forward propagation pass on RTL@. The pass tries to combine two
8461 instructions and checks if the result can be simplified. If loop unrolling
8462 is active, two passes are performed and the second is scheduled after
8465 This option is enabled by default at optimization levels @option{-O},
8466 @option{-O2}, @option{-O3}, @option{-Os}.
8468 @item -ffp-contract=@var{style}
8469 @opindex ffp-contract
8470 @option{-ffp-contract=off} disables floating-point expression contraction.
8471 @option{-ffp-contract=fast} enables floating-point expression contraction
8472 such as forming of fused multiply-add operations if the target has
8473 native support for them.
8474 @option{-ffp-contract=on} enables floating-point expression contraction
8475 if allowed by the language standard. This is currently not implemented
8476 and treated equal to @option{-ffp-contract=off}.
8478 The default is @option{-ffp-contract=fast}.
8480 @item -fomit-frame-pointer
8481 @opindex fomit-frame-pointer
8482 Omit the frame pointer in functions that don't need one. This avoids the
8483 instructions to save, set up and restore the frame pointer; on many targets
8484 it also makes an extra register available.
8486 On some targets this flag has no effect because the standard calling sequence
8487 always uses a frame pointer, so it cannot be omitted.
8489 Note that @option{-fno-omit-frame-pointer} doesn't guarantee the frame pointer
8490 is used in all functions. Several targets always omit the frame pointer in
8493 Enabled by default at @option{-O} and higher.
8495 @item -foptimize-sibling-calls
8496 @opindex foptimize-sibling-calls
8497 Optimize sibling and tail recursive calls.
8499 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8501 @item -foptimize-strlen
8502 @opindex foptimize-strlen
8503 Optimize various standard C string functions (e.g.@: @code{strlen},
8504 @code{strchr} or @code{strcpy}) and
8505 their @code{_FORTIFY_SOURCE} counterparts into faster alternatives.
8507 Enabled at levels @option{-O2}, @option{-O3}.
8512 Do not expand any functions inline apart from those marked with
8513 the @code{always_inline} attribute. This is the default when not
8516 Single functions can be exempted from inlining by marking them
8517 with the @code{noinline} attribute.
8519 @item -finline-small-functions
8520 @opindex finline-small-functions
8521 Integrate functions into their callers when their body is smaller than expected
8522 function call code (so overall size of program gets smaller). The compiler
8523 heuristically decides which functions are simple enough to be worth integrating
8524 in this way. This inlining applies to all functions, even those not declared
8527 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8529 @item -findirect-inlining
8530 @opindex findirect-inlining
8531 Inline also indirect calls that are discovered to be known at compile
8532 time thanks to previous inlining. This option has any effect only
8533 when inlining itself is turned on by the @option{-finline-functions}
8534 or @option{-finline-small-functions} options.
8536 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8538 @item -finline-functions
8539 @opindex finline-functions
8540 Consider all functions for inlining, even if they are not declared inline.
8541 The compiler heuristically decides which functions are worth integrating
8544 If all calls to a given function are integrated, and the function is
8545 declared @code{static}, then the function is normally not output as
8546 assembler code in its own right.
8548 Enabled at levels @option{-O3}, @option{-Os}. Also enabled
8549 by @option{-fprofile-use} and @option{-fauto-profile}.
8551 @item -finline-functions-called-once
8552 @opindex finline-functions-called-once
8553 Consider all @code{static} functions called once for inlining into their
8554 caller even if they are not marked @code{inline}. If a call to a given
8555 function is integrated, then the function is not output as assembler code
8558 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os},
8559 but not @option{-Og}.
8561 @item -fearly-inlining
8562 @opindex fearly-inlining
8563 Inline functions marked by @code{always_inline} and functions whose body seems
8564 smaller than the function call overhead early before doing
8565 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
8566 makes profiling significantly cheaper and usually inlining faster on programs
8567 having large chains of nested wrapper functions.
8573 Perform interprocedural scalar replacement of aggregates, removal of
8574 unused parameters and replacement of parameters passed by reference
8575 by parameters passed by value.
8577 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
8579 @item -finline-limit=@var{n}
8580 @opindex finline-limit
8581 By default, GCC limits the size of functions that can be inlined. This flag
8582 allows coarse control of this limit. @var{n} is the size of functions that
8583 can be inlined in number of pseudo instructions.
8585 Inlining is actually controlled by a number of parameters, which may be
8586 specified individually by using @option{--param @var{name}=@var{value}}.
8587 The @option{-finline-limit=@var{n}} option sets some of these parameters
8591 @item max-inline-insns-single
8592 is set to @var{n}/2.
8593 @item max-inline-insns-auto
8594 is set to @var{n}/2.
8597 See below for a documentation of the individual
8598 parameters controlling inlining and for the defaults of these parameters.
8600 @emph{Note:} there may be no value to @option{-finline-limit} that results
8601 in default behavior.
8603 @emph{Note:} pseudo instruction represents, in this particular context, an
8604 abstract measurement of function's size. In no way does it represent a count
8605 of assembly instructions and as such its exact meaning might change from one
8606 release to an another.
8608 @item -fno-keep-inline-dllexport
8609 @opindex fno-keep-inline-dllexport
8610 @opindex fkeep-inline-dllexport
8611 This is a more fine-grained version of @option{-fkeep-inline-functions},
8612 which applies only to functions that are declared using the @code{dllexport}
8613 attribute or declspec. @xref{Function Attributes,,Declaring Attributes of
8616 @item -fkeep-inline-functions
8617 @opindex fkeep-inline-functions
8618 In C, emit @code{static} functions that are declared @code{inline}
8619 into the object file, even if the function has been inlined into all
8620 of its callers. This switch does not affect functions using the
8621 @code{extern inline} extension in GNU C90@. In C++, emit any and all
8622 inline functions into the object file.
8624 @item -fkeep-static-functions
8625 @opindex fkeep-static-functions
8626 Emit @code{static} functions into the object file, even if the function
8629 @item -fkeep-static-consts
8630 @opindex fkeep-static-consts
8631 Emit variables declared @code{static const} when optimization isn't turned
8632 on, even if the variables aren't referenced.
8634 GCC enables this option by default. If you want to force the compiler to
8635 check if a variable is referenced, regardless of whether or not
8636 optimization is turned on, use the @option{-fno-keep-static-consts} option.
8638 @item -fmerge-constants
8639 @opindex fmerge-constants
8640 Attempt to merge identical constants (string constants and floating-point
8641 constants) across compilation units.
8643 This option is the default for optimized compilation if the assembler and
8644 linker support it. Use @option{-fno-merge-constants} to inhibit this
8647 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8649 @item -fmerge-all-constants
8650 @opindex fmerge-all-constants
8651 Attempt to merge identical constants and identical variables.
8653 This option implies @option{-fmerge-constants}. In addition to
8654 @option{-fmerge-constants} this considers e.g.@: even constant initialized
8655 arrays or initialized constant variables with integral or floating-point
8656 types. Languages like C or C++ require each variable, including multiple
8657 instances of the same variable in recursive calls, to have distinct locations,
8658 so using this option results in non-conforming
8661 @item -fmodulo-sched
8662 @opindex fmodulo-sched
8663 Perform swing modulo scheduling immediately before the first scheduling
8664 pass. This pass looks at innermost loops and reorders their
8665 instructions by overlapping different iterations.
8667 @item -fmodulo-sched-allow-regmoves
8668 @opindex fmodulo-sched-allow-regmoves
8669 Perform more aggressive SMS-based modulo scheduling with register moves
8670 allowed. By setting this flag certain anti-dependences edges are
8671 deleted, which triggers the generation of reg-moves based on the
8672 life-range analysis. This option is effective only with
8673 @option{-fmodulo-sched} enabled.
8675 @item -fno-branch-count-reg
8676 @opindex fno-branch-count-reg
8677 @opindex fbranch-count-reg
8678 Disable the optimization pass that scans for opportunities to use
8679 ``decrement and branch'' instructions on a count register instead of
8680 instruction sequences that decrement a register, compare it against zero, and
8681 then branch based upon the result. This option is only meaningful on
8682 architectures that support such instructions, which include x86, PowerPC,
8683 IA-64 and S/390. Note that the @option{-fno-branch-count-reg} option
8684 doesn't remove the decrement and branch instructions from the generated
8685 instruction stream introduced by other optimization passes.
8687 The default is @option{-fbranch-count-reg} at @option{-O1} and higher,
8688 except for @option{-Og}.
8690 @item -fno-function-cse
8691 @opindex fno-function-cse
8692 @opindex ffunction-cse
8693 Do not put function addresses in registers; make each instruction that
8694 calls a constant function contain the function's address explicitly.
8696 This option results in less efficient code, but some strange hacks
8697 that alter the assembler output may be confused by the optimizations
8698 performed when this option is not used.
8700 The default is @option{-ffunction-cse}
8702 @item -fno-zero-initialized-in-bss
8703 @opindex fno-zero-initialized-in-bss
8704 @opindex fzero-initialized-in-bss
8705 If the target supports a BSS section, GCC by default puts variables that
8706 are initialized to zero into BSS@. This can save space in the resulting
8709 This option turns off this behavior because some programs explicitly
8710 rely on variables going to the data section---e.g., so that the
8711 resulting executable can find the beginning of that section and/or make
8712 assumptions based on that.
8714 The default is @option{-fzero-initialized-in-bss}.
8716 @item -fthread-jumps
8717 @opindex fthread-jumps
8718 Perform optimizations that check to see if a jump branches to a
8719 location where another comparison subsumed by the first is found. If
8720 so, the first branch is redirected to either the destination of the
8721 second branch or a point immediately following it, depending on whether
8722 the condition is known to be true or false.
8724 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8726 @item -fsplit-wide-types
8727 @opindex fsplit-wide-types
8728 When using a type that occupies multiple registers, such as @code{long
8729 long} on a 32-bit system, split the registers apart and allocate them
8730 independently. This normally generates better code for those types,
8731 but may make debugging more difficult.
8733 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
8736 @item -fsplit-wide-types-early
8737 @opindex fsplit-wide-types-early
8738 Fully split wide types early, instead of very late.
8739 This option has no effect unless @option{-fsplit-wide-types} is turned on.
8741 This is the default on some targets.
8743 @item -fcse-follow-jumps
8744 @opindex fcse-follow-jumps
8745 In common subexpression elimination (CSE), scan through jump instructions
8746 when the target of the jump is not reached by any other path. For
8747 example, when CSE encounters an @code{if} statement with an
8748 @code{else} clause, CSE follows the jump when the condition
8751 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8753 @item -fcse-skip-blocks
8754 @opindex fcse-skip-blocks
8755 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
8756 follow jumps that conditionally skip over blocks. When CSE
8757 encounters a simple @code{if} statement with no else clause,
8758 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
8759 body of the @code{if}.
8761 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8763 @item -frerun-cse-after-loop
8764 @opindex frerun-cse-after-loop
8765 Re-run common subexpression elimination after loop optimizations are
8768 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8772 Perform a global common subexpression elimination pass.
8773 This pass also performs global constant and copy propagation.
8775 @emph{Note:} When compiling a program using computed gotos, a GCC
8776 extension, you may get better run-time performance if you disable
8777 the global common subexpression elimination pass by adding
8778 @option{-fno-gcse} to the command line.
8780 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8784 When @option{-fgcse-lm} is enabled, global common subexpression elimination
8785 attempts to move loads that are only killed by stores into themselves. This
8786 allows a loop containing a load/store sequence to be changed to a load outside
8787 the loop, and a copy/store within the loop.
8789 Enabled by default when @option{-fgcse} is enabled.
8793 When @option{-fgcse-sm} is enabled, a store motion pass is run after
8794 global common subexpression elimination. This pass attempts to move
8795 stores out of loops. When used in conjunction with @option{-fgcse-lm},
8796 loops containing a load/store sequence can be changed to a load before
8797 the loop and a store after the loop.
8799 Not enabled at any optimization level.
8803 When @option{-fgcse-las} is enabled, the global common subexpression
8804 elimination pass eliminates redundant loads that come after stores to the
8805 same memory location (both partial and full redundancies).
8807 Not enabled at any optimization level.
8809 @item -fgcse-after-reload
8810 @opindex fgcse-after-reload
8811 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
8812 pass is performed after reload. The purpose of this pass is to clean up
8815 Enabled by @option{-fprofile-use} and @option{-fauto-profile}.
8817 @item -faggressive-loop-optimizations
8818 @opindex faggressive-loop-optimizations
8819 This option tells the loop optimizer to use language constraints to
8820 derive bounds for the number of iterations of a loop. This assumes that
8821 loop code does not invoke undefined behavior by for example causing signed
8822 integer overflows or out-of-bound array accesses. The bounds for the
8823 number of iterations of a loop are used to guide loop unrolling and peeling
8824 and loop exit test optimizations.
8825 This option is enabled by default.
8827 @item -funconstrained-commons
8828 @opindex funconstrained-commons
8829 This option tells the compiler that variables declared in common blocks
8830 (e.g.@: Fortran) may later be overridden with longer trailing arrays. This
8831 prevents certain optimizations that depend on knowing the array bounds.
8833 @item -fcrossjumping
8834 @opindex fcrossjumping
8835 Perform cross-jumping transformation.
8836 This transformation unifies equivalent code and saves code size. The
8837 resulting code may or may not perform better than without cross-jumping.
8839 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8841 @item -fauto-inc-dec
8842 @opindex fauto-inc-dec
8843 Combine increments or decrements of addresses with memory accesses.
8844 This pass is always skipped on architectures that do not have
8845 instructions to support this. Enabled by default at @option{-O} and
8846 higher on architectures that support this.
8850 Perform dead code elimination (DCE) on RTL@.
8851 Enabled by default at @option{-O} and higher.
8855 Perform dead store elimination (DSE) on RTL@.
8856 Enabled by default at @option{-O} and higher.
8858 @item -fif-conversion
8859 @opindex fif-conversion
8860 Attempt to transform conditional jumps into branch-less equivalents. This
8861 includes use of conditional moves, min, max, set flags and abs instructions, and
8862 some tricks doable by standard arithmetics. The use of conditional execution
8863 on chips where it is available is controlled by @option{-fif-conversion2}.
8865 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}, but
8866 not with @option{-Og}.
8868 @item -fif-conversion2
8869 @opindex fif-conversion2
8870 Use conditional execution (where available) to transform conditional jumps into
8871 branch-less equivalents.
8873 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}, but
8874 not with @option{-Og}.
8876 @item -fdeclone-ctor-dtor
8877 @opindex fdeclone-ctor-dtor
8878 The C++ ABI requires multiple entry points for constructors and
8879 destructors: one for a base subobject, one for a complete object, and
8880 one for a virtual destructor that calls operator delete afterwards.
8881 For a hierarchy with virtual bases, the base and complete variants are
8882 clones, which means two copies of the function. With this option, the
8883 base and complete variants are changed to be thunks that call a common
8886 Enabled by @option{-Os}.
8888 @item -fdelete-null-pointer-checks
8889 @opindex fdelete-null-pointer-checks
8890 Assume that programs cannot safely dereference null pointers, and that
8891 no code or data element resides at address zero.
8892 This option enables simple constant
8893 folding optimizations at all optimization levels. In addition, other
8894 optimization passes in GCC use this flag to control global dataflow
8895 analyses that eliminate useless checks for null pointers; these assume
8896 that a memory access to address zero always results in a trap, so
8897 that if a pointer is checked after it has already been dereferenced,
8900 Note however that in some environments this assumption is not true.
8901 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
8902 for programs that depend on that behavior.
8904 This option is enabled by default on most targets. On Nios II ELF, it
8905 defaults to off. On AVR, CR16, and MSP430, this option is completely disabled.
8907 Passes that use the dataflow information
8908 are enabled independently at different optimization levels.
8910 @item -fdevirtualize
8911 @opindex fdevirtualize
8912 Attempt to convert calls to virtual functions to direct calls. This
8913 is done both within a procedure and interprocedurally as part of
8914 indirect inlining (@option{-findirect-inlining}) and interprocedural constant
8915 propagation (@option{-fipa-cp}).
8916 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8918 @item -fdevirtualize-speculatively
8919 @opindex fdevirtualize-speculatively
8920 Attempt to convert calls to virtual functions to speculative direct calls.
8921 Based on the analysis of the type inheritance graph, determine for a given call
8922 the set of likely targets. If the set is small, preferably of size 1, change
8923 the call into a conditional deciding between direct and indirect calls. The
8924 speculative calls enable more optimizations, such as inlining. When they seem
8925 useless after further optimization, they are converted back into original form.
8927 @item -fdevirtualize-at-ltrans
8928 @opindex fdevirtualize-at-ltrans
8929 Stream extra information needed for aggressive devirtualization when running
8930 the link-time optimizer in local transformation mode.
8931 This option enables more devirtualization but
8932 significantly increases the size of streamed data. For this reason it is
8933 disabled by default.
8935 @item -fexpensive-optimizations
8936 @opindex fexpensive-optimizations
8937 Perform a number of minor optimizations that are relatively expensive.
8939 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8943 Attempt to remove redundant extension instructions. This is especially
8944 helpful for the x86-64 architecture, which implicitly zero-extends in 64-bit
8945 registers after writing to their lower 32-bit half.
8947 Enabled for Alpha, AArch64 and x86 at levels @option{-O2},
8948 @option{-O3}, @option{-Os}.
8950 @item -fno-lifetime-dse
8951 @opindex fno-lifetime-dse
8952 @opindex flifetime-dse
8953 In C++ the value of an object is only affected by changes within its
8954 lifetime: when the constructor begins, the object has an indeterminate
8955 value, and any changes during the lifetime of the object are dead when
8956 the object is destroyed. Normally dead store elimination will take
8957 advantage of this; if your code relies on the value of the object
8958 storage persisting beyond the lifetime of the object, you can use this
8959 flag to disable this optimization. To preserve stores before the
8960 constructor starts (e.g.@: because your operator new clears the object
8961 storage) but still treat the object as dead after the destructor you,
8962 can use @option{-flifetime-dse=1}. The default behavior can be
8963 explicitly selected with @option{-flifetime-dse=2}.
8964 @option{-flifetime-dse=0} is equivalent to @option{-fno-lifetime-dse}.
8966 @item -flive-range-shrinkage
8967 @opindex flive-range-shrinkage
8968 Attempt to decrease register pressure through register live range
8969 shrinkage. This is helpful for fast processors with small or moderate
8972 @item -fira-algorithm=@var{algorithm}
8973 @opindex fira-algorithm
8974 Use the specified coloring algorithm for the integrated register
8975 allocator. The @var{algorithm} argument can be @samp{priority}, which
8976 specifies Chow's priority coloring, or @samp{CB}, which specifies
8977 Chaitin-Briggs coloring. Chaitin-Briggs coloring is not implemented
8978 for all architectures, but for those targets that do support it, it is
8979 the default because it generates better code.
8981 @item -fira-region=@var{region}
8982 @opindex fira-region
8983 Use specified regions for the integrated register allocator. The
8984 @var{region} argument should be one of the following:
8989 Use all loops as register allocation regions.
8990 This can give the best results for machines with a small and/or
8991 irregular register set.
8994 Use all loops except for loops with small register pressure
8995 as the regions. This value usually gives
8996 the best results in most cases and for most architectures,
8997 and is enabled by default when compiling with optimization for speed
8998 (@option{-O}, @option{-O2}, @dots{}).
9001 Use all functions as a single region.
9002 This typically results in the smallest code size, and is enabled by default for
9003 @option{-Os} or @option{-O0}.
9007 @item -fira-hoist-pressure
9008 @opindex fira-hoist-pressure
9009 Use IRA to evaluate register pressure in the code hoisting pass for
9010 decisions to hoist expressions. This option usually results in smaller
9011 code, but it can slow the compiler down.
9013 This option is enabled at level @option{-Os} for all targets.
9015 @item -fira-loop-pressure
9016 @opindex fira-loop-pressure
9017 Use IRA to evaluate register pressure in loops for decisions to move
9018 loop invariants. This option usually results in generation
9019 of faster and smaller code on machines with large register files (>= 32
9020 registers), but it can slow the compiler down.
9022 This option is enabled at level @option{-O3} for some targets.
9024 @item -fno-ira-share-save-slots
9025 @opindex fno-ira-share-save-slots
9026 @opindex fira-share-save-slots
9027 Disable sharing of stack slots used for saving call-used hard
9028 registers living through a call. Each hard register gets a
9029 separate stack slot, and as a result function stack frames are
9032 @item -fno-ira-share-spill-slots
9033 @opindex fno-ira-share-spill-slots
9034 @opindex fira-share-spill-slots
9035 Disable sharing of stack slots allocated for pseudo-registers. Each
9036 pseudo-register that does not get a hard register gets a separate
9037 stack slot, and as a result function stack frames are larger.
9041 Enable CFG-sensitive rematerialization in LRA. Instead of loading
9042 values of spilled pseudos, LRA tries to rematerialize (recalculate)
9043 values if it is profitable.
9045 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
9047 @item -fdelayed-branch
9048 @opindex fdelayed-branch
9049 If supported for the target machine, attempt to reorder instructions
9050 to exploit instruction slots available after delayed branch
9053 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os},
9054 but not at @option{-Og}.
9056 @item -fschedule-insns
9057 @opindex fschedule-insns
9058 If supported for the target machine, attempt to reorder instructions to
9059 eliminate execution stalls due to required data being unavailable. This
9060 helps machines that have slow floating point or memory load instructions
9061 by allowing other instructions to be issued until the result of the load
9062 or floating-point instruction is required.
9064 Enabled at levels @option{-O2}, @option{-O3}.
9066 @item -fschedule-insns2
9067 @opindex fschedule-insns2
9068 Similar to @option{-fschedule-insns}, but requests an additional pass of
9069 instruction scheduling after register allocation has been done. This is
9070 especially useful on machines with a relatively small number of
9071 registers and where memory load instructions take more than one cycle.
9073 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
9075 @item -fno-sched-interblock
9076 @opindex fno-sched-interblock
9077 @opindex fsched-interblock
9078 Disable instruction scheduling across basic blocks, which
9079 is normally enabled when scheduling before register allocation, i.e.@:
9080 with @option{-fschedule-insns} or at @option{-O2} or higher.
9082 @item -fno-sched-spec
9083 @opindex fno-sched-spec
9084 @opindex fsched-spec
9085 Disable speculative motion of non-load instructions, which
9086 is normally enabled when scheduling before register allocation, i.e.@:
9087 with @option{-fschedule-insns} or at @option{-O2} or higher.
9089 @item -fsched-pressure
9090 @opindex fsched-pressure
9091 Enable register pressure sensitive insn scheduling before register
9092 allocation. This only makes sense when scheduling before register
9093 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
9094 @option{-O2} or higher. Usage of this option can improve the
9095 generated code and decrease its size by preventing register pressure
9096 increase above the number of available hard registers and subsequent
9097 spills in register allocation.
9099 @item -fsched-spec-load
9100 @opindex fsched-spec-load
9101 Allow speculative motion of some load instructions. This only makes
9102 sense when scheduling before register allocation, i.e.@: with
9103 @option{-fschedule-insns} or at @option{-O2} or higher.
9105 @item -fsched-spec-load-dangerous
9106 @opindex fsched-spec-load-dangerous
9107 Allow speculative motion of more load instructions. This only makes
9108 sense when scheduling before register allocation, i.e.@: with
9109 @option{-fschedule-insns} or at @option{-O2} or higher.
9111 @item -fsched-stalled-insns
9112 @itemx -fsched-stalled-insns=@var{n}
9113 @opindex fsched-stalled-insns
9114 Define how many insns (if any) can be moved prematurely from the queue
9115 of stalled insns into the ready list during the second scheduling pass.
9116 @option{-fno-sched-stalled-insns} means that no insns are moved
9117 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
9118 on how many queued insns can be moved prematurely.
9119 @option{-fsched-stalled-insns} without a value is equivalent to
9120 @option{-fsched-stalled-insns=1}.
9122 @item -fsched-stalled-insns-dep
9123 @itemx -fsched-stalled-insns-dep=@var{n}
9124 @opindex fsched-stalled-insns-dep
9125 Define how many insn groups (cycles) are examined for a dependency
9126 on a stalled insn that is a candidate for premature removal from the queue
9127 of stalled insns. This has an effect only during the second scheduling pass,
9128 and only if @option{-fsched-stalled-insns} is used.
9129 @option{-fno-sched-stalled-insns-dep} is equivalent to
9130 @option{-fsched-stalled-insns-dep=0}.
9131 @option{-fsched-stalled-insns-dep} without a value is equivalent to
9132 @option{-fsched-stalled-insns-dep=1}.
9134 @item -fsched2-use-superblocks
9135 @opindex fsched2-use-superblocks
9136 When scheduling after register allocation, use superblock scheduling.
9137 This allows motion across basic block boundaries,
9138 resulting in faster schedules. This option is experimental, as not all machine
9139 descriptions used by GCC model the CPU closely enough to avoid unreliable
9140 results from the algorithm.
9142 This only makes sense when scheduling after register allocation, i.e.@: with
9143 @option{-fschedule-insns2} or at @option{-O2} or higher.
9145 @item -fsched-group-heuristic
9146 @opindex fsched-group-heuristic
9147 Enable the group heuristic in the scheduler. This heuristic favors
9148 the instruction that belongs to a schedule group. This is enabled
9149 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
9150 or @option{-fschedule-insns2} or at @option{-O2} or higher.
9152 @item -fsched-critical-path-heuristic
9153 @opindex fsched-critical-path-heuristic
9154 Enable the critical-path heuristic in the scheduler. This heuristic favors
9155 instructions on the critical path. This is enabled by default when
9156 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
9157 or @option{-fschedule-insns2} or at @option{-O2} or higher.
9159 @item -fsched-spec-insn-heuristic
9160 @opindex fsched-spec-insn-heuristic
9161 Enable the speculative instruction heuristic in the scheduler. This
9162 heuristic favors speculative instructions with greater dependency weakness.
9163 This is enabled by default when scheduling is enabled, i.e.@:
9164 with @option{-fschedule-insns} or @option{-fschedule-insns2}
9165 or at @option{-O2} or higher.
9167 @item -fsched-rank-heuristic
9168 @opindex fsched-rank-heuristic
9169 Enable the rank heuristic in the scheduler. This heuristic favors
9170 the instruction belonging to a basic block with greater size or frequency.
9171 This is enabled by default when scheduling is enabled, i.e.@:
9172 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
9173 at @option{-O2} or higher.
9175 @item -fsched-last-insn-heuristic
9176 @opindex fsched-last-insn-heuristic
9177 Enable the last-instruction heuristic in the scheduler. This heuristic
9178 favors the instruction that is less dependent on the last instruction
9179 scheduled. This is enabled by default when scheduling is enabled,
9180 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
9181 at @option{-O2} or higher.
9183 @item -fsched-dep-count-heuristic
9184 @opindex fsched-dep-count-heuristic
9185 Enable the dependent-count heuristic in the scheduler. This heuristic
9186 favors the instruction that has more instructions depending on it.
9187 This is enabled by default when scheduling is enabled, i.e.@:
9188 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
9189 at @option{-O2} or higher.
9191 @item -freschedule-modulo-scheduled-loops
9192 @opindex freschedule-modulo-scheduled-loops
9193 Modulo scheduling is performed before traditional scheduling. If a loop
9194 is modulo scheduled, later scheduling passes may change its schedule.
9195 Use this option to control that behavior.
9197 @item -fselective-scheduling
9198 @opindex fselective-scheduling
9199 Schedule instructions using selective scheduling algorithm. Selective
9200 scheduling runs instead of the first scheduler pass.
9202 @item -fselective-scheduling2
9203 @opindex fselective-scheduling2
9204 Schedule instructions using selective scheduling algorithm. Selective
9205 scheduling runs instead of the second scheduler pass.
9207 @item -fsel-sched-pipelining
9208 @opindex fsel-sched-pipelining
9209 Enable software pipelining of innermost loops during selective scheduling.
9210 This option has no effect unless one of @option{-fselective-scheduling} or
9211 @option{-fselective-scheduling2} is turned on.
9213 @item -fsel-sched-pipelining-outer-loops
9214 @opindex fsel-sched-pipelining-outer-loops
9215 When pipelining loops during selective scheduling, also pipeline outer loops.
9216 This option has no effect unless @option{-fsel-sched-pipelining} is turned on.
9218 @item -fsemantic-interposition
9219 @opindex fsemantic-interposition
9220 Some object formats, like ELF, allow interposing of symbols by the
9222 This means that for symbols exported from the DSO, the compiler cannot perform
9223 interprocedural propagation, inlining and other optimizations in anticipation
9224 that the function or variable in question may change. While this feature is
9225 useful, for example, to rewrite memory allocation functions by a debugging
9226 implementation, it is expensive in the terms of code quality.
9227 With @option{-fno-semantic-interposition} the compiler assumes that
9228 if interposition happens for functions the overwriting function will have
9229 precisely the same semantics (and side effects).
9230 Similarly if interposition happens
9231 for variables, the constructor of the variable will be the same. The flag
9232 has no effect for functions explicitly declared inline
9233 (where it is never allowed for interposition to change semantics)
9234 and for symbols explicitly declared weak.
9237 @opindex fshrink-wrap
9238 Emit function prologues only before parts of the function that need it,
9239 rather than at the top of the function. This flag is enabled by default at
9240 @option{-O} and higher.
9242 @item -fshrink-wrap-separate
9243 @opindex fshrink-wrap-separate
9244 Shrink-wrap separate parts of the prologue and epilogue separately, so that
9245 those parts are only executed when needed.
9246 This option is on by default, but has no effect unless @option{-fshrink-wrap}
9247 is also turned on and the target supports this.
9249 @item -fcaller-saves
9250 @opindex fcaller-saves
9251 Enable allocation of values to registers that are clobbered by
9252 function calls, by emitting extra instructions to save and restore the
9253 registers around such calls. Such allocation is done only when it
9254 seems to result in better code.
9256 This option is always enabled by default on certain machines, usually
9257 those which have no call-preserved registers to use instead.
9259 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
9261 @item -fcombine-stack-adjustments
9262 @opindex fcombine-stack-adjustments
9263 Tracks stack adjustments (pushes and pops) and stack memory references
9264 and then tries to find ways to combine them.
9266 Enabled by default at @option{-O1} and higher.
9270 Use caller save registers for allocation if those registers are not used by
9271 any called function. In that case it is not necessary to save and restore
9272 them around calls. This is only possible if called functions are part of
9273 same compilation unit as current function and they are compiled before it.
9275 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}, however the option
9276 is disabled if generated code will be instrumented for profiling
9277 (@option{-p}, or @option{-pg}) or if callee's register usage cannot be known
9278 exactly (this happens on targets that do not expose prologues
9279 and epilogues in RTL).
9281 @item -fconserve-stack
9282 @opindex fconserve-stack
9283 Attempt to minimize stack usage. The compiler attempts to use less
9284 stack space, even if that makes the program slower. This option
9285 implies setting the @option{large-stack-frame} parameter to 100
9286 and the @option{large-stack-frame-growth} parameter to 400.
9288 @item -ftree-reassoc
9289 @opindex ftree-reassoc
9290 Perform reassociation on trees. This flag is enabled by default
9291 at @option{-O} and higher.
9293 @item -fcode-hoisting
9294 @opindex fcode-hoisting
9295 Perform code hoisting. Code hoisting tries to move the
9296 evaluation of expressions executed on all paths to the function exit
9297 as early as possible. This is especially useful as a code size
9298 optimization, but it often helps for code speed as well.
9299 This flag is enabled by default at @option{-O2} and higher.
9303 Perform partial redundancy elimination (PRE) on trees. This flag is
9304 enabled by default at @option{-O2} and @option{-O3}.
9306 @item -ftree-partial-pre
9307 @opindex ftree-partial-pre
9308 Make partial redundancy elimination (PRE) more aggressive. This flag is
9309 enabled by default at @option{-O3}.
9311 @item -ftree-forwprop
9312 @opindex ftree-forwprop
9313 Perform forward propagation on trees. This flag is enabled by default
9314 at @option{-O} and higher.
9318 Perform full redundancy elimination (FRE) on trees. The difference
9319 between FRE and PRE is that FRE only considers expressions
9320 that are computed on all paths leading to the redundant computation.
9321 This analysis is faster than PRE, though it exposes fewer redundancies.
9322 This flag is enabled by default at @option{-O} and higher.
9324 @item -ftree-phiprop
9325 @opindex ftree-phiprop
9326 Perform hoisting of loads from conditional pointers on trees. This
9327 pass is enabled by default at @option{-O} and higher.
9329 @item -fhoist-adjacent-loads
9330 @opindex fhoist-adjacent-loads
9331 Speculatively hoist loads from both branches of an if-then-else if the
9332 loads are from adjacent locations in the same structure and the target
9333 architecture has a conditional move instruction. This flag is enabled
9334 by default at @option{-O2} and higher.
9336 @item -ftree-copy-prop
9337 @opindex ftree-copy-prop
9338 Perform copy propagation on trees. This pass eliminates unnecessary
9339 copy operations. This flag is enabled by default at @option{-O} and
9342 @item -fipa-pure-const
9343 @opindex fipa-pure-const
9344 Discover which functions are pure or constant.
9345 Enabled by default at @option{-O} and higher.
9347 @item -fipa-reference
9348 @opindex fipa-reference
9349 Discover which static variables do not escape the
9351 Enabled by default at @option{-O} and higher.
9353 @item -fipa-reference-addressable
9354 @opindex fipa-reference-addressable
9355 Discover read-only, write-only and non-addressable static variables.
9356 Enabled by default at @option{-O} and higher.
9358 @item -fipa-stack-alignment
9359 @opindex fipa-stack-alignment
9360 Reduce stack alignment on call sites if possible.
9365 Perform interprocedural pointer analysis and interprocedural modification
9366 and reference analysis. This option can cause excessive memory and
9367 compile-time usage on large compilation units. It is not enabled by
9368 default at any optimization level.
9371 @opindex fipa-profile
9372 Perform interprocedural profile propagation. The functions called only from
9373 cold functions are marked as cold. Also functions executed once (such as
9374 @code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
9375 functions and loop less parts of functions executed once are then optimized for
9377 Enabled by default at @option{-O} and higher.
9381 Perform interprocedural constant propagation.
9382 This optimization analyzes the program to determine when values passed
9383 to functions are constants and then optimizes accordingly.
9384 This optimization can substantially increase performance
9385 if the application has constants passed to functions.
9386 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
9387 It is also enabled by @option{-fprofile-use} and @option{-fauto-profile}.
9389 @item -fipa-cp-clone
9390 @opindex fipa-cp-clone
9391 Perform function cloning to make interprocedural constant propagation stronger.
9392 When enabled, interprocedural constant propagation performs function cloning
9393 when externally visible function can be called with constant arguments.
9394 Because this optimization can create multiple copies of functions,
9395 it may significantly increase code size
9396 (see @option{--param ipcp-unit-growth=@var{value}}).
9397 This flag is enabled by default at @option{-O3}.
9398 It is also enabled by @option{-fprofile-use} and @option{-fauto-profile}.
9401 @opindex fipa-bit-cp
9402 When enabled, perform interprocedural bitwise constant
9403 propagation. This flag is enabled by default at @option{-O2} and
9404 by @option{-fprofile-use} and @option{-fauto-profile}.
9405 It requires that @option{-fipa-cp} is enabled.
9409 When enabled, perform interprocedural propagation of value
9410 ranges. This flag is enabled by default at @option{-O2}. It requires
9411 that @option{-fipa-cp} is enabled.
9415 Perform Identical Code Folding for functions and read-only variables.
9416 The optimization reduces code size and may disturb unwind stacks by replacing
9417 a function by equivalent one with a different name. The optimization works
9418 more effectively with link-time optimization enabled.
9420 Although the behavior is similar to the Gold Linker's ICF optimization, GCC ICF
9421 works on different levels and thus the optimizations are not same - there are
9422 equivalences that are found only by GCC and equivalences found only by Gold.
9424 This flag is enabled by default at @option{-O2} and @option{-Os}.
9426 @item -flive-patching=@var{level}
9427 @opindex flive-patching
9428 Control GCC's optimizations to produce output suitable for live-patching.
9430 If the compiler's optimization uses a function's body or information extracted
9431 from its body to optimize/change another function, the latter is called an
9432 impacted function of the former. If a function is patched, its impacted
9433 functions should be patched too.
9435 The impacted functions are determined by the compiler's interprocedural
9436 optimizations. For example, a caller is impacted when inlining a function
9438 cloning a function and changing its caller to call this new clone,
9439 or extracting a function's pureness/constness information to optimize
9440 its direct or indirect callers, etc.
9442 Usually, the more IPA optimizations enabled, the larger the number of
9443 impacted functions for each function. In order to control the number of
9444 impacted functions and more easily compute the list of impacted function,
9445 IPA optimizations can be partially enabled at two different levels.
9447 The @var{level} argument should be one of the following:
9453 Only enable inlining and cloning optimizations, which includes inlining,
9454 cloning, interprocedural scalar replacement of aggregates and partial inlining.
9455 As a result, when patching a function, all its callers and its clones'
9456 callers are impacted, therefore need to be patched as well.
9458 @option{-flive-patching=inline-clone} disables the following optimization flags:
9459 @gccoptlist{-fwhole-program -fipa-pta -fipa-reference -fipa-ra @gol
9460 -fipa-icf -fipa-icf-functions -fipa-icf-variables @gol
9461 -fipa-bit-cp -fipa-vrp -fipa-pure-const -fipa-reference-addressable @gol
9462 -fipa-stack-alignment}
9464 @item inline-only-static
9466 Only enable inlining of static functions.
9467 As a result, when patching a static function, all its callers are impacted
9468 and so need to be patched as well.
9470 In addition to all the flags that @option{-flive-patching=inline-clone}
9472 @option{-flive-patching=inline-only-static} disables the following additional
9474 @gccoptlist{-fipa-cp-clone -fipa-sra -fpartial-inlining -fipa-cp}
9478 When @option{-flive-patching} is specified without any value, the default value
9479 is @var{inline-clone}.
9481 This flag is disabled by default.
9483 Note that @option{-flive-patching} is not supported with link-time optimization
9486 @item -fisolate-erroneous-paths-dereference
9487 @opindex fisolate-erroneous-paths-dereference
9488 Detect paths that trigger erroneous or undefined behavior due to
9489 dereferencing a null pointer. Isolate those paths from the main control
9490 flow and turn the statement with erroneous or undefined behavior into a trap.
9491 This flag is enabled by default at @option{-O2} and higher and depends on
9492 @option{-fdelete-null-pointer-checks} also being enabled.
9494 @item -fisolate-erroneous-paths-attribute
9495 @opindex fisolate-erroneous-paths-attribute
9496 Detect paths that trigger erroneous or undefined behavior due to a null value
9497 being used in a way forbidden by a @code{returns_nonnull} or @code{nonnull}
9498 attribute. Isolate those paths from the main control flow and turn the
9499 statement with erroneous or undefined behavior into a trap. This is not
9500 currently enabled, but may be enabled by @option{-O2} in the future.
9504 Perform forward store motion on trees. This flag is
9505 enabled by default at @option{-O} and higher.
9507 @item -ftree-bit-ccp
9508 @opindex ftree-bit-ccp
9509 Perform sparse conditional bit constant propagation on trees and propagate
9510 pointer alignment information.
9511 This pass only operates on local scalar variables and is enabled by default
9512 at @option{-O1} and higher, except for @option{-Og}.
9513 It requires that @option{-ftree-ccp} is enabled.
9517 Perform sparse conditional constant propagation (CCP) on trees. This
9518 pass only operates on local scalar variables and is enabled by default
9519 at @option{-O} and higher.
9521 @item -fssa-backprop
9522 @opindex fssa-backprop
9523 Propagate information about uses of a value up the definition chain
9524 in order to simplify the definitions. For example, this pass strips
9525 sign operations if the sign of a value never matters. The flag is
9526 enabled by default at @option{-O} and higher.
9529 @opindex fssa-phiopt
9530 Perform pattern matching on SSA PHI nodes to optimize conditional
9531 code. This pass is enabled by default at @option{-O1} and higher,
9532 except for @option{-Og}.
9534 @item -ftree-switch-conversion
9535 @opindex ftree-switch-conversion
9536 Perform conversion of simple initializations in a switch to
9537 initializations from a scalar array. This flag is enabled by default
9538 at @option{-O2} and higher.
9540 @item -ftree-tail-merge
9541 @opindex ftree-tail-merge
9542 Look for identical code sequences. When found, replace one with a jump to the
9543 other. This optimization is known as tail merging or cross jumping. This flag
9544 is enabled by default at @option{-O2} and higher. The compilation time
9546 be limited using @option{max-tail-merge-comparisons} parameter and
9547 @option{max-tail-merge-iterations} parameter.
9551 Perform dead code elimination (DCE) on trees. This flag is enabled by
9552 default at @option{-O} and higher.
9554 @item -ftree-builtin-call-dce
9555 @opindex ftree-builtin-call-dce
9556 Perform conditional dead code elimination (DCE) for calls to built-in functions
9557 that may set @code{errno} but are otherwise free of side effects. This flag is
9558 enabled by default at @option{-O2} and higher if @option{-Os} is not also
9561 @item -ffinite-loops
9562 @opindex ffinite-loops
9563 @opindex fno-finite-loops
9564 Assume that a loop with an exit will eventually take the exit and not loop
9565 indefinitely. This allows the compiler to remove loops that otherwise have
9566 no side-effects, not considering eventual endless looping as such.
9568 This option is enabled by default at @option{-O2}.
9570 @item -ftree-dominator-opts
9571 @opindex ftree-dominator-opts
9572 Perform a variety of simple scalar cleanups (constant/copy
9573 propagation, redundancy elimination, range propagation and expression
9574 simplification) based on a dominator tree traversal. This also
9575 performs jump threading (to reduce jumps to jumps). This flag is
9576 enabled by default at @option{-O} and higher.
9580 Perform dead store elimination (DSE) on trees. A dead store is a store into
9581 a memory location that is later overwritten by another store without
9582 any intervening loads. In this case the earlier store can be deleted. This
9583 flag is enabled by default at @option{-O} and higher.
9587 Perform loop header copying on trees. This is beneficial since it increases
9588 effectiveness of code motion optimizations. It also saves one jump. This flag
9589 is enabled by default at @option{-O} and higher. It is not enabled
9590 for @option{-Os}, since it usually increases code size.
9592 @item -ftree-loop-optimize
9593 @opindex ftree-loop-optimize
9594 Perform loop optimizations on trees. This flag is enabled by default
9595 at @option{-O} and higher.
9597 @item -ftree-loop-linear
9598 @itemx -floop-strip-mine
9600 @opindex ftree-loop-linear
9601 @opindex floop-strip-mine
9602 @opindex floop-block
9603 Perform loop nest optimizations. Same as
9604 @option{-floop-nest-optimize}. To use this code transformation, GCC has
9605 to be configured with @option{--with-isl} to enable the Graphite loop
9606 transformation infrastructure.
9608 @item -fgraphite-identity
9609 @opindex fgraphite-identity
9610 Enable the identity transformation for graphite. For every SCoP we generate
9611 the polyhedral representation and transform it back to gimple. Using
9612 @option{-fgraphite-identity} we can check the costs or benefits of the
9613 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
9614 are also performed by the code generator isl, like index splitting and
9615 dead code elimination in loops.
9617 @item -floop-nest-optimize
9618 @opindex floop-nest-optimize
9619 Enable the isl based loop nest optimizer. This is a generic loop nest
9620 optimizer based on the Pluto optimization algorithms. It calculates a loop
9621 structure optimized for data-locality and parallelism. This option
9624 @item -floop-parallelize-all
9625 @opindex floop-parallelize-all
9626 Use the Graphite data dependence analysis to identify loops that can
9627 be parallelized. Parallelize all the loops that can be analyzed to
9628 not contain loop carried dependences without checking that it is
9629 profitable to parallelize the loops.
9631 @item -ftree-coalesce-vars
9632 @opindex ftree-coalesce-vars
9633 While transforming the program out of the SSA representation, attempt to
9634 reduce copying by coalescing versions of different user-defined
9635 variables, instead of just compiler temporaries. This may severely
9636 limit the ability to debug an optimized program compiled with
9637 @option{-fno-var-tracking-assignments}. In the negated form, this flag
9638 prevents SSA coalescing of user variables. This option is enabled by
9639 default if optimization is enabled, and it does very little otherwise.
9641 @item -ftree-loop-if-convert
9642 @opindex ftree-loop-if-convert
9643 Attempt to transform conditional jumps in the innermost loops to
9644 branch-less equivalents. The intent is to remove control-flow from
9645 the innermost loops in order to improve the ability of the
9646 vectorization pass to handle these loops. This is enabled by default
9647 if vectorization is enabled.
9649 @item -ftree-loop-distribution
9650 @opindex ftree-loop-distribution
9651 Perform loop distribution. This flag can improve cache performance on
9652 big loop bodies and allow further loop optimizations, like
9653 parallelization or vectorization, to take place. For example, the loop
9669 This flag is enabled by default at @option{-O3}.
9670 It is also enabled by @option{-fprofile-use} and @option{-fauto-profile}.
9672 @item -ftree-loop-distribute-patterns
9673 @opindex ftree-loop-distribute-patterns
9674 Perform loop distribution of patterns that can be code generated with
9675 calls to a library. This flag is enabled by default at @option{-O3}, and
9676 by @option{-fprofile-use} and @option{-fauto-profile}.
9678 This pass distributes the initialization loops and generates a call to
9679 memset zero. For example, the loop
9695 and the initialization loop is transformed into a call to memset zero.
9696 This flag is enabled by default at @option{-O3}.
9697 It is also enabled by @option{-fprofile-use} and @option{-fauto-profile}.
9699 @item -floop-interchange
9700 @opindex floop-interchange
9701 Perform loop interchange outside of graphite. This flag can improve cache
9702 performance on loop nest and allow further loop optimizations, like
9703 vectorization, to take place. For example, the loop
9705 for (int i = 0; i < N; i++)
9706 for (int j = 0; j < N; j++)
9707 for (int k = 0; k < N; k++)
9708 c[i][j] = c[i][j] + a[i][k]*b[k][j];
9712 for (int i = 0; i < N; i++)
9713 for (int k = 0; k < N; k++)
9714 for (int j = 0; j < N; j++)
9715 c[i][j] = c[i][j] + a[i][k]*b[k][j];
9717 This flag is enabled by default at @option{-O3}.
9718 It is also enabled by @option{-fprofile-use} and @option{-fauto-profile}.
9720 @item -floop-unroll-and-jam
9721 @opindex floop-unroll-and-jam
9722 Apply unroll and jam transformations on feasible loops. In a loop
9723 nest this unrolls the outer loop by some factor and fuses the resulting
9724 multiple inner loops. This flag is enabled by default at @option{-O3}.
9725 It is also enabled by @option{-fprofile-use} and @option{-fauto-profile}.
9727 @item -ftree-loop-im
9728 @opindex ftree-loop-im
9729 Perform loop invariant motion on trees. This pass moves only invariants that
9730 are hard to handle at RTL level (function calls, operations that expand to
9731 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
9732 operands of conditions that are invariant out of the loop, so that we can use
9733 just trivial invariantness analysis in loop unswitching. The pass also includes
9736 @item -ftree-loop-ivcanon
9737 @opindex ftree-loop-ivcanon
9738 Create a canonical counter for number of iterations in loops for which
9739 determining number of iterations requires complicated analysis. Later
9740 optimizations then may determine the number easily. Useful especially
9741 in connection with unrolling.
9743 @item -ftree-scev-cprop
9744 @opindex ftree-scev-cprop
9745 Perform final value replacement. If a variable is modified in a loop
9746 in such a way that its value when exiting the loop can be determined using
9747 only its initial value and the number of loop iterations, replace uses of
9748 the final value by such a computation, provided it is sufficiently cheap.
9749 This reduces data dependencies and may allow further simplifications.
9750 Enabled by default at @option{-O} and higher.
9754 Perform induction variable optimizations (strength reduction, induction
9755 variable merging and induction variable elimination) on trees.
9757 @item -ftree-parallelize-loops=n
9758 @opindex ftree-parallelize-loops
9759 Parallelize loops, i.e., split their iteration space to run in n threads.
9760 This is only possible for loops whose iterations are independent
9761 and can be arbitrarily reordered. The optimization is only
9762 profitable on multiprocessor machines, for loops that are CPU-intensive,
9763 rather than constrained e.g.@: by memory bandwidth. This option
9764 implies @option{-pthread}, and thus is only supported on targets
9765 that have support for @option{-pthread}.
9769 Perform function-local points-to analysis on trees. This flag is
9770 enabled by default at @option{-O1} and higher, except for @option{-Og}.
9774 Perform scalar replacement of aggregates. This pass replaces structure
9775 references with scalars to prevent committing structures to memory too
9776 early. This flag is enabled by default at @option{-O1} and higher,
9777 except for @option{-Og}.
9779 @item -fstore-merging
9780 @opindex fstore-merging
9781 Perform merging of narrow stores to consecutive memory addresses. This pass
9782 merges contiguous stores of immediate values narrower than a word into fewer
9783 wider stores to reduce the number of instructions. This is enabled by default
9784 at @option{-O2} and higher as well as @option{-Os}.
9788 Perform temporary expression replacement during the SSA->normal phase. Single
9789 use/single def temporaries are replaced at their use location with their
9790 defining expression. This results in non-GIMPLE code, but gives the expanders
9791 much more complex trees to work on resulting in better RTL generation. This is
9792 enabled by default at @option{-O} and higher.
9796 Perform straight-line strength reduction on trees. This recognizes related
9797 expressions involving multiplications and replaces them by less expensive
9798 calculations when possible. This is enabled by default at @option{-O} and
9801 @item -ftree-vectorize
9802 @opindex ftree-vectorize
9803 Perform vectorization on trees. This flag enables @option{-ftree-loop-vectorize}
9804 and @option{-ftree-slp-vectorize} if not explicitly specified.
9806 @item -ftree-loop-vectorize
9807 @opindex ftree-loop-vectorize
9808 Perform loop vectorization on trees. This flag is enabled by default at
9809 @option{-O3} and by @option{-ftree-vectorize}, @option{-fprofile-use},
9810 and @option{-fauto-profile}.
9812 @item -ftree-slp-vectorize
9813 @opindex ftree-slp-vectorize
9814 Perform basic block vectorization on trees. This flag is enabled by default at
9815 @option{-O3} and by @option{-ftree-vectorize}, @option{-fprofile-use},
9816 and @option{-fauto-profile}.
9818 @item -fvect-cost-model=@var{model}
9819 @opindex fvect-cost-model
9820 Alter the cost model used for vectorization. The @var{model} argument
9821 should be one of @samp{unlimited}, @samp{dynamic} or @samp{cheap}.
9822 With the @samp{unlimited} model the vectorized code-path is assumed
9823 to be profitable while with the @samp{dynamic} model a runtime check
9824 guards the vectorized code-path to enable it only for iteration
9825 counts that will likely execute faster than when executing the original
9826 scalar loop. The @samp{cheap} model disables vectorization of
9827 loops where doing so would be cost prohibitive for example due to
9828 required runtime checks for data dependence or alignment but otherwise
9829 is equal to the @samp{dynamic} model.
9830 The default cost model depends on other optimization flags and is
9831 either @samp{dynamic} or @samp{cheap}.
9833 @item -fsimd-cost-model=@var{model}
9834 @opindex fsimd-cost-model
9835 Alter the cost model used for vectorization of loops marked with the OpenMP
9836 simd directive. The @var{model} argument should be one of
9837 @samp{unlimited}, @samp{dynamic}, @samp{cheap}. All values of @var{model}
9838 have the same meaning as described in @option{-fvect-cost-model} and by
9839 default a cost model defined with @option{-fvect-cost-model} is used.
9843 Perform Value Range Propagation on trees. This is similar to the
9844 constant propagation pass, but instead of values, ranges of values are
9845 propagated. This allows the optimizers to remove unnecessary range
9846 checks like array bound checks and null pointer checks. This is
9847 enabled by default at @option{-O2} and higher. Null pointer check
9848 elimination is only done if @option{-fdelete-null-pointer-checks} is
9852 @opindex fsplit-paths
9853 Split paths leading to loop backedges. This can improve dead code
9854 elimination and common subexpression elimination. This is enabled by
9855 default at @option{-O3} and above.
9857 @item -fsplit-ivs-in-unroller
9858 @opindex fsplit-ivs-in-unroller
9859 Enables expression of values of induction variables in later iterations
9860 of the unrolled loop using the value in the first iteration. This breaks
9861 long dependency chains, thus improving efficiency of the scheduling passes.
9863 A combination of @option{-fweb} and CSE is often sufficient to obtain the
9864 same effect. However, that is not reliable in cases where the loop body
9865 is more complicated than a single basic block. It also does not work at all
9866 on some architectures due to restrictions in the CSE pass.
9868 This optimization is enabled by default.
9870 @item -fvariable-expansion-in-unroller
9871 @opindex fvariable-expansion-in-unroller
9872 With this option, the compiler creates multiple copies of some
9873 local variables when unrolling a loop, which can result in superior code.
9875 This optimization is enabled by default for PowerPC targets, but disabled
9876 by default otherwise.
9878 @item -fpartial-inlining
9879 @opindex fpartial-inlining
9880 Inline parts of functions. This option has any effect only
9881 when inlining itself is turned on by the @option{-finline-functions}
9882 or @option{-finline-small-functions} options.
9884 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
9886 @item -fpredictive-commoning
9887 @opindex fpredictive-commoning
9888 Perform predictive commoning optimization, i.e., reusing computations
9889 (especially memory loads and stores) performed in previous
9890 iterations of loops.
9892 This option is enabled at level @option{-O3}.
9893 It is also enabled by @option{-fprofile-use} and @option{-fauto-profile}.
9895 @item -fprefetch-loop-arrays
9896 @opindex fprefetch-loop-arrays
9897 If supported by the target machine, generate instructions to prefetch
9898 memory to improve the performance of loops that access large arrays.
9900 This option may generate better or worse code; results are highly
9901 dependent on the structure of loops within the source code.
9903 Disabled at level @option{-Os}.
9905 @item -fno-printf-return-value
9906 @opindex fno-printf-return-value
9907 @opindex fprintf-return-value
9908 Do not substitute constants for known return value of formatted output
9909 functions such as @code{sprintf}, @code{snprintf}, @code{vsprintf}, and
9910 @code{vsnprintf} (but not @code{printf} of @code{fprintf}). This
9911 transformation allows GCC to optimize or even eliminate branches based
9912 on the known return value of these functions called with arguments that
9913 are either constant, or whose values are known to be in a range that
9914 makes determining the exact return value possible. For example, when
9915 @option{-fprintf-return-value} is in effect, both the branch and the
9916 body of the @code{if} statement (but not the call to @code{snprint})
9917 can be optimized away when @code{i} is a 32-bit or smaller integer
9918 because the return value is guaranteed to be at most 8.
9922 if (snprintf (buf, "%08x", i) >= sizeof buf)
9926 The @option{-fprintf-return-value} option relies on other optimizations
9927 and yields best results with @option{-O2} and above. It works in tandem
9928 with the @option{-Wformat-overflow} and @option{-Wformat-truncation}
9929 options. The @option{-fprintf-return-value} option is enabled by default.
9932 @itemx -fno-peephole2
9933 @opindex fno-peephole
9935 @opindex fno-peephole2
9937 Disable any machine-specific peephole optimizations. The difference
9938 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
9939 are implemented in the compiler; some targets use one, some use the
9940 other, a few use both.
9942 @option{-fpeephole} is enabled by default.
9943 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
9945 @item -fno-guess-branch-probability
9946 @opindex fno-guess-branch-probability
9947 @opindex fguess-branch-probability
9948 Do not guess branch probabilities using heuristics.
9950 GCC uses heuristics to guess branch probabilities if they are
9951 not provided by profiling feedback (@option{-fprofile-arcs}). These
9952 heuristics are based on the control flow graph. If some branch probabilities
9953 are specified by @code{__builtin_expect}, then the heuristics are
9954 used to guess branch probabilities for the rest of the control flow graph,
9955 taking the @code{__builtin_expect} info into account. The interactions
9956 between the heuristics and @code{__builtin_expect} can be complex, and in
9957 some cases, it may be useful to disable the heuristics so that the effects
9958 of @code{__builtin_expect} are easier to understand.
9960 It is also possible to specify expected probability of the expression
9961 with @code{__builtin_expect_with_probability} built-in function.
9963 The default is @option{-fguess-branch-probability} at levels
9964 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
9966 @item -freorder-blocks
9967 @opindex freorder-blocks
9968 Reorder basic blocks in the compiled function in order to reduce number of
9969 taken branches and improve code locality.
9971 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
9973 @item -freorder-blocks-algorithm=@var{algorithm}
9974 @opindex freorder-blocks-algorithm
9975 Use the specified algorithm for basic block reordering. The
9976 @var{algorithm} argument can be @samp{simple}, which does not increase
9977 code size (except sometimes due to secondary effects like alignment),
9978 or @samp{stc}, the ``software trace cache'' algorithm, which tries to
9979 put all often executed code together, minimizing the number of branches
9980 executed by making extra copies of code.
9982 The default is @samp{simple} at levels @option{-O}, @option{-Os}, and
9983 @samp{stc} at levels @option{-O2}, @option{-O3}.
9985 @item -freorder-blocks-and-partition
9986 @opindex freorder-blocks-and-partition
9987 In addition to reordering basic blocks in the compiled function, in order
9988 to reduce number of taken branches, partitions hot and cold basic blocks
9989 into separate sections of the assembly and @file{.o} files, to improve
9990 paging and cache locality performance.
9992 This optimization is automatically turned off in the presence of
9993 exception handling or unwind tables (on targets using setjump/longjump or target specific scheme), for linkonce sections, for functions with a user-defined
9994 section attribute and on any architecture that does not support named
9995 sections. When @option{-fsplit-stack} is used this option is not
9996 enabled by default (to avoid linker errors), but may be enabled
9997 explicitly (if using a working linker).
9999 Enabled for x86 at levels @option{-O2}, @option{-O3}, @option{-Os}.
10001 @item -freorder-functions
10002 @opindex freorder-functions
10003 Reorder functions in the object file in order to
10004 improve code locality. This is implemented by using special
10005 subsections @code{.text.hot} for most frequently executed functions and
10006 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
10007 the linker so object file format must support named sections and linker must
10008 place them in a reasonable way.
10010 This option isn't effective unless you either provide profile feedback
10011 (see @option{-fprofile-arcs} for details) or manually annotate functions with
10012 @code{hot} or @code{cold} attributes (@pxref{Common Function Attributes}).
10014 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
10016 @item -fstrict-aliasing
10017 @opindex fstrict-aliasing
10018 Allow the compiler to assume the strictest aliasing rules applicable to
10019 the language being compiled. For C (and C++), this activates
10020 optimizations based on the type of expressions. In particular, an
10021 object of one type is assumed never to reside at the same address as an
10022 object of a different type, unless the types are almost the same. For
10023 example, an @code{unsigned int} can alias an @code{int}, but not a
10024 @code{void*} or a @code{double}. A character type may alias any other
10027 @anchor{Type-punning}Pay special attention to code like this:
10040 The practice of reading from a different union member than the one most
10041 recently written to (called ``type-punning'') is common. Even with
10042 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
10043 is accessed through the union type. So, the code above works as
10044 expected. @xref{Structures unions enumerations and bit-fields
10045 implementation}. However, this code might not:
10056 Similarly, access by taking the address, casting the resulting pointer
10057 and dereferencing the result has undefined behavior, even if the cast
10058 uses a union type, e.g.:
10062 return ((union a_union *) &d)->i;
10066 The @option{-fstrict-aliasing} option is enabled at levels
10067 @option{-O2}, @option{-O3}, @option{-Os}.
10069 @item -falign-functions
10070 @itemx -falign-functions=@var{n}
10071 @itemx -falign-functions=@var{n}:@var{m}
10072 @itemx -falign-functions=@var{n}:@var{m}:@var{n2}
10073 @itemx -falign-functions=@var{n}:@var{m}:@var{n2}:@var{m2}
10074 @opindex falign-functions
10075 Align the start of functions to the next power-of-two greater than
10076 @var{n}, skipping up to @var{m}-1 bytes. This ensures that at least
10077 the first @var{m} bytes of the function can be fetched by the CPU
10078 without crossing an @var{n}-byte alignment boundary.
10080 If @var{m} is not specified, it defaults to @var{n}.
10082 Examples: @option{-falign-functions=32} aligns functions to the next
10083 32-byte boundary, @option{-falign-functions=24} aligns to the next
10084 32-byte boundary only if this can be done by skipping 23 bytes or less,
10085 @option{-falign-functions=32:7} aligns to the next
10086 32-byte boundary only if this can be done by skipping 6 bytes or less.
10088 The second pair of @var{n2}:@var{m2} values allows you to specify
10089 a secondary alignment: @option{-falign-functions=64:7:32:3} aligns to
10090 the next 64-byte boundary if this can be done by skipping 6 bytes or less,
10091 otherwise aligns to the next 32-byte boundary if this can be done
10092 by skipping 2 bytes or less.
10093 If @var{m2} is not specified, it defaults to @var{n2}.
10095 Some assemblers only support this flag when @var{n} is a power of two;
10096 in that case, it is rounded up.
10098 @option{-fno-align-functions} and @option{-falign-functions=1} are
10099 equivalent and mean that functions are not aligned.
10101 If @var{n} is not specified or is zero, use a machine-dependent default.
10102 The maximum allowed @var{n} option value is 65536.
10104 Enabled at levels @option{-O2}, @option{-O3}.
10106 @item -flimit-function-alignment
10107 If this option is enabled, the compiler tries to avoid unnecessarily
10108 overaligning functions. It attempts to instruct the assembler to align
10109 by the amount specified by @option{-falign-functions}, but not to
10110 skip more bytes than the size of the function.
10112 @item -falign-labels
10113 @itemx -falign-labels=@var{n}
10114 @itemx -falign-labels=@var{n}:@var{m}
10115 @itemx -falign-labels=@var{n}:@var{m}:@var{n2}
10116 @itemx -falign-labels=@var{n}:@var{m}:@var{n2}:@var{m2}
10117 @opindex falign-labels
10118 Align all branch targets to a power-of-two boundary.
10120 Parameters of this option are analogous to the @option{-falign-functions} option.
10121 @option{-fno-align-labels} and @option{-falign-labels=1} are
10122 equivalent and mean that labels are not aligned.
10124 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
10125 are greater than this value, then their values are used instead.
10127 If @var{n} is not specified or is zero, use a machine-dependent default
10128 which is very likely to be @samp{1}, meaning no alignment.
10129 The maximum allowed @var{n} option value is 65536.
10131 Enabled at levels @option{-O2}, @option{-O3}.
10133 @item -falign-loops
10134 @itemx -falign-loops=@var{n}
10135 @itemx -falign-loops=@var{n}:@var{m}
10136 @itemx -falign-loops=@var{n}:@var{m}:@var{n2}
10137 @itemx -falign-loops=@var{n}:@var{m}:@var{n2}:@var{m2}
10138 @opindex falign-loops
10139 Align loops to a power-of-two boundary. If the loops are executed
10140 many times, this makes up for any execution of the dummy padding
10143 Parameters of this option are analogous to the @option{-falign-functions} option.
10144 @option{-fno-align-loops} and @option{-falign-loops=1} are
10145 equivalent and mean that loops are not aligned.
10146 The maximum allowed @var{n} option value is 65536.
10148 If @var{n} is not specified or is zero, use a machine-dependent default.
10150 Enabled at levels @option{-O2}, @option{-O3}.
10152 @item -falign-jumps
10153 @itemx -falign-jumps=@var{n}
10154 @itemx -falign-jumps=@var{n}:@var{m}
10155 @itemx -falign-jumps=@var{n}:@var{m}:@var{n2}
10156 @itemx -falign-jumps=@var{n}:@var{m}:@var{n2}:@var{m2}
10157 @opindex falign-jumps
10158 Align branch targets to a power-of-two boundary, for branch targets
10159 where the targets can only be reached by jumping. In this case,
10160 no dummy operations need be executed.
10162 Parameters of this option are analogous to the @option{-falign-functions} option.
10163 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
10164 equivalent and mean that loops are not aligned.
10166 If @var{n} is not specified or is zero, use a machine-dependent default.
10167 The maximum allowed @var{n} option value is 65536.
10169 Enabled at levels @option{-O2}, @option{-O3}.
10171 @item -funit-at-a-time
10172 @opindex funit-at-a-time
10173 This option is left for compatibility reasons. @option{-funit-at-a-time}
10174 has no effect, while @option{-fno-unit-at-a-time} implies
10175 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
10177 Enabled by default.
10179 @item -fno-toplevel-reorder
10180 @opindex fno-toplevel-reorder
10181 @opindex ftoplevel-reorder
10182 Do not reorder top-level functions, variables, and @code{asm}
10183 statements. Output them in the same order that they appear in the
10184 input file. When this option is used, unreferenced static variables
10185 are not removed. This option is intended to support existing code
10186 that relies on a particular ordering. For new code, it is better to
10187 use attributes when possible.
10189 @option{-ftoplevel-reorder} is the default at @option{-O1} and higher, and
10190 also at @option{-O0} if @option{-fsection-anchors} is explicitly requested.
10191 Additionally @option{-fno-toplevel-reorder} implies
10192 @option{-fno-section-anchors}.
10196 Constructs webs as commonly used for register allocation purposes and assign
10197 each web individual pseudo register. This allows the register allocation pass
10198 to operate on pseudos directly, but also strengthens several other optimization
10199 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
10200 however, make debugging impossible, since variables no longer stay in a
10203 Enabled by default with @option{-funroll-loops}.
10205 @item -fwhole-program
10206 @opindex fwhole-program
10207 Assume that the current compilation unit represents the whole program being
10208 compiled. All public functions and variables with the exception of @code{main}
10209 and those merged by attribute @code{externally_visible} become static functions
10210 and in effect are optimized more aggressively by interprocedural optimizers.
10212 This option should not be used in combination with @option{-flto}.
10213 Instead relying on a linker plugin should provide safer and more precise
10216 @item -flto[=@var{n}]
10218 This option runs the standard link-time optimizer. When invoked
10219 with source code, it generates GIMPLE (one of GCC's internal
10220 representations) and writes it to special ELF sections in the object
10221 file. When the object files are linked together, all the function
10222 bodies are read from these ELF sections and instantiated as if they
10223 had been part of the same translation unit.
10225 To use the link-time optimizer, @option{-flto} and optimization
10226 options should be specified at compile time and during the final link.
10227 It is recommended that you compile all the files participating in the
10228 same link with the same options and also specify those options at
10233 gcc -c -O2 -flto foo.c
10234 gcc -c -O2 -flto bar.c
10235 gcc -o myprog -flto -O2 foo.o bar.o
10238 The first two invocations to GCC save a bytecode representation
10239 of GIMPLE into special ELF sections inside @file{foo.o} and
10240 @file{bar.o}. The final invocation reads the GIMPLE bytecode from
10241 @file{foo.o} and @file{bar.o}, merges the two files into a single
10242 internal image, and compiles the result as usual. Since both
10243 @file{foo.o} and @file{bar.o} are merged into a single image, this
10244 causes all the interprocedural analyses and optimizations in GCC to
10245 work across the two files as if they were a single one. This means,
10246 for example, that the inliner is able to inline functions in
10247 @file{bar.o} into functions in @file{foo.o} and vice-versa.
10249 Another (simpler) way to enable link-time optimization is:
10252 gcc -o myprog -flto -O2 foo.c bar.c
10255 The above generates bytecode for @file{foo.c} and @file{bar.c},
10256 merges them together into a single GIMPLE representation and optimizes
10257 them as usual to produce @file{myprog}.
10259 The important thing to keep in mind is that to enable link-time
10260 optimizations you need to use the GCC driver to perform the link step.
10261 GCC automatically performs link-time optimization if any of the
10262 objects involved were compiled with the @option{-flto} command-line option.
10263 You can always override
10264 the automatic decision to do link-time optimization
10265 by passing @option{-fno-lto} to the link command.
10267 To make whole program optimization effective, it is necessary to make
10268 certain whole program assumptions. The compiler needs to know
10269 what functions and variables can be accessed by libraries and runtime
10270 outside of the link-time optimized unit. When supported by the linker,
10271 the linker plugin (see @option{-fuse-linker-plugin}) passes information
10272 to the compiler about used and externally visible symbols. When
10273 the linker plugin is not available, @option{-fwhole-program} should be
10274 used to allow the compiler to make these assumptions, which leads
10275 to more aggressive optimization decisions.
10277 When a file is compiled with @option{-flto} without
10278 @option{-fuse-linker-plugin}, the generated object file is larger than
10279 a regular object file because it contains GIMPLE bytecodes and the usual
10280 final code (see @option{-ffat-lto-objects}. This means that
10281 object files with LTO information can be linked as normal object
10282 files; if @option{-fno-lto} is passed to the linker, no
10283 interprocedural optimizations are applied. Note that when
10284 @option{-fno-fat-lto-objects} is enabled the compile stage is faster
10285 but you cannot perform a regular, non-LTO link on them.
10287 When producing the final binary, GCC only
10288 applies link-time optimizations to those files that contain bytecode.
10289 Therefore, you can mix and match object files and libraries with
10290 GIMPLE bytecodes and final object code. GCC automatically selects
10291 which files to optimize in LTO mode and which files to link without
10292 further processing.
10294 Generally, options specified at link time override those
10295 specified at compile time, although in some cases GCC attempts to infer
10296 link-time options from the settings used to compile the input files.
10298 If you do not specify an optimization level option @option{-O} at
10299 link time, then GCC uses the highest optimization level
10300 used when compiling the object files. Note that it is generally
10301 ineffective to specify an optimization level option only at link time and
10302 not at compile time, for two reasons. First, compiling without
10303 optimization suppresses compiler passes that gather information
10304 needed for effective optimization at link time. Second, some early
10305 optimization passes can be performed only at compile time and
10308 There are some code generation flags preserved by GCC when
10309 generating bytecodes, as they need to be used during the final link.
10310 Currently, the following options and their settings are taken from
10311 the first object file that explicitly specifies them:
10312 @option{-fPIC}, @option{-fpic}, @option{-fpie}, @option{-fcommon},
10313 @option{-fexceptions}, @option{-fnon-call-exceptions}, @option{-fgnu-tm}
10314 and all the @option{-m} target flags.
10316 Certain ABI-changing flags are required to match in all compilation units,
10317 and trying to override this at link time with a conflicting value
10318 is ignored. This includes options such as @option{-freg-struct-return}
10319 and @option{-fpcc-struct-return}.
10321 Other options such as @option{-ffp-contract}, @option{-fno-strict-overflow},
10322 @option{-fwrapv}, @option{-fno-trapv} or @option{-fno-strict-aliasing}
10323 are passed through to the link stage and merged conservatively for
10324 conflicting translation units. Specifically
10325 @option{-fno-strict-overflow}, @option{-fwrapv} and @option{-fno-trapv} take
10326 precedence; and for example @option{-ffp-contract=off} takes precedence
10327 over @option{-ffp-contract=fast}. You can override them at link time.
10329 If LTO encounters objects with C linkage declared with incompatible
10330 types in separate translation units to be linked together (undefined
10331 behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
10332 issued. The behavior is still undefined at run time. Similar
10333 diagnostics may be raised for other languages.
10335 Another feature of LTO is that it is possible to apply interprocedural
10336 optimizations on files written in different languages:
10340 g++ -c -flto bar.cc
10341 gfortran -c -flto baz.f90
10342 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
10345 Notice that the final link is done with @command{g++} to get the C++
10346 runtime libraries and @option{-lgfortran} is added to get the Fortran
10347 runtime libraries. In general, when mixing languages in LTO mode, you
10348 should use the same link command options as when mixing languages in a
10349 regular (non-LTO) compilation.
10351 If object files containing GIMPLE bytecode are stored in a library archive, say
10352 @file{libfoo.a}, it is possible to extract and use them in an LTO link if you
10353 are using a linker with plugin support. To create static libraries suitable
10354 for LTO, use @command{gcc-ar} and @command{gcc-ranlib} instead of @command{ar}
10355 and @command{ranlib};
10356 to show the symbols of object files with GIMPLE bytecode, use
10357 @command{gcc-nm}. Those commands require that @command{ar}, @command{ranlib}
10358 and @command{nm} have been compiled with plugin support. At link time, use the
10359 flag @option{-fuse-linker-plugin} to ensure that the library participates in
10360 the LTO optimization process:
10363 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
10366 With the linker plugin enabled, the linker extracts the needed
10367 GIMPLE files from @file{libfoo.a} and passes them on to the running GCC
10368 to make them part of the aggregated GIMPLE image to be optimized.
10370 If you are not using a linker with plugin support and/or do not
10371 enable the linker plugin, then the objects inside @file{libfoo.a}
10372 are extracted and linked as usual, but they do not participate
10373 in the LTO optimization process. In order to make a static library suitable
10374 for both LTO optimization and usual linkage, compile its object files with
10375 @option{-flto} @option{-ffat-lto-objects}.
10377 Link-time optimizations do not require the presence of the whole program to
10378 operate. If the program does not require any symbols to be exported, it is
10379 possible to combine @option{-flto} and @option{-fwhole-program} to allow
10380 the interprocedural optimizers to use more aggressive assumptions which may
10381 lead to improved optimization opportunities.
10382 Use of @option{-fwhole-program} is not needed when linker plugin is
10383 active (see @option{-fuse-linker-plugin}).
10385 The current implementation of LTO makes no
10386 attempt to generate bytecode that is portable between different
10387 types of hosts. The bytecode files are versioned and there is a
10388 strict version check, so bytecode files generated in one version of
10389 GCC do not work with an older or newer version of GCC.
10391 Link-time optimization does not work well with generation of debugging
10392 information on systems other than those using a combination of ELF and
10395 If you specify the optional @var{n}, the optimization and code
10396 generation done at link time is executed in parallel using @var{n}
10397 parallel jobs by utilizing an installed @command{make} program. The
10398 environment variable @env{MAKE} may be used to override the program
10399 used. The default value for @var{n} is automatically detected based
10400 on number of cores.
10402 You can also specify @option{-flto=jobserver} to use GNU make's
10403 job server mode to determine the number of parallel jobs. This
10404 is useful when the Makefile calling GCC is already executing in parallel.
10405 You must prepend a @samp{+} to the command recipe in the parent Makefile
10406 for this to work. This option likely only works if @env{MAKE} is
10409 @item -flto-partition=@var{alg}
10410 @opindex flto-partition
10411 Specify the partitioning algorithm used by the link-time optimizer.
10412 The value is either @samp{1to1} to specify a partitioning mirroring
10413 the original source files or @samp{balanced} to specify partitioning
10414 into equally sized chunks (whenever possible) or @samp{max} to create
10415 new partition for every symbol where possible. Specifying @samp{none}
10416 as an algorithm disables partitioning and streaming completely.
10417 The default value is @samp{balanced}. While @samp{1to1} can be used
10418 as an workaround for various code ordering issues, the @samp{max}
10419 partitioning is intended for internal testing only.
10420 The value @samp{one} specifies that exactly one partition should be
10421 used while the value @samp{none} bypasses partitioning and executes
10422 the link-time optimization step directly from the WPA phase.
10424 @item -flto-compression-level=@var{n}
10425 @opindex flto-compression-level
10426 This option specifies the level of compression used for intermediate
10427 language written to LTO object files, and is only meaningful in
10428 conjunction with LTO mode (@option{-flto}). Valid
10429 values are 0 (no compression) to 9 (maximum compression). Values
10430 outside this range are clamped to either 0 or 9. If the option is not
10431 given, a default balanced compression setting is used.
10433 @item -fuse-linker-plugin
10434 @opindex fuse-linker-plugin
10435 Enables the use of a linker plugin during link-time optimization. This
10436 option relies on plugin support in the linker, which is available in gold
10437 or in GNU ld 2.21 or newer.
10439 This option enables the extraction of object files with GIMPLE bytecode out
10440 of library archives. This improves the quality of optimization by exposing
10441 more code to the link-time optimizer. This information specifies what
10442 symbols can be accessed externally (by non-LTO object or during dynamic
10443 linking). Resulting code quality improvements on binaries (and shared
10444 libraries that use hidden visibility) are similar to @option{-fwhole-program}.
10445 See @option{-flto} for a description of the effect of this flag and how to
10448 This option is enabled by default when LTO support in GCC is enabled
10449 and GCC was configured for use with
10450 a linker supporting plugins (GNU ld 2.21 or newer or gold).
10452 @item -ffat-lto-objects
10453 @opindex ffat-lto-objects
10454 Fat LTO objects are object files that contain both the intermediate language
10455 and the object code. This makes them usable for both LTO linking and normal
10456 linking. This option is effective only when compiling with @option{-flto}
10457 and is ignored at link time.
10459 @option{-fno-fat-lto-objects} improves compilation time over plain LTO, but
10460 requires the complete toolchain to be aware of LTO. It requires a linker with
10461 linker plugin support for basic functionality. Additionally,
10462 @command{nm}, @command{ar} and @command{ranlib}
10463 need to support linker plugins to allow a full-featured build environment
10464 (capable of building static libraries etc). GCC provides the @command{gcc-ar},
10465 @command{gcc-nm}, @command{gcc-ranlib} wrappers to pass the right options
10466 to these tools. With non fat LTO makefiles need to be modified to use them.
10468 Note that modern binutils provide plugin auto-load mechanism.
10469 Installing the linker plugin into @file{$libdir/bfd-plugins} has the same
10470 effect as usage of the command wrappers (@command{gcc-ar}, @command{gcc-nm} and
10471 @command{gcc-ranlib}).
10473 The default is @option{-fno-fat-lto-objects} on targets with linker plugin
10476 @item -fcompare-elim
10477 @opindex fcompare-elim
10478 After register allocation and post-register allocation instruction splitting,
10479 identify arithmetic instructions that compute processor flags similar to a
10480 comparison operation based on that arithmetic. If possible, eliminate the
10481 explicit comparison operation.
10483 This pass only applies to certain targets that cannot explicitly represent
10484 the comparison operation before register allocation is complete.
10486 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
10488 @item -fcprop-registers
10489 @opindex fcprop-registers
10490 After register allocation and post-register allocation instruction splitting,
10491 perform a copy-propagation pass to try to reduce scheduling dependencies
10492 and occasionally eliminate the copy.
10494 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
10496 @item -fprofile-correction
10497 @opindex fprofile-correction
10498 Profiles collected using an instrumented binary for multi-threaded programs may
10499 be inconsistent due to missed counter updates. When this option is specified,
10500 GCC uses heuristics to correct or smooth out such inconsistencies. By
10501 default, GCC emits an error message when an inconsistent profile is detected.
10503 This option is enabled by @option{-fauto-profile}.
10505 @item -fprofile-use
10506 @itemx -fprofile-use=@var{path}
10507 @opindex fprofile-use
10508 Enable profile feedback-directed optimizations,
10509 and the following optimizations, many of which
10510 are generally profitable only with profile feedback available:
10512 @gccoptlist{-fbranch-probabilities -fprofile-values @gol
10513 -funroll-loops -fpeel-loops -ftracer -fvpt @gol
10514 -finline-functions -fipa-cp -fipa-cp-clone -fipa-bit-cp @gol
10515 -fpredictive-commoning -fsplit-loops -funswitch-loops @gol
10516 -fgcse-after-reload -ftree-loop-vectorize -ftree-slp-vectorize @gol
10517 -fvect-cost-model=dynamic -ftree-loop-distribute-patterns @gol
10518 -fprofile-reorder-functions}
10520 Before you can use this option, you must first generate profiling information.
10521 @xref{Instrumentation Options}, for information about the
10522 @option{-fprofile-generate} option.
10524 By default, GCC emits an error message if the feedback profiles do not
10525 match the source code. This error can be turned into a warning by using
10526 @option{-Wno-error=coverage-mismatch}. Note this may result in poorly
10527 optimized code. Additionally, by default, GCC also emits a warning message if
10528 the feedback profiles do not exist (see @option{-Wmissing-profile}).
10530 If @var{path} is specified, GCC looks at the @var{path} to find
10531 the profile feedback data files. See @option{-fprofile-dir}.
10533 @item -fauto-profile
10534 @itemx -fauto-profile=@var{path}
10535 @opindex fauto-profile
10536 Enable sampling-based feedback-directed optimizations,
10537 and the following optimizations,
10538 many of which are generally profitable only with profile feedback available:
10540 @gccoptlist{-fbranch-probabilities -fprofile-values @gol
10541 -funroll-loops -fpeel-loops -ftracer -fvpt @gol
10542 -finline-functions -fipa-cp -fipa-cp-clone -fipa-bit-cp @gol
10543 -fpredictive-commoning -fsplit-loops -funswitch-loops @gol
10544 -fgcse-after-reload -ftree-loop-vectorize -ftree-slp-vectorize @gol
10545 -fvect-cost-model=dynamic -ftree-loop-distribute-patterns @gol
10546 -fprofile-correction}
10548 @var{path} is the name of a file containing AutoFDO profile information.
10549 If omitted, it defaults to @file{fbdata.afdo} in the current directory.
10551 Producing an AutoFDO profile data file requires running your program
10552 with the @command{perf} utility on a supported GNU/Linux target system.
10553 For more information, see @uref{https://perf.wiki.kernel.org/}.
10557 perf record -e br_inst_retired:near_taken -b -o perf.data \
10561 Then use the @command{create_gcov} tool to convert the raw profile data
10562 to a format that can be used by GCC.@ You must also supply the
10563 unstripped binary for your program to this tool.
10564 See @uref{https://github.com/google/autofdo}.
10568 create_gcov --binary=your_program.unstripped --profile=perf.data \
10569 --gcov=profile.afdo
10573 The following options control compiler behavior regarding floating-point
10574 arithmetic. These options trade off between speed and
10575 correctness. All must be specifically enabled.
10578 @item -ffloat-store
10579 @opindex ffloat-store
10580 Do not store floating-point variables in registers, and inhibit other
10581 options that might change whether a floating-point value is taken from a
10582 register or memory.
10584 @cindex floating-point precision
10585 This option prevents undesirable excess precision on machines such as
10586 the 68000 where the floating registers (of the 68881) keep more
10587 precision than a @code{double} is supposed to have. Similarly for the
10588 x86 architecture. For most programs, the excess precision does only
10589 good, but a few programs rely on the precise definition of IEEE floating
10590 point. Use @option{-ffloat-store} for such programs, after modifying
10591 them to store all pertinent intermediate computations into variables.
10593 @item -fexcess-precision=@var{style}
10594 @opindex fexcess-precision
10595 This option allows further control over excess precision on machines
10596 where floating-point operations occur in a format with more precision or
10597 range than the IEEE standard and interchange floating-point types. By
10598 default, @option{-fexcess-precision=fast} is in effect; this means that
10599 operations may be carried out in a wider precision than the types specified
10600 in the source if that would result in faster code, and it is unpredictable
10601 when rounding to the types specified in the source code takes place.
10602 When compiling C, if @option{-fexcess-precision=standard} is specified then
10603 excess precision follows the rules specified in ISO C99; in particular,
10604 both casts and assignments cause values to be rounded to their
10605 semantic types (whereas @option{-ffloat-store} only affects
10606 assignments). This option is enabled by default for C if a strict
10607 conformance option such as @option{-std=c99} is used.
10608 @option{-ffast-math} enables @option{-fexcess-precision=fast} by default
10609 regardless of whether a strict conformance option is used.
10612 @option{-fexcess-precision=standard} is not implemented for languages
10613 other than C. On the x86, it has no effect if @option{-mfpmath=sse}
10614 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
10615 semantics apply without excess precision, and in the latter, rounding
10619 @opindex ffast-math
10620 Sets the options @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
10621 @option{-ffinite-math-only}, @option{-fno-rounding-math},
10622 @option{-fno-signaling-nans}, @option{-fcx-limited-range} and
10623 @option{-fexcess-precision=fast}.
10625 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
10627 This option is not turned on by any @option{-O} option besides
10628 @option{-Ofast} since it can result in incorrect output for programs
10629 that depend on an exact implementation of IEEE or ISO rules/specifications
10630 for math functions. It may, however, yield faster code for programs
10631 that do not require the guarantees of these specifications.
10633 @item -fno-math-errno
10634 @opindex fno-math-errno
10635 @opindex fmath-errno
10636 Do not set @code{errno} after calling math functions that are executed
10637 with a single instruction, e.g., @code{sqrt}. A program that relies on
10638 IEEE exceptions for math error handling may want to use this flag
10639 for speed while maintaining IEEE arithmetic compatibility.
10641 This option is not turned on by any @option{-O} option since
10642 it can result in incorrect output for programs that depend on
10643 an exact implementation of IEEE or ISO rules/specifications for
10644 math functions. It may, however, yield faster code for programs
10645 that do not require the guarantees of these specifications.
10647 The default is @option{-fmath-errno}.
10649 On Darwin systems, the math library never sets @code{errno}. There is
10650 therefore no reason for the compiler to consider the possibility that
10651 it might, and @option{-fno-math-errno} is the default.
10653 @item -funsafe-math-optimizations
10654 @opindex funsafe-math-optimizations
10656 Allow optimizations for floating-point arithmetic that (a) assume
10657 that arguments and results are valid and (b) may violate IEEE or
10658 ANSI standards. When used at link time, it may include libraries
10659 or startup files that change the default FPU control word or other
10660 similar optimizations.
10662 This option is not turned on by any @option{-O} option since
10663 it can result in incorrect output for programs that depend on
10664 an exact implementation of IEEE or ISO rules/specifications for
10665 math functions. It may, however, yield faster code for programs
10666 that do not require the guarantees of these specifications.
10667 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
10668 @option{-fassociative-math} and @option{-freciprocal-math}.
10670 The default is @option{-fno-unsafe-math-optimizations}.
10672 @item -fassociative-math
10673 @opindex fassociative-math
10675 Allow re-association of operands in series of floating-point operations.
10676 This violates the ISO C and C++ language standard by possibly changing
10677 computation result. NOTE: re-ordering may change the sign of zero as
10678 well as ignore NaNs and inhibit or create underflow or overflow (and
10679 thus cannot be used on code that relies on rounding behavior like
10680 @code{(x + 2**52) - 2**52}. May also reorder floating-point comparisons
10681 and thus may not be used when ordered comparisons are required.
10682 This option requires that both @option{-fno-signed-zeros} and
10683 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
10684 much sense with @option{-frounding-math}. For Fortran the option
10685 is automatically enabled when both @option{-fno-signed-zeros} and
10686 @option{-fno-trapping-math} are in effect.
10688 The default is @option{-fno-associative-math}.
10690 @item -freciprocal-math
10691 @opindex freciprocal-math
10693 Allow the reciprocal of a value to be used instead of dividing by
10694 the value if this enables optimizations. For example @code{x / y}
10695 can be replaced with @code{x * (1/y)}, which is useful if @code{(1/y)}
10696 is subject to common subexpression elimination. Note that this loses
10697 precision and increases the number of flops operating on the value.
10699 The default is @option{-fno-reciprocal-math}.
10701 @item -ffinite-math-only
10702 @opindex ffinite-math-only
10703 Allow optimizations for floating-point arithmetic that assume
10704 that arguments and results are not NaNs or +-Infs.
10706 This option is not turned on by any @option{-O} option since
10707 it can result in incorrect output for programs that depend on
10708 an exact implementation of IEEE or ISO rules/specifications for
10709 math functions. It may, however, yield faster code for programs
10710 that do not require the guarantees of these specifications.
10712 The default is @option{-fno-finite-math-only}.
10714 @item -fno-signed-zeros
10715 @opindex fno-signed-zeros
10716 @opindex fsigned-zeros
10717 Allow optimizations for floating-point arithmetic that ignore the
10718 signedness of zero. IEEE arithmetic specifies the behavior of
10719 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
10720 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
10721 This option implies that the sign of a zero result isn't significant.
10723 The default is @option{-fsigned-zeros}.
10725 @item -fno-trapping-math
10726 @opindex fno-trapping-math
10727 @opindex ftrapping-math
10728 Compile code assuming that floating-point operations cannot generate
10729 user-visible traps. These traps include division by zero, overflow,
10730 underflow, inexact result and invalid operation. This option requires
10731 that @option{-fno-signaling-nans} be in effect. Setting this option may
10732 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
10734 This option should never be turned on by any @option{-O} option since
10735 it can result in incorrect output for programs that depend on
10736 an exact implementation of IEEE or ISO rules/specifications for
10739 The default is @option{-ftrapping-math}.
10741 @item -frounding-math
10742 @opindex frounding-math
10743 Disable transformations and optimizations that assume default floating-point
10744 rounding behavior. This is round-to-zero for all floating point
10745 to integer conversions, and round-to-nearest for all other arithmetic
10746 truncations. This option should be specified for programs that change
10747 the FP rounding mode dynamically, or that may be executed with a
10748 non-default rounding mode. This option disables constant folding of
10749 floating-point expressions at compile time (which may be affected by
10750 rounding mode) and arithmetic transformations that are unsafe in the
10751 presence of sign-dependent rounding modes.
10753 The default is @option{-fno-rounding-math}.
10755 This option is experimental and does not currently guarantee to
10756 disable all GCC optimizations that are affected by rounding mode.
10757 Future versions of GCC may provide finer control of this setting
10758 using C99's @code{FENV_ACCESS} pragma. This command-line option
10759 will be used to specify the default state for @code{FENV_ACCESS}.
10761 @item -fsignaling-nans
10762 @opindex fsignaling-nans
10763 Compile code assuming that IEEE signaling NaNs may generate user-visible
10764 traps during floating-point operations. Setting this option disables
10765 optimizations that may change the number of exceptions visible with
10766 signaling NaNs. This option implies @option{-ftrapping-math}.
10768 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
10771 The default is @option{-fno-signaling-nans}.
10773 This option is experimental and does not currently guarantee to
10774 disable all GCC optimizations that affect signaling NaN behavior.
10776 @item -fno-fp-int-builtin-inexact
10777 @opindex fno-fp-int-builtin-inexact
10778 @opindex ffp-int-builtin-inexact
10779 Do not allow the built-in functions @code{ceil}, @code{floor},
10780 @code{round} and @code{trunc}, and their @code{float} and @code{long
10781 double} variants, to generate code that raises the ``inexact''
10782 floating-point exception for noninteger arguments. ISO C99 and C11
10783 allow these functions to raise the ``inexact'' exception, but ISO/IEC
10784 TS 18661-1:2014, the C bindings to IEEE 754-2008, does not allow these
10785 functions to do so.
10787 The default is @option{-ffp-int-builtin-inexact}, allowing the
10788 exception to be raised. This option does nothing unless
10789 @option{-ftrapping-math} is in effect.
10791 Even if @option{-fno-fp-int-builtin-inexact} is used, if the functions
10792 generate a call to a library function then the ``inexact'' exception
10793 may be raised if the library implementation does not follow TS 18661.
10795 @item -fsingle-precision-constant
10796 @opindex fsingle-precision-constant
10797 Treat floating-point constants as single precision instead of
10798 implicitly converting them to double-precision constants.
10800 @item -fcx-limited-range
10801 @opindex fcx-limited-range
10802 When enabled, this option states that a range reduction step is not
10803 needed when performing complex division. Also, there is no checking
10804 whether the result of a complex multiplication or division is @code{NaN
10805 + I*NaN}, with an attempt to rescue the situation in that case. The
10806 default is @option{-fno-cx-limited-range}, but is enabled by
10807 @option{-ffast-math}.
10809 This option controls the default setting of the ISO C99
10810 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
10813 @item -fcx-fortran-rules
10814 @opindex fcx-fortran-rules
10815 Complex multiplication and division follow Fortran rules. Range
10816 reduction is done as part of complex division, but there is no checking
10817 whether the result of a complex multiplication or division is @code{NaN
10818 + I*NaN}, with an attempt to rescue the situation in that case.
10820 The default is @option{-fno-cx-fortran-rules}.
10824 The following options control optimizations that may improve
10825 performance, but are not enabled by any @option{-O} options. This
10826 section includes experimental options that may produce broken code.
10829 @item -fbranch-probabilities
10830 @opindex fbranch-probabilities
10831 After running a program compiled with @option{-fprofile-arcs}
10832 (@pxref{Instrumentation Options}),
10833 you can compile it a second time using
10834 @option{-fbranch-probabilities}, to improve optimizations based on
10835 the number of times each branch was taken. When a program
10836 compiled with @option{-fprofile-arcs} exits, it saves arc execution
10837 counts to a file called @file{@var{sourcename}.gcda} for each source
10838 file. The information in this data file is very dependent on the
10839 structure of the generated code, so you must use the same source code
10840 and the same optimization options for both compilations.
10842 With @option{-fbranch-probabilities}, GCC puts a
10843 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
10844 These can be used to improve optimization. Currently, they are only
10845 used in one place: in @file{reorg.c}, instead of guessing which path a
10846 branch is most likely to take, the @samp{REG_BR_PROB} values are used to
10847 exactly determine which path is taken more often.
10849 Enabled by @option{-fprofile-use} and @option{-fauto-profile}.
10851 @item -fprofile-values
10852 @opindex fprofile-values
10853 If combined with @option{-fprofile-arcs}, it adds code so that some
10854 data about values of expressions in the program is gathered.
10856 With @option{-fbranch-probabilities}, it reads back the data gathered
10857 from profiling values of expressions for usage in optimizations.
10859 Enabled by @option{-fprofile-generate}, @option{-fprofile-use}, and
10860 @option{-fauto-profile}.
10862 @item -fprofile-reorder-functions
10863 @opindex fprofile-reorder-functions
10864 Function reordering based on profile instrumentation collects
10865 first time of execution of a function and orders these functions
10866 in ascending order.
10868 Enabled with @option{-fprofile-use}.
10872 If combined with @option{-fprofile-arcs}, this option instructs the compiler
10873 to add code to gather information about values of expressions.
10875 With @option{-fbranch-probabilities}, it reads back the data gathered
10876 and actually performs the optimizations based on them.
10877 Currently the optimizations include specialization of division operations
10878 using the knowledge about the value of the denominator.
10880 Enabled with @option{-fprofile-use} and @option{-fauto-profile}.
10882 @item -frename-registers
10883 @opindex frename-registers
10884 Attempt to avoid false dependencies in scheduled code by making use
10885 of registers left over after register allocation. This optimization
10886 most benefits processors with lots of registers. Depending on the
10887 debug information format adopted by the target, however, it can
10888 make debugging impossible, since variables no longer stay in
10889 a ``home register''.
10891 Enabled by default with @option{-funroll-loops}.
10893 @item -fschedule-fusion
10894 @opindex fschedule-fusion
10895 Performs a target dependent pass over the instruction stream to schedule
10896 instructions of same type together because target machine can execute them
10897 more efficiently if they are adjacent to each other in the instruction flow.
10899 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
10903 Perform tail duplication to enlarge superblock size. This transformation
10904 simplifies the control flow of the function allowing other optimizations to do
10907 Enabled by @option{-fprofile-use} and @option{-fauto-profile}.
10909 @item -funroll-loops
10910 @opindex funroll-loops
10911 Unroll loops whose number of iterations can be determined at compile time or
10912 upon entry to the loop. @option{-funroll-loops} implies
10913 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
10914 It also turns on complete loop peeling (i.e.@: complete removal of loops with
10915 a small constant number of iterations). This option makes code larger, and may
10916 or may not make it run faster.
10918 Enabled by @option{-fprofile-use} and @option{-fauto-profile}.
10920 @item -funroll-all-loops
10921 @opindex funroll-all-loops
10922 Unroll all loops, even if their number of iterations is uncertain when
10923 the loop is entered. This usually makes programs run more slowly.
10924 @option{-funroll-all-loops} implies the same options as
10925 @option{-funroll-loops}.
10928 @opindex fpeel-loops
10929 Peels loops for which there is enough information that they do not
10930 roll much (from profile feedback or static analysis). It also turns on
10931 complete loop peeling (i.e.@: complete removal of loops with small constant
10932 number of iterations).
10934 Enabled by @option{-O3}, @option{-fprofile-use}, and @option{-fauto-profile}.
10936 @item -fmove-loop-invariants
10937 @opindex fmove-loop-invariants
10938 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
10939 at level @option{-O1} and higher, except for @option{-Og}.
10941 @item -fsplit-loops
10942 @opindex fsplit-loops
10943 Split a loop into two if it contains a condition that's always true
10944 for one side of the iteration space and false for the other.
10946 Enabled by @option{-fprofile-use} and @option{-fauto-profile}.
10948 @item -funswitch-loops
10949 @opindex funswitch-loops
10950 Move branches with loop invariant conditions out of the loop, with duplicates
10951 of the loop on both branches (modified according to result of the condition).
10953 Enabled by @option{-fprofile-use} and @option{-fauto-profile}.
10955 @item -fversion-loops-for-strides
10956 @opindex fversion-loops-for-strides
10957 If a loop iterates over an array with a variable stride, create another
10958 version of the loop that assumes the stride is always one. For example:
10961 for (int i = 0; i < n; ++i)
10962 x[i * stride] = @dots{};
10969 for (int i = 0; i < n; ++i)
10972 for (int i = 0; i < n; ++i)
10973 x[i * stride] = @dots{};
10976 This is particularly useful for assumed-shape arrays in Fortran where
10977 (for example) it allows better vectorization assuming contiguous accesses.
10978 This flag is enabled by default at @option{-O3}.
10979 It is also enabled by @option{-fprofile-use} and @option{-fauto-profile}.
10981 @item -ffunction-sections
10982 @itemx -fdata-sections
10983 @opindex ffunction-sections
10984 @opindex fdata-sections
10985 Place each function or data item into its own section in the output
10986 file if the target supports arbitrary sections. The name of the
10987 function or the name of the data item determines the section's name
10988 in the output file.
10990 Use these options on systems where the linker can perform optimizations to
10991 improve locality of reference in the instruction space. Most systems using the
10992 ELF object format have linkers with such optimizations. On AIX, the linker
10993 rearranges sections (CSECTs) based on the call graph. The performance impact
10996 Together with a linker garbage collection (linker @option{--gc-sections}
10997 option) these options may lead to smaller statically-linked executables (after
11000 On ELF/DWARF systems these options do not degenerate the quality of the debug
11001 information. There could be issues with other object files/debug info formats.
11003 Only use these options when there are significant benefits from doing so. When
11004 you specify these options, the assembler and linker create larger object and
11005 executable files and are also slower. These options affect code generation.
11006 They prevent optimizations by the compiler and assembler using relative
11007 locations inside a translation unit since the locations are unknown until
11008 link time. An example of such an optimization is relaxing calls to short call
11011 @item -fbranch-target-load-optimize
11012 @opindex fbranch-target-load-optimize
11013 Perform branch target register load optimization before prologue / epilogue
11015 The use of target registers can typically be exposed only during reload,
11016 thus hoisting loads out of loops and doing inter-block scheduling needs
11017 a separate optimization pass.
11019 @item -fbranch-target-load-optimize2
11020 @opindex fbranch-target-load-optimize2
11021 Perform branch target register load optimization after prologue / epilogue
11024 @item -fbtr-bb-exclusive
11025 @opindex fbtr-bb-exclusive
11026 When performing branch target register load optimization, don't reuse
11027 branch target registers within any basic block.
11030 @opindex fstdarg-opt
11031 Optimize the prologue of variadic argument functions with respect to usage of
11034 @item -fsection-anchors
11035 @opindex fsection-anchors
11036 Try to reduce the number of symbolic address calculations by using
11037 shared ``anchor'' symbols to address nearby objects. This transformation
11038 can help to reduce the number of GOT entries and GOT accesses on some
11041 For example, the implementation of the following function @code{foo}:
11044 static int a, b, c;
11045 int foo (void) @{ return a + b + c; @}
11049 usually calculates the addresses of all three variables, but if you
11050 compile it with @option{-fsection-anchors}, it accesses the variables
11051 from a common anchor point instead. The effect is similar to the
11052 following pseudocode (which isn't valid C):
11057 register int *xr = &x;
11058 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
11062 Not all targets support this option.
11064 @item --param @var{name}=@var{value}
11066 In some places, GCC uses various constants to control the amount of
11067 optimization that is done. For example, GCC does not inline functions
11068 that contain more than a certain number of instructions. You can
11069 control some of these constants on the command line using the
11070 @option{--param} option.
11072 The names of specific parameters, and the meaning of the values, are
11073 tied to the internals of the compiler, and are subject to change
11074 without notice in future releases.
11076 In order to get minimal, maximal and default value of a parameter,
11077 one can use @option{--help=param -Q} options.
11079 In each case, the @var{value} is an integer. The allowable choices for
11083 @item predictable-branch-outcome
11084 When branch is predicted to be taken with probability lower than this threshold
11085 (in percent), then it is considered well predictable.
11087 @item max-rtl-if-conversion-insns
11088 RTL if-conversion tries to remove conditional branches around a block and
11089 replace them with conditionally executed instructions. This parameter
11090 gives the maximum number of instructions in a block which should be
11091 considered for if-conversion. The compiler will
11092 also use other heuristics to decide whether if-conversion is likely to be
11095 @item max-rtl-if-conversion-predictable-cost
11096 @itemx max-rtl-if-conversion-unpredictable-cost
11097 RTL if-conversion will try to remove conditional branches around a block
11098 and replace them with conditionally executed instructions. These parameters
11099 give the maximum permissible cost for the sequence that would be generated
11100 by if-conversion depending on whether the branch is statically determined
11101 to be predictable or not. The units for this parameter are the same as
11102 those for the GCC internal seq_cost metric. The compiler will try to
11103 provide a reasonable default for this parameter using the BRANCH_COST
11106 @item max-crossjump-edges
11107 The maximum number of incoming edges to consider for cross-jumping.
11108 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
11109 the number of edges incoming to each block. Increasing values mean
11110 more aggressive optimization, making the compilation time increase with
11111 probably small improvement in executable size.
11113 @item min-crossjump-insns
11114 The minimum number of instructions that must be matched at the end
11115 of two blocks before cross-jumping is performed on them. This
11116 value is ignored in the case where all instructions in the block being
11117 cross-jumped from are matched.
11119 @item max-grow-copy-bb-insns
11120 The maximum code size expansion factor when copying basic blocks
11121 instead of jumping. The expansion is relative to a jump instruction.
11123 @item max-goto-duplication-insns
11124 The maximum number of instructions to duplicate to a block that jumps
11125 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
11126 passes, GCC factors computed gotos early in the compilation process,
11127 and unfactors them as late as possible. Only computed jumps at the
11128 end of a basic blocks with no more than max-goto-duplication-insns are
11131 @item max-delay-slot-insn-search
11132 The maximum number of instructions to consider when looking for an
11133 instruction to fill a delay slot. If more than this arbitrary number of
11134 instructions are searched, the time savings from filling the delay slot
11135 are minimal, so stop searching. Increasing values mean more
11136 aggressive optimization, making the compilation time increase with probably
11137 small improvement in execution time.
11139 @item max-delay-slot-live-search
11140 When trying to fill delay slots, the maximum number of instructions to
11141 consider when searching for a block with valid live register
11142 information. Increasing this arbitrarily chosen value means more
11143 aggressive optimization, increasing the compilation time. This parameter
11144 should be removed when the delay slot code is rewritten to maintain the
11145 control-flow graph.
11147 @item max-gcse-memory
11148 The approximate maximum amount of memory that can be allocated in
11149 order to perform the global common subexpression elimination
11150 optimization. If more memory than specified is required, the
11151 optimization is not done.
11153 @item max-gcse-insertion-ratio
11154 If the ratio of expression insertions to deletions is larger than this value
11155 for any expression, then RTL PRE inserts or removes the expression and thus
11156 leaves partially redundant computations in the instruction stream.
11158 @item max-pending-list-length
11159 The maximum number of pending dependencies scheduling allows
11160 before flushing the current state and starting over. Large functions
11161 with few branches or calls can create excessively large lists which
11162 needlessly consume memory and resources.
11164 @item max-modulo-backtrack-attempts
11165 The maximum number of backtrack attempts the scheduler should make
11166 when modulo scheduling a loop. Larger values can exponentially increase
11169 @item max-inline-insns-single
11170 Several parameters control the tree inliner used in GCC@.
11171 This number sets the maximum number of instructions (counted in GCC's
11172 internal representation) in a single function that the tree inliner
11173 considers for inlining. This only affects functions declared
11174 inline and methods implemented in a class declaration (C++).
11176 @item max-inline-insns-auto
11177 When you use @option{-finline-functions} (included in @option{-O3}),
11178 a lot of functions that would otherwise not be considered for inlining
11179 by the compiler are investigated. To those functions, a different
11180 (more restrictive) limit compared to functions declared inline can
11183 @item max-inline-insns-small
11184 This is bound applied to calls which are considered relevant with
11185 @option{-finline-small-functions}.
11187 @item max-inline-insns-size
11188 This is bound applied to calls which are optimized for size. Small growth
11189 may be desirable to anticipate optimization oppurtunities exposed by inlining.
11191 @item uninlined-function-insns
11192 Number of instructions accounted by inliner for function overhead such as
11193 function prologue and epilogue.
11195 @item uninlined-function-time
11196 Extra time accounted by inliner for function overhead such as time needed to
11197 execute function prologue and epilogue
11199 @item uninlined-thunk-insns
11200 @item uninlined-thunk-time
11201 Same as @option{--param uninlined-function-insns} and
11202 @option{--param uninlined-function-time} but applied to function thunks
11204 @item inline-min-speedup
11205 When estimated performance improvement of caller + callee runtime exceeds this
11206 threshold (in percent), the function can be inlined regardless of the limit on
11207 @option{--param max-inline-insns-single} and @option{--param
11208 max-inline-insns-auto}.
11210 @item large-function-insns
11211 The limit specifying really large functions. For functions larger than this
11212 limit after inlining, inlining is constrained by
11213 @option{--param large-function-growth}. This parameter is useful primarily
11214 to avoid extreme compilation time caused by non-linear algorithms used by the
11217 @item large-function-growth
11218 Specifies maximal growth of large function caused by inlining in percents.
11219 For example, parameter value 100 limits large function growth to 2.0 times
11222 @item large-unit-insns
11223 The limit specifying large translation unit. Growth caused by inlining of
11224 units larger than this limit is limited by @option{--param inline-unit-growth}.
11225 For small units this might be too tight.
11226 For example, consider a unit consisting of function A
11227 that is inline and B that just calls A three times. If B is small relative to
11228 A, the growth of unit is 300\% and yet such inlining is very sane. For very
11229 large units consisting of small inlineable functions, however, the overall unit
11230 growth limit is needed to avoid exponential explosion of code size. Thus for
11231 smaller units, the size is increased to @option{--param large-unit-insns}
11232 before applying @option{--param inline-unit-growth}.
11234 @item inline-unit-growth
11235 Specifies maximal overall growth of the compilation unit caused by inlining.
11236 For example, parameter value 20 limits unit growth to 1.2 times the original
11237 size. Cold functions (either marked cold via an attribute or by profile
11238 feedback) are not accounted into the unit size.
11240 @item ipcp-unit-growth
11241 Specifies maximal overall growth of the compilation unit caused by
11242 interprocedural constant propagation. For example, parameter value 10 limits
11243 unit growth to 1.1 times the original size.
11245 @item large-stack-frame
11246 The limit specifying large stack frames. While inlining the algorithm is trying
11247 to not grow past this limit too much.
11249 @item large-stack-frame-growth
11250 Specifies maximal growth of large stack frames caused by inlining in percents.
11251 For example, parameter value 1000 limits large stack frame growth to 11 times
11254 @item max-inline-insns-recursive
11255 @itemx max-inline-insns-recursive-auto
11256 Specifies the maximum number of instructions an out-of-line copy of a
11257 self-recursive inline
11258 function can grow into by performing recursive inlining.
11260 @option{--param max-inline-insns-recursive} applies to functions
11262 For functions not declared inline, recursive inlining
11263 happens only when @option{-finline-functions} (included in @option{-O3}) is
11264 enabled; @option{--param max-inline-insns-recursive-auto} applies instead.
11266 @item max-inline-recursive-depth
11267 @itemx max-inline-recursive-depth-auto
11268 Specifies the maximum recursion depth used for recursive inlining.
11270 @option{--param max-inline-recursive-depth} applies to functions
11271 declared inline. For functions not declared inline, recursive inlining
11272 happens only when @option{-finline-functions} (included in @option{-O3}) is
11273 enabled; @option{--param max-inline-recursive-depth-auto} applies instead.
11275 @item min-inline-recursive-probability
11276 Recursive inlining is profitable only for function having deep recursion
11277 in average and can hurt for function having little recursion depth by
11278 increasing the prologue size or complexity of function body to other
11281 When profile feedback is available (see @option{-fprofile-generate}) the actual
11282 recursion depth can be guessed from the probability that function recurses
11283 via a given call expression. This parameter limits inlining only to call
11284 expressions whose probability exceeds the given threshold (in percents).
11286 @item early-inlining-insns
11287 Specify growth that the early inliner can make. In effect it increases
11288 the amount of inlining for code having a large abstraction penalty.
11290 @item max-early-inliner-iterations
11291 Limit of iterations of the early inliner. This basically bounds
11292 the number of nested indirect calls the early inliner can resolve.
11293 Deeper chains are still handled by late inlining.
11295 @item comdat-sharing-probability
11296 Probability (in percent) that C++ inline function with comdat visibility
11297 are shared across multiple compilation units.
11299 @item profile-func-internal-id
11300 A parameter to control whether to use function internal id in profile
11301 database lookup. If the value is 0, the compiler uses an id that
11302 is based on function assembler name and filename, which makes old profile
11303 data more tolerant to source changes such as function reordering etc.
11305 @item min-vect-loop-bound
11306 The minimum number of iterations under which loops are not vectorized
11307 when @option{-ftree-vectorize} is used. The number of iterations after
11308 vectorization needs to be greater than the value specified by this option
11309 to allow vectorization.
11311 @item gcse-cost-distance-ratio
11312 Scaling factor in calculation of maximum distance an expression
11313 can be moved by GCSE optimizations. This is currently supported only in the
11314 code hoisting pass. The bigger the ratio, the more aggressive code hoisting
11315 is with simple expressions, i.e., the expressions that have cost
11316 less than @option{gcse-unrestricted-cost}. Specifying 0 disables
11317 hoisting of simple expressions.
11319 @item gcse-unrestricted-cost
11320 Cost, roughly measured as the cost of a single typical machine
11321 instruction, at which GCSE optimizations do not constrain
11322 the distance an expression can travel. This is currently
11323 supported only in the code hoisting pass. The lesser the cost,
11324 the more aggressive code hoisting is. Specifying 0
11325 allows all expressions to travel unrestricted distances.
11327 @item max-hoist-depth
11328 The depth of search in the dominator tree for expressions to hoist.
11329 This is used to avoid quadratic behavior in hoisting algorithm.
11330 The value of 0 does not limit on the search, but may slow down compilation
11333 @item max-tail-merge-comparisons
11334 The maximum amount of similar bbs to compare a bb with. This is used to
11335 avoid quadratic behavior in tree tail merging.
11337 @item max-tail-merge-iterations
11338 The maximum amount of iterations of the pass over the function. This is used to
11339 limit compilation time in tree tail merging.
11341 @item store-merging-allow-unaligned
11342 Allow the store merging pass to introduce unaligned stores if it is legal to
11345 @item max-stores-to-merge
11346 The maximum number of stores to attempt to merge into wider stores in the store
11349 @item max-unrolled-insns
11350 The maximum number of instructions that a loop may have to be unrolled.
11351 If a loop is unrolled, this parameter also determines how many times
11352 the loop code is unrolled.
11354 @item max-average-unrolled-insns
11355 The maximum number of instructions biased by probabilities of their execution
11356 that a loop may have to be unrolled. If a loop is unrolled,
11357 this parameter also determines how many times the loop code is unrolled.
11359 @item max-unroll-times
11360 The maximum number of unrollings of a single loop.
11362 @item max-peeled-insns
11363 The maximum number of instructions that a loop may have to be peeled.
11364 If a loop is peeled, this parameter also determines how many times
11365 the loop code is peeled.
11367 @item max-peel-times
11368 The maximum number of peelings of a single loop.
11370 @item max-peel-branches
11371 The maximum number of branches on the hot path through the peeled sequence.
11373 @item max-completely-peeled-insns
11374 The maximum number of insns of a completely peeled loop.
11376 @item max-completely-peel-times
11377 The maximum number of iterations of a loop to be suitable for complete peeling.
11379 @item max-completely-peel-loop-nest-depth
11380 The maximum depth of a loop nest suitable for complete peeling.
11382 @item max-unswitch-insns
11383 The maximum number of insns of an unswitched loop.
11385 @item max-unswitch-level
11386 The maximum number of branches unswitched in a single loop.
11388 @item lim-expensive
11389 The minimum cost of an expensive expression in the loop invariant motion.
11391 @item iv-consider-all-candidates-bound
11392 Bound on number of candidates for induction variables, below which
11393 all candidates are considered for each use in induction variable
11394 optimizations. If there are more candidates than this,
11395 only the most relevant ones are considered to avoid quadratic time complexity.
11397 @item iv-max-considered-uses
11398 The induction variable optimizations give up on loops that contain more
11399 induction variable uses.
11401 @item iv-always-prune-cand-set-bound
11402 If the number of candidates in the set is smaller than this value,
11403 always try to remove unnecessary ivs from the set
11404 when adding a new one.
11406 @item avg-loop-niter
11407 Average number of iterations of a loop.
11409 @item dse-max-object-size
11410 Maximum size (in bytes) of objects tracked bytewise by dead store elimination.
11411 Larger values may result in larger compilation times.
11413 @item dse-max-alias-queries-per-store
11414 Maximum number of queries into the alias oracle per store.
11415 Larger values result in larger compilation times and may result in more
11416 removed dead stores.
11418 @item scev-max-expr-size
11419 Bound on size of expressions used in the scalar evolutions analyzer.
11420 Large expressions slow the analyzer.
11422 @item scev-max-expr-complexity
11423 Bound on the complexity of the expressions in the scalar evolutions analyzer.
11424 Complex expressions slow the analyzer.
11426 @item max-tree-if-conversion-phi-args
11427 Maximum number of arguments in a PHI supported by TREE if conversion
11428 unless the loop is marked with simd pragma.
11430 @item vect-max-version-for-alignment-checks
11431 The maximum number of run-time checks that can be performed when
11432 doing loop versioning for alignment in the vectorizer.
11434 @item vect-max-version-for-alias-checks
11435 The maximum number of run-time checks that can be performed when
11436 doing loop versioning for alias in the vectorizer.
11438 @item vect-max-peeling-for-alignment
11439 The maximum number of loop peels to enhance access alignment
11440 for vectorizer. Value -1 means no limit.
11442 @item max-iterations-to-track
11443 The maximum number of iterations of a loop the brute-force algorithm
11444 for analysis of the number of iterations of the loop tries to evaluate.
11446 @item hot-bb-count-ws-permille
11447 A basic block profile count is considered hot if it contributes to
11448 the given permillage (i.e.@: 0...1000) of the entire profiled execution.
11450 @item hot-bb-frequency-fraction
11451 Select fraction of the entry block frequency of executions of basic block in
11452 function given basic block needs to have to be considered hot.
11454 @item max-predicted-iterations
11455 The maximum number of loop iterations we predict statically. This is useful
11456 in cases where a function contains a single loop with known bound and
11457 another loop with unknown bound.
11458 The known number of iterations is predicted correctly, while
11459 the unknown number of iterations average to roughly 10. This means that the
11460 loop without bounds appears artificially cold relative to the other one.
11462 @item builtin-expect-probability
11463 Control the probability of the expression having the specified value. This
11464 parameter takes a percentage (i.e.@: 0 ... 100) as input.
11466 @item builtin-string-cmp-inline-length
11467 The maximum length of a constant string for a builtin string cmp call
11468 eligible for inlining.
11470 @item align-threshold
11472 Select fraction of the maximal frequency of executions of a basic block in
11473 a function to align the basic block.
11475 @item align-loop-iterations
11477 A loop expected to iterate at least the selected number of iterations is
11480 @item tracer-dynamic-coverage
11481 @itemx tracer-dynamic-coverage-feedback
11483 This value is used to limit superblock formation once the given percentage of
11484 executed instructions is covered. This limits unnecessary code size
11487 The @option{tracer-dynamic-coverage-feedback} parameter
11488 is used only when profile
11489 feedback is available. The real profiles (as opposed to statically estimated
11490 ones) are much less balanced allowing the threshold to be larger value.
11492 @item tracer-max-code-growth
11493 Stop tail duplication once code growth has reached given percentage. This is
11494 a rather artificial limit, as most of the duplicates are eliminated later in
11495 cross jumping, so it may be set to much higher values than is the desired code
11498 @item tracer-min-branch-ratio
11500 Stop reverse growth when the reverse probability of best edge is less than this
11501 threshold (in percent).
11503 @item tracer-min-branch-probability
11504 @itemx tracer-min-branch-probability-feedback
11506 Stop forward growth if the best edge has probability lower than this
11509 Similarly to @option{tracer-dynamic-coverage} two parameters are
11510 provided. @option{tracer-min-branch-probability-feedback} is used for
11511 compilation with profile feedback and @option{tracer-min-branch-probability}
11512 compilation without. The value for compilation with profile feedback
11513 needs to be more conservative (higher) in order to make tracer
11516 @item stack-clash-protection-guard-size
11517 Specify the size of the operating system provided stack guard as
11518 2 raised to @var{num} bytes. Higher values may reduce the
11519 number of explicit probes, but a value larger than the operating system
11520 provided guard will leave code vulnerable to stack clash style attacks.
11522 @item stack-clash-protection-probe-interval
11523 Stack clash protection involves probing stack space as it is allocated. This
11524 param controls the maximum distance between probes into the stack as 2 raised
11525 to @var{num} bytes. Higher values may reduce the number of explicit probes, but a value
11526 larger than the operating system provided guard will leave code vulnerable to
11527 stack clash style attacks.
11529 @item max-cse-path-length
11531 The maximum number of basic blocks on path that CSE considers.
11533 @item max-cse-insns
11534 The maximum number of instructions CSE processes before flushing.
11536 @item ggc-min-expand
11538 GCC uses a garbage collector to manage its own memory allocation. This
11539 parameter specifies the minimum percentage by which the garbage
11540 collector's heap should be allowed to expand between collections.
11541 Tuning this may improve compilation speed; it has no effect on code
11544 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
11545 RAM >= 1GB@. If @code{getrlimit} is available, the notion of ``RAM'' is
11546 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
11547 GCC is not able to calculate RAM on a particular platform, the lower
11548 bound of 30% is used. Setting this parameter and
11549 @option{ggc-min-heapsize} to zero causes a full collection to occur at
11550 every opportunity. This is extremely slow, but can be useful for
11553 @item ggc-min-heapsize
11555 Minimum size of the garbage collector's heap before it begins bothering
11556 to collect garbage. The first collection occurs after the heap expands
11557 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
11558 tuning this may improve compilation speed, and has no effect on code
11561 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit that
11562 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
11563 with a lower bound of 4096 (four megabytes) and an upper bound of
11564 131072 (128 megabytes). If GCC is not able to calculate RAM on a
11565 particular platform, the lower bound is used. Setting this parameter
11566 very large effectively disables garbage collection. Setting this
11567 parameter and @option{ggc-min-expand} to zero causes a full collection
11568 to occur at every opportunity.
11570 @item max-reload-search-insns
11571 The maximum number of instruction reload should look backward for equivalent
11572 register. Increasing values mean more aggressive optimization, making the
11573 compilation time increase with probably slightly better performance.
11575 @item max-cselib-memory-locations
11576 The maximum number of memory locations cselib should take into account.
11577 Increasing values mean more aggressive optimization, making the compilation time
11578 increase with probably slightly better performance.
11580 @item max-sched-ready-insns
11581 The maximum number of instructions ready to be issued the scheduler should
11582 consider at any given time during the first scheduling pass. Increasing
11583 values mean more thorough searches, making the compilation time increase
11584 with probably little benefit.
11586 @item max-sched-region-blocks
11587 The maximum number of blocks in a region to be considered for
11588 interblock scheduling.
11590 @item max-pipeline-region-blocks
11591 The maximum number of blocks in a region to be considered for
11592 pipelining in the selective scheduler.
11594 @item max-sched-region-insns
11595 The maximum number of insns in a region to be considered for
11596 interblock scheduling.
11598 @item max-pipeline-region-insns
11599 The maximum number of insns in a region to be considered for
11600 pipelining in the selective scheduler.
11602 @item min-spec-prob
11603 The minimum probability (in percents) of reaching a source block
11604 for interblock speculative scheduling.
11606 @item max-sched-extend-regions-iters
11607 The maximum number of iterations through CFG to extend regions.
11608 A value of 0 disables region extensions.
11610 @item max-sched-insn-conflict-delay
11611 The maximum conflict delay for an insn to be considered for speculative motion.
11613 @item sched-spec-prob-cutoff
11614 The minimal probability of speculation success (in percents), so that
11615 speculative insns are scheduled.
11617 @item sched-state-edge-prob-cutoff
11618 The minimum probability an edge must have for the scheduler to save its
11621 @item sched-mem-true-dep-cost
11622 Minimal distance (in CPU cycles) between store and load targeting same
11625 @item selsched-max-lookahead
11626 The maximum size of the lookahead window of selective scheduling. It is a
11627 depth of search for available instructions.
11629 @item selsched-max-sched-times
11630 The maximum number of times that an instruction is scheduled during
11631 selective scheduling. This is the limit on the number of iterations
11632 through which the instruction may be pipelined.
11634 @item selsched-insns-to-rename
11635 The maximum number of best instructions in the ready list that are considered
11636 for renaming in the selective scheduler.
11639 The minimum value of stage count that swing modulo scheduler
11642 @item max-last-value-rtl
11643 The maximum size measured as number of RTLs that can be recorded in an expression
11644 in combiner for a pseudo register as last known value of that register.
11646 @item max-combine-insns
11647 The maximum number of instructions the RTL combiner tries to combine.
11649 @item integer-share-limit
11650 Small integer constants can use a shared data structure, reducing the
11651 compiler's memory usage and increasing its speed. This sets the maximum
11652 value of a shared integer constant.
11654 @item ssp-buffer-size
11655 The minimum size of buffers (i.e.@: arrays) that receive stack smashing
11656 protection when @option{-fstack-protection} is used.
11658 @item min-size-for-stack-sharing
11659 The minimum size of variables taking part in stack slot sharing when not
11662 @item max-jump-thread-duplication-stmts
11663 Maximum number of statements allowed in a block that needs to be
11664 duplicated when threading jumps.
11666 @item max-fields-for-field-sensitive
11667 Maximum number of fields in a structure treated in
11668 a field sensitive manner during pointer analysis.
11670 @item prefetch-latency
11671 Estimate on average number of instructions that are executed before
11672 prefetch finishes. The distance prefetched ahead is proportional
11673 to this constant. Increasing this number may also lead to less
11674 streams being prefetched (see @option{simultaneous-prefetches}).
11676 @item simultaneous-prefetches
11677 Maximum number of prefetches that can run at the same time.
11679 @item l1-cache-line-size
11680 The size of cache line in L1 data cache, in bytes.
11682 @item l1-cache-size
11683 The size of L1 data cache, in kilobytes.
11685 @item l2-cache-size
11686 The size of L2 data cache, in kilobytes.
11688 @item prefetch-dynamic-strides
11689 Whether the loop array prefetch pass should issue software prefetch hints
11690 for strides that are non-constant. In some cases this may be
11691 beneficial, though the fact the stride is non-constant may make it
11692 hard to predict when there is clear benefit to issuing these hints.
11694 Set to 1 if the prefetch hints should be issued for non-constant
11695 strides. Set to 0 if prefetch hints should be issued only for strides that
11696 are known to be constant and below @option{prefetch-minimum-stride}.
11698 @item prefetch-minimum-stride
11699 Minimum constant stride, in bytes, to start using prefetch hints for. If
11700 the stride is less than this threshold, prefetch hints will not be issued.
11702 This setting is useful for processors that have hardware prefetchers, in
11703 which case there may be conflicts between the hardware prefetchers and
11704 the software prefetchers. If the hardware prefetchers have a maximum
11705 stride they can handle, it should be used here to improve the use of
11706 software prefetchers.
11708 A value of -1 means we don't have a threshold and therefore
11709 prefetch hints can be issued for any constant stride.
11711 This setting is only useful for strides that are known and constant.
11713 @item loop-interchange-max-num-stmts
11714 The maximum number of stmts in a loop to be interchanged.
11716 @item loop-interchange-stride-ratio
11717 The minimum ratio between stride of two loops for interchange to be profitable.
11719 @item min-insn-to-prefetch-ratio
11720 The minimum ratio between the number of instructions and the
11721 number of prefetches to enable prefetching in a loop.
11723 @item prefetch-min-insn-to-mem-ratio
11724 The minimum ratio between the number of instructions and the
11725 number of memory references to enable prefetching in a loop.
11727 @item use-canonical-types
11728 Whether the compiler should use the ``canonical'' type system.
11729 Should always be 1, which uses a more efficient internal
11730 mechanism for comparing types in C++ and Objective-C++. However, if
11731 bugs in the canonical type system are causing compilation failures,
11732 set this value to 0 to disable canonical types.
11734 @item switch-conversion-max-branch-ratio
11735 Switch initialization conversion refuses to create arrays that are
11736 bigger than @option{switch-conversion-max-branch-ratio} times the number of
11737 branches in the switch.
11739 @item max-partial-antic-length
11740 Maximum length of the partial antic set computed during the tree
11741 partial redundancy elimination optimization (@option{-ftree-pre}) when
11742 optimizing at @option{-O3} and above. For some sorts of source code
11743 the enhanced partial redundancy elimination optimization can run away,
11744 consuming all of the memory available on the host machine. This
11745 parameter sets a limit on the length of the sets that are computed,
11746 which prevents the runaway behavior. Setting a value of 0 for
11747 this parameter allows an unlimited set length.
11749 @item rpo-vn-max-loop-depth
11750 Maximum loop depth that is value-numbered optimistically.
11751 When the limit hits the innermost
11752 @var{rpo-vn-max-loop-depth} loops and the outermost loop in the
11753 loop nest are value-numbered optimistically and the remaining ones not.
11755 @item sccvn-max-alias-queries-per-access
11756 Maximum number of alias-oracle queries we perform when looking for
11757 redundancies for loads and stores. If this limit is hit the search
11758 is aborted and the load or store is not considered redundant. The
11759 number of queries is algorithmically limited to the number of
11760 stores on all paths from the load to the function entry.
11762 @item ira-max-loops-num
11763 IRA uses regional register allocation by default. If a function
11764 contains more loops than the number given by this parameter, only at most
11765 the given number of the most frequently-executed loops form regions
11766 for regional register allocation.
11768 @item ira-max-conflict-table-size
11769 Although IRA uses a sophisticated algorithm to compress the conflict
11770 table, the table can still require excessive amounts of memory for
11771 huge functions. If the conflict table for a function could be more
11772 than the size in MB given by this parameter, the register allocator
11773 instead uses a faster, simpler, and lower-quality
11774 algorithm that does not require building a pseudo-register conflict table.
11776 @item ira-loop-reserved-regs
11777 IRA can be used to evaluate more accurate register pressure in loops
11778 for decisions to move loop invariants (see @option{-O3}). The number
11779 of available registers reserved for some other purposes is given
11780 by this parameter. Default of the parameter
11781 is the best found from numerous experiments.
11783 @item lra-inheritance-ebb-probability-cutoff
11784 LRA tries to reuse values reloaded in registers in subsequent insns.
11785 This optimization is called inheritance. EBB is used as a region to
11786 do this optimization. The parameter defines a minimal fall-through
11787 edge probability in percentage used to add BB to inheritance EBB in
11788 LRA. The default value was chosen
11789 from numerous runs of SPEC2000 on x86-64.
11791 @item loop-invariant-max-bbs-in-loop
11792 Loop invariant motion can be very expensive, both in compilation time and
11793 in amount of needed compile-time memory, with very large loops. Loops
11794 with more basic blocks than this parameter won't have loop invariant
11795 motion optimization performed on them.
11797 @item loop-max-datarefs-for-datadeps
11798 Building data dependencies is expensive for very large loops. This
11799 parameter limits the number of data references in loops that are
11800 considered for data dependence analysis. These large loops are no
11801 handled by the optimizations using loop data dependencies.
11803 @item max-vartrack-size
11804 Sets a maximum number of hash table slots to use during variable
11805 tracking dataflow analysis of any function. If this limit is exceeded
11806 with variable tracking at assignments enabled, analysis for that
11807 function is retried without it, after removing all debug insns from
11808 the function. If the limit is exceeded even without debug insns, var
11809 tracking analysis is completely disabled for the function. Setting
11810 the parameter to zero makes it unlimited.
11812 @item max-vartrack-expr-depth
11813 Sets a maximum number of recursion levels when attempting to map
11814 variable names or debug temporaries to value expressions. This trades
11815 compilation time for more complete debug information. If this is set too
11816 low, value expressions that are available and could be represented in
11817 debug information may end up not being used; setting this higher may
11818 enable the compiler to find more complex debug expressions, but compile
11819 time and memory use may grow.
11821 @item max-debug-marker-count
11822 Sets a threshold on the number of debug markers (e.g.@: begin stmt
11823 markers) to avoid complexity explosion at inlining or expanding to RTL.
11824 If a function has more such gimple stmts than the set limit, such stmts
11825 will be dropped from the inlined copy of a function, and from its RTL
11828 @item min-nondebug-insn-uid
11829 Use uids starting at this parameter for nondebug insns. The range below
11830 the parameter is reserved exclusively for debug insns created by
11831 @option{-fvar-tracking-assignments}, but debug insns may get
11832 (non-overlapping) uids above it if the reserved range is exhausted.
11834 @item ipa-sra-ptr-growth-factor
11835 IPA-SRA replaces a pointer to an aggregate with one or more new
11836 parameters only when their cumulative size is less or equal to
11837 @option{ipa-sra-ptr-growth-factor} times the size of the original
11840 @item sra-max-scalarization-size-Ospeed
11841 @itemx sra-max-scalarization-size-Osize
11842 The two Scalar Reduction of Aggregates passes (SRA and IPA-SRA) aim to
11843 replace scalar parts of aggregates with uses of independent scalar
11844 variables. These parameters control the maximum size, in storage units,
11845 of aggregate which is considered for replacement when compiling for
11847 (@option{sra-max-scalarization-size-Ospeed}) or size
11848 (@option{sra-max-scalarization-size-Osize}) respectively.
11850 @item tm-max-aggregate-size
11851 When making copies of thread-local variables in a transaction, this
11852 parameter specifies the size in bytes after which variables are
11853 saved with the logging functions as opposed to save/restore code
11854 sequence pairs. This option only applies when using
11857 @item graphite-max-nb-scop-params
11858 To avoid exponential effects in the Graphite loop transforms, the
11859 number of parameters in a Static Control Part (SCoP) is bounded.
11860 A value of zero can be used to lift
11861 the bound. A variable whose value is unknown at compilation time and
11862 defined outside a SCoP is a parameter of the SCoP.
11864 @item loop-block-tile-size
11865 Loop blocking or strip mining transforms, enabled with
11866 @option{-floop-block} or @option{-floop-strip-mine}, strip mine each
11867 loop in the loop nest by a given number of iterations. The strip
11868 length can be changed using the @option{loop-block-tile-size}
11871 @item ipa-cp-value-list-size
11872 IPA-CP attempts to track all possible values and types passed to a function's
11873 parameter in order to propagate them and perform devirtualization.
11874 @option{ipa-cp-value-list-size} is the maximum number of values and types it
11875 stores per one formal parameter of a function.
11877 @item ipa-cp-eval-threshold
11878 IPA-CP calculates its own score of cloning profitability heuristics
11879 and performs those cloning opportunities with scores that exceed
11880 @option{ipa-cp-eval-threshold}.
11882 @item ipa-cp-recursion-penalty
11883 Percentage penalty the recursive functions will receive when they
11884 are evaluated for cloning.
11886 @item ipa-cp-single-call-penalty
11887 Percentage penalty functions containing a single call to another
11888 function will receive when they are evaluated for cloning.
11890 @item ipa-max-agg-items
11891 IPA-CP is also capable to propagate a number of scalar values passed
11892 in an aggregate. @option{ipa-max-agg-items} controls the maximum
11893 number of such values per one parameter.
11895 @item ipa-cp-loop-hint-bonus
11896 When IPA-CP determines that a cloning candidate would make the number
11897 of iterations of a loop known, it adds a bonus of
11898 @option{ipa-cp-loop-hint-bonus} to the profitability score of
11901 @item ipa-max-aa-steps
11902 During its analysis of function bodies, IPA-CP employs alias analysis
11903 in order to track values pointed to by function parameters. In order
11904 not spend too much time analyzing huge functions, it gives up and
11905 consider all memory clobbered after examining
11906 @option{ipa-max-aa-steps} statements modifying memory.
11908 @item lto-partitions
11909 Specify desired number of partitions produced during WHOPR compilation.
11910 The number of partitions should exceed the number of CPUs used for compilation.
11912 @item lto-min-partition
11913 Size of minimal partition for WHOPR (in estimated instructions).
11914 This prevents expenses of splitting very small programs into too many
11917 @item lto-max-partition
11918 Size of max partition for WHOPR (in estimated instructions).
11919 to provide an upper bound for individual size of partition.
11920 Meant to be used only with balanced partitioning.
11922 @item lto-max-streaming-parallelism
11923 Maximal number of parallel processes used for LTO streaming.
11925 @item cxx-max-namespaces-for-diagnostic-help
11926 The maximum number of namespaces to consult for suggestions when C++
11927 name lookup fails for an identifier.
11929 @item sink-frequency-threshold
11930 The maximum relative execution frequency (in percents) of the target block
11931 relative to a statement's original block to allow statement sinking of a
11932 statement. Larger numbers result in more aggressive statement sinking.
11933 A small positive adjustment is applied for
11934 statements with memory operands as those are even more profitable so sink.
11936 @item max-stores-to-sink
11937 The maximum number of conditional store pairs that can be sunk. Set to 0
11938 if either vectorization (@option{-ftree-vectorize}) or if-conversion
11939 (@option{-ftree-loop-if-convert}) is disabled.
11941 @item allow-store-data-races
11942 Allow optimizers to introduce new data races on stores.
11943 Set to 1 to allow, otherwise to 0.
11945 @item case-values-threshold
11946 The smallest number of different values for which it is best to use a
11947 jump-table instead of a tree of conditional branches. If the value is
11948 0, use the default for the machine.
11950 @item jump-table-max-growth-ratio-for-size
11951 The maximum code size growth ratio when expanding
11952 into a jump table (in percent). The parameter is used when
11953 optimizing for size.
11955 @item jump-table-max-growth-ratio-for-speed
11956 The maximum code size growth ratio when expanding
11957 into a jump table (in percent). The parameter is used when
11958 optimizing for speed.
11960 @item tree-reassoc-width
11961 Set the maximum number of instructions executed in parallel in
11962 reassociated tree. This parameter overrides target dependent
11963 heuristics used by default if has non zero value.
11965 @item sched-pressure-algorithm
11966 Choose between the two available implementations of
11967 @option{-fsched-pressure}. Algorithm 1 is the original implementation
11968 and is the more likely to prevent instructions from being reordered.
11969 Algorithm 2 was designed to be a compromise between the relatively
11970 conservative approach taken by algorithm 1 and the rather aggressive
11971 approach taken by the default scheduler. It relies more heavily on
11972 having a regular register file and accurate register pressure classes.
11973 See @file{haifa-sched.c} in the GCC sources for more details.
11975 The default choice depends on the target.
11977 @item max-slsr-cand-scan
11978 Set the maximum number of existing candidates that are considered when
11979 seeking a basis for a new straight-line strength reduction candidate.
11982 Enable buffer overflow detection for global objects. This kind
11983 of protection is enabled by default if you are using
11984 @option{-fsanitize=address} option.
11985 To disable global objects protection use @option{--param asan-globals=0}.
11988 Enable buffer overflow detection for stack objects. This kind of
11989 protection is enabled by default when using @option{-fsanitize=address}.
11990 To disable stack protection use @option{--param asan-stack=0} option.
11992 @item asan-instrument-reads
11993 Enable buffer overflow detection for memory reads. This kind of
11994 protection is enabled by default when using @option{-fsanitize=address}.
11995 To disable memory reads protection use
11996 @option{--param asan-instrument-reads=0}.
11998 @item asan-instrument-writes
11999 Enable buffer overflow detection for memory writes. This kind of
12000 protection is enabled by default when using @option{-fsanitize=address}.
12001 To disable memory writes protection use
12002 @option{--param asan-instrument-writes=0} option.
12004 @item asan-memintrin
12005 Enable detection for built-in functions. This kind of protection
12006 is enabled by default when using @option{-fsanitize=address}.
12007 To disable built-in functions protection use
12008 @option{--param asan-memintrin=0}.
12010 @item asan-use-after-return
12011 Enable detection of use-after-return. This kind of protection
12012 is enabled by default when using the @option{-fsanitize=address} option.
12013 To disable it use @option{--param asan-use-after-return=0}.
12015 Note: By default the check is disabled at run time. To enable it,
12016 add @code{detect_stack_use_after_return=1} to the environment variable
12017 @env{ASAN_OPTIONS}.
12019 @item asan-instrumentation-with-call-threshold
12020 If number of memory accesses in function being instrumented
12021 is greater or equal to this number, use callbacks instead of inline checks.
12022 E.g. to disable inline code use
12023 @option{--param asan-instrumentation-with-call-threshold=0}.
12025 @item use-after-scope-direct-emission-threshold
12026 If the size of a local variable in bytes is smaller or equal to this
12027 number, directly poison (or unpoison) shadow memory instead of using
12028 run-time callbacks.
12030 @item max-fsm-thread-path-insns
12031 Maximum number of instructions to copy when duplicating blocks on a
12032 finite state automaton jump thread path.
12034 @item max-fsm-thread-length
12035 Maximum number of basic blocks on a finite state automaton jump thread
12038 @item max-fsm-thread-paths
12039 Maximum number of new jump thread paths to create for a finite state
12042 @item parloops-chunk-size
12043 Chunk size of omp schedule for loops parallelized by parloops.
12045 @item parloops-schedule
12046 Schedule type of omp schedule for loops parallelized by parloops (static,
12047 dynamic, guided, auto, runtime).
12049 @item parloops-min-per-thread
12050 The minimum number of iterations per thread of an innermost parallelized
12051 loop for which the parallelized variant is preferred over the single threaded
12052 one. Note that for a parallelized loop nest the
12053 minimum number of iterations of the outermost loop per thread is two.
12055 @item max-ssa-name-query-depth
12056 Maximum depth of recursion when querying properties of SSA names in things
12057 like fold routines. One level of recursion corresponds to following a
12060 @item hsa-gen-debug-stores
12061 Enable emission of special debug stores within HSA kernels which are
12062 then read and reported by libgomp plugin. Generation of these stores
12063 is disabled by default, use @option{--param hsa-gen-debug-stores=1} to
12066 @item max-speculative-devirt-maydefs
12067 The maximum number of may-defs we analyze when looking for a must-def
12068 specifying the dynamic type of an object that invokes a virtual call
12069 we may be able to devirtualize speculatively.
12071 @item max-vrp-switch-assertions
12072 The maximum number of assertions to add along the default edge of a switch
12073 statement during VRP.
12075 @item unroll-jam-min-percent
12076 The minimum percentage of memory references that must be optimized
12077 away for the unroll-and-jam transformation to be considered profitable.
12079 @item unroll-jam-max-unroll
12080 The maximum number of times the outer loop should be unrolled by
12081 the unroll-and-jam transformation.
12083 @item max-rtl-if-conversion-unpredictable-cost
12084 Maximum permissible cost for the sequence that would be generated
12085 by the RTL if-conversion pass for a branch that is considered unpredictable.
12087 @item max-variable-expansions-in-unroller
12088 If @option{-fvariable-expansion-in-unroller} is used, the maximum number
12089 of times that an individual variable will be expanded during loop unrolling.
12091 @item tracer-min-branch-probability-feedback
12092 Stop forward growth if the probability of best edge is less than
12093 this threshold (in percent). Used when profile feedback is available.
12095 @item partial-inlining-entry-probability
12096 Maximum probability of the entry BB of split region
12097 (in percent relative to entry BB of the function)
12098 to make partial inlining happen.
12100 @item max-tracked-strlens
12101 Maximum number of strings for which strlen optimization pass will
12102 track string lengths.
12104 @item gcse-after-reload-partial-fraction
12105 The threshold ratio for performing partial redundancy
12106 elimination after reload.
12108 @item gcse-after-reload-critical-fraction
12109 The threshold ratio of critical edges execution count that
12110 permit performing redundancy elimination after reload.
12112 @item max-loop-header-insns
12113 The maximum number of insns in loop header duplicated
12114 by the copy loop headers pass.
12116 @item vect-epilogues-nomask
12117 Enable loop epilogue vectorization using smaller vector size.
12119 @item slp-max-insns-in-bb
12120 Maximum number of instructions in basic block to be
12121 considered for SLP vectorization.
12123 @item avoid-fma-max-bits
12124 Maximum number of bits for which we avoid creating FMAs.
12126 @item sms-loop-average-count-threshold
12127 A threshold on the average loop count considered by the swing modulo scheduler.
12129 @item sms-dfa-history
12130 The number of cycles the swing modulo scheduler considers when checking
12131 conflicts using DFA.
12133 @item hot-bb-count-fraction
12134 Select fraction of the maximal count of repetitions of basic block
12135 in program given basic block needs
12136 to have to be considered hot (used in non-LTO mode)
12138 @item max-inline-insns-recursive-auto
12139 The maximum number of instructions non-inline function
12140 can grow to via recursive inlining.
12142 @item graphite-allow-codegen-errors
12143 Whether codegen errors should be ICEs when @option{-fchecking}.
12145 @item sms-max-ii-factor
12146 A factor for tuning the upper bound that swing modulo scheduler
12147 uses for scheduling a loop.
12149 @item lra-max-considered-reload-pseudos
12150 The max number of reload pseudos which are considered during
12151 spilling a non-reload pseudo.
12153 @item max-pow-sqrt-depth
12154 Maximum depth of sqrt chains to use when synthesizing exponentiation
12155 by a real constant.
12157 @item max-dse-active-local-stores
12158 Maximum number of active local stores in RTL dead store elimination.
12160 @item asan-instrument-allocas
12161 Enable asan allocas/VLAs protection.
12163 @item max-iterations-computation-cost
12164 Bound on the cost of an expression to compute the number of iterations.
12166 @item max-isl-operations
12167 Maximum number of isl operations, 0 means unlimited.
12169 @item graphite-max-arrays-per-scop
12170 Maximum number of arrays per scop.
12172 @item max-vartrack-reverse-op-size
12173 Max. size of loc list for which reverse ops should be added.
12175 @item unlikely-bb-count-fraction
12176 The minimum fraction of profile runs a given basic block execution count
12177 must be not to be considered unlikely.
12179 @item tracer-dynamic-coverage-feedback
12180 The percentage of function, weighted by execution frequency,
12181 that must be covered by trace formation.
12182 Used when profile feedback is available.
12184 @item max-inline-recursive-depth-auto
12185 The maximum depth of recursive inlining for non-inline functions.
12187 @item fsm-scale-path-stmts
12188 Scale factor to apply to the number of statements in a threading path
12189 when comparing to the number of (scaled) blocks.
12191 @item fsm-maximum-phi-arguments
12192 Maximum number of arguments a PHI may have before the FSM threader
12193 will not try to thread through its block.
12195 @item uninit-control-dep-attempts
12196 Maximum number of nested calls to search for control dependencies
12197 during uninitialized variable analysis.
12199 @item max-once-peeled-insns
12200 The maximum number of insns of a peeled loop that rolls only once.
12202 @item sra-max-scalarization-size-Osize
12203 Maximum size, in storage units, of an aggregate
12204 which should be considered for scalarization when compiling for size.
12206 @item fsm-scale-path-blocks
12207 Scale factor to apply to the number of blocks in a threading path
12208 when comparing to the number of (scaled) statements.
12210 @item sched-autopref-queue-depth
12211 Hardware autoprefetcher scheduler model control flag.
12212 Number of lookahead cycles the model looks into; at '
12213 ' only enable instruction sorting heuristic.
12215 @item loop-versioning-max-inner-insns
12216 The maximum number of instructions that an inner loop can have
12217 before the loop versioning pass considers it too big to copy.
12219 @item loop-versioning-max-outer-insns
12220 The maximum number of instructions that an outer loop can have
12221 before the loop versioning pass considers it too big to copy,
12222 discounting any instructions in inner loops that directly benefit
12225 @item ssa-name-def-chain-limit
12226 The maximum number of SSA_NAME assignments to follow in determining
12227 a property of a variable such as its value. This limits the number
12228 of iterations or recursive calls GCC performs when optimizing certain
12229 statements or when determining their validity prior to issuing
12235 @node Instrumentation Options
12236 @section Program Instrumentation Options
12237 @cindex instrumentation options
12238 @cindex program instrumentation options
12239 @cindex run-time error checking options
12240 @cindex profiling options
12241 @cindex options, program instrumentation
12242 @cindex options, run-time error checking
12243 @cindex options, profiling
12245 GCC supports a number of command-line options that control adding
12246 run-time instrumentation to the code it normally generates.
12247 For example, one purpose of instrumentation is collect profiling
12248 statistics for use in finding program hot spots, code coverage
12249 analysis, or profile-guided optimizations.
12250 Another class of program instrumentation is adding run-time checking
12251 to detect programming errors like invalid pointer
12252 dereferences or out-of-bounds array accesses, as well as deliberately
12253 hostile attacks such as stack smashing or C++ vtable hijacking.
12254 There is also a general hook which can be used to implement other
12255 forms of tracing or function-level instrumentation for debug or
12256 program analysis purposes.
12259 @cindex @command{prof}
12260 @cindex @command{gprof}
12265 Generate extra code to write profile information suitable for the
12266 analysis program @command{prof} (for @option{-p}) or @command{gprof}
12267 (for @option{-pg}). You must use this option when compiling
12268 the source files you want data about, and you must also use it when
12271 You can use the function attribute @code{no_instrument_function} to
12272 suppress profiling of individual functions when compiling with these options.
12273 @xref{Common Function Attributes}.
12275 @item -fprofile-arcs
12276 @opindex fprofile-arcs
12277 Add code so that program flow @dfn{arcs} are instrumented. During
12278 execution the program records how many times each branch and call is
12279 executed and how many times it is taken or returns. On targets that support
12280 constructors with priority support, profiling properly handles constructors,
12281 destructors and C++ constructors (and destructors) of classes which are used
12282 as a type of a global variable.
12285 program exits it saves this data to a file called
12286 @file{@var{auxname}.gcda} for each source file. The data may be used for
12287 profile-directed optimizations (@option{-fbranch-probabilities}), or for
12288 test coverage analysis (@option{-ftest-coverage}). Each object file's
12289 @var{auxname} is generated from the name of the output file, if
12290 explicitly specified and it is not the final executable, otherwise it is
12291 the basename of the source file. In both cases any suffix is removed
12292 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
12293 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
12294 @xref{Cross-profiling}.
12296 @cindex @command{gcov}
12300 This option is used to compile and link code instrumented for coverage
12301 analysis. The option is a synonym for @option{-fprofile-arcs}
12302 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
12303 linking). See the documentation for those options for more details.
12308 Compile the source files with @option{-fprofile-arcs} plus optimization
12309 and code generation options. For test coverage analysis, use the
12310 additional @option{-ftest-coverage} option. You do not need to profile
12311 every source file in a program.
12314 Compile the source files additionally with @option{-fprofile-abs-path}
12315 to create absolute path names in the @file{.gcno} files. This allows
12316 @command{gcov} to find the correct sources in projects where compilations
12317 occur with different working directories.
12320 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
12321 (the latter implies the former).
12324 Run the program on a representative workload to generate the arc profile
12325 information. This may be repeated any number of times. You can run
12326 concurrent instances of your program, and provided that the file system
12327 supports locking, the data files will be correctly updated. Unless
12328 a strict ISO C dialect option is in effect, @code{fork} calls are
12329 detected and correctly handled without double counting.
12332 For profile-directed optimizations, compile the source files again with
12333 the same optimization and code generation options plus
12334 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
12335 Control Optimization}).
12338 For test coverage analysis, use @command{gcov} to produce human readable
12339 information from the @file{.gcno} and @file{.gcda} files. Refer to the
12340 @command{gcov} documentation for further information.
12344 With @option{-fprofile-arcs}, for each function of your program GCC
12345 creates a program flow graph, then finds a spanning tree for the graph.
12346 Only arcs that are not on the spanning tree have to be instrumented: the
12347 compiler adds code to count the number of times that these arcs are
12348 executed. When an arc is the only exit or only entrance to a block, the
12349 instrumentation code can be added to the block; otherwise, a new basic
12350 block must be created to hold the instrumentation code.
12353 @item -ftest-coverage
12354 @opindex ftest-coverage
12355 Produce a notes file that the @command{gcov} code-coverage utility
12356 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
12357 show program coverage. Each source file's note file is called
12358 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
12359 above for a description of @var{auxname} and instructions on how to
12360 generate test coverage data. Coverage data matches the source files
12361 more closely if you do not optimize.
12363 @item -fprofile-abs-path
12364 @opindex fprofile-abs-path
12365 Automatically convert relative source file names to absolute path names
12366 in the @file{.gcno} files. This allows @command{gcov} to find the correct
12367 sources in projects where compilations occur with different working
12370 @item -fprofile-dir=@var{path}
12371 @opindex fprofile-dir
12373 Set the directory to search for the profile data files in to @var{path}.
12374 This option affects only the profile data generated by
12375 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
12376 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
12377 and its related options. Both absolute and relative paths can be used.
12378 By default, GCC uses the current directory as @var{path}, thus the
12379 profile data file appears in the same directory as the object file.
12380 In order to prevent the file name clashing, if the object file name is
12381 not an absolute path, we mangle the absolute path of the
12382 @file{@var{sourcename}.gcda} file and use it as the file name of a
12383 @file{.gcda} file. See similar option @option{-fprofile-note}.
12385 When an executable is run in a massive parallel environment, it is recommended
12386 to save profile to different folders. That can be done with variables
12387 in @var{path} that are exported during run-time:
12395 value of environment variable @var{VAR}
12399 @item -fprofile-generate
12400 @itemx -fprofile-generate=@var{path}
12401 @opindex fprofile-generate
12403 Enable options usually used for instrumenting application to produce
12404 profile useful for later recompilation with profile feedback based
12405 optimization. You must use @option{-fprofile-generate} both when
12406 compiling and when linking your program.
12408 The following options are enabled:
12409 @option{-fprofile-arcs}, @option{-fprofile-values},
12410 @option{-finline-functions}, and @option{-fipa-bit-cp}.
12412 If @var{path} is specified, GCC looks at the @var{path} to find
12413 the profile feedback data files. See @option{-fprofile-dir}.
12415 To optimize the program based on the collected profile information, use
12416 @option{-fprofile-use}. @xref{Optimize Options}, for more information.
12418 @item -fprofile-note=@var{path}
12419 @opindex fprofile-note
12421 If @var{path} is specified, GCC saves @file{.gcno} file into @var{path}
12422 location. If you combine the option with multiple source files,
12423 the @file{.gcno} file will be overwritten.
12425 @item -fprofile-update=@var{method}
12426 @opindex fprofile-update
12428 Alter the update method for an application instrumented for profile
12429 feedback based optimization. The @var{method} argument should be one of
12430 @samp{single}, @samp{atomic} or @samp{prefer-atomic}.
12431 The first one is useful for single-threaded applications,
12432 while the second one prevents profile corruption by emitting thread-safe code.
12434 @strong{Warning:} When an application does not properly join all threads
12435 (or creates an detached thread), a profile file can be still corrupted.
12437 Using @samp{prefer-atomic} would be transformed either to @samp{atomic},
12438 when supported by a target, or to @samp{single} otherwise. The GCC driver
12439 automatically selects @samp{prefer-atomic} when @option{-pthread}
12440 is present in the command line.
12442 @item -fprofile-filter-files=@var{regex}
12443 @opindex fprofile-filter-files
12445 Instrument only functions from files where names match
12446 any regular expression (separated by a semi-colon).
12448 For example, @option{-fprofile-filter-files=main.c;module.*.c} will instrument
12449 only @file{main.c} and all C files starting with 'module'.
12451 @item -fprofile-exclude-files=@var{regex}
12452 @opindex fprofile-exclude-files
12454 Instrument only functions from files where names do not match
12455 all the regular expressions (separated by a semi-colon).
12457 For example, @option{-fprofile-exclude-files=/usr/*} will prevent instrumentation
12458 of all files that are located in @file{/usr/} folder.
12460 @item -fsanitize=address
12461 @opindex fsanitize=address
12462 Enable AddressSanitizer, a fast memory error detector.
12463 Memory access instructions are instrumented to detect
12464 out-of-bounds and use-after-free bugs.
12465 The option enables @option{-fsanitize-address-use-after-scope}.
12466 See @uref{https://github.com/google/sanitizers/wiki/AddressSanitizer} for
12467 more details. The run-time behavior can be influenced using the
12468 @env{ASAN_OPTIONS} environment variable. When set to @code{help=1},
12469 the available options are shown at startup of the instrumented program. See
12470 @url{https://github.com/google/sanitizers/wiki/AddressSanitizerFlags#run-time-flags}
12471 for a list of supported options.
12472 The option cannot be combined with @option{-fsanitize=thread}.
12474 @item -fsanitize=kernel-address
12475 @opindex fsanitize=kernel-address
12476 Enable AddressSanitizer for Linux kernel.
12477 See @uref{https://github.com/google/kasan/wiki} for more details.
12479 @item -fsanitize=pointer-compare
12480 @opindex fsanitize=pointer-compare
12481 Instrument comparison operation (<, <=, >, >=) with pointer operands.
12482 The option must be combined with either @option{-fsanitize=kernel-address} or
12483 @option{-fsanitize=address}
12484 The option cannot be combined with @option{-fsanitize=thread}.
12485 Note: By default the check is disabled at run time. To enable it,
12486 add @code{detect_invalid_pointer_pairs=2} to the environment variable
12487 @env{ASAN_OPTIONS}. Using @code{detect_invalid_pointer_pairs=1} detects
12488 invalid operation only when both pointers are non-null.
12490 @item -fsanitize=pointer-subtract
12491 @opindex fsanitize=pointer-subtract
12492 Instrument subtraction with pointer operands.
12493 The option must be combined with either @option{-fsanitize=kernel-address} or
12494 @option{-fsanitize=address}
12495 The option cannot be combined with @option{-fsanitize=thread}.
12496 Note: By default the check is disabled at run time. To enable it,
12497 add @code{detect_invalid_pointer_pairs=2} to the environment variable
12498 @env{ASAN_OPTIONS}. Using @code{detect_invalid_pointer_pairs=1} detects
12499 invalid operation only when both pointers are non-null.
12501 @item -fsanitize=thread
12502 @opindex fsanitize=thread
12503 Enable ThreadSanitizer, a fast data race detector.
12504 Memory access instructions are instrumented to detect
12505 data race bugs. See @uref{https://github.com/google/sanitizers/wiki#threadsanitizer} for more
12506 details. The run-time behavior can be influenced using the @env{TSAN_OPTIONS}
12507 environment variable; see
12508 @url{https://github.com/google/sanitizers/wiki/ThreadSanitizerFlags} for a list of
12510 The option cannot be combined with @option{-fsanitize=address},
12511 @option{-fsanitize=leak}.
12513 Note that sanitized atomic builtins cannot throw exceptions when
12514 operating on invalid memory addresses with non-call exceptions
12515 (@option{-fnon-call-exceptions}).
12517 @item -fsanitize=leak
12518 @opindex fsanitize=leak
12519 Enable LeakSanitizer, a memory leak detector.
12520 This option only matters for linking of executables and
12521 the executable is linked against a library that overrides @code{malloc}
12522 and other allocator functions. See
12523 @uref{https://github.com/google/sanitizers/wiki/AddressSanitizerLeakSanitizer} for more
12524 details. The run-time behavior can be influenced using the
12525 @env{LSAN_OPTIONS} environment variable.
12526 The option cannot be combined with @option{-fsanitize=thread}.
12528 @item -fsanitize=undefined
12529 @opindex fsanitize=undefined
12530 Enable UndefinedBehaviorSanitizer, a fast undefined behavior detector.
12531 Various computations are instrumented to detect undefined behavior
12532 at runtime. Current suboptions are:
12536 @item -fsanitize=shift
12537 @opindex fsanitize=shift
12538 This option enables checking that the result of a shift operation is
12539 not undefined. Note that what exactly is considered undefined differs
12540 slightly between C and C++, as well as between ISO C90 and C99, etc.
12541 This option has two suboptions, @option{-fsanitize=shift-base} and
12542 @option{-fsanitize=shift-exponent}.
12544 @item -fsanitize=shift-exponent
12545 @opindex fsanitize=shift-exponent
12546 This option enables checking that the second argument of a shift operation
12547 is not negative and is smaller than the precision of the promoted first
12550 @item -fsanitize=shift-base
12551 @opindex fsanitize=shift-base
12552 If the second argument of a shift operation is within range, check that the
12553 result of a shift operation is not undefined. Note that what exactly is
12554 considered undefined differs slightly between C and C++, as well as between
12555 ISO C90 and C99, etc.
12557 @item -fsanitize=integer-divide-by-zero
12558 @opindex fsanitize=integer-divide-by-zero
12559 Detect integer division by zero as well as @code{INT_MIN / -1} division.
12561 @item -fsanitize=unreachable
12562 @opindex fsanitize=unreachable
12563 With this option, the compiler turns the @code{__builtin_unreachable}
12564 call into a diagnostics message call instead. When reaching the
12565 @code{__builtin_unreachable} call, the behavior is undefined.
12567 @item -fsanitize=vla-bound
12568 @opindex fsanitize=vla-bound
12569 This option instructs the compiler to check that the size of a variable
12570 length array is positive.
12572 @item -fsanitize=null
12573 @opindex fsanitize=null
12574 This option enables pointer checking. Particularly, the application
12575 built with this option turned on will issue an error message when it
12576 tries to dereference a NULL pointer, or if a reference (possibly an
12577 rvalue reference) is bound to a NULL pointer, or if a method is invoked
12578 on an object pointed by a NULL pointer.
12580 @item -fsanitize=return
12581 @opindex fsanitize=return
12582 This option enables return statement checking. Programs
12583 built with this option turned on will issue an error message
12584 when the end of a non-void function is reached without actually
12585 returning a value. This option works in C++ only.
12587 @item -fsanitize=signed-integer-overflow
12588 @opindex fsanitize=signed-integer-overflow
12589 This option enables signed integer overflow checking. We check that
12590 the result of @code{+}, @code{*}, and both unary and binary @code{-}
12591 does not overflow in the signed arithmetics. Note, integer promotion
12592 rules must be taken into account. That is, the following is not an
12595 signed char a = SCHAR_MAX;
12599 @item -fsanitize=bounds
12600 @opindex fsanitize=bounds
12601 This option enables instrumentation of array bounds. Various out of bounds
12602 accesses are detected. Flexible array members, flexible array member-like
12603 arrays, and initializers of variables with static storage are not instrumented.
12605 @item -fsanitize=bounds-strict
12606 @opindex fsanitize=bounds-strict
12607 This option enables strict instrumentation of array bounds. Most out of bounds
12608 accesses are detected, including flexible array members and flexible array
12609 member-like arrays. Initializers of variables with static storage are not
12612 @item -fsanitize=alignment
12613 @opindex fsanitize=alignment
12615 This option enables checking of alignment of pointers when they are
12616 dereferenced, or when a reference is bound to insufficiently aligned target,
12617 or when a method or constructor is invoked on insufficiently aligned object.
12619 @item -fsanitize=object-size
12620 @opindex fsanitize=object-size
12621 This option enables instrumentation of memory references using the
12622 @code{__builtin_object_size} function. Various out of bounds pointer
12623 accesses are detected.
12625 @item -fsanitize=float-divide-by-zero
12626 @opindex fsanitize=float-divide-by-zero
12627 Detect floating-point division by zero. Unlike other similar options,
12628 @option{-fsanitize=float-divide-by-zero} is not enabled by
12629 @option{-fsanitize=undefined}, since floating-point division by zero can
12630 be a legitimate way of obtaining infinities and NaNs.
12632 @item -fsanitize=float-cast-overflow
12633 @opindex fsanitize=float-cast-overflow
12634 This option enables floating-point type to integer conversion checking.
12635 We check that the result of the conversion does not overflow.
12636 Unlike other similar options, @option{-fsanitize=float-cast-overflow} is
12637 not enabled by @option{-fsanitize=undefined}.
12638 This option does not work well with @code{FE_INVALID} exceptions enabled.
12640 @item -fsanitize=nonnull-attribute
12641 @opindex fsanitize=nonnull-attribute
12643 This option enables instrumentation of calls, checking whether null values
12644 are not passed to arguments marked as requiring a non-null value by the
12645 @code{nonnull} function attribute.
12647 @item -fsanitize=returns-nonnull-attribute
12648 @opindex fsanitize=returns-nonnull-attribute
12650 This option enables instrumentation of return statements in functions
12651 marked with @code{returns_nonnull} function attribute, to detect returning
12652 of null values from such functions.
12654 @item -fsanitize=bool
12655 @opindex fsanitize=bool
12657 This option enables instrumentation of loads from bool. If a value other
12658 than 0/1 is loaded, a run-time error is issued.
12660 @item -fsanitize=enum
12661 @opindex fsanitize=enum
12663 This option enables instrumentation of loads from an enum type. If
12664 a value outside the range of values for the enum type is loaded,
12665 a run-time error is issued.
12667 @item -fsanitize=vptr
12668 @opindex fsanitize=vptr
12670 This option enables instrumentation of C++ member function calls, member
12671 accesses and some conversions between pointers to base and derived classes,
12672 to verify the referenced object has the correct dynamic type.
12674 @item -fsanitize=pointer-overflow
12675 @opindex fsanitize=pointer-overflow
12677 This option enables instrumentation of pointer arithmetics. If the pointer
12678 arithmetics overflows, a run-time error is issued.
12680 @item -fsanitize=builtin
12681 @opindex fsanitize=builtin
12683 This option enables instrumentation of arguments to selected builtin
12684 functions. If an invalid value is passed to such arguments, a run-time
12685 error is issued. E.g.@ passing 0 as the argument to @code{__builtin_ctz}
12686 or @code{__builtin_clz} invokes undefined behavior and is diagnosed
12691 While @option{-ftrapv} causes traps for signed overflows to be emitted,
12692 @option{-fsanitize=undefined} gives a diagnostic message.
12693 This currently works only for the C family of languages.
12695 @item -fno-sanitize=all
12696 @opindex fno-sanitize=all
12698 This option disables all previously enabled sanitizers.
12699 @option{-fsanitize=all} is not allowed, as some sanitizers cannot be used
12702 @item -fasan-shadow-offset=@var{number}
12703 @opindex fasan-shadow-offset
12704 This option forces GCC to use custom shadow offset in AddressSanitizer checks.
12705 It is useful for experimenting with different shadow memory layouts in
12706 Kernel AddressSanitizer.
12708 @item -fsanitize-sections=@var{s1},@var{s2},...
12709 @opindex fsanitize-sections
12710 Sanitize global variables in selected user-defined sections. @var{si} may
12713 @item -fsanitize-recover@r{[}=@var{opts}@r{]}
12714 @opindex fsanitize-recover
12715 @opindex fno-sanitize-recover
12716 @option{-fsanitize-recover=} controls error recovery mode for sanitizers
12717 mentioned in comma-separated list of @var{opts}. Enabling this option
12718 for a sanitizer component causes it to attempt to continue
12719 running the program as if no error happened. This means multiple
12720 runtime errors can be reported in a single program run, and the exit
12721 code of the program may indicate success even when errors
12722 have been reported. The @option{-fno-sanitize-recover=} option
12723 can be used to alter
12724 this behavior: only the first detected error is reported
12725 and program then exits with a non-zero exit code.
12727 Currently this feature only works for @option{-fsanitize=undefined} (and its suboptions
12728 except for @option{-fsanitize=unreachable} and @option{-fsanitize=return}),
12729 @option{-fsanitize=float-cast-overflow}, @option{-fsanitize=float-divide-by-zero},
12730 @option{-fsanitize=bounds-strict},
12731 @option{-fsanitize=kernel-address} and @option{-fsanitize=address}.
12732 For these sanitizers error recovery is turned on by default,
12733 except @option{-fsanitize=address}, for which this feature is experimental.
12734 @option{-fsanitize-recover=all} and @option{-fno-sanitize-recover=all} is also
12735 accepted, the former enables recovery for all sanitizers that support it,
12736 the latter disables recovery for all sanitizers that support it.
12738 Even if a recovery mode is turned on the compiler side, it needs to be also
12739 enabled on the runtime library side, otherwise the failures are still fatal.
12740 The runtime library defaults to @code{halt_on_error=0} for
12741 ThreadSanitizer and UndefinedBehaviorSanitizer, while default value for
12742 AddressSanitizer is @code{halt_on_error=1}. This can be overridden through
12743 setting the @code{halt_on_error} flag in the corresponding environment variable.
12745 Syntax without an explicit @var{opts} parameter is deprecated. It is
12746 equivalent to specifying an @var{opts} list of:
12749 undefined,float-cast-overflow,float-divide-by-zero,bounds-strict
12752 @item -fsanitize-address-use-after-scope
12753 @opindex fsanitize-address-use-after-scope
12754 Enable sanitization of local variables to detect use-after-scope bugs.
12755 The option sets @option{-fstack-reuse} to @samp{none}.
12757 @item -fsanitize-undefined-trap-on-error
12758 @opindex fsanitize-undefined-trap-on-error
12759 The @option{-fsanitize-undefined-trap-on-error} option instructs the compiler to
12760 report undefined behavior using @code{__builtin_trap} rather than
12761 a @code{libubsan} library routine. The advantage of this is that the
12762 @code{libubsan} library is not needed and is not linked in, so this
12763 is usable even in freestanding environments.
12765 @item -fsanitize-coverage=trace-pc
12766 @opindex fsanitize-coverage=trace-pc
12767 Enable coverage-guided fuzzing code instrumentation.
12768 Inserts a call to @code{__sanitizer_cov_trace_pc} into every basic block.
12770 @item -fsanitize-coverage=trace-cmp
12771 @opindex fsanitize-coverage=trace-cmp
12772 Enable dataflow guided fuzzing code instrumentation.
12773 Inserts a call to @code{__sanitizer_cov_trace_cmp1},
12774 @code{__sanitizer_cov_trace_cmp2}, @code{__sanitizer_cov_trace_cmp4} or
12775 @code{__sanitizer_cov_trace_cmp8} for integral comparison with both operands
12776 variable or @code{__sanitizer_cov_trace_const_cmp1},
12777 @code{__sanitizer_cov_trace_const_cmp2},
12778 @code{__sanitizer_cov_trace_const_cmp4} or
12779 @code{__sanitizer_cov_trace_const_cmp8} for integral comparison with one
12780 operand constant, @code{__sanitizer_cov_trace_cmpf} or
12781 @code{__sanitizer_cov_trace_cmpd} for float or double comparisons and
12782 @code{__sanitizer_cov_trace_switch} for switch statements.
12784 @item -fcf-protection=@r{[}full@r{|}branch@r{|}return@r{|}none@r{]}
12785 @opindex fcf-protection
12786 Enable code instrumentation of control-flow transfers to increase
12787 program security by checking that target addresses of control-flow
12788 transfer instructions (such as indirect function call, function return,
12789 indirect jump) are valid. This prevents diverting the flow of control
12790 to an unexpected target. This is intended to protect against such
12791 threats as Return-oriented Programming (ROP), and similarly
12792 call/jmp-oriented programming (COP/JOP).
12794 The value @code{branch} tells the compiler to implement checking of
12795 validity of control-flow transfer at the point of indirect branch
12796 instructions, i.e.@: call/jmp instructions. The value @code{return}
12797 implements checking of validity at the point of returning from a
12798 function. The value @code{full} is an alias for specifying both
12799 @code{branch} and @code{return}. The value @code{none} turns off
12802 The macro @code{__CET__} is defined when @option{-fcf-protection} is
12803 used. The first bit of @code{__CET__} is set to 1 for the value
12804 @code{branch} and the second bit of @code{__CET__} is set to 1 for
12807 You can also use the @code{nocf_check} attribute to identify
12808 which functions and calls should be skipped from instrumentation
12809 (@pxref{Function Attributes}).
12811 Currently the x86 GNU/Linux target provides an implementation based
12812 on Intel Control-flow Enforcement Technology (CET).
12814 @item -fstack-protector
12815 @opindex fstack-protector
12816 Emit extra code to check for buffer overflows, such as stack smashing
12817 attacks. This is done by adding a guard variable to functions with
12818 vulnerable objects. This includes functions that call @code{alloca}, and
12819 functions with buffers larger than 8 bytes. The guards are initialized
12820 when a function is entered and then checked when the function exits.
12821 If a guard check fails, an error message is printed and the program exits.
12823 @item -fstack-protector-all
12824 @opindex fstack-protector-all
12825 Like @option{-fstack-protector} except that all functions are protected.
12827 @item -fstack-protector-strong
12828 @opindex fstack-protector-strong
12829 Like @option{-fstack-protector} but includes additional functions to
12830 be protected --- those that have local array definitions, or have
12831 references to local frame addresses.
12833 @item -fstack-protector-explicit
12834 @opindex fstack-protector-explicit
12835 Like @option{-fstack-protector} but only protects those functions which
12836 have the @code{stack_protect} attribute.
12838 @item -fstack-check
12839 @opindex fstack-check
12840 Generate code to verify that you do not go beyond the boundary of the
12841 stack. You should specify this flag if you are running in an
12842 environment with multiple threads, but you only rarely need to specify it in
12843 a single-threaded environment since stack overflow is automatically
12844 detected on nearly all systems if there is only one stack.
12846 Note that this switch does not actually cause checking to be done; the
12847 operating system or the language runtime must do that. The switch causes
12848 generation of code to ensure that they see the stack being extended.
12850 You can additionally specify a string parameter: @samp{no} means no
12851 checking, @samp{generic} means force the use of old-style checking,
12852 @samp{specific} means use the best checking method and is equivalent
12853 to bare @option{-fstack-check}.
12855 Old-style checking is a generic mechanism that requires no specific
12856 target support in the compiler but comes with the following drawbacks:
12860 Modified allocation strategy for large objects: they are always
12861 allocated dynamically if their size exceeds a fixed threshold. Note this
12862 may change the semantics of some code.
12865 Fixed limit on the size of the static frame of functions: when it is
12866 topped by a particular function, stack checking is not reliable and
12867 a warning is issued by the compiler.
12870 Inefficiency: because of both the modified allocation strategy and the
12871 generic implementation, code performance is hampered.
12874 Note that old-style stack checking is also the fallback method for
12875 @samp{specific} if no target support has been added in the compiler.
12877 @samp{-fstack-check=} is designed for Ada's needs to detect infinite recursion
12878 and stack overflows. @samp{specific} is an excellent choice when compiling
12879 Ada code. It is not generally sufficient to protect against stack-clash
12880 attacks. To protect against those you want @samp{-fstack-clash-protection}.
12882 @item -fstack-clash-protection
12883 @opindex fstack-clash-protection
12884 Generate code to prevent stack clash style attacks. When this option is
12885 enabled, the compiler will only allocate one page of stack space at a time
12886 and each page is accessed immediately after allocation. Thus, it prevents
12887 allocations from jumping over any stack guard page provided by the
12890 Most targets do not fully support stack clash protection. However, on
12891 those targets @option{-fstack-clash-protection} will protect dynamic stack
12892 allocations. @option{-fstack-clash-protection} may also provide limited
12893 protection for static stack allocations if the target supports
12894 @option{-fstack-check=specific}.
12896 @item -fstack-limit-register=@var{reg}
12897 @itemx -fstack-limit-symbol=@var{sym}
12898 @itemx -fno-stack-limit
12899 @opindex fstack-limit-register
12900 @opindex fstack-limit-symbol
12901 @opindex fno-stack-limit
12902 Generate code to ensure that the stack does not grow beyond a certain value,
12903 either the value of a register or the address of a symbol. If a larger
12904 stack is required, a signal is raised at run time. For most targets,
12905 the signal is raised before the stack overruns the boundary, so
12906 it is possible to catch the signal without taking special precautions.
12908 For instance, if the stack starts at absolute address @samp{0x80000000}
12909 and grows downwards, you can use the flags
12910 @option{-fstack-limit-symbol=__stack_limit} and
12911 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
12912 of 128KB@. Note that this may only work with the GNU linker.
12914 You can locally override stack limit checking by using the
12915 @code{no_stack_limit} function attribute (@pxref{Function Attributes}).
12917 @item -fsplit-stack
12918 @opindex fsplit-stack
12919 Generate code to automatically split the stack before it overflows.
12920 The resulting program has a discontiguous stack which can only
12921 overflow if the program is unable to allocate any more memory. This
12922 is most useful when running threaded programs, as it is no longer
12923 necessary to calculate a good stack size to use for each thread. This
12924 is currently only implemented for the x86 targets running
12927 When code compiled with @option{-fsplit-stack} calls code compiled
12928 without @option{-fsplit-stack}, there may not be much stack space
12929 available for the latter code to run. If compiling all code,
12930 including library code, with @option{-fsplit-stack} is not an option,
12931 then the linker can fix up these calls so that the code compiled
12932 without @option{-fsplit-stack} always has a large stack. Support for
12933 this is implemented in the gold linker in GNU binutils release 2.21
12936 @item -fvtable-verify=@r{[}std@r{|}preinit@r{|}none@r{]}
12937 @opindex fvtable-verify
12938 This option is only available when compiling C++ code.
12939 It turns on (or off, if using @option{-fvtable-verify=none}) the security
12940 feature that verifies at run time, for every virtual call, that
12941 the vtable pointer through which the call is made is valid for the type of
12942 the object, and has not been corrupted or overwritten. If an invalid vtable
12943 pointer is detected at run time, an error is reported and execution of the
12944 program is immediately halted.
12946 This option causes run-time data structures to be built at program startup,
12947 which are used for verifying the vtable pointers.
12948 The options @samp{std} and @samp{preinit}
12949 control the timing of when these data structures are built. In both cases the
12950 data structures are built before execution reaches @code{main}. Using
12951 @option{-fvtable-verify=std} causes the data structures to be built after
12952 shared libraries have been loaded and initialized.
12953 @option{-fvtable-verify=preinit} causes them to be built before shared
12954 libraries have been loaded and initialized.
12956 If this option appears multiple times in the command line with different
12957 values specified, @samp{none} takes highest priority over both @samp{std} and
12958 @samp{preinit}; @samp{preinit} takes priority over @samp{std}.
12961 @opindex fvtv-debug
12962 When used in conjunction with @option{-fvtable-verify=std} or
12963 @option{-fvtable-verify=preinit}, causes debug versions of the
12964 runtime functions for the vtable verification feature to be called.
12965 This flag also causes the compiler to log information about which
12966 vtable pointers it finds for each class.
12967 This information is written to a file named @file{vtv_set_ptr_data.log}
12968 in the directory named by the environment variable @env{VTV_LOGS_DIR}
12969 if that is defined or the current working directory otherwise.
12971 Note: This feature @emph{appends} data to the log file. If you want a fresh log
12972 file, be sure to delete any existing one.
12975 @opindex fvtv-counts
12976 This is a debugging flag. When used in conjunction with
12977 @option{-fvtable-verify=std} or @option{-fvtable-verify=preinit}, this
12978 causes the compiler to keep track of the total number of virtual calls
12979 it encounters and the number of verifications it inserts. It also
12980 counts the number of calls to certain run-time library functions
12981 that it inserts and logs this information for each compilation unit.
12982 The compiler writes this information to a file named
12983 @file{vtv_count_data.log} in the directory named by the environment
12984 variable @env{VTV_LOGS_DIR} if that is defined or the current working
12985 directory otherwise. It also counts the size of the vtable pointer sets
12986 for each class, and writes this information to @file{vtv_class_set_sizes.log}
12987 in the same directory.
12989 Note: This feature @emph{appends} data to the log files. To get fresh log
12990 files, be sure to delete any existing ones.
12992 @item -finstrument-functions
12993 @opindex finstrument-functions
12994 Generate instrumentation calls for entry and exit to functions. Just
12995 after function entry and just before function exit, the following
12996 profiling functions are called with the address of the current
12997 function and its call site. (On some platforms,
12998 @code{__builtin_return_address} does not work beyond the current
12999 function, so the call site information may not be available to the
13000 profiling functions otherwise.)
13003 void __cyg_profile_func_enter (void *this_fn,
13005 void __cyg_profile_func_exit (void *this_fn,
13009 The first argument is the address of the start of the current function,
13010 which may be looked up exactly in the symbol table.
13012 This instrumentation is also done for functions expanded inline in other
13013 functions. The profiling calls indicate where, conceptually, the
13014 inline function is entered and exited. This means that addressable
13015 versions of such functions must be available. If all your uses of a
13016 function are expanded inline, this may mean an additional expansion of
13017 code size. If you use @code{extern inline} in your C code, an
13018 addressable version of such functions must be provided. (This is
13019 normally the case anyway, but if you get lucky and the optimizer always
13020 expands the functions inline, you might have gotten away without
13021 providing static copies.)
13023 A function may be given the attribute @code{no_instrument_function}, in
13024 which case this instrumentation is not done. This can be used, for
13025 example, for the profiling functions listed above, high-priority
13026 interrupt routines, and any functions from which the profiling functions
13027 cannot safely be called (perhaps signal handlers, if the profiling
13028 routines generate output or allocate memory).
13029 @xref{Common Function Attributes}.
13031 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
13032 @opindex finstrument-functions-exclude-file-list
13034 Set the list of functions that are excluded from instrumentation (see
13035 the description of @option{-finstrument-functions}). If the file that
13036 contains a function definition matches with one of @var{file}, then
13037 that function is not instrumented. The match is done on substrings:
13038 if the @var{file} parameter is a substring of the file name, it is
13039 considered to be a match.
13044 -finstrument-functions-exclude-file-list=/bits/stl,include/sys
13048 excludes any inline function defined in files whose pathnames
13049 contain @file{/bits/stl} or @file{include/sys}.
13051 If, for some reason, you want to include letter @samp{,} in one of
13052 @var{sym}, write @samp{\,}. For example,
13053 @option{-finstrument-functions-exclude-file-list='\,\,tmp'}
13054 (note the single quote surrounding the option).
13056 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
13057 @opindex finstrument-functions-exclude-function-list
13059 This is similar to @option{-finstrument-functions-exclude-file-list},
13060 but this option sets the list of function names to be excluded from
13061 instrumentation. The function name to be matched is its user-visible
13062 name, such as @code{vector<int> blah(const vector<int> &)}, not the
13063 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
13064 match is done on substrings: if the @var{sym} parameter is a substring
13065 of the function name, it is considered to be a match. For C99 and C++
13066 extended identifiers, the function name must be given in UTF-8, not
13067 using universal character names.
13069 @item -fpatchable-function-entry=@var{N}[,@var{M}]
13070 @opindex fpatchable-function-entry
13071 Generate @var{N} NOPs right at the beginning
13072 of each function, with the function entry point before the @var{M}th NOP.
13073 If @var{M} is omitted, it defaults to @code{0} so the
13074 function entry points to the address just at the first NOP.
13075 The NOP instructions reserve extra space which can be used to patch in
13076 any desired instrumentation at run time, provided that the code segment
13077 is writable. The amount of space is controllable indirectly via
13078 the number of NOPs; the NOP instruction used corresponds to the instruction
13079 emitted by the internal GCC back-end interface @code{gen_nop}. This behavior
13080 is target-specific and may also depend on the architecture variant and/or
13081 other compilation options.
13083 For run-time identification, the starting addresses of these areas,
13084 which correspond to their respective function entries minus @var{M},
13085 are additionally collected in the @code{__patchable_function_entries}
13086 section of the resulting binary.
13088 Note that the value of @code{__attribute__ ((patchable_function_entry
13089 (N,M)))} takes precedence over command-line option
13090 @option{-fpatchable-function-entry=N,M}. This can be used to increase
13091 the area size or to remove it completely on a single function.
13092 If @code{N=0}, no pad location is recorded.
13094 The NOP instructions are inserted at---and maybe before, depending on
13095 @var{M}---the function entry address, even before the prologue.
13100 @node Preprocessor Options
13101 @section Options Controlling the Preprocessor
13102 @cindex preprocessor options
13103 @cindex options, preprocessor
13105 These options control the C preprocessor, which is run on each C source
13106 file before actual compilation.
13108 If you use the @option{-E} option, nothing is done except preprocessing.
13109 Some of these options make sense only together with @option{-E} because
13110 they cause the preprocessor output to be unsuitable for actual
13113 In addition to the options listed here, there are a number of options
13114 to control search paths for include files documented in
13115 @ref{Directory Options}.
13116 Options to control preprocessor diagnostics are listed in
13117 @ref{Warning Options}.
13120 @include cppopts.texi
13122 @item -Wp,@var{option}
13124 You can use @option{-Wp,@var{option}} to bypass the compiler driver
13125 and pass @var{option} directly through to the preprocessor. If
13126 @var{option} contains commas, it is split into multiple options at the
13127 commas. However, many options are modified, translated or interpreted
13128 by the compiler driver before being passed to the preprocessor, and
13129 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
13130 interface is undocumented and subject to change, so whenever possible
13131 you should avoid using @option{-Wp} and let the driver handle the
13134 @item -Xpreprocessor @var{option}
13135 @opindex Xpreprocessor
13136 Pass @var{option} as an option to the preprocessor. You can use this to
13137 supply system-specific preprocessor options that GCC does not
13140 If you want to pass an option that takes an argument, you must use
13141 @option{-Xpreprocessor} twice, once for the option and once for the argument.
13143 @item -no-integrated-cpp
13144 @opindex no-integrated-cpp
13145 Perform preprocessing as a separate pass before compilation.
13146 By default, GCC performs preprocessing as an integrated part of
13147 input tokenization and parsing.
13148 If this option is provided, the appropriate language front end
13149 (@command{cc1}, @command{cc1plus}, or @command{cc1obj} for C, C++,
13150 and Objective-C, respectively) is instead invoked twice,
13151 once for preprocessing only and once for actual compilation
13152 of the preprocessed input.
13153 This option may be useful in conjunction with the @option{-B} or
13154 @option{-wrapper} options to specify an alternate preprocessor or
13155 perform additional processing of the program source between
13156 normal preprocessing and compilation.
13160 @node Assembler Options
13161 @section Passing Options to the Assembler
13163 @c prevent bad page break with this line
13164 You can pass options to the assembler.
13167 @item -Wa,@var{option}
13169 Pass @var{option} as an option to the assembler. If @var{option}
13170 contains commas, it is split into multiple options at the commas.
13172 @item -Xassembler @var{option}
13173 @opindex Xassembler
13174 Pass @var{option} as an option to the assembler. You can use this to
13175 supply system-specific assembler options that GCC does not
13178 If you want to pass an option that takes an argument, you must use
13179 @option{-Xassembler} twice, once for the option and once for the argument.
13184 @section Options for Linking
13185 @cindex link options
13186 @cindex options, linking
13188 These options come into play when the compiler links object files into
13189 an executable output file. They are meaningless if the compiler is
13190 not doing a link step.
13194 @item @var{object-file-name}
13195 A file name that does not end in a special recognized suffix is
13196 considered to name an object file or library. (Object files are
13197 distinguished from libraries by the linker according to the file
13198 contents.) If linking is done, these object files are used as input
13207 If any of these options is used, then the linker is not run, and
13208 object file names should not be used as arguments. @xref{Overall
13211 @item -flinker-output=@var{type}
13212 @opindex flinker-output
13213 This option controls code generation of the link-time optimizer. By
13214 default the linker output is automatically determined by the linker
13215 plugin. For debugging the compiler and if incremental linking with a
13216 non-LTO object file is desired, it may be useful to control the type
13219 If @var{type} is @samp{exec}, code generation produces a static
13220 binary. In this case @option{-fpic} and @option{-fpie} are both
13223 If @var{type} is @samp{dyn}, code generation produces a shared
13224 library. In this case @option{-fpic} or @option{-fPIC} is preserved,
13225 but not enabled automatically. This allows to build shared libraries
13226 without position-independent code on architectures where this is
13227 possible, i.e.@: on x86.
13229 If @var{type} is @samp{pie}, code generation produces an @option{-fpie}
13230 executable. This results in similar optimizations as @samp{exec}
13231 except that @option{-fpie} is not disabled if specified at compilation
13234 If @var{type} is @samp{rel}, the compiler assumes that incremental linking is
13235 done. The sections containing intermediate code for link-time optimization are
13236 merged, pre-optimized, and output to the resulting object file. In addition, if
13237 @option{-ffat-lto-objects} is specified, binary code is produced for future
13238 non-LTO linking. The object file produced by incremental linking is smaller
13239 than a static library produced from the same object files. At link time the
13240 result of incremental linking also loads faster than a static
13241 library assuming that the majority of objects in the library are used.
13243 Finally @samp{nolto-rel} configures the compiler for incremental linking where
13244 code generation is forced, a final binary is produced, and the intermediate
13245 code for later link-time optimization is stripped. When multiple object files
13246 are linked together the resulting code is better optimized than with
13247 link-time optimizations disabled (for example, cross-module inlining
13248 happens), but most of benefits of whole program optimizations are lost.
13250 During the incremental link (by @option{-r}) the linker plugin defaults to
13251 @option{rel}. With current interfaces to GNU Binutils it is however not
13252 possible to incrementally link LTO objects and non-LTO objects into a single
13253 mixed object file. If any of object files in incremental link cannot
13254 be used for link-time optimization, the linker plugin issues a warning and
13255 uses @samp{nolto-rel}. To maintain whole program optimization, it is
13256 recommended to link such objects into static library instead. Alternatively it
13257 is possible to use H.J. Lu's binutils with support for mixed objects.
13260 @opindex fuse-ld=bfd
13261 Use the @command{bfd} linker instead of the default linker.
13263 @item -fuse-ld=gold
13264 @opindex fuse-ld=gold
13265 Use the @command{gold} linker instead of the default linker.
13268 @opindex fuse-ld=lld
13269 Use the LLVM @command{lld} linker instead of the default linker.
13272 @item -l@var{library}
13273 @itemx -l @var{library}
13275 Search the library named @var{library} when linking. (The second
13276 alternative with the library as a separate argument is only for
13277 POSIX compliance and is not recommended.)
13279 The @option{-l} option is passed directly to the linker by GCC. Refer
13280 to your linker documentation for exact details. The general
13281 description below applies to the GNU linker.
13283 The linker searches a standard list of directories for the library.
13284 The directories searched include several standard system directories
13285 plus any that you specify with @option{-L}.
13287 Static libraries are archives of object files, and have file names
13288 like @file{lib@var{library}.a}. Some targets also support shared
13289 libraries, which typically have names like @file{lib@var{library}.so}.
13290 If both static and shared libraries are found, the linker gives
13291 preference to linking with the shared library unless the
13292 @option{-static} option is used.
13294 It makes a difference where in the command you write this option; the
13295 linker searches and processes libraries and object files in the order they
13296 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
13297 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
13298 to functions in @samp{z}, those functions may not be loaded.
13302 You need this special case of the @option{-l} option in order to
13303 link an Objective-C or Objective-C++ program.
13305 @item -nostartfiles
13306 @opindex nostartfiles
13307 Do not use the standard system startup files when linking.
13308 The standard system libraries are used normally, unless @option{-nostdlib},
13309 @option{-nolibc}, or @option{-nodefaultlibs} is used.
13311 @item -nodefaultlibs
13312 @opindex nodefaultlibs
13313 Do not use the standard system libraries when linking.
13314 Only the libraries you specify are passed to the linker, and options
13315 specifying linkage of the system libraries, such as @option{-static-libgcc}
13316 or @option{-shared-libgcc}, are ignored.
13317 The standard startup files are used normally, unless @option{-nostartfiles}
13320 The compiler may generate calls to @code{memcmp},
13321 @code{memset}, @code{memcpy} and @code{memmove}.
13322 These entries are usually resolved by entries in
13323 libc. These entry points should be supplied through some other
13324 mechanism when this option is specified.
13328 Do not use the C library or system libraries tightly coupled with it when
13329 linking. Still link with the startup files, @file{libgcc} or toolchain
13330 provided language support libraries such as @file{libgnat}, @file{libgfortran}
13331 or @file{libstdc++} unless options preventing their inclusion are used as
13332 well. This typically removes @option{-lc} from the link command line, as well
13333 as system libraries that normally go with it and become meaningless when
13334 absence of a C library is assumed, for example @option{-lpthread} or
13335 @option{-lm} in some configurations. This is intended for bare-board
13336 targets when there is indeed no C library available.
13340 Do not use the standard system startup files or libraries when linking.
13341 No startup files and only the libraries you specify are passed to
13342 the linker, and options specifying linkage of the system libraries, such as
13343 @option{-static-libgcc} or @option{-shared-libgcc}, are ignored.
13345 The compiler may generate calls to @code{memcmp}, @code{memset},
13346 @code{memcpy} and @code{memmove}.
13347 These entries are usually resolved by entries in
13348 libc. These entry points should be supplied through some other
13349 mechanism when this option is specified.
13351 @cindex @option{-lgcc}, use with @option{-nostdlib}
13352 @cindex @option{-nostdlib} and unresolved references
13353 @cindex unresolved references and @option{-nostdlib}
13354 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
13355 @cindex @option{-nodefaultlibs} and unresolved references
13356 @cindex unresolved references and @option{-nodefaultlibs}
13357 One of the standard libraries bypassed by @option{-nostdlib} and
13358 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
13359 which GCC uses to overcome shortcomings of particular machines, or special
13360 needs for some languages.
13361 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
13362 Collection (GCC) Internals},
13363 for more discussion of @file{libgcc.a}.)
13364 In most cases, you need @file{libgcc.a} even when you want to avoid
13365 other standard libraries. In other words, when you specify @option{-nostdlib}
13366 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
13367 This ensures that you have no unresolved references to internal GCC
13368 library subroutines.
13369 (An example of such an internal subroutine is @code{__main}, used to ensure C++
13370 constructors are called; @pxref{Collect2,,@code{collect2}, gccint,
13371 GNU Compiler Collection (GCC) Internals}.)
13373 @item -e @var{entry}
13374 @itemx --entry=@var{entry}
13378 Specify that the program entry point is @var{entry}. The argument is
13379 interpreted by the linker; the GNU linker accepts either a symbol name
13384 Produce a dynamically linked position independent executable on targets
13385 that support it. For predictable results, you must also specify the same
13386 set of options used for compilation (@option{-fpie}, @option{-fPIE},
13387 or model suboptions) when you specify this linker option.
13391 Don't produce a dynamically linked position independent executable.
13394 @opindex static-pie
13395 Produce a static position independent executable on targets that support
13396 it. A static position independent executable is similar to a static
13397 executable, but can be loaded at any address without a dynamic linker.
13398 For predictable results, you must also specify the same set of options
13399 used for compilation (@option{-fpie}, @option{-fPIE}, or model
13400 suboptions) when you specify this linker option.
13404 Link with the POSIX threads library. This option is supported on
13405 GNU/Linux targets, most other Unix derivatives, and also on
13406 x86 Cygwin and MinGW targets. On some targets this option also sets
13407 flags for the preprocessor, so it should be used consistently for both
13408 compilation and linking.
13412 Produce a relocatable object as output. This is also known as partial
13417 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
13418 that support it. This instructs the linker to add all symbols, not
13419 only used ones, to the dynamic symbol table. This option is needed
13420 for some uses of @code{dlopen} or to allow obtaining backtraces
13421 from within a program.
13425 Remove all symbol table and relocation information from the executable.
13429 On systems that support dynamic linking, this overrides @option{-pie}
13430 and prevents linking with the shared libraries. On other systems, this
13431 option has no effect.
13435 Produce a shared object which can then be linked with other objects to
13436 form an executable. Not all systems support this option. For predictable
13437 results, you must also specify the same set of options used for compilation
13438 (@option{-fpic}, @option{-fPIC}, or model suboptions) when
13439 you specify this linker option.@footnote{On some systems, @samp{gcc -shared}
13440 needs to build supplementary stub code for constructors to work. On
13441 multi-libbed systems, @samp{gcc -shared} must select the correct support
13442 libraries to link against. Failing to supply the correct flags may lead
13443 to subtle defects. Supplying them in cases where they are not necessary
13446 @item -shared-libgcc
13447 @itemx -static-libgcc
13448 @opindex shared-libgcc
13449 @opindex static-libgcc
13450 On systems that provide @file{libgcc} as a shared library, these options
13451 force the use of either the shared or static version, respectively.
13452 If no shared version of @file{libgcc} was built when the compiler was
13453 configured, these options have no effect.
13455 There are several situations in which an application should use the
13456 shared @file{libgcc} instead of the static version. The most common
13457 of these is when the application wishes to throw and catch exceptions
13458 across different shared libraries. In that case, each of the libraries
13459 as well as the application itself should use the shared @file{libgcc}.
13461 Therefore, the G++ driver automatically adds @option{-shared-libgcc}
13462 whenever you build a shared library or a main executable, because C++
13463 programs typically use exceptions, so this is the right thing to do.
13465 If, instead, you use the GCC driver to create shared libraries, you may
13466 find that they are not always linked with the shared @file{libgcc}.
13467 If GCC finds, at its configuration time, that you have a non-GNU linker
13468 or a GNU linker that does not support option @option{--eh-frame-hdr},
13469 it links the shared version of @file{libgcc} into shared libraries
13470 by default. Otherwise, it takes advantage of the linker and optimizes
13471 away the linking with the shared version of @file{libgcc}, linking with
13472 the static version of libgcc by default. This allows exceptions to
13473 propagate through such shared libraries, without incurring relocation
13474 costs at library load time.
13476 However, if a library or main executable is supposed to throw or catch
13477 exceptions, you must link it using the G++ driver, or using the option
13478 @option{-shared-libgcc}, such that it is linked with the shared
13481 @item -static-libasan
13482 @opindex static-libasan
13483 When the @option{-fsanitize=address} option is used to link a program,
13484 the GCC driver automatically links against @option{libasan}. If
13485 @file{libasan} is available as a shared library, and the @option{-static}
13486 option is not used, then this links against the shared version of
13487 @file{libasan}. The @option{-static-libasan} option directs the GCC
13488 driver to link @file{libasan} statically, without necessarily linking
13489 other libraries statically.
13491 @item -static-libtsan
13492 @opindex static-libtsan
13493 When the @option{-fsanitize=thread} option is used to link a program,
13494 the GCC driver automatically links against @option{libtsan}. If
13495 @file{libtsan} is available as a shared library, and the @option{-static}
13496 option is not used, then this links against the shared version of
13497 @file{libtsan}. The @option{-static-libtsan} option directs the GCC
13498 driver to link @file{libtsan} statically, without necessarily linking
13499 other libraries statically.
13501 @item -static-liblsan
13502 @opindex static-liblsan
13503 When the @option{-fsanitize=leak} option is used to link a program,
13504 the GCC driver automatically links against @option{liblsan}. If
13505 @file{liblsan} is available as a shared library, and the @option{-static}
13506 option is not used, then this links against the shared version of
13507 @file{liblsan}. The @option{-static-liblsan} option directs the GCC
13508 driver to link @file{liblsan} statically, without necessarily linking
13509 other libraries statically.
13511 @item -static-libubsan
13512 @opindex static-libubsan
13513 When the @option{-fsanitize=undefined} option is used to link a program,
13514 the GCC driver automatically links against @option{libubsan}. If
13515 @file{libubsan} is available as a shared library, and the @option{-static}
13516 option is not used, then this links against the shared version of
13517 @file{libubsan}. The @option{-static-libubsan} option directs the GCC
13518 driver to link @file{libubsan} statically, without necessarily linking
13519 other libraries statically.
13521 @item -static-libstdc++
13522 @opindex static-libstdc++
13523 When the @command{g++} program is used to link a C++ program, it
13524 normally automatically links against @option{libstdc++}. If
13525 @file{libstdc++} is available as a shared library, and the
13526 @option{-static} option is not used, then this links against the
13527 shared version of @file{libstdc++}. That is normally fine. However, it
13528 is sometimes useful to freeze the version of @file{libstdc++} used by
13529 the program without going all the way to a fully static link. The
13530 @option{-static-libstdc++} option directs the @command{g++} driver to
13531 link @file{libstdc++} statically, without necessarily linking other
13532 libraries statically.
13536 Bind references to global symbols when building a shared object. Warn
13537 about any unresolved references (unless overridden by the link editor
13538 option @option{-Xlinker -z -Xlinker defs}). Only a few systems support
13541 @item -T @var{script}
13543 @cindex linker script
13544 Use @var{script} as the linker script. This option is supported by most
13545 systems using the GNU linker. On some targets, such as bare-board
13546 targets without an operating system, the @option{-T} option may be required
13547 when linking to avoid references to undefined symbols.
13549 @item -Xlinker @var{option}
13551 Pass @var{option} as an option to the linker. You can use this to
13552 supply system-specific linker options that GCC does not recognize.
13554 If you want to pass an option that takes a separate argument, you must use
13555 @option{-Xlinker} twice, once for the option and once for the argument.
13556 For example, to pass @option{-assert definitions}, you must write
13557 @option{-Xlinker -assert -Xlinker definitions}. It does not work to write
13558 @option{-Xlinker "-assert definitions"}, because this passes the entire
13559 string as a single argument, which is not what the linker expects.
13561 When using the GNU linker, it is usually more convenient to pass
13562 arguments to linker options using the @option{@var{option}=@var{value}}
13563 syntax than as separate arguments. For example, you can specify
13564 @option{-Xlinker -Map=output.map} rather than
13565 @option{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
13566 this syntax for command-line options.
13568 @item -Wl,@var{option}
13570 Pass @var{option} as an option to the linker. If @var{option} contains
13571 commas, it is split into multiple options at the commas. You can use this
13572 syntax to pass an argument to the option.
13573 For example, @option{-Wl,-Map,output.map} passes @option{-Map output.map} to the
13574 linker. When using the GNU linker, you can also get the same effect with
13575 @option{-Wl,-Map=output.map}.
13577 @item -u @var{symbol}
13579 Pretend the symbol @var{symbol} is undefined, to force linking of
13580 library modules to define it. You can use @option{-u} multiple times with
13581 different symbols to force loading of additional library modules.
13583 @item -z @var{keyword}
13585 @option{-z} is passed directly on to the linker along with the keyword
13586 @var{keyword}. See the section in the documentation of your linker for
13587 permitted values and their meanings.
13590 @node Directory Options
13591 @section Options for Directory Search
13592 @cindex directory options
13593 @cindex options, directory search
13594 @cindex search path
13596 These options specify directories to search for header files, for
13597 libraries and for parts of the compiler:
13600 @include cppdiropts.texi
13602 @item -iplugindir=@var{dir}
13603 @opindex iplugindir=
13604 Set the directory to search for plugins that are passed
13605 by @option{-fplugin=@var{name}} instead of
13606 @option{-fplugin=@var{path}/@var{name}.so}. This option is not meant
13607 to be used by the user, but only passed by the driver.
13611 Add directory @var{dir} to the list of directories to be searched
13614 @item -B@var{prefix}
13616 This option specifies where to find the executables, libraries,
13617 include files, and data files of the compiler itself.
13619 The compiler driver program runs one or more of the subprograms
13620 @command{cpp}, @command{cc1}, @command{as} and @command{ld}. It tries
13621 @var{prefix} as a prefix for each program it tries to run, both with and
13622 without @samp{@var{machine}/@var{version}/} for the corresponding target
13623 machine and compiler version.
13625 For each subprogram to be run, the compiler driver first tries the
13626 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
13627 is not specified, the driver tries two standard prefixes,
13628 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
13629 those results in a file name that is found, the unmodified program
13630 name is searched for using the directories specified in your
13631 @env{PATH} environment variable.
13633 The compiler checks to see if the path provided by @option{-B}
13634 refers to a directory, and if necessary it adds a directory
13635 separator character at the end of the path.
13637 @option{-B} prefixes that effectively specify directory names also apply
13638 to libraries in the linker, because the compiler translates these
13639 options into @option{-L} options for the linker. They also apply to
13640 include files in the preprocessor, because the compiler translates these
13641 options into @option{-isystem} options for the preprocessor. In this case,
13642 the compiler appends @samp{include} to the prefix.
13644 The runtime support file @file{libgcc.a} can also be searched for using
13645 the @option{-B} prefix, if needed. If it is not found there, the two
13646 standard prefixes above are tried, and that is all. The file is left
13647 out of the link if it is not found by those means.
13649 Another way to specify a prefix much like the @option{-B} prefix is to use
13650 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
13653 As a special kludge, if the path provided by @option{-B} is
13654 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
13655 9, then it is replaced by @file{[dir/]include}. This is to help
13656 with boot-strapping the compiler.
13658 @item -no-canonical-prefixes
13659 @opindex no-canonical-prefixes
13660 Do not expand any symbolic links, resolve references to @samp{/../}
13661 or @samp{/./}, or make the path absolute when generating a relative
13664 @item --sysroot=@var{dir}
13666 Use @var{dir} as the logical root directory for headers and libraries.
13667 For example, if the compiler normally searches for headers in
13668 @file{/usr/include} and libraries in @file{/usr/lib}, it instead
13669 searches @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
13671 If you use both this option and the @option{-isysroot} option, then
13672 the @option{--sysroot} option applies to libraries, but the
13673 @option{-isysroot} option applies to header files.
13675 The GNU linker (beginning with version 2.16) has the necessary support
13676 for this option. If your linker does not support this option, the
13677 header file aspect of @option{--sysroot} still works, but the
13678 library aspect does not.
13680 @item --no-sysroot-suffix
13681 @opindex no-sysroot-suffix
13682 For some targets, a suffix is added to the root directory specified
13683 with @option{--sysroot}, depending on the other options used, so that
13684 headers may for example be found in
13685 @file{@var{dir}/@var{suffix}/usr/include} instead of
13686 @file{@var{dir}/usr/include}. This option disables the addition of
13691 @node Code Gen Options
13692 @section Options for Code Generation Conventions
13693 @cindex code generation conventions
13694 @cindex options, code generation
13695 @cindex run-time options
13697 These machine-independent options control the interface conventions
13698 used in code generation.
13700 Most of them have both positive and negative forms; the negative form
13701 of @option{-ffoo} is @option{-fno-foo}. In the table below, only
13702 one of the forms is listed---the one that is not the default. You
13703 can figure out the other form by either removing @samp{no-} or adding
13707 @item -fstack-reuse=@var{reuse-level}
13708 @opindex fstack_reuse
13709 This option controls stack space reuse for user declared local/auto variables
13710 and compiler generated temporaries. @var{reuse_level} can be @samp{all},
13711 @samp{named_vars}, or @samp{none}. @samp{all} enables stack reuse for all
13712 local variables and temporaries, @samp{named_vars} enables the reuse only for
13713 user defined local variables with names, and @samp{none} disables stack reuse
13714 completely. The default value is @samp{all}. The option is needed when the
13715 program extends the lifetime of a scoped local variable or a compiler generated
13716 temporary beyond the end point defined by the language. When a lifetime of
13717 a variable ends, and if the variable lives in memory, the optimizing compiler
13718 has the freedom to reuse its stack space with other temporaries or scoped
13719 local variables whose live range does not overlap with it. Legacy code extending
13720 local lifetime is likely to break with the stack reuse optimization.
13739 if (*p == 10) // out of scope use of local1
13750 A(int k) : i(k), j(k) @{ @}
13757 void foo(const A& ar)
13764 foo(A(10)); // temp object's lifetime ends when foo returns
13770 ap->i+= 10; // ap references out of scope temp whose space
13771 // is reused with a. What is the value of ap->i?
13776 The lifetime of a compiler generated temporary is well defined by the C++
13777 standard. When a lifetime of a temporary ends, and if the temporary lives
13778 in memory, the optimizing compiler has the freedom to reuse its stack
13779 space with other temporaries or scoped local variables whose live range
13780 does not overlap with it. However some of the legacy code relies on
13781 the behavior of older compilers in which temporaries' stack space is
13782 not reused, the aggressive stack reuse can lead to runtime errors. This
13783 option is used to control the temporary stack reuse optimization.
13787 This option generates traps for signed overflow on addition, subtraction,
13788 multiplication operations.
13789 The options @option{-ftrapv} and @option{-fwrapv} override each other, so using
13790 @option{-ftrapv} @option{-fwrapv} on the command-line results in
13791 @option{-fwrapv} being effective. Note that only active options override, so
13792 using @option{-ftrapv} @option{-fwrapv} @option{-fno-wrapv} on the command-line
13793 results in @option{-ftrapv} being effective.
13797 This option instructs the compiler to assume that signed arithmetic
13798 overflow of addition, subtraction and multiplication wraps around
13799 using twos-complement representation. This flag enables some optimizations
13800 and disables others.
13801 The options @option{-ftrapv} and @option{-fwrapv} override each other, so using
13802 @option{-ftrapv} @option{-fwrapv} on the command-line results in
13803 @option{-fwrapv} being effective. Note that only active options override, so
13804 using @option{-ftrapv} @option{-fwrapv} @option{-fno-wrapv} on the command-line
13805 results in @option{-ftrapv} being effective.
13807 @item -fwrapv-pointer
13808 @opindex fwrapv-pointer
13809 This option instructs the compiler to assume that pointer arithmetic
13810 overflow on addition and subtraction wraps around using twos-complement
13811 representation. This flag disables some optimizations which assume
13812 pointer overflow is invalid.
13814 @item -fstrict-overflow
13815 @opindex fstrict-overflow
13816 This option implies @option{-fno-wrapv} @option{-fno-wrapv-pointer} and when
13817 negated implies @option{-fwrapv} @option{-fwrapv-pointer}.
13820 @opindex fexceptions
13821 Enable exception handling. Generates extra code needed to propagate
13822 exceptions. For some targets, this implies GCC generates frame
13823 unwind information for all functions, which can produce significant data
13824 size overhead, although it does not affect execution. If you do not
13825 specify this option, GCC enables it by default for languages like
13826 C++ that normally require exception handling, and disables it for
13827 languages like C that do not normally require it. However, you may need
13828 to enable this option when compiling C code that needs to interoperate
13829 properly with exception handlers written in C++. You may also wish to
13830 disable this option if you are compiling older C++ programs that don't
13831 use exception handling.
13833 @item -fnon-call-exceptions
13834 @opindex fnon-call-exceptions
13835 Generate code that allows trapping instructions to throw exceptions.
13836 Note that this requires platform-specific runtime support that does
13837 not exist everywhere. Moreover, it only allows @emph{trapping}
13838 instructions to throw exceptions, i.e.@: memory references or floating-point
13839 instructions. It does not allow exceptions to be thrown from
13840 arbitrary signal handlers such as @code{SIGALRM}.
13842 @item -fdelete-dead-exceptions
13843 @opindex fdelete-dead-exceptions
13844 Consider that instructions that may throw exceptions but don't otherwise
13845 contribute to the execution of the program can be optimized away.
13846 This option is enabled by default for the Ada front end, as permitted by
13847 the Ada language specification.
13848 Optimization passes that cause dead exceptions to be removed are enabled independently at different optimization levels.
13850 @item -funwind-tables
13851 @opindex funwind-tables
13852 Similar to @option{-fexceptions}, except that it just generates any needed
13853 static data, but does not affect the generated code in any other way.
13854 You normally do not need to enable this option; instead, a language processor
13855 that needs this handling enables it on your behalf.
13857 @item -fasynchronous-unwind-tables
13858 @opindex fasynchronous-unwind-tables
13859 Generate unwind table in DWARF format, if supported by target machine. The
13860 table is exact at each instruction boundary, so it can be used for stack
13861 unwinding from asynchronous events (such as debugger or garbage collector).
13863 @item -fno-gnu-unique
13864 @opindex fno-gnu-unique
13865 @opindex fgnu-unique
13866 On systems with recent GNU assembler and C library, the C++ compiler
13867 uses the @code{STB_GNU_UNIQUE} binding to make sure that definitions
13868 of template static data members and static local variables in inline
13869 functions are unique even in the presence of @code{RTLD_LOCAL}; this
13870 is necessary to avoid problems with a library used by two different
13871 @code{RTLD_LOCAL} plugins depending on a definition in one of them and
13872 therefore disagreeing with the other one about the binding of the
13873 symbol. But this causes @code{dlclose} to be ignored for affected
13874 DSOs; if your program relies on reinitialization of a DSO via
13875 @code{dlclose} and @code{dlopen}, you can use
13876 @option{-fno-gnu-unique}.
13878 @item -fpcc-struct-return
13879 @opindex fpcc-struct-return
13880 Return ``short'' @code{struct} and @code{union} values in memory like
13881 longer ones, rather than in registers. This convention is less
13882 efficient, but it has the advantage of allowing intercallability between
13883 GCC-compiled files and files compiled with other compilers, particularly
13884 the Portable C Compiler (pcc).
13886 The precise convention for returning structures in memory depends
13887 on the target configuration macros.
13889 Short structures and unions are those whose size and alignment match
13890 that of some integer type.
13892 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13893 switch is not binary compatible with code compiled with the
13894 @option{-freg-struct-return} switch.
13895 Use it to conform to a non-default application binary interface.
13897 @item -freg-struct-return
13898 @opindex freg-struct-return
13899 Return @code{struct} and @code{union} values in registers when possible.
13900 This is more efficient for small structures than
13901 @option{-fpcc-struct-return}.
13903 If you specify neither @option{-fpcc-struct-return} nor
13904 @option{-freg-struct-return}, GCC defaults to whichever convention is
13905 standard for the target. If there is no standard convention, GCC
13906 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13907 the principal compiler. In those cases, we can choose the standard, and
13908 we chose the more efficient register return alternative.
13910 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13911 switch is not binary compatible with code compiled with the
13912 @option{-fpcc-struct-return} switch.
13913 Use it to conform to a non-default application binary interface.
13915 @item -fshort-enums
13916 @opindex fshort-enums
13917 Allocate to an @code{enum} type only as many bytes as it needs for the
13918 declared range of possible values. Specifically, the @code{enum} type
13919 is equivalent to the smallest integer type that has enough room.
13921 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13922 code that is not binary compatible with code generated without that switch.
13923 Use it to conform to a non-default application binary interface.
13925 @item -fshort-wchar
13926 @opindex fshort-wchar
13927 Override the underlying type for @code{wchar_t} to be @code{short
13928 unsigned int} instead of the default for the target. This option is
13929 useful for building programs to run under WINE@.
13931 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13932 code that is not binary compatible with code generated without that switch.
13933 Use it to conform to a non-default application binary interface.
13936 @opindex fno-common
13938 @cindex tentative definitions
13939 In C code, this option controls the placement of global variables
13940 defined without an initializer, known as @dfn{tentative definitions}
13941 in the C standard. Tentative definitions are distinct from declarations
13942 of a variable with the @code{extern} keyword, which do not allocate storage.
13944 Unix C compilers have traditionally allocated storage for
13945 uninitialized global variables in a common block. This allows the
13946 linker to resolve all tentative definitions of the same variable
13947 in different compilation units to the same object, or to a non-tentative
13949 This is the behavior specified by @option{-fcommon}, and is the default for
13950 GCC on most targets.
13951 On the other hand, this behavior is not required by ISO
13952 C, and on some targets may carry a speed or code size penalty on
13953 variable references.
13955 The @option{-fno-common} option specifies that the compiler should instead
13956 place uninitialized global variables in the BSS section of the object file.
13957 This inhibits the merging of tentative definitions by the linker so
13958 you get a multiple-definition error if the same
13959 variable is defined in more than one compilation unit.
13960 Compiling with @option{-fno-common} is useful on targets for which
13961 it provides better performance, or if you wish to verify that the
13962 program will work on other systems that always treat uninitialized
13963 variable definitions this way.
13968 Ignore the @code{#ident} directive.
13970 @item -finhibit-size-directive
13971 @opindex finhibit-size-directive
13972 Don't output a @code{.size} assembler directive, or anything else that
13973 would cause trouble if the function is split in the middle, and the
13974 two halves are placed at locations far apart in memory. This option is
13975 used when compiling @file{crtstuff.c}; you should not need to use it
13978 @item -fverbose-asm
13979 @opindex fverbose-asm
13980 Put extra commentary information in the generated assembly code to
13981 make it more readable. This option is generally only of use to those
13982 who actually need to read the generated assembly code (perhaps while
13983 debugging the compiler itself).
13985 @option{-fno-verbose-asm}, the default, causes the
13986 extra information to be omitted and is useful when comparing two assembler
13989 The added comments include:
13994 information on the compiler version and command-line options,
13997 the source code lines associated with the assembly instructions,
13998 in the form FILENAME:LINENUMBER:CONTENT OF LINE,
14001 hints on which high-level expressions correspond to
14002 the various assembly instruction operands.
14006 For example, given this C source file:
14014 for (i = 0; i < n; i++)
14021 compiling to (x86_64) assembly via @option{-S} and emitting the result
14022 direct to stdout via @option{-o} @option{-}
14025 gcc -S test.c -fverbose-asm -Os -o -
14028 gives output similar to this:
14032 # GNU C11 (GCC) version 7.0.0 20160809 (experimental) (x86_64-pc-linux-gnu)
14039 .type test, @@function
14043 # test.c:4: int total = 0;
14044 xorl %eax, %eax # <retval>
14045 # test.c:6: for (i = 0; i < n; i++)
14046 xorl %edx, %edx # i
14048 # test.c:6: for (i = 0; i < n; i++)
14049 cmpl %edi, %edx # n, i
14051 # test.c:7: total += i * i;
14052 movl %edx, %ecx # i, tmp92
14053 imull %edx, %ecx # i, tmp92
14054 # test.c:6: for (i = 0; i < n; i++)
14056 # test.c:7: total += i * i;
14057 addl %ecx, %eax # tmp92, <retval>
14065 .ident "GCC: (GNU) 7.0.0 20160809 (experimental)"
14066 .section .note.GNU-stack,"",@@progbits
14069 The comments are intended for humans rather than machines and hence the
14070 precise format of the comments is subject to change.
14072 @item -frecord-gcc-switches
14073 @opindex frecord-gcc-switches
14074 This switch causes the command line used to invoke the
14075 compiler to be recorded into the object file that is being created.
14076 This switch is only implemented on some targets and the exact format
14077 of the recording is target and binary file format dependent, but it
14078 usually takes the form of a section containing ASCII text. This
14079 switch is related to the @option{-fverbose-asm} switch, but that
14080 switch only records information in the assembler output file as
14081 comments, so it never reaches the object file.
14082 See also @option{-grecord-gcc-switches} for another
14083 way of storing compiler options into the object file.
14087 @cindex global offset table
14089 Generate position-independent code (PIC) suitable for use in a shared
14090 library, if supported for the target machine. Such code accesses all
14091 constant addresses through a global offset table (GOT)@. The dynamic
14092 loader resolves the GOT entries when the program starts (the dynamic
14093 loader is not part of GCC; it is part of the operating system). If
14094 the GOT size for the linked executable exceeds a machine-specific
14095 maximum size, you get an error message from the linker indicating that
14096 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
14097 instead. (These maximums are 8k on the SPARC, 28k on AArch64 and 32k
14098 on the m68k and RS/6000. The x86 has no such limit.)
14100 Position-independent code requires special support, and therefore works
14101 only on certain machines. For the x86, GCC supports PIC for System V
14102 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
14103 position-independent.
14105 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14110 If supported for the target machine, emit position-independent code,
14111 suitable for dynamic linking and avoiding any limit on the size of the
14112 global offset table. This option makes a difference on AArch64, m68k,
14113 PowerPC and SPARC@.
14115 Position-independent code requires special support, and therefore works
14116 only on certain machines.
14118 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14125 These options are similar to @option{-fpic} and @option{-fPIC}, but the
14126 generated position-independent code can be only linked into executables.
14127 Usually these options are used to compile code that will be linked using
14128 the @option{-pie} GCC option.
14130 @option{-fpie} and @option{-fPIE} both define the macros
14131 @code{__pie__} and @code{__PIE__}. The macros have the value 1
14132 for @option{-fpie} and 2 for @option{-fPIE}.
14137 Do not use the PLT for external function calls in position-independent code.
14138 Instead, load the callee address at call sites from the GOT and branch to it.
14139 This leads to more efficient code by eliminating PLT stubs and exposing
14140 GOT loads to optimizations. On architectures such as 32-bit x86 where
14141 PLT stubs expect the GOT pointer in a specific register, this gives more
14142 register allocation freedom to the compiler.
14143 Lazy binding requires use of the PLT;
14144 with @option{-fno-plt} all external symbols are resolved at load time.
14146 Alternatively, the function attribute @code{noplt} can be used to avoid calls
14147 through the PLT for specific external functions.
14149 In position-dependent code, a few targets also convert calls to
14150 functions that are marked to not use the PLT to use the GOT instead.
14152 @item -fno-jump-tables
14153 @opindex fno-jump-tables
14154 @opindex fjump-tables
14155 Do not use jump tables for switch statements even where it would be
14156 more efficient than other code generation strategies. This option is
14157 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14158 building code that forms part of a dynamic linker and cannot
14159 reference the address of a jump table. On some targets, jump tables
14160 do not require a GOT and this option is not needed.
14162 @item -ffixed-@var{reg}
14164 Treat the register named @var{reg} as a fixed register; generated code
14165 should never refer to it (except perhaps as a stack pointer, frame
14166 pointer or in some other fixed role).
14168 @var{reg} must be the name of a register. The register names accepted
14169 are machine-specific and are defined in the @code{REGISTER_NAMES}
14170 macro in the machine description macro file.
14172 This flag does not have a negative form, because it specifies a
14175 @item -fcall-used-@var{reg}
14176 @opindex fcall-used
14177 Treat the register named @var{reg} as an allocable register that is
14178 clobbered by function calls. It may be allocated for temporaries or
14179 variables that do not live across a call. Functions compiled this way
14180 do not save and restore the register @var{reg}.
14182 It is an error to use this flag with the frame pointer or stack pointer.
14183 Use of this flag for other registers that have fixed pervasive roles in
14184 the machine's execution model produces disastrous results.
14186 This flag does not have a negative form, because it specifies a
14189 @item -fcall-saved-@var{reg}
14190 @opindex fcall-saved
14191 Treat the register named @var{reg} as an allocable register saved by
14192 functions. It may be allocated even for temporaries or variables that
14193 live across a call. Functions compiled this way save and restore
14194 the register @var{reg} if they use it.
14196 It is an error to use this flag with the frame pointer or stack pointer.
14197 Use of this flag for other registers that have fixed pervasive roles in
14198 the machine's execution model produces disastrous results.
14200 A different sort of disaster results from the use of this flag for
14201 a register in which function values may be returned.
14203 This flag does not have a negative form, because it specifies a
14206 @item -fpack-struct[=@var{n}]
14207 @opindex fpack-struct
14208 Without a value specified, pack all structure members together without
14209 holes. When a value is specified (which must be a small power of two), pack
14210 structure members according to this value, representing the maximum
14211 alignment (that is, objects with default alignment requirements larger than
14212 this are output potentially unaligned at the next fitting location.
14214 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
14215 code that is not binary compatible with code generated without that switch.
14216 Additionally, it makes the code suboptimal.
14217 Use it to conform to a non-default application binary interface.
14219 @item -fleading-underscore
14220 @opindex fleading-underscore
14221 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
14222 change the way C symbols are represented in the object file. One use
14223 is to help link with legacy assembly code.
14225 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
14226 generate code that is not binary compatible with code generated without that
14227 switch. Use it to conform to a non-default application binary interface.
14228 Not all targets provide complete support for this switch.
14230 @item -ftls-model=@var{model}
14231 @opindex ftls-model
14232 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
14233 The @var{model} argument should be one of @samp{global-dynamic},
14234 @samp{local-dynamic}, @samp{initial-exec} or @samp{local-exec}.
14235 Note that the choice is subject to optimization: the compiler may use
14236 a more efficient model for symbols not visible outside of the translation
14237 unit, or if @option{-fpic} is not given on the command line.
14239 The default without @option{-fpic} is @samp{initial-exec}; with
14240 @option{-fpic} the default is @samp{global-dynamic}.
14242 @item -ftrampolines
14243 @opindex ftrampolines
14244 For targets that normally need trampolines for nested functions, always
14245 generate them instead of using descriptors. Otherwise, for targets that
14246 do not need them, like for example HP-PA or IA-64, do nothing.
14248 A trampoline is a small piece of code that is created at run time on the
14249 stack when the address of a nested function is taken, and is used to call
14250 the nested function indirectly. Therefore, it requires the stack to be
14251 made executable in order for the program to work properly.
14253 @option{-fno-trampolines} is enabled by default on a language by language
14254 basis to let the compiler avoid generating them, if it computes that this
14255 is safe, and replace them with descriptors. Descriptors are made up of data
14256 only, but the generated code must be prepared to deal with them. As of this
14257 writing, @option{-fno-trampolines} is enabled by default only for Ada.
14259 Moreover, code compiled with @option{-ftrampolines} and code compiled with
14260 @option{-fno-trampolines} are not binary compatible if nested functions are
14261 present. This option must therefore be used on a program-wide basis and be
14262 manipulated with extreme care.
14264 @item -fvisibility=@r{[}default@r{|}internal@r{|}hidden@r{|}protected@r{]}
14265 @opindex fvisibility
14266 Set the default ELF image symbol visibility to the specified option---all
14267 symbols are marked with this unless overridden within the code.
14268 Using this feature can very substantially improve linking and
14269 load times of shared object libraries, produce more optimized
14270 code, provide near-perfect API export and prevent symbol clashes.
14271 It is @strong{strongly} recommended that you use this in any shared objects
14274 Despite the nomenclature, @samp{default} always means public; i.e.,
14275 available to be linked against from outside the shared object.
14276 @samp{protected} and @samp{internal} are pretty useless in real-world
14277 usage so the only other commonly used option is @samp{hidden}.
14278 The default if @option{-fvisibility} isn't specified is
14279 @samp{default}, i.e., make every symbol public.
14281 A good explanation of the benefits offered by ensuring ELF
14282 symbols have the correct visibility is given by ``How To Write
14283 Shared Libraries'' by Ulrich Drepper (which can be found at
14284 @w{@uref{https://www.akkadia.org/drepper/}})---however a superior
14285 solution made possible by this option to marking things hidden when
14286 the default is public is to make the default hidden and mark things
14287 public. This is the norm with DLLs on Windows and with @option{-fvisibility=hidden}
14288 and @code{__attribute__ ((visibility("default")))} instead of
14289 @code{__declspec(dllexport)} you get almost identical semantics with
14290 identical syntax. This is a great boon to those working with
14291 cross-platform projects.
14293 For those adding visibility support to existing code, you may find
14294 @code{#pragma GCC visibility} of use. This works by you enclosing
14295 the declarations you wish to set visibility for with (for example)
14296 @code{#pragma GCC visibility push(hidden)} and
14297 @code{#pragma GCC visibility pop}.
14298 Bear in mind that symbol visibility should be viewed @strong{as
14299 part of the API interface contract} and thus all new code should
14300 always specify visibility when it is not the default; i.e., declarations
14301 only for use within the local DSO should @strong{always} be marked explicitly
14302 as hidden as so to avoid PLT indirection overheads---making this
14303 abundantly clear also aids readability and self-documentation of the code.
14304 Note that due to ISO C++ specification requirements, @code{operator new} and
14305 @code{operator delete} must always be of default visibility.
14307 Be aware that headers from outside your project, in particular system
14308 headers and headers from any other library you use, may not be
14309 expecting to be compiled with visibility other than the default. You
14310 may need to explicitly say @code{#pragma GCC visibility push(default)}
14311 before including any such headers.
14313 @code{extern} declarations are not affected by @option{-fvisibility}, so
14314 a lot of code can be recompiled with @option{-fvisibility=hidden} with
14315 no modifications. However, this means that calls to @code{extern}
14316 functions with no explicit visibility use the PLT, so it is more
14317 effective to use @code{__attribute ((visibility))} and/or
14318 @code{#pragma GCC visibility} to tell the compiler which @code{extern}
14319 declarations should be treated as hidden.
14321 Note that @option{-fvisibility} does affect C++ vague linkage
14322 entities. This means that, for instance, an exception class that is
14323 be thrown between DSOs must be explicitly marked with default
14324 visibility so that the @samp{type_info} nodes are unified between
14327 An overview of these techniques, their benefits and how to use them
14328 is at @uref{http://gcc.gnu.org/@/wiki/@/Visibility}.
14330 @item -fstrict-volatile-bitfields
14331 @opindex fstrict-volatile-bitfields
14332 This option should be used if accesses to volatile bit-fields (or other
14333 structure fields, although the compiler usually honors those types
14334 anyway) should use a single access of the width of the
14335 field's type, aligned to a natural alignment if possible. For
14336 example, targets with memory-mapped peripheral registers might require
14337 all such accesses to be 16 bits wide; with this flag you can
14338 declare all peripheral bit-fields as @code{unsigned short} (assuming short
14339 is 16 bits on these targets) to force GCC to use 16-bit accesses
14340 instead of, perhaps, a more efficient 32-bit access.
14342 If this option is disabled, the compiler uses the most efficient
14343 instruction. In the previous example, that might be a 32-bit load
14344 instruction, even though that accesses bytes that do not contain
14345 any portion of the bit-field, or memory-mapped registers unrelated to
14346 the one being updated.
14348 In some cases, such as when the @code{packed} attribute is applied to a
14349 structure field, it may not be possible to access the field with a single
14350 read or write that is correctly aligned for the target machine. In this
14351 case GCC falls back to generating multiple accesses rather than code that
14352 will fault or truncate the result at run time.
14354 Note: Due to restrictions of the C/C++11 memory model, write accesses are
14355 not allowed to touch non bit-field members. It is therefore recommended
14356 to define all bits of the field's type as bit-field members.
14358 The default value of this option is determined by the application binary
14359 interface for the target processor.
14361 @item -fsync-libcalls
14362 @opindex fsync-libcalls
14363 This option controls whether any out-of-line instance of the @code{__sync}
14364 family of functions may be used to implement the C++11 @code{__atomic}
14365 family of functions.
14367 The default value of this option is enabled, thus the only useful form
14368 of the option is @option{-fno-sync-libcalls}. This option is used in
14369 the implementation of the @file{libatomic} runtime library.
14373 @node Developer Options
14374 @section GCC Developer Options
14375 @cindex developer options
14376 @cindex debugging GCC
14377 @cindex debug dump options
14378 @cindex dump options
14379 @cindex compilation statistics
14381 This section describes command-line options that are primarily of
14382 interest to GCC developers, including options to support compiler
14383 testing and investigation of compiler bugs and compile-time
14384 performance problems. This includes options that produce debug dumps
14385 at various points in the compilation; that print statistics such as
14386 memory use and execution time; and that print information about GCC's
14387 configuration, such as where it searches for libraries. You should
14388 rarely need to use any of these options for ordinary compilation and
14391 Many developer options that cause GCC to dump output to a file take an
14392 optional @samp{=@var{filename}} suffix. You can specify @samp{stdout}
14393 or @samp{-} to dump to standard output, and @samp{stderr} for standard
14396 If @samp{=@var{filename}} is omitted, a default dump file name is
14397 constructed by concatenating the base dump file name, a pass number,
14398 phase letter, and pass name. The base dump file name is the name of
14399 output file produced by the compiler if explicitly specified and not
14400 an executable; otherwise it is the source file name.
14401 The pass number is determined by the order passes are registered with
14402 the compiler's pass manager.
14403 This is generally the same as the order of execution, but passes
14404 registered by plugins, target-specific passes, or passes that are
14405 otherwise registered late are numbered higher than the pass named
14406 @samp{final}, even if they are executed earlier. The phase letter is
14407 one of @samp{i} (inter-procedural analysis), @samp{l}
14408 (language-specific), @samp{r} (RTL), or @samp{t} (tree).
14409 The files are created in the directory of the output file.
14413 @item -d@var{letters}
14414 @itemx -fdump-rtl-@var{pass}
14415 @itemx -fdump-rtl-@var{pass}=@var{filename}
14417 @opindex fdump-rtl-@var{pass}
14418 Says to make debugging dumps during compilation at times specified by
14419 @var{letters}. This is used for debugging the RTL-based passes of the
14422 Some @option{-d@var{letters}} switches have different meaning when
14423 @option{-E} is used for preprocessing. @xref{Preprocessor Options},
14424 for information about preprocessor-specific dump options.
14426 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
14427 @option{-d} option @var{letters}. Here are the possible
14428 letters for use in @var{pass} and @var{letters}, and their meanings:
14432 @item -fdump-rtl-alignments
14433 @opindex fdump-rtl-alignments
14434 Dump after branch alignments have been computed.
14436 @item -fdump-rtl-asmcons
14437 @opindex fdump-rtl-asmcons
14438 Dump after fixing rtl statements that have unsatisfied in/out constraints.
14440 @item -fdump-rtl-auto_inc_dec
14441 @opindex fdump-rtl-auto_inc_dec
14442 Dump after auto-inc-dec discovery. This pass is only run on
14443 architectures that have auto inc or auto dec instructions.
14445 @item -fdump-rtl-barriers
14446 @opindex fdump-rtl-barriers
14447 Dump after cleaning up the barrier instructions.
14449 @item -fdump-rtl-bbpart
14450 @opindex fdump-rtl-bbpart
14451 Dump after partitioning hot and cold basic blocks.
14453 @item -fdump-rtl-bbro
14454 @opindex fdump-rtl-bbro
14455 Dump after block reordering.
14457 @item -fdump-rtl-btl1
14458 @itemx -fdump-rtl-btl2
14459 @opindex fdump-rtl-btl2
14460 @opindex fdump-rtl-btl2
14461 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
14462 after the two branch
14463 target load optimization passes.
14465 @item -fdump-rtl-bypass
14466 @opindex fdump-rtl-bypass
14467 Dump after jump bypassing and control flow optimizations.
14469 @item -fdump-rtl-combine
14470 @opindex fdump-rtl-combine
14471 Dump after the RTL instruction combination pass.
14473 @item -fdump-rtl-compgotos
14474 @opindex fdump-rtl-compgotos
14475 Dump after duplicating the computed gotos.
14477 @item -fdump-rtl-ce1
14478 @itemx -fdump-rtl-ce2
14479 @itemx -fdump-rtl-ce3
14480 @opindex fdump-rtl-ce1
14481 @opindex fdump-rtl-ce2
14482 @opindex fdump-rtl-ce3
14483 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
14484 @option{-fdump-rtl-ce3} enable dumping after the three
14485 if conversion passes.
14487 @item -fdump-rtl-cprop_hardreg
14488 @opindex fdump-rtl-cprop_hardreg
14489 Dump after hard register copy propagation.
14491 @item -fdump-rtl-csa
14492 @opindex fdump-rtl-csa
14493 Dump after combining stack adjustments.
14495 @item -fdump-rtl-cse1
14496 @itemx -fdump-rtl-cse2
14497 @opindex fdump-rtl-cse1
14498 @opindex fdump-rtl-cse2
14499 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
14500 the two common subexpression elimination passes.
14502 @item -fdump-rtl-dce
14503 @opindex fdump-rtl-dce
14504 Dump after the standalone dead code elimination passes.
14506 @item -fdump-rtl-dbr
14507 @opindex fdump-rtl-dbr
14508 Dump after delayed branch scheduling.
14510 @item -fdump-rtl-dce1
14511 @itemx -fdump-rtl-dce2
14512 @opindex fdump-rtl-dce1
14513 @opindex fdump-rtl-dce2
14514 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
14515 the two dead store elimination passes.
14517 @item -fdump-rtl-eh
14518 @opindex fdump-rtl-eh
14519 Dump after finalization of EH handling code.
14521 @item -fdump-rtl-eh_ranges
14522 @opindex fdump-rtl-eh_ranges
14523 Dump after conversion of EH handling range regions.
14525 @item -fdump-rtl-expand
14526 @opindex fdump-rtl-expand
14527 Dump after RTL generation.
14529 @item -fdump-rtl-fwprop1
14530 @itemx -fdump-rtl-fwprop2
14531 @opindex fdump-rtl-fwprop1
14532 @opindex fdump-rtl-fwprop2
14533 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
14534 dumping after the two forward propagation passes.
14536 @item -fdump-rtl-gcse1
14537 @itemx -fdump-rtl-gcse2
14538 @opindex fdump-rtl-gcse1
14539 @opindex fdump-rtl-gcse2
14540 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
14541 after global common subexpression elimination.
14543 @item -fdump-rtl-init-regs
14544 @opindex fdump-rtl-init-regs
14545 Dump after the initialization of the registers.
14547 @item -fdump-rtl-initvals
14548 @opindex fdump-rtl-initvals
14549 Dump after the computation of the initial value sets.
14551 @item -fdump-rtl-into_cfglayout
14552 @opindex fdump-rtl-into_cfglayout
14553 Dump after converting to cfglayout mode.
14555 @item -fdump-rtl-ira
14556 @opindex fdump-rtl-ira
14557 Dump after iterated register allocation.
14559 @item -fdump-rtl-jump
14560 @opindex fdump-rtl-jump
14561 Dump after the second jump optimization.
14563 @item -fdump-rtl-loop2
14564 @opindex fdump-rtl-loop2
14565 @option{-fdump-rtl-loop2} enables dumping after the rtl
14566 loop optimization passes.
14568 @item -fdump-rtl-mach
14569 @opindex fdump-rtl-mach
14570 Dump after performing the machine dependent reorganization pass, if that
14573 @item -fdump-rtl-mode_sw
14574 @opindex fdump-rtl-mode_sw
14575 Dump after removing redundant mode switches.
14577 @item -fdump-rtl-rnreg
14578 @opindex fdump-rtl-rnreg
14579 Dump after register renumbering.
14581 @item -fdump-rtl-outof_cfglayout
14582 @opindex fdump-rtl-outof_cfglayout
14583 Dump after converting from cfglayout mode.
14585 @item -fdump-rtl-peephole2
14586 @opindex fdump-rtl-peephole2
14587 Dump after the peephole pass.
14589 @item -fdump-rtl-postreload
14590 @opindex fdump-rtl-postreload
14591 Dump after post-reload optimizations.
14593 @item -fdump-rtl-pro_and_epilogue
14594 @opindex fdump-rtl-pro_and_epilogue
14595 Dump after generating the function prologues and epilogues.
14597 @item -fdump-rtl-sched1
14598 @itemx -fdump-rtl-sched2
14599 @opindex fdump-rtl-sched1
14600 @opindex fdump-rtl-sched2
14601 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
14602 after the basic block scheduling passes.
14604 @item -fdump-rtl-ree
14605 @opindex fdump-rtl-ree
14606 Dump after sign/zero extension elimination.
14608 @item -fdump-rtl-seqabstr
14609 @opindex fdump-rtl-seqabstr
14610 Dump after common sequence discovery.
14612 @item -fdump-rtl-shorten
14613 @opindex fdump-rtl-shorten
14614 Dump after shortening branches.
14616 @item -fdump-rtl-sibling
14617 @opindex fdump-rtl-sibling
14618 Dump after sibling call optimizations.
14620 @item -fdump-rtl-split1
14621 @itemx -fdump-rtl-split2
14622 @itemx -fdump-rtl-split3
14623 @itemx -fdump-rtl-split4
14624 @itemx -fdump-rtl-split5
14625 @opindex fdump-rtl-split1
14626 @opindex fdump-rtl-split2
14627 @opindex fdump-rtl-split3
14628 @opindex fdump-rtl-split4
14629 @opindex fdump-rtl-split5
14630 These options enable dumping after five rounds of
14631 instruction splitting.
14633 @item -fdump-rtl-sms
14634 @opindex fdump-rtl-sms
14635 Dump after modulo scheduling. This pass is only run on some
14638 @item -fdump-rtl-stack
14639 @opindex fdump-rtl-stack
14640 Dump after conversion from GCC's ``flat register file'' registers to the
14641 x87's stack-like registers. This pass is only run on x86 variants.
14643 @item -fdump-rtl-subreg1
14644 @itemx -fdump-rtl-subreg2
14645 @opindex fdump-rtl-subreg1
14646 @opindex fdump-rtl-subreg2
14647 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
14648 the two subreg expansion passes.
14650 @item -fdump-rtl-unshare
14651 @opindex fdump-rtl-unshare
14652 Dump after all rtl has been unshared.
14654 @item -fdump-rtl-vartrack
14655 @opindex fdump-rtl-vartrack
14656 Dump after variable tracking.
14658 @item -fdump-rtl-vregs
14659 @opindex fdump-rtl-vregs
14660 Dump after converting virtual registers to hard registers.
14662 @item -fdump-rtl-web
14663 @opindex fdump-rtl-web
14664 Dump after live range splitting.
14666 @item -fdump-rtl-regclass
14667 @itemx -fdump-rtl-subregs_of_mode_init
14668 @itemx -fdump-rtl-subregs_of_mode_finish
14669 @itemx -fdump-rtl-dfinit
14670 @itemx -fdump-rtl-dfinish
14671 @opindex fdump-rtl-regclass
14672 @opindex fdump-rtl-subregs_of_mode_init
14673 @opindex fdump-rtl-subregs_of_mode_finish
14674 @opindex fdump-rtl-dfinit
14675 @opindex fdump-rtl-dfinish
14676 These dumps are defined but always produce empty files.
14679 @itemx -fdump-rtl-all
14681 @opindex fdump-rtl-all
14682 Produce all the dumps listed above.
14686 Annotate the assembler output with miscellaneous debugging information.
14690 Dump all macro definitions, at the end of preprocessing, in addition to
14695 Produce a core dump whenever an error occurs.
14699 Annotate the assembler output with a comment indicating which
14700 pattern and alternative is used. The length and cost of each instruction are
14705 Dump the RTL in the assembler output as a comment before each instruction.
14706 Also turns on @option{-dp} annotation.
14710 Just generate RTL for a function instead of compiling it. Usually used
14711 with @option{-fdump-rtl-expand}.
14715 @opindex fdump-debug
14716 Dump debugging information generated during the debug
14719 @item -fdump-earlydebug
14720 @opindex fdump-earlydebug
14721 Dump debugging information generated during the early debug
14724 @item -fdump-noaddr
14725 @opindex fdump-noaddr
14726 When doing debugging dumps, suppress address output. This makes it more
14727 feasible to use diff on debugging dumps for compiler invocations with
14728 different compiler binaries and/or different
14729 text / bss / data / heap / stack / dso start locations.
14732 @opindex freport-bug
14733 Collect and dump debug information into a temporary file if an
14734 internal compiler error (ICE) occurs.
14736 @item -fdump-unnumbered
14737 @opindex fdump-unnumbered
14738 When doing debugging dumps, suppress instruction numbers and address output.
14739 This makes it more feasible to use diff on debugging dumps for compiler
14740 invocations with different options, in particular with and without
14743 @item -fdump-unnumbered-links
14744 @opindex fdump-unnumbered-links
14745 When doing debugging dumps (see @option{-d} option above), suppress
14746 instruction numbers for the links to the previous and next instructions
14749 @item -fdump-ipa-@var{switch}
14750 @itemx -fdump-ipa-@var{switch}-@var{options}
14752 Control the dumping at various stages of inter-procedural analysis
14753 language tree to a file. The file name is generated by appending a
14754 switch specific suffix to the source file name, and the file is created
14755 in the same directory as the output file. The following dumps are
14760 Enables all inter-procedural analysis dumps.
14763 Dumps information about call-graph optimization, unused function removal,
14764 and inlining decisions.
14767 Dump after function inlining.
14771 Additionally, the options @option{-optimized}, @option{-missed},
14772 @option{-note}, and @option{-all} can be provided, with the same meaning
14773 as for @option{-fopt-info}, defaulting to @option{-optimized}.
14775 For example, @option{-fdump-ipa-inline-optimized-missed} will emit
14776 information on callsites that were inlined, along with callsites
14777 that were not inlined.
14779 By default, the dump will contain messages about successful
14780 optimizations (equivalent to @option{-optimized}) together with
14781 low-level details about the analysis.
14783 @item -fdump-lang-all
14784 @itemx -fdump-lang-@var{switch}
14785 @itemx -fdump-lang-@var{switch}-@var{options}
14786 @itemx -fdump-lang-@var{switch}-@var{options}=@var{filename}
14787 @opindex fdump-lang-all
14788 @opindex fdump-lang
14789 Control the dumping of language-specific information. The @var{options}
14790 and @var{filename} portions behave as described in the
14791 @option{-fdump-tree} option. The following @var{switch} values are
14797 Enable all language-specific dumps.
14800 Dump class hierarchy information. Virtual table information is emitted
14801 unless '@option{slim}' is specified. This option is applicable to C++ only.
14804 Dump the raw internal tree data. This option is applicable to C++ only.
14808 @item -fdump-passes
14809 @opindex fdump-passes
14810 Print on @file{stderr} the list of optimization passes that are turned
14811 on and off by the current command-line options.
14813 @item -fdump-statistics-@var{option}
14814 @opindex fdump-statistics
14815 Enable and control dumping of pass statistics in a separate file. The
14816 file name is generated by appending a suffix ending in
14817 @samp{.statistics} to the source file name, and the file is created in
14818 the same directory as the output file. If the @samp{-@var{option}}
14819 form is used, @samp{-stats} causes counters to be summed over the
14820 whole compilation unit while @samp{-details} dumps every event as
14821 the passes generate them. The default with no option is to sum
14822 counters for each function compiled.
14824 @item -fdump-tree-all
14825 @itemx -fdump-tree-@var{switch}
14826 @itemx -fdump-tree-@var{switch}-@var{options}
14827 @itemx -fdump-tree-@var{switch}-@var{options}=@var{filename}
14828 @opindex fdump-tree-all
14829 @opindex fdump-tree
14830 Control the dumping at various stages of processing the intermediate
14831 language tree to a file. If the @samp{-@var{options}}
14832 form is used, @var{options} is a list of @samp{-} separated options
14833 which control the details of the dump. Not all options are applicable
14834 to all dumps; those that are not meaningful are ignored. The
14835 following options are available
14839 Print the address of each node. Usually this is not meaningful as it
14840 changes according to the environment and source file. Its primary use
14841 is for tying up a dump file with a debug environment.
14843 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
14844 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
14845 use working backward from mangled names in the assembly file.
14847 When dumping front-end intermediate representations, inhibit dumping
14848 of members of a scope or body of a function merely because that scope
14849 has been reached. Only dump such items when they are directly reachable
14850 by some other path.
14852 When dumping pretty-printed trees, this option inhibits dumping the
14853 bodies of control structures.
14855 When dumping RTL, print the RTL in slim (condensed) form instead of
14856 the default LISP-like representation.
14858 Print a raw representation of the tree. By default, trees are
14859 pretty-printed into a C-like representation.
14861 Enable more detailed dumps (not honored by every dump option). Also
14862 include information from the optimization passes.
14864 Enable dumping various statistics about the pass (not honored by every dump
14867 Enable showing basic block boundaries (disabled in raw dumps).
14869 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
14870 dump a representation of the control flow graph suitable for viewing with
14871 GraphViz to @file{@var{file}.@var{passid}.@var{pass}.dot}. Each function in
14872 the file is pretty-printed as a subgraph, so that GraphViz can render them
14873 all in a single plot.
14875 This option currently only works for RTL dumps, and the RTL is always
14876 dumped in slim form.
14878 Enable showing virtual operands for every statement.
14880 Enable showing line numbers for statements.
14882 Enable showing the unique ID (@code{DECL_UID}) for each variable.
14884 Enable showing the tree dump for each statement.
14886 Enable showing the EH region number holding each statement.
14888 Enable showing scalar evolution analysis details.
14890 Enable showing optimization information (only available in certain
14893 Enable showing missed optimization information (only available in certain
14896 Enable other detailed optimization information (only available in
14899 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
14900 and @option{lineno}.
14902 Turn on all optimization options, i.e., @option{optimized},
14903 @option{missed}, and @option{note}.
14906 To determine what tree dumps are available or find the dump for a pass
14907 of interest follow the steps below.
14911 Invoke GCC with @option{-fdump-passes} and in the @file{stderr} output
14912 look for a code that corresponds to the pass you are interested in.
14913 For example, the codes @code{tree-evrp}, @code{tree-vrp1}, and
14914 @code{tree-vrp2} correspond to the three Value Range Propagation passes.
14915 The number at the end distinguishes distinct invocations of the same pass.
14917 To enable the creation of the dump file, append the pass code to
14918 the @option{-fdump-} option prefix and invoke GCC with it. For example,
14919 to enable the dump from the Early Value Range Propagation pass, invoke
14920 GCC with the @option{-fdump-tree-evrp} option. Optionally, you may
14921 specify the name of the dump file. If you don't specify one, GCC
14922 creates as described below.
14924 Find the pass dump in a file whose name is composed of three components
14925 separated by a period: the name of the source file GCC was invoked to
14926 compile, a numeric suffix indicating the pass number followed by the
14927 letter @samp{t} for tree passes (and the letter @samp{r} for RTL passes),
14928 and finally the pass code. For example, the Early VRP pass dump might
14929 be in a file named @file{myfile.c.038t.evrp} in the current working
14930 directory. Note that the numeric codes are not stable and may change
14931 from one version of GCC to another.
14935 @itemx -fopt-info-@var{options}
14936 @itemx -fopt-info-@var{options}=@var{filename}
14938 Controls optimization dumps from various optimization passes. If the
14939 @samp{-@var{options}} form is used, @var{options} is a list of
14940 @samp{-} separated option keywords to select the dump details and
14943 The @var{options} can be divided into three groups:
14946 options describing what kinds of messages should be emitted,
14948 options describing the verbosity of the dump, and
14950 options describing which optimizations should be included.
14952 The options from each group can be freely mixed as they are
14953 non-overlapping. However, in case of any conflicts,
14954 the later options override the earlier options on the command
14957 The following options control which kinds of messages should be emitted:
14961 Print information when an optimization is successfully applied. It is
14962 up to a pass to decide which information is relevant. For example, the
14963 vectorizer passes print the source location of loops which are
14964 successfully vectorized.
14966 Print information about missed optimizations. Individual passes
14967 control which information to include in the output.
14969 Print verbose information about optimizations, such as certain
14970 transformations, more detailed messages about decisions etc.
14972 Print detailed optimization information. This includes
14973 @samp{optimized}, @samp{missed}, and @samp{note}.
14976 The following option controls the dump verbosity:
14980 By default, only ``high-level'' messages are emitted. This option enables
14981 additional, more detailed, messages, which are likely to only be of interest
14985 One or more of the following option keywords can be used to describe a
14986 group of optimizations:
14990 Enable dumps from all interprocedural optimizations.
14992 Enable dumps from all loop optimizations.
14994 Enable dumps from all inlining optimizations.
14996 Enable dumps from all OMP (Offloading and Multi Processing) optimizations.
14998 Enable dumps from all vectorization optimizations.
15000 Enable dumps from all optimizations. This is a superset of
15001 the optimization groups listed above.
15004 If @var{options} is
15005 omitted, it defaults to @samp{optimized-optall}, which means to dump messages
15006 about successful optimizations from all the passes, omitting messages
15007 that are treated as ``internals''.
15009 If the @var{filename} is provided, then the dumps from all the
15010 applicable optimizations are concatenated into the @var{filename}.
15011 Otherwise the dump is output onto @file{stderr}. Though multiple
15012 @option{-fopt-info} options are accepted, only one of them can include
15013 a @var{filename}. If other filenames are provided then all but the
15014 first such option are ignored.
15016 Note that the output @var{filename} is overwritten
15017 in case of multiple translation units. If a combined output from
15018 multiple translation units is desired, @file{stderr} should be used
15021 In the following example, the optimization info is output to
15030 gcc -O3 -fopt-info-missed=missed.all
15034 outputs missed optimization report from all the passes into
15035 @file{missed.all}, and this one:
15038 gcc -O2 -ftree-vectorize -fopt-info-vec-missed
15042 prints information about missed optimization opportunities from
15043 vectorization passes on @file{stderr}.
15044 Note that @option{-fopt-info-vec-missed} is equivalent to
15045 @option{-fopt-info-missed-vec}. The order of the optimization group
15046 names and message types listed after @option{-fopt-info} does not matter.
15048 As another example,
15050 gcc -O3 -fopt-info-inline-optimized-missed=inline.txt
15054 outputs information about missed optimizations as well as
15055 optimized locations from all the inlining passes into
15061 gcc -fopt-info-vec-missed=vec.miss -fopt-info-loop-optimized=loop.opt
15065 Here the two output filenames @file{vec.miss} and @file{loop.opt} are
15066 in conflict since only one output file is allowed. In this case, only
15067 the first option takes effect and the subsequent options are
15068 ignored. Thus only @file{vec.miss} is produced which contains
15069 dumps from the vectorizer about missed opportunities.
15071 @item -fsave-optimization-record
15072 @opindex fsave-optimization-record
15073 Write a SRCFILE.opt-record.json.gz file detailing what optimizations
15074 were performed, for those optimizations that support @option{-fopt-info}.
15076 This option is experimental and the format of the data within the
15077 compressed JSON file is subject to change.
15079 It is roughly equivalent to a machine-readable version of
15080 @option{-fopt-info-all}, as a collection of messages with source file,
15081 line number and column number, with the following additional data for
15087 the execution count of the code being optimized, along with metadata about
15088 whether this was from actual profile data, or just an estimate, allowing
15089 consumers to prioritize messages by code hotness,
15092 the function name of the code being optimized, where applicable,
15095 the ``inlining chain'' for the code being optimized, so that when
15096 a function is inlined into several different places (which might
15097 themselves be inlined), the reader can distinguish between the copies,
15100 objects identifying those parts of the message that refer to expressions,
15101 statements or symbol-table nodes, which of these categories they are, and,
15102 when available, their source code location,
15105 the GCC pass that emitted the message, and
15108 the location in GCC's own code from which the message was emitted
15112 Additionally, some messages are logically nested within other
15113 messages, reflecting implementation details of the optimization
15116 @item -fsched-verbose=@var{n}
15117 @opindex fsched-verbose
15118 On targets that use instruction scheduling, this option controls the
15119 amount of debugging output the scheduler prints to the dump files.
15121 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
15122 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
15123 For @var{n} greater than one, it also output basic block probabilities,
15124 detailed ready list information and unit/insn info. For @var{n} greater
15125 than two, it includes RTL at abort point, control-flow and regions info.
15126 And for @var{n} over four, @option{-fsched-verbose} also includes
15131 @item -fenable-@var{kind}-@var{pass}
15132 @itemx -fdisable-@var{kind}-@var{pass}=@var{range-list}
15136 This is a set of options that are used to explicitly disable/enable
15137 optimization passes. These options are intended for use for debugging GCC.
15138 Compiler users should use regular options for enabling/disabling
15143 @item -fdisable-ipa-@var{pass}
15144 Disable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
15145 statically invoked in the compiler multiple times, the pass name should be
15146 appended with a sequential number starting from 1.
15148 @item -fdisable-rtl-@var{pass}
15149 @itemx -fdisable-rtl-@var{pass}=@var{range-list}
15150 Disable RTL pass @var{pass}. @var{pass} is the pass name. If the same pass is
15151 statically invoked in the compiler multiple times, the pass name should be
15152 appended with a sequential number starting from 1. @var{range-list} is a
15153 comma-separated list of function ranges or assembler names. Each range is a number
15154 pair separated by a colon. The range is inclusive in both ends. If the range
15155 is trivial, the number pair can be simplified as a single number. If the
15156 function's call graph node's @var{uid} falls within one of the specified ranges,
15157 the @var{pass} is disabled for that function. The @var{uid} is shown in the
15158 function header of a dump file, and the pass names can be dumped by using
15159 option @option{-fdump-passes}.
15161 @item -fdisable-tree-@var{pass}
15162 @itemx -fdisable-tree-@var{pass}=@var{range-list}
15163 Disable tree pass @var{pass}. See @option{-fdisable-rtl} for the description of
15166 @item -fenable-ipa-@var{pass}
15167 Enable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
15168 statically invoked in the compiler multiple times, the pass name should be
15169 appended with a sequential number starting from 1.
15171 @item -fenable-rtl-@var{pass}
15172 @itemx -fenable-rtl-@var{pass}=@var{range-list}
15173 Enable RTL pass @var{pass}. See @option{-fdisable-rtl} for option argument
15174 description and examples.
15176 @item -fenable-tree-@var{pass}
15177 @itemx -fenable-tree-@var{pass}=@var{range-list}
15178 Enable tree pass @var{pass}. See @option{-fdisable-rtl} for the description
15179 of option arguments.
15183 Here are some examples showing uses of these options.
15187 # disable ccp1 for all functions
15188 -fdisable-tree-ccp1
15189 # disable complete unroll for function whose cgraph node uid is 1
15190 -fenable-tree-cunroll=1
15191 # disable gcse2 for functions at the following ranges [1,1],
15192 # [300,400], and [400,1000]
15193 # disable gcse2 for functions foo and foo2
15194 -fdisable-rtl-gcse2=foo,foo2
15195 # disable early inlining
15196 -fdisable-tree-einline
15197 # disable ipa inlining
15198 -fdisable-ipa-inline
15199 # enable tree full unroll
15200 -fenable-tree-unroll
15205 @itemx -fchecking=@var{n}
15207 @opindex fno-checking
15208 Enable internal consistency checking. The default depends on
15209 the compiler configuration. @option{-fchecking=2} enables further
15210 internal consistency checking that might affect code generation.
15212 @item -frandom-seed=@var{string}
15213 @opindex frandom-seed
15214 This option provides a seed that GCC uses in place of
15215 random numbers in generating certain symbol names
15216 that have to be different in every compiled file. It is also used to
15217 place unique stamps in coverage data files and the object files that
15218 produce them. You can use the @option{-frandom-seed} option to produce
15219 reproducibly identical object files.
15221 The @var{string} can either be a number (decimal, octal or hex) or an
15222 arbitrary string (in which case it's converted to a number by
15225 The @var{string} should be different for every file you compile.
15228 @itemx -save-temps=cwd
15229 @opindex save-temps
15230 Store the usual ``temporary'' intermediate files permanently; place them
15231 in the current directory and name them based on the source file. Thus,
15232 compiling @file{foo.c} with @option{-c -save-temps} produces files
15233 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
15234 preprocessed @file{foo.i} output file even though the compiler now
15235 normally uses an integrated preprocessor.
15237 When used in combination with the @option{-x} command-line option,
15238 @option{-save-temps} is sensible enough to avoid over writing an
15239 input source file with the same extension as an intermediate file.
15240 The corresponding intermediate file may be obtained by renaming the
15241 source file before using @option{-save-temps}.
15243 If you invoke GCC in parallel, compiling several different source
15244 files that share a common base name in different subdirectories or the
15245 same source file compiled for multiple output destinations, it is
15246 likely that the different parallel compilers will interfere with each
15247 other, and overwrite the temporary files. For instance:
15250 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
15251 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
15254 may result in @file{foo.i} and @file{foo.o} being written to
15255 simultaneously by both compilers.
15257 @item -save-temps=obj
15258 @opindex save-temps=obj
15259 Store the usual ``temporary'' intermediate files permanently. If the
15260 @option{-o} option is used, the temporary files are based on the
15261 object file. If the @option{-o} option is not used, the
15262 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
15267 gcc -save-temps=obj -c foo.c
15268 gcc -save-temps=obj -c bar.c -o dir/xbar.o
15269 gcc -save-temps=obj foobar.c -o dir2/yfoobar
15273 creates @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
15274 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
15275 @file{dir2/yfoobar.o}.
15277 @item -time@r{[}=@var{file}@r{]}
15279 Report the CPU time taken by each subprocess in the compilation
15280 sequence. For C source files, this is the compiler proper and assembler
15281 (plus the linker if linking is done).
15283 Without the specification of an output file, the output looks like this:
15290 The first number on each line is the ``user time'', that is time spent
15291 executing the program itself. The second number is ``system time'',
15292 time spent executing operating system routines on behalf of the program.
15293 Both numbers are in seconds.
15295 With the specification of an output file, the output is appended to the
15296 named file, and it looks like this:
15299 0.12 0.01 cc1 @var{options}
15300 0.00 0.01 as @var{options}
15303 The ``user time'' and the ``system time'' are moved before the program
15304 name, and the options passed to the program are displayed, so that one
15305 can later tell what file was being compiled, and with which options.
15307 @item -fdump-final-insns@r{[}=@var{file}@r{]}
15308 @opindex fdump-final-insns
15309 Dump the final internal representation (RTL) to @var{file}. If the
15310 optional argument is omitted (or if @var{file} is @code{.}), the name
15311 of the dump file is determined by appending @code{.gkd} to the
15312 compilation output file name.
15314 @item -fcompare-debug@r{[}=@var{opts}@r{]}
15315 @opindex fcompare-debug
15316 @opindex fno-compare-debug
15317 If no error occurs during compilation, run the compiler a second time,
15318 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
15319 passed to the second compilation. Dump the final internal
15320 representation in both compilations, and print an error if they differ.
15322 If the equal sign is omitted, the default @option{-gtoggle} is used.
15324 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
15325 and nonzero, implicitly enables @option{-fcompare-debug}. If
15326 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
15327 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
15330 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
15331 is equivalent to @option{-fno-compare-debug}, which disables the dumping
15332 of the final representation and the second compilation, preventing even
15333 @env{GCC_COMPARE_DEBUG} from taking effect.
15335 To verify full coverage during @option{-fcompare-debug} testing, set
15336 @env{GCC_COMPARE_DEBUG} to say @option{-fcompare-debug-not-overridden},
15337 which GCC rejects as an invalid option in any actual compilation
15338 (rather than preprocessing, assembly or linking). To get just a
15339 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
15340 not overridden} will do.
15342 @item -fcompare-debug-second
15343 @opindex fcompare-debug-second
15344 This option is implicitly passed to the compiler for the second
15345 compilation requested by @option{-fcompare-debug}, along with options to
15346 silence warnings, and omitting other options that would cause the compiler
15347 to produce output to files or to standard output as a side effect. Dump
15348 files and preserved temporary files are renamed so as to contain the
15349 @code{.gk} additional extension during the second compilation, to avoid
15350 overwriting those generated by the first.
15352 When this option is passed to the compiler driver, it causes the
15353 @emph{first} compilation to be skipped, which makes it useful for little
15354 other than debugging the compiler proper.
15358 Turn off generation of debug info, if leaving out this option
15359 generates it, or turn it on at level 2 otherwise. The position of this
15360 argument in the command line does not matter; it takes effect after all
15361 other options are processed, and it does so only once, no matter how
15362 many times it is given. This is mainly intended to be used with
15363 @option{-fcompare-debug}.
15365 @item -fvar-tracking-assignments-toggle
15366 @opindex fvar-tracking-assignments-toggle
15367 @opindex fno-var-tracking-assignments-toggle
15368 Toggle @option{-fvar-tracking-assignments}, in the same way that
15369 @option{-gtoggle} toggles @option{-g}.
15373 Makes the compiler print out each function name as it is compiled, and
15374 print some statistics about each pass when it finishes.
15376 @item -ftime-report
15377 @opindex ftime-report
15378 Makes the compiler print some statistics about the time consumed by each
15379 pass when it finishes.
15381 @item -ftime-report-details
15382 @opindex ftime-report-details
15383 Record the time consumed by infrastructure parts separately for each pass.
15385 @item -fira-verbose=@var{n}
15386 @opindex fira-verbose
15387 Control the verbosity of the dump file for the integrated register allocator.
15388 The default value is 5. If the value @var{n} is greater or equal to 10,
15389 the dump output is sent to stderr using the same format as @var{n} minus 10.
15392 @opindex flto-report
15393 Prints a report with internal details on the workings of the link-time
15394 optimizer. The contents of this report vary from version to version.
15395 It is meant to be useful to GCC developers when processing object
15396 files in LTO mode (via @option{-flto}).
15398 Disabled by default.
15400 @item -flto-report-wpa
15401 @opindex flto-report-wpa
15402 Like @option{-flto-report}, but only print for the WPA phase of link-time
15406 @opindex fmem-report
15407 Makes the compiler print some statistics about permanent memory
15408 allocation when it finishes.
15410 @item -fmem-report-wpa
15411 @opindex fmem-report-wpa
15412 Makes the compiler print some statistics about permanent memory
15413 allocation for the WPA phase only.
15415 @item -fpre-ipa-mem-report
15416 @opindex fpre-ipa-mem-report
15417 @item -fpost-ipa-mem-report
15418 @opindex fpost-ipa-mem-report
15419 Makes the compiler print some statistics about permanent memory
15420 allocation before or after interprocedural optimization.
15422 @item -fprofile-report
15423 @opindex fprofile-report
15424 Makes the compiler print some statistics about consistency of the
15425 (estimated) profile and effect of individual passes.
15427 @item -fstack-usage
15428 @opindex fstack-usage
15429 Makes the compiler output stack usage information for the program, on a
15430 per-function basis. The filename for the dump is made by appending
15431 @file{.su} to the @var{auxname}. @var{auxname} is generated from the name of
15432 the output file, if explicitly specified and it is not an executable,
15433 otherwise it is the basename of the source file. An entry is made up
15438 The name of the function.
15442 One or more qualifiers: @code{static}, @code{dynamic}, @code{bounded}.
15445 The qualifier @code{static} means that the function manipulates the stack
15446 statically: a fixed number of bytes are allocated for the frame on function
15447 entry and released on function exit; no stack adjustments are otherwise made
15448 in the function. The second field is this fixed number of bytes.
15450 The qualifier @code{dynamic} means that the function manipulates the stack
15451 dynamically: in addition to the static allocation described above, stack
15452 adjustments are made in the body of the function, for example to push/pop
15453 arguments around function calls. If the qualifier @code{bounded} is also
15454 present, the amount of these adjustments is bounded at compile time and
15455 the second field is an upper bound of the total amount of stack used by
15456 the function. If it is not present, the amount of these adjustments is
15457 not bounded at compile time and the second field only represents the
15462 Emit statistics about front-end processing at the end of the compilation.
15463 This option is supported only by the C++ front end, and
15464 the information is generally only useful to the G++ development team.
15466 @item -fdbg-cnt-list
15467 @opindex fdbg-cnt-list
15468 Print the name and the counter upper bound for all debug counters.
15471 @item -fdbg-cnt=@var{counter-value-list}
15473 Set the internal debug counter lower and upper bound. @var{counter-value-list}
15474 is a comma-separated list of @var{name}:@var{lower_bound}:@var{upper_bound}
15475 tuples which sets the lower and the upper bound of each debug
15476 counter @var{name}. The @var{lower_bound} is optional and is zero
15477 initialized if not set.
15478 All debug counters have the initial upper bound of @code{UINT_MAX};
15479 thus @code{dbg_cnt} returns true always unless the upper bound
15480 is set by this option.
15481 For example, with @option{-fdbg-cnt=dce:2:4,tail_call:10},
15482 @code{dbg_cnt(dce)} returns true only for third and fourth invocation.
15483 For @code{dbg_cnt(tail_call)} true is returned for first 10 invocations.
15485 @item -print-file-name=@var{library}
15486 @opindex print-file-name
15487 Print the full absolute name of the library file @var{library} that
15488 would be used when linking---and don't do anything else. With this
15489 option, GCC does not compile or link anything; it just prints the
15492 @item -print-multi-directory
15493 @opindex print-multi-directory
15494 Print the directory name corresponding to the multilib selected by any
15495 other switches present in the command line. This directory is supposed
15496 to exist in @env{GCC_EXEC_PREFIX}.
15498 @item -print-multi-lib
15499 @opindex print-multi-lib
15500 Print the mapping from multilib directory names to compiler switches
15501 that enable them. The directory name is separated from the switches by
15502 @samp{;}, and each switch starts with an @samp{@@} instead of the
15503 @samp{-}, without spaces between multiple switches. This is supposed to
15504 ease shell processing.
15506 @item -print-multi-os-directory
15507 @opindex print-multi-os-directory
15508 Print the path to OS libraries for the selected
15509 multilib, relative to some @file{lib} subdirectory. If OS libraries are
15510 present in the @file{lib} subdirectory and no multilibs are used, this is
15511 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
15512 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
15513 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
15514 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
15516 @item -print-multiarch
15517 @opindex print-multiarch
15518 Print the path to OS libraries for the selected multiarch,
15519 relative to some @file{lib} subdirectory.
15521 @item -print-prog-name=@var{program}
15522 @opindex print-prog-name
15523 Like @option{-print-file-name}, but searches for a program such as @command{cpp}.
15525 @item -print-libgcc-file-name
15526 @opindex print-libgcc-file-name
15527 Same as @option{-print-file-name=libgcc.a}.
15529 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
15530 but you do want to link with @file{libgcc.a}. You can do:
15533 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
15536 @item -print-search-dirs
15537 @opindex print-search-dirs
15538 Print the name of the configured installation directory and a list of
15539 program and library directories @command{gcc} searches---and don't do anything else.
15541 This is useful when @command{gcc} prints the error message
15542 @samp{installation problem, cannot exec cpp0: No such file or directory}.
15543 To resolve this you either need to put @file{cpp0} and the other compiler
15544 components where @command{gcc} expects to find them, or you can set the environment
15545 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
15546 Don't forget the trailing @samp{/}.
15547 @xref{Environment Variables}.
15549 @item -print-sysroot
15550 @opindex print-sysroot
15551 Print the target sysroot directory that is used during
15552 compilation. This is the target sysroot specified either at configure
15553 time or using the @option{--sysroot} option, possibly with an extra
15554 suffix that depends on compilation options. If no target sysroot is
15555 specified, the option prints nothing.
15557 @item -print-sysroot-headers-suffix
15558 @opindex print-sysroot-headers-suffix
15559 Print the suffix added to the target sysroot when searching for
15560 headers, or give an error if the compiler is not configured with such
15561 a suffix---and don't do anything else.
15564 @opindex dumpmachine
15565 Print the compiler's target machine (for example,
15566 @samp{i686-pc-linux-gnu})---and don't do anything else.
15569 @opindex dumpversion
15570 Print the compiler version (for example, @code{3.0}, @code{6.3.0} or @code{7})---and don't do
15571 anything else. This is the compiler version used in filesystem paths and
15572 specs. Depending on how the compiler has been configured it can be just
15573 a single number (major version), two numbers separated by a dot (major and
15574 minor version) or three numbers separated by dots (major, minor and patchlevel
15577 @item -dumpfullversion
15578 @opindex dumpfullversion
15579 Print the full compiler version---and don't do anything else. The output is
15580 always three numbers separated by dots, major, minor and patchlevel version.
15584 Print the compiler's built-in specs---and don't do anything else. (This
15585 is used when GCC itself is being built.) @xref{Spec Files}.
15588 @node Submodel Options
15589 @section Machine-Dependent Options
15590 @cindex submodel options
15591 @cindex specifying hardware config
15592 @cindex hardware models and configurations, specifying
15593 @cindex target-dependent options
15594 @cindex machine-dependent options
15596 Each target machine supported by GCC can have its own options---for
15597 example, to allow you to compile for a particular processor variant or
15598 ABI, or to control optimizations specific to that machine. By
15599 convention, the names of machine-specific options start with
15602 Some configurations of the compiler also support additional target-specific
15603 options, usually for compatibility with other compilers on the same
15606 @c This list is ordered alphanumerically by subsection name.
15607 @c It should be the same order and spelling as these options are listed
15608 @c in Machine Dependent Options
15611 * AArch64 Options::
15612 * Adapteva Epiphany Options::
15613 * AMD GCN Options::
15617 * Blackfin Options::
15623 * DEC Alpha Options::
15627 * GNU/Linux Options::
15637 * MicroBlaze Options::
15640 * MN10300 Options::
15644 * Nios II Options::
15645 * Nvidia PTX Options::
15646 * OpenRISC Options::
15648 * picoChip Options::
15649 * PowerPC Options::
15653 * RS/6000 and PowerPC Options::
15655 * S/390 and zSeries Options::
15658 * Solaris 2 Options::
15661 * System V Options::
15662 * TILE-Gx Options::
15663 * TILEPro Options::
15668 * VxWorks Options::
15670 * x86 Windows Options::
15671 * Xstormy16 Options::
15673 * zSeries Options::
15676 @node AArch64 Options
15677 @subsection AArch64 Options
15678 @cindex AArch64 Options
15680 These options are defined for AArch64 implementations:
15684 @item -mabi=@var{name}
15686 Generate code for the specified data model. Permissible values
15687 are @samp{ilp32} for SysV-like data model where int, long int and pointers
15688 are 32 bits, and @samp{lp64} for SysV-like data model where int is 32 bits,
15689 but long int and pointers are 64 bits.
15691 The default depends on the specific target configuration. Note that
15692 the LP64 and ILP32 ABIs are not link-compatible; you must compile your
15693 entire program with the same ABI, and link with a compatible set of libraries.
15696 @opindex mbig-endian
15697 Generate big-endian code. This is the default when GCC is configured for an
15698 @samp{aarch64_be-*-*} target.
15700 @item -mgeneral-regs-only
15701 @opindex mgeneral-regs-only
15702 Generate code which uses only the general-purpose registers. This will prevent
15703 the compiler from using floating-point and Advanced SIMD registers but will not
15704 impose any restrictions on the assembler.
15706 @item -mlittle-endian
15707 @opindex mlittle-endian
15708 Generate little-endian code. This is the default when GCC is configured for an
15709 @samp{aarch64-*-*} but not an @samp{aarch64_be-*-*} target.
15711 @item -mcmodel=tiny
15712 @opindex mcmodel=tiny
15713 Generate code for the tiny code model. The program and its statically defined
15714 symbols must be within 1MB of each other. Programs can be statically or
15715 dynamically linked.
15717 @item -mcmodel=small
15718 @opindex mcmodel=small
15719 Generate code for the small code model. The program and its statically defined
15720 symbols must be within 4GB of each other. Programs can be statically or
15721 dynamically linked. This is the default code model.
15723 @item -mcmodel=large
15724 @opindex mcmodel=large
15725 Generate code for the large code model. This makes no assumptions about
15726 addresses and sizes of sections. Programs can be statically linked only.
15728 @item -mstrict-align
15729 @itemx -mno-strict-align
15730 @opindex mstrict-align
15731 @opindex mno-strict-align
15732 Avoid or allow generating memory accesses that may not be aligned on a natural
15733 object boundary as described in the architecture specification.
15735 @item -momit-leaf-frame-pointer
15736 @itemx -mno-omit-leaf-frame-pointer
15737 @opindex momit-leaf-frame-pointer
15738 @opindex mno-omit-leaf-frame-pointer
15739 Omit or keep the frame pointer in leaf functions. The former behavior is the
15742 @item -mstack-protector-guard=@var{guard}
15743 @itemx -mstack-protector-guard-reg=@var{reg}
15744 @itemx -mstack-protector-guard-offset=@var{offset}
15745 @opindex mstack-protector-guard
15746 @opindex mstack-protector-guard-reg
15747 @opindex mstack-protector-guard-offset
15748 Generate stack protection code using canary at @var{guard}. Supported
15749 locations are @samp{global} for a global canary or @samp{sysreg} for a
15750 canary in an appropriate system register.
15752 With the latter choice the options
15753 @option{-mstack-protector-guard-reg=@var{reg}} and
15754 @option{-mstack-protector-guard-offset=@var{offset}} furthermore specify
15755 which system register to use as base register for reading the canary,
15756 and from what offset from that base register. There is no default
15757 register or offset as this is entirely for use within the Linux
15760 @item -mstack-protector-guard=@var{guard}
15761 @itemx -mstack-protector-guard-reg=@var{reg}
15762 @itemx -mstack-protector-guard-offset=@var{offset}
15763 @opindex mstack-protector-guard
15764 @opindex mstack-protector-guard-reg
15765 @opindex mstack-protector-guard-offset
15766 Generate stack protection code using canary at @var{guard}. Supported
15767 locations are @samp{global} for a global canary or @samp{sysreg} for a
15768 canary in an appropriate system register.
15770 With the latter choice the options
15771 @option{-mstack-protector-guard-reg=@var{reg}} and
15772 @option{-mstack-protector-guard-offset=@var{offset}} furthermore specify
15773 which system register to use as base register for reading the canary,
15774 and from what offset from that base register. There is no default
15775 register or offset as this is entirely for use within the Linux
15778 @item -mtls-dialect=desc
15779 @opindex mtls-dialect=desc
15780 Use TLS descriptors as the thread-local storage mechanism for dynamic accesses
15781 of TLS variables. This is the default.
15783 @item -mtls-dialect=traditional
15784 @opindex mtls-dialect=traditional
15785 Use traditional TLS as the thread-local storage mechanism for dynamic accesses
15788 @item -mtls-size=@var{size}
15790 Specify bit size of immediate TLS offsets. Valid values are 12, 24, 32, 48.
15791 This option requires binutils 2.26 or newer.
15793 @item -mfix-cortex-a53-835769
15794 @itemx -mno-fix-cortex-a53-835769
15795 @opindex mfix-cortex-a53-835769
15796 @opindex mno-fix-cortex-a53-835769
15797 Enable or disable the workaround for the ARM Cortex-A53 erratum number 835769.
15798 This involves inserting a NOP instruction between memory instructions and
15799 64-bit integer multiply-accumulate instructions.
15801 @item -mfix-cortex-a53-843419
15802 @itemx -mno-fix-cortex-a53-843419
15803 @opindex mfix-cortex-a53-843419
15804 @opindex mno-fix-cortex-a53-843419
15805 Enable or disable the workaround for the ARM Cortex-A53 erratum number 843419.
15806 This erratum workaround is made at link time and this will only pass the
15807 corresponding flag to the linker.
15809 @item -mlow-precision-recip-sqrt
15810 @itemx -mno-low-precision-recip-sqrt
15811 @opindex mlow-precision-recip-sqrt
15812 @opindex mno-low-precision-recip-sqrt
15813 Enable or disable the reciprocal square root approximation.
15814 This option only has an effect if @option{-ffast-math} or
15815 @option{-funsafe-math-optimizations} is used as well. Enabling this reduces
15816 precision of reciprocal square root results to about 16 bits for
15817 single precision and to 32 bits for double precision.
15819 @item -mlow-precision-sqrt
15820 @itemx -mno-low-precision-sqrt
15821 @opindex mlow-precision-sqrt
15822 @opindex mno-low-precision-sqrt
15823 Enable or disable the square root approximation.
15824 This option only has an effect if @option{-ffast-math} or
15825 @option{-funsafe-math-optimizations} is used as well. Enabling this reduces
15826 precision of square root results to about 16 bits for
15827 single precision and to 32 bits for double precision.
15828 If enabled, it implies @option{-mlow-precision-recip-sqrt}.
15830 @item -mlow-precision-div
15831 @itemx -mno-low-precision-div
15832 @opindex mlow-precision-div
15833 @opindex mno-low-precision-div
15834 Enable or disable the division approximation.
15835 This option only has an effect if @option{-ffast-math} or
15836 @option{-funsafe-math-optimizations} is used as well. Enabling this reduces
15837 precision of division results to about 16 bits for
15838 single precision and to 32 bits for double precision.
15840 @item -mtrack-speculation
15841 @itemx -mno-track-speculation
15842 Enable or disable generation of additional code to track speculative
15843 execution through conditional branches. The tracking state can then
15844 be used by the compiler when expanding calls to
15845 @code{__builtin_speculation_safe_copy} to permit a more efficient code
15846 sequence to be generated.
15848 @item -march=@var{name}
15850 Specify the name of the target architecture and, optionally, one or
15851 more feature modifiers. This option has the form
15852 @option{-march=@var{arch}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}.
15854 The permissible values for @var{arch} are @samp{armv8-a},
15855 @samp{armv8.1-a}, @samp{armv8.2-a}, @samp{armv8.3-a}, @samp{armv8.4-a},
15856 @samp{armv8.5-a} or @var{native}.
15858 The value @samp{armv8.5-a} implies @samp{armv8.4-a} and enables compiler
15859 support for the ARMv8.5-A architecture extensions.
15861 The value @samp{armv8.4-a} implies @samp{armv8.3-a} and enables compiler
15862 support for the ARMv8.4-A architecture extensions.
15864 The value @samp{armv8.3-a} implies @samp{armv8.2-a} and enables compiler
15865 support for the ARMv8.3-A architecture extensions.
15867 The value @samp{armv8.2-a} implies @samp{armv8.1-a} and enables compiler
15868 support for the ARMv8.2-A architecture extensions.
15870 The value @samp{armv8.1-a} implies @samp{armv8-a} and enables compiler
15871 support for the ARMv8.1-A architecture extension. In particular, it
15872 enables the @samp{+crc}, @samp{+lse}, and @samp{+rdma} features.
15874 The value @samp{native} is available on native AArch64 GNU/Linux and
15875 causes the compiler to pick the architecture of the host system. This
15876 option has no effect if the compiler is unable to recognize the
15877 architecture of the host system,
15879 The permissible values for @var{feature} are listed in the sub-section
15880 on @ref{aarch64-feature-modifiers,,@option{-march} and @option{-mcpu}
15881 Feature Modifiers}. Where conflicting feature modifiers are
15882 specified, the right-most feature is used.
15884 GCC uses @var{name} to determine what kind of instructions it can emit
15885 when generating assembly code. If @option{-march} is specified
15886 without either of @option{-mtune} or @option{-mcpu} also being
15887 specified, the code is tuned to perform well across a range of target
15888 processors implementing the target architecture.
15890 @item -mtune=@var{name}
15892 Specify the name of the target processor for which GCC should tune the
15893 performance of the code. Permissible values for this option are:
15894 @samp{generic}, @samp{cortex-a35}, @samp{cortex-a53}, @samp{cortex-a55},
15895 @samp{cortex-a57}, @samp{cortex-a72}, @samp{cortex-a73}, @samp{cortex-a75},
15896 @samp{cortex-a76}, @samp{ares}, @samp{exynos-m1}, @samp{emag}, @samp{falkor},
15897 @samp{neoverse-e1},@samp{neoverse-n1},@samp{qdf24xx}, @samp{saphira},
15898 @samp{phecda}, @samp{xgene1}, @samp{vulcan}, @samp{octeontx},
15899 @samp{octeontx81}, @samp{octeontx83}, @samp{thunderx}, @samp{thunderxt88},
15900 @samp{thunderxt88p1}, @samp{thunderxt81}, @samp{tsv110},
15901 @samp{thunderxt83}, @samp{thunderx2t99},
15902 @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53},
15903 @samp{cortex-a73.cortex-a35}, @samp{cortex-a73.cortex-a53},
15904 @samp{cortex-a75.cortex-a55}, @samp{cortex-a76.cortex-a55}
15907 The values @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53},
15908 @samp{cortex-a73.cortex-a35}, @samp{cortex-a73.cortex-a53},
15909 @samp{cortex-a75.cortex-a55}, @samp{cortex-a76.cortex-a55} specify that GCC
15910 should tune for a big.LITTLE system.
15912 Additionally on native AArch64 GNU/Linux systems the value
15913 @samp{native} tunes performance to the host system. This option has no effect
15914 if the compiler is unable to recognize the processor of the host system.
15916 Where none of @option{-mtune=}, @option{-mcpu=} or @option{-march=}
15917 are specified, the code is tuned to perform well across a range
15918 of target processors.
15920 This option cannot be suffixed by feature modifiers.
15922 @item -mcpu=@var{name}
15924 Specify the name of the target processor, optionally suffixed by one
15925 or more feature modifiers. This option has the form
15926 @option{-mcpu=@var{cpu}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}, where
15927 the permissible values for @var{cpu} are the same as those available
15928 for @option{-mtune}. The permissible values for @var{feature} are
15929 documented in the sub-section on
15930 @ref{aarch64-feature-modifiers,,@option{-march} and @option{-mcpu}
15931 Feature Modifiers}. Where conflicting feature modifiers are
15932 specified, the right-most feature is used.
15934 GCC uses @var{name} to determine what kind of instructions it can emit when
15935 generating assembly code (as if by @option{-march}) and to determine
15936 the target processor for which to tune for performance (as if
15937 by @option{-mtune}). Where this option is used in conjunction
15938 with @option{-march} or @option{-mtune}, those options take precedence
15939 over the appropriate part of this option.
15941 @item -moverride=@var{string}
15943 Override tuning decisions made by the back-end in response to a
15944 @option{-mtune=} switch. The syntax, semantics, and accepted values
15945 for @var{string} in this option are not guaranteed to be consistent
15948 This option is only intended to be useful when developing GCC.
15950 @item -mverbose-cost-dump
15951 @opindex mverbose-cost-dump
15952 Enable verbose cost model dumping in the debug dump files. This option is
15953 provided for use in debugging the compiler.
15955 @item -mpc-relative-literal-loads
15956 @itemx -mno-pc-relative-literal-loads
15957 @opindex mpc-relative-literal-loads
15958 @opindex mno-pc-relative-literal-loads
15959 Enable or disable PC-relative literal loads. With this option literal pools are
15960 accessed using a single instruction and emitted after each function. This
15961 limits the maximum size of functions to 1MB. This is enabled by default for
15962 @option{-mcmodel=tiny}.
15964 @item -msign-return-address=@var{scope}
15965 @opindex msign-return-address
15966 Select the function scope on which return address signing will be applied.
15967 Permissible values are @samp{none}, which disables return address signing,
15968 @samp{non-leaf}, which enables pointer signing for functions which are not leaf
15969 functions, and @samp{all}, which enables pointer signing for all functions. The
15970 default value is @samp{none}. This option has been deprecated by
15971 -mbranch-protection.
15973 @item -mbranch-protection=@var{none}|@var{standard}|@var{pac-ret}[+@var{leaf}+@var{b-key}]|@var{bti}
15974 @opindex mbranch-protection
15975 Select the branch protection features to use.
15976 @samp{none} is the default and turns off all types of branch protection.
15977 @samp{standard} turns on all types of branch protection features. If a feature
15978 has additional tuning options, then @samp{standard} sets it to its standard
15980 @samp{pac-ret[+@var{leaf}]} turns on return address signing to its standard
15981 level: signing functions that save the return address to memory (non-leaf
15982 functions will practically always do this) using the a-key. The optional
15983 argument @samp{leaf} can be used to extend the signing to include leaf
15984 functions. The optional argument @samp{b-key} can be used to sign the functions
15985 with the B-key instead of the A-key.
15986 @samp{bti} turns on branch target identification mechanism.
15988 @item -msve-vector-bits=@var{bits}
15989 @opindex msve-vector-bits
15990 Specify the number of bits in an SVE vector register. This option only has
15991 an effect when SVE is enabled.
15993 GCC supports two forms of SVE code generation: ``vector-length
15994 agnostic'' output that works with any size of vector register and
15995 ``vector-length specific'' output that allows GCC to make assumptions
15996 about the vector length when it is useful for optimization reasons.
15997 The possible values of @samp{bits} are: @samp{scalable}, @samp{128},
15998 @samp{256}, @samp{512}, @samp{1024} and @samp{2048}.
15999 Specifying @samp{scalable} selects vector-length agnostic
16000 output. At present @samp{-msve-vector-bits=128} also generates vector-length
16001 agnostic output. All other values generate vector-length specific code.
16002 The behavior of these values may change in future releases and no value except
16003 @samp{scalable} should be relied on for producing code that is portable across
16004 different hardware SVE vector lengths.
16006 The default is @samp{-msve-vector-bits=scalable}, which produces
16007 vector-length agnostic code.
16010 @subsubsection @option{-march} and @option{-mcpu} Feature Modifiers
16011 @anchor{aarch64-feature-modifiers}
16012 @cindex @option{-march} feature modifiers
16013 @cindex @option{-mcpu} feature modifiers
16014 Feature modifiers used with @option{-march} and @option{-mcpu} can be any of
16015 the following and their inverses @option{no@var{feature}}:
16019 Enable CRC extension. This is on by default for
16020 @option{-march=armv8.1-a}.
16022 Enable Crypto extension. This also enables Advanced SIMD and floating-point
16025 Enable floating-point instructions. This is on by default for all possible
16026 values for options @option{-march} and @option{-mcpu}.
16028 Enable Advanced SIMD instructions. This also enables floating-point
16029 instructions. This is on by default for all possible values for options
16030 @option{-march} and @option{-mcpu}.
16032 Enable Scalable Vector Extension instructions. This also enables Advanced
16033 SIMD and floating-point instructions.
16035 Enable Large System Extension instructions. This is on by default for
16036 @option{-march=armv8.1-a}.
16038 Enable Round Double Multiply Accumulate instructions. This is on by default
16039 for @option{-march=armv8.1-a}.
16041 Enable FP16 extension. This also enables floating-point instructions.
16043 Enable FP16 fmla extension. This also enables FP16 extensions and
16044 floating-point instructions. This option is enabled by default for @option{-march=armv8.4-a}. Use of this option with architectures prior to Armv8.2-A is not supported.
16047 Enable the RcPc extension. This does not change code generation from GCC,
16048 but is passed on to the assembler, enabling inline asm statements to use
16049 instructions from the RcPc extension.
16051 Enable the Dot Product extension. This also enables Advanced SIMD instructions.
16053 Enable the Armv8-a aes and pmull crypto extension. This also enables Advanced
16056 Enable the Armv8-a sha2 crypto extension. This also enables Advanced SIMD instructions.
16058 Enable the sha512 and sha3 crypto extension. This also enables Advanced SIMD
16059 instructions. Use of this option with architectures prior to Armv8.2-A is not supported.
16061 Enable the sm3 and sm4 crypto extension. This also enables Advanced SIMD instructions.
16062 Use of this option with architectures prior to Armv8.2-A is not supported.
16064 Enable the Statistical Profiling extension. This option is only to enable the
16065 extension at the assembler level and does not affect code generation.
16067 Enable the Armv8.5-a Random Number instructions. This option is only to
16068 enable the extension at the assembler level and does not affect code
16071 Enable the Armv8.5-a Memory Tagging Extensions. This option is only to
16072 enable the extension at the assembler level and does not affect code
16075 Enable the Armv8-a Speculation Barrier instruction. This option is only to
16076 enable the extension at the assembler level and does not affect code
16077 generation. This option is enabled by default for @option{-march=armv8.5-a}.
16079 Enable the Armv8-a Speculative Store Bypass Safe instruction. This option
16080 is only to enable the extension at the assembler level and does not affect code
16081 generation. This option is enabled by default for @option{-march=armv8.5-a}.
16083 Enable the Armv8-a Execution and Data Prediction Restriction instructions.
16084 This option is only to enable the extension at the assembler level and does
16085 not affect code generation. This option is enabled by default for
16087 Enable the Armv8-a Scalable Vector Extension 2. This also enables SVE
16090 Enable SVE2 bitperm instructions. This also enables SVE2 instructions.
16092 Enable SVE2 sm4 instructions. This also enables SVE2 instructions.
16094 Enable SVE2 aes instructions. This also enables SVE2 instructions.
16096 Enable SVE2 sha3 instructions. This also enables SVE2 instructions.
16097 @option{-march=armv8.5-a}.
16099 Enable the Transactional Memory Extension.
16103 Feature @option{crypto} implies @option{aes}, @option{sha2}, and @option{simd},
16104 which implies @option{fp}.
16105 Conversely, @option{nofp} implies @option{nosimd}, which implies
16106 @option{nocrypto}, @option{noaes} and @option{nosha2}.
16108 @node Adapteva Epiphany Options
16109 @subsection Adapteva Epiphany Options
16111 These @samp{-m} options are defined for Adapteva Epiphany:
16114 @item -mhalf-reg-file
16115 @opindex mhalf-reg-file
16116 Don't allocate any register in the range @code{r32}@dots{}@code{r63}.
16117 That allows code to run on hardware variants that lack these registers.
16119 @item -mprefer-short-insn-regs
16120 @opindex mprefer-short-insn-regs
16121 Preferentially allocate registers that allow short instruction generation.
16122 This can result in increased instruction count, so this may either reduce or
16123 increase overall code size.
16125 @item -mbranch-cost=@var{num}
16126 @opindex mbranch-cost
16127 Set the cost of branches to roughly @var{num} ``simple'' instructions.
16128 This cost is only a heuristic and is not guaranteed to produce
16129 consistent results across releases.
16133 Enable the generation of conditional moves.
16135 @item -mnops=@var{num}
16137 Emit @var{num} NOPs before every other generated instruction.
16139 @item -mno-soft-cmpsf
16140 @opindex mno-soft-cmpsf
16141 @opindex msoft-cmpsf
16142 For single-precision floating-point comparisons, emit an @code{fsub} instruction
16143 and test the flags. This is faster than a software comparison, but can
16144 get incorrect results in the presence of NaNs, or when two different small
16145 numbers are compared such that their difference is calculated as zero.
16146 The default is @option{-msoft-cmpsf}, which uses slower, but IEEE-compliant,
16147 software comparisons.
16149 @item -mstack-offset=@var{num}
16150 @opindex mstack-offset
16151 Set the offset between the top of the stack and the stack pointer.
16152 E.g., a value of 8 means that the eight bytes in the range @code{sp+0@dots{}sp+7}
16153 can be used by leaf functions without stack allocation.
16154 Values other than @samp{8} or @samp{16} are untested and unlikely to work.
16155 Note also that this option changes the ABI; compiling a program with a
16156 different stack offset than the libraries have been compiled with
16157 generally does not work.
16158 This option can be useful if you want to evaluate if a different stack
16159 offset would give you better code, but to actually use a different stack
16160 offset to build working programs, it is recommended to configure the
16161 toolchain with the appropriate @option{--with-stack-offset=@var{num}} option.
16163 @item -mno-round-nearest
16164 @opindex mno-round-nearest
16165 @opindex mround-nearest
16166 Make the scheduler assume that the rounding mode has been set to
16167 truncating. The default is @option{-mround-nearest}.
16170 @opindex mlong-calls
16171 If not otherwise specified by an attribute, assume all calls might be beyond
16172 the offset range of the @code{b} / @code{bl} instructions, and therefore load the
16173 function address into a register before performing a (otherwise direct) call.
16174 This is the default.
16176 @item -mshort-calls
16177 @opindex short-calls
16178 If not otherwise specified by an attribute, assume all direct calls are
16179 in the range of the @code{b} / @code{bl} instructions, so use these instructions
16180 for direct calls. The default is @option{-mlong-calls}.
16184 Assume addresses can be loaded as 16-bit unsigned values. This does not
16185 apply to function addresses for which @option{-mlong-calls} semantics
16188 @item -mfp-mode=@var{mode}
16190 Set the prevailing mode of the floating-point unit.
16191 This determines the floating-point mode that is provided and expected
16192 at function call and return time. Making this mode match the mode you
16193 predominantly need at function start can make your programs smaller and
16194 faster by avoiding unnecessary mode switches.
16196 @var{mode} can be set to one the following values:
16200 Any mode at function entry is valid, and retained or restored when
16201 the function returns, and when it calls other functions.
16202 This mode is useful for compiling libraries or other compilation units
16203 you might want to incorporate into different programs with different
16204 prevailing FPU modes, and the convenience of being able to use a single
16205 object file outweighs the size and speed overhead for any extra
16206 mode switching that might be needed, compared with what would be needed
16207 with a more specific choice of prevailing FPU mode.
16210 This is the mode used for floating-point calculations with
16211 truncating (i.e.@: round towards zero) rounding mode. That includes
16212 conversion from floating point to integer.
16214 @item round-nearest
16215 This is the mode used for floating-point calculations with
16216 round-to-nearest-or-even rounding mode.
16219 This is the mode used to perform integer calculations in the FPU, e.g.@:
16220 integer multiply, or integer multiply-and-accumulate.
16223 The default is @option{-mfp-mode=caller}
16225 @item -mno-split-lohi
16226 @itemx -mno-postinc
16227 @itemx -mno-postmodify
16228 @opindex mno-split-lohi
16229 @opindex msplit-lohi
16230 @opindex mno-postinc
16232 @opindex mno-postmodify
16233 @opindex mpostmodify
16234 Code generation tweaks that disable, respectively, splitting of 32-bit
16235 loads, generation of post-increment addresses, and generation of
16236 post-modify addresses. The defaults are @option{msplit-lohi},
16237 @option{-mpost-inc}, and @option{-mpost-modify}.
16239 @item -mnovect-double
16240 @opindex mno-vect-double
16241 @opindex mvect-double
16242 Change the preferred SIMD mode to SImode. The default is
16243 @option{-mvect-double}, which uses DImode as preferred SIMD mode.
16245 @item -max-vect-align=@var{num}
16246 @opindex max-vect-align
16247 The maximum alignment for SIMD vector mode types.
16248 @var{num} may be 4 or 8. The default is 8.
16249 Note that this is an ABI change, even though many library function
16250 interfaces are unaffected if they don't use SIMD vector modes
16251 in places that affect size and/or alignment of relevant types.
16253 @item -msplit-vecmove-early
16254 @opindex msplit-vecmove-early
16255 Split vector moves into single word moves before reload. In theory this
16256 can give better register allocation, but so far the reverse seems to be
16257 generally the case.
16259 @item -m1reg-@var{reg}
16261 Specify a register to hold the constant @minus{}1, which makes loading small negative
16262 constants and certain bitmasks faster.
16263 Allowable values for @var{reg} are @samp{r43} and @samp{r63},
16264 which specify use of that register as a fixed register,
16265 and @samp{none}, which means that no register is used for this
16266 purpose. The default is @option{-m1reg-none}.
16270 @node AMD GCN Options
16271 @subsection AMD GCN Options
16272 @cindex AMD GCN Options
16274 These options are defined specifically for the AMD GCN port.
16278 @item -march=@var{gpu}
16280 @itemx -mtune=@var{gpu}
16282 Set architecture type or tuning for @var{gpu}. Supported values for @var{gpu}
16288 Compile for GCN3 Fiji devices (gfx803).
16291 Compile for GCN5 Vega 10 devices (gfx900).
16294 Compile for GCN5 Vega 20 devices (gfx906).
16298 @item -mstack-size=@var{bytes}
16299 @opindex mstack-size
16300 Specify how many @var{bytes} of stack space will be requested for each GPU
16301 thread (wave-front). Beware that there may be many threads and limited memory
16302 available. The size of the stack allocation may also have an impact on
16303 run-time performance. The default is 32KB when using OpenACC or OpenMP, and
16309 @subsection ARC Options
16310 @cindex ARC options
16312 The following options control the architecture variant for which code
16315 @c architecture variants
16318 @item -mbarrel-shifter
16319 @opindex mbarrel-shifter
16320 Generate instructions supported by barrel shifter. This is the default
16321 unless @option{-mcpu=ARC601} or @samp{-mcpu=ARCEM} is in effect.
16324 @opindex mjli-alawys
16325 Force to call a function using jli_s instruction. This option is
16326 valid only for ARCv2 architecture.
16328 @item -mcpu=@var{cpu}
16330 Set architecture type, register usage, and instruction scheduling
16331 parameters for @var{cpu}. There are also shortcut alias options
16332 available for backward compatibility and convenience. Supported
16333 values for @var{cpu} are
16339 Compile for ARC600. Aliases: @option{-mA6}, @option{-mARC600}.
16343 Compile for ARC601. Alias: @option{-mARC601}.
16348 Compile for ARC700. Aliases: @option{-mA7}, @option{-mARC700}.
16349 This is the default when configured with @option{--with-cpu=arc700}@.
16352 Compile for ARC EM.
16355 Compile for ARC HS.
16358 Compile for ARC EM CPU with no hardware extensions.
16361 Compile for ARC EM4 CPU.
16364 Compile for ARC EM4 DMIPS CPU.
16367 Compile for ARC EM4 DMIPS CPU with the single-precision floating-point
16371 Compile for ARC EM4 DMIPS CPU with single-precision floating-point and
16372 double assist instructions.
16375 Compile for ARC HS CPU with no hardware extensions except the atomic
16379 Compile for ARC HS34 CPU.
16382 Compile for ARC HS38 CPU.
16385 Compile for ARC HS38 CPU with all hardware extensions on.
16388 Compile for ARC 600 CPU with @code{norm} instructions enabled.
16390 @item arc600_mul32x16
16391 Compile for ARC 600 CPU with @code{norm} and 32x16-bit multiply
16392 instructions enabled.
16395 Compile for ARC 600 CPU with @code{norm} and @code{mul64}-family
16396 instructions enabled.
16399 Compile for ARC 601 CPU with @code{norm} instructions enabled.
16401 @item arc601_mul32x16
16402 Compile for ARC 601 CPU with @code{norm} and 32x16-bit multiply
16403 instructions enabled.
16406 Compile for ARC 601 CPU with @code{norm} and @code{mul64}-family
16407 instructions enabled.
16410 Compile for ARC 700 on NPS400 chip.
16413 Compile for ARC EM minimalist configuration featuring reduced register
16420 @itemx -mdpfp-compact
16421 @opindex mdpfp-compact
16422 Generate double-precision FPX instructions, tuned for the compact
16426 @opindex mdpfp-fast
16427 Generate double-precision FPX instructions, tuned for the fast
16430 @item -mno-dpfp-lrsr
16431 @opindex mno-dpfp-lrsr
16432 Disable @code{lr} and @code{sr} instructions from using FPX extension
16437 Generate extended arithmetic instructions. Currently only
16438 @code{divaw}, @code{adds}, @code{subs}, and @code{sat16} are
16439 supported. This is always enabled for @option{-mcpu=ARC700}.
16444 Do not generate @code{mpy}-family instructions for ARC700. This option is
16449 Generate 32x16-bit multiply and multiply-accumulate instructions.
16453 Generate @code{mul64} and @code{mulu64} instructions.
16454 Only valid for @option{-mcpu=ARC600}.
16458 Generate @code{norm} instructions. This is the default if @option{-mcpu=ARC700}
16463 @itemx -mspfp-compact
16464 @opindex mspfp-compact
16465 Generate single-precision FPX instructions, tuned for the compact
16469 @opindex mspfp-fast
16470 Generate single-precision FPX instructions, tuned for the fast
16475 Enable generation of ARC SIMD instructions via target-specific
16476 builtins. Only valid for @option{-mcpu=ARC700}.
16479 @opindex msoft-float
16480 This option ignored; it is provided for compatibility purposes only.
16481 Software floating-point code is emitted by default, and this default
16482 can overridden by FPX options; @option{-mspfp}, @option{-mspfp-compact}, or
16483 @option{-mspfp-fast} for single precision, and @option{-mdpfp},
16484 @option{-mdpfp-compact}, or @option{-mdpfp-fast} for double precision.
16488 Generate @code{swap} instructions.
16492 This enables use of the locked load/store conditional extension to implement
16493 atomic memory built-in functions. Not available for ARC 6xx or ARC
16498 Enable @code{div} and @code{rem} instructions for ARCv2 cores.
16500 @item -mcode-density
16501 @opindex mcode-density
16502 Enable code density instructions for ARC EM.
16503 This option is on by default for ARC HS.
16507 Enable double load/store operations for ARC HS cores.
16509 @item -mtp-regno=@var{regno}
16511 Specify thread pointer register number.
16513 @item -mmpy-option=@var{multo}
16514 @opindex mmpy-option
16515 Compile ARCv2 code with a multiplier design option. You can specify
16516 the option using either a string or numeric value for @var{multo}.
16517 @samp{wlh1} is the default value. The recognized values are:
16522 No multiplier available.
16526 16x16 multiplier, fully pipelined.
16527 The following instructions are enabled: @code{mpyw} and @code{mpyuw}.
16531 32x32 multiplier, fully
16532 pipelined (1 stage). The following instructions are additionally
16533 enabled: @code{mpy}, @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
16537 32x32 multiplier, fully pipelined
16538 (2 stages). The following instructions are additionally enabled: @code{mpy},
16539 @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
16543 Two 16x16 multipliers, blocking,
16544 sequential. The following instructions are additionally enabled: @code{mpy},
16545 @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
16549 One 16x16 multiplier, blocking,
16550 sequential. The following instructions are additionally enabled: @code{mpy},
16551 @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
16555 One 32x4 multiplier, blocking,
16556 sequential. The following instructions are additionally enabled: @code{mpy},
16557 @code{mpyu}, @code{mpym}, @code{mpymu}, and @code{mpy_s}.
16561 ARC HS SIMD support.
16565 ARC HS SIMD support.
16569 ARC HS SIMD support.
16573 This option is only available for ARCv2 cores@.
16575 @item -mfpu=@var{fpu}
16577 Enables support for specific floating-point hardware extensions for ARCv2
16578 cores. Supported values for @var{fpu} are:
16583 Enables support for single-precision floating-point hardware
16587 Enables support for double-precision floating-point hardware
16588 extensions. The single-precision floating-point extension is also
16589 enabled. Not available for ARC EM@.
16592 Enables support for double-precision floating-point hardware
16593 extensions using double-precision assist instructions. The single-precision
16594 floating-point extension is also enabled. This option is
16595 only available for ARC EM@.
16598 Enables support for double-precision floating-point hardware
16599 extensions using double-precision assist instructions.
16600 The single-precision floating-point, square-root, and divide
16601 extensions are also enabled. This option is
16602 only available for ARC EM@.
16605 Enables support for double-precision floating-point hardware
16606 extensions using double-precision assist instructions.
16607 The single-precision floating-point and fused multiply and add
16608 hardware extensions are also enabled. This option is
16609 only available for ARC EM@.
16612 Enables support for double-precision floating-point hardware
16613 extensions using double-precision assist instructions.
16614 All single-precision floating-point hardware extensions are also
16615 enabled. This option is only available for ARC EM@.
16618 Enables support for single-precision floating-point, square-root and divide
16619 hardware extensions@.
16622 Enables support for double-precision floating-point, square-root and divide
16623 hardware extensions. This option
16624 includes option @samp{fpus_div}. Not available for ARC EM@.
16627 Enables support for single-precision floating-point and
16628 fused multiply and add hardware extensions@.
16631 Enables support for double-precision floating-point and
16632 fused multiply and add hardware extensions. This option
16633 includes option @samp{fpus_fma}. Not available for ARC EM@.
16636 Enables support for all single-precision floating-point hardware
16640 Enables support for all single- and double-precision floating-point
16641 hardware extensions. Not available for ARC EM@.
16645 @item -mirq-ctrl-saved=@var{register-range}, @var{blink}, @var{lp_count}
16646 @opindex mirq-ctrl-saved
16647 Specifies general-purposes registers that the processor automatically
16648 saves/restores on interrupt entry and exit. @var{register-range} is
16649 specified as two registers separated by a dash. The register range
16650 always starts with @code{r0}, the upper limit is @code{fp} register.
16651 @var{blink} and @var{lp_count} are optional. This option is only
16652 valid for ARC EM and ARC HS cores.
16654 @item -mrgf-banked-regs=@var{number}
16655 @opindex mrgf-banked-regs
16656 Specifies the number of registers replicated in second register bank
16657 on entry to fast interrupt. Fast interrupts are interrupts with the
16658 highest priority level P0. These interrupts save only PC and STATUS32
16659 registers to avoid memory transactions during interrupt entry and exit
16660 sequences. Use this option when you are using fast interrupts in an
16661 ARC V2 family processor. Permitted values are 4, 8, 16, and 32.
16663 @item -mlpc-width=@var{width}
16664 @opindex mlpc-width
16665 Specify the width of the @code{lp_count} register. Valid values for
16666 @var{width} are 8, 16, 20, 24, 28 and 32 bits. The default width is
16667 fixed to 32 bits. If the width is less than 32, the compiler does not
16668 attempt to transform loops in your program to use the zero-delay loop
16669 mechanism unless it is known that the @code{lp_count} register can
16670 hold the required loop-counter value. Depending on the width
16671 specified, the compiler and run-time library might continue to use the
16672 loop mechanism for various needs. This option defines macro
16673 @code{__ARC_LPC_WIDTH__} with the value of @var{width}.
16677 This option instructs the compiler to generate code for a 16-entry
16678 register file. This option defines the @code{__ARC_RF16__}
16679 preprocessor macro.
16681 @item -mbranch-index
16682 @opindex mbranch-index
16683 Enable use of @code{bi} or @code{bih} instructions to implement jump
16688 The following options are passed through to the assembler, and also
16689 define preprocessor macro symbols.
16691 @c Flags used by the assembler, but for which we define preprocessor
16692 @c macro symbols as well.
16695 @opindex mdsp-packa
16696 Passed down to the assembler to enable the DSP Pack A extensions.
16697 Also sets the preprocessor symbol @code{__Xdsp_packa}. This option is
16702 Passed down to the assembler to enable the dual Viterbi butterfly
16703 extension. Also sets the preprocessor symbol @code{__Xdvbf}. This
16704 option is deprecated.
16706 @c ARC700 4.10 extension instruction
16709 Passed down to the assembler to enable the locked load/store
16710 conditional extension. Also sets the preprocessor symbol
16715 Passed down to the assembler. Also sets the preprocessor symbol
16716 @code{__Xxmac_d16}. This option is deprecated.
16720 Passed down to the assembler. Also sets the preprocessor symbol
16721 @code{__Xxmac_24}. This option is deprecated.
16723 @c ARC700 4.10 extension instruction
16726 Passed down to the assembler to enable the 64-bit time-stamp counter
16727 extension instruction. Also sets the preprocessor symbol
16728 @code{__Xrtsc}. This option is deprecated.
16730 @c ARC700 4.10 extension instruction
16733 Passed down to the assembler to enable the swap byte ordering
16734 extension instruction. Also sets the preprocessor symbol
16738 @opindex mtelephony
16739 Passed down to the assembler to enable dual- and single-operand
16740 instructions for telephony. Also sets the preprocessor symbol
16741 @code{__Xtelephony}. This option is deprecated.
16745 Passed down to the assembler to enable the XY memory extension. Also
16746 sets the preprocessor symbol @code{__Xxy}.
16750 The following options control how the assembly code is annotated:
16752 @c Assembly annotation options
16756 Annotate assembler instructions with estimated addresses.
16758 @item -mannotate-align
16759 @opindex mannotate-align
16760 Explain what alignment considerations lead to the decision to make an
16761 instruction short or long.
16765 The following options are passed through to the linker:
16767 @c options passed through to the linker
16771 Passed through to the linker, to specify use of the @code{arclinux} emulation.
16772 This option is enabled by default in tool chains built for
16773 @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets
16774 when profiling is not requested.
16776 @item -marclinux_prof
16777 @opindex marclinux_prof
16778 Passed through to the linker, to specify use of the
16779 @code{arclinux_prof} emulation. This option is enabled by default in
16780 tool chains built for @w{@code{arc-linux-uclibc}} and
16781 @w{@code{arceb-linux-uclibc}} targets when profiling is requested.
16785 The following options control the semantics of generated code:
16787 @c semantically relevant code generation options
16790 @opindex mlong-calls
16791 Generate calls as register indirect calls, thus providing access
16792 to the full 32-bit address range.
16794 @item -mmedium-calls
16795 @opindex mmedium-calls
16796 Don't use less than 25-bit addressing range for calls, which is the
16797 offset available for an unconditional branch-and-link
16798 instruction. Conditional execution of function calls is suppressed, to
16799 allow use of the 25-bit range, rather than the 21-bit range with
16800 conditional branch-and-link. This is the default for tool chains built
16801 for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets.
16805 Put definitions of externally-visible data in a small data section if
16806 that data is no bigger than @var{num} bytes. The default value of
16807 @var{num} is 4 for any ARC configuration, or 8 when we have double
16808 load/store operations.
16813 Do not generate sdata references. This is the default for tool chains
16814 built for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}}
16817 @item -mvolatile-cache
16818 @opindex mvolatile-cache
16819 Use ordinarily cached memory accesses for volatile references. This is the
16822 @item -mno-volatile-cache
16823 @opindex mno-volatile-cache
16824 @opindex mvolatile-cache
16825 Enable cache bypass for volatile references.
16829 The following options fine tune code generation:
16830 @c code generation tuning options
16833 @opindex malign-call
16834 Do alignment optimizations for call instructions.
16836 @item -mauto-modify-reg
16837 @opindex mauto-modify-reg
16838 Enable the use of pre/post modify with register displacement.
16840 @item -mbbit-peephole
16841 @opindex mbbit-peephole
16842 Enable bbit peephole2.
16846 This option disables a target-specific pass in @file{arc_reorg} to
16847 generate compare-and-branch (@code{br@var{cc}}) instructions.
16848 It has no effect on
16849 generation of these instructions driven by the combiner pass.
16851 @item -mcase-vector-pcrel
16852 @opindex mcase-vector-pcrel
16853 Use PC-relative switch case tables to enable case table shortening.
16854 This is the default for @option{-Os}.
16856 @item -mcompact-casesi
16857 @opindex mcompact-casesi
16858 Enable compact @code{casesi} pattern. This is the default for @option{-Os},
16859 and only available for ARCv1 cores. This option is deprecated.
16861 @item -mno-cond-exec
16862 @opindex mno-cond-exec
16863 Disable the ARCompact-specific pass to generate conditional
16864 execution instructions.
16866 Due to delay slot scheduling and interactions between operand numbers,
16867 literal sizes, instruction lengths, and the support for conditional execution,
16868 the target-independent pass to generate conditional execution is often lacking,
16869 so the ARC port has kept a special pass around that tries to find more
16870 conditional execution generation opportunities after register allocation,
16871 branch shortening, and delay slot scheduling have been done. This pass
16872 generally, but not always, improves performance and code size, at the cost of
16873 extra compilation time, which is why there is an option to switch it off.
16874 If you have a problem with call instructions exceeding their allowable
16875 offset range because they are conditionalized, you should consider using
16876 @option{-mmedium-calls} instead.
16878 @item -mearly-cbranchsi
16879 @opindex mearly-cbranchsi
16880 Enable pre-reload use of the @code{cbranchsi} pattern.
16882 @item -mexpand-adddi
16883 @opindex mexpand-adddi
16884 Expand @code{adddi3} and @code{subdi3} at RTL generation time into
16885 @code{add.f}, @code{adc} etc. This option is deprecated.
16887 @item -mindexed-loads
16888 @opindex mindexed-loads
16889 Enable the use of indexed loads. This can be problematic because some
16890 optimizers then assume that indexed stores exist, which is not
16895 Enable Local Register Allocation. This is still experimental for ARC,
16896 so by default the compiler uses standard reload
16897 (i.e.@: @option{-mno-lra}).
16899 @item -mlra-priority-none
16900 @opindex mlra-priority-none
16901 Don't indicate any priority for target registers.
16903 @item -mlra-priority-compact
16904 @opindex mlra-priority-compact
16905 Indicate target register priority for r0..r3 / r12..r15.
16907 @item -mlra-priority-noncompact
16908 @opindex mlra-priority-noncompact
16909 Reduce target register priority for r0..r3 / r12..r15.
16912 @opindex mmillicode
16913 When optimizing for size (using @option{-Os}), prologues and epilogues
16914 that have to save or restore a large number of registers are often
16915 shortened by using call to a special function in libgcc; this is
16916 referred to as a @emph{millicode} call. As these calls can pose
16917 performance issues, and/or cause linking issues when linking in a
16918 nonstandard way, this option is provided to turn on or off millicode
16921 @item -mcode-density-frame
16922 @opindex mcode-density-frame
16923 This option enable the compiler to emit @code{enter} and @code{leave}
16924 instructions. These instructions are only valid for CPUs with
16925 code-density feature.
16928 @opindex mmixed-code
16929 Tweak register allocation to help 16-bit instruction generation.
16930 This generally has the effect of decreasing the average instruction size
16931 while increasing the instruction count.
16935 Enable @samp{q} instruction alternatives.
16936 This is the default for @option{-Os}.
16940 Enable @samp{Rcq} constraint handling.
16941 Most short code generation depends on this.
16942 This is the default.
16946 Enable @samp{Rcw} constraint handling.
16947 Most ccfsm condexec mostly depends on this.
16948 This is the default.
16950 @item -msize-level=@var{level}
16951 @opindex msize-level
16952 Fine-tune size optimization with regards to instruction lengths and alignment.
16953 The recognized values for @var{level} are:
16956 No size optimization. This level is deprecated and treated like @samp{1}.
16959 Short instructions are used opportunistically.
16962 In addition, alignment of loops and of code after barriers are dropped.
16965 In addition, optional data alignment is dropped, and the option @option{Os} is enabled.
16969 This defaults to @samp{3} when @option{-Os} is in effect. Otherwise,
16970 the behavior when this is not set is equivalent to level @samp{1}.
16972 @item -mtune=@var{cpu}
16974 Set instruction scheduling parameters for @var{cpu}, overriding any implied
16975 by @option{-mcpu=}.
16977 Supported values for @var{cpu} are
16981 Tune for ARC600 CPU.
16984 Tune for ARC601 CPU.
16987 Tune for ARC700 CPU with standard multiplier block.
16990 Tune for ARC700 CPU with XMAC block.
16993 Tune for ARC725D CPU.
16996 Tune for ARC750D CPU.
17000 @item -mmultcost=@var{num}
17002 Cost to assume for a multiply instruction, with @samp{4} being equal to a
17003 normal instruction.
17005 @item -munalign-prob-threshold=@var{probability}
17006 @opindex munalign-prob-threshold
17007 Set probability threshold for unaligning branches.
17008 When tuning for @samp{ARC700} and optimizing for speed, branches without
17009 filled delay slot are preferably emitted unaligned and long, unless
17010 profiling indicates that the probability for the branch to be taken
17011 is below @var{probability}. @xref{Cross-profiling}.
17012 The default is (REG_BR_PROB_BASE/2), i.e.@: 5000.
17016 The following options are maintained for backward compatibility, but
17017 are now deprecated and will be removed in a future release:
17019 @c Deprecated options
17027 @opindex mbig-endian
17030 Compile code for big-endian targets. Use of these options is now
17031 deprecated. Big-endian code is supported by configuring GCC to build
17032 @w{@code{arceb-elf32}} and @w{@code{arceb-linux-uclibc}} targets,
17033 for which big endian is the default.
17035 @item -mlittle-endian
17036 @opindex mlittle-endian
17039 Compile code for little-endian targets. Use of these options is now
17040 deprecated. Little-endian code is supported by configuring GCC to build
17041 @w{@code{arc-elf32}} and @w{@code{arc-linux-uclibc}} targets,
17042 for which little endian is the default.
17044 @item -mbarrel_shifter
17045 @opindex mbarrel_shifter
17046 Replaced by @option{-mbarrel-shifter}.
17048 @item -mdpfp_compact
17049 @opindex mdpfp_compact
17050 Replaced by @option{-mdpfp-compact}.
17053 @opindex mdpfp_fast
17054 Replaced by @option{-mdpfp-fast}.
17057 @opindex mdsp_packa
17058 Replaced by @option{-mdsp-packa}.
17062 Replaced by @option{-mea}.
17066 Replaced by @option{-mmac-24}.
17070 Replaced by @option{-mmac-d16}.
17072 @item -mspfp_compact
17073 @opindex mspfp_compact
17074 Replaced by @option{-mspfp-compact}.
17077 @opindex mspfp_fast
17078 Replaced by @option{-mspfp-fast}.
17080 @item -mtune=@var{cpu}
17082 Values @samp{arc600}, @samp{arc601}, @samp{arc700} and
17083 @samp{arc700-xmac} for @var{cpu} are replaced by @samp{ARC600},
17084 @samp{ARC601}, @samp{ARC700} and @samp{ARC700-xmac} respectively.
17086 @item -multcost=@var{num}
17088 Replaced by @option{-mmultcost}.
17093 @subsection ARM Options
17094 @cindex ARM options
17096 These @samp{-m} options are defined for the ARM port:
17099 @item -mabi=@var{name}
17101 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
17102 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
17105 @opindex mapcs-frame
17106 Generate a stack frame that is compliant with the ARM Procedure Call
17107 Standard for all functions, even if this is not strictly necessary for
17108 correct execution of the code. Specifying @option{-fomit-frame-pointer}
17109 with this option causes the stack frames not to be generated for
17110 leaf functions. The default is @option{-mno-apcs-frame}.
17111 This option is deprecated.
17115 This is a synonym for @option{-mapcs-frame} and is deprecated.
17118 @c not currently implemented
17119 @item -mapcs-stack-check
17120 @opindex mapcs-stack-check
17121 Generate code to check the amount of stack space available upon entry to
17122 every function (that actually uses some stack space). If there is
17123 insufficient space available then either the function
17124 @code{__rt_stkovf_split_small} or @code{__rt_stkovf_split_big} is
17125 called, depending upon the amount of stack space required. The runtime
17126 system is required to provide these functions. The default is
17127 @option{-mno-apcs-stack-check}, since this produces smaller code.
17129 @c not currently implemented
17130 @item -mapcs-reentrant
17131 @opindex mapcs-reentrant
17132 Generate reentrant, position-independent code. The default is
17133 @option{-mno-apcs-reentrant}.
17136 @item -mthumb-interwork
17137 @opindex mthumb-interwork
17138 Generate code that supports calling between the ARM and Thumb
17139 instruction sets. Without this option, on pre-v5 architectures, the
17140 two instruction sets cannot be reliably used inside one program. The
17141 default is @option{-mno-thumb-interwork}, since slightly larger code
17142 is generated when @option{-mthumb-interwork} is specified. In AAPCS
17143 configurations this option is meaningless.
17145 @item -mno-sched-prolog
17146 @opindex mno-sched-prolog
17147 @opindex msched-prolog
17148 Prevent the reordering of instructions in the function prologue, or the
17149 merging of those instruction with the instructions in the function's
17150 body. This means that all functions start with a recognizable set
17151 of instructions (or in fact one of a choice from a small set of
17152 different function prologues), and this information can be used to
17153 locate the start of functions inside an executable piece of code. The
17154 default is @option{-msched-prolog}.
17156 @item -mfloat-abi=@var{name}
17157 @opindex mfloat-abi
17158 Specifies which floating-point ABI to use. Permissible values
17159 are: @samp{soft}, @samp{softfp} and @samp{hard}.
17161 Specifying @samp{soft} causes GCC to generate output containing
17162 library calls for floating-point operations.
17163 @samp{softfp} allows the generation of code using hardware floating-point
17164 instructions, but still uses the soft-float calling conventions.
17165 @samp{hard} allows generation of floating-point instructions
17166 and uses FPU-specific calling conventions.
17168 The default depends on the specific target configuration. Note that
17169 the hard-float and soft-float ABIs are not link-compatible; you must
17170 compile your entire program with the same ABI, and link with a
17171 compatible set of libraries.
17173 @item -mgeneral-regs-only
17174 @opindex mgeneral-regs-only
17175 Generate code which uses only the general-purpose registers. This will prevent
17176 the compiler from using floating-point and Advanced SIMD registers but will not
17177 impose any restrictions on the assembler.
17179 @item -mlittle-endian
17180 @opindex mlittle-endian
17181 Generate code for a processor running in little-endian mode. This is
17182 the default for all standard configurations.
17185 @opindex mbig-endian
17186 Generate code for a processor running in big-endian mode; the default is
17187 to compile code for a little-endian processor.
17192 When linking a big-endian image select between BE8 and BE32 formats.
17193 The option has no effect for little-endian images and is ignored. The
17194 default is dependent on the selected target architecture. For ARMv6
17195 and later architectures the default is BE8, for older architectures
17196 the default is BE32. BE32 format has been deprecated by ARM.
17198 @item -march=@var{name}@r{[}+extension@dots{}@r{]}
17200 This specifies the name of the target ARM architecture. GCC uses this
17201 name to determine what kind of instructions it can emit when generating
17202 assembly code. This option can be used in conjunction with or instead
17203 of the @option{-mcpu=} option.
17205 Permissible names are:
17207 @samp{armv5t}, @samp{armv5te},
17208 @samp{armv6}, @samp{armv6j}, @samp{armv6k}, @samp{armv6kz}, @samp{armv6t2},
17209 @samp{armv6z}, @samp{armv6zk},
17210 @samp{armv7}, @samp{armv7-a}, @samp{armv7ve},
17211 @samp{armv8-a}, @samp{armv8.1-a}, @samp{armv8.2-a}, @samp{armv8.3-a},
17216 @samp{armv6-m}, @samp{armv6s-m},
17217 @samp{armv7-m}, @samp{armv7e-m},
17218 @samp{armv8-m.base}, @samp{armv8-m.main},
17219 @samp{iwmmxt} and @samp{iwmmxt2}.
17221 Additionally, the following architectures, which lack support for the
17222 Thumb execution state, are recognized but support is deprecated: @samp{armv4}.
17224 Many of the architectures support extensions. These can be added by
17225 appending @samp{+@var{extension}} to the architecture name. Extension
17226 options are processed in order and capabilities accumulate. An extension
17227 will also enable any necessary base extensions
17228 upon which it depends. For example, the @samp{+crypto} extension
17229 will always enable the @samp{+simd} extension. The exception to the
17230 additive construction is for extensions that are prefixed with
17231 @samp{+no@dots{}}: these extensions disable the specified option and
17232 any other extensions that may depend on the presence of that
17235 For example, @samp{-march=armv7-a+simd+nofp+vfpv4} is equivalent to
17236 writing @samp{-march=armv7-a+vfpv4} since the @samp{+simd} option is
17237 entirely disabled by the @samp{+nofp} option that follows it.
17239 Most extension names are generically named, but have an effect that is
17240 dependent upon the architecture to which it is applied. For example,
17241 the @samp{+simd} option can be applied to both @samp{armv7-a} and
17242 @samp{armv8-a} architectures, but will enable the original ARMv7-A
17243 Advanced SIMD (Neon) extensions for @samp{armv7-a} and the ARMv8-A
17244 variant for @samp{armv8-a}.
17246 The table below lists the supported extensions for each architecture.
17247 Architectures not mentioned do not support any extensions.
17260 The VFPv2 floating-point instructions. The extension @samp{+vfpv2} can be
17261 used as an alias for this extension.
17264 Disable the floating-point instructions.
17268 The common subset of the ARMv7-A, ARMv7-R and ARMv7-M architectures.
17271 The VFPv3 floating-point instructions, with 16 double-precision
17272 registers. The extension @samp{+vfpv3-d16} can be used as an alias
17273 for this extension. Note that floating-point is not supported by the
17274 base ARMv7-M architecture, but is compatible with both the ARMv7-A and
17275 ARMv7-R architectures.
17278 Disable the floating-point instructions.
17284 The multiprocessing extension.
17287 The security extension.
17290 The VFPv3 floating-point instructions, with 16 double-precision
17291 registers. The extension @samp{+vfpv3-d16} can be used as an alias
17292 for this extension.
17295 The Advanced SIMD (Neon) v1 and the VFPv3 floating-point instructions.
17296 The extensions @samp{+neon} and @samp{+neon-vfpv3} can be used as aliases
17297 for this extension.
17300 The VFPv3 floating-point instructions, with 32 double-precision
17303 @item +vfpv3-d16-fp16
17304 The VFPv3 floating-point instructions, with 16 double-precision
17305 registers and the half-precision floating-point conversion operations.
17308 The VFPv3 floating-point instructions, with 32 double-precision
17309 registers and the half-precision floating-point conversion operations.
17312 The VFPv4 floating-point instructions, with 16 double-precision
17316 The VFPv4 floating-point instructions, with 32 double-precision
17320 The Advanced SIMD (Neon) v1 and the VFPv3 floating-point instructions, with
17321 the half-precision floating-point conversion operations.
17324 The Advanced SIMD (Neon) v2 and the VFPv4 floating-point instructions.
17327 Disable the Advanced SIMD instructions (does not disable floating point).
17330 Disable the floating-point and Advanced SIMD instructions.
17334 The extended version of the ARMv7-A architecture with support for
17338 The VFPv4 floating-point instructions, with 16 double-precision registers.
17339 The extension @samp{+vfpv4-d16} can be used as an alias for this extension.
17342 The Advanced SIMD (Neon) v2 and the VFPv4 floating-point instructions. The
17343 extension @samp{+neon-vfpv4} can be used as an alias for this extension.
17346 The VFPv3 floating-point instructions, with 16 double-precision
17350 The VFPv3 floating-point instructions, with 32 double-precision
17353 @item +vfpv3-d16-fp16
17354 The VFPv3 floating-point instructions, with 16 double-precision
17355 registers and the half-precision floating-point conversion operations.
17358 The VFPv3 floating-point instructions, with 32 double-precision
17359 registers and the half-precision floating-point conversion operations.
17362 The VFPv4 floating-point instructions, with 16 double-precision
17366 The VFPv4 floating-point instructions, with 32 double-precision
17370 The Advanced SIMD (Neon) v1 and the VFPv3 floating-point instructions.
17371 The extension @samp{+neon-vfpv3} can be used as an alias for this extension.
17374 The Advanced SIMD (Neon) v1 and the VFPv3 floating-point instructions, with
17375 the half-precision floating-point conversion operations.
17378 Disable the Advanced SIMD instructions (does not disable floating point).
17381 Disable the floating-point and Advanced SIMD instructions.
17387 The Cyclic Redundancy Check (CRC) instructions.
17389 The ARMv8-A Advanced SIMD and floating-point instructions.
17391 The cryptographic instructions.
17393 Disable the cryptographic instructions.
17395 Disable the floating-point, Advanced SIMD and cryptographic instructions.
17397 Speculation Barrier Instruction.
17399 Execution and Data Prediction Restriction Instructions.
17405 The ARMv8.1-A Advanced SIMD and floating-point instructions.
17408 The cryptographic instructions. This also enables the Advanced SIMD and
17409 floating-point instructions.
17412 Disable the cryptographic instructions.
17415 Disable the floating-point, Advanced SIMD and cryptographic instructions.
17418 Speculation Barrier Instruction.
17421 Execution and Data Prediction Restriction Instructions.
17428 The half-precision floating-point data processing instructions.
17429 This also enables the Advanced SIMD and floating-point instructions.
17432 The half-precision floating-point fmla extension. This also enables
17433 the half-precision floating-point extension and Advanced SIMD and
17434 floating-point instructions.
17437 The ARMv8.1-A Advanced SIMD and floating-point instructions.
17440 The cryptographic instructions. This also enables the Advanced SIMD and
17441 floating-point instructions.
17444 Enable the Dot Product extension. This also enables Advanced SIMD instructions.
17447 Disable the cryptographic extension.
17450 Disable the floating-point, Advanced SIMD and cryptographic instructions.
17453 Speculation Barrier Instruction.
17456 Execution and Data Prediction Restriction Instructions.
17462 The half-precision floating-point data processing instructions.
17463 This also enables the Advanced SIMD and floating-point instructions as well
17464 as the Dot Product extension and the half-precision floating-point fmla
17468 The ARMv8.3-A Advanced SIMD and floating-point instructions as well as the
17469 Dot Product extension.
17472 The cryptographic instructions. This also enables the Advanced SIMD and
17473 floating-point instructions as well as the Dot Product extension.
17476 Disable the cryptographic extension.
17479 Disable the floating-point, Advanced SIMD and cryptographic instructions.
17482 Speculation Barrier Instruction.
17485 Execution and Data Prediction Restriction Instructions.
17491 The half-precision floating-point data processing instructions.
17492 This also enables the Advanced SIMD and floating-point instructions as well
17493 as the Dot Product extension and the half-precision floating-point fmla
17497 The ARMv8.3-A Advanced SIMD and floating-point instructions as well as the
17498 Dot Product extension.
17501 The cryptographic instructions. This also enables the Advanced SIMD and
17502 floating-point instructions as well as the Dot Product extension.
17505 Disable the cryptographic extension.
17508 Disable the floating-point, Advanced SIMD and cryptographic instructions.
17514 The single-precision VFPv3 floating-point instructions. The extension
17515 @samp{+vfpv3xd} can be used as an alias for this extension.
17518 The VFPv3 floating-point instructions with 16 double-precision registers.
17519 The extension +vfpv3-d16 can be used as an alias for this extension.
17521 @item +vfpv3xd-d16-fp16
17522 The single-precision VFPv3 floating-point instructions with 16 double-precision
17523 registers and the half-precision floating-point conversion operations.
17525 @item +vfpv3-d16-fp16
17526 The VFPv3 floating-point instructions with 16 double-precision
17527 registers and the half-precision floating-point conversion operations.
17530 Disable the floating-point extension.
17533 The ARM-state integer division instructions.
17536 Disable the ARM-state integer division extension.
17542 The single-precision VFPv4 floating-point instructions.
17545 The single-precision FPv5 floating-point instructions.
17548 The single- and double-precision FPv5 floating-point instructions.
17551 Disable the floating-point extensions.
17557 The DSP instructions.
17560 Disable the DSP extension.
17563 The single-precision floating-point instructions.
17566 The single- and double-precision floating-point instructions.
17569 Disable the floating-point extension.
17575 The Cyclic Redundancy Check (CRC) instructions.
17577 The single-precision FPv5 floating-point instructions.
17579 The ARMv8-A Advanced SIMD and floating-point instructions.
17581 The cryptographic instructions.
17583 Disable the cryptographic instructions.
17585 Disable the floating-point, Advanced SIMD and cryptographic instructions.
17590 @option{-march=native} causes the compiler to auto-detect the architecture
17591 of the build computer. At present, this feature is only supported on
17592 GNU/Linux, and not all architectures are recognized. If the auto-detect
17593 is unsuccessful the option has no effect.
17595 @item -mtune=@var{name}
17597 This option specifies the name of the target ARM processor for
17598 which GCC should tune the performance of the code.
17599 For some ARM implementations better performance can be obtained by using
17601 Permissible names are: @samp{arm7tdmi}, @samp{arm7tdmi-s}, @samp{arm710t},
17602 @samp{arm720t}, @samp{arm740t}, @samp{strongarm}, @samp{strongarm110},
17603 @samp{strongarm1100}, 0@samp{strongarm1110}, @samp{arm8}, @samp{arm810},
17604 @samp{arm9}, @samp{arm9e}, @samp{arm920}, @samp{arm920t}, @samp{arm922t},
17605 @samp{arm946e-s}, @samp{arm966e-s}, @samp{arm968e-s}, @samp{arm926ej-s},
17606 @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi}, @samp{arm1020t},
17607 @samp{arm1026ej-s}, @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
17608 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
17609 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
17610 @samp{generic-armv7-a}, @samp{cortex-a5}, @samp{cortex-a7}, @samp{cortex-a8},
17611 @samp{cortex-a9}, @samp{cortex-a12}, @samp{cortex-a15}, @samp{cortex-a17},
17612 @samp{cortex-a32}, @samp{cortex-a35}, @samp{cortex-a53}, @samp{cortex-a55},
17613 @samp{cortex-a57}, @samp{cortex-a72}, @samp{cortex-a73}, @samp{cortex-a75},
17614 @samp{cortex-a76}, @samp{ares}, @samp{cortex-r4}, @samp{cortex-r4f},
17615 @samp{cortex-r5}, @samp{cortex-r7}, @samp{cortex-r8}, @samp{cortex-r52},
17616 @samp{cortex-m0}, @samp{cortex-m0plus}, @samp{cortex-m1}, @samp{cortex-m3},
17617 @samp{cortex-m4}, @samp{cortex-m7}, @samp{cortex-m23}, @samp{cortex-m33},
17618 @samp{cortex-m1.small-multiply}, @samp{cortex-m0.small-multiply},
17619 @samp{cortex-m0plus.small-multiply}, @samp{exynos-m1}, @samp{marvell-pj4},
17620 @samp{neoverse-n1}, @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2},
17621 @samp{ep9312}, @samp{fa526}, @samp{fa626}, @samp{fa606te}, @samp{fa626te},
17622 @samp{fmp626}, @samp{fa726te}, @samp{xgene1}.
17624 Additionally, this option can specify that GCC should tune the performance
17625 of the code for a big.LITTLE system. Permissible names are:
17626 @samp{cortex-a15.cortex-a7}, @samp{cortex-a17.cortex-a7},
17627 @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53},
17628 @samp{cortex-a72.cortex-a35}, @samp{cortex-a73.cortex-a53},
17629 @samp{cortex-a75.cortex-a55}, @samp{cortex-a76.cortex-a55}.
17631 @option{-mtune=generic-@var{arch}} specifies that GCC should tune the
17632 performance for a blend of processors within architecture @var{arch}.
17633 The aim is to generate code that run well on the current most popular
17634 processors, balancing between optimizations that benefit some CPUs in the
17635 range, and avoiding performance pitfalls of other CPUs. The effects of
17636 this option may change in future GCC versions as CPU models come and go.
17638 @option{-mtune} permits the same extension options as @option{-mcpu}, but
17639 the extension options do not affect the tuning of the generated code.
17641 @option{-mtune=native} causes the compiler to auto-detect the CPU
17642 of the build computer. At present, this feature is only supported on
17643 GNU/Linux, and not all architectures are recognized. If the auto-detect is
17644 unsuccessful the option has no effect.
17646 @item -mcpu=@var{name}@r{[}+extension@dots{}@r{]}
17648 This specifies the name of the target ARM processor. GCC uses this name
17649 to derive the name of the target ARM architecture (as if specified
17650 by @option{-march}) and the ARM processor type for which to tune for
17651 performance (as if specified by @option{-mtune}). Where this option
17652 is used in conjunction with @option{-march} or @option{-mtune},
17653 those options take precedence over the appropriate part of this option.
17655 Many of the supported CPUs implement optional architectural
17656 extensions. Where this is so the architectural extensions are
17657 normally enabled by default. If implementations that lack the
17658 extension exist, then the extension syntax can be used to disable
17659 those extensions that have been omitted. For floating-point and
17660 Advanced SIMD (Neon) instructions, the settings of the options
17661 @option{-mfloat-abi} and @option{-mfpu} must also be considered:
17662 floating-point and Advanced SIMD instructions will only be used if
17663 @option{-mfloat-abi} is not set to @samp{soft}; and any setting of
17664 @option{-mfpu} other than @samp{auto} will override the available
17665 floating-point and SIMD extension instructions.
17667 For example, @samp{cortex-a9} can be found in three major
17668 configurations: integer only, with just a floating-point unit or with
17669 floating-point and Advanced SIMD. The default is to enable all the
17670 instructions, but the extensions @samp{+nosimd} and @samp{+nofp} can
17671 be used to disable just the SIMD or both the SIMD and floating-point
17672 instructions respectively.
17674 Permissible names for this option are the same as those for
17677 The following extension options are common to the listed CPUs:
17681 Disable the DSP instructions on @samp{cortex-m33}.
17684 Disables the floating-point instructions on @samp{arm9e},
17685 @samp{arm946e-s}, @samp{arm966e-s}, @samp{arm968e-s}, @samp{arm10e},
17686 @samp{arm1020e}, @samp{arm1022e}, @samp{arm926ej-s},
17687 @samp{arm1026ej-s}, @samp{cortex-r5}, @samp{cortex-r7}, @samp{cortex-r8},
17688 @samp{cortex-m4}, @samp{cortex-m7} and @samp{cortex-m33}.
17689 Disables the floating-point and SIMD instructions on
17690 @samp{generic-armv7-a}, @samp{cortex-a5}, @samp{cortex-a7},
17691 @samp{cortex-a8}, @samp{cortex-a9}, @samp{cortex-a12},
17692 @samp{cortex-a15}, @samp{cortex-a17}, @samp{cortex-a15.cortex-a7},
17693 @samp{cortex-a17.cortex-a7}, @samp{cortex-a32}, @samp{cortex-a35},
17694 @samp{cortex-a53} and @samp{cortex-a55}.
17697 Disables the double-precision component of the floating-point instructions
17698 on @samp{cortex-r5}, @samp{cortex-r7}, @samp{cortex-r8}, @samp{cortex-r52} and
17702 Disables the SIMD (but not floating-point) instructions on
17703 @samp{generic-armv7-a}, @samp{cortex-a5}, @samp{cortex-a7}
17704 and @samp{cortex-a9}.
17707 Enables the cryptographic instructions on @samp{cortex-a32},
17708 @samp{cortex-a35}, @samp{cortex-a53}, @samp{cortex-a55}, @samp{cortex-a57},
17709 @samp{cortex-a72}, @samp{cortex-a73}, @samp{cortex-a75}, @samp{exynos-m1},
17710 @samp{xgene1}, @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53},
17711 @samp{cortex-a73.cortex-a35}, @samp{cortex-a73.cortex-a53} and
17712 @samp{cortex-a75.cortex-a55}.
17715 Additionally the @samp{generic-armv7-a} pseudo target defaults to
17716 VFPv3 with 16 double-precision registers. It supports the following
17717 extension options: @samp{mp}, @samp{sec}, @samp{vfpv3-d16},
17718 @samp{vfpv3}, @samp{vfpv3-d16-fp16}, @samp{vfpv3-fp16},
17719 @samp{vfpv4-d16}, @samp{vfpv4}, @samp{neon}, @samp{neon-vfpv3},
17720 @samp{neon-fp16}, @samp{neon-vfpv4}. The meanings are the same as for
17721 the extensions to @option{-march=armv7-a}.
17723 @option{-mcpu=generic-@var{arch}} is also permissible, and is
17724 equivalent to @option{-march=@var{arch} -mtune=generic-@var{arch}}.
17725 See @option{-mtune} for more information.
17727 @option{-mcpu=native} causes the compiler to auto-detect the CPU
17728 of the build computer. At present, this feature is only supported on
17729 GNU/Linux, and not all architectures are recognized. If the auto-detect
17730 is unsuccessful the option has no effect.
17732 @item -mfpu=@var{name}
17734 This specifies what floating-point hardware (or hardware emulation) is
17735 available on the target. Permissible names are: @samp{auto}, @samp{vfpv2},
17737 @samp{vfpv3-fp16}, @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd},
17738 @samp{vfpv3xd-fp16}, @samp{neon-vfpv3}, @samp{neon-fp16}, @samp{vfpv4},
17739 @samp{vfpv4-d16}, @samp{fpv4-sp-d16}, @samp{neon-vfpv4},
17740 @samp{fpv5-d16}, @samp{fpv5-sp-d16},
17741 @samp{fp-armv8}, @samp{neon-fp-armv8} and @samp{crypto-neon-fp-armv8}.
17742 Note that @samp{neon} is an alias for @samp{neon-vfpv3} and @samp{vfp}
17743 is an alias for @samp{vfpv2}.
17745 The setting @samp{auto} is the default and is special. It causes the
17746 compiler to select the floating-point and Advanced SIMD instructions
17747 based on the settings of @option{-mcpu} and @option{-march}.
17749 If the selected floating-point hardware includes the NEON extension
17750 (e.g.@: @option{-mfpu=neon}), note that floating-point
17751 operations are not generated by GCC's auto-vectorization pass unless
17752 @option{-funsafe-math-optimizations} is also specified. This is
17753 because NEON hardware does not fully implement the IEEE 754 standard for
17754 floating-point arithmetic (in particular denormal values are treated as
17755 zero), so the use of NEON instructions may lead to a loss of precision.
17757 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}).
17759 @item -mfp16-format=@var{name}
17760 @opindex mfp16-format
17761 Specify the format of the @code{__fp16} half-precision floating-point type.
17762 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
17763 the default is @samp{none}, in which case the @code{__fp16} type is not
17764 defined. @xref{Half-Precision}, for more information.
17766 @item -mstructure-size-boundary=@var{n}
17767 @opindex mstructure-size-boundary
17768 The sizes of all structures and unions are rounded up to a multiple
17769 of the number of bits set by this option. Permissible values are 8, 32
17770 and 64. The default value varies for different toolchains. For the COFF
17771 targeted toolchain the default value is 8. A value of 64 is only allowed
17772 if the underlying ABI supports it.
17774 Specifying a larger number can produce faster, more efficient code, but
17775 can also increase the size of the program. Different values are potentially
17776 incompatible. Code compiled with one value cannot necessarily expect to
17777 work with code or libraries compiled with another value, if they exchange
17778 information using structures or unions.
17780 This option is deprecated.
17782 @item -mabort-on-noreturn
17783 @opindex mabort-on-noreturn
17784 Generate a call to the function @code{abort} at the end of a
17785 @code{noreturn} function. It is executed if the function tries to
17789 @itemx -mno-long-calls
17790 @opindex mlong-calls
17791 @opindex mno-long-calls
17792 Tells the compiler to perform function calls by first loading the
17793 address of the function into a register and then performing a subroutine
17794 call on this register. This switch is needed if the target function
17795 lies outside of the 64-megabyte addressing range of the offset-based
17796 version of subroutine call instruction.
17798 Even if this switch is enabled, not all function calls are turned
17799 into long calls. The heuristic is that static functions, functions
17800 that have the @code{short_call} attribute, functions that are inside
17801 the scope of a @code{#pragma no_long_calls} directive, and functions whose
17802 definitions have already been compiled within the current compilation
17803 unit are not turned into long calls. The exceptions to this rule are
17804 that weak function definitions, functions with the @code{long_call}
17805 attribute or the @code{section} attribute, and functions that are within
17806 the scope of a @code{#pragma long_calls} directive are always
17807 turned into long calls.
17809 This feature is not enabled by default. Specifying
17810 @option{-mno-long-calls} restores the default behavior, as does
17811 placing the function calls within the scope of a @code{#pragma
17812 long_calls_off} directive. Note these switches have no effect on how
17813 the compiler generates code to handle function calls via function
17816 @item -msingle-pic-base
17817 @opindex msingle-pic-base
17818 Treat the register used for PIC addressing as read-only, rather than
17819 loading it in the prologue for each function. The runtime system is
17820 responsible for initializing this register with an appropriate value
17821 before execution begins.
17823 @item -mpic-register=@var{reg}
17824 @opindex mpic-register
17825 Specify the register to be used for PIC addressing.
17826 For standard PIC base case, the default is any suitable register
17827 determined by compiler. For single PIC base case, the default is
17828 @samp{R9} if target is EABI based or stack-checking is enabled,
17829 otherwise the default is @samp{R10}.
17831 @item -mpic-data-is-text-relative
17832 @opindex mpic-data-is-text-relative
17833 Assume that the displacement between the text and data segments is fixed
17834 at static link time. This permits using PC-relative addressing
17835 operations to access data known to be in the data segment. For
17836 non-VxWorks RTP targets, this option is enabled by default. When
17837 disabled on such targets, it will enable @option{-msingle-pic-base} by
17840 @item -mpoke-function-name
17841 @opindex mpoke-function-name
17842 Write the name of each function into the text section, directly
17843 preceding the function prologue. The generated code is similar to this:
17847 .ascii "arm_poke_function_name", 0
17850 .word 0xff000000 + (t1 - t0)
17851 arm_poke_function_name
17853 stmfd sp!, @{fp, ip, lr, pc@}
17857 When performing a stack backtrace, code can inspect the value of
17858 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
17859 location @code{pc - 12} and the top 8 bits are set, then we know that
17860 there is a function name embedded immediately preceding this location
17861 and has length @code{((pc[-3]) & 0xff000000)}.
17868 Select between generating code that executes in ARM and Thumb
17869 states. The default for most configurations is to generate code
17870 that executes in ARM state, but the default can be changed by
17871 configuring GCC with the @option{--with-mode=}@var{state}
17874 You can also override the ARM and Thumb mode for each function
17875 by using the @code{target("thumb")} and @code{target("arm")} function attributes
17876 (@pxref{ARM Function Attributes}) or pragmas (@pxref{Function Specific Option Pragmas}).
17879 @opindex mflip-thumb
17880 Switch ARM/Thumb modes on alternating functions.
17881 This option is provided for regression testing of mixed Thumb/ARM code
17882 generation, and is not intended for ordinary use in compiling code.
17885 @opindex mtpcs-frame
17886 Generate a stack frame that is compliant with the Thumb Procedure Call
17887 Standard for all non-leaf functions. (A leaf function is one that does
17888 not call any other functions.) The default is @option{-mno-tpcs-frame}.
17890 @item -mtpcs-leaf-frame
17891 @opindex mtpcs-leaf-frame
17892 Generate a stack frame that is compliant with the Thumb Procedure Call
17893 Standard for all leaf functions. (A leaf function is one that does
17894 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
17896 @item -mcallee-super-interworking
17897 @opindex mcallee-super-interworking
17898 Gives all externally visible functions in the file being compiled an ARM
17899 instruction set header which switches to Thumb mode before executing the
17900 rest of the function. This allows these functions to be called from
17901 non-interworking code. This option is not valid in AAPCS configurations
17902 because interworking is enabled by default.
17904 @item -mcaller-super-interworking
17905 @opindex mcaller-super-interworking
17906 Allows calls via function pointers (including virtual functions) to
17907 execute correctly regardless of whether the target code has been
17908 compiled for interworking or not. There is a small overhead in the cost
17909 of executing a function pointer if this option is enabled. This option
17910 is not valid in AAPCS configurations because interworking is enabled
17913 @item -mtp=@var{name}
17915 Specify the access model for the thread local storage pointer. The valid
17916 models are @samp{soft}, which generates calls to @code{__aeabi_read_tp},
17917 @samp{cp15}, which fetches the thread pointer from @code{cp15} directly
17918 (supported in the arm6k architecture), and @samp{auto}, which uses the
17919 best available method for the selected processor. The default setting is
17922 @item -mtls-dialect=@var{dialect}
17923 @opindex mtls-dialect
17924 Specify the dialect to use for accessing thread local storage. Two
17925 @var{dialect}s are supported---@samp{gnu} and @samp{gnu2}. The
17926 @samp{gnu} dialect selects the original GNU scheme for supporting
17927 local and global dynamic TLS models. The @samp{gnu2} dialect
17928 selects the GNU descriptor scheme, which provides better performance
17929 for shared libraries. The GNU descriptor scheme is compatible with
17930 the original scheme, but does require new assembler, linker and
17931 library support. Initial and local exec TLS models are unaffected by
17932 this option and always use the original scheme.
17934 @item -mword-relocations
17935 @opindex mword-relocations
17936 Only generate absolute relocations on word-sized values (i.e.@: R_ARM_ABS32).
17937 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
17938 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
17939 is specified. This option conflicts with @option{-mslow-flash-data}.
17941 @item -mfix-cortex-m3-ldrd
17942 @opindex mfix-cortex-m3-ldrd
17943 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
17944 with overlapping destination and base registers are used. This option avoids
17945 generating these instructions. This option is enabled by default when
17946 @option{-mcpu=cortex-m3} is specified.
17948 @item -munaligned-access
17949 @itemx -mno-unaligned-access
17950 @opindex munaligned-access
17951 @opindex mno-unaligned-access
17952 Enables (or disables) reading and writing of 16- and 32- bit values
17953 from addresses that are not 16- or 32- bit aligned. By default
17954 unaligned access is disabled for all pre-ARMv6, all ARMv6-M and for
17955 ARMv8-M Baseline architectures, and enabled for all other
17956 architectures. If unaligned access is not enabled then words in packed
17957 data structures are accessed a byte at a time.
17959 The ARM attribute @code{Tag_CPU_unaligned_access} is set in the
17960 generated object file to either true or false, depending upon the
17961 setting of this option. If unaligned access is enabled then the
17962 preprocessor symbol @code{__ARM_FEATURE_UNALIGNED} is also
17965 @item -mneon-for-64bits
17966 @opindex mneon-for-64bits
17967 Enables using Neon to handle scalar 64-bits operations. This is
17968 disabled by default since the cost of moving data from core registers
17971 @item -mslow-flash-data
17972 @opindex mslow-flash-data
17973 Assume loading data from flash is slower than fetching instruction.
17974 Therefore literal load is minimized for better performance.
17975 This option is only supported when compiling for ARMv7 M-profile and
17976 off by default. It conflicts with @option{-mword-relocations}.
17978 @item -masm-syntax-unified
17979 @opindex masm-syntax-unified
17980 Assume inline assembler is using unified asm syntax. The default is
17981 currently off which implies divided syntax. This option has no impact
17982 on Thumb2. However, this may change in future releases of GCC.
17983 Divided syntax should be considered deprecated.
17985 @item -mrestrict-it
17986 @opindex mrestrict-it
17987 Restricts generation of IT blocks to conform to the rules of ARMv8-A.
17988 IT blocks can only contain a single 16-bit instruction from a select
17989 set of instructions. This option is on by default for ARMv8-A Thumb mode.
17991 @item -mprint-tune-info
17992 @opindex mprint-tune-info
17993 Print CPU tuning information as comment in assembler file. This is
17994 an option used only for regression testing of the compiler and not
17995 intended for ordinary use in compiling code. This option is disabled
17998 @item -mverbose-cost-dump
17999 @opindex mverbose-cost-dump
18000 Enable verbose cost model dumping in the debug dump files. This option is
18001 provided for use in debugging the compiler.
18004 @opindex mpure-code
18005 Do not allow constant data to be placed in code sections.
18006 Additionally, when compiling for ELF object format give all text sections the
18007 ELF processor-specific section attribute @code{SHF_ARM_PURECODE}. This option
18008 is only available when generating non-pic code for M-profile targets with the
18013 Generate secure code as per the "ARMv8-M Security Extensions: Requirements on
18014 Development Tools Engineering Specification", which can be found on
18015 @url{http://infocenter.arm.com/help/topic/com.arm.doc.ecm0359818/ECM0359818_armv8m_security_extensions_reqs_on_dev_tools_1_0.pdf}.
18019 @subsection AVR Options
18020 @cindex AVR Options
18022 These options are defined for AVR implementations:
18025 @item -mmcu=@var{mcu}
18027 Specify Atmel AVR instruction set architectures (ISA) or MCU type.
18029 The default for this option is@tie{}@samp{avr2}.
18031 GCC supports the following AVR devices and ISAs:
18033 @include avr-mmcu.texi
18038 Assume that all data in static storage can be accessed by LDS / STS
18039 instructions. This option has only an effect on reduced Tiny devices like
18040 ATtiny40. See also the @code{absdata}
18041 @ref{AVR Variable Attributes,variable attribute}.
18043 @item -maccumulate-args
18044 @opindex maccumulate-args
18045 Accumulate outgoing function arguments and acquire/release the needed
18046 stack space for outgoing function arguments once in function
18047 prologue/epilogue. Without this option, outgoing arguments are pushed
18048 before calling a function and popped afterwards.
18050 Popping the arguments after the function call can be expensive on
18051 AVR so that accumulating the stack space might lead to smaller
18052 executables because arguments need not be removed from the
18053 stack after such a function call.
18055 This option can lead to reduced code size for functions that perform
18056 several calls to functions that get their arguments on the stack like
18057 calls to printf-like functions.
18059 @item -mbranch-cost=@var{cost}
18060 @opindex mbranch-cost
18061 Set the branch costs for conditional branch instructions to
18062 @var{cost}. Reasonable values for @var{cost} are small, non-negative
18063 integers. The default branch cost is 0.
18065 @item -mcall-prologues
18066 @opindex mcall-prologues
18067 Functions prologues/epilogues are expanded as calls to appropriate
18068 subroutines. Code size is smaller.
18070 @item -mgas-isr-prologues
18071 @opindex mgas-isr-prologues
18072 Interrupt service routines (ISRs) may use the @code{__gcc_isr} pseudo
18073 instruction supported by GNU Binutils.
18074 If this option is on, the feature can still be disabled for individual
18075 ISRs by means of the @ref{AVR Function Attributes,,@code{no_gccisr}}
18076 function attribute. This feature is activated per default
18077 if optimization is on (but not with @option{-Og}, @pxref{Optimize Options}),
18078 and if GNU Binutils support @w{@uref{https://sourceware.org/PR21683,PR21683}}.
18082 Assume @code{int} to be 8-bit integer. This affects the sizes of all types: a
18083 @code{char} is 1 byte, an @code{int} is 1 byte, a @code{long} is 2 bytes,
18084 and @code{long long} is 4 bytes. Please note that this option does not
18085 conform to the C standards, but it results in smaller code
18088 @item -mmain-is-OS_task
18089 @opindex mmain-is-OS_task
18090 Do not save registers in @code{main}. The effect is the same like
18091 attaching attribute @ref{AVR Function Attributes,,@code{OS_task}}
18092 to @code{main}. It is activated per default if optimization is on.
18094 @item -mn-flash=@var{num}
18096 Assume that the flash memory has a size of
18097 @var{num} times 64@tie{}KiB.
18099 @item -mno-interrupts
18100 @opindex mno-interrupts
18101 Generated code is not compatible with hardware interrupts.
18102 Code size is smaller.
18106 Try to replace @code{CALL} resp.@: @code{JMP} instruction by the shorter
18107 @code{RCALL} resp.@: @code{RJMP} instruction if applicable.
18108 Setting @option{-mrelax} just adds the @option{--mlink-relax} option to
18109 the assembler's command line and the @option{--relax} option to the
18110 linker's command line.
18112 Jump relaxing is performed by the linker because jump offsets are not
18113 known before code is located. Therefore, the assembler code generated by the
18114 compiler is the same, but the instructions in the executable may
18115 differ from instructions in the assembler code.
18117 Relaxing must be turned on if linker stubs are needed, see the
18118 section on @code{EIND} and linker stubs below.
18122 Assume that the device supports the Read-Modify-Write
18123 instructions @code{XCH}, @code{LAC}, @code{LAS} and @code{LAT}.
18125 @item -mshort-calls
18126 @opindex mshort-calls
18128 Assume that @code{RJMP} and @code{RCALL} can target the whole
18131 This option is used internally for multilib selection. It is
18132 not an optimization option, and you don't need to set it by hand.
18136 Treat the stack pointer register as an 8-bit register,
18137 i.e.@: assume the high byte of the stack pointer is zero.
18138 In general, you don't need to set this option by hand.
18140 This option is used internally by the compiler to select and
18141 build multilibs for architectures @code{avr2} and @code{avr25}.
18142 These architectures mix devices with and without @code{SPH}.
18143 For any setting other than @option{-mmcu=avr2} or @option{-mmcu=avr25}
18144 the compiler driver adds or removes this option from the compiler
18145 proper's command line, because the compiler then knows if the device
18146 or architecture has an 8-bit stack pointer and thus no @code{SPH}
18151 Use address register @code{X} in a way proposed by the hardware. This means
18152 that @code{X} is only used in indirect, post-increment or
18153 pre-decrement addressing.
18155 Without this option, the @code{X} register may be used in the same way
18156 as @code{Y} or @code{Z} which then is emulated by additional
18158 For example, loading a value with @code{X+const} addressing with a
18159 small non-negative @code{const < 64} to a register @var{Rn} is
18163 adiw r26, const ; X += const
18164 ld @var{Rn}, X ; @var{Rn} = *X
18165 sbiw r26, const ; X -= const
18169 @opindex mtiny-stack
18170 Only change the lower 8@tie{}bits of the stack pointer.
18172 @item -mfract-convert-truncate
18173 @opindex mfract-convert-truncate
18174 Allow to use truncation instead of rounding towards zero for fractional fixed-point types.
18177 @opindex nodevicelib
18178 Don't link against AVR-LibC's device specific library @code{lib<mcu>.a}.
18180 @item -Waddr-space-convert
18181 @opindex Waddr-space-convert
18182 @opindex Wno-addr-space-convert
18183 Warn about conversions between address spaces in the case where the
18184 resulting address space is not contained in the incoming address space.
18186 @item -Wmisspelled-isr
18187 @opindex Wmisspelled-isr
18188 @opindex Wno-misspelled-isr
18189 Warn if the ISR is misspelled, i.e.@: without __vector prefix.
18190 Enabled by default.
18193 @subsubsection @code{EIND} and Devices with More Than 128 Ki Bytes of Flash
18194 @cindex @code{EIND}
18195 Pointers in the implementation are 16@tie{}bits wide.
18196 The address of a function or label is represented as word address so
18197 that indirect jumps and calls can target any code address in the
18198 range of 64@tie{}Ki words.
18200 In order to facilitate indirect jump on devices with more than 128@tie{}Ki
18201 bytes of program memory space, there is a special function register called
18202 @code{EIND} that serves as most significant part of the target address
18203 when @code{EICALL} or @code{EIJMP} instructions are used.
18205 Indirect jumps and calls on these devices are handled as follows by
18206 the compiler and are subject to some limitations:
18211 The compiler never sets @code{EIND}.
18214 The compiler uses @code{EIND} implicitly in @code{EICALL}/@code{EIJMP}
18215 instructions or might read @code{EIND} directly in order to emulate an
18216 indirect call/jump by means of a @code{RET} instruction.
18219 The compiler assumes that @code{EIND} never changes during the startup
18220 code or during the application. In particular, @code{EIND} is not
18221 saved/restored in function or interrupt service routine
18225 For indirect calls to functions and computed goto, the linker
18226 generates @emph{stubs}. Stubs are jump pads sometimes also called
18227 @emph{trampolines}. Thus, the indirect call/jump jumps to such a stub.
18228 The stub contains a direct jump to the desired address.
18231 Linker relaxation must be turned on so that the linker generates
18232 the stubs correctly in all situations. See the compiler option
18233 @option{-mrelax} and the linker option @option{--relax}.
18234 There are corner cases where the linker is supposed to generate stubs
18235 but aborts without relaxation and without a helpful error message.
18238 The default linker script is arranged for code with @code{EIND = 0}.
18239 If code is supposed to work for a setup with @code{EIND != 0}, a custom
18240 linker script has to be used in order to place the sections whose
18241 name start with @code{.trampolines} into the segment where @code{EIND}
18245 The startup code from libgcc never sets @code{EIND}.
18246 Notice that startup code is a blend of code from libgcc and AVR-LibC.
18247 For the impact of AVR-LibC on @code{EIND}, see the
18248 @w{@uref{http://nongnu.org/avr-libc/user-manual/,AVR-LibC user manual}}.
18251 It is legitimate for user-specific startup code to set up @code{EIND}
18252 early, for example by means of initialization code located in
18253 section @code{.init3}. Such code runs prior to general startup code
18254 that initializes RAM and calls constructors, but after the bit
18255 of startup code from AVR-LibC that sets @code{EIND} to the segment
18256 where the vector table is located.
18258 #include <avr/io.h>
18261 __attribute__((section(".init3"),naked,used,no_instrument_function))
18262 init3_set_eind (void)
18264 __asm volatile ("ldi r24,pm_hh8(__trampolines_start)\n\t"
18265 "out %i0,r24" :: "n" (&EIND) : "r24","memory");
18270 The @code{__trampolines_start} symbol is defined in the linker script.
18273 Stubs are generated automatically by the linker if
18274 the following two conditions are met:
18277 @item The address of a label is taken by means of the @code{gs} modifier
18278 (short for @emph{generate stubs}) like so:
18280 LDI r24, lo8(gs(@var{func}))
18281 LDI r25, hi8(gs(@var{func}))
18283 @item The final location of that label is in a code segment
18284 @emph{outside} the segment where the stubs are located.
18288 The compiler emits such @code{gs} modifiers for code labels in the
18289 following situations:
18291 @item Taking address of a function or code label.
18292 @item Computed goto.
18293 @item If prologue-save function is used, see @option{-mcall-prologues}
18294 command-line option.
18295 @item Switch/case dispatch tables. If you do not want such dispatch
18296 tables you can specify the @option{-fno-jump-tables} command-line option.
18297 @item C and C++ constructors/destructors called during startup/shutdown.
18298 @item If the tools hit a @code{gs()} modifier explained above.
18302 Jumping to non-symbolic addresses like so is @emph{not} supported:
18307 /* Call function at word address 0x2 */
18308 return ((int(*)(void)) 0x2)();
18312 Instead, a stub has to be set up, i.e.@: the function has to be called
18313 through a symbol (@code{func_4} in the example):
18318 extern int func_4 (void);
18320 /* Call function at byte address 0x4 */
18325 and the application be linked with @option{-Wl,--defsym,func_4=0x4}.
18326 Alternatively, @code{func_4} can be defined in the linker script.
18329 @subsubsection Handling of the @code{RAMPD}, @code{RAMPX}, @code{RAMPY} and @code{RAMPZ} Special Function Registers
18330 @cindex @code{RAMPD}
18331 @cindex @code{RAMPX}
18332 @cindex @code{RAMPY}
18333 @cindex @code{RAMPZ}
18334 Some AVR devices support memories larger than the 64@tie{}KiB range
18335 that can be accessed with 16-bit pointers. To access memory locations
18336 outside this 64@tie{}KiB range, the content of a @code{RAMP}
18337 register is used as high part of the address:
18338 The @code{X}, @code{Y}, @code{Z} address register is concatenated
18339 with the @code{RAMPX}, @code{RAMPY}, @code{RAMPZ} special function
18340 register, respectively, to get a wide address. Similarly,
18341 @code{RAMPD} is used together with direct addressing.
18345 The startup code initializes the @code{RAMP} special function
18346 registers with zero.
18349 If a @ref{AVR Named Address Spaces,named address space} other than
18350 generic or @code{__flash} is used, then @code{RAMPZ} is set
18351 as needed before the operation.
18354 If the device supports RAM larger than 64@tie{}KiB and the compiler
18355 needs to change @code{RAMPZ} to accomplish an operation, @code{RAMPZ}
18356 is reset to zero after the operation.
18359 If the device comes with a specific @code{RAMP} register, the ISR
18360 prologue/epilogue saves/restores that SFR and initializes it with
18361 zero in case the ISR code might (implicitly) use it.
18364 RAM larger than 64@tie{}KiB is not supported by GCC for AVR targets.
18365 If you use inline assembler to read from locations outside the
18366 16-bit address range and change one of the @code{RAMP} registers,
18367 you must reset it to zero after the access.
18371 @subsubsection AVR Built-in Macros
18373 GCC defines several built-in macros so that the user code can test
18374 for the presence or absence of features. Almost any of the following
18375 built-in macros are deduced from device capabilities and thus
18376 triggered by the @option{-mmcu=} command-line option.
18378 For even more AVR-specific built-in macros see
18379 @ref{AVR Named Address Spaces} and @ref{AVR Built-in Functions}.
18384 Build-in macro that resolves to a decimal number that identifies the
18385 architecture and depends on the @option{-mmcu=@var{mcu}} option.
18386 Possible values are:
18388 @code{2}, @code{25}, @code{3}, @code{31}, @code{35},
18389 @code{4}, @code{5}, @code{51}, @code{6}
18391 for @var{mcu}=@code{avr2}, @code{avr25}, @code{avr3}, @code{avr31},
18392 @code{avr35}, @code{avr4}, @code{avr5}, @code{avr51}, @code{avr6},
18397 @code{102}, @code{103}, @code{104},
18398 @code{105}, @code{106}, @code{107}
18400 for @var{mcu}=@code{avrtiny},
18401 @code{avrxmega2}, @code{avrxmega3}, @code{avrxmega4},
18402 @code{avrxmega5}, @code{avrxmega6}, @code{avrxmega7}, respectively.
18403 If @var{mcu} specifies a device, this built-in macro is set
18404 accordingly. For example, with @option{-mmcu=atmega8} the macro is
18405 defined to @code{4}.
18407 @item __AVR_@var{Device}__
18408 Setting @option{-mmcu=@var{device}} defines this built-in macro which reflects
18409 the device's name. For example, @option{-mmcu=atmega8} defines the
18410 built-in macro @code{__AVR_ATmega8__}, @option{-mmcu=attiny261a} defines
18411 @code{__AVR_ATtiny261A__}, etc.
18413 The built-in macros' names follow
18414 the scheme @code{__AVR_@var{Device}__} where @var{Device} is
18415 the device name as from the AVR user manual. The difference between
18416 @var{Device} in the built-in macro and @var{device} in
18417 @option{-mmcu=@var{device}} is that the latter is always lowercase.
18419 If @var{device} is not a device but only a core architecture like
18420 @samp{avr51}, this macro is not defined.
18422 @item __AVR_DEVICE_NAME__
18423 Setting @option{-mmcu=@var{device}} defines this built-in macro to
18424 the device's name. For example, with @option{-mmcu=atmega8} the macro
18425 is defined to @code{atmega8}.
18427 If @var{device} is not a device but only a core architecture like
18428 @samp{avr51}, this macro is not defined.
18430 @item __AVR_XMEGA__
18431 The device / architecture belongs to the XMEGA family of devices.
18433 @item __AVR_HAVE_ELPM__
18434 The device has the @code{ELPM} instruction.
18436 @item __AVR_HAVE_ELPMX__
18437 The device has the @code{ELPM R@var{n},Z} and @code{ELPM
18438 R@var{n},Z+} instructions.
18440 @item __AVR_HAVE_MOVW__
18441 The device has the @code{MOVW} instruction to perform 16-bit
18442 register-register moves.
18444 @item __AVR_HAVE_LPMX__
18445 The device has the @code{LPM R@var{n},Z} and
18446 @code{LPM R@var{n},Z+} instructions.
18448 @item __AVR_HAVE_MUL__
18449 The device has a hardware multiplier.
18451 @item __AVR_HAVE_JMP_CALL__
18452 The device has the @code{JMP} and @code{CALL} instructions.
18453 This is the case for devices with more than 8@tie{}KiB of program
18456 @item __AVR_HAVE_EIJMP_EICALL__
18457 @itemx __AVR_3_BYTE_PC__
18458 The device has the @code{EIJMP} and @code{EICALL} instructions.
18459 This is the case for devices with more than 128@tie{}KiB of program memory.
18460 This also means that the program counter
18461 (PC) is 3@tie{}bytes wide.
18463 @item __AVR_2_BYTE_PC__
18464 The program counter (PC) is 2@tie{}bytes wide. This is the case for devices
18465 with up to 128@tie{}KiB of program memory.
18467 @item __AVR_HAVE_8BIT_SP__
18468 @itemx __AVR_HAVE_16BIT_SP__
18469 The stack pointer (SP) register is treated as 8-bit respectively
18470 16-bit register by the compiler.
18471 The definition of these macros is affected by @option{-mtiny-stack}.
18473 @item __AVR_HAVE_SPH__
18475 The device has the SPH (high part of stack pointer) special function
18476 register or has an 8-bit stack pointer, respectively.
18477 The definition of these macros is affected by @option{-mmcu=} and
18478 in the cases of @option{-mmcu=avr2} and @option{-mmcu=avr25} also
18481 @item __AVR_HAVE_RAMPD__
18482 @itemx __AVR_HAVE_RAMPX__
18483 @itemx __AVR_HAVE_RAMPY__
18484 @itemx __AVR_HAVE_RAMPZ__
18485 The device has the @code{RAMPD}, @code{RAMPX}, @code{RAMPY},
18486 @code{RAMPZ} special function register, respectively.
18488 @item __NO_INTERRUPTS__
18489 This macro reflects the @option{-mno-interrupts} command-line option.
18491 @item __AVR_ERRATA_SKIP__
18492 @itemx __AVR_ERRATA_SKIP_JMP_CALL__
18493 Some AVR devices (AT90S8515, ATmega103) must not skip 32-bit
18494 instructions because of a hardware erratum. Skip instructions are
18495 @code{SBRS}, @code{SBRC}, @code{SBIS}, @code{SBIC} and @code{CPSE}.
18496 The second macro is only defined if @code{__AVR_HAVE_JMP_CALL__} is also
18499 @item __AVR_ISA_RMW__
18500 The device has Read-Modify-Write instructions (XCH, LAC, LAS and LAT).
18502 @item __AVR_SFR_OFFSET__=@var{offset}
18503 Instructions that can address I/O special function registers directly
18504 like @code{IN}, @code{OUT}, @code{SBI}, etc.@: may use a different
18505 address as if addressed by an instruction to access RAM like @code{LD}
18506 or @code{STS}. This offset depends on the device architecture and has
18507 to be subtracted from the RAM address in order to get the
18508 respective I/O@tie{}address.
18510 @item __AVR_SHORT_CALLS__
18511 The @option{-mshort-calls} command line option is set.
18513 @item __AVR_PM_BASE_ADDRESS__=@var{addr}
18514 Some devices support reading from flash memory by means of @code{LD*}
18515 instructions. The flash memory is seen in the data address space
18516 at an offset of @code{__AVR_PM_BASE_ADDRESS__}. If this macro
18517 is not defined, this feature is not available. If defined,
18518 the address space is linear and there is no need to put
18519 @code{.rodata} into RAM. This is handled by the default linker
18520 description file, and is currently available for
18521 @code{avrtiny} and @code{avrxmega3}. Even more convenient,
18522 there is no need to use address spaces like @code{__flash} or
18523 features like attribute @code{progmem} and @code{pgm_read_*}.
18525 @item __WITH_AVRLIBC__
18526 The compiler is configured to be used together with AVR-Libc.
18527 See the @option{--with-avrlibc} configure option.
18531 @node Blackfin Options
18532 @subsection Blackfin Options
18533 @cindex Blackfin Options
18536 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
18538 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
18539 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
18540 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
18541 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
18542 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
18543 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
18544 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
18545 @samp{bf561}, @samp{bf592}.
18547 The optional @var{sirevision} specifies the silicon revision of the target
18548 Blackfin processor. Any workarounds available for the targeted silicon revision
18549 are enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
18550 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
18551 are enabled. The @code{__SILICON_REVISION__} macro is defined to two
18552 hexadecimal digits representing the major and minor numbers in the silicon
18553 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
18554 is not defined. If @var{sirevision} is @samp{any}, the
18555 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
18556 If this optional @var{sirevision} is not used, GCC assumes the latest known
18557 silicon revision of the targeted Blackfin processor.
18559 GCC defines a preprocessor macro for the specified @var{cpu}.
18560 For the @samp{bfin-elf} toolchain, this option causes the hardware BSP
18561 provided by libgloss to be linked in if @option{-msim} is not given.
18563 Without this option, @samp{bf532} is used as the processor by default.
18565 Note that support for @samp{bf561} is incomplete. For @samp{bf561},
18566 only the preprocessor macro is defined.
18570 Specifies that the program will be run on the simulator. This causes
18571 the simulator BSP provided by libgloss to be linked in. This option
18572 has effect only for @samp{bfin-elf} toolchain.
18573 Certain other options, such as @option{-mid-shared-library} and
18574 @option{-mfdpic}, imply @option{-msim}.
18576 @item -momit-leaf-frame-pointer
18577 @opindex momit-leaf-frame-pointer
18578 Don't keep the frame pointer in a register for leaf functions. This
18579 avoids the instructions to save, set up and restore frame pointers and
18580 makes an extra register available in leaf functions.
18582 @item -mspecld-anomaly
18583 @opindex mspecld-anomaly
18584 When enabled, the compiler ensures that the generated code does not
18585 contain speculative loads after jump instructions. If this option is used,
18586 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
18588 @item -mno-specld-anomaly
18589 @opindex mno-specld-anomaly
18590 @opindex mspecld-anomaly
18591 Don't generate extra code to prevent speculative loads from occurring.
18593 @item -mcsync-anomaly
18594 @opindex mcsync-anomaly
18595 When enabled, the compiler ensures that the generated code does not
18596 contain CSYNC or SSYNC instructions too soon after conditional branches.
18597 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
18599 @item -mno-csync-anomaly
18600 @opindex mno-csync-anomaly
18601 @opindex mcsync-anomaly
18602 Don't generate extra code to prevent CSYNC or SSYNC instructions from
18603 occurring too soon after a conditional branch.
18607 When enabled, the compiler is free to take advantage of the knowledge that
18608 the entire program fits into the low 64k of memory.
18611 @opindex mno-low64k
18612 Assume that the program is arbitrarily large. This is the default.
18614 @item -mstack-check-l1
18615 @opindex mstack-check-l1
18616 Do stack checking using information placed into L1 scratchpad memory by the
18619 @item -mid-shared-library
18620 @opindex mid-shared-library
18621 Generate code that supports shared libraries via the library ID method.
18622 This allows for execute in place and shared libraries in an environment
18623 without virtual memory management. This option implies @option{-fPIC}.
18624 With a @samp{bfin-elf} target, this option implies @option{-msim}.
18626 @item -mno-id-shared-library
18627 @opindex mno-id-shared-library
18628 @opindex mid-shared-library
18629 Generate code that doesn't assume ID-based shared libraries are being used.
18630 This is the default.
18632 @item -mleaf-id-shared-library
18633 @opindex mleaf-id-shared-library
18634 Generate code that supports shared libraries via the library ID method,
18635 but assumes that this library or executable won't link against any other
18636 ID shared libraries. That allows the compiler to use faster code for jumps
18639 @item -mno-leaf-id-shared-library
18640 @opindex mno-leaf-id-shared-library
18641 @opindex mleaf-id-shared-library
18642 Do not assume that the code being compiled won't link against any ID shared
18643 libraries. Slower code is generated for jump and call insns.
18645 @item -mshared-library-id=n
18646 @opindex mshared-library-id
18647 Specifies the identification number of the ID-based shared library being
18648 compiled. Specifying a value of 0 generates more compact code; specifying
18649 other values forces the allocation of that number to the current
18650 library but is no more space- or time-efficient than omitting this option.
18654 Generate code that allows the data segment to be located in a different
18655 area of memory from the text segment. This allows for execute in place in
18656 an environment without virtual memory management by eliminating relocations
18657 against the text section.
18659 @item -mno-sep-data
18660 @opindex mno-sep-data
18662 Generate code that assumes that the data segment follows the text segment.
18663 This is the default.
18666 @itemx -mno-long-calls
18667 @opindex mlong-calls
18668 @opindex mno-long-calls
18669 Tells the compiler to perform function calls by first loading the
18670 address of the function into a register and then performing a subroutine
18671 call on this register. This switch is needed if the target function
18672 lies outside of the 24-bit addressing range of the offset-based
18673 version of subroutine call instruction.
18675 This feature is not enabled by default. Specifying
18676 @option{-mno-long-calls} restores the default behavior. Note these
18677 switches have no effect on how the compiler generates code to handle
18678 function calls via function pointers.
18682 Link with the fast floating-point library. This library relaxes some of
18683 the IEEE floating-point standard's rules for checking inputs against
18684 Not-a-Number (NAN), in the interest of performance.
18687 @opindex minline-plt
18688 Enable inlining of PLT entries in function calls to functions that are
18689 not known to bind locally. It has no effect without @option{-mfdpic}.
18692 @opindex mmulticore
18693 Build a standalone application for multicore Blackfin processors.
18694 This option causes proper start files and link scripts supporting
18695 multicore to be used, and defines the macro @code{__BFIN_MULTICORE}.
18696 It can only be used with @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}.
18698 This option can be used with @option{-mcorea} or @option{-mcoreb}, which
18699 selects the one-application-per-core programming model. Without
18700 @option{-mcorea} or @option{-mcoreb}, the single-application/dual-core
18701 programming model is used. In this model, the main function of Core B
18702 should be named as @code{coreb_main}.
18704 If this option is not used, the single-core application programming
18709 Build a standalone application for Core A of BF561 when using
18710 the one-application-per-core programming model. Proper start files
18711 and link scripts are used to support Core A, and the macro
18712 @code{__BFIN_COREA} is defined.
18713 This option can only be used in conjunction with @option{-mmulticore}.
18717 Build a standalone application for Core B of BF561 when using
18718 the one-application-per-core programming model. Proper start files
18719 and link scripts are used to support Core B, and the macro
18720 @code{__BFIN_COREB} is defined. When this option is used, @code{coreb_main}
18721 should be used instead of @code{main}.
18722 This option can only be used in conjunction with @option{-mmulticore}.
18726 Build a standalone application for SDRAM. Proper start files and
18727 link scripts are used to put the application into SDRAM, and the macro
18728 @code{__BFIN_SDRAM} is defined.
18729 The loader should initialize SDRAM before loading the application.
18733 Assume that ICPLBs are enabled at run time. This has an effect on certain
18734 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
18735 are enabled; for standalone applications the default is off.
18739 @subsection C6X Options
18740 @cindex C6X Options
18743 @item -march=@var{name}
18745 This specifies the name of the target architecture. GCC uses this
18746 name to determine what kind of instructions it can emit when generating
18747 assembly code. Permissible names are: @samp{c62x},
18748 @samp{c64x}, @samp{c64x+}, @samp{c67x}, @samp{c67x+}, @samp{c674x}.
18751 @opindex mbig-endian
18752 Generate code for a big-endian target.
18754 @item -mlittle-endian
18755 @opindex mlittle-endian
18756 Generate code for a little-endian target. This is the default.
18760 Choose startup files and linker script suitable for the simulator.
18762 @item -msdata=default
18763 @opindex msdata=default
18764 Put small global and static data in the @code{.neardata} section,
18765 which is pointed to by register @code{B14}. Put small uninitialized
18766 global and static data in the @code{.bss} section, which is adjacent
18767 to the @code{.neardata} section. Put small read-only data into the
18768 @code{.rodata} section. The corresponding sections used for large
18769 pieces of data are @code{.fardata}, @code{.far} and @code{.const}.
18772 @opindex msdata=all
18773 Put all data, not just small objects, into the sections reserved for
18774 small data, and use addressing relative to the @code{B14} register to
18778 @opindex msdata=none
18779 Make no use of the sections reserved for small data, and use absolute
18780 addresses to access all data. Put all initialized global and static
18781 data in the @code{.fardata} section, and all uninitialized data in the
18782 @code{.far} section. Put all constant data into the @code{.const}
18787 @subsection CRIS Options
18788 @cindex CRIS Options
18790 These options are defined specifically for the CRIS ports.
18793 @item -march=@var{architecture-type}
18794 @itemx -mcpu=@var{architecture-type}
18797 Generate code for the specified architecture. The choices for
18798 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
18799 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
18800 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
18803 @item -mtune=@var{architecture-type}
18805 Tune to @var{architecture-type} everything applicable about the generated
18806 code, except for the ABI and the set of available instructions. The
18807 choices for @var{architecture-type} are the same as for
18808 @option{-march=@var{architecture-type}}.
18810 @item -mmax-stack-frame=@var{n}
18811 @opindex mmax-stack-frame
18812 Warn when the stack frame of a function exceeds @var{n} bytes.
18818 The options @option{-metrax4} and @option{-metrax100} are synonyms for
18819 @option{-march=v3} and @option{-march=v8} respectively.
18821 @item -mmul-bug-workaround
18822 @itemx -mno-mul-bug-workaround
18823 @opindex mmul-bug-workaround
18824 @opindex mno-mul-bug-workaround
18825 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
18826 models where it applies. This option is active by default.
18830 Enable CRIS-specific verbose debug-related information in the assembly
18831 code. This option also has the effect of turning off the @samp{#NO_APP}
18832 formatted-code indicator to the assembler at the beginning of the
18837 Do not use condition-code results from previous instruction; always emit
18838 compare and test instructions before use of condition codes.
18840 @item -mno-side-effects
18841 @opindex mno-side-effects
18842 @opindex mside-effects
18843 Do not emit instructions with side effects in addressing modes other than
18846 @item -mstack-align
18847 @itemx -mno-stack-align
18848 @itemx -mdata-align
18849 @itemx -mno-data-align
18850 @itemx -mconst-align
18851 @itemx -mno-const-align
18852 @opindex mstack-align
18853 @opindex mno-stack-align
18854 @opindex mdata-align
18855 @opindex mno-data-align
18856 @opindex mconst-align
18857 @opindex mno-const-align
18858 These options (@samp{no-} options) arrange (eliminate arrangements) for the
18859 stack frame, individual data and constants to be aligned for the maximum
18860 single data access size for the chosen CPU model. The default is to
18861 arrange for 32-bit alignment. ABI details such as structure layout are
18862 not affected by these options.
18870 Similar to the stack- data- and const-align options above, these options
18871 arrange for stack frame, writable data and constants to all be 32-bit,
18872 16-bit or 8-bit aligned. The default is 32-bit alignment.
18874 @item -mno-prologue-epilogue
18875 @itemx -mprologue-epilogue
18876 @opindex mno-prologue-epilogue
18877 @opindex mprologue-epilogue
18878 With @option{-mno-prologue-epilogue}, the normal function prologue and
18879 epilogue which set up the stack frame are omitted and no return
18880 instructions or return sequences are generated in the code. Use this
18881 option only together with visual inspection of the compiled code: no
18882 warnings or errors are generated when call-saved registers must be saved,
18883 or storage for local variables needs to be allocated.
18887 @opindex mno-gotplt
18889 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
18890 instruction sequences that load addresses for functions from the PLT part
18891 of the GOT rather than (traditional on other architectures) calls to the
18892 PLT@. The default is @option{-mgotplt}.
18896 Legacy no-op option only recognized with the cris-axis-elf and
18897 cris-axis-linux-gnu targets.
18901 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
18905 This option, recognized for the cris-axis-elf, arranges
18906 to link with input-output functions from a simulator library. Code,
18907 initialized data and zero-initialized data are allocated consecutively.
18911 Like @option{-sim}, but pass linker options to locate initialized data at
18912 0x40000000 and zero-initialized data at 0x80000000.
18916 @subsection CR16 Options
18917 @cindex CR16 Options
18919 These options are defined specifically for the CR16 ports.
18925 Enable the use of multiply-accumulate instructions. Disabled by default.
18929 @opindex mcr16cplus
18931 Generate code for CR16C or CR16C+ architecture. CR16C+ architecture
18936 Links the library libsim.a which is in compatible with simulator. Applicable
18937 to ELF compiler only.
18941 Choose integer type as 32-bit wide.
18945 Generates @code{sbit}/@code{cbit} instructions for bit manipulations.
18947 @item -mdata-model=@var{model}
18948 @opindex mdata-model
18949 Choose a data model. The choices for @var{model} are @samp{near},
18950 @samp{far} or @samp{medium}. @samp{medium} is default.
18951 However, @samp{far} is not valid with @option{-mcr16c}, as the
18952 CR16C architecture does not support the far data model.
18955 @node C-SKY Options
18956 @subsection C-SKY Options
18957 @cindex C-SKY Options
18959 GCC supports these options when compiling for C-SKY V2 processors.
18963 @item -march=@var{arch}
18965 Specify the C-SKY target architecture. Valid values for @var{arch} are:
18966 @samp{ck801}, @samp{ck802}, @samp{ck803}, @samp{ck807}, and @samp{ck810}.
18967 The default is @samp{ck810}.
18969 @item -mcpu=@var{cpu}
18971 Specify the C-SKY target processor. Valid values for @var{cpu} are:
18972 @samp{ck801}, @samp{ck801t},
18973 @samp{ck802}, @samp{ck802t}, @samp{ck802j},
18974 @samp{ck803}, @samp{ck803h}, @samp{ck803t}, @samp{ck803ht},
18975 @samp{ck803f}, @samp{ck803fh}, @samp{ck803e}, @samp{ck803eh},
18976 @samp{ck803et}, @samp{ck803eht}, @samp{ck803ef}, @samp{ck803efh},
18977 @samp{ck803ft}, @samp{ck803eft}, @samp{ck803efht}, @samp{ck803r1},
18978 @samp{ck803hr1}, @samp{ck803tr1}, @samp{ck803htr1}, @samp{ck803fr1},
18979 @samp{ck803fhr1}, @samp{ck803er1}, @samp{ck803ehr1}, @samp{ck803etr1},
18980 @samp{ck803ehtr1}, @samp{ck803efr1}, @samp{ck803efhr1}, @samp{ck803ftr1},
18981 @samp{ck803eftr1}, @samp{ck803efhtr1},
18982 @samp{ck803s}, @samp{ck803st}, @samp{ck803se}, @samp{ck803sf},
18983 @samp{ck803sef}, @samp{ck803seft},
18984 @samp{ck807e}, @samp{ck807ef}, @samp{ck807}, @samp{ck807f},
18985 @samp{ck810e}, @samp{ck810et}, @samp{ck810ef}, @samp{ck810eft},
18986 @samp{ck810}, @samp{ck810v}, @samp{ck810f}, @samp{ck810t}, @samp{ck810fv},
18987 @samp{ck810tv}, @samp{ck810ft}, and @samp{ck810ftv}.
18990 @opindex mbig-endian
18993 @itemx -mlittle-endian
18994 @opindex mlittle-endian
18998 Select big- or little-endian code. The default is little-endian.
19001 @opindex mhard-float
19002 @itemx -msoft-float
19003 @opindex msoft-float
19005 Select hardware or software floating-point implementations.
19006 The default is soft float.
19008 @item -mdouble-float
19009 @itemx -mno-double-float
19010 @opindex mdouble-float
19011 When @option{-mhard-float} is in effect, enable generation of
19012 double-precision float instructions. This is the default except
19013 when compiling for CK803.
19018 When @option{-mhard-float} is in effect, enable generation of
19019 @code{frecipd}, @code{fsqrtd}, and @code{fdivd} instructions.
19020 This is the default except when compiling for CK803.
19022 @item -mfpu=@var{fpu}
19024 Select the floating-point processor. This option can only be used with
19025 @option{-mhard-float}.
19026 Values for @var{fpu} are
19027 @samp{fpv2_sf} (equivalent to @samp{-mno-double-float -mno-fdivdu}),
19028 @samp{fpv2} (@samp{-mdouble-float -mno-divdu}), and
19029 @samp{fpv2_divd} (@samp{-mdouble-float -mdivdu}).
19034 Enable the extended @code{lrw} instruction. This option defaults to on
19035 for CK801 and off otherwise.
19040 Enable interrupt stack instructions; the default is off.
19042 The @option{-mistack} option is required to handle the
19043 @code{interrupt} and @code{isr} function attributes
19044 (@pxref{C-SKY Function Attributes}).
19048 Enable multiprocessor instructions; the default is off.
19052 Enable coprocessor instructions; the default is off.
19056 Enable coprocessor instructions; the default is off.
19060 Enable C-SKY security instructions; the default is off.
19064 Enable C-SKY trust instructions; the default is off.
19072 Enable C-SKY DSP, Enhanced DSP, or Vector DSP instructions, respectively.
19073 All of these options default to off.
19078 Generate divide instructions. Default is off.
19083 Generate code for Smart Mode, using only registers numbered 0-7 to allow
19084 use of 16-bit instructions. This option is ignored for CK801 where this
19085 is the required behavior, and it defaults to on for CK802.
19086 For other targets, the default is off.
19088 @item -mhigh-registers
19089 @itemx -mno-high-registers
19090 @opindex mhigh-registers
19091 Generate code using the high registers numbered 16-31. This option
19092 is not supported on CK801, CK802, or CK803, and is enabled by default
19093 for other processors.
19098 Generate code using global anchor symbol addresses.
19101 @itemx -mno-pushpop
19103 Generate code using @code{push} and @code{pop} instructions. This option
19106 @item -mmultiple-stld
19108 @itemx -mno-multiple-stld
19110 @opindex mmultiple-stld
19111 Generate code using @code{stm} and @code{ldm} instructions. This option
19112 isn't supported on CK801 but is enabled by default on other processors.
19115 @itemx -mno-constpool
19116 @opindex mconstpool
19117 Create constant pools in the compiler instead of deferring it to the
19118 assembler. This option is the default and required for correct code
19119 generation on CK801 and CK802, and is optional on other processors.
19122 @item -mno-stack-size
19123 @opindex mstack-size
19124 Emit @code{.stack_size} directives for each function in the assembly
19125 output. This option defaults to off.
19130 Generate code for the C-SKY compiler runtime instead of libgcc. This
19131 option defaults to off.
19133 @item -mbranch-cost=@var{n}
19134 @opindex mbranch-cost=
19135 Set the branch costs to roughly @code{n} instructions. The default is 1.
19137 @item -msched-prolog
19138 @itemx -mno-sched-prolog
19139 @opindex msched-prolog
19140 Permit scheduling of function prologue and epilogue sequences. Using
19141 this option can result in code that is not compliant with the C-SKY V2 ABI
19142 prologue requirements and that cannot be debugged or backtraced.
19143 It is disabled by default.
19147 @node Darwin Options
19148 @subsection Darwin Options
19149 @cindex Darwin options
19151 These options are defined for all architectures running the Darwin operating
19154 FSF GCC on Darwin does not create ``fat'' object files; it creates
19155 an object file for the single architecture that GCC was built to
19156 target. Apple's GCC on Darwin does create ``fat'' files if multiple
19157 @option{-arch} options are used; it does so by running the compiler or
19158 linker multiple times and joining the results together with
19161 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
19162 @samp{i686}) is determined by the flags that specify the ISA
19163 that GCC is targeting, like @option{-mcpu} or @option{-march}. The
19164 @option{-force_cpusubtype_ALL} option can be used to override this.
19166 The Darwin tools vary in their behavior when presented with an ISA
19167 mismatch. The assembler, @file{as}, only permits instructions to
19168 be used that are valid for the subtype of the file it is generating,
19169 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
19170 The linker for shared libraries, @file{/usr/bin/libtool}, fails
19171 and prints an error if asked to create a shared library with a less
19172 restrictive subtype than its input files (for instance, trying to put
19173 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
19174 for executables, @command{ld}, quietly gives the executable the most
19175 restrictive subtype of any of its input files.
19180 Add the framework directory @var{dir} to the head of the list of
19181 directories to be searched for header files. These directories are
19182 interleaved with those specified by @option{-I} options and are
19183 scanned in a left-to-right order.
19185 A framework directory is a directory with frameworks in it. A
19186 framework is a directory with a @file{Headers} and/or
19187 @file{PrivateHeaders} directory contained directly in it that ends
19188 in @file{.framework}. The name of a framework is the name of this
19189 directory excluding the @file{.framework}. Headers associated with
19190 the framework are found in one of those two directories, with
19191 @file{Headers} being searched first. A subframework is a framework
19192 directory that is in a framework's @file{Frameworks} directory.
19193 Includes of subframework headers can only appear in a header of a
19194 framework that contains the subframework, or in a sibling subframework
19195 header. Two subframeworks are siblings if they occur in the same
19196 framework. A subframework should not have the same name as a
19197 framework; a warning is issued if this is violated. Currently a
19198 subframework cannot have subframeworks; in the future, the mechanism
19199 may be extended to support this. The standard frameworks can be found
19200 in @file{/System/Library/Frameworks} and
19201 @file{/Library/Frameworks}. An example include looks like
19202 @code{#include <Framework/header.h>}, where @file{Framework} denotes
19203 the name of the framework and @file{header.h} is found in the
19204 @file{PrivateHeaders} or @file{Headers} directory.
19206 @item -iframework@var{dir}
19207 @opindex iframework
19208 Like @option{-F} except the directory is a treated as a system
19209 directory. The main difference between this @option{-iframework} and
19210 @option{-F} is that with @option{-iframework} the compiler does not
19211 warn about constructs contained within header files found via
19212 @var{dir}. This option is valid only for the C family of languages.
19216 Emit debugging information for symbols that are used. For stabs
19217 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
19218 This is by default ON@.
19222 Emit debugging information for all symbols and types.
19224 @item -mmacosx-version-min=@var{version}
19225 The earliest version of MacOS X that this executable will run on
19226 is @var{version}. Typical values of @var{version} include @code{10.1},
19227 @code{10.2}, and @code{10.3.9}.
19229 If the compiler was built to use the system's headers by default,
19230 then the default for this option is the system version on which the
19231 compiler is running, otherwise the default is to make choices that
19232 are compatible with as many systems and code bases as possible.
19236 Enable kernel development mode. The @option{-mkernel} option sets
19237 @option{-static}, @option{-fno-common}, @option{-fno-use-cxa-atexit},
19238 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
19239 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
19240 applicable. This mode also sets @option{-mno-altivec},
19241 @option{-msoft-float}, @option{-fno-builtin} and
19242 @option{-mlong-branch} for PowerPC targets.
19244 @item -mone-byte-bool
19245 @opindex mone-byte-bool
19246 Override the defaults for @code{bool} so that @code{sizeof(bool)==1}.
19247 By default @code{sizeof(bool)} is @code{4} when compiling for
19248 Darwin/PowerPC and @code{1} when compiling for Darwin/x86, so this
19249 option has no effect on x86.
19251 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
19252 to generate code that is not binary compatible with code generated
19253 without that switch. Using this switch may require recompiling all
19254 other modules in a program, including system libraries. Use this
19255 switch to conform to a non-default data model.
19257 @item -mfix-and-continue
19258 @itemx -ffix-and-continue
19259 @itemx -findirect-data
19260 @opindex mfix-and-continue
19261 @opindex ffix-and-continue
19262 @opindex findirect-data
19263 Generate code suitable for fast turnaround development, such as to
19264 allow GDB to dynamically load @file{.o} files into already-running
19265 programs. @option{-findirect-data} and @option{-ffix-and-continue}
19266 are provided for backwards compatibility.
19270 Loads all members of static archive libraries.
19271 See man ld(1) for more information.
19273 @item -arch_errors_fatal
19274 @opindex arch_errors_fatal
19275 Cause the errors having to do with files that have the wrong architecture
19278 @item -bind_at_load
19279 @opindex bind_at_load
19280 Causes the output file to be marked such that the dynamic linker will
19281 bind all undefined references when the file is loaded or launched.
19285 Produce a Mach-o bundle format file.
19286 See man ld(1) for more information.
19288 @item -bundle_loader @var{executable}
19289 @opindex bundle_loader
19290 This option specifies the @var{executable} that will load the build
19291 output file being linked. See man ld(1) for more information.
19294 @opindex dynamiclib
19295 When passed this option, GCC produces a dynamic library instead of
19296 an executable when linking, using the Darwin @file{libtool} command.
19298 @item -force_cpusubtype_ALL
19299 @opindex force_cpusubtype_ALL
19300 This causes GCC's output file to have the @samp{ALL} subtype, instead of
19301 one controlled by the @option{-mcpu} or @option{-march} option.
19303 @item -allowable_client @var{client_name}
19304 @itemx -client_name
19305 @itemx -compatibility_version
19306 @itemx -current_version
19308 @itemx -dependency-file
19310 @itemx -dylinker_install_name
19312 @itemx -exported_symbols_list
19315 @itemx -flat_namespace
19316 @itemx -force_flat_namespace
19317 @itemx -headerpad_max_install_names
19320 @itemx -install_name
19321 @itemx -keep_private_externs
19322 @itemx -multi_module
19323 @itemx -multiply_defined
19324 @itemx -multiply_defined_unused
19327 @itemx -no_dead_strip_inits_and_terms
19328 @itemx -nofixprebinding
19329 @itemx -nomultidefs
19331 @itemx -noseglinkedit
19332 @itemx -pagezero_size
19334 @itemx -prebind_all_twolevel_modules
19335 @itemx -private_bundle
19337 @itemx -read_only_relocs
19339 @itemx -sectobjectsymbols
19343 @itemx -sectobjectsymbols
19346 @itemx -segs_read_only_addr
19348 @itemx -segs_read_write_addr
19349 @itemx -seg_addr_table
19350 @itemx -seg_addr_table_filename
19351 @itemx -seglinkedit
19353 @itemx -segs_read_only_addr
19354 @itemx -segs_read_write_addr
19355 @itemx -single_module
19357 @itemx -sub_library
19359 @itemx -sub_umbrella
19360 @itemx -twolevel_namespace
19363 @itemx -unexported_symbols_list
19364 @itemx -weak_reference_mismatches
19365 @itemx -whatsloaded
19366 @opindex allowable_client
19367 @opindex client_name
19368 @opindex compatibility_version
19369 @opindex current_version
19370 @opindex dead_strip
19371 @opindex dependency-file
19372 @opindex dylib_file
19373 @opindex dylinker_install_name
19375 @opindex exported_symbols_list
19377 @opindex flat_namespace
19378 @opindex force_flat_namespace
19379 @opindex headerpad_max_install_names
19380 @opindex image_base
19382 @opindex install_name
19383 @opindex keep_private_externs
19384 @opindex multi_module
19385 @opindex multiply_defined
19386 @opindex multiply_defined_unused
19387 @opindex noall_load
19388 @opindex no_dead_strip_inits_and_terms
19389 @opindex nofixprebinding
19390 @opindex nomultidefs
19392 @opindex noseglinkedit
19393 @opindex pagezero_size
19395 @opindex prebind_all_twolevel_modules
19396 @opindex private_bundle
19397 @opindex read_only_relocs
19399 @opindex sectobjectsymbols
19402 @opindex sectcreate
19403 @opindex sectobjectsymbols
19406 @opindex segs_read_only_addr
19407 @opindex segs_read_write_addr
19408 @opindex seg_addr_table
19409 @opindex seg_addr_table_filename
19410 @opindex seglinkedit
19412 @opindex segs_read_only_addr
19413 @opindex segs_read_write_addr
19414 @opindex single_module
19416 @opindex sub_library
19417 @opindex sub_umbrella
19418 @opindex twolevel_namespace
19421 @opindex unexported_symbols_list
19422 @opindex weak_reference_mismatches
19423 @opindex whatsloaded
19424 These options are passed to the Darwin linker. The Darwin linker man page
19425 describes them in detail.
19428 @node DEC Alpha Options
19429 @subsection DEC Alpha Options
19431 These @samp{-m} options are defined for the DEC Alpha implementations:
19434 @item -mno-soft-float
19435 @itemx -msoft-float
19436 @opindex mno-soft-float
19437 @opindex msoft-float
19438 Use (do not use) the hardware floating-point instructions for
19439 floating-point operations. When @option{-msoft-float} is specified,
19440 functions in @file{libgcc.a} are used to perform floating-point
19441 operations. Unless they are replaced by routines that emulate the
19442 floating-point operations, or compiled in such a way as to call such
19443 emulations routines, these routines issue floating-point
19444 operations. If you are compiling for an Alpha without floating-point
19445 operations, you must ensure that the library is built so as not to call
19448 Note that Alpha implementations without floating-point operations are
19449 required to have floating-point registers.
19452 @itemx -mno-fp-regs
19454 @opindex mno-fp-regs
19455 Generate code that uses (does not use) the floating-point register set.
19456 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
19457 register set is not used, floating-point operands are passed in integer
19458 registers as if they were integers and floating-point results are passed
19459 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
19460 so any function with a floating-point argument or return value called by code
19461 compiled with @option{-mno-fp-regs} must also be compiled with that
19464 A typical use of this option is building a kernel that does not use,
19465 and hence need not save and restore, any floating-point registers.
19469 The Alpha architecture implements floating-point hardware optimized for
19470 maximum performance. It is mostly compliant with the IEEE floating-point
19471 standard. However, for full compliance, software assistance is
19472 required. This option generates code fully IEEE-compliant code
19473 @emph{except} that the @var{inexact-flag} is not maintained (see below).
19474 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
19475 defined during compilation. The resulting code is less efficient but is
19476 able to correctly support denormalized numbers and exceptional IEEE
19477 values such as not-a-number and plus/minus infinity. Other Alpha
19478 compilers call this option @option{-ieee_with_no_inexact}.
19480 @item -mieee-with-inexact
19481 @opindex mieee-with-inexact
19482 This is like @option{-mieee} except the generated code also maintains
19483 the IEEE @var{inexact-flag}. Turning on this option causes the
19484 generated code to implement fully-compliant IEEE math. In addition to
19485 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
19486 macro. On some Alpha implementations the resulting code may execute
19487 significantly slower than the code generated by default. Since there is
19488 very little code that depends on the @var{inexact-flag}, you should
19489 normally not specify this option. Other Alpha compilers call this
19490 option @option{-ieee_with_inexact}.
19492 @item -mfp-trap-mode=@var{trap-mode}
19493 @opindex mfp-trap-mode
19494 This option controls what floating-point related traps are enabled.
19495 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
19496 The trap mode can be set to one of four values:
19500 This is the default (normal) setting. The only traps that are enabled
19501 are the ones that cannot be disabled in software (e.g., division by zero
19505 In addition to the traps enabled by @samp{n}, underflow traps are enabled
19509 Like @samp{u}, but the instructions are marked to be safe for software
19510 completion (see Alpha architecture manual for details).
19513 Like @samp{su}, but inexact traps are enabled as well.
19516 @item -mfp-rounding-mode=@var{rounding-mode}
19517 @opindex mfp-rounding-mode
19518 Selects the IEEE rounding mode. Other Alpha compilers call this option
19519 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
19524 Normal IEEE rounding mode. Floating-point numbers are rounded towards
19525 the nearest machine number or towards the even machine number in case
19529 Round towards minus infinity.
19532 Chopped rounding mode. Floating-point numbers are rounded towards zero.
19535 Dynamic rounding mode. A field in the floating-point control register
19536 (@var{fpcr}, see Alpha architecture reference manual) controls the
19537 rounding mode in effect. The C library initializes this register for
19538 rounding towards plus infinity. Thus, unless your program modifies the
19539 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
19542 @item -mtrap-precision=@var{trap-precision}
19543 @opindex mtrap-precision
19544 In the Alpha architecture, floating-point traps are imprecise. This
19545 means without software assistance it is impossible to recover from a
19546 floating trap and program execution normally needs to be terminated.
19547 GCC can generate code that can assist operating system trap handlers
19548 in determining the exact location that caused a floating-point trap.
19549 Depending on the requirements of an application, different levels of
19550 precisions can be selected:
19554 Program precision. This option is the default and means a trap handler
19555 can only identify which program caused a floating-point exception.
19558 Function precision. The trap handler can determine the function that
19559 caused a floating-point exception.
19562 Instruction precision. The trap handler can determine the exact
19563 instruction that caused a floating-point exception.
19566 Other Alpha compilers provide the equivalent options called
19567 @option{-scope_safe} and @option{-resumption_safe}.
19569 @item -mieee-conformant
19570 @opindex mieee-conformant
19571 This option marks the generated code as IEEE conformant. You must not
19572 use this option unless you also specify @option{-mtrap-precision=i} and either
19573 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
19574 is to emit the line @samp{.eflag 48} in the function prologue of the
19575 generated assembly file.
19577 @item -mbuild-constants
19578 @opindex mbuild-constants
19579 Normally GCC examines a 32- or 64-bit integer constant to
19580 see if it can construct it from smaller constants in two or three
19581 instructions. If it cannot, it outputs the constant as a literal and
19582 generates code to load it from the data segment at run time.
19584 Use this option to require GCC to construct @emph{all} integer constants
19585 using code, even if it takes more instructions (the maximum is six).
19587 You typically use this option to build a shared library dynamic
19588 loader. Itself a shared library, it must relocate itself in memory
19589 before it can find the variables and constants in its own data segment.
19607 Indicate whether GCC should generate code to use the optional BWX,
19608 CIX, FIX and MAX instruction sets. The default is to use the instruction
19609 sets supported by the CPU type specified via @option{-mcpu=} option or that
19610 of the CPU on which GCC was built if none is specified.
19613 @itemx -mfloat-ieee
19614 @opindex mfloat-vax
19615 @opindex mfloat-ieee
19616 Generate code that uses (does not use) VAX F and G floating-point
19617 arithmetic instead of IEEE single and double precision.
19619 @item -mexplicit-relocs
19620 @itemx -mno-explicit-relocs
19621 @opindex mexplicit-relocs
19622 @opindex mno-explicit-relocs
19623 Older Alpha assemblers provided no way to generate symbol relocations
19624 except via assembler macros. Use of these macros does not allow
19625 optimal instruction scheduling. GNU binutils as of version 2.12
19626 supports a new syntax that allows the compiler to explicitly mark
19627 which relocations should apply to which instructions. This option
19628 is mostly useful for debugging, as GCC detects the capabilities of
19629 the assembler when it is built and sets the default accordingly.
19632 @itemx -mlarge-data
19633 @opindex msmall-data
19634 @opindex mlarge-data
19635 When @option{-mexplicit-relocs} is in effect, static data is
19636 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
19637 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
19638 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
19639 16-bit relocations off of the @code{$gp} register. This limits the
19640 size of the small data area to 64KB, but allows the variables to be
19641 directly accessed via a single instruction.
19643 The default is @option{-mlarge-data}. With this option the data area
19644 is limited to just below 2GB@. Programs that require more than 2GB of
19645 data must use @code{malloc} or @code{mmap} to allocate the data in the
19646 heap instead of in the program's data segment.
19648 When generating code for shared libraries, @option{-fpic} implies
19649 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
19652 @itemx -mlarge-text
19653 @opindex msmall-text
19654 @opindex mlarge-text
19655 When @option{-msmall-text} is used, the compiler assumes that the
19656 code of the entire program (or shared library) fits in 4MB, and is
19657 thus reachable with a branch instruction. When @option{-msmall-data}
19658 is used, the compiler can assume that all local symbols share the
19659 same @code{$gp} value, and thus reduce the number of instructions
19660 required for a function call from 4 to 1.
19662 The default is @option{-mlarge-text}.
19664 @item -mcpu=@var{cpu_type}
19666 Set the instruction set and instruction scheduling parameters for
19667 machine type @var{cpu_type}. You can specify either the @samp{EV}
19668 style name or the corresponding chip number. GCC supports scheduling
19669 parameters for the EV4, EV5 and EV6 family of processors and
19670 chooses the default values for the instruction set from the processor
19671 you specify. If you do not specify a processor type, GCC defaults
19672 to the processor on which the compiler was built.
19674 Supported values for @var{cpu_type} are
19680 Schedules as an EV4 and has no instruction set extensions.
19684 Schedules as an EV5 and has no instruction set extensions.
19688 Schedules as an EV5 and supports the BWX extension.
19693 Schedules as an EV5 and supports the BWX and MAX extensions.
19697 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
19701 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
19704 Native toolchains also support the value @samp{native},
19705 which selects the best architecture option for the host processor.
19706 @option{-mcpu=native} has no effect if GCC does not recognize
19709 @item -mtune=@var{cpu_type}
19711 Set only the instruction scheduling parameters for machine type
19712 @var{cpu_type}. The instruction set is not changed.
19714 Native toolchains also support the value @samp{native},
19715 which selects the best architecture option for the host processor.
19716 @option{-mtune=native} has no effect if GCC does not recognize
19719 @item -mmemory-latency=@var{time}
19720 @opindex mmemory-latency
19721 Sets the latency the scheduler should assume for typical memory
19722 references as seen by the application. This number is highly
19723 dependent on the memory access patterns used by the application
19724 and the size of the external cache on the machine.
19726 Valid options for @var{time} are
19730 A decimal number representing clock cycles.
19736 The compiler contains estimates of the number of clock cycles for
19737 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
19738 (also called Dcache, Scache, and Bcache), as well as to main memory.
19739 Note that L3 is only valid for EV5.
19745 @subsection FR30 Options
19746 @cindex FR30 Options
19748 These options are defined specifically for the FR30 port.
19752 @item -msmall-model
19753 @opindex msmall-model
19754 Use the small address space model. This can produce smaller code, but
19755 it does assume that all symbolic values and addresses fit into a
19760 Assume that runtime support has been provided and so there is no need
19761 to include the simulator library (@file{libsim.a}) on the linker
19767 @subsection FT32 Options
19768 @cindex FT32 Options
19770 These options are defined specifically for the FT32 port.
19776 Specifies that the program will be run on the simulator. This causes
19777 an alternate runtime startup and library to be linked.
19778 You must not use this option when generating programs that will run on
19779 real hardware; you must provide your own runtime library for whatever
19780 I/O functions are needed.
19784 Enable Local Register Allocation. This is still experimental for FT32,
19785 so by default the compiler uses standard reload.
19789 Do not use div and mod instructions.
19793 Enable use of the extended instructions of the FT32B processor.
19797 Compress all code using the Ft32B code compression scheme.
19801 Do not generate code that reads program memory.
19806 @subsection FRV Options
19807 @cindex FRV Options
19813 Only use the first 32 general-purpose registers.
19818 Use all 64 general-purpose registers.
19823 Use only the first 32 floating-point registers.
19828 Use all 64 floating-point registers.
19831 @opindex mhard-float
19833 Use hardware instructions for floating-point operations.
19836 @opindex msoft-float
19838 Use library routines for floating-point operations.
19843 Dynamically allocate condition code registers.
19848 Do not try to dynamically allocate condition code registers, only
19849 use @code{icc0} and @code{fcc0}.
19854 Change ABI to use double word insns.
19860 Do not use double word instructions.
19865 Use floating-point double instructions.
19868 @opindex mno-double
19870 Do not use floating-point double instructions.
19875 Use media instructions.
19880 Do not use media instructions.
19885 Use multiply and add/subtract instructions.
19888 @opindex mno-muladd
19890 Do not use multiply and add/subtract instructions.
19895 Select the FDPIC ABI, which uses function descriptors to represent
19896 pointers to functions. Without any PIC/PIE-related options, it
19897 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
19898 assumes GOT entries and small data are within a 12-bit range from the
19899 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
19900 are computed with 32 bits.
19901 With a @samp{bfin-elf} target, this option implies @option{-msim}.
19904 @opindex minline-plt
19906 Enable inlining of PLT entries in function calls to functions that are
19907 not known to bind locally. It has no effect without @option{-mfdpic}.
19908 It's enabled by default if optimizing for speed and compiling for
19909 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
19910 optimization option such as @option{-O3} or above is present in the
19916 Assume a large TLS segment when generating thread-local code.
19921 Do not assume a large TLS segment when generating thread-local code.
19926 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
19927 that is known to be in read-only sections. It's enabled by default,
19928 except for @option{-fpic} or @option{-fpie}: even though it may help
19929 make the global offset table smaller, it trades 1 instruction for 4.
19930 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
19931 one of which may be shared by multiple symbols, and it avoids the need
19932 for a GOT entry for the referenced symbol, so it's more likely to be a
19933 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
19935 @item -multilib-library-pic
19936 @opindex multilib-library-pic
19938 Link with the (library, not FD) pic libraries. It's implied by
19939 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
19940 @option{-fpic} without @option{-mfdpic}. You should never have to use
19944 @opindex mlinked-fp
19946 Follow the EABI requirement of always creating a frame pointer whenever
19947 a stack frame is allocated. This option is enabled by default and can
19948 be disabled with @option{-mno-linked-fp}.
19951 @opindex mlong-calls
19953 Use indirect addressing to call functions outside the current
19954 compilation unit. This allows the functions to be placed anywhere
19955 within the 32-bit address space.
19957 @item -malign-labels
19958 @opindex malign-labels
19960 Try to align labels to an 8-byte boundary by inserting NOPs into the
19961 previous packet. This option only has an effect when VLIW packing
19962 is enabled. It doesn't create new packets; it merely adds NOPs to
19965 @item -mlibrary-pic
19966 @opindex mlibrary-pic
19968 Generate position-independent EABI code.
19973 Use only the first four media accumulator registers.
19978 Use all eight media accumulator registers.
19983 Pack VLIW instructions.
19988 Do not pack VLIW instructions.
19991 @opindex mno-eflags
19993 Do not mark ABI switches in e_flags.
19996 @opindex mcond-move
19998 Enable the use of conditional-move instructions (default).
20000 This switch is mainly for debugging the compiler and will likely be removed
20001 in a future version.
20003 @item -mno-cond-move
20004 @opindex mno-cond-move
20006 Disable the use of conditional-move instructions.
20008 This switch is mainly for debugging the compiler and will likely be removed
20009 in a future version.
20014 Enable the use of conditional set instructions (default).
20016 This switch is mainly for debugging the compiler and will likely be removed
20017 in a future version.
20022 Disable the use of conditional set instructions.
20024 This switch is mainly for debugging the compiler and will likely be removed
20025 in a future version.
20028 @opindex mcond-exec
20030 Enable the use of conditional execution (default).
20032 This switch is mainly for debugging the compiler and will likely be removed
20033 in a future version.
20035 @item -mno-cond-exec
20036 @opindex mno-cond-exec
20038 Disable the use of conditional execution.
20040 This switch is mainly for debugging the compiler and will likely be removed
20041 in a future version.
20043 @item -mvliw-branch
20044 @opindex mvliw-branch
20046 Run a pass to pack branches into VLIW instructions (default).
20048 This switch is mainly for debugging the compiler and will likely be removed
20049 in a future version.
20051 @item -mno-vliw-branch
20052 @opindex mno-vliw-branch
20054 Do not run a pass to pack branches into VLIW instructions.
20056 This switch is mainly for debugging the compiler and will likely be removed
20057 in a future version.
20059 @item -mmulti-cond-exec
20060 @opindex mmulti-cond-exec
20062 Enable optimization of @code{&&} and @code{||} in conditional execution
20065 This switch is mainly for debugging the compiler and will likely be removed
20066 in a future version.
20068 @item -mno-multi-cond-exec
20069 @opindex mno-multi-cond-exec
20071 Disable optimization of @code{&&} and @code{||} in conditional execution.
20073 This switch is mainly for debugging the compiler and will likely be removed
20074 in a future version.
20076 @item -mnested-cond-exec
20077 @opindex mnested-cond-exec
20079 Enable nested conditional execution optimizations (default).
20081 This switch is mainly for debugging the compiler and will likely be removed
20082 in a future version.
20084 @item -mno-nested-cond-exec
20085 @opindex mno-nested-cond-exec
20087 Disable nested conditional execution optimizations.
20089 This switch is mainly for debugging the compiler and will likely be removed
20090 in a future version.
20092 @item -moptimize-membar
20093 @opindex moptimize-membar
20095 This switch removes redundant @code{membar} instructions from the
20096 compiler-generated code. It is enabled by default.
20098 @item -mno-optimize-membar
20099 @opindex mno-optimize-membar
20100 @opindex moptimize-membar
20102 This switch disables the automatic removal of redundant @code{membar}
20103 instructions from the generated code.
20105 @item -mtomcat-stats
20106 @opindex mtomcat-stats
20108 Cause gas to print out tomcat statistics.
20110 @item -mcpu=@var{cpu}
20113 Select the processor type for which to generate code. Possible values are
20114 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
20115 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
20119 @node GNU/Linux Options
20120 @subsection GNU/Linux Options
20122 These @samp{-m} options are defined for GNU/Linux targets:
20127 Use the GNU C library. This is the default except
20128 on @samp{*-*-linux-*uclibc*}, @samp{*-*-linux-*musl*} and
20129 @samp{*-*-linux-*android*} targets.
20133 Use uClibc C library. This is the default on
20134 @samp{*-*-linux-*uclibc*} targets.
20138 Use the musl C library. This is the default on
20139 @samp{*-*-linux-*musl*} targets.
20143 Use Bionic C library. This is the default on
20144 @samp{*-*-linux-*android*} targets.
20148 Compile code compatible with Android platform. This is the default on
20149 @samp{*-*-linux-*android*} targets.
20151 When compiling, this option enables @option{-mbionic}, @option{-fPIC},
20152 @option{-fno-exceptions} and @option{-fno-rtti} by default. When linking,
20153 this option makes the GCC driver pass Android-specific options to the linker.
20154 Finally, this option causes the preprocessor macro @code{__ANDROID__}
20157 @item -tno-android-cc
20158 @opindex tno-android-cc
20159 Disable compilation effects of @option{-mandroid}, i.e., do not enable
20160 @option{-mbionic}, @option{-fPIC}, @option{-fno-exceptions} and
20161 @option{-fno-rtti} by default.
20163 @item -tno-android-ld
20164 @opindex tno-android-ld
20165 Disable linking effects of @option{-mandroid}, i.e., pass standard Linux
20166 linking options to the linker.
20170 @node H8/300 Options
20171 @subsection H8/300 Options
20173 These @samp{-m} options are defined for the H8/300 implementations:
20178 Shorten some address references at link time, when possible; uses the
20179 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
20180 ld, Using ld}, for a fuller description.
20184 Generate code for the H8/300H@.
20188 Generate code for the H8S@.
20192 Generate code for the H8S and H8/300H in the normal mode. This switch
20193 must be used either with @option{-mh} or @option{-ms}.
20197 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
20201 Extended registers are stored on stack before execution of function
20202 with monitor attribute. Default option is @option{-mexr}.
20203 This option is valid only for H8S targets.
20208 Extended registers are not stored on stack before execution of function
20209 with monitor attribute. Default option is @option{-mno-exr}.
20210 This option is valid only for H8S targets.
20214 Make @code{int} data 32 bits by default.
20217 @opindex malign-300
20218 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
20219 The default for the H8/300H and H8S is to align longs and floats on
20221 @option{-malign-300} causes them to be aligned on 2-byte boundaries.
20222 This option has no effect on the H8/300.
20226 @subsection HPPA Options
20227 @cindex HPPA Options
20229 These @samp{-m} options are defined for the HPPA family of computers:
20232 @item -march=@var{architecture-type}
20234 Generate code for the specified architecture. The choices for
20235 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
20236 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
20237 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
20238 architecture option for your machine. Code compiled for lower numbered
20239 architectures runs on higher numbered architectures, but not the
20242 @item -mpa-risc-1-0
20243 @itemx -mpa-risc-1-1
20244 @itemx -mpa-risc-2-0
20245 @opindex mpa-risc-1-0
20246 @opindex mpa-risc-1-1
20247 @opindex mpa-risc-2-0
20248 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
20250 @item -mcaller-copies
20251 @opindex mcaller-copies
20252 The caller copies function arguments passed by hidden reference. This
20253 option should be used with care as it is not compatible with the default
20254 32-bit runtime. However, only aggregates larger than eight bytes are
20255 passed by hidden reference and the option provides better compatibility
20258 @item -mjump-in-delay
20259 @opindex mjump-in-delay
20260 This option is ignored and provided for compatibility purposes only.
20262 @item -mdisable-fpregs
20263 @opindex mdisable-fpregs
20264 Prevent floating-point registers from being used in any manner. This is
20265 necessary for compiling kernels that perform lazy context switching of
20266 floating-point registers. If you use this option and attempt to perform
20267 floating-point operations, the compiler aborts.
20269 @item -mdisable-indexing
20270 @opindex mdisable-indexing
20271 Prevent the compiler from using indexing address modes. This avoids some
20272 rather obscure problems when compiling MIG generated code under MACH@.
20274 @item -mno-space-regs
20275 @opindex mno-space-regs
20276 @opindex mspace-regs
20277 Generate code that assumes the target has no space registers. This allows
20278 GCC to generate faster indirect calls and use unscaled index address modes.
20280 Such code is suitable for level 0 PA systems and kernels.
20282 @item -mfast-indirect-calls
20283 @opindex mfast-indirect-calls
20284 Generate code that assumes calls never cross space boundaries. This
20285 allows GCC to emit code that performs faster indirect calls.
20287 This option does not work in the presence of shared libraries or nested
20290 @item -mfixed-range=@var{register-range}
20291 @opindex mfixed-range
20292 Generate code treating the given register range as fixed registers.
20293 A fixed register is one that the register allocator cannot use. This is
20294 useful when compiling kernel code. A register range is specified as
20295 two registers separated by a dash. Multiple register ranges can be
20296 specified separated by a comma.
20298 @item -mlong-load-store
20299 @opindex mlong-load-store
20300 Generate 3-instruction load and store sequences as sometimes required by
20301 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
20304 @item -mportable-runtime
20305 @opindex mportable-runtime
20306 Use the portable calling conventions proposed by HP for ELF systems.
20310 Enable the use of assembler directives only GAS understands.
20312 @item -mschedule=@var{cpu-type}
20314 Schedule code according to the constraints for the machine type
20315 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
20316 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
20317 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
20318 proper scheduling option for your machine. The default scheduling is
20322 @opindex mlinker-opt
20323 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
20324 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
20325 linkers in which they give bogus error messages when linking some programs.
20328 @opindex msoft-float
20329 Generate output containing library calls for floating point.
20330 @strong{Warning:} the requisite libraries are not available for all HPPA
20331 targets. Normally the facilities of the machine's usual C compiler are
20332 used, but this cannot be done directly in cross-compilation. You must make
20333 your own arrangements to provide suitable library functions for
20336 @option{-msoft-float} changes the calling convention in the output file;
20337 therefore, it is only useful if you compile @emph{all} of a program with
20338 this option. In particular, you need to compile @file{libgcc.a}, the
20339 library that comes with GCC, with @option{-msoft-float} in order for
20344 Generate the predefine, @code{_SIO}, for server IO@. The default is
20345 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
20346 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
20347 options are available under HP-UX and HI-UX@.
20351 Use options specific to GNU @command{ld}.
20352 This passes @option{-shared} to @command{ld} when
20353 building a shared library. It is the default when GCC is configured,
20354 explicitly or implicitly, with the GNU linker. This option does not
20355 affect which @command{ld} is called; it only changes what parameters
20356 are passed to that @command{ld}.
20357 The @command{ld} that is called is determined by the
20358 @option{--with-ld} configure option, GCC's program search path, and
20359 finally by the user's @env{PATH}. The linker used by GCC can be printed
20360 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
20361 on the 64-bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
20365 Use options specific to HP @command{ld}.
20366 This passes @option{-b} to @command{ld} when building
20367 a shared library and passes @option{+Accept TypeMismatch} to @command{ld} on all
20368 links. It is the default when GCC is configured, explicitly or
20369 implicitly, with the HP linker. This option does not affect
20370 which @command{ld} is called; it only changes what parameters are passed to that
20372 The @command{ld} that is called is determined by the @option{--with-ld}
20373 configure option, GCC's program search path, and finally by the user's
20374 @env{PATH}. The linker used by GCC can be printed using @samp{which
20375 `gcc -print-prog-name=ld`}. This option is only available on the 64-bit
20376 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
20379 @opindex mno-long-calls
20380 @opindex mlong-calls
20381 Generate code that uses long call sequences. This ensures that a call
20382 is always able to reach linker generated stubs. The default is to generate
20383 long calls only when the distance from the call site to the beginning
20384 of the function or translation unit, as the case may be, exceeds a
20385 predefined limit set by the branch type being used. The limits for
20386 normal calls are 7,600,000 and 240,000 bytes, respectively for the
20387 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
20390 Distances are measured from the beginning of functions when using the
20391 @option{-ffunction-sections} option, or when using the @option{-mgas}
20392 and @option{-mno-portable-runtime} options together under HP-UX with
20395 It is normally not desirable to use this option as it degrades
20396 performance. However, it may be useful in large applications,
20397 particularly when partial linking is used to build the application.
20399 The types of long calls used depends on the capabilities of the
20400 assembler and linker, and the type of code being generated. The
20401 impact on systems that support long absolute calls, and long pic
20402 symbol-difference or pc-relative calls should be relatively small.
20403 However, an indirect call is used on 32-bit ELF systems in pic code
20404 and it is quite long.
20406 @item -munix=@var{unix-std}
20408 Generate compiler predefines and select a startfile for the specified
20409 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
20410 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
20411 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
20412 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
20413 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
20416 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
20417 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
20418 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
20419 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
20420 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
20421 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
20423 It is @emph{important} to note that this option changes the interfaces
20424 for various library routines. It also affects the operational behavior
20425 of the C library. Thus, @emph{extreme} care is needed in using this
20428 Library code that is intended to operate with more than one UNIX
20429 standard must test, set and restore the variable @code{__xpg4_extended_mask}
20430 as appropriate. Most GNU software doesn't provide this capability.
20434 Suppress the generation of link options to search libdld.sl when the
20435 @option{-static} option is specified on HP-UX 10 and later.
20439 The HP-UX implementation of setlocale in libc has a dependency on
20440 libdld.sl. There isn't an archive version of libdld.sl. Thus,
20441 when the @option{-static} option is specified, special link options
20442 are needed to resolve this dependency.
20444 On HP-UX 10 and later, the GCC driver adds the necessary options to
20445 link with libdld.sl when the @option{-static} option is specified.
20446 This causes the resulting binary to be dynamic. On the 64-bit port,
20447 the linkers generate dynamic binaries by default in any case. The
20448 @option{-nolibdld} option can be used to prevent the GCC driver from
20449 adding these link options.
20453 Add support for multithreading with the @dfn{dce thread} library
20454 under HP-UX@. This option sets flags for both the preprocessor and
20458 @node IA-64 Options
20459 @subsection IA-64 Options
20460 @cindex IA-64 Options
20462 These are the @samp{-m} options defined for the Intel IA-64 architecture.
20466 @opindex mbig-endian
20467 Generate code for a big-endian target. This is the default for HP-UX@.
20469 @item -mlittle-endian
20470 @opindex mlittle-endian
20471 Generate code for a little-endian target. This is the default for AIX5
20477 @opindex mno-gnu-as
20478 Generate (or don't) code for the GNU assembler. This is the default.
20479 @c Also, this is the default if the configure option @option{--with-gnu-as}
20485 @opindex mno-gnu-ld
20486 Generate (or don't) code for the GNU linker. This is the default.
20487 @c Also, this is the default if the configure option @option{--with-gnu-ld}
20492 Generate code that does not use a global pointer register. The result
20493 is not position independent code, and violates the IA-64 ABI@.
20495 @item -mvolatile-asm-stop
20496 @itemx -mno-volatile-asm-stop
20497 @opindex mvolatile-asm-stop
20498 @opindex mno-volatile-asm-stop
20499 Generate (or don't) a stop bit immediately before and after volatile asm
20502 @item -mregister-names
20503 @itemx -mno-register-names
20504 @opindex mregister-names
20505 @opindex mno-register-names
20506 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
20507 the stacked registers. This may make assembler output more readable.
20513 Disable (or enable) optimizations that use the small data section. This may
20514 be useful for working around optimizer bugs.
20516 @item -mconstant-gp
20517 @opindex mconstant-gp
20518 Generate code that uses a single constant global pointer value. This is
20519 useful when compiling kernel code.
20523 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
20524 This is useful when compiling firmware code.
20526 @item -minline-float-divide-min-latency
20527 @opindex minline-float-divide-min-latency
20528 Generate code for inline divides of floating-point values
20529 using the minimum latency algorithm.
20531 @item -minline-float-divide-max-throughput
20532 @opindex minline-float-divide-max-throughput
20533 Generate code for inline divides of floating-point values
20534 using the maximum throughput algorithm.
20536 @item -mno-inline-float-divide
20537 @opindex mno-inline-float-divide
20538 Do not generate inline code for divides of floating-point values.
20540 @item -minline-int-divide-min-latency
20541 @opindex minline-int-divide-min-latency
20542 Generate code for inline divides of integer values
20543 using the minimum latency algorithm.
20545 @item -minline-int-divide-max-throughput
20546 @opindex minline-int-divide-max-throughput
20547 Generate code for inline divides of integer values
20548 using the maximum throughput algorithm.
20550 @item -mno-inline-int-divide
20551 @opindex mno-inline-int-divide
20552 @opindex minline-int-divide
20553 Do not generate inline code for divides of integer values.
20555 @item -minline-sqrt-min-latency
20556 @opindex minline-sqrt-min-latency
20557 Generate code for inline square roots
20558 using the minimum latency algorithm.
20560 @item -minline-sqrt-max-throughput
20561 @opindex minline-sqrt-max-throughput
20562 Generate code for inline square roots
20563 using the maximum throughput algorithm.
20565 @item -mno-inline-sqrt
20566 @opindex mno-inline-sqrt
20567 Do not generate inline code for @code{sqrt}.
20570 @itemx -mno-fused-madd
20571 @opindex mfused-madd
20572 @opindex mno-fused-madd
20573 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
20574 instructions. The default is to use these instructions.
20576 @item -mno-dwarf2-asm
20577 @itemx -mdwarf2-asm
20578 @opindex mno-dwarf2-asm
20579 @opindex mdwarf2-asm
20580 Don't (or do) generate assembler code for the DWARF line number debugging
20581 info. This may be useful when not using the GNU assembler.
20583 @item -mearly-stop-bits
20584 @itemx -mno-early-stop-bits
20585 @opindex mearly-stop-bits
20586 @opindex mno-early-stop-bits
20587 Allow stop bits to be placed earlier than immediately preceding the
20588 instruction that triggered the stop bit. This can improve instruction
20589 scheduling, but does not always do so.
20591 @item -mfixed-range=@var{register-range}
20592 @opindex mfixed-range
20593 Generate code treating the given register range as fixed registers.
20594 A fixed register is one that the register allocator cannot use. This is
20595 useful when compiling kernel code. A register range is specified as
20596 two registers separated by a dash. Multiple register ranges can be
20597 specified separated by a comma.
20599 @item -mtls-size=@var{tls-size}
20601 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
20604 @item -mtune=@var{cpu-type}
20606 Tune the instruction scheduling for a particular CPU, Valid values are
20607 @samp{itanium}, @samp{itanium1}, @samp{merced}, @samp{itanium2},
20608 and @samp{mckinley}.
20614 Generate code for a 32-bit or 64-bit environment.
20615 The 32-bit environment sets int, long and pointer to 32 bits.
20616 The 64-bit environment sets int to 32 bits and long and pointer
20617 to 64 bits. These are HP-UX specific flags.
20619 @item -mno-sched-br-data-spec
20620 @itemx -msched-br-data-spec
20621 @opindex mno-sched-br-data-spec
20622 @opindex msched-br-data-spec
20623 (Dis/En)able data speculative scheduling before reload.
20624 This results in generation of @code{ld.a} instructions and
20625 the corresponding check instructions (@code{ld.c} / @code{chk.a}).
20626 The default setting is disabled.
20628 @item -msched-ar-data-spec
20629 @itemx -mno-sched-ar-data-spec
20630 @opindex msched-ar-data-spec
20631 @opindex mno-sched-ar-data-spec
20632 (En/Dis)able data speculative scheduling after reload.
20633 This results in generation of @code{ld.a} instructions and
20634 the corresponding check instructions (@code{ld.c} / @code{chk.a}).
20635 The default setting is enabled.
20637 @item -mno-sched-control-spec
20638 @itemx -msched-control-spec
20639 @opindex mno-sched-control-spec
20640 @opindex msched-control-spec
20641 (Dis/En)able control speculative scheduling. This feature is
20642 available only during region scheduling (i.e.@: before reload).
20643 This results in generation of the @code{ld.s} instructions and
20644 the corresponding check instructions @code{chk.s}.
20645 The default setting is disabled.
20647 @item -msched-br-in-data-spec
20648 @itemx -mno-sched-br-in-data-spec
20649 @opindex msched-br-in-data-spec
20650 @opindex mno-sched-br-in-data-spec
20651 (En/Dis)able speculative scheduling of the instructions that
20652 are dependent on the data speculative loads before reload.
20653 This is effective only with @option{-msched-br-data-spec} enabled.
20654 The default setting is enabled.
20656 @item -msched-ar-in-data-spec
20657 @itemx -mno-sched-ar-in-data-spec
20658 @opindex msched-ar-in-data-spec
20659 @opindex mno-sched-ar-in-data-spec
20660 (En/Dis)able speculative scheduling of the instructions that
20661 are dependent on the data speculative loads after reload.
20662 This is effective only with @option{-msched-ar-data-spec} enabled.
20663 The default setting is enabled.
20665 @item -msched-in-control-spec
20666 @itemx -mno-sched-in-control-spec
20667 @opindex msched-in-control-spec
20668 @opindex mno-sched-in-control-spec
20669 (En/Dis)able speculative scheduling of the instructions that
20670 are dependent on the control speculative loads.
20671 This is effective only with @option{-msched-control-spec} enabled.
20672 The default setting is enabled.
20674 @item -mno-sched-prefer-non-data-spec-insns
20675 @itemx -msched-prefer-non-data-spec-insns
20676 @opindex mno-sched-prefer-non-data-spec-insns
20677 @opindex msched-prefer-non-data-spec-insns
20678 If enabled, data-speculative instructions are chosen for schedule
20679 only if there are no other choices at the moment. This makes
20680 the use of the data speculation much more conservative.
20681 The default setting is disabled.
20683 @item -mno-sched-prefer-non-control-spec-insns
20684 @itemx -msched-prefer-non-control-spec-insns
20685 @opindex mno-sched-prefer-non-control-spec-insns
20686 @opindex msched-prefer-non-control-spec-insns
20687 If enabled, control-speculative instructions are chosen for schedule
20688 only if there are no other choices at the moment. This makes
20689 the use of the control speculation much more conservative.
20690 The default setting is disabled.
20692 @item -mno-sched-count-spec-in-critical-path
20693 @itemx -msched-count-spec-in-critical-path
20694 @opindex mno-sched-count-spec-in-critical-path
20695 @opindex msched-count-spec-in-critical-path
20696 If enabled, speculative dependencies are considered during
20697 computation of the instructions priorities. This makes the use of the
20698 speculation a bit more conservative.
20699 The default setting is disabled.
20701 @item -msched-spec-ldc
20702 @opindex msched-spec-ldc
20703 Use a simple data speculation check. This option is on by default.
20705 @item -msched-control-spec-ldc
20706 @opindex msched-spec-ldc
20707 Use a simple check for control speculation. This option is on by default.
20709 @item -msched-stop-bits-after-every-cycle
20710 @opindex msched-stop-bits-after-every-cycle
20711 Place a stop bit after every cycle when scheduling. This option is on
20714 @item -msched-fp-mem-deps-zero-cost
20715 @opindex msched-fp-mem-deps-zero-cost
20716 Assume that floating-point stores and loads are not likely to cause a conflict
20717 when placed into the same instruction group. This option is disabled by
20720 @item -msel-sched-dont-check-control-spec
20721 @opindex msel-sched-dont-check-control-spec
20722 Generate checks for control speculation in selective scheduling.
20723 This flag is disabled by default.
20725 @item -msched-max-memory-insns=@var{max-insns}
20726 @opindex msched-max-memory-insns
20727 Limit on the number of memory insns per instruction group, giving lower
20728 priority to subsequent memory insns attempting to schedule in the same
20729 instruction group. Frequently useful to prevent cache bank conflicts.
20730 The default value is 1.
20732 @item -msched-max-memory-insns-hard-limit
20733 @opindex msched-max-memory-insns-hard-limit
20734 Makes the limit specified by @option{msched-max-memory-insns} a hard limit,
20735 disallowing more than that number in an instruction group.
20736 Otherwise, the limit is ``soft'', meaning that non-memory operations
20737 are preferred when the limit is reached, but memory operations may still
20743 @subsection LM32 Options
20744 @cindex LM32 options
20746 These @option{-m} options are defined for the LatticeMico32 architecture:
20749 @item -mbarrel-shift-enabled
20750 @opindex mbarrel-shift-enabled
20751 Enable barrel-shift instructions.
20753 @item -mdivide-enabled
20754 @opindex mdivide-enabled
20755 Enable divide and modulus instructions.
20757 @item -mmultiply-enabled
20758 @opindex multiply-enabled
20759 Enable multiply instructions.
20761 @item -msign-extend-enabled
20762 @opindex msign-extend-enabled
20763 Enable sign extend instructions.
20765 @item -muser-enabled
20766 @opindex muser-enabled
20767 Enable user-defined instructions.
20772 @subsection M32C Options
20773 @cindex M32C options
20776 @item -mcpu=@var{name}
20778 Select the CPU for which code is generated. @var{name} may be one of
20779 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
20780 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
20781 the M32C/80 series.
20785 Specifies that the program will be run on the simulator. This causes
20786 an alternate runtime library to be linked in which supports, for
20787 example, file I/O@. You must not use this option when generating
20788 programs that will run on real hardware; you must provide your own
20789 runtime library for whatever I/O functions are needed.
20791 @item -memregs=@var{number}
20793 Specifies the number of memory-based pseudo-registers GCC uses
20794 during code generation. These pseudo-registers are used like real
20795 registers, so there is a tradeoff between GCC's ability to fit the
20796 code into available registers, and the performance penalty of using
20797 memory instead of registers. Note that all modules in a program must
20798 be compiled with the same value for this option. Because of that, you
20799 must not use this option with GCC's default runtime libraries.
20803 @node M32R/D Options
20804 @subsection M32R/D Options
20805 @cindex M32R/D options
20807 These @option{-m} options are defined for Renesas M32R/D architectures:
20812 Generate code for the M32R/2@.
20816 Generate code for the M32R/X@.
20820 Generate code for the M32R@. This is the default.
20822 @item -mmodel=small
20823 @opindex mmodel=small
20824 Assume all objects live in the lower 16MB of memory (so that their addresses
20825 can be loaded with the @code{ld24} instruction), and assume all subroutines
20826 are reachable with the @code{bl} instruction.
20827 This is the default.
20829 The addressability of a particular object can be set with the
20830 @code{model} attribute.
20832 @item -mmodel=medium
20833 @opindex mmodel=medium
20834 Assume objects may be anywhere in the 32-bit address space (the compiler
20835 generates @code{seth/add3} instructions to load their addresses), and
20836 assume all subroutines are reachable with the @code{bl} instruction.
20838 @item -mmodel=large
20839 @opindex mmodel=large
20840 Assume objects may be anywhere in the 32-bit address space (the compiler
20841 generates @code{seth/add3} instructions to load their addresses), and
20842 assume subroutines may not be reachable with the @code{bl} instruction
20843 (the compiler generates the much slower @code{seth/add3/jl}
20844 instruction sequence).
20847 @opindex msdata=none
20848 Disable use of the small data area. Variables are put into
20849 one of @code{.data}, @code{.bss}, or @code{.rodata} (unless the
20850 @code{section} attribute has been specified).
20851 This is the default.
20853 The small data area consists of sections @code{.sdata} and @code{.sbss}.
20854 Objects may be explicitly put in the small data area with the
20855 @code{section} attribute using one of these sections.
20857 @item -msdata=sdata
20858 @opindex msdata=sdata
20859 Put small global and static data in the small data area, but do not
20860 generate special code to reference them.
20863 @opindex msdata=use
20864 Put small global and static data in the small data area, and generate
20865 special instructions to reference them.
20869 @cindex smaller data references
20870 Put global and static objects less than or equal to @var{num} bytes
20871 into the small data or BSS sections instead of the normal data or BSS
20872 sections. The default value of @var{num} is 8.
20873 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
20874 for this option to have any effect.
20876 All modules should be compiled with the same @option{-G @var{num}} value.
20877 Compiling with different values of @var{num} may or may not work; if it
20878 doesn't the linker gives an error message---incorrect code is not
20883 Makes the M32R-specific code in the compiler display some statistics
20884 that might help in debugging programs.
20886 @item -malign-loops
20887 @opindex malign-loops
20888 Align all loops to a 32-byte boundary.
20890 @item -mno-align-loops
20891 @opindex mno-align-loops
20892 Do not enforce a 32-byte alignment for loops. This is the default.
20894 @item -missue-rate=@var{number}
20895 @opindex missue-rate=@var{number}
20896 Issue @var{number} instructions per cycle. @var{number} can only be 1
20899 @item -mbranch-cost=@var{number}
20900 @opindex mbranch-cost=@var{number}
20901 @var{number} can only be 1 or 2. If it is 1 then branches are
20902 preferred over conditional code, if it is 2, then the opposite applies.
20904 @item -mflush-trap=@var{number}
20905 @opindex mflush-trap=@var{number}
20906 Specifies the trap number to use to flush the cache. The default is
20907 12. Valid numbers are between 0 and 15 inclusive.
20909 @item -mno-flush-trap
20910 @opindex mno-flush-trap
20911 Specifies that the cache cannot be flushed by using a trap.
20913 @item -mflush-func=@var{name}
20914 @opindex mflush-func=@var{name}
20915 Specifies the name of the operating system function to call to flush
20916 the cache. The default is @samp{_flush_cache}, but a function call
20917 is only used if a trap is not available.
20919 @item -mno-flush-func
20920 @opindex mno-flush-func
20921 Indicates that there is no OS function for flushing the cache.
20925 @node M680x0 Options
20926 @subsection M680x0 Options
20927 @cindex M680x0 options
20929 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
20930 The default settings depend on which architecture was selected when
20931 the compiler was configured; the defaults for the most common choices
20935 @item -march=@var{arch}
20937 Generate code for a specific M680x0 or ColdFire instruction set
20938 architecture. Permissible values of @var{arch} for M680x0
20939 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
20940 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
20941 architectures are selected according to Freescale's ISA classification
20942 and the permissible values are: @samp{isaa}, @samp{isaaplus},
20943 @samp{isab} and @samp{isac}.
20945 GCC defines a macro @code{__mcf@var{arch}__} whenever it is generating
20946 code for a ColdFire target. The @var{arch} in this macro is one of the
20947 @option{-march} arguments given above.
20949 When used together, @option{-march} and @option{-mtune} select code
20950 that runs on a family of similar processors but that is optimized
20951 for a particular microarchitecture.
20953 @item -mcpu=@var{cpu}
20955 Generate code for a specific M680x0 or ColdFire processor.
20956 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
20957 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
20958 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
20959 below, which also classifies the CPUs into families:
20961 @multitable @columnfractions 0.20 0.80
20962 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
20963 @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}
20964 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
20965 @item @samp{5206e} @tab @samp{5206e}
20966 @item @samp{5208} @tab @samp{5207} @samp{5208}
20967 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
20968 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
20969 @item @samp{5216} @tab @samp{5214} @samp{5216}
20970 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
20971 @item @samp{5225} @tab @samp{5224} @samp{5225}
20972 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
20973 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
20974 @item @samp{5249} @tab @samp{5249}
20975 @item @samp{5250} @tab @samp{5250}
20976 @item @samp{5271} @tab @samp{5270} @samp{5271}
20977 @item @samp{5272} @tab @samp{5272}
20978 @item @samp{5275} @tab @samp{5274} @samp{5275}
20979 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
20980 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
20981 @item @samp{5307} @tab @samp{5307}
20982 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
20983 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
20984 @item @samp{5407} @tab @samp{5407}
20985 @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}
20988 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
20989 @var{arch} is compatible with @var{cpu}. Other combinations of
20990 @option{-mcpu} and @option{-march} are rejected.
20992 GCC defines the macro @code{__mcf_cpu_@var{cpu}} when ColdFire target
20993 @var{cpu} is selected. It also defines @code{__mcf_family_@var{family}},
20994 where the value of @var{family} is given by the table above.
20996 @item -mtune=@var{tune}
20998 Tune the code for a particular microarchitecture within the
20999 constraints set by @option{-march} and @option{-mcpu}.
21000 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
21001 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
21002 and @samp{cpu32}. The ColdFire microarchitectures
21003 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
21005 You can also use @option{-mtune=68020-40} for code that needs
21006 to run relatively well on 68020, 68030 and 68040 targets.
21007 @option{-mtune=68020-60} is similar but includes 68060 targets
21008 as well. These two options select the same tuning decisions as
21009 @option{-m68020-40} and @option{-m68020-60} respectively.
21011 GCC defines the macros @code{__mc@var{arch}} and @code{__mc@var{arch}__}
21012 when tuning for 680x0 architecture @var{arch}. It also defines
21013 @code{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
21014 option is used. If GCC is tuning for a range of architectures,
21015 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
21016 it defines the macros for every architecture in the range.
21018 GCC also defines the macro @code{__m@var{uarch}__} when tuning for
21019 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
21020 of the arguments given above.
21026 Generate output for a 68000. This is the default
21027 when the compiler is configured for 68000-based systems.
21028 It is equivalent to @option{-march=68000}.
21030 Use this option for microcontrollers with a 68000 or EC000 core,
21031 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
21035 Generate output for a 68010. This is the default
21036 when the compiler is configured for 68010-based systems.
21037 It is equivalent to @option{-march=68010}.
21043 Generate output for a 68020. This is the default
21044 when the compiler is configured for 68020-based systems.
21045 It is equivalent to @option{-march=68020}.
21049 Generate output for a 68030. This is the default when the compiler is
21050 configured for 68030-based systems. It is equivalent to
21051 @option{-march=68030}.
21055 Generate output for a 68040. This is the default when the compiler is
21056 configured for 68040-based systems. It is equivalent to
21057 @option{-march=68040}.
21059 This option inhibits the use of 68881/68882 instructions that have to be
21060 emulated by software on the 68040. Use this option if your 68040 does not
21061 have code to emulate those instructions.
21065 Generate output for a 68060. This is the default when the compiler is
21066 configured for 68060-based systems. It is equivalent to
21067 @option{-march=68060}.
21069 This option inhibits the use of 68020 and 68881/68882 instructions that
21070 have to be emulated by software on the 68060. Use this option if your 68060
21071 does not have code to emulate those instructions.
21075 Generate output for a CPU32. This is the default
21076 when the compiler is configured for CPU32-based systems.
21077 It is equivalent to @option{-march=cpu32}.
21079 Use this option for microcontrollers with a
21080 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
21081 68336, 68340, 68341, 68349 and 68360.
21085 Generate output for a 520X ColdFire CPU@. This is the default
21086 when the compiler is configured for 520X-based systems.
21087 It is equivalent to @option{-mcpu=5206}, and is now deprecated
21088 in favor of that option.
21090 Use this option for microcontroller with a 5200 core, including
21091 the MCF5202, MCF5203, MCF5204 and MCF5206.
21095 Generate output for a 5206e ColdFire CPU@. The option is now
21096 deprecated in favor of the equivalent @option{-mcpu=5206e}.
21100 Generate output for a member of the ColdFire 528X family.
21101 The option is now deprecated in favor of the equivalent
21102 @option{-mcpu=528x}.
21106 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
21107 in favor of the equivalent @option{-mcpu=5307}.
21111 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
21112 in favor of the equivalent @option{-mcpu=5407}.
21116 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
21117 This includes use of hardware floating-point instructions.
21118 The option is equivalent to @option{-mcpu=547x}, and is now
21119 deprecated in favor of that option.
21123 Generate output for a 68040, without using any of the new instructions.
21124 This results in code that can run relatively efficiently on either a
21125 68020/68881 or a 68030 or a 68040. The generated code does use the
21126 68881 instructions that are emulated on the 68040.
21128 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
21132 Generate output for a 68060, without using any of the new instructions.
21133 This results in code that can run relatively efficiently on either a
21134 68020/68881 or a 68030 or a 68040. The generated code does use the
21135 68881 instructions that are emulated on the 68060.
21137 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
21141 @opindex mhard-float
21143 Generate floating-point instructions. This is the default for 68020
21144 and above, and for ColdFire devices that have an FPU@. It defines the
21145 macro @code{__HAVE_68881__} on M680x0 targets and @code{__mcffpu__}
21146 on ColdFire targets.
21149 @opindex msoft-float
21150 Do not generate floating-point instructions; use library calls instead.
21151 This is the default for 68000, 68010, and 68832 targets. It is also
21152 the default for ColdFire devices that have no FPU.
21158 Generate (do not generate) ColdFire hardware divide and remainder
21159 instructions. If @option{-march} is used without @option{-mcpu},
21160 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
21161 architectures. Otherwise, the default is taken from the target CPU
21162 (either the default CPU, or the one specified by @option{-mcpu}). For
21163 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
21164 @option{-mcpu=5206e}.
21166 GCC defines the macro @code{__mcfhwdiv__} when this option is enabled.
21170 Consider type @code{int} to be 16 bits wide, like @code{short int}.
21171 Additionally, parameters passed on the stack are also aligned to a
21172 16-bit boundary even on targets whose API mandates promotion to 32-bit.
21176 Do not consider type @code{int} to be 16 bits wide. This is the default.
21179 @itemx -mno-bitfield
21180 @opindex mnobitfield
21181 @opindex mno-bitfield
21182 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
21183 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
21187 Do use the bit-field instructions. The @option{-m68020} option implies
21188 @option{-mbitfield}. This is the default if you use a configuration
21189 designed for a 68020.
21193 Use a different function-calling convention, in which functions
21194 that take a fixed number of arguments return with the @code{rtd}
21195 instruction, which pops their arguments while returning. This
21196 saves one instruction in the caller since there is no need to pop
21197 the arguments there.
21199 This calling convention is incompatible with the one normally
21200 used on Unix, so you cannot use it if you need to call libraries
21201 compiled with the Unix compiler.
21203 Also, you must provide function prototypes for all functions that
21204 take variable numbers of arguments (including @code{printf});
21205 otherwise incorrect code is generated for calls to those
21208 In addition, seriously incorrect code results if you call a
21209 function with too many arguments. (Normally, extra arguments are
21210 harmlessly ignored.)
21212 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
21213 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
21215 The default is @option{-mno-rtd}.
21218 @itemx -mno-align-int
21219 @opindex malign-int
21220 @opindex mno-align-int
21221 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
21222 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
21223 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
21224 Aligning variables on 32-bit boundaries produces code that runs somewhat
21225 faster on processors with 32-bit busses at the expense of more memory.
21227 @strong{Warning:} if you use the @option{-malign-int} switch, GCC
21228 aligns structures containing the above types differently than
21229 most published application binary interface specifications for the m68k.
21233 Use the pc-relative addressing mode of the 68000 directly, instead of
21234 using a global offset table. At present, this option implies @option{-fpic},
21235 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
21236 not presently supported with @option{-mpcrel}, though this could be supported for
21237 68020 and higher processors.
21239 @item -mno-strict-align
21240 @itemx -mstrict-align
21241 @opindex mno-strict-align
21242 @opindex mstrict-align
21243 Do not (do) assume that unaligned memory references are handled by
21247 Generate code that allows the data segment to be located in a different
21248 area of memory from the text segment. This allows for execute-in-place in
21249 an environment without virtual memory management. This option implies
21252 @item -mno-sep-data
21253 Generate code that assumes that the data segment follows the text segment.
21254 This is the default.
21256 @item -mid-shared-library
21257 Generate code that supports shared libraries via the library ID method.
21258 This allows for execute-in-place and shared libraries in an environment
21259 without virtual memory management. This option implies @option{-fPIC}.
21261 @item -mno-id-shared-library
21262 Generate code that doesn't assume ID-based shared libraries are being used.
21263 This is the default.
21265 @item -mshared-library-id=n
21266 Specifies the identification number of the ID-based shared library being
21267 compiled. Specifying a value of 0 generates more compact code; specifying
21268 other values forces the allocation of that number to the current
21269 library, but is no more space- or time-efficient than omitting this option.
21275 When generating position-independent code for ColdFire, generate code
21276 that works if the GOT has more than 8192 entries. This code is
21277 larger and slower than code generated without this option. On M680x0
21278 processors, this option is not needed; @option{-fPIC} suffices.
21280 GCC normally uses a single instruction to load values from the GOT@.
21281 While this is relatively efficient, it only works if the GOT
21282 is smaller than about 64k. Anything larger causes the linker
21283 to report an error such as:
21285 @cindex relocation truncated to fit (ColdFire)
21287 relocation truncated to fit: R_68K_GOT16O foobar
21290 If this happens, you should recompile your code with @option{-mxgot}.
21291 It should then work with very large GOTs. However, code generated with
21292 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
21293 the value of a global symbol.
21295 Note that some linkers, including newer versions of the GNU linker,
21296 can create multiple GOTs and sort GOT entries. If you have such a linker,
21297 you should only need to use @option{-mxgot} when compiling a single
21298 object file that accesses more than 8192 GOT entries. Very few do.
21300 These options have no effect unless GCC is generating
21301 position-independent code.
21303 @item -mlong-jump-table-offsets
21304 @opindex mlong-jump-table-offsets
21305 Use 32-bit offsets in @code{switch} tables. The default is to use
21310 @node MCore Options
21311 @subsection MCore Options
21312 @cindex MCore options
21314 These are the @samp{-m} options defined for the Motorola M*Core
21320 @itemx -mno-hardlit
21322 @opindex mno-hardlit
21323 Inline constants into the code stream if it can be done in two
21324 instructions or less.
21330 Use the divide instruction. (Enabled by default).
21332 @item -mrelax-immediate
21333 @itemx -mno-relax-immediate
21334 @opindex mrelax-immediate
21335 @opindex mno-relax-immediate
21336 Allow arbitrary-sized immediates in bit operations.
21338 @item -mwide-bitfields
21339 @itemx -mno-wide-bitfields
21340 @opindex mwide-bitfields
21341 @opindex mno-wide-bitfields
21342 Always treat bit-fields as @code{int}-sized.
21344 @item -m4byte-functions
21345 @itemx -mno-4byte-functions
21346 @opindex m4byte-functions
21347 @opindex mno-4byte-functions
21348 Force all functions to be aligned to a 4-byte boundary.
21350 @item -mcallgraph-data
21351 @itemx -mno-callgraph-data
21352 @opindex mcallgraph-data
21353 @opindex mno-callgraph-data
21354 Emit callgraph information.
21357 @itemx -mno-slow-bytes
21358 @opindex mslow-bytes
21359 @opindex mno-slow-bytes
21360 Prefer word access when reading byte quantities.
21362 @item -mlittle-endian
21363 @itemx -mbig-endian
21364 @opindex mlittle-endian
21365 @opindex mbig-endian
21366 Generate code for a little-endian target.
21372 Generate code for the 210 processor.
21376 Assume that runtime support has been provided and so omit the
21377 simulator library (@file{libsim.a)} from the linker command line.
21379 @item -mstack-increment=@var{size}
21380 @opindex mstack-increment
21381 Set the maximum amount for a single stack increment operation. Large
21382 values can increase the speed of programs that contain functions
21383 that need a large amount of stack space, but they can also trigger a
21384 segmentation fault if the stack is extended too much. The default
21390 @subsection MeP Options
21391 @cindex MeP options
21397 Enables the @code{abs} instruction, which is the absolute difference
21398 between two registers.
21402 Enables all the optional instructions---average, multiply, divide, bit
21403 operations, leading zero, absolute difference, min/max, clip, and
21409 Enables the @code{ave} instruction, which computes the average of two
21412 @item -mbased=@var{n}
21414 Variables of size @var{n} bytes or smaller are placed in the
21415 @code{.based} section by default. Based variables use the @code{$tp}
21416 register as a base register, and there is a 128-byte limit to the
21417 @code{.based} section.
21421 Enables the bit operation instructions---bit test (@code{btstm}), set
21422 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
21423 test-and-set (@code{tas}).
21425 @item -mc=@var{name}
21427 Selects which section constant data is placed in. @var{name} may
21428 be @samp{tiny}, @samp{near}, or @samp{far}.
21432 Enables the @code{clip} instruction. Note that @option{-mclip} is not
21433 useful unless you also provide @option{-mminmax}.
21435 @item -mconfig=@var{name}
21437 Selects one of the built-in core configurations. Each MeP chip has
21438 one or more modules in it; each module has a core CPU and a variety of
21439 coprocessors, optional instructions, and peripherals. The
21440 @code{MeP-Integrator} tool, not part of GCC, provides these
21441 configurations through this option; using this option is the same as
21442 using all the corresponding command-line options. The default
21443 configuration is @samp{default}.
21447 Enables the coprocessor instructions. By default, this is a 32-bit
21448 coprocessor. Note that the coprocessor is normally enabled via the
21449 @option{-mconfig=} option.
21453 Enables the 32-bit coprocessor's instructions.
21457 Enables the 64-bit coprocessor's instructions.
21461 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
21465 Causes constant variables to be placed in the @code{.near} section.
21469 Enables the @code{div} and @code{divu} instructions.
21473 Generate big-endian code.
21477 Generate little-endian code.
21479 @item -mio-volatile
21480 @opindex mio-volatile
21481 Tells the compiler that any variable marked with the @code{io}
21482 attribute is to be considered volatile.
21486 Causes variables to be assigned to the @code{.far} section by default.
21490 Enables the @code{leadz} (leading zero) instruction.
21494 Causes variables to be assigned to the @code{.near} section by default.
21498 Enables the @code{min} and @code{max} instructions.
21502 Enables the multiplication and multiply-accumulate instructions.
21506 Disables all the optional instructions enabled by @option{-mall-opts}.
21510 Enables the @code{repeat} and @code{erepeat} instructions, used for
21511 low-overhead looping.
21515 Causes all variables to default to the @code{.tiny} section. Note
21516 that there is a 65536-byte limit to this section. Accesses to these
21517 variables use the @code{%gp} base register.
21521 Enables the saturation instructions. Note that the compiler does not
21522 currently generate these itself, but this option is included for
21523 compatibility with other tools, like @code{as}.
21527 Link the SDRAM-based runtime instead of the default ROM-based runtime.
21531 Link the simulator run-time libraries.
21535 Link the simulator runtime libraries, excluding built-in support
21536 for reset and exception vectors and tables.
21540 Causes all functions to default to the @code{.far} section. Without
21541 this option, functions default to the @code{.near} section.
21543 @item -mtiny=@var{n}
21545 Variables that are @var{n} bytes or smaller are allocated to the
21546 @code{.tiny} section. These variables use the @code{$gp} base
21547 register. The default for this option is 4, but note that there's a
21548 65536-byte limit to the @code{.tiny} section.
21552 @node MicroBlaze Options
21553 @subsection MicroBlaze Options
21554 @cindex MicroBlaze Options
21559 @opindex msoft-float
21560 Use software emulation for floating point (default).
21563 @opindex mhard-float
21564 Use hardware floating-point instructions.
21568 Do not optimize block moves, use @code{memcpy}.
21570 @item -mno-clearbss
21571 @opindex mno-clearbss
21572 This option is deprecated. Use @option{-fno-zero-initialized-in-bss} instead.
21574 @item -mcpu=@var{cpu-type}
21576 Use features of, and schedule code for, the given CPU.
21577 Supported values are in the format @samp{v@var{X}.@var{YY}.@var{Z}},
21578 where @var{X} is a major version, @var{YY} is the minor version, and
21579 @var{Z} is compatibility code. Example values are @samp{v3.00.a},
21580 @samp{v4.00.b}, @samp{v5.00.a}, @samp{v5.00.b}, @samp{v6.00.a}.
21582 @item -mxl-soft-mul
21583 @opindex mxl-soft-mul
21584 Use software multiply emulation (default).
21586 @item -mxl-soft-div
21587 @opindex mxl-soft-div
21588 Use software emulation for divides (default).
21590 @item -mxl-barrel-shift
21591 @opindex mxl-barrel-shift
21592 Use the hardware barrel shifter.
21594 @item -mxl-pattern-compare
21595 @opindex mxl-pattern-compare
21596 Use pattern compare instructions.
21598 @item -msmall-divides
21599 @opindex msmall-divides
21600 Use table lookup optimization for small signed integer divisions.
21602 @item -mxl-stack-check
21603 @opindex mxl-stack-check
21604 This option is deprecated. Use @option{-fstack-check} instead.
21607 @opindex mxl-gp-opt
21608 Use GP-relative @code{.sdata}/@code{.sbss} sections.
21610 @item -mxl-multiply-high
21611 @opindex mxl-multiply-high
21612 Use multiply high instructions for high part of 32x32 multiply.
21614 @item -mxl-float-convert
21615 @opindex mxl-float-convert
21616 Use hardware floating-point conversion instructions.
21618 @item -mxl-float-sqrt
21619 @opindex mxl-float-sqrt
21620 Use hardware floating-point square root instruction.
21623 @opindex mbig-endian
21624 Generate code for a big-endian target.
21626 @item -mlittle-endian
21627 @opindex mlittle-endian
21628 Generate code for a little-endian target.
21631 @opindex mxl-reorder
21632 Use reorder instructions (swap and byte reversed load/store).
21634 @item -mxl-mode-@var{app-model}
21635 Select application model @var{app-model}. Valid models are
21638 normal executable (default), uses startup code @file{crt0.o}.
21640 @item -mpic-data-is-text-relative
21641 @opindex mpic-data-is-text-relative
21642 Assume that the displacement between the text and data segments is fixed
21643 at static link time. This allows data to be referenced by offset from start of
21644 text address instead of GOT since PC-relative addressing is not supported.
21647 for use with Xilinx Microprocessor Debugger (XMD) based
21648 software intrusive debug agent called xmdstub. This uses startup file
21649 @file{crt1.o} and sets the start address of the program to 0x800.
21652 for applications that are loaded using a bootloader.
21653 This model uses startup file @file{crt2.o} which does not contain a processor
21654 reset vector handler. This is suitable for transferring control on a
21655 processor reset to the bootloader rather than the application.
21658 for applications that do not require any of the
21659 MicroBlaze vectors. This option may be useful for applications running
21660 within a monitoring application. This model uses @file{crt3.o} as a startup file.
21663 Option @option{-xl-mode-@var{app-model}} is a deprecated alias for
21664 @option{-mxl-mode-@var{app-model}}.
21669 @subsection MIPS Options
21670 @cindex MIPS options
21676 Generate big-endian code.
21680 Generate little-endian code. This is the default for @samp{mips*el-*-*}
21683 @item -march=@var{arch}
21685 Generate code that runs on @var{arch}, which can be the name of a
21686 generic MIPS ISA, or the name of a particular processor.
21688 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
21689 @samp{mips32}, @samp{mips32r2}, @samp{mips32r3}, @samp{mips32r5},
21690 @samp{mips32r6}, @samp{mips64}, @samp{mips64r2}, @samp{mips64r3},
21691 @samp{mips64r5} and @samp{mips64r6}.
21692 The processor names are:
21693 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
21694 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
21695 @samp{5kc}, @samp{5kf},
21697 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
21698 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
21699 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1}, @samp{34kn},
21700 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
21701 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
21702 @samp{i6400}, @samp{i6500},
21704 @samp{loongson2e}, @samp{loongson2f}, @samp{loongson3a}, @samp{gs464},
21705 @samp{gs464e}, @samp{gs264e},
21707 @samp{m14k}, @samp{m14kc}, @samp{m14ke}, @samp{m14kec},
21708 @samp{m5100}, @samp{m5101},
21709 @samp{octeon}, @samp{octeon+}, @samp{octeon2}, @samp{octeon3},
21711 @samp{p5600}, @samp{p6600},
21712 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
21713 @samp{r4600}, @samp{r4650}, @samp{r4700}, @samp{r5900},
21714 @samp{r6000}, @samp{r8000},
21715 @samp{rm7000}, @samp{rm9000},
21716 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
21719 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
21720 @samp{vr5000}, @samp{vr5400}, @samp{vr5500},
21721 @samp{xlr} and @samp{xlp}.
21722 The special value @samp{from-abi} selects the
21723 most compatible architecture for the selected ABI (that is,
21724 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
21726 The native Linux/GNU toolchain also supports the value @samp{native},
21727 which selects the best architecture option for the host processor.
21728 @option{-march=native} has no effect if GCC does not recognize
21731 In processor names, a final @samp{000} can be abbreviated as @samp{k}
21732 (for example, @option{-march=r2k}). Prefixes are optional, and
21733 @samp{vr} may be written @samp{r}.
21735 Names of the form @samp{@var{n}f2_1} refer to processors with
21736 FPUs clocked at half the rate of the core, names of the form
21737 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
21738 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
21739 processors with FPUs clocked a ratio of 3:2 with respect to the core.
21740 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
21741 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
21742 accepted as synonyms for @samp{@var{n}f1_1}.
21744 GCC defines two macros based on the value of this option. The first
21745 is @code{_MIPS_ARCH}, which gives the name of target architecture, as
21746 a string. The second has the form @code{_MIPS_ARCH_@var{foo}},
21747 where @var{foo} is the capitalized value of @code{_MIPS_ARCH}@.
21748 For example, @option{-march=r2000} sets @code{_MIPS_ARCH}
21749 to @code{"r2000"} and defines the macro @code{_MIPS_ARCH_R2000}.
21751 Note that the @code{_MIPS_ARCH} macro uses the processor names given
21752 above. In other words, it has the full prefix and does not
21753 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
21754 the macro names the resolved architecture (either @code{"mips1"} or
21755 @code{"mips3"}). It names the default architecture when no
21756 @option{-march} option is given.
21758 @item -mtune=@var{arch}
21760 Optimize for @var{arch}. Among other things, this option controls
21761 the way instructions are scheduled, and the perceived cost of arithmetic
21762 operations. The list of @var{arch} values is the same as for
21765 When this option is not used, GCC optimizes for the processor
21766 specified by @option{-march}. By using @option{-march} and
21767 @option{-mtune} together, it is possible to generate code that
21768 runs on a family of processors, but optimize the code for one
21769 particular member of that family.
21771 @option{-mtune} defines the macros @code{_MIPS_TUNE} and
21772 @code{_MIPS_TUNE_@var{foo}}, which work in the same way as the
21773 @option{-march} ones described above.
21777 Equivalent to @option{-march=mips1}.
21781 Equivalent to @option{-march=mips2}.
21785 Equivalent to @option{-march=mips3}.
21789 Equivalent to @option{-march=mips4}.
21793 Equivalent to @option{-march=mips32}.
21797 Equivalent to @option{-march=mips32r3}.
21801 Equivalent to @option{-march=mips32r5}.
21805 Equivalent to @option{-march=mips32r6}.
21809 Equivalent to @option{-march=mips64}.
21813 Equivalent to @option{-march=mips64r2}.
21817 Equivalent to @option{-march=mips64r3}.
21821 Equivalent to @option{-march=mips64r5}.
21825 Equivalent to @option{-march=mips64r6}.
21830 @opindex mno-mips16
21831 Generate (do not generate) MIPS16 code. If GCC is targeting a
21832 MIPS32 or MIPS64 architecture, it makes use of the MIPS16e ASE@.
21834 MIPS16 code generation can also be controlled on a per-function basis
21835 by means of @code{mips16} and @code{nomips16} attributes.
21836 @xref{Function Attributes}, for more information.
21838 @item -mflip-mips16
21839 @opindex mflip-mips16
21840 Generate MIPS16 code on alternating functions. This option is provided
21841 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
21842 not intended for ordinary use in compiling user code.
21844 @item -minterlink-compressed
21845 @itemx -mno-interlink-compressed
21846 @opindex minterlink-compressed
21847 @opindex mno-interlink-compressed
21848 Require (do not require) that code using the standard (uncompressed) MIPS ISA
21849 be link-compatible with MIPS16 and microMIPS code, and vice versa.
21851 For example, code using the standard ISA encoding cannot jump directly
21852 to MIPS16 or microMIPS code; it must either use a call or an indirect jump.
21853 @option{-minterlink-compressed} therefore disables direct jumps unless GCC
21854 knows that the target of the jump is not compressed.
21856 @item -minterlink-mips16
21857 @itemx -mno-interlink-mips16
21858 @opindex minterlink-mips16
21859 @opindex mno-interlink-mips16
21860 Aliases of @option{-minterlink-compressed} and
21861 @option{-mno-interlink-compressed}. These options predate the microMIPS ASE
21862 and are retained for backwards compatibility.
21874 Generate code for the given ABI@.
21876 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
21877 generates 64-bit code when you select a 64-bit architecture, but you
21878 can use @option{-mgp32} to get 32-bit code instead.
21880 For information about the O64 ABI, see
21881 @uref{http://gcc.gnu.org/@/projects/@/mipso64-abi.html}.
21883 GCC supports a variant of the o32 ABI in which floating-point registers
21884 are 64 rather than 32 bits wide. You can select this combination with
21885 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @code{mthc1}
21886 and @code{mfhc1} instructions and is therefore only supported for
21887 MIPS32R2, MIPS32R3 and MIPS32R5 processors.
21889 The register assignments for arguments and return values remain the
21890 same, but each scalar value is passed in a single 64-bit register
21891 rather than a pair of 32-bit registers. For example, scalar
21892 floating-point values are returned in @samp{$f0} only, not a
21893 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
21894 remains the same in that the even-numbered double-precision registers
21897 Two additional variants of the o32 ABI are supported to enable
21898 a transition from 32-bit to 64-bit registers. These are FPXX
21899 (@option{-mfpxx}) and FP64A (@option{-mfp64} @option{-mno-odd-spreg}).
21900 The FPXX extension mandates that all code must execute correctly
21901 when run using 32-bit or 64-bit registers. The code can be interlinked
21902 with either FP32 or FP64, but not both.
21903 The FP64A extension is similar to the FP64 extension but forbids the
21904 use of odd-numbered single-precision registers. This can be used
21905 in conjunction with the @code{FRE} mode of FPUs in MIPS32R5
21906 processors and allows both FP32 and FP64A code to interlink and
21907 run in the same process without changing FPU modes.
21910 @itemx -mno-abicalls
21912 @opindex mno-abicalls
21913 Generate (do not generate) code that is suitable for SVR4-style
21914 dynamic objects. @option{-mabicalls} is the default for SVR4-based
21919 Generate (do not generate) code that is fully position-independent,
21920 and that can therefore be linked into shared libraries. This option
21921 only affects @option{-mabicalls}.
21923 All @option{-mabicalls} code has traditionally been position-independent,
21924 regardless of options like @option{-fPIC} and @option{-fpic}. However,
21925 as an extension, the GNU toolchain allows executables to use absolute
21926 accesses for locally-binding symbols. It can also use shorter GP
21927 initialization sequences and generate direct calls to locally-defined
21928 functions. This mode is selected by @option{-mno-shared}.
21930 @option{-mno-shared} depends on binutils 2.16 or higher and generates
21931 objects that can only be linked by the GNU linker. However, the option
21932 does not affect the ABI of the final executable; it only affects the ABI
21933 of relocatable objects. Using @option{-mno-shared} generally makes
21934 executables both smaller and quicker.
21936 @option{-mshared} is the default.
21942 Assume (do not assume) that the static and dynamic linkers
21943 support PLTs and copy relocations. This option only affects
21944 @option{-mno-shared -mabicalls}. For the n64 ABI, this option
21945 has no effect without @option{-msym32}.
21947 You can make @option{-mplt} the default by configuring
21948 GCC with @option{--with-mips-plt}. The default is
21949 @option{-mno-plt} otherwise.
21955 Lift (do not lift) the usual restrictions on the size of the global
21958 GCC normally uses a single instruction to load values from the GOT@.
21959 While this is relatively efficient, it only works if the GOT
21960 is smaller than about 64k. Anything larger causes the linker
21961 to report an error such as:
21963 @cindex relocation truncated to fit (MIPS)
21965 relocation truncated to fit: R_MIPS_GOT16 foobar
21968 If this happens, you should recompile your code with @option{-mxgot}.
21969 This works with very large GOTs, although the code is also
21970 less efficient, since it takes three instructions to fetch the
21971 value of a global symbol.
21973 Note that some linkers can create multiple GOTs. If you have such a
21974 linker, you should only need to use @option{-mxgot} when a single object
21975 file accesses more than 64k's worth of GOT entries. Very few do.
21977 These options have no effect unless GCC is generating position
21982 Assume that general-purpose registers are 32 bits wide.
21986 Assume that general-purpose registers are 64 bits wide.
21990 Assume that floating-point registers are 32 bits wide.
21994 Assume that floating-point registers are 64 bits wide.
21998 Do not assume the width of floating-point registers.
22001 @opindex mhard-float
22002 Use floating-point coprocessor instructions.
22005 @opindex msoft-float
22006 Do not use floating-point coprocessor instructions. Implement
22007 floating-point calculations using library calls instead.
22011 Equivalent to @option{-msoft-float}, but additionally asserts that the
22012 program being compiled does not perform any floating-point operations.
22013 This option is presently supported only by some bare-metal MIPS
22014 configurations, where it may select a special set of libraries
22015 that lack all floating-point support (including, for example, the
22016 floating-point @code{printf} formats).
22017 If code compiled with @option{-mno-float} accidentally contains
22018 floating-point operations, it is likely to suffer a link-time
22019 or run-time failure.
22021 @item -msingle-float
22022 @opindex msingle-float
22023 Assume that the floating-point coprocessor only supports single-precision
22026 @item -mdouble-float
22027 @opindex mdouble-float
22028 Assume that the floating-point coprocessor supports double-precision
22029 operations. This is the default.
22032 @itemx -mno-odd-spreg
22033 @opindex modd-spreg
22034 @opindex mno-odd-spreg
22035 Enable the use of odd-numbered single-precision floating-point registers
22036 for the o32 ABI. This is the default for processors that are known to
22037 support these registers. When using the o32 FPXX ABI, @option{-mno-odd-spreg}
22041 @itemx -mabs=legacy
22043 @opindex mabs=legacy
22044 These options control the treatment of the special not-a-number (NaN)
22045 IEEE 754 floating-point data with the @code{abs.@i{fmt}} and
22046 @code{neg.@i{fmt}} machine instructions.
22048 By default or when @option{-mabs=legacy} is used the legacy
22049 treatment is selected. In this case these instructions are considered
22050 arithmetic and avoided where correct operation is required and the
22051 input operand might be a NaN. A longer sequence of instructions that
22052 manipulate the sign bit of floating-point datum manually is used
22053 instead unless the @option{-ffinite-math-only} option has also been
22056 The @option{-mabs=2008} option selects the IEEE 754-2008 treatment. In
22057 this case these instructions are considered non-arithmetic and therefore
22058 operating correctly in all cases, including in particular where the
22059 input operand is a NaN. These instructions are therefore always used
22060 for the respective operations.
22063 @itemx -mnan=legacy
22065 @opindex mnan=legacy
22066 These options control the encoding of the special not-a-number (NaN)
22067 IEEE 754 floating-point data.
22069 The @option{-mnan=legacy} option selects the legacy encoding. In this
22070 case quiet NaNs (qNaNs) are denoted by the first bit of their trailing
22071 significand field being 0, whereas signaling NaNs (sNaNs) are denoted
22072 by the first bit of their trailing significand field being 1.
22074 The @option{-mnan=2008} option selects the IEEE 754-2008 encoding. In
22075 this case qNaNs are denoted by the first bit of their trailing
22076 significand field being 1, whereas sNaNs are denoted by the first bit of
22077 their trailing significand field being 0.
22079 The default is @option{-mnan=legacy} unless GCC has been configured with
22080 @option{--with-nan=2008}.
22086 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
22087 implement atomic memory built-in functions. When neither option is
22088 specified, GCC uses the instructions if the target architecture
22091 @option{-mllsc} is useful if the runtime environment can emulate the
22092 instructions and @option{-mno-llsc} can be useful when compiling for
22093 nonstandard ISAs. You can make either option the default by
22094 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
22095 respectively. @option{--with-llsc} is the default for some
22096 configurations; see the installation documentation for details.
22102 Use (do not use) revision 1 of the MIPS DSP ASE@.
22103 @xref{MIPS DSP Built-in Functions}. This option defines the
22104 preprocessor macro @code{__mips_dsp}. It also defines
22105 @code{__mips_dsp_rev} to 1.
22111 Use (do not use) revision 2 of the MIPS DSP ASE@.
22112 @xref{MIPS DSP Built-in Functions}. This option defines the
22113 preprocessor macros @code{__mips_dsp} and @code{__mips_dspr2}.
22114 It also defines @code{__mips_dsp_rev} to 2.
22117 @itemx -mno-smartmips
22118 @opindex msmartmips
22119 @opindex mno-smartmips
22120 Use (do not use) the MIPS SmartMIPS ASE.
22122 @item -mpaired-single
22123 @itemx -mno-paired-single
22124 @opindex mpaired-single
22125 @opindex mno-paired-single
22126 Use (do not use) paired-single floating-point instructions.
22127 @xref{MIPS Paired-Single Support}. This option requires
22128 hardware floating-point support to be enabled.
22134 Use (do not use) MIPS Digital Media Extension instructions.
22135 This option can only be used when generating 64-bit code and requires
22136 hardware floating-point support to be enabled.
22141 @opindex mno-mips3d
22142 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
22143 The option @option{-mips3d} implies @option{-mpaired-single}.
22146 @itemx -mno-micromips
22147 @opindex mmicromips
22148 @opindex mno-mmicromips
22149 Generate (do not generate) microMIPS code.
22151 MicroMIPS code generation can also be controlled on a per-function basis
22152 by means of @code{micromips} and @code{nomicromips} attributes.
22153 @xref{Function Attributes}, for more information.
22159 Use (do not use) MT Multithreading instructions.
22165 Use (do not use) the MIPS MCU ASE instructions.
22171 Use (do not use) the MIPS Enhanced Virtual Addressing instructions.
22177 Use (do not use) the MIPS Virtualization (VZ) instructions.
22183 Use (do not use) the MIPS eXtended Physical Address (XPA) instructions.
22189 Use (do not use) the MIPS Cyclic Redundancy Check (CRC) instructions.
22195 Use (do not use) the MIPS Global INValidate (GINV) instructions.
22197 @item -mloongson-mmi
22198 @itemx -mno-loongson-mmi
22199 @opindex mloongson-mmi
22200 @opindex mno-loongson-mmi
22201 Use (do not use) the MIPS Loongson MultiMedia extensions Instructions (MMI).
22203 @item -mloongson-ext
22204 @itemx -mno-loongson-ext
22205 @opindex mloongson-ext
22206 @opindex mno-loongson-ext
22207 Use (do not use) the MIPS Loongson EXTensions (EXT) instructions.
22209 @item -mloongson-ext2
22210 @itemx -mno-loongson-ext2
22211 @opindex mloongson-ext2
22212 @opindex mno-loongson-ext2
22213 Use (do not use) the MIPS Loongson EXTensions r2 (EXT2) instructions.
22217 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
22218 an explanation of the default and the way that the pointer size is
22223 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
22225 The default size of @code{int}s, @code{long}s and pointers depends on
22226 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
22227 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
22228 32-bit @code{long}s. Pointers are the same size as @code{long}s,
22229 or the same size as integer registers, whichever is smaller.
22235 Assume (do not assume) that all symbols have 32-bit values, regardless
22236 of the selected ABI@. This option is useful in combination with
22237 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
22238 to generate shorter and faster references to symbolic addresses.
22242 Put definitions of externally-visible data in a small data section
22243 if that data is no bigger than @var{num} bytes. GCC can then generate
22244 more efficient accesses to the data; see @option{-mgpopt} for details.
22246 The default @option{-G} option depends on the configuration.
22248 @item -mlocal-sdata
22249 @itemx -mno-local-sdata
22250 @opindex mlocal-sdata
22251 @opindex mno-local-sdata
22252 Extend (do not extend) the @option{-G} behavior to local data too,
22253 such as to static variables in C@. @option{-mlocal-sdata} is the
22254 default for all configurations.
22256 If the linker complains that an application is using too much small data,
22257 you might want to try rebuilding the less performance-critical parts with
22258 @option{-mno-local-sdata}. You might also want to build large
22259 libraries with @option{-mno-local-sdata}, so that the libraries leave
22260 more room for the main program.
22262 @item -mextern-sdata
22263 @itemx -mno-extern-sdata
22264 @opindex mextern-sdata
22265 @opindex mno-extern-sdata
22266 Assume (do not assume) that externally-defined data is in
22267 a small data section if the size of that data is within the @option{-G} limit.
22268 @option{-mextern-sdata} is the default for all configurations.
22270 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
22271 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
22272 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
22273 is placed in a small data section. If @var{Var} is defined by another
22274 module, you must either compile that module with a high-enough
22275 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
22276 definition. If @var{Var} is common, you must link the application
22277 with a high-enough @option{-G} setting.
22279 The easiest way of satisfying these restrictions is to compile
22280 and link every module with the same @option{-G} option. However,
22281 you may wish to build a library that supports several different
22282 small data limits. You can do this by compiling the library with
22283 the highest supported @option{-G} setting and additionally using
22284 @option{-mno-extern-sdata} to stop the library from making assumptions
22285 about externally-defined data.
22291 Use (do not use) GP-relative accesses for symbols that are known to be
22292 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
22293 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
22296 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
22297 might not hold the value of @code{_gp}. For example, if the code is
22298 part of a library that might be used in a boot monitor, programs that
22299 call boot monitor routines pass an unknown value in @code{$gp}.
22300 (In such situations, the boot monitor itself is usually compiled
22301 with @option{-G0}.)
22303 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
22304 @option{-mno-extern-sdata}.
22306 @item -membedded-data
22307 @itemx -mno-embedded-data
22308 @opindex membedded-data
22309 @opindex mno-embedded-data
22310 Allocate variables to the read-only data section first if possible, then
22311 next in the small data section if possible, otherwise in data. This gives
22312 slightly slower code than the default, but reduces the amount of RAM required
22313 when executing, and thus may be preferred for some embedded systems.
22315 @item -muninit-const-in-rodata
22316 @itemx -mno-uninit-const-in-rodata
22317 @opindex muninit-const-in-rodata
22318 @opindex mno-uninit-const-in-rodata
22319 Put uninitialized @code{const} variables in the read-only data section.
22320 This option is only meaningful in conjunction with @option{-membedded-data}.
22322 @item -mcode-readable=@var{setting}
22323 @opindex mcode-readable
22324 Specify whether GCC may generate code that reads from executable sections.
22325 There are three possible settings:
22328 @item -mcode-readable=yes
22329 Instructions may freely access executable sections. This is the
22332 @item -mcode-readable=pcrel
22333 MIPS16 PC-relative load instructions can access executable sections,
22334 but other instructions must not do so. This option is useful on 4KSc
22335 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
22336 It is also useful on processors that can be configured to have a dual
22337 instruction/data SRAM interface and that, like the M4K, automatically
22338 redirect PC-relative loads to the instruction RAM.
22340 @item -mcode-readable=no
22341 Instructions must not access executable sections. This option can be
22342 useful on targets that are configured to have a dual instruction/data
22343 SRAM interface but that (unlike the M4K) do not automatically redirect
22344 PC-relative loads to the instruction RAM.
22347 @item -msplit-addresses
22348 @itemx -mno-split-addresses
22349 @opindex msplit-addresses
22350 @opindex mno-split-addresses
22351 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
22352 relocation operators. This option has been superseded by
22353 @option{-mexplicit-relocs} but is retained for backwards compatibility.
22355 @item -mexplicit-relocs
22356 @itemx -mno-explicit-relocs
22357 @opindex mexplicit-relocs
22358 @opindex mno-explicit-relocs
22359 Use (do not use) assembler relocation operators when dealing with symbolic
22360 addresses. The alternative, selected by @option{-mno-explicit-relocs},
22361 is to use assembler macros instead.
22363 @option{-mexplicit-relocs} is the default if GCC was configured
22364 to use an assembler that supports relocation operators.
22366 @item -mcheck-zero-division
22367 @itemx -mno-check-zero-division
22368 @opindex mcheck-zero-division
22369 @opindex mno-check-zero-division
22370 Trap (do not trap) on integer division by zero.
22372 The default is @option{-mcheck-zero-division}.
22374 @item -mdivide-traps
22375 @itemx -mdivide-breaks
22376 @opindex mdivide-traps
22377 @opindex mdivide-breaks
22378 MIPS systems check for division by zero by generating either a
22379 conditional trap or a break instruction. Using traps results in
22380 smaller code, but is only supported on MIPS II and later. Also, some
22381 versions of the Linux kernel have a bug that prevents trap from
22382 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
22383 allow conditional traps on architectures that support them and
22384 @option{-mdivide-breaks} to force the use of breaks.
22386 The default is usually @option{-mdivide-traps}, but this can be
22387 overridden at configure time using @option{--with-divide=breaks}.
22388 Divide-by-zero checks can be completely disabled using
22389 @option{-mno-check-zero-division}.
22391 @item -mload-store-pairs
22392 @itemx -mno-load-store-pairs
22393 @opindex mload-store-pairs
22394 @opindex mno-load-store-pairs
22395 Enable (disable) an optimization that pairs consecutive load or store
22396 instructions to enable load/store bonding. This option is enabled by
22397 default but only takes effect when the selected architecture is known
22398 to support bonding.
22403 @opindex mno-memcpy
22404 Force (do not force) the use of @code{memcpy} for non-trivial block
22405 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
22406 most constant-sized copies.
22409 @itemx -mno-long-calls
22410 @opindex mlong-calls
22411 @opindex mno-long-calls
22412 Disable (do not disable) use of the @code{jal} instruction. Calling
22413 functions using @code{jal} is more efficient but requires the caller
22414 and callee to be in the same 256 megabyte segment.
22416 This option has no effect on abicalls code. The default is
22417 @option{-mno-long-calls}.
22423 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
22424 instructions, as provided by the R4650 ISA@.
22430 Enable (disable) use of the @code{madd} and @code{msub} integer
22431 instructions. The default is @option{-mimadd} on architectures
22432 that support @code{madd} and @code{msub} except for the 74k
22433 architecture where it was found to generate slower code.
22436 @itemx -mno-fused-madd
22437 @opindex mfused-madd
22438 @opindex mno-fused-madd
22439 Enable (disable) use of the floating-point multiply-accumulate
22440 instructions, when they are available. The default is
22441 @option{-mfused-madd}.
22443 On the R8000 CPU when multiply-accumulate instructions are used,
22444 the intermediate product is calculated to infinite precision
22445 and is not subject to the FCSR Flush to Zero bit. This may be
22446 undesirable in some circumstances. On other processors the result
22447 is numerically identical to the equivalent computation using
22448 separate multiply, add, subtract and negate instructions.
22452 Tell the MIPS assembler to not run its preprocessor over user
22453 assembler files (with a @samp{.s} suffix) when assembling them.
22456 @itemx -mno-fix-24k
22458 @opindex mno-fix-24k
22459 Work around the 24K E48 (lost data on stores during refill) errata.
22460 The workarounds are implemented by the assembler rather than by GCC@.
22463 @itemx -mno-fix-r4000
22464 @opindex mfix-r4000
22465 @opindex mno-fix-r4000
22466 Work around certain R4000 CPU errata:
22469 A double-word or a variable shift may give an incorrect result if executed
22470 immediately after starting an integer division.
22472 A double-word or a variable shift may give an incorrect result if executed
22473 while an integer multiplication is in progress.
22475 An integer division may give an incorrect result if started in a delay slot
22476 of a taken branch or a jump.
22480 @itemx -mno-fix-r4400
22481 @opindex mfix-r4400
22482 @opindex mno-fix-r4400
22483 Work around certain R4400 CPU errata:
22486 A double-word or a variable shift may give an incorrect result if executed
22487 immediately after starting an integer division.
22491 @itemx -mno-fix-r10000
22492 @opindex mfix-r10000
22493 @opindex mno-fix-r10000
22494 Work around certain R10000 errata:
22497 @code{ll}/@code{sc} sequences may not behave atomically on revisions
22498 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
22501 This option can only be used if the target architecture supports
22502 branch-likely instructions. @option{-mfix-r10000} is the default when
22503 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
22507 @itemx -mno-fix-r5900
22508 @opindex mfix-r5900
22509 Do not attempt to schedule the preceding instruction into the delay slot
22510 of a branch instruction placed at the end of a short loop of six
22511 instructions or fewer and always schedule a @code{nop} instruction there
22512 instead. The short loop bug under certain conditions causes loops to
22513 execute only once or twice, due to a hardware bug in the R5900 chip. The
22514 workaround is implemented by the assembler rather than by GCC@.
22517 @itemx -mno-fix-rm7000
22518 @opindex mfix-rm7000
22519 Work around the RM7000 @code{dmult}/@code{dmultu} errata. The
22520 workarounds are implemented by the assembler rather than by GCC@.
22523 @itemx -mno-fix-vr4120
22524 @opindex mfix-vr4120
22525 Work around certain VR4120 errata:
22528 @code{dmultu} does not always produce the correct result.
22530 @code{div} and @code{ddiv} do not always produce the correct result if one
22531 of the operands is negative.
22533 The workarounds for the division errata rely on special functions in
22534 @file{libgcc.a}. At present, these functions are only provided by
22535 the @code{mips64vr*-elf} configurations.
22537 Other VR4120 errata require a NOP to be inserted between certain pairs of
22538 instructions. These errata are handled by the assembler, not by GCC itself.
22541 @opindex mfix-vr4130
22542 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
22543 workarounds are implemented by the assembler rather than by GCC,
22544 although GCC avoids using @code{mflo} and @code{mfhi} if the
22545 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
22546 instructions are available instead.
22549 @itemx -mno-fix-sb1
22551 Work around certain SB-1 CPU core errata.
22552 (This flag currently works around the SB-1 revision 2
22553 ``F1'' and ``F2'' floating-point errata.)
22555 @item -mr10k-cache-barrier=@var{setting}
22556 @opindex mr10k-cache-barrier
22557 Specify whether GCC should insert cache barriers to avoid the
22558 side effects of speculation on R10K processors.
22560 In common with many processors, the R10K tries to predict the outcome
22561 of a conditional branch and speculatively executes instructions from
22562 the ``taken'' branch. It later aborts these instructions if the
22563 predicted outcome is wrong. However, on the R10K, even aborted
22564 instructions can have side effects.
22566 This problem only affects kernel stores and, depending on the system,
22567 kernel loads. As an example, a speculatively-executed store may load
22568 the target memory into cache and mark the cache line as dirty, even if
22569 the store itself is later aborted. If a DMA operation writes to the
22570 same area of memory before the ``dirty'' line is flushed, the cached
22571 data overwrites the DMA-ed data. See the R10K processor manual
22572 for a full description, including other potential problems.
22574 One workaround is to insert cache barrier instructions before every memory
22575 access that might be speculatively executed and that might have side
22576 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
22577 controls GCC's implementation of this workaround. It assumes that
22578 aborted accesses to any byte in the following regions does not have
22583 the memory occupied by the current function's stack frame;
22586 the memory occupied by an incoming stack argument;
22589 the memory occupied by an object with a link-time-constant address.
22592 It is the kernel's responsibility to ensure that speculative
22593 accesses to these regions are indeed safe.
22595 If the input program contains a function declaration such as:
22601 then the implementation of @code{foo} must allow @code{j foo} and
22602 @code{jal foo} to be executed speculatively. GCC honors this
22603 restriction for functions it compiles itself. It expects non-GCC
22604 functions (such as hand-written assembly code) to do the same.
22606 The option has three forms:
22609 @item -mr10k-cache-barrier=load-store
22610 Insert a cache barrier before a load or store that might be
22611 speculatively executed and that might have side effects even
22614 @item -mr10k-cache-barrier=store
22615 Insert a cache barrier before a store that might be speculatively
22616 executed and that might have side effects even if aborted.
22618 @item -mr10k-cache-barrier=none
22619 Disable the insertion of cache barriers. This is the default setting.
22622 @item -mflush-func=@var{func}
22623 @itemx -mno-flush-func
22624 @opindex mflush-func
22625 Specifies the function to call to flush the I and D caches, or to not
22626 call any such function. If called, the function must take the same
22627 arguments as the common @code{_flush_func}, that is, the address of the
22628 memory range for which the cache is being flushed, the size of the
22629 memory range, and the number 3 (to flush both caches). The default
22630 depends on the target GCC was configured for, but commonly is either
22631 @code{_flush_func} or @code{__cpu_flush}.
22633 @item mbranch-cost=@var{num}
22634 @opindex mbranch-cost
22635 Set the cost of branches to roughly @var{num} ``simple'' instructions.
22636 This cost is only a heuristic and is not guaranteed to produce
22637 consistent results across releases. A zero cost redundantly selects
22638 the default, which is based on the @option{-mtune} setting.
22640 @item -mbranch-likely
22641 @itemx -mno-branch-likely
22642 @opindex mbranch-likely
22643 @opindex mno-branch-likely
22644 Enable or disable use of Branch Likely instructions, regardless of the
22645 default for the selected architecture. By default, Branch Likely
22646 instructions may be generated if they are supported by the selected
22647 architecture. An exception is for the MIPS32 and MIPS64 architectures
22648 and processors that implement those architectures; for those, Branch
22649 Likely instructions are not be generated by default because the MIPS32
22650 and MIPS64 architectures specifically deprecate their use.
22652 @item -mcompact-branches=never
22653 @itemx -mcompact-branches=optimal
22654 @itemx -mcompact-branches=always
22655 @opindex mcompact-branches=never
22656 @opindex mcompact-branches=optimal
22657 @opindex mcompact-branches=always
22658 These options control which form of branches will be generated. The
22659 default is @option{-mcompact-branches=optimal}.
22661 The @option{-mcompact-branches=never} option ensures that compact branch
22662 instructions will never be generated.
22664 The @option{-mcompact-branches=always} option ensures that a compact
22665 branch instruction will be generated if available. If a compact branch
22666 instruction is not available, a delay slot form of the branch will be
22669 This option is supported from MIPS Release 6 onwards.
22671 The @option{-mcompact-branches=optimal} option will cause a delay slot
22672 branch to be used if one is available in the current ISA and the delay
22673 slot is successfully filled. If the delay slot is not filled, a compact
22674 branch will be chosen if one is available.
22676 @item -mfp-exceptions
22677 @itemx -mno-fp-exceptions
22678 @opindex mfp-exceptions
22679 Specifies whether FP exceptions are enabled. This affects how
22680 FP instructions are scheduled for some processors.
22681 The default is that FP exceptions are
22684 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
22685 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
22688 @item -mvr4130-align
22689 @itemx -mno-vr4130-align
22690 @opindex mvr4130-align
22691 The VR4130 pipeline is two-way superscalar, but can only issue two
22692 instructions together if the first one is 8-byte aligned. When this
22693 option is enabled, GCC aligns pairs of instructions that it
22694 thinks should execute in parallel.
22696 This option only has an effect when optimizing for the VR4130.
22697 It normally makes code faster, but at the expense of making it bigger.
22698 It is enabled by default at optimization level @option{-O3}.
22703 Enable (disable) generation of @code{synci} instructions on
22704 architectures that support it. The @code{synci} instructions (if
22705 enabled) are generated when @code{__builtin___clear_cache} is
22708 This option defaults to @option{-mno-synci}, but the default can be
22709 overridden by configuring GCC with @option{--with-synci}.
22711 When compiling code for single processor systems, it is generally safe
22712 to use @code{synci}. However, on many multi-core (SMP) systems, it
22713 does not invalidate the instruction caches on all cores and may lead
22714 to undefined behavior.
22716 @item -mrelax-pic-calls
22717 @itemx -mno-relax-pic-calls
22718 @opindex mrelax-pic-calls
22719 Try to turn PIC calls that are normally dispatched via register
22720 @code{$25} into direct calls. This is only possible if the linker can
22721 resolve the destination at link time and if the destination is within
22722 range for a direct call.
22724 @option{-mrelax-pic-calls} is the default if GCC was configured to use
22725 an assembler and a linker that support the @code{.reloc} assembly
22726 directive and @option{-mexplicit-relocs} is in effect. With
22727 @option{-mno-explicit-relocs}, this optimization can be performed by the
22728 assembler and the linker alone without help from the compiler.
22730 @item -mmcount-ra-address
22731 @itemx -mno-mcount-ra-address
22732 @opindex mmcount-ra-address
22733 @opindex mno-mcount-ra-address
22734 Emit (do not emit) code that allows @code{_mcount} to modify the
22735 calling function's return address. When enabled, this option extends
22736 the usual @code{_mcount} interface with a new @var{ra-address}
22737 parameter, which has type @code{intptr_t *} and is passed in register
22738 @code{$12}. @code{_mcount} can then modify the return address by
22739 doing both of the following:
22742 Returning the new address in register @code{$31}.
22744 Storing the new address in @code{*@var{ra-address}},
22745 if @var{ra-address} is nonnull.
22748 The default is @option{-mno-mcount-ra-address}.
22750 @item -mframe-header-opt
22751 @itemx -mno-frame-header-opt
22752 @opindex mframe-header-opt
22753 Enable (disable) frame header optimization in the o32 ABI. When using the
22754 o32 ABI, calling functions will allocate 16 bytes on the stack for the called
22755 function to write out register arguments. When enabled, this optimization
22756 will suppress the allocation of the frame header if it can be determined that
22759 This optimization is off by default at all optimization levels.
22762 @itemx -mno-lxc1-sxc1
22763 @opindex mlxc1-sxc1
22764 When applicable, enable (disable) the generation of @code{lwxc1},
22765 @code{swxc1}, @code{ldxc1}, @code{sdxc1} instructions. Enabled by default.
22770 When applicable, enable (disable) the generation of 4-operand @code{madd.s},
22771 @code{madd.d} and related instructions. Enabled by default.
22776 @subsection MMIX Options
22777 @cindex MMIX Options
22779 These options are defined for the MMIX:
22783 @itemx -mno-libfuncs
22785 @opindex mno-libfuncs
22786 Specify that intrinsic library functions are being compiled, passing all
22787 values in registers, no matter the size.
22790 @itemx -mno-epsilon
22792 @opindex mno-epsilon
22793 Generate floating-point comparison instructions that compare with respect
22794 to the @code{rE} epsilon register.
22796 @item -mabi=mmixware
22798 @opindex mabi=mmixware
22800 Generate code that passes function parameters and return values that (in
22801 the called function) are seen as registers @code{$0} and up, as opposed to
22802 the GNU ABI which uses global registers @code{$231} and up.
22804 @item -mzero-extend
22805 @itemx -mno-zero-extend
22806 @opindex mzero-extend
22807 @opindex mno-zero-extend
22808 When reading data from memory in sizes shorter than 64 bits, use (do not
22809 use) zero-extending load instructions by default, rather than
22810 sign-extending ones.
22813 @itemx -mno-knuthdiv
22815 @opindex mno-knuthdiv
22816 Make the result of a division yielding a remainder have the same sign as
22817 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
22818 remainder follows the sign of the dividend. Both methods are
22819 arithmetically valid, the latter being almost exclusively used.
22821 @item -mtoplevel-symbols
22822 @itemx -mno-toplevel-symbols
22823 @opindex mtoplevel-symbols
22824 @opindex mno-toplevel-symbols
22825 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
22826 code can be used with the @code{PREFIX} assembly directive.
22830 Generate an executable in the ELF format, rather than the default
22831 @samp{mmo} format used by the @command{mmix} simulator.
22833 @item -mbranch-predict
22834 @itemx -mno-branch-predict
22835 @opindex mbranch-predict
22836 @opindex mno-branch-predict
22837 Use (do not use) the probable-branch instructions, when static branch
22838 prediction indicates a probable branch.
22840 @item -mbase-addresses
22841 @itemx -mno-base-addresses
22842 @opindex mbase-addresses
22843 @opindex mno-base-addresses
22844 Generate (do not generate) code that uses @emph{base addresses}. Using a
22845 base address automatically generates a request (handled by the assembler
22846 and the linker) for a constant to be set up in a global register. The
22847 register is used for one or more base address requests within the range 0
22848 to 255 from the value held in the register. The generally leads to short
22849 and fast code, but the number of different data items that can be
22850 addressed is limited. This means that a program that uses lots of static
22851 data may require @option{-mno-base-addresses}.
22853 @item -msingle-exit
22854 @itemx -mno-single-exit
22855 @opindex msingle-exit
22856 @opindex mno-single-exit
22857 Force (do not force) generated code to have a single exit point in each
22861 @node MN10300 Options
22862 @subsection MN10300 Options
22863 @cindex MN10300 options
22865 These @option{-m} options are defined for Matsushita MN10300 architectures:
22870 Generate code to avoid bugs in the multiply instructions for the MN10300
22871 processors. This is the default.
22873 @item -mno-mult-bug
22874 @opindex mno-mult-bug
22875 Do not generate code to avoid bugs in the multiply instructions for the
22876 MN10300 processors.
22880 Generate code using features specific to the AM33 processor.
22884 Do not generate code using features specific to the AM33 processor. This
22889 Generate code using features specific to the AM33/2.0 processor.
22893 Generate code using features specific to the AM34 processor.
22895 @item -mtune=@var{cpu-type}
22897 Use the timing characteristics of the indicated CPU type when
22898 scheduling instructions. This does not change the targeted processor
22899 type. The CPU type must be one of @samp{mn10300}, @samp{am33},
22900 @samp{am33-2} or @samp{am34}.
22902 @item -mreturn-pointer-on-d0
22903 @opindex mreturn-pointer-on-d0
22904 When generating a function that returns a pointer, return the pointer
22905 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
22906 only in @code{a0}, and attempts to call such functions without a prototype
22907 result in errors. Note that this option is on by default; use
22908 @option{-mno-return-pointer-on-d0} to disable it.
22912 Do not link in the C run-time initialization object file.
22916 Indicate to the linker that it should perform a relaxation optimization pass
22917 to shorten branches, calls and absolute memory addresses. This option only
22918 has an effect when used on the command line for the final link step.
22920 This option makes symbolic debugging impossible.
22924 Allow the compiler to generate @emph{Long Instruction Word}
22925 instructions if the target is the @samp{AM33} or later. This is the
22926 default. This option defines the preprocessor macro @code{__LIW__}.
22930 Do not allow the compiler to generate @emph{Long Instruction Word}
22931 instructions. This option defines the preprocessor macro
22936 Allow the compiler to generate the @emph{SETLB} and @emph{Lcc}
22937 instructions if the target is the @samp{AM33} or later. This is the
22938 default. This option defines the preprocessor macro @code{__SETLB__}.
22942 Do not allow the compiler to generate @emph{SETLB} or @emph{Lcc}
22943 instructions. This option defines the preprocessor macro
22944 @code{__NO_SETLB__}.
22948 @node Moxie Options
22949 @subsection Moxie Options
22950 @cindex Moxie Options
22956 Generate big-endian code. This is the default for @samp{moxie-*-*}
22961 Generate little-endian code.
22965 Generate mul.x and umul.x instructions. This is the default for
22966 @samp{moxiebox-*-*} configurations.
22970 Do not link in the C run-time initialization object file.
22974 @node MSP430 Options
22975 @subsection MSP430 Options
22976 @cindex MSP430 Options
22978 These options are defined for the MSP430:
22984 Force assembly output to always use hex constants. Normally such
22985 constants are signed decimals, but this option is available for
22986 testsuite and/or aesthetic purposes.
22990 Select the MCU to target. This is used to create a C preprocessor
22991 symbol based upon the MCU name, converted to upper case and pre- and
22992 post-fixed with @samp{__}. This in turn is used by the
22993 @file{msp430.h} header file to select an MCU-specific supplementary
22996 The option also sets the ISA to use. If the MCU name is one that is
22997 known to only support the 430 ISA then that is selected, otherwise the
22998 430X ISA is selected. A generic MCU name of @samp{msp430} can also be
22999 used to select the 430 ISA. Similarly the generic @samp{msp430x} MCU
23000 name selects the 430X ISA.
23002 In addition an MCU-specific linker script is added to the linker
23003 command line. The script's name is the name of the MCU with
23004 @file{.ld} appended. Thus specifying @option{-mmcu=xxx} on the @command{gcc}
23005 command line defines the C preprocessor symbol @code{__XXX__} and
23006 cause the linker to search for a script called @file{xxx.ld}.
23008 This option is also passed on to the assembler.
23011 @itemx -mno-warn-mcu
23013 @opindex mno-warn-mcu
23014 This option enables or disables warnings about conflicts between the
23015 MCU name specified by the @option{-mmcu} option and the ISA set by the
23016 @option{-mcpu} option and/or the hardware multiply support set by the
23017 @option{-mhwmult} option. It also toggles warnings about unrecognized
23018 MCU names. This option is on by default.
23022 Specifies the ISA to use. Accepted values are @samp{msp430},
23023 @samp{msp430x} and @samp{msp430xv2}. This option is deprecated. The
23024 @option{-mmcu=} option should be used to select the ISA.
23028 Link to the simulator runtime libraries and linker script. Overrides
23029 any scripts that would be selected by the @option{-mmcu=} option.
23033 Use large-model addressing (20-bit pointers, 32-bit @code{size_t}).
23037 Use small-model addressing (16-bit pointers, 16-bit @code{size_t}).
23041 This option is passed to the assembler and linker, and allows the
23042 linker to perform certain optimizations that cannot be done until
23047 Describes the type of hardware multiply supported by the target.
23048 Accepted values are @samp{none} for no hardware multiply, @samp{16bit}
23049 for the original 16-bit-only multiply supported by early MCUs.
23050 @samp{32bit} for the 16/32-bit multiply supported by later MCUs and
23051 @samp{f5series} for the 16/32-bit multiply supported by F5-series MCUs.
23052 A value of @samp{auto} can also be given. This tells GCC to deduce
23053 the hardware multiply support based upon the MCU name provided by the
23054 @option{-mmcu} option. If no @option{-mmcu} option is specified or if
23055 the MCU name is not recognized then no hardware multiply support is
23056 assumed. @code{auto} is the default setting.
23058 Hardware multiplies are normally performed by calling a library
23059 routine. This saves space in the generated code. When compiling at
23060 @option{-O3} or higher however the hardware multiplier is invoked
23061 inline. This makes for bigger, but faster code.
23063 The hardware multiply routines disable interrupts whilst running and
23064 restore the previous interrupt state when they finish. This makes
23065 them safe to use inside interrupt handlers as well as in normal code.
23069 Enable the use of a minimum runtime environment - no static
23070 initializers or constructors. This is intended for memory-constrained
23071 devices. The compiler includes special symbols in some objects
23072 that tell the linker and runtime which code fragments are required.
23074 @item -mcode-region=
23075 @itemx -mdata-region=
23076 @opindex mcode-region
23077 @opindex mdata-region
23078 These options tell the compiler where to place functions and data that
23079 do not have one of the @code{lower}, @code{upper}, @code{either} or
23080 @code{section} attributes. Possible values are @code{lower},
23081 @code{upper}, @code{either} or @code{any}. The first three behave
23082 like the corresponding attribute. The fourth possible value -
23083 @code{any} - is the default. It leaves placement entirely up to the
23084 linker script and how it assigns the standard sections
23085 (@code{.text}, @code{.data}, etc) to the memory regions.
23087 @item -msilicon-errata=
23088 @opindex msilicon-errata
23089 This option passes on a request to assembler to enable the fixes for
23090 the named silicon errata.
23092 @item -msilicon-errata-warn=
23093 @opindex msilicon-errata-warn
23094 This option passes on a request to the assembler to enable warning
23095 messages when a silicon errata might need to be applied.
23099 @node NDS32 Options
23100 @subsection NDS32 Options
23101 @cindex NDS32 Options
23103 These options are defined for NDS32 implementations:
23108 @opindex mbig-endian
23109 Generate code in big-endian mode.
23111 @item -mlittle-endian
23112 @opindex mlittle-endian
23113 Generate code in little-endian mode.
23115 @item -mreduced-regs
23116 @opindex mreduced-regs
23117 Use reduced-set registers for register allocation.
23120 @opindex mfull-regs
23121 Use full-set registers for register allocation.
23125 Generate conditional move instructions.
23129 Do not generate conditional move instructions.
23133 Generate performance extension instructions.
23135 @item -mno-ext-perf
23136 @opindex mno-ext-perf
23137 Do not generate performance extension instructions.
23140 @opindex mext-perf2
23141 Generate performance extension 2 instructions.
23143 @item -mno-ext-perf2
23144 @opindex mno-ext-perf2
23145 Do not generate performance extension 2 instructions.
23148 @opindex mext-string
23149 Generate string extension instructions.
23151 @item -mno-ext-string
23152 @opindex mno-ext-string
23153 Do not generate string extension instructions.
23157 Generate v3 push25/pop25 instructions.
23160 @opindex mno-v3push
23161 Do not generate v3 push25/pop25 instructions.
23165 Generate 16-bit instructions.
23168 @opindex mno-16-bit
23169 Do not generate 16-bit instructions.
23171 @item -misr-vector-size=@var{num}
23172 @opindex misr-vector-size
23173 Specify the size of each interrupt vector, which must be 4 or 16.
23175 @item -mcache-block-size=@var{num}
23176 @opindex mcache-block-size
23177 Specify the size of each cache block,
23178 which must be a power of 2 between 4 and 512.
23180 @item -march=@var{arch}
23182 Specify the name of the target architecture.
23184 @item -mcmodel=@var{code-model}
23186 Set the code model to one of
23189 All the data and read-only data segments must be within 512KB addressing space.
23190 The text segment must be within 16MB addressing space.
23191 @item @samp{medium}
23192 The data segment must be within 512KB while the read-only data segment can be
23193 within 4GB addressing space. The text segment should be still within 16MB
23196 All the text and data segments can be within 4GB addressing space.
23200 @opindex mctor-dtor
23201 Enable constructor/destructor feature.
23205 Guide linker to relax instructions.
23209 @node Nios II Options
23210 @subsection Nios II Options
23211 @cindex Nios II options
23212 @cindex Altera Nios II options
23214 These are the options defined for the Altera Nios II processor.
23220 @cindex smaller data references
23221 Put global and static objects less than or equal to @var{num} bytes
23222 into the small data or BSS sections instead of the normal data or BSS
23223 sections. The default value of @var{num} is 8.
23225 @item -mgpopt=@var{option}
23230 Generate (do not generate) GP-relative accesses. The following
23231 @var{option} names are recognized:
23236 Do not generate GP-relative accesses.
23239 Generate GP-relative accesses for small data objects that are not
23240 external, weak, or uninitialized common symbols.
23241 Also use GP-relative addressing for objects that
23242 have been explicitly placed in a small data section via a @code{section}
23246 As for @samp{local}, but also generate GP-relative accesses for
23247 small data objects that are external, weak, or common. If you use this option,
23248 you must ensure that all parts of your program (including libraries) are
23249 compiled with the same @option{-G} setting.
23252 Generate GP-relative accesses for all data objects in the program. If you
23253 use this option, the entire data and BSS segments
23254 of your program must fit in 64K of memory and you must use an appropriate
23255 linker script to allocate them within the addressable range of the
23259 Generate GP-relative addresses for function pointers as well as data
23260 pointers. If you use this option, the entire text, data, and BSS segments
23261 of your program must fit in 64K of memory and you must use an appropriate
23262 linker script to allocate them within the addressable range of the
23267 @option{-mgpopt} is equivalent to @option{-mgpopt=local}, and
23268 @option{-mno-gpopt} is equivalent to @option{-mgpopt=none}.
23270 The default is @option{-mgpopt} except when @option{-fpic} or
23271 @option{-fPIC} is specified to generate position-independent code.
23272 Note that the Nios II ABI does not permit GP-relative accesses from
23275 You may need to specify @option{-mno-gpopt} explicitly when building
23276 programs that include large amounts of small data, including large
23277 GOT data sections. In this case, the 16-bit offset for GP-relative
23278 addressing may not be large enough to allow access to the entire
23279 small data section.
23281 @item -mgprel-sec=@var{regexp}
23282 @opindex mgprel-sec
23283 This option specifies additional section names that can be accessed via
23284 GP-relative addressing. It is most useful in conjunction with
23285 @code{section} attributes on variable declarations
23286 (@pxref{Common Variable Attributes}) and a custom linker script.
23287 The @var{regexp} is a POSIX Extended Regular Expression.
23289 This option does not affect the behavior of the @option{-G} option, and
23290 the specified sections are in addition to the standard @code{.sdata}
23291 and @code{.sbss} small-data sections that are recognized by @option{-mgpopt}.
23293 @item -mr0rel-sec=@var{regexp}
23294 @opindex mr0rel-sec
23295 This option specifies names of sections that can be accessed via a
23296 16-bit offset from @code{r0}; that is, in the low 32K or high 32K
23297 of the 32-bit address space. It is most useful in conjunction with
23298 @code{section} attributes on variable declarations
23299 (@pxref{Common Variable Attributes}) and a custom linker script.
23300 The @var{regexp} is a POSIX Extended Regular Expression.
23302 In contrast to the use of GP-relative addressing for small data,
23303 zero-based addressing is never generated by default and there are no
23304 conventional section names used in standard linker scripts for sections
23305 in the low or high areas of memory.
23311 Generate little-endian (default) or big-endian (experimental) code,
23314 @item -march=@var{arch}
23316 This specifies the name of the target Nios II architecture. GCC uses this
23317 name to determine what kind of instructions it can emit when generating
23318 assembly code. Permissible names are: @samp{r1}, @samp{r2}.
23320 The preprocessor macro @code{__nios2_arch__} is available to programs,
23321 with value 1 or 2, indicating the targeted ISA level.
23323 @item -mbypass-cache
23324 @itemx -mno-bypass-cache
23325 @opindex mno-bypass-cache
23326 @opindex mbypass-cache
23327 Force all load and store instructions to always bypass cache by
23328 using I/O variants of the instructions. The default is not to
23331 @item -mno-cache-volatile
23332 @itemx -mcache-volatile
23333 @opindex mcache-volatile
23334 @opindex mno-cache-volatile
23335 Volatile memory access bypass the cache using the I/O variants of
23336 the load and store instructions. The default is not to bypass the cache.
23338 @item -mno-fast-sw-div
23339 @itemx -mfast-sw-div
23340 @opindex mno-fast-sw-div
23341 @opindex mfast-sw-div
23342 Do not use table-based fast divide for small numbers. The default
23343 is to use the fast divide at @option{-O3} and above.
23347 @itemx -mno-hw-mulx
23351 @opindex mno-hw-mul
23353 @opindex mno-hw-mulx
23355 @opindex mno-hw-div
23357 Enable or disable emitting @code{mul}, @code{mulx} and @code{div} family of
23358 instructions by the compiler. The default is to emit @code{mul}
23359 and not emit @code{div} and @code{mulx}.
23365 Enable or disable generation of Nios II R2 BMX (bit manipulation) and
23366 CDX (code density) instructions. Enabling these instructions also
23367 requires @option{-march=r2}. Since these instructions are optional
23368 extensions to the R2 architecture, the default is not to emit them.
23370 @item -mcustom-@var{insn}=@var{N}
23371 @itemx -mno-custom-@var{insn}
23372 @opindex mcustom-@var{insn}
23373 @opindex mno-custom-@var{insn}
23374 Each @option{-mcustom-@var{insn}=@var{N}} option enables use of a
23375 custom instruction with encoding @var{N} when generating code that uses
23376 @var{insn}. For example, @option{-mcustom-fadds=253} generates custom
23377 instruction 253 for single-precision floating-point add operations instead
23378 of the default behavior of using a library call.
23380 The following values of @var{insn} are supported. Except as otherwise
23381 noted, floating-point operations are expected to be implemented with
23382 normal IEEE 754 semantics and correspond directly to the C operators or the
23383 equivalent GCC built-in functions (@pxref{Other Builtins}).
23385 Single-precision floating point:
23388 @item @samp{fadds}, @samp{fsubs}, @samp{fdivs}, @samp{fmuls}
23389 Binary arithmetic operations.
23395 Unary absolute value.
23397 @item @samp{fcmpeqs}, @samp{fcmpges}, @samp{fcmpgts}, @samp{fcmples}, @samp{fcmplts}, @samp{fcmpnes}
23398 Comparison operations.
23400 @item @samp{fmins}, @samp{fmaxs}
23401 Floating-point minimum and maximum. These instructions are only
23402 generated if @option{-ffinite-math-only} is specified.
23404 @item @samp{fsqrts}
23405 Unary square root operation.
23407 @item @samp{fcoss}, @samp{fsins}, @samp{ftans}, @samp{fatans}, @samp{fexps}, @samp{flogs}
23408 Floating-point trigonometric and exponential functions. These instructions
23409 are only generated if @option{-funsafe-math-optimizations} is also specified.
23413 Double-precision floating point:
23416 @item @samp{faddd}, @samp{fsubd}, @samp{fdivd}, @samp{fmuld}
23417 Binary arithmetic operations.
23423 Unary absolute value.
23425 @item @samp{fcmpeqd}, @samp{fcmpged}, @samp{fcmpgtd}, @samp{fcmpled}, @samp{fcmpltd}, @samp{fcmpned}
23426 Comparison operations.
23428 @item @samp{fmind}, @samp{fmaxd}
23429 Double-precision minimum and maximum. These instructions are only
23430 generated if @option{-ffinite-math-only} is specified.
23432 @item @samp{fsqrtd}
23433 Unary square root operation.
23435 @item @samp{fcosd}, @samp{fsind}, @samp{ftand}, @samp{fatand}, @samp{fexpd}, @samp{flogd}
23436 Double-precision trigonometric and exponential functions. These instructions
23437 are only generated if @option{-funsafe-math-optimizations} is also specified.
23443 @item @samp{fextsd}
23444 Conversion from single precision to double precision.
23446 @item @samp{ftruncds}
23447 Conversion from double precision to single precision.
23449 @item @samp{fixsi}, @samp{fixsu}, @samp{fixdi}, @samp{fixdu}
23450 Conversion from floating point to signed or unsigned integer types, with
23451 truncation towards zero.
23454 Conversion from single-precision floating point to signed integer,
23455 rounding to the nearest integer and ties away from zero.
23456 This corresponds to the @code{__builtin_lroundf} function when
23457 @option{-fno-math-errno} is used.
23459 @item @samp{floatis}, @samp{floatus}, @samp{floatid}, @samp{floatud}
23460 Conversion from signed or unsigned integer types to floating-point types.
23464 In addition, all of the following transfer instructions for internal
23465 registers X and Y must be provided to use any of the double-precision
23466 floating-point instructions. Custom instructions taking two
23467 double-precision source operands expect the first operand in the
23468 64-bit register X. The other operand (or only operand of a unary
23469 operation) is given to the custom arithmetic instruction with the
23470 least significant half in source register @var{src1} and the most
23471 significant half in @var{src2}. A custom instruction that returns a
23472 double-precision result returns the most significant 32 bits in the
23473 destination register and the other half in 32-bit register Y.
23474 GCC automatically generates the necessary code sequences to write
23475 register X and/or read register Y when double-precision floating-point
23476 instructions are used.
23481 Write @var{src1} into the least significant half of X and @var{src2} into
23482 the most significant half of X.
23485 Write @var{src1} into Y.
23487 @item @samp{frdxhi}, @samp{frdxlo}
23488 Read the most or least (respectively) significant half of X and store it in
23492 Read the value of Y and store it into @var{dest}.
23495 Note that you can gain more local control over generation of Nios II custom
23496 instructions by using the @code{target("custom-@var{insn}=@var{N}")}
23497 and @code{target("no-custom-@var{insn}")} function attributes
23498 (@pxref{Function Attributes})
23499 or pragmas (@pxref{Function Specific Option Pragmas}).
23501 @item -mcustom-fpu-cfg=@var{name}
23502 @opindex mcustom-fpu-cfg
23504 This option enables a predefined, named set of custom instruction encodings
23505 (see @option{-mcustom-@var{insn}} above).
23506 Currently, the following sets are defined:
23508 @option{-mcustom-fpu-cfg=60-1} is equivalent to:
23509 @gccoptlist{-mcustom-fmuls=252 @gol
23510 -mcustom-fadds=253 @gol
23511 -mcustom-fsubs=254 @gol
23512 -fsingle-precision-constant}
23514 @option{-mcustom-fpu-cfg=60-2} is equivalent to:
23515 @gccoptlist{-mcustom-fmuls=252 @gol
23516 -mcustom-fadds=253 @gol
23517 -mcustom-fsubs=254 @gol
23518 -mcustom-fdivs=255 @gol
23519 -fsingle-precision-constant}
23521 @option{-mcustom-fpu-cfg=72-3} is equivalent to:
23522 @gccoptlist{-mcustom-floatus=243 @gol
23523 -mcustom-fixsi=244 @gol
23524 -mcustom-floatis=245 @gol
23525 -mcustom-fcmpgts=246 @gol
23526 -mcustom-fcmples=249 @gol
23527 -mcustom-fcmpeqs=250 @gol
23528 -mcustom-fcmpnes=251 @gol
23529 -mcustom-fmuls=252 @gol
23530 -mcustom-fadds=253 @gol
23531 -mcustom-fsubs=254 @gol
23532 -mcustom-fdivs=255 @gol
23533 -fsingle-precision-constant}
23535 Custom instruction assignments given by individual
23536 @option{-mcustom-@var{insn}=} options override those given by
23537 @option{-mcustom-fpu-cfg=}, regardless of the
23538 order of the options on the command line.
23540 Note that you can gain more local control over selection of a FPU
23541 configuration by using the @code{target("custom-fpu-cfg=@var{name}")}
23542 function attribute (@pxref{Function Attributes})
23543 or pragma (@pxref{Function Specific Option Pragmas}).
23547 These additional @samp{-m} options are available for the Altera Nios II
23548 ELF (bare-metal) target:
23554 Link with HAL BSP. This suppresses linking with the GCC-provided C runtime
23555 startup and termination code, and is typically used in conjunction with
23556 @option{-msys-crt0=} to specify the location of the alternate startup code
23557 provided by the HAL BSP.
23561 Link with a limited version of the C library, @option{-lsmallc}, rather than
23564 @item -msys-crt0=@var{startfile}
23566 @var{startfile} is the file name of the startfile (crt0) to use
23567 when linking. This option is only useful in conjunction with @option{-mhal}.
23569 @item -msys-lib=@var{systemlib}
23571 @var{systemlib} is the library name of the library that provides
23572 low-level system calls required by the C library,
23573 e.g.@: @code{read} and @code{write}.
23574 This option is typically used to link with a library provided by a HAL BSP.
23578 @node Nvidia PTX Options
23579 @subsection Nvidia PTX Options
23580 @cindex Nvidia PTX options
23581 @cindex nvptx options
23583 These options are defined for Nvidia PTX:
23591 Generate code for 32-bit or 64-bit ABI.
23593 @item -misa=@var{ISA-string}
23595 Generate code for given the specified PTX ISA (e.g.@: @samp{sm_35}). ISA
23596 strings must be lower-case. Valid ISA strings include @samp{sm_30} and
23597 @samp{sm_35}. The default ISA is sm_30.
23600 @opindex mmainkernel
23601 Link in code for a __main kernel. This is for stand-alone instead of
23602 offloading execution.
23606 Apply partitioned execution optimizations. This is the default when any
23607 level of optimization is selected.
23610 @opindex msoft-stack
23611 Generate code that does not use @code{.local} memory
23612 directly for stack storage. Instead, a per-warp stack pointer is
23613 maintained explicitly. This enables variable-length stack allocation (with
23614 variable-length arrays or @code{alloca}), and when global memory is used for
23615 underlying storage, makes it possible to access automatic variables from other
23616 threads, or with atomic instructions. This code generation variant is used
23617 for OpenMP offloading, but the option is exposed on its own for the purpose
23618 of testing the compiler; to generate code suitable for linking into programs
23619 using OpenMP offloading, use option @option{-mgomp}.
23621 @item -muniform-simt
23622 @opindex muniform-simt
23623 Switch to code generation variant that allows to execute all threads in each
23624 warp, while maintaining memory state and side effects as if only one thread
23625 in each warp was active outside of OpenMP SIMD regions. All atomic operations
23626 and calls to runtime (malloc, free, vprintf) are conditionally executed (iff
23627 current lane index equals the master lane index), and the register being
23628 assigned is copied via a shuffle instruction from the master lane. Outside of
23629 SIMD regions lane 0 is the master; inside, each thread sees itself as the
23630 master. Shared memory array @code{int __nvptx_uni[]} stores all-zeros or
23631 all-ones bitmasks for each warp, indicating current mode (0 outside of SIMD
23632 regions). Each thread can bitwise-and the bitmask at position @code{tid.y}
23633 with current lane index to compute the master lane index.
23637 Generate code for use in OpenMP offloading: enables @option{-msoft-stack} and
23638 @option{-muniform-simt} options, and selects corresponding multilib variant.
23642 @node OpenRISC Options
23643 @subsection OpenRISC Options
23644 @cindex OpenRISC Options
23646 These options are defined for OpenRISC:
23650 @item -mboard=@var{name}
23652 Configure a board specific runtime. This will be passed to the linker for
23653 newlib board library linking. The default is @code{or1ksim}.
23657 This option is ignored; it is for compatibility purposes only. This used to
23658 select linker and preprocessor options for use with newlib.
23664 Select software or hardware divide (@code{l.div}, @code{l.divu}) instructions.
23665 This default is hardware divide.
23671 Select software or hardware multiply (@code{l.mul}, @code{l.muli}) instructions.
23672 This default is hardware multiply.
23675 @itemx -mhard-float
23676 @opindex msoft-float
23677 @opindex mhard-float
23678 Select software or hardware for floating point operations.
23679 The default is software.
23681 @item -mdouble-float
23682 @opindex mdouble-float
23683 When @option{-mhard-float} is selected, enables generation of double-precision
23684 floating point instructions. By default functions from @file{libgcc} are used
23685 to perform double-precision floating point operations.
23687 @item -munordered-float
23688 @opindex munordered-float
23689 When @option{-mhard-float} is selected, enables generation of unordered
23690 floating point compare and set flag (@code{lf.sfun*}) instructions. By default
23691 functions from @file{libgcc} are used to perform unordered floating point
23692 compare and set flag operations.
23696 Enable generation of conditional move (@code{l.cmov}) instructions. By
23697 default the equivalent will be generated using using set and branch.
23701 Enable generation of rotate right (@code{l.ror}) instructions. By default
23702 functions from @file{libgcc} are used to perform rotate right operations.
23706 Enable generation of rotate right with immediate (@code{l.rori}) instructions.
23707 By default functions from @file{libgcc} are used to perform rotate right with
23708 immediate operations.
23712 Enable generation of sign extension (@code{l.ext*}) instructions. By default
23713 memory loads are used to perform sign extension.
23717 Enable generation of compare and set flag with immediate (@code{l.sf*i})
23718 instructions. By default extra instructions will be generated to store the
23719 immediate to a register first.
23723 Enable generation of shift with immediate (@code{l.srai}, @code{l.srli},
23724 @code{l.slli}) instructions. By default extra instructions will be generated
23725 to store the immediate to a register first.
23730 @node PDP-11 Options
23731 @subsection PDP-11 Options
23732 @cindex PDP-11 Options
23734 These options are defined for the PDP-11:
23739 Use hardware FPP floating point. This is the default. (FIS floating
23740 point on the PDP-11/40 is not supported.) Implies -m45.
23743 @opindex msoft-float
23744 Do not use hardware floating point.
23748 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
23752 Return floating-point results in memory. This is the default.
23756 Generate code for a PDP-11/40. Implies -msoft-float -mno-split.
23760 Generate code for a PDP-11/45. This is the default.
23764 Generate code for a PDP-11/10. Implies -msoft-float -mno-split.
23770 Use 16-bit @code{int}. This is the default.
23776 Use 32-bit @code{int}.
23780 Target has split instruction and data space. Implies -m45.
23784 Use Unix assembler syntax.
23788 Use DEC assembler syntax.
23792 Use GNU assembler syntax. This is the default.
23796 Use the new LRA register allocator. By default, the old ``reload''
23800 @node picoChip Options
23801 @subsection picoChip Options
23802 @cindex picoChip options
23804 These @samp{-m} options are defined for picoChip implementations:
23808 @item -mae=@var{ae_type}
23810 Set the instruction set, register set, and instruction scheduling
23811 parameters for array element type @var{ae_type}. Supported values
23812 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
23814 @option{-mae=ANY} selects a completely generic AE type. Code
23815 generated with this option runs on any of the other AE types. The
23816 code is not as efficient as it would be if compiled for a specific
23817 AE type, and some types of operation (e.g., multiplication) do not
23818 work properly on all types of AE.
23820 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
23821 for compiled code, and is the default.
23823 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
23824 option may suffer from poor performance of byte (char) manipulation,
23825 since the DSP AE does not provide hardware support for byte load/stores.
23827 @item -msymbol-as-address
23828 Enable the compiler to directly use a symbol name as an address in a
23829 load/store instruction, without first loading it into a
23830 register. Typically, the use of this option generates larger
23831 programs, which run faster than when the option isn't used. However, the
23832 results vary from program to program, so it is left as a user option,
23833 rather than being permanently enabled.
23835 @item -mno-inefficient-warnings
23836 Disables warnings about the generation of inefficient code. These
23837 warnings can be generated, for example, when compiling code that
23838 performs byte-level memory operations on the MAC AE type. The MAC AE has
23839 no hardware support for byte-level memory operations, so all byte
23840 load/stores must be synthesized from word load/store operations. This is
23841 inefficient and a warning is generated to indicate
23842 that you should rewrite the code to avoid byte operations, or to target
23843 an AE type that has the necessary hardware support. This option disables
23848 @node PowerPC Options
23849 @subsection PowerPC Options
23850 @cindex PowerPC options
23852 These are listed under @xref{RS/6000 and PowerPC Options}.
23855 @subsection PRU Options
23856 @cindex PRU Options
23858 These command-line options are defined for PRU target:
23863 Link with a minimum runtime environment, with no support for static
23864 initializers and constructors. Using this option can significantly reduce
23865 the size of the final ELF binary. Beware that the compiler could still
23866 generate code with static initializers and constructors. It is up to the
23867 programmer to ensure that the source program will not use those features.
23869 @item -mmcu=@var{mcu}
23871 Specify the PRU MCU variant to use. Check Newlib for the exact list of
23876 Make GCC pass the @option{--no-relax} command-line option to the linker
23877 instead of the @option{--relax} option.
23881 Allow (or do not allow) GCC to use the LOOP instruction.
23883 @item -mabi=@var{variant}
23885 Specify the ABI variant to output code for. @option{-mabi=ti} selects the
23886 unmodified TI ABI while @option{-mabi=gnu} selects a GNU variant that copes
23887 more naturally with certain GCC assumptions. These are the differences:
23890 @item Function Pointer Size
23891 TI ABI specifies that function (code) pointers are 16-bit, whereas GNU
23892 supports only 32-bit data and code pointers.
23894 @item Optional Return Value Pointer
23895 Function return values larger than 64 bits are passed by using a hidden
23896 pointer as the first argument of the function. TI ABI, though, mandates that
23897 the pointer can be NULL in case the caller is not using the returned value.
23898 GNU always passes and expects a valid return value pointer.
23902 The current @option{-mabi=ti} implementation simply raises a compile error
23903 when any of the above code constructs is detected. As a consequence
23904 the standard C library cannot be built and it is omitted when linking with
23907 Relaxation is a GNU feature and for safety reasons is disabled when using
23908 @option{-mabi=ti}. The TI toolchain does not emit relocations for QBBx
23909 instructions, so the GNU linker cannot adjust them when shortening adjacent
23910 LDI32 pseudo instructions.
23914 @node RISC-V Options
23915 @subsection RISC-V Options
23916 @cindex RISC-V Options
23918 These command-line options are defined for RISC-V targets:
23921 @item -mbranch-cost=@var{n}
23922 @opindex mbranch-cost
23923 Set the cost of branches to roughly @var{n} instructions.
23928 When generating PIC code, do or don't allow the use of PLTs. Ignored for
23929 non-PIC. The default is @option{-mplt}.
23931 @item -mabi=@var{ABI-string}
23933 Specify integer and floating-point calling convention. @var{ABI-string}
23934 contains two parts: the size of integer types and the registers used for
23935 floating-point types. For example @samp{-march=rv64ifd -mabi=lp64d} means that
23936 @samp{long} and pointers are 64-bit (implicitly defining @samp{int} to be
23937 32-bit), and that floating-point values up to 64 bits wide are passed in F
23938 registers. Contrast this with @samp{-march=rv64ifd -mabi=lp64f}, which still
23939 allows the compiler to generate code that uses the F and D extensions but only
23940 allows floating-point values up to 32 bits long to be passed in registers; or
23941 @samp{-march=rv64ifd -mabi=lp64}, in which no floating-point arguments will be
23942 passed in registers.
23944 The default for this argument is system dependent, users who want a specific
23945 calling convention should specify one explicitly. The valid calling
23946 conventions are: @samp{ilp32}, @samp{ilp32f}, @samp{ilp32d}, @samp{lp64},
23947 @samp{lp64f}, and @samp{lp64d}. Some calling conventions are impossible to
23948 implement on some ISAs: for example, @samp{-march=rv32if -mabi=ilp32d} is
23949 invalid because the ABI requires 64-bit values be passed in F registers, but F
23950 registers are only 32 bits wide. There is also the @samp{ilp32e} ABI that can
23951 only be used with the @samp{rv32e} architecture. This ABI is not well
23952 specified at present, and is subject to change.
23957 Do or don't use hardware floating-point divide and square root instructions.
23958 This requires the F or D extensions for floating-point registers. The default
23959 is to use them if the specified architecture has these instructions.
23964 Do or don't use hardware instructions for integer division. This requires the
23965 M extension. The default is to use them if the specified architecture has
23966 these instructions.
23968 @item -march=@var{ISA-string}
23970 Generate code for given RISC-V ISA (e.g.@: @samp{rv64im}). ISA strings must be
23971 lower-case. Examples include @samp{rv64i}, @samp{rv32g}, @samp{rv32e}, and
23974 @item -mtune=@var{processor-string}
23976 Optimize the output for the given processor, specified by microarchitecture
23977 name. Permissible values for this option are: @samp{rocket},
23978 @samp{sifive-3-series}, @samp{sifive-5-series}, @samp{sifive-7-series},
23981 When @option{-mtune=} is not specified, the default is @samp{rocket}.
23983 The @samp{size} choice is not intended for use by end-users. This is used
23984 when @option{-Os} is specified. It overrides the instruction cost info
23985 provided by @option{-mtune=}, but does not override the pipeline info. This
23986 helps reduce code size while still giving good performance.
23988 @item -mpreferred-stack-boundary=@var{num}
23989 @opindex mpreferred-stack-boundary
23990 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
23991 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
23992 the default is 4 (16 bytes or 128-bits).
23994 @strong{Warning:} If you use this switch, then you must build all modules with
23995 the same value, including any libraries. This includes the system libraries
23996 and startup modules.
23998 @item -msmall-data-limit=@var{n}
23999 @opindex msmall-data-limit
24000 Put global and static data smaller than @var{n} bytes into a special section
24003 @item -msave-restore
24004 @itemx -mno-save-restore
24005 @opindex msave-restore
24006 Do or don't use smaller but slower prologue and epilogue code that uses
24007 library function calls. The default is to use fast inline prologues and
24010 @item -mstrict-align
24011 @itemx -mno-strict-align
24012 @opindex mstrict-align
24013 Do not or do generate unaligned memory accesses. The default is set depending
24014 on whether the processor we are optimizing for supports fast unaligned access
24017 @item -mcmodel=medlow
24018 @opindex mcmodel=medlow
24019 Generate code for the medium-low code model. The program and its statically
24020 defined symbols must lie within a single 2 GiB address range and must lie
24021 between absolute addresses @minus{}2 GiB and +2 GiB. Programs can be
24022 statically or dynamically linked. This is the default code model.
24024 @item -mcmodel=medany
24025 @opindex mcmodel=medany
24026 Generate code for the medium-any code model. The program and its statically
24027 defined symbols must be within any single 2 GiB address range. Programs can be
24028 statically or dynamically linked.
24030 @item -mexplicit-relocs
24031 @itemx -mno-exlicit-relocs
24032 Use or do not use assembler relocation operators when dealing with symbolic
24033 addresses. The alternative is to use assembler macros instead, which may
24034 limit optimization.
24038 Take advantage of linker relaxations to reduce the number of instructions
24039 required to materialize symbol addresses. The default is to take advantage of
24040 linker relaxations.
24042 @item -memit-attribute
24043 @itemx -mno-emit-attribute
24044 Emit (do not emit) RISC-V attribute to record extra information into ELF
24045 objects. This feature requires at least binutils 2.32.
24047 @item -malign-data=@var{type}
24048 @opindex malign-data
24049 Control how GCC aligns variables and constants of array, structure, or union
24050 types. Supported values for @var{type} are @samp{xlen} which uses x register
24051 width as the alignment value, and @samp{natural} which uses natural alignment.
24052 @samp{xlen} is the default.
24056 @subsection RL78 Options
24057 @cindex RL78 Options
24063 Links in additional target libraries to support operation within a
24072 Specifies the type of hardware multiplication and division support to
24073 be used. The simplest is @code{none}, which uses software for both
24074 multiplication and division. This is the default. The @code{g13}
24075 value is for the hardware multiply/divide peripheral found on the
24076 RL78/G13 (S2 core) targets. The @code{g14} value selects the use of
24077 the multiplication and division instructions supported by the RL78/G14
24078 (S3 core) parts. The value @code{rl78} is an alias for @code{g14} and
24079 the value @code{mg10} is an alias for @code{none}.
24081 In addition a C preprocessor macro is defined, based upon the setting
24082 of this option. Possible values are: @code{__RL78_MUL_NONE__},
24083 @code{__RL78_MUL_G13__} or @code{__RL78_MUL_G14__}.
24090 Specifies the RL78 core to target. The default is the G14 core, also
24091 known as an S3 core or just RL78. The G13 or S2 core does not have
24092 multiply or divide instructions, instead it uses a hardware peripheral
24093 for these operations. The G10 or S1 core does not have register
24094 banks, so it uses a different calling convention.
24096 If this option is set it also selects the type of hardware multiply
24097 support to use, unless this is overridden by an explicit
24098 @option{-mmul=none} option on the command line. Thus specifying
24099 @option{-mcpu=g13} enables the use of the G13 hardware multiply
24100 peripheral and specifying @option{-mcpu=g10} disables the use of
24101 hardware multiplications altogether.
24103 Note, although the RL78/G14 core is the default target, specifying
24104 @option{-mcpu=g14} or @option{-mcpu=rl78} on the command line does
24105 change the behavior of the toolchain since it also enables G14
24106 hardware multiply support. If these options are not specified on the
24107 command line then software multiplication routines will be used even
24108 though the code targets the RL78 core. This is for backwards
24109 compatibility with older toolchains which did not have hardware
24110 multiply and divide support.
24112 In addition a C preprocessor macro is defined, based upon the setting
24113 of this option. Possible values are: @code{__RL78_G10__},
24114 @code{__RL78_G13__} or @code{__RL78_G14__}.
24124 These are aliases for the corresponding @option{-mcpu=} option. They
24125 are provided for backwards compatibility.
24129 Allow the compiler to use all of the available registers. By default
24130 registers @code{r24..r31} are reserved for use in interrupt handlers.
24131 With this option enabled these registers can be used in ordinary
24134 @item -m64bit-doubles
24135 @itemx -m32bit-doubles
24136 @opindex m64bit-doubles
24137 @opindex m32bit-doubles
24138 Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
24139 or 32 bits (@option{-m32bit-doubles}) in size. The default is
24140 @option{-m32bit-doubles}.
24142 @item -msave-mduc-in-interrupts
24143 @itemx -mno-save-mduc-in-interrupts
24144 @opindex msave-mduc-in-interrupts
24145 @opindex mno-save-mduc-in-interrupts
24146 Specifies that interrupt handler functions should preserve the
24147 MDUC registers. This is only necessary if normal code might use
24148 the MDUC registers, for example because it performs multiplication
24149 and division operations. The default is to ignore the MDUC registers
24150 as this makes the interrupt handlers faster. The target option -mg13
24151 needs to be passed for this to work as this feature is only available
24152 on the G13 target (S2 core). The MDUC registers will only be saved
24153 if the interrupt handler performs a multiplication or division
24154 operation or it calls another function.
24158 @node RS/6000 and PowerPC Options
24159 @subsection IBM RS/6000 and PowerPC Options
24160 @cindex RS/6000 and PowerPC Options
24161 @cindex IBM RS/6000 and PowerPC Options
24163 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
24165 @item -mpowerpc-gpopt
24166 @itemx -mno-powerpc-gpopt
24167 @itemx -mpowerpc-gfxopt
24168 @itemx -mno-powerpc-gfxopt
24171 @itemx -mno-powerpc64
24175 @itemx -mno-popcntb
24177 @itemx -mno-popcntd
24184 @itemx -mno-hard-dfp
24185 @opindex mpowerpc-gpopt
24186 @opindex mno-powerpc-gpopt
24187 @opindex mpowerpc-gfxopt
24188 @opindex mno-powerpc-gfxopt
24189 @opindex mpowerpc64
24190 @opindex mno-powerpc64
24194 @opindex mno-popcntb
24196 @opindex mno-popcntd
24202 @opindex mno-hard-dfp
24203 You use these options to specify which instructions are available on the
24204 processor you are using. The default value of these options is
24205 determined when configuring GCC@. Specifying the
24206 @option{-mcpu=@var{cpu_type}} overrides the specification of these
24207 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
24208 rather than the options listed above.
24210 Specifying @option{-mpowerpc-gpopt} allows
24211 GCC to use the optional PowerPC architecture instructions in the
24212 General Purpose group, including floating-point square root. Specifying
24213 @option{-mpowerpc-gfxopt} allows GCC to
24214 use the optional PowerPC architecture instructions in the Graphics
24215 group, including floating-point select.
24217 The @option{-mmfcrf} option allows GCC to generate the move from
24218 condition register field instruction implemented on the POWER4
24219 processor and other processors that support the PowerPC V2.01
24221 The @option{-mpopcntb} option allows GCC to generate the popcount and
24222 double-precision FP reciprocal estimate instruction implemented on the
24223 POWER5 processor and other processors that support the PowerPC V2.02
24225 The @option{-mpopcntd} option allows GCC to generate the popcount
24226 instruction implemented on the POWER7 processor and other processors
24227 that support the PowerPC V2.06 architecture.
24228 The @option{-mfprnd} option allows GCC to generate the FP round to
24229 integer instructions implemented on the POWER5+ processor and other
24230 processors that support the PowerPC V2.03 architecture.
24231 The @option{-mcmpb} option allows GCC to generate the compare bytes
24232 instruction implemented on the POWER6 processor and other processors
24233 that support the PowerPC V2.05 architecture.
24234 The @option{-mhard-dfp} option allows GCC to generate the decimal
24235 floating-point instructions implemented on some POWER processors.
24237 The @option{-mpowerpc64} option allows GCC to generate the additional
24238 64-bit instructions that are found in the full PowerPC64 architecture
24239 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
24240 @option{-mno-powerpc64}.
24242 @item -mcpu=@var{cpu_type}
24244 Set architecture type, register usage, and
24245 instruction scheduling parameters for machine type @var{cpu_type}.
24246 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
24247 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
24248 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
24249 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
24250 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
24251 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
24252 @samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{e5500},
24253 @samp{e6500}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
24254 @samp{titan}, @samp{power3}, @samp{power4}, @samp{power5}, @samp{power5+},
24255 @samp{power6}, @samp{power6x}, @samp{power7}, @samp{power8},
24256 @samp{power9}, @samp{future}, @samp{powerpc}, @samp{powerpc64},
24257 @samp{powerpc64le}, @samp{rs64}, and @samp{native}.
24259 @option{-mcpu=powerpc}, @option{-mcpu=powerpc64}, and
24260 @option{-mcpu=powerpc64le} specify pure 32-bit PowerPC (either
24261 endian), 64-bit big endian PowerPC and 64-bit little endian PowerPC
24262 architecture machine types, with an appropriate, generic processor
24263 model assumed for scheduling purposes.
24265 Specifying @samp{native} as cpu type detects and selects the
24266 architecture option that corresponds to the host processor of the
24267 system performing the compilation.
24268 @option{-mcpu=native} has no effect if GCC does not recognize the
24271 The other options specify a specific processor. Code generated under
24272 those options runs best on that processor, and may not run at all on
24275 The @option{-mcpu} options automatically enable or disable the
24278 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
24279 -mpopcntb -mpopcntd -mpowerpc64 @gol
24280 -mpowerpc-gpopt -mpowerpc-gfxopt @gol
24281 -mmulhw -mdlmzb -mmfpgpr -mvsx @gol
24282 -mcrypto -mhtm -mpower8-fusion -mpower8-vector @gol
24283 -mquad-memory -mquad-memory-atomic -mfloat128 -mfloat128-hardware}
24285 The particular options set for any particular CPU varies between
24286 compiler versions, depending on what setting seems to produce optimal
24287 code for that CPU; it doesn't necessarily reflect the actual hardware's
24288 capabilities. If you wish to set an individual option to a particular
24289 value, you may specify it after the @option{-mcpu} option, like
24290 @option{-mcpu=970 -mno-altivec}.
24292 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
24293 not enabled or disabled by the @option{-mcpu} option at present because
24294 AIX does not have full support for these options. You may still
24295 enable or disable them individually if you're sure it'll work in your
24298 @item -mtune=@var{cpu_type}
24300 Set the instruction scheduling parameters for machine type
24301 @var{cpu_type}, but do not set the architecture type or register usage,
24302 as @option{-mcpu=@var{cpu_type}} does. The same
24303 values for @var{cpu_type} are used for @option{-mtune} as for
24304 @option{-mcpu}. If both are specified, the code generated uses the
24305 architecture and registers set by @option{-mcpu}, but the
24306 scheduling parameters set by @option{-mtune}.
24308 @item -mcmodel=small
24309 @opindex mcmodel=small
24310 Generate PowerPC64 code for the small model: The TOC is limited to
24313 @item -mcmodel=medium
24314 @opindex mcmodel=medium
24315 Generate PowerPC64 code for the medium model: The TOC and other static
24316 data may be up to a total of 4G in size. This is the default for 64-bit
24319 @item -mcmodel=large
24320 @opindex mcmodel=large
24321 Generate PowerPC64 code for the large model: The TOC may be up to 4G
24322 in size. Other data and code is only limited by the 64-bit address
24326 @itemx -mno-altivec
24328 @opindex mno-altivec
24329 Generate code that uses (does not use) AltiVec instructions, and also
24330 enable the use of built-in functions that allow more direct access to
24331 the AltiVec instruction set. You may also need to set
24332 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
24335 When @option{-maltivec} is used, the element order for AltiVec intrinsics
24336 such as @code{vec_splat}, @code{vec_extract}, and @code{vec_insert}
24337 match array element order corresponding to the endianness of the
24338 target. That is, element zero identifies the leftmost element in a
24339 vector register when targeting a big-endian platform, and identifies
24340 the rightmost element in a vector register when targeting a
24341 little-endian platform.
24346 @opindex mno-vrsave
24347 Generate VRSAVE instructions when generating AltiVec code.
24350 @opindex msecure-plt
24351 Generate code that allows @command{ld} and @command{ld.so}
24352 to build executables and shared
24353 libraries with non-executable @code{.plt} and @code{.got} sections.
24355 32-bit SYSV ABI option.
24359 Generate code that uses a BSS @code{.plt} section that @command{ld.so}
24361 requires @code{.plt} and @code{.got}
24362 sections that are both writable and executable.
24363 This is a PowerPC 32-bit SYSV ABI option.
24369 This switch enables or disables the generation of ISEL instructions.
24375 Generate code that uses (does not use) vector/scalar (VSX)
24376 instructions, and also enable the use of built-in functions that allow
24377 more direct access to the VSX instruction set.
24382 @opindex mno-crypto
24383 Enable the use (disable) of the built-in functions that allow direct
24384 access to the cryptographic instructions that were added in version
24385 2.07 of the PowerPC ISA.
24391 Enable (disable) the use of the built-in functions that allow direct
24392 access to the Hardware Transactional Memory (HTM) instructions that
24393 were added in version 2.07 of the PowerPC ISA.
24395 @item -mpower8-fusion
24396 @itemx -mno-power8-fusion
24397 @opindex mpower8-fusion
24398 @opindex mno-power8-fusion
24399 Generate code that keeps (does not keeps) some integer operations
24400 adjacent so that the instructions can be fused together on power8 and
24403 @item -mpower8-vector
24404 @itemx -mno-power8-vector
24405 @opindex mpower8-vector
24406 @opindex mno-power8-vector
24407 Generate code that uses (does not use) the vector and scalar
24408 instructions that were added in version 2.07 of the PowerPC ISA. Also
24409 enable the use of built-in functions that allow more direct access to
24410 the vector instructions.
24412 @item -mquad-memory
24413 @itemx -mno-quad-memory
24414 @opindex mquad-memory
24415 @opindex mno-quad-memory
24416 Generate code that uses (does not use) the non-atomic quad word memory
24417 instructions. The @option{-mquad-memory} option requires use of
24420 @item -mquad-memory-atomic
24421 @itemx -mno-quad-memory-atomic
24422 @opindex mquad-memory-atomic
24423 @opindex mno-quad-memory-atomic
24424 Generate code that uses (does not use) the atomic quad word memory
24425 instructions. The @option{-mquad-memory-atomic} option requires use of
24429 @itemx -mno-float128
24431 @opindex mno-float128
24432 Enable/disable the @var{__float128} keyword for IEEE 128-bit floating point
24433 and use either software emulation for IEEE 128-bit floating point or
24434 hardware instructions.
24436 The VSX instruction set (@option{-mvsx}, @option{-mcpu=power7},
24437 @option{-mcpu=power8}), or @option{-mcpu=power9} must be enabled to
24438 use the IEEE 128-bit floating point support. The IEEE 128-bit
24439 floating point support only works on PowerPC Linux systems.
24441 The default for @option{-mfloat128} is enabled on PowerPC Linux
24442 systems using the VSX instruction set, and disabled on other systems.
24444 If you use the ISA 3.0 instruction set (@option{-mpower9-vector} or
24445 @option{-mcpu=power9}) on a 64-bit system, the IEEE 128-bit floating
24446 point support will also enable the generation of ISA 3.0 IEEE 128-bit
24447 floating point instructions. Otherwise, if you do not specify to
24448 generate ISA 3.0 instructions or you are targeting a 32-bit big endian
24449 system, IEEE 128-bit floating point will be done with software
24452 @item -mfloat128-hardware
24453 @itemx -mno-float128-hardware
24454 @opindex mfloat128-hardware
24455 @opindex mno-float128-hardware
24456 Enable/disable using ISA 3.0 hardware instructions to support the
24457 @var{__float128} data type.
24459 The default for @option{-mfloat128-hardware} is enabled on PowerPC
24460 Linux systems using the ISA 3.0 instruction set, and disabled on other
24467 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
24468 targets (including GNU/Linux). The 32-bit environment sets int, long
24469 and pointer to 32 bits and generates code that runs on any PowerPC
24470 variant. The 64-bit environment sets int to 32 bits and long and
24471 pointer to 64 bits, and generates code for PowerPC64, as for
24472 @option{-mpowerpc64}.
24475 @itemx -mno-fp-in-toc
24476 @itemx -mno-sum-in-toc
24477 @itemx -mminimal-toc
24479 @opindex mno-fp-in-toc
24480 @opindex mno-sum-in-toc
24481 @opindex mminimal-toc
24482 Modify generation of the TOC (Table Of Contents), which is created for
24483 every executable file. The @option{-mfull-toc} option is selected by
24484 default. In that case, GCC allocates at least one TOC entry for
24485 each unique non-automatic variable reference in your program. GCC
24486 also places floating-point constants in the TOC@. However, only
24487 16,384 entries are available in the TOC@.
24489 If you receive a linker error message that saying you have overflowed
24490 the available TOC space, you can reduce the amount of TOC space used
24491 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
24492 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
24493 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
24494 generate code to calculate the sum of an address and a constant at
24495 run time instead of putting that sum into the TOC@. You may specify one
24496 or both of these options. Each causes GCC to produce very slightly
24497 slower and larger code at the expense of conserving TOC space.
24499 If you still run out of space in the TOC even when you specify both of
24500 these options, specify @option{-mminimal-toc} instead. This option causes
24501 GCC to make only one TOC entry for every file. When you specify this
24502 option, GCC produces code that is slower and larger but which
24503 uses extremely little TOC space. You may wish to use this option
24504 only on files that contain less frequently-executed code.
24510 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
24511 @code{long} type, and the infrastructure needed to support them.
24512 Specifying @option{-maix64} implies @option{-mpowerpc64},
24513 while @option{-maix32} disables the 64-bit ABI and
24514 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
24517 @itemx -mno-xl-compat
24518 @opindex mxl-compat
24519 @opindex mno-xl-compat
24520 Produce code that conforms more closely to IBM XL compiler semantics
24521 when using AIX-compatible ABI@. Pass floating-point arguments to
24522 prototyped functions beyond the register save area (RSA) on the stack
24523 in addition to argument FPRs. Do not assume that most significant
24524 double in 128-bit long double value is properly rounded when comparing
24525 values and converting to double. Use XL symbol names for long double
24528 The AIX calling convention was extended but not initially documented to
24529 handle an obscure K&R C case of calling a function that takes the
24530 address of its arguments with fewer arguments than declared. IBM XL
24531 compilers access floating-point arguments that do not fit in the
24532 RSA from the stack when a subroutine is compiled without
24533 optimization. Because always storing floating-point arguments on the
24534 stack is inefficient and rarely needed, this option is not enabled by
24535 default and only is necessary when calling subroutines compiled by IBM
24536 XL compilers without optimization.
24540 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
24541 application written to use message passing with special startup code to
24542 enable the application to run. The system must have PE installed in the
24543 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
24544 must be overridden with the @option{-specs=} option to specify the
24545 appropriate directory location. The Parallel Environment does not
24546 support threads, so the @option{-mpe} option and the @option{-pthread}
24547 option are incompatible.
24549 @item -malign-natural
24550 @itemx -malign-power
24551 @opindex malign-natural
24552 @opindex malign-power
24553 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
24554 @option{-malign-natural} overrides the ABI-defined alignment of larger
24555 types, such as floating-point doubles, on their natural size-based boundary.
24556 The option @option{-malign-power} instructs GCC to follow the ABI-specified
24557 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
24559 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
24563 @itemx -mhard-float
24564 @opindex msoft-float
24565 @opindex mhard-float
24566 Generate code that does not use (uses) the floating-point register set.
24567 Software floating-point emulation is provided if you use the
24568 @option{-msoft-float} option, and pass the option to GCC when linking.
24571 @itemx -mno-multiple
24573 @opindex mno-multiple
24574 Generate code that uses (does not use) the load multiple word
24575 instructions and the store multiple word instructions. These
24576 instructions are generated by default on POWER systems, and not
24577 generated on PowerPC systems. Do not use @option{-mmultiple} on little-endian
24578 PowerPC systems, since those instructions do not work when the
24579 processor is in little-endian mode. The exceptions are PPC740 and
24580 PPC750 which permit these instructions in little-endian mode.
24585 @opindex mno-update
24586 Generate code that uses (does not use) the load or store instructions
24587 that update the base register to the address of the calculated memory
24588 location. These instructions are generated by default. If you use
24589 @option{-mno-update}, there is a small window between the time that the
24590 stack pointer is updated and the address of the previous frame is
24591 stored, which means code that walks the stack frame across interrupts or
24592 signals may get corrupted data.
24594 @item -mavoid-indexed-addresses
24595 @itemx -mno-avoid-indexed-addresses
24596 @opindex mavoid-indexed-addresses
24597 @opindex mno-avoid-indexed-addresses
24598 Generate code that tries to avoid (not avoid) the use of indexed load
24599 or store instructions. These instructions can incur a performance
24600 penalty on Power6 processors in certain situations, such as when
24601 stepping through large arrays that cross a 16M boundary. This option
24602 is enabled by default when targeting Power6 and disabled otherwise.
24605 @itemx -mno-fused-madd
24606 @opindex mfused-madd
24607 @opindex mno-fused-madd
24608 Generate code that uses (does not use) the floating-point multiply and
24609 accumulate instructions. These instructions are generated by default
24610 if hardware floating point is used. The machine-dependent
24611 @option{-mfused-madd} option is now mapped to the machine-independent
24612 @option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
24613 mapped to @option{-ffp-contract=off}.
24619 Generate code that uses (does not use) the half-word multiply and
24620 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
24621 These instructions are generated by default when targeting those
24628 Generate code that uses (does not use) the string-search @samp{dlmzb}
24629 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
24630 generated by default when targeting those processors.
24632 @item -mno-bit-align
24634 @opindex mno-bit-align
24635 @opindex mbit-align
24636 On System V.4 and embedded PowerPC systems do not (do) force structures
24637 and unions that contain bit-fields to be aligned to the base type of the
24640 For example, by default a structure containing nothing but 8
24641 @code{unsigned} bit-fields of length 1 is aligned to a 4-byte
24642 boundary and has a size of 4 bytes. By using @option{-mno-bit-align},
24643 the structure is aligned to a 1-byte boundary and is 1 byte in
24646 @item -mno-strict-align
24647 @itemx -mstrict-align
24648 @opindex mno-strict-align
24649 @opindex mstrict-align
24650 On System V.4 and embedded PowerPC systems do not (do) assume that
24651 unaligned memory references are handled by the system.
24653 @item -mrelocatable
24654 @itemx -mno-relocatable
24655 @opindex mrelocatable
24656 @opindex mno-relocatable
24657 Generate code that allows (does not allow) a static executable to be
24658 relocated to a different address at run time. A simple embedded
24659 PowerPC system loader should relocate the entire contents of
24660 @code{.got2} and 4-byte locations listed in the @code{.fixup} section,
24661 a table of 32-bit addresses generated by this option. For this to
24662 work, all objects linked together must be compiled with
24663 @option{-mrelocatable} or @option{-mrelocatable-lib}.
24664 @option{-mrelocatable} code aligns the stack to an 8-byte boundary.
24666 @item -mrelocatable-lib
24667 @itemx -mno-relocatable-lib
24668 @opindex mrelocatable-lib
24669 @opindex mno-relocatable-lib
24670 Like @option{-mrelocatable}, @option{-mrelocatable-lib} generates a
24671 @code{.fixup} section to allow static executables to be relocated at
24672 run time, but @option{-mrelocatable-lib} does not use the smaller stack
24673 alignment of @option{-mrelocatable}. Objects compiled with
24674 @option{-mrelocatable-lib} may be linked with objects compiled with
24675 any combination of the @option{-mrelocatable} options.
24681 On System V.4 and embedded PowerPC systems do not (do) assume that
24682 register 2 contains a pointer to a global area pointing to the addresses
24683 used in the program.
24686 @itemx -mlittle-endian
24688 @opindex mlittle-endian
24689 On System V.4 and embedded PowerPC systems compile code for the
24690 processor in little-endian mode. The @option{-mlittle-endian} option is
24691 the same as @option{-mlittle}.
24694 @itemx -mbig-endian
24696 @opindex mbig-endian
24697 On System V.4 and embedded PowerPC systems compile code for the
24698 processor in big-endian mode. The @option{-mbig-endian} option is
24699 the same as @option{-mbig}.
24701 @item -mdynamic-no-pic
24702 @opindex mdynamic-no-pic
24703 On Darwin and Mac OS X systems, compile code so that it is not
24704 relocatable, but that its external references are relocatable. The
24705 resulting code is suitable for applications, but not shared
24708 @item -msingle-pic-base
24709 @opindex msingle-pic-base
24710 Treat the register used for PIC addressing as read-only, rather than
24711 loading it in the prologue for each function. The runtime system is
24712 responsible for initializing this register with an appropriate value
24713 before execution begins.
24715 @item -mprioritize-restricted-insns=@var{priority}
24716 @opindex mprioritize-restricted-insns
24717 This option controls the priority that is assigned to
24718 dispatch-slot restricted instructions during the second scheduling
24719 pass. The argument @var{priority} takes the value @samp{0}, @samp{1},
24720 or @samp{2} to assign no, highest, or second-highest (respectively)
24721 priority to dispatch-slot restricted
24724 @item -msched-costly-dep=@var{dependence_type}
24725 @opindex msched-costly-dep
24726 This option controls which dependences are considered costly
24727 by the target during instruction scheduling. The argument
24728 @var{dependence_type} takes one of the following values:
24732 No dependence is costly.
24735 All dependences are costly.
24737 @item @samp{true_store_to_load}
24738 A true dependence from store to load is costly.
24740 @item @samp{store_to_load}
24741 Any dependence from store to load is costly.
24744 Any dependence for which the latency is greater than or equal to
24745 @var{number} is costly.
24748 @item -minsert-sched-nops=@var{scheme}
24749 @opindex minsert-sched-nops
24750 This option controls which NOP insertion scheme is used during
24751 the second scheduling pass. The argument @var{scheme} takes one of the
24759 Pad with NOPs any dispatch group that has vacant issue slots,
24760 according to the scheduler's grouping.
24762 @item @samp{regroup_exact}
24763 Insert NOPs to force costly dependent insns into
24764 separate groups. Insert exactly as many NOPs as needed to force an insn
24765 to a new group, according to the estimated processor grouping.
24768 Insert NOPs to force costly dependent insns into
24769 separate groups. Insert @var{number} NOPs to force an insn to a new group.
24773 @opindex mcall-sysv
24774 On System V.4 and embedded PowerPC systems compile code using calling
24775 conventions that adhere to the March 1995 draft of the System V
24776 Application Binary Interface, PowerPC processor supplement. This is the
24777 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
24779 @item -mcall-sysv-eabi
24781 @opindex mcall-sysv-eabi
24782 @opindex mcall-eabi
24783 Specify both @option{-mcall-sysv} and @option{-meabi} options.
24785 @item -mcall-sysv-noeabi
24786 @opindex mcall-sysv-noeabi
24787 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
24789 @item -mcall-aixdesc
24791 On System V.4 and embedded PowerPC systems compile code for the AIX
24795 @opindex mcall-linux
24796 On System V.4 and embedded PowerPC systems compile code for the
24797 Linux-based GNU system.
24799 @item -mcall-freebsd
24800 @opindex mcall-freebsd
24801 On System V.4 and embedded PowerPC systems compile code for the
24802 FreeBSD operating system.
24804 @item -mcall-netbsd
24805 @opindex mcall-netbsd
24806 On System V.4 and embedded PowerPC systems compile code for the
24807 NetBSD operating system.
24809 @item -mcall-openbsd
24810 @opindex mcall-netbsd
24811 On System V.4 and embedded PowerPC systems compile code for the
24812 OpenBSD operating system.
24814 @item -mtraceback=@var{traceback_type}
24815 @opindex mtraceback
24816 Select the type of traceback table. Valid values for @var{traceback_type}
24817 are @samp{full}, @samp{part}, and @samp{no}.
24819 @item -maix-struct-return
24820 @opindex maix-struct-return
24821 Return all structures in memory (as specified by the AIX ABI)@.
24823 @item -msvr4-struct-return
24824 @opindex msvr4-struct-return
24825 Return structures smaller than 8 bytes in registers (as specified by the
24828 @item -mabi=@var{abi-type}
24830 Extend the current ABI with a particular extension, or remove such extension.
24831 Valid values are @samp{altivec}, @samp{no-altivec},
24832 @samp{ibmlongdouble}, @samp{ieeelongdouble},
24833 @samp{elfv1}, @samp{elfv2}@.
24835 @item -mabi=ibmlongdouble
24836 @opindex mabi=ibmlongdouble
24837 Change the current ABI to use IBM extended-precision long double.
24838 This is not likely to work if your system defaults to using IEEE
24839 extended-precision long double. If you change the long double type
24840 from IEEE extended-precision, the compiler will issue a warning unless
24841 you use the @option{-Wno-psabi} option. Requires @option{-mlong-double-128}
24844 @item -mabi=ieeelongdouble
24845 @opindex mabi=ieeelongdouble
24846 Change the current ABI to use IEEE extended-precision long double.
24847 This is not likely to work if your system defaults to using IBM
24848 extended-precision long double. If you change the long double type
24849 from IBM extended-precision, the compiler will issue a warning unless
24850 you use the @option{-Wno-psabi} option. Requires @option{-mlong-double-128}
24854 @opindex mabi=elfv1
24855 Change the current ABI to use the ELFv1 ABI.
24856 This is the default ABI for big-endian PowerPC 64-bit Linux.
24857 Overriding the default ABI requires special system support and is
24858 likely to fail in spectacular ways.
24861 @opindex mabi=elfv2
24862 Change the current ABI to use the ELFv2 ABI.
24863 This is the default ABI for little-endian PowerPC 64-bit Linux.
24864 Overriding the default ABI requires special system support and is
24865 likely to fail in spectacular ways.
24867 @item -mgnu-attribute
24868 @itemx -mno-gnu-attribute
24869 @opindex mgnu-attribute
24870 @opindex mno-gnu-attribute
24871 Emit .gnu_attribute assembly directives to set tag/value pairs in a
24872 .gnu.attributes section that specify ABI variations in function
24873 parameters or return values.
24876 @itemx -mno-prototype
24877 @opindex mprototype
24878 @opindex mno-prototype
24879 On System V.4 and embedded PowerPC systems assume that all calls to
24880 variable argument functions are properly prototyped. Otherwise, the
24881 compiler must insert an instruction before every non-prototyped call to
24882 set or clear bit 6 of the condition code register (@code{CR}) to
24883 indicate whether floating-point values are passed in the floating-point
24884 registers in case the function takes variable arguments. With
24885 @option{-mprototype}, only calls to prototyped variable argument functions
24886 set or clear the bit.
24890 On embedded PowerPC systems, assume that the startup module is called
24891 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
24892 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
24897 On embedded PowerPC systems, assume that the startup module is called
24898 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
24903 On embedded PowerPC systems, assume that the startup module is called
24904 @file{crt0.o} and the standard C libraries are @file{libads.a} and
24907 @item -myellowknife
24908 @opindex myellowknife
24909 On embedded PowerPC systems, assume that the startup module is called
24910 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
24915 On System V.4 and embedded PowerPC systems, specify that you are
24916 compiling for a VxWorks system.
24920 On embedded PowerPC systems, set the @code{PPC_EMB} bit in the ELF flags
24921 header to indicate that @samp{eabi} extended relocations are used.
24927 On System V.4 and embedded PowerPC systems do (do not) adhere to the
24928 Embedded Applications Binary Interface (EABI), which is a set of
24929 modifications to the System V.4 specifications. Selecting @option{-meabi}
24930 means that the stack is aligned to an 8-byte boundary, a function
24931 @code{__eabi} is called from @code{main} to set up the EABI
24932 environment, and the @option{-msdata} option can use both @code{r2} and
24933 @code{r13} to point to two separate small data areas. Selecting
24934 @option{-mno-eabi} means that the stack is aligned to a 16-byte boundary,
24935 no EABI initialization function is called from @code{main}, and the
24936 @option{-msdata} option only uses @code{r13} to point to a single
24937 small data area. The @option{-meabi} option is on by default if you
24938 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
24941 @opindex msdata=eabi
24942 On System V.4 and embedded PowerPC systems, put small initialized
24943 @code{const} global and static data in the @code{.sdata2} section, which
24944 is pointed to by register @code{r2}. Put small initialized
24945 non-@code{const} global and static data in the @code{.sdata} section,
24946 which is pointed to by register @code{r13}. Put small uninitialized
24947 global and static data in the @code{.sbss} section, which is adjacent to
24948 the @code{.sdata} section. The @option{-msdata=eabi} option is
24949 incompatible with the @option{-mrelocatable} option. The
24950 @option{-msdata=eabi} option also sets the @option{-memb} option.
24953 @opindex msdata=sysv
24954 On System V.4 and embedded PowerPC systems, put small global and static
24955 data in the @code{.sdata} section, which is pointed to by register
24956 @code{r13}. Put small uninitialized global and static data in the
24957 @code{.sbss} section, which is adjacent to the @code{.sdata} section.
24958 The @option{-msdata=sysv} option is incompatible with the
24959 @option{-mrelocatable} option.
24961 @item -msdata=default
24963 @opindex msdata=default
24965 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
24966 compile code the same as @option{-msdata=eabi}, otherwise compile code the
24967 same as @option{-msdata=sysv}.
24970 @opindex msdata=data
24971 On System V.4 and embedded PowerPC systems, put small global
24972 data in the @code{.sdata} section. Put small uninitialized global
24973 data in the @code{.sbss} section. Do not use register @code{r13}
24974 to address small data however. This is the default behavior unless
24975 other @option{-msdata} options are used.
24979 @opindex msdata=none
24981 On embedded PowerPC systems, put all initialized global and static data
24982 in the @code{.data} section, and all uninitialized data in the
24983 @code{.bss} section.
24985 @item -mreadonly-in-sdata
24986 @opindex mreadonly-in-sdata
24987 @opindex mno-readonly-in-sdata
24988 Put read-only objects in the @code{.sdata} section as well. This is the
24991 @item -mblock-move-inline-limit=@var{num}
24992 @opindex mblock-move-inline-limit
24993 Inline all block moves (such as calls to @code{memcpy} or structure
24994 copies) less than or equal to @var{num} bytes. The minimum value for
24995 @var{num} is 32 bytes on 32-bit targets and 64 bytes on 64-bit
24996 targets. The default value is target-specific.
24998 @item -mblock-compare-inline-limit=@var{num}
24999 @opindex mblock-compare-inline-limit
25000 Generate non-looping inline code for all block compares (such as calls
25001 to @code{memcmp} or structure compares) less than or equal to @var{num}
25002 bytes. If @var{num} is 0, all inline expansion (non-loop and loop) of
25003 block compare is disabled. The default value is target-specific.
25005 @item -mblock-compare-inline-loop-limit=@var{num}
25006 @opindex mblock-compare-inline-loop-limit
25007 Generate an inline expansion using loop code for all block compares that
25008 are less than or equal to @var{num} bytes, but greater than the limit
25009 for non-loop inline block compare expansion. If the block length is not
25010 constant, at most @var{num} bytes will be compared before @code{memcmp}
25011 is called to compare the remainder of the block. The default value is
25014 @item -mstring-compare-inline-limit=@var{num}
25015 @opindex mstring-compare-inline-limit
25016 Compare at most @var{num} string bytes with inline code.
25017 If the difference or end of string is not found at the
25018 end of the inline compare a call to @code{strcmp} or @code{strncmp} will
25019 take care of the rest of the comparison. The default is 64 bytes.
25023 @cindex smaller data references (PowerPC)
25024 @cindex .sdata/.sdata2 references (PowerPC)
25025 On embedded PowerPC systems, put global and static items less than or
25026 equal to @var{num} bytes into the small data or BSS sections instead of
25027 the normal data or BSS section. By default, @var{num} is 8. The
25028 @option{-G @var{num}} switch is also passed to the linker.
25029 All modules should be compiled with the same @option{-G @var{num}} value.
25032 @itemx -mno-regnames
25034 @opindex mno-regnames
25035 On System V.4 and embedded PowerPC systems do (do not) emit register
25036 names in the assembly language output using symbolic forms.
25039 @itemx -mno-longcall
25041 @opindex mno-longcall
25042 By default assume that all calls are far away so that a longer and more
25043 expensive calling sequence is required. This is required for calls
25044 farther than 32 megabytes (33,554,432 bytes) from the current location.
25045 A short call is generated if the compiler knows
25046 the call cannot be that far away. This setting can be overridden by
25047 the @code{shortcall} function attribute, or by @code{#pragma
25050 Some linkers are capable of detecting out-of-range calls and generating
25051 glue code on the fly. On these systems, long calls are unnecessary and
25052 generate slower code. As of this writing, the AIX linker can do this,
25053 as can the GNU linker for PowerPC/64. It is planned to add this feature
25054 to the GNU linker for 32-bit PowerPC systems as well.
25056 On PowerPC64 ELFv2 and 32-bit PowerPC systems with newer GNU linkers,
25057 GCC can generate long calls using an inline PLT call sequence (see
25058 @option{-mpltseq}). PowerPC with @option{-mbss-plt} and PowerPC64
25059 ELFv1 (big-endian) do not support inline PLT calls.
25061 On Darwin/PPC systems, @code{#pragma longcall} generates @code{jbsr
25062 callee, L42}, plus a @dfn{branch island} (glue code). The two target
25063 addresses represent the callee and the branch island. The
25064 Darwin/PPC linker prefers the first address and generates a @code{bl
25065 callee} if the PPC @code{bl} instruction reaches the callee directly;
25066 otherwise, the linker generates @code{bl L42} to call the branch
25067 island. The branch island is appended to the body of the
25068 calling function; it computes the full 32-bit address of the callee
25071 On Mach-O (Darwin) systems, this option directs the compiler emit to
25072 the glue for every direct call, and the Darwin linker decides whether
25073 to use or discard it.
25075 In the future, GCC may ignore all longcall specifications
25076 when the linker is known to generate glue.
25081 @opindex mno-pltseq
25082 Implement (do not implement) -fno-plt and long calls using an inline
25083 PLT call sequence that supports lazy linking and long calls to
25084 functions in dlopen'd shared libraries. Inline PLT calls are only
25085 supported on PowerPC64 ELFv2 and 32-bit PowerPC systems with newer GNU
25086 linkers, and are enabled by default if the support is detected when
25087 configuring GCC, and, in the case of 32-bit PowerPC, if GCC is
25088 configured with @option{--enable-secureplt}. @option{-mpltseq} code
25089 and @option{-mbss-plt} 32-bit PowerPC relocatable objects may not be
25092 @item -mtls-markers
25093 @itemx -mno-tls-markers
25094 @opindex mtls-markers
25095 @opindex mno-tls-markers
25096 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
25097 specifying the function argument. The relocation allows the linker to
25098 reliably associate function call with argument setup instructions for
25099 TLS optimization, which in turn allows GCC to better schedule the
25105 This option enables use of the reciprocal estimate and
25106 reciprocal square root estimate instructions with additional
25107 Newton-Raphson steps to increase precision instead of doing a divide or
25108 square root and divide for floating-point arguments. You should use
25109 the @option{-ffast-math} option when using @option{-mrecip} (or at
25110 least @option{-funsafe-math-optimizations},
25111 @option{-ffinite-math-only}, @option{-freciprocal-math} and
25112 @option{-fno-trapping-math}). Note that while the throughput of the
25113 sequence is generally higher than the throughput of the non-reciprocal
25114 instruction, the precision of the sequence can be decreased by up to 2
25115 ulp (i.e.@: the inverse of 1.0 equals 0.99999994) for reciprocal square
25118 @item -mrecip=@var{opt}
25119 @opindex mrecip=opt
25120 This option controls which reciprocal estimate instructions
25121 may be used. @var{opt} is a comma-separated list of options, which may
25122 be preceded by a @code{!} to invert the option:
25127 Enable all estimate instructions.
25130 Enable the default instructions, equivalent to @option{-mrecip}.
25133 Disable all estimate instructions, equivalent to @option{-mno-recip}.
25136 Enable the reciprocal approximation instructions for both
25137 single and double precision.
25140 Enable the single-precision reciprocal approximation instructions.
25143 Enable the double-precision reciprocal approximation instructions.
25146 Enable the reciprocal square root approximation instructions for both
25147 single and double precision.
25150 Enable the single-precision reciprocal square root approximation instructions.
25153 Enable the double-precision reciprocal square root approximation instructions.
25157 So, for example, @option{-mrecip=all,!rsqrtd} enables
25158 all of the reciprocal estimate instructions, except for the
25159 @code{FRSQRTE}, @code{XSRSQRTEDP}, and @code{XVRSQRTEDP} instructions
25160 which handle the double-precision reciprocal square root calculations.
25162 @item -mrecip-precision
25163 @itemx -mno-recip-precision
25164 @opindex mrecip-precision
25165 Assume (do not assume) that the reciprocal estimate instructions
25166 provide higher-precision estimates than is mandated by the PowerPC
25167 ABI. Selecting @option{-mcpu=power6}, @option{-mcpu=power7} or
25168 @option{-mcpu=power8} automatically selects @option{-mrecip-precision}.
25169 The double-precision square root estimate instructions are not generated by
25170 default on low-precision machines, since they do not provide an
25171 estimate that converges after three steps.
25173 @item -mveclibabi=@var{type}
25174 @opindex mveclibabi
25175 Specifies the ABI type to use for vectorizing intrinsics using an
25176 external library. The only type supported at present is @samp{mass},
25177 which specifies to use IBM's Mathematical Acceleration Subsystem
25178 (MASS) libraries for vectorizing intrinsics using external libraries.
25179 GCC currently emits calls to @code{acosd2}, @code{acosf4},
25180 @code{acoshd2}, @code{acoshf4}, @code{asind2}, @code{asinf4},
25181 @code{asinhd2}, @code{asinhf4}, @code{atan2d2}, @code{atan2f4},
25182 @code{atand2}, @code{atanf4}, @code{atanhd2}, @code{atanhf4},
25183 @code{cbrtd2}, @code{cbrtf4}, @code{cosd2}, @code{cosf4},
25184 @code{coshd2}, @code{coshf4}, @code{erfcd2}, @code{erfcf4},
25185 @code{erfd2}, @code{erff4}, @code{exp2d2}, @code{exp2f4},
25186 @code{expd2}, @code{expf4}, @code{expm1d2}, @code{expm1f4},
25187 @code{hypotd2}, @code{hypotf4}, @code{lgammad2}, @code{lgammaf4},
25188 @code{log10d2}, @code{log10f4}, @code{log1pd2}, @code{log1pf4},
25189 @code{log2d2}, @code{log2f4}, @code{logd2}, @code{logf4},
25190 @code{powd2}, @code{powf4}, @code{sind2}, @code{sinf4}, @code{sinhd2},
25191 @code{sinhf4}, @code{sqrtd2}, @code{sqrtf4}, @code{tand2},
25192 @code{tanf4}, @code{tanhd2}, and @code{tanhf4} when generating code
25193 for power7. Both @option{-ftree-vectorize} and
25194 @option{-funsafe-math-optimizations} must also be enabled. The MASS
25195 libraries must be specified at link time.
25200 Generate (do not generate) the @code{friz} instruction when the
25201 @option{-funsafe-math-optimizations} option is used to optimize
25202 rounding of floating-point values to 64-bit integer and back to floating
25203 point. The @code{friz} instruction does not return the same value if
25204 the floating-point number is too large to fit in an integer.
25206 @item -mpointers-to-nested-functions
25207 @itemx -mno-pointers-to-nested-functions
25208 @opindex mpointers-to-nested-functions
25209 Generate (do not generate) code to load up the static chain register
25210 (@code{r11}) when calling through a pointer on AIX and 64-bit Linux
25211 systems where a function pointer points to a 3-word descriptor giving
25212 the function address, TOC value to be loaded in register @code{r2}, and
25213 static chain value to be loaded in register @code{r11}. The
25214 @option{-mpointers-to-nested-functions} is on by default. You cannot
25215 call through pointers to nested functions or pointers
25216 to functions compiled in other languages that use the static chain if
25217 you use @option{-mno-pointers-to-nested-functions}.
25219 @item -msave-toc-indirect
25220 @itemx -mno-save-toc-indirect
25221 @opindex msave-toc-indirect
25222 Generate (do not generate) code to save the TOC value in the reserved
25223 stack location in the function prologue if the function calls through
25224 a pointer on AIX and 64-bit Linux systems. If the TOC value is not
25225 saved in the prologue, it is saved just before the call through the
25226 pointer. The @option{-mno-save-toc-indirect} option is the default.
25228 @item -mcompat-align-parm
25229 @itemx -mno-compat-align-parm
25230 @opindex mcompat-align-parm
25231 Generate (do not generate) code to pass structure parameters with a
25232 maximum alignment of 64 bits, for compatibility with older versions
25235 Older versions of GCC (prior to 4.9.0) incorrectly did not align a
25236 structure parameter on a 128-bit boundary when that structure contained
25237 a member requiring 128-bit alignment. This is corrected in more
25238 recent versions of GCC. This option may be used to generate code
25239 that is compatible with functions compiled with older versions of
25242 The @option{-mno-compat-align-parm} option is the default.
25244 @item -mstack-protector-guard=@var{guard}
25245 @itemx -mstack-protector-guard-reg=@var{reg}
25246 @itemx -mstack-protector-guard-offset=@var{offset}
25247 @itemx -mstack-protector-guard-symbol=@var{symbol}
25248 @opindex mstack-protector-guard
25249 @opindex mstack-protector-guard-reg
25250 @opindex mstack-protector-guard-offset
25251 @opindex mstack-protector-guard-symbol
25252 Generate stack protection code using canary at @var{guard}. Supported
25253 locations are @samp{global} for global canary or @samp{tls} for per-thread
25254 canary in the TLS block (the default with GNU libc version 2.4 or later).
25256 With the latter choice the options
25257 @option{-mstack-protector-guard-reg=@var{reg}} and
25258 @option{-mstack-protector-guard-offset=@var{offset}} furthermore specify
25259 which register to use as base register for reading the canary, and from what
25260 offset from that base register. The default for those is as specified in the
25261 relevant ABI. @option{-mstack-protector-guard-symbol=@var{symbol}} overrides
25262 the offset with a symbol reference to a canary in the TLS block.
25268 Generate (do not generate) pc-relative addressing when the option
25269 @option{-mcpu=future} is used.
25273 @subsection RX Options
25276 These command-line options are defined for RX targets:
25279 @item -m64bit-doubles
25280 @itemx -m32bit-doubles
25281 @opindex m64bit-doubles
25282 @opindex m32bit-doubles
25283 Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
25284 or 32 bits (@option{-m32bit-doubles}) in size. The default is
25285 @option{-m32bit-doubles}. @emph{Note} RX floating-point hardware only
25286 works on 32-bit values, which is why the default is
25287 @option{-m32bit-doubles}.
25293 Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
25294 floating-point hardware. The default is enabled for the RX600
25295 series and disabled for the RX200 series.
25297 Floating-point instructions are only generated for 32-bit floating-point
25298 values, however, so the FPU hardware is not used for doubles if the
25299 @option{-m64bit-doubles} option is used.
25301 @emph{Note} If the @option{-fpu} option is enabled then
25302 @option{-funsafe-math-optimizations} is also enabled automatically.
25303 This is because the RX FPU instructions are themselves unsafe.
25305 @item -mcpu=@var{name}
25307 Selects the type of RX CPU to be targeted. Currently three types are
25308 supported, the generic @samp{RX600} and @samp{RX200} series hardware and
25309 the specific @samp{RX610} CPU. The default is @samp{RX600}.
25311 The only difference between @samp{RX600} and @samp{RX610} is that the
25312 @samp{RX610} does not support the @code{MVTIPL} instruction.
25314 The @samp{RX200} series does not have a hardware floating-point unit
25315 and so @option{-nofpu} is enabled by default when this type is
25318 @item -mbig-endian-data
25319 @itemx -mlittle-endian-data
25320 @opindex mbig-endian-data
25321 @opindex mlittle-endian-data
25322 Store data (but not code) in the big-endian format. The default is
25323 @option{-mlittle-endian-data}, i.e.@: to store data in the little-endian
25326 @item -msmall-data-limit=@var{N}
25327 @opindex msmall-data-limit
25328 Specifies the maximum size in bytes of global and static variables
25329 which can be placed into the small data area. Using the small data
25330 area can lead to smaller and faster code, but the size of area is
25331 limited and it is up to the programmer to ensure that the area does
25332 not overflow. Also when the small data area is used one of the RX's
25333 registers (usually @code{r13}) is reserved for use pointing to this
25334 area, so it is no longer available for use by the compiler. This
25335 could result in slower and/or larger code if variables are pushed onto
25336 the stack instead of being held in this register.
25338 Note, common variables (variables that have not been initialized) and
25339 constants are not placed into the small data area as they are assigned
25340 to other sections in the output executable.
25342 The default value is zero, which disables this feature. Note, this
25343 feature is not enabled by default with higher optimization levels
25344 (@option{-O2} etc) because of the potentially detrimental effects of
25345 reserving a register. It is up to the programmer to experiment and
25346 discover whether this feature is of benefit to their program. See the
25347 description of the @option{-mpid} option for a description of how the
25348 actual register to hold the small data area pointer is chosen.
25354 Use the simulator runtime. The default is to use the libgloss
25355 board-specific runtime.
25357 @item -mas100-syntax
25358 @itemx -mno-as100-syntax
25359 @opindex mas100-syntax
25360 @opindex mno-as100-syntax
25361 When generating assembler output use a syntax that is compatible with
25362 Renesas's AS100 assembler. This syntax can also be handled by the GAS
25363 assembler, but it has some restrictions so it is not generated by default.
25365 @item -mmax-constant-size=@var{N}
25366 @opindex mmax-constant-size
25367 Specifies the maximum size, in bytes, of a constant that can be used as
25368 an operand in a RX instruction. Although the RX instruction set does
25369 allow constants of up to 4 bytes in length to be used in instructions,
25370 a longer value equates to a longer instruction. Thus in some
25371 circumstances it can be beneficial to restrict the size of constants
25372 that are used in instructions. Constants that are too big are instead
25373 placed into a constant pool and referenced via register indirection.
25375 The value @var{N} can be between 0 and 4. A value of 0 (the default)
25376 or 4 means that constants of any size are allowed.
25380 Enable linker relaxation. Linker relaxation is a process whereby the
25381 linker attempts to reduce the size of a program by finding shorter
25382 versions of various instructions. Disabled by default.
25384 @item -mint-register=@var{N}
25385 @opindex mint-register
25386 Specify the number of registers to reserve for fast interrupt handler
25387 functions. The value @var{N} can be between 0 and 4. A value of 1
25388 means that register @code{r13} is reserved for the exclusive use
25389 of fast interrupt handlers. A value of 2 reserves @code{r13} and
25390 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
25391 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
25392 A value of 0, the default, does not reserve any registers.
25394 @item -msave-acc-in-interrupts
25395 @opindex msave-acc-in-interrupts
25396 Specifies that interrupt handler functions should preserve the
25397 accumulator register. This is only necessary if normal code might use
25398 the accumulator register, for example because it performs 64-bit
25399 multiplications. The default is to ignore the accumulator as this
25400 makes the interrupt handlers faster.
25406 Enables the generation of position independent data. When enabled any
25407 access to constant data is done via an offset from a base address
25408 held in a register. This allows the location of constant data to be
25409 determined at run time without requiring the executable to be
25410 relocated, which is a benefit to embedded applications with tight
25411 memory constraints. Data that can be modified is not affected by this
25414 Note, using this feature reserves a register, usually @code{r13}, for
25415 the constant data base address. This can result in slower and/or
25416 larger code, especially in complicated functions.
25418 The actual register chosen to hold the constant data base address
25419 depends upon whether the @option{-msmall-data-limit} and/or the
25420 @option{-mint-register} command-line options are enabled. Starting
25421 with register @code{r13} and proceeding downwards, registers are
25422 allocated first to satisfy the requirements of @option{-mint-register},
25423 then @option{-mpid} and finally @option{-msmall-data-limit}. Thus it
25424 is possible for the small data area register to be @code{r8} if both
25425 @option{-mint-register=4} and @option{-mpid} are specified on the
25428 By default this feature is not enabled. The default can be restored
25429 via the @option{-mno-pid} command-line option.
25431 @item -mno-warn-multiple-fast-interrupts
25432 @itemx -mwarn-multiple-fast-interrupts
25433 @opindex mno-warn-multiple-fast-interrupts
25434 @opindex mwarn-multiple-fast-interrupts
25435 Prevents GCC from issuing a warning message if it finds more than one
25436 fast interrupt handler when it is compiling a file. The default is to
25437 issue a warning for each extra fast interrupt handler found, as the RX
25438 only supports one such interrupt.
25440 @item -mallow-string-insns
25441 @itemx -mno-allow-string-insns
25442 @opindex mallow-string-insns
25443 @opindex mno-allow-string-insns
25444 Enables or disables the use of the string manipulation instructions
25445 @code{SMOVF}, @code{SCMPU}, @code{SMOVB}, @code{SMOVU}, @code{SUNTIL}
25446 @code{SWHILE} and also the @code{RMPA} instruction. These
25447 instructions may prefetch data, which is not safe to do if accessing
25448 an I/O register. (See section 12.2.7 of the RX62N Group User's Manual
25449 for more information).
25451 The default is to allow these instructions, but it is not possible for
25452 GCC to reliably detect all circumstances where a string instruction
25453 might be used to access an I/O register, so their use cannot be
25454 disabled automatically. Instead it is reliant upon the programmer to
25455 use the @option{-mno-allow-string-insns} option if their program
25456 accesses I/O space.
25458 When the instructions are enabled GCC defines the C preprocessor
25459 symbol @code{__RX_ALLOW_STRING_INSNS__}, otherwise it defines the
25460 symbol @code{__RX_DISALLOW_STRING_INSNS__}.
25466 Use only (or not only) @code{JSR} instructions to access functions.
25467 This option can be used when code size exceeds the range of @code{BSR}
25468 instructions. Note that @option{-mno-jsr} does not mean to not use
25469 @code{JSR} but instead means that any type of branch may be used.
25472 @emph{Note:} The generic GCC command-line option @option{-ffixed-@var{reg}}
25473 has special significance to the RX port when used with the
25474 @code{interrupt} function attribute. This attribute indicates a
25475 function intended to process fast interrupts. GCC ensures
25476 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
25477 and/or @code{r13} and only provided that the normal use of the
25478 corresponding registers have been restricted via the
25479 @option{-ffixed-@var{reg}} or @option{-mint-register} command-line
25482 @node S/390 and zSeries Options
25483 @subsection S/390 and zSeries Options
25484 @cindex S/390 and zSeries Options
25486 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
25490 @itemx -msoft-float
25491 @opindex mhard-float
25492 @opindex msoft-float
25493 Use (do not use) the hardware floating-point instructions and registers
25494 for floating-point operations. When @option{-msoft-float} is specified,
25495 functions in @file{libgcc.a} are used to perform floating-point
25496 operations. When @option{-mhard-float} is specified, the compiler
25497 generates IEEE floating-point instructions. This is the default.
25500 @itemx -mno-hard-dfp
25502 @opindex mno-hard-dfp
25503 Use (do not use) the hardware decimal-floating-point instructions for
25504 decimal-floating-point operations. When @option{-mno-hard-dfp} is
25505 specified, functions in @file{libgcc.a} are used to perform
25506 decimal-floating-point operations. When @option{-mhard-dfp} is
25507 specified, the compiler generates decimal-floating-point hardware
25508 instructions. This is the default for @option{-march=z9-ec} or higher.
25510 @item -mlong-double-64
25511 @itemx -mlong-double-128
25512 @opindex mlong-double-64
25513 @opindex mlong-double-128
25514 These switches control the size of @code{long double} type. A size
25515 of 64 bits makes the @code{long double} type equivalent to the @code{double}
25516 type. This is the default.
25519 @itemx -mno-backchain
25520 @opindex mbackchain
25521 @opindex mno-backchain
25522 Store (do not store) the address of the caller's frame as backchain pointer
25523 into the callee's stack frame.
25524 A backchain may be needed to allow debugging using tools that do not understand
25525 DWARF call frame information.
25526 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
25527 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
25528 the backchain is placed into the topmost word of the 96/160 byte register
25531 In general, code compiled with @option{-mbackchain} is call-compatible with
25532 code compiled with @option{-mmo-backchain}; however, use of the backchain
25533 for debugging purposes usually requires that the whole binary is built with
25534 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
25535 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
25536 to build a linux kernel use @option{-msoft-float}.
25538 The default is to not maintain the backchain.
25540 @item -mpacked-stack
25541 @itemx -mno-packed-stack
25542 @opindex mpacked-stack
25543 @opindex mno-packed-stack
25544 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
25545 specified, the compiler uses the all fields of the 96/160 byte register save
25546 area only for their default purpose; unused fields still take up stack space.
25547 When @option{-mpacked-stack} is specified, register save slots are densely
25548 packed at the top of the register save area; unused space is reused for other
25549 purposes, allowing for more efficient use of the available stack space.
25550 However, when @option{-mbackchain} is also in effect, the topmost word of
25551 the save area is always used to store the backchain, and the return address
25552 register is always saved two words below the backchain.
25554 As long as the stack frame backchain is not used, code generated with
25555 @option{-mpacked-stack} is call-compatible with code generated with
25556 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
25557 S/390 or zSeries generated code that uses the stack frame backchain at run
25558 time, not just for debugging purposes. Such code is not call-compatible
25559 with code compiled with @option{-mpacked-stack}. Also, note that the
25560 combination of @option{-mbackchain},
25561 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
25562 to build a linux kernel use @option{-msoft-float}.
25564 The default is to not use the packed stack layout.
25567 @itemx -mno-small-exec
25568 @opindex msmall-exec
25569 @opindex mno-small-exec
25570 Generate (or do not generate) code using the @code{bras} instruction
25571 to do subroutine calls.
25572 This only works reliably if the total executable size does not
25573 exceed 64k. The default is to use the @code{basr} instruction instead,
25574 which does not have this limitation.
25580 When @option{-m31} is specified, generate code compliant to the
25581 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
25582 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
25583 particular to generate 64-bit instructions. For the @samp{s390}
25584 targets, the default is @option{-m31}, while the @samp{s390x}
25585 targets default to @option{-m64}.
25591 When @option{-mzarch} is specified, generate code using the
25592 instructions available on z/Architecture.
25593 When @option{-mesa} is specified, generate code using the
25594 instructions available on ESA/390. Note that @option{-mesa} is
25595 not possible with @option{-m64}.
25596 When generating code compliant to the GNU/Linux for S/390 ABI,
25597 the default is @option{-mesa}. When generating code compliant
25598 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
25604 The @option{-mhtm} option enables a set of builtins making use of
25605 instructions available with the transactional execution facility
25606 introduced with the IBM zEnterprise EC12 machine generation
25607 @ref{S/390 System z Built-in Functions}.
25608 @option{-mhtm} is enabled by default when using @option{-march=zEC12}.
25614 When @option{-mvx} is specified, generate code using the instructions
25615 available with the vector extension facility introduced with the IBM
25616 z13 machine generation.
25617 This option changes the ABI for some vector type values with regard to
25618 alignment and calling conventions. In case vector type values are
25619 being used in an ABI-relevant context a GAS @samp{.gnu_attribute}
25620 command will be added to mark the resulting binary with the ABI used.
25621 @option{-mvx} is enabled by default when using @option{-march=z13}.
25624 @itemx -mno-zvector
25626 @opindex mno-zvector
25627 The @option{-mzvector} option enables vector language extensions and
25628 builtins using instructions available with the vector extension
25629 facility introduced with the IBM z13 machine generation.
25630 This option adds support for @samp{vector} to be used as a keyword to
25631 define vector type variables and arguments. @samp{vector} is only
25632 available when GNU extensions are enabled. It will not be expanded
25633 when requesting strict standard compliance e.g.@: with @option{-std=c99}.
25634 In addition to the GCC low-level builtins @option{-mzvector} enables
25635 a set of builtins added for compatibility with AltiVec-style
25636 implementations like Power and Cell. In order to make use of these
25637 builtins the header file @file{vecintrin.h} needs to be included.
25638 @option{-mzvector} is disabled by default.
25644 Generate (or do not generate) code using the @code{mvcle} instruction
25645 to perform block moves. When @option{-mno-mvcle} is specified,
25646 use a @code{mvc} loop instead. This is the default unless optimizing for
25653 Print (or do not print) additional debug information when compiling.
25654 The default is to not print debug information.
25656 @item -march=@var{cpu-type}
25658 Generate code that runs on @var{cpu-type}, which is the name of a
25659 system representing a certain processor type. Possible values for
25660 @var{cpu-type} are @samp{z900}/@samp{arch5}, @samp{z990}/@samp{arch6},
25661 @samp{z9-109}, @samp{z9-ec}/@samp{arch7}, @samp{z10}/@samp{arch8},
25662 @samp{z196}/@samp{arch9}, @samp{zEC12}, @samp{z13}/@samp{arch11},
25663 @samp{z14}/@samp{arch12}, and @samp{native}.
25665 The default is @option{-march=z900}.
25667 Specifying @samp{native} as cpu type can be used to select the best
25668 architecture option for the host processor.
25669 @option{-march=native} has no effect if GCC does not recognize the
25672 @item -mtune=@var{cpu-type}
25674 Tune to @var{cpu-type} everything applicable about the generated code,
25675 except for the ABI and the set of available instructions.
25676 The list of @var{cpu-type} values is the same as for @option{-march}.
25677 The default is the value used for @option{-march}.
25680 @itemx -mno-tpf-trace
25681 @opindex mtpf-trace
25682 @opindex mno-tpf-trace
25683 Generate code that adds (does not add) in TPF OS specific branches to trace
25684 routines in the operating system. This option is off by default, even
25685 when compiling for the TPF OS@.
25688 @itemx -mno-fused-madd
25689 @opindex mfused-madd
25690 @opindex mno-fused-madd
25691 Generate code that uses (does not use) the floating-point multiply and
25692 accumulate instructions. These instructions are generated by default if
25693 hardware floating point is used.
25695 @item -mwarn-framesize=@var{framesize}
25696 @opindex mwarn-framesize
25697 Emit a warning if the current function exceeds the given frame size. Because
25698 this is a compile-time check it doesn't need to be a real problem when the program
25699 runs. It is intended to identify functions that most probably cause
25700 a stack overflow. It is useful to be used in an environment with limited stack
25701 size e.g.@: the linux kernel.
25703 @item -mwarn-dynamicstack
25704 @opindex mwarn-dynamicstack
25705 Emit a warning if the function calls @code{alloca} or uses dynamically-sized
25706 arrays. This is generally a bad idea with a limited stack size.
25708 @item -mstack-guard=@var{stack-guard}
25709 @itemx -mstack-size=@var{stack-size}
25710 @opindex mstack-guard
25711 @opindex mstack-size
25712 If these options are provided the S/390 back end emits additional instructions in
25713 the function prologue that trigger a trap if the stack size is @var{stack-guard}
25714 bytes above the @var{stack-size} (remember that the stack on S/390 grows downward).
25715 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
25716 the frame size of the compiled function is chosen.
25717 These options are intended to be used to help debugging stack overflow problems.
25718 The additionally emitted code causes only little overhead and hence can also be
25719 used in production-like systems without greater performance degradation. The given
25720 values have to be exact powers of 2 and @var{stack-size} has to be greater than
25721 @var{stack-guard} without exceeding 64k.
25722 In order to be efficient the extra code makes the assumption that the stack starts
25723 at an address aligned to the value given by @var{stack-size}.
25724 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
25726 @item -mhotpatch=@var{pre-halfwords},@var{post-halfwords}
25728 If the hotpatch option is enabled, a ``hot-patching'' function
25729 prologue is generated for all functions in the compilation unit.
25730 The funtion label is prepended with the given number of two-byte
25731 NOP instructions (@var{pre-halfwords}, maximum 1000000). After
25732 the label, 2 * @var{post-halfwords} bytes are appended, using the
25733 largest NOP like instructions the architecture allows (maximum
25736 If both arguments are zero, hotpatching is disabled.
25738 This option can be overridden for individual functions with the
25739 @code{hotpatch} attribute.
25742 @node Score Options
25743 @subsection Score Options
25744 @cindex Score Options
25746 These options are defined for Score implementations:
25751 Compile code for big-endian mode. This is the default.
25755 Compile code for little-endian mode.
25759 Disable generation of @code{bcnz} instructions.
25763 Enable generation of unaligned load and store instructions.
25767 Enable the use of multiply-accumulate instructions. Disabled by default.
25771 Specify the SCORE5 as the target architecture.
25775 Specify the SCORE5U of the target architecture.
25779 Specify the SCORE7 as the target architecture. This is the default.
25783 Specify the SCORE7D as the target architecture.
25787 @subsection SH Options
25789 These @samp{-m} options are defined for the SH implementations:
25794 Generate code for the SH1.
25798 Generate code for the SH2.
25801 Generate code for the SH2e.
25805 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
25806 that the floating-point unit is not used.
25808 @item -m2a-single-only
25809 @opindex m2a-single-only
25810 Generate code for the SH2a-FPU, in such a way that no double-precision
25811 floating-point operations are used.
25814 @opindex m2a-single
25815 Generate code for the SH2a-FPU assuming the floating-point unit is in
25816 single-precision mode by default.
25820 Generate code for the SH2a-FPU assuming the floating-point unit is in
25821 double-precision mode by default.
25825 Generate code for the SH3.
25829 Generate code for the SH3e.
25833 Generate code for the SH4 without a floating-point unit.
25835 @item -m4-single-only
25836 @opindex m4-single-only
25837 Generate code for the SH4 with a floating-point unit that only
25838 supports single-precision arithmetic.
25842 Generate code for the SH4 assuming the floating-point unit is in
25843 single-precision mode by default.
25847 Generate code for the SH4.
25851 Generate code for SH4-100.
25853 @item -m4-100-nofpu
25854 @opindex m4-100-nofpu
25855 Generate code for SH4-100 in such a way that the
25856 floating-point unit is not used.
25858 @item -m4-100-single
25859 @opindex m4-100-single
25860 Generate code for SH4-100 assuming the floating-point unit is in
25861 single-precision mode by default.
25863 @item -m4-100-single-only
25864 @opindex m4-100-single-only
25865 Generate code for SH4-100 in such a way that no double-precision
25866 floating-point operations are used.
25870 Generate code for SH4-200.
25872 @item -m4-200-nofpu
25873 @opindex m4-200-nofpu
25874 Generate code for SH4-200 without in such a way that the
25875 floating-point unit is not used.
25877 @item -m4-200-single
25878 @opindex m4-200-single
25879 Generate code for SH4-200 assuming the floating-point unit is in
25880 single-precision mode by default.
25882 @item -m4-200-single-only
25883 @opindex m4-200-single-only
25884 Generate code for SH4-200 in such a way that no double-precision
25885 floating-point operations are used.
25889 Generate code for SH4-300.
25891 @item -m4-300-nofpu
25892 @opindex m4-300-nofpu
25893 Generate code for SH4-300 without in such a way that the
25894 floating-point unit is not used.
25896 @item -m4-300-single
25897 @opindex m4-300-single
25898 Generate code for SH4-300 in such a way that no double-precision
25899 floating-point operations are used.
25901 @item -m4-300-single-only
25902 @opindex m4-300-single-only
25903 Generate code for SH4-300 in such a way that no double-precision
25904 floating-point operations are used.
25908 Generate code for SH4-340 (no MMU, no FPU).
25912 Generate code for SH4-500 (no FPU). Passes @option{-isa=sh4-nofpu} to the
25917 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
25918 floating-point unit is not used.
25920 @item -m4a-single-only
25921 @opindex m4a-single-only
25922 Generate code for the SH4a, in such a way that no double-precision
25923 floating-point operations are used.
25926 @opindex m4a-single
25927 Generate code for the SH4a assuming the floating-point unit is in
25928 single-precision mode by default.
25932 Generate code for the SH4a.
25936 Same as @option{-m4a-nofpu}, except that it implicitly passes
25937 @option{-dsp} to the assembler. GCC doesn't generate any DSP
25938 instructions at the moment.
25942 Compile code for the processor in big-endian mode.
25946 Compile code for the processor in little-endian mode.
25950 Align doubles at 64-bit boundaries. Note that this changes the calling
25951 conventions, and thus some functions from the standard C library do
25952 not work unless you recompile it first with @option{-mdalign}.
25956 Shorten some address references at link time, when possible; uses the
25957 linker option @option{-relax}.
25961 Use 32-bit offsets in @code{switch} tables. The default is to use
25966 Enable the use of bit manipulation instructions on SH2A.
25970 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
25971 alignment constraints.
25975 Comply with the calling conventions defined by Renesas.
25978 @opindex mno-renesas
25979 Comply with the calling conventions defined for GCC before the Renesas
25980 conventions were available. This option is the default for all
25981 targets of the SH toolchain.
25984 @opindex mnomacsave
25985 Mark the @code{MAC} register as call-clobbered, even if
25986 @option{-mrenesas} is given.
25992 Control the IEEE compliance of floating-point comparisons, which affects the
25993 handling of cases where the result of a comparison is unordered. By default
25994 @option{-mieee} is implicitly enabled. If @option{-ffinite-math-only} is
25995 enabled @option{-mno-ieee} is implicitly set, which results in faster
25996 floating-point greater-equal and less-equal comparisons. The implicit settings
25997 can be overridden by specifying either @option{-mieee} or @option{-mno-ieee}.
25999 @item -minline-ic_invalidate
26000 @opindex minline-ic_invalidate
26001 Inline code to invalidate instruction cache entries after setting up
26002 nested function trampolines.
26003 This option has no effect if @option{-musermode} is in effect and the selected
26004 code generation option (e.g.@: @option{-m4}) does not allow the use of the @code{icbi}
26006 If the selected code generation option does not allow the use of the @code{icbi}
26007 instruction, and @option{-musermode} is not in effect, the inlined code
26008 manipulates the instruction cache address array directly with an associative
26009 write. This not only requires privileged mode at run time, but it also
26010 fails if the cache line had been mapped via the TLB and has become unmapped.
26014 Dump instruction size and location in the assembly code.
26017 @opindex mpadstruct
26018 This option is deprecated. It pads structures to multiple of 4 bytes,
26019 which is incompatible with the SH ABI@.
26021 @item -matomic-model=@var{model}
26022 @opindex matomic-model=@var{model}
26023 Sets the model of atomic operations and additional parameters as a comma
26024 separated list. For details on the atomic built-in functions see
26025 @ref{__atomic Builtins}. The following models and parameters are supported:
26030 Disable compiler generated atomic sequences and emit library calls for atomic
26031 operations. This is the default if the target is not @code{sh*-*-linux*}.
26034 Generate GNU/Linux compatible gUSA software atomic sequences for the atomic
26035 built-in functions. The generated atomic sequences require additional support
26036 from the interrupt/exception handling code of the system and are only suitable
26037 for SH3* and SH4* single-core systems. This option is enabled by default when
26038 the target is @code{sh*-*-linux*} and SH3* or SH4*. When the target is SH4A,
26039 this option also partially utilizes the hardware atomic instructions
26040 @code{movli.l} and @code{movco.l} to create more efficient code, unless
26041 @samp{strict} is specified.
26044 Generate software atomic sequences that use a variable in the thread control
26045 block. This is a variation of the gUSA sequences which can also be used on
26046 SH1* and SH2* targets. The generated atomic sequences require additional
26047 support from the interrupt/exception handling code of the system and are only
26048 suitable for single-core systems. When using this model, the @samp{gbr-offset=}
26049 parameter has to be specified as well.
26052 Generate software atomic sequences that temporarily disable interrupts by
26053 setting @code{SR.IMASK = 1111}. This model works only when the program runs
26054 in privileged mode and is only suitable for single-core systems. Additional
26055 support from the interrupt/exception handling code of the system is not
26056 required. This model is enabled by default when the target is
26057 @code{sh*-*-linux*} and SH1* or SH2*.
26060 Generate hardware atomic sequences using the @code{movli.l} and @code{movco.l}
26061 instructions only. This is only available on SH4A and is suitable for
26062 multi-core systems. Since the hardware instructions support only 32 bit atomic
26063 variables access to 8 or 16 bit variables is emulated with 32 bit accesses.
26064 Code compiled with this option is also compatible with other software
26065 atomic model interrupt/exception handling systems if executed on an SH4A
26066 system. Additional support from the interrupt/exception handling code of the
26067 system is not required for this model.
26070 This parameter specifies the offset in bytes of the variable in the thread
26071 control block structure that should be used by the generated atomic sequences
26072 when the @samp{soft-tcb} model has been selected. For other models this
26073 parameter is ignored. The specified value must be an integer multiple of four
26074 and in the range 0-1020.
26077 This parameter prevents mixed usage of multiple atomic models, even if they
26078 are compatible, and makes the compiler generate atomic sequences of the
26079 specified model only.
26085 Generate the @code{tas.b} opcode for @code{__atomic_test_and_set}.
26086 Notice that depending on the particular hardware and software configuration
26087 this can degrade overall performance due to the operand cache line flushes
26088 that are implied by the @code{tas.b} instruction. On multi-core SH4A
26089 processors the @code{tas.b} instruction must be used with caution since it
26090 can result in data corruption for certain cache configurations.
26093 @opindex mprefergot
26094 When generating position-independent code, emit function calls using
26095 the Global Offset Table instead of the Procedure Linkage Table.
26098 @itemx -mno-usermode
26100 @opindex mno-usermode
26101 Don't allow (allow) the compiler generating privileged mode code. Specifying
26102 @option{-musermode} also implies @option{-mno-inline-ic_invalidate} if the
26103 inlined code would not work in user mode. @option{-musermode} is the default
26104 when the target is @code{sh*-*-linux*}. If the target is SH1* or SH2*
26105 @option{-musermode} has no effect, since there is no user mode.
26107 @item -multcost=@var{number}
26108 @opindex multcost=@var{number}
26109 Set the cost to assume for a multiply insn.
26111 @item -mdiv=@var{strategy}
26112 @opindex mdiv=@var{strategy}
26113 Set the division strategy to be used for integer division operations.
26114 @var{strategy} can be one of:
26119 Calls a library function that uses the single-step division instruction
26120 @code{div1} to perform the operation. Division by zero calculates an
26121 unspecified result and does not trap. This is the default except for SH4,
26122 SH2A and SHcompact.
26125 Calls a library function that performs the operation in double precision
26126 floating point. Division by zero causes a floating-point exception. This is
26127 the default for SHcompact with FPU. Specifying this for targets that do not
26128 have a double precision FPU defaults to @code{call-div1}.
26131 Calls a library function that uses a lookup table for small divisors and
26132 the @code{div1} instruction with case distinction for larger divisors. Division
26133 by zero calculates an unspecified result and does not trap. This is the default
26134 for SH4. Specifying this for targets that do not have dynamic shift
26135 instructions defaults to @code{call-div1}.
26139 When a division strategy has not been specified the default strategy is
26140 selected based on the current target. For SH2A the default strategy is to
26141 use the @code{divs} and @code{divu} instructions instead of library function
26144 @item -maccumulate-outgoing-args
26145 @opindex maccumulate-outgoing-args
26146 Reserve space once for outgoing arguments in the function prologue rather
26147 than around each call. Generally beneficial for performance and size. Also
26148 needed for unwinding to avoid changing the stack frame around conditional code.
26150 @item -mdivsi3_libfunc=@var{name}
26151 @opindex mdivsi3_libfunc=@var{name}
26152 Set the name of the library function used for 32-bit signed division to
26154 This only affects the name used in the @samp{call} division strategies, and
26155 the compiler still expects the same sets of input/output/clobbered registers as
26156 if this option were not present.
26158 @item -mfixed-range=@var{register-range}
26159 @opindex mfixed-range
26160 Generate code treating the given register range as fixed registers.
26161 A fixed register is one that the register allocator cannot use. This is
26162 useful when compiling kernel code. A register range is specified as
26163 two registers separated by a dash. Multiple register ranges can be
26164 specified separated by a comma.
26166 @item -mbranch-cost=@var{num}
26167 @opindex mbranch-cost=@var{num}
26168 Assume @var{num} to be the cost for a branch instruction. Higher numbers
26169 make the compiler try to generate more branch-free code if possible.
26170 If not specified the value is selected depending on the processor type that
26171 is being compiled for.
26174 @itemx -mno-zdcbranch
26175 @opindex mzdcbranch
26176 @opindex mno-zdcbranch
26177 Assume (do not assume) that zero displacement conditional branch instructions
26178 @code{bt} and @code{bf} are fast. If @option{-mzdcbranch} is specified, the
26179 compiler prefers zero displacement branch code sequences. This is
26180 enabled by default when generating code for SH4 and SH4A. It can be explicitly
26181 disabled by specifying @option{-mno-zdcbranch}.
26183 @item -mcbranch-force-delay-slot
26184 @opindex mcbranch-force-delay-slot
26185 Force the usage of delay slots for conditional branches, which stuffs the delay
26186 slot with a @code{nop} if a suitable instruction cannot be found. By default
26187 this option is disabled. It can be enabled to work around hardware bugs as
26188 found in the original SH7055.
26191 @itemx -mno-fused-madd
26192 @opindex mfused-madd
26193 @opindex mno-fused-madd
26194 Generate code that uses (does not use) the floating-point multiply and
26195 accumulate instructions. These instructions are generated by default
26196 if hardware floating point is used. The machine-dependent
26197 @option{-mfused-madd} option is now mapped to the machine-independent
26198 @option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
26199 mapped to @option{-ffp-contract=off}.
26205 Allow or disallow the compiler to emit the @code{fsca} instruction for sine
26206 and cosine approximations. The option @option{-mfsca} must be used in
26207 combination with @option{-funsafe-math-optimizations}. It is enabled by default
26208 when generating code for SH4A. Using @option{-mno-fsca} disables sine and cosine
26209 approximations even if @option{-funsafe-math-optimizations} is in effect.
26215 Allow or disallow the compiler to emit the @code{fsrra} instruction for
26216 reciprocal square root approximations. The option @option{-mfsrra} must be used
26217 in combination with @option{-funsafe-math-optimizations} and
26218 @option{-ffinite-math-only}. It is enabled by default when generating code for
26219 SH4A. Using @option{-mno-fsrra} disables reciprocal square root approximations
26220 even if @option{-funsafe-math-optimizations} and @option{-ffinite-math-only} are
26223 @item -mpretend-cmove
26224 @opindex mpretend-cmove
26225 Prefer zero-displacement conditional branches for conditional move instruction
26226 patterns. This can result in faster code on the SH4 processor.
26230 Generate code using the FDPIC ABI.
26234 @node Solaris 2 Options
26235 @subsection Solaris 2 Options
26236 @cindex Solaris 2 options
26238 These @samp{-m} options are supported on Solaris 2:
26241 @item -mclear-hwcap
26242 @opindex mclear-hwcap
26243 @option{-mclear-hwcap} tells the compiler to remove the hardware
26244 capabilities generated by the Solaris assembler. This is only necessary
26245 when object files use ISA extensions not supported by the current
26246 machine, but check at runtime whether or not to use them.
26248 @item -mimpure-text
26249 @opindex mimpure-text
26250 @option{-mimpure-text}, used in addition to @option{-shared}, tells
26251 the compiler to not pass @option{-z text} to the linker when linking a
26252 shared object. Using this option, you can link position-dependent
26253 code into a shared object.
26255 @option{-mimpure-text} suppresses the ``relocations remain against
26256 allocatable but non-writable sections'' linker error message.
26257 However, the necessary relocations trigger copy-on-write, and the
26258 shared object is not actually shared across processes. Instead of
26259 using @option{-mimpure-text}, you should compile all source code with
26260 @option{-fpic} or @option{-fPIC}.
26264 These switches are supported in addition to the above on Solaris 2:
26269 This is a synonym for @option{-pthread}.
26272 @node SPARC Options
26273 @subsection SPARC Options
26274 @cindex SPARC options
26276 These @samp{-m} options are supported on the SPARC:
26279 @item -mno-app-regs
26281 @opindex mno-app-regs
26283 Specify @option{-mapp-regs} to generate output using the global registers
26284 2 through 4, which the SPARC SVR4 ABI reserves for applications. Like the
26285 global register 1, each global register 2 through 4 is then treated as an
26286 allocable register that is clobbered by function calls. This is the default.
26288 To be fully SVR4 ABI-compliant at the cost of some performance loss,
26289 specify @option{-mno-app-regs}. You should compile libraries and system
26290 software with this option.
26296 With @option{-mflat}, the compiler does not generate save/restore instructions
26297 and uses a ``flat'' or single register window model. This model is compatible
26298 with the regular register window model. The local registers and the input
26299 registers (0--5) are still treated as ``call-saved'' registers and are
26300 saved on the stack as needed.
26302 With @option{-mno-flat} (the default), the compiler generates save/restore
26303 instructions (except for leaf functions). This is the normal operating mode.
26306 @itemx -mhard-float
26308 @opindex mhard-float
26309 Generate output containing floating-point instructions. This is the
26313 @itemx -msoft-float
26315 @opindex msoft-float
26316 Generate output containing library calls for floating point.
26317 @strong{Warning:} the requisite libraries are not available for all SPARC
26318 targets. Normally the facilities of the machine's usual C compiler are
26319 used, but this cannot be done directly in cross-compilation. You must make
26320 your own arrangements to provide suitable library functions for
26321 cross-compilation. The embedded targets @samp{sparc-*-aout} and
26322 @samp{sparclite-*-*} do provide software floating-point support.
26324 @option{-msoft-float} changes the calling convention in the output file;
26325 therefore, it is only useful if you compile @emph{all} of a program with
26326 this option. In particular, you need to compile @file{libgcc.a}, the
26327 library that comes with GCC, with @option{-msoft-float} in order for
26330 @item -mhard-quad-float
26331 @opindex mhard-quad-float
26332 Generate output containing quad-word (long double) floating-point
26335 @item -msoft-quad-float
26336 @opindex msoft-quad-float
26337 Generate output containing library calls for quad-word (long double)
26338 floating-point instructions. The functions called are those specified
26339 in the SPARC ABI@. This is the default.
26341 As of this writing, there are no SPARC implementations that have hardware
26342 support for the quad-word floating-point instructions. They all invoke
26343 a trap handler for one of these instructions, and then the trap handler
26344 emulates the effect of the instruction. Because of the trap handler overhead,
26345 this is much slower than calling the ABI library routines. Thus the
26346 @option{-msoft-quad-float} option is the default.
26348 @item -mno-unaligned-doubles
26349 @itemx -munaligned-doubles
26350 @opindex mno-unaligned-doubles
26351 @opindex munaligned-doubles
26352 Assume that doubles have 8-byte alignment. This is the default.
26354 With @option{-munaligned-doubles}, GCC assumes that doubles have 8-byte
26355 alignment only if they are contained in another type, or if they have an
26356 absolute address. Otherwise, it assumes they have 4-byte alignment.
26357 Specifying this option avoids some rare compatibility problems with code
26358 generated by other compilers. It is not the default because it results
26359 in a performance loss, especially for floating-point code.
26362 @itemx -mno-user-mode
26363 @opindex muser-mode
26364 @opindex mno-user-mode
26365 Do not generate code that can only run in supervisor mode. This is relevant
26366 only for the @code{casa} instruction emitted for the LEON3 processor. This
26369 @item -mfaster-structs
26370 @itemx -mno-faster-structs
26371 @opindex mfaster-structs
26372 @opindex mno-faster-structs
26373 With @option{-mfaster-structs}, the compiler assumes that structures
26374 should have 8-byte alignment. This enables the use of pairs of
26375 @code{ldd} and @code{std} instructions for copies in structure
26376 assignment, in place of twice as many @code{ld} and @code{st} pairs.
26377 However, the use of this changed alignment directly violates the SPARC
26378 ABI@. Thus, it's intended only for use on targets where the developer
26379 acknowledges that their resulting code is not directly in line with
26380 the rules of the ABI@.
26382 @item -mstd-struct-return
26383 @itemx -mno-std-struct-return
26384 @opindex mstd-struct-return
26385 @opindex mno-std-struct-return
26386 With @option{-mstd-struct-return}, the compiler generates checking code
26387 in functions returning structures or unions to detect size mismatches
26388 between the two sides of function calls, as per the 32-bit ABI@.
26390 The default is @option{-mno-std-struct-return}. This option has no effect
26397 Enable Local Register Allocation. This is the default for SPARC since GCC 7
26398 so @option{-mno-lra} needs to be passed to get old Reload.
26400 @item -mcpu=@var{cpu_type}
26402 Set the instruction set, register set, and instruction scheduling parameters
26403 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
26404 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{hypersparc},
26405 @samp{leon}, @samp{leon3}, @samp{leon3v7}, @samp{sparclite}, @samp{f930},
26406 @samp{f934}, @samp{sparclite86x}, @samp{sparclet}, @samp{tsc701}, @samp{v9},
26407 @samp{ultrasparc}, @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2},
26408 @samp{niagara3}, @samp{niagara4}, @samp{niagara7} and @samp{m8}.
26410 Native Solaris and GNU/Linux toolchains also support the value @samp{native},
26411 which selects the best architecture option for the host processor.
26412 @option{-mcpu=native} has no effect if GCC does not recognize
26415 Default instruction scheduling parameters are used for values that select
26416 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
26417 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
26419 Here is a list of each supported architecture and their supported
26427 supersparc, hypersparc, leon, leon3
26430 f930, f934, sparclite86x
26436 ultrasparc, ultrasparc3, niagara, niagara2, niagara3, niagara4,
26440 By default (unless configured otherwise), GCC generates code for the V7
26441 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
26442 additionally optimizes it for the Cypress CY7C602 chip, as used in the
26443 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
26444 SPARCStation 1, 2, IPX etc.
26446 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
26447 architecture. The only difference from V7 code is that the compiler emits
26448 the integer multiply and integer divide instructions which exist in SPARC-V8
26449 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
26450 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
26453 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
26454 the SPARC architecture. This adds the integer multiply, integer divide step
26455 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
26456 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
26457 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
26458 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
26459 MB86934 chip, which is the more recent SPARClite with FPU@.
26461 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
26462 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
26463 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
26464 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
26465 optimizes it for the TEMIC SPARClet chip.
26467 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
26468 architecture. This adds 64-bit integer and floating-point move instructions,
26469 3 additional floating-point condition code registers and conditional move
26470 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
26471 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
26472 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
26473 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
26474 @option{-mcpu=niagara}, the compiler additionally optimizes it for
26475 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
26476 additionally optimizes it for Sun UltraSPARC T2 chips. With
26477 @option{-mcpu=niagara3}, the compiler additionally optimizes it for Sun
26478 UltraSPARC T3 chips. With @option{-mcpu=niagara4}, the compiler
26479 additionally optimizes it for Sun UltraSPARC T4 chips. With
26480 @option{-mcpu=niagara7}, the compiler additionally optimizes it for
26481 Oracle SPARC M7 chips. With @option{-mcpu=m8}, the compiler
26482 additionally optimizes it for Oracle M8 chips.
26484 @item -mtune=@var{cpu_type}
26486 Set the instruction scheduling parameters for machine type
26487 @var{cpu_type}, but do not set the instruction set or register set that the
26488 option @option{-mcpu=@var{cpu_type}} does.
26490 The same values for @option{-mcpu=@var{cpu_type}} can be used for
26491 @option{-mtune=@var{cpu_type}}, but the only useful values are those
26492 that select a particular CPU implementation. Those are
26493 @samp{cypress}, @samp{supersparc}, @samp{hypersparc}, @samp{leon},
26494 @samp{leon3}, @samp{leon3v7}, @samp{f930}, @samp{f934},
26495 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
26496 @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2}, @samp{niagara3},
26497 @samp{niagara4}, @samp{niagara7} and @samp{m8}. With native Solaris
26498 and GNU/Linux toolchains, @samp{native} can also be used.
26503 @opindex mno-v8plus
26504 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
26505 difference from the V8 ABI is that the global and out registers are
26506 considered 64 bits wide. This is enabled by default on Solaris in 32-bit
26507 mode for all SPARC-V9 processors.
26513 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
26514 Visual Instruction Set extensions. The default is @option{-mno-vis}.
26520 With @option{-mvis2}, GCC generates code that takes advantage of
26521 version 2.0 of the UltraSPARC Visual Instruction Set extensions. The
26522 default is @option{-mvis2} when targeting a cpu that supports such
26523 instructions, such as UltraSPARC-III and later. Setting @option{-mvis2}
26524 also sets @option{-mvis}.
26530 With @option{-mvis3}, GCC generates code that takes advantage of
26531 version 3.0 of the UltraSPARC Visual Instruction Set extensions. The
26532 default is @option{-mvis3} when targeting a cpu that supports such
26533 instructions, such as niagara-3 and later. Setting @option{-mvis3}
26534 also sets @option{-mvis2} and @option{-mvis}.
26540 With @option{-mvis4}, GCC generates code that takes advantage of
26541 version 4.0 of the UltraSPARC Visual Instruction Set extensions. The
26542 default is @option{-mvis4} when targeting a cpu that supports such
26543 instructions, such as niagara-7 and later. Setting @option{-mvis4}
26544 also sets @option{-mvis3}, @option{-mvis2} and @option{-mvis}.
26550 With @option{-mvis4b}, GCC generates code that takes advantage of
26551 version 4.0 of the UltraSPARC Visual Instruction Set extensions, plus
26552 the additional VIS instructions introduced in the Oracle SPARC
26553 Architecture 2017. The default is @option{-mvis4b} when targeting a
26554 cpu that supports such instructions, such as m8 and later. Setting
26555 @option{-mvis4b} also sets @option{-mvis4}, @option{-mvis3},
26556 @option{-mvis2} and @option{-mvis}.
26561 @opindex mno-cbcond
26562 With @option{-mcbcond}, GCC generates code that takes advantage of the UltraSPARC
26563 Compare-and-Branch-on-Condition instructions. The default is @option{-mcbcond}
26564 when targeting a CPU that supports such instructions, such as Niagara-4 and
26571 With @option{-mfmaf}, GCC generates code that takes advantage of the UltraSPARC
26572 Fused Multiply-Add Floating-point instructions. The default is @option{-mfmaf}
26573 when targeting a CPU that supports such instructions, such as Niagara-3 and
26579 @opindex mno-fsmuld
26580 With @option{-mfsmuld}, GCC generates code that takes advantage of the
26581 Floating-point Multiply Single to Double (FsMULd) instruction. The default is
26582 @option{-mfsmuld} when targeting a CPU supporting the architecture versions V8
26583 or V9 with FPU except @option{-mcpu=leon}.
26589 With @option{-mpopc}, GCC generates code that takes advantage of the UltraSPARC
26590 Population Count instruction. The default is @option{-mpopc}
26591 when targeting a CPU that supports such an instruction, such as Niagara-2 and
26598 With @option{-msubxc}, GCC generates code that takes advantage of the UltraSPARC
26599 Subtract-Extended-with-Carry instruction. The default is @option{-msubxc}
26600 when targeting a CPU that supports such an instruction, such as Niagara-7 and
26604 @opindex mfix-at697f
26605 Enable the documented workaround for the single erratum of the Atmel AT697F
26606 processor (which corresponds to erratum #13 of the AT697E processor).
26609 @opindex mfix-ut699
26610 Enable the documented workarounds for the floating-point errata and the data
26611 cache nullify errata of the UT699 processor.
26614 @opindex mfix-ut700
26615 Enable the documented workaround for the back-to-back store errata of
26616 the UT699E/UT700 processor.
26618 @item -mfix-gr712rc
26619 @opindex mfix-gr712rc
26620 Enable the documented workaround for the back-to-back store errata of
26621 the GR712RC processor.
26624 These @samp{-m} options are supported in addition to the above
26625 on SPARC-V9 processors in 64-bit environments:
26632 Generate code for a 32-bit or 64-bit environment.
26633 The 32-bit environment sets int, long and pointer to 32 bits.
26634 The 64-bit environment sets int to 32 bits and long and pointer
26637 @item -mcmodel=@var{which}
26639 Set the code model to one of
26643 The Medium/Low code model: 64-bit addresses, programs
26644 must be linked in the low 32 bits of memory. Programs can be statically
26645 or dynamically linked.
26648 The Medium/Middle code model: 64-bit addresses, programs
26649 must be linked in the low 44 bits of memory, the text and data segments must
26650 be less than 2GB in size and the data segment must be located within 2GB of
26654 The Medium/Anywhere code model: 64-bit addresses, programs
26655 may be linked anywhere in memory, the text and data segments must be less
26656 than 2GB in size and the data segment must be located within 2GB of the
26660 The Medium/Anywhere code model for embedded systems:
26661 64-bit addresses, the text and data segments must be less than 2GB in
26662 size, both starting anywhere in memory (determined at link time). The
26663 global register %g4 points to the base of the data segment. Programs
26664 are statically linked and PIC is not supported.
26667 @item -mmemory-model=@var{mem-model}
26668 @opindex mmemory-model
26669 Set the memory model in force on the processor to one of
26673 The default memory model for the processor and operating system.
26676 Relaxed Memory Order
26679 Partial Store Order
26685 Sequential Consistency
26688 These memory models are formally defined in Appendix D of the SPARC-V9
26689 architecture manual, as set in the processor's @code{PSTATE.MM} field.
26692 @itemx -mno-stack-bias
26693 @opindex mstack-bias
26694 @opindex mno-stack-bias
26695 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
26696 frame pointer if present, are offset by @minus{}2047 which must be added back
26697 when making stack frame references. This is the default in 64-bit mode.
26698 Otherwise, assume no such offset is present.
26702 @subsection SPU Options
26703 @cindex SPU options
26705 These @samp{-m} options are supported on the SPU:
26709 @itemx -merror-reloc
26710 @opindex mwarn-reloc
26711 @opindex merror-reloc
26713 The loader for SPU does not handle dynamic relocations. By default, GCC
26714 gives an error when it generates code that requires a dynamic
26715 relocation. @option{-mno-error-reloc} disables the error,
26716 @option{-mwarn-reloc} generates a warning instead.
26719 @itemx -munsafe-dma
26721 @opindex munsafe-dma
26723 Instructions that initiate or test completion of DMA must not be
26724 reordered with respect to loads and stores of the memory that is being
26726 With @option{-munsafe-dma} you must use the @code{volatile} keyword to protect
26727 memory accesses, but that can lead to inefficient code in places where the
26728 memory is known to not change. Rather than mark the memory as volatile,
26729 you can use @option{-msafe-dma} to tell the compiler to treat
26730 the DMA instructions as potentially affecting all memory.
26732 @item -mbranch-hints
26733 @opindex mbranch-hints
26735 By default, GCC generates a branch hint instruction to avoid
26736 pipeline stalls for always-taken or probably-taken branches. A hint
26737 is not generated closer than 8 instructions away from its branch.
26738 There is little reason to disable them, except for debugging purposes,
26739 or to make an object a little bit smaller.
26743 @opindex msmall-mem
26744 @opindex mlarge-mem
26746 By default, GCC generates code assuming that addresses are never larger
26747 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
26748 a full 32-bit address.
26753 By default, GCC links against startup code that assumes the SPU-style
26754 main function interface (which has an unconventional parameter list).
26755 With @option{-mstdmain}, GCC links your program against startup
26756 code that assumes a C99-style interface to @code{main}, including a
26757 local copy of @code{argv} strings.
26759 @item -mfixed-range=@var{register-range}
26760 @opindex mfixed-range
26761 Generate code treating the given register range as fixed registers.
26762 A fixed register is one that the register allocator cannot use. This is
26763 useful when compiling kernel code. A register range is specified as
26764 two registers separated by a dash. Multiple register ranges can be
26765 specified separated by a comma.
26771 Compile code assuming that pointers to the PPU address space accessed
26772 via the @code{__ea} named address space qualifier are either 32 or 64
26773 bits wide. The default is 32 bits. As this is an ABI-changing option,
26774 all object code in an executable must be compiled with the same setting.
26776 @item -maddress-space-conversion
26777 @itemx -mno-address-space-conversion
26778 @opindex maddress-space-conversion
26779 @opindex mno-address-space-conversion
26780 Allow/disallow treating the @code{__ea} address space as superset
26781 of the generic address space. This enables explicit type casts
26782 between @code{__ea} and generic pointer as well as implicit
26783 conversions of generic pointers to @code{__ea} pointers. The
26784 default is to allow address space pointer conversions.
26786 @item -mcache-size=@var{cache-size}
26787 @opindex mcache-size
26788 This option controls the version of libgcc that the compiler links to an
26789 executable and selects a software-managed cache for accessing variables
26790 in the @code{__ea} address space with a particular cache size. Possible
26791 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
26792 and @samp{128}. The default cache size is 64KB.
26794 @item -matomic-updates
26795 @itemx -mno-atomic-updates
26796 @opindex matomic-updates
26797 @opindex mno-atomic-updates
26798 This option controls the version of libgcc that the compiler links to an
26799 executable and selects whether atomic updates to the software-managed
26800 cache of PPU-side variables are used. If you use atomic updates, changes
26801 to a PPU variable from SPU code using the @code{__ea} named address space
26802 qualifier do not interfere with changes to other PPU variables residing
26803 in the same cache line from PPU code. If you do not use atomic updates,
26804 such interference may occur; however, writing back cache lines is
26805 more efficient. The default behavior is to use atomic updates.
26808 @itemx -mdual-nops=@var{n}
26809 @opindex mdual-nops
26810 By default, GCC inserts NOPs to increase dual issue when it expects
26811 it to increase performance. @var{n} can be a value from 0 to 10. A
26812 smaller @var{n} inserts fewer NOPs. 10 is the default, 0 is the
26813 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
26815 @item -mhint-max-nops=@var{n}
26816 @opindex mhint-max-nops
26817 Maximum number of NOPs to insert for a branch hint. A branch hint must
26818 be at least 8 instructions away from the branch it is affecting. GCC
26819 inserts up to @var{n} NOPs to enforce this, otherwise it does not
26820 generate the branch hint.
26822 @item -mhint-max-distance=@var{n}
26823 @opindex mhint-max-distance
26824 The encoding of the branch hint instruction limits the hint to be within
26825 256 instructions of the branch it is affecting. By default, GCC makes
26826 sure it is within 125.
26829 @opindex msafe-hints
26830 Work around a hardware bug that causes the SPU to stall indefinitely.
26831 By default, GCC inserts the @code{hbrp} instruction to make sure
26832 this stall won't happen.
26836 @node System V Options
26837 @subsection Options for System V
26839 These additional options are available on System V Release 4 for
26840 compatibility with other compilers on those systems:
26845 Create a shared object.
26846 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
26850 Identify the versions of each tool used by the compiler, in a
26851 @code{.ident} assembler directive in the output.
26855 Refrain from adding @code{.ident} directives to the output file (this is
26858 @item -YP,@var{dirs}
26860 Search the directories @var{dirs}, and no others, for libraries
26861 specified with @option{-l}.
26863 @item -Ym,@var{dir}
26865 Look in the directory @var{dir} to find the M4 preprocessor.
26866 The assembler uses this option.
26867 @c This is supposed to go with a -Yd for predefined M4 macro files, but
26868 @c the generic assembler that comes with Solaris takes just -Ym.
26871 @node TILE-Gx Options
26872 @subsection TILE-Gx Options
26873 @cindex TILE-Gx options
26875 These @samp{-m} options are supported on the TILE-Gx:
26878 @item -mcmodel=small
26879 @opindex mcmodel=small
26880 Generate code for the small model. The distance for direct calls is
26881 limited to 500M in either direction. PC-relative addresses are 32
26882 bits. Absolute addresses support the full address range.
26884 @item -mcmodel=large
26885 @opindex mcmodel=large
26886 Generate code for the large model. There is no limitation on call
26887 distance, pc-relative addresses, or absolute addresses.
26889 @item -mcpu=@var{name}
26891 Selects the type of CPU to be targeted. Currently the only supported
26892 type is @samp{tilegx}.
26898 Generate code for a 32-bit or 64-bit environment. The 32-bit
26899 environment sets int, long, and pointer to 32 bits. The 64-bit
26900 environment sets int to 32 bits and long and pointer to 64 bits.
26903 @itemx -mlittle-endian
26904 @opindex mbig-endian
26905 @opindex mlittle-endian
26906 Generate code in big/little endian mode, respectively.
26909 @node TILEPro Options
26910 @subsection TILEPro Options
26911 @cindex TILEPro options
26913 These @samp{-m} options are supported on the TILEPro:
26916 @item -mcpu=@var{name}
26918 Selects the type of CPU to be targeted. Currently the only supported
26919 type is @samp{tilepro}.
26923 Generate code for a 32-bit environment, which sets int, long, and
26924 pointer to 32 bits. This is the only supported behavior so the flag
26925 is essentially ignored.
26929 @subsection V850 Options
26930 @cindex V850 Options
26932 These @samp{-m} options are defined for V850 implementations:
26936 @itemx -mno-long-calls
26937 @opindex mlong-calls
26938 @opindex mno-long-calls
26939 Treat all calls as being far away (near). If calls are assumed to be
26940 far away, the compiler always loads the function's address into a
26941 register, and calls indirect through the pointer.
26947 Do not optimize (do optimize) basic blocks that use the same index
26948 pointer 4 or more times to copy pointer into the @code{ep} register, and
26949 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
26950 option is on by default if you optimize.
26952 @item -mno-prolog-function
26953 @itemx -mprolog-function
26954 @opindex mno-prolog-function
26955 @opindex mprolog-function
26956 Do not use (do use) external functions to save and restore registers
26957 at the prologue and epilogue of a function. The external functions
26958 are slower, but use less code space if more than one function saves
26959 the same number of registers. The @option{-mprolog-function} option
26960 is on by default if you optimize.
26964 Try to make the code as small as possible. At present, this just turns
26965 on the @option{-mep} and @option{-mprolog-function} options.
26967 @item -mtda=@var{n}
26969 Put static or global variables whose size is @var{n} bytes or less into
26970 the tiny data area that register @code{ep} points to. The tiny data
26971 area can hold up to 256 bytes in total (128 bytes for byte references).
26973 @item -msda=@var{n}
26975 Put static or global variables whose size is @var{n} bytes or less into
26976 the small data area that register @code{gp} points to. The small data
26977 area can hold up to 64 kilobytes.
26979 @item -mzda=@var{n}
26981 Put static or global variables whose size is @var{n} bytes or less into
26982 the first 32 kilobytes of memory.
26986 Specify that the target processor is the V850.
26990 Specify that the target processor is the V850E3V5. The preprocessor
26991 constant @code{__v850e3v5__} is defined if this option is used.
26995 Specify that the target processor is the V850E3V5. This is an alias for
26996 the @option{-mv850e3v5} option.
27000 Specify that the target processor is the V850E2V3. The preprocessor
27001 constant @code{__v850e2v3__} is defined if this option is used.
27005 Specify that the target processor is the V850E2. The preprocessor
27006 constant @code{__v850e2__} is defined if this option is used.
27010 Specify that the target processor is the V850E1. The preprocessor
27011 constants @code{__v850e1__} and @code{__v850e__} are defined if
27012 this option is used.
27016 Specify that the target processor is the V850ES. This is an alias for
27017 the @option{-mv850e1} option.
27021 Specify that the target processor is the V850E@. The preprocessor
27022 constant @code{__v850e__} is defined if this option is used.
27024 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
27025 nor @option{-mv850e2} nor @option{-mv850e2v3} nor @option{-mv850e3v5}
27026 are defined then a default target processor is chosen and the
27027 relevant @samp{__v850*__} preprocessor constant is defined.
27029 The preprocessor constants @code{__v850} and @code{__v851__} are always
27030 defined, regardless of which processor variant is the target.
27032 @item -mdisable-callt
27033 @itemx -mno-disable-callt
27034 @opindex mdisable-callt
27035 @opindex mno-disable-callt
27036 This option suppresses generation of the @code{CALLT} instruction for the
27037 v850e, v850e1, v850e2, v850e2v3 and v850e3v5 flavors of the v850
27040 This option is enabled by default when the RH850 ABI is
27041 in use (see @option{-mrh850-abi}), and disabled by default when the
27042 GCC ABI is in use. If @code{CALLT} instructions are being generated
27043 then the C preprocessor symbol @code{__V850_CALLT__} is defined.
27049 Pass on (or do not pass on) the @option{-mrelax} command-line option
27053 @itemx -mno-long-jumps
27054 @opindex mlong-jumps
27055 @opindex mno-long-jumps
27056 Disable (or re-enable) the generation of PC-relative jump instructions.
27059 @itemx -mhard-float
27060 @opindex msoft-float
27061 @opindex mhard-float
27062 Disable (or re-enable) the generation of hardware floating point
27063 instructions. This option is only significant when the target
27064 architecture is @samp{V850E2V3} or higher. If hardware floating point
27065 instructions are being generated then the C preprocessor symbol
27066 @code{__FPU_OK__} is defined, otherwise the symbol
27067 @code{__NO_FPU__} is defined.
27071 Enables the use of the e3v5 LOOP instruction. The use of this
27072 instruction is not enabled by default when the e3v5 architecture is
27073 selected because its use is still experimental.
27077 @opindex mrh850-abi
27079 Enables support for the RH850 version of the V850 ABI. This is the
27080 default. With this version of the ABI the following rules apply:
27084 Integer sized structures and unions are returned via a memory pointer
27085 rather than a register.
27088 Large structures and unions (more than 8 bytes in size) are passed by
27092 Functions are aligned to 16-bit boundaries.
27095 The @option{-m8byte-align} command-line option is supported.
27098 The @option{-mdisable-callt} command-line option is enabled by
27099 default. The @option{-mno-disable-callt} command-line option is not
27103 When this version of the ABI is enabled the C preprocessor symbol
27104 @code{__V850_RH850_ABI__} is defined.
27108 Enables support for the old GCC version of the V850 ABI. With this
27109 version of the ABI the following rules apply:
27113 Integer sized structures and unions are returned in register @code{r10}.
27116 Large structures and unions (more than 8 bytes in size) are passed by
27120 Functions are aligned to 32-bit boundaries, unless optimizing for
27124 The @option{-m8byte-align} command-line option is not supported.
27127 The @option{-mdisable-callt} command-line option is supported but not
27128 enabled by default.
27131 When this version of the ABI is enabled the C preprocessor symbol
27132 @code{__V850_GCC_ABI__} is defined.
27134 @item -m8byte-align
27135 @itemx -mno-8byte-align
27136 @opindex m8byte-align
27137 @opindex mno-8byte-align
27138 Enables support for @code{double} and @code{long long} types to be
27139 aligned on 8-byte boundaries. The default is to restrict the
27140 alignment of all objects to at most 4-bytes. When
27141 @option{-m8byte-align} is in effect the C preprocessor symbol
27142 @code{__V850_8BYTE_ALIGN__} is defined.
27145 @opindex mbig-switch
27146 Generate code suitable for big switch tables. Use this option only if
27147 the assembler/linker complain about out of range branches within a switch
27152 This option causes r2 and r5 to be used in the code generated by
27153 the compiler. This setting is the default.
27155 @item -mno-app-regs
27156 @opindex mno-app-regs
27157 This option causes r2 and r5 to be treated as fixed registers.
27162 @subsection VAX Options
27163 @cindex VAX options
27165 These @samp{-m} options are defined for the VAX:
27170 Do not output certain jump instructions (@code{aobleq} and so on)
27171 that the Unix assembler for the VAX cannot handle across long
27176 Do output those jump instructions, on the assumption that the
27177 GNU assembler is being used.
27181 Output code for G-format floating-point numbers instead of D-format.
27184 @node Visium Options
27185 @subsection Visium Options
27186 @cindex Visium options
27192 A program which performs file I/O and is destined to run on an MCM target
27193 should be linked with this option. It causes the libraries libc.a and
27194 libdebug.a to be linked. The program should be run on the target under
27195 the control of the GDB remote debugging stub.
27199 A program which performs file I/O and is destined to run on the simulator
27200 should be linked with option. This causes libraries libc.a and libsim.a to
27204 @itemx -mhard-float
27206 @opindex mhard-float
27207 Generate code containing floating-point instructions. This is the
27211 @itemx -msoft-float
27213 @opindex msoft-float
27214 Generate code containing library calls for floating-point.
27216 @option{-msoft-float} changes the calling convention in the output file;
27217 therefore, it is only useful if you compile @emph{all} of a program with
27218 this option. In particular, you need to compile @file{libgcc.a}, the
27219 library that comes with GCC, with @option{-msoft-float} in order for
27222 @item -mcpu=@var{cpu_type}
27224 Set the instruction set, register set, and instruction scheduling parameters
27225 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
27226 @samp{mcm}, @samp{gr5} and @samp{gr6}.
27228 @samp{mcm} is a synonym of @samp{gr5} present for backward compatibility.
27230 By default (unless configured otherwise), GCC generates code for the GR5
27231 variant of the Visium architecture.
27233 With @option{-mcpu=gr6}, GCC generates code for the GR6 variant of the Visium
27234 architecture. The only difference from GR5 code is that the compiler will
27235 generate block move instructions.
27237 @item -mtune=@var{cpu_type}
27239 Set the instruction scheduling parameters for machine type @var{cpu_type},
27240 but do not set the instruction set or register set that the option
27241 @option{-mcpu=@var{cpu_type}} would.
27245 Generate code for the supervisor mode, where there are no restrictions on
27246 the access to general registers. This is the default.
27249 @opindex muser-mode
27250 Generate code for the user mode, where the access to some general registers
27251 is forbidden: on the GR5, registers r24 to r31 cannot be accessed in this
27252 mode; on the GR6, only registers r29 to r31 are affected.
27256 @subsection VMS Options
27258 These @samp{-m} options are defined for the VMS implementations:
27261 @item -mvms-return-codes
27262 @opindex mvms-return-codes
27263 Return VMS condition codes from @code{main}. The default is to return POSIX-style
27264 condition (e.g.@: error) codes.
27266 @item -mdebug-main=@var{prefix}
27267 @opindex mdebug-main=@var{prefix}
27268 Flag the first routine whose name starts with @var{prefix} as the main
27269 routine for the debugger.
27273 Default to 64-bit memory allocation routines.
27275 @item -mpointer-size=@var{size}
27276 @opindex mpointer-size=@var{size}
27277 Set the default size of pointers. Possible options for @var{size} are
27278 @samp{32} or @samp{short} for 32 bit pointers, @samp{64} or @samp{long}
27279 for 64 bit pointers, and @samp{no} for supporting only 32 bit pointers.
27280 The later option disables @code{pragma pointer_size}.
27283 @node VxWorks Options
27284 @subsection VxWorks Options
27285 @cindex VxWorks Options
27287 The options in this section are defined for all VxWorks targets.
27288 Options specific to the target hardware are listed with the other
27289 options for that target.
27294 GCC can generate code for both VxWorks kernels and real time processes
27295 (RTPs). This option switches from the former to the latter. It also
27296 defines the preprocessor macro @code{__RTP__}.
27299 @opindex non-static
27300 Link an RTP executable against shared libraries rather than static
27301 libraries. The options @option{-static} and @option{-shared} can
27302 also be used for RTPs (@pxref{Link Options}); @option{-static}
27309 These options are passed down to the linker. They are defined for
27310 compatibility with Diab.
27313 @opindex Xbind-lazy
27314 Enable lazy binding of function calls. This option is equivalent to
27315 @option{-Wl,-z,now} and is defined for compatibility with Diab.
27319 Disable lazy binding of function calls. This option is the default and
27320 is defined for compatibility with Diab.
27324 @subsection x86 Options
27325 @cindex x86 Options
27327 These @samp{-m} options are defined for the x86 family of computers.
27331 @item -march=@var{cpu-type}
27333 Generate instructions for the machine type @var{cpu-type}. In contrast to
27334 @option{-mtune=@var{cpu-type}}, which merely tunes the generated code
27335 for the specified @var{cpu-type}, @option{-march=@var{cpu-type}} allows GCC
27336 to generate code that may not run at all on processors other than the one
27337 indicated. Specifying @option{-march=@var{cpu-type}} implies
27338 @option{-mtune=@var{cpu-type}}.
27340 The choices for @var{cpu-type} are:
27344 This selects the CPU to generate code for at compilation time by determining
27345 the processor type of the compiling machine. Using @option{-march=native}
27346 enables all instruction subsets supported by the local machine (hence
27347 the result might not run on different machines). Using @option{-mtune=native}
27348 produces code optimized for the local machine under the constraints
27349 of the selected instruction set.
27352 A generic CPU with 64-bit extensions.
27355 Original Intel i386 CPU@.
27358 Intel i486 CPU@. (No scheduling is implemented for this chip.)
27362 Intel Pentium CPU with no MMX support.
27365 Intel Lakemont MCU, based on Intel Pentium CPU.
27368 Intel Pentium MMX CPU, based on Pentium core with MMX instruction set support.
27371 Intel Pentium Pro CPU@.
27374 When used with @option{-march}, the Pentium Pro
27375 instruction set is used, so the code runs on all i686 family chips.
27376 When used with @option{-mtune}, it has the same meaning as @samp{generic}.
27379 Intel Pentium II CPU, based on Pentium Pro core with MMX instruction set
27384 Intel Pentium III CPU, based on Pentium Pro core with MMX and SSE instruction
27388 Intel Pentium M; low-power version of Intel Pentium III CPU
27389 with MMX, SSE and SSE2 instruction set support. Used by Centrino notebooks.
27393 Intel Pentium 4 CPU with MMX, SSE and SSE2 instruction set support.
27396 Improved version of Intel Pentium 4 CPU with MMX, SSE, SSE2 and SSE3 instruction
27400 Improved version of Intel Pentium 4 CPU with 64-bit extensions, MMX, SSE,
27401 SSE2 and SSE3 instruction set support.
27404 Intel Core 2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
27405 instruction set support.
27408 Intel Nehalem CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
27409 SSE4.1, SSE4.2 and POPCNT instruction set support.
27412 Intel Westmere CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
27413 SSE4.1, SSE4.2, POPCNT, AES and PCLMUL instruction set support.
27416 Intel Sandy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
27417 SSE4.1, SSE4.2, POPCNT, AVX, AES and PCLMUL instruction set support.
27420 Intel Ivy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
27421 SSE4.1, SSE4.2, POPCNT, AVX, AES, PCLMUL, FSGSBASE, RDRND and F16C
27422 instruction set support.
27425 Intel Haswell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
27426 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
27427 BMI, BMI2 and F16C instruction set support.
27430 Intel Broadwell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
27431 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
27432 BMI, BMI2, F16C, RDSEED, ADCX and PREFETCHW instruction set support.
27435 Intel Skylake CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
27436 SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
27437 BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC and
27438 XSAVES instruction set support.
27441 Intel Bonnell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3 and SSSE3
27442 instruction set support.
27445 Intel Silvermont CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
27446 SSE4.1, SSE4.2, POPCNT, AES, PCLMUL and RDRND instruction set support.
27449 Intel Goldmont CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
27450 SSE4.1, SSE4.2, POPCNT, AES, PCLMUL, RDRND, XSAVE, XSAVEOPT and FSGSBASE
27451 instruction set support.
27453 @item goldmont-plus
27454 Intel Goldmont Plus CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
27455 SSSE3, SSE4.1, SSE4.2, POPCNT, AES, PCLMUL, RDRND, XSAVE, XSAVEOPT, FSGSBASE,
27456 PTWRITE, RDPID, SGX and UMIP instruction set support.
27459 Intel Tremont CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
27460 SSE4.1, SSE4.2, POPCNT, AES, PCLMUL, RDRND, XSAVE, XSAVEOPT, FSGSBASE, PTWRITE,
27461 RDPID, SGX, UMIP, GFNI-SSE, CLWB and ENCLV instruction set support.
27464 Intel Knight's Landing CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
27465 SSSE3, SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
27466 BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, AVX512F, AVX512PF, AVX512ER and
27467 AVX512CD instruction set support.
27470 Intel Knights Mill CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
27471 SSSE3, SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
27472 BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, AVX512F, AVX512PF, AVX512ER, AVX512CD,
27473 AVX5124VNNIW, AVX5124FMAPS and AVX512VPOPCNTDQ instruction set support.
27475 @item skylake-avx512
27476 Intel Skylake Server CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
27477 SSSE3, SSE4.1, SSE4.2, POPCNT, PKU, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
27478 BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC, XSAVES, AVX512F,
27479 CLWB, AVX512VL, AVX512BW, AVX512DQ and AVX512CD instruction set support.
27482 Intel Cannonlake Server CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2,
27483 SSE3, SSSE3, SSE4.1, SSE4.2, POPCNT, PKU, AVX, AVX2, AES, PCLMUL, FSGSBASE,
27484 RDRND, FMA, BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC,
27485 XSAVES, AVX512F, AVX512VL, AVX512BW, AVX512DQ, AVX512CD, AVX512VBMI,
27486 AVX512IFMA, SHA and UMIP instruction set support.
27488 @item icelake-client
27489 Intel Icelake Client CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2,
27490 SSE3, SSSE3, SSE4.1, SSE4.2, POPCNT, PKU, AVX, AVX2, AES, PCLMUL, FSGSBASE,
27491 RDRND, FMA, BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC,
27492 XSAVES, AVX512F, AVX512VL, AVX512BW, AVX512DQ, AVX512CD, AVX512VBMI,
27493 AVX512IFMA, SHA, CLWB, UMIP, RDPID, GFNI, AVX512VBMI2, AVX512VPOPCNTDQ,
27494 AVX512BITALG, AVX512VNNI, VPCLMULQDQ, VAES instruction set support.
27496 @item icelake-server
27497 Intel Icelake Server CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2,
27498 SSE3, SSSE3, SSE4.1, SSE4.2, POPCNT, PKU, AVX, AVX2, AES, PCLMUL, FSGSBASE,
27499 RDRND, FMA, BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC,
27500 XSAVES, AVX512F, AVX512VL, AVX512BW, AVX512DQ, AVX512CD, AVX512VBMI,
27501 AVX512IFMA, SHA, CLWB, UMIP, RDPID, GFNI, AVX512VBMI2, AVX512VPOPCNTDQ,
27502 AVX512BITALG, AVX512VNNI, VPCLMULQDQ, VAES, PCONFIG and WBNOINVD instruction
27506 Intel Cascadelake CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
27507 SSE4.1, SSE4.2, POPCNT, PKU, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA, BMI,
27508 BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC, XSAVES, AVX512F, CLWB,
27509 AVX512VL, AVX512BW, AVX512DQ, AVX512CD and AVX512VNNI instruction set support.
27512 AMD K6 CPU with MMX instruction set support.
27516 Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
27519 @itemx athlon-tbird
27520 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
27526 Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
27527 instruction set support.
27533 Processors based on the AMD K8 core with x86-64 instruction set support,
27534 including the AMD Opteron, Athlon 64, and Athlon 64 FX processors.
27535 (This supersets MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit
27536 instruction set extensions.)
27539 @itemx opteron-sse3
27540 @itemx athlon64-sse3
27541 Improved versions of AMD K8 cores with SSE3 instruction set support.
27545 CPUs based on AMD Family 10h cores with x86-64 instruction set support. (This
27546 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
27547 instruction set extensions.)
27550 CPUs based on AMD Family 15h cores with x86-64 instruction set support. (This
27551 supersets FMA4, AVX, XOP, LWP, AES, PCLMUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A,
27552 SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set extensions.)
27554 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
27555 supersets BMI, TBM, F16C, FMA, FMA4, AVX, XOP, LWP, AES, PCLMUL, CX16, MMX,
27556 SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set
27559 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
27560 supersets BMI, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, XOP, LWP, AES,
27561 PCLMUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and
27562 64-bit instruction set extensions.
27564 AMD Family 15h core based CPUs with x86-64 instruction set support. (This
27565 supersets BMI, BMI2, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, AVX2, XOP, LWP,
27566 AES, PCLMUL, CX16, MOVBE, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1,
27567 SSE4.2, ABM and 64-bit instruction set extensions.
27570 AMD Family 17h core based CPUs with x86-64 instruction set support. (This
27571 supersets BMI, BMI2, F16C, FMA, FSGSBASE, AVX, AVX2, ADCX, RDSEED, MWAITX,
27572 SHA, CLZERO, AES, PCLMUL, CX16, MOVBE, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3,
27573 SSE4.1, SSE4.2, ABM, XSAVEC, XSAVES, CLFLUSHOPT, POPCNT, and 64-bit
27574 instruction set extensions.
27576 AMD Family 17h core based CPUs with x86-64 instruction set support. (This
27577 supersets BMI, BMI2, ,CLWB, F16C, FMA, FSGSBASE, AVX, AVX2, ADCX, RDSEED,
27578 MWAITX, SHA, CLZERO, AES, PCLMUL, CX16, MOVBE, MMX, SSE, SSE2, SSE3, SSE4A,
27579 SSSE3, SSE4.1, SSE4.2, ABM, XSAVEC, XSAVES, CLFLUSHOPT, POPCNT, and 64-bit
27580 instruction set extensions.)
27584 CPUs based on AMD Family 14h cores with x86-64 instruction set support. (This
27585 supersets MMX, SSE, SSE2, SSE3, SSSE3, SSE4A, CX16, ABM and 64-bit
27586 instruction set extensions.)
27589 CPUs based on AMD Family 16h cores with x86-64 instruction set support. This
27590 includes MOVBE, F16C, BMI, AVX, PCLMUL, AES, SSE4.2, SSE4.1, CX16, ABM,
27591 SSE4A, SSSE3, SSE3, SSE2, SSE, MMX and 64-bit instruction set extensions.
27594 IDT WinChip C6 CPU, dealt in same way as i486 with additional MMX instruction
27598 IDT WinChip 2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
27599 instruction set support.
27602 VIA C3 CPU with MMX and 3DNow!@: instruction set support.
27603 (No scheduling is implemented for this chip.)
27606 VIA C3-2 (Nehemiah/C5XL) CPU with MMX and SSE instruction set support.
27607 (No scheduling is implemented for this chip.)
27610 VIA C7 (Esther) CPU with MMX, SSE, SSE2 and SSE3 instruction set support.
27611 (No scheduling is implemented for this chip.)
27614 VIA Eden Samuel 2 CPU with MMX and 3DNow!@: instruction set support.
27615 (No scheduling is implemented for this chip.)
27618 VIA Eden Nehemiah CPU with MMX and SSE instruction set support.
27619 (No scheduling is implemented for this chip.)
27622 VIA Eden Esther CPU with MMX, SSE, SSE2 and SSE3 instruction set support.
27623 (No scheduling is implemented for this chip.)
27626 VIA Eden X2 CPU with x86-64, MMX, SSE, SSE2 and SSE3 instruction set support.
27627 (No scheduling is implemented for this chip.)
27630 VIA Eden X4 CPU with x86-64, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2,
27631 AVX and AVX2 instruction set support.
27632 (No scheduling is implemented for this chip.)
27635 Generic VIA Nano CPU with x86-64, MMX, SSE, SSE2, SSE3 and SSSE3
27636 instruction set support.
27637 (No scheduling is implemented for this chip.)
27640 VIA Nano 1xxx CPU with x86-64, MMX, SSE, SSE2, SSE3 and SSSE3
27641 instruction set support.
27642 (No scheduling is implemented for this chip.)
27645 VIA Nano 2xxx CPU with x86-64, MMX, SSE, SSE2, SSE3 and SSSE3
27646 instruction set support.
27647 (No scheduling is implemented for this chip.)
27650 VIA Nano 3xxx CPU with x86-64, MMX, SSE, SSE2, SSE3, SSSE3 and SSE4.1
27651 instruction set support.
27652 (No scheduling is implemented for this chip.)
27655 VIA Nano Dual Core CPU with x86-64, MMX, SSE, SSE2, SSE3, SSSE3 and SSE4.1
27656 instruction set support.
27657 (No scheduling is implemented for this chip.)
27660 VIA Nano Quad Core CPU with x86-64, MMX, SSE, SSE2, SSE3, SSSE3 and SSE4.1
27661 instruction set support.
27662 (No scheduling is implemented for this chip.)
27665 AMD Geode embedded processor with MMX and 3DNow!@: instruction set support.
27668 @item -mtune=@var{cpu-type}
27670 Tune to @var{cpu-type} everything applicable about the generated code, except
27671 for the ABI and the set of available instructions.
27672 While picking a specific @var{cpu-type} schedules things appropriately
27673 for that particular chip, the compiler does not generate any code that
27674 cannot run on the default machine type unless you use a
27675 @option{-march=@var{cpu-type}} option.
27676 For example, if GCC is configured for i686-pc-linux-gnu
27677 then @option{-mtune=pentium4} generates code that is tuned for Pentium 4
27678 but still runs on i686 machines.
27680 The choices for @var{cpu-type} are the same as for @option{-march}.
27681 In addition, @option{-mtune} supports 2 extra choices for @var{cpu-type}:
27685 Produce code optimized for the most common IA32/@/AMD64/@/EM64T processors.
27686 If you know the CPU on which your code will run, then you should use
27687 the corresponding @option{-mtune} or @option{-march} option instead of
27688 @option{-mtune=generic}. But, if you do not know exactly what CPU users
27689 of your application will have, then you should use this option.
27691 As new processors are deployed in the marketplace, the behavior of this
27692 option will change. Therefore, if you upgrade to a newer version of
27693 GCC, code generation controlled by this option will change to reflect
27695 that are most common at the time that version of GCC is released.
27697 There is no @option{-march=generic} option because @option{-march}
27698 indicates the instruction set the compiler can use, and there is no
27699 generic instruction set applicable to all processors. In contrast,
27700 @option{-mtune} indicates the processor (or, in this case, collection of
27701 processors) for which the code is optimized.
27704 Produce code optimized for the most current Intel processors, which are
27705 Haswell and Silvermont for this version of GCC. If you know the CPU
27706 on which your code will run, then you should use the corresponding
27707 @option{-mtune} or @option{-march} option instead of @option{-mtune=intel}.
27708 But, if you want your application performs better on both Haswell and
27709 Silvermont, then you should use this option.
27711 As new Intel processors are deployed in the marketplace, the behavior of
27712 this option will change. Therefore, if you upgrade to a newer version of
27713 GCC, code generation controlled by this option will change to reflect
27714 the most current Intel processors at the time that version of GCC is
27717 There is no @option{-march=intel} option because @option{-march} indicates
27718 the instruction set the compiler can use, and there is no common
27719 instruction set applicable to all processors. In contrast,
27720 @option{-mtune} indicates the processor (or, in this case, collection of
27721 processors) for which the code is optimized.
27724 @item -mcpu=@var{cpu-type}
27726 A deprecated synonym for @option{-mtune}.
27728 @item -mfpmath=@var{unit}
27730 Generate floating-point arithmetic for selected unit @var{unit}. The choices
27731 for @var{unit} are:
27735 Use the standard 387 floating-point coprocessor present on the majority of chips and
27736 emulated otherwise. Code compiled with this option runs almost everywhere.
27737 The temporary results are computed in 80-bit precision instead of the precision
27738 specified by the type, resulting in slightly different results compared to most
27739 of other chips. See @option{-ffloat-store} for more detailed description.
27741 This is the default choice for non-Darwin x86-32 targets.
27744 Use scalar floating-point instructions present in the SSE instruction set.
27745 This instruction set is supported by Pentium III and newer chips,
27746 and in the AMD line
27747 by Athlon-4, Athlon XP and Athlon MP chips. The earlier version of the SSE
27748 instruction set supports only single-precision arithmetic, thus the double and
27749 extended-precision arithmetic are still done using 387. A later version, present
27750 only in Pentium 4 and AMD x86-64 chips, supports double-precision
27753 For the x86-32 compiler, you must use @option{-march=@var{cpu-type}}, @option{-msse}
27754 or @option{-msse2} switches to enable SSE extensions and make this option
27755 effective. For the x86-64 compiler, these extensions are enabled by default.
27757 The resulting code should be considerably faster in the majority of cases and avoid
27758 the numerical instability problems of 387 code, but may break some existing
27759 code that expects temporaries to be 80 bits.
27761 This is the default choice for the x86-64 compiler, Darwin x86-32 targets,
27762 and the default choice for x86-32 targets with the SSE2 instruction set
27763 when @option{-ffast-math} is enabled.
27768 Attempt to utilize both instruction sets at once. This effectively doubles the
27769 amount of available registers, and on chips with separate execution units for
27770 387 and SSE the execution resources too. Use this option with care, as it is
27771 still experimental, because the GCC register allocator does not model separate
27772 functional units well, resulting in unstable performance.
27775 @item -masm=@var{dialect}
27776 @opindex masm=@var{dialect}
27777 Output assembly instructions using selected @var{dialect}. Also affects
27778 which dialect is used for basic @code{asm} (@pxref{Basic Asm}) and
27779 extended @code{asm} (@pxref{Extended Asm}). Supported choices (in dialect
27780 order) are @samp{att} or @samp{intel}. The default is @samp{att}. Darwin does
27781 not support @samp{intel}.
27784 @itemx -mno-ieee-fp
27786 @opindex mno-ieee-fp
27787 Control whether or not the compiler uses IEEE floating-point
27788 comparisons. These correctly handle the case where the result of a
27789 comparison is unordered.
27792 @itemx -mhard-float
27794 @opindex mhard-float
27795 Generate output containing 80387 instructions for floating point.
27798 @itemx -msoft-float
27800 @opindex msoft-float
27801 Generate output containing library calls for floating point.
27803 @strong{Warning:} the requisite libraries are not part of GCC@.
27804 Normally the facilities of the machine's usual C compiler are used, but
27805 this cannot be done directly in cross-compilation. You must make your
27806 own arrangements to provide suitable library functions for
27809 On machines where a function returns floating-point results in the 80387
27810 register stack, some floating-point opcodes may be emitted even if
27811 @option{-msoft-float} is used.
27813 @item -mno-fp-ret-in-387
27814 @opindex mno-fp-ret-in-387
27815 @opindex mfp-ret-in-387
27816 Do not use the FPU registers for return values of functions.
27818 The usual calling convention has functions return values of types
27819 @code{float} and @code{double} in an FPU register, even if there
27820 is no FPU@. The idea is that the operating system should emulate
27823 The option @option{-mno-fp-ret-in-387} causes such values to be returned
27824 in ordinary CPU registers instead.
27826 @item -mno-fancy-math-387
27827 @opindex mno-fancy-math-387
27828 @opindex mfancy-math-387
27829 Some 387 emulators do not support the @code{sin}, @code{cos} and
27830 @code{sqrt} instructions for the 387. Specify this option to avoid
27831 generating those instructions.
27832 This option is overridden when @option{-march}
27833 indicates that the target CPU always has an FPU and so the
27834 instruction does not need emulation. These
27835 instructions are not generated unless you also use the
27836 @option{-funsafe-math-optimizations} switch.
27838 @item -malign-double
27839 @itemx -mno-align-double
27840 @opindex malign-double
27841 @opindex mno-align-double
27842 Control whether GCC aligns @code{double}, @code{long double}, and
27843 @code{long long} variables on a two-word boundary or a one-word
27844 boundary. Aligning @code{double} variables on a two-word boundary
27845 produces code that runs somewhat faster on a Pentium at the
27846 expense of more memory.
27848 On x86-64, @option{-malign-double} is enabled by default.
27850 @strong{Warning:} if you use the @option{-malign-double} switch,
27851 structures containing the above types are aligned differently than
27852 the published application binary interface specifications for the x86-32
27853 and are not binary compatible with structures in code compiled
27854 without that switch.
27856 @item -m96bit-long-double
27857 @itemx -m128bit-long-double
27858 @opindex m96bit-long-double
27859 @opindex m128bit-long-double
27860 These switches control the size of @code{long double} type. The x86-32
27861 application binary interface specifies the size to be 96 bits,
27862 so @option{-m96bit-long-double} is the default in 32-bit mode.
27864 Modern architectures (Pentium and newer) prefer @code{long double}
27865 to be aligned to an 8- or 16-byte boundary. In arrays or structures
27866 conforming to the ABI, this is not possible. So specifying
27867 @option{-m128bit-long-double} aligns @code{long double}
27868 to a 16-byte boundary by padding the @code{long double} with an additional
27871 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
27872 its ABI specifies that @code{long double} is aligned on 16-byte boundary.
27874 Notice that neither of these options enable any extra precision over the x87
27875 standard of 80 bits for a @code{long double}.
27877 @strong{Warning:} if you override the default value for your target ABI, this
27878 changes the size of
27879 structures and arrays containing @code{long double} variables,
27880 as well as modifying the function calling convention for functions taking
27881 @code{long double}. Hence they are not binary-compatible
27882 with code compiled without that switch.
27884 @item -mlong-double-64
27885 @itemx -mlong-double-80
27886 @itemx -mlong-double-128
27887 @opindex mlong-double-64
27888 @opindex mlong-double-80
27889 @opindex mlong-double-128
27890 These switches control the size of @code{long double} type. A size
27891 of 64 bits makes the @code{long double} type equivalent to the @code{double}
27892 type. This is the default for 32-bit Bionic C library. A size
27893 of 128 bits makes the @code{long double} type equivalent to the
27894 @code{__float128} type. This is the default for 64-bit Bionic C library.
27896 @strong{Warning:} if you override the default value for your target ABI, this
27897 changes the size of
27898 structures and arrays containing @code{long double} variables,
27899 as well as modifying the function calling convention for functions taking
27900 @code{long double}. Hence they are not binary-compatible
27901 with code compiled without that switch.
27903 @item -malign-data=@var{type}
27904 @opindex malign-data
27905 Control how GCC aligns variables. Supported values for @var{type} are
27906 @samp{compat} uses increased alignment value compatible uses GCC 4.8
27907 and earlier, @samp{abi} uses alignment value as specified by the
27908 psABI, and @samp{cacheline} uses increased alignment value to match
27909 the cache line size. @samp{compat} is the default.
27911 @item -mlarge-data-threshold=@var{threshold}
27912 @opindex mlarge-data-threshold
27913 When @option{-mcmodel=medium} is specified, data objects larger than
27914 @var{threshold} are placed in the large data section. This value must be the
27915 same across all objects linked into the binary, and defaults to 65535.
27919 Use a different function-calling convention, in which functions that
27920 take a fixed number of arguments return with the @code{ret @var{num}}
27921 instruction, which pops their arguments while returning. This saves one
27922 instruction in the caller since there is no need to pop the arguments
27925 You can specify that an individual function is called with this calling
27926 sequence with the function attribute @code{stdcall}. You can also
27927 override the @option{-mrtd} option by using the function attribute
27928 @code{cdecl}. @xref{Function Attributes}.
27930 @strong{Warning:} this calling convention is incompatible with the one
27931 normally used on Unix, so you cannot use it if you need to call
27932 libraries compiled with the Unix compiler.
27934 Also, you must provide function prototypes for all functions that
27935 take variable numbers of arguments (including @code{printf});
27936 otherwise incorrect code is generated for calls to those
27939 In addition, seriously incorrect code results if you call a
27940 function with too many arguments. (Normally, extra arguments are
27941 harmlessly ignored.)
27943 @item -mregparm=@var{num}
27945 Control how many registers are used to pass integer arguments. By
27946 default, no registers are used to pass arguments, and at most 3
27947 registers can be used. You can control this behavior for a specific
27948 function by using the function attribute @code{regparm}.
27949 @xref{Function Attributes}.
27951 @strong{Warning:} if you use this switch, and
27952 @var{num} is nonzero, then you must build all modules with the same
27953 value, including any libraries. This includes the system libraries and
27957 @opindex msseregparm
27958 Use SSE register passing conventions for float and double arguments
27959 and return values. You can control this behavior for a specific
27960 function by using the function attribute @code{sseregparm}.
27961 @xref{Function Attributes}.
27963 @strong{Warning:} if you use this switch then you must build all
27964 modules with the same value, including any libraries. This includes
27965 the system libraries and startup modules.
27967 @item -mvect8-ret-in-mem
27968 @opindex mvect8-ret-in-mem
27969 Return 8-byte vectors in memory instead of MMX registers. This is the
27970 default on VxWorks to match the ABI of the Sun Studio compilers until
27971 version 12. @emph{Only} use this option if you need to remain
27972 compatible with existing code produced by those previous compiler
27973 versions or older versions of GCC@.
27982 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
27983 is specified, the significands of results of floating-point operations are
27984 rounded to 24 bits (single precision); @option{-mpc64} rounds the
27985 significands of results of floating-point operations to 53 bits (double
27986 precision) and @option{-mpc80} rounds the significands of results of
27987 floating-point operations to 64 bits (extended double precision), which is
27988 the default. When this option is used, floating-point operations in higher
27989 precisions are not available to the programmer without setting the FPU
27990 control word explicitly.
27992 Setting the rounding of floating-point operations to less than the default
27993 80 bits can speed some programs by 2% or more. Note that some mathematical
27994 libraries assume that extended-precision (80-bit) floating-point operations
27995 are enabled by default; routines in such libraries could suffer significant
27996 loss of accuracy, typically through so-called ``catastrophic cancellation'',
27997 when this option is used to set the precision to less than extended precision.
27999 @item -mstackrealign
28000 @opindex mstackrealign
28001 Realign the stack at entry. On the x86, the @option{-mstackrealign}
28002 option generates an alternate prologue and epilogue that realigns the
28003 run-time stack if necessary. This supports mixing legacy codes that keep
28004 4-byte stack alignment with modern codes that keep 16-byte stack alignment for
28005 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
28006 applicable to individual functions.
28008 @item -mpreferred-stack-boundary=@var{num}
28009 @opindex mpreferred-stack-boundary
28010 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
28011 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
28012 the default is 4 (16 bytes or 128 bits).
28014 @strong{Warning:} When generating code for the x86-64 architecture with
28015 SSE extensions disabled, @option{-mpreferred-stack-boundary=3} can be
28016 used to keep the stack boundary aligned to 8 byte boundary. Since
28017 x86-64 ABI require 16 byte stack alignment, this is ABI incompatible and
28018 intended to be used in controlled environment where stack space is
28019 important limitation. This option leads to wrong code when functions
28020 compiled with 16 byte stack alignment (such as functions from a standard
28021 library) are called with misaligned stack. In this case, SSE
28022 instructions may lead to misaligned memory access traps. In addition,
28023 variable arguments are handled incorrectly for 16 byte aligned
28024 objects (including x87 long double and __int128), leading to wrong
28025 results. You must build all modules with
28026 @option{-mpreferred-stack-boundary=3}, including any libraries. This
28027 includes the system libraries and startup modules.
28029 @item -mincoming-stack-boundary=@var{num}
28030 @opindex mincoming-stack-boundary
28031 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
28032 boundary. If @option{-mincoming-stack-boundary} is not specified,
28033 the one specified by @option{-mpreferred-stack-boundary} is used.
28035 On Pentium and Pentium Pro, @code{double} and @code{long double} values
28036 should be aligned to an 8-byte boundary (see @option{-malign-double}) or
28037 suffer significant run time performance penalties. On Pentium III, the
28038 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
28039 properly if it is not 16-byte aligned.
28041 To ensure proper alignment of this values on the stack, the stack boundary
28042 must be as aligned as that required by any value stored on the stack.
28043 Further, every function must be generated such that it keeps the stack
28044 aligned. Thus calling a function compiled with a higher preferred
28045 stack boundary from a function compiled with a lower preferred stack
28046 boundary most likely misaligns the stack. It is recommended that
28047 libraries that use callbacks always use the default setting.
28049 This extra alignment does consume extra stack space, and generally
28050 increases code size. Code that is sensitive to stack space usage, such
28051 as embedded systems and operating system kernels, may want to reduce the
28052 preferred alignment to @option{-mpreferred-stack-boundary=2}.
28109 @itemx -mavx512ifma
28110 @opindex mavx512ifma
28112 @itemx -mavx512vbmi
28113 @opindex mavx512vbmi
28124 @itemx -mclflushopt
28125 @opindex mclflushopt
28160 @itemx -mprefetchwt1
28161 @opindex mprefetchwt1
28232 @itemx -mavx512vbmi2
28233 @opindex mavx512vbmi2
28235 @itemx -mavx512bf16
28236 @opindex mavx512bf16
28247 @itemx -mvpclmulqdq
28248 @opindex mvpclmulqdq
28250 @itemx -mavx512bitalg
28251 @opindex mavx512bitalg
28257 @opindex mmovdir64b
28262 @itemx -mavx512vpopcntdq
28263 @opindex mavx512vpopcntdq
28265 @itemx -mavx512vp2intersect
28266 @opindex mavx512vp2intersect
28268 @itemx -mavx5124fmaps
28269 @opindex mavx5124fmaps
28271 @itemx -mavx512vnni
28272 @opindex mavx512vnni
28274 @itemx -mavx5124vnniw
28275 @opindex mavx5124vnniw
28279 These switches enable the use of instructions in the MMX, SSE,
28280 SSE2, SSE3, SSSE3, SSE4, SSE4A, SSE4.1, SSE4.2, AVX, AVX2, AVX512F, AVX512PF,
28281 AVX512ER, AVX512CD, AVX512VL, AVX512BW, AVX512DQ, AVX512IFMA, AVX512VBMI, SHA,
28282 AES, PCLMUL, CLFLUSHOPT, CLWB, FSGSBASE, PTWRITE, RDRND, F16C, FMA, PCONFIG,
28283 WBNOINVD, FMA4, PREFETCHW, RDPID, PREFETCHWT1, RDSEED, SGX, XOP, LWP,
28284 3DNow!@:, enhanced 3DNow!@:, POPCNT, ABM, ADX, BMI, BMI2, LZCNT, FXSR, XSAVE,
28285 XSAVEOPT, XSAVEC, XSAVES, RTM, HLE, TBM, MWAITX, CLZERO, PKU, AVX512VBMI2,
28286 GFNI, VAES, WAITPKG, VPCLMULQDQ, AVX512BITALG, MOVDIRI, MOVDIR64B, AVX512BF16,
28287 ENQCMD, AVX512VPOPCNTDQ, AVX5124FMAPS, AVX512VNNI, AVX5124VNNIW, or CLDEMOTE
28288 extended instruction sets. Each has a corresponding @option{-mno-} option to
28289 disable use of these instructions.
28291 These extensions are also available as built-in functions: see
28292 @ref{x86 Built-in Functions}, for details of the functions enabled and
28293 disabled by these switches.
28295 To generate SSE/SSE2 instructions automatically from floating-point
28296 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
28298 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
28299 generates new AVX instructions or AVX equivalence for all SSEx instructions
28302 These options enable GCC to use these extended instructions in
28303 generated code, even without @option{-mfpmath=sse}. Applications that
28304 perform run-time CPU detection must compile separate files for each
28305 supported architecture, using the appropriate flags. In particular,
28306 the file containing the CPU detection code should be compiled without
28309 @item -mdump-tune-features
28310 @opindex mdump-tune-features
28311 This option instructs GCC to dump the names of the x86 performance
28312 tuning features and default settings. The names can be used in
28313 @option{-mtune-ctrl=@var{feature-list}}.
28315 @item -mtune-ctrl=@var{feature-list}
28316 @opindex mtune-ctrl=@var{feature-list}
28317 This option is used to do fine grain control of x86 code generation features.
28318 @var{feature-list} is a comma separated list of @var{feature} names. See also
28319 @option{-mdump-tune-features}. When specified, the @var{feature} is turned
28320 on if it is not preceded with @samp{^}, otherwise, it is turned off.
28321 @option{-mtune-ctrl=@var{feature-list}} is intended to be used by GCC
28322 developers. Using it may lead to code paths not covered by testing and can
28323 potentially result in compiler ICEs or runtime errors.
28326 @opindex mno-default
28327 This option instructs GCC to turn off all tunable features. See also
28328 @option{-mtune-ctrl=@var{feature-list}} and @option{-mdump-tune-features}.
28332 This option instructs GCC to emit a @code{cld} instruction in the prologue
28333 of functions that use string instructions. String instructions depend on
28334 the DF flag to select between autoincrement or autodecrement mode. While the
28335 ABI specifies the DF flag to be cleared on function entry, some operating
28336 systems violate this specification by not clearing the DF flag in their
28337 exception dispatchers. The exception handler can be invoked with the DF flag
28338 set, which leads to wrong direction mode when string instructions are used.
28339 This option can be enabled by default on 32-bit x86 targets by configuring
28340 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
28341 instructions can be suppressed with the @option{-mno-cld} compiler option
28345 @opindex mvzeroupper
28346 This option instructs GCC to emit a @code{vzeroupper} instruction
28347 before a transfer of control flow out of the function to minimize
28348 the AVX to SSE transition penalty as well as remove unnecessary @code{zeroupper}
28351 @item -mprefer-avx128
28352 @opindex mprefer-avx128
28353 This option instructs GCC to use 128-bit AVX instructions instead of
28354 256-bit AVX instructions in the auto-vectorizer.
28356 @item -mprefer-vector-width=@var{opt}
28357 @opindex mprefer-vector-width
28358 This option instructs GCC to use @var{opt}-bit vector width in instructions
28359 instead of default on the selected platform.
28363 No extra limitations applied to GCC other than defined by the selected platform.
28366 Prefer 128-bit vector width for instructions.
28369 Prefer 256-bit vector width for instructions.
28372 Prefer 512-bit vector width for instructions.
28377 This option enables GCC to generate @code{CMPXCHG16B} instructions in 64-bit
28378 code to implement compare-and-exchange operations on 16-byte aligned 128-bit
28379 objects. This is useful for atomic updates of data structures exceeding one
28380 machine word in size. The compiler uses this instruction to implement
28381 @ref{__sync Builtins}. However, for @ref{__atomic Builtins} operating on
28382 128-bit integers, a library call is always used.
28386 This option enables generation of @code{SAHF} instructions in 64-bit code.
28387 Early Intel Pentium 4 CPUs with Intel 64 support,
28388 prior to the introduction of Pentium 4 G1 step in December 2005,
28389 lacked the @code{LAHF} and @code{SAHF} instructions
28390 which are supported by AMD64.
28391 These are load and store instructions, respectively, for certain status flags.
28392 In 64-bit mode, the @code{SAHF} instruction is used to optimize @code{fmod},
28393 @code{drem}, and @code{remainder} built-in functions;
28394 see @ref{Other Builtins} for details.
28398 This option enables use of the @code{movbe} instruction to implement
28399 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
28403 The @option{-mshstk} option enables shadow stack built-in functions
28404 from x86 Control-flow Enforcement Technology (CET).
28408 This option enables built-in functions @code{__builtin_ia32_crc32qi},
28409 @code{__builtin_ia32_crc32hi}, @code{__builtin_ia32_crc32si} and
28410 @code{__builtin_ia32_crc32di} to generate the @code{crc32} machine instruction.
28414 This option enables use of @code{RCPSS} and @code{RSQRTSS} instructions
28415 (and their vectorized variants @code{RCPPS} and @code{RSQRTPS})
28416 with an additional Newton-Raphson step
28417 to increase precision instead of @code{DIVSS} and @code{SQRTSS}
28418 (and their vectorized
28419 variants) for single-precision floating-point arguments. These instructions
28420 are generated only when @option{-funsafe-math-optimizations} is enabled
28421 together with @option{-ffinite-math-only} and @option{-fno-trapping-math}.
28422 Note that while the throughput of the sequence is higher than the throughput
28423 of the non-reciprocal instruction, the precision of the sequence can be
28424 decreased by up to 2 ulp (i.e.@: the inverse of 1.0 equals 0.99999994).
28426 Note that GCC implements @code{1.0f/sqrtf(@var{x})} in terms of @code{RSQRTSS}
28427 (or @code{RSQRTPS}) already with @option{-ffast-math} (or the above option
28428 combination), and doesn't need @option{-mrecip}.
28430 Also note that GCC emits the above sequence with additional Newton-Raphson step
28431 for vectorized single-float division and vectorized @code{sqrtf(@var{x})}
28432 already with @option{-ffast-math} (or the above option combination), and
28433 doesn't need @option{-mrecip}.
28435 @item -mrecip=@var{opt}
28436 @opindex mrecip=opt
28437 This option controls which reciprocal estimate instructions
28438 may be used. @var{opt} is a comma-separated list of options, which may
28439 be preceded by a @samp{!} to invert the option:
28443 Enable all estimate instructions.
28446 Enable the default instructions, equivalent to @option{-mrecip}.
28449 Disable all estimate instructions, equivalent to @option{-mno-recip}.
28452 Enable the approximation for scalar division.
28455 Enable the approximation for vectorized division.
28458 Enable the approximation for scalar square root.
28461 Enable the approximation for vectorized square root.
28464 So, for example, @option{-mrecip=all,!sqrt} enables
28465 all of the reciprocal approximations, except for square root.
28467 @item -mveclibabi=@var{type}
28468 @opindex mveclibabi
28469 Specifies the ABI type to use for vectorizing intrinsics using an
28470 external library. Supported values for @var{type} are @samp{svml}
28471 for the Intel short
28472 vector math library and @samp{acml} for the AMD math core library.
28473 To use this option, both @option{-ftree-vectorize} and
28474 @option{-funsafe-math-optimizations} have to be enabled, and an SVML or ACML
28475 ABI-compatible library must be specified at link time.
28477 GCC currently emits calls to @code{vmldExp2},
28478 @code{vmldLn2}, @code{vmldLog102}, @code{vmldPow2},
28479 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
28480 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
28481 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
28482 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4},
28483 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
28484 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
28485 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
28486 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
28487 function type when @option{-mveclibabi=svml} is used, and @code{__vrd2_sin},
28488 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
28489 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
28490 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
28491 @code{__vrs4_log10f} and @code{__vrs4_powf} for the corresponding function type
28492 when @option{-mveclibabi=acml} is used.
28494 @item -mabi=@var{name}
28496 Generate code for the specified calling convention. Permissible values
28497 are @samp{sysv} for the ABI used on GNU/Linux and other systems, and
28498 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
28499 ABI when targeting Microsoft Windows and the SysV ABI on all other systems.
28500 You can control this behavior for specific functions by
28501 using the function attributes @code{ms_abi} and @code{sysv_abi}.
28502 @xref{Function Attributes}.
28504 @item -mforce-indirect-call
28505 @opindex mforce-indirect-call
28506 Force all calls to functions to be indirect. This is useful
28507 when using Intel Processor Trace where it generates more precise timing
28508 information for function calls.
28510 @item -mmanual-endbr
28511 @opindex mmanual-endbr
28512 Insert ENDBR instruction at function entry only via the @code{cf_check}
28513 function attribute. This is useful when used with the option
28514 @option{-fcf-protection=branch} to control ENDBR insertion at the
28517 @item -mcall-ms2sysv-xlogues
28518 @opindex mcall-ms2sysv-xlogues
28519 @opindex mno-call-ms2sysv-xlogues
28520 Due to differences in 64-bit ABIs, any Microsoft ABI function that calls a
28521 System V ABI function must consider RSI, RDI and XMM6-15 as clobbered. By
28522 default, the code for saving and restoring these registers is emitted inline,
28523 resulting in fairly lengthy prologues and epilogues. Using
28524 @option{-mcall-ms2sysv-xlogues} emits prologues and epilogues that
28525 use stubs in the static portion of libgcc to perform these saves and restores,
28526 thus reducing function size at the cost of a few extra instructions.
28528 @item -mtls-dialect=@var{type}
28529 @opindex mtls-dialect
28530 Generate code to access thread-local storage using the @samp{gnu} or
28531 @samp{gnu2} conventions. @samp{gnu} is the conservative default;
28532 @samp{gnu2} is more efficient, but it may add compile- and run-time
28533 requirements that cannot be satisfied on all systems.
28536 @itemx -mno-push-args
28537 @opindex mpush-args
28538 @opindex mno-push-args
28539 Use PUSH operations to store outgoing parameters. This method is shorter
28540 and usually equally fast as method using SUB/MOV operations and is enabled
28541 by default. In some cases disabling it may improve performance because of
28542 improved scheduling and reduced dependencies.
28544 @item -maccumulate-outgoing-args
28545 @opindex maccumulate-outgoing-args
28546 If enabled, the maximum amount of space required for outgoing arguments is
28547 computed in the function prologue. This is faster on most modern CPUs
28548 because of reduced dependencies, improved scheduling and reduced stack usage
28549 when the preferred stack boundary is not equal to 2. The drawback is a notable
28550 increase in code size. This switch implies @option{-mno-push-args}.
28554 Support thread-safe exception handling on MinGW. Programs that rely
28555 on thread-safe exception handling must compile and link all code with the
28556 @option{-mthreads} option. When compiling, @option{-mthreads} defines
28557 @option{-D_MT}; when linking, it links in a special thread helper library
28558 @option{-lmingwthrd} which cleans up per-thread exception-handling data.
28560 @item -mms-bitfields
28561 @itemx -mno-ms-bitfields
28562 @opindex mms-bitfields
28563 @opindex mno-ms-bitfields
28565 Enable/disable bit-field layout compatible with the native Microsoft
28568 If @code{packed} is used on a structure, or if bit-fields are used,
28569 it may be that the Microsoft ABI lays out the structure differently
28570 than the way GCC normally does. Particularly when moving packed
28571 data between functions compiled with GCC and the native Microsoft compiler
28572 (either via function call or as data in a file), it may be necessary to access
28575 This option is enabled by default for Microsoft Windows
28576 targets. This behavior can also be controlled locally by use of variable
28577 or type attributes. For more information, see @ref{x86 Variable Attributes}
28578 and @ref{x86 Type Attributes}.
28580 The Microsoft structure layout algorithm is fairly simple with the exception
28581 of the bit-field packing.
28582 The padding and alignment of members of structures and whether a bit-field
28583 can straddle a storage-unit boundary are determine by these rules:
28586 @item Structure members are stored sequentially in the order in which they are
28587 declared: the first member has the lowest memory address and the last member
28590 @item Every data object has an alignment requirement. The alignment requirement
28591 for all data except structures, unions, and arrays is either the size of the
28592 object or the current packing size (specified with either the
28593 @code{aligned} attribute or the @code{pack} pragma),
28594 whichever is less. For structures, unions, and arrays,
28595 the alignment requirement is the largest alignment requirement of its members.
28596 Every object is allocated an offset so that:
28599 offset % alignment_requirement == 0
28602 @item Adjacent bit-fields are packed into the same 1-, 2-, or 4-byte allocation
28603 unit if the integral types are the same size and if the next bit-field fits
28604 into the current allocation unit without crossing the boundary imposed by the
28605 common alignment requirements of the bit-fields.
28608 MSVC interprets zero-length bit-fields in the following ways:
28611 @item If a zero-length bit-field is inserted between two bit-fields that
28612 are normally coalesced, the bit-fields are not coalesced.
28619 unsigned long bf_1 : 12;
28621 unsigned long bf_2 : 12;
28626 The size of @code{t1} is 8 bytes with the zero-length bit-field. If the
28627 zero-length bit-field were removed, @code{t1}'s size would be 4 bytes.
28629 @item If a zero-length bit-field is inserted after a bit-field, @code{foo}, and the
28630 alignment of the zero-length bit-field is greater than the member that follows it,
28631 @code{bar}, @code{bar} is aligned as the type of the zero-length bit-field.
28652 For @code{t2}, @code{bar} is placed at offset 2, rather than offset 1.
28653 Accordingly, the size of @code{t2} is 4. For @code{t3}, the zero-length
28654 bit-field does not affect the alignment of @code{bar} or, as a result, the size
28657 Taking this into account, it is important to note the following:
28660 @item If a zero-length bit-field follows a normal bit-field, the type of the
28661 zero-length bit-field may affect the alignment of the structure as whole. For
28662 example, @code{t2} has a size of 4 bytes, since the zero-length bit-field follows a
28663 normal bit-field, and is of type short.
28665 @item Even if a zero-length bit-field is not followed by a normal bit-field, it may
28666 still affect the alignment of the structure:
28677 Here, @code{t4} takes up 4 bytes.
28680 @item Zero-length bit-fields following non-bit-field members are ignored:
28692 Here, @code{t5} takes up 2 bytes.
28696 @item -mno-align-stringops
28697 @opindex mno-align-stringops
28698 @opindex malign-stringops
28699 Do not align the destination of inlined string operations. This switch reduces
28700 code size and improves performance in case the destination is already aligned,
28701 but GCC doesn't know about it.
28703 @item -minline-all-stringops
28704 @opindex minline-all-stringops
28705 By default GCC inlines string operations only when the destination is
28706 known to be aligned to least a 4-byte boundary.
28707 This enables more inlining and increases code
28708 size, but may improve performance of code that depends on fast
28709 @code{memcpy} and @code{memset} for short lengths.
28710 The option enables inline expansion of @code{strlen} for all
28711 pointer alignments.
28713 @item -minline-stringops-dynamically
28714 @opindex minline-stringops-dynamically
28715 For string operations of unknown size, use run-time checks with
28716 inline code for small blocks and a library call for large blocks.
28718 @item -mstringop-strategy=@var{alg}
28719 @opindex mstringop-strategy=@var{alg}
28720 Override the internal decision heuristic for the particular algorithm to use
28721 for inlining string operations. The allowed values for @var{alg} are:
28727 Expand using i386 @code{rep} prefix of the specified size.
28731 @itemx unrolled_loop
28732 Expand into an inline loop.
28735 Always use a library call.
28738 @item -mmemcpy-strategy=@var{strategy}
28739 @opindex mmemcpy-strategy=@var{strategy}
28740 Override the internal decision heuristic to decide if @code{__builtin_memcpy}
28741 should be inlined and what inline algorithm to use when the expected size
28742 of the copy operation is known. @var{strategy}
28743 is a comma-separated list of @var{alg}:@var{max_size}:@var{dest_align} triplets.
28744 @var{alg} is specified in @option{-mstringop-strategy}, @var{max_size} specifies
28745 the max byte size with which inline algorithm @var{alg} is allowed. For the last
28746 triplet, the @var{max_size} must be @code{-1}. The @var{max_size} of the triplets
28747 in the list must be specified in increasing order. The minimal byte size for
28748 @var{alg} is @code{0} for the first triplet and @code{@var{max_size} + 1} of the
28751 @item -mmemset-strategy=@var{strategy}
28752 @opindex mmemset-strategy=@var{strategy}
28753 The option is similar to @option{-mmemcpy-strategy=} except that it is to control
28754 @code{__builtin_memset} expansion.
28756 @item -momit-leaf-frame-pointer
28757 @opindex momit-leaf-frame-pointer
28758 Don't keep the frame pointer in a register for leaf functions. This
28759 avoids the instructions to save, set up, and restore frame pointers and
28760 makes an extra register available in leaf functions. The option
28761 @option{-fomit-leaf-frame-pointer} removes the frame pointer for leaf functions,
28762 which might make debugging harder.
28764 @item -mtls-direct-seg-refs
28765 @itemx -mno-tls-direct-seg-refs
28766 @opindex mtls-direct-seg-refs
28767 Controls whether TLS variables may be accessed with offsets from the
28768 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
28769 or whether the thread base pointer must be added. Whether or not this
28770 is valid depends on the operating system, and whether it maps the
28771 segment to cover the entire TLS area.
28773 For systems that use the GNU C Library, the default is on.
28776 @itemx -mno-sse2avx
28778 Specify that the assembler should encode SSE instructions with VEX
28779 prefix. The option @option{-mavx} turns this on by default.
28784 If profiling is active (@option{-pg}), put the profiling
28785 counter call before the prologue.
28786 Note: On x86 architectures the attribute @code{ms_hook_prologue}
28787 isn't possible at the moment for @option{-mfentry} and @option{-pg}.
28789 @item -mrecord-mcount
28790 @itemx -mno-record-mcount
28791 @opindex mrecord-mcount
28792 If profiling is active (@option{-pg}), generate a __mcount_loc section
28793 that contains pointers to each profiling call. This is useful for
28794 automatically patching and out calls.
28797 @itemx -mno-nop-mcount
28798 @opindex mnop-mcount
28799 If profiling is active (@option{-pg}), generate the calls to
28800 the profiling functions as NOPs. This is useful when they
28801 should be patched in later dynamically. This is likely only
28802 useful together with @option{-mrecord-mcount}.
28804 @item -minstrument-return=@var{type}
28805 @opindex minstrument-return
28806 Instrument function exit in -pg -mfentry instrumented functions with
28807 call to specified function. This only instruments true returns ending
28808 with ret, but not sibling calls ending with jump. Valid types
28809 are @var{none} to not instrument, @var{call} to generate a call to __return__,
28810 or @var{nop5} to generate a 5 byte nop.
28812 @item -mrecord-return
28813 @itemx -mno-record-return
28814 @opindex mrecord-return
28815 Generate a __return_loc section pointing to all return instrumentation code.
28817 @item -mfentry-name=@var{name}
28818 @opindex mfentry-name
28819 Set name of __fentry__ symbol called at function entry for -pg -mfentry functions.
28821 @item -mfentry-section=@var{name}
28822 @opindex mfentry-section
28823 Set name of section to record -mrecord-mcount calls (default __mcount_loc).
28825 @item -mskip-rax-setup
28826 @itemx -mno-skip-rax-setup
28827 @opindex mskip-rax-setup
28828 When generating code for the x86-64 architecture with SSE extensions
28829 disabled, @option{-mskip-rax-setup} can be used to skip setting up RAX
28830 register when there are no variable arguments passed in vector registers.
28832 @strong{Warning:} Since RAX register is used to avoid unnecessarily
28833 saving vector registers on stack when passing variable arguments, the
28834 impacts of this option are callees may waste some stack space,
28835 misbehave or jump to a random location. GCC 4.4 or newer don't have
28836 those issues, regardless the RAX register value.
28839 @itemx -mno-8bit-idiv
28840 @opindex m8bit-idiv
28841 On some processors, like Intel Atom, 8-bit unsigned integer divide is
28842 much faster than 32-bit/64-bit integer divide. This option generates a
28843 run-time check. If both dividend and divisor are within range of 0
28844 to 255, 8-bit unsigned integer divide is used instead of
28845 32-bit/64-bit integer divide.
28847 @item -mavx256-split-unaligned-load
28848 @itemx -mavx256-split-unaligned-store
28849 @opindex mavx256-split-unaligned-load
28850 @opindex mavx256-split-unaligned-store
28851 Split 32-byte AVX unaligned load and store.
28853 @item -mstack-protector-guard=@var{guard}
28854 @itemx -mstack-protector-guard-reg=@var{reg}
28855 @itemx -mstack-protector-guard-offset=@var{offset}
28856 @opindex mstack-protector-guard
28857 @opindex mstack-protector-guard-reg
28858 @opindex mstack-protector-guard-offset
28859 Generate stack protection code using canary at @var{guard}. Supported
28860 locations are @samp{global} for global canary or @samp{tls} for per-thread
28861 canary in the TLS block (the default). This option has effect only when
28862 @option{-fstack-protector} or @option{-fstack-protector-all} is specified.
28864 With the latter choice the options
28865 @option{-mstack-protector-guard-reg=@var{reg}} and
28866 @option{-mstack-protector-guard-offset=@var{offset}} furthermore specify
28867 which segment register (@code{%fs} or @code{%gs}) to use as base register
28868 for reading the canary, and from what offset from that base register.
28869 The default for those is as specified in the relevant ABI.
28871 @item -mgeneral-regs-only
28872 @opindex mgeneral-regs-only
28873 Generate code that uses only the general-purpose registers. This
28874 prevents the compiler from using floating-point, vector, mask and bound
28877 @item -mindirect-branch=@var{choice}
28878 @opindex mindirect-branch
28879 Convert indirect call and jump with @var{choice}. The default is
28880 @samp{keep}, which keeps indirect call and jump unmodified.
28881 @samp{thunk} converts indirect call and jump to call and return thunk.
28882 @samp{thunk-inline} converts indirect call and jump to inlined call
28883 and return thunk. @samp{thunk-extern} converts indirect call and jump
28884 to external call and return thunk provided in a separate object file.
28885 You can control this behavior for a specific function by using the
28886 function attribute @code{indirect_branch}. @xref{Function Attributes}.
28888 Note that @option{-mcmodel=large} is incompatible with
28889 @option{-mindirect-branch=thunk} and
28890 @option{-mindirect-branch=thunk-extern} since the thunk function may
28891 not be reachable in the large code model.
28893 Note that @option{-mindirect-branch=thunk-extern} is incompatible with
28894 @option{-fcf-protection=branch} since the external thunk cannot be modified
28895 to disable control-flow check.
28897 @item -mfunction-return=@var{choice}
28898 @opindex mfunction-return
28899 Convert function return with @var{choice}. The default is @samp{keep},
28900 which keeps function return unmodified. @samp{thunk} converts function
28901 return to call and return thunk. @samp{thunk-inline} converts function
28902 return to inlined call and return thunk. @samp{thunk-extern} converts
28903 function return to external call and return thunk provided in a separate
28904 object file. You can control this behavior for a specific function by
28905 using the function attribute @code{function_return}.
28906 @xref{Function Attributes}.
28908 Note that @option{-mcmodel=large} is incompatible with
28909 @option{-mfunction-return=thunk} and
28910 @option{-mfunction-return=thunk-extern} since the thunk function may
28911 not be reachable in the large code model.
28914 @item -mindirect-branch-register
28915 @opindex mindirect-branch-register
28916 Force indirect call and jump via register.
28920 These @samp{-m} switches are supported in addition to the above
28921 on x86-64 processors in 64-bit environments.
28934 Generate code for a 16-bit, 32-bit or 64-bit environment.
28935 The @option{-m32} option sets @code{int}, @code{long}, and pointer types
28937 generates code that runs on any i386 system.
28939 The @option{-m64} option sets @code{int} to 32 bits and @code{long} and pointer
28940 types to 64 bits, and generates code for the x86-64 architecture.
28941 For Darwin only the @option{-m64} option also turns off the @option{-fno-pic}
28942 and @option{-mdynamic-no-pic} options.
28944 The @option{-mx32} option sets @code{int}, @code{long}, and pointer types
28946 generates code for the x86-64 architecture.
28948 The @option{-m16} option is the same as @option{-m32}, except for that
28949 it outputs the @code{.code16gcc} assembly directive at the beginning of
28950 the assembly output so that the binary can run in 16-bit mode.
28952 The @option{-miamcu} option generates code which conforms to Intel MCU
28953 psABI. It requires the @option{-m32} option to be turned on.
28955 @item -mno-red-zone
28956 @opindex mno-red-zone
28958 Do not use a so-called ``red zone'' for x86-64 code. The red zone is mandated
28959 by the x86-64 ABI; it is a 128-byte area beyond the location of the
28960 stack pointer that is not modified by signal or interrupt handlers
28961 and therefore can be used for temporary data without adjusting the stack
28962 pointer. The flag @option{-mno-red-zone} disables this red zone.
28964 @item -mcmodel=small
28965 @opindex mcmodel=small
28966 Generate code for the small code model: the program and its symbols must
28967 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
28968 Programs can be statically or dynamically linked. This is the default
28971 @item -mcmodel=kernel
28972 @opindex mcmodel=kernel
28973 Generate code for the kernel code model. The kernel runs in the
28974 negative 2 GB of the address space.
28975 This model has to be used for Linux kernel code.
28977 @item -mcmodel=medium
28978 @opindex mcmodel=medium
28979 Generate code for the medium model: the program is linked in the lower 2
28980 GB of the address space. Small symbols are also placed there. Symbols
28981 with sizes larger than @option{-mlarge-data-threshold} are put into
28982 large data or BSS sections and can be located above 2GB. Programs can
28983 be statically or dynamically linked.
28985 @item -mcmodel=large
28986 @opindex mcmodel=large
28987 Generate code for the large model. This model makes no assumptions
28988 about addresses and sizes of sections.
28990 @item -maddress-mode=long
28991 @opindex maddress-mode=long
28992 Generate code for long address mode. This is only supported for 64-bit
28993 and x32 environments. It is the default address mode for 64-bit
28996 @item -maddress-mode=short
28997 @opindex maddress-mode=short
28998 Generate code for short address mode. This is only supported for 32-bit
28999 and x32 environments. It is the default address mode for 32-bit and
29003 @node x86 Windows Options
29004 @subsection x86 Windows Options
29005 @cindex x86 Windows Options
29006 @cindex Windows Options for x86
29008 These additional options are available for Microsoft Windows targets:
29014 specifies that a console application is to be generated, by
29015 instructing the linker to set the PE header subsystem type
29016 required for console applications.
29017 This option is available for Cygwin and MinGW targets and is
29018 enabled by default on those targets.
29022 This option is available for Cygwin and MinGW targets. It
29023 specifies that a DLL---a dynamic link library---is to be
29024 generated, enabling the selection of the required runtime
29025 startup object and entry point.
29027 @item -mnop-fun-dllimport
29028 @opindex mnop-fun-dllimport
29029 This option is available for Cygwin and MinGW targets. It
29030 specifies that the @code{dllimport} attribute should be ignored.
29034 This option is available for MinGW targets. It specifies
29035 that MinGW-specific thread support is to be used.
29039 This option is available for MinGW-w64 targets. It causes
29040 the @code{UNICODE} preprocessor macro to be predefined, and
29041 chooses Unicode-capable runtime startup code.
29045 This option is available for Cygwin and MinGW targets. It
29046 specifies that the typical Microsoft Windows predefined macros are to
29047 be set in the pre-processor, but does not influence the choice
29048 of runtime library/startup code.
29052 This option is available for Cygwin and MinGW targets. It
29053 specifies that a GUI application is to be generated by
29054 instructing the linker to set the PE header subsystem type
29057 @item -fno-set-stack-executable
29058 @opindex fno-set-stack-executable
29059 @opindex fset-stack-executable
29060 This option is available for MinGW targets. It specifies that
29061 the executable flag for the stack used by nested functions isn't
29062 set. This is necessary for binaries running in kernel mode of
29063 Microsoft Windows, as there the User32 API, which is used to set executable
29064 privileges, isn't available.
29066 @item -fwritable-relocated-rdata
29067 @opindex fno-writable-relocated-rdata
29068 @opindex fwritable-relocated-rdata
29069 This option is available for MinGW and Cygwin targets. It specifies
29070 that relocated-data in read-only section is put into the @code{.data}
29071 section. This is a necessary for older runtimes not supporting
29072 modification of @code{.rdata} sections for pseudo-relocation.
29074 @item -mpe-aligned-commons
29075 @opindex mpe-aligned-commons
29076 This option is available for Cygwin and MinGW targets. It
29077 specifies that the GNU extension to the PE file format that
29078 permits the correct alignment of COMMON variables should be
29079 used when generating code. It is enabled by default if
29080 GCC detects that the target assembler found during configuration
29081 supports the feature.
29084 See also under @ref{x86 Options} for standard options.
29086 @node Xstormy16 Options
29087 @subsection Xstormy16 Options
29088 @cindex Xstormy16 Options
29090 These options are defined for Xstormy16:
29095 Choose startup files and linker script suitable for the simulator.
29098 @node Xtensa Options
29099 @subsection Xtensa Options
29100 @cindex Xtensa Options
29102 These options are supported for Xtensa targets:
29106 @itemx -mno-const16
29108 @opindex mno-const16
29109 Enable or disable use of @code{CONST16} instructions for loading
29110 constant values. The @code{CONST16} instruction is currently not a
29111 standard option from Tensilica. When enabled, @code{CONST16}
29112 instructions are always used in place of the standard @code{L32R}
29113 instructions. The use of @code{CONST16} is enabled by default only if
29114 the @code{L32R} instruction is not available.
29117 @itemx -mno-fused-madd
29118 @opindex mfused-madd
29119 @opindex mno-fused-madd
29120 Enable or disable use of fused multiply/add and multiply/subtract
29121 instructions in the floating-point option. This has no effect if the
29122 floating-point option is not also enabled. Disabling fused multiply/add
29123 and multiply/subtract instructions forces the compiler to use separate
29124 instructions for the multiply and add/subtract operations. This may be
29125 desirable in some cases where strict IEEE 754-compliant results are
29126 required: the fused multiply add/subtract instructions do not round the
29127 intermediate result, thereby producing results with @emph{more} bits of
29128 precision than specified by the IEEE standard. Disabling fused multiply
29129 add/subtract instructions also ensures that the program output is not
29130 sensitive to the compiler's ability to combine multiply and add/subtract
29133 @item -mserialize-volatile
29134 @itemx -mno-serialize-volatile
29135 @opindex mserialize-volatile
29136 @opindex mno-serialize-volatile
29137 When this option is enabled, GCC inserts @code{MEMW} instructions before
29138 @code{volatile} memory references to guarantee sequential consistency.
29139 The default is @option{-mserialize-volatile}. Use
29140 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
29142 @item -mforce-no-pic
29143 @opindex mforce-no-pic
29144 For targets, like GNU/Linux, where all user-mode Xtensa code must be
29145 position-independent code (PIC), this option disables PIC for compiling
29148 @item -mtext-section-literals
29149 @itemx -mno-text-section-literals
29150 @opindex mtext-section-literals
29151 @opindex mno-text-section-literals
29152 These options control the treatment of literal pools. The default is
29153 @option{-mno-text-section-literals}, which places literals in a separate
29154 section in the output file. This allows the literal pool to be placed
29155 in a data RAM/ROM, and it also allows the linker to combine literal
29156 pools from separate object files to remove redundant literals and
29157 improve code size. With @option{-mtext-section-literals}, the literals
29158 are interspersed in the text section in order to keep them as close as
29159 possible to their references. This may be necessary for large assembly
29160 files. Literals for each function are placed right before that function.
29162 @item -mauto-litpools
29163 @itemx -mno-auto-litpools
29164 @opindex mauto-litpools
29165 @opindex mno-auto-litpools
29166 These options control the treatment of literal pools. The default is
29167 @option{-mno-auto-litpools}, which places literals in a separate
29168 section in the output file unless @option{-mtext-section-literals} is
29169 used. With @option{-mauto-litpools} the literals are interspersed in
29170 the text section by the assembler. Compiler does not produce explicit
29171 @code{.literal} directives and loads literals into registers with
29172 @code{MOVI} instructions instead of @code{L32R} to let the assembler
29173 do relaxation and place literals as necessary. This option allows
29174 assembler to create several literal pools per function and assemble
29175 very big functions, which may not be possible with
29176 @option{-mtext-section-literals}.
29178 @item -mtarget-align
29179 @itemx -mno-target-align
29180 @opindex mtarget-align
29181 @opindex mno-target-align
29182 When this option is enabled, GCC instructs the assembler to
29183 automatically align instructions to reduce branch penalties at the
29184 expense of some code density. The assembler attempts to widen density
29185 instructions to align branch targets and the instructions following call
29186 instructions. If there are not enough preceding safe density
29187 instructions to align a target, no widening is performed. The
29188 default is @option{-mtarget-align}. These options do not affect the
29189 treatment of auto-aligned instructions like @code{LOOP}, which the
29190 assembler always aligns, either by widening density instructions or
29191 by inserting NOP instructions.
29194 @itemx -mno-longcalls
29195 @opindex mlongcalls
29196 @opindex mno-longcalls
29197 When this option is enabled, GCC instructs the assembler to translate
29198 direct calls to indirect calls unless it can determine that the target
29199 of a direct call is in the range allowed by the call instruction. This
29200 translation typically occurs for calls to functions in other source
29201 files. Specifically, the assembler translates a direct @code{CALL}
29202 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
29203 The default is @option{-mno-longcalls}. This option should be used in
29204 programs where the call target can potentially be out of range. This
29205 option is implemented in the assembler, not the compiler, so the
29206 assembly code generated by GCC still shows direct call
29207 instructions---look at the disassembled object code to see the actual
29208 instructions. Note that the assembler uses an indirect call for
29209 every cross-file call, not just those that really are out of range.
29212 @node zSeries Options
29213 @subsection zSeries Options
29214 @cindex zSeries options
29216 These are listed under @xref{S/390 and zSeries Options}.
29222 @section Specifying Subprocesses and the Switches to Pass to Them
29225 @command{gcc} is a driver program. It performs its job by invoking a
29226 sequence of other programs to do the work of compiling, assembling and
29227 linking. GCC interprets its command-line parameters and uses these to
29228 deduce which programs it should invoke, and which command-line options
29229 it ought to place on their command lines. This behavior is controlled
29230 by @dfn{spec strings}. In most cases there is one spec string for each
29231 program that GCC can invoke, but a few programs have multiple spec
29232 strings to control their behavior. The spec strings built into GCC can
29233 be overridden by using the @option{-specs=} command-line switch to specify
29236 @dfn{Spec files} are plain-text files that are used to construct spec
29237 strings. They consist of a sequence of directives separated by blank
29238 lines. The type of directive is determined by the first non-whitespace
29239 character on the line, which can be one of the following:
29242 @item %@var{command}
29243 Issues a @var{command} to the spec file processor. The commands that can
29247 @item %include <@var{file}>
29248 @cindex @code{%include}
29249 Search for @var{file} and insert its text at the current point in the
29252 @item %include_noerr <@var{file}>
29253 @cindex @code{%include_noerr}
29254 Just like @samp{%include}, but do not generate an error message if the include
29255 file cannot be found.
29257 @item %rename @var{old_name} @var{new_name}
29258 @cindex @code{%rename}
29259 Rename the spec string @var{old_name} to @var{new_name}.
29263 @item *[@var{spec_name}]:
29264 This tells the compiler to create, override or delete the named spec
29265 string. All lines after this directive up to the next directive or
29266 blank line are considered to be the text for the spec string. If this
29267 results in an empty string then the spec is deleted. (Or, if the
29268 spec did not exist, then nothing happens.) Otherwise, if the spec
29269 does not currently exist a new spec is created. If the spec does
29270 exist then its contents are overridden by the text of this
29271 directive, unless the first character of that text is the @samp{+}
29272 character, in which case the text is appended to the spec.
29274 @item [@var{suffix}]:
29275 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
29276 and up to the next directive or blank line are considered to make up the
29277 spec string for the indicated suffix. When the compiler encounters an
29278 input file with the named suffix, it processes the spec string in
29279 order to work out how to compile that file. For example:
29283 z-compile -input %i
29286 This says that any input file whose name ends in @samp{.ZZ} should be
29287 passed to the program @samp{z-compile}, which should be invoked with the
29288 command-line switch @option{-input} and with the result of performing the
29289 @samp{%i} substitution. (See below.)
29291 As an alternative to providing a spec string, the text following a
29292 suffix directive can be one of the following:
29295 @item @@@var{language}
29296 This says that the suffix is an alias for a known @var{language}. This is
29297 similar to using the @option{-x} command-line switch to GCC to specify a
29298 language explicitly. For example:
29305 Says that .ZZ files are, in fact, C++ source files.
29308 This causes an error messages saying:
29311 @var{name} compiler not installed on this system.
29315 GCC already has an extensive list of suffixes built into it.
29316 This directive adds an entry to the end of the list of suffixes, but
29317 since the list is searched from the end backwards, it is effectively
29318 possible to override earlier entries using this technique.
29322 GCC has the following spec strings built into it. Spec files can
29323 override these strings or create their own. Note that individual
29324 targets can also add their own spec strings to this list.
29327 asm Options to pass to the assembler
29328 asm_final Options to pass to the assembler post-processor
29329 cpp Options to pass to the C preprocessor
29330 cc1 Options to pass to the C compiler
29331 cc1plus Options to pass to the C++ compiler
29332 endfile Object files to include at the end of the link
29333 link Options to pass to the linker
29334 lib Libraries to include on the command line to the linker
29335 libgcc Decides which GCC support library to pass to the linker
29336 linker Sets the name of the linker
29337 predefines Defines to be passed to the C preprocessor
29338 signed_char Defines to pass to CPP to say whether @code{char} is signed
29340 startfile Object files to include at the start of the link
29343 Here is a small example of a spec file:
29346 %rename lib old_lib
29349 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
29352 This example renames the spec called @samp{lib} to @samp{old_lib} and
29353 then overrides the previous definition of @samp{lib} with a new one.
29354 The new definition adds in some extra command-line options before
29355 including the text of the old definition.
29357 @dfn{Spec strings} are a list of command-line options to be passed to their
29358 corresponding program. In addition, the spec strings can contain
29359 @samp{%}-prefixed sequences to substitute variable text or to
29360 conditionally insert text into the command line. Using these constructs
29361 it is possible to generate quite complex command lines.
29363 Here is a table of all defined @samp{%}-sequences for spec
29364 strings. Note that spaces are not generated automatically around the
29365 results of expanding these sequences. Therefore you can concatenate them
29366 together or combine them with constant text in a single argument.
29370 Substitute one @samp{%} into the program name or argument.
29373 Substitute the name of the input file being processed.
29376 Substitute the basename of the input file being processed.
29377 This is the substring up to (and not including) the last period
29378 and not including the directory.
29381 This is the same as @samp{%b}, but include the file suffix (text after
29385 Marks the argument containing or following the @samp{%d} as a
29386 temporary file name, so that that file is deleted if GCC exits
29387 successfully. Unlike @samp{%g}, this contributes no text to the
29390 @item %g@var{suffix}
29391 Substitute a file name that has suffix @var{suffix} and is chosen
29392 once per compilation, and mark the argument in the same way as
29393 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
29394 name is now chosen in a way that is hard to predict even when previously
29395 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
29396 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
29397 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
29398 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
29399 was simply substituted with a file name chosen once per compilation,
29400 without regard to any appended suffix (which was therefore treated
29401 just like ordinary text), making such attacks more likely to succeed.
29403 @item %u@var{suffix}
29404 Like @samp{%g}, but generates a new temporary file name
29405 each time it appears instead of once per compilation.
29407 @item %U@var{suffix}
29408 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
29409 new one if there is no such last file name. In the absence of any
29410 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
29411 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
29412 involves the generation of two distinct file names, one
29413 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
29414 simply substituted with a file name chosen for the previous @samp{%u},
29415 without regard to any appended suffix.
29417 @item %j@var{suffix}
29418 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
29419 writable, and if @option{-save-temps} is not used;
29420 otherwise, substitute the name
29421 of a temporary file, just like @samp{%u}. This temporary file is not
29422 meant for communication between processes, but rather as a junk
29423 disposal mechanism.
29425 @item %|@var{suffix}
29426 @itemx %m@var{suffix}
29427 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
29428 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
29429 all. These are the two most common ways to instruct a program that it
29430 should read from standard input or write to standard output. If you
29431 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
29432 construct: see for example @file{gcc/fortran/lang-specs.h}.
29434 @item %.@var{SUFFIX}
29435 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
29436 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
29437 terminated by the next space or %.
29440 Marks the argument containing or following the @samp{%w} as the
29441 designated output file of this compilation. This puts the argument
29442 into the sequence of arguments that @samp{%o} substitutes.
29445 Substitutes the names of all the output files, with spaces
29446 automatically placed around them. You should write spaces
29447 around the @samp{%o} as well or the results are undefined.
29448 @samp{%o} is for use in the specs for running the linker.
29449 Input files whose names have no recognized suffix are not compiled
29450 at all, but they are included among the output files, so they are
29454 Substitutes the suffix for object files. Note that this is
29455 handled specially when it immediately follows @samp{%g, %u, or %U},
29456 because of the need for those to form complete file names. The
29457 handling is such that @samp{%O} is treated exactly as if it had already
29458 been substituted, except that @samp{%g, %u, and %U} do not currently
29459 support additional @var{suffix} characters following @samp{%O} as they do
29460 following, for example, @samp{.o}.
29463 Substitutes the standard macro predefinitions for the
29464 current target machine. Use this when running @command{cpp}.
29467 Like @samp{%p}, but puts @samp{__} before and after the name of each
29468 predefined macro, except for macros that start with @samp{__} or with
29469 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
29473 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
29474 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
29475 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
29476 and @option{-imultilib} as necessary.
29479 Current argument is the name of a library or startup file of some sort.
29480 Search for that file in a standard list of directories and substitute
29481 the full name found. The current working directory is included in the
29482 list of directories scanned.
29485 Current argument is the name of a linker script. Search for that file
29486 in the current list of directories to scan for libraries. If the file
29487 is located insert a @option{--script} option into the command line
29488 followed by the full path name found. If the file is not found then
29489 generate an error message. Note: the current working directory is not
29493 Print @var{str} as an error message. @var{str} is terminated by a newline.
29494 Use this when inconsistent options are detected.
29496 @item %(@var{name})
29497 Substitute the contents of spec string @var{name} at this point.
29499 @item %x@{@var{option}@}
29500 Accumulate an option for @samp{%X}.
29503 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
29507 Output the accumulated assembler options specified by @option{-Wa}.
29510 Output the accumulated preprocessor options specified by @option{-Wp}.
29513 Process the @code{asm} spec. This is used to compute the
29514 switches to be passed to the assembler.
29517 Process the @code{asm_final} spec. This is a spec string for
29518 passing switches to an assembler post-processor, if such a program is
29522 Process the @code{link} spec. This is the spec for computing the
29523 command line passed to the linker. Typically it makes use of the
29524 @samp{%L %G %S %D and %E} sequences.
29527 Dump out a @option{-L} option for each directory that GCC believes might
29528 contain startup files. If the target supports multilibs then the
29529 current multilib directory is prepended to each of these paths.
29532 Process the @code{lib} spec. This is a spec string for deciding which
29533 libraries are included on the command line to the linker.
29536 Process the @code{libgcc} spec. This is a spec string for deciding
29537 which GCC support library is included on the command line to the linker.
29540 Process the @code{startfile} spec. This is a spec for deciding which
29541 object files are the first ones passed to the linker. Typically
29542 this might be a file named @file{crt0.o}.
29545 Process the @code{endfile} spec. This is a spec string that specifies
29546 the last object files that are passed to the linker.
29549 Process the @code{cpp} spec. This is used to construct the arguments
29550 to be passed to the C preprocessor.
29553 Process the @code{cc1} spec. This is used to construct the options to be
29554 passed to the actual C compiler (@command{cc1}).
29557 Process the @code{cc1plus} spec. This is used to construct the options to be
29558 passed to the actual C++ compiler (@command{cc1plus}).
29561 Substitute the variable part of a matched option. See below.
29562 Note that each comma in the substituted string is replaced by
29566 Remove all occurrences of @code{-S} from the command line. Note---this
29567 command is position dependent. @samp{%} commands in the spec string
29568 before this one see @code{-S}, @samp{%} commands in the spec string
29569 after this one do not.
29571 @item %:@var{function}(@var{args})
29572 Call the named function @var{function}, passing it @var{args}.
29573 @var{args} is first processed as a nested spec string, then split
29574 into an argument vector in the usual fashion. The function returns
29575 a string which is processed as if it had appeared literally as part
29576 of the current spec.
29578 The following built-in spec functions are provided:
29581 @item @code{getenv}
29582 The @code{getenv} spec function takes two arguments: an environment
29583 variable name and a string. If the environment variable is not
29584 defined, a fatal error is issued. Otherwise, the return value is the
29585 value of the environment variable concatenated with the string. For
29586 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
29589 %:getenv(TOPDIR /include)
29592 expands to @file{/path/to/top/include}.
29594 @item @code{if-exists}
29595 The @code{if-exists} spec function takes one argument, an absolute
29596 pathname to a file. If the file exists, @code{if-exists} returns the
29597 pathname. Here is a small example of its usage:
29601 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
29604 @item @code{if-exists-else}
29605 The @code{if-exists-else} spec function is similar to the @code{if-exists}
29606 spec function, except that it takes two arguments. The first argument is
29607 an absolute pathname to a file. If the file exists, @code{if-exists-else}
29608 returns the pathname. If it does not exist, it returns the second argument.
29609 This way, @code{if-exists-else} can be used to select one file or another,
29610 based on the existence of the first. Here is a small example of its usage:
29614 crt0%O%s %:if-exists(crti%O%s) \
29615 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
29618 @item @code{replace-outfile}
29619 The @code{replace-outfile} spec function takes two arguments. It looks for the
29620 first argument in the outfiles array and replaces it with the second argument. Here
29621 is a small example of its usage:
29624 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
29627 @item @code{remove-outfile}
29628 The @code{remove-outfile} spec function takes one argument. It looks for the
29629 first argument in the outfiles array and removes it. Here is a small example
29633 %:remove-outfile(-lm)
29636 @item @code{pass-through-libs}
29637 The @code{pass-through-libs} spec function takes any number of arguments. It
29638 finds any @option{-l} options and any non-options ending in @file{.a} (which it
29639 assumes are the names of linker input library archive files) and returns a
29640 result containing all the found arguments each prepended by
29641 @option{-plugin-opt=-pass-through=} and joined by spaces. This list is
29642 intended to be passed to the LTO linker plugin.
29645 %:pass-through-libs(%G %L %G)
29648 @item @code{print-asm-header}
29649 The @code{print-asm-header} function takes no arguments and simply
29650 prints a banner like:
29656 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
29659 It is used to separate compiler options from assembler options
29660 in the @option{--target-help} output.
29664 Substitutes the @code{-S} switch, if that switch is given to GCC@.
29665 If that switch is not specified, this substitutes nothing. Note that
29666 the leading dash is omitted when specifying this option, and it is
29667 automatically inserted if the substitution is performed. Thus the spec
29668 string @samp{%@{foo@}} matches the command-line option @option{-foo}
29669 and outputs the command-line option @option{-foo}.
29672 Like %@{@code{S}@} but mark last argument supplied within as a file to be
29673 deleted on failure.
29676 Substitutes all the switches specified to GCC whose names start
29677 with @code{-S}, but which also take an argument. This is used for
29678 switches like @option{-o}, @option{-D}, @option{-I}, etc.
29679 GCC considers @option{-o foo} as being
29680 one switch whose name starts with @samp{o}. %@{o*@} substitutes this
29681 text, including the space. Thus two arguments are generated.
29684 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
29685 (the order of @code{S} and @code{T} in the spec is not significant).
29686 There can be any number of ampersand-separated variables; for each the
29687 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
29690 Substitutes @code{X}, if the @option{-S} switch is given to GCC@.
29693 Substitutes @code{X}, if the @option{-S} switch is @emph{not} given to GCC@.
29696 Substitutes @code{X} if one or more switches whose names start with
29697 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
29698 once, no matter how many such switches appeared. However, if @code{%*}
29699 appears somewhere in @code{X}, then @code{X} is substituted once
29700 for each matching switch, with the @code{%*} replaced by the part of
29701 that switch matching the @code{*}.
29703 If @code{%*} appears as the last part of a spec sequence then a space
29704 is added after the end of the last substitution. If there is more
29705 text in the sequence, however, then a space is not generated. This
29706 allows the @code{%*} substitution to be used as part of a larger
29707 string. For example, a spec string like this:
29710 %@{mcu=*:--script=%*/memory.ld@}
29714 when matching an option like @option{-mcu=newchip} produces:
29717 --script=newchip/memory.ld
29721 Substitutes @code{X}, if processing a file with suffix @code{S}.
29724 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
29727 Substitutes @code{X}, if processing a file for language @code{S}.
29730 Substitutes @code{X}, if not processing a file for language @code{S}.
29733 Substitutes @code{X} if either @code{-S} or @code{-P} is given to
29734 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
29735 @code{*} sequences as well, although they have a stronger binding than
29736 the @samp{|}. If @code{%*} appears in @code{X}, all of the
29737 alternatives must be starred, and only the first matching alternative
29740 For example, a spec string like this:
29743 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
29747 outputs the following command-line options from the following input
29748 command-line options:
29753 -d fred.c -foo -baz -boggle
29754 -d jim.d -bar -baz -boggle
29757 @item %@{S:X; T:Y; :D@}
29759 If @code{S} is given to GCC, substitutes @code{X}; else if @code{T} is
29760 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
29761 be as many clauses as you need. This may be combined with @code{.},
29762 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
29767 The switch matching text @code{S} in a @samp{%@{S@}}, @samp{%@{S:X@}}
29768 or similar construct can use a backslash to ignore the special meaning
29769 of the character following it, thus allowing literal matching of a
29770 character that is otherwise specially treated. For example,
29771 @samp{%@{std=iso9899\:1999:X@}} substitutes @code{X} if the
29772 @option{-std=iso9899:1999} option is given.
29774 The conditional text @code{X} in a @samp{%@{S:X@}} or similar
29775 construct may contain other nested @samp{%} constructs or spaces, or
29776 even newlines. They are processed as usual, as described above.
29777 Trailing white space in @code{X} is ignored. White space may also
29778 appear anywhere on the left side of the colon in these constructs,
29779 except between @code{.} or @code{*} and the corresponding word.
29781 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
29782 handled specifically in these constructs. If another value of
29783 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
29784 @option{-W} switch is found later in the command line, the earlier
29785 switch value is ignored, except with @{@code{S}*@} where @code{S} is
29786 just one letter, which passes all matching options.
29788 The character @samp{|} at the beginning of the predicate text is used to
29789 indicate that a command should be piped to the following command, but
29790 only if @option{-pipe} is specified.
29792 It is built into GCC which switches take arguments and which do not.
29793 (You might think it would be useful to generalize this to allow each
29794 compiler's spec to say which switches take arguments. But this cannot
29795 be done in a consistent fashion. GCC cannot even decide which input
29796 files have been specified without knowing which switches take arguments,
29797 and it must know which input files to compile in order to tell which
29800 GCC also knows implicitly that arguments starting in @option{-l} are to be
29801 treated as compiler output files, and passed to the linker in their
29802 proper position among the other output files.
29804 @node Environment Variables
29805 @section Environment Variables Affecting GCC
29806 @cindex environment variables
29808 @c man begin ENVIRONMENT
29809 This section describes several environment variables that affect how GCC
29810 operates. Some of them work by specifying directories or prefixes to use
29811 when searching for various kinds of files. Some are used to specify other
29812 aspects of the compilation environment.
29814 Note that you can also specify places to search using options such as
29815 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
29816 take precedence over places specified using environment variables, which
29817 in turn take precedence over those specified by the configuration of GCC@.
29818 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
29819 GNU Compiler Collection (GCC) Internals}.
29824 @c @itemx LC_COLLATE
29826 @c @itemx LC_MONETARY
29827 @c @itemx LC_NUMERIC
29832 @c @findex LC_COLLATE
29833 @findex LC_MESSAGES
29834 @c @findex LC_MONETARY
29835 @c @findex LC_NUMERIC
29839 These environment variables control the way that GCC uses
29840 localization information which allows GCC to work with different
29841 national conventions. GCC inspects the locale categories
29842 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
29843 so. These locale categories can be set to any value supported by your
29844 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
29845 Kingdom encoded in UTF-8.
29847 The @env{LC_CTYPE} environment variable specifies character
29848 classification. GCC uses it to determine the character boundaries in
29849 a string; this is needed for some multibyte encodings that contain quote
29850 and escape characters that are otherwise interpreted as a string
29853 The @env{LC_MESSAGES} environment variable specifies the language to
29854 use in diagnostic messages.
29856 If the @env{LC_ALL} environment variable is set, it overrides the value
29857 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
29858 and @env{LC_MESSAGES} default to the value of the @env{LANG}
29859 environment variable. If none of these variables are set, GCC
29860 defaults to traditional C English behavior.
29864 If @env{TMPDIR} is set, it specifies the directory to use for temporary
29865 files. GCC uses temporary files to hold the output of one stage of
29866 compilation which is to be used as input to the next stage: for example,
29867 the output of the preprocessor, which is the input to the compiler
29870 @item GCC_COMPARE_DEBUG
29871 @findex GCC_COMPARE_DEBUG
29872 Setting @env{GCC_COMPARE_DEBUG} is nearly equivalent to passing
29873 @option{-fcompare-debug} to the compiler driver. See the documentation
29874 of this option for more details.
29876 @item GCC_EXEC_PREFIX
29877 @findex GCC_EXEC_PREFIX
29878 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
29879 names of the subprograms executed by the compiler. No slash is added
29880 when this prefix is combined with the name of a subprogram, but you can
29881 specify a prefix that ends with a slash if you wish.
29883 If @env{GCC_EXEC_PREFIX} is not set, GCC attempts to figure out
29884 an appropriate prefix to use based on the pathname it is invoked with.
29886 If GCC cannot find the subprogram using the specified prefix, it
29887 tries looking in the usual places for the subprogram.
29889 The default value of @env{GCC_EXEC_PREFIX} is
29890 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
29891 the installed compiler. In many cases @var{prefix} is the value
29892 of @code{prefix} when you ran the @file{configure} script.
29894 Other prefixes specified with @option{-B} take precedence over this prefix.
29896 This prefix is also used for finding files such as @file{crt0.o} that are
29899 In addition, the prefix is used in an unusual way in finding the
29900 directories to search for header files. For each of the standard
29901 directories whose name normally begins with @samp{/usr/local/lib/gcc}
29902 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
29903 replacing that beginning with the specified prefix to produce an
29904 alternate directory name. Thus, with @option{-Bfoo/}, GCC searches
29905 @file{foo/bar} just before it searches the standard directory
29906 @file{/usr/local/lib/bar}.
29907 If a standard directory begins with the configured
29908 @var{prefix} then the value of @var{prefix} is replaced by
29909 @env{GCC_EXEC_PREFIX} when looking for header files.
29911 @item COMPILER_PATH
29912 @findex COMPILER_PATH
29913 The value of @env{COMPILER_PATH} is a colon-separated list of
29914 directories, much like @env{PATH}. GCC tries the directories thus
29915 specified when searching for subprograms, if it cannot find the
29916 subprograms using @env{GCC_EXEC_PREFIX}.
29919 @findex LIBRARY_PATH
29920 The value of @env{LIBRARY_PATH} is a colon-separated list of
29921 directories, much like @env{PATH}. When configured as a native compiler,
29922 GCC tries the directories thus specified when searching for special
29923 linker files, if it cannot find them using @env{GCC_EXEC_PREFIX}. Linking
29924 using GCC also uses these directories when searching for ordinary
29925 libraries for the @option{-l} option (but directories specified with
29926 @option{-L} come first).
29930 @cindex locale definition
29931 This variable is used to pass locale information to the compiler. One way in
29932 which this information is used is to determine the character set to be used
29933 when character literals, string literals and comments are parsed in C and C++.
29934 When the compiler is configured to allow multibyte characters,
29935 the following values for @env{LANG} are recognized:
29939 Recognize JIS characters.
29941 Recognize SJIS characters.
29943 Recognize EUCJP characters.
29946 If @env{LANG} is not defined, or if it has some other value, then the
29947 compiler uses @code{mblen} and @code{mbtowc} as defined by the default locale to
29948 recognize and translate multibyte characters.
29952 Some additional environment variables affect the behavior of the
29955 @include cppenv.texi
29959 @node Precompiled Headers
29960 @section Using Precompiled Headers
29961 @cindex precompiled headers
29962 @cindex speed of compilation
29964 Often large projects have many header files that are included in every
29965 source file. The time the compiler takes to process these header files
29966 over and over again can account for nearly all of the time required to
29967 build the project. To make builds faster, GCC allows you to
29968 @dfn{precompile} a header file.
29970 To create a precompiled header file, simply compile it as you would any
29971 other file, if necessary using the @option{-x} option to make the driver
29972 treat it as a C or C++ header file. You may want to use a
29973 tool like @command{make} to keep the precompiled header up-to-date when
29974 the headers it contains change.
29976 A precompiled header file is searched for when @code{#include} is
29977 seen in the compilation. As it searches for the included file
29978 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
29979 compiler looks for a precompiled header in each directory just before it
29980 looks for the include file in that directory. The name searched for is
29981 the name specified in the @code{#include} with @samp{.gch} appended. If
29982 the precompiled header file cannot be used, it is ignored.
29984 For instance, if you have @code{#include "all.h"}, and you have
29985 @file{all.h.gch} in the same directory as @file{all.h}, then the
29986 precompiled header file is used if possible, and the original
29987 header is used otherwise.
29989 Alternatively, you might decide to put the precompiled header file in a
29990 directory and use @option{-I} to ensure that directory is searched
29991 before (or instead of) the directory containing the original header.
29992 Then, if you want to check that the precompiled header file is always
29993 used, you can put a file of the same name as the original header in this
29994 directory containing an @code{#error} command.
29996 This also works with @option{-include}. So yet another way to use
29997 precompiled headers, good for projects not designed with precompiled
29998 header files in mind, is to simply take most of the header files used by
29999 a project, include them from another header file, precompile that header
30000 file, and @option{-include} the precompiled header. If the header files
30001 have guards against multiple inclusion, they are skipped because
30002 they've already been included (in the precompiled header).
30004 If you need to precompile the same header file for different
30005 languages, targets, or compiler options, you can instead make a
30006 @emph{directory} named like @file{all.h.gch}, and put each precompiled
30007 header in the directory, perhaps using @option{-o}. It doesn't matter
30008 what you call the files in the directory; every precompiled header in
30009 the directory is considered. The first precompiled header
30010 encountered in the directory that is valid for this compilation is
30011 used; they're searched in no particular order.
30013 There are many other possibilities, limited only by your imagination,
30014 good sense, and the constraints of your build system.
30016 A precompiled header file can be used only when these conditions apply:
30020 Only one precompiled header can be used in a particular compilation.
30023 A precompiled header cannot be used once the first C token is seen. You
30024 can have preprocessor directives before a precompiled header; you cannot
30025 include a precompiled header from inside another header.
30028 The precompiled header file must be produced for the same language as
30029 the current compilation. You cannot use a C precompiled header for a C++
30033 The precompiled header file must have been produced by the same compiler
30034 binary as the current compilation is using.
30037 Any macros defined before the precompiled header is included must
30038 either be defined in the same way as when the precompiled header was
30039 generated, or must not affect the precompiled header, which usually
30040 means that they don't appear in the precompiled header at all.
30042 The @option{-D} option is one way to define a macro before a
30043 precompiled header is included; using a @code{#define} can also do it.
30044 There are also some options that define macros implicitly, like
30045 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
30048 @item If debugging information is output when using the precompiled
30049 header, using @option{-g} or similar, the same kind of debugging information
30050 must have been output when building the precompiled header. However,
30051 a precompiled header built using @option{-g} can be used in a compilation
30052 when no debugging information is being output.
30054 @item The same @option{-m} options must generally be used when building
30055 and using the precompiled header. @xref{Submodel Options},
30056 for any cases where this rule is relaxed.
30058 @item Each of the following options must be the same when building and using
30059 the precompiled header:
30061 @gccoptlist{-fexceptions}
30064 Some other command-line options starting with @option{-f},
30065 @option{-p}, or @option{-O} must be defined in the same way as when
30066 the precompiled header was generated. At present, it's not clear
30067 which options are safe to change and which are not; the safest choice
30068 is to use exactly the same options when generating and using the
30069 precompiled header. The following are known to be safe:
30071 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
30072 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
30073 -fsched-verbose=@var{number} -fschedule-insns -fvisibility= @gol
30078 For all of these except the last, the compiler automatically
30079 ignores the precompiled header if the conditions aren't met. If you
30080 find an option combination that doesn't work and doesn't cause the
30081 precompiled header to be ignored, please consider filing a bug report,
30084 If you do use differing options when generating and using the
30085 precompiled header, the actual behavior is a mixture of the
30086 behavior for the options. For instance, if you use @option{-g} to
30087 generate the precompiled header but not when using it, you may or may
30088 not get debugging information for routines in the precompiled header.